Full Paper

A Mathematical Model of Heat Transfer in a Rotary Kiln

Thermo-Reactor
By S.-Q. Li*, L.-B. Ma, W. Wan, and Q. Yao
Rotary kilns are used ubiquitously in the chemical and metallurgical industries. The mechanism of heat transfer in a rotary
kiln is discussed in this paper, in which the effect of rotation is considered in determining heat transfer coefficients. In particular, an extended penetration theory is successfully developed
to describe the heat transfer coefficient of the covered wall
p
to the bulk solid in a rotary kiln, i.e., hcwcb = (vdp/kg + 0.5/ 2kb rb cpb n=u0 )1 (0.096 < v < 0.198). A one-dimensional axial
heat transfer model for an internally heated rotary kiln has been developed. Both predicted temperature profiles and heat
transfer fluxes agree well with the experimental data of Barr et al. The simulated results are used to successfully explain for
the first time the coupling phenomenon of the bulk bed and covered wall temperatures discussed in previous publications.

1 Introduction
Rotary kilns are ubiquitous fixtures of the chemical, metallurgical, and pharmaceutical process industries. They are
commonly used for three purposes: heating, reacting, and
drying of solid material, and in many cases they are used to
achieve a combination of these aims [1]. Within recent years,
rotary kilns have become widely used for the thermal treatment of waste materials (e.g., the incineration of hazardous
waste [24], the gasification of waste tires or wood to obtain
activated carbon [5, 6], and the thermal desorption of contaminated soils [7]). This widespread usage can be attributed
to factors such as the ability to handle varied feedstock, for
example slurries or granular materials having large variations in particle size, or the ability to maintain distinct environments, for example reducing conditions within the bed
coexisting with an oxidizing freeboard [8]. Slow rotation of
an inclined kiln enables the thorough mixing of wastes during their transport from inlet to outlet, and flexible adjustment of the residence time can yield the optimum conditions
for the thermal destruction of solid wastes [9, 10]. In the design or modeling of rotary kilns, four important aspects
should be considered from a process engineering point of
view: heat transfer, flow of material through the rotary kiln,
gas-solid mass transfer, and reaction kinetics. Heat transfer
is the most important of these aspects because in many practical cases heat transfer limits the performance of the rotary
kiln [8, 11].
Compared with other gas/solid reactors, such as packed
bed and fluidized bed, the intrinsic features of heat transfer
in a rotary kiln include:
I.
Heat transfer coefficients between the gas and rotating
wall, and between the gas and rolling bed surface are
all influenced by the rotation rate of the drum.
II. The inner wall contacts periodically with the high tem
[*]

Shui-Qing Li (author to whom correspondence should be addressed,

[email protected]), L.-B. Ma, W. Wan, Q. Yao, Key
Laboratory for Thermal Science and Power Engineering of the Ministry
of Education, Department of Thermal Engineering, Tsinghua University,
Beijing, 100084, China.

1480

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

perature gas and at intervals with the bulk bed, so heat

from the high temperature gas absorbed by the wall is
delivered indirectly to the bed. This heat transfer process is described as a storage-release mechanism,
which is dependent on the heat transfer mechanism between the inner wall and bulk bed [12]. Thus, it is very
important to determine the heat transfer coefficient between the inner wall and solid bed.
III. Radiation heat transfer can not be ignored in a relatively higher temperature environment, especially when the
temperature exceeds 1000 C.
In general, a rotary kiln can be classified as internally
heated or externally heated. The internally heated mode
(see Fig. 1a)), mainly used as a waste incinerator, has high
temperature flue gas as its heat source. This kind of heating
mode makes use of coal, gas, or oil directly. The fuel goes
through a combustor or burner where it is mixed with air or
oxygen to generate high temperature gas. The high temperature gas is then introduced into the drum in a direction
either co-current or counter-current to the solid flow. However, the externally heated rotary kiln (see Fig. 1b)) can use
either electric heat flux or the outer wall that is indirectly
heated by the high temperature flow as its heat source. The
externally heated kiln is often adopted as a pyrolyzer or gasifier of a special kind of waste. The largest difference between the two modes is that the heat loss from the outer wall
to the environment should be considered in an internally
heated mode, while the heat sourced from outer wall should
be considered in an externally heated mode. Heat transfer in
rotary kiln occurs via the following five mechanisms:1)
Qcwcb, heat transfer between the covered wall and covered lower bed. This is an integral process including the
heat convection via unsteady state heat conduction, the
contact of solid particles and the covered wall, and their
radiation.
Qgew, heat transfer between the exposed wall and freeboard gas, including both convection term Qcg ew and radiation term Qrg ew .

1)

List of symbols at the end of the paper.

