Numerical study of heat transfer in pulsating turbulent air flow
1 like480 views
The document presents a numerical study on the heat transfer characteristics of pulsating turbulent air flow in a circular tube, aimed at understanding how pulsation affects heat transfer rates under varying conditions. Using computational fluid dynamics (CFD) with a Fluent code, the study investigates the flow and temperature fields while addressing the influence of parameters like frequency and Reynolds number on the Nusselt number. Results indicate minimal changes in mean Nusselt number compared to steady flows, highlighting the complexities and varying effects of pulsating flow on heat transfer.
![Thermal Issues in Emerging Technologies, ThETA 1, Cairo, Egypt, Jan 3-6th
2007
1-4244-0897-0/07/$20.00 ©2007 IEEE ThETA01/068 63
NUMERICAL STUDY OF HEAT TRANSFER IN PULSATING TURBULENT AIR FLOW
E. A. M. Elshafei a
, M. Safwat Mohamed b
, H. Mansour; M. Sakrc
Mechanical Power Department, Mansoura University
Faculty of Engineering, El-Mansoura, EGYPT, (35516)
eelshafei@mans.edu.eg a
, msafwat@mans.edu.eg b
, moh_saker@mans.edu.egc
ABSTRACT
A numerical investigation of heat transfer
characteristics of pulsating turbulent flow in a circular
tube is carried out. The flow is thermally and
hydrodynamically fully developed and the tube wall is
subjected to a uniform heat flux. The flow inlet to the
pipe consists of fixed component and pulsating
component that varies sinusoidally with time. The flow
and temperature fields are computed numerically using
computational fluid dynamics (CFD) Fluent code.
Prediction of heat transfer characteristics is performed
over a range of 104
≤ Re ≤ 4x104
and 0 ≤ ƒ ≤ 70 are
observed. Results showed little reduction in the mean
time-averaged Nusselt number with respect to that of
steady flow. However, in the fully developed
established region, the local Nusselt number either
increases or decreases over the steady flow-values
depending on the frequency parameter. These noticed
deviations are rather small in magnitude for the
computed parameter ranges. The characteristics of heat
transfer are qualitatively consistent with the available
experimental and numerical predictions.
1- INTRODUCTION
The study of pulsating flow in pipe has been a subject of
interest among many researches. The problem is
important in biological applications in relation to blood
flow and in industrial applications related to heat
exchange efficiency, cavitation in hydraulic pipe lines,
pressure surges and reciprocating machines. In general,
pulsating flow filed is assumed to consist of a steady
poiseuille flow part and purely oscillating part. Many
experimental, analytical and numerical works studied
the heat transfer in pulsating laminar flow but little in
turbulent flow. One of the key issues concerning
pulsating convection heat transfer in tubes is whether a
superposed flow pulsation effects on heat transfers in
the original steady flow. The answer to this question in
the previous studies can be classified into four different
opinions: (1) flow pulsation enhances heat transfer [1, 2,
3]; (2) it deteriorates heat transfer [4, 5, 6, 7]; (3) it has
no effect on heat transfer [8]; and (4) it either enhances
or deteriorates heat transfer, depending on flow
parameters [9-28].
Numerical study by Xuefeng Wang and Nengli Zhang
[2] showed that in a pulsating turbulent flow there is an
optimum Womersley number at which heat transfer is
maximally enhanced. Experimental studies by [9, 11,
13-18, 25-28] showed that increment and reductions in
mean Nusselt number depends on pulsation frequency,
the turbulent bursting frequency, the amplitude, the
axial location, Reynolds number and Prandtl number. In
order to understand these phenomena and to resolve the
contradictory results, different models of turbulence for
pulsating flows were considered. Two of these models
are well known, the quasi-steady flow model [24] and
the bursting model discussed by [25-28].
Due to its complexity, studies on heat transfer in
pulsating turbulent flow are far from matured. The aim
of the present work is to investigate the mechanics of
heat transfer in pulsating pipe flow numerically in order
to show clearly how pulsation affects the rate of heat
transfer and how does this phenomenon depend on the
pulsation frequency and Reynolds number.
A problem of pulsating flow in a pipe, externally
subjected to a constant heat flux is considered in the
present study. The governing equations are solved with
the aid of the Fluent code. The numerical results can be
served as a useful source material against which the
outcome of future analytical and experimental
investigations can be checked.
2- MATHEMATICAL FORMULATIONS
2.1 Pipe Geometry
In the present study, the numerical solution of the
pulsating turbulent flow through an externally heated
pipe is analyzed. Air is selected as a working fluid. The
radius and length of the pipe are R and L respectively.
The thickness of the pipe is neglected and the thermal
boundary condition on the pipe wall is assumed to be a
uniform heat flux. The two-dimensional axisymmetric
model of the considered pipe is shown in Fig.1.
The pipe geometry at the inlet is long enough to assure a
hydrodynamically fully developed flow at the heated
section and simulated to the experiential facility of [32].
Since the pipe cross-section is circular, it is assumed
that the flow is axisymmetric. In cylindrical polar
coordinates, this means that the flow variables depend
only on the axial coordinate x and radial coordinate r.
The pipe wall section is modeled as: unheated section of
2.6 m length and a heated one of 1.16 m long.
Sinusoidal pulsating flow is assumed to be entering to
the pipe of fixed period (the velocity is only oscillates)
as can be seen in Fig.2.](https://image.slidesharecdn.com/numericalstudyofheattransferinpulsatingturbulentairflow-140804083108-phpapp02/85/Numerical-study-of-heat-transfer-in-pulsating-turbulent-air-flow-1-320.jpg)
![Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al.
ThETA01/06864
2.2 The Governing Conservation Equations
The governing equations for transient turbulent
incompressible flow and heat transfer in the flow region
(0 ≤ x ≤ L and 0 ≤ r ≤ R) can be written as follows [32]:
Continuity:
0
∂∂
∂
=
∂
++
r
v
r
v
x
u
(1)
Momentum in x-direction:
.
1
3
2
2
1
∂
∂
-
)(
∂
∂1
)(
∂
∂1
∂
∂
∂
∂
+
∂
∂
∂
∂
+
+
∂
∂
+
∂
∂
−
∂
∂
∂
∂
+=
++
x
v
r
u
r
rr
r
v
r
v
x
u
x
u
r
xrx
p
rvu
rr
ruu
xrt
u
µ
µ
ρ
Momentum in r-direction:
( ) ( )
.
3
2
2
3
2
2
1
1
11
2
+
∂
∂
+
∂
∂
+−
+
∂
∂
+
∂
∂
−
∂
∂
∂
∂
+
∂
∂
+
∂
∂
∂
∂
+
∂
∂
−=
∂
∂
+
∂
∂
+
∂
∂
r
v
r
v
x
u
rr
v
r
v
r
v
x
u
r
v
r
rr
r
u
x
v
r
rrr
P
rvv
rr
ruv
xrt
v
µ
µ
µ
µ
ρ
Energy:
φ
r
T
r
rrx
T
r
T
v
x
T
u
t
T
Cp
eff µλ
ρ
+
∂
∂
∂
∂
+
∂
∂
=
++
1
∂
∂
∂
∂
∂
∂
2
2
Where φ is viscous dissipation term, which is
φ
and λeff is the effective thermal conductivity, and for
standard k-ε model, is given by
λeff = λ+ λt (3. c),
λt = Cp µt /Prt (3. d)
2.3 Turbulence Model
Among the available turbulent models, those based on
eddy viscosity are widely applied to the turbulent
kinetic energy – turbulent energy dissipation rate (k-ε).
Two additional equations for the standard k-ε turbulence
model: the turbulence kinetic energy, k, and the
dissipation rate, ε, are determine using the following
transport equations, respectively.
k – Equation:
Where
ε- equation:
( )ρε
ε
ε
σ
µ
µ
ε
σ
µ
µ
ε
ρ
εε
εε
21
∂
∂
∂
∂1
∂
∂
∂
∂
CGC
k
r
r
rrxxDt
D tt
−+
++
+=
The turbulent viscosity, µt is computed by combining K
- ε as follows:
)/( 2
ερµ µ kCt = (6)
Cµ, C1ε, and C2ε are the model constant; σk and σε are
turbulent Prandtl numbers for k and ε, respectively.
They have the following default values (Launder
&Spalding 1972): Cµ= 0.09, C1ε= 1.44, C2ε= 1.92, σk
=1.0, σε=1.3,
(3.a)
(2.a)
(2.b)
Uin
T in
Entrance length (2.6 m)
Symmetry axis
Fluid
(Air)
Pipe wall
R
L heated (1.16 m)
x
Uniform wall heat flux q"
r
Fig. 1 Schematic diagram for the studied problem.
