ISIJ International, Vol, 35 (1 995). No. 12, pp.

1438-1443

The Estimation of Convective Heat Transfer Coefficients between

a Spherical Particle and Fluid at Lower Reynolds Number

Takeshi IMAI. Takeaki MURAYAMA1)

and Yoichi ONO1)
Formerly Graduate School, Kyushu University. Nowat Yawata Works, Nippon Steel Corporation,
Tobihatacho. Tobata-ku,
Kitakyushu. Fukuoka-ken. 804 Japan. 1) Department of Materials Science and Engineering. Faculty of Engineering,
Kyushu University. Hakozaki. Higashi-ku. Fukuoka. Fukuoka-ken, 81 2 Japan. E-mail: murayama@zaiko.kyushu-u.ac.j
p
(Received on April 14. 1995.• accepted in final form on June 19. 1995)

Convective heat transfer coefficients between a spherical particle and fluid for two flow systems
(solid-liquid,
-gas) were measuredand were calculated with computer simulation at relatively low Reynolds
numbers, and the applicabilities of empirical equation, for example, Ranz
& Marshall's equation, to the
estimations of convective, heat transfer coefficients of these systems
were investigated.
An aluminum sphere, 32 in diameter, which
mm was kept at about 60'C initially for solid-water system
and at 30'C for solid-air system, wasdipped into each fluid flow and the relationships between the Nusselt
number, Nu, and the Reynolds number, Rep, were obtained. Convective heat transfer coefficients
were
calculated with computer simulation and were comparedwith experimental results.
For solid-water systems, convective heat transfer coefficient
was as large as the value estimated by Ranz
& Marshal l's eq uation for the range 300 Rep, but for lower Reynolds numberreg ion, heat tra nsfer coefficients
were slightly larger than the value calculated by Ranz & Marshall's equation. Although Reynolds number
measured in solid-air system was small, heat transfer coefficient was as large as ones calculated with
following empirical equation,

Nu= 2.0 + 0.6 Prll*Re~!' + 0.43 (PrGr) '/4

KEYWORDS: convective heat transfer coefficient; Ranz & Marshall's equation; Yuge's equation; Reynolds
number; Prandtl number; Nusselt number; Grashof number; Biot number; computer simulation; finite
deference method.

1. Introduction Nu= 2.0 + 0.6 Prl/3 (9 x Rep)1/2 . ... , ....

.(2)

The process simulation for the blast furnace with a Nu= 2.0 + I I prl/3
.
Re~•6 ..........(3)
mathematical model has been extensively investigat-
edl 4) and the knowledge resulted from the calculation However, the applicability of Eqs. (1), (2), and (3) at
lower Reynolds numberthan 300 was not investigated
is available for the stable operation. For this simulation,
it is
enough. Moreover, in such lower Reynolds number
necessary to use proper process parameters and region, the effect of natural convection
manyequations to evaluate these parameters have been Qn the equation
is also supposed. Therefore, in this
required until today. work, convective heat
transfer coefficients between particle and fiuid were
In the previous work,5) convective heat transfer co-
measured at relatively low Reynolds numbers and the
efficientsbetween a spherical particle and liquid were
measured and the applicability of some empirical
applicabilityof Ranz &
Marshall's equation, Eq. (1),
equations for this system wasexamined. It wasconfirmed was examined. Convective heat transfer coefficients for
the wide range of Reynolds number
that the convective heat transfer coefficients between were calculated
a with the computer simulation and were comparedwith
spherical particle and liquid could be estimated with
experimental data and the applicability of the empirical
Ranz & Marshall's equation,6) Eq. (1), for the range
equation. Eq. (1), was examined.
300 Rep.

Nu=2.0+0.6Prl/3Rel/2 ..........(1) 2. Principle

For packed beds, it wasconfirmed that convective heat In the experiment and the numerical analysis, the
transfer coefficients between particle and liquid could be following phenomenonwas utilized. metallic sphere A
estimated with the equation by Ranz for packed beds,5) in a uniform temperature, initially To, was dipped into
Eq. (2), and Wakao's equation,6) Eq. (3). the flows of water or air of certain temperature, T,.,
a
and a certain particle Reynolds numberand was cooled.

C 1995 ISIJ 1438

 ISIJ International, Vol. 35 (1995), No. 12

In this system Brot number, Bi, which was calculated

300(mm)
with use of convective heat transfer coefficient estimated
I
by Eq. (1), wasless than O. Therefore, it wasconsidered
. o o
that the sphere had uniform temperature.9) From the l OO(mm)
relation of the unsteady state heat transfer between the o ~~) o
particle and the fluid fiow, fundamental Eq. (4) was
derived.
o o

-pCVddtT*=hA(T.-T*) .....
..........(4)

I.C. T*=To at t=0 ......

