Numerical simulation of heat and mass transfer

in uidised bed heat treatment furnaces

W.M. Gao
a,*
, L.X. Kong
b
, P.D. Hodgson
a
a
School of Engineering and Technology, Deakin University, Geelong, Vic. 3217, Australia
b
Centre for Advanced Manufacturing and Research, University of South Australia, Mawson Lakes, SA 5095, Australia
Received 2 December 2001; accepted 6 February 2002
Abstract
Inthis paper, a novel combinedtheoretical andcomputational model is developedtosimulate the heat and mass transfer betweena uidised bed
and a workpiece surface, and within the workpiece by considering the uidised bed as a mediumconsisting of a double-particle layer and an even
porous layer. The heat and mass-transfer ux from the uidised bed to the workpiece surface is contributed by dense and bubble phases,
respectively. The convective heat and mass transfer is simulated by analysing the gas dynamics in the uidised bed, while radiative heat transfer is
modelled by simulating photon emission in a three-dimensional particle array. The simulation shows that convection is approximately constant,
while radiation contributes signicantly to the heat transfer. The heat-transfer coefcient on an immersed surface near particles is about 610
times that on other areas. The transient heat and mass-transfer coefcient, heat and mass-transfer ux on any surface of the workpiece, transient
temperature and carbon distributions at any position of the workpiece during the metal carburising process are studied with the simulation.
#2002 Elsevier Science B.V. All rights reserved.
Keywords: Heat treatment; Fluidised beds; Simulation; Heat and mass transfer; Radiation; Convection
1. Introduction
The use of uidised bed furnaces for the heat treatment of
metals is now a widely accepted technique because it has a
rapid heating rate, the atmosphere can be changed rapidly and
the temperature distribution in the bed is uniform [1,2]. Up to
now, various models have been developed to predict heat-
transfer coefcients between immersed surfaces and uidised
beds and there have been reviewed by Saxena [3], Sharma and
Turton [4], Yoshida et al. [5], and Tien [6]. Most of these
models focused on the heat transfer between bubbling ui-
dised beds andimmersed tubes, which are used tosimulate the
heat exchange parts in coal combustors or boilers. However,
these models cannot be directly applied to heat treatment in
high temperature uidised bed furnaces because the metal
workpieces to be treated have quite complex congurations. It
is necessary to simulate heat and mass transfers between the
workpiece surface and uidised beds and within the work-
piece in the heating or cooling process in order to further
expand the application of uidised beds in heat treatment,
explore new heat treatment processes, intensify heat and
mass transfers between the heating surface and the uidised
bed, control uidised bed heat treatment processes, select and
optimise the operation parameters and improve the quality of
products.
For small particles, the dominant heat transfer between
the immersed surface and the uidised beds is contributed by
emulsion and bubble phases, alternately. The former is
controlled by the particle residence time t
r
, and the particle
concentration near the heat-transfer surface, while the latter
is dominated by the fraction of bubble time f
b
, when the
bubble is in contact with the immersed surface. The heat
transferred between the emulsion phase and the immersed
object surface at high temperatures in gas-uidised beds is
by transient convection q
ce
, and radiation q
re
. Between the
bubble phase and the surface of the immersed object, the
heat transferred is by radiation q
rb
. The overall heat ux from
uidised beds to the workpiece surface is, therefore,
q = (q
ce
q
re
)(1 f
b
) q
rb
f
b
(1)
From the heat uxes identied above, the corresponding
heat-transfer coefcients can be dened as
h
w
= (h
ce
h
re
)(1 f
b
) h
rb
f
b
(2)
Very little experimental and theoretical research on mass
transfer has been reported [1,7], although uidised beds have
been widely used in the surface modication processing of
steels such as in carburising, carbonitriding, nitriding, and
Journal of Materials Processing Technology 125126 (2002) 170178
*
Corresponding author. Tel.: 61-3-5227-2908; fax: 61-3-5227-2167.
E-mail address: weimin@deakin.edu.au (W.M. Gao).
0924-0136/02/$ see front matter # 2002 Elsevier Science B.V. All rights reserved.
