See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/259521275

Hydrodynamically and thermally developing

laminar flow in spiral coil tubes

Article in International Journal of Thermal Sciences March 2014

DOI: 10.1016/j.ijthermalsci.2013.10.020

CITATIONS READS

8 197

2 authors, including:

zge Altun
Eskisehir Osmangazi University
17 PUBLICATIONS 32 CITATIONS

SEE PROFILE

All content following this page was uploaded by zge Altun on 26 August 2015.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
 International Journal of Thermal Sciences 77 (2014) 96e107

Contents lists available at ScienceDirect

International Journal of Thermal Sciences

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

Hydrodynamically and thermally developing laminar ow in spiral

coil tubes
Zekeriya Alta*, zge Altun
Department of Mechanical Engineering, School of Engineering and Architecture, Eskisehir Osmangazi University, Bat-Meselik 26480, Eskisehir, Turkey

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

Article history: In this study, steady state combined developing ow and heat transfer in spiral tube coils is numerically
Received 9 April 2013 investigated. The spiral coil is isothermal and the uid ow is laminar. The spiral coils with four different
Received in revised form curvature ratios of dened as the radius of the spiral at the outlet (Ro) to the radius at the inlet (Ri), Ro/
30 October 2013
Ri 2.5, 5, 7 and 17, are simulated for Pr 0.7 (air) and Pr 7 (water). The cold uid is assumed to enter
Accepted 30 October 2013
the coil from the innermost turn of the spiral tube. The three-dimensional steady-state continuity, Navier
Available online
eStokes and energy equations are solved using the commercially available CFD softwaredFluent
v6.1.22. The effects of the spiral tube pitch, curvature ratio, the Prandtl and the Dean number on the
Keywords:
Combined developing
friction factor and the heat transfer are investigated for combined developing ow. With increasing
Laminar ow Reynolds number, the heat transfer is enhanced 2e4 times over straight tubes of the same length due to
Heat transfer secondary ow and centrifugal forces. The friction losses also increase to some extent. As the number of
Friction factor turns is increased, the normalized friction factor tends to decline towards the straight tube values. Useful
Spiral coil correlations for the normalized apparent friction factor and the mean axial Nusselt number were
generated.
2013 Elsevier Masson SAS. All rights reserved.

1. Introduction experimental [6e11] and numerical [12e24] studies have been

Spiral tubes or spiral coils are widely used in various thermal
engineering applications such as heat exchangers, electronic cool-
ing, chemical reactors, etc. They have better heat transfer perfor-
mance and compactness which result in occupying less space. The
heat transfer phenomena occurring in the spiral tubes are more
complicated than those in straight ducts. Secondary ows observed
in the ow patterns and the curvature of the spiral coils facilitates
centrifugal forces which in turn signicantly affect the ow eld
and heat transfer.
In an experimental study on the viscosity of air, Grindley and
Gibson [1] were the rst who noticed the effect of curvature on the
ow through a coiled pipe. Eustice [2,3] experimentally demon-
strated the existence of the secondary ow by ink injection to
water owing in a coiled pipe. A mathematical model for the
uid ow in a curved duct of constant radius (or constant curva-
ture) was introduced by Dean [4,5]. His studies revealed that a
secondary ow develop in curved tubes when so called the Dean
number exceeds a certain critical value. Since then several
* Corresponding author. Tel.: 90 222 2393750; fax: 90 222 2393613.
E-mail addresses: zaltac@ogu.edu.tr, altacz@gmail.com (Z. Alta), okutlu@ogu.
edu.tr (. Altun).
published which examined the ow and heat transfer phenomena
in the spiral tubes.
In the literature, most of the studies on curved tubes are focused
on helically coiled tubes. Very little information is available on
horizontal spiral coiled tubes (HSCT) and HSCT heat exchangers.
Researchers have studied different vertical helical coils. Naphon
et al. [13] have reviewed the literature for three main categories of
curved tubes: vertical helical coiled tubes, spirally coiled tubes and
others curved tube arrangements. Although numerous studies on
the hydrodynamic and the heat transfer performance are available
in literature, there are very few published studies on the horizontal
spiral coils where the curvature ratio variation (D/Dc) is 0.1 and
below where Dc is the diameter of curvature. Naphon [14] experi-
mentally studied coils having coil diameter varied from 150 mm to
400 mm and the tube diameter 9.5 mm outer diameter. Coils having
curvature ratio (D/Dc) variation of 0.24e0.067 were investigated for
the hydrodynamic characteristics and compared with vertical he-
lical coil.
Ho and Wijeysundera [15,16] numerically studied a spiral coil
heat exchanger consisting of a number of horizontal layers of
spirally wound, nned tubes connected to vertical manifolds at the
inner and outermost turns of each coil. Alammar [17] studied tur-
bulent ow and heat transfer. The local skin friction coefcient, the
Nusselt number, and the wall temperature along the tube wall were

1290-0729/$ e see front matter 2013 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ijthermalsci.2013.10.020
 Z. Alta, . Altun / International Journal of Thermal Sciences 77 (2014) 96e107 97

