International Journal of Heat and Mass Transfer 76 (2014) 128143

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

journal homepage: www.elsevier.com/locate/ijhmt

Review

Numerical simulation of the heat transfer from a heated plate

with surface variations to an impinging jet
J. Ortega-Casanova a,, F.J. Granados-Ortiz b
a
b

Fluid Mechanics Group, E.T.S. Ingeniera Industrial, Universidad de Mlaga, C/Dr. Ortiz Ramos s/n, 29071 Mlaga, Spain
School of Computing and Mathematical Sciences, University of Greenwich, Old Royal Naval College, Park Row, London SE10 9LS, UK

a r t i c l e

i n f o

Article history:
Received 20 December 2013
Received in revised form 6 March 2014
Accepted 9 April 2014
Available online 15 May 2014
Keywords:
Heat transfer
Numerical simulations
Impinging jets
Turbulent ows
Heated plates
Swirling jets
Dimpled plates
Bumped plates

a b s t r a c t
The study of heat transfer between impinging jets and non-uniform heated plates is presented here to
analyse if surface variations along the plates, (i.e. dimples, bumps, and bumps&dimples, as we study
here), can improve the heat transfer phenomenon. To that end, numerical simulations of the impingement of two different types of axisymmetric turbulent jets on a non-at plate, located at a known
distance H from the jet exit, have been conducted. The cylindrical jet used, of diameter D, is created by
a swirl generator nozzle that, depending on its conguration, can produce jets with high or low swirl
intensity levels. Different values of non-dimensional nozzle-to-plate distance, H=D, have been studied,
as well as different values of the Reynolds number, Re. To know whether or not surface variations along
the plate improve the heat transfer between the impinging jet and the plate, our results are compared
with those obtained when a at plate is used.
2014 Elsevier Ltd. All rights reserved.

Contents
1.
2.
3.
4.
5.

6.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Preliminary definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The impinging jet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Numerical considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1.
Nozzle-to-plate distance effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.
Surface variations effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.3.
Nozzle-type effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Conflict of interest statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction
Jet impingement on heated surfaces is frequently used as a tool
to enhance heat transfer between them, especially at the stagnation
point. Heat transfer has also a major effect at the region where the
jet impinges on the surface, due to the development of the
Corresponding author. Tel.: +34 951952382.
E-mail addresses: jortega@uma.es (J. Ortega-Casanova), F.J.GranadosOrtiz@
greenwich.ac.uk (F.J. Granados-Ortiz).
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2014.04.022
0017-9310/ 2014 Elsevier Ltd. All rights reserved.

128
130
131
131
134
135
137
139
143
143
143

boundary layer along the surface, where heat exchange processes

take place. The interest of studying this heat transfer mechanism
through impinging jets is owing to its many engineering
applications, such as the heat transfer of blades in gas turbines
[1], the cooling of electronic devices [26], the heat transfer on
windshields of vehicles [7], or cooling in grinding processes [8],
between others.
In all the industrial applications presented above, a jet impinges
on a surface whose form depends on the particular application
under study. Normally, the surface is considered as a at plate with

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

129

Nomenclature
cp
CFD
ds
D
e
G
GCI
h
H
k
K
L
LDA
n
Nr
Nu
Nu
Nz
p
PIV
Pr
q
Q
r; h; z
R
Re
S
T
u; v ; w
UDF

v0

~
V
Wc
y
Y

uid heat capacity

computational uid dynamics
differential surface element
diameter of the jet
specic internal energy
production of turbulence due to mean velocity gradients
grid convergence index
enthalpy
distance from the nozzle to plate
turbulent kinetic energy
thermal conductivity of the uid
swirl parameter
laser doppler anemometry
nth computational grid
radial number of nodes
Nusselt number
area-weighted average Nusselt number
axial number of nodes
pressure
particle image velocimetry
Prandtl number
heat ux
ow rate
radial, azimuthal and axial coordinates
radius of the impinged surface
Reynolds number
impinged surface
temperature
radial, azimuthal and axial velocities
user-dened function
velocity uctuations
velocity vector
characteristic velocity
dimensionless wall distance
turbulent dissipation

Symbols
N
Nusselt coefcients ratio
N
area-weighted average Nusselt coefcients ratio

the jet impinging perpendicular on it (see 9], for a review). However, less studies have been carried out on non-uniform surfaces.
In Ekkad and Kontrovitz [10], can be found an experimentally
study of the effect of dimple location in a plate and the effect of
dimple depth, for different jets with different Reynold numbers.
A similar study can be found in Kanokjaruvijit and MartinezBotas
[11], where various jets impinge on a staggered array of hemispherical dimples with the consideration of various parametric
effects, such as Reynolds number, jet-to-plate distance, depth of
the dimples, and curvature of the dimples for both impinging on
dimples and impinging on at separation portions, showing that
the variations on the shape of the plate (concretely with shallow
dimples) are able to enhance the heat transfer up to a 70% with
respect to the plate one and showing that dimples are more
effective when a strong crossow is present.
Other interesting studies of the shape variation, specically in
terms of seeing how one single dimple/bump is relevant, could
be the one by Imbriale et al. [12], where the heat transfer between
a concave surface and a row of air jets impinging on it is studied by
an experimental study by varying the inclination of the jets, pitch,
impinging distance, Mach and Reynolds numbers. In ztekin et al.
[13], an experimental and numerical study is carried out to

R
S2

nozzle #1
nozzle #2

Greek symbols
d
Diracs delta
D
representative mesh size
Dzp
position of the rst node from the plate

turbulent kinetic energy dissipation
c
grid renement factor
C
effective diffusivity
l
uid viscosity
m
uid kinematic viscosity
x
specic turbulent dissipation rate
q
uid density
r
standard deviation. Heat transfer uniformity
s
computational time
e
relative error
.
observed order of accuracy
Subscript
0
a
coarse
e
ne
it
eff
ext
i; j
j
p
t

magnitude evaluated at r 0
approximate
coarse grid
Richardson extrapolation value
ne grid
iterations
effective
extrapolate
coordinate direction in compact tensor notation
jet
plate
turbulent