,2
222
∂
∂
+
∂
∂
+
+
∂
∂
+
∂
∂
=
r
u
x
v
r
v
r
v
x
u
(4.a)
ρε
σ
µ
µ
σ
µ
µρ
-
∂
∂
∂
∂1
∂
∂
∂
∂
G
r
k
r
rr
x
k
xDt
Dk
k
t
k
t
+
++
+=
(4.b)
++
+
+
=
2222
∂
∂
∂
∂
2
∂
∂
2
∂
∂
2
x
v
r
u
r
v
r
v
x
u
G tµ
0
10
20
30
40
50
U(t)[m/s]
Um
UA
Л 2Л
Fig .2. Variations of inlet velocity
(3.b)
(5)](https://image.slidesharecdn.com/numericalstudyofheattransferinpulsatingturbulentairflow-140804083108-phpapp02/85/Numerical-study-of-heat-transfer-in-pulsating-turbulent-air-flow-2-320.jpg)
![Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al.
ThETA01/068 65
2-4 Heat Transfer Coefficient and Nusselt
Number
Due to the unique characteristics of the pulsating flow,
several heat transfer coefficients have to be carefully
defined.
The instantaneous local heat transfer coefficient is given
by
avw TT
txh
-
q′′
=),( (7-a),
and the local heat transfer coefficient, defined as the
average of heat transfer coefficient over a cycle, is
formulated as
∫=
π
ω
π
2
0
)( )(
2
1
thdh xx (7-b)
The instantaneous heat transfer coefficient, defined as
the average of heat transfer coefficient over the whole
length of pipe at a given moment, is expressed as
∫=
L
tt hdx
L
h
0
)( .
1
(7-c)
Subsequently, several relevant Nusselt numbers are
rationally introduced. The instantaneous local Nusselt
number is
K
hD
txNu tx =),(, (7-d)
The overall Nusselt number; Nu is given by
∫ ∫=
L
tx dxtdNu
L
Nu
0
2
0
, .)(
2
1
π
ω
π
(7-e)
To ensure sufficient temporal resolution of the
numerical solution, five to seven cycles are required to
obtain a steady periodic solution with an acceptable
accuracy.
2-5 Boundary Conditions
The solution domain of the considered 2D,
axisymmetric pipe flow is geometrically quite simple,
which is a rectangle on the x-r plane, enclosed by the
inlet, outlet, symmetry and wall boundaries .On walls
no-slip condition are assumed for the momentum
equations. The inlet velocity values have been derived
from given Reynolds numbers. Inaccuracies due to an
uncertainty in the shape of the inlet velocity profiles are
not expected to play an important role, since all
calculations were carried out at places sufficiently down
stream from the inlet, so that quite fully-developed
conditions should be expected at the calculating section.
For pulsating cases, the inlet velocity is varied with
time. The outlet boundary condition is called "outflow",
which implies zero-gradient condition at the outlet.
At the pipe inlet, (x=0):
For steady flow condition, U (t) = Uin, and for
sinusoidal pulsating flow, U (t) = Um+ UA sin (ωt)
At the pipe exit plane, (x = L):
At the pipe axis, (r = 0):
At the pipe wall, (r = R): (u, v k, ε) = 0, a uniform heat
flux is imposed, q" = qο for heated section and q" = 0
for the adiabatic wall.
Values of k and ε are not known at the inlet, but, if they
are not given by experimental data, some reasonable
assumptions can be made. The kinetic energy of
turbulence according to a certain value of the square of
the average inlet velocity is expressed by [ 33]
k=0.03 Uin
2
In addition, dissipation is calculated according to the
equation
R
k
C
03.0
2
3
µε =
2-6 Simulation Values
In the analysis, the following simulation values are
taken as:
Uin = (12.7 to 42.8) m/s, Um =11.07 m/s
UA = (1.63 to 31.81) m/s, ω = (41.9 to 427.25) rad/s,
and qο =922 W/m2
3- COMPUTATIONAL PROCEDURE
3-1 Calculation Tools
Even the difficult general differential equations now
yield to the approximating technique known as
numerical analysis, whereby the derivates are simulated
by algebraic relations between a finite numbers of grid
points in the flow field which are then solved on a
digital computer. A suitable CFD computer code is used
to solve the governing equations numerically along with
the boundary and the initial conditions. The CFD
program is a process by which the fluid flow
characteristics can be predicted through arbitrary
geometries giving such information as: flow speed,
pressure, residence time, flow patterns, etc. The Fluent
6·1 (Fluent 2003) program is chosen as the CFD
computer code for this work because of the ease with
which the analysis model can be created, and because
the software allows users to modify the code for special
analysis conditions through the use of user subroutines.
The Fluent computer code uses a finite-volume
procedure to solve the governing equations of fluid flow
in primitive variables such as velocity components in
0=
∂
∂
r
φ
0=
∂
∂
x
φ](https://image.slidesharecdn.com/numericalstudyofheattransferinpulsatingturbulentairflow-140804083108-phpapp02/85/Numerical-study-of-heat-transfer-in-pulsating-turbulent-air-flow-3-320.jpg)
![Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al.
ThETA01/06866
axial and radial directions and pressure. A variety of
turbulence models is offered by the Fluent computer
code. The standard k–ε model was used as a turbulence
model in this study. This model is a semi-empirical one,
based on model transport equations for the turbulent
kinetic energy (k) and its dissipation rate (ε). The model
transport equation for k is derived from the exact
equation while the model transport equation for ε is
obtained using physical reasoning and bears little
resemblance to its mathematically exact counterpart. In
the derivation of the k–ε model, it was assumed that the
flow is fully turbulent, and the effects of molecular
viscosity are negligible. The standard k–ε model is,
therefore, valid only for fully turbulent flows. A detailed
description of turbulence models and its applications to
turbulence can be found in "Fluent 2003; Launder &
Spalding 1972". In the case of the standard k–ε models,
two additional transport equations, (for the turbulent
kinetic energy and the turbulence dissipation rate) are
solved, and turbulent viscosity, µt, is computed as a
function of k and ε. The solution method for this study
is axisymmetric. In order to define the pulsating inlet
velocities in all cases, an UDF (User-Defined Function)
file was introduced to the prepared fluent case file. By
using the results of the calculations performed with the
fluent code. This program, written in FORTRAN 77
language, calculates numerically the local and average
Nusselt Number.
Grid size: Grid-independent tests were carried out to
ensure grid independence of the calculated results;
consequently, the grid size and the grid orientation
giving grid independent results were selected, and thus a
total cell number of 45000 cells (300× 150) were
adopted. The grid points are clustered in the radial
direction so that the finer spacing is formed near the
wall and at the inlet. Fig.3 shows the grid.
4- RESULTS AND DISCUSSION
4-1 Steady Flow
To validate the numerical calculations, the
computational results for the fully developed steady
turbulent pipe flow are compared with the experimental
data observed by Elshafei et al. [31]. For two different
values of Reynolds number, the numerical results are in
consistent with the experimental results; within about
±5 % as can be noticed in Fig 4. The flow was
observed to be thermally fully developed for X/D ≥ 20.
4-2 Pulsating Flow
In the present work, investigations are performed within
the ranges of 104
≤ Re ≤ 4x104
, and 0 ≤ ƒ ≤ 70.
The effect of pulsating frequency on the local time
average Nusselt number at fixed Reynolds number of
16.8×103
is shown in Fig. 5. It can be seen that, as the
axial location moves downstream, the time average
local Nusselt number for both pulsating and steady
flows decays rapidly from a very large value near the
entrance and the axial decrease becomes less step in the
bulk of the pipe soon after leaving this entrance region.
The variation of the time averaged local Nusselt number
for pulsating turbulent air flow exhibits a similar trend
to that of the steady air flow case, but generally with
lower values than that of the steady flow for f < 61.13.
The maximum reduction of about 6% in the local
Nusselt number was predicted in the thermal entrance
region. For f ≥ 61.13, little enhancement in the local
time average Nusselt number of about 8% for X/D ≥ 50
was detected.
Figure 6 illustrates the variation of the ratio of the local
Nusselt number value for pulsating flow to that of the
steady flow value; η with the axial distance at Re =
22.5×103
. For f ≤ 39.3 Hz, it can be observed from Fig.
6-a that the ratio η starts rising rapidly in the entrance
zone of the thermally developed region (X/D<20) and
becomes nearly fixed below unity where its value in the
bulk of the pipe soon after leaving the entrance length
(X/D≥20) implies that the effect of pulsation on the
local heat transfer diminishes.
For f ≥ 42.5 Hz, the value of η close to the end of the
pipe enhanced by about 4%as shown in Fig. 6-b.