Nomenclature x, y, z cartesian coordinate system [m]

z, s dimensionless axial length [e]
A surface area [m2]
b coil pitch [m] Greek symbol
Cf apparent friction factor [e] a thermal diffusivity [m2/s]
d tube diameter [m] 4 polar angle [rad]
D coil diameter [m] m dynamic viscosity [N s/m2]
De Dean number [e] y kinematic viscosity [m2/s]
Gz Graetz number [e] r density [kg/m3]
h heat transfer coefcient [W/m2 K] q circumferential angle [rad]
hs mean local heat transfer coefcient [W/m2 K]
k thermal conductivity [W/m K] Subscripts
L length of the (spiral) tube [m] app apparent
Nu Nusselt number [e] c curvature
P pressure [N/m2] fd fully developed straight tube
Pr Prandtl number [e] i inlet
R radius of curvature [m] o outlet
Re Reynolds number [e] s spiral coil
s axial tube length [m] st straight tube
T temperature [K] dst developing straight tube
u, v, w velocity components [m/s] w wall
V uid velocity [m/s]

presented. Kurnia et al. [18] investigated the heat transfer perfor- that the local friction factor and heat transfer varied continuously
mance of various congurations of coils of non-circular tubes. The along the tube length.
effects of the Re, Pr numbers and the coil diameter are discussed. Even though the interest in spiral coiled systems is on the rise,
Their results have been compared with those of the straight square there are very few published articles on horizontally arranged
tube of the same length. Kurnia et al. [19] have also studied three- spiral coil tubes. These articles deal with the laminar uid ows
dimensional laminar ow of a Newtonian uid in a square cross- (Re < 2100) and generally with the parabolic velocity BC at the inlet.
section ducts. Naphon [20] numerically and experimentally inves- There is very little information and/or correlations on the friction
tigated the heat transfer and ow characteristics of the horizontal factor and Nusselt numbers as well. In the absence of appropriate
spiral-coil tube. It was found that the induced centrifugal force in
the spiral-coil tube had a signicant effect on the enhancement of
heat transfer and the pressure drop increased. Yang and Chiang [21]
experimentally investigated the heat transfer for water owing
through a curved pipe with varying-curvature. The results were
compared with those of the straight pipe. It was found that the
friction coefcient for curved pipe was also increased by less than
40%.
Sasmito et al. [22] conducted a numerical study of laminar
ow heat transfer in-plane spiral ducts with rectangular, square,
triangular, trapezoidal, circular and half circular cross sections,
and compared them to straight ducts of the same cross sections
and at the same length as the coiled ducts. It was reported that
the in-plane spiral ducts have higher heat transfer performance
than straight ducts. Sasmito et al. [23] also studied laminar
nanouid ow and heat transfer enhancement in square cross
section horizontal tubes. Shih and Huang [24] studied the ther-
mal design and model analysis of Swiss-roll recuperator. The
results showed that the effectiveness of the recuperator
increased with the number of turns and number of transfer unit.
Yoo et al. [25] numerically studied horizontal spiral coils of six
turns in which radius of curvature was increased exponentially
with the polar angle. They found that the effect of Re number was
stronger than that of the curvature. Bowman and Park [26] using
Fluent v6.2 numerically investigated the pressure drop and heat
transfer characteristics of coiled systems. They have found that
up to 10% of the additional pressure drop and 40% of the
enhanced heat transfer characteristics were obtained from the
spiral coil system over the toroidal. Using the same CFD software,
the same authors [27] numerically studied laminar ow and heat
transfer performance of spiral coil tube heat exchangers with
coil-to-tube radius ratios (5e45) and spiral pitch. They showed Fig. 1. (a) Geometry and the coordinate system (b) spiral pipe conguration.
98 Z. Alta, . Altun / International Journal of Thermal Sciences 77 (2014) 96e107

correlations, traditional approach is to use correlations developed

for circular or helical tubes with an average curvature. The aim of
this study is to investigate the heat transfer enhancement and
laminar ow in the entry-region and develop useful design corre-
lations for the Nusselt number and the apparent friction factor. For
this purpose, four spiral coil congurations (Case A, B, C and D) of
four-turns with a constant pitch are studied numerically. Useful
correlations for the combined entry region were generated to aid
the researchers in this eld.

2. Model, numerical solution and validations

The spiral tube coil geometry and the coordinate system are
illustrated in Fig. 1. The radius of the coil at the inlet and outlet are
denoted by Ri and Ro, respectively. The spiral tube is described by
the polar equation R(4) Ri b4/(2p) where b is the pitch of the
coil and 4 is the polar angle. Thus, a constant pitch between any two
spiral turns is always assured; however, the radius of curvature is
not constant. The spiral coil of four turns is studied which is
determined to provide sufcient axial length to be able to examine
the combined phenomena in the entrance region. (a)
The steady-state equations for Newtonian uids are given as,
for the continuity

V$V 0 (1)
for the momentum equations

1
V$VV  VP nV2 V (2)
r
and for the energy

V$VT aV2 T (3)

The wall of the coil and the uid inlet are maintained at con-
stant temperatures T Tw and T Ti, respectively. The velocity
components at the non-slip walls are set to zero (u v w 0),
and the uid velocity at the inlet is also assumed to be
constantdu ui. Therefore, both the temperature and the ow
eld are allowed to develop within the spiral tube coil. At the
spiral duct exit, zero pressure outlet boundary condition is
employed. On the symmetry plane, we impose the symmetry BC;
i.e., vV/vq 0 and vT/vq 0.
Due to a lack of data for spiral coil congurations in a wide range (b)
of geometrical parameters, rst, the problem of combined devel-
oping ow in a straight tube was tackled to determine the grid Fig. 2. For straight tubes, the comparisons of the theoretical and computational values
sensitivity and to validate the accuracy of Nusselt and skin friction of (a) Nusselt number (b) apparent friction factor at Re 1000.