..........(5)

ThenEq. wasintegrated under the initial condition

(4)
E
(5) and Eq. was derived. The relation between the
(6) lurninum
sphere E
~'
temperature. T*, of a metallic sphere and time, t, is o
obtained from this equation.
200(mmc) o
T.
- T~ = exp hA (6) Alumina ball

- t
To T~
~ pCV
Equation (7) is derived by the substitution of Eq. (6)
for the definition of the Nusselt number.
2
2PCR T,- T~
t ~
Gas

Nu=- In (7) Fig. l. Schematic diagram of air fiow system for cooling
3kLt To T~
~ particle.

If the relation between t and

given based on theT is
temperature variation curve of the metallic sphere in the Markin Recorder
cI mmThermocou
Cu-Constantan
ple
(Personal computer)
fluid fiow of a certain Reynolds number, convective heat
Solder RS-232C
transfer coefficient, h, or Nusselt number, Nu, is obtained tp
2mmHole
from Eqs. (6) and (7). Multi Logging Meter

Thermocouples
3. Experiment mSphere
Al umin u

3.1. Flow Apparatus and Velocity Measurement in R=i 6.00mm

Solid-Liquid System
For the measurements of convective heat transfer
coefficient of solid-1iquid system, the same circulating
water tank as one reported in the previous paper5) was
used. The water flow wasprepared with the screw-motor
and the velocity was given by the measurementof the Hot Water Immerseinto Each
Flow systems
time required for the float to pass through the measuring
area. From the result, particle Reynolds number, Rep Fig. 2. Equipment for temperature measurement.
was calculated.
3.2. Flow Apparatus and Velocity Measurement in measurementshownin the previous paper.5) The sheath
Solid-Gas System thermocouple of the copper constantan of I mmin
diameter was inserted in the center of the aluminum
Figure I shows the fiow apparatus for measurement
sphere of 32mm in diameter and was soldered.
of convective heat transfer coefficients of solid-gas
system. This was madeof the circular acrylic cylinder, The temperature change of the sphere was measured
in the case of solid-water system,57 as follows. Initially,
whoseheight was I OOO and radius was 200 mm.The
mm the sphere was preheated in hot water of about 60'C
alumina sphere, whose radius was lOmm,was packed
until the temperature of the sphere becameequal to that
at the lower part of this tube in order to regulate the air
of the hot water, Next, the sphere was subrnerged in the
flow profile. The air flow wasgiven by an air compressor.
The velocity of air flow was defined by the entrance water flow of about 20'C (room temperature). The
velocity into the tube and was measuredwith gas flow temperature drop of the metallic sphere was recorded
directly by a personal-computer through a multi-10gging
meter. Before the measurementof temperature drop
meter.
curve of the solid particle, the position where the particle
In the case of solid-air system, the measurementwas
was held in the tube must be decided. Therefore, this carried out as follows. Initially, the sphere waspreheated
problem was solved with the computer simulation
in hot water of about 30'C until the temperature of the
explained in the later section.
sphere becameequal to that of the hot water. Next, the
3.3. Measurementof Temperature Drop of a Sphere sphere was put into the air flow of about 25'C (room
Figure 2 shows the equipment for temperature temperature) as shownin Fig. The temperature drop I .

1439 C 1995 ISIJ

 ISIJ International, Vol. 35 (1995). No. 12

of the metallic sphere was recorded by the method and the converged heat transfer coefficient wascalculated
mentioned above. in every time iteration with Semi-Implicit-Method for
In this way, the temperature drop of the sphere sub- Pressure-Linked Equations (SIMPLEmethod).
merged in the water flow or the air flow of a desired auj
Reynolds number was measured and convective heat =0 ..........(1 l)
ax
transfer coefficients at several flow systems were cal- j
culated by Eq. (7). au, a (uJul) e~j (v ;~j I ap
J i