Superscript
b
bumped plate
d
dimpled plate
db
dimpled&bumped plate
f
at plate

investigate the turbulent slot jet impingement cooling characteristics on concave plates (big dimple) by varying the surface
curvature and the Reynolds number (around ten times lower
Reynolds numbers than those used in the present paper). The more
relevant outcomes of this research were that both the average and
stagnation point Nusselt numbers decrease when the nozzle-tosurface distance increases, both the average and stagnation point
Nusselt numbers increase when Reynolds number increases, and
it is disclosed that the surface curvature increases the average
Nusselt number from a depth value.
The analysis of a bump in a at plate can be seen in Zhang et al.
[14], combining both PIV and numerical simulation. In this paper,
the single jet impinges on the protrusion and the local Nusselt
number increases with its presence, obtaining relevant conclusions
such as the local Nusselt number increases when the depth
increases, and the average Nusselt number increases with bump
relative depth and the jet Reynolds number.
Another interesting variation on plate, that also certies how
important the selection of its conguration is, in heat transfer
terms, is the inclination. This particular variation is not going to
be treated in the present research, but can be noticed in studies
such as Beitelmal et al. [15], where an experimental analysis of

130

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

the effect of the inclination of an impinging two-dimensional air

jet on the heat transfer from a uniformly heated at plate was
developed, showing that the maximum Nusselt number decreases
when the inclination decreases; or the already mentioned Roy and
Patel [7].
Due to the relevant impact of the shape on heat transfer by
using dimples/bumps noticed in the literature explained above,
the study presented in this paper is considered by the authors as
a relevant research in the heat transfer eld. This relevance is
supported by the combination of the shape variation itself and
the addition of an impinging swirling jet to the test cases, which
was identied as a type of jet that enhances heat transfer in
comparison with the non-swirling one (see [16]).
In this article, we mainly focus our attention on how the heat
transferred from the plate to the jet can be affected by four aspects:
the shape variation of the plate surface, the type of jet created by
the nozzle, the nozzle-to-plate distance, and the Reynolds number
of the jet. Once the effects of these parameters have been studied,
we will be able to know how the heat transfer can be enhanced,
comparing our results with those of a at plate under the same
conditions.
Regarding the plate surface, it will be modied by using small
axisymmetric dimples, bumps, or both on it. With these changes
on the plate surface, we will seek the conguration that provides
the highest heat exchange in comparison with a at plate, when
the same type of jet and nozzle-to-plate distance are used.
Respecting to the jet, two kinds of nozzle congurations are
used, so two families of jets can be generated, each of them with
seven different ow rates, or Reynolds numbers. These jets have
been previously experimentally characterised by the LDA (Laser
Doppler Anemometry) technique and mathematically modelled
(see [17]), and whose generator nozzle is shown in Fig. 1. The
mathematical modelling of the jet is used in the numerical simulations as a boundary condition, such as it was also done in the heat
transfer study by Ortega-Casanova [16].
Regarding the nozzle-to-plate distance, three different values of
the non-dimensional parameter H=D were used: H=D 5; 10; 30.
Some preliminary denitions are given in the next section. A
brief description of the experimental device to generate the jet will
be given in Section 3, together with some details of its mathematical model. Next, in Section 4, some details of the computational
considerations taken into account, will be described, as well as, a
description of the geometry under study. Section 5 will be dedicated to present and discuss the results. Finally, the conclusions
will be presented in Section 6.
2. Preliminary denitions
In this section, different parameters and variables appearing
along the document will be introduced, as well as different characteristic magnitudes and dimensionless variables.
The physical problem is going to be considered symmetric along
the azimuthal direction h, that is, the problem will be axisymmetric. This kind of problems are usually described by cylindrical polar
coordinates r; h; z, with the velocity vector components dened as
~
Vr; h; z u; v ; w, where u; v and w are the velocities in radial,
azimuthal and axial direction, respectively. As characteristic
length, the exit nozzle diameter D is used, while the characteristic
velocity W c is based on the ow rate Q through the nozzle and
dened as

Wc

4Q

pD2

Regarding the uid, it will be considered incompressible with constant physical properties, such as the density q, the viscosity l or

Fig. 1. Nozzle used to generate the jets (dimensions in mm).

the thermal conductivity K. All these magnitudes already dened

will allow us to introduce one of the dimensionless numbers
governing the problem, the Reynolds number:

Re

qW c D
:
l

As in Ortega-Casanova et al. [17], for each nozzle blade conguration, seven ow rates were used, which also gives seven Reynolds
numbers:

Re  0:7; 0:9; 1:1; 1:3; 1:5; 1:7; 1:9  104 :

Usually, in heat transfer problems, the heat ux q between two

media (the solid hot plate and the uid, in our case) is quantied,
in a dimensionless way, by means of the Nusselt number Nu dened
as

Nur

qrD
;
KT p  T j

where T p is the temperature of the heated plate (constant) and T j is

the temperature of the jet once it leaves the nozzle (constant, too).
As can be seen, in (4), both the Nusselt number and q present radial
dependence, so two values of Nu will be taken into account in order
to characterise the heat transfer process: the Nusselt number at the
stagnation point Nu0  Nur 0, and the area-weighted average
Nusselt number Nu, dened as

Nu

1
S

Nurds;

where S represents the surface on which the jet impinges (with ds a

differential surface element on it). Nu0 is a local measure of the heat
transfer at r 0 on the plate, while Nu is a global measure of the
heat transfer taking into account the surface S of the plate.
Since the aim of this study is to know whether a plate with
surface variations works better than a at one, two more parameters will be dened to quantify the comparison: the stagnation
Nusselt coefcients ratio,
k

N0

Nuk0
Nuf0

with k b; d; db;

