Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/223209187

Velocity measurements in the near field of a radial swirler

Article  in  Experimental Thermal and Fluid Science (EXP THERM FLUID SCI) · March 1998
DOI: 10.1016/S0894-1777(97)10027-9

CITATIONS READS

17 62

1 author:

Lyes Khezzar
Khalifa University of Science & Technology, Abu Dhabi, UAE
131 PUBLICATIONS   522 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Impinging jets atomization View project

Gas-Liquid Swirling Flows View project

All content following this page was uploaded by Lyes Khezzar on 08 May 2018.

The user has requested enhancement of the downloaded file.


Experimental Thermal and Fluid Science 16 (1998) 230±236

Velocity measurements in the near ®eld of a radial swirler


1
L. Khezzar
Mechanical Engineering Department, Sultan Qaboos University, Muscat, Oman
Received 8 May 1996; received in revised form 19 February 1997; accepted 25 March 1997

Abstract

Con®ned turbulent air swirling ¯ows are investigated for two con®gurations where swirl is produced by a conventional radial type
swirler. The ¯ow Reynolds number based on the bulk velocity in the downstream pipe was equal to 15,000 and the swirl number
evaluated using velocity pro®les in the downstream pipe was equal to 1.5. The ¯ow downstream of the swirler has been quanti®ed by
measurements of three velocity components and their corresponding rms values along the radial direction and for several axial po-
sitions using laser Doppler anemometry. The downstream ¯ow structure revealed the existence of an o€-axis reverse ¯ow region and
a forced-vortex type of ¯ow in the inner region and free-vortex in the outer one. The strength of the inner vortex is maintained with
downstream distance. The results represent a set of benchmark data for validation of calculation methods. Ó 1998 Elsevier Science
Inc. All rights reserved.

Keywords: Swirling ¯ow; Swirlers; Turbulent ¯ow; Fluid velocity measurements; Recirculation bubble

1. Introduction pressure gradient on the axis. More speci®cally, Escu-


dier and Keller [6] found that isothermal supercritical
Swirlers are used in gas turbine combustors to impart and sub-critical swirling ¯ows exhibit di€erent sensitivi-
rotation to the internal air ¯ow. The resulting swirling ties to changes in exit conditions, and this is particularly
motion ensures ¯ame stabilisation and improved mixing relevant to ¯ow computations. Generally the resulting
between the fuel and air and thus increased combustion ¯ow topology depends on the strength of the swirl mo-
eciency. Better economical design is currently motivat- tion, the particular design of the swirler itself and how
ed by the need to obtain low levels of pollutants and im- swirl is imparted and the boundary conditions of the
prove the exit temperature distribution from the geometry as experienced in [7]. Under certain circum-
combustor. stances however, strong swirl may lead to damping of
Although real swirling combustor ¯ows involve com- turbulence transport and thus swirl e€ects can become
bustion with combined heat and mass transfer, it is undesirable and designers need therefore to understand
nonetheless instructive to study model swirl combustors them for better control of the resulting ¯ow features.
under noncombusting conditions. The absence of com- The research e€ort to date in the aerothermodynamic
bustion allows to isolate the e€ect of real ¯ow dynamics design of combustors has had the objective of achieving
phenomena. Several arrangements representing di€erent a better understanding of the complex phenomena asso-
concepts of the combustor geometry have been investi- ciated with swirling ¯ows and an improvement in the ac-
gated earlier, under isothermal con®ned turbulent con- curacy and predictive capabilities of turbulence models
ditions, such as co-axial con®ned jets of So et al. [1], and computational methods as can be seen in [8±10].
Habib and Whitelaw [2], Vu and Gouldin [3], Altgeld It is now recognised that calculations of strongly swirl-
et al. [4] or a single con®ned jet as that of Rhode et al. ing ¯ows are better handled through the use of a second
[5]. A shared characteristic of con®ned swirling ¯ows moment closure turbulence model, see [11], and that the
is the presence of a recirculation bubble in the core of k±e model is inadequate to compute such ¯ows as it pre-
the ¯ow, either on- or o€-axis, as a result of the adverse dicts too high rates of mixing.
The present work reports on the air ¯ow velocity ®eld
1
Address: College of Engineering, Sultan Qaboos University, P.O. under isothermal, con®ned turbulent conditions down-
Box 33, Al-Khood, Muscat 123, Oman. Tel.: +968 515 353; fax: +968 stream of a speci®c design of a model radial in¯ow swirl-
513 416. er of industrial relevance. Relatively few studies have