The impacts of other relevant parameters on heat
transfer are illustrated in Fig.7 (a, b). The detected
values of η due to the variation of pulsation frequency at
higher value of Reynolds number (Re = 31.6×10 3
). At
lower frequency (ƒ ≤ 39.3 Hz), a small change in heat
transfer ratio is visible near the entrance region. At
(ƒ=13.3 Hz) a rapid increases in η can be observed up to
X/D =20, after which it seems to be unchanged and
generally, the value of η is lees than unity. The same
trends can be seen for ƒ =28.3 Hz and ƒ =39.3 Hz.
However, in the downstream region the value of η
reaches about unity.
Velocity
Inlet
Axisymmetric
Adiabatic section Heated section
X
R
Outlet
Flow
Fig.3 Computational grid for pipe](https://image.slidesharecdn.com/numericalstudyofheattransferinpulsatingturbulentairflow-140804083108-phpapp02/85/Numerical-study-of-heat-transfer-in-pulsating-turbulent-air-flow-4-320.jpg)
![Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al.
ThETA01/068 67
Fig.4. Variation of local Nusselt number at different
Reynolds number
The effect of changing pulsation frequency on the ratio
η seems to be negligible specially, for higher values of
pulsation frequency as can be noticed in Fig. 7.b.
At high values of Reynolds number; Re ≥ 37×103
,
certain qualitative trends can be shown in Figs. 8(a, b),
which are similar to those described in the previous
figures. For all studied values of pulsation frequency,
the values of the ratio η are always less than unity.
This global behavior of pulsating flow is consistent with
the analytical and numerical results of [5, 8, 20, 24, 29,
30 and 31]. The overall patterns of η curves are
qualitatively similar as discussed earlier.
Fig.6.Distrbution of η at different frequency, Re=22500
Fig.7.Distrbution of ηx at different frequency, Re=31600
X/D
Nux
0 20 40 60
0
100
200
300
400
Numerical Re=22500
Numerical Re=31600
Expermental Re=22500
Expermental Re=31600
η
η
0 20 40 60
0.84
0.88
0.92
0.96
1
1.04
Re=22500
f=6.67
f=13.3
f=20.3
f=28.3
f=39.3
0 20 40 60 80
0.88
0.92
0.96
1
1.04
1.08
RE=22500
f=42.5
f=45.7
f=53.13
f=61.13
f=68
X/D
X/D
(b)
(a)
Nux
Fig.5.Distrbution of Nux at different frequency, Re=16800
X/D
0 20 40 60
0
50
100
150
200
250
300
Re=16800
steady
f=42.5
f=45.7
f=49.13
f=53.13
f=61.13
f=68
0 20 40 60
0
50
100
150
200
250
300
Re=16800
steady
f=6.67
f=13.3
f=20.3
f=39.3
Nux
X/D
(b)
(a)
0 20 40 60
0.92
0.94
0.96
0.98
1
1.02
Re=31600
f=13.3
f=28.3
f=39.3
0 20 40 60
0.94
0.96
0.98
1
Re=31600
f=42.5
f=45.7
f=53.13
X/D
X/D
η
η
(b)
(a)](https://image.slidesharecdn.com/numericalstudyofheattransferinpulsatingturbulentairflow-140804083108-phpapp02/85/Numerical-study-of-heat-transfer-in-pulsating-turbulent-air-flow-5-320.jpg)
![Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al.
ThETA01/06868
Fig.8. Distribution of η at different frequency, Re=37000
Fig.9. Heat transfer ratio for different Reynolds
Number and pulsation frequency.
The third set of results pertains to the effect of the
Reynolds number on the relative mean time- averaged
Nusselt; ηm in case of pulsating flow in pipes. The
calculated data for ηm versus Re are described in Figs.9
(a, b). In the range of pulsating frequency; ƒ ≤ 39.3, it
is found that Reynolds number strongly affects the heat
transfer ratio; ηm when Reynolds number is relatively
low; Re < 25 x 103
as can be noticed in Fig. 9-a. The
heat transfer ratio is sharply increases with Reynolds
number up to nearly unity. However, when Reynolds
number exceeds this value, the effect of increasing
Reynolds number diminishes and the value of ηm
remains nearly constant. It is also noticed that the
variation of the imposed pulsating frequency has a
relatively small effect on ηm. These computed results are
in consistent with that reported by Xuefeng. W. and
Nengli Z. [2]
Fig.10. Comparison of the experimental data
with numerical data at different (ω*)
For higher pulsation frequency; ƒ ≥ 42.5, the value of
the ratio ηm increases also with the increase of Reynolds
number. As ηm reaches to its peak value at Re ≈ 22.5 x
103
, it then smoothly decreases with increasing
Reynolds number up to 33 x 103
where ηm value is held
nearly constant for higher values of Reynolds number as
can be noticed in Fig. 9-b.
The behavior of Nusselt number corresponding to the
variation in both Reynolds number and pulsation
frequency can also be explained in view of the bursting
phenomena (Loiao and Wang [25] and Genin et al.
[26]). The bursting model defines certain regions on
frequency-Reynolds number plane, where, the bursting
frequency can be dependent or independent of the
pulsation frequency.
The present results showed that the heat transfer
coefficient may be increased or decreased with the
change of both Reynolds number and frequency. The
mean bursting frequency may be subdued to the
pulsation frequency leading to the occurring of
resonance that is dependent only on the pulsation
0 20 40 60
0.94
0.96
0.98
1
Re=37000
f=6.67
f=13.3
f=20.3
f=28.3
f=39.3
0 20 40 60
0.92
0.94
0.96
0.98
1
Re=37000
f=42.5
f=45.7
f=53.13
f=61.13
X/D
X/D
η
η
(b)
(a)
15000 20000 25000 30000 35000 40000
0.84
0.88
0.92
0.96
1
f=6.67
f=13.3
f=20.7
f=39.3
Re
(a)
ηm
Re
(b)
ηm
15000 20000 25000 30000 35000 40000
0.88
0.92
0.96
1
1.04
f=42.5
f=45.7
f=53.13
f=61.13
f=68
ω*
ω*
ηm
(a)
ω*
ηm
(b)
ηm
(c)
0 1 2 3
0.8
0.9
1
1.1
1.2
Re=37000
Experimental
Numerical
0 2 4 6
0.8
0.9
1
1.1
1.2
Re=16800
Experimental
Numerical
0 1 2 3 4 5
0.8
0.9
1
1.1
1.2
Re=22500
Experimental
Numerical](https://image.slidesharecdn.com/numericalstudyofheattransferinpulsatingturbulentairflow-140804083108-phpapp02/85/Numerical-study-of-heat-transfer-in-pulsating-turbulent-air-flow-6-320.jpg)
![Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al.
ThETA01/068 69
frequency. Accordingly, the heat transfer process is
expected to be affected resulting in either reduction or
enhancement in the rate of heat transfer. These results
are in consistent with those reported by Habibi et al [17]
and Liao and Wang [25].
A comparison between the present numerical results
with the experimental ones [32] over a ranges of 16 x
103
< Re< 38x103
and 6.67 ≤ ƒ ≤ 68 Hz are described in
Fig. 10. The data are presented in the form of ηm as a
function of bursting frequency ω*
. It can be observed
that for all values of Reynolds number, the numerical
predictions of ηm as a function of ω*
have the sametrend
as those for the experimental data. The discrepancies
between the experimental data and the computed ones at
low bursting turbulent frequency are of about 10%. As
ω*
increases, the computed results closes to the
experimental ones, specially, at higher value of
Reynolds number.
In summary, the present elaborated numerical results for
heat transfer characteristics of turbulent pulsating flow
are useful. It can serve as a source of materials against
which further analytical data may be checked for
consistency.
5-CONCLUSTION
Simulation was performed to describe heat transfer in
pulsating turbulent air flow in a pipe using CFD code.
The simulation results on the fully developed pulsating
turbulent flow and heat transfer are compared with the
available experimental data. The findings suggest that
the heat transfer augmentation in the established flow
regions takes place only for a certain band of
frequencies. For extreme low and high frequencies, heat
transfer is actually reduced below that of steady flow.
The following may be concluded:
At all values of pulsation frequency, the
variations in the local time-averaged Nusselt
number exhibit similar trends to the steady
flows.
The local time-averaged Nusselt number either
increases or decreases than steady flow values
depending on the frequency and Reynolds
number.
Frequency has little effect on the values of
mean time-averaged Nusselt number especially
at higher values of pulsation frequency.
The presence of any pulsation frequency has a
noticed effect on the heat transfer coefficient in
comparison with that of steady flow, even at a
very small value of pulsation frequency.
However, the change of the value of pulsation
frequency has relativity small effect on the heat
transfer rates.