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

and the area-weighted average Nusselt coefcients ratio,

Nk

Nu
with k b; d; db;
Nuf

where f stands for at plate and b, d, db for bumped, dimpled,

dimpled&bumped plate. Therefore, values of these parameters
above unity mean that the heat transferred (locally or globally,
depending on the parameter analysed) from the plate with surface
variations is greater than from a at plate, whereas values under
unity mean just the opposite.
In order to have comparable results, the radius of the surface S
on which Nu was calculated by means of (5), was the same for all
Nu calculated in the work. Obviously, when dimples and/or bumps
are used, the surface S is slightly higher than for a at plate. In
particular, Sb Sd 1:056Sf and Sdb 1:123Sf .
Another aspect to take into account is that, due to the non-uniform impinged surface, a bad performance in terms of heat transfer
uniformity is expected. This uniformity is typically measured (see
[18], among others) by the evaluation of the standard deviation
percentage of the Nusselt number on the impinged area:

r 100

q
2
R 
1
Nur  Nur ds
S S
Nur

This characteristic will be discussed later for a particular

conguration.
3. The impinging jet
As we have previously mentioned in the Introduction, the jet
used in this work is exactly the same used to study the heat transfer from a at plate in Ortega-Casanova [16] (also used in [17], but
in an underwater excavation study). The jet is generated by a nozzle (see Fig. 1) and and it can also have azimuthal motion, which is
given to the uid by means of eight rotatable blades located at nozzle bottom and depending the jet swirl intensity on the blade rotation (see Fig. 2). Therefore, if the blades are radially oriented, as
shown in Fig. 2(a), the swirl intensity is the lowest (nozzle R),
while for their maximum rotation, see Fig. 2(b), the jet swirl intensity is the highest (nozzle S2). It depends on the Reynolds number,
as can be seen in Fig. 3, where a swirl parameter L [see 19] based on
the nozzle exit axial and azimuthal velocity proles, and calculated
as

R1 2
r wv dr
0
R
L
1
D=2 0 rw2  1=2v 2 dr

is depicted. Those two nozzle congurations are the ones used in

this study. Although the jet swirl intensity levels of the nozzle R
are not completely nulls, this type of jet will be referred hereinafter
as the non-swirling jet.

Fig. 2. Plan view of the blades: (a) radially oriented (R conguration); (b) with the
maximum rotation (S2 conguration).

131

The swirling jet characteristics at the nozzle exit (i.e. the

velocity and turbulent magnitude proles), will be used in the
numerical simulations as an inlet boundary condition. They were
all previously experimentally measured using the LDA technique,
and mathematically modelled both the velocity proles and its
turbulence (see [17], for details).
All we need to know is that, for each ow rate through the
nozzle, i.e. for each Reynolds number, a mathematical function to
model the jet emerging from the nozzle is totally known, and it will
be used in the numerical simulations as an inlet boundary
condition.

4. Numerical considerations
In this section, different considerations will be presented. Those
considerations are mainly related to computational aspects of the
study, such as computational geometries, boundary conditions,
governing equations and a study of the grid convergence.
According to the azimuthal symmetry of the impinging jet, the
physical geometry will be considered also axisymmetric, so only
the ow on a bi-dimensional plane is numerically simulated. This
plane extends radially, from the symmetry axis of the jet to a
dimensionless distance R=D, far enough to not affect the development of the jet both from the nozzle and along the plate; and axially, to a dimensionless distance H=D from the exit of the nozzle to
the plate. All these measurements are depicted in Fig. 4, where the
geometry of the at plate without surface variations is shown.
Once the computational plane have been commented, next the
surface variations on the uniform plate will be described. The
reader should keep in mind that the geometry under study is axisymmetric, so a 3D image of it can be visualised if the 2D axisymmetric plane is rotated 2p radians. The diameter of both the
dimples and bumps has been xed to 0:25D, while the separation
between their centres was xed to D=2, as it is shown in Fig. 6.
The joining between the at pieces and the dimples or bumps
was rounded with a radius of 0:05D. A total of 5 dimples, bumps,
and dimples&bumps are made on the plate which means that
the non-uniform zone of the plate extends radially to 2:5D. These
all plate congurations with dimples, bumps and dimples&bumps
are shown in Fig. 6. The total length R of the plate is R  5D, actually it is equals to 21D=4, when H=D 5; 10, and R 15D when
H=D 30, because in this last case, the jet needs a higher radial
domain due to its spreading from the nozzle to the plate.
Fig. 4 can help us to introduce the different boundary conditions
used in the numerical simulations. Once the problem under study
is considered to be axisymmetric, the left line corresponds to the
axis of symmetry of the geometry; the lower boundary corresponds to the at/non-at heated plate, where a no-slip boundary
condition with a prescribed known temperature is used; the right
line corresponds to a boundary condition where the ow is
allowed to exit. Regarding the upper boundary, it is split in two
parts, the left one, where the mathematical model of the jet is used
as boundary condition and ranging from r 0 to r D, and the
right one (ranging from r D to the right side of the domain),
where the same kind of boundary condition used at the right hand
side of the domain is imposed. It must be noted that the boundary
condition of the jet extends to D instead of D=2 since the velocity
prole at the nozzle exit was measured until that radial position,
where the uid velocities are practically nulls. According to the
CFD software used to solve the problem, the commercial software
Fluentin our case, the different physical boundary conditions correspond to the next ones in Fluents terminology: the axis of symmetry is an axis boundary condition; the no-slip plate is a wall
with known temperature; the right boundary condition and the
right part of the upper boundary are a pressure-outlet boundary

132

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

Fig. 3. Swirl parameter versus Re for the nozzle congurations used in the study.

proles of the jet are mathematically modelled for each Reynolds

number (or ow rate) and blade conguration, using the mathematical functions given in Ortega-Casanova et al. [17].
The governing equations, solved numerically by the CFD
software in order to get the evolution of the jet from the nozzle
to the plate, are those corresponding to an axisymmetric incompressible turbulent swirling ow that, in Cartesian tensor notation,
can be written as:
the continuity equation:

