Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Full Paper

Modeling Gas-Solid Reactions in the Bed of a Rotary Kiln


By Klaus Elgeti*

When dealing with gas-solid reactions in rotary kilns, it is necessary to realize that the total particle surface within the granular
bed can be much larger than the outer surface of the bed. Depending on the reaction conditions this inner surface can contribute
considerably to the chemical conversion in the kiln. In this paper, a model is presented, which describes the reaction within the
bed for the case of bed movement according to the cascade mode. In this case, gas is drawn into the rotating bed together with the
particles. As a key quantity, an effectiveness factor g of the bed is defined. It is the ratio of the actual conversion to the conversion
that would occur if the concentration of the reacting component remained unchanged throughout the bed, i.e. at its entrance
concentration. An evaluation for reactions of order m shows this factor to be more than 25 % when the Damköhler number is
smaller than 2. It approaches 100 % as the Damköhler number approaches 0. The Damköhler number used in this paper contains
the void fraction of the particle bed in its denominator.

1 Introduction

In many practical applications the solid in a rotary kiln kiln tube rolling particles
reactor consists of flowable granular material. The transverse
motion of the particles caused by the rotation of the kiln can
have quite different forms, depending on the Froude number,
the filling degree and the friction properties. This has been
extensively studied by Henein, Watkinson and Brima-
combe[1]. Quite recently, Mellmann and Specht [2] presented
equations which allow to calculate the criteria for the
transition between the different modes of motion. In practical rotating bed
applications the cascade mode is predominant. In this mode
the particles in the bed rest against the kiln wall. Every particle
moves along a circumferential path with the angular velocity
Figure 1. Movement of particles in the cascade mode.
of the cylinder until it reaches the surface. From there, it rolls
down the slope entering the bed again at some arbitrary spot in
In the present work, a theoretical model will be developed
the lower half of the slope [3] (Fig. 1). Together with the
to describe the reaction in such a rotary kiln reactor.
particles re-entering the bed, gas will be drawn into the void
The layer of well-mixed rolling particles on top of the bed
space of the bed. This gas can react with the solid phase. Since
may have reaction properties that are different from those of
the total surface of the granular material within the bed is
the rotating part, because the total particle surface is exposed
much larger than the surface of the particles in direct contact
to the gas, and the mass transfer between the gas and the
with the gaseous phase at the bed surface, this ªinnerº reaction
particle surface may be better. The model in this paper
can contribute considerably to the conversion in the kiln. The
neglects such a difference, assuming the layer to be rather thin.
degree of this contribution will depend on whether the reagent
Whether or not this is true, depends largely on the rotational
in the gas is converted quickly ± in this case only a small
speed. If this assumption for a practical application is not
fraction of the inner surface contributes to the reaction ± or
valid, the model applies then to the part of the bed below the
persists partly during its way through the bed.
rolling layer, and the contribution of the rolling layer has to be
Various models for the movement of the granular flow in
added to obtain the total conversion.
radial [4] and in axial directions [5,6] in a rotary kiln reactor
The concentration of the gas entering the bed is the
have been proposed. Despite a number of investigations on the
concentration of the boundary layer between the bed and
thermal conduction within the fixed bed [7±9] and various mass
the gas phase in the kiln. If it differs from the concentration of
transport models for the gaseous phase above the surface of the
the bulk phase, this has to be taken into account by a usual
bed [10±12], the reaction process in the voids of the granular bed
mass transfer calculation.
has not been examined and included in the existing models of
particle movement. Generally, it is assumed that reaction and
mass transfer occur exclusively at the bed surface or in the layer 2 Definition and Calculation of an Effectiveness Factor for the
of rolling particles at the bed surface [12,13]. Total Particle Surface
±
[*] Prof. Dr.-Ing. K. Elgeti, Am Hermannshof 8, D-51467 Bergisch Gladbach, We consider at some given time the process in a small
Germany. section of length DL of the kiln. It is assumed that the particles

