You are on page 1of 8

Nuclear Engineering and Design 237 (2007) 10251032

Investigation of mixed convection in a large rectangular enclosure


Fenglei Niu
a
, Haihua Zhao
b
, Per F. Peterson
b,
, Joel Woodcock
c
, Robert E. Henry
d
a
Condensed Matter and Thermal Physics Group MS K764, Los Alamos National Laboratory, Los Alamos, NM 87545, USA
b
Department of Nuclear Engineering, University of California, Berkeley, CA 94720-1730, USA
c
Westinghouse Electric Company, Monroeville, PA 15146, USA
d
Fauske & Associates, Inc., Burr Ridge, IL 60527, USA
Received 25 August 2006; received in revised form 14 December 2006; accepted 15 December 2006
Abstract
This experimental research investigates mixed convection and heat transfer augmentation by gaseous forced jets in a large enclosure, at conditions
simulating those of passive containment cooling systems for Gen III+ passively safe reactors. The experiment is designed to measure the key
parameters governing heat transfer augmentation by forced jets, and to investigate the effects of geometric factors, including the jet diameter, jet
injection orientation, interior structures, and enclosure aspect ratio. The tests cover a variety of injection modes leading to ow congurations of
interest for mixing and stratication phenomena in containments under accident conditions. Correlations for heat transfer augmentation by forced
jets are developed and compared with experimental data. The characteristic recirculation speed inside the enclosure is introduced and analyzed.
Steady stratied temperature distributions are compared with model simulations of the BMIX++ code.
2007 Elsevier B.V. All rights reserved.
1. Introduction
Passive containment cooling systems (PCCS) provide the
safety-related ultimate heat sink for a new generation of pas-
sively safe reactors such as AP1000 and ESBWR. During a
LOCA accident, natural forces, such as gravity, natural circu-
lation, and a small number of automatic valves make the safety
systemwork. The natural circulation and the pipe break injection
cause the combined natural and forced convection heat trans-
fer in the containment. The steel containment vessel, in some
designs like the AP1000, provides the heat transfer surface that
removes heat from inside the containment to the outside.
Mixed convection ows have received considerable attention
since the late 1970s, and comprehensive literature reviews were
given by Incropera and Dewitt (1996). Fox et al. (1992) and
Smith et al. (1992) found that experimental results for transient
stratication of BWR pressure suppression pools could be pre-
dicted using numerical solutions of one-dimensional differential
equations describing the effect of buoyant jets on the vertical
temperature distribution. In University of California at Berke-
ley, some mixed convection and related researches have since

Corresponding author. Tel.: +1 510 643 7749; fax: +1 510 643 9685.
E-mail address: peterson@nuc.berkeley.edu (P.F. Peterson).
1990s been performed for both gaseous and liquid uids. Some
accomplishments related to gas-jet mixed convection include:
Kuhn et al. (2002) studied mixing processes and heat transfer
augmentation by a hot-air jet in a large cylindrical enclosure
heated from the bottom, and gave a correlation as a function
of Archimedes number, a uid property factor, and a geomet-
ric factor; Peterson et al. (1991) studied experimentally and
numerically transient thermal stratication in pools with shal-
low buoyant jets; Peterson (1994) showed that large enclosures
mixed by buoyant plumes and wall jets can often be expected to
stratify, and provided a criterion for assessing when the momen-
tum injected by forced jets would break down stratication in
large enclosures; Peterson and Gamble (1998) presented a scal-
ing method that could provide the basis for the design of scaled
experiments for studying jet-induced heat and mass transfer in
large enclosures.
Mixed convection is of interest and importance in a wide
variety of engineering applications. However, mixed-convection
in large rectangular enclosures has not been investigated at great
length. Limited work has been performed on natural-convection
augmentation by forced jets. Few experimental data have been
obtained on mixing and stratication phenomena inside large
three-dimensional enclosures agitated by forced-jet ows.
Much of the research has concerned laminar ow in simple
congurations and geometries. Many experiment geometries
0029-5493/$ see front matter 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.nucengdes.2006.12.011
1026 F. Niu et al. / Nuclear Engineering and Design 237 (2007) 10251032
Nomenclature
A surface area (m
2
)
Ar Archimedes number
b width parameter (m)
C coefcient
d jet diameter (cm)
D
t
diameter of the block tube (cm)
Gr
L
enclosure Grashof number
K empirical loss coefcient
L enclosure characteristic length (m)
m mass (kg)
M momentum ux (kg m/s
2
)
n unit normal vector to surface
Nu Nusselt number
P mechanical energy ux (kg m
2
/s
3
)
r radial coordinate (m)
Re Reynolds number
S surface area (m
2
)
T temperature (K)
u x-velocity (m/s)
u

