International Journal of Thermal Sciences 120 (2017) 41e49

Contents lists available at ScienceDirect

International Journal of Thermal Sciences

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

Laminar forced convective heat transfer in helical pipe flow

Anup Kumer Datta a, *, Shinichiro Yanase a, Toshinori Kouchi a, Mohammed M.E. Shatat b
a
Graduate School of Natural Science and Technology, Okayama University, Okayama 700-8530, Japan
b
Faculty of Engineering, Port Said University, Egypt

a r t i c l e i n f o a b s t r a c t

Article history: Laminar forced convective heat transfer in a helical pipe with circular cross section subjected to wall
Received 24 February 2017 heating was investigated numerically by three-dimensional (3D) direct numerical simulations (DNS)
Received in revised form comparing with the experimental data obtained by Shatat (2010). The study was performed for three
18 April 2017
Prandtl numbers, Pr ¼ 8.5, 7.5 and 4.02, over the wide range of torsion. In 3D steady calculations, we
Accepted 29 May 2017
Available online 9 June 2017
found the appearance of fully-developed axially invariant flow regions, where the averaged Nusselt
number (averaged over the peripheral of the pipe cross section) were calculated, being in good agree-
ment with the experimental data. Because of the effect of torsion on the heat transfer characteristics, the
Keywords:
Helical pipe
averaged Nusselt number exhibits repetition of decrease and increases as torsion increases from zero for
Prandtl number all Reynolds numbers. It was found that there exists two maximums and two minimums of the averaged
DNS Nusselt number. It is interesting that the global minimum of the Nusselt number occurs at by0:1 and the
Forced convective heat transfer global maximum at by0:55.
Torsion © 2017 Elsevier Masson SAS. All rights reserved.

1. Introduction Yang et al. [2] numerically studied convective heat transfer of

The study of heat transfer in a helical pipe has been performed
extensively, both experimentally and numerically, because of a lot
of engineering applications and scientific interest. Due to higher
heat transfer efficiency compared to straight pipe configuration,
heat transfer in a helical pipe is frequently used in power genera-
tion, nuclear industries, refrigeration and air-conditioning systems,
heat recovery systems, food and dairy processes and other thermal
processing plants. A main flow feature in a helical pipe is the
presence of the secondary flow motion generated by the combined
effects of centrifugal force and pitch-induced torsion. The second-
ary vortices promote fluid mixing and in turn enhance heat transfer
performance resulting in larger pressure drop in the flow passage.
Austen and Soliman [1] experimentally studied the influence of
pitch on the pressure drop and heat transfer characteristics of he-
lical pipe flow for the condition of uniform input of heat flux. The
results included the isothermal and adiabatic friction factors, the
wall temperature, and the local Nusselt number. Significant pitch
effects were noted in the friction factor and the Nusselt number at
low Reynolds numbers. These effects were attributed to the natural
convection, and so they diminished as the Reynolds number
increased.
* Corresponding author.
the fully developed laminar flow in a helical pipe with a finite pitch.
They found that it was significantly affected by the Dean number,
torsion and the Prandtl number. The energy equation for the helical
pipe flow was derived on the basic law of the energy conservation.
In the convective heat transfer, they concluded that secondary flow
was stronger when the Dean number increased. For a large Prandtl
number, they found the temperature gradient near the bottom half
of the pipe increased markedly and that near the top half of the
pipe decreased.
Lin et al. [3] numerically studied 3D laminar forced convective
fluid flow and heat transfer in the entrance region (0e270 ) of a
helical pipe with a finite pitch (0e0.6) over the wide range of the
Reynolds number from 250 to 2000. By the analysis of the averaged
friction factor and the Nusselt number, they found the fully
developed heat transfer region. They also found that the effect of
curvature on the development of the averaged friction factor was
similar to that on the Nusselt number averaged over the pipe
circumference in the entrance region of the pipe and the wavy
behavior of the averaged Nusselt number increased in the whole
region of the helical pipe due to the increased of the Reynolds
number.
Rindt et al. [4] conducted numerical simulations of the evolution
of both free and forced convective flow in a helically coiled heat
exchanger with an axially varying wall temperature boundary
condition for the Reynolds number of 500, with comparison to the