Chem. Eng. Technol. 25 (2002) 6, Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2002 0930-7516/02/0606-0651
0930-7516/02/0606-0651$$17.50+.50/0
17.50+.50/0 651
Full Paper

stay practically unchanged in size and reaction properties Here, H is one half of the opening angle of the particle bed as
during one turn through the bed, i.e., that the reaction time defined in Fig. 2.
constant is large compared to the corresponding residence Inserting Eqs. (2) and (3) results in
time. In the case of a heterogeneous catalytic reaction there is,
ZX
of course, no change at all. In both cases it is sufficient to e 
r ˆ ci ui … x† co  uo … x† d x
consider the reaction in the gas phase. Fig. 2 shows the X 2 …H sinHcosH†
definition of geometric variables and parameters in a cross xˆRcosH (5)
section of the kiln.
We now define an effectiveness factor g of the bed as the
ratio of the actual conversion to the conversion that would
occur if the concentration of the reacting component
remained unchanged throughout the bed, i.e. at its entrance
value. This would be the case for n ® ¥.
X The corresponding rate of reaction is
r¥ = r(ci)
Combining with Eq. (5), we have
ϕ Θ r ci e
gˆ ˆ 
r1 X 2 r ci …H sinHcosH†
ψ
x
dx ZX  
co
ui … x† u … x† d x (6)
ci o
xˆRcosH
Figure 2. Definition of geometric variables and parameters in a cross section of
the kiln. Introducing the angle j, defined in Fig. 2, instead of the
radius x, we obtain with
The total volume flow rate of the gas entering the rotating
granular bed in the section DL is1)
x cosj ˆ XcosH
ZX
V_ Gi ˆ DL  e ui … x†d x (1)
sinj
xˆXcosH and dx ˆ XcosH dj
cos2 j
where
e is the void fraction in the bed, r c ecosH
gˆ ˆ  i
ui the flow velocity at the entrance to the bed, r1 Xr ci …H sinHcosH†
x the radial variable as defined in Fig. 2, and
X the radius of the tube. ZH  
cA sinj
The entering flow rate of the educt is ui …j† uo …j† dj (7)
cE cos2 j
ZX jˆ0

n_ i ˆ DL  e  ci ui … x†d x (2)
From this effectiveness factor, which, like ci, may vary along
xˆXcosH
the length of the kiln, one obtains the reaction rate at the cross
where ci is the molar concentration of the educt in the gaseous section considered:
phase at the entrance to the bed.
The total flow rate of educt leaving the particle bed is r = g ´ r(ci)
X
Z
and, inserting this into Eq. (4) the amount of educt converted
n_ o ˆ DL  e co  uo … x†d x (3)
per unit time in the section of length DL
xˆXcosH

where uo is the gas velocity at the exit of the bed. n_ o n_ i ˆ g  r ci  X 2 …H sinHcosH†DL (8)
The effective reaction rate related to the total bed volume
including the void fraction is In order to evaluate Eq. (7), it is necessary to know the
reaction kinetics and the value of the two velocities ui and uo. If
n_ o n_ i
r ˆ (4) there is no volume change during the reaction, both velocities
X 2 …H sinHcosH†DL
are equal and are determined by the rotation of the kiln and
± the bed. Otherwise, they are also influenced by the reaction
1) List of symbols at the end of the paper. process in a complex way.

652 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2002 0930-7516/02/0606-0652 $ 17.50+.50/0 Chem. Eng. Technol. 25 (2002) 6
Full Paper

3 Reaction in the Particle Bed rate will be zero at some point halfway along the path and from
there go in both directions towards the entrance and exit of the
We assume that the gas in the gaps moves along bed, the sign depending on whether the volume increases or
circumferential paths with radius x without mass exchange shrinks. The distribution will be such that the pressure drops to
with neighboring voids. Because neighboring voids differ only the entrance and to the exit are equal. By modeling this
slightly in concentration, this should be a good approximation. pressure drop we could obtain the needed additional equation.
We now consider a differential fixed volume at the radius x However, because of the equality of the pressure drops, the
with radial depth dx and angular depth dw. The mass balance following approximation, which greatly simplifies the solu-
for a steady state process yields tion, will be sufficient for practical application. We start from
the assumption that ± in the case of an increase of volume ± the

