Journal of Heat Transfer.

Received December 07, 2016;

Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

Natural Convection Heat Transfer from a Short or Long, Solid or Hollow

Horizontal Cylinder Suspended in Air or Placed on Ground
Swastik Acharya* and Sukanta K Dash
Department of Mechanical Engineering
Indian Institute of Technology Kharagpur, India, 721 302
*
Email: swastik.acharya8@gmail.com

Abstract:

d
ite
Numerical simulations have been conducted to study natural convection heat transfer from
solid or hollow cylinders in the laminar range of Ra spanning from 104 to 108 for L/D in the

ed
𝐿𝐿
range of 0.05 ≤ ≤ 20. Interesting flow structures around the thin hollow cylinder have
𝐷𝐷

py
been observed for small and large L/D. It has been found that the average Nu for solid or
hollow horizontal cylinders in air is marginally higher than when they are on ground for the

Co
entire range of L/D and Ra limited up to 107. Up to a Ra of 107 Nu for a solid cylinder in air
is higher than that of Nu for a hollow cylinder in air but when Ra exceeds 107 Nu for a
ot
hollow cylinder is marginally higher than that of the solid cylinder until an L/D of 0.2. When,
tN
L/D rises beyond 0.2 the situation reveres causing Nu for a solid cylinder to be again higher
than that of the hollow cylinder when suspended in air. A solid cylinder on ground has higher
rip

Nu compared to that of a hollow cylinder on ground up to a Ra of 106. However, for higher

sc

Ra of 108 a hollow cylinder on ground has higher Nu compared to that of a solid cylinder on
ground until an L/D of 5 and after that the situation reverses again.
nu

.
Ma

Key words: Short or long cylinder, Hollow or solid cylinder, Natural convection, Inner
surface heat loss, Outer surface heat loss
ed
pt
ce
Ac

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

Nomenclature :

At total area of convective surface, m2

D diameter of horizontal cylinder, m

g acceleration due to gravity, m/s2

h average convective heat transfer coefficient, W/m2 - k

d
k thermal conductivity of fluid, W/m-K

ite
L length of the cylinder, m

ed
Nu average surface Nusselt number based on diameter

py
Pr Prandtl number

p pressure, N/m2

Co
Q total heat transfer rate, W

T fluid temperature, K
ot
tN
Tw cylinder surface temperature, K

T∞ ambient temperature, K
r ip

t time, sec
sc

U flow velocity, m/sec

nu

x Cartesian coordinate
Ma

Ra Rayleigh number based on diameter

L/D length to diameter ratio of cylinder

ed

Greek symbols
pt

α thermal diffusivity, m/s2

ce

β thermal expansion coefficient, 1/K

Ac

μ dynamic viscosity, kg/m-s

ν kinematic viscosity, m/s2

ρ density, kg/m3

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

1. Introduction

To improve the performance as well as reliability of components natural convection heat

transfer is preferable over forced convection as it does not require any external source for
fluid motion. Applications of natural convection are found in numerous engineering
applications such as cooling of pipe lines, cooling of hot wires, heat exchanger design,
cooling of electronic components, boiler design, air conditioning etc. Natural convection heat

d
transfer from a heated horizontal cylinder has been a challenging topic for researchers since

ite
ten decades because of its huge application in industrial problems. Natural convection heat

ed
transfer from horizontal solid cylinders has been studied in detail and a summary is given
here where as literature on short solid and hollow cylinders are rarely available.

py
1.1 Early investigation on horizontal solid cylinder

Co
In long past, studies on solid cylinders were done mainly in the form of experiments [1-11].
As the main objective was to find the heat transfer from the heated cylindrical surface, many
ot
experiments have been done for different cases by varying aspect ratios, working mediums,
tN

materials of cylinder within the constraint of Grashof number and Prandtl number to give a
suitable correlation for average Nusselt number. Generally the correlated average Nusselt
ip

number for all the experiments of early investigations were of the form of
r
sc

𝑁𝑁𝑁𝑁 = 𝑎𝑎 (𝐺𝐺𝑟𝑟𝐷𝐷 Pr)b

nu

Where, a and b were constants. All the early investigations ended by about 1966 and from
Ma

there onwards new types of investigations started in the area of natural convection.

1.2 Modern investigations

ed

Compared to the early investigations the modern studies (after 1970) brought about fairly
pt

accurate and complicated correlations for Nu. The limiting value of Nusselt number
ce

correlation from Saville and Churchill [10] as well as Tsuboucghi and Masuda [9] were used
by Churchill and Chu [12] to predict the most accurate Nu for horizontal cylinders which was
Ac

valid for all Rayleigh and Prandtl numbers. The correlation was in the form of

(1/6)
 Ra 
Nu = 0.6 + 0.387 
1/2
9/16 16/9 
 [1 + (0.559 / Pr) ] 

and is valid for all Prandtl number and Ra within the range of 10-11≤ Ra≤ 109 .

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

Nakii and Okazaki [14] analytically calculated Nusselt number for natural convection over a

heated horizontal cylinder at a low Grashof number and compared their results with

experimental data. Merkin [15] presented a complete solution of full Navier - Stokes

equation using Blasius and Gortler series expansion method along with integral method and

finite difference scheme and solved natural convection boundary layer flow over a cylinder of

d
elliptical cross section. Kuhne and Goldstein [16] numerically solved the complete Navier –

ite
Stokes equation along with the energy equation for natural convection heat transfer from a

ed
horizontal heated cylinder with the conclusion that boundary layer assumptions are not valid

for a low and moderate Rayleigh number. Farouk and Guceri [17] studied both theoretically

py
and numerically the natural convection heat transfer from a horizontal cylinder and validated

Co
their results against Kuhne and Goldstein [16] with the conclusion that the average Nusselt

number decreases from bottom to top of the cylinder. In 1983 De Docio [18] performed
ot
experiment on natural convection from a horizontal cylinderwhich was partly submerged in
tN

ice and came out with a relation between Nu and Ra (Ra in the range of 104 to 107). Al -
ip

Arabi and Khamis [19] came out with correlation for Nu versus Ra for horizontal cylinders
r
sc

with different lengths and diameters having many inclination angles to the vertical with a

conclusion that transition to turbulence occurs near (GrLPr)Cr = 2.6×109 + 1.1×109 tanθ.
nu
Ma

Most of the earlier studies as well as the modern studies were carried out in steady state.

Wang et al. [20] numerically investigated transient natural convection from heated cylinders
ed

with spline fractional method. Characteristics related to boundary layer thickness were
pt

obtained by scaling analysis and compared with numerical results. Overshoot and oscillatory
ce

behavior of local Nusseltnumbers have been observed to be decayed as steady state reached.
Ac

Saitoh et al. [21] solved natural convection heat transfer from horizontal cylinder with five

different kind of methods such as Finite difference, Multi point FDM with uniform mesh,

Multi point FDM with two computational domain and Multi point FDM with logarithmic

coordinate transformation. In their study local as well as average Nusselt number has been

plotted against Rayleigh number of 103, 104 and 105. Chouikh et al. [22] numerically solved

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

natural convection from an isothermal horizontal cylinder with stream function - vorticity

approach for laminar flow in a 2D domain with Ra varying from 10 to 106. Md. Mamun

Molla et al. [24] solved the non-dimensional boundary layer equations for a horizontal

circular cylinder with uniform heat flux in presence of heat generation.

d
Earlier investigation has been done with constant thermo physical properties but Molla et al.

ite
[23] investigated natural convection heat transfer from a heated horizontal cylinder with

ed
temperature dependent viscosity. Viscosity was assumed to be a linearly inverse function of

py
temperature. Their result indicated local heat transfer was found to be increasing and local

skin friction coefficient was decreasing with increase in viscosity. Atayilmaz and Teke [25]