uctuating x-velocity (m/s)


U velocity (m/s)
V volume (m
3
)
W mass rate (kg/s)
X distance from jet entrance (m)
z vertical coordinate (m)
Greek symbols

T
Taylors jet entrainment constant
coefcient of thermal expansion (K
1
)
dynamic viscosity (Ns/m
2
)
kinematic viscosity (m
2
/s)
density (kg/m
3
)
Subscripts
a ambient
b bulk value
bj free buoyant jet
C centerline
D drag
J jet
L using enclosure characteristic length
m mixed convection
n natural convection
o jet outlet
R recirculation
w wall
are simplied into parallel-plate channels or rectangular cavities
and the thermal boundary conditions are set to be symmetrical.
Extensive mixing experiments in large containment enclosure
are thus needed to improve key scaling, experimental, and
modeling tools for predicting mixing and transport in passive
containments and connement enclosures.
This experimental research studies mixed-convection and
heat transfer augmentation by forced jets in various directions
inside a large enclosure with a vertical cooling surface. The
experiments are performed by varying several geometric fac-
tors, including the jet diameter, jet injection orientation, ow
obstructions, and enclosure aspect ratio. The correlations of heat
transfer augmentation by forced jets are developed and tested by
experimental data. The recirculation speed inside the enclosure
is dened and analyzed. The steady stratied temperature distri-
butions are compared to the predictions made by BMIX++ code
simulation. Both scaling and modeling of stratied mixing in
large enclosures require detailed and accurate empirical models
for wall and free jets. This research effort provides experimental
results to support the development of a new, computation-
ally efcient model for mixing under the stratied conditions
that characterize large volumes in passive systems, and for
assessing the heat transfer augmentation produced by forced-jet
injection.
2. Experimental facilities and test procedures
Fig. 1 shows a schematic diagram of the experimental sys-
tems. They are an open loop composed of air supplies, heating
systems, a test section with a large insulated rectangular enclo-
sure, cooling systems, and data acquisition.
The large rectangular enclosure was constructed with the
size of 2.29 m2.29 m2.29 m. One of the walls was made
of a 0.32-cm-thick copper plate. Cooling water was circulated
through copper tubes, welded to the backside of the copper
plate, to generate a nearly isothermal surface. All the walls,
ceiling, and oor were surrounded with insulation materials,
making an adiabatic test section. The wall opposite the verti-
cal cooling surface could be moved to change the enclosure
aspect ratio. The jet tubing inserted horizontally into the enclo-
sure through this wall at different locations. The hot air entered
the enclosure from the jet and left through a 53-mm i.d. open-
ing at the bottom of the movable wall. Compressed air was
heated by helical heaters before being injected into the test
section.
Thermocouples and heat ux sensors were embedded in the
copper plate to measure the temperature and heat ux of the
cooling surface. Some thermocouples were mounted in the inlet
and outlet of the cooling water to measure the temperature differ-
ence of the cooling water for the calculation of the cooling rate.
A calibrated orice with a differential pressure transducer was
installed in the cooling water loop to measure the water ow
rate. Heating rates by injected hot air were calculated based
on the ow rate and the temperature difference between the
air inlet and outlet. The air ow rate at inlet and outlet were
measured using calibrated rotameters. The ambient temperature
distribution inside the enclosure was measured by thermocou-
ple matrix. The enclosure was looked as a control volume in
mass balance. There is no signicant pressure change in the
enclosure.
To start the experiment, the water valves and the air valves
were slowly opened until the desired air and water ow rates
were reached. Heaters were turned on and checked regularly.
F. Niu et al. / Nuclear Engineering and Design 237 (2007) 10251032 1027
Fig. 1. Schematic diagram of experimental system.
The heater outlet temperature should be monitored carefully to
avoid overheating. When the temperatures inside the enclosure
did not have signicant changes for 15 min, the systemachieved
a steady state. Data were recorded every 10 s.
Experiments were rst performed to investigate the natural
convective heat transfer in the enclosure, and the results were
compared to those from combined natural and forced convec-
tion. The natural circulation currents could develop in the large
enclosure when the wall is cooled without jet injection.
3. Experimental results and analysis
3.1. Effects of injection orientation and jet diameter on
heat transfer augmentation
To investigate how injection orientations and jet diameters
affect heat transfer augmentation, air is injected into the enclo-
sure with four directions: (A) horizontal injection toward the
cooling plate, (B) 45

