International Journal of Thermal Sciences 134 (2018) 40–53

Contents lists available at ScienceDirect

International Journal of Thermal Sciences

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

Natural convection heat transfer from a horizontal hollow cylinder with T

internal longitudinal fins
Swastik Acharya∗, Sukanta K. Dash
Department of Mechanical Engineering, Indian Institute of Technology Kharagpur, 721 302, India

A R T I C LE I N FO A B S T R A C T

Keywords: Three dimensional numerical simulations have been carried out by solving the continuity, momentum and the
Natural convection energy equations to predict the flow and the temperature field around an internally finned horizontal cylinder in
Internally finned cylinder natural convection. The fin height, fin spacing or the fin number and the length of the cylinder were varied at
Conjugate heat transfer different Ra to predict the heat loss from the finned cylinder and interesting findings were obtained. The heat
Correlation for Nu
loss from a short cylinder (L/D < 1) can be maximum for certain height of the fin and certain fin number after
which the heat loss can reduce if the fin height increases or the fin spacing decreases in the laminar range of Ra
to within 107. When the length of the cylinder increases (L/D > 1) the maximum point of heat loss vanishes for
all heights of the fins and fin spacing. A table has been generated to show the optimum configuration of fin
height, fin number where maximum heat loss can take place. The average Nu of the finned cylinder was seen to
be decreasing with H/D, L/D and increasing with Ra. From the enormous numerical simulations that has been
done in the present study a general correlation for the Nu as a function of L/D, H/D, fin number, N and Ra has
been developed to within an accuracy of ± 6% which can be beneficial to the industry and practical designers.

1. Introduction
Natural convection heat transfer plays a significant role in cooling of
electronic components [1], hot solid or hollow cylinders lying in in-
dustrial bay for cooling, compact heat exchanger design [2,3], cooling
of nuclear reactor [4,5] in case of sudden power failure or accidents.
Natural convection is preferred over forced convection as it requires no
external agent for fluid motion around the hot body for cooling and
hence helps improving the reliability of the components in a system in
an efficient and cost effective manner with additional advantages of
having less noise and vibration. Since the heat transfer coefficient in
natural convection is quite low, the enhancement of heat transfer in
natural convection has been a challenging topic for most of the re-
searchers in the present days. The most promising solution for enhan-
cing the heat transfer is by employing fins (extended part) on the heated
surface which provides the additional surface area for a better heat
transfer. While many studies have been performed on natural convec-
tion heat transfer from cylinder with external fins [6–9] however, a few
literatures are available on natural convection heat transfer from cy-
linders with internal fins and almost no literature is available on a
horizontal hollow cylinder with internal longitudinal fins. Internal fins
find its application in many practical field such as electrical heating of
cylinders, nuclear heating, heat exchanger devices [10], cooling of
annular nuclear reactor in sudden power failure or at the time of closing
the power plant, ventilators and radiators [11], annular combustion
chamber, hollow electrical conductor of cylindrical shape carrying huge
current, where the inner surface of the cylinder is to lose most of the
heat. When a heat source is present on the outer surface of the cylinder,
it is difficult to attach fins on the outer surface, so the employment of
internal fins becomes essential. Since the inner surface of a heated
horizontal hollow cylinder losses more heat than that of the outer
surface, (when L/D is less than 1 as has been explained by the present
authors in their previous work [12,13]), so it would be prudent to use
internal fins instead of external fins for a cylinder with aspect ratio less
than 1. Many researches have been conducted on forced convection
from an internally finned cylinder [14–17] but a few are only available
regarding natural convection heat transfer from an internally finned
cylinder. One of the earliest in this genre was the work done by Prakash
and Liu [18] who investigated the buoyancy induced flow in a verti-
cally internally finned circular duct and reported that there is sig-
nificant improvements of thermal performance of a finned tube over an
un-finned tube. Joo and Kim [19] experimentally and numerically in-
vestigated the natural convection heat transfer from internally finned
vertical tube and optimized the fin configuration for maximum heat
transfer rate. They optimized fin height, fin number and fin thickness
under a given diameter and length of the tube and proposed a

∗
Corresponding author.
E-mail address: swastik.acharya8@gmail.com (S. Acharya).

https://doi.org/10.1016/j.ijthermalsci.2018.07.039
Received 10 October 2017; Received in revised form 10 June 2018; Accepted 29 July 2018
1290-0729/ © 2018 Elsevier Masson SAS. All rights reserved.
S. Acharya, S.K. Dash International Journal of Thermal Sciences 134 (2018) 40–53

Nomenclature Qmax maximum heat transfer, W

T fluid temperature, K
At total area of convective surface, m2 Tf local fin surface temperature, K
Ac cylinder surface area, m2 Tw cylinder surface temperature, K
Af fin surface area, m2 T∞ ambient temperature, K
D diameter of horizontal cylinder, m t time, sec
g acceleration due to gravity, m/s2 U flow velocity, m/sec
H fin height, m x Cartesian coordinate
h average convective heat transfer coefficient, W/m2 -K Ra Rayleigh number based on diameter
hc average convective heat transfer coefficient of the cylinder H/D fin height to diameter ratio
wall, W/m2 -K L/D length to diameter ratio
hf local convective heat transfer coefficient of the fin surface, s/D fin spacing to diameter ratio
W/m2 -K
hf avg average convective heat transfer coefficient of the fin Greek symbols
surface, W/m2 -K
kf thermal conductivity of fluid, W/m-K α thermal diffusivity, m/s2
ks thermal conductivity of solid (fin), W/m-K β thermal expansion coefficient, 1/K
L length of the cylinder, m ε effectiveness of the finned cylinder
Nu average surface Nusselt number of the finned cylinder ƞ efficiency of the finned cylinder
based on diameter μ dynamic viscosity, kg/m-s
N number of fins ν kinematic viscosity, m/s2
Pr Prandtl number ρ density of fluid, kg/m3
p pressure, N/m2
Q total heat transfer rate, W