factor predictions against those of given in the literature. For the

three-dimensional tube model, the continuity, the momentum and
where n is the direction normal to the wall, k is the thermal con-
the energy equations were solved using commercially available
ductivity, As is the wall surface area up to the corresponding the
software Fluent 6.1.22 [28]. The difference scheme of the transport
axial distance z, hz is the mean local heat transfer coefcient over
terms was chosen as Second Order Upwind, and the SIMPLE algo-
the prescribed surface area up to z.
rithm was adapted. Due to the symmetry nature of the ow and
One of the most commonly used mean Nusselt number corre-
temperature elds observed in smooth curved tubes [29e36], only
lation (Pr  5) for the combined entry region is given as [37]
semi-spiral tube above the symmetry plane was used in the nu-
merical simulations. Axially (Nz 200) and circumferentially
(Nq 16) homogeneous grid spacing were used while radially
(Nr 24) clustered grids by the wall were adapted to be able to hz d 0:0688Gz
adequately capture temperature and velocity gradients. The mean Nuz 3:66 (5)
k 1 0:04Gz2=3
Nusselt number is computed over the circumference of the cross
section at any axial distance z as
where Gz is the Graetz number dened as Gz d/z Re Pr.
Z The Nusselt number, given in Eq. (5), is only a function of the
hz d 1 vT
Nuz dAs (4) Graetz number. However, for the combined entrance region, it
k As vn
As known that the Nusselt number depends on not only the Graetz
 Z. Alta, . Altun / International Journal of Thermal Sciences 77 (2014) 96e107 99

number but also the Prandtl number. For Pr  0.1, Baehr and Ste- Table 1
phan [38] derived an improved correlation in the following form The convergence of the mean Nusselt number and the friction factor with the grid
conguration for Re 1200 and Pr 0.7.

3:66 tanh Gz1 Number of nodes Number of elements Nu Cf

0:0499
tanh 2:264Gz1=3 1:7Gz2=3  Gz1
Nuz   1=6  (6) Case A
tanh 2:432 Pr Gz1 225,488 206,040 14.842 0.14675
388,865 319,160 14.834 0.14666
1,038,818 771,640 14.844 0.14671
which also accounts for the variations in Nusselt number at the 3,631,523 2,581,560 14.841 0.14675
thermal entrance region. Case B
258,816 245,440 5.850 0.10613
In Fig. 2(a), the mean Nusselt number variations for straight
415,365 352,820 5.846 0.10595
tubes are depicted comparatively. Numerical solutions for both 1,035,262 782,340 5.829 0.10595
Pr 0.7 and 7 are in perfect agreement with those of predicted with 3,502,255 2,500,420 5.829 0.10595
Eq. (6). It is also clear that the numerical solutions are signicantly Case C
122,668 116,160 12.473 0.12779
better than those obtained from the predictions of Eq. (5).
196,801 166,980 12.476 0.12788
Another parameter which is an engineering interest in tube 490,266 370,260 12.526 0.12792
ows is the friction factor. In order to determine the frictional 1,657,995 1,183,380 12.577 0.12794
losses, the skin (or Fanning) friction factor needs to be known. The Case D
Fanning friction coefcient is also referred to as friction factor in the 291,962 276,720 3.899 0.09828
527,255 438,140 3.894 0.09784
literature which is related to the Darcy friction factor by the factor
1,459,040 1,083,820 3.876 0.09781
of one-fourth in fully developed straight tube ows. For fully
developed laminar ow in straight circular tubes, the friction factor
is known to be 16/Re. In the hydrodynamic entry region, the friction
3. Results and discussions
factor is referred to as the apparent friction factor. An expression for
the mean apparent friction factor developed from theoretical
The spiral coil congurations considered in these numerical
analysis of laminar ow entrance region of straight tubes [39] is
simulations are Di/d 2, Do/d 14, b/d 1.5 (Case A), Di/d 10, Do/
given as
d 50, b/d 5 (Case B), Di/d 8, Do/d 20, b/d 1.5 (Case C) and
" Di/d 5, Do/d 85, b/d 10 (Case D). The uid velocity and the
   1:1 #0:46
0:046 temperature elds are allowed to develop within the spiral tube
C f Re 16 1 (7)
app z congurations given above.
It should be noted that the critical Reynolds number, which
where z (z/d)/Re. Another expression for the apparent friction determines the transition from laminar to turbulent ow, in curved
p
factor cited in the literature is given by Muzychka [40] as or helicoidal tubes is dened as Recrit 21001 12 d=D where
D is the coil diameter [41]. However, for spiral tubes, no single Recrit
"  # exists due to varying curvature. The minimum value of critical
  3:44 2  2 1=2 Reynolds number is obtained by using Ro in the above expression
C f Re p C f Re (8)
app z fd (as D 2Ro) which yields Recrit > 2100.
Before presenting our ndings, it is worth mentioning that in
where C f Refd 16 is for the fully developed laminar ow in the spiral coils, the radius of curvature and curvature are a function
straight tubes [37]. of the polar angle (or axial length). To develop an expression for
Using the numerical solutions for the ow eld, the apparent both quantities of interest, we consider the arc element ds which
p
mean skin friction factors up to various cross sections along the can be stated as ds R2 4 b2 =4p2 d4. Working in polar co-
axial length, z, of a straight tube is also computed as follows: ordinate system, the radius of curvature is found as
Rc 4 R2 4 b2 =4p2 3=2 =R2 4 b2 =2p2 . As 4 increases,
  ZZ
1 Rc grows larger. Thus, it is clear that beyond about a half-turn the
Cf Cf dAs (9)
app As
For Re 1000, the variation of the apparent friction factor along
the axial length z obtained from both the analytical (Eqs. (7) and
(8)) and Fluent are comparatively given in Fig. 2(b). It is
observed that the computed mean friction factor variation along
the axial length of the straight tube is consistent with the theo-
retical analysis.
In addition to the foregoing analysis on the straight tubes, the
grid sensitivity analysis for all the spiral coil congurations tackled
in this study was performed. The mean Nusselt number and the
friction coefcients evaluated with different grids are tabulated for
Re 1200 and Pr 0.7 in Table 1. The grid-independent converged
values of the parameters of interest are obtained with 388,865,
415,365, 196,801 and 527,255 nodes for Case A, B, C and D (see
Section 3), respectively. With the guidance of the foregoing vali-
dation analysis, the grid spacing, mesh type and mesh size criteria
adapted for the straight duct solutions were maintained for the
spiral coils which led to up to about 800,000e1,000,000
nodes (Fig. 3). Fig. 3. Illustration of a typical three dimensional grid conguration.
100 Z. Alta, . Altun / International Journal of Thermal Sciences 77 (2014) 96e107