4. Derivation of Corrected Equation

at
+ ax
j
=
Tref)+aa~ :~j
7 ax*

possible that the effect of forced convection on

It is +gp(T ••
(12)

the equation for the estimation of Nusselt numbermight

be smaller than that of natural convection at lower
Reynolds numbers. In such case, Ranz Marshall's &
aT
at
+ axj
a (u.T)
J
a
= axj ~~, ( v aT
axj
(13)

equation, Eq. (1), should be corrected for the effect of

Figure 3 shows the boundary conditions around a
natural convection. The Yuge's equation, Eq. (9), is
spherical particle used in this simulation.
famousand widely used whenthe natural convection is
dominated.
Table I
shows the calculation condition. This cal-
culation was based on the two-dimensional unsteady
Nu=2,0+0.43(PrGr)1/4 ..........(9)
heat transfer problem and the time steps used in the

In the right hand side of Eq. (9), the first term represents
calculations were I sec for solid-water system and
5sec for solid-air system respectively. The calculation
the effect of heat conduction in fiuid and the second term
was carried out until 40 time steps for each system.
represents the effect of natural convection.
If it can be assumedthat the relation between the 5.2. Thermal Properties Used for the Simulation
second term in the right-hand side of Eq. (1), which Thermal properties of fluids might vary with the
transfer, and
meansthe effect of forced convection heat variation of the fiuid temperature at lower Reynolds
the second term in the right side of Eq. (9), which means number. Therefore, in this study, the properties of fluids
linear,10)
the effect of natural convection heat transfer, is
were glven by the functions of fluid temperature. Table
the corrected equation is given by Eq. (lO).
2showsthe functions of thermal properties of each fluid.
Nu=2.0+0.6Prl/3Rrl/2 +0.43(PrGr)1/4 .....(lO) These functions were obtained by the regression analyses

of Eq. (lO) as well as

at the pressure O, I MPa.
In this study, the applicabilities
5.3. Position of Particle in the Tube for the Measure-
Eq. (1) to the measured and the calculated data were
examined. ment of Convective Heat Transfer Coefficient be-
tween Particle and Air
5. Simulation Before the measurementof convective heat transfer
coefficient between a spherical particle and air, the
5.1. Governing Equation positions where the particle should be held in the tube
The ideal flow field was not always given in this work must be decided. Therefore, the profile of the air fiow
and it was impossible to measure the convective heat
transfer coefficients in wide range of Reynolds numbers. calculation of
Table l. Conditions for unsteady state
Therefore, heat transfer coefficient was calculated with convective heat transfer coefficients between a
computer simulation. spherical particle and fluids.
The equations of continuity, momentum and energy
(11), (12) and (13) in differential Time Time Reynolds
given by
were Eqs. Fluids
steps(-) increments (s) number(-)
forms in Cartesian coordinates. These equations were
transformed into two-dimensional cylindrical coordinate Water 40 l .O O. 1-lO OOO
form and the finite difference method on the basis of Air 40 5.0 O,
1 IOOOO
control volume methodll) was used in this simulation

tp2c

o ddy2 = O
drv
v u=o
d2tp ay
i~
= o dx~ - O
ax

y, r
u=1 v=0
du dc
v=0 =0 _O
Aluminum ~y ay
sphere

Fig. 3. Boundary conditions around a sphere for this simulation.

C 1995 ISIJ 1440

 ISIJ International, Vol. 35 (1995), No. 12

Table 2. Thermal properties of fluids. Table 3. Conditions for steady state calculation of air fiow
in circular tube.
Water
217.99 +6.9107T- 1.9239 x 10~2T2+ I .618 x 10~5T3 Time Time Reynolds
p (kglm3) Fluid
(-) increments
steps (s) number(-)
;.
(W/mK) -0.60328 + 6.5612 x l0~3 T-8.3923 x 106T2
Pr (-) 989.94-8.5399T+ 2.4691 x l0~2 T2 -2.3862 x 10~5T3 Air lOOOO
G" (-)
9.8067 x 0.0323 x AT
(1.159 x l0~4-9.9325 x 10 T+28583 x 10 9T2)2 x Tf
0.3
Air
-~~ - A
-
A~
p (kg/m3) 3.2959
- I .0180 x lO~ T+I .0399 x lO~ T2
2 5 o
B
-
BI
~(W/mK) 4.5146 10~3 +7,2287 x lO~s T
> 0.2
C
-
CI
Pr (-) 0.80433 -4.7076 > l0~4T+ 5.5244 x l0~7 T2 ~:~'

Gr (-)
9.8067 x 0.0323 x AT :~

(- l.3819 x l0~5 +9.8961 > 10~BT)2x Tf 0.1

AT; Temperature difference between particle and fiuid (K)