Co
experimentally as well as numerically studied natural convection from a heated cylinder in
ot
different environmental temperature condition. They took two cylinders of diameter 4.8 mm

and 9.45 mm for their experimental study and varied the cylinder surface temperature in the
tN

range of 100C to 400C and 200C to 600C respectively. They proposed correlation for Nusselt
ip

number in the range of Ra 74 to 3400 and validated their result with Churchill's correlation.
r
sc

Ashjaee et al. [26] calculated Nusselt number of a heated horizontal cylinder located at a

certain height from an adiabatic surface both experimentally and numerically. They solved it
nu

using a 2-D finite volume method with a SIMPLE scheme for pressure correction equation
Ma

where Ra was in the range of 102 to 105 and L/D in the range of 0.1 to 17 where L was the

distance from the adiabatic surface to the bottom point of the cylinder.They could observe for
ed

a particular L/D and Ra the local Nusselt number was increasing at the lower surface of the
pt

cylinder and also the average Nu was increasing with increase in L/D.
ce

A two dimensional heat transfer solution of natural convection around a horizontal cylinder
Ac

located beneath an isothermal wall was studied by Meherizi et al. [27] using Lattice-

Boltzmann method. They found that heat transfer decreased by reducing the distance of the

cylinder from the isothermal wall. In the recent study Kuehner et al. [28] investigated

experimentally velocity boundary layer and plume formation region over a heated horizontal

cylinder in water. Particle image velocimetry measurement was obtained for free convection

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

around the heated horizontal cylinder for a Rayleigh number of 1.33×106 and Prandtl number

of 5.98. In his study the effect of plume swaying on the velocity boundary layer and plume

development were investigated. It was observed that plume swaying decayed as H/D ratio

increased where H was the distance from the extreme upper point of cylinder to the free

surface of water.

d
A steady two dimensional laminar natural convection from a horizontal cylinder enclosed in

ite
the corner of two adiabatic walls were studied both numerically and experientially by

ed
Sedaghat et al. [29]. They varied Rayleigh number from 3×105 to 6×105 experimentally and

104 to 106 numerically. Computations were made varying the vertical cylinder spacing and

py
horizontal cylinder spacing to predict the heat transfer and show the thermal plume.

Co
1.3 Modern investigations on Hollow cylinder
ot
Natural convection research on hollow cylinder started in 1991 by Tsuang-Yen Na [30].
tN

Natural convection over a hollow cylinder with constant inner surface temperature was
ip

solved where the outer surface temperature was to be found out. Effect of wall conduction on
r

natural convection was studied in his work. In 2008 Cheng - Long Chang [31] numerically
sc

investigated natural convection of micro polar fluid flow along slender hollow cylinder with
nu

wall conduction effect. This study investigated the effect of conjugate heat transfer
Ma

parameter, micro polar parameter, and the Prandtl number of fluids. The effect of wall

conduction on local heat transfer coefficient, interfacial temperature was found. In 2010
ed

Mamun et al. [32] studied mixed convection heat transfer characteristics within a ventilated
pt

cavity having heated hollow cylinder. Effect of thermometric parameter on heat transfer
ce

coefficients were found in this study.

Ac

Bilah et al. [33] numerically solved the fluid flow due to mixed convection in a lid driven

cavity having a heated cylinder. A hollow cylinder was positioned at the center of a lid driven

cavity and the governing equations were solved by residual finite element method with

Newton's Raphson Technique. It was found that the flow field was strongly dependent upon

the cylinder diameter. Rani and Reddy [34] investigated the conjugate transient free

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

convective heat transfer from a vertical slender hollow cylinder with heat generation effect

where the inner cylinder was kept at a constant temperature. CFD technique was

implemented to solve the governing equations with unconditionally C-N type of implicit

finite difference scheme.

After a thorough literature survey it has been found out that the natural convection flow and

d
the temperature field around a short or long, solid or hollow horizontal cylinder is not readily

ite
available in the literature which can give a pictorial view of the plume around the cylinder

ed
and hence help us to understand the heat transfer from it in a much better way.

py
When a solid or a hollow cylinder is in air or lying on the ground the flow and the thermal

Co
plume around it as a result of natural convection can be visualized experimentally although it
would be a more difficult and expensive task. However, the same can be visualized by
ot
performing a 3D numerical simulation of the Navier - Stokes equation along with the energy
equation in a relatively very short time with too less expenses. 3D simulations are needed
tN

because of the shortness of the cylinder otherwise a 2D simulation could have been fine.
ip

Here, we see the natural convection flow vector around a short isothermal hollow cylinder
r

and a short solid cylinder (end faces insulated) when they are placed on the ground.
sc
nu

0.035 0.07 0.105 0.14 0.175 0.21 0.245 0.28

Ma
ed
pt
ce
Ac

(a) (b)

Fig. 1. Flow vector around a (a) hollow and (b) solid isothermal cylinder on ground, L/D =1, Ra=10

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

Such a pictorial view is possible today due to the immense computing power of the modern
day computers which was not possible even 15 years ago. Natural convection cooling from
short and long cylinders in air or ground is an everyday phenomenon in steel making
industries. Pipes in hot condition are placed on the ground to cool naturally and ingots also
the same way. So naturally one becomes curious to study such phenomenon around a short or
long hollow cylinder and short or long solid cylinder either in air or placed on the ground.

d
The present study is motivated in this direction where it is aimed to solve the 3D governing

ite
equations in the flow domain and show the flow and thermal plume pictorially, while

ed
estimating the average Nu and the heat loss from the inner and the outer surface of the hollow

py
cylinder for different L/D ratio when it is either in air or lying on the ground.

Co
ot
tN
r ip
sc
nu
Ma
ed
pt
ce
Ac

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

2. Physical description of the problem

Fig. 2 (a) Fig. 2 (b)

d
Fig. 2. (a) A solid or thin hollow cylinder in air with the computational domain around it, (b) solid or the thin hollow

ite
cylinder lying on ground with its computational domain (a schematic representation)

Fig. 2 (a) shows an isothermal solid or thin hollow cylinder suspended in air and Fig. 2 (b)

ed
shows the same cylinder lying on ground. The objective of the problem is to find out the

py
natural convection flow and temperature field around the cylinder when it is in air or lying on
ground for various aspect ratio (L/D) of the cylinder while Ra is in the laminar range of 104 to

Co
108. After obtaining the temperature field the objective is again to find out the surface
average Nu for the cylinder in various conditions and the heat loss from the inner and outer
ot
surface of the hollow cylinder. In order to solve this problem a 3D solution of the momentum
tN
and energy equations are needed for which a computational domain around the cylinder has
to be built in. The size of the domain would influence the flow field and the result to some
ip

extent. Hence, a domain independent study along with grid independence test also has to be
r
sc

performed. Since, we plan to show the flow field around a short thin hollow cylinder or a
short solid cylinder either in air or on ground, so a 3D solution is required to be performed.
nu

For simulation of the flow and temperature field around the cylinder when it is in air, the
Ma

computational domain has to surround the cylinder and the space above the cylinder has to be
much more compared to that at its sides and bottom. All the six faces of the computational
domain would serve as pressure outlet boundary so that the flow can either come into the
ed

domain or leave the domain. When the cylinder is on the ground the computational domain
pt

has to merge with the bottom curved surface of the cylinder and there would be space on the
ce

side flat surface, curved surface and the top curve surface of the cylinder. The bottom surface
of the computational domain for this case would serve as an adiabatic wall while all other 5
Ac

faces would be pressure outlet boundary so as to allow the flow to either come into the
domain or leave the domain.

3. Mathematical formulation:

The starting point for the solution of the flow and temperature field around the cylinder is the
continuity and the Navier - Stokes equation along with the energy equation which we write

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

here in tensorial form in a rectangular coordinate system, since this is the best compact way
of representing them.