No increase in the mean time averaged-Nusselt
number for frequency f ≤39.34. For f ≥ 42.5,
little increase in time averaged-Nusselt number
is detected at Re ≈ 22.5x103
Nomenclature
Cp specific heat at constant pressure
, (J/Kg K)
C1, C2, Cµ empirical constant of k-ε model.
D diameter of pipe (m)
G the production of turbulent kinetic
energy (Kg/m s3
)
h instantaneous local heat transfer
coefficient, (W/m2
K)
ht instantaneous heat transfer coefficient
which is the average of heat transfer
coefficient over the whole length of
pipe at given moment, (W/m2
K)
hx local heat transfer coefficient which is
the average of heat transfer coefficient
over a cycle, (W/m2
k)
P pressure (pa)
K thermal conductivity, (W/m K)
k turbulent kinetic energy, (m2
/s2
)
L length of pipe, (m)
Nu overall Nusselt number
Nu t instantaneous Nusselt number
Nu x local Nusselt number
Pr Prandtl number, υ/α
q" heat flux per unit area(W/m2
)
r radial coordinate, (m)
R radius of pipe, (m)
Re Reynolds number, UD/υ
Tav average bulk temperature along cross-
sectional area of the pipe, (o
C)
Tin inlet temperature of the pipe (o
C)
Uin average of mean velocity at x=0, (m/s)
Uin =Um+ UA
Um Mean velocity of pulsating
component, (m/s)
UA Amplitude velocity of pulsating
component, (m/s)
UDF user defied function
u velocity component in the axial
direction (m/s)
v velocity component in the radial
direction (m/s)
x axial distance(m)
t time(s)
Greek symbols
α thermal diffusivity of fluid K/ρCp
(m2
/s)
ε turbulent energy dissipation rate
(W/kg)
φ viscous dissipation
µ dynamic viscosity (kg/m s)
µt eddy (or turbulent) viscosity (kg/m s)
υ kinematics viscosity (m2
/s)
ρ density (Kg/m3
)
σk turbulent Prandtl number for k
σT turbulent Prandtl number for T](https://image.slidesharecdn.com/numericalstudyofheattransferinpulsatingturbulentairflow-140804083108-phpapp02/85/Numerical-study-of-heat-transfer-in-pulsating-turbulent-air-flow-7-320.jpg)
![Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al.
ThETA01/06870
σε turbulent Prandtl number for ε
ω angular frequency (1/s)
ω* dimensionless frequency; (ω
D/U*
), U* is the fraction velocity
described as U* =0.199Um /Re-0.125
η Relative local Nusselt number =
(Nupx/Nusx)
Subscripts
s Steady state
p pulsation
m mean
REFERENCES
[1] J.Zheng, D. Zeng, W. Ping, G. Hong, 2004, ‘’An
Experimental Study of Heat Transfer Enhancement With
Pulsating Flow’’, Heat transfer-Asian research, Vol,
23(5), pp. 279- 286
[2] W.Xuefeng, Z.Nengli, 2005 “Numerical Analysis of Heat
Transfer in Pulsating Turbulent Flow in a Pipe’’, J Heat
Mass Transfer, Vol,. 48, pp.3957-3970.
[3] Faghri. M, Javdani. K, Faghri. A, 1979 “Heat Transfer
with Laminar Pulsating Flow in Pipe’’. Lett Heat Mass
Transfer, Vol, 6(4), pp. 259–270
[4] Hesham M. Mostafa, A. M. Torki,K. M .Abd-Elsalam ,
2005 “Experimental Study for Forced Convection Heat
Transfer of Pulsating Flow Inside Horizontal
Tube’’.4th
International Conference on Heat Transfer,
Fluid Mechanics and Thermodynamics,Cario,Ak3,
[5] Hesham M. Mostafa, A. M. Torki,K. M .Abd-Elsalam,
2005, “Theoretical study of laminar pulsating flow in
pipe’’4thInternational Conference on Heat Transfer
,Fluid Mechanics and Thermodynamics,Cario.
[6] Hemeada, H. N, Sabry, M. N. AbdelRahim. A. Mansour
H, 2002, “Theoretical analysis of heat transfer in laminar
pulsating Flow’’, J Heat Mass Transfer, Vol, 45, pp.
1767-1780.
[7] Guo, Z. and Sung, H.j, 1997,”Analysis of the Nusselt
Number in pulsating Pipe Flow’’, J Heat Mass Transfer,
Vol, 40(10), pp .2486-2489.
[8] H.Chattopadhyay, F.Durst, S.Ray, 2006, “Analysis of
heat transfer in simultaneously developing pulsating
laminar flow in a pipe with constant wall temperature’’, J
Heat Mass Transfer, Vol, 33, pp. 475-481.
[9] Al-Haddad A, Al-Binally N, 1989, “Prediction of Heat
Transfer Coefficient in Pulsating Flow’’. Int J Heat Fluid
Flow, Vol, 10(2), pp. 131–133.
[10] T. S. Zhao and P. Cheng, 1998, “Heat Transfer in
Oscillatory Flows’’. Annual Review of Heat Transfer,
Vol, IX, chapter 7, pp. 1-64.
[11] Habib M.A, Said S.A.M, Al-Dini S.A, Asghar A,
Gbadebo S.A, 1999 ,”Heat transfer characteristics of
pulsated turbulent pipe flow’’. J Heat Mass Transfer, Vol,
34(5), pp.413–421.
[12] S.K.Gupa,R. D. Patel, R. C. Ackerberg, 1982, “ Wall
heat /Mass Transfer in pulsating Flow’’, chemical
Engineering Science, Vol, 37(12), pp.1727-173
[13] M. H. I. Barid, G. J. Duncan, J. I. Smith, J. Taylor, 1996,
“Heat Transfer in pulsed Turbulent Flow’’, chemical
Engineering Science, Vol,. 21, pp. 197-199.
[14] S. A. Gbadebo, S.A.M. Said, SAM, M.A. Habbib, 1999,
“ Average Nusselt number correlation in the thermal
entrance region of steady and pulsating turbulent pipe
flows’’. J Heat Mass Transfer, Vol, 35, pp.377–381.
[15] A. E. Zohir.M. M.A. Habbi. A. M. Attya. A.l. Eid, 2005,
“An Experimental Investigation of Heat Transfer to
Pulsating Pipe Air Flow with Different Amplitudes’’. J
Heat Mass Transfer, Vol, 42 (7), pp. 625-635.
[16] O.k.Erdal, J. L. Gainer, 1979, “The effect of pulsation on
heat transfer’’ ,Int. Eng. Chem. Fundam, Vol,. 18(10), pp.
11-15.
[17] M.A. Habbib, A.M. Attya, S.A.M. Said, A.l. Eid, A.Z.
Aly,2004, “Heat transfer characteristics and Nusselt
number correlation of turbulent pulsating air flows’’. J
Heat Mass Transfer, Vol,. 40, pp. 307-318.
[18] M.A. Habbib, A.M. Attya, S.A.M. Said, A.l. Eid, and
A.Z. Ally, 2002, “convective Heat transfer characteristics
of laminar pulsating air flow’’. J Heat Mass Transfer,
Vol,. 38, pp. 221-232.
[19] H. W. Cho and J. M. Hyun, 1990, “Numerical solutions
of pulsating flow and heat transfer characteristics in a
pipe’’ ,Int. J. Heat and fluid flow, Vol,.11,(4), pp. 321-
330.
[20] D,Y.Lee,S.J.Park,S.T.Ro, 1998, “ Heat transfer in the
thermally region of a laminar oscillating pipe flow’’
Ctyogenics, Vol,. 38, pp. 585-594.
[21] Moschandreou, T and Zamir, 1997, “ Heat transfer in a
tube with pulsating flow and constant heat flux’’, J Heat
Mass Transfer. Vol,. 35(10), pp. 2461-2466.
[22] A.Scotti,U.Piomelli, 2001, “ Numerical simulation of
pulsating turbulent channel flow’’, Physics of fluid,
Vol,.13(5), pp. 1367-1384.
[23] S.A.M.Said,M.A.Habib,M,O,Iqal, 2003, “Heat transfer to
pulsating flow in an abrupt pipe
expansion’’,Int,J,Num.Heat,Fluidflow, Vol,. 13(3), pp.
286,308.
[24] Lev.shemer, 1985, “Laminar-turbulent transition in a
slowly pulsating pipe flow’’,Phys.Fluids, Vol,. 28(12),
pp. 3506-3509.
[25] Liao Ns, Wang CC, 1985, “An investigation of the heat
transfer in pulsating turbulent pipe flow’’. Fundamentals
of forced and Mixed Convection. The 23th Nat Heat
Transfer Conf Denver.Co.U.S.A. Aug. 4-7,53-59.
