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

net/publication/233365496

Experimental and Numerical Study of Free Convection on an Isothermal

Downward Cone

Article  in  Experimental Heat Transfer · October 2007

DOI: 10.1080/08916150701418286

CITATIONS READS

6 391

4 authors, including:

Mehdi Ashjaee Tooraj Yousefi

University of Tehran Razi University
181 PUBLICATIONS   1,214 CITATIONS    36 PUBLICATIONS   704 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Thermophysical Properties Measurment of Hybrid Nanofluid View project

Haydrolic electrical jacks View project

All content following this page was uploaded by Mehdi Ashjaee on 01 January 2014.

The user has requested enhancement of the downloaded file.

Proceedings of the 4th WSEAS Int. Conf. on HEAT TRANSFER, THERMAL ENGINEERING and ENVIRONMENT, Elounda, Greece, August 21-23, 2006 (pp159-165)

Experimental and Numerical Study of Free Convection on an Isothermal

Downward Cone
M. ASHJAEE, T. YOUSEFI, M. ARZAGHI, M. JARAHI
Mechanical Engineering Department
University of Tehran
Tehran 11365-4563
IRAN

Abstract: The laminar free convection heat transfer from an isothermal downward cone is investigated
experimentally and numerically. The experimental investigation is carried out by Mach-Zehnder interferometry
technique and the numerical simulation was done by Fluent. The cone tip angle has been kept constant to 45˚ and it
has been suspended from its base throughout the experiment. This paper focuses on the effect of Rayleigh number
variation on the local and average free convection heat transfer coefficient over the conical surface. The local and
average Nusselt numbers were determined for the Rayleigh number range of 4.9 × 10 5 to 1.1 × 10 6 . Also the experiment
and the numerical simulation were carried out on a vertical isothermal cylinder of circular cross section in order to
compare results with other researchers for the verification of our experimental and numerical results. The significant
influence of the upper end surface of the cone, both in the experimental and numerical studies indicated a
recirculation region above the upper end surface which affects the local convection heat transfer at the slant trailing
edge and causes it to increase. Also a correlation for the calculation of the local Nusselt number over the cone is
proposed.

Key-Words: Free Convection, Downward Cone, Mach-Zehnder, Interferometry, Vertical Cylinder, Conical Surface

1. Introduction
The industrial applications of convection heat transfer experimental results for the average heat transfer rates
from cones could be found in civil engineering, from the cones with different tip angles, and the two
chemical industry (bottom of tanks), and electronics limiting cases of vertical cylinder and circular disc as
(transistors, resistors, diodes, lamps). special cases. In their analytical work, they assumed
The problem of laminar free convection flow along a the parabolic temperature profile in the boundary layer
vertical faced up or down cone has been treated which is applicable only to thin boundary layer flow
theoretically in the literature by many researchers and should be modified to take into account the effect
since 1953. Merk and Prins [1] found the similarity of thick boundary layer resulting at low Rayleigh
solution for the case of an isothermal cone whereas numbers. Besides, several mixed convection studies
Hering and Grosh [2] have obtained a number of from cones are done by other researchers.
similarity solutions for cones with prescribed wall An experimental measurement of mean heat transfer
temperatures being a power function of the distance rates from vertical cones has been done by P.H.
from the apex along the generator. Further results were Oosthuizen and E. Donaldson [10] in which the model
obtained by Hering [3] and Sparrow and Guinle [4] for has been treated as a lumped object and they have
small values of Prandtl number (Pr) and by Roy [5] for concentrated on finding the average free convection
large values of Prandtl number, respectively. Also, heat transfer coefficient. Natural convection heat
Alamgir [6] has investigated the overall heat transfer transfer from vertical cylinder in the unbounded space
from vertical cones using the integral method. Further, has been investigated experimentally and numerically,
Pop and Takhar [7] have studied the compressibility previously [11].
effects in laminar free convection from a vertical cone, A literature search reveals that no experimental studies
while Hossain and Paul [8] have considered the effect have been carried out for the determination of the local
of suction and injection when the cone surface is heat transfer coefficients on the surface of downward
permeable. In all these analytical papers, the isothermal cones where the boundary layer thickness is
boundary-layer thickness is small compared to the comparable to the diameter of the cone.
local radius of the cone. Also, Lewandowski et al. [9] The present study focuses on finding the variation of
studied the problem of laminar free convection from the local and average heat transfer coefficients on the
an upward cone theoretically. Their work also include surface of a downward isothermal cone with respect to

1
Proceedings of the 4th WSEAS Int. Conf. on HEAT TRANSFER, THERMAL ENGINEERING and ENVIRONMENT, Elounda, Greece, August 21-23, 2006 (pp159-165)

different Rayleigh numbers. The study is carried out

experimentally using Mach-Zehnder interferometer
and numerically by Fluent.