http://dx.doi.org/10.1016/j.ijthermalsci.2017.05.026
1290-0729/© 2017 Elsevier Masson SAS. All rights reserved.
42 A.K. Datta et al. / International Journal of Thermal Sciences 120 (2017) 41e49
Nomenclature
dp diameter of the pipe
H pitch of the helical pipe
t unit tangential vector
n unit normal vector
b unit binormal vector
R distance of the center line from the axis of the whole
system
Pr Prandtl number
Tw wall temperature
Rh radius of curvature
P pressure
Tb bulk temperature
Dn Dean Number
Re Reynolds number
Nu Nusselt number
constant wall temperature boundary condition. The influence of
buoyancy force on the heat transfer and secondary flow was
addressed. It was found that, the wavy behavior of the Nusselt
number existed in a helical pipe due to the effects of both buoyancy
force and constant wall temperature boundary conditions. It was to
be noted that for higher buoyancy effects, however, this wavy
phenomena of the local Nusselt number diminished for the case
with an axially varying wall temperature due to the stabilizing ef-
fect of stratification.
Zheng et al. [5] performed numerical computations to study
heat transfer under the combined effects of the convection and
thermal radiation of laminar flow in a helical pipe with a finite
pitch. It was observed that the influence of the thermal radiation on
the axial velocity, secondary velocity and temperature field were
negligible, and that an increase in the total heat transfer rate due to
the presence of the thermal radiation was remarkable, but the
friction factor was not affected by the thermal radiation.
Shatat [6] experimentally investigated the influence of heat
transfer on fluid flow through straight and helical pipes, the wall of
which is heated, where the pipe wall temperature was always
greater than the inlet temperature. He found a wavy behavior of the
dimensionless mean temperature of the pipe wall in the entrance
region of the pipe until the temperature field became fully devel-
oped. The amplitude of these wavy behavior was found to increase
by the increase of the Dean number. He found that the averaged
Nusselt number remarkably increased by increasing the curvature
in the laminar as well as the turbulent flow regions, as a result of
the secondary flow generated from the centrifugal force presence
due to coil curvature.
Jayakumar et al. [7] numerically investigated a correlation for
the prediction of the local Nusselt number (the development of
heat flow along any side of the pipe wall) of fluid flow inside
vertically oriented helically coiled pipes. It was pointed out that the
boundary condition of wall temperature of this study was lower
than the inflow temperature. It was concluded that as the diameter
of the helical pipe increased, the effect of the centrifugal force due
to coil curvature decreased. The wavy behavior of the local Nusselt
number was also observed due to the rotational force induced by
the torsion of the helical pipe, which, however, did not affect the
averaged Nusselt number over the entire heat transfer surface.
Datta et al. [8] numerically studied the non-thermal helical pipe
flow with circular cross section, where they found a good agree-
ment between the numerical results and experimental data. They
qw rate of heat flux
u velocity component in the x-direction
v velocity component in the y-direction
w velocity component in the z-direction
x horizontal axis
y vertical axis
z axis in the direction of the main flow
a radius of the pipe
Greek letters
d curvature
b torsion parameter
a thermal diffusivity
m viscosity
t torsion
y kinematic viscosity
r density
generated an appropriate 3D grid structure of the helical pipe by
the helical orthogonal coordinates system, where OpenFOAM was
used. To confirm accuracy of the simulation, they conducted nu-
merical simulations by increasing grid points of the mesh system
and found the simulation to be accurate enough for the study of
laminar-turbulent transition. It should be remarked that we used
the same coordinates system of Datta et al. [8] for the present
numerical simulations, where the wall of the pipe is homoge-
neously heated.
Although many researches have been conducted on the forced
convection in a helical pipe, there is no investigation where com-
parison of the numerical and experimental results are conducted
and a comprehensive study is performed over the wide range of
torsion to the authors' knowledge.
In the present paper, we study laminar forced convection in a
helical pipe with circular cross section experimentally and
numerically. The numerical study is performed by use of the open
source code (OpenFOAM [9]), in the same manner as Datta et al. [8].
The objective of this study is to explore the effect of torsion over the
wide range, on the averaged Nusselt number for a helical pipe with
heated pipe wall based on the comparison with experimental re-