0894-1777/98/$19.00 Ó 1998 Elsevier Science Inc. All rights reserved.


PII: S 0 8 9 4 - 1 7 7 7 ( 9 7 ) 1 0 0 2 7 - 9
L. Khezzar / Experimental Thermal and Fluid Science 16 (1998) 230±236 231

been conducted on turbulent ¯ows generated by radial


in¯ow swirlers. Emphasis is on the near ¯ow ®eld gener-
ated by such a swirler in two geometrical con®gurations.
Laser Doppler anemometry is used to obtain the de-
tailed velocity measurements in the downstream swirling
¯ow. A further objective of this work is to obtain accu-
rate ¯uid velocity measurements that can be used as a
benchmark set of data to assess the accuracy of calcula-
tion methods.
The experimental setup and ¯ow geometry are de-
scribed in the following section together with the mea-
Fig. 2. Test section description, dimensions in mm.
surement technique. The results are presented and
discussed next, followed by the conclusions.
swirler is directly connected to a fairing, then a perspex
downstream pipe with a length of 350 mm, and internal
2. The experimental method and outside diameter of 84 and 90 mm, respectively. In
the second one, referred to as case 2, the con®guration is
2.1. Flow con®guration similar to case 1 in all respects except that a short per-
spex pipe, shown in Fig. 2 by the dotted line, of 46
The geometrical details of the model swirler, which mm inside diameter and 27 mm in length, is inserted be-
was carefully machined from Plexiglas, are given in tween the swirler and the fairing. The short-pipe insert
Fig. 1. The swirler possesses eight arms, equally distrib- or spool pipe is supposed to represent the pre-chamber.
uted around the circumference of its plate, as shown in In order to provide an outlet boundary without re-
Fig. 1. The axial width of each radial inlet passage, de- verse ¯ow an aluminium ba‚e of 54 mm diameter and
®ned in the ®gure, is 4 mm. 4 mm thickness was located at a distance of 100 mm
The steady ¯ow air circuit is a suction system driven from the end of the downstream pipe in all cases. The
by a centrifugal exhauster. The air, drawn from the stag- ba‚e was rigidly located with a central rod of 100 mm
nant surrounding room atmosphere, enters the test sec- in length which was ®rmly maintained in its position
tion and hence the swirler in the radial direction. It by a number of rods arranged in a spider form and ®xed
consists of a test section, an adapter followed by a 2 in to a circular ring pushed in the downstream pipe. The
plastic pipe and a standard BS ori®ce plate meter. A contraction between the 84 mm diameter pipe and the
U-tube water manometer was connected to pressure 2 in. extract pipe was a€orded by a Plexiglas adapter
taps located both upstream and downstream of the plate (not shown).
and allowed monitoring of the ¯ow rate through the test
section. The exhauster was operated at constant speed
and the air ¯ow rate through the test section was regu- 2.2. Experimental technique
lated by a valve located between the ori®ce meter and
the exhauster. Two test sections were used and Fig. 2 il- The laser Doppler anemometer consisted of a Spec-
lustrates both of them. Both test sections incorporated tra-Physics Argon-Ion air cooled laser operating at
the radial swirler in two di€erent manners. In the ®rst 514.5 nm with a power output of 250 mW, a transmit-
one, which will be referred to hereafter as case 1, the ting optical unit and a receiving unit, and was operated
with forward collected scattered light o€-axis. The sys-
tem was mounted on a milling table which could be tra-
versed in the three spatial directions with a precision of
0.05 mm. A rotating di€raction grating-based unit was
used. It consisted of a focusing, a collimating and an
imaging lens. The collection part of the system consisted
of a DANTEC (55 ´ 34) PM optics and a Burle Indus-
tries 4526V1 photomultiplier tube. The output of the
photomultiplier tube was ampli®ed prior to processing.
The principal characteristics of the optical arrangements
are summarised in Table 1.
The output signal of the photomultiplier was pro-
cessed by a frequency counter signal processor (TSI
1990C) which was interfaced to a 386 IBM PC to obtain
ensemble averaged mean and rms velocities from a sta-
tistically sucient number of samples (4000). Correc-
tions for the refraction of the beams at the cylindrical
walls of the test section were applied for the swirl com-
Fig. 1. Details of the swirler, dimensions in mm. ponent of velocity.
232 L. Khezzar / Experimental Thermal and Fluid Science 16 (1998) 230±236