𝜕𝜕
Continuity : (ρ𝑈𝑈𝑖𝑖 ) = 0 (1)
𝜕𝜕𝑥𝑥 𝑖𝑖

𝐷𝐷(ρ𝑈𝑈 𝑖𝑖 ) 𝜕𝜕𝜕𝜕 𝜕𝜕 𝜕𝜕 𝑈𝑈 𝑖𝑖 𝜕𝜕𝑈𝑈 𝑗𝑗

Momentum : = − 𝜕𝜕𝑥𝑥 + 𝜕𝜕𝑥𝑥 �µ � + �� − ρβ𝑔𝑔𝑖𝑖 (𝑇𝑇 − 𝑇𝑇∞ ) (2)
𝐷𝐷𝐷𝐷 𝑖𝑖 𝑗𝑗 𝜕𝜕 𝑥𝑥 𝑗𝑗 𝜕𝜕 𝑥𝑥 𝑖𝑖

𝐷𝐷(ρ𝑇𝑇) 𝜕𝜕 µ 𝜕𝜕𝜕𝜕
= 𝜕𝜕𝑥𝑥 �𝑃𝑃𝑃𝑃 𝜕𝜕𝑥𝑥 � (3)

d
Energy: 𝐷𝐷𝐷𝐷 𝑖𝑖 𝑖𝑖

ite
Since the flow field is buoyancy driven hence, it is not assumed to be compressible. So the

ed
value of ρ is computed at the mean reference temperature while the buoyancy force term is
computed from the source term in Eqn.(2) through Boussinesque approximation where β is

py
also computed at the mean temperature. Instead of assuming Boussinesque approximation

Co
one could have taken the ideal gas law to directly compute the density term in the source term
of Eqn.(2) and in that case β (𝑇𝑇 − 𝑇𝑇∞ ) would have been dropped.
ot
4. Heat Transfer parameters:
tN

To compute the average Nu for the cylinder we defined,

ip

𝑔𝑔 β (𝑇𝑇𝑤𝑤 −𝑇𝑇∞ )𝐷𝐷 3

𝑅𝑅𝑅𝑅 = and 𝑄𝑄 = ℎ𝐴𝐴𝑡𝑡 (𝑇𝑇𝑤𝑤 − 𝑇𝑇∞ ) (property evaluated at mean temp) (4)
να
r
sc

Where At is the total surface of the cylinder from where heat is being lost. 𝐴𝐴𝑡𝑡 = 2𝜋𝜋𝜋𝜋𝜋𝜋 for the
nu

case of a hollow cylinder and 𝐴𝐴𝑡𝑡 = 𝜋𝜋𝜋𝜋𝜋𝜋 for the case of a solid cylinder with the ends
𝐷𝐷 2
insulated. When the ends of the solid cylinder is not insulated then, 𝐴𝐴𝑡𝑡 = 𝜋𝜋𝜋𝜋𝜋𝜋 + π
Ma

. Q is
2
𝜕𝜕𝜕𝜕
directly computed from the solution of the temperature field by integrating − ∫ 𝑘𝑘 𝑑𝑑𝑑𝑑 over
𝜕𝜕𝜕𝜕
ed

the entire area from where heat is being lost from the cylinder. So the unknown value of h
(heat transfer coefficient) now can be computed like this:
pt
ce

𝑄𝑄 ℎ𝐷𝐷 𝑄𝑄𝑄𝑄
ℎ= and 𝑁𝑁𝑁𝑁 = = (5)
𝐴𝐴𝑡𝑡 (𝑇𝑇𝑤𝑤 −𝑇𝑇∞ ) 𝑘𝑘 𝐴𝐴𝑡𝑡 (𝑇𝑇𝑤𝑤 −𝑇𝑇∞ )𝑘𝑘
Ac

So it can now be seen from Eqn.(5) that our computed value of Nu is proportional to the net
heat loss from the cylinder and a prescription of h is not needed for the computation of Q or
Nu in our work.

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

5. Boundary conditions:

In order to solve Eqn.(1) to (3) we need to specify the boundary conditions on the domain as
has been shown in Fig. 2 (a) or Fig. 2 (b). It 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, rather than stating the condition on a specified coordinate system.

The surface of the cylinder is given a wall boundary where the velocity normal to it or

d
ite
tangential to it is set to be zero. The cylinder surface temperature is set to a constant value of
𝑇𝑇𝑤𝑤 .

ed
For the solid cylinder the flat surfaces are given an adiabatic condition for the temperature.

py
Another condition of constant 𝑇𝑇𝑤𝑤 on the flat surface has also been given to generate different

Co
flow pattern and heat loss from the cylinder. For the hollow cylinder this situation does not
arise since the entire cylinder is at a constant temperature of 𝑇𝑇𝑤𝑤 . Other than this the cylinder
does not need any boundary condition while conditions are to be specified on the domain
ot
boundary.
tN

For the cases of suspended cylinder in air, the domain surfaces are not touching the cylinder
ip

wall. Since the cylinder is hot the plume is expected to rise from it and exit from the top of
r

the domain while the flow would come into the domain from the sides of it. So all the six
sc

faces of the domain are given a pressure outlet boundary where the pressure is set to
nu

atmospheric pressure. This condition is more realistic than anything else because this would
allow the flow to either go out of the domain or come into the domain depending upon the
Ma

inside pressure condition in the domain.

ed

For the case of the cylinder lying on the ground, the domain bottom surface is in contact with
the cylinder. So no flow is supposed to cross this surface. So the bottom surface of the
pt

domain is given an adiabatic wall condition which would not allow any flow to cross this
ce

surface and heat is also not expected to cross this surface. However, from mathematical point
Ac

of view this surface could have been given a constant temp condition representing the earth
surface, but that would have called for more parametric variation in the present solution
which we avoided in this present work.

6. Numerical solution methodology

Equations (1-3) are to be solved in a domain which encompasses the cylinder so that we can
obtain the flow and the temperature field around the cylinder. For this purpose we need to

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

discretize the domain containing the cylinder into small cells over which our equations can be
solved. First all the equations are integrated over small cells and the integral equation is then
discretized using finite volume technique which yields a system of non-linear algebraic
equations. These algebraic equations are then solved by the multi grid solver of Fluent 15.0
[37] using appropriate boundary conditions which have been stated above. From the
geometry of a solid or hollow cylinder either in air or on ground it is clear that there would be

d
symmetry around the longitudinal cross sectional plane as well as the transversal cross

ite
sectional plane of the cylinder which has been used in the present simulation to reduce the
number of computational cells greatly. The convective terms in the momentum and energy

ed
equations have been discretized to first order upwind scheme for a trial convergence and then

py
switched over to second order upwind for the final converged solution. The pressure
correction equation has been coupled with the flow equations as per the SIMPLE algorithm.

Co
Initial guesses for all the three velocities were set to zero while the temperature for the flow
field was set to that of the ambient and the surface temperature of the cylinder was set to the
ot
wall temperature as a part of the boundary condition. Under relaxation parameters such as 0.3
tN
for the continuity and 0.7 for the momentum and 1 for the energy equations were used for the
iterative solution of the equations. Details of these solution processes can be found in the
ip

author’s previous work [35-36]. Up to a Ra of 106 a direct steady state solution of the
r
sc

equations are possible but however after that a transient solution has be done in order to
obtain the steady state and a time step of 0.01s was found to be adequate for convergence.
nu

Normalized residuals for all iterations were checked and when that fell below 10-4 for the
continuity and the momentum equations and 10-6 for the energy equation the solution was
Ma

said to have been converged. After the convergence of the solution grid independence and
domain independence tests were carried out before the final results were used for further
ed

computations and plotting.

pt

6.1 Domain independence test

ce

The cylinder was placed at a location in the rectangular domain such that the bottom of the
Ac

cylinder was away by D or 2D from the bottom plane of the domain. The side surfaces of the
cylinder was away by 2D from the side planes of the domain while the top surface of the
cylinder was away by nD (n is a number) from the top surface of the domain. From lots of
simulations we found that the distance of the side faces of the cylinder from the domain side
surfaces and the distance of the bottom surface of the cylinder from the bottom plane of the
domain do not influence the result at all, so we kept them fixed at 2D while the domain

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

height was influencing the result to some extent so we kept on varying the domain height to
see the effect on Nu.

d
ite
ed
py
Fig. 3. Influence of domain height on average Nu for a solid cylinder in air

From Fig.3 it is clear that as the domain height increases from 4D onwards the change in Nu

Co
diminishes and finally that becomes a constant at about 8D. Since the flow is a suction flow
from the sides of the domain surface (on most part of it) so the distance of the domain surface
ot
from the side surface of the cylinder does not play a vital role in changing the flow structure.
tN
Whereas the top surface of the domain receives the exit plume coming from the cylinder
surface and the top surface is also a pressure outlet boundary which influences the internal
ip

pressure in the domain to some extent. So the distance of the top surface of the domain from
r
sc

the cylinder top edge is relatively important compared to any other distance from the
cylinder. So in all simulations (cylinder in air or on ground) the height of the domain was
nu

kept more than 8D or at least 8D.