sults obtained by Shatat [6]. We also studied the effect of torsion on
the secondary flow, the projection of the trajectory of fluid particles
on the cross section, the temperature field and the local Nusselt
number along the peripheral circle over the cross section.
2. Numerical simulation
2.1. Helical pipe geometry
Consider a laminar incompressible viscous Newtonian fluid
streaming through a helical pipe with constant curvature and pitch
length as shown in Fig. 1(a). The cross section of the helical pipe is a
circle with the internal diameter dp (2a), and H is one pitch length of
the helical pipe. The unit tangential vector, t, the unit normal
vector, n, and the unit binormal vector, b of the center line of the
pipe are defined, where b ¼ t  n (see Fig. 1(b)). The x-axis is to the
direction of ­n, the y-axis to the direction of ­b and the z-axis is a
taken along the center line of the pipe. The x, y and z coordinates
constitute the orthogonal coordinate system. The flow is driven by
constant pressure gradient along the center line of the pipe and the
main flow direction is to the positive z. A thick broken line denotes
the center line of the helical pipe and R is the distance of the center
 A.K. Datta et al. / International Journal of Thermal Sciences 120 (2017) 41e49
Fig. 1. (a) Helical pipe with circular cross section, (b) coordinate system.
line from the axis of the whole system.
The radius of curvature of the center line of the helical pipe
Rh ¼ R½1 þ ðH=2pRÞ2 . Then, the non-dimensional curvature of the
pipe d ¼ a=Rh , the non-dimensional
pffiffiffiffiffiffi torsion t ¼ aðH=2pÞ=Rh 2 and
the torsion parameter b ¼ t= 2d as introduced by Kao [10]. We
took the curvature of the helical pipe d ¼ 0:018 and b ¼ 0:0012 in
order to compare the present DNS results with the experimental
data by Shatat [6]. We made the same configuration of the helical
pipe geometry as the experimental results.
2.2. Governing equations
The continuity equation, momentum equation and temperature
equation for the laminar viscous incompressible fluid are
V$U ¼ 0;
vU 1 OpenFOAM reduces Eqs. (1)e(3) into a set of algebraic equations
þ ðU$VÞU ¼  VP þ vV2 U;
vt r
vT
þ ðU$VÞT ¼ aV2 T :
vt
Here, U ¼ ðu; v; wÞ is the velocity vector, where u, v, w are the
velocity of the x, y and z components, respectively. t the time, P the
pressure, v ð ¼ m=rÞ the kinematic viscosity, r the density. T the
temperature and a the thermal diffusivity of the fluid. The Prandtl
number is given by
y
Pr ¼
a and a fixed value (atmospheric pressure) for the pressure were
2.3. Numerical simulation method
The open source code (OpenFOAM) was utilized to calculate
laminar forced convective flow in a helical pipe. By using Open-
FOAM [9], Eqs. (1)e(3) are solved by way of the cell-centered finite
volume method (FVM) without turbulence model, which means
that we conducted DNS by way of OpenFOAM. To apply OpenFOAM,
it is necessary to generate a proper mesh in the flow domain. The
cross sectional mesh in the x-y plane was generated as illustrated in
Fig. 2, and the uniform mesh was generated in the z-direction. A
careful mesh generation study was carried out in order to provide a
43
Fig. 2. Mesh on the cross section of the helical pipe.
(1)
good numerical accuracy.
(2) and solves them. Numerical solutions are iteratively sought by
pressure correction method called SIMPLE algorithm [11]. The nu-
merical solution is assumed to be convergent if the maximum re-
(3) sidual error in any of the primitive variables such as velocity
components, pressure, over the whole grids is less that 106 be-
tween one successive iteration. It should be remarked that we
developed a new OpenFOAM code to calculate the forced convec-
tion, because no code is available for the forced convection in the
circulating OpenFOAM.
As the inflow boundary condition, we imposed a uniform axial
flow determined by the Reynolds number for the velocity and the
(4) Neumann condition for the pressure. Zero gradient for the velocity
imposed as the outflow boundary conditions. No slip boundary
condition was imposed on the wall. As for the temperature
boundary conditions, a uniform heated temperature Tw is assumed
on the pipe wall. Uniform axial flow at temperature T0 enters into
the helical pipe at the inlet and zero gradient was imposed for
temperature at the outlet. The physical properties of working fluid
depends on the bulk temperature given at each axial position.
The computational mesh numbers have been varied in order to
check the numerical accuracy, where the number of mesh in the
radial direction is M, and N in the circumferential direction. The
uniform mesh was generated in the z-direction, where the number
of mesh is L. We obtained the steady solutions at Re ¼ 1633 and d ¼
0:018 for b ¼ 0:0012 and Pr ¼ 8.5 for four types of the mesh, which
44 A.K. Datta et al. / International Journal of Thermal Sciences 120 (2017) 41e49

