Sunteți pe pagina 1din 13

Experimental Thermal and Fluid Science 20 (1999) 6678

www.elsevier.nl/locate/etfs

Flow visualization of a round jet impinging on cylindrical surfaces


C. Cornaro 1, A.S. Fleischer, R.J. Goldstein *

Department of Mechanical Engineering, Institute of Technology, University of Minnesota, Bld. 125, 111, Church Street S.E., Minneapolis,
MN 55455-0111, USA
Received 15 December 1998; received in revised form 28 June 1999; accepted 19 July 1999

Abstract

Smoke wire ow visualization is used to investigate the behavior of a round jet issuing from a straight tube and impinging on
concave and convex surfaces with high relative curvature values. Local velocity and turbulence intensity measurements of the free jet
are correlated to the formation of the visualized ordered-ow structures. Visualization of the impinging jet shows the initiation and
growth of ring vortices in the jet shear layer and their interaction with the cylindrical surfaces. Eects of relative curvature, nozzle
diameter, nozzle-to-surface distance and Reynolds number on the jet ow structure are described. 1999 Elsevier Science Inc. All
rights reserved.

Keywords: Jets; Jet impingement; Vortex; Ring vortex; Curved surface; Flow visualization

1. Introduction described in terms of a at surface. For this reason, the


eect of surface geometry on impingement cooling
The wide use of jets to enhance heat transfer has led warrants investigation. Yet few works on jet impinge-
to many studies of the ow and heat transfer charac- ment on curved surfaces have been published.
teristics of jet impingement. Jet impingement is used as a Some of the earliest works that do focus on jet im-
cooling method in many applications ranging from pingement on curved surfaces are Chupp et al. [4], who
textiles to electronic components, and the eects of jet studied heat transfer from an array of small round jets
diameter, orientation, Reynolds number, and jet-surface impinging on a concave surface, and Metzger et al. [5]
spacing on ow and heat transfer on at plates have who looked at a similar geometry. Tabako and Cle-
been thoroughly investigated. Comprehensive reviews venger [6] compared the cooling of a single slot jet to
have been published by Jambunathan et al. [1] and arrays of round jets on a concave surface.
Viskanta [2]. Popiel and Trass [3] report the results of Hrycak [7] studied the heat transfer from concave
ow visualization of a round free jet and of a jet im- hemispherical plates with Reynolds numbers of 12,000
pinging on a at plate. The entrainment process, due to to 88,000 and small relative curvature values
the ring vortices, was studied using a smoke wire (d=D 0.0340.1). His results indicated that at the
method. Clear details of the ow pattern and of the stagnation point, the heat transfer was higher on the
vortex structures are presented. concave plate than for an equivalent jet impinging on a
Although many jet impingement studies have been at plate.
reported, most investigate the eect of impingement on Yang et al. [8] studied heat transfer on semicylindri-
at surfaces. However, a wide variety of components, cal concave surfaces. Gau and Chung [9] investigated
including turbine blades, which can benet from im- the heat transfer from semicylindrical concave and
pingement cooling have a geometry which cannot be convex surfaces with slot jets at Reynolds numbers from
6000 to 35,000 and slot width to surface diameter ratios
of 845. Correlations for stagnation point and average
* Nusselt number are presented. Their results indicate that
Corresponding author. Tel. : +1-612-625-5552; fax: +1-612-625-
3434.
heat transfer is enhanced with increasing relative cur-
E-mail address: rjg@me.umn.edu (R.J. Goldstein) vature (d=D) for both convex and concave surfaces, and
1
Currently at Department of Mechanical Engineering, University in both cases is higher than for impingement on a at
of Rome ``Tor Vergata'', Via di Tor Vergata, 00133 Rome, Italy. plate under the same conditions. Flow visualization is

0894-1777/99/$ - see front matter 1999 Elsevier Science Inc. All rights reserved.
PII: S 0 8 9 4 - 1 7 7 7 ( 9 9 ) 0 0 0 3 2 - 1
C. Cornaro et al. / Experimental Thermal and Fluid Science 20 (1999) 6678 67