Ma

6.2 Grid independence test

ed
pt
ce
Ac

(a) (b) (c)

Fig. 4. (a) cell arrangement for hollow cylinder in air, (b) cross sectional view, (c) Blown up view near the
cylinder wall

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

Fig.4 shows the cell arrangement around a hollow cylinder which is in air. Fig.4 (b) shows
the cells in transversal cross section and Fig.4 (c) is a blown up view of the cells near the
cylinder wall and inside the cylinder. This is to be noted here that the cell arrangement shown
in Fig.4 is also a cut away view from the entire domain and not the total picture of the entire
domain (since it is not possible to show the cells in the entire domain in a small area).

d
ite
ed
py
Co
Fig. 5. Average Nu for a hollow cylinder in air as a function of number of cells
ot
Fig.5 shows the variation of Nu as we keep on refining the cells near the cylinder surface and
tN
away from it. It can be seen that after a cells of 4x105 the average Nu remained almost
constant. For all other simulations with L/D =1 we used similar cells and whenever L/D
ip

changed from 1 we did similar grid independence test before reporting the result. It is to be
r
sc

noted that the minimum cell length was less than D/80 or sometime about D/100 near the
cylinder wall and away from the wall it was around D/40.
nu

6.3 A test of validation

Ma
ed
pt
ce
Ac

Fig. 6. Average Nu for a solid cylinder in air, a comparison with Experimental correlation

Fig.6 shows the average Nu for a solid cylinder (end faces insulated) in air for different
aspect ratio (L/D) of the cylinder. It can be seen that for the laminar range of Ra going from
104 to 107 our present computation matches well with the correlation of Churchill and Chu

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

[12] and also with Morgan [13]. It has to be noted here that the correlations of Churchill and
Morgan are only valid for long cylinders and only up to a Ra of 107. So our computation of
Nu at Ra of 108 does not match with the correlation very well. For short solid cylinders the
Nu is expected to be more compared to that of a long cylinder and our present computation
shows this. Reason for this and explanation on this would be given little later in this work.

7. Results and Discussions

d
ite
We have envisaged showing the plume and flow structure around a short or long cylinder
either in air or lying on ground. So we would do that in this section systematically while

ed
bringing comparisons on Nu for many different cases.

py
7.1 Solid (end faces insulated) or hollow cylinder in air or ground

Co
ot
tN
r ip
sc
nu

Fig. 7. Average Nu for a solid cylinder in air or lying Fig. 7a. Average Nu for a hollow cylinder in air or
on ground as a function of L/D lying on ground as a function of L/D
Ma

Fig.7 and Fig.7a show the average Nu for a solid and hollow cylinder respectively when they
ed

are either in air or on ground. It has to be noted here that when the cylinder is on the ground it
pt

is numerically difficult to go to an L/D of 0.05 so we could only do the computations up to an

ce

L/D of 0.1. As L/D increases the Nu decreases and becomes almost constant after L/D of 5
for a solid cylinder (adiabatic ends) in air or ground. But for a hollow cylinder constancy in
Ac

Nu is not achieved even after an L/D of 20. The edge effect of flow and heat transfer from the
cylinder is predominant at low L/D compared to high L/D so the Nu at low L/D is higher
compared to that of high L/D. Since, the flow enters the hollow cylinder and leaves from
within the inside so, the edge effect in case of a hollow cylinder is higher compared to a solid
cylinder for which the fall in Nu is marginally higher at low L/D in case of a hollow cylinder
compared to that of a solid cylinder.

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

It can be seen from the plots that for the range of Ra in between 104 to 107 a cylinder in air
has marginally higher Nu compared to that of a cylinder on ground (both cases of solid or
hollow). However for a Ra of 108 there is a considerable difference in Nu for a cylinder (for
both solid and hollow) in air to that of a cylinder on ground. When the cylinder is on ground
the natural convection plume from the bottom of the cylinder is impaired due to the presence
of the ground while in air the plume develops well from the bottom of the cylinder and

d
proceeds to the top surface of the cylinder without any hindrance. This is the main reason

ite
why the Nu is marginally higher when the cylinder is in air compared to that of the case when
it is on the ground. As Ra becomes high the plume strength becomes high for a cylinder in

ed
air, but when it is on the ground the plume strength is relatively low compared to that when it

py
was in air. So the average Nu becomes low for the case when a cylinder is on ground and the
effect becomes prominent at high Ra which we see in Fig.7. When the cylinder is hollow

Co
there would be a plume entering into the cylinder as well as there would be a plume outside
the cylinder and this would happen for both the cases of the cylinder either in air or on
ot
ground. Both the inside plume and the outside plume would take away heat from the hollow
tN
cylinder for which the Nu for a hollow cylinder in air or ground would not be very different
from each other for low L/D up to 1. At low L/D the plume would enter the hollow cylinder
ip

up to its middle but for higher L/D of 5 or more the plume would not go up to the center of
r
sc

the cylinder or even if it enters the strength would remain quite low. It would only enter some
distance in to the cylinder and then return back to exit on the same face through which it had
nu

entered. So in this case the heat loss from the hollow cylinder at high Ra from its inside
Ma

portion would be less and hence as a whole the Nu would be less for the case of a hollow
cylinder on ground (since the stronger buoyant plume would penetrate less in to the cylinder
when it is on ground compared to that of the case when it in air) compared to that of a hollow
ed

cylinder in air and the difference at Ra = 108 would be easily visible which we see in Fig.7a.
pt
ce
Ac

(a) (b)
301 303.5 306 308.5 311 313.5 316 318.5 321 323.5 326

Fig. 8. Temperature plume near a hollow cylinder on (a) ground and in (b) air, Ra = 108, L/D = 5

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

Fig.8 shows the temperature contour or thermal plume around a hollow cylinder in air or on
ground. It is clear from Fig.8a that the outside air enters less into the cylinder core compared
to that of the case when the cylinder is in air. When in air more air enters the cylinder, so the
core has more blue zone (low temperature zone) and hence more heat is lost from the inside
of the cylinder, as a result the cylinder loses more heat compared to the case when it is on
ground. From Fig.8 (b) it is clearly visible that the average temperature gradient on the

d
outside of the cylinder is more compared to the case when it is on ground, which results in

ite
more heat loss from the cylinder when it is in air. At this high Ra a transient simulation gives
better result and a steady state simulation is too hard to give a converged solution. We used a

ed
time step of 0.01 for this case and have tried also many other time steps like 0.005, .001 and

py
0.05. All these time steps give the same figure at a particular time which is considered to
have attained steady state.