correlation for optimum efficiency as a function of D/L and Rayleigh
number based on length of the cylinder. Myhren and Holmberg [11]
analyzed the thermal performance of the ventilators radiator with in-
ternal convection fins and reported a significant improvement in ef-
fectiveness of the ventilation radiator which would save the electrical
energy in a notable amount. Soliman et al. [20] analyzed laminar heat
transfer in internally finned tube with uniform outside wall tempera-
ture. They presented the distribution of temperature in fins and fluid
and concluded these were strongly dependent on fin configurations.
They highlighted some important applications for the justification of
their boundary condition applied to their problem and interestingly this
boundary condition suites the present study too.
After a thorough literature survey, it has been found that almost no
studies have been conducted on natural convection heat transfer from a
horizontal internally finned cylinder where heat transfer occurs from
both the inner and outer surfaces. Variation of the heat transfer char-
acteristics such as Nu, effectiveness and efficiency of the internally
finned horizontal cylinder as a function of various fin configurations,
aspect ratio and Rayleigh number have been studied thoroughly with
the help of vector plots and temperature contours in the present work
which has not been reported in any literature available till now, which
reflects the novelty of the present study. The present study deals with a
detailed analysis of the natural convection heat transfer from an iso-
thermal hollow horizontal internally finned cylinder. Numerical simu-
lations of full Navier - Stokes equations along with the energy equation
have been performed in a laminar range of Ra spanning from 104 to 108
to analyze the effect of fin height, fin number and cylinder length on the
heat transfer from the cylinder. 3-D heat conduction equation is solved
in the fin material taking conjugate heat transfer into account while
solving the temperature field in the fluid medium, air. Thickness of the
cylinder is set to be zero since it is too less compared to the height of the
fin and the length of the cylinder. Two independent parameters of fin
configuration such as fin number and fin height are considered in a
wide range to obtain an exploded effect on the heat transfer while the
fin thickness to cylinder diameter was kept constant at 0.01262 for all
the cases. Highly conductive aluminum fins were employed to obtain
maximum fin efficiency. There are two competitive opposing mechan-
isms which affect the thermal performance of the cylinder, namely the
increase in heat transfer due to the provision of additional surface area
with a simultaneous decrease in heat transfer due to the obstruction of

Fig. 1. Schematic representation of a thin hollow horizontal cylinder with internal fins and the computational domain, (a) lateral cross section, (b) longitudinal cross
section.

41
S. Acharya, S.K. Dash
Fig. 2. Influence of domain height on average Nu for an internal finned cy-
linder.
the air flow in presence of fins. Therefore, one can expect an optimal
configuration of fin for the maximum heat transfer as a consequence of
the interplay of the two competitive parameters which needs to be
figured out in this study. In the present study, a reliable analysis is
performed in an adequate manner to find the optimum fin configura-
tions for maximum heat transfer for different aspect ratios of the cy-
linder and Rayleigh numbers, which has not been reported in the lit-
erature so far.
2. Physical description of the problem
Fig. 1(a) shows the lateral cross section and 1 (b) shows the long-
itudinal cross section of an internally finned isothermal horizontal cy-
linder placed in air with a rectangular domain around it. Fins are em-
ployed longitudinally at different radial locations of the cylinder. Fins
are placed in such a way that the symmetry in the longitudinal plane
would be preserved so that one can consider the symmetry of the whole
domain for the computation to reduce the computational time. The
thickness of the fin to cylinder diameter ratio is set to be 0.01262 for all
the simulations. The temperature difference between the surroundings
and the cylinder wall has been taken to be 26 K and the properties of air
surrounding the finned cylinder have been calculated at the film tem-
perature. The pertinent parameters for heat transfer like the fin height,
fin spacing and the cylinder length are varied in a wide range with
respect to the cylinder diameter D. The fin height used in the present
study has been taken in the range of D/12 to D/3. A uniform fin spacing
has been maintained for all the cases with fin spacing to diameter ratio
Fig. 3. Cell arrangement in the hollow cylinder with fins, (a) longitudinal cross section, (b) lateral cross section, (c) Blown up view near the cylinder and fin.
42
International Journal of Thermal Sciences 134 (2018) 40–53
Fig. 4. Average Nu for an internally finned horizontal cylinder as a function of
number of cells.
varied from π/4 to π/40. The main objective of the present study is to
figure out the optimum fin configurations for maximum heat transfer
for different aspect ratios of the cylinder and Rayleigh number. In order
to solve the problem 3-D full Navier-stokes equations along with the
energy equation have to be considered in the computational domain
and a 3-D conduction equation has to be solved in the fin geometry. At
the fin wall - air interface, heat flux and temperature continuity have to
be satisfied. Flow field and thermal field has to be found around the
cylinder, so that Nusselt number of the finned cylinder could be cal-
culated. The size of domain would influence the flow field and the re-
sult to some extent. Hence a domain independent and a grid in-
dependent test have to be performed to get a domain and grid
independent solution. A rectangular domain has to be present around
the cylinder, all six faces of which would serve as pressure outlet
boundary condition so that the flow can either come into the domain or
leave the domain.
3. Mathematical formulation:
The Continuity, Navier- Stokes and the energy equations are re-
quired for the solution of the flow and temperature field around the
finned cylinder. A 3-D conduction equation is required to find tem-
perature distribution inside the fin. Heat flux and temperature con-
tinuity at the interface is required to solve the conjugate heat transfer at
the solid - fluid interface. We write the relevant equations here in
tensorial form in a rectangular coordinate system, since it is the best
compact way of representing them
S. Acharya, S.K. Dash International Journal of Thermal Sciences 134 (2018) 40–53

Table 1
A comparison of the present simulation with the experimental data of Joo and Kim [19].
Case Tube diameter Tube length Fin thickness(m) Fin height Fin number Heat transfer rate (W) (Joo Heat transfer rate (W) (Present Variation (in %)
(m) (m) (m) (N) and Kim) simulation)

1 0.050 0.060 0.0003 0.020 12 6.73 6.1396 8.77

2 0.150 0.060 0.0008 0.055 36 42.2 43.7290 3.62
3 0.150 0.5 0.0008 0.057 22 144.4 138.4386 4.13

Fig. 5. Internally finned vertical cylinder used in the experimental analysis of Joo and Kim [19].