presented, but interpretation of the ow pattern is


complicated by the resolution of the presented pictures.
The authors do suggest that counter-rotating vortices
are formed at the stagnation point.
Lee et al. [10] studied the eects of curvature on local
heat transfer from a round jet with Reynolds numbers of
11,00050,000 and relative curvature values from 0.034
to 0.089. The impingement surface was a convex hemi-
sphere. Their results showed that the stagnation point
Nusselt number increases with increasing values of d=D.
Stagnation point and average Nusselt number correla-
tions are presented. Fig. 1. Experimental apparatus.
Kornblum and Goldstein [11] examined the ow of
circular jets impinging on concave and convex semicyl-
indrical surfaces for small d=D (0.01970.0394). Their the pipe entrance to smooth the ow shorten the length
ow visualization results, while not detailed enough to of the tube used to develop the boundary layer. The
show the motion of the ring vortices, indicate a sub- actual L=d is 8.38 for d 47.2 mm, 8.95 for d 72.6 mm
stantial recirculation of the ow exiting the concave and 9.22 for d 98.6 mm. A thermocouple in the tube,
surface into the main jet ow. approximately 10 cm from the jet exit, measures the jet
These studies, while providing valuable information air temperature.
on the eects of relative curvature on jet impingement, The jet impinges on a curved surface which consists
were all conducted at low values of relative curvature. of a convex or concave semicylinder of 252 mm inner
The present study was undertaken to provide insight diameter and 256.76 mm outer diameter.
into the jet vortex structure during impingement on The average velocity at the jet exit is calculated from
convex and concave surfaces with high relative curva- the air ow rate measured upstream of the plenum by an
ture (d=D 0.180.38). A smoke wire technique similar orice plate and manometer. Local velocity distribu-
to that of Corke et al. [12] is used. For completeness and tions of the free jet are measured using a hot wire ane-
contrast, visualizations of a free jet and of a jet im- mometer. The radial distribution of the axial velocity is
pinging on a at surface are also presented. measured approximately 5 mm downstream of the jet
The jet issues from straight tubes to impinge normal exit. The output signal is ltered at 5 kHz and sampled
to at plates and semicylindrical convex and concave with a frequency of 10 kHz for the radial distribution.
surfaces. The experiment is conducted at four dierent The axial velocity distribution is determined by data
tube Reynolds numbers (6000, 10,000, 15,000, and sampled at 4 kHz with a low pass ltering at 2 kHz. The
20,000) for jet exit-to-surface spacing from 1 to 4 tube data sampling rate was modied for the axial velocity
diameters. Jet centerline velocities varied from 4 to 6 m/ distribution to correct for an external disturbance at the
s. The relative curvature is varied by changing the tube sampling frequency.
diameter. Tube diameters of 47.2, 72.6 and 98.6 mm A smoke wire technique is used to visualize the ow
used with an impingement surface diameter of 252 mm structures of the jets. A 0.1 mm diameter stainless steel
result in a relative curvature range of 0.180.39. wire is located approximately 1 mm downstream of the
Local radial distribution of the axial velocity and jet exit. Before each test the wire is coated with mineral
turbulence intensity distributions at the jet exit and of oil. Small evenly spaced droplets form on its surface. A
axial velocity along the jet axis are measured and cor- power supply provides the energy necessary to heat the
related to the observed ow structures. The eects of wire by the Joule eect. The heated oil droplets pro-
surface curvature, nozzle diameter, nozzle-to-surface duce thin white smoke laments that trace the air ow
spacing, Reynolds number and turbulence intensity issuing from the tubes. A black background located
level on the ow behavior and vortex formation are behind the jet improves the contrast for the visualiza-
analyzed. tion. The curved surface is painted black for the same
reason.
A record of the resulting ow eld is captured using a
2. Experimental apparatus Nikon FG camera with 1:2.8 105 mm AF Micro Nikkor
Lens. A powerful short duration ash is essential to get
The air jet is produced by the apparatus shown in satisfactory pictures. A Nikon SB-15 ash is used for
Fig. 1. The air is supplied by the building compressor this experiment. The camera shutter release drives the
and controlled by a pressure regulator. Upstream of the ash light while the power supply to the wire and the
jet, the air passes through a plenum and a exible tubing shutter release are synchronized manually. All signi-
of 25.4 cm diameter that allows movement of the tube in cant data, such as vortex size and frequency, is scaled o
order to vary the jet exit-to-surface distance. Three tubes the photos. Because the photos capture only an instant
of dierent diameters (47.2, 72.6 and 98.6 mm) are used. in time, a signicant numbers of photos were taken and
The tube length to diameter ratio (L=d) is 10 for all the the data ranges presented cover the extremes of the
tubes, however honeycomb and mesh screens inserted in photographic information.
68 C. Cornaro et al. / Experimental Thermal and Fluid Science 20 (1999) 6678