Fig. 4. Velocity magnitude (left) and temperature eld (right) contours for (a) Re 200, (b) Re 750, (c) Re 1200, (d) Re 2000 of Case A and Pr 0.7.
 Z. Alta, . Altun / International Journal of Thermal Sciences 77 (2014) 96e107
Fig. 5. Temperature eld contours for (a) Re 200, (b) Re 750, (c) Re 1200, (d) Re 2000 of Case A and Pr 7.
term with pitch will be negligible for compact coils, (b/R)2/
(4p2) << 1; thus, for the radius of curvature, we can justify writing
Rc y R(4) and dRc bd4/(2p). In view of this simplication,
ds y R(4)d4 becomes ds y 2pRcdRc/b where Ri  Rc  Ro. Then, for a
coil of several or more turns, the arc length in terms of the radius of
curvature could be written as sRc yp=bR2c  R2i . In this study,
we will refer to Ro/Ri ratio as the curvature ratio of a spiral coil.
3.1. Flow and temperature eld developments
Due to the curvature and no slip boundary condition at the spiral
walls, the development of axial velocity is affected by the axial and
secondary ows. The numerical simulation results indicate that the
axial ow pattern is different for low and high Reynolds numbers
depending on the spiral tube conguration; thereby affecting the
temperature eld.
In this study, extensive discussions on the ow eld were not
intended since there has been a plenty of published articles on the
ow eld of curved and helical tubes. Our discussions will be strictly
concerned with the combined entry ow and axial variations of the
ow eld combined with the research ndings in this subject.
The velocity magnitude,jVj=uavg , and the dimensionless tem-
perature distribution, Q (T  Ti)/(Tw  Ti), along z 0 plane
(symmetry plane) for Pr 0.7 and Case A (Ro/Ri 7) are depicted in
Fig. 4 for various Reynolds numbers. This is an extreme case in
which the relative variation in the radius of curvature of the
innermost and outermost walls of the spiral at the entrance is
signicant. The uid particles along the center of curvature have
larger centrifugal force than other regions at the inlet, and this force
pushes the uid away from the center of curvature. This motion
results in maximum velocity to occur near the inner wall of the
spiral tube at very close to the entrance. However, after a short
distance into the tube, the maximum velocity shifts towards the
101
outer wall of the spiral due to acting centrifugal forces and
decreasing curvature. It is clear in the velocity contours of Fig. 4 that
uid further downstream, accumulated uid near the innermost
wall discharges towards the outer wall of the spiral tube. This
phenomenon can be explained by the viscous, inertia and centrif-
ugal forces developing at the entrance region. Due to large viscous
forces, centrifugal forces become dominant at the entrance causing
the uid accumulation near the innermost wall of the spiral tube
where the maximum velocity is observed. As the Reynolds number
increases, the amount of centrifugal forces near the innermost wall
decreases while centrifugal forces increase near the outermost
wall. This discharge process is completed at about 130e180 from
the inlet of the spiral. Due to the secondary ow caused by the
nature of the ow and the curvature changes, the heat transfer
process from the walls are also affected which results in not only
axial but also circumferential boundary layer formations. Well into
the spiral tube, the uid is heated more in regions where the uid is
relatively motionless causing larger thermal boundary layers at the
inner side of the spiral walls while very thin thermal boundary
layers are observed at the outer side of the spiral wall. The form of
hydrodynamic boundary layer is established within the rst turn
indicating relatively short hydrodynamic entrance length. On the
other hand, the thermal boundary grows from the inner towards
the outer wall of the spiral pipe wall. As expected for low Pr uids,
the thermal entrance length is shorter. In Fig. 4, it is observed that
the thermal entry lengths are about 1, 2, 23/4 and 3 coil for
Reynolds numbers of 200, 750, 1200 and 2000, respectively.
The dimensionless temperature elds along z 0 plane (sym-
metry plane) for Pr 7 of Case A are depicted in Fig. 5 for various
Reynolds numbers. The ow eld is the same; however, due to uid
having ten times larger Pr value, the thermal entrance length is
longer than that of Pr 0.7. As it can be observed from Fig. 5(a), the
thermal entrance length for Re 200 is completed at about 3.5e4
turn. On the other hand, as the Reynolds number is increased above
102 Z. Alta, . Altun / International Journal of Thermal Sciences 77 (2014) 96e107