are named as M1, M2, M3 and M4 as listed in Table 1. For each mesh, The heat transfer experiments were carried out by heating the
the maximum axial velocity wmax, the averaged Nusselt number, outer surface of the helical pipe by electric current when the fluid
Nu over the circumference of the pipe wall at a fully developed passed the test section. A heat exchanger was used to control the
region and the total number of grids are shown in Table 1. Based on inlet temperature of the working fluid within the range
Table 1, we concluded that M3 is sufficient for the computation of 9.6  Ce10.8  C before entering the test section. Pipe wall temper-
the force convective heat transfer in a helical pipe. ature was measured at equally spaced 14 axial positions along the
helical pipe axis utilizing T-type thermocouples mounted on top,
bottom, outer and inner sides of the helical pipe at each position. In
2.4. Nusselt number
addition to wall thermocouples, another two thermocouples were
mounted in the inlet and outlet of the test sections to measure the
In the present study, the mean axial velocity w is defined by the
bulk temperatures. Linear increase of the bulk temperature was
volumetric flux Q as
expected as the heat added was uniformly distributed over the pipe
Q wall. All thermocouples were connected to a data logger (Advantest
w¼ : (5) R7326-B). Before installation, all thermocouples were calibrated
p a2
against a standard precision thermometer using a constant tem-
The Reynolds number, Re, and the Dean number, Dn, is defined perature water bath. After thermocouples were installed, the test
by section was covered with 100 mm thickness layer of glass wool to
pffiffiffi minimize the heat loss to the environment.
w 2a
Re ¼ ; Dn ¼ Re d: (6)
n
4. Results
The Nusselt number, Nu, is defined as an index of the heat
transfer from the heated wall to fluid as In this section, the results of DNS are presented. In the first part,
steady solutions obtained by DNS are compared with the experi-
qw ð2aÞ
Nu ¼ ; (7) mental data (Shatat [6]) in the fully developed region of the helical
k ðTw  Tb Þ pipe for b ¼ 0:0012 and d ¼ 0:018. In the second part, we obtained
the Nusselt number averaged over the pipe circumference in
where qw the rate of heat flux and k the thermal conductivity of the consideration of the effect of torsion.
fluid. Tb the bulk temperature of the fluid defined by