Table 1 and downstream of it, for case 2. As shown in Fig. 2,


Optical characteristics of LDA con®gurations the reference position corresponds to the intersection
Laser
of the fairing with the downstream pipe. The velocity re-
Wavelength (nm) 514.5 sults have been normalised by the bulk velocity Ub , cal-
Power (mW) 250 culated from the pro®le at the exit annular space, in the
Beam exit diameter 0.8 downstream pipe which was equal to 2.53 m/s for cases 1
Optics and 2. Prior to each set of measurements the symmetry
Focusing lens (f/mm) 80 of the ¯ow was checked at selected stations downstream
Collimating lens (f/mm) 300 and upstream of the reference location and found to be
Imaging lens (f/mm) 300 within 2% in the mean velocity and therefore only half-
Frequency shift (MHz) 6 diameter pro®les are shown.
Half angle intersection () 4.73
The axial velocity component for case 1 and its cor-
Measuring volume
Length (lm) 776
responding rms are depicted in Fig. 3 for di€erent axial
E€ective length (lm) 300 locations. They reveal the existence of an o€-axis recir-
Diameter (lm) 64 culation region with a maximum reverse velocity equal
Fringe spacing (lm) 3.122 to around twice the bulk velocity at, x ˆ 10 mm, and de-
cays subsequently to levels equal to the bulk velocity.
The reverse ¯ow region extends at x ˆ 10 mm from a ra-
The air ¯ow was seeded with silicone oil droplets of dial position of r/R ˆ 0.2 to a position of r/R ˆ 0.7. This
up to 3 lm diameter, generated in a blast-type atomiser radial extent subsequently narrows in the downstream
and introduced with almost zero momentum in the ple- direction. The radial pro®les of the axial mean velocity
num region surrounding the inlet to the swirler so as to suggest, thus, the presence of two recirculating bubbles
produce minimum disturbance to the inlet ¯ow. of annular shape occupying over 80% of the radius of
Possible sources of measurement uncertainties, see the pipe at x ˆ 10 mm as observed by Faler and Leibov-
[12], include velocity bias and velocity gradient broaden- itch [13]. The inner weak vortex is formed of relatively
ing which is only important in regions of large mean ve- slow moving ¯uid. This structure of two nested vortices
locity gradients. Employing o€-axis collection, high which circulate in an opposite manner was also ob-
frequency shift and a small pinhole on the photomulti- served, albeit in laminar water ¯ow, by Bornstein and
plier these uncertainties were reduced to insigni®cant Escudier [14], who used a radial in¯ow van swirler.
levels. The timing uncertainty in the TSI counter The maximum axial positive velocities occur in the outer
equipped with a 1 GHz clock is also negligible. The sta- part of the ¯ow near the wall, whereas the central region
tistical uncertainty due to ®nite sample size was at most is characterised by small absolute positive velocities,
equal to 2.5% for the mean velocity in regions of high
turbulence intensity and up to 3% for the rms velocities.
The overall uncertainty in the measurement of mean and
rms velocities is estimated to be not more than 5% and
10%, respectively.