DOI: 10.1002/ceat.200500241

Chem. Eng. Technol. 2005, 28, No. 12

Full Paper
penetration model, mostly used in packed beds and fluidized
beds, is introduced to evaluate the heat transfer coefficient
between the rotating covered wall and the solid bed. The predicted results agree well with the experimental data of Lehmberg et al. [13]. Based on the analysis of heat transfer mechanisms, a one-dimensional axial heat transfer model for an
internally heated rotary kiln is presented. The theoretical predictions and empirical equations of the heat transfer coefficients are validated by the experimental data in the literature.

Figure 1. Modes of heat transfer in a rotary kiln: (a) internally heated; (b) externally heated.

2 Heat Transfer Between the Covered Wall and

Bulk Bed

Qgeb, heat transfer between the exposed upper bed and

freeboard gas, including both convection term Qcg eb and
radiation term Qrg eb .
Qeweb, heat transfer between the exposed bed and exposed wall, which is only in terms of radiation.
Qsh, heat loss of wall for an internally heated rotary kiln,
or So, external heat resource for an externally heated rotary kiln.
In this paper, the heat transfer mechanisms in an internally
heated rotary kiln, including heat transfer coefficients for
both convection and radiation, are discussed. An extended

The hcwcb heat transfer coefficient between the covered

wall and contact bed plays an important role in the heat
transfer model of a rotary kiln. Tab. 1 summarizes several
previous investigations on the heat transfer between the wall
and particle bed.
In general, the overall coefficient of heat transfer between
the covered wall and bulk bed includes not only the heat
transfer coefficient of the bed surface to bulk solids (hcd
b ),
but that of the wall surface to bed surface (hwb). The penetration theory developed by Schlnder [14] is used to evalutate the former term, hcd
b . In order to evaluate the latter

Table 1. Summary of heat transfer coefficient equations between the wall and bulk solid.
Researcher
Wes et al. [16]
(1974)

Validity

Heat transfer coefficient hcw cb

p
(1) hcw cb 2 kb rb cpb =ptc 2kb 2n=ab u0 1=2
p 1=2
(2) Nu 2 2Pe of which: Nu hcw cb lw0 =kb Pe nR2 u0 =ab

Tcheng & Watkinson [17]

(1979)

(1) hcw

Lehmberg et al. [13]

(1976)

hcw

cb

11:6kb nR2 =ab u0 0:3 =lw0

Rotary kiln

(2) Nu 11:6Pe0:3 of which: Nu hcw

cb

Rotary kiln

s
kb rb cpb 2
p
tc
p

0
cb lw =kb

Pe nR2 u0 =ab

1
1
p p exph2 ab tc erfchab tc
h ab tc h ab tc

Rotary kiln

of which: parameter h ag =kb

Wachters & Kambers [18]
(1964)

hcw

Ferron et al. [19]

(1991)

(1) hcw

cb

p
p
pab tc =2kb pd0 =2kb
cb

2 Nu

p
conditions: n < 10rpm; d 0:00112 u0

Nu
kb =lw0

Rotary kiln

p
2 2Pe1=2

1 h
P
Bj u0
j1

pa1

j2 pu2

exp 2Pe a20

p 1 1
kb rb cpb =ptc
p
of which: hwb up hwp 1 rp kg = 2Rp r hrad ;
h
i
2k
4rs
R
hwp Rg 1 rd
1 ; hrad 1=ew 1=e
Tw3
R ln1 rd
so 1

Schlnder [14]: Heat

penetration model (1982)

hcw

cb

1=awb 2

Schlnder [14]: simplified

penetration model (1982)

hcw

cb

vdp =kg 2

p 1
kb rb cpb =ptc

Basu [15]: surface renewal model

(1994)

hcw

cb

vdp =kg 2

p 1
kb rb cpb =ptc

Chem. Eng. Technol. 2005, 28, No. 12

Rotary kiln

gas film thickness: v 0:02871

http://www.cet-journal.de

eb

parameter: v 0:085

Packed bed and fluidized bed

Packed bed and fluidized bed

Circulating fluidized bed

0:581

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1481

Full Paper
term, hwb, Schlnder assumed there is a thin layer of gas film
existing at the contact point of the round particle and slippery wall, which is less than the mean free path of the gas
molecule. Schlnder's work has been widely used in the heat
transfer model of fluid beds and packed beds [14, 15]. Wes et
al. [16] used the penetration theory in a rotary kiln reactor,
and assumed that the the wall temperature and bulk temperature at the contact point were equal. Thus, the coefficient
hwb was ignored, resulting in an excessively large estimate of
the total heat transfer coeffiecient between the covered wall
and bulk bed. Lehmberg et al. postulated that there was a
small gap between the wall and bed, and a temperature jump
occured at the gap [13]. The gap turned out to be much
greater than the mean free path of a gas molecule, which is
different from Schlnder's assumption. They incorporated
an equation of heat flux across the thin gap into the penetration theory and obtained an analytic result (as seen in
Tab. 1). However, it is too complicated and cannot be used
flexibly. Silimilar attempts had been conducted by Tscheng
and Watkinson,Wachters and Kramers, and Ferron et al.
[1719]. However, a general formula has not existed till now.
In this paper, an extended model based on penetration
theory is developed to describe the wall-solid heat transfer
in a rotary kiln reactor. As shown in Fig. 2, the heat transfer
of the covered wall and bulk bed is mainly controlled by
three aspects:
1/hwb, contact resistance caused by the gas film between
the covered wall and bed surface.
1=hcd
b , thermal resistance due to unsteady state heat conduction from the bed surface to the bulk bed, in which the
temperature reduces from Tb(0) to Tb().
1=had
b , thermal resistance due to advection heat transfer
within bulk solids [10].