2. Experimental Setup

2.1 Interferometer
A Mach-Zehnder Interferometer (MZI) with
100 mm optics was used in the experimental study. Fig.1 Plan view of the Mach-Zehnder interferometer
The output of an interferometer is an interference
fringe field that is related directly to the variation of
the refractive index in the test model. Although
restricted to fluids that are transparent, this optical
technique can give nonintrusive temperature field
measurement without calibration and with no thermal
inertia. A basic MZI consists of a light source, two
mirrors and two beam splitters. Figure 1 shows a
schematic of the interferometer used in the
experiment. Beam splitters BS1 and BS2, along with
plane mirrors M1 and M2 constituted the basic MZI.
Further information about MZI can be found in [27,
28, 29]. The light source which was used is a 100mW
Helium-Neon laser, λ = 632.8nm .All the MZI
components were positioned so that the light beam
Fig.2 A sample Interferogram of the vertical cone at
was incident on each of them at an angle of 30 D . This finite fringe mode
is different from the conventional MZI setting of 45 D .
The reason for using 30 D is that a wider beam can be
obtained with this angle, thereby increasing the field of 2.2 Experiment Test Section
view. All the interferograms were digitized with a The tested vertical cylinder and cone were made of
"Panasonic WV-CP410" 1/3" CCD camera with aluminum and are shown schematically in figures 3
440000 pixel. To acquire the interferogram the camera and 4 respectively. The cone base diameter is
was connected to a video recorder through a PC. At 43mm with the tip angle of 45 D . The length and
each surface temperature we recorded three diameter of the vertical cylinder was chosen
interferograms. Figure 2 shows one of the images 12 .7 mm and 92 mm respectively in order to provide
which are recorded by the CCD camera. For results comparable to the previous work [26, 32] for
quantitative measurement of local free convection heat verifying the method of the experimental data
transfer coefficients over lateral surface of the cone, a reduction which has been used. Also wooden end caps
reference fringe mode was employed adjusting the with thermal conductivity of 0.05 W mK were
mirrors of the interferometer so that the fringes outside
installed on the bases of both the cylinder and the cone
the thermal boundary layer extend perpendicular to the
to minimize the end effect.
axis of symmetry. The reason for applying reference
The surface of both objects was highly polished to
fringe mode in regions with low local heat transfer
assure smoothness and the reduction of the radiation
rates is that there is insufficient number of isotherms
heat transfer. Both objects, the cylinder and the cone,
near the surface, and it is desirable to measure the
were hollow; therefore a heater could be placed inside
surface temperature gradient directly from the
them in order to achieve different surface temperature.
interference fringe field, where the reference fringe
The vertical cylinder was filled with magnesium oxide
shifts represents the variation of refractive index in the
powder. By passing electricity through its heater by a
flow field.
20V − 2 A power supply and considering relatively
thick walled aluminum tube, we could achieve constant
cylinder surface temperature. For heating the cone thin
strip sheet heaters were wrapped on the surface of a

2
Proceedings of the 4th WSEAS Int. Conf. on HEAT TRANSFER, THERMAL ENGINEERING and ENVIRONMENT, Elounda, Greece, August 21-23, 2006 (pp159-165)

temperature over the surface of the cone. The variation

of temperature on both surfaces was below 0.3D C ,
which is reasonable for constant surface temperature
assumption on both models. The electrical power
supplied to the heater was controlled by a variable
transformer, therefore obtaining different surface
temperatures. The supplied electrical power was
measured via a digital wattmeter. The local surface
temperatures of the heated cylinder and cone were
measured via type K thermocouples, embedded in the
surface by means of pinned-junction method. The
thermocouples location on the cylinder was 50mm far
from one end, and on the cone it has been mounted at
15mm from its base. Two other thermocouples of the
same type were used to measure the ambient and the
reference temperatures for data reduction. All the
temperatures were monitored continuously by a
"TESTO 177" four channel data logger which was
Fig.3 Internal arrangement of the cylinder and connected to a PC. The thermocouples and the data
thermocouple locations logger set were calibrated before installation. In order
to calibrate the thermocouples at two data points, the
junction readings were recorded in the mixture of ice-
water and boiling water at the laboratory pressure. The
laboratory pressure and relative humidity were recorded
during all the experiments. The cone and the cylinder
were suspended from their bases such that their vertical
axis along the gravity was perpendicular to the laser
beam.
In order to eliminate the effect of any other air
disturbances on the experimental test section, the entire
setup was housed within a top open Plexiglas box of
dimensions 0.5 × 0.5 × 0 .5m . Also the box had two side
openings for the laser beam passage.

3. Experimental Data Reduction Method

The 3-D nature of the temperature field around both
objects, the vertical cylinder and the downward cone,
arise a need for a quantitative method other than 2-D
conventional data reduction method [12] to investigate
Fig4. Internal arrangement of the cone and the distribution of change of refractive index in the
thermocouple locations radial direction.
The direction of the laser beam which passes through
wooden cone with a cap assembly. The distances the cone thermal boundary layer is parallel to the X
between these strips were kept small enough to achieve axis, figure 5, which is being perpendicular to the cone
uniform surface temperature. The wooden wrapped axis Z .In this case we have a variation of index of
cone is then covered by a fire resistant cloth to prevent refraction in the radial direction because of the
the electrical contact between the heaters and the presence of a disturbance due to buoyancy induced
hollow aluminum cone. Finally, the wooden cone with flow over the cone and the optical path is no longer
the wrapped heater covered by the cloth was placed uniform. The fringe shift calculation for axisymmetric
inside the aluminum hollow cone. By passing electricity objects is fully explained by Ostrovsky [15], Hauf &
through the heater by a 30V − 10 A power supply and Grigull [12], and Vest [14]. The reference fringe
mode has been employed for measurement of
considering relatively thick walled aluminum conical
temperature gradient over lateral surface of the cone.
surface, we could achieve uniform and constant surface

3
Proceedings of the 4th WSEAS Int. Conf. on HEAT TRANSFER, THERMAL ENGINEERING and ENVIRONMENT, Elounda, Greece, August 21-23, 2006 (pp159-165)

In infinite fringe mode the information fringe shift is values of δ and integrate the function fitted to
gained in a discontinuous manner; i.e., δ = 0 , 0.5, 1, . . dδ (r ) / dr . If a proper method of differentiation
2 2

were applied, these functions are suitably well-

behaved and satisfy Eqs. (2) and (3). Another solution
is to fit curve to δ (r 2 ) then differentiate the curve and
calculate the integral. This procedure also needs a
great accuracy in curve fitting; whereas a little
deviation from the actual function of δ will result in
an incorrect temperature profile on the surface.
Finally, there are several transform methods which
have been referred to by Vest [[13].
Considering a transform method introduced by James
W. Bradley [16] for obtaining the density distribution
from an axisymmetric interferogram, let
Fig.5 Disturbed phase in a plane of the Z=constant t = 1 − y / R . From what is known of the properties
2 2

of the fringe shift, it is reasonable to assume that for

The fringe shifts δ is the difference, in wavelengths,
constant z, δ / t 0.5 can be expanded in a power series in
of the optical paths of two light beams, one of which
t:
passes through undisturbed and the other through a ∞
disturbed medium. Under the assumption that δ (t ) = t 0.5 ∑ Bnt n
n − n∞ = K ( ρ − ρ ∞ ), this difference is given by [15]: n =0

K x2 (4)
δ ( y, z ) =
λ ∫ x1
[ ρ ( x, y , z ) − ρ ∞ ]dx (1) Using a Laplace transform, Bradley [16] assumed that
for a constant z, δ / t 0.5 can be represented adequately
In a plane z=constant let the disturbance be bounded over the unit interval by, say, a cubic in t:
by the circle r =R. Then we see from the figure 5 that
along any light ray with y=constant, Eq. (1) becomes
δ / t 0.5 = B0 + B1t + B2t 2 + B3t 3
(5)
K R2 [ ρ ( r 2 ) − ρ ∞ ]dr 2
δ ( y2 ) =
λ ∫ y2 ( r 2 − y 2 )1 / 2
(2) We determine the coefficients by a least squares fit of
the fringe shift values and obtain ρ at once from
This is an Abel –type integral equation, with the Eq.(5):
solution: 1 ⎡ 3B 15 B2 2 35B3 3 ⎤
2 2 2 ρ − ρ∞ = ⎢ B0 ( 1 )t + ( )t + ( )t ⎥ (6)
λ R2 [dδ (r ) / dr ]dr 2A ⎣ 2 8 16 ⎦
ρ ( y 2 ) − ρ∞ = −
πK ∫y 2
(r 2 − y 2 )1 / 2
(3)
KR
To summarize, several methods are available for the where A = and finally the temperature profile
analysis of the above integral. In the usual reduction λ
process, classical methods, to obtain the density can be found as follows:
( −1)
distribution in a plane of constant z the equation can ⎛ 1 3.R0 .( ρ − ρ ∞ ) ⎞
be approximated by a finite summation involving the T = ⎜⎜ − ⎟⎟ (7)
⎝ T∞ 2.P∞ ⎠
discrete values of δ along the corresponding line of
constant z in the interferogram. These classical A MATLAB code has been developed to calculate the
methods have been explained by Ostrovsky [15], Hauf complete temperature distribution in the radial
& Grigull [12], and Vest [14]. All such methods are direction around the cone in seven imaginary
numerical approximations of equation (2) or (3). The horizontal planes which are perpendicular to the axis
axisymmetric distribution is divided into a number of of symmetry of the cone. The ineterferograms were
concentric annular zones, and integration is replaced read and the fringe shifts where calculated in 0.018mm
with summation. Thus, the task of finding the radial radial intervals within the test area by the method
described in [[13].
distribution of ρ ( y 2 ) consists of solving a system of
The Laplace transform was then applied to the
algebraic equations. Since Eq. (3) involves a derivative available fringe shifts in the same way that was
of δ , there are inherent dangers in such an outlined above to determine temperature distribution.
approximation. The problem can be alleviated, Consequently, the free convection heat transfer
however, by applying a differentiation scheme to the coefficient over the lateral surface is determined

4
Proceedings of the 4th WSEAS Int. Conf. on HEAT TRANSFER, THERMAL ENGINEERING and ENVIRONMENT, Elounda, Greece, August 21-23, 2006 (pp159-165)

locally by reading the temperature gradient at the 4. Numerical Simulation

surface as follow: The physical model under consideration is depicted in
dT 1 figure 7. A cone with tip angle θ and slant length L
hξ = − k ⋅ (8)
dr r =0 (Tw − T∞ ) suspended vertically downward. The Numerical
calculations are carried out using cylindrical
where h ξ is the local heat transfer coefficient, and ξ coordinate system (z, r ) with its origin located at the
is the distance measured from the tip of the cone and tip of the cone as indicated in the figure. The laminar,
the cylinder base along their lateral surfaces and k is steady state, two dimensional and segregated model
the air thermal conductivity. Therefore: was used from Fluent.
hξ ξ ξ dT
Nuξ = =− ⋅ (9)
k (Tw − T∞ ) dr r =0
Where ξ is the characteristic length which has been
the local distance from the cone tip.
The average free convection heat transfer
coefficient, hL , is calculated from:
L
1
hL =
L ∫
hξ ⋅ dξ
0
(10)

where L is the cone slant length. Fig. 7 Physical model and its boundary conditions
The thermophysical properties of the air were
estimated at film temperature, which is defined as: By employing the Boussinesq approximation for the
T + T∞ buoyancy force, the governing equations are given by:
Tf = w (11)
2
∂
For the calculation domain only one half of the Continuity: (ru ) + ∂ (rυ ) = 0 (12)
meridian plane of the cone is taken because of the ∂z ∂r
symmetry. ∂ ⎛ 2 ∂u ⎞ 1 ∂ ⎛ ∂u ⎞
⎜u − ν ⎟+ r ⎜υu − ν ⎟=
In order to check the accuracy of the experiment and Momentum: ∂ z ⎝ ∂z ⎠ r ∂r ⎝ ∂r ⎠ (13a)
the method of data reduction the experiment also was 1 ∂
− (P + ρgz ) + gβ (T − T∞ )
carried out on a vertical isothermal cylinder and the ρ ∂z
local free convection coefficient were compared with ∂ ⎛ ∂v ⎞ 1 ∂ ⎛ 2 ∂v ⎞
⎜ uυ − υ ⎟ + r ⎜υ − ν ⎟=
the numerical study and the previous works [26, 32]. ∂z ⎝ ∂z ⎠ r ∂r ⎝ ∂r ⎠ (13b)
Figure 6 shows these results which are in good −
1 ∂
(P + ρgz ) − υv2
ρ ∂r r
agreement with the numerical simulation and the
previous works.

Energy: ∂⎛ ν ∂T ⎞ 1 ∂ ⎛ ν ∂T ⎞ (14)
⎜ uT − ⎟+ r ⎜υT − ⎟=0
∂z ⎜⎝ pr ∂z ⎟⎠ r ∂r ⎜⎝ pr ∂r ⎟⎠

In the above equations, u and υ are the velocity

components in the z and r directions while T and P
denote the temperature and pressure.
The cone was placed in an extended space for the
numerical solution domain. At left, right and bottom
sides of this extended zone we applied pressure inlet
condition. The pressure outlet condition was applied at
the top side. For the cone lateral and base surfaces the
constant temperature and adiabatic wall have been
applied respectively. These boundary conditions are
Fig. 6 Local heat transfer coefficient versus shown in figure 7.The problem was solved for
dimensionless length for cylinder different cone surface temperatures. An unstructured
grid was applied for this model. On the cone surface
the boundary layer mesh was used. The numerical

5
Proceedings of the 4th WSEAS Int. Conf. on HEAT TRANSFER, THERMAL ENGINEERING and ENVIRONMENT, Elounda, Greece, August 21-23, 2006 (pp159-165)

results which are in excellent agreement with the Fig.8 Local heat transfer coefficient versus
experimental data point are shown in figures 8,9, 10. dimensionless length for cone at Ra = 7.9 × 10 5

5. Results and Discussion

Figures 8, 9, 10 indicate the variation of the local
convection heat transfer coefficient from the tip of
the cone along the lateral surface for a range of
Rayleigh numbers between 4.9 × 10 5 to 1.1× 106 . All
the figures show a decreasing trend up to
approximately 5 mm from the base due to the
growth of thermal boundary layer.
The significant influence of the base of the cone,
both in the experimental and numerical studies,
indicates a recirculation region above the cone
base which affects the convection heat transfer at
Fig. 9 Local heat transfer coefficient versus
the trailing edge and causes it to increase. This
effect can be confirmed with the presentation of dimensionless length for cone at Ra = 8.7 × 10 5
the velocity field around the cone obtained from
the numerical simulation which is indicated in
figure 11. The same phenomenon has been
observed by Koyama et al. [18].
A correlation has been obtained for the local
Nusselt number variation along ξ for different
values of the Rayleigh number:
Nu ξ Raξ 0.7 = 0.001125(ξ L )−1.5 (15)
This correlation is based on the both experimental
and numerical data points which are in excellent
agreement with each other, as shown in figure 12.
hL L
Also an average Nusselt number, Nu L = ,
K
based on the Eq.(10) has been obtained as Fig. 10 Local heat transfer coefficient versus
following: dimensionless length for cone at Ra = 9.9 × 10 5
Nu L = 0.000703Ra L 0.7 (16)

Fig.11 Effect of trailing edge on velocity vector

6
Proceedings of the 4th WSEAS Int. Conf. on HEAT TRANSFER, THERMAL ENGINEERING and ENVIRONMENT, Elounda, Greece, August 21-23, 2006 (pp159-165)

[7] I. Pop, and H.S. Takhar, Compressibility effects in

laminar free convection from a vertical cone, Applied
Scientific Research, vol. 78, pp. 1-82, 1991.
[8] M.A. Hossain, and S.C. Paul, Free convection from
a vertical permeable circular cone with non-uniform
surface temperature, Acta Mechanica, vol. 151, pp.
103-114, 2001.
[9] W.M. Lewandowski, S. Szymański, P. Kubski, E.
Radziemska, H. Bieszk and T. Wilczewski, Natural
convective heat transfer from isothermal conic, Int. J.
of Heat and Mass Transfer, vol.42, Issue 10, pp. 1895-
1907, 1999.
[10] P.H. Oosthuizen and E. Donaldson, Free
Convective Heat Transfer from Vertical
Cones, Transactions of the ASME, pp. 3[19]-331,
1972.
Fig. 12 Local Nusselt number versus dimensionless [11] T. Choi, An Interferometric and numerical study
length of free-, forced- , and mixed convection heat transfer
6. Conclusions from heated vertical cylinders, Ph.D. thesis, University
The laminar free convection from an isothermal of Wisconsin-Madison, U.S., 1985.
downward cone of circular cross section was studied [12] W. Hauf, , and U. Grigull, Optical Methods in
both experimentally and numerically. Experiment was Heat Transfer, In Advances in Heat Transfer, vol. 6,
carried out using Mach-Zehnder interferometer. The pp. 133-366, Academic Press, New York, 1970.
purpose of the study was to determine the effect of the [[13] E.R.G. Eckert., and R.J. Goldstein,
variations of Rayleigh numbers on the Nusselt Measurements in Heat Transfer, 2nd ed. McGraw-Hill,
number. Correlations have been obtained for local and New York, 1972.
average Nusselt numbers. The effect of the base of the [14] M. Charles Vest., Holographic Interferometry,
cone on the heat transfer characteristics was studied John Wiley & sons, 1979.
and it was indicated a recirculation region above the [15] Yu.I. Ostrovsky, M.M. Butusov and G.V.
cone base which affects the convection heat transfer at Ostrovskaya Interferometry by Holography, Springer-
the trailing edge and causes it to increase. Verlag Berlin Heidelberg, New York, 1980.
[16] J.W. Bradley, Density Determination from
Axisymmetric Inteferograms, AIAA, vol. 6, pp. 1190-
References 1192, 1968.
[1] J.H. Merk, and J.A. Prins, Thermal convection in [17] John H. Lienhard IV, and John H. Lienhard V, A
laminar boundary layer, Applied Scientific Research, Heat Transfer Text Book, 3rd ed. , Phlogiston Press,
vol. 4, pp. 10-24, 1953. Cambridge, Massachusetts, 2002.
[2] R.G. Hering and R.J. Grosh, Laminar free [18] H. Koyama, A. Nakayama, S. Ohsawa and H.
convection from a non-isothermal cone, Int. J. of Heat Yamada, Theoretical and experimental study of free
and Mass Transfer, vol.5, pp. 1059-1068, 1962. convection from a vertical frustum of a cone of a finite
[3] R.G. Hering, Laminar free convection from a non- length, Int. J. of Heat and Mass Transfer, vol. 28, Issue
isothermal cone, Int. J. of Heat and Mass Transfer, vol. 5, pp. 969-976, 1985.
8,pp. 1059-1068, 1965. [[19]] D. Naylor, Recent developments in the
[4] E.M. Sparrow, and L.D.F. Guinle, Deviation from measurement of convective heat transfer rates by laser
classical free convection boundary layer theory at low interferometry, Int. J. Heat and Fluid Flow, vol. 24,
Prandtl numbers, Int. J. of Heat and Mass Transfer, Issue 3, pp. 345-355, 2003.
vol. 11,pp. 1403-1415, 1968. [20] F.N. Lin, and B.T. Chao, Laminar Free
[5] S. Roy, Free convection from a vertical cone at Convection Over Two Dimensional and Axisymmetric
high Prandtl numbers, ASME J. of Heat Transfer, Bodies of Arbitrary Contour, ASME J. Heat Transfer,
vol.96, pp.115-117, 1974. vol. 96, pp. 435-442, 1974.
[6] M. Alamgir, Overall heat transfer from vertical
cones in laminar free convection: an approximate
method, ASME J. of Heat Transfer, vol. 101, pp. 174-
6, 1989.

View publication stats