3. Results

The experimental results consisted of measurements


of averaged mean and rms components in the axial, ra-
dial and tangential directions, for the two geometrical
con®gurations studied of cases 1 and 2. All the measure-
ments were obtained at a Reynolds number based on the
downstream pipe diameter and ¯uid bulk velocity equal
to 15,000. The swirl number obtained from the pro®les
at x ˆ 10 mm was equal to 1.5 and was calculated using
the relation
R
UWr2 dr
R : …1†
R U 2 r dr
Integration of the axial velocity pro®les in all cases
produced volume ¯owrates which varied by less than
10% and values of this magnitude were found only for
pro®les with steep velocity gradients. The velocity pro-
®les were obtained at axial positions downstream of
the reference location for case 1 and both upstream Fig. 3. Axial velocity, Mean and rms, case 1.
L. Khezzar / Experimental Thermal and Fluid Science 16 (1998) 230±236 233

which vary from 0.316Ub at x ˆ 10 mm to negative near The approximation to the Burger's velocity distribution
zero at x ˆ 200 mm, suggesting a rear stagnation point as given by the equation
for the bubbles structure in a similar fashion to the ¯ows R …r2 =R2 †

of Faler and Leibovitch [13]. The gradual pipe wall ex- W ˆA 1 ÿ eÿB …2†
pansion and the recirculating toroidal bubbles which r
provide blockage to the ¯ow are both responsible for was only satisfactory for the inner core region, when a
the acceleration of the axial ¯ow near the wall of the curve ®t was attempted. In the outer region, there was
downstream pipe and the removal of the expansion re- a notable discrepancy between the measured data and
circulation zone, see [3,5]. In Fig. 3, it can be observed the above equation as noticed previously in [14]. The
that, the radial distribution of the rms velocity is not swirl velocity pro®le approaching but never completely
uniform. Because of high gradients of mean velocity attaining a Burger's vortex is a common feature of single
near the wall and along the centre line, the correspond- swirling ¯ows with radial in¯ow, see [3,14], which, from
ing rms velocities reach a maximum of 3.04Ub at x ˆ 10 a calculation scheme's point of view, can be captured by
mm in the outer part of the ¯ow, show two relative a full Reynolds stress closure but one that remains inad-
peaks near the wall and along the centre line, and decay equately accounted for by the k±e turbulence model, as
in the downstream direction to adopt a uniform pro®le shown by Hogg and Leschziner [8]. Examination of the
with a marked decrease near the axis at x ˆ 200 mm with axial development of the swirl velocity pro®les reveal
a value of around 0.9Ub . The decay is relatively rapid that the strength of the inner vortex is maintained up
and is indicative of di€usion and dissipation. In the case to a downstream distance of x ˆ 150 mm. If the inner
of combustion being present, predicting its e€ect would core is delimited by the radial position of the local max-
be tentative, however, as previously observed by other imum swirl velocity as postulated above, one can ob-
workers, see [7], it can be expected to enlarge the recir- serve that almost no expansion of the inner core can
culation zone and lead to higher levels of turbulence in- be detected. In the outer region, however, the axial gra-
tensities. dient of the tangential velocity is signi®cant and is re-
The maximum mean swirl velocity in Fig. 4 has a val- sponsible for the onset of reverse ¯ow in that region.
ue of about 14Ub at x ˆ 150 mm. The pro®les are similar A simpli®ed analysis based on the quasi-cylindrical ap-
in shape at all the stations and adopt a peak down- proximation to the radial component of the momentum
stream at a radial position of 0.4R, which marks the equation leads to the following expression for the axial
end of a forced-vortex region. The swirl velocity pro®les pressure gradient
    Z
clearly exhibit an inner solid-body rotation and an outer @P @P @ R qx2
free-vortex structure. The combination of the inner and  ÿ dx: …3†
@x rˆ0 @x rˆR @x 0 r
outer structure tends to but never approaches a Burger's
vortex velocity distribution with downstream distance. If the second term on the right-hand side decays contin-
uously with axial distance as is the case here, an adverse
pressure gradient is set up, with the consequence of re-
verse ¯ow. Various other explanations, not entirely sat-
isfactory, for the occurrence of a recirculating bubble in
swirling ¯ows have been formulated in addition to the
simpli®ed one given above. One that particularly de-
serves mention is given by Brown and Lopez [15], who
have shown, on the basis of theoretical considerations,
that in an axisymmetric swirling pipe ¯ow, the produc-
tion of negative azimuthal vorticity which is dominated
by inviscid mechanisms leads to divergence of the
streamlines and consequently to a recirculating bubble.
The corresponding rms velocities reach maximum
values of 2.5Ub in the outer part of the ¯ow over the ®rst
30 mm and decay with downstream distance to less than
0.9Ub at x ˆ 150 mm. In a similar fashion to the axial
component the centre line values of the rms velocities re-
main relatively high, with downstream decay only im-
portant in the outer core region, where viscous
di€usion it seems, is more predominant.
The radial velocity component and its corresponding
rms values are represented in Fig. 5 for three axial sta-
tions. The mean pro®les are similar in shape, with posi-
tive and negative values and a maximum nearly equal to
the bulk velocity. The distribution of the negative and
positive values along the radius con®rms the presence
Fig. 4. Swirl velocity, mean and rms, case 1. of the nested counter-rotating vortices. The rms pro®les
234 L. Khezzar / Experimental Thermal and Fluid Science 16 (1998) 230±236