∂ −kf
∂Tf
Continuity : (Ui ) = 0 Qc
∂x i (1) hc =
Ac (Tw − T∞)
, ∫ hf dAf = ∫ (Tf − ∂Tn∞) dAf
D (Ui ) ∂p ∂ ⎡ ⎛ ∂Ui ∂Uj ⎞ ⎤ hence
Momentum :ρ = − +μ ⎜ + ⎟ − ρβgi (T − T∞)
Dt ∂x i ∂x j ⎢ ⎝ ∂x j ∂x i ⎠ ⎥ ∂Tf
⎣ ⎦ ⎛ Qc −kf ⎞
1 hD
(2) h=
(Ac + Af ) ⎜⎜ Tw − T∞
+ ∫ (Tf − ∂Tn∞) dAf ⎟⎟ and Nu =
kf
⎝ ⎠ (6)
D (T ) μ ∂ ⎛ ∂T ⎞
Energy: ρ = ⎜ ⎟
So it can now be seen from Eqn. (6) that our computed value of Nu
Dt Pr ∂x i ⎝ ∂x i ⎠ (3)
is proportional to the net heat loss from the finned cylinder and in-
It is assumed that the fluid around the cylinder is incompressible, so versely proportional to the net area of heat loss. Therefore, as per the
Boussinesq approximation [21] has been taken into account which definition, Nu has become proportional to the average heat flux from
would compute the buoyancy in the source term of the momentum the finned cylinder. So, a prescription of h is not needed for the com-
equation. One could have used ideal gas equation to take care of the putation of Q or Nu in our work.
density variation with respect to temperature, in that case the
β (T − T∞) would have been dropped. All the thermo physical proper- 5. Boundary conditions
ties along with the density are used at the mean film temperature and
assumed to be constant throughout the solution. The fin material is In order to solve Eqns. (1)–(4) we need to specify the boundary
considered as aluminum and all material properties of fin are assumed conditions on the domain as has been shown in Fig. 1 (a) or Fig. 1 (b). It
to be constant with respect to temperature. can be seen that it is difficult to specify the condition in any coordinate
system, so we specify it through objects as has been described below,
∂ ⎛ ∂T ⎞ rather than stating the condition on a specified coordinate system.
Energy equation for conduction in solid : ⎜ ⎟ =0
∂x i ⎝ ∂x i ⎠ (4) The surface of the cylinder is given a wall boundary where the ve-
locity normal to it or tangential to it is set to be zero. The cylinder
surface temperature is set to a constant value of Tw.
4. Heat transfer parameters Since the fin base is at a temperature same as the cylinder surface,
fin base temperature is set to be Tw .
To compute the average Nu for the cylinder [22] we defined, Since the cylinder is hot the plume is expected to rise from it and
gβ (Tw − T∞ ) D3 exit from the top of the domain while the flow would come into the
Ra = (property evaluated at mean temp) domain from the sides of it. So all the six faces of the domain are given a
να (5)
pressure outlet boundary where the pressure is set to atmospheric or the
hAt = ∫ hc dAc + ∫ hf dAf , where, h is the average heat transfer gauge pressure set to zero. This condition is more realistic than any-
coefficient of the finned cylinder, dAc is the elemental area over the thing else because this would allow the flow to either go out of the
cylinder surface dAf is the elemental area over fin surface. Generally, domain or come into the domain depending upon the inside pressure
the total cylinder area can be calculated as Ac = 2∗ (πDL − tNL) , where condition in the domain which is normally computed in the process of
t is the thickness of the fin. Qc is directly computed from the solution of solution through the pressure linked equations.
∂T
the temperature field by integrating − ∫ kf ∂n dAc over the entire area of
the finned cylinder from where heat is being lost to the surroundings. At the fin-air interface the condition for heat flux and temperature
So the unknown value of h (heat transfer coefficient) now can be continuity should be satisfied and is given as:
computed like this:

43
S. Acharya, S.K. Dash International Journal of Thermal Sciences 134 (2018) 40–53

Fig. 6. Effectiveness (ε) of finned cylinder as a function of fin number, fin height for a Ra of 104(a) L/D = 0.2, (c) L/D = 0.5,(e) L/D = 2, (g) L/D = 5 and Ra of 107
(b) L/D = 0.2, (d) L/D = 0.5,(f) L/D = 2, (h) L/D = 5.