Co
7.2 Solid cylinder (end faces isothermal) in air or ground
ot
Fig.9 shows the variation of Nu as a function of L/D for a solid cylinder in air or lying on
tN
ground when its end faces are isothermal. In Fig.7 we saw the plot for Nu when the end faces
were insulated or adiabatic. It can be seen that for all Ra and over all the entire range of L/D
ip

the Nu for the solid cylinder in air is marginally higher than that of the case when the cylinder
r

is on ground. At Ra of 108 the Nu at L/D = 0.1 is same for the case of a solid cylinder in air or
sc

on ground. As L/D increases Nu for the case of solid cylinder on ground becomes less and
nu

less and the difference to that of a solid cylinder in air becomes more and more. It can be
Ma

marked from Fig.7 that the average Nu for this case is always less in comparison to the case
of a solid cylinder when the end faces were insulated. When the end faces are isothermal the
cylinder losses more heat but the area over which the heat transfer occurs has increased so as
ed

result the Nu can be less compared to the case of a solid cylinder with end faces insulated.
pt
ce
Ac

Fig. 9. Average Nu for a solid cylinder in air or ground when end faces are isothermal

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

It is to be noted that only in this section we are showing the case of a solid cylinder that has
isothermal end faces otherwise elsewhere in the section we only consider the case of a solid
cylinder with end faces insulated, if not mentioned otherwise.

7.3 Solid and hollow cylinder in air

In the previous section we compared the Nu for a solid cylinder when in air or ground and a
similar comparison was shown for the hollow cylinder. In this section we would compare the

d
ite
Nu between a solid and hollow cylinder when in air. Fig. 10 shows the Nu as a function of
L/D. Up to a Ra of 107 the Nu for a hollow cylinder is lower compared to that of a solid

ed
cylinder for all the range of L/D going from 0.05 to 20. The hollow cylinder losses heat from
its inner surface as well as from its outer surface whereas the solid cylinder losses heat from

py
its outer surface only (ends are adiabatic). So the heat loss from the hollow cylinder is near to

Co
double that of the solid cylinder but the heat transfer area for the hollow cylinder is also
double, that of the solid cylinder. As a result the Nu from for a hollow cylinder falls
marginally short compared to that of a solid cylinder.
ot
tN
r ip
sc
nu
Ma
ed

Fig. 10. Comparison of Nu between a solid and hollow cylinder in air

pt
ce

When the length of the hollow cylinder increases (L/D more) the natural convection in the
Ac

inside of the cylinder becomes weak resulting in poor heat transfer from the inner surface
causing the Nu to be much less compared to a solid cylinder which we see in Fig.10 clearly at
high L/D. As Ra increases the difference in Nu between the solid and the hollow cylinder at
high L/D becomes even more prominent. However, at high Ra of 108 and low L/D the hollow
cylinder has more Nu compared to that of a solid cylinder. When the length of the cylinder is
short (L/D too less) the convection plume enters the inside of the cylinder very easily and

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

helps to take away more heat from the inner surface and this phenomenon is just absent in the
case of a solid cylinder. So this is exactly called the edge effect and this helps to improve the
Nu for the case of a hollow cylinder over the solid cylinder up to an L/D of 0.2. However,
after an L/D of 0.2 the inner convection in the hollow cylinder becomes weak and that results
in lower heat loss from the inner portion and hence the overall Nu becomes less compared to
that of a solid cylinder.

d
7.4 Solid and hollow cylinder on ground

ite
ed
py
Co
ot
tN
ip

Fig. 11. Comparison of Nu between a solid and hollow cylinder on ground

r
sc

Fig.11 shows a comparison of Nu between a solid and hollow cylinder when they are placed
nu

on ground. Up to a Ra of 106 the Nu for a hollow cylinder is marginally lower compared to

Ma

that of a solid cylinder. As L/D increases the difference between the Nu for a solid and
hollow cylinder rises. After a Ra of 107 the difference becomes even more prominent. The
ed

reason for this is given in the previous section which still holds here. For higher Ra (more
than 107) at lower L/D the Nu for a hollow cylinder is more compared to that of a solid
pt

cylinder. At a Ra of 108 the hollow cylinder has higher Nu compared to that of a solid
ce

cylinder up to an L/D of 5. When on ground the solid cylinder cannot lose heat from its
bottom part more effectively whereas the hollow cylinder can lose heat from its inner portion
Ac

more effectively due to buoyancy currents entering into the hollow cylinder. So the Nu for a
hollow cylinder can be higher compared to a solid cylinder at high Ra and low L/D less than
5. But for high L/D more than 5 the heat loss from the inner portion of the hollow cylinder
reduces due to poor buoyancy currents entering into the cylinder, as a result the overall Nu
falls down and becomes less than that of the solid cylinder.

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

7.5 Heat loss from hollow cylinder in air or ground

d
ite
ed
Fig. 12. Comparison of heat loss from the outer and Fig. 13. Comparison of heat loss from the outer and inner
inner surface of a hollow cylinder when placed in air surface of a hollow cylinder when placed on ground

py
Co
Fig. 12 and Fig. 13 show the heat loss from the inner and outer surface of a hollow cylinder
when it is either on ground or in air. For the purpose of reproducing Q one has to use surface
temperature of the cylinder to be 326K and the ambient temperature to be 300K, whereas the
ot
thermal properties to find Ra has to be evaluated at the film temperature. For a low Ra of 104
tN

the heat loss from the inner surface is less compared to that of the outer when the cylinder is
ip

in air. But when, the cylinder is on ground the heat loss from the inner and outer are about the
same until an L/D of 0.4. When, L/D becomes more the inner surface losses less heat because
r
sc

the convection current entering the inside of the hollow cylinder becomes weak. So we see a
very large difference between the heat loss from the inner to outer surface at low Ra. When
nu

Ra increase the overall buoyancy or natural convection plume becomes stronger and that
Ma

helps to remove more heat from the inner surface of the cylinder causing the difference
between the inner and outer heat loss to be less and less with Ra and L/D as is clear from Fig.
ed

12 and Fig.13. When the hollow cylinder is on ground the inner surface losses more heat
compared to the outer surface up to an L/D of 0.8 for Ra ranging from 106 to 108. Since the
pt

bottom surface of the cylinder cannot lose heat effectively due to the presence of the ground
ce

whereas the inner surface can lose heat due to the buoyancy plume penetrating into the inside
Ac

portion of the cylinder so such a cross over can happen for lower L/D. As L/D increases the
heat loss from the inner surface falls because the penetrating buoyancy plume into the inside
of the cylinder becomes weak and the heat loss from the outer surface remains high.

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

7.6 Velocity and thermal plume around a hollow or solid cylinder

It is quite interesting to see the velocity plume around a hollow cylinder when it is on ground.
Such pictures are normally not found in any literature, so we intend to show this here since
this would help to enhance the imagination of the readers.

d
ite
ed
py
Co
ot
tN
r ip
sc

(a) (b)
nu
Ma
ed
pt
ce
Ac

(c) (d)

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

d
ite
ed
py
Co
(e) 0 0.035 0.07 0.105 0.14 0.175 0.21 0.245 0.28
(f)
ot
Fig. 14. Velocity vector around a hollow cylinder placed on ground for Ra= 106, L/D (a) 0.5, (c) 1 and (e) 2
tN
Fig. 14 (a) , (c ) and (e) show the velocity vector around a hollow cylinder for different L/D
ratio on the longitudinal cross sectional plane while Fig. 14 (b), (d) and (f) show the plume on
ip

the transversal mid cross sectional plane. It can be seen from the vector plot that the
r

surroundings air enter the inside of the hollow cylinder from both its side ends and exit also
sc

from the side ends. This penetrating buoyant plume helps to remove heat from the bottom
nu

part of the cylinder wall and also the exiting plume takes heat from the top part of the
cylinder. The plume visualization on the mid cross sectional plane looks quite interesting.
Ma

The rising plume inside the cylinder forms two rolls near the top wall of the cylinder for L/D
more than 1 while for L/D of 0.5 there is no roll formation inside the cylinder, since it is too
ed

short. This helps to remove more heat from the hollow cylinder since roll formation would
pt

hinder the heat removal. This is the main reason why for a short cylinder the Nu is higher
compared to a long cylinder. When the cylinder is in air the plume around the cylinder looks
ce

almost similar to what we have shown when it is on ground only with the exception of the
Ac

plume entry at the side edges of the cylinder. So we do not show them here for saving space.