case 2 and are similar in shape to those of case 1 in


the downstream pipe. They reveal the existence of an
o€-axis reverse ¯ow region and indicate also the pres-
ence of counter-rotating vortices, which extend up-
stream of the fairing inside the spool pipe. Because of
blockage provided by the latter, large positive values
of the axial velocity are observed near the wall and rel-
atively low positive values in the central part of the ¯ow.
The mean velocity on the axis, this time, is greater than
zero approaching twice the bulk velocity value at
x ˆ )25 mm. The rms velocity values in the spool pipe
are higher than in the downstream pipe and decay there-
after in the downstream direction.
The swirl mean velocities with their corresponding
rms values for case 2 are shown in Fig. 7. Large mean
swirl velocities exist in the spool pipe with maximum
values equal to about 20 times the bulk velocity and a
shape signi®cantly di€erent from those which exist in
the downstream pipe. The pro®les measured at
x ˆ )25 mm and x ˆ )38mm are almost identical. It
can be seen that the tangential mean velocity distribu-
tion is of forced-vortex type in the entire spool pipe.
Downstream of the fairing, the mean velocity distribu-
Fig. 5. Radial velocity, mean and rms, case 1.
tion adopts a pro®le similar to the one in case 1, see
Fig. 4, i.e. forced vortex in a smaller inner region up
to r/R ˆ 0.2 and thereafter a free-vortex shape. The
have peaks on the axis and in the vicinity of the outer mean swirl velocity pro®les are steeper in case 2 than
wall which are associated with gradients in the corre- in case 1, but the strength of the inner core vortex, is still
sponding mean velocity pro®les. The levels are almost well preserved at x ˆ 30 mm. The drastic change in
equal to those of the swirl rms velocities at the same ax- shape of the mean swirl velocity pro®le in passing from
ial locations. the spool pipe to the downstream pipe results because of
Fig. 6 presents mean axial velocity pro®les for three the e€ect of ¯ow divergence encourged by the slopping
axial locations one upstream at x ˆ )25 mm and two wall of the fairing, the larger diameter of the down-
downstream of the fairing at x ˆ 10 and x ˆ 50 mm for stream pipe and the blockage provided by the bubble.