and the much thicker plug flow' region where the material
is carried upward by the rotating wall of the kiln. Particles
roll down continually and form a constant layer on the bed
surface, then enter the stagnant plug-flow region where they
are carried upward by the rotating wall until they re-enter
the active layer and roll again. The active layer itself is characterized by vigorous mixing, and hence a high rate of surface renewal which promotes heat transfer from the freeboard. The intensive mixing of particles in the active layer
makes the temperature uniform, which has been verified by
experiment [8]. So, we can assume that the temperature
within the bulk solids is uniform, denoted as Tb(). Thus,
the advection heat coefficient had
b can be considered infinite,
while resistance 1=had
can
be
approximated
to zero.
b
2.2 Thermal Resistance from the Bed Surface
to the Bulk Bed, 1/hcd
b
The heat transfer between the bed surface and bulk bed
can be described by a one-dimensional transition conduction
problem [16], which is shown in Fig. 3. The analytical result
for this problem is:
q
hcd
2 kb rb cpb =ptc
b

(1)

The contact time of the material and covered wall can be

replaced by tc = u0/2pn, then:
q

2
2kb rb cpb n=u0
hcd
b

(2)

Figure 2. Mechanism of heat transfer between the covered wall and bulk bed
in a rotary kiln.

2.1 Thermal Resistance Caused by Solid Mixing, 1/had

b
When the rotary kiln is operated in a rolling mode, the behavior of the bed material can be characterized by two distinct regions: the thinner active layer, which is formed as the
granular material flows down the sloping upper bed surface,
1482

Figure 3. Unsteady heat conduction in the penetration boundary layer of the

bed.

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

http://www.cet-journal.de

Chem. Eng. Technol. 2005, 28, No. 12

Full Paper
Among the previous experiments on hcwcb, the data obtained by Lehmberg et al. [13] are representative, as cited by
other researchers [17, 19]. The data from Lehmberg et al.
are also used to verify the validity of the wall-to-bed heat
transfer model in this paper.
Tab. 2 presents the thermophysical properties of the materials used by Lehmberg et al. Fig. 4 shows the comparison
between extended penetration theory (Eqs. (67)) and the
experimental results of Lehmberg et al., and the non-dimensional terms are given in Fig. 5. As the particle size increases

2.3 The Contact Resistance Between the Wall and Bed

Surface, 1/hwb
Schlnder postulated that the contact resistance of the gas
film is composed of three parts: conduction within the gas
film, conduction between single particles and the covered
wall, and radiation from the wall to particles [14]. Thus, hwb
can be expressed as:

hwb rp hwp 1
in which hwp
and hrad

2kg
R


p
rp kg = 2Rp r hrad

 

rd
R
ln 1
1
R
rd

(3)

1

Table 2. Thermophysical properties of the experimental materials (Lehmberg

et al. [13]).

Cs
4
T3 .
1=ew 1=es 1 w

Sand

No.

157.5

323.5

794.0

1038

137.0

1240

1320

1480

1480

675

dp [lm]
3

rb [kg/m ]

For a plug-flow packed bed, Schlnder gave a simplified

empirical formula for hwb:
hwb

Materials

1
1

k =d

k =d
v g p 0:085 g p

(4)

Soda

cpb [J/kg K]

776

776

776

776

1144

kb [W/m K]

0.213

0.223

0.251

0.257

0.122

ab [106 m2/s]

0.221

0.218

0.218

0.224

0.158

When it comes to a fluidized bed, Lints and Glicksman

[20] proposed a similar empirical formula:
hwb

1
1

k =d
v g p 0:02871 eb

0:581

kg =dp 0.1< v < 0.5 (5)

Both Schlnder and Lehmberg et al. have proposed the

contact resistance due to the gas film, but they have paradoxical conclusions as to the depth of film. As suggested by
Schlnder, Eq. (3) can be used provided the film thickness is
less than that of the free path of a gas molecule, which differs from Lehmberg et al. The empirical Eqs. (4) or (5) that
were mostly used in modeling packed beds and fluidized
beds are more suitable for the contact resistance in a rotary
kiln reactor.