upward injection, (C) vertical/up injec-


tion, and (D) 180

backward injection away from the cooling


plate. Three different size jet nozzles are employed for each
injection direction.
As illustrated in Fig. 2, the experimental data for augmenta-
tion can be well correlated by
Nu
m
Nu
n
= (1 +C
J
Ar
J
)
1/3
(1a)
where the jet Archimedes number is a function of the jet
Reynolds number, Re
J
=U
J
d
J
/, is the kinematic viscosity of
the air, and the natural convection Grashof number, Gr
L
=g
(T
b
T
w
)L
3
x
/
2
:
Ar
J
=
Re
2
J
Gr
L
x
(2a)
where U
J
is the jet velocity, d
J
the jet diameter, and L
x
the enclo-
sure characteristic length. The temperatureaverage correlation
for the natural convection is
Nu
n
= C
1
Gr
1/3
L
x
Pr
1/3
where C
1
=0.128. Eq. (1a) is the correlation of forced-jet aug-
mentation of natural convection heat transfer, which can be
derived using a combining rule for mixed convection and appro-
priate forced and natural convection models. It is a function of
the jet Archimedes number and the coefcient C
J
, which is in
the range of 5.48.5 and includes the effects of the jet mode,
injection orientation, and enclosure aspect ratio.
The magnitude of heat transfer augmentation is largest for
horizontal injection toward the cooling plate (A), and followed
by 45

injection (B), vertical/up injection (C), and 180

back-
ward injection (D). Compared to vertical/up injection, horizontal
injection has better mixing effects, and the injection toward
the cooling plate can signicantly increase the wall jet veloc-
ity at the heat transfer surface, which in turn raises heat transfer.
The data for all injection orientations asymptotically approach
1 at low Archimedes number, where heat transfer is dominated
by natural convection. It is found (more clearly for vertical/up
injection) that the experimental data, regardless of the jet diam-
eter, are clustered into groups of trend lines in accordance with
Fig. 2. Effects of injection orientation and jet diameter.
1028 F. Niu et al. / Nuclear Engineering and Design 237 (2007) 10251032
Fig. 3. Effects of enclosure aspect ratio on heat transfer augmentation.
their injection orientations, which implies that the effect of jet
diameter is weak.
3.2. Effects of enclosure aspect ratio on heat transfer
augmentation
Fig. 3 gives a comparison of the effects of varying the enclo-
sure aspect ratio. Experiments were performed with vertical/up
jet injections in a large enclosure (2.11 m2.27 m2.18 m)
and a tall narrow enclosure (1.30 m2.27 m2.18 m).
It can be seen the heat transfer augmentation increases with
reduction of enclosure aspect ratio. C
J
is equal to 5.9, and 7.7 for
the large enclosure and medium enclosure, respectively. Some
experiments (Kuhn et al., 2002) have shown that the average
velocities in the enclosure induced by jet injection increase with
reduction of enclosure aspect ratio, which can explain these
experimental results. After the buoyant jet reaches the insulated
ceiling, it transforms to a ceiling jet ow spreading out with
decreasing velocity. In a large enclosure, due to the friction,
the ceiling jet ow may dissipate completely before reaching
the vertical cooling plate. In the narrow enclosure, however,
the ceiling jet ow could have sufcient kinetic energy to reach
the corner and then turn down to accelerate the wall jet on the
cooling surface, which will assist heat transfer between air and
cooling plate. So the smaller enclosure can receive higher heat
transfer augmentation.
3.3. Effects of obstructions on heat transfer augmentation
The experimental data from a simplied, empty enclosure
will inevitably be quite different from those from a real contain-
ment with complex pipes, stairways, and other interior structures
that perturb large-scale recirculation ows. To investigate the
effects of structures on heat transfer augmentation, experiments
were performed with some ow obstructions inside the enclo-
sure.
Experiments were rst performed with the jet directly
impinging upon a 4.2-cm diameter cylinder, which was placed
between the jet nozzle and the cooling plate. With this con-
guration, the cylinder experienced large drag forces and thus
extracted substantial momentum from the jet.
If the jet hits a blocking structure such as the cylinder before
reaching the opposite wall, the loss of jet momentum can sig-
nicantly affect the effectiveness of the mixing, and thus reduce
the heat transfer rate. Except for the small region near the jet
exit or impinged surface, the jet will transit to become a fully
developed jet if it does not impinge on a blocking structure. In
this region, the distribution of the streamwise velocity across a
free expanded jet has been given by List (1982):
u(r) = U
C
e
(r/b)
2
(3)
where U
C
is the local streamwise velocity at the centerline and
b the width parameter. The fractional momentum loss due to
impingement on a cylinder can then be evaluated by
M
loss
M
=
2
_