[26] GeninLg, KovalAp, ManachkhaSp, SviridovVG, 1992,
“Hydrodynamics and heat transfer with pulsating fluid
flow in tubes’’. Thermal Eng Vol,.39950, pp. 30-34.
[27] Havemann, H, A and Rao, N,N, 1954, “heat transfer in
pulsating flow’’.Nature , Vol,. 7(4418), pp. 41.
[28] M.artineelli R.C;Boeleter LMK;Weinberg Eb;Yakahi S,
1943, “Heat transfer to a fluid flowing periodical at low
frequencies in a vertical tube’’. Trans. ASME Oct, pp.
789-798.
[29] S. Y. Kim, B. H. Kang, J. M. Hyun, 1993, “Heat Transfer
in The thermally Developed Region of a Pulsating
Channel Flow”, J Heat Mass Transfer, Vol,. 36 (17), pp.
4257-4266.
[30] Jie-Chang Yu,Zhi-Xin Li, T. s. Zhaoan , 2004,
“analytical study of pulsating laminar heat convection in
a circular tube with constant heat flux'', J Heat Mass
Transfer, Vol,. 47, pp. 5297-5301.
[31]E. A.M.Elshafei, M. Safwat.Mohamed, H. Mansour;
M.Sakr, 2006, “Experimental Study of Heat Transfer in
Pulsating Turbulent Flow in a Pipe”, publications in
process (8th
International Congress of Fluid Dynamics &
Propulsion, 2006.
[32] Y.HÜSEYİN, B. GAMZE, K. NESRİN and Y. ŞENAY,
2005, “Numerical Study on Transient Local Entropy
Generation in Pulsating Turbulent Flow through an
Externally Hated Pipe” Sădhană, Vol. 30, Part 5, October
2005, pp. 625–648.](https://image.slidesharecdn.com/numericalstudyofheattransferinpulsatingturbulentairflow-140804083108-phpapp02/85/Numerical-study-of-heat-transfer-in-pulsating-turbulent-air-flow-8-320.jpg)
Numerical study of heat transfer in pulsating turbulent air flow
- 1. Thermal Issues in Emerging Technologies, ThETA 1, Cairo, Egypt, Jan 3-6th 2007 1-4244-0897-0/07/$20.00 ©2007 IEEE ThETA01/068 63 NUMERICAL STUDY OF HEAT TRANSFER IN PULSATING TURBULENT AIR FLOW E. A. M. Elshafei a , M. Safwat Mohamed b , H. Mansour; M. Sakrc Mechanical Power Department, Mansoura University Faculty of Engineering, El-Mansoura, EGYPT, (35516) eelshafei@mans.edu.eg a , msafwat@mans.edu.eg b , moh_saker@mans.edu.egc ABSTRACT A numerical investigation of heat transfer characteristics of pulsating turbulent flow in a circular tube is carried out. The flow is thermally and hydrodynamically fully developed and the tube wall is subjected to a uniform heat flux. The flow inlet to the pipe consists of fixed component and pulsating component that varies sinusoidally with time. The flow and temperature fields are computed numerically using computational fluid dynamics (CFD) Fluent code. Prediction of heat transfer characteristics is performed over a range of 104 ≤ Re ≤ 4x104 and 0 ≤ ƒ ≤ 70 are observed. Results showed little reduction in the mean time-averaged Nusselt number with respect to that of steady flow. However, in the fully developed established region, the local Nusselt number either increases or decreases over the steady flow-values depending on the frequency parameter. These noticed deviations are rather small in magnitude for the computed parameter ranges. The characteristics of heat transfer are qualitatively consistent with the available experimental and numerical predictions. 1- INTRODUCTION The study of pulsating flow in pipe has been a subject of interest among many researches. The problem is important in biological applications in relation to blood flow and in industrial applications related to heat exchange efficiency, cavitation in hydraulic pipe lines, pressure surges and reciprocating machines. In general, pulsating flow filed is assumed to consist of a steady poiseuille flow part and purely oscillating part. Many experimental, analytical and numerical works studied the heat transfer in pulsating laminar flow but little in turbulent flow. One of the key issues concerning pulsating convection heat transfer in tubes is whether a superposed flow pulsation effects on heat transfers in the original steady flow. The answer to this question in the previous studies can be classified into four different opinions: (1) flow pulsation enhances heat transfer [1, 2, 3]; (2) it deteriorates heat transfer [4, 5, 6, 7]; (3) it has no effect on heat transfer [8]; and (4) it either enhances or deteriorates heat transfer, depending on flow parameters [9-28]. Numerical study by Xuefeng Wang and Nengli Zhang [2] showed that in a pulsating turbulent flow there is an optimum Womersley number at which heat transfer is maximally enhanced. Experimental studies by [9, 11, 13-18, 25-28] showed that increment and reductions in mean Nusselt number depends on pulsation frequency, the turbulent bursting frequency, the amplitude, the axial location, Reynolds number and Prandtl number. In order to understand these phenomena and to resolve the contradictory results, different models of turbulence for pulsating flows were considered. Two of these models are well known, the quasi-steady flow model [24] and the bursting model discussed by [25-28]. Due to its complexity, studies on heat transfer in pulsating turbulent flow are far from matured. The aim of the present work is to investigate the mechanics of heat transfer in pulsating pipe flow numerically in order to show clearly how pulsation affects the rate of heat transfer and how does this phenomenon depend on the pulsation frequency and Reynolds number. A problem of pulsating flow in a pipe, externally subjected to a constant heat flux is considered in the present study. The governing equations are solved with the aid of the Fluent code. The numerical results can be served as a useful source material against which the outcome of future analytical and experimental investigations can be checked. 2- MATHEMATICAL FORMULATIONS 2.1 Pipe Geometry In the present study, the numerical solution of the pulsating turbulent flow through an externally heated pipe is analyzed. Air is selected as a working fluid. The radius and length of the pipe are R and L respectively. The thickness of the pipe is neglected and the thermal boundary condition on the pipe wall is assumed to be a uniform heat flux. The two-dimensional axisymmetric model of the considered pipe is shown in Fig.1. The pipe geometry at the inlet is long enough to assure a hydrodynamically fully developed flow at the heated section and simulated to the experiential facility of [32]. Since the pipe cross-section is circular, it is assumed that the flow is axisymmetric. In cylindrical polar coordinates, this means that the flow variables depend only on the axial coordinate x and radial coordinate r. The pipe wall section is modeled as: unheated section of 2.6 m length and a heated one of 1.16 m long. Sinusoidal pulsating flow is assumed to be entering to the pipe of fixed period (the velocity is only oscillates) as can be seen in Fig.2.