44
S. Acharya, S.K. Dash International Journal of Thermal Sciences 134 (2018) 40–53

Table 2 fluid near the wall. The cell size adjacent to the wall has been taken to
Optimization of fin configurations for maximum effectiveness as a function of be D/160 and a growth factor has been set to 1.21 for gradually in-
L/D and Ra. creasing the cell size from wall to the interior of the domain.
L/D Ra Optimum (N) Optimum (H/D) Optimum Effectiveness (ε) Fig. 4 shows the Nu variation with respect to number of cells. The
cell number was increased by refining the grid at the wall and domain.
0.2 104 40 0.25 1.35 It can be seen that up to a cell number of 5 × 105 Nu decreases gra-
0.5 104 8 0.2917 1.133
dually then it becomes almost constant. So for all the simulations of L/
1 104 8 0.2917 1.051
2 104 4 0.3333 1.003 D = 2 similar cells were used and when L/D changed, similar grid in-
5 104 4 0.3333 0.996 dependent tests were conducted to arrive at the required cell numbers.
0.2 105 24 0.25 1.7054
0.5 105 20 0.25 1.4465
6.3. A test of validation
1 105 12 0.2917 1.295
2 105 8 0.3333 1.1016
5 105 4 0.3333 0.9794 Table 1 shows three cases with detailed fin configurations used for
0.2 106 40 0.25 2.271 an internally finned vertical cylinder by Joo and Kim [19] in their ex-
0.5 106 24 0.25 2.12 periment which can be visualized from Fig. 5. Joo and Kim experi-
1 106 20 0.2917 1.782
mentally calculated the heat transfer rate from the vertically oriented
2 106 12 0.3333 1.24
5 106 4 0.3333 1.078 internally finned cylinder subjected to natural convection heat transfer.
0.2 107 40 0.25 3.25 For their calculation, they insulated the outer wall of the cylinder,
0.5 107 40 0.25 2.9214 performed experiment for a temperature difference of 30 °C and cal-
1 107 24 0.3333 2.559
culated heat transfer solely due to natural convection from the inner
2 107 24 0.3333 2.073
5 107 8 0.3333 1.292
part of the cylinder. In the present simulation, the cases mentioned in
0.2 108 40 0.2917 4.2419 Table 1 were computed and compared with the experimental data of
0.5 108 40 0.2917 3.509 Joo and Kim which shows a good agreement between both of them and
1 108 40 0.2917 2.916 hence confirming our present solution procedure and grid arrangement.
2 108 40 0.2917 2.539
5 108 20 0.3333 1.744
7. Results and discussions

∂T ∂Tfluid ⎞ 7.1. Effect of fin height, fin number, length of the cylinder and Rayleigh
ks ⎛ solid ⎞ = kf ⎛
⎜ ⎟
number on the effectiveness(ε) of the cylinder
⎝ ∂n ⎠interface ⎝ ∂n ⎠interface (7)

Tsolid = Tfluid (8)
6. Numerical solution methodology
Numerical solution methodology adopted for the present solution is
pretty standard and can be found from the authors previous work [12-
13,23-24]. Multi grid solver Fluent 15 [25] has been used for the pre-
sent simulations. We do not preset the same thing here for the sake of
saving space.
6.1. Domain independence test
The cylinder with internal fins was placed in a rectangular domain
such that the cylinder was away by 3.5D from the bottom of the do-
main. From a lot of simulations it was found that the distances of the
side faces of the domain from the cylinder did not influence the results
at all, so we kept it fixed for all the simulations as 2D. Since the height
of the domain would influence the result to some extent, we kept
varying the domain height by nD (n is an integer) to conduct a domain
independent test for each case.
From Fig. 2 it is clear that Nu increases slightly as the domain height
increased from 4D to 7D then remained constant. Since the plume from
the cylinder is received on the top surface of the domain, so the domain
height has an influence on the flow field to some extent. So the varia-
tion of domain height was more important than the distances of the
cylinder side faces from the domain. So in all other simulations, the
domain height was kept more than 8D or at least to 8D.
6.2. Grid independence test
Fig. 3 shows the cut away view from the entire domain with grid
arrangement around the hollow cylinder with internal fins and Fig. 3c
shows the blown up view of cells near the cylinder wall and the fin
surface. Very fine grids are taken near the cylinder wall and the fin
surface to capture the flow physics and boundary layer behavior of the
Effectiveness (ε ) of the finned cylinder is defined as the ratio of total
heat transfer from a finned cylinder to that of an un-finned cylinder.
Heat transfer from a cylinder explicitly depends upon the surface area
and the temperature gradient prevailing on the outer side and the inner
side of the cylinder with fins. The heat transfer is quantified by two
competitive parameters, one is the increase in surface area due to ad-
dition of fins and the other is the decrease in volume flow rate of air due
to flow obstruction generated by the presence of fins. With the increase
in fin number the surface area increases but the flow resistance offered
by the inner side of the cylinder also increases reducing the heat
transfer from the inner portion of the cylinder. Thus it is expected that
an optimal configuration could exist to give maximum heat transfer
from the cylinder. Such a picture can be seen from Fig. 6a–d where we
can see maximum heat transfer takes place at various H/D as the
number of fins change from 4 to 40 at a Ra of 104 and 107. For lower Ra
of 104 and L/D less than 0.5 we see maximum heat transfer takes place
at certain fin numbers and H/D. When the fins are only 4 the convection
loss from the fins do not decrease although the area increases with H/D.
So we do not see any maximum point of heat loss and the same thing
happens at Ra of 107. As the fin number increases the area of heat loss
increases and the flow resistance for inter fin convection also increases
which causes the heat loss to be reduced. So we get to see maximum
point of heat loss for fins more than 8 at L/D less than 0.5. For higher Ra
of 107 the convection around the inner portion of the cylinder where
the fins are there remains strong. So the inter fin convection does not
get reduced due to the rise in fin area through a rise in fin number or fin
height. So we do not see any maximum heat loss point for 4, 8, and 12
fins with H/D. Only when the fin number exceeds 24 we get to see a
maximum heat loss point (Fig. b, d) at H/D of 0.25. When the fins are
too many the area has increased a lot but the inter fin convection has
suffered a lot so the heat loss reduces from the inner portion causing
maximum heat loss to take place at a certain fin area and fin spacing for
low L/D of 0.2 or 0.5.
As the L/D increases the flow rate into the inner portion of the cy-
linder falls a lot and hence the convective heat loss from the fins also
reduces a lot. So at low Ra of 104 we never see any maximum heat loss