Fig. 6. Axial velocity, mean and rms case 2. Fig. 7. Swirl velocity, mean and rms, case 2.
L. Khezzar / Experimental Thermal and Fluid Science 16 (1998) 230±236 235

The implication for a ¯uid particle originating from the


spool pipe is a reduction of the mean swirl velocity in the
outer diverging part of the ¯ow in the downstream pipe
by virtue of conservation of angular momentum. The
corresponding rms velocities in the spool pipe have a
maximum near the pipe wall and decay downstream of
the fairing, to reach similar levels as in case 1.
The mean and rms values of the radial velocity are
shown in Fig. 8. The mean velocity is important in the
spool pipe, because of the close proximity of the outlet
of the swirler, where its positive maximum is up to four
times the bulk velocity at a radial position of r/R ˆ 0.8
and then decays downstream.
Examination of the level of the rms velocity compo-
nents in the axial, tangential and radial directions, as
shown in Figs. 3±5, reveals the existence of an anisotro-
pic turbulent ®eld especially in the inner region of the
¯ow up to r/R ˆ 0.4. This is usually the case with strong-
ly swirling ¯ows and implies for the calculation of such
¯ows the use of a turbulence model that can mimic the
presence of anisotropy in the turbulent ®eld.
Pro®les of the mean and rms values of three compo-
nents of velocity in the exit plane represented by the an-
nular space between the ba‚e and tube wall are
presented in Fig. 9 for case 1. The same pro®les within
experimental errors were measured for case 2. It can
be seen that the ¯ow is predominantly in the down-
stream direction with a uniform swirl velocity distribu-
tion of the tangential component in the annular space
between the ba‚e and the wall of the pipe. The boun-
dary values of velocity are thus given in sucient detail
to be used in numerical simulation exercises.
Fig. 9. Velocity distribution at exit plane.

4. Practical signi®cance

Evidently, isothermal swirling ¯ows have been and


are still being investigated and although much progress
has been achieved, many of their features remain poorly
understood. For example, questions on what governs
the presence of an on-axis or an o€-axis recirculating
bubble and its spatial extent cannot be easily answered.
It is also known that such ¯ows remain dicult to pre-
dict in all their details (see [11,16]).
The practical signi®cance of the present work can be
discussed under the above two comments. First, data is
presented that describes the velocity ®eld generated by a
typical model swirler of industrial relevance. Second,
most of the previous work on the numerical simulation
of swirling ¯ows made use of inlet boundary conditions
located downstream of the swirler, in axisymmetric ge-
ometries. The present con®guration allows the computa-
tions to be performed with inlet boundary conditions
prior to entering the swirler and thus the solution do-
main would include the swirler. Although, it is generally
accepted that predictions can be sensitive to the location
at which computations are initiated. Nonetheless, from
Fig. 8. Radial velocity, mean and rms, case 2. a practical point of view, this represents a viable prac-
236 L. Khezzar / Experimental Thermal and Fluid Science 16 (1998) 230±236

tice, since in general, detailed inlet conditions will not be Acknowledgements


available. It is therefore hoped, that such an exercise
should permit assessment of the extent to which reliable This work was carried at Imperial College of Science
¯ow predictions in the absence of combustion, including Technology and Medicine while the author was a visit-
the swirler in the solution domain, can be obtained. ing researcher. Financial support provided by the DTI
is gratefully acknowledged. Thanks go to Professor
J.H. Whitelaw of Imperial College, London and Mr.
Roy Norster of European Gas Turbines Ltd., for useful
5. Conclusions discussions.