d cV_ G ˆ r…c†  DL  xdw  d x (9) flow velocity of the gas is reduced at the entrance by the same
amount as it is increased at the exit, and that its average value
where r(c) is the reaction rate related to the total volume is equal to the rotation speed. This will not be the case if the
including the void fraction, and conversion is completed immediately after the gas has entered
dVÇ G = ue . DL d x is the flow rate of the gas through the the bed. In this case, however, only a small fraction of the inner
area DL dx, which is caused by the rotation and, surface contributes to the reaction and the main contribution
possibly, by the volume change. will come from the outer surface. The flow is considered to be
A change of volume can be caused by the stoichiometry of one-dimensional. The average flow rate then is, as in the case
the reaction and by a temperature change. Here, we will only of constant volume,
consider the stoichiometric influence since a temperature _ ˆ u
dV   dxDLe ˆ n  2xp  dxDLe
change would certainly be small. G

In order to describe the relation between the concentration where


change during reaction and the volume change, we use the
cosH
following equation u
 ˆ n  2xp ˆ 2np  X (13)
cosj
!
d V_ G c dV_ G According to Eq. (11), the relationship between the
ˆ1‡b 1 (10)
d V_ G0
c dV_ 0 G0
velocities at the entrance and exit is
uo 1‡b
which is exactly valid when the gas has ideal behavior. The ˆ c
ui 1‡b o
subscript 0 refers to a reference point at the beginning of or ci

during the reaction. b is a dimensionless constant, which can From the symmetry assumption, it follows
be calculated from the stoichiometry. From Eq. (10) we obtain 
1
u
 ˆ ui ‡ uo
dV_ G 1‡b 2
ˆ c
dV_ G0 1‡b From this we obtain
c0
co
or, introducing the corresponding flow velocities, since the 2‡2b
ci
cross section is constant ui ˆ u
 co (14)
2‡b‡b
ci
u 1‡b and
ˆ c (11)
u0 1‡b
c0 2‡2b
uo ˆ u
 c (15)
Introducing this into Eq. (9) and, substituting c0 = ci and 2‡b‡b o
ci
u0 = ui, we obtain
0 1
Thus, Eq. (12) becomes
B c C r …c † c
…1 ‡ b†d@ cA ˆ xdw   2‡b‡b o
1‡b ui e …1‡b† c r …c † ci
ci  2 d ˆ co dw
c c i ci 2pne 2‡2b
…1‡b† 1‡b ci
or  2   (12) ci
c c r …c †
1‡b d ˆ xdw This first-order differential equation can be integrated
ci ci ci ui e
numerically, if the function r(c) is known. Integrating from
This equation describes the change of concentration c
w =0 to w = 2 j results in
during the movement of the gaseous phase through the
co
granular bed along the radius x. However, since ui is unknown, 2‡2b co  
ci R 1 c 2j
some additional information is necessary. p…1 ‡ b† co  2 d ˆ (16)
The volume change results in a flow rate that is superimposed 2‡b‡b cˆci c ci ci 2ne
ci 1‡b r…c†
on the flow rate due to the rotation. This superimposed flow ci

Chem. Eng. Technol. 25 (2002) 6, Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2002 0930-7516/02/0606-0653 $ 17.50+.50/0 653
Full Paper

Further, from Eq. (7) we obtain the effectiveness factor by After inserting Eq. (21) this equation can be directly
inserting Eqs. (14),(15), and (13): integrated numerically. Fig. 3 shows the result for a first-order
reaction.
r 4ci npe cos2 H
gˆ ˆ  1
r1 r ci …H sinHcosH†
0,9
Θ = 20°
0 co 1 0,8
ZH 1 η 0,7 Θ = 60°
B ci
C sinj (17) Θ = 80°
@ co A cos3 j dj 0,6
2‡b‡b
jˆ0 ci 0,5
0,4
0,3
4 Reactions of Order m 0,2
0,1
The function r(c) can have quite different forms depending 0
on the kind of the chemical reaction and a possible influence of 0 2 4 6 8 10
Da
mass transfer from the gas in the voids to the particle surface.
Figure 3. Effectiveness factor according to Eq. (20) for a first-order reaction,
As an example, we consider a reaction of order m. constant volume (b = 0) and two different values of H.