0
C
D
u
2
(r)/2D
t
dr
_

0
u
2
(r)2r dr
(4)
where the drag coefcient is C
D
0.9 in the range of Reynolds
number of the experiments, D
t
the diameter of the blocking tub-
ing, and the air density, assumed to be constant. Integrating
Eq. (4) yields:
M
loss
M
=
C
D
D
t

2b
(5)
For a given self-similar distribution of mean velocity and
pressure, the width parameter b is related to the local jet
expanded diameter d
bj
(List, 1982):
b =
d
bj
2

2
(6)
where d
bj
was given by Peterson and Gamble (1998):
d
bj
= 4

2
T
x +d
bj0
(7)
where
T
is the Taylors jet entrainment constant, typically tak-
ing a value around 0.05, d
bj0
the jet nozzle diameter, and x is the
distance between the jet exit and the obstruction.
Then from Eqs. (5)(7) the fraction of momentum loss can
be estimated in terms of the jet nozzle diameter, blocking tubing
diameter, andthe distance betweenthe jet originandthe blocking
tubing, as
M
loss
M
=
2C
D
D
t

(4

2
T
x +d
bj0
)
(8a)
This equation is only valid in the linear decay region far from
jet exit and cooling plate. In the regions near the jet exit or
cooling surface, the ow eld has large spatial variations that
make it complicated to estimate momentumloss. Eq. (8a) shows
the fraction of jet momentum loss increases with the projected
frontal size of the blocking structure and decreases with the jet
nozzle diameter and the distance from the jet origin.
Fig. 4 shows one of the experimental results. In this experi-
ment, a 2.2 cm i.d. jet nozzle was used and the distance between
the jet exit and the cylinder was 30 cm. The experimental
results were plotted compared to those from non-obstruction
experiments. The heat transfer augmentation decreased due to
the obstruction, giving C
J
=5.5, compared to C
J
=8.5 for non-
obstruction experiment. This can be explained by the effect of
F. Niu et al. / Nuclear Engineering and Design 237 (2007) 10251032 1029
Fig. 4. Effects of large drag forces on heat transfer augmentation.
the obstruction on the jet Archimedes number. FromEq. (1a) the
heat transfer augmentation is a function of the jet Archimedes
number:
Ar
j
=
Re
2
j
Gr
L
=

2
u
2
j
d
2
bj0

2
Gr
L
(8b)
The jet momentum ux is conserved along the path of the
free jet and can be expressed as
M
d
2
bj0
u
2
j
4
(8c)
Eqs. (8b) and (8c) imply that, if density variation is neglected,
the jet momentumis proportional to the jet Archimedes number:
M
loss
M