The most interesting part is to see the thermal plume for short cylinders when they are either
on ground or in air. We show that here for the case of Ra 106 for a hollow cylinder of various
L/D.

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

d
ite
ed
(a) (b) (c)

py
L/D =1 L/D =2 L/D =5 (a’) (b’) (c’)

Co
301 303.5 306 308.5 311 313.5 316 318.5 321 323.5 326
ot
Fig. 15. Temperature plume around a hollow cylinder at Ra=106 and L/D (a) 1, (b) 2, and (c) 5 when the cylinder is in air
tN
r ip
sc
nu
Ma

(a) (b) (c)

L/D =1 L/D =2 L/D =5 (a’) (b’) (c’)
ed

Fig. 16. Temperature plume around a hollow cylinder at Ra=106 and L/D (a) 1, (b) 2, and (c) 5 when the cylinder is on
pt

ground
ce

Fig.15 and Fig.16 show the thermal plume around a hollow cylinder while the figures with
prime (for example a’ corresponds to a) show the plume on a mid-transversal cross section of
Ac

the cylinder. From the picture it is clear that the atmospheric air enters the side faces of the
cylinder and leaves the side faces so the blue zone of atmospheric temperature forms inside
the cylinder. The plume which rises inside the cylinder takes away the heat from the top part
of the cylinder. For L/D of 5 the air cannot enter up to the center of the cylinder so the core
remains hot. When the cylinder is in air the core seems to be relatively cooler compared to

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

 Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

the case when it is on ground and this is the reason the cylinder in air losses more heat
compared to the case when it is on ground

301 303.5 306 308.5 311 313.5 316 318.5 321 323.5 326
d
ite
(a) (b) (c) (d)

ed
L/D=1 L/D=2 L/D=1 L/D=2

py
Fig. 17. Thermal plume around a short solid cylinder in air (a-b) and ground (c-d), Ra=106

Co
ot
tN
ip

0.035 0.07 0.105 0.14 0.175 0.21 0.245 0.28

r
sc
nu
Ma
ed

0
pt
ce

1
Ac

Fig. 18. velocity vector around a short solid cylinder in air (a-b) and ground (c-d), Ra=106

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

Fig.17 show the thermal plume for a solid cylinder of L/D =1 and 2 when it is either is air or
on ground. When, it is in air it losses heat from its bottom part so a thermal layer at the
bottom part can be seen (green color). But when the solid cylinder is on ground it does not
lose any heat from its bottom part and hence no formation of thermal layer. A striking feature
in the velocity field can be seen (Fig.18) when the plume tries to glide along both the side
faces of the cylinder. When in air the gliding velocity along the side face is more compared to

d
the case when it is on ground. Otherwise the look of the velocity field in both the cases is

ite
almost similar to each other, with the fluid being entrained towards the solid cylinder from
the side ambient and rising on top of the cylinder as a thick plume.

ed
Fig.19 shows the thermal buoyant plume around a solid or hollow cylinder of L/D =1 when it

py
is either in air or on ground for two widely varying Ra of 104 and 107. Such types of figures

Co
or analysis are almost absent in literature so it is prudent to show it in achieves which we plan
to do here. The figures are put into a table so that reference to the figure would be a simple
ot
task without calling a figure number. The first row shows a solid cylinder in air and the effect
of Ra on the plume. When Ra is low the plume is too thick around the cylinder and for high
tN

Ra it is thin at the bottom and then breaks on the top to four major plumes. Column two and
ip

four show the view from the side or it is a view in transversal mid cross sectional plane. The
r

second row shows the thermal plume for the solid cylinder when it is on ground. Clearly the
sc

plume is thicker here compared to the case of the cylinder when it was in air and for this
nu

reason the net heat loss from this was becoming little lower. The third row shows the plume
for a hollow cylinder in air. The surroundings air gets sucked in to the cylinder and then
Ma

comes out of it through its side face as another plume for which the plume looks to be thicker
on the top part of the cylinder for low Ra of 104. The air is sucked until the mid of the
ed

cylinder but does not effectively take away the heat, for which the core remains hot. But
when Ra is 107 the suction velocity into the core of the cylinder is high and heat loss is also
pt

high from the cylinder wall (the plume gets out of the inside of the cylinder at a higher
ce

velocity) as a result the core remains much cooler. The plume on top of the cylinder breaks
Ac

into several smaller plumes which rise to the top of the domain. The fourth row show the
thermal plume for the hollow cylinder when it is on ground. The cylinder has thicker thermal
layer as it is connected with the ground and this is the reason for which it loses less heat at
low Ra. At high Ra the plume goes into the side face of the cylinder and takes away heat
from the cylinder wall and leaves the side-face again at high velocity for which the core
remains much cooler compared to a low Ra case.

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

7.6 Effect of Ra on buoyant plume around a short cylinder

Ra=104 , L/D =1 Ra=107, L/D=1

d
ite
301 303.5 306 308.5 311 313.5 316 318.5 321 323.5 326
ed
py
Co
ot
tN
r ip
sc
nu
Ma
ed
pt
ce

Fig. 19. Thermal buoyant plume around a solid and hollow cylinder in air or ground
Ac

8. Development of correlations for Nu

There are several empirical correlations in the literature for Nu which we have stated in our
literature review section. However, correlations of Nu for low L/D are not available for which
we plan to present this here. Moreover, for a hollow cylinder constancy in Nu is not even
attended at an L/D of 20 which happens easily for a solid cylinder. So a correlation of Nu

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

against Ra and L/D would be beneficial to the academicians as well as to the practicing
engineers in the industries. In order to develop a numerical correlation we have performed
CFD studies on a solid or hollow cylinder and then found out the Nu as a function of Ra and
L/D. Then we tried to fit the best equation to it in a polynomial form with power indices.
Such types of equations have very low error in it when we try to predict back the Nu (predict)
to match with the computed value. Correlations of Nu are developed for a solid or hollow
cylinder either in air or on ground in the range of Ra going from 104 to 108 and L/D in the

d
ite
range of 0.1 to 20 for a cylinder on ground and 0.05 to 20 for a cylinder in air.

ed
Here the accuracy of such a relation is shown which has been developed for a solid cylinder
in air with 90 computational cases. It can be seen from Fig.20 that 82 out of 90 data points

py
are in the accuracy range of ±6%. Similarly other Nu correlations for solid cylinder on

Co
ground and hollow cylinder in air and ground are developed and all of them show a very
similar picture like that of Fig.20 with accuracy range of ±6%, for which we do not show
ot
them here for the sake of saving space.Table-1 shows the correlations of Nu developed for all
tN
the cases along with the constants used in it which can be certainly very beneficial to the
academics and industries.
r ip
sc
nu
Ma
ed
pt
ce
Ac

Fig. 20. A comparison of predicted Nu with that of the computed value for a solid cylinder in air

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

Table-1

Nu correlations for the cylinder for various cases, L/D valid in the range 0.1 to 20 for cylinder on ground and 0.05 to 20 for
cylinder in air and Ra in the range of 104 to 108