2.4 The Total Heat Transfer Coefficient between the Covered

Wall and Bulk Bed, hcwcb

Figure 4. Comparison between extended penetration theory (Eq. (6)) and experimental results.

In general, the total heat transfer coefficient between the

covered wall and bed (hcwcb) can be expressed as:
hcw

cb

1=1=hwb 1=hcd
1=had

b
b
q 1
vdp =kg 2 2kb rb cpb n=u0

(6)

Non-dimensionalisation of Eq. (6) gave:

Nud

pr
p
1
2

(7)

Ped

where Nud = hcwcbdp/kg, Ped = (dp/kg)2rbcpbkb1/tc, and

tc = u0/2pn.
Chem. Eng. Technol. 2005, 28, No. 12

http://www.cet-journal.de

Figure 5. Non-dimensional comparison between extended penetration theory

(Eq. (7)) and experimental results.
 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1483

Full Paper
from 157.5 mm to 1038 mm, the v value increases from
0.096 to 0.198. As all five kinds of material are concerned,
the optimum parameter v is 0.10 (see Fig. 5). The extended
model of hcwcb agrees well with experimental data, while
the model by Wes et al. overestimates the values as it does
not take account of the contact resistance. The v values for
the packed bed and fluidized bed are 0.085 and 0.21.0, respectively [14, 20]. The rotary kiln has a better mixing than
the packed bed, but a worse mixing than the fluidized bed,
so a v value between 0.0960.198 is appropriate. It is
concluded that the self-extended penetration theory can be
successfully used to evaluate the wall-to-bed heat transfer
coefficient hcwcb.

3 The Convective Heat Transfer Coefficient in a

Rotary Kiln
The convective heat transfer includes two aspects: heat
transfer between the freeboard gas and exposed wall
(hcg ew ), and heat transfer between the freeboard gas
and exposed upper bed surface (hcg eb ). The classic equation
for turbulent convection in a non-rotating drum,
hgw(b) = 0.023(kg/D)Reg0.8Pr0.4, is not suitable for the calculation of hcg eb and hcg ew in a rotary kiln.
Tscheng and Watkinson [17] conducted a series of experiments on the influence of the rotation rate, gas flow rate,
and percent fill on the convective heat transfer in a rotary
kiln. Regression analysis of the experimental data led to the
following dimensionless form:
De =kg 1:54 Re0:575
Rew 0:292
g

Nug

ew

hcg

ew

Nug

eb

hg

eb De =kg

0:46 Re0:535
Re0:104
g
g
w

(8)

0:341

(9)

Where the flow Reynolds number Reg = VgDe/m,

rotational Reynolds number Rew = De2w/m, percent fill
g = (u0 sinu0)/2p, and equivalent diameter De = 0.5D(2p
u0 + sinu0)/(p u0/2 + sinu0/2) were introduced. Actually,
De is a function of the percent fill of solids. Provided 1600 <
Reg < 7800 and 20 < Rew < 800, then Eqs. (89) are valid.

the freeboard gas, inside wall, and bed surface [22]. It is difficult to directly simulate the radiation in a rotary kiln. In general, radiative transfer is negligible at < ~ 300400 C, but
comparable with convective heat transfer at ~700900 C, and
dominant at > 1000 C [22, 23].
The radiative conditions that exist within the kiln freeboard were simplified by making the following assumptions
[23]:
The surfaces of the wall and bed are taken to be radiatively gray because the spectral emissivity of the solid and
wall refractory are not well known.
The inlet or outlet surfaces of the kiln are adiabatic.
The gas is taken to be radiantly gray, even though it contains CO2 and H2O which emit and absorb radiation in distinct bands. The error due to this assumption is about 20 %.
The material is mixed sufficiently. The temperature of the
surface and gas is uniform at any axial position. The effect
of any radial temperature gradient is ignored, while the
circumferential wall temperature is approximated by four
zones to eliminate its effect.
The effect of any axial temperature gradient in the solid,
wall, and freeboard gas is negligible. The error brought by
this approximation is small due to the large L/D ratio of
the kiln.
No fire or flame exists in the rotary kiln.
On the basis of the above assumptions, the one-zone wall
model to approximate the circumferentially changing inside
wall temperature was used to determine the radiant heat
flows, and hence the radiative heat transfer coefficients within the freeboard area [22, 23]. The resistance analogs of the
models are shown in Fig. 6, while the view factors are given
in Tab. 3.
The radiant heat transfer coefficient from the gas to the
bed surface is represented by:
hrg

eb

Qr
E J =Rg b
 g b  g b

Ab Tg Tb
Ab Tg Tb

(10)