@V i
0;
@xi

10

the momentum equations:




 @ v 0 v 0
i j
@V i V j
1 @p
@ @V i @V j 2 @V l


m

 dij
;
@xj
@xj @xj @xi 3 @xl
q @xi
@xj

11

and the energy equation:



@
@
@T
;
V i qe p
K eff
@xi
@xj
@xj

12

where e, dened as

eh
Fig. 4. Sketch of the geometry. Dimensions and boundary conditions.

condition; and, nally, the left part of the upper boundary is a

velocity inlet boundary condition, where the turbulent axisymmetric jet generated by the nozzle is imposed by means of a Fluent
UDF (User Dened Function), by which the velocity and turbulence

~
V ~
V
;
2

13

is the internal energy per unit mass, T is the temperature, q is the

density, p is the pressure, m is the kinematic viscosity, h is the
enthalpy, d is the Diracs delta, K is the thermal conductivity and
K eff K K t is the effective thermal conductivity that takes into
account the turbulent thermal conductivity K t : K t cp lt =Pr t . cp is
the uid heat capacity, lt is the turbulent dynamic viscosity and
Prt is the turbulent Prandtl number. However, due to the unknown

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

velocity uctuations v 0 , the problem closure is achieved by using a

turbulent model, such as the Shear Stress Transport (SST) kx in
this study. Its use is motivated by the work done by Sagot et al.
[20], who found a good agreement between numerical and experimental heat transfer results by using the SST kx model. In addition to that, it is also used a range of Reynold numbers close to
those employed in the referred paper, therefore the use of the chosen model is properly justied. Nevertheless, our own validation
has been carried out to have an idea about the goodness of the turbulent model SST kx is. To that end, we did a comparison of the
solution given by 3 different turbulent models with experimental
data extracted from literature. In particular, we solved numerically

the impingement of an axial jet, with Reynolds numbers equals to

23,000, against a at plate heated with a constant heat ux and
located at a distance H (=2D) from the jet exit. The turbulent models
under study were the SST and the standard k  x, and the enhanced
k, while the experimental data were extracted from Lee et al. [21]
and Baughn and Shimizu [22]. Fig. 5 shows all the proles of the Nu,
together with a 10% deviation bands of the SST results. As one can
see, the experimental data are within these bands: the numerical
values with the SST model agree very well with the experimental
data both at the stagnation point and for large values of r=D, being
the maximum discrepancies (of the order of 8%) around the second
peak of the Nusselt prole; any other tested turbulent model gives

Fig. 5. Validation of the turbulent model.

0.

05

D/2

12
0.

5D

5D

0.

12

D/2

D/2

0.0

5D

0.

12

5D

5D

0.0

D/2

133

0.
5D
12

Fig. 6. Congurations and dimensions of plates.

134

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

worse results than the SST one. Due to all these reasons, the SST
model is the one used nally for the simulations. Despite of using
the same turbulent model than Sagot et al. [20], their Nusselt evolution in the validation study they show is slightly different from
our. This is because the grid sed in both studies is not the same,
although globally both Nusselt proles are quite similar.
Therefore, two closure equations are required: one to know the
turbulent kinetic energy k and another one to know the specic
turbulent dissipation rate x. In compact notation, they can be written as

@
@
@k
Gk  Y k ;
kV i
Ck
@xi
@xj
@xj

@
@
@x
Gx  Y x :
q xV i
Cx
@xi
@xj
@xj

14
15

Ck and Cx are the effective diffusivity of k and x, while Gk and Gx

are the generation of k and x respectively, due to mean velocity
gradients. Finally, Y k and Y x are the dissipation of k and x respectively. To know more about their denition and implementation in
Fluent, the reader is referred to Fluent 6.2 Users Guide [23].
In order to have an idea of the computational quality of the
numerical solutions, as well as, to chose the optimum grid, a grid
convergence study was carried out. In particular, we solved numerically the ow with different meshes for our shortest distance from
the nozzle to the plate, H=D 5, the highest Reynolds number,
Re  1:9  104 , and a bumped plate. This conguration is thought
to be the most unfavourable (with the highest velocities), so the
conclusions of the grid convergence study, in terms of both the
optimum grid and its discretization error, can be used for the other
remaining congurations. This study of the grid convergence was
mainly focused on the number of grid points along the plate in
r-direction, Nr, and how close the rst node must be from the plate
in z-direction, i.e. the value Dzp , since that region is, clearly, the
more interesting of the domain and where the heat transfer takes
place. In that sense, the rst thing we looked for was the optimum
value of Dzp , which also gives the value of the dimensionless turbulent distance y . It will tell us if the turbulent boundary layer along
the plate is being solved properly. Thus, the study of y was carried
out for three different distances of the rst node from the plate:
Dzp
0:001; 0:002; 0:003. Once the ow was solved and a steady
D
state reached, the values we obtained were y 1:1; 2:2; 3:3. To
solve the turbulent boundary layer along the plate, values of y
of unity order are needed, so the rst node from the plate was
chosen at 0:001D.
With the st row of nodes from the plates chosen, the next step
is to nd the optimum number of the grid nodes for the computational domain. To that end, and as before, we have solved numerically the most unfavourable case (H=D 5; Re  1:9  104 and a
bumped plate) with three different mesh grids to choose the optimum one. The total number of grid nodes along radial and axial
directions was Nr 335; 435; 735 and Nz 80; 110; 150,
respectively, with the nodes concentrated around the plate (with
Dzp 0:001D), the nozzle exit, and the transitions between
dimples/bumps and at regions. Following Celik et al. [24], we
can estimate the error of the discretization by using the grids
previously commented. First, we have dened an average representative mesh size h taking into account that the region of interest