45
S. Acharya, S.K. Dash
Fig. 7. Finned cylinder effectiveness as a function of H/D and number of fins (N), (a) Ra = 104 and L/D = 0.5, (b) Ra = 107and L/D = 0.5, (c) Ra = 104and L/D = 1,
(d) Ra = 107 and L/D = 1, (e)) Ra = 104 and L/D = 5, (f) Ra = 107 and L/D = 5 for a hollow cylinder with internal fins and a hollow cylinder with external fins.
point and the effectiveness also remains lower than 1 (Fig. 6 e, g). But
for higher Ra of 107 the effectiveness remains higher than 1 for L/D
more than 2 although we do not see maximum heat loss points again,
except for the case of 40 fins at H/D of 0.3. However, it should be
marked that at 40 fins the effectiveness remains lower compared to that
of 4, 8, 12 and 24 fins for L/D of 5. At an L/D of 2 the effectiveness with
12 and 24 fins remains high compared to all other fin numbers. All
these phenomena happen simply due to the interplay between the fin
area which helps to take away more heat from the cylinder and on the
other hand the rise in the fin area causes to reduce the inter-fin con-
vection which reduces the heat transfer from the fins. So a maximum
heat loss configuration becomes possible.
We have provided here a table to know the configuration where
maximum heat loss can take place from the internally finned horizontal
cylinder. Table 2 shows the optimum configuration where maximum
heat loss can take place for some L/D in the range of 0.2–5 and Ra in the
46
International Journal of Thermal Sciences 134 (2018) 40–53
range of 104 to 108. This table can be helpful to industrial designers.
From Table 2 it can be observed that maximum effectiveness for dif-
ferent L/D and Ra lies between 1 and 4 and the finned cylinder is more
effective at a lower L/D for all Ra. Therefore, for some applications such
as cooling of annular nuclear reactor at the time of closing of power
plant, cooling of electric or vacuum furnace, oven wall after power
turned off, cooling of LED [26-27], shell and tube heat exchangers and
air ventilator [11], it would be a better option to have internal fin ar-
rangement for enhancement of the heat transfer, since the length to
diameter ratio generally varies from 0.2 to 2 in all the applications.
Internal fin arrangement is preferable only when the heat source is
present on the outer surface of the cylinder and it is difficult to attach
fins on the outer surface [19,20], otherwise both internal and external
fins could be employed on the inner and outer surface of the heated
hollow cylinder to maximize the heat transfer from the cylinder.
S. Acharya, S.K. Dash
Fig. 8. Variation of Nusselt number as a function of H/D and number of fins (N), (a) Ra = 104and L/D = 0.5, (b)) Ra = 107and L/D = 0.5, (c)) Ra = 104and L/D = 2,
(d)) Ra = 107 and L/D = 2, (e)) Ra = 104 and L/D = 5, (f) Ra = 107 and L/D = 5.
7.2. A comparison of effectiveness of a hollow cylinder with internal fins
with a hollow cylinder with external fins
Fig. 7 shows the comparison of effectiveness of an internally finned
hollow cylinder with an externally finned hollow cylinder as a function
of fin height and number of fins (N). For a high Ra of 107 the effec-
tiveness of an internally finned hollow cylinder is marginally higher
than that of a hollow cylinder with external fins at low L/D of 0.5, and
number of fins (N > 8) (Fig. 7 b). When H/D increases beyond 0.15,
effectiveness of a hollow cylinder with external fins is more than that of
the cylinder with internal fins. However, when L/D is greater than 0.5,
effectiveness of a hollow cylinder with internal fins is lesser than that of
the cylinder with external fins except at low fin number (N = 4) and
low H/D (Fig. 7 (d), (f)). For a low Ra, effectiveness of an internally
47
International Journal of Thermal Sciences 134 (2018) 40–53
finned hollow cylinder is higher than that of the externally finned cy-
linder for low H/D and less fin numbers when L/D is less than 0.5
(Fig. 7 (a)). For other cases effectiveness of an externally finned hollow
cylinder is more than that of the internally finned cylinder. However,
there is no constraint of increasing the height of the fin in case of a
hollow cylinder with external fins, which could help in carrying more
heat away from the cylinder and in the case of an internally finned
cylinder the fin height can not be increased beyond its radius. But, in
the present study, we consider the cases of the hollow cylinder with the
arrangement of internal fins only. Therefore, the heat transfer phe-
nomenon around a hollow cylinder with external longitudinal fins has
not been discussed here. Generally the internally finned cylinder is
more effective at low L/D and high Ra. Since at high Ra and low L/D the
entrainment of air into the cylinder through the side surface of the
S. Acharya, S.K. Dash
Fig. 9. Velocity vector around the cylinder with (a) N = 12, H/D = 0.0833, (c) N = 12, H/D = 0.3333 in longitudinal cross section, (b)N = 12, H/D = 0.0833, (d)
N = 12, H/D = 0.3333 in lateral cross section (zoomed view) for a Ra = 104 and L/D = 2.
cylinder is high, the air makes maximum surface contact with the inner
wall of the cylinder and carries more heat out of the cylinder. There-
fore, in this case, provision of internal fins become more effective which
can be seen from Fig. 7.2 (b).
7.3. Effect of fin height, number of fins, length of the cylinder and Rayleigh
number on the Nusselt number for internally finned horizontal cylinder
Fig. 8 shows the variation of the Nusselt number as a function of H/
D and number of fins placed inside the cylinder for different aspect ratio
and Rayleigh number of the cylinder. As H/D increases the area of the
fins increases at a particular L/D and the heat loss increases but attains
a maximum value (for L/D less than 1) and then decreases due to the
inter fin convection becoming less and less with the decreasing fin
spacing as because of increase in fin numbers. There is always an in-
crease in area of the fin with the increase in H/D and fin number. But
the heat loss does not increase indefinitely, rather attains a peak and
then falls for lower L/D and higher fin number. For lower fin number
the heat loss increases mostly. For higher L/D (more than 1) the heat
loss mostly reduces with H/D for Ra of 104 and increases for Ra of 107.
But Nu (which is a heat flux in the present study) is a ratio of heat loss
to total area of the fins and bare area of the cylinder. So if the area
increases faster compared to the heat loss then the Nu would fall
48
International Journal of Thermal Sciences 134 (2018) 40–53
always. This is what we see from Fig. 8 where Nu falls with H/D. It can
be seen from Fig. 8 that Nu is higher for lower L/D compared to that at
higher L/D. At low L/D the cylinder with internal fins have more heat
flux compared to that at high L/D which can be seen from Fig. 6
through effectiveness being higher at lower L/D compared to that at
higher L/D. So the Nu remains high at low L/D compared to that at high
L/D. As Ra increases the Nu also increases sharply at all H/D and L/D
compared to the corresponding H/D and L/D at lower Ra. This happens
because at higher Ra the buoyancy driven flow around the internally