PII: S0 9 2 4 - 0 1 3 6 ( 0 2 ) 0 0 3 7 2 - 2
nitrocarburising. The transfer of carbon from uidised beds
to the workpiece involves the following steps:
1. The transport of the molecules which contain carbon
(CO, CH
4
etc.) through the carburising atmosphere to
the steel surface.
2. The diffusion of the molecules which contain carbon
through the Nernst layer.
3. Reaction in the steel surface through the decomposition
of gaseous molecules at the interface between the steel
and the gaseous atmosphere.
4. The absorption and dissolution of carbon atom on the
steel surface.
5. The carbon diffusion in the steel.
In this paper, convective and radiative heat transfer, carbon
transfer between the uidised bed and the immersed surface,
and heat and mass transfer within the workpiece are simu-
lated using a model developed in this work that considers the
bed as a combination of a double-particle layer and an even
porous medium.
2. Heat and mass transfer between fluidised
bed and immersed surface
2.1. Double-particle layer and porous medium model
The particles in the emulsion affect the gas dynamic and
thermal boundary layers. The behaviour of particles in
contact with the surface of the immersed object is signicant
for heat and mass transfer, especially, in the uidised bed
heat treatment furnace in which radiation contributes sig-
nicantly to overall heat transfer because of the high tem-
perature. It has been reported that the arrangement of the
particles is very important in simulating the heat transfer
[6,8]. However, it is difcult to congure the array of
particles near the immersed surface during the uidisation,
because of the chaos of particle motion and the difculty in
measuring the local distribution of voids near the immersed
surface. The distribution of voids near the vessel wall in a
xed bed was based on a rhombohedral array of particles and
the multiple-layer model was used to simulate heat transfer
between uidised beds and immersed surfaces [8]. However,
under the same conditions, the proportion of voids near the
surface immersed in the bed is larger than that on the side
surface of the apparatus [9]. It is not necessary to model the
conguration of more particle layers, because the heat
transfer between the immersed surface and emulsion during
the residence time occurs in a distance of not more than one
particle diameter from immersed surface.
In the present work, a domain consisting of two layers of
particles and a medium of uniform porosity, shown in Fig. 1
was used to simulate the gas dynamics and heat and mass
transfers near the immersed surface. The triangular layout of
each array of particles was chosen based on dynamic
simulation in which the geometry of the particle array should
generate the lowest pressure drop among all possible particle
arrays (rhombohedral, orthorhombic and cubic), and based
on the photograph of particle arrays near the immersed
surface during uidising [10]. The geometric parameters
corresponding to the particle array were determined fromthe
measurement of the distribution of local porosity near the
immersed surface during uidisation [9]. The porous med-
ium begins at the centre line of the second-particle layer and
the porosity of this medium remains constant throughout.
The distribution of voids near the immersed surface corre-
sponding to the double-particle layer and porous medium
model is very accurate compared with the experimental data
measured by Korolev et al. [9] using X-rays (Fig. 2).
2.2. Governing equations for momentum, heat and
mass transfer
The gas ow equation in the porous medium can be
formulated as follows.
@(rV)
@t
\ (rVV) = \p \
~
T rF F
S
(3)
The external body force or momentum source term of the
porous medium F
S
consists of two components, a viscous
term and an inertial loss term based on the Ergun equation:
F
S
=
150m(1 d)
2
(F
S
d
p
)
2
d
2
V
1:75r(1 d)
F
S
d
p
d
VV (4)
It should be noted that V is the velocity of gas in the bed,
instead of the velocity in an empty bed as in the Ergun
equation.
The gas energy equation in the emulsion phase can be
formulated as
@(rC
Pg
T)
@t
\ (rC
Pg
VT) = \ (K
g
\T) (5)
The energy conservation for particles in the emulsion is
rC
Pp
@T
@t
= K
p
\
2
T (6)
Fig. 1. Double-particle layer and porous medium model.
W.M. Gao et al. / Journal of Materials Processing Technology 125126 (2002) 170178 171
and the convective heat ux at the immersed surface is
q
ce
= K
e
@T
@n
(7)
The convective heat-transfer coefcient is dened by
h
ce
=
q
ce
(T
w
T
B
)
(8)
In the simulation, a carbon conservation equation in the
emulsion and bubble, which is similar to the energy equa-
tion, needs to be constructed. The relation between the
carbon diffusion transfer ux and coefcient b
1
is given by
j
C
= D
Cg
@C
@n
= b
1
(C
B
C
g
) (9)
where C
g
is carbon concentration in hydrocarbon complexes
near the workpiece surface.