1 4.1. Nusselt number

Tb ¼ ∬ w T dx dy
pa2 w cross section
where the integration is performed over the cross section.
3. Experimental setup
3.1. Flow and heat transfer measurements
A schematic diagram of the experimental setup used for flow
and heat transfer measurements is shown in Fig. 3.
Water was used as working fluid, which are circulated by a
centrifugal pump 2 controlled by a variable-speed controller from a
water tank 1 to a heat exchanger 3 in order to control the fluid
temperature. Then it entered the test section 5 after going through
a 2-m long straight entrance region 4 to be nearly fully developed.
After the test section, the working fluid was directed to a flowmeter
6 to measure the flow flux rate and returned to the tank 1. In case of
relatively high temperature of the fluid, it was directed to the water
sink to remove the heat added in the test section and replaced by
fresh water from the water source. The helical pipe used in the test
section of the experiment was one the inner diameter of which is
14.4 mm, the outer diameter 16 mm. The pitch of the helical pipe is
32 mm with three turns, and so its non-dimensional curvature
d ¼ 0:018, torsion parameter b ¼ 0:0012.
(8)
The Nusselt number, Nu, defined in Eq. (7), is used as an index of
the heat transfer of forced convection from the uniformly heated
wall. It should be pointed out that since the experiment was per-
formed for Pr ¼ 8.5, we compared the present DNS results with the
experimental data for this case. To validate the present DNS, we
obtained Nu for two cases of Pr. They are plotted with the symbol C
for Pr ¼ 8.5 and : for Pr ¼ 7.5 as a function of the Reynolds number
in Fig. 4, where Manlapaz and Churchill's [12] correlation formula
2 3
!3  3=2
48 51=11 Dn
Nu ¼ 4 þ 2 þ 1:816 5
11 1 þ 1342=Pr Dn2 1 þ 1:15=Pr
(9)
is also shown with a solid line and the experimental data with a
symbol -. It should be remarked that Eq. (9) is valid for negligible
torsion for b ≪ 1.
As shown in Fig. 4, the averaged Nusselt number increases as Re
increases. It is found that the present DNS results for Pr ¼ 8.5 is very
close to the experimental data, and Eq. (9) over the wide range of
the Reynolds number, 1000  Re  4100. It should be pointed out
that Yamamoto et al. [13] experimentally found the critical Rey-
nolds number of the laminar to turbulent transition, Rec ¼ 2600, for
b ¼ 0:002 and d ¼ 0:01 with no thermal effect, by observing the

Table 1
Grid sensitivity analysis.