finned cylinder and also on the external part of the cylinder becomes
much stronger to that at lower Ra. So the net heat loss increases com-
pared to that at lower Ra and as a result the Nu increases compared to
the corresponding lower Ra case. From Fig. 8 it can also be seen that
higher Nu is obtained for lower fin number and as the fin number in-
creases the Nu decreases. The reduction in Nu becomes too drastic at
higher fin number, higher L/D and higher H/D. At low fin number the
cylinder losses less heat compared to that at high fin number for Ra of
104 when L/D is less than 1. For L/D more than 1 the heat loss again
becomes more for less fin number compared to that at high fin number.
But Nu is a ratio of heat loss to total fin area, so the value of Nu reduces
with fin number always. Similar trends in heat loss can be seen for Ra of
107 with fin number and hence the Nu always decreases with fin
number as can be seen from Fig. 8.
S. Acharya, S.K. Dash
Fig. 10. Temperature contour around the cylinder for an Ra of 104 and L/D = 0.5, longitudinal and lateral cross section for (a) H/D = 0.0833 (b) H/D = 0.333 at the
mid section of the cylinder.
Fig. 11. Velocity vector around the cylinder with (a)N = 4, (b) N = 8 and (c) N = 40 fins in longitudinal cross section for H/D = 0.2083,L/D = 0.5 and Ra = 104.
7.4. Effect of fin height on the flow and temperature field for internal finned
cylinder
Fig. 9 shows effect of fin height on the flow field which is quantified
by the magnitude of the velocity around the cylinder for N = 12. Fig. 9a
shows the velocity field for lower fin height while Fig. 9c shows that for
higher fin height. As the fin height increases, it can be seen that the
plume from the side of the cylinder gets more resistance and can not
penetrate much into the cylinder while for lower fin height there is
considerable amount of plume penetration or air flow into the cylinder
which helps to take away more heat from the cylinder and the fin. This
results in higher Nu for the case of lower H/D compared to that at
higher H/D which can be seen in Fig. 8. Fig. 10 shows the corre-
sponding contour plot of temperature for two different fin heights at the
same Ra. Fig. 10a shows the temperature contour inside the cylinder
and outside of it for lower fin height and Fig. 10b shows that for higher
fin height. When the fin height is lower more air is sucked into the
cylinder and as a result the air temperature remains lower inside the
49
International Journal of Thermal Sciences 134 (2018) 40–53
cylinder while for the case of higher fin height there is more resistance
to air flow and hence less air flows into the cylinder causing the air
temperature to remain high inside the cylinder which finally results
into lower heat transfer.
7.5. Effect of fin number on the flow and temperature field for internal
finned cylinder
Fig. 11 shows the flow field around the finned cylinder at a Ra of
104 when the number of fins increases from 4 to 40 in the cylinder. As
the fins increases in number we can see that the flow of air into the
cylinder for the case of 40 fins has decreased compared to that of 4 fins
or 8 fins. More area of the inner portion of the cylinder is subjected to
higher air velocity in case of 8 fins compared to that of 4 and 40 fins
(Fig. 11). So the case of 8 fins gives rise to more heat transfer as we saw
from Fig. 6 (c). Fig. 12 shows the temperature contour for this case. We
can see that for the case of 4 fins the core of the cylinder is cooler
compared to that of 8 or 40 fins. For the case of 8 and 40 fins, the core
S. Acharya, S.K. Dash
Fig. 12. Temperature contour around the cylinder for a Ra = 104 and L/D = 0.5, longitudinal and lateral cross section for (a) N = 4, (b) N = 8 and (c) N = 40 at the
mid section of the cylinder.
can be seen to be much hotter where heat loss from the 8 fins has in-
creased and heat loss from the 40 fins has decreased. Although the core
with 8 fins is hotter compared to the case of 4 fins still the heat loss can
be more due to the rise in surface area. In case of 40 fins the core is even
hotter but the heat loss has decreased a lot since the inter-fin convection
has fallen although the surface area has increased (Fig. 6c).
7.6. Effect of Rayleigh number on the thermal plume for internal finned
cylinder
Fig. 13 shows the thermal plume around the internally finned hor-
izontal cylinder for L/D of 0.5 at different height of the fin when the fin
number is 40 and Ra is 104 and 107. When the fin height is more the
internal core of the cylinder is seen to be much hotter compared to the
case of lower fin height at the same Ra. When the fin height is less it
attracts more ambient air towards the core of the cylinder compared to
the case of higher fin height as can be seen from Fig. 13b which shows
the plume to be attracted to the inner core of the cylinder causing the
core to remain cooler. For higher fin height the plume does not get
attracted towards the inner core of the cylinder. When Ra increases to
107 the plume gets even more attracted to the inner core of the cylinder
(Fig. 13 c) and keeps the core cooler for lower fin height. When the fin
height is more the core remains hotter although there is plume pene-
tration to the core of the cylinder. At higher Ra, the core of the cylinder
has more area filled with lower temperature air compared to the case of
lower Ra. The plume structure around the internally finned horizontal
cylinder would help readers to understand the flow and heat transfer
and hence come out with better industrial design.
50
International Journal of Thermal Sciences 134 (2018) 40–53
7.7. Effect of fin height, fin spacing, length of the cylinder and Rayleigh
number on the fin efficiency (ƞ)
In the present study, the efficiency of the finned cylinder is calcu-
lated byη = Q . Where Q is the total heat transfer from the finned
Qmax
cylinder and Qmax is the heat transfer calculated when the fin surface is
at the cylinder base temperatureTw .Where Qmax can be calculated as
Qmax = (hc Ac + hfavg Af )(Tw − T∞) (9)
Q Qc + Qf
So efficiency η = =
Qmax (hc Ac + hfavg Af )(Tw − T∞) (10)
Fig. 14 shows the variation of fin efficiency for an internally finned
horizontal cylinder as a function of fin spacing and fin height for two
different Ra of 104 and 107at two different L/Ds of 0.5 and 2. Since
highly conductive aluminum fins are used, it is expected that higher
efficiency can be obtained which is reflected in Fig. 14. The efficiency at
lower Ra of 104 is slightly higher than that at higher Ra of 107. Taller
fins always have a lower efficiency than that of the shorter fins and the
efficiency at H/D = 0.0833 is approximately 1for all Ra. At high Ra the
fin efficiency is seen to have minimum values when the fin number is
high and the fins are longer. As the fins become shorter and shorter the
minima on the fin efficiency becomes lesser and lesser and at lowest fin
height the minima almost becomes negligible. For lower Ra of 104 al-
though there exists minimum on the fin efficiency curve but practically
that can be ignored since the efficiency is very close to 1.
S. Acharya, S.K. Dash International Journal of Thermal Sciences 134 (2018) 40–53