2.3. Radiative heat transfer
The local radiative energy of an innitesimal particle
surface is absorbed by the immersed surface and other
particle surfaces. For most media used in uidised beds
for heat treatment, such as alumina, silica sand and carbon,
the particle is opaque with a grey emitting diffusely reect-
ing surface. The difference in temperature among the par-
ticles near the immersed surface is very low, because of the
short emulsion residence time, so that the heat transfer
between particle surfaces can be ignored as a consequence
of Kirchhoff's law. The void gas is considered to be trans-
parent. Radiative heat exchange between the emulsion and
the heat-transfer surface is contributed by the particle sur-
faces exposed to the surface of the immersed object, corre-
sponding to the present double-particle layer and porous
medium model.
For a grey diffuse surface, the effective bed emissivity
depends solely on the surface emissivity of the particles and
the geometry of the particle array. If the radiative viewfactor
from an innitesimal particle surface dF(x) of emissivity e
p
,
and temperature T, to an immersed surface of emissivity e
w
and temperature T
w
is j
pw
, then the efcient area radiative
heat rate between this innitesimal particle surface and the
immersed surface is
dQ
r
(x) =
s
1=e
w
j
pw
(1=e
p
1)
(T
4
T
4
w
) dF(x) (10)
The total area radiative heat rate from a particle is then
Q
r
=
Z
o=2p
r
2
s
1=e
w
j
pw
(1=e
p
1)
(T
4
T
4
w
) do (11)
From this, the total radiative heat rate between the emulsion
and the immersed surface is obtained by integrating the Q
r
over all the particles in the double-particle layer of the model
introduced in this work. If the number particles in the double
layers near the immersed surface of area A
w
is N, then the
average radiative heat ux and the radiative heat-transfer
coefcient are, respectively, given by
q
re
=
P
i=N
Q
r
A
w
(12)
h
re
=
q
re
T
B
T
w
(13)
The view factor is evaluated numerically using the photon
emission simulation method on the geometric shape of the
particle array presented in the present work. Actually, the
viewfactor fromthe innitesimal particle surfaces located at
the same distance from the immersed surface is the same.
Fig. 3 shows the result of simulation.
2.4. Bubble phase heat transfer
Abubble in contact with the immersed surface is enclosed
by only two surfaces, the concave spherical surface of the
Fig. 2. Variation local porosity in the vicinity of the surface of the object.
172 W.M. Gao et al. / Journal of Materials Processing Technology 125126 (2002) 170178
emulsion phase with an effective emissivity of e
e
and a
temperature of T
B
, and the surface of the immersed object
with an emissivity of e
w
and a temperature of T
w
. Yoshida
et al. [5] considered the bubbles to be transparent, because
the radiation mean-free-path length in the gas phase was
much greater than the usual diameter of bubbles. Therefore,
the heat transfer between the bubbles and the surface of an
immersed object is due to the radiation place between the
internal concave surface of the bubbles and the immersed
surface. The radiative heat ux exchanged between the two
surfaces through the transparent bubble is
q
rb
=
s(T
4
B
T
4
w
)
1=e
w
j
Bw
(1=e
e
1)
(14)
j
Bw
is the view factor from the internal concave surface of
the bubbles to the immersed surface. In order to determine
j
Bw
, it is necessary to know the shape of bubbles in contact
with the surface of the immersed object. Yoshida et al. [5]
adopted a geometrical simplication that the bubbles in
contact with the immersed object are hemispherical, i.e.,
j
Bw
is 0.5. The radiative heat-transfer coefcient from the
bubble phase is given by
h
rb
=
q
rb
T
B
T
w
(15)
3. Heat and mass transfers within the workpiece
3.1. Energy conservation
The three-dimensional energy equation that includes
phase change potential within workpiece is
rC
P
@T
@t
= k\
2
T q
p
(16)
with the boundary condition
t _ 0; k
@T
@n
i

w
= h
w
(T
B
T
w
) (17)
The thermal properties, including heat conductivity and
specic heat capacity, change with temperature and the
phases in the workpiece.