3. Flow characterization presence of organized ow structures observed within


the ow.
The velocity and turbulence intensity distributions at The radial distributions of the axial velocity and
the jet exit and along the jet axis are necessary to un- relative turbulence intensities for the 47.2 mm jet are
derstand the dynamics of the ow and to explain the shown in Fig. 2. The turbulence intensity is presented as
behavior of the jet far downstream. The ow dynamics the percentage of the ratio of the RMS of the uctuating
also inuence the heat transfer from the impingement component if the local velocity over the centerline ve-
surface. In general, a characterization of the ow is re- locity at the jet exit. The values of the turbulence in-
quired to understand the heat transfer mechanisms. As tensity at both the jet center and in the shear layer at the
will be shown later, these velocity and turbulence in- jet edge are presented in Table 1. Figs. 3 and 4 present
tensity proles at the jet exit can be related to the the radial distributions of the axial velocity and turbu-
lence intensity levels for the 72.6 mm jet and the 98.6
mm jet. As can be seen from these velocity proles, the
ow is symmetric and free of disturbances at the exit of
the jet.
The axial velocity distribution was obtained by po-
sitioning the probe at the centerline of the jet exit and
measuring the velocity at various distances away from
the jet exit. Due to laboratory space constraints the axial
velocity prole for the 98.6 mm jet was not measured.
Fig. 5 shows the axial velocity prole for the 47.2 mm
jet.
The potential core region of the jet can be determined
from the axial velocity distribution using the intersection
of a straight line t to each distinct slope of the prole.
The potential core length increases as the Reynolds
number is increased. (See Table 2.)
For the 47.2 mm jet, a nonuniformity with x in the
velocity distribution is observed at Re 6000 (see Fig.
5a). No nonuniformity is observed at Re 10,000 (see
Fig. 5b), 15,000 or 20,000. Similar nonuniformities (not
shown) are observed for the 72.6 mm jet at Re 6000
and 10,000, but not at 15,000 or 20,000. These nonuni-
formities are consistently reproduced for several runs at
the same Reynolds number and may be explained by the
presence of large vortex structures that perturb the jet
centerline at low Reynolds numbers. As will be seen
from the ow visualization, as the Reynolds number is
increased, the inuence of the ring vortices on the center
of the jet is reduced.

4. Free jet

A free jet was visualized to verify the imaging pro-


cedure and to use as baseline information for later
comparison to the behavior of the impinging jets. The
most signicant aspects of the free jet behavior are a
distinct Reynolds number eect and a radial contrac-
tion/expansion of the potential core region due to vortex
Fig. 2. Radial distribution of: (a) axial velocity; (b) turbulence inten- entrainment. Interestingly, dierent ow situations ap-
sity for 47.2 mm jet (x 5 mm). pear to occur for the same Reynolds number developed

Table 1
Turbulence intensity levels in jet central region/shear layer at jet edge
Jet diameter (mm) Re 6000 Re 10,000 Re 15,000 Re 20,000
47.2 0.5%/9% 0.5%/14% 1%/14% 1%/13%
72.6 0.15%/0.6% 0.2%/0.6% 0.5%/15% 0.6%/14%
98.6 0.2%/0.5% 0.2%/0.6% 0.2%/2% 0.8%/16%
C. Cornaro et al. / Experimental Thermal and Fluid Science 20 (1999) 6678 69

Fig. 3. Radial distribution of: (a) axial velocity; (b) turbulence inten- Fig. 4. Radial distribution of: (a) axial velocity; (b) turbulence inten-
sity for 72.6 mm jet (x 5 mm). sity for 98.6 mm jet (x 5 mm).

in dierent jet diameters. This situation results from the (turbulence level of 0.6% in the shear layer at the jet
higher turbulence levels in the shear layer at the jet edge edge) and Fig. 6d shows the 98.6 mm jet at Re 6000
for the smaller diameter jets. (turbulence level of 0.5% in the shear layer at the jet
Well-organized vortex structures are present for the edge). In both these cases, large vortices are present.
47.2 mm jet at Re 6000 (see Fig. 6a). The relatively low The ring vortices leaving the nozzle form from an
turbulence intensity in the shear layer at the jet edge instability in the low turbulence shear layer. This in-
(9%) for this case (see Table 1) allows the formation of stability creates disturbances in the ow which result in
these large vortices. Fig. 6b shows the 47.2 mm jet at small vortices which roll up and grow in size as they are
Re 10,000. In this case, the turbulence intensity in the convected downstream. As can be seen in Fig. 6a the
shear layer at the jet edge has increased to 14% and the vortices start to break down at the end of the potential
large vortex structures are no longer present. We con- core where the radial oscillation of the jet is high. This
sistently found that large vortex structures were present method of vortex formation was also observed by Popiel
when the turbulence intensity level in the shear layer at and Trass [3]. As can be seen in Table 3, the spacing
the jet edge was less than 10% and were not present between adjacent vortices is a function of both Reynolds
when the turbulence intensity level was greater than number and jet diameter. The spacing between adjacent
14%. Fig. 6c shows the 72.6 mm jet at Re 6000 vortices was estimated by scaling o a series of 46
70 C. Cornaro et al. / Experimental Thermal and Fluid Science 20 (1999) 6678

c and d in the ring vortex entrainment regions. The


spacing between the smoke streaklines is reduced where
the ring vortex rotates towards the jet centerline and is
expanded where the ring vortex rotates away from the
jet centerline. This contraction and expansion may be
responsible for the axial velocity oscillations observed in
the potential core region and exhibited in the axial ve-
locity distribution at Re 6000 in Fig. 5a. The uid is
accelerated and retarded as the potential core is radially
contracted and expanded by the vortex structures. As
can be seen in Fig. 6b, no ring vortices are present for
the 47.2 mm jet operating at Re 10,000, due to the
high turbulence level in the shear layer at the jet edge for
this case (14%), and in this case the potential core is not
subject to the expansions and contractions of the pre-
vious situations. In addition, the axial velocity distri-
bution in this case (Fig. 5b, Re 10,000) does not show
any variation in the potential core region.