Mesh name Mesh size Nu wmax [m/s] Total number of grid

M N L

M1 8  11  360 15.38 938 1.66 007 249 600

M2 10  13  360 16.53 598 1.67 090 372 000
M3 12  16  360 17.82 339 1.68 025 527 400
M4 14  18  360 17.84 031 1.68 158 722 400
 A.K. Datta et al. / International Journal of Thermal Sciences 120 (2017) 41e49 45

Fig. 3. Experimental setup.

Fig. 5. The Nusselt number for d ¼ 0:018 at Pr ¼ 4.02. The symbol ; represents the
Fig. 4. The Nusselt number for b ¼ 0:0012 and d ¼ 0:018. The symbols C and : Nu for b ¼ 0:1, B for b ¼ 0:3, : for b ¼ 0:45 and , for b ¼ 0:8.
represents the present DNS results for Pr ¼ 8.5 and 7.5, respectively, and the experi-
mental data by Shatat [6] with the symbol - for Pr ¼ 8.5.

4.2. Torsion effect on the Nusselt number

unsteady behavior of secondary and axial flow. Since the velocity
In order to study the effect of torsion on the Nusselt number
field is hardly affected by the temperature field in the case of forced
over the range of Reynolds number 1000  Re  4100, we con-
convection, the critical Reynolds number may be nearly the same
ducted numerical calculations for b ¼ 0:10; 0:30; 0:45 and 0:80 at
for the present forced convection.
For Pr ¼ 8.5, we found that for higher Reynolds numbers
d ¼ 0:018 and Pr ¼ 4:02 without turbulence model. It should be
remarked that no experimental data nor numerical result is avail-
(Re > 2600), the Nusselt numbers obtained by the present DNS are a
able for higher torsion case to the author's knowledge of literature.
little higher than the values obtained experimentally, while the
Since the present DNS results agree with the experimental data
Nusselt numbers by DNS are in an excellent agreement with
pretty well for b ¼ 0:0012 as shown in Fig. 4, the accuracy of the
the experimental data for Re < 2600. The small differences
present calculations is assured.
for Re  2600 are probably due to the turbulence effect. We also
It should be remarked that Shatat [6] experimentally found
found that Nu slightly decreases as the Pr decreases, as shown in
that the temperature averaged over the peripheral circle of the
Fig. 4 for Pr ¼ 7.5. For Pr ¼ 7.5, Nu is in fairly good agreement with
pipe wall varies linearly from the inlet to the outlet of the helical
Eq. (9) although the curve of Eq. (9) for Pr ¼ 7.5 is not shown in
pipe, which is in good agreement with the present DNS result.
Fig. 4.
46 A.K. Datta et al. / International Journal of Thermal Sciences 120 (2017) 41e49
Shatat [6] found that the fully developed region exists in one turn
after two turns from the inlet of the helical pipe except for the
region near the outlet of the helical pipe. On the other hand, we
numerically found the fully developed region in two turns after
three turns from the inlet of the helical pipe. We compared the
experimental results with DNS data in the fully developed region
for each study.
In Fig. 5, we show DNS results for five values of b, where Nu
decreases as b increases from zero for 1000  Re  4100, and a
minimum of Nu occurs at b ¼ 0.1. It is interesting that Nu tends to
increase from a minimum as b increases for b > 0:1. For a wide
range of b, it is found that a maximum of Nu occurs at b ¼ 0:45. If b
further increases, Nu decreases again.
Yang et al. [3] found that Nu significantly decreases as b in-
creases from zero (a toroidal pipe) by the 2D numerical simulation.
The present 3D DNS result is in good agreement with the result of
Yang et al. [3] for b  0:1.
The variation of Nu as a function of b is shown in Fig. 6 for three
Reynolds numbers (Re ¼ 1047, 1633 and 2600) at Pr ¼ 4.02 and
d ¼ 0:018. It should be remarked that turbulence may affect the
value of Nu for Re > 2600 in experiments (Yamamoto et al. [13]). In
Fig. 6, it is clear that Nu increases as Re increases. It is found in Fig. 6
that each curve of Nu has two local minimums and two local
maximums, where the global minimum exists at by0:1 and the
global maximum at by0:55 for any value of Re.
4.3. Torsion effect on the secondary flow and temperature field
In the present 3D steady calculations, we found the fully
developed flow region, where the flow and temperature profiles do
not change appreciably with the pressure gradient vp=vz being
nearly constant in the z-direction. This region lies in the part of the
helical pipe after three turns from the inlet of the helical pipe
except for the region near the outlet. It should be noted that forced
convection in a helical pipe is affected by the secondary flow in the
cross section very much since the temperature field is strongly
dependent on it. Therefore, it is important to examine the sec-
ondary flow pattern in the forced convection.
A helical pipe is characterized by the value of b. When b is zero,