4 The Radiative Heat Transfer Coefficient in a

Rotary Kiln
The radiation existing within the freeboard of a rotary kiln
is enclosed by the exposed bed surface, inner exposed wall,
freeboard gas, and inlet/outlet surface [21]. As shown in
Fig. 1, radiative heat transfer occurs between the gas volume
and its boundary surfaces (the inside wall and exposed bed).
Additional radiative exchanges occur between area elements
on the inside wall and other areas on both the wall and the exposed bed surface. All these heat transfers are influenced by
the emissivity characteristics and temperature distribution of
1484

Figure 6. Radiation network within the freeboard of a rotary kiln using onezone wall model.
Table 3. View factors for a one-zone wall model of radiation.
Fbg = Fwg = Fbw = 1.0
Aw = R(2p-u0)

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Ab = Ru0

http://www.cet-journal.de

Chem. Eng. Technol. 2005, 28, No. 12

Full Paper
The radiant heat transfer coefficient from the gas to the
exposed wall is given by:
P r
P
Qg wi
Eg Jwi =Rg wi
r





(11)
hg ew

w
Aw Tg Tw
Aw Tg T
and the radiant heat transfer coefficient from the exposed
wall to the bed surface is expressed as:
P r
P
Qwi b
Jwi Jb =Rb wi
r

(12)
hew eb
w T
w T
Ab T
Ab T
b
b

Figure 7. Schematic of an internally heated, counter-current rotary kiln reactor.

5 Heat Loss of the Outer Wall

Tb(L). The counter-current gas of Tg(L) enters at z = L and

is cooled to Tg(0).
As shown in Fig. 7, the thermal component of these onedimensional models can be derived by considering a transverse slice which divides the section into separate control
volumes of freeboard gas and solid bed. Under steady state
conditions, the energy conservation for any control volume
requires that:

The heat loss from the inner wall to the outer environment
includes three heat transfer process: the conduction resistance between the inner and outer wall (1=hcd
w sh ), the convective resistance between the outer wall and the environment (1=hcsh a ), and the radiant resistance (1=hrsh a ). So the
total resistance of heat loss can be presented by (calculated
on the basis of the outer wall area):
1=hw

1=hcd
w

sh

1=hrsh

1=hcsh

The conduction resistance 1=hcd

w
outer wall was calculated by:
1=hcd
w sh

between the inner and

sh

(14)

3
C0 esh rTsh

Where it satisfies C0 1 TTa

sh

2  3
TTa TTa
sh

(15)

The convection heat transfer coefficient hcsh a between the

outer wall and environment can be calculated from [22, 24]:


Re
w
p

0:2 ,
when:
Gr
0:36 
0:35
0:11ka Pr
hcsh a
(16a)
0:5Re2w Re2a Gr
D


0:3
Rew
k Pr
N
when: p
(16b)
< 0:2 , hcsh a a
CRea
D
Gr
where Rew, Rea, and Gr are the separately rotating Reynolds, gas Reynolds, and Grashof numbers.

http://www.cet-journal.de

Qcw

n_ i DHg

(17)

m
_ b Cps

dTb
Qg
dz

cb

n_ i DHb

(18)

ew

Qeb

ew

Qcb

(19)

cw

Gas phase:
P

m
_ g Cpg

dTg
hcg
dz

ew

hcg

hrg
eb

ew

hrg

Aew Tg
eb

Aeb Tg

Tw
Tb

(20)

Bed phase:

6 Validation of the Steady State Heat Transfer

Model in an Internally Heated Rotary Kiln

Chem. Eng. Technol. 2005, 28, No. 12

eb Qew eb

Where in Eqs. (1718), n_i is the production rate for various species involved in the chemical reaction, while DH is
the corresponding enthalpy of the reaction.
For the inertia material, the chemical reaction can be ignored. Using the heat transfer coefficients in Eqs. (1719)
leads to:

Fig. 7 shows a general schematic of an internally heated,

counter-current rotary kiln reactor. A cold material of Tb(0)
enters the inclined kiln at z = 0 and has to be heated to

ew Qg eb

Wall phase: Qsh Qg

sh

6.1 Steady State Model Development

dTg
Qg
dz

m
_ g Cpg

One additional condition is that no net energy accumulation occurs within the wall, so the energy conservation at
wall is:

between the

Bed phase:
P

D
D
o ln o
2Kw
Di

The radiation heat transfer coefficient hrsh

outer wall and ambience was:
hrsh

Gas phase:

(13)

m
_ b Cps

dTb
hcg
dz

hrew

eb

eb

hrg

Aeb Tw

eb

Aeb Tg

Tb hccw

cb

Tb

Acw Tw

Tb

(21)

Wall phase:
hw

T0 hcg

sh Ash Tw

hrew

eb

Aeb Tb

ew

hrg

ew

Tw hccw

Aew Tg

Tw

Acw Tb

Tw

cb

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

(22)
1485

Full Paper
The above differential equations can be solved using the
four-order Runge-Kutta method combined with the inlet
conditions of the freeboard gas and bulk solid.