r(c) = ±kcm (18) The diagram shows that in the range considered (0 < Da < 9),
the values for the effectiveness factor lie rather closely together,
Eq. (16) then transforms to i.e., they follow approximately an equation of the form
co
2‡2b co =ci g = f(Da, m, b)
2j ci R dz
Da ˆ …1 ‡ b † co 2 (19) We find the same result for other reaction orders. For
H 2‡b‡b zˆ1 zm …1‡bz†
ci
sufficiently large values of Da (Da > 5), practically all of the
where z is a variable of integration and entering educt is converted. Then co(j) ® 0 and from Eq. (20)
follows
kcm 1
H 2
Da = i 1 Hsin H 2
e 2pn gDa!1 ˆ
2Da …H sinHcosH† 2‡b
is defined as the Damköhler number for the reaction, which
In this case Eq. (8) simplifies to
uses a representative residence time
2 2
n_ i n_ o ˆ n_ i ˆ ci nepX 2 DLsin H  (22)
H 2‡b
2pn ni now is the total converted amount. Eq. (22) does not contain
Similarly, we introduce Eq. (18) into Eq. (17): any kinetic constant since it simply expresses the amount of
0 c
1 educt drawn into the particle bed. However, it is only valid for
H 1 o Da > 5.
1 2Hcos H 2 R B ci C sinj
gˆ @ c A dj (20) For the general situation Figs. 4 and 5 demonstrate the
Da H sinHcosH jˆ0 2‡b‡b o cos3 j
ci dependency of g from Da and the influence of the reaction
order and the volume change respectively.
With constant volume, i.e. with b = 0 , Eq. (19) results in
1
  1=…m 1†
co j 0,9
ˆ 1 ‡ 2…m 1†  Da (21)
ci H 0,8
m = 0,25
0,7
For m < 1, this relation is valid as long as η m = 1,0
0,6 m = 2,0
j 1 0,5
Da <
H 2…1 m† 0,4

j 0,3
For larger values of Da we have co ˆ 0.
H 0,2
0,1
Eq. (20) for b = 0 simplifies to
0

2 ZH   0 2 4 6 8 10
1 H  cos H co sinj Da
gˆ 1 dj
Da H sinHcosH ci cos3 j Figure 4. Effectiveness factor according to Eq. (20) for different reaction orders,
jˆ0 constant volume (b = 0) and H = 60.

654 Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2002 0930-7516/02/0606-0654 $ 17.50+.50/0 Chem. Eng. Technol. 25 (2002) 6
Full Paper

1 n number of revolutions
0,9 n molar flow rate of the educt in the particle bed
0,8 r reaction rate related to the total volume of solid plus
η 0,7 b=0 voids
b = - 0,5
0,6 b=1
r average reaction rate in the particle bed
0,5 u flow velocity
0,4 VG volumetric flow rate of the gas in the particle bed
0,3 x radial variable, Fig. 2
0,2 X radius of the tube
0,1
z variable for integration
0
0 2 4 6 8 10
Da Greek symbols
Figure 5. Effectiveness factor according to Eq. (20) in the case of a volume
change during reaction for a first order reaction and H = 60. g effectiveness factor, Eq. (6)
e void fraction in the bed
H half opening angle of the particle bed, Fig. 2
j half opening angle up to the radius x, Fig. 2
5 Conclusion w angular variable, Fig. 2
DL length of tube section considered
In rotary kilns, where the granular motion follows the
cascade mode, the particle surface within the bed can play a
substantial role in gas-solid reactions. For reactions of order Indices
m, the effectiveness factor for the inner surface depends
mainly on the Damköhler number i at the entrance of the particle bed
kcm 1
H o at the outlet of the particle bed
Da = i
" 2pn
For Da < 2, the effectiveness factor is greater than 