the helical pipe reduces to a toroidal pipe, and if b tends to infinity,
it approaches a straight pipe. One of the important differences
between a laminar flow in a toroidal pipe and that in a helical pipe
Fig. 6. The Nusselt number as a function of b for d ¼ 0:018 at Pr ¼ 4.02. The line of C
for Re ¼ 1047, - for Re ¼ 1633 and : for Re ¼ 2600.
is that two symmetrical vortices appear in the cross section for the
toroidal pipe, while one dominates another for the helical pipe due
to the effect of the finite value of b.
In order to the study the cases of the minimal and maximal
averaged Nusselt number, we show the vector plots of the sec-
ondary flow for b ¼ 0:002 (small or negligible torsion case), 0.1
(minimal averaged Nusselt number), 0.55 (maximal averaged
Nusselt number) at the fully developed region for Re ¼ 1047,
d ¼ 0:018 and Pr ¼ 4.02 in Fig. 7. We also show the cut section of the
contours of the Q-value (the second invariant of the velocity
gradient tensor) surface (see [14]) on the same figure.
For the small torsion case (b ¼ 0:002), the axial flow yields the
centrifugal force toward outside of the pipe, and the fluid is moved
from the inner to outer side. Because of mass conservation, the fluid
returns to the inner side around the upper and lower region of the
cross section. As a result, secondary flow is formed in the cross
section exhibiting two symmetric circulating cells shown in
Fig. 7(a). If torsion effect begins the work, the symmetric two-cell
pattern changes into the asymmetric two-cell pattern, as shown
in Fig. 7(b).
In Fig. 7(b), we observe asymmetric secondary flow with a
slightly weak anti-clockwise vortex in the upper semicircle and a
slightly strong clockwise vortex in the lower semicircle for b ¼ 0:1.
For b ¼ 0:55 in Fig. 7(c), a large and strong vortex is produced in the
lower part of the cross section and the vortex becomes small and
weak in the upper part. This shows that, torsion of the pipe in-
creases the strength of the vortex rotating in the same direction as
torsion. It should be noted that since torsion works as a rotational
force, its effect is to induce a circulating motion in the fluid. It is
seen in Fig. 7(c) that for b ¼ 0:55, the secondary flow approxi-
mately consists of one large vortex, the center of which lies in the
middle of third and fourth quadrant of the cross section.
To supply additional insight into the vector plot of secondary
flow, we show the 2D stream lines, the projection of the trajectory
of fluid particles on the cross section for several values of b as
shown in Fig. 8, where the vortical structure is seen more clearly.
Then, we study the effect of torsion on the temperature field,
where the temperature is non-dimensionalized by TTT w
. Fig. 9 dis-
w Tb
plays the temperature field in the cross section of the helical pipe
obtained by DNS at the fully developed flow region for
b ¼ 0:002; 0:1 and 0.55 at Re ¼ 1047, d ¼ 0:018 and Pr ¼ 4.02,
where the wall of the helical pipe is homogenously heated. The
contour lines are drawn at every 10 of the temperature profile for
b ¼ 0:002 and 0:1 and every 20 of the temperature profile for
b ¼ 0:55.
It should be noted that the minimum temperature region exists
inside the pipe because fluid particles gain heat on the pipe wall
and move into the inner pipe region convected by the axial and
secondary flow. It is found that a contour of the minimum tem-
perature lies in the middle of the first and second quadrant on the
upper semi-circle for b ¼ 0:002 and the second minimum lies in the
middle of the third and fourth quadrant, as shown in Fig. 9(a). In
Fig. 9(b) for b ¼ 0:1, we notice that the temperature field becomes
slightly asymmetric similarly as the secondary flow in Fig. 7(b). As b
increases, the asymmetry of the temperature field becomes con-
spicuous as shown in Fig. 9(c) for b ¼ 0:55. We find the minimum
temperature region lies in the second quadrant on the upper
semicircle of the cross section.
4.4. Local Nusselt number
Fig. 10 shows the distribution of the local Nusselt number along
the peripheral circle over the cross section at the fully developed
flow region for Re ¼ 1047, d ¼ 0:018 and Pr ¼ 4.02. The torsion
parameter b, which controls the flow characteristics by the
 A.K. Datta et al. / International Journal of Thermal Sciences 120 (2017) 41e49 47

Fig. 7. Vector plots of the secondary flow and contours of the Q-value for Re ¼ 1047, d ¼ 0:018 and Pr ¼ 4.02, where an outer wall is on the right-hand side of each figure. The thick
lines denote the contour of the Q-value, where Q ¼ 1.0 for all figures. Fig. 7(a) is for b ¼ 0.002 at z ¼ 0.16 m, (b) b ¼ 0.1 at z ¼ 5.95 m and (c) b ¼ 0.55 at z ¼ 2.01 m.

