Professional Documents
Culture Documents
net/publication/258076520
CITATIONS READS
10 399
4 authors, including:
15 PUBLICATIONS 130 CITATIONS
Universidad de Jaén
80 PUBLICATIONS 1,405 CITATIONS
SEE PROFILE
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Stability, sensitivity and control of jets and wakes behind bluff bodies View project
All content following this page was uploaded by C. Martínez-Bazán on 30 May 2014.
Confined jet flows are of interest for many practical en- slender jet of characteristic length R j a that can be described
gineering applications. The configuration shown in Fig. 1, in with relative errors of order R ⫺2 j with the boundary-layer
which an axisymmetric jet of radius a, with Ⰶ1, dis- 共BL兲 approximation. The same incompressible fluid of den-
charges into a coaxial ducted stream of radius a, is relevant sity and kinematic viscosity is assumed to be present in
in particular to ejector systems and combustion chambers. both streams. In the analysis, the ratio of the inner to the
Most of the previous works have been devoted to the case of outer radii will be assumed to be very small, as done in
turbulent flows, corresponding to the high-Reynolds-number Ref. 5 for the confined jet with no coflow, and the momen-
jets most often encountered in applications. Of particular rel- tum flux of the coflow J c will be assumed to be comparable
evance are the pioneering experimental and theoretical to that of the jet, i.e., Craya–Curtet numbers C⫽(J c /J j ) 1/2
analyses of Craya and Curtet1 and Curtet,2 which addressed of order unity, the latter being the appropriate distinguished
as a central issue the emergence of regions of reverse flow limit for studying the emergence of reverse flow. Although
near the confining wall when sufficiently weak coflow is much research has been done on confined laminar jets 共see,
present. A dimensionless parameter based on similarity con- e.g., the literature review given in Ref. 5兲, the study of co-
siderations was proposed to characterize the resulting flow, flow laminar jets in this limit has not been attempted in the
similar to that previously proposed by Thring and Newby in past. As shown below, the use of the BL approximation pro-
their study of turbulent, coflow diffusion flames,3 for which vides a simple way to evaluate, for the moderately large
the recirculating flow provides a key stabilizing mechanism. values of R j typical of many small-scale flow devices, the
For the case of uniform coflow investigated by Craya and critical conditions for existence of recirculating flow, a result
Curtet, their parameter reduces to C⫽(J c /J j ) 1/2 in the limit that is independent of R j . The resulting flow structure de-
→0 of small inner jet radius, where J c and J j represent the pends on the shape of the initial coflow velocity profile, but
momentum fluxes of the coflow and of the jet, respectively. A is independent of the downstream boundary conditions, pro-
detailed account of experimental results, theoretical analyses vided the confining duct is longer than the resulting recircu-
and approximate integral solutions for turbulent Craya– lating region. The BL analysis also gives the dependence of
Curtet jets can be found in Ref. 4. the flow scales, e.g., length of recirculating region, on R j .
Unlike the above mentioned studies, we address here the The characteristic lengths R j a and a will be used as
laminar flow arising for values of the jet Reynolds number scales to define the dimensionless axial and radial coordi-
R j ⫽ 关 J j /( ) 兴 1/2/ Ⰷ1 below a certain critical value, when nates, x and r, while the characteristic velocities u 0
the resulting symmetric solution remains steady, giving a ⫽ 关 J j /( ) 兴 1/2/a and /a will be used as scales for the di-
mensionless axial and radial velocities u and v and the char-
a兲
Electronic mail: a.revuelta@ciemat.es acteristic dynamic pressure u 20 will be employed to scale
冉 冊
K 2 ⫽⫺17.0961, on the other hand, needs to account for the
u u dp 1 u following correction f 2 in the wall boundary layer, and also
u ⫹ v ⫽⫺ ⫹ r , 共2兲
x r dx r r r for the fluid volume ⫺8 x entrained by the jet. Combining
to be integrated with boundary conditions the uniform core, the boundary layer and the jet yields
and
u/ r⫽ v ⫽0 at r⫽0
u⫽
3
8x冋1⫹
3 r
64 x 冉 冊册2 ⫺2
⫹
C
冋 冉冊
ḟ ⫹
2 1
C
x
1/2
ḟ 2 ⫹ . . . 册 共5兲
first approximation as a point source of momentum with neg- represents uniformly the velocity distribution for ⰆxⰆ1.