Fig. 13. Temperature contour around the cylinder for L/D = 0.5 longitudinal and lateral cross section for (a) H/D = 0.0833, Ra = 104, (b)H/D = 0.3333, Ra = 104,
(c) H/D = 0.0833, Ra = 107 and (d) H/D = 0.3333, Ra = 107.

8. Development of correlation for Nu: a = 0.3402545, b = 0.1905148, c = −0.1793514, d

= −2.543332, e = 0.4104193, f = 1.178002 g = −50.49396, h
Since lots of numerical results were obtained for many practical
situations with internal fins in the cylinder, so it was thought to be = 4.308947, i = 0.3664362, j = 0.2509347, k = −0.2173071, l
prudent to generate a general correlation for the Nu as a function of the = 0.0581678, m = −0.0565596, n = 0.0673154, o = −4.018556
pertinent input non-dimensional parameters such as H/D, N, L/D and
Ra. The correlation is valid for a wide range of the input parameters Fig. 15 shows the predicted values are within ± 6% of the calculated
where H/D, N, L/D and Ra are in the range of 0.0833–0.3333, 4 to 40, value. This is true for most of the data points when the Nu is more than
0.2 to 5 and 105 to 108respectively which is represented in the fol- 10. However, for Nu less than 10 the error band comes to within
lowing form: ± 10%. The proposed correlation can be used for both industrial ap-
plications and academic purposes.
L d H h L k H m
Nu = a (Rab) + c ⎛ ⎞ + e (N f ) + g ⎛ ⎞ + iRa j ⎛ ⎞ N l ⎛ ⎞
⎝D⎠ ⎝D⎠ ⎝D⎠ ⎝D⎠
9. Conclusions
H
+ nln ⎛Ra N ⎞ (o − N )
⎝ D ⎠ (11)
The full Navier-Stokes equations along with the energy equation
have been solved in the laminar range of Ra spanning from 104 to 108
Where, the values of the coefficients of constants are:
for a horizontal cylinder with internal fins to predict the heat loss from
it for different configurations of the fin and thus arrive at a surface
averaged Nu for the system. The following conclusions have been de-
rived from the present study.

51
S. Acharya, S.K. Dash International Journal of Thermal Sciences 134 (2018) 40–53

Fig. 14. Variation of fin efficiency as a function of fin spacing, fin height for (a) Ra = 104,L/D = 0.5 (b) Ra = 104,L/D = 2, (c) Ra = 107,L/D = 0.5 and (d)
Ra = 107,L/D = 2.

➢ Nu for a low L/D is found to be more than that of high L/D for all H/
D, N and Ra.
➢ The fin efficiency was found to be decreasing with H/D for all range
of Ra and fin number, N.
➢ A very useful correlation of Nu has been developed which can be
used in the industrial applications as well as for academic purposes
to find the natural convection heat transfer from an internally finned
horizontal cylinder.