- 2. Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al. ThETA01/06864 2.2 The Governing Conservation Equations The governing equations for transient turbulent incompressible flow and heat transfer in the flow region (0 ≤ x ≤ L and 0 ≤ r ≤ R) can be written as follows [32]: Continuity: 0 ∂∂ ∂ = ∂ ++ r v r v x u (1) Momentum in x-direction: . 1 3 2 2 1 ∂ ∂ - )( ∂ ∂1 )( ∂ ∂1 ∂ ∂ ∂ ∂ + ∂ ∂ ∂ ∂ + + ∂ ∂ + ∂ ∂ − ∂ ∂ ∂ ∂ += ++ x v r u r rr r v r v x u x u r xrx p rvu rr ruu xrt u µ µ ρ Momentum in r-direction: ( ) ( ) . 3 2 2 3 2 2 1 1 11 2 + ∂ ∂ + ∂ ∂ +− + ∂ ∂ + ∂ ∂ − ∂ ∂ ∂ ∂ + ∂ ∂ + ∂ ∂ ∂ ∂ + ∂ ∂ −= ∂ ∂ + ∂ ∂ + ∂ ∂ r v r v x u rr v r v r v x u r v r rr r u x v r rrr P rvv rr ruv xrt v µ µ µ µ ρ Energy: φ r T r rrx T r T v x T u t T Cp eff µλ ρ + ∂ ∂ ∂ ∂ + ∂ ∂ = ++ 1 ∂ ∂ ∂ ∂ ∂ ∂ 2 2 Where φ is viscous dissipation term, which is φ and λeff is the effective thermal conductivity, and for standard k-ε model, is given by λeff = λ+ λt (3. c), λt = Cp µt /Prt (3. d) 2.3 Turbulence Model Among the available turbulent models, those based on eddy viscosity are widely applied to the turbulent kinetic energy – turbulent energy dissipation rate (k-ε). Two additional equations for the standard k-ε turbulence model: the turbulence kinetic energy, k, and the dissipation rate, ε, are determine using the following transport equations, respectively. k – Equation: Where ε- equation: ( )ρε ε ε σ µ µ ε σ µ µ ε ρ εε εε 21 ∂ ∂ ∂ ∂1 ∂ ∂ ∂ ∂ CGC k r r rrxxDt D tt −+ ++ += The turbulent viscosity, µt is computed by combining K - ε as follows: )/( 2 ερµ µ kCt = (6) Cµ, C1ε, and C2ε are the model constant; σk and σε are turbulent Prandtl numbers for k and ε, respectively. They have the following default values (Launder &Spalding 1972): Cµ= 0.09, C1ε= 1.44, C2ε= 1.92, σk =1.0, σε=1.3, (3.a) (2.a) (2.b) Uin T in Entrance length (2.6 m) Symmetry axis Fluid (Air) Pipe wall R L heated (1.16 m) x Uniform wall heat flux q" r Fig. 1 Schematic diagram for the studied problem. ,2 222 ∂ ∂ + ∂ ∂ + + ∂ ∂ + ∂ ∂ = r u x v r v r v x u (4.a) ρε σ µ µ σ µ µρ - ∂ ∂ ∂ ∂1 ∂ ∂ ∂ ∂ G r k r rr x k xDt Dk k t k t + ++ += (4.b) ++ + + = 2222 ∂ ∂ ∂ ∂ 2 ∂ ∂ 2 ∂ ∂ 2 x v r u r v r v x u G tµ 0 10 20 30 40 50 U(t)[m/s] Um UA Л 2Л Fig .2. Variations of inlet velocity (3.b) (5)
- 3. Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al. ThETA01/068 65 2-4 Heat Transfer Coefficient and Nusselt Number Due to the unique characteristics of the pulsating flow, several heat transfer coefficients have to be carefully defined. The instantaneous local heat transfer coefficient is given by avw TT txh - q′′ =),( (7-a), and the local heat transfer coefficient, defined as the average of heat transfer coefficient over a cycle, is formulated as ∫= π ω π 2 0 )( )( 2 1 thdh xx (7-b) The instantaneous heat transfer coefficient, defined as the average of heat transfer coefficient over the whole length of pipe at a given moment, is expressed as ∫= L tt hdx L h 0 )( . 1 (7-c) Subsequently, several relevant Nusselt numbers are rationally introduced. The instantaneous local Nusselt number is K hD txNu tx =),(, (7-d) The overall Nusselt number; Nu is given by ∫ ∫= L tx dxtdNu L Nu 0 2 0 , .)( 2 1 π ω π (7-e) To ensure sufficient temporal resolution of the numerical solution, five to seven cycles are required to obtain a steady periodic solution with an acceptable accuracy. 2-5 Boundary Conditions The solution domain of the considered 2D, axisymmetric pipe flow is geometrically quite simple, which is a rectangle on the x-r plane, enclosed by the inlet, outlet, symmetry and wall boundaries .On walls no-slip condition are assumed for the momentum equations. The inlet velocity values have been derived from given Reynolds numbers. Inaccuracies due to an uncertainty in the shape of the inlet velocity profiles are not expected to play an important role, since all calculations were carried out at places sufficiently down stream from the inlet, so that quite fully-developed conditions should be expected at the calculating section. For pulsating cases, the inlet velocity is varied with time. The outlet boundary condition is called "outflow", which implies zero-gradient condition at the outlet. At the pipe inlet, (x=0): For steady flow condition, U (t) = Uin, and for sinusoidal pulsating flow, U (t) = Um+ UA sin (ωt) At the pipe exit plane, (x = L): At the pipe axis, (r = 0): At the pipe wall, (r = R): (u, v k, ε) = 0, a uniform heat flux is imposed, q" = qο for heated section and q" = 0 for the adiabatic wall. Values of k and ε are not known at the inlet, but, if they are not given by experimental data, some reasonable assumptions can be made. The kinetic energy of turbulence according to a certain value of the square of the average inlet velocity is expressed by [ 33] k=0.03 Uin 2 In addition, dissipation is calculated according to the equation R k C 03.0 2 3 µε = 2-6 Simulation Values In the analysis, the following simulation values are taken as: Uin = (12.7 to 42.8) m/s, Um =11.07 m/s UA = (1.63 to 31.81) m/s, ω = (41.9 to 427.25) rad/s, and qο =922 W/m2 3- COMPUTATIONAL PROCEDURE 3-1 Calculation Tools Even the difficult general differential equations now yield to the approximating technique known as numerical analysis, whereby the derivates are simulated by algebraic relations between a finite numbers of grid points in the flow field which are then solved on a digital computer. A suitable CFD computer code is used to solve the governing equations numerically along with the boundary and the initial conditions. The CFD program is a process by which the fluid flow characteristics can be predicted through arbitrary geometries giving such information as: flow speed, pressure, residence time, flow patterns, etc. The Fluent 6·1 (Fluent 2003) program is chosen as the CFD computer code for this work because of the ease with which the analysis model can be created, and because the software allows users to modify the code for special analysis conditions through the use of user subroutines. The Fluent computer code uses a finite-volume procedure to solve the governing equations of fluid flow in primitive variables such as velocity components in 0= ∂ ∂ r φ 0= ∂ ∂ x φ
- 4. Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al. ThETA01/06866 axial and radial directions and pressure. A variety of turbulence models is offered by the Fluent computer code. The standard k–ε model was used as a turbulence model in this study. This model is a semi-empirical one, based on model transport equations for the turbulent kinetic energy (k) and its dissipation rate (ε). The model transport equation for k is derived from the exact equation while the model transport equation for ε is obtained using physical reasoning and bears little resemblance to its mathematically exact counterpart. In the derivation of the k–ε model, it was assumed that the flow is fully turbulent, and the effects of molecular viscosity are negligible. The standard k–ε model is, therefore, valid only for fully turbulent flows. A detailed description of turbulence models and its applications to turbulence can be found in "Fluent 2003; Launder & Spalding 1972". In the case of the standard k–ε models, two additional transport equations, (for the turbulent kinetic energy and the turbulence dissipation rate) are solved, and turbulent viscosity, µt, is computed as a function of k and ε. The solution method for this study is axisymmetric. In order to define the pulsating inlet velocities in all cases, an UDF (User-Defined Function) file was introduced to the prepared fluent case file. By using the results of the calculations performed with the fluent code. This program, written in FORTRAN 77 language, calculates numerically the local and average Nusselt Number. Grid size: Grid-independent tests were carried out to ensure grid independence of the calculated results; consequently, the grid size and the grid orientation giving grid independent results were selected, and thus a total cell number of 45000 cells (300× 150) were adopted. The grid points are clustered in the radial direction so that the finer spacing is formed near the wall and at the inlet. Fig.3 shows the grid. 4- RESULTS AND DISCUSSION 4-1 Steady Flow To validate the numerical calculations, the computational results for the fully developed steady turbulent pipe flow are compared with the experimental data observed by Elshafei et al. [31]. For two different values of Reynolds number, the numerical results are in consistent with the experimental results; within about ±5 % as can be noticed in Fig 4. The flow was observed to be thermally fully developed for X/D ≥ 20. 