ligible volume flux. Initially, the jet develops over a distance Hence, the limit of large jet Reynolds numbers R j Ⰷ1
of order x⬃Ⰶ1 as a free jet discharging into a stagnant and large expansion ratios Ⰶ1 reduces the problem in the
atmosphere, approaching for distances x in the range Ⰶx first approximation to that of integrating 共1兲 and 共2兲 with
Ⰶ1 the self-similar Schlichting solution6 u⫽3/(8x) 关 1 boundary conditions 共3兲 and with the initial conditions given
⫹(3/64)(r/x) 2 兴 ⫺2 independent of the shape of the initial in 共4兲 replaced in the integrations with the profiles 共5兲 共uni-
velocity profile u j . This solution must be supplemented with form coflow兲 or 共6兲 共fully developed coflow兲 evaluated at x
the solution emerging in the coflow stream at small distances Ⰶ1. As anticipated above, the Craya–Curtet number C
ⰆxⰆ1, which will be obtained below for two limiting emerges as the single governing parameter, which was varied
210 Phys. Fluids, Vol. 16, No. 1, January 2004 Revuelta et al.
FIG. 3. The variation with C of the upstream and downstream ends of the
recirculation region and of the maximum value of the stream function with
uniform and parabolic coflow; the dashed lines are the results with parabolic
coflow represented in terms of C * .
the curves p(x) relating the pressure jump, the duct length, the plots for C⫽0.40, C⫽0.28, and C⫽0.17. It can be ex-
and the momentum flux of the inner jet with the resulting pected that the accuracy of the BL approximation would im-
coflow. prove for larger R j , a point supported by comparisons with
As seen in the comparison of Fig. 3, the flow structure integrations of the complete Navier–Stokes equations com-
depends on the velocity distribution in the coflow. When puted in Ref. 5 for confined jets with no coflow. Neverthe-
parabolic coflow is considered, the existence of relatively less, considering the moderate values of the parameters R j
small velocities in the vicinity of r⫽1 clearly facilitates the ⫽35 and ⫺1 ⯝34.8 used in the experiment, the comparisons
appearance of reverse flow for a larger C c . Use of an alter- of Fig. 2 are sufficiently satisfactory, and demonstrate the
native governing parameter, C * , incorporating the influence applicability of the BL approximation combined with the
of the nonuniform coflow momentum distribution, may re- limit ⫽0 to describe laminar Craya–Curtet jets with finite
duce this dependence. Among the different possibilities, one Reynolds number and finite expansion ratios.
may for instance use the coflow volume flux Q c to define It is worth mentioning that the boundary-layer computa-
C * ⫽ 关 Q c /( 冑 a) 兴 /(J j / ) 1/2 as the ratio of the characteristic tion given above also applies to turbulent Craya–Curtet jets,
coflow velocity Q c /( a 2 ) to the characteristic velocity u 0 when use is made of a constant turbulent viscosity T to
⫽(J j / ) 1/2/a. This alternative definition yields C * ⫽C for model turbulent stresses. The value C c ⫽0.65 determined
uniform coflow and C * ⫽)C/2 for parabolic coflow, so that with this simple Boussinesq approach agrees reasonably well
in terms of C * criticality is achieved at C c* ⯝0.65 for uni- with the value C c ⯝0.75 obtained experimentally10 for turbu-
form coflow and at C * c ⯝0.67 for parabolic coflow. Also re- lent confined jets with uniform coflow and Ⰶ1. In using
markable is the agreement observed in Fig. 3 when the re- this approximation, it seems appropriate to select for the con-
sults of the parabolic coflow are represented in terms of C * . stant turbulent viscosity the value T ⫽0.0316关 J j /( ) 兴 1/2
To validate the BL results given above, experiments for which the associated Schlichting solution matches the
were performed for moderately large values of R j and ⫺1 . spreading angle measured experimentally for a free turbulent
A water jet was injected through a hypodermic vertical jet.11 Note that the apparent Reynolds number based on the
needle of inner radius equal to a⫽0.597 mm into a coaxial turbulent viscosity becomes R j ⫽ 关 J j /( ) 兴 1/2/ T ⫽31.62, so
water stream confined by a cylindrical glass duct of radius that in this approximation the streamlines given in Fig. 2 for
a⫽20 mm (⬇0.03). The flow rates were accurately con- R j ⫽35 represent approximately the mean motion of turbu-
trolled by separate flowmeters provided with high-precision lent Craya–Curtet jets.