Type of cylinder Nu correlation Constants of the correlation

Solid in air 𝐿𝐿 𝐶𝐶3 𝐴𝐴1 = −0.1, 𝐴𝐴2 = 0.059, 𝐴𝐴3
𝑁𝑁𝑁𝑁 = 𝐴𝐴1 𝑅𝑅𝑅𝑅𝐵𝐵1 �𝐶𝐶1 − 𝐶𝐶2 � � � = 0.998, 𝐵𝐵1 = 0.08418, 𝐶𝐶1
𝐷𝐷
= −31.9084, 𝐶𝐶2 = 1.288, 𝐶𝐶3
𝐿𝐿 𝐸𝐸1
+ 𝐴𝐴2 𝑅𝑅𝑅𝑅 𝐷𝐷1
− 𝐴𝐴3 � � = −1.4236, 𝐷𝐷1 = 0.34861, 𝐸𝐸1
𝐷𝐷 = 0.5239

d
ite
Solid on ground 𝐴𝐴1 = −10, 𝐴𝐴2 = −15, 𝐴𝐴3
= 0.1, 𝐵𝐵1
𝐿𝐿 𝐶𝐶3 = 0.17307, 𝐶𝐶1

ed
𝑁𝑁𝑁𝑁 = 𝐴𝐴1 𝑅𝑅𝑅𝑅𝐵𝐵1 �𝐶𝐶1 − 𝐶𝐶2 � � � = −1.5166, 𝐶𝐶2
𝐷𝐷
𝐿𝐿 = 0.0199, 𝐶𝐶3

py
+ 𝐴𝐴2 𝑅𝑅𝑅𝑅𝐷𝐷1 − 𝐴𝐴3 � � = −0.7226, 𝐷𝐷1
𝐷𝐷 = 0.16765

Co
Hollow in air 𝐿𝐿 𝐶𝐶3 𝐴𝐴1 = −0.1, 𝐴𝐴2 = −0.6, 𝐴𝐴3
𝑁𝑁𝑁𝑁 = 𝐴𝐴1 𝑅𝑅𝑅𝑅𝐵𝐵1 �𝐶𝐶1 − 𝐶𝐶2 � � � = 0.998, 𝐵𝐵1 = 0.25966, 𝐶𝐶1
𝐷𝐷
= −5.31948, 𝐶𝐶2 = 0.998, 𝐶𝐶3
𝐿𝐿 𝐸𝐸1
ot
+ 𝐴𝐴2 𝑅𝑅𝑅𝑅𝐷𝐷1 + 𝐴𝐴3 � � = −0.38213, 𝐷𝐷1 = 0.20876, 𝐸𝐸1
𝐷𝐷 = −0.95664
tN
Hollow on ground 𝐴𝐴1 = −0.1, 𝐴𝐴2 = −0.6, 𝐴𝐴3
𝐶𝐶3 = 0.998, 𝐵𝐵1 = 0.41168, 𝐶𝐶1
𝐿𝐿
𝑁𝑁𝑁𝑁 = 𝐴𝐴1 𝑅𝑅𝑅𝑅𝐵𝐵1 �𝐶𝐶1 − 𝐶𝐶2 � � � = −5.58372, 𝐶𝐶2 = 1.3, 𝐶𝐶3
ip

𝐷𝐷
= −0.02321, 𝐷𝐷1 = 0.41729, 𝐸𝐸1
𝐿𝐿 𝐸𝐸1
r

+ 𝐴𝐴2 𝑅𝑅𝑅𝑅𝐷𝐷1 + � � = −0.990177

𝐷𝐷
sc
nu

9. Conclusions
Ma

Three dimensional conservation equations for mass, momentum and energy have been solved
to predict the natural convection flow and temperature field around a solid or hollow cylinder
ed

either in air or on ground. The solution shows very nice thermal and flow plumes around a
pt

short and long cylinder which have not been reported in a pictorial manner in the literature so
ce

far. From the 3D CFD analysis the following things can be concluded:

•
Ac

A solid cylinder in air attains constancy in Nu at an L/D around 1 for up to a Ra of 108

whereas on ground it attains constancy in Nu around L/D of 1 only up to a Ra of 107
and for Ra of 108 the constancy in Nu comes at much larger L/D.

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

• For a hollow cylinder in air or ground the constancy in Nu comes at around L/D of 1
only for Ra up to 104 and for higher Ra the constancy in Nu is never achieved even up
to an L/D of 20.

• The heat loss from the outer surface of a hollow cylinder suspended in air is always
more compared to that of the inner surface for all Ra up to 108 for a range of L/D

d
between 0.1 to 20 but for L/D within 0.1 the inner surface marginally loses more heat

ite
compared to the outer one when Ra is more than 104. When the hollow cylinder is on
ground the inner surface loses more heat compared to the outer surface for Ra more

ed
than 106 up to an L/D of 0.6 and after an L/D of 1 the outer surface loses more heat

py
compared to the inner one.

Co
• The heat loss from the inner surface of the hollow cylinder either in air or on ground
becomes constant at L/D of 1 for low Ra of 104. As Ra increases the constancy in heat
ot
loss is attended at higher and higher L/D, whereas the heat loss from the outer surface
tN
goes on increasing with Ra and L/D and never attains a constant value.
ip

• Up to a Ra of 106 the Nu for a solid cylinder on ground is marginally higher than that
r
sc

of the hollow cylinder on ground for L/D up to 20. For Ra of 107 a hollow cylinder on
ground has marginally higher Nu compared to a solid cylinder on ground up to an L/D
nu

of 0.5 and after that the situation reverses. But at a Ra of 108 a hollow cylinder on
Ma

ground has higher Nu compared to that of a solid cylinder on ground up to an L/D of

5 and after that the situation reverses.
ed
pt
ce
Ac

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

References :

1. W. Ayrton, H. Kilgour, (1892) The thermal emissivity of thin wires in air, Philos TR
Soc A, vol. 183, pp. 371–405
2. J. Petavel, (1898) On the heat dissipated by a platinum surface at high temperatures,
Philos T R Soc A, vol. 191: pp. 501–524
3. A. Kennelly, C. Wright, JV. Bylevelt, (1909) The convection of heat from small
copper wires, Trans Amer Inst Elec Eng, vol. 28: pp. 363–393
4. I. Langmuir, (1912) The convection and conduction of heat in gases, Trans Am Inst
Electr En, vol. 31:pp. 1229–1240

d
5. A. Davis, (1922) Natural convective cooling in fluids, Phil Mag¸ vol. 44: pp. 920–940

ite
6. R. Nelson, (1924) Free convection heat in liquids, Phys Rev, vol. 23: pp. 94–103
7. W. Nusselt, (1929) Heat dissipation from horizontal tubes and wires to gases and

ed
liquids, Ver DeutIng, vol. 73: pp. 1475–1478
8. C. Lander, (1942) A review of recent progress in heat transfer, J Inst Mech Eng, vol.

py
148:pp. 81–112
9. T. Tsubouchi, H. Masuda, (1966–1967) Natural convection heat transfer from a

Co
horizontal circular cylinder with small rectangular grooves, Sci Rep Res Inst Tohaku
Univ Ser B, vol. 18: pp. 211–242
10. WR. Penney, TB. Jefferson, (1966) Heat transfer from an oscillating horizontal wire to
water and ethylene glycol, J Heat Trans, vol. 88, pp. 359–366 ot
11. D. Saville, S. Churchill, (1967) Laminar free convection in boundary layers near
tN
horizontal cylinders and vertical axisymmetric bodies, J Fluid Mech, vol. 29, pp. 391–
399
12. S. Churchill, H. Chu, (1975) Correlating equations for laminar and turbulent free
ip

convection from a horizontal cylinder, Int J Heat Mass Trans, vol. 18, pp. 1049–1053
13. V. T. Morgan, The Overal Convective Heat Transfer from Smooth Circular Cylinders,
r
sc

Academic Press, New York,1975.