5. Impinging jet at plate

Limited visualization of a jet impinging on a at plate


is performed to contrast with the jet impingement on
curved surfaces. This visualization is performed for the
47.2 mm jet (Re 6000) and for the 72.6 mm jet
(Re 6000 and 10,000). For both jets, jet-to-surface
spacings of H =d 4; 3; 2; 1 were examined. The visual-
ization of the 72.6 mm jet at Re 6000 is shown in Fig.
7ad. All four jet-to-surface spacings are shown. This
situation is typical of all the at plate impingements and
will be the only situation examined in detail.
For the 72.6 mm jet (Re 6000), the potential core
length is 4d (See Table 2), and thus at H =d 4, the
impingement on the surface occurs at the end of the
potential core (See Fig. 7a). The location of the im-
pingement of the vortices on the surface can be seen
from this photo and occurs approximately at a distance
equivalent to the jet radius from the stagnation point.
For the case of Fig. 7a, the vortex structures break down
Fig. 5. Axial velocity distributions for 47.2 mm jet at jet centerline: (a) quickly on the surface due to a strong radial oscillation
Re 6000; (b) Re 10,000. of the impingement region of the jet. This oscillation is
due to the location of the impingement point near the
end of the potential core of the jet. This oscillation was
photos for each situation. The uncertainty due to both observed during the experimental runs, and on video-
measurement and repeatability is 11% with a con- tape of the experiments. However, as the still photos
dence level of 95%. As the Reynolds number increases, only capture an instant in time, it is dicult to show the
the vortex spacing decreases and the initial disturbance oscillation within the context of this paper. This oscil-
in the ow that causes the vortex structures appears lation has also been observed by Popiel and Trass [3].
earlier. Reducing the nozzle-to-plate distance to 3 diameters,
Contractions and expansions of the potential core Fig. 7b, results in a longer duration of the vortices as
region in the radial direction can be observed in Fig. 6a, they move radially along the plate (see Table 4). Minor

Table 2
Potential core length
Jet diameter (mm) Re 6000 Re 10,000 Re 15,000 Re 20,000
47.2 4:1d 4:4d 5:1d 5:5d
72.6 4:0d 4:0d 4:4d 5:6d
C. Cornaro et al. / Experimental Thermal and Fluid Science 20 (1999) 6678 71

Fig. 6. Free jet at: (a) Re 6000, d 47.2 mm; (b) Re 10,000, d 47.2 mm; (c) Re 6000, d 72.6 mm; (d) Re 6000, d 98.6 mm.

radial oscillation of the stagnation point can still be the surface as they are carried by the main ow. A dis-
observed in the videotape of the experiments, but the tortion of the smoke laments can be observed at the
ow along the plate remains attached and unaected by stagnation point.
the radial oscillation.
At H =d 2, Fig. 7c, the jet does not exhibit any well-
formed vortex structures before hitting the surface be- 6. Impinging jet convex surface
cause the vortex structures have not had the time to
form prior to impingement. However, a disturbance is In this experiment, jet impingement on a cylindrical
visible in the ow upstream of the plate. The vortices convex surface was extensively studied. The eects of
appear to be smaller in diameter than at the larger H =d relative curvature, nozzle-to-surface distance and Rey-
spacings. The stagnation point is not oscillating as in the nolds number on the ow structures were investigated.
previous cases. The visualization was completed for the 47.2 mm jet
At H =d 1, although the stagnation point is not (Re 6000), the 72.6 mm jet (Re 6000 and 10,000) and
oscillating radially, the ow along the plate oscillates the 98.6 mm jet (Re 6000, 10,000 and 15,000). In each
strongly on and o the surface in the axial direction as case, nozzle-to-surface spacings of H =d 4; 3; 2; 1 were
can be seen for one snapshot in time in Fig. 7d. Only used.
tiny vortices are observed, as there was not sucient
distance or time for the larger vortices to develop prior 6.1. Nozzle-to-surface distance eect
to the impingement region, and they oscillate on and o
A strong inuence of jet-to-surface spacing on the
ow dynamics was observed. This eect is similar to that
Table 3
found with the at plate and was the same regardless of
Axial spacing of ring vortices in free jet the jet diameter or Reynolds number, so only one con-
dition will be considered in depth (72.6 mm jet,
Jet diameter (mm) Re 6000 Re 10,000 Re 15,000
Re 6000). In general, the preservation of stable co-
47.2 0:9d NA NA herent vortices on the surface is a strong function of jet-
72.6 0:8d 0:6d NA to-surface spacing (see Table 5) and the presence of
98.6 0:7d 0:5d 0:4d
the convex surface aects the jet ow upstream of the
72 C. Cornaro et al. / Experimental Thermal and Fluid Science 20 (1999) 6678