6.2 Verification of the Steady State Model by Experiment

In the present work, the data from Barr et al. [25] is used
for model verification. In their experiments the internallyheated rotary kiln has an internal diameter of 0.406 m and
length of 5.5 m, and the diameter of the outer wall is
0.610 m. A gas burner is fitted in the end of the rotary kiln
to produce heat flow. The run conditions of the pilot kiln
trials in model validation, as well as relevant physical properties of the gas and solid, are summarized in Tab. 4.

tion cannot meet the required precision. This is mainly because the heat loss was not considered in Imber and Paschkis's model. In addition, this analytical solution cannot be
extended to an externally heated rotary kiln.
Fig. 9 displays the comparison of the measured temperature profiles and those predicted by the current one-dimensional model. It can be seen that the calculated data agree
with the experimental data very well. At the kiln inlet the
predicted bed temperature is a little lower than the experimental value because of the inlet effect of the kiln. As
shown in Fig. 9, the bed temperature increased rapidly within the initial 1.5 m of the kiln inlet, and then the increment
with axial position is very small. It can be explained that the
two parameters (hcwcb and hcg eb ) play a crucial role in the
heating process of solids in the inlet zone.

Table 4. Run conditions of rotary kiln trials in model validation from Barr [25].
Struction
condition

range

Operation
condition

range

Physical
characters

range

5.5

1.5 rpm

rg

0.69 kg/m3

0.406

Wg

430 g/s

mg

4.0106 m2/s

12 %

Vg

0.22 m/s

kg

0.042 W/m k

20

Wb

614 g/s

rb

1650 kg/m3

< 0.1

Vb

1.72.0 mm/s kb

0.27 W/m k

Previously, Imber and Paschkis gave a simplified analytical solution for the axial temperatures of the wall, bed, and
freeboard gas in a one-dimensional rotary kiln heat exchanger [26]. This analytical solution has been used to calculate
the heat balance in the design of rotary kilns, e.g., Kiang and
Metry used it in the design of a hazardous waste incinerator
[27]. Fig. 8 shows the comparison between the simplified analytical results and experimental data by Barr et al. [25]. It is
apparent that the actual temperatures of the wall surface
and bulk bed are much lower than the analytical results at
the exit section, which indicates that the simplified predic-

Figure 8. Comparison of predicted axial temperature profiles by simplified

analysis and experimental data.

1486

Figure 9. Comparison of predicted axial temperature profiles by the one-dimensional model and experimental data.

It is noted that the temperatures of the solid bed and inside wall were closely coupled at all axial locations in Barr's
experiments [25]. Barr et al. emphasized that this close coupling of temperatures was a feature of all trials, but didn't offer an explanation of this fact. A similar phenomenon was
also reported in other literature [17]. The predicted temperature profile results are the first to verify the coupling phenomenon. Boateng conducted similar modeling work, but
didn't mention and verify the coupling in his study [8]. It can
be inferred that the haphazard coincidence of all heat transfer coefficients leads to the coupling of the wall and bed
temperatures. The excellent agreement of both predications
and experiments on the temperature coupling phenomenon
implies that proper selection of heat transfer coefficients
(conductive or radiative) is very important in the simulation
of a rotary kiln.
The comparison of the predictions and measurements for
the three heat transfer rates, i.e., the net rate of heat transfer
to bed Qb, heat loss from outer wall Qsh, and heat transfer
from the covered wall to the contacting solid bed Qcwcb is
shown in Fig. 10. The good agreement between them shows
that the one-dimensional model developed in this paper is
satisfactory. The net input to the bed, Qb, declined rapidly

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

http://www.cet-journal.de

Chem. Eng. Technol. 2005, 28, No. 12

Full Paper

Figure 10. Predicted and measured net heat transfer rate variation as a function of axial position.

Figure 11. Net heat fluxes to the solid bed from different modes of transfer.

with axial distance adjacent to the charge end and then

varied smoothly. The experimental data of Qb at inlet zone
(L < 1.5 m) is not provided by Barr et al. When the length is
in the range of 25 m, the predicted Qb retains a steady value between 1.01.2 kW/m, which fits well with the experiments. The variation of Qcwcb is similar to that of Qb. However, heat loss from the outer wall Qsh increased along the
axial location. It reached 4 kW/m at the kiln outlet.
In the initial zone of the kiln, the net input to the bed, Qb,
is much larger than heat loss from the outer wall, Qsh. Therefore, the bed and wall temperature increased rapidly within
this zone, as shown in Fig. 9. The heat loss is much higher
than the net heat input to the bed in the posterior part of the
rotary kiln (L > 2 m). The efficiency of heat transfer becomes low, and then the incremental changes of the bed and
wall temperatures are small. The above non-uniformity of
heat transfer as a function of axial position must be considered in the design of a rotary kiln reactor. In order to optimize kiln design, different thicknesses of the refractory and
insulating layers should be considered for different zones of
rotary kiln.
The net heat transfer to the solid bed is obtained from the
four contributions, i.e., Qcw cb , Qrew eb , Qcg eb , and Qrg eb .
The net fluxes from the above four heat transfer modes to
the bulk bed are shown in Fig. 11, and relative percentages
of each contribution are shown in Fig. 12. In the low temperature kiln inlet zone, the heat transfer from the covered wall
Qcwcb and that from gas convection Qcg eb mainly occupy
higher percentages of the net heat input to the bed Qb,
which are 45 % and 42 %, respectively. The proportion of
radiative heat transfer can be negligible within this zone because of the quite low temperature. Since radiative heat
transfer is linear to temperature raised to the fourth power,
the thermal resistance associated with radiative exchanges
between the gas and the exposed bed and wall surface declines rapidly with increasing temperature. So the percentage of radiative heat transfer increases rapidly along the