4-2 Pulsating Flow In the present work, investigations are performed within the ranges of 104 ≤ Re ≤ 4x104 , and 0 ≤ ƒ ≤ 70. The effect of pulsating frequency on the local time average Nusselt number at fixed Reynolds number of 16.8×103 is shown in Fig. 5. It can be seen that, as the axial location moves downstream, the time average local Nusselt number for both pulsating and steady flows decays rapidly from a very large value near the entrance and the axial decrease becomes less step in the bulk of the pipe soon after leaving this entrance region. The variation of the time averaged local Nusselt number for pulsating turbulent air flow exhibits a similar trend to that of the steady air flow case, but generally with lower values than that of the steady flow for f < 61.13. The maximum reduction of about 6% in the local Nusselt number was predicted in the thermal entrance region. For f ≥ 61.13, little enhancement in the local time average Nusselt number of about 8% for X/D ≥ 50 was detected. Figure 6 illustrates the variation of the ratio of the local Nusselt number value for pulsating flow to that of the steady flow value; η with the axial distance at Re = 22.5×103 . For f ≤ 39.3 Hz, it can be observed from Fig. 6-a that the ratio η starts rising rapidly in the entrance zone of the thermally developed region (X/D<20) and becomes nearly fixed below unity where its value in the bulk of the pipe soon after leaving the entrance length (X/D≥20) implies that the effect of pulsation on the local heat transfer diminishes. For f ≥ 42.5 Hz, the value of η close to the end of the pipe enhanced by about 4%as shown in Fig. 6-b. The impacts of other relevant parameters on heat transfer are illustrated in Fig.7 (a, b). The detected values of η due to the variation of pulsation frequency at higher value of Reynolds number (Re = 31.6×10 3 ). At lower frequency (ƒ ≤ 39.3 Hz), a small change in heat transfer ratio is visible near the entrance region. At (ƒ=13.3 Hz) a rapid increases in η can be observed up to X/D =20, after which it seems to be unchanged and generally, the value of η is lees than unity. The same trends can be seen for ƒ =28.3 Hz and ƒ =39.3 Hz. However, in the downstream region the value of η reaches about unity. Velocity Inlet Axisymmetric Adiabatic section Heated section X R Outlet Flow Fig.3 Computational grid for pipe
- 5. Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al. ThETA01/068 67 Fig.4. Variation of local Nusselt number at different Reynolds number The effect of changing pulsation frequency on the ratio η seems to be negligible specially, for higher values of pulsation frequency as can be noticed in Fig. 7.b. At high values of Reynolds number; Re ≥ 37×103 , certain qualitative trends can be shown in Figs. 8(a, b), which are similar to those described in the previous figures. For all studied values of pulsation frequency, the values of the ratio η are always less than unity. This global behavior of pulsating flow is consistent with the analytical and numerical results of [5, 8, 20, 24, 29, 30 and 31]. The overall patterns of η curves are qualitatively similar as discussed earlier. Fig.6.Distrbution of η at different frequency, Re=22500 Fig.7.Distrbution of ηx at different frequency, Re=31600 X/D Nux 0 20 40 60 0 100 200 300 400 Numerical Re=22500 Numerical Re=31600 Expermental Re=22500 Expermental Re=31600 η η 0 20 40 60 0.84 0.88 0.92 0.96 1 1.04 Re=22500 f=6.67 f=13.3 f=20.3 f=28.3 f=39.3 0 20 40 60 80 0.88 0.92 0.96 1 1.04 1.08 RE=22500 f=42.5 f=45.7 f=53.13 f=61.13 f=68 X/D X/D (b) (a) Nux Fig.5.Distrbution of Nux at different frequency, Re=16800 X/D 0 20 40 60 0 50 100 150 200 250 300 Re=16800 steady f=42.5 f=45.7 f=49.13 f=53.13 f=61.13 f=68 0 20 40 60 0 50 100 150 200 250 300 Re=16800 steady f=6.67 f=13.3 f=20.3 f=39.3 Nux X/D (b) (a) 0 20 40 60 0.92 0.94 0.96 0.98 1 1.02 Re=31600 f=13.3 f=28.3 f=39.3 0 20 40 60 0.94 0.96 0.98 1 Re=31600 f=42.5 f=45.7 f=53.13 X/D X/D η η (b) (a)
- 6. Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al. ThETA01/06868 Fig.8. Distribution of η at different frequency, Re=37000 Fig.9. Heat transfer ratio for different Reynolds Number and pulsation frequency. The third set of results pertains to the effect of the Reynolds number on the relative mean time- averaged Nusselt; ηm in case of pulsating flow in pipes. The calculated data for ηm versus Re are described in Figs.9 (a, b). In the range of pulsating frequency; ƒ ≤ 39.3, it is found that Reynolds number strongly affects the heat transfer ratio; ηm when Reynolds number is relatively low; Re < 25 x 103 as can be noticed in Fig. 9-a. The heat transfer ratio is sharply increases with Reynolds number up to nearly unity. However, when Reynolds number exceeds this value, the effect of increasing Reynolds number diminishes and the value of ηm remains nearly constant. It is also noticed that the variation of the imposed pulsating frequency has a relatively small effect on ηm. These computed results are in consistent with that reported by Xuefeng. W. and Nengli Z. [2] Fig.10. Comparison of the experimental data with numerical data at different (ω*) For higher pulsation frequency; ƒ ≥ 42.5, the value of the ratio ηm increases also with the increase of Reynolds number. As ηm reaches to its peak value at Re ≈ 22.5 x 103 , it then smoothly decreases with increasing Reynolds number up to 33 x 103 where ηm value is held nearly constant for higher values of Reynolds number as can be noticed in Fig. 9-b. The behavior of Nusselt number corresponding to the variation in both Reynolds number and pulsation frequency can also be explained in view of the bursting phenomena (Loiao and Wang [25] and Genin et al. [26]). The bursting model defines certain regions on frequency-Reynolds number plane, where, the bursting frequency can be dependent or independent of the pulsation frequency. The present results showed that the heat transfer coefficient may be increased or decreased with the change of both Reynolds number and frequency. The mean bursting frequency may be subdued to the pulsation frequency leading to the occurring of resonance that is dependent only on the pulsation 0 20 40 60 0.94 0.96 0.98 1 Re=37000 f=6.67 f=13.3 f=20.3 f=28.3 f=39.3 0 20 40 60 0.92 0.94 0.96 0.98 1 Re=37000 f=42.5 f=45.7 f=53.13 f=61.13 X/D X/D η η (b) (a) 15000 20000 25000 30000 35000 40000 0.84 0.88 0.92 0.96 1 f=6.67 f=13.3 f=20.7 f=39.3 Re (a) ηm Re (b) ηm 15000 20000 25000 30000 35000 40000 0.88 0.92 0.96 1 1.04 f=42.5 f=45.7 f=53.13 f=61.13 f=68 ω* ω* ηm (a) ω* ηm (b) ηm (c) 0 1 2 3 0.8 0.9 1 1.1 1.2 Re=37000 Experimental Numerical 0 2 4 6 0.8 0.9 1 1.1 1.2 Re=16800 Experimental Numerical 0 1 2 3 4 5 0.8 0.9 1 1.1 1.2 Re=22500 Experimental Numerical
- 7. Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al. ThETA01/068 69 frequency. Accordingly, the heat transfer process is expected to be affected resulting in either reduction or enhancement in the rate of heat transfer. These results are in consistent with those reported by Habibi et al [17] and Liao and Wang [25]. A comparison between the present numerical results with the experimental ones [32] over a ranges of 16 x 103 < Re< 38x103 and 6.67 ≤ ƒ ≤ 68 Hz are described in Fig. 10. The data are presented in the form of ηm as a function of bursting frequency ω* . It can be observed that for all values of Reynolds number, the numerical predictions of ηm as a function of ω* have the sametrend as those for the experimental data. The discrepancies between the experimental data and the computed ones at low bursting turbulent frequency are of about 10%. As ω* increases, the computed results closes to the experimental ones, specially, at higher value of Reynolds number. In summary, the present elaborated numerical results for heat transfer characteristics of turbulent pulsating flow are useful. It can serve as a source of materials against which further analytical data may be checked for consistency. 