Fig. 6. Velocity magnitude (left) and temperature eld (right) contours for (a) Re 200, (b) Re 750, (c) Re 1200, (d) Re 2000 of Case C and Pr 0.7.
 Z. Alta, . Altun / International Journal of Thermal Sciences 77 (2014) 96e107
500, thermally fully development is not observed. The axial tem-
perature eld (thermal boundary layer) remains more less constant
which in turn this is desirable since it increases the heat transfer
rate.
The velocity magnitude,jVj=uavg , and the dimensionless tem-
perature distribution, Q (T  Ti)/(Tw  Ti), along z 0 plane
(symmetry plane) for Pr 0.7 and Case C (Ro/Ri 2.5) are depicted
in Fig. 6 for various Reynolds numbers. This is a case in which the
radius of curvature of the innermost and outermost walls of the
spiral is more less the same, and the pitch is the same as Case A. The
maximum velocity at the entrance region is observed to be at the
center of curvature of the spiral tube; however, further down-
stream, the maximum velocity shifts towards the outer wall of the
spiral pipe. This phenomenon is due to smaller curvature ratio (or
larger radius of curvature, Ri) at the inlet. Thus, the magnitudes of
the centrifugal forces acting on both sides of the tube axis are
comparable. As the uid discharges towards the outer side wall of
the spiral tube, this process is completed at about 50e80 rotation
from the inlet. The hydrodynamic boundary layer is established at
about 90 rotation indicating the hydrodynamic entrance length is
relatively short for Re 200 (Fig. 6a). The thermal boundary, as in
Case A, grows from the inner towards the outer wall of the spiral
walls under the inuence of the velocity eld. In Fig. 6, it is
observed that the thermal entry lengths are about 5/8, 9/8, 1 and
13/4 coil for Reynolds numbers of 200, 750, 1200 and 2000,
respectively.
The dimensionless temperature distribution along z 0 plane
(symmetry plane) for Pr 7 and Case C are depicted in Fig. 7 for
various Reynolds numbers. Because of larger Pr value, the thermal
entrance length extends over longer section of the spiral. The
thermal boundary layer at the innermost wall is thicker than the
outermost wall which is consistent with the hydrodynamic
boundary layer.
Fig. 7. Temperature eld contours for (a) Re 200, (b) Re 750, (c) Re 1200, (d) Re 2000 of Case C and Pr 7.
103
3.2. Heat transfer aspects
In order to examine the heat transfer enhancement aspect of
the spiral coil ow, the mean Nusselt numbers up to various
axial locations (s) along the spiral tube are evaluated. The spiral
coil is divided into minimum 32 axial locations, especially
condensed in the rst 90 turn. The mean Nusselt number, Nus , is
evaluated at these axial locations with Eq. (4). The mean Nusselt
number variations for each case are plotted against 1/Gz to be
consistent with the heat transfer analysis of straight tubes. For
straight tubes and for large values of 1/Gz, the mean Nusselt
number is independent of Prandtl number. However, for the
combined entrance region, the Nusselt number depends on the
developing ow eld which is highly sensitive to the uid vis-
cosity. Hence, the heat transfer results depend on the Prandtl
number for the combined entry length cases as shown for the
straight tubes in Fig. 2.
For all spiral coil cases studied, the axial variation of the mean
Nusselt number with respect to 1/Gzdwhich Graetz number is
dened as Gz (d/s)RePr where s is the axial length of spiral
tubedis depicted in Fig. 8. In Case A and Case C, in Fig. 8(a and
c), for Pr 0.7 and 7, the mean Nusselt number sharply de-
creases at the entry and then uctuates as it increases before
stabilizing. This observed phenomenon is different than that of
straight tubes, and these oscillations are caused by the ow
pattern which is affected by the centrifugal and viscous forces
inducing thermal boundary layer formations at the entrance as
discussed in the previous section. It is more pronounced in Case
A and C since the radius of curvature is smaller than the other
cases, and this relatively small radius of curvature is maintained
throughout the rst two turns due to small pitch (b/d 1.5).
These oscillations are not observed in Case B and Case D (in
Fig. 8(c and d)); however, for Pr 0.7 and Re > 750 of Case B, the
104 Z. Alta, . Altun / International Journal of Thermal Sciences 77 (2014) 96e107

Fig. 8. The axial variation of the mean Nusselt number for (a) Case A, (b) Case B, (c) Case C and (d) Case D.