Appendix A. Supplementary data

Supplementary data related to this article can be found at https://

doi.org/10.1016/j.ijthermalsci.2018.07.039

References

Fig. 15. A comparison of predicted Nu with that of the computed Nu for an
internally finned horizontal cylinder.
➢ For L/D less than 0.5 and for a low Ra of 104 there exists a point of
maximum heat transfer for all H/D and fin number except 4. For a
high Ra of 107 maximum heat loss takes place at H/D of 0.25 with
40 fins.
➢ For a low Ra of 104 and L/D > 2, the effectiveness of an internally
finned horizontal cylinder is found to be less than 1 for all H/D and
N suggesting that fins are not useful for enhancement of heat
transfer for these conditions.
➢ Average surface Nusselt number of the internally finned horizontal
cylinder is found to be continuously decreasing with increase of the
H/D and N for all Ra and L/D.
[1] R.J. McGlen, R. Jachuck, S. Lin, Integrated thermal management technique for high
power electronics devices, Appl. Therm. Eng. 24 (2004) 1143–1156.
[2] T. Aicher, H. Martin, Natural Convection in Heat Exchangers–a New Incentive for
More Compact Heat Exchangers, (2005).
[3] Q. Lin, S.J. Harrison, M. Lagerquist, Analysis and modeling of compact heat ex-
changers for natural convection application, Proceedings of EuroSun Conference,
Copenhagen, Denmark, 2000, June.
[4] M. Hidaka, S. Hatamiya, Y. Kataoka, T. Nakao, S.I. Kashiwai, Natural Convection
Type Nuclear Reactor, (1991).
[5] Y. Masuhara, K. Tominaga, Natural Convection Type Nuclear Reactor, (1992).
[6] T. Tsubouchi, H. Masuda, Natural convection heat transfer from horizontal cylin-
ders with circular fins, International Heat Transfer Conference, vol. 4, Begel House
Inc., 197017.
[7] J.A. Edwards, J.B. Chaddock, An experimental investigation of the radiation and
free convection heat transfer from a cylindrical disk extended surface, Trans. Am.
Soc. Heat. Refrig. Air-Cond. Eng 69 (1963) 313–322.
[8] E.M. Sparrow, P.A. Bahrami, Experiments on natural convection heat transfer on
the fins of a finned horizontal tube, Int. J. Heat Mass Tran. 23 (11) (1980)
1555–1560.
[9] E. Hahne, D. Zhu, Natural convection heat transfer on finned tubes in air, Int. J.

52
S. Acharya, S.K. Dash
Heat Mass Tran. 37 (1994) 59–63.
[10] R.L. Webb, M.J. Scott, A parametric analysis of the performance of internally finned
tubes for heat exchanger application, ASME J. Heat Transfer 102 (1980) 38–43.
[11] J.A. Myhren, S. Holmberg, Improving the thermal performance of ventilation
radiators–The role of internal convection fins, Int. J. Therm. Sci. 50 (2) (2011)
115–123.
[12] S. Acharya, S.K. Dash, Natural convection heat transfer from a short or long, solid or
hollow horizontal cylinder suspended in air or placed on ground, J. Heat Tran. 139
(7) (2017) 072501.
[13] S. Acharya, S.K. Dash, Natural convection heat transfer from perforated hollow
cylinder with inline and staggered holes, J. Heat Tran. (2017), https://doi.org/10.
1115/1.4037875.
[14] I.M. Rustum, H.M. Soliman, Numerical analysis of laminar mixed convection in
horizontal internally finned tubes, Int. J. Heat Mass Tran. 33 (7) (1990) 1485–1496.
[15] W.Q. Tao, Conjugated laminar forced convective heat transfer from internally
finned tubes, J Heat Transfer (Transactions of the ASME (American Society of
Mechanical Engineers), Series C);(United States) 109 (3) (1987).
[16] X. Liu, M.K. Jensen, Geometry effects on turbulent flow and heat transfer in in-
ternally finned tubes, J. Heat Tran. 123 (6) (2001) 1035–1044.
[17] I.M. Rustum, H.M. Soliman, Experimental investigation of laminar mixed convec-
tion in tubes with longitudinal internal fins, J. Heat Tran. 110 (2) (1988) 366–372.
[18] C. Prakash, Y.D. Liu, Buoyancy induced flow in a vertical internally finned circular
53
International Journal of Thermal Sciences 134 (2018) 40–53
duct, J. Heat Tran. 107 (1) (1985) 118–123.
[19] Y. Joo, S.J. Kim, Thermal optimization of vertically oriented, internally finned tubes
in natural convection, Int. J. Heat Mass Tran. 93 (2016) 991–999.
[20] H.M. Soliman, T.S. Chau, A.C. Trupp, Analysis of laminar heat transfer in internally
finned tubes with uniform outside wall temperature, J. Heat Tran. 102 (4) (1980)
598–604.
[21] D.D. Gray, A. Giorgini, The validity of the Boussinesq approximation for liquids and
gases, Int. J. Heat Mass Tran. 19 (1976) pp545–551.
[22] A. Bejan, Convection Heat Transfer, John wiley & sons, 2013.
[23] J.R. Senapati, S.K. Dash, S. Roy, Numerical investigation of natural convection heat
transfer over annular finned horizontal cylinder, Int. J. Heat Mass Tran. 96 (2016)
330–345.
[24] S. Acharya, S.K. Dash, Numerical analysis of natural convection heat transfer from a
vertical hollow cylinder suspended in air, J. Heat Tran. (2017), https://doi.org/10.
1115/1.4038478.
[25] (15), A.N.S.Y.S. Fluent, User's Guide, (2014) USA.
[26] J. Petroski, Advanced natural convection cooling designs for LED bulb systems,
ASME 2013 International Technical Conference and Exhibition on Packaging and
Integration of Electronic and Photonic Microsystems, 2013 V002T08A035-
V002T08A035.
[27] Bi LI, L. Xiong, C. Lai, Y. Tang, B. Li, L. Xiong, C. Lai, Y. Tang, Investigation of
internally finned LED heat sinks, J. Semiconduct. 39 (3) (2018) 034002.