5-CONCLUSTION Simulation was performed to describe heat transfer in pulsating turbulent air flow in a pipe using CFD code. The simulation results on the fully developed pulsating turbulent flow and heat transfer are compared with the available experimental data. The findings suggest that the heat transfer augmentation in the established flow regions takes place only for a certain band of frequencies. For extreme low and high frequencies, heat transfer is actually reduced below that of steady flow. The following may be concluded: At all values of pulsation frequency, the variations in the local time-averaged Nusselt number exhibit similar trends to the steady flows. The local time-averaged Nusselt number either increases or decreases than steady flow values depending on the frequency and Reynolds number. Frequency has little effect on the values of mean time-averaged Nusselt number especially at higher values of pulsation frequency. The presence of any pulsation frequency has a noticed effect on the heat transfer coefficient in comparison with that of steady flow, even at a very small value of pulsation frequency. However, the change of the value of pulsation frequency has relativity small effect on the heat transfer rates. No increase in the mean time averaged-Nusselt number for frequency f ≤39.34. For f ≥ 42.5, little increase in time averaged-Nusselt number is detected at Re ≈ 22.5x103 Nomenclature Cp specific heat at constant pressure , (J/Kg K) C1, C2, Cµ empirical constant of k-ε model. D diameter of pipe (m) G the production of turbulent kinetic energy (Kg/m s3 ) h instantaneous local heat transfer coefficient, (W/m2 K) ht instantaneous heat transfer coefficient which is the average of heat transfer coefficient over the whole length of pipe at given moment, (W/m2 K) hx local heat transfer coefficient which is the average of heat transfer coefficient over a cycle, (W/m2 k) P pressure (pa) K thermal conductivity, (W/m K) k turbulent kinetic energy, (m2 /s2 ) L length of pipe, (m) Nu overall Nusselt number Nu t instantaneous Nusselt number Nu x local Nusselt number Pr Prandtl number, υ/α q" heat flux per unit area(W/m2 ) r radial coordinate, (m) R radius of pipe, (m) Re Reynolds number, UD/υ Tav average bulk temperature along cross- sectional area of the pipe, (o C) Tin inlet temperature of the pipe (o C) Uin average of mean velocity at x=0, (m/s) Uin =Um+ UA Um Mean velocity of pulsating component, (m/s) UA Amplitude velocity of pulsating component, (m/s) UDF user defied function u velocity component in the axial direction (m/s) v velocity component in the radial direction (m/s) x axial distance(m) t time(s) Greek symbols α thermal diffusivity of fluid K/ρCp (m2 /s) ε turbulent energy dissipation rate (W/kg) φ viscous dissipation µ dynamic viscosity (kg/m s) µt eddy (or turbulent) viscosity (kg/m s) υ kinematics viscosity (m2 /s) ρ density (Kg/m3 ) σk turbulent Prandtl number for k σT turbulent Prandtl number for T
- 8. Numerical study of heat transfer in pulsating turbulent air flow, E. A.M.Elshafei et al. ThETA01/06870 σε turbulent Prandtl number for ε ω angular frequency (1/s) ω* dimensionless frequency; (ω D/U* ), U* is the fraction velocity described as U* =0.199Um /Re-0.125 η Relative local Nusselt number = (Nupx/Nusx) Subscripts s Steady state p pulsation m mean REFERENCES [1] J.Zheng, D. Zeng, W. Ping, G. Hong, 2004, ‘’An Experimental Study of Heat Transfer Enhancement With Pulsating Flow’’, Heat transfer-Asian research, Vol, 23(5), pp. 279- 286 [2] W.Xuefeng, Z.Nengli, 2005 “Numerical Analysis of Heat Transfer in Pulsating Turbulent Flow in a Pipe’’, J Heat Mass Transfer, Vol,. 48, pp.3957-3970. [3] Faghri. M, Javdani. K, Faghri. A, 1979 “Heat Transfer with Laminar Pulsating Flow in Pipe’’. Lett Heat Mass Transfer, Vol, 6(4), pp. 259–270 [4] Hesham M. Mostafa, A. M. Torki,K. M .Abd-Elsalam , 2005 “Experimental Study for Forced Convection Heat Transfer of Pulsating Flow Inside Horizontal Tube’’.4th International Conference on Heat Transfer, Fluid Mechanics and Thermodynamics,Cario,Ak3, [5] Hesham M. Mostafa, A. M. Torki,K. M .Abd-Elsalam, 2005, “Theoretical study of laminar pulsating flow in pipe’’4thInternational Conference on Heat Transfer ,Fluid Mechanics and Thermodynamics,Cario. [6] Hemeada, H. N, Sabry, M. N. AbdelRahim. A. Mansour H, 2002, “Theoretical analysis of heat transfer in laminar pulsating Flow’’, J Heat Mass Transfer, Vol, 45, pp. 1767-1780. [7] Guo, Z. and Sung, H.j, 1997,”Analysis of the Nusselt Number in pulsating Pipe Flow’’, J Heat Mass Transfer, Vol, 40(10), pp .2486-2489. [8] H.Chattopadhyay, F.Durst, S.Ray, 2006, “Analysis of heat transfer in simultaneously developing pulsating laminar flow in a pipe with constant wall temperature’’, J Heat Mass Transfer, Vol, 33, pp. 475-481. [9] Al-Haddad A, Al-Binally N, 1989, “Prediction of Heat Transfer Coefficient in Pulsating Flow’’. Int J Heat Fluid Flow, Vol, 10(2), pp. 131–133. [10] T. S. Zhao and P. Cheng, 1998, “Heat Transfer in Oscillatory Flows’’. Annual Review of Heat Transfer, Vol, IX, chapter 7, pp. 1-64. [11] Habib M.A, Said S.A.M, Al-Dini S.A, Asghar A, Gbadebo S.A, 1999 ,”Heat transfer characteristics of pulsated turbulent pipe flow’’. J Heat Mass Transfer, Vol, 34(5), pp.413–421. [12] S.K.Gupa,R. D. Patel, R. C. Ackerberg, 1982, “ Wall heat /Mass Transfer in pulsating Flow’’, chemical Engineering Science, Vol, 37(12), pp.1727-173 [13] M. H. I. Barid, G. J. Duncan, J. I. Smith, J. Taylor, 1996, “Heat Transfer in pulsed Turbulent Flow’’, chemical Engineering Science, Vol,. 21, pp. 197-199. [14] S. A. Gbadebo, S.A.M. Said, SAM, M.A. Habbib, 1999, “ Average Nusselt number correlation in the thermal entrance region of steady and pulsating turbulent pipe flows’’. J Heat Mass Transfer, Vol, 35, pp.377–381. [15] A. E. Zohir.M. M.A. Habbi. A. M. Attya. A.l. Eid, 2005, “An Experimental Investigation of Heat Transfer to Pulsating Pipe Air Flow with Different Amplitudes’’. J Heat Mass Transfer, Vol, 42 (7), pp. 625-635. [16] O.k.Erdal, J. L. Gainer, 1979, “The effect of pulsation on heat transfer’’ ,Int. Eng. Chem. Fundam, Vol,. 18(10), pp. 11-15. [17] M.A. Habbib, A.M. Attya, S.A.M. Said, A.l. Eid, A.Z. Aly,2004, “Heat transfer characteristics and Nusselt number correlation of turbulent pulsating air flows’’. J Heat Mass Transfer, Vol,. 40, pp. 307-318. [18] M.A. Habbib, A.M. Attya, S.A.M. Said, A.l. Eid, and A.Z. Ally, 2002, “convective Heat transfer characteristics of laminar pulsating air flow’’. J Heat Mass Transfer, Vol,. 38, pp. 221-232. [19] H. W. Cho and J. M. Hyun, 1990, “Numerical solutions of pulsating flow and heat transfer characteristics in a pipe’’ ,Int. J. Heat and fluid flow, Vol,.11,(4), pp. 321- 330. [20] D,Y.Lee,S.J.Park,S.T.Ro, 1998, “ Heat transfer in the thermally region of a laminar oscillating pipe flow’’ Ctyogenics, Vol,. 38, pp. 585-594. [21] Moschandreou, T and Zamir, 1997, “ Heat transfer in a tube with pulsating flow and constant heat flux’’, J Heat Mass Transfer. Vol,. 35(10), pp. 2461-2466. [22] A.Scotti,U.Piomelli, 2001, “ Numerical simulation of pulsating turbulent channel flow’’, Physics of fluid, Vol,.13(5), pp. 1367-1384. [23] S.A.M.Said,M.A.Habib,M,O,Iqal, 2003, “Heat transfer to pulsating flow in an abrupt pipe expansion’’,Int,J,Num.Heat,Fluidflow, Vol,. 13(3), pp. 286,308. [24] Lev.shemer, 1985, “Laminar-turbulent transition in a slowly pulsating pipe flow’’,Phys.Fluids, Vol,. 28(12), pp. 3506-3509. [25] Liao Ns, Wang CC, 1985, “An investigation of the heat transfer in pulsating turbulent pipe flow’’. Fundamentals of forced and Mixed Convection. The 23th Nat Heat Transfer Conf Denver.Co.U.S.A. Aug. 4-7,53-59. [26] GeninLg, KovalAp, ManachkhaSp, SviridovVG, 1992, “Hydrodynamics and heat transfer with pulsating fluid flow in tubes’’. Thermal Eng Vol,.39950, pp. 30-34. [27] Havemann, H, A and Rao, N,N, 1954, “heat transfer in pulsating flow’’.Nature , Vol,. 7(4418), pp. 41. [28] M.artineelli R.C;Boeleter LMK;Weinberg Eb;Yakahi S, 1943, “Heat transfer to a fluid flowing periodical at low frequencies in a vertical tube’’. Trans. ASME Oct, pp. 789-798. [29] S. Y. Kim, B. H. Kang, J. M. Hyun, 1993, “Heat Transfer in The thermally Developed Region of a Pulsating Channel Flow”, J Heat Mass Transfer, Vol,. 36 (17), pp. 4257-4266. [30] Jie-Chang Yu,Zhi-Xin Li, T. s. Zhaoan , 2004, “analytical study of pulsating laminar heat convection in a circular tube with constant heat flux'', J Heat Mass Transfer, Vol,. 47, pp. 5297-5301. [31]E. A.M.Elshafei, M. Safwat.Mohamed, H. Mansour; M.Sakr, 2006, “Experimental Study of Heat Transfer in Pulsating Turbulent Flow in a Pipe”, publications in process (8th International Congress of Fluid Dynamics & Propulsion, 2006. [32] Y.HÜSEYİN, B. GAMZE, K. NESRİN and Y. ŞENAY, 2005, “Numerical Study on Transient Local Entropy Generation in Pulsating Turbulent Flow through an Externally Hated Pipe” Sădhană, Vol. 30, Part 5, October 2005, pp. 625–648.