Ar
j,loss
Ar
j
(8d)
So the loss of jet momentum due to obstruction will decrease
the jet Archimedes number, and therefore decrease the heat
transfer augmentation.
The interior structures, even though not impinged by jets
directly, can affect the mixed convection in the enclosure as well.
In order to study the degradation of forced convection augmenta-
tion due to the interior structures, 36 horizontal obstruction pipes
were installed in four vertical arrays inside the enclosure, each
array having nine pipes. Experiments were repeated to compare
the results with those from non-obstruction experiments. Fig. 5
shows the heat transfer augmentation decreases by up to 2030%
due to the ow obstructions.
3.4. Steady temperature prole measurements and
BMIX++ code simulation
Stratication is the formation of horizontal layers of constant
density. Stratication exists in a containment atmosphere if the
density of the layers decreases in the upward vertical direction
and if forced convection mixing is not sufciently strong to dis-
rupt the stable uid layers. Based on the derivations provided by
Peterson (1994), a jet or plume is not able to disturb the stable
Fig. 5. Degradation of forced convection augmentation due to obstructions.
vertical stratication if:
Ar
J
16
<<
_
1 +
d
bj0
4

2
T
L
_
2
(9)
In parallel with experimental work, we have developed and
validated the BMIX++ code (Berkeley mechanistic MIXing
code in C++) for predicting transient mixing in stratied enclo-
sures. BMIX++ code is a one-dimensional Lagrangian transient
ow and heat transfer code. It is only used for low Ar
j
cases.
For high Ar
j
, the enclosure is well mixed and can be treated
as a lumped mass. By applying the Zuber scaling methodology
(1991), scaling rules were developed showing that under strat-
ied conditions in an enclosure the governing conservation
equations for mass, momentum, energy and species reduce to
simpler one-dimensional forms (Peterson, 1994)this scaling
provides the coupled, ordinary differential equations that are
solved numerically by BMIX++.
The modeling of mixing and stratication in a large stratied
enclosure consists of two parts: modeling the ambient volume,
which can be calculated using a one-dimensional Lagrangian
method (tracking movable control volumes), and modeling sub-
structures, such as the jets, plumes, and wall boundary ows,
which can be calculated with one-dimensional integral meth-
ods or analytical methods. The two parts are coupled through
entrainment and discharge processes.
In each simulation of the experiments, four basic models are
employed in the BMIX++ code: free buoyant jet, isothermal
wall jet, small vent, and wall conduction models. Some sec-
ondary effects, such as the ceiling jet caused by impingement
of the free buoyant jet, wall jets along insulated vertical walls,
oor jets caused by impingement of the wall jets, and radiation
heat transfer, are neglected and contribute to differences between
experimental and model results. The direct interactions between
jets and walls are not considered because they typically are only
important at higher Ar
j
, where the enclosure is well mixed. A
detailed description of the BMIX++ code can be found in Zhaos
(2003).
Fig. 6 shows the temperature proles for horizontal injection
with different jet inlet temperatures. The hot air was injected
1030 F. Niu et al. / Nuclear Engineering and Design 237 (2007) 10251032
Fig. 6. Steady temperature proles for experiment and BMIX++ code simula-
tion: horizontal injection.
horizontally into the enclosure from a point near the movable
wall and at different elevations. A series of experiments with
0.14 m inlet diameter and 1/3 H inlet location were simulated
by the BMIX++ code. The cooling plate temperature was kept
at approximately 15.5