The potential for the transformed phase, i.e., the energy
absorbed or released in phase transformation processes, is
calculated from
q
p
=
X
n
i=1
L
i
V
i
(18)
In the present study, the isothermal timetemperature-trans-
formation (TTT) diagram and Scheil's nucleation `additivity
principle' were used to calculate the fraction of the phase
transformed. The volume fraction of austenite transformed
on heating or of austenite transformed into pro-eutectoid
ferrite, cementite, pearlite, bainite and martensite in cool-
ing can be given [11,12] by (except for the martensitic
transformation):
V
i
= 1 exp(b
i
t
n
i
) (19)
The values of b
i
and n
i
depend on the transformation
temperature and are tted for each individual grade of steel
from the transformation curves plotted in the isothermal
TTT diagram. i = 16 refer to austenite, ferrite, cementite,
pearlite, bainite, and martensite, respectively.
Because the athermal transformation of austenite into
martensite below the martensite starting temperature M
S
does not depend on time, it can be described by
V
M
= 1 exp[a(M
S
T)[ (20)
where a is a coefcient depending on the components of
steel, and a = 0:011 K
1
for most steels.
Fig. 3. The variation of the radiative view factor.
W.M. Gao et al. / Journal of Materials Processing Technology 125126 (2002) 170178 173
3.2. Carbon transfer
The transfer of carbon in the workpiece surface is gover-
ned by the difference between carbon concentration in the
gas near the workpiece surface C
g
and the carbon concen-
tration C
S
at the steel surface. The carbon transfer ux is
dened by
j
C
= b
2
(C
g
C
S
) (21)
where the mass-transfer coefcient b
2
is determined by
1. The absorption behaviour of hydrocarbon complexes
and other kinds of gaseous molecules, such as CO, CH
4
,
etc., on the steel surface.
2. The decomposition of hydrocarbon complexes into
atomic carbon.
3. The desorption of oxygen obtained through the decom-
position of CO.
4. The reaction of oxygen with H
2
or CO.
Because of the high temperature in carburising, the absorp-
tion of hydrocarbon and desorption of oxygen in the above
processes are signicant for the mass transfer. A universal
relation determining the mass-transfer coefcient b
2
is
b
2
= b
0
exp
E
RT

(22)
If the surface mass-transfer coefcient is dened by
b =
j
C
B
C
S
(23)
b is given by
1
b
=
1
b
1

1
b
2
(24)
because the carbon transfer ux in Eq. (21) should equal the
diffusion transfer ux in Eq. (9). Fick's 2nd law was applied
to simulate the mass transfer within the workpiece.
4. Results and discussion
The mathematical model given above was integrated into
FLUENT 5.5, a nite difference CFD software developed by
Fluent Incorporation. A mesh size of 10 mm was chosen for
both the particle layers, while 50 mm was used for both
porous medium and the immersed object. The wall of the
immersed surface and that of the emulsion or bubble phase
were then analysed at the same heat and mass-transfer uxes
and temperature. The minimum uidising velocity U
mf
used
was the gas velocity in the emulsion. The time for which the
emulsion phase or bubble phase contacted the immersed
surface was determined by the emulsion residence time t
r
and bubble time fraction f
b
proposed by Thring [13]. The
time step in the calculation was chosen by dividing the
residence time t
r
into 10 intervals. Normally the time step Dt
was 0.040.1 s depending on bed conditions.
4.1. Heat transfer
The same operation and uidising medium parameters as
were used in the experiments reported in the literature were
used in the present work to validate the model. Table 1 lists
the experimental conditions used in this work, which were
adopted from the various sources cited in the table.