mean Nusselt number increases before stabilizing as seen in  q q

 0:10 exp 0:70 Pr  0:1528 Gz
Fig. 8(b). This is caused by short thermal entry length of low Pr Nu Rc Gz Pr
0:77 0:07De0:66
c  1=4 q
uids, but for Pr 7 the mean Nusselt value steadily decreases. Nust Ri d Gz
1  0:031
Due to increased radius of curvature at the inlet region, the ow 2Rc Pr
eld is a bit more stable. For Case D where the radius of cur- (10)
vature increases signicantly with each turn, the ow stabilizes
in relatively short distance, thus oscillations are not so signi- where Rc is the radius of the curvature along the axial path of the
p
cant. We also observe that the mean Nusselt number along the spiral tube, Ri is the coil radius at the inlet, Dec Re d=2Rc is the
ow axis for specic Pr number tends to become linear in a loge curvature dependent Dean number, b/d is the pitch to tube diam-
log graph in the entrance (very small 1/Gz values) of the tube. eter ratio. The estimated variance and the correlation coefcient of
For increasing Reynolds numbers, the mean Nusselt number this correlation are 0.0167 and 0.998, respectively.
branch out from this linear behavior. In Fig. 9, for all cases, the normalized Nusselt number and
The mean Nusselt numbers up to predetermined axial dis- the tteddEq. (10)dvariation with 1/Gz are illustrated. The
tances of the spiral tube were computed for all cases considered computed values are represented by symbols while the tted
in the study. A normalized mean Nusselt number dened as equations are depicted with a straight curve (for Pr 0.7) or
the ratio of the mean Nusselt number of spiral tube Nu to dashed curves (for Pr 7). It is observed that at the entrance for
that of the straight tube Nust dEq. (6)dis computed. Total of very large Graetz values, the normalized value approaches unity
1185 normalized Nusselt number ratio data were tted using which implies that the mean Nusselt variation is identical to that
non-linear regression, and the following correlation for com- of straight tubes. Some oscillations are observed as the uid
bined entry ow was obtained for 1.5  b/d  10, 1  Ri/d  5, progresses inside the duct as seen in Fig. 9(a) due to ow pat-
7  Ro/d  45, 2.5  Ro/Ri17, 200  Re  2000 and terns formed in very small curvature ratio. Overall a fairly good
0.70  Pr  7: t is observed.
 Z. Alta, . Altun / International Journal of Thermal Sciences 77 (2014) 96e107
(a)
(b)
Fig. 9. The normalized Nusselt number variation along the ow axis for (a) Case A, (b) Case B, (c) Case C and (d) Case D (computed values indicated by symbols, for Pr 0.7 straight
and for Pr 7 dashed lines are obtained from Eq. (10)).
3.3. Skin friction factor
In curved tubes, circumferential and axial hydrodynamic
boundary layers develop due to secondary ows caused by the
centrifugal forces; thus, the friction factor as a result of the curva-
ture is higher than that of the straight pipe. The mean friction
factors, C f , are computed with Eq. (9) at 32 axial locations as
indicated in Section 3.2. The friction factor in curved tubes such as
torus and helically wound tubes are often presented as a normal-
ized friction factorda ratio of the curved to that of the straight tube.
The published correlations for fully developed ow, this ratio is
generally stated a function of Dean number and/or the tube
diameter to coil diameter ratio (d/D). However, in combined entry
ow, it is expected this ratio to be a function of dimensionless axial
distance s as well.
A normalized factor dened as the ratio of the apparent mean
friction factor of spiral C f to that of the straight tube C f ;st dEq.
(8)dis considered for the regression and for comparison purposes.
The normalized data were tted using non-linear regression, and
the following correlation for combined entry ow was obtained:
! 
Cf s Rc 0:11 0:016b=d
0:644 4:45De0:50
c
e (11)
C f;st 50s 1 Ri
105
(c)
(d)
The estimated variance and the correlation coefcient of this
correlation are 0.0024 and 0.9991, respectively. The normalized
expression is a function of radius of curvature (Rc) which is also a
function of s, variable curvature ratio dened as the radius of
curvature at the inlet (Rc/Ri), pitch to tube diameter (b/d)
and the dimensionless axial distance (s). In this correlation, it
is clear that for large values of b/d, the magnitude of exp[0.016(b/
d)] term is not in the order of unity. On the other hand, the
dimensionless tube length and the Dean number p can be approxi-
mated as s ypR2c  R2i =bdRe and Dec Re= 2=Ri =d2 Re=p
b=ds 1=4 . Thus, it is clear that the radius of curvature and the
Dean number are functions of Ri/d and b/d.
In Fig. 10, the normalized skin friction factor is illustrated for all
cases. In Case A and C where the radius of curvature is relatively
small at the tube entrance, the normalized friction factor increases
about 2.5 times or more for Re 2000 (Fig. 9(a and c)). For Case B
and D where the tube length is longer than the other two cases, a
maximum is observed (Fig. 9(b and d)). Thus the numerical pre-
dictions reveal that if the spiral tube length is sufcient enough,
then the normalized friction factor increases to a maximum and
then decreases. In Case A and C, the maximum can be observed by
adding more turns to the present congurations. To nd the length
which the maximum occur, we seek the maximum of Eq. (11). One
can easily obtain an analytical expression as follows:
106 Z. Alta, . Altun / International Journal of Thermal Sciences 77 (2014) 96e107

(a) (c)

(b) (d)
Fig. 10. The normalized friction factor along the ow axis for (a) Case A, (b) Case B, (c) Case C and (d) Case D (computed values indicated by symbols, for Pr 0.7 straight and for
Pr 7 dashed lines are obtained from Eq. (11)).