C and the inlet mass ow rate was kept


at 0.004 kg/s. Two layers exist inside the enclosure: an almost
homogenous cold lower layer and a linear stratied hot upper
layer. The overall agreement between experiment and simulation
is acceptable, but the upper layer shape predicted by the code
has smaller gradient than the experimental shape.
The following neglected phenomena in BMIX++ code may
explain the discrepancy between the experimental results and the
code prediction. The free buoyant jet impinges directly on the
cooling plate. The heat transfer within the impingement zone is
enhanced, but the heat transfer in the region above the impinge-
ment zone is reduced because the upward wall jet driven by the
impinging jet decreases the velocity of the downward natural
boundary ow along the cooling plate. Therefore, the simula-
tion gives higher temperature just above the injection point and
lower temperature for the upper layer relative tothe experimental
data.
3.5. Empirical loss coefcients and recirculation speed
correlation
When a forced jet is injected into an enclosure, it induces a
large-scale recirculating ow due to the entrainment of ambient
uid into the jet. These large-scale ows can augment heat and
mass transfer. Consideration of the strength of this recirculating
ow can allow the heat and mass transfer augmentation to be
predicted. Most of the jet kinetic energy is lost due to:
irreversible entrainment into the injected jet;
turbulent ow losses in the corners;
drag losses over structures;
wall shear along enclosure surfaces.
The loss of jet kinetic energy by entrainment is relatively insensi-
tive to the enclosure geometry, but the other losses depend upon
the speed of the large-scale recirculation ow induced in the
enclosure. The speed of this recirculation ow will adjust until
the rate of kinetic energy injected by the jet is balanced by these
losses. The strength of this recirculating ow can be character-
ized by a velocity scale. Here this characteristic velocity scale
is called the recirculation speed. The recirculation speed can
be evaluated through an enclosure mechanical energy balance.
If compression work is neglected, and the enclosure Reynolds
number is large so that the Reynolds stress work is much larger
than the viscous stress work, a steady-status mechanical energy
conservation equation for the enclosure can be given by
_
S
_
n
_
1
2
u
2
u +p u
__
dS
+
_
V
( u [ u

])dV +
_
V
g( u e
z
) dV = 0 (10)
Eq. (10) permits the recirculation speed for the enclosure to be
dened as
U
2
R
=
1
m
_
V
(u
2
) dV (11)
The mechanical energy uxes entering and leaving the enclo-
sure across its surfaces can be represented by two terms:
_
S
_
n
_
1
2
u
2
u +p u
__
dS = P
E
P
J
(12)
where P
E
is the mechanical energy ux leaving from the enclo-
sure and P
J
is the mechanical energy ux from the break jet.
Likewise, the volumetric mechanical energy sink terms can
be expressed as
_
V
( u [ u

])dV = P
D
+P
W
+
m

k=1
P
S,k
(13)
where P
D
is the mechanical energy dissipated by uid entrain-
ment in the break jet, P
W
the mechanical energy dissipated by
ow losses to the enclosure walls (both shear stress and form
losses in enclosure corners), and P
S,k
is the mechanical energy
dissipated by drag forces from ow over structures k =1 to m.
Neglecting the buoyancy work, Eq. (10) can then be
expressed as
(P
J
P
D
) +P
E
+P
W
+
m

k=1
P
S,k
= 0 (14)
If the jet does not impinge directly upon structures that would
experience large drag forces from the jet, the momentum of the
jet W
J
U
J
is conserved. If the jet does impinge and loses some
momentum, the loss can be treated with an empirical coefcient.
Thus, at the location where entrainment has equilibrated the
jet mechanically with the large-scale recirculating ow, the net
mechanical energy delivered by the jet is
P
J
P
D
=
K
J
(W
J
+W
entrained
)U
2
R
2
=
K
J
W
J
U
J
U
R
2
(15)
where K
J
is an empirical break-jet loss coefcient.
F. Niu et al. / Nuclear Engineering and Design 237 (2007) 10251032 1031
For ow leaving the enclosure, the energy loss is
P
E
=
W
J
U
2
R
2
(16)
Inside the enclosure, mechanical energy is dissipated by ow
losses to walls and corners, as well as by drag forces due to
ow over structures. For a given enclosure geometry and break
injection orientation, ow losses to walls and corners can be
correlated empirically as
P
W
= K
W
A
W

a
U
3
R
2
(17)
where A
W
is the wall surface area and K
W
an empirical enclosure
loss coefcient.
Flow losses from drag forces due to ow over structures can
be correlated empirically as
P
S,k
= C
D,k
A
S,k

a
U
3
R
2
(18)
where A
S,k
is frontal area (the area projected perpendicular to
the recirculation ow) of structure k and C
D,k
an empirical drag
coefcient.
With Eqs. (14)(18), the enclosure mechanical energy bal-
ance can then be written as
_
K
W
A
W

a
+
m

k=1
C
D,k
A
S,k

a
_
_
U
R
U
J
_
2
+
J
A
J
_
U
R
U
J
_
K
J

J
A
J
= 0 (19)
The second termof above represents the outowenergy loss that
can be neglected under low exit velocity. Then the relationship
between the recirculation speed and jet velocity can be obtained:
U
R
U
J
=
_
K
J