of our problem is the non-uniform plate, and that some magnitudes evaluated on it will be used to choose the optimum
mesh.
p
So, the representative mesh size is chosen as D pR2 =Nr, thus
the grid renement factor is given by c Dcoarse =Dfine
Nr fine =Nr coarse . This last value, together with the area-weighted average Nusselt Nu on the plate obtained from the numerical solutions,
are used to calculate the apparent (or observed) order of accuracy .
of the method, the area-weighted average Nusselt extrapolated
value Nue (by means of the generalized Richardson extrapolation),
the approximate relative error ea , the extrapolated relative error
eext and the grid convergence index GCI (for the expressions we
used see [24]). They are all shown in Table 1, where the computational time of the software to carry out 1000 iterations (s1000it ) is
also shown. Hence, according to these results, the numerical uncertainty in the ne(medium)-grid solution for the area-weighted
average Nusselt number can be reported as 0.4(1.0)%. Since both
are quite low,
1%, being the computational time more than three
times greater with the ne grid than with the medium, we nally
decide to choose the grid having Nr 435 and Nz 110 nodes as
the optimum. Specically, this radial grid distribution has been
used in all the simulations when H=D 5; 10 because of the radial
dimension of the domain is the same. However, when H=D 30,
the radial domain is higher and more nodes in the radial direction
were used. In particular, for H=D 30; Nr 510 were used, being
from node 1 until 435 in the same radial locations than when
H=D 5; 10. Regarding the axial dimension for H=D 10;
Nz 160, and for H=D 30; Nz 230. These grid points were
non-uniformly distributed along the axial direction having all
meshes not only Dzp 0:001D but also the same grid distribution
from the plate to a distance equals to D=2, in order to assure that
the uncertainty study should be valid for all nozzle-to-plate
distances. In Fig. 7, one can see a detail of how the grid looks
around an arbitrary dimple.
Regarding the numerical methods used in the simulations, it
must be said that a typical simulation requires about 70  103
iterations to converge, being around one fth of the total iterations
done with rst-order error methods, while the remaining iterations were done with the second-order error schemes PRESTO
(PREssure STaggering Option) and QUICK (Quadratic Upwind Interpolation for Convective Kinematics), while the Pressure-Velocity
Coupling were carried out with the SIMPLE (Semi-Implicit Method
for Pressure-Linked Equations) scheme. Finally, the gravity effects
were not taken into account since the inertial forces are much bigger than the gravitational ones, because of Froude number is much
greater than one.
5. Results and discussion
Due to the big amount of data available once the simulations
were carried out, this section will be split in three subsections in
order to analyse independently the effect that the nozzle-to-plate
distance, type of plate, and type of nozzle used to generate the
jet, have on the heat transferred from the plate. All the results will
d;b;db
be shown in terms of N 0
and N d;b;db , which help us to compare
the heat transfer between the at plate and non-at ones. In order
to have comparable results, the surface S used to calculate the
area-weighted average Nusselt number in (5) was the same for
all geometries and it extends radially 5D.

Table 1
Discretization error results.
n

Nr

c hn1 =hn

Nu

Nun1;n
e

en1;n
(%)
a

en1;n
(%)
ext

GCIn1;n (%)

s1000it (s)

1
2
3

735
435
335

1.7
1.3

104.7
104.9
105.1

1.8

104.6
104.6

0.2
0.2

0.1
0.3

0.4
1.0

1,380
450
300

135

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

Fig. 7. Grid detail around a dimple.

5.1. Nozzle-to-plate distance effect

All the information from the simulations is summarised in
Figs. 8, 10 and 14, for H=D 5; 10; 30, respectively. Different curves
correspond to different combinations of plate and nozzle-type, as
indicated in the legend.
When the smallest distance between the nozzle and the plate is
analysed, see Fig. 8, at the stagnation point only the dimpled plate

impinged by a non-swirling jet (nozzle R) works slightly better

than the at plate (N 0 > 1), whereas impinged by a swirling jet
(nozzle S2) works as good as the at plate (N 0  1). For any other
combination of nozzle and surface type, the heat transferred at the
stagnation point is lower than when a at plate is used (N 0 < 1),
see Fig. 8(a). It is also remarkable that the lowest heat transfer
coefcients ratio at the stagnation point occurs when the jet swirl
intensity is the highest (nozzle S2 and Re  0:9  104 , as shown in
[17]), being the heat transfer at the stagnation point of a at plate
almost twice that of a bumped plate, playing the bumps an important role at high jet swirl intensities. Regarding the global coefcients ratio N , see Fig. 8(b), it is always greater than unity, so
that, any of the plates tested works better (higher heat transfer)
than the at one. In addition to this, for this small separation, the
better option to get the maximum heat transfer from the plate,
both locally and globally, is by means of a dimpled plate (instead
of a at one) and a non-swirling jet (nozzle R) for any of the
Reynolds numbers under study. On the other hand, the worst
behaviour at the stagnation point is obtained when the bumped
plate is used, regardless of whether swirling or non-swirling jets
are used. This can be explained in terms of the velocity magnitude
contours when a at and a bumped plate are impinged by the same
jet, as shown in Fig. 9: the presence of the rst bump creates a stagnation region between it and the axis of symmetry greater than for
the case without the bump. Due to this, the ow has smaller velocities around the bumped plate than around the at one. This means
that both the heat transfer and the Nusselt number coefcient on
that small region are also smaller on the bumped plate than on
the at one.
For intermediate separations, in our case H=D 10, whose
results are shown in Fig. 10, at the stagnation point none of the

(a)

(b)

Fig. 8. (a) N 0 ; (b) N . When H=D 5 and for the plate-nozzle combination indicated in the legend.

136

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

Fig. 9. Contours of the dimensionless velocity magnitude. Comparison between the impingement on a at (left) and a axisymmetric bumped (right) plate for
H=D 5; Re  1:5  104 and nozzle R. The shown region extends 2D axially and 5D radially. The velocity magnitude has been made dimensionless with W c .