The present work investigated a turbulent con®ned


swirling ¯ow in the near ®eld of a model radial swirler References
with two di€erent geometrical con®gurations. A combi-
nation of an o€-axis recirculation zone with low positive [1] R.M.C. So, A. Ahmed, H.C. Mongia, Jet characteristics in
mean velocities on the axis of the tube was observed in con®ned swirling ¯ow, Experiments in Fluids 3 (1985) 221±230.
both con®gurations of cases 1 and 2. The resulting com- [2] M.A. Habib, J.H. Whitelaw, Velocity characteristics of con®ned
coaxial jets with and without swirl, J. Fluids Engineering 102
plex ¯ow®eld contained a pair of counter-rotating vor-
(1980) 47±53.
tices with relatively high turbulence intensity but low [3] B.T. Vu, F.C. Gouldin, Flow measurements in a model swirl
mean velocities inside the inner core. The swirl velocity combustor, AIAA Journal 20 (5) (1982) 642±651.
pro®les revealed a forced vortex in the inner region [4] H. Altgeld, W.P. Jones, J. Wilhelmi, Velocity measurements in a
and a free vortex in the outer one. The strength of the con®ned swirl driven recirculating ¯ow, Experiments in Fluids 1
forced vortex was maintained with downstream distance (1983) 73±78.
in the measurement region. In case 2, the ¯ow ®eld in the [5] D.L. Rhode, D.G. Lilley, D.K. Mclaughlin, Mean ¯ow®elds in
spool pipe assumed a forced-vortex distribution over the axisymmetric combustor geometries with swirl, AIAA Journal 21
entire pipe with high swirl velocities and adopted down- (4) (1983) 593±600.
stream a smaller radius inner core than in case 1. The [6] M.P. Escuider, J.J. Keller, Recirculation in swirling ¯ow: A
manifestation of vortex breakdown, AIAA Journal 23 (1) (1985)
pro®les of the rms components of the velocity con®rm
111±116.
the anisotropy of the turbulence ®eld especially within [7] A.S. Nejad, S.P. Vanka, S.C. Favaloro, M. Samimy, C.
the vortices region. Langenfeld, Application of laser velocimetry for characterization
The ¯ow is of sub-critical nature and this fact as- of con®ned swirling ¯ow, Journal of Eng. for Gas turbines and
sumes particular signi®cance for the location of the out- Power 111 (1989) 36±45.
let boundary conditions in calculation exercises. [8] S. Hogg, M.A. Leschziner, Computation of highly swirling
However, detailed velocity distributions at the exit plane con®ned ¯ow with a Reynolds stress turbulence model, AIAA
in the annular space between the ba‚e and pipe wall, Journal 27 (1) (1989) 57±63.
were obtained and can thus be used to set up computa- [9] S.N. Singh, D.P. Agrawal, Velocity predictions of contraswirling
tional models for validation of calculation methods of jets in a suddenly expanding con®nement, AIAA Journal 25 (1)
(1987) 161±163.
such ¯ows.
[10] M.T. Abujelala, D.G. Lilley, Con®ned swirling ¯ow predictions,
AIAA paper 83-0136, 1983.
[11] B.E. Launder, Modelling the formation and dispersal of stream-
Nomenclature
wise vortices in turbulent ¯ow, Aeronautical Journal 99 (990)
(1995) 419±431.
A Constant [12] F. Durst, A. Melling, J.H. Whitelaw, Principles and Practice of
B Constant Laser-Doppler Anemometry, Academic Press, New York, 1981.
r Radial position [13] J.H. Faler, S. Leibovich, An experimental map of the internal
R Downstream pipe radius structure of a vortex breakdown, J. Fluid Mech. 86 (2) (1978)
R0 Upstream pipe radius 313±335.
U Mean axial velocity component [14] J. Bornstein, M.P. Escudier, LDA measurements within a vortex
u0 Axial rms velocity component breakdown bubble, International Symposium on Application of
Laser-Doppler Anemometry to Fluid Mechanics, Lisbon, Portu-
Ub Bulk velocity
gal, paper 10.3, 5±7 July, 1982, pp. 253±263.
V Mean radial velocity component [15] G.L. Brown, J.M. Lopez, Axisymmetric vortex break down, J.
v0 Radial rms velocity component Fluid Mech. 221 (1990) 553±576.
W Mean swirl velocity component [16] D.G. Sloan, P.J. Smith, L.D. Smoot, Modeling of swirl in
w0 Swirl rms velocity component turbulent ¯ow systems, in: Progress in Energy Combust. Sci., 12
x Axial position (1986) 163±250.

View publication stats

You might also like