J
A
J
K
W
A
W

a
+

m
k=1
C
D,k
A
S,k

a
_
1/2
(20)
So the recirculation Reynolds number and the jet Reynolds can
be related by
Re
R,L
x
= Re
J,d
L
x

J
2
a
_
K
J

J
K
W
A
W

a
+

m
k=1
C
D,k
A
S,k

a
_
1/2
(21)
The large-scale recirculating ow augments heat transfer to
surfaces in the enclosure, and the strength of recirculating ow
can be used as a parameter for correlating experimental data for
forced-convection augmentation. Similar to Eq. (1a), the heat
transfer augmentation can also be expressed as
Nu
m
Nu
n
= (1 +C
R
Ar
R
)
1/3
(1b)
where the recirculation Archimedes number is dened by
Ar
R
=
Re
2
R
Gr
L
x
(2b)
With Eqs. (1b), (2b), and (21), we can get the newexpression
of the heat transfer augmentation:
Nu
m
Nu
n
=
_
1 +C
R
L
2
x
_

a
_
2

_
K
J

J
K
W
A
W

a
+

m
k=1
C
D,k
A
S,k

a
_
Re
2
J,d
Gr
L
x
_
1/3
(22)
Eq. (22) gives the quantitative effects of injection conditions
and enclosure geometry. K
J
varies from1.0 for jets which do not
impinge upon structures that generate signicant drag forces,
to a value approaching 0 for jets which are introduced into an
enclosure through highly effective momentum diffusers. K
W
is
a function of enclosure aspect ratio. The drag coefcient C
D
is
a function of Reynolds number and can take the empirical value
for ow over cylinders. C
R
is a strong function of the injection
direction.
Here, the empirical loss coefcients are estimated using Eq.
(20) andthe data of heat transfer augmentationobtainedhere. We
rst consider a large enclosure, which does not contain imping-
ing obstructions or interior structures. In this condition, Eq. (22)
can be simplied into:
Nu
m
Nu
n
=
_
1 +C
R
L
2
x
_

a
_
2
_

J
K
W
A
W

a
_
Re
2
J,d
Gr
L
x
_
1/3
(23)
K
W
is constant for the xed aspect ratio and C
R
varies with injec-
tion direction. We can nd values for these parameters which
give the best correlation, using Eq. (23), for the experimental
values of heat transfer augmentation that have been measured.
After nding C
R
for each injection direction, in the same way we
can continue to nd each K
W
by changing enclosure aspect ratio,
K
J
by adding impinging obstructions, and C
D
by adding interior
structures. Table 1 gives optimal values for the loss coefcients
and Figs. 79 give plots of predicted values versus experimental
data of the heat transfer augmentation, which show good agree-
ment. The heat loss from insulated walls was not included in the
experimental data, which made the heat transfer rates obtained
from experimental date consistently less than predicted values.
3.6. Average relative errors
The error analysis for the experiments was performed using
the standard error propagation methods. The relative errors for
all parameters can be calculated from the uncertainties of the
apparatus and instruments used in the measurement. Table 2 lists
Table 1
Optimal values of loss coefcients
Loss coefcients Optimal values
K
J
1.0 (no blocks); 0.3 (4.2 cm o.d. cylinder
a
); 0
(momentum diffusers)
K
W
0.3 (1.7 m2.27 m2.18 m); 0.14
(0.85 m2.27 m2.18 m)
C
D,k
1.2 (2 in. o.d. horizontal tubing)
C
R
300 (0

injection); 120 (45

injection); 100
(90

injection); 65 (180

injection)
a
At 30 cm from the jet outlet.
1032 F. Niu et al. / Nuclear Engineering and Design 237 (2007) 10251032
Fig. 7. Large enclosure, 0

injection, no blocks, no interior structures.