(a)

(b)

Fig. 10. As in Fig. 8 but for H=D 10.

congurations studied works better than the at plate, at the most,

they behave as good as it, giving again the bumped plate the worst
heat transfer regardless of the jet used, as shown in Fig. 10(a).
Moreover, when the heat transfer is analysed globally by means
of N , as shown in Fig. 10(b), we nd two congurations working
worse than the at plate: the dimpled, and the dimpled&bumped,
when a non-swirling jet impinges against them. It must be also
noted that the impingement of a swirling jet against the bumped
plate gives globally the best results, whereas gives the worst
results locally at the stagnation point. As for H=D 5, the rst
bump creates a stagnation region with low velocities, and this
gives rise to a smaller stagnation Nusselt numbers than for a atplate. The reason why the bumped plate gives the highest heat
transfer is because of the friction of the radial jet with the bumps,
as can be seen in Fig. 11, where some streamlines are depicted.

Those impacts give rise to peaks in the radial distribution of

the Nusselt number coefcient, as shown in Fig. 12, where it is also
sketched the radial distribution of bumps and dimples. Moreover,
although stagnation regions appear between bumps, where the
Nusselt number diminishes, when the area-weighted average Nusselt number is obtained, it is greater than the one of the at plate,
that is, N > 1. In Fig. 12, it is also depicted the radial evolution of
Nusselt number of the dimpled, and dimpled&bumped plate. As
one may observe, when the radial wall jet reaches the bumps,
the Nusselt number increases, whereas when it gets to the dimples,
the Nusselt number decreases. Despite of that behaviour, the joining of the right hand side of the dimple with the next at piece of
the plate generates local peaks in the Nusselt number, as shown in
Fig. 12. These behaviours can be explained using Fig. 13, which
shows some of the streamlines of the ow when a dimpled or a

137

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

Fig. 11. Detail of the bumps region where some streamlines starting at the nozzle
exit are shown. H=D 10; Re  1:5  104 and nozzle S2. The shown region extends
2D axially and 3D radially.

dimpled&bumped plate is located at the bottom. The friction of the

radial jet with the bumps is observed again, giving rise to high local
peaks in the Nusselt number shown in Fig. 12, and it is also
observed how the radial jet owing on the dimples slightly hit
the dimple right side, giving rise to a peak in the Nusselt number,
as shown in Fig. 12. In summary, for intermediate distances, the
use of a at plate always gives better heat transfer at the stagnation point than any other combination of nozzle and plate, being
remarkable that the heat transfer can be highly reduced, even
halved, if bumped plates are used. In relation to global results,
non-swirling jets (nozzle R) give the worst results, although their

impingement against a bumped plate is better than against a at

one. The impingement of swirling jets against any plate always
give better results than impinging against a at one.
Fig. 14 shows the results when the largest nozzle-to-plate separation is studied: H=D 30. Analysing the local heat transfer coefcients ratio at the stagnation point, it is clear that only the
dimpled plate works better than the at one, regardless of the type
of jet used. The remaining nozzle and plate combinations produce
worse results at the stagnation point than the at plate, as can be
seen in Fig. 19(a). Regarding the global heat transfer coefcients
ratio, none of the nozzle and plate combinations is much better
than the at plate, at the most, they work as good as it when
bumped plates are analysed, as can be seen in Fig. 14(b). In
Fig. 15, the radial Nusselt number prole, obtained when a swirling
jet (nozzle S2) impinges against the four different surfaces under
comparison, is depicted, and where a behaviour similar to the
one described when H=D 10 is observed. In this gure it is also
shown the radial heat transfer uniformity r, Fig. 15(c), where, as
expected, its high value for non-uniform plates shows a bad performance in comparison with a at plate. The reasons of this are the
Nusselt number oscillations associated with the dimples and/or
bumps on the plate.
In summary, for large distances, the use of plates with surface
variations has no benets in enhancing heat transfer in comparison
with a at one, except at the stagnation point when dimpled plates
are used, regardless of the type of jet used.
5.2. Surface variations effect
In the present section, results are plotted in a different way to
explain how surface variations on the plate affect the heat
transferred from it, in comparison with a at one. The results are

(a)

(b)

Fig. 12. (a) Sketch of the different plates; (b) Nu. When H=D 10; Re  1:5  104 , nozzle S2 and the plates are the indicated in the legend.

138

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

Fig. 13. Streamlines for H=D 10; Re  1:5  104 and nozzle S2, when the jet impinges against a dimpled&bumped (left part) and a dimpled plate (right part). The limits of
the shown region in each part are the same than in Fig. 11.

Fig. 14. As in Fig. 8 but for H=D 30.

depicted in Figs. 1618 for the bumped, dimpled, and dimpled&bumped plate, respectively.
Regarding the bumped plate (see Fig. 16), at the stagnation
point the heat transfer coefcients ratio is always below unity,
which means that any combination of nozzle and nozzle-to-plate
distance gives worse heat transfer than a at plate, as shown in
Fig. 16(a): the rst bump generates a stagnation region with low
heat transfer at the stagnation point (see Fig. 9). However, in
regard to the global heat transfer coefcients ratio, in Fig. 16(b)
one can observed that its value is always greater (for
H=D 5; 10) or practically equal to (for H=D 30) unity. This is
due to the friction of the radial wall jet with the bumps (see

Fig. 11), where high local values of heat transfer is obtained. In

summary, globally, a bumped plate will always work better than,
or at the most as, a at plate, regardless of the type of jet used,
whereas locally it will always work worse.
When the dimpled plate is analysed (see Fig. 17), with respect
to the local heat transfer coefcients ratio N 0 , it can be said that
some congurations are slightly better than the at one, such as
H=D 30 with any type of jet, and H=D 5 and nozzle R, whereas
the most unfavourable conguration is that one with H=D 10,
regardless of the type of jet, for which the heat transfer is a little
worse than with a at plate (for most Reynolds numbers), as can
be seen in Fig. 17(a). With respect to the global heat transfer

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

139

(a)

(b)

(c)