valves. Constant-head water supplies were employed to
avoid pressure fluctuations in the feeding lines. To visualize ACKNOWLEDGMENTS
the streamlines, the coflow stream was seeded with neutrally
buoyant particles of mean diameter equal to 5 m. A 2 W This collaborative research was supported by the Span-
argon–ion laser source was employed to produce a laser ish MCyT under Project No. DPI2002-04550-C07. Fruitful
sheet using a cylindrical lens. The light scattered by the par- discussions with Dr. Marcos Vera are gratefully acknowl-
ticles was captured with a charge coupled device camera and edged.
digitally enhanced afterwards, producing the images shown
in Fig. 2. Streamlines are clearly visible in regions of suffi- 1
A. Craya and R. Curtet, ‘‘Sur l’evolution d’un jet en espace confine,’’ C.
ciently large velocity, whereas in the vicinity of the wall and R. Acad. Sci., Paris 241, 621 共1955兲.
also within the toroidal vortex for 0⬍C c ⫺CⰆ1, where the 2
R. Curtet, ‘‘Confined jets and recirculation phenomena with cold air,’’
flow is nearly stagnant, the existence of still standing par- Combust. Flame 2, 383 共1958兲.
3
M. W. Thring and M. P. Newby, ‘‘Combustion length of enclosed turbulent
ticles results in a grainy region. For the experimental results jet flames,’’ Proc. Combust. Institute 4, 789 共1953兲.
reported here, the Reynolds number of the jet was main- 4
N. Rajaratnam, Turbulent Jets 共Elsevier, Amsterdam, 1976兲, pp. 148 –183.
tained at the constant value R j ⫽35, while the coflow was 5
A. Revuelta, A. L. Sánchez, and A. Liñán, ‘‘Confined axisymmetric lami-
reduced to obtain decreasing values of C. Recirculation was nar jets with large expansion ratios,’’ J. Fluid Mech. 456, 319 共2002兲.
6
H. Schlichting, ‘‘Laminare strahlausbreitung,’’ Z. Angew. Math. Mech. 13,
first observed for C c ⫽0.68, a value slightly above that pre- 260 共1933兲.
dicted by the BL calculation. For smaller values of C both 7
L. Rosenhead, Laminar Boundary Layers 共Dover, New York, 1963兲, pp.
the length and the width of the recirculating region are seen 439– 446.
8
S. Goldstein, Modern Developments in Fluid Mechanics 共Clarendon, Ox-
to increase, in agreement with the numerical results. The
ford, 1938兲.
discrepancy in C c results in larger differences in flow struc- 9
H. Blasius, ‘‘Grenzschichten in Flüssigkeiten mit kleiner Reibung,’’ Z.
ture in comparisons for C⫽0.68 and C⫽0.54, with no re- Math. Phys. 56, 1 共1908兲.
10
verse flow appearing in the numerical results for C⫽0.68 H. A. Becker, H. C. Hottel, and G. C. Williams, ‘‘Mixing and flow in
ducted turbulent jets,’’ Proc. Combust. Institute 9, 7 共1962兲.
and with the recirculating region determined experimentally 11
H. J. Hussein, S. C. Capp, and W. K. George, ‘‘Velocity measurements in
for C⫽0.54 being larger and located upstream from that de- a high-Reynolds-number, momentum-conserving, axisymmetric, turbulent
termined numerically. Much better agreement is observed in jet,’’ J. Fluid Mech. 258, 31 共1994兲.
View publication stats