Figs. 4 and 5 allow comparison to be made of the total
heat-transfer coefcient (h
w
) and radiative heat-transfer
coefcient (h
r
) predicted from the model developed in this
work and the experimental results. It is found that the
predictions by the present model are very close to the
experimental results of H. S. Li et al. [15] and Saxena [3]
for a ne particle uidised bed. H. S. Li et al. [15] measured
the spatial average heat-transfer ux for particles of mean
diameter of 622 mm. Saxena and Mathur measured the total
heat-transfer and radiative heat-transfer coefcient for beds
of 559 and 751 mm sand particles at ve different bed
temperatures. A deviation of 25% predicted with the model
developed in this work is much lower than the 50%predicted
by other models [3]. For a large particle uidised bed, the
model over-predicts the total heat-transfer coefcient, as
suggested by comparison with the experimental results of
Pidwerbecki and Welty [14], who measured the total and
radiative heat ux around an immersed bronze tube of outer
diameter of 51 mm and inner diameter of 32 mm in a
uidised bed of large particles with a size of 2.9 mm and
temperature of 7001003 8C, probably due to different
uidising materials being used. The particles of 53.5%
silica, 43.8% alumina, 2.3% titania, and 0.4% other sub-
stances were used as the bed material in that work. In the
present work, pure silica sand was chosen as the uidising
medium.
In this work, the radiative heat transfer between emulsion
and immersed surface is simulated by directly calculating
Table 1
The experimental parameters of reference
Reference Immersed object Bed media Diameters (mm) Bed temperatures (8C) U/U
mf
Pidwerbecki and Welty [14] Horizontal bronze tube,
51/32 mm
Silica alumina others 2.900 700, 810, 908, 1003 1.01.5
H.S. Li et al. [15] Horizontal stainless tube,
40/20 mm
Silica sand 1.815, 1.103, 0.622 424, 618, 735, 872, 1028 1.02.0
Saxena [3] Sand 0.559, 0.751 112, 222, 472, 547, 712 1.02.0
174 W.M. Gao et al. / Journal of Materials Processing Technology 125126 (2002) 170178
the particle radiation based on the particle array assumed in
the double-particle layer and porous medium model. This
results in improvement in the accuracy of prediction of the
radiative heat-transfer coefcient. Fig. 5 shows that the
predictions from this model are in good agreement with
the experimental data for all particle sizes, mostly being
within 25%.
The heat transfer between a uidised bed of 200 mm silica
sand particles with an emissivity of 0.6 and the surface of an
immersed object with a temperature of 300 K and an emis-
sivity of 0.8 was simulated with different bed temperatures
and air uidisation number of 2.0 to enhance the under-
standing of the heat-transfer mechanism. Fig. 6 shows how
the variation in average temperatures of the gas and particles
in the emulsion change with distance from the immersed
surface at different residence times. It is clear that a change
in gas temperature only occurs within a distance of approxi-
mately 2/3-particle diameter and is less than one particle
diameter for the longest residence time, 10 s. The change in
the temperature near the immersed surface decreases loga-
rithmically with the increasing of the residence time.
The temperature of particles only change very slightly for
all residence times. Therefore, heat transfer between particle
surfaces facing the immersed surface is signicant to the
total heat transfer. Fig. 7 shows the distribution of heat-
transfer coefcients between the emulsion and the immersed
surface around a particle. It shows that the average heat-
transfer coefcient on the immersed surface at the point in
contact with particles is about 610 times that at the other
surfaces. The section with a high heat transfer is within a
Fig. 4. Comparison of h
w
predicted by the present model with experiments by Saxena [3], Hong Shun et al. [15], and Pidwerbecki and Welty [14].
Fig. 5. Comparison of h
r
predicted by the present model with the experiments by Saxena [3], and Pidwerbecki and Welty [14].
W.M. Gao et al. / Journal of Materials Processing Technology 125126 (2002) 170178 175
circle of diameter 0.5d
p
. Gas dynamics clearly show that the
gas is still in this section. The heat transfer can therefore also
be called conduction, as in the analyses of heat transfer by
most researchers [3] who consider that heat transfer consists
of conduction and convection. However, a uniform gas lm
of about 0.1d
p
has to be introduced into conduction models
to agree with experimental results. From the result of the
present simulation, the mean gas lm thickness in the
circular section of diameter 0.5d
p
is about 0.064d
p
.