spiral coils of constant pitch, the design parameters can be

s identied as Ri/d, b/d and Re. This analysis reveals that if the
93 173 2p Ri =d2 1
s (12)
700 9800 7 b=d Re
It is clear that for Ri =d2 =b=d=Re  1 the location of
maximum will be s x 0.265 which can be attained by large (b/
d)Re values. The pitch of the coil clearly inuences the apparent
normalized friction factor. In Case A, the maximum occurs at
0.267 for all Re values since the geometry dependent terms in
Eq. (12) are very small. In Case B, the maximum occurs at 0.333
and 0.273 for Re 200 and 2000, respectively since the geom-
etry dependent term becomes larger. In Case C, the maximum is
located at 0.388 and 0.28 for Re 200 and 2000, respectively.
This result is interesting in that Case A and Case C have the same
pitch to diameter ratio. In Case D, which has the longest length,
the maximum feature is observed in all Reynolds values and the
maximum is located at 0.276 and 0.266 for Re 200 and 2000,
respectively. These observations are consistent with the nu-
merical solutions as seen in Fig. 10. It can be deduced that the
spiral tube length is sufcient enough (a case with more turns),
the normalized friction factor will decrease after having a
maximum.
4. Conclusions
In this study, the convection heat transfer and the laminar
friction losses in combined entry region of horizontal spiral tube
coils are investigated numerically. The centrifugal force and the
secondary ow have signicant effects on the enhancement of heat
transfer and the friction losses. The ndings of this study can be
summarized as follows:
1) In comparison to the straight tubes, the combined heat transfer
is a function of Prandtl and Dean numbers, the variable curva-
ture and Graetz number. Unlike the straight tube ows, a single
asymptotic value is not obtained.
 Z. Alta, . Altun / International Journal of Thermal Sciences 77 (2014) 96e107
2) The mean Nusselt number obeys the straight tube ow at the
very entrance of the tube, but then branches out to stabilize at
various values for decreasing Graetz number.
3) The friction factor also increases with the spiral tube length;
however, for spiral coils of more than four turns or spiral coils of
sufcient length (s > 0.25), the normalized friction factor de-
creases. The normalized friction factor increases by as much as
2.5 times for coils of small radius of curvatures.
4) Correlations for the mean Nusselt number and the friction fac-
tors for a wide range of design parameters are obtained for the
combined entry ows in spiral coils.
References
[1] J.H. Grindley, A.H. Gibson, On the frictional resistance to the ow of air
through a pipe, Proc. R. Soc. A 80 (1908) 114e139.
[2] J. Eustice, Flow of water in curved pipes, Proc. R. Soc. A 84 (1910) 107e118.
[3] J. Eustice, Experiments of streamline motion in curved pipes, Proc. R. Soc. A 85
(1911) 119e131.
[4] W.R. Dean, Note on the motion of uid in a curved Pipe, Philos. Mag. Ser. 4
(20) (1927) 208e223.
[5] W.R. Dean, The stream-line motion of uid in a curved pipe, Philos. Mag. Ser. 5
(1928) 673e695.
[6] P. Naphon, J. Suwagrai, Effect of curvature ratios on the heat transfer and ow
developments in the horizontal spirally coiled tubes, Int. J. Heat Mass Transfer
50 (2007) 444e451.
[7] S. Ramachandran, P. Kalaichelvi, S. Sundaram, Heat transfer studies in a spiral
plate heat exchanger for water-palm oil two phase systems, Braz. J. Chem.
Eng. 25 (2008) 483e490.
[8] R. Balakrishnan, J.S. Santhappan, M.L. Dhasan, Heat transfer correlation for a
refrigerant mixture in a vertical helical coil evaporator, Therm. Sci. 13 (2009)
197e206.
[9] R. Rajavel, K. Saravanan, Heat transfer studies on spiral plate heat exchanger,
Therm. Sci. 12 (2008) 85e90.
[10] R. Vijayan, P. Srinivasan, Experimental evaluation of internal heat exchanger
inuence on R-22 window air conditioner retrotted with R-407C, Therm. Sci.
14 (2010) 39e47.
[11] M.K. Khan, R. Kumar, P.K. Sahoo, Experimental investigation on diabatic ow
of R-134a through spiral capillary tube, Int. J. Refrig. 32 (2009) 261e271.
[12] A. Nakayama, N. Kokubo, T. Ishida, F. Kuwahara, Conjugate numerical model
for cooling a uid owing through a spiral coil immersed in a chilled water
container, Numer. Heat Transfer Part A 37 (2000) 155e165.
[13] P. Naphon, S. Wongwises, A review of ow and heat transfer characteristics in
curved tubes, Renew. Sustain. Energy Rev. 10 (2006) 463e490.
[14] P. Naphon, S. Wongwises, An experimental study on the in-tube convective
heat transfer coefcients in a spiral coil heat exchanger, Int. Commun. Heat