Fig. 8. Large enclosure, 0

injection, impinging a block, no interior structures.


Fig. 9. Large enclosure, 90

injection, no blocks, 36 interior pipes.


Table 2
Average relative errors of major dimensionless parameters
Name Relative error (%)
Jet Reynolds number 3.1
Enclosure Grashof number 8.7
Average Nusselt number (natural convection) 2.9
Average Nusselt number (mixed convection) 28.9
Jet Archimedes number 10.6
Heat transfer augmentation 29.1
the average relative errors of major dimensionless parameters
used in the experiments.
4. Conclusions
The experimental studies have investigated heat transfer
under combined natural and forced convection with a variety of
jet injection modes. The heat transfer augmentation by forced
jets is controlled by jet Archimedes number, uid properties,
injection orientation, ow obstructions, and enclosure aspect
ratio. The experimental data are well correlated by developed
correlations. The strength of the recirculating ow inside the
enclosure can be characterized by the recirculation speed. The
optimal empirical loss coefcients in the correlation of heat
transfer augmentation were selected, and based on the selected
parameters, the predicted heat transfer augmentation values
agree well with experimental data. For buoyant jets (low Ar
j
)
the steadystratiedtemperature distributions inside anenclosure
can be predicted by BMIX++ code simulation.
Acknowledgements
Funding for this research was provided by the U.S. Depart-
ment of Energy(DOE) under the NEERResearchGrant Program
and by Westinghouse Electric Company, as part of the research
to improve understanding of mixing and heat transfer augmenta-
tion by buoyant and forced jets in reactor containments. The rst
author gratefully acknowledges all the members in the Thermal-
Hydraulics Laboratory, Department of Nuclear Engineering,
University of California at Berkeley, who provided so much kind
help and made construction of the experimental facility possible.
References
Fox, R.J., et al., 1992. Temperature distribution in pools with shallow buoyant
jets. In: Proceedings of the Fifth International Topical Meeting on Nuclear
Reactor Thermal Hydraulics (NURETH-5), September 2124, Salt Lake
City, Utah, pp. 12271234.
Incropera, F.P., Dewitt, D.P., 1996. Fundamentals of Heat and Mass Transfer.
John Wiley & Sons, New York, pp. 515516.
Kuhn, S.Z., Kang, H.K., Peterson, P.F., 2002. Study of mixing and augmentation
of natural convection heat transfer by a forced jet in a large enclosure. J. Heat
Transfer 124 (4), 660666.
List, E.J., 1982. Turbulent jets and plumes. Ann. Rev. Fluid Mech. 14, 189212.
Peterson, P.F., Rao, I.J., Schrock, V.E., 1991. Transient Thermal Stratication
in Pools with Shallow Buoyant Jets, Symposium on Nuclear Reactor Ther-
mal Hydraulics. In: Hassan, Y.A., Hochreiter, L.E. (Eds.), Nuclear Reactor
Thermal Hydraulics, HTD-Vol. 190, ASME, New York, pp. 5562.
Peterson, P.F., 1994. Scaling and analysis of mixing in large stratied volumes.
Int. J. Heat Mass Transfer 37 (1), 97106.
Peterson, P.F., Gamble, R., 1998. Scaling for forced-convection augmentation
of heat and mass transfer in large enclosures by injected jets. Trans. Am.
Nucl. Soc. 78, 265266.
Smith, B.L., et al., 1992. Analysis of single-phase mixing experiments in open
pools. Thermal Hydraulics of Advanced and Special Purpose Reactors,
ASME HTD-Vol. 209, pp. 91100.
Zhao, H., 2003. Computation of mixing in large stably stratied enclosures.
Ph.D. Dissertation. University of California, Berkeley.
Zuber, N., 1991. An Integrated Structure and Scaling Methodology for Severe
Accident Technical Issues Resolution, Appendix D, NUREG/CR-5809, U.S.
Nuclear Regulatory Commission.

You might also like