Fig. 8. 2D stream lines of the secondary flow for Re ¼ 1047, d ¼ 0:018 and Pr ¼ 4.02, where an outer wall is on the right-hand side of each figure. Fig. 7(a) is for b ¼ 0.002 at z ¼ 0.16,
m (b) b ¼ 0.1 at z ¼ 5.95 m and (c) b ¼ 0.55 at z ¼ 2.01 m.

Fig. 9. Contours of the temperature field for Re ¼ 1047, d ¼ 0:018 and Pr ¼ 4.02, where the outer wall is on the right-hand side of each figure. Fig. 8(a) is for b ¼ 0.002 at z ¼ 0.16 m,
(b) b ¼ 0.1 at z ¼ 5.95 m and (c) b ¼ 0.55 at z ¼ 2.01 m. The increment of the temperature is 10 for b ¼ 0.002 and 0.1 and 20 for b ¼ 0.55.

rotational force varies from 0.002 to 0.55. The abscissa in Fig. 10 is
the angle along the peripheral circle of the cross section and the
ordinate the local Nusselt number at different positions on the
circle.
For the small torsion case (b ¼ 0:002), we found the wavy
behavior of the local Nusselt number on the circle as shown with a
solid line in Fig. 10. This result is similar to the previous study of
Yang et al. [2]. In its wavy behavior, it is clearly seen that two
maximal peaks of the local Nusselt number exist in the upper and
lower semicircles and the minimum value of the local Nusselt
number is located on the inside wall. We notice that the peak of the
local Nusselt number is slightly higher at the upper wall than at the
lower wall. It is found that the main contribution to the averaged
Nusselt number comes from these two peaks.
It is interesting that for b ¼ 0:1, any peak of the local Nusselt
number disappears and the distribution of the local Nusselt num-
ber becomes rather flat as shown by a dashed line in Fig. 10. The
rotational force due to increase of torsion parameter weakened the
two-cell pattern of the secondary flow and consequently wiped out
the peaks of the local Nusselt number. As a result the averaged
Nusselt number takes a bottom at by0:1.
However, if b further increases and becomes 0.55, the distri-
bution of the local Nusselt number goes upward as shown by a
dotted line in Fig. 10, so that the averaged Nusselt number recovers
its value and becomes greater than that for b ¼ 0:002. It is caused
by a strong swirling motion of the secondary flow at large torsion
parameter. The moderate peak of the local Nusselt number lies
around the inside wall.
In summary, for small torsion parameter, heat transfer from
the pipe wall to fluid occurs at two local regions on the wall due to
the two-cell pattern of the secondary flow. For large torsion
parameter, on the other hand, heat transfer occurs globally on the
wall due to the large one cell pattern of the secondary flow. For the
intermediate torsion parameter, total heat transfer becomes
48 A.K. Datta et al. / International Journal of Thermal Sciences 120 (2017) 41e49
Fig. 10. Local Nusselt number distribution for Re ¼ 1047, d ¼ 0:018 and Pr ¼ 4.02. The solid is for b ¼ 0.002 at z ¼ 0.16 m, a dashed line for b ¼ 0.1 at z ¼ 5.95 m and a dotted line for
b ¼ 0.55 at z ¼ 2.01 m.
minimum because both the local and global heat transfer are not
effective.
5. Conclusion
3D DNS study of the flow through helical pipes has been con-
ducted and the averaged Nusselt number was obtained for the wide
range of torsion parameter, b, three Reynolds numbers, Re ¼ 1047,
1633 and 2600, three Prandtl numbers, Pr ¼ 8.5, 7 and 4.02 and
curvature, d ¼ 0:018. Obtained results were compared with the
experimental data (Shatat [6]) and also the previous study (Yang
et al. [2]).
In 3D steady calculations with homogeneously heated wall
temperature, we found the appearance of the fully developed flow
region, where we compared the averaged Nusselt number obtained
by DNS with the experimental data and the formula by Manlapaz
and Churchill's [12]. For low torsion parameter, b ¼ 0:0012, we
found the Nusselt number of the present DNS was in good agree-
ment with the experimental data for Pr ¼ 8.5. If the Prandtl number
decreases, we found that the Nusselt number slightly decreases at
low torsion case, which is in good agreement with the results by
Yang et al. [2].
If b increases from zero (a toroidal pipe), then we found the
averaged Nusselt number of the helical pipe begins to decrease
from that of a toroidal pipe, and increases after taking the global
minimum at by0:1. If b further increases, we found a local
maximum of the averaged Nusselt number, which decreases again
and then takes the global maximum at by0:55 for any value of the
Reynolds number. For the wide range b, we found that the torsion
effect is remarkable on the averaged Nusselt number for any value
of the Prandtl number, and torsion parameter was found to play an
important role in the forced convection of helical pipe flow.
In order to obtain physical insight into the reason of variation of
the averaged Nusselt number as b changes, we plotted the sec-
ondary flow, 2D stream lines and the temperature field on the cross
section for several values of b. It was found that the secondary flow
exhibits nearly symmetric two-cell pattern for small b case, then
becomes asymmetric and finally changes into one-cell pattern as b
increases. The temperature field changes from a slightly asym-
metric form to a strongly asymmetric form via an asymmetric form.
We also plotted the local Nusselt number over the peripheral circle
and studied the topological change of the temperature distribution
as b increases. It was found that the minimum of the averaged
Nusselt number occurred when both the effect of the secondary
flow with symmetric two-cell pattern and that with one-cell
pattern become feeble for by0:1. We hope the present result
may contribute to the improvement of heat exchanger using helical
pipes.
Acknowledgments
A. K. Datta would like to acknowledge gratefully the financial
support from the Japanese Ministry of Education, Culture, Sports,
Science and Technology (MEXT) for study in Japan. S. Yanase would
like to give their cordial thanks to the MEXT for the financial sup-
port through the Grant-in-Aid for Scientific Research, No.
15K05814.
References
[1] Austen DS, Soliman HM. Laminar flow and heat transfer in helically coiled
tubes with substantial pitch. Exp Therm Fluid Sci 1988;1:183e94.
[2] Yang G, Dong ZF, Ebadian MA. Laminar forced convection in a helicoidal pipe
with finite pitch. Int J Heat Mass Transf 1995;38(5):853e62.
[3] Lin CX, Zhang P, Ebadian MA. Lamianr forced convection in the entrance re-
gion of helical pipes. Int J Heat Mass Transf 1997;40(14):3293e304.
[4] Rindt CCM, Sillekens JJM, van Steenhoven AA. The influence of the wall
temperature on the development of heat transfer and secondary flow in a
coiled heat exchanger. Int Comun Heat mass Transf 1999;26(2):187e98.
[5] Zheng B, Lin CX, Ebadian MA. Combined laminar forced convection and
thermal radiation in a helical pipe. Int J Heat Mass Transf 2000;43:1067e78.
[6] Shatat MME. Influence of micro-bubbles on the heat transfer and pressure
drop characteristics of water flow in straight and helical pipes. PhD thesis.
Japan: Okayama University; 2010.
[7] Jayakumar JS, Mahajani SM, Mandal JC, Iyer KN, Vijayan PK. CFD analysis of
single-phase flows inside helically coiled tubes. Comput Chem Engi 2010;34:
430e46.
[8] Datta AK, Hayamizu Y, Kouchi T, Nagata Y, Yamamoto K, Yanase Y. Numerical
 A.K. Datta et al. / International Journal of Thermal Sciences 120 (2017) 41e49 49

study of turbulent helical pipe flow with comparison to the experimental.
ASME J Fluid Eng 2017. http://dx.doi.org/10.1115/1.4036477.
[9] OpenFOAM. OpenFOAM 2.3.0 user guide. available at: http://www.openfoam.
com/, ; 2014.
[10] Kao HC. Torsion effect on fully developed flow in a helical pipe. J Fluid Mech
1987;184:335e56.
[11] Patankar SV, Spalding DB. A calculation procedure for heat, mass and mo-
mentum transfer in three-dimensional parabolic flows. Int J Heat Mass Transf
1972;115:1787e803.
[12] Manlapaz RL, Churchill SEW. Fully developed laminar convection from a he-
lical coil. Chem Eng Commun 1981;9:185e200.
[13] Yamamoto K, Akita T, Ikeuchi H, Kita Y. Experiment study of the flow in a
helical circular tube. Fluid Dyn Res 1995;22:153e70.
[14] Lesieur M, Metais O, Comte P. Large-eddy simulation of turbulence. UK:
Cambridge University Press; 2005.