14. S. Nakai, T. Okazaki, (1975) Heat transfer form a horizontal circular wire at small
Reynolds and Grashof numbers-I, Int J Heat Mass Trans, vol. 18, pp. 387–396
nu

15. J. Merkin, (1977) Free convection boundary layers on cylinders of elliptic cross
section, ASME J Heat Trans, vol. 99, pp. 453–457
Ma

16. T. Kuehn, R. Goldstein, (1980) Numerical solution to the Navier - Stokes equations for
laminar natural convection about a horizontal isothermal circular cylinder, Int J Heat
Mass Trans, vol. 23, pp. 971–979
ed

17. B. Farouk, SI. Güçeri, (1981) Natural convection from a horizontal cylinder-laminar
regime, J Heat Trans, vol. 103, pp. 522–527
pt

18. L. De Socio, (1983) Laminar free convection around horizontal circular cylinders, Int J
Heat Mass Trans, vol. 26, pp. 1669–1677
ce

19. M. Al-Arabi, M. Khamis, (1982) Natural convection heat transfer from inclined
cylinders, Int J Heat Mass Trans, vol. 25:3–15
Ac

20. P. Wang, R. Khawita, D. L. Nguyen, (1990) Transient laminar natural convection from
horizontal cylinders, International Journal of Heat and Mass Transfer, vol. 34, pp.
1429-1442
21. T. Saitoh, T. Sajiki, K. Maruhara, (1993) Bench mark solutions to natural convection
heat transfer problem around a horizontal circular cylinder, Int J Heat Mass Trans, vol.
36, pp. 1251–1259
22. T. Chouikh, A. Guizani, M. Maalej, (1997) Numerical study of laminar natural
convection flow around horizontal isothermal cylinder, Renewable Energy, vol. 13, pp.
71–78

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

23. M.M. Molla, M.A. Hossain, R.S.R. Gorla, (2005) natural convection flow from
isothermal horizontal circular cylinder with temperature dependent viscosity, Heat
Mass Transfer, vol. 41, pp. 594 - 598
24. M.M. Molla, S.C. Paul, A. Hossain, (2008) Natural Convection Flow for a horizontal
cylinder with uniform heat flux in presence of heat generation, Applied Mathematical
Modeling, vol. 33, pp. 3226 - 3236
25. SO. Atayilmaz, I. Teke, (2009) Experimental and numerical study of the natural
convection from a heated horizontal cylinder, Int Commu Heat and Mass, vol. 36, pp.
731–738

d
26. M. Ashjaee, S. Yazdani, S. Bigham, T. Yousefi, (2011) Experimental and numerical
investigation on free convection from horizontal cylinder located above an adiabatic

ite
surface, Heat Transfer Engineering, vol. 33, pp. 213–214
27. A.A. Mehrizi, M. Farhadi, H.H. Afrouzi, S. Shayamehr, H. Lotfizadeh, (2013) Lattice

ed
Boltzmann simulation of natural convection flow around a horizontal cylinder located
beneath an insulated plate, Journal of Theoretical and Applied Physics, vol. 51, pp.

py
729–739
28. J.P Kuehner, A.M Hamed, J.D Mitchell, (2015) Experimental investigation of free

Co
convection velocity boundary layer and plume formation region for a heated
horizontal cylinder, International Journal of Heat and Mass Transfer, vol. 82, pp. 78–
97
ot
29. M.H Sedaghat, M. Yaghoubi, M.J. Mafhrebi, (2015) Analysis of natural convection
heat transfer from a cylinder enclosed in a corner of two adiabatic wall, Experimental
tN
Thermal and Fluid Science, vol. 62, pp. 78–97
30. Na. Tsung-Yen, (1993) Effect of wall condition on natural convection over a vertical
slender hollow circular cylinder, Applied Scientific Research, vol. 54, pp. 39–50
ip

31. L.C. Chang, (2008) Numerical simulation of natural convection of micro polar fluid
flow along slender hollow cylinder withy wall conduction effect, Non linear Science
r
sc

and Numerical Simulation, vol. 13, pp. 9-20

32. M.M.H Mamun, M.M Rahman, M.M Billah, R Saidur, (2010) A numerical study on
the effect of a hollow cylinder on mixed convection in a ventilated cavity, International
nu

Communication in Heat and Mass Transfer, vol. 37, pp. 1326–1334

33. M.M Billah, M.M Rahman, M.U. Sharif, N.A. Rahim, R. Saidur, M. Hassanuzzaman,
Ma

(2011) Numerical analysis of fluid flow due to mixed convection in a li driven cavity
having a heated circular hollow cylinder, International Communication in Heat and
Mass Transfer, vol. 38, pp. 1093-1103
ed

34. H.P. Rani, G.J. Reddy, (2011) Conjugate transient free convective heat transfer from a
vertical slender hollow cylinder with heat generation effect, Applied Mathematics, vol.
pt

2, pp. 90-98
35. J.R. Senapati, S.K. Dash and Roy, (2016) Numerical investigation of natural
ce

convection heat transfer over annular finned horizontal cylinder, Int Journal of Heat
and Mass Transfer, Vol. 96, pp. 330-345.
Ac

36. J.R. Senapati , S.K. Dash and S. Roy, (2016), 3D numerical study of the effect of
eccentricity on heat transfer characteristics over horizontal cylinder fitted with annular
fins, International Journal of Thermal sciences, Vol. 108, pp. 28-39
37. Ansys Fluent, Release 15.0, user manual

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-

Journal of Heat Transfer. Received December 07, 2016;
Accepted manuscript posted February 8, 2017. doi:10.1115/1.4035919
Copyright (c) 2017 by ASME

Figure captions list :

Fig. 1. Flow vector around a (a) hollow and (b) solid isothermal cylinder on ground, L/D =1, Ra=10

Fig. 2. (a) A solid or thin hollow cylinder in air with the computational domain around it, (b) solid or the thin
hollow cylinder lying on ground with its computational domain (a schematic representation)

Fig. 3. Influence of domain height on average Nu for a solid cylinder in air

Fig. 4. (a) cell arrangement for hollow cylinder in air, (b) cross sectional view, (c) Blown up view near the
cylinder wall

d
Fig. 5. Average Nu for a hollow cylinder in air as a function of number of cells

ite
Fig. 6. Average Nu for a solid cylinder in air, a comparison with Experimental correlation

ed
Fig. 7. Average Nu for a solid cylinder in air or lying on ground as a function of L/D

py
Fig. 7a. Average Nu for a hollow cylinder in air or lying on ground as a function of L/D

Fig. 8. Temperature plume near a hollow cylinder on (a) ground and in (b) air, Ra = 108, L/D = 5

Co
Fig. 9. Average Nu for a solid cylinder in air or ground when end faces are isothermal

Fig. 10. Comparison of Nu between a solid and hollow cylinder in air ot

Fig. 11. Comparison of Nu between a solid and hollow cylinder on ground
tN

Fig. 12. Comparison of heat loss from the outer and inner surface of a hollow cylinder when placed in air
ip

Fig. 13. Comparison of heat loss from the outer and inner surface of a hollow cylinder when placed on ground
r

Fig. 14. Velocity vector around a hollow cylinder placed on ground for Ra= 106, L/D (a) 0.5, (c) 1 and (e) 2
sc

Fig. 15. Temperature plume around a hollow cylinder at Ra=106 and L/D (a) 1, (b) 2, and (c) 5 when the
nu

cylinder is in air

Fig. 16. Temperature plume around a hollow cylinder at Ra=106 and L/D (a) 1, (b) 2, and (c) 5 when the
Ma

cylinder is on ground

Fig. 17. Thermal plume around a short solid cylinder in air (a-b) and ground (c-d), Ra=106
ed

Fig. 18. velocity vector around a short solid cylinder in air (a-b) and ground (c-d), Ra=106

Fig. 19. Thermal buoyant plume around a solid and hollow cylinder in air or ground
pt

Fig. 20. A comparison of predicted Nu with that of the computed value for a solid cylinder in air
ce
Ac

Table caption list :

Table 1. Nu correlations for the cylinder for various cases, L/D valid in the range 0.1 to 20 for cylinder on
ground and 0.05 to 20 for cylinder in air and Ra in the range of 104 to 108

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jhtrao/0/ on 02/14/2017 Terms of Use: http://www.asme.org/about-