Fig. 7. Impingement on a at plate (d 72.6 mm, Re 6000): (a) H =d 4; (b) H=d 3; (c) H =d 2; (d) H=d 1:

surface in certain low Reynolds number cases, causing potential core length is 4d (See Table 2), and thus at
the ring vortices to be spaced more closely together than H =d 4, the impingement on the surface occurs at the
for the free jet. However, as jet diameter and Reynolds end of the potential core. The vortex structures are
number increases, this eect disappears as demonstrated typically already beginning to break down by the end of
in Table 6. The spacing between adjacent vortices was the potential core, and the radial oscillation of the jet
estimated by scaling o a series of 46 photos for each at the stagnation point acts to hasten this break down of
situation. The uncertainty due to both measurement and the vortices that hit the surface. In addition, the sym-
repeatability is 11% with a condence level of 95%. It metry of the jet upstream of the surface is destroyed,
would also be interesting to compare the ring vortex unlike the situation of impingement on a at plate.
spacing dierences between impingement on a at plate The interaction of the jet with the convex cylindrical
and on convex surfaces. Unfortunately, for this experi- surface thus results in dierences in the jet behavior at
ment, not enough at plate data were available to make H =d 4 when compared with the free jet and at plate
an accurate comparison. impingement cases. When the convex cylindrical surface
A strong radial oscillation of the stagnation point is is present, the ring vortices do not exhibit the symmetry
the main feature observed when H =d 4. Fig. 8a, at one of and are spaced more closely together than for the free
instant in time, captures a strong oscillation to the left of jet and at plate impingement visualization.
the photo. The videotape of the experiment shows that At H =d 3 (Fig. 8b), unlike H =d 4, dened vortex
the jet moves randomly in this way around the im- structures can be observed in the jet shear layer. The
pingement point. This is similar to, but an even stronger vortices roll down the surface until they dissipate in an
eect, than was found with impingement on the at unsteady ow situation, and the videotape shows that
plate (see Fig. 7a). For the 72.6 mm jet at Re 6000, the the stagnation point oscillates less dramatically radially

Table 4
Duration of stable, coherent vortex rings on at plate
Flow situation H=d 4 H =d 3 H=d 2 H=d 1
d 47.2 mm, Re 6000 Immediate breakdown 1.5d Inconclusive Oscillates on and o surface
d 72.6 mm, Re 6000 Immediate breakdown 1.5d 1.5d Oscillates on and o surface
d 72.6 mm, Re 10,000 Immediate breakdown 1d 1d Oscillates on and o surface
C. Cornaro et al. / Experimental Thermal and Fluid Science 20 (1999) 6678 73

Table 5
Duration of stable, coherent vortex rings on cylindrical convex surface
Flow situation H =d 4 H =d 3 H=d 2 H=d 1
47.2 mm, Re 6000 Immediate breakdown Immediate breakdown 30 Oscillates on and o surface
72.6 mm, Re 6000 Immediate breakdown 40 Remain as long as smoke is visible Oscillates on and o surface
72.6 mm, Re 10,000 Immediate breakdown Immediate breakdown 35 Oscillates on and o surface
98.6 mm, Re 6000 Immediate breakdown Immediate breakdown Remain as long as smoke is visible Oscillates on and o surface
98.6 mm, Re 10,000 Immediate breakdown Immediate breakdown Remain as long as smoke is visible Oscillates on and o surface
98.6 mm, Re 15,000 No data avail. Immediate breakdown 35 Oscillates on and o surface

Table 6 layer on the cylindrical surface. These vortices appear to


Spacing of ring vortices (free jet/upstream of convex surface for remain intact as long as the smoke is visible.
H =d 3) At H =d 1 (Fig. 8d) although the stagnation point is
Jet diameter (mm) Re 6000 Re 10,000 Re 15,000 not oscillating radially, a strong oscillation of the sur-
47.2 mm 0:9d=0:7d NA NA face ow on and o the convex surface in the axial di-
72.6 mm 0:8d=0:6d 0:6d=0:6d NA rections is observed. Fig. 8d captures the instant where
98.6 mm 0:7d=0:7d 0:5d=0:5d 0:4d=0:4d the ow along the left side of the surface is beginning to
lift o. This is similar to that observed for the at plate
impingement. Only tiny vortices are generated in the
oscillating shear layer, and similar to the at plate sit-
than at H =d 4. In this case, the jet ow upstream of uation, a distortion of the smoke laments can be ob-
the convex surface preserves its symmetry. served at the stagnation point. It is dicult to determine
At H =d 2 (Fig. 8c) the stagnation point no longer how long these small vortices remain on the surface due
oscillates and well-dened vortices appear in the shear to their axial oscillation.