Fig. 8 shows the effect of the bed temperature T
B
on the
total heat-transfer coefcient h
w
, convective heat-transfer
coefcient h
c
, radiative heat-transfer coefcient h
r
, and
relative radiative contribution h
r
/h
w
. The radiative compo-
nent h
r
, increases at a higher rate as the bed temperature
increases. In contrast, the total heat-transfer coefcient h
w
increases more slowly with bed temperature. The convective
heat-transfer coefcient h
c
increases with bed temperature
initially (less then 1000 K) at a smaller rate, and then
decreases when bed temperature is above 1000 K. A rela-
tively stable convective heat-transfer coefcient is found in
the high temperature uidised bed. The change in total heat-
transfer coefcient is mainly affected by radiation. Saxena
[3] summarised the relations between heat-transfer coef-
cient and other parameters:
h
w
= (1 f
b
)
k
eo
r
e
C
PS
t
r

1=2
f
b
se
t
(T
2
B
T
2
w
)(T
B
T
w
)
(25)
h
e
= (1 f
b
) k
eo

8
9
d
p
sT
3
B

r
e
C
PS
pt
r

1=2

f
b
q
rb
T
B
T
w
(26)
The effective emulsion density is assumed to be r
e
=
r
S
(1 d
mf
) and does not change with bed temperature.
Fig. 6. Variation of gas temperature in emulsion.
Fig. 7. Distribution of heat-transfer coefficient.
176 W.M. Gao et al. / Journal of Materials Processing Technology 125126 (2002) 170178
The emulsion residence time and bubble time fraction were
calculated from the correlations by Thring [13], and the
effective thermal conductivity was calculated from the
model of Kunii and Smith [16] at ambient temperature.
The component h
c
of the convective heat-transfer coefcient
in the total heat-transfer coefcient h
w
in Eqs. (24) and (25)
does not therefore vary with bed temperature.
Radiation becomes important with increase in bed tem-
perature. About 25% of the total heat ux comes from
radiation at a bed temperature of 1300 K, which was also
reported by Shafey et al. [17] who simulated conductive
and radiative heat transfers using an unsteady state packet
model.
4.2. Mass transfer
At present the diffusion coefcients D
g
of hydrocarbon
complexes and other kinds of gaseous molecules in uidised
beds, and the diffusion coefcients D
C
of carbon in most
alloy steels are not well documented and the mass-transfer
coefcient b
2
can only be determined under some special
experimental conditions. The so-called ``inverse method''
should therefore be used, by putting the measured distribu-
tion of the concentration of elements in the thermomecha-
nically treated layer and other heat treating parameters into
the mathematical model and then repeating the calculation
until D and b
2
are extracted.
In the present work, D
g
was assumed to be the diffusion
coefcient of CH
4
in air, D
C
be a constant of 0:248
10
4
mm
2
/s, while b
2
was determined using the model
developed by Zhu et al. [18].
The carbon distribution in a uidised bed with 1.2%
carbon and within an immersed object at different carburis-
ing times are shown in Fig. 9. This shows that the change in
carbon concentration in the emulsion and the bubble near the
surface of the immersed object is very small. The carbon
Fig. 8. Effect of bed temperature on the heat-transfer coefficients.
Fig. 9. Carbon profile curves in fluidised bed and within workpiece.
W.M. Gao et al. / Journal of Materials Processing Technology 125126 (2002) 170178 177
transfer is mainly controlled by the mass-transfer coefcient
b
2
and diffusion coefcient of carbon D
C
in the workpiece.
Because the mass-transfer coefcient b
2
on the surface of
the workpiece is only determined by the temperature as
indicated in Eq. (22), a result similar to that for heat transfer
is found. When the immersed object is at a steady carburis-
ing temperature, a constant surface mass-transfer coefcient
is obtained. However, b is actually affected by the carburis-
ing atmosphere and the uidisation parameters, namely
particle diameter, uidisation number, and the bed tempera-
ture. Further experimental work is needed to provide the
basis for an adequate understanding of the mass-transfer
mechanism in the uidised bed carburising process to allow
modication of the mass-transfer model.
5. Conclusions
The gas momentum, heat and mass transfers have been
calculated based on the model consisting of a double layer of
particles and a porous medium presented in this work.