Chem. Eng. Technol. 2005, 28, No. 12

http://www.cet-journal.de

Figure 12. Contribution of various heat transfer modes to the solid bed.

axial position. As shown in Fig. 12, the proportions of Qcwcb

and Qrew eb in Qb reduce gradually because the difference
between the temperatures of the material and wall reduces
more and more. Reversed heat transfer even appeared from
the solid bed to the wall.
In addition, all the heat transfer coefficients used in the
one-dimensional model simulation in this paper are shown
in Tab. 5. The axial gradients of radiative coefficients hrg ew,
hrg eb , hrcw eb , and hrsh o are shown in Fig. 13.

Table 5. Summary of heat transfer coefficients of various modes used in onedimensional model.
Convective
coefficients

hcgew

hcgeb

hccweb

hcsho

hcwsh

W/m2K

~ 8.45

~ 32.0

~ 121

~ 6.70

~ 4.43

Radiative
coefficients

hrg

hrg

hrcw

hrsh

W/m2K

7.041.4

6.310.1

ew

eb

5.842.3

eb

7.2151

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1487

Full Paper

h ew-eb

100

Radiative Heat Coefficient /W/m k

Symbols used
80
60
40

[m]
[m2s1]
[]
[kJ kg1K1]
[m]
[m]

Di
Do
dp
E
F
Gr
h
DH
J
k
L
lw

m
MRT
MVF
Nu
n
n
Pe
Pr
Q
R
Re
r
So
T
tc

[m]
[m]
[m]
[W m2]
[]
[]
[W m2K1]
[J mol1]
[W m2]
[W m1K1]
[m]
[m]
[kg s1]
[min]
[min]
[]
[rpm]
[mol m1s1]
[]
[]
[W m1]
[m]
[]
[m]
[W m1]
[K]
[s]

h g-ew

20

h sh-o

h g-eb

10
8
6
0

L /m
Figure 13. The gradients of the radiation heat transfer coefficient as a function
of axial position.

7 Conclusions
The modes of heat transfer, conductive or radiative, in a
rotary kiln are discussed, and the effect of rotation on heat
transfer is considered. In particular, the penetration theory
widely used for fluidized beds and fixed beds is successfully
developed to evaluate the heat transfer coefficient of the
covered wall and bulk solid in a rotating drum. This empirical formula is expressed as:
q
hcwcb = (vdp/kg + 0.5/ 2kb rb cpb n=u0 )1 (0.096 < v < 0.198)
A one-dimensional axial heat transfer model for an internally heated rotary kiln has been developed. Both predicted
temperature profiles and heat transfer fluxes agree well with
experimental data published by Barr et al. The simulated
results explain for the first time the interesting coupling phenomenon of the bed and wall temperatures observed in previous literature. It is also verified by the model that both
Qcw cb and Qcg eb play a crucial role in the fast heating of
solids at the kiln inlet.

Acknowledgements
This work was partially funded by the Nation Natural
Science Funds of China (No.50306012) and the National
Key Basic Research and Development Program
(No.2002CB211606). Part of the revised work on this paper
was carried out during a visit to the Institute of Particle
Science and Engineering, University of Leeds. S.-Q. Li is
grateful to Dr. Y.-L. Ding for his hospitality and encouragement during his stay at Leeds. The kind help with this work
by Professors J.-H. Yan and Y. Chi, Zhejiang University, is
greatly acknowledged.
Received: July 17, 2005 [CET 0241]

1488

A
ab
C
Cp
D
De

heat transfer area per unit kiln length

effect thermal diffusivity
correction factor
specific heat capacity
diameter of rotary kiln
equivalent diameter of the freeboard
region
internal diameter of rotary kiln
external diameter of rotary kiln
diameter of particles
radiometric force
view factor
Grasholf number
heat transfer coefficient
reaction enthalpy
effective radiation
thermal conduction coefficient
length of rotary kiln
exposed wall circumference
mass flow of gas or solid
mean residence time of solid
material volumetric flow of solid
Nusselt number
rotational rate of rotary kiln
species ratio of component
Peclet number
Prandtl number
net heat transfer
Radius of rotary kiln
Reynolds number
radius of solid
external heat resource
absolute temperature
contact time between bulk solid and
covered wall