Fig. 8. Impingement on a convex surface. Eect of jet-to-surface spacing (d 72.6 mm, Re 6000): (a) H=d 4; (b) H =d 3; (c) H =d 2;
(d) H =d 1:
74 C. Cornaro et al. / Experimental Thermal and Fluid Science 20 (1999) 6678

6.2. Curvature eect 0.38) vortices can be observed on the surface, even
though they break down rapidly.
The eect of curvature was investigated by varying At H =d 4 and H =d 1 (not shown), the curvature
the nozzle diameter of the impinging jet and maintaining eects are reduced. The strong radial oscillation at the
a xed Reynolds number and surface diameter, thus jet impingement region for H =d 4 and the strong axial
eectively changing d=D. This methodology is consistent oscillation of the surface ow for H =d 1 overwhelm
with other studies that examined the eect of surface any relative curvature eects.
curvature, including Hrycak [7], Gau and Chung [9],
and Lee et al. [10]. When considering the eect of rela- 6.3. 3-D visualization
tive curvature on the ow structures, nozzle-to-surface
spacing must also be considered due to its strong eect The visualizations in all the earlier gures show the
on the ow structure. behavior of a jet section parallel to the radial axis of the
First considering the curvature eects at spacing of cylinder. In this case, only a cross-section of the ow is
H =d 2, a transition from laminar to unsteady ow observed. To address this limitation, the smoke wire was
along the surface is observed for the lowest relative rotated 45 and 90 with respect to the original position
curvature (d=D 0:18). (See Fig. 9a). When the relative to investigate the behavior of the ow at other cross-
curvature is increased, the ow remains laminar along sections. As an example, the visualizations for the 72.6
the surface as can be seen in Fig. 9b, d=D 0:28, and mm jet at H =d 2 and Re 6000 are shown in Fig. 10.
Fig. 9c, d=D 0:38. This may due to the stabilizing At 90 orientation (Fig. 10a) the section parallel to the
eect on the ow of the increasing centrifugal force, axial axis of the cylinder is visualized. The ow structure
resulting in a longer period of laminar ow for higher in this orientation is similar to that observed for a jet
relative curvatures [13]. impinging on a at plate. The visualization made for 45
The situation at H =d 3 (not shown) is similar to (Fig. 10b) show ow structures which appear similar to
that at H =d 2. The ow along the surface is unsteady those with the original wire orientation, although it is
for the lowest relative curvature (d=D 0:18) while for clear that this an intermediary step between the two
the larger relative curvature values (d=D 0:28 and orientations.

Fig. 9. Impingement on a convex surface. Eect of curvature (H =d 2, Re 6000): (a) d=D 0:18; (b) d=D 0:28; (c) d=D 0:38:
C. Cornaro et al. / Experimental Thermal and Fluid Science 20 (1999) 6678 75

ow over the convex surface and is qualitatively the


same for all diameters and Reynolds numbers, so only
one situation will be considered (72.6 mm jet,
Re 6000). Stable coherent vortices exist on the surface
only for low Reynolds numbers for H =d 3 and
H =d 2 (see Table 7) While vortex structures do exist
for H =d 1, the ow along the surface oscillates on and
o the surface in the axial direction, as with impinge-
ment on convex and at surfaces.
At H =d 4, similar to the ow over convex surfaces,
the stagnation point oscillates strongly in the radial di-
rection, dissipating the upstream vortex structures and
jet symmetry. Fig. 11a captures this oscillation as the jet
moves to the right edge of the surface. The videotape of
the experiment shows that the jet oscillates randomly
about the impingement point. For the 72.6 mm jet
(Re 6000), potential core length is 4d (See Table 2);
thus at H =d 4, the impingement on the surface occurs
at the end of the potential core. The vortex structures
are already beginning to break down by the end of the
potential core, and the radial oscillation of the jet at
the stagnation point acts to hasten this breakdown of
the vortices that approach the surface. No stable vor-
tices are observed on the surface.
At H =d 3 (Fig 11b), the videotape of the experi-
ment shows that the radial oscillation of the stagnation
point is less intense than at H =d 4, but unlike the
convex surface situation, vortices cannot be seen along
Fig. 10. Impingement on a convex surface. 3-D eects (d 72.6 mm, the surface.
H =d 2, Re 6000): (a) 90 smoke wire orientation; (b) 45 smoke At H =d 2, small vortices appear on the wall, and
wire orientation. maintain integrity for quite a distance from the stagna-
tion point before breakdown. The recirculation of the
ow into the main stream can be clearly seen in Fig. 11c.
At H =d 1, small vortices appears in the wall jet
7. Impinging jet concave surface region and maintain integrity until almost 90 (see Fig.
11d). However, although the stagnation point is not
The jet impinging on a concave surface is dicult to oscillating radially, a strong oscillation of the surface
observe in that the surface shape limits the illumination ow on and o the convex surface in the axial direction
of the ow eld. In some cases, the lighting and shadows is observed. This is similar to the situation for ow
on the surface interfere with a clear interpretation of the impinging on the at plate and convex surfaces. Fig. 11d
ow structures. The visualization was completed for the captures an instant in time where the ow on the right
47.2 mm jet (Re 6000), the 72.6 mm jet (Re 6000 and side of the surface has lifted o the surface while the ow
10,000) and the 98.6 mm jet (Re 6000, 10,000 and on the left has touched down to the surface.
15,000). In each case, nozzle-to-surface spacings of
H =d 4, 3, 2 and 1 were used. 7.2. Curvature eect
In general, the visualization shows that the ow on a
concave surface is more unsteady than ow on a convex When considering the eect of curvature on the ow
surface. The centrifugal forces act to destabilize the ow structures, nozzle-to-surface spacing must also be
over concave surfaces while they act to stabilize the ow considered due to its strong inuence on the ow
over convex surfaces [13]. The ow upstream of the structure.
concave surface is strongly aected by the ow exiting At H =d 3, the lower the relative curvature, the
the surface at 90. This exhaust ow becomes entrained longer the vortices retain their integrity. This is opposite
in the primary jet ow, reducing the likelihood of stable to that observed for the convex surface. For the highest
ring vortices upstream of the surface. This is most relative curvature, d=D 0:39 (Fig. 12a), no structures
clearly shown in Fig. 11c. are seen on the surface. As the relative curvature de-
creases, some vortices can be seen in the shear layer
7.1. Nozzle-to-surface distance eect upstream of the surface, but they break down as soon
they reach the surface (Fig. 12b). At the lowest relative
The jet-to-surface spacing strongly inuences the curvature, d=D 0:19 (Fig. 12c), distinct vortices ap-
ow dynamics. This eect is similar to that observed for pear on the surface and break down at approximately
76 C. Cornaro et al. / Experimental Thermal and Fluid Science 20 (1999) 6678