Radiative heat transfer was investigated by simulating
photon emission in the particle array. The heat-transfer
model developed here shows a high accuracy for predicting
the coefcients of total and radiative heat transfer. The
convection is approximately constant, 25% of the total heat
rate is radiation in a heat treatment uidised bed at a
temperature of 900 8C. The heat-transfer coefcient on
the immersed surface near the particles is about 610 times
of that on the other surface. The heat and mass-transfer
model based on the simulation of gas dynamics can help to
understand the heat and mass-transfer mechanisms.
A theoretical model with coupling heat transfer, phase
transformation and potential for phase transformation in the
workpiece has been developed. Fick's 2nd lawwas applied to
simulate mass transfer within workpiece in thermo-chemical
processes. The simulation gives the transient heat and mass-
transfer coefcient, heat and mass-transfer ux on any sur-
face of workpiece, non-steady distributions of temperature
and carbon at any position of the workpiece during the heat
treatment process. This provides an insight into the optimisa-
tion of heat treatment processes in auidised bed furnace.
References
[1] R.W. Reynoldson, Heat treatment in fluidized bed furnaces, ASM
International, Materials Park, OH, 1993.
[2] T. Fukuda, H. Hattori, Surface heat treatment using fluidized beds,
Heat Treat. Met. 15 (1988) 5358.
[3] S.C. Saxena, Heat transfer between immersed surfaces and gas-
fluidized beds, Adv. Heat Transfer 19 (1989) 97190.
[4] K.R. Sharma, R. Turton, Mesoscopic approach to correlate surface
heat transfer coefficients with pressure fluctuations in dense gas
solid fluidized beds, Powd. Technol. 99 (2) (1998) 109118.
[5] K. Yoshida, D. Kunii, O. Levenspiel, Heat transfer mechanisms
between wall surface and fluidized bed, Int. J. Heat Mass Transfer 12
(1969) 529536.
[6] C.L. Tien, Thermal radiation in packed and fluidized bed, J. Heat
Transfer 110 (1988) 12311242.
[7] D. Dragomir, L. Druga, The advantages of fluidized bed carburizing,
Mater. Sci. Eng. A 302 (1) (2001) 115119.
[8] G.D. Mazza, C.A. Berto, G.F. Barreto, Evalution of radiative heat
transfer properties in debse particulate media, Powd. Technol. 62 (2)
(1991) 137144.
[9] V.N. Korolev, N.Y. Syromyatnikov, W.M. Tolmachev, The structure
of fixed and fluidized granular beds near immersed walls, J. Eng.
Phys. 21 (6) (1971) 973.
[10] S.C. Saxena, P.D.C. Patel, Variation of solids population profile
around horizontal tube surfaces if different diameters immersed in a
three-dimensional gas-fluidized bed, AIChE Symp. Sec. 84 (262)
(1988) 3649.
[11] W.A. Johnson, R.F. Mehl, Reaction kinetics of nucleation and
growth, Trans. AIME 135 (1939) 416518.
[12] M. Avrami, Kinetics of phase change. Part 1. General theory,
J. Chem. Phys. 7 (1939) 11031112.
[13] R.H. Thring, Fluidized bed combustor for the stirling engine, Int. J.
Heat Mass Transfer 4 (1977) 911918.
[14] D. Pidwerbecki, J.R. Welty, Splash-zone heat transfer in bubbling
fluidized bedsan experimental study of temperature effects,
Exp. Therm. Fluid Sci. 9 (3) (1994) 356365.
[15] H.S. Li et al., An investigation on instantaneous local heat transfer
coefficients in high-temperature fluidized beds: experimental results,
Int. J. Heat Mass Transfer 36 (1993) 43894395.
[16] D. Kunii, J.M. Smith, Heat transfer characteristics of porous rocks,
AIChE J. 6 (1) (1960) 7178.
[17] H.-M. Shafey, A.M. Abd-El-Ghany, A.M. Nassib, An analysis of the
combined conductive-radiative heat transfer between a surface and a
gas-fluidized bed at high temperature, Int. J. Heat Mass Transfer 36
(1993) 22812292.
[18] B. Zhu et al., Experiment study and mathematics model of carbon
transfer coefficient in a fluidized bed furnace, J. Heat Treat. Met. 19
(3) (1998) 26 (in Chinese).
178 W.M. Gao et al. / Journal of Materials Processing Technology 125126 (2002) 170178