Tf; Film temperature (K)
0,0
0.0 0.2 0.4 0.6 0.8 1.O
50mm LongitUdinal distance(m)
LIH)I
y, v
Fig. 5. The w-velocity profile of air flow in circular tube.
x, u
zl
w 3 Eq.(1
)
I
Eq.( O)
e Oalculated
O
l
I O
= 2 o Experimental data
I
I T=0'o
l
CO
U,
1:,

aS
~
o)
l C:
o C9
1:,
1 e ee e
l O)
C (Ts-Tw)/(To-Tw)
Pr = 5.89
= O, 1
JO
I
I ~ O-1 O 1
solid

2
-
Iiquid system

3 4
l l log(Rep)
E
E L=0'35 Fig. Convective heat transfer coefficients
o
6. between a particle
and water.

w=wll
A' B', C'
t position shown in Fig. 4. From the u and v velocity
Fig. 4. Boundary conditions for the simulation of a air flow
profiles, it
was found that the velocities vary sharply in
the packed bed for the range
in circular tube. z O.2, but both velocities
approach to zero toward the top of this tube. The w
velocity profile shows that the velocity reduces slowly
in the tube was obtained with computer simulation. and approaches to the constant velocity in the range in
Equations (11), (12) and (13) were rewritten into the which the longitudinal distance is greater than about 0.6
equations in three dimensional Cartesian coordinates for both Reynolds numbers. So in this work, the position
form and then the air flow in the tube was calculated where the particle should be held in the tube wasselected
with the same way mentioned above. Figure 4 shows in this range.
boundary conditions used in this simulation. The air
flows into the tube at the low temperature, 25'C. The 6. Results
tube was not kept in adiabatic condition, because this
system was kept at the lower temperature. The air flows
6.1. Convective Heat Transfer Coefficient between a
through the packed bed, which exists at the lower part Spherical Particle and Water
of the tube, thus the pressure drop increases. In this Figure 6 shows the obtained relationship between Nu
simulation, this drop was estimated by Ergun's equa- and Rep for solidwater system in the range 40 Rep
tion.12) 3900. In this case, the temperature difference between
particle and liquid, T* T~, was I 'C. The calculated data
Table 3 shows calculation condition. Steady state
-
calculation was carried out at two Reynolds numbers, agreed well with the experimental data, but these values
lOO and 10. were slightly larger than the ones estimated by Eq. (1)
Figure 5
shows w velocity componentsprofile at each for the lower Reynolds number region. The calculated

1441 C 1995 ISIJ

 ISIJ International. Vol. 35 (1 995), No. 12

with the ones estimated by Eq. (lO). Similar to the

results showthat Nusselt numberwould approach to the
limiting value of about 10 because of the existence of phenomenaat low Reynolds number in particle-water
natural convection. Fromthe knowledge of the previous system, natural convection heat transfer becomes
work,5) there existed the tendency that the smaller the dominant at this temperature difference between the
particle and air and this effect on the convective heat
temperature difference became, the better the experi-
transfer coefficient must be taken into consideration.
mental data agreed with Eq. (1). As the temperature
difference approaches to zero, convective heat transfer 6.3. Velocity Vector in This Flow Field
coefficient betweena particle and liquid can be estimated
Figure velocity vector in a hemi-
8 shows calculated
by Ranz Marshall's equation, whennatural convection
& spherical particlewater system. There exists a large
would not happen. Here, the applicability of corrected steady state vortex behind the particle at Rep= 3OOO,but
equation, Eq. (10) was not confirmed, because Grashof
this vortex becomessmaller at Rep 1778 and vanishes
numbermight be estimated too large in this system. =
in the range Rep I OOO. The same phenomenonwas
6.2. Convective Heat Transfer Coefficient between a confirmed in solid-air system. It was found in Secs. 6.1
Spherical Particle and Air and 6.2 that convective heat transfer coefficient becomes
shows the obtained relationship between Nu larger in the range Rep 300 because of the existence of
Figure 7 natural convection. Fromthis result, it is supposed that
and Repfor solid-air system for the range 4.4 Rep 60.
In this case, the temperature difference between particle the existence of vortex affects on the convective heat
and liquid. T*-T "' was 5'C. In this system, convective transfer phenomena.
heat transfer coefficient could not be measured at the
wide range of Reynolds numberbecause of the limita- 7. Conclusion
tions of present apparatus. The calculated data agreed Convective heat transfer coefficients in a particle-
well with the experimental data, but these values were liquid, -gas systems were measuredexperimentally and
greatly larger than the ones estimated by Eq. (1) and it calculated with computer simulation at relatively
were
seemsthat Nusselt numberwould also approach to
the
low Reynolds numbers. Following results were obtained.
limiting value of about 10. Both data agreed rather well
(1) Convective heat transfer coefficients between
a particle and water could be estimated by Ranz &
Marshall's equation for the range 300 Rep, but those
2.0
Eq.(1)
e becamea little larger at low Reynolds numberand the
Eq.(10) e calculated Nusselt number would approach to the
1.5 e Calculated
e limiting value of 10 because of the existence of natural
o Experimental data e convection.
Convective heat transfer coefficients between a
(2)
1.o e e particle and air could not be estimated by Ranz &
Marshall's but the corrected equation. Eq. (lO) for the
0.5
Pr = 0.71 3 60. In this Reynolds number range,
range 4.4
(Ts-Tw)/(To-Tw) = 0.5 natural convection was dominant and Nusselt number
gas - solid system would also approach to the limiting value of 10.
0.0 -1
O 1 2 3 4 (3) It
was found from the computer simulation that
the effect of the vortex existing behind the particle on
Iog(Rep)
convective heat transfer coefficient might bc significant.
Fig. 7. Convective heat transfer coefficients between a particle
and air.