Fig. 15. (a) Sketch of the different plates; (b) Nu; (c)

r. When H=D 30; Re  1:5  104 , nozzle S2 and the plates indicated.

coefcients ratio, the separation H=D 5 with both types of jets is

better than using a at plate, and also H=D 10 with a nozzle S2,
whereas the worst results are with H=D 30 and any type of jet,
and H=D 10 coupled with non-swirling jets for high Reynolds
numbers, as can be seen in Fig. 17(b). To sum up, locally, few benets can be obtained using dimpled plates whereas, globally, the
best congurations are H=D 5, and H=D 10 with nozzle S2.
For the dimpled&bumped plate, whose results are shown in
Fig. 18, it is easy to see that at the stagnation point the heat transfer coefcients ratio is always below unity, that is to say, the at
plate gives always better results than the dimpled&bumped one,
regardless of the separation H=D and the type of nozzle used, see
Fig. 18(a): the rst bump works as in the bumped plate, creating
a low velocity region around the stagnation point. On the other
hand, with regards to the global coefcients ratio, it can be seen
in Fig. 18(b) how with H=D 30 and any type of jet, and with
H=D 10 and non-swirling jets, the heat transfer is worse than
using at plates, whereas for the remaining combinations of separation and nozzle, the heat transfer coefcients ratio is slightly better than with the at plate. In summary, locally, none of the
dimpled&bumped plates works better than the at one, whereas
globally, the best ones are obtained when H=D 5, and H=D 10
with nozzle S2.
5.3. Nozzle-type effect
All the results used to explain the effect of the nozzle on the
heat transfer are depicted in Figs. 19 and 20 for nozzles R and

S2, respectively. How the nozzle type affects the heat transfer coefcients ratios is analysed in what follows.
Regarding the nozzle R and the local heat transfer coefcients
ratio (see Fig. 19(a)), only the dimpled plates give better or similar
d
results than the at one, i.e. N 0 J 1. The remaining combinations
of separation and plate always give worse results than the at plate
at the stagnation point. When the global heat transfer coefcients
ratio is analysed, there are some situations where a non-at plate
is more preferable than a at one. In particular, that is so for the
separation H=D 5 and any type of surface variations, and also
for H=D 10 and a bumped plate. Nevertheless, the results with
H=D 30 and a bumped plate, and H=D 10 and a dimpled plate
are quite similar to those corresponding to a at plate. The
combinations for which the global heat transfer is worse than
for a at plate are H=D 30 together with a dimpled and a dimpled&bumped plate, and H=D 10 with a dimpled&bumped plate.
In summary, locally, the best results at the stagnation point are
obtained with a dimpled plate, whereas globally, the best ones
are obtained when H=D 5.
On the other hand, with respect to the nozzle S2 (see Fig. 20),
with relation to the local heat transfer coefcients ratio, the dimpled plate gives better results than the at one, when H=D 30,
and quite similar to them, when H=D 5; 10. When bumped or
dimpled&bumped plates are used, the results are worse than with
the at one, as can be seen in Fig. 20(a). When the global heat
transfer coefcients ratio is analysed, see Fig. 20(b), it is observed
that only when H=D 30, the results are worse than for a at plate,
whereas any other separation works better than the at one. There

140

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

(a)

(b)

Fig. 16. (a) N 0 ; (b) N . When a bumped plate is used and for the nozzle-to-plate distances and nozzle types indicated in the legend.

(a)

(b)

Fig. 17. As in Fig. 16 but for a dimpled plate.

is, however, a special combination: the bumped plate and the

separation H=D 10, for which, globally, the results are the best
whereas, locally, they are the worst. As it was said previously,
the use of a bumped plate decreases the Nusselt number

coefcients ratio at the stagnation point. This can also be

conrmed by having a look at Fig. 21, where the radial prole of
Nusselt number is shown for bumped and at plates and all the
nozzle-to-plate distances studied for an intermediate Reynolds

141

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

(a)

(b)

Fig. 18. As in Fig. 16 but for a dimpled&bumped plate.

(a)

(b)

Fig. 19. (a) N 0 ; (b) N . When a nozzle of type R is used and for the nozzle-to-plate distances and plates indicated in the legend.

142

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

(a)

(b)

Fig. 20. As in Fig. 19 but for a nozzle S2.

(a)

(b)

Fig. 21. (a) Sketch of the bumped plate; (b) Nu. When Re  1:5  104 , nozzle S2 and for the plates and H=D values indicated in the legend.

J. Ortega-Casanova, F.J. Granados-Ortiz / International Journal of Heat and Mass Transfer 76 (2014) 128143

number. It can be also seen how, at the stagnation point, the separation H=D 10 is the worst: the highest difference exists
between Nub0 and Nuf0 ; whereas it is not so high for the other values
of H=D. The analysis of Fig. 21 also conrms that, when the radial
jet ows on any bump, a peak always appears in the Nusselt number, independently of the separation. Obviously, the farther the
separation, the lower the peak, being the area-weighted average
Nusselt number for H=D 10 and the bumped plate the highest
of the six congurations shown in Fig. 21.
6. Conclusions
A numerical study has been conducted in order to identify
whether certain types of non-at plates (at plates with axisymmetric perturbations on them) can transfer more heat than a at
one, to an impinging jet (generated by specic nozzle). Different
non-at plates, nozzle-to-plate distances and Reynolds numbers
(ranged from about 0:7  104 to 1:9  104 ), as well as two type of
jets, swirling and non-swirling have been analysed in the study.
Although a general rule to improve the heat transfer from the plate
can not be proposed, we are able to give some suggestions which
could help to a better design of devices from which heat must be
transferred by means of impinging jets. The suggestions will
depend on where the heat transfer must be increased, if at the
stagnation point or on the whole plate:
At the stagnation point: In comparison with a at plate, the heat
transfer can be increased, or at least be as good as the at plate,
by using dimpled plates, regardless of the type of both the jet
used to impinge the plate and the nozzle-to-plate separation.
Bumped and dimpled&bumped plates always give worse heat
transfer at the stagnation point than a at plate.
On the whole plate: When the heat transfer must be increased in
the whole plate, it is recommended to use bumped plates for
which the global heat transfer coefcient is greater than, or at
least as high as, that of a at plate. The same can be said for
dimpled plates, although in this case there are some combinations of jets and nozzle-to-plate separations giving slightly
worse results than a bumped plate. The dimpled&bumped plate
gives better or clearly worse results than the at plate depending on the jet and nozzle-to-plate separation used.
Obviously, all these suggestions are restricted to the ranges of
the parameters used in this study.
The results and conclusions reported in this paper can be interesting, specically, for future research in cooling processes of
devices where the temperature has an important role in their
working behaviour, e.g. electronic components or industrial
components in thermodynamic cycles, between others.
Conict of interest statement
The authors declare that there are not known concts of
interest associated with this publication.