Mass Transfer 29 (2002) 797e809.
[15] J.C. Ho, N.E. Wijeysundera, Study of compact spiral-coil cooling and dehu-
midifying heat exchanger, Appl. Therm. Eng. 16 (1996) 777e790.
[16] J.C. Ho, N.E. Wijeysundera, An unmixed-air ow model of a spiral coil cooling
dehumidifying unit, Appl. Therm. Eng. 19 (1999) 865e883.
[17] K.N. Alammar, Turbulent ow and heat transfer characteristics in U-tubes: a
numerical study, Therm. Sci. 13 (2009) 175e181.
[18] J.C. Kurnia, A.P. Sasmito, A.S. Mujumdar, Evaluation of the heat transfer per-
formance of helical coils of non-circular tubes, J. Zhejiang Univ. Sci. A 12
(2011) 63e70.
View publication stats
107
[19] J.C. Kurnia, A.P. Sasmito, A.S. Mujumdar, Numerical investigation of laminar
heat transfer performance of various cooling channel designs, Appl. Therm.
Eng. 31 (2011) 1293e1304.
[20] P. Naphon, Study on the heat transfer and ow characteristics in a spiral-coil
tube, Int. Commun. Heat Mass Transfer 38 (2011) 69e74.
[21] R. Yang, F.P. Chiang, An experimental heat transfer study for periodically
varying-curvature curved-pipe, Int. J. Heat Mass Transfer 45 (2002) 3199e
3204.
[22] A.P. Sasmito, J.C. Kurnia, W. Wang, V. Jangam, A.S. Mujumdar, Numerical
analysis of laminar heat transfer performance of in-plane spiral ducts with
various cross-sections at xed cross-section area, Int. J. Heat Mass Transfer 55
(2012) 5882e5890.
[23] A.P. Sasmito, J.C. Kurnia, A.S. Mujumdar, Numerical evaluation of laminar heat
transfer enhancement in nanouid ow in coiled square tubes, Nanoscale Res.
Lett. 6 (1) (2011) 376e385.
[24] H.Y. Shih, Y.C. Huang, Thermal design and model analysis of the Swiss-roll
recuperator for an innovative micro gas turbine, Appl. Therm. Eng. 29
(2009) 1493e1499.
[25] G. Yoo, H. Choi, W. Dong, Fluid ow and heat transfer of spiral coiled tube:
effect of Reynolds number and curvature ratio, J. Cent. South Univ. 19 (2012)
471e476.
[26] A.J. Bowman, H. Park, CFD study on laminar ow pressure drop and heat
transfer characteristics in toroidal and spiral coil system, in: ASME 2004 In-
ternational Mechanical Engineering Congress and Exposition, Anaheim, Cali-
fornia, November 13e19, 2004. IMECE2004-59879.
[27] A.J. Bowman, H. Park, CFD investigation of developing and redeveloping
laminar ow and heat transfer characteristics in coiled tube systems for
constant wall temperature heating and cooling: part IIdspiral congurations,
in: ASME 2006 International Mechanical Engineering Congress and Exposi-
tion, Heat Transfer, vol. 1, 2006. Chicago, November 5e10, IMECE2006-15572.
[28] Fluent, FLUENT 6.1 Users Guide, FLUENT, Inc., Lebanon: NH, 2003.
[29] S.V. Patankar, V.S. Pratap, D.B. Spalding, Prediction of turbulent ow in pipes,
J. Fluid Mech. 67 (1970) 583e595.
[30] D.J. McConalogue, R.S. Srivastava, Motion of a uid in a curved Tube, Proc. R.
Soc. A 307 (1968) 37e53.
[31] H.G. Sharma, A. Prakash, Flow of a second order uid in a curved pipe, Indian J.
Pure Appl. Math. 8 (1977) 546e557.
[32] Z. Zapryanov, C. Christov, E. Toshev, Fully developed laminar ow and heat
transfer in curved tubes, Int. J. Heat Mass Transfer 23 (1980) 873e880.
[33] Z. Yang, H.B. Keller, Multiple laminar ows through curved pipes, Appl.
Numer. Math. 2 (1986) 257e271.
[34] R.J. Silva, R.M. Valle, M. Ziviani, Numerical hydrodynamic ad thermal analysis
of laminar ow in curved elliptic and rectangular ducts, Int. J. Therm. Sci. 38
(1999) 585e594.
[35] M.R.H. Nobari, B.R. Ahrabi, G. Akbari, A numerical analysis of developing ow
and heat transfer in a curved annular pipe, Int. J. Therm. Sci. 48 (2009) 1542e
1551.
[36] E. Amani, M.R.H. Nobari, A numerical investigation of entropy generation in
the entrance region of curved pipes at constant wall temperature, Energy 36
(2011) 4909e4918.
[37] T.L. Bergman, A.S. Lavine, F.P. Incropera, D.P. Dewitt, Fundamentals of Heat
and Mass Transfer, seventh ed., John Wiley & Sons, 2011.
[38] D.H. Baehr, K. Stephan, Heat and Mass Transfer, revised second ed., Springer-
Verlag, Berlin, 2006.
[39] J.P. Plessis, M.R. Collins, A new denition for laminar ow entrance lengths of
straight ducts, J. South Afr. Inst. Mech. Eng. R&D 8 (1992) 11e16.
[40] Y.S. Muzychka, Analytical and Experimental Study of Fluid Friction and Heat
Transfer in Low Reynolds Number Flow Heat Exchangers (Ph.D. Thesis),
University of Waterloo, Canada, 1999, http://www.engr.mun.ca/muzychka/
Muzychka__Thesis.pdf.
[41] W.M. Rohsenow, J.P. Hartnett, Y.I. Cho, Handbook of Heat Transfer, third ed.,
McGraw Hill, New York, 1998 (Chapter 5).