Greek Symbols
a
b
d
e
eb
x
g
h
r
r
C
J

[rad]
[rad]
[m]
[]
[]
[rad s1]
[]
[rad]
[kg m3]
[W m2K4]
[]
[m]

u
v

[rad]
[]

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

slope angle of rotary kiln

slope angle of bulk bed
roughness
emissivity
particle concentration in the bulk bed
angular velocity of rotary kiln
packing ratio of solid
packing angle of bed
density
blackbody radiation constant
longitudinal fill ration of solid bed
modifying value of gas molecular mean
free path
half central angle of sectional solid bed
thickness of gas film

http://www.cet-journal.de

Chem. Eng. Technol. 2005, 28, No. 12

Full Paper
t
rp

[m2 s1]
[rad]

kinematic viscosity
ratio of covered particle on the wall

Superscripts and subscripts

a
b, s
cb
cw
eb
ew
sh
w
wb
wp
f
g
sh
ad
c
cd
r, rad

environment air
bulk bed, solid
bed contacted with wall
covered wall
exposed bed surface
exposed wall surface
outer wall (shell)
inside wall
wall to bed
wall to particle
heating furnace for externally heated
kiln
freeboard gas
outer wall (shell)
advection heat transfer
convection heat transfer
conduction heat transfer
radiation heat transfer

References
[1] R. J. Spurling, Ph.D. Thesis, Cambridge University 2000.
[2] H. W. Schenck, J. O. L. Wendt, A. R. Kerstein, Combust. Sci. Technol.
1996, 116117, 427.
[3] W. D. Owens, Combust. Flame 1991, 86, 101.
[4] F. Marias, Comput. Chem. Eng. 2003, 27, 813.
[5] I. Abe, Carbon 2001, 39, 1485.
[6] G. San Miguel, G. D. Fowler, M. Dall'Orso, C. J. Sollars, J. Chem.
Technol. Biotechnol. 2001, 77, 1.
[7] V. A. Cundy, F. L. Larsen, J. S. Lighty, Can. J. Chem. Eng. 1996, 74, 63.
[8] A. A. Boateng, P. V. Barr, Int. J. Heat Mass Transfer 1996, 39, 2131.
[9] S.-Q. Li et al., Powder Technol. 2002, 126, 217.
[10] S.-Q. Li et al., Powder Technol. 2002, 126, 228.
[11] L. Yang, B. Farouk, J. Air Waste Manage. Assoc. 1997, 47, 1189.
[12] J. Kern, Int. J. Heat Mass Transfer 1974, 17, 981.
[13] J. Lehmberg, M. Hehl, K. Schgerl, Powder Technol. 1997, 18, 149.
[14] E. V. Schlnder, in Proc. of the 4th Int. Heat Transfer Conf. (Eds:
U. Grigulll, E. Hahne, K. Stephan, J. Straub), Hemisphere, Mnchen
1982.
[15] P. Basu, Chem. Eng. Sci. 1990, 45, 3123.
[16] G. W. J. Wes, A. A. H. Drinkenburg, S. Stemerding, Powder Technol.
1976, 13, 185.
[17] S. H. Tscheng, A. P. Watkinson, Can. J. Chem. Eng. 1979, 57, 433.
[18] L. H. J. Wachters, H. Kramers, in Chem. Reaction Eng.: Proc. 3rd
European Symposium, Pergamon Press, Oxford 1964.
[19] J. R. Ferron, D. K. Singh, AIChE J. 1991, 37, 747.
[20] M. C. Lints, L. R. Glicksman, AIChE J. 1994, 40, 297.
[21] B. G. Jenkins, F. D. Moles, Trans. IchemE 1981, 59, 17.
[22] J. P. Gorog, J. K. Brimacombe, Metallurgical Trans. B 1981, 12B, 55.
[23] J. P. Gorog, T. N. Adams, J. K. Brimacombe, Metallurgical Trans. B
1982, 13B, 153.
[24] N. K. Kim, J. E. Lyon, N. V. Suryanarayana, Ind. Eng. Chem. Process.
Des. Dev. 1986, 25, 843.
[25] P. V. Barr, J. K. Brimacombe, Metallurgical Trans. B 1989, 20B, 391.
[26] M. Imber, V. Paschkis, Int. J. Heat Mass Transfer 1962, 5, 623.
[27] Y. H. Kiang, A. A Metry, Hazardous Waste Processing Technology, Ann
Arbor Science Publisher Inc., Michigan 1982.

______________________

Chem. Eng. Technol. 2005, 28, No. 12

http://www.cet-journal.de

 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1489