143

References
[1] B. Han, R.J. Goldstein, Jet impingement heat transfer in gas turbine systems,
Ann. New York Acad. Sci. 934 (1) (2001) 147161.
[2] B.R. Hollworth, M. Durbin, Impingement cooling of electronics, J. Heat Transfer
(Trans. ASME (American Society of Mechanical Engineers), Ser. C) United States
114 (3) (1992).
[3] D.H. Lee, Y.S. Chung, P.M. Ligrani, Jet impingement cooling of chips equipped
with multiple cylindrical pedestal ns, J. Electron. Packag. 129 (3) (2007)
221228.
[4] S.W. Chang, S.F. Chiou, S.F. Chang, Heat transfer of impinging jet array over
concave-dimpled surface with applications to cooling of electronic chipsets,
Exp. Therm. Fluid Sci. 31 (7) (2007) 625640.
[5] C.J. Chang, C.H. Shen, C. Gau, Flow and heat transfer of a micro jet impinging on
a heated chip. Part I Micro free and impinging jet ow, Nanoscale Microscale
Thermophys. Eng. 17 (1) (2013) 5068.
[6] C.J. Chang, H.T. Chen, C. Gau, Flow and heat transfer of a microjet impinging on
a heated chip. Part II Heat transfer, Nanoscale Microscale Thermophys. Eng.
17 (2) (2013) 92111.
[7] S. Roy, P. Patel, Study of heat transfer for a pair of rectangular jets impinging on
an inclined surface, Int. J. Heat Mass Transfer 46 (3) (2003) 411425.
[8] D. Babic, D.B. Murray, A.A. Torrance, Mist jet cooling of grinding processes, Int.
J. Mach. Tools Manuf. 45 (10) (2005) 11711177.
[9] N. Zuckerman, N. Lior, Jet impingement heat transfer: physics, correlations,
and numerical modeling, Adv. Heat Transfer 39 (2006) 565631.
[10] S.V. Ekkad, D. Kontrovitz, Jet impingement heat transfer on dimpled target
surfaces, Int. J. Heat Fluid Flow 23 (1) (2002) 2228.
[11] K. Kanokjaruvijit, R.F. Martinez-Botas, Heat transfer correlations of
perpendicularly impinging jets on a hemispherical-dimpled surface, Int. J.
Heat Mass Transfer 53 (15) (2010) 30453056.
[12] M. Imbriale, A. Ianiro, C. Meola, G. Cardone, Convective heat transfer by a row
of jets impinging on a concave surface, Int. J. Therm. Sci. 75 (2014) 153163.
[13] E. ztekin, O. Aydin, M. Avc, Heat transfer in a turbulent slot jet ow
impinging on concave surfaces, Int. Commun. Heat Mass Transfer 44 (2013)
7782.
[14] D. Zhang, H. Qu, J. Lan, J. Chen, Y. Xie, Flow and heat transfer characteristics of
single jet impinging on protrusioned surface, Int. J. Heat Mass Transfer 58 (1)
(2013) 1828.
[15] A.H. Beitelmal, M.A. Saad, C.D. Patel, The effect of inclination on the heat
transfer between a at surface and an impinging two-dimensional air jet, Int. J.
Heat Fluid Flow 21 (2) (2000) 156163.
[16] J. Ortega-Casanova, CFD and correlations of the heat transfer from a wall at
constant temperature to an impinging swirling jet, Int. J. Heat Mass Transfer
55 (2012) 58365845.
[17] J. Ortega-Casanova, N. Campos, R. Fernandez-Feria, Experimental study on
sand bed excavation by impinging swirling jets, J. Hydraul. Res. 49 (5) (2011)
601610.
[18] A. Ianiro, G. Cardone, Heat transfer rate and uniformity in multichannel
swirling impinging jets, Appl. Therm. Eng. 49 (2012) 8998.
[19] N.A. Chigier, A. Chervinsky, Experimental investigation of swirling vortex
motion in jets, J. Appl. Mech. 34 (1967) 443451.
[20] B. Sagot, G. Antonini, A. Christgen, F. Buron, Jet impingement heat transfer on a
at plate at a constant wall temperature, Int. J. Therm. Sci. 47 (12) (2008)
16101619.
[21] D.H. Lee, S.Y. Won, Y.T. Kim, Y.S. Chung, Turbulent heat transfer from a at
surface to a swirling round impinging jet, Int. J. Heat Mass Transfer 45 (1)
(2002) 223227.
[22] J.W. Baughn, S. Shimizu, Heat transfer measurements from a surface with
uniform heat ux and an impinging jet, J. Heat Transfer (Trans ASME
(American Society of Mechanical Engineers) Ser. C) United States 111 (4)
(1989).
[23] Fluent 6.2 Users Guide, Fluent Incorporated, Centerra Resource Park, 10,
Cavendish Court, Lebanon (NH) 03766 USA, 2005.
[24] I.B. Celik, U. Ghia, P.J. Roache, Procedure for estimation and reporting of
uncertainty due to discretization in CFD applications, J. Fluids Eng. Trans.
ASME 130 (7) (2008).