Fig. 11. Impingement on a concave surface. Eect of jet-to-surface spacing (d 72.6 mm, Re 6000): (a) H =d 4; (b) H=d 3; (c) H =d 2;
(d) H =d 1:

40 from the stagnation point. The stagnation point and concave surfaces of a semicylinder were performed.
oscillates and the recirculating ow does not aect the A low turbulence intensity in the shear layer at the jet
jet stream. At H =d 2 (not shown), the curvature eect edge at the tube exit is related to the development of
is the similar to that observed for H =d 3. organized vortex structures in the free jet shear layer,
As with ow over the convex surface, at H =d 4 and while a high turbulence intensity in the shear layer at the
H =d 1 (not shown), the relative curvature eects are jet edge inhibits the development of organized vortex
reduced. The strong radial oscillation in the jet im- structures. An oscillation in the axial velocity in the
pingement region for H =d 4 and the strong axial os- potential core region at Re 6000 is associated with the
cillation of the surface ow for H =d 1 mask any vortex evolution along the free jet shear layer in that
curvature eects. region. The ow visualization of a free jet shows that the
spacing between the smoke streaklines is reduced where
the ring vortex rotates towards the jet centerline and is
expanded where the ring vortex rotates away from the
8. Conclusions jet centerline. The uid is accelerated and retarded as the
potential core is radially contracted and expanded by
Detailed ow visualizations of a free jet, a jet im- the vortex structures resulting in axial velocity oscilla-
pinging on a at plate, and a jet impinging on convex tions.

Table 7
Duration of stable, coherent vortex rings on cylindrical concave surface
Flow situation H=d 4 H=d 3 H=d 2 H=d 1
47.2 mm, Re 6000 Immediate breakdown 40 30 Oscillates on and o surface to ~90
72.6 mm, Re 6000 Immediate breakdown Immediate breakdown 60 Oscillates on and o surface to ~90
72.6 mm, Re 10,000 Immediate breakdown Immediate breakdown Immediate breakdown Oscillates on and o surface to ~90
98.6 mm, Re 6000 Immediate breakdown Immediate breakdown Immediate breakdown Oscillates on and o surface to ~90
98.6 mm, Re 10,000 Immediate breakdown Immediate breakdown Immediate breakdown Oscillates on and o surface to ~90
C. Cornaro et al. / Experimental Thermal and Fluid Science 20 (1999) 6678 77

Fig. 12. Impingement on a concave surface. Eect of curvature (H =d 3, Re 6000): (a) d=D 0:39; (b) d=D 0:29; (c) d=D 0:19:

The ring vortices appear to form from an instability in axial oscillation of the ow on and o the surface is
the low turbulence shear layer which result in small initiated. This axial oscillation carries the vortex struc-
vortices which roll up and grow in size as they are con- tures periodically on and o the surface.
vected downstream. The vortices start to break down at An increase in relative curvature for jet impingement
the end of the potential core where the radial oscillation on a convex surface delays a transition to turbulence as
of the jet is high. The spacing between adjacent vortices is the ow along the surface remains stable and shows
a function of Reynolds number. As the Reynolds num- well-formed vortex structures. This may due to the sta-
ber increases, the vortex spacing decreases. bilizing eect on the ow of the increasing centrifugal
A strong inuence of jet-to-surface spacing was also force, resulting in a longer period of laminar ow for
observed for impingement on a convex surface. The higher relative curvatures.
preservation of stable coherent vortices on the surface is Flow over a concave surface is more unsteady than
a strong function of jet-to-surface spacing, with the ow on a convex surface as the ow upstream of the
likelihood of vortices occurring on the surface decreas- concave surface is strongly aected by the ow exiting
ing with increases in jet-to-surface spacing. The presence the surface into recirculation. This exhaust ow becomes
of the convex surface also aects the jet ow upstream of entrained in the primary jet ow, reducing the likelihood
the surface causing the ring vortices to be spaced more of stable ring vortices.
closely together than for the free jet. However, as the jet The jet-to-surface spacing also strongly inuences
diameter increases, the spacing between vortex rings the ow dynamics for jet impingement on a concave
tends to approach and even surpass that of the free jet. surface. The eect is similar to that observed for ow
A strong radial oscillation of the stagnation point is over a convex surface and is qualitatively the same for
the main feature observed for large jet-to-surface spac- the all diameters and Reynolds numbers. Radial oscil-
ing with impingement on a convex surface. The radial lation of the impingement point occurs and decreases
oscillation of the jet acts to hasten the breakdown of the with H =d. Vortex structures are found at low H =d
vortices that reaches the surface and destroys the sym- values. However, in contrast to the convex surface, the
metry of the jet upstream of the surface. ow along the concave surface become less stable and
As the jet-to-surface spacing is decreased, the radial show fewer vortex structures as the relative curvature
oscillation of the impingement point decreases, but an increases.
78 C. Cornaro et al. / Experimental Thermal and Fluid Science 20 (1999) 6678

Nomenclature [2] R. Viskanta, Heat transfer to impinging isothermal gas and ame
D curved surface outer diameter for convex jets, Exp. Thermal Fluid Sci. 6 (1993) 111134.
[3] C.O. Popiel, O. Trass, Visualization of a free and impinging
surface (256.76 mm), curved surface inner
round jet, Exp. Thermal Fluid Sci. 4 (1991) 253264.
diameter for concave surface (252 mm) [4] R. Chupp, H. Helms, P. McFadden, T. Brown, Evaluation of
H jet-to-surface distance, m internal heat transfer coecients for impingement cooled turbine
L tube length, m airfoil, J. Aircraft 6 (1969) 203208.
Re Reynolds number, dimensionless [5] D. Metzger, T. Yamashita, C. Jenkins, Impingement cooling of
U local velocity of jet, m/s concave surfaces with lines of circular air jets, J. Eng. Power 91
UC velocity at jet exit centerline, m/s (1969) 149158.
UO average velocity at jet exit, m/s [6] W. Tabako, W. Clevenger, Gas turbine blade heat transfer
UP local velocity uctuation, m/s augmentation by impingement of air jets having various con-
d jet tube diameter, mm gurations, J. Eng. Power 94 (1972) 5160.
[7] P. Hrycak, Heat transfer from a row of jets impinging on concave
r radial coordinate of jet, m
semi-cylindrical surfaces, Int. J. Heat Mass Transfer 28 (1981)
x axial coordinate of jet measured from tube 175181.
exit, m [8] G. Yang, M. Choi, J.S. Lee, A study of jet impingement cooling
on the semicircular concave surface: eects of two dierent nozzle
congurations, in: 1995 National Heat Transfer Conference,
HTD-5 307, 1995, pp. 131138.
Acknowledgements [9] C. Gau, C.M. Chung, Surface curvature eect on slot-air-jet
impingement cooling ow and heat transfer process, J. Heat
This material is based upon work completed while Transfer 113 (1991) 858864.
C. Cornaro was supported by a fellowship of the Italian [10] D.H. Lee, Y.S. Chung, D.S. Kim, Turbulent ow and heat
National Research Council and while A.S. Fleischer was transfer measurements on a curved surface with a fully developed
supported under a National Science Foundation Grad- round impinging jet, Int. J. Heat Fluid Flow 18 (1997) 160169.
[11] Y. Kornblum, R.J. Goldstein, Jet impingement on semicylindrical
uate Research Fellowship. Support from the Engineer-
concave and convex surfaces: Part Two heat transfer, Interna-
ing Research Program of the Department of Energy is tional Symposium on Physics of Heat Transfer in boiling and
gratefully acknowledged. condensation (1997) 597602.
[12] T. Corke, D. Koga, R. Drubka, H. Nagib, A new technique for
introducing controlled sheets of smoke streaklines in wind
References tunnels, in: Proc. Int. Congress on Instrumentation in Aerospace
Simulation Facilities, IEEE Publication 77CH 1251-8 AES, 1977,
[1] K. Jambunathan, E. Lai, M.A. Moss, B.L. Button, A review of pp. 7480.
heat transfer data for singular jet impingement, Int. J. Heat Fluid [13] H. Schlicting, Boundary Layer Theory, 6th ed., McGraw-Hill,
Flow 13 (1992) 106115. New York, pp. 488491 and 504507, 1968.

S-ar putea să vă placă și