(F~_~~piIOOO:*;

Fig. 8. Velocity vector in a spherical particle-water system.

C 1995 ISIJ 1442

 ISIJ International, Vol. 35 (1 995). No. 12

Acknowledgments x, y, z: X, Y, Z axis, respectively (m)

Theauthors are grateful to the membersof Committee p: Compressibility (-)
(Chairman is Professor Dr. J. Yagi) on the Tr~nsport Void fraction of bed (-)
Phenomena of Four Fluids for their comments. Kinetic viscosity (m2/s)
p: Density of aluminum (kg/m3)
Nomenclature c: Dependentvariable (-)
A: Surface area of sphere (m2)
Bi : hDlk*; Biot number(-) REFERENCES
C: Specific heat of aluminum (J/kg K) l) M. Hatano and K. Kurita:
Tetsu-to-Hagan~, 66 (1980), 1898,
Gr : Grashof number(-) 2) S. Taguchi. H. Kubo, N, Tsuchiya, K. Ichifuji and K. Okabe:
g: Gravitational acceleration (m/s2) Tetsu-to-Hagan~, 68 (1982), 239, 247.
h: Convective heat transfer coefficient (W/m2) 3) Transport Phenomena in the Lower Part of Blast Furnace, ed.
by Committee on Transport Phenomenain Gas-Solid-Liquid
Nu: Nusselt number(-) Packed Bed, ISIJ, Tokyo, (1992).
kL : Thermal conductivity of liquid (W/mK) 4) K. Takatani, T. Inada and Y. Ujidawa: CAMP-ISIJ, 7 (1994),
k. : Thermal conductivity of solid (W/mK) 50.

p : Pressure of fiuid (Pa) 5)

6)
T. Imai, T.
W. E.
Murayamaand Y. Ono: ISIJ
Ranz and W. R. Marshall,
Inl., 34 (1994), 777.
Pr : Prandtle number(-) Jr.: Chem. Eng. Prog., 48
(1951), No. 3, 141.
R: Radius of a sphere (m) 7) W. E. Ranz: Chem.Eng.Prog., 48 (1952), No. 5, 247.
Rep: Particle Reynolds number(-) 8) N. Wakao, S. Kaguei and T. Funazkri: Chem. Eng, Sci., 34
t: Time (s) (1979), 325.
T: Temperature of fluid (K) 9) D. R. Pitts and L. E. Sisson: Schaum's Outline of Theory and
To : Initial temperature of sphere (K) Problems of Heat Transfer, MacGraw~Hill, NewYork, (1977),
75.
T* : Temperature of sphere (K)
T~: Temperature of water (K) l O) T. Shirotsuka. A. Hirata
and A. Murakami: Kagaku-Gijutsusha-
notameno-Ido-Sokudoron, Ohm-Sha,Tokyo, (1966), 238.
v, w: Velocity of fluid (x,y, z direction, respectively.
u, l l) S. V. Patankar: Numerical Heat Transfer and Fluid Flow,
Refer to Fig. 4.) (m/s) Hemisphere. Washington D.C., (1980).
V: Volumeof sphere (m3) l2) S. Ergun: Chem.Eng. Prog., 48 (1952), 89.

wo : Value of w at inlet (Refer to Fig. 4.) (m/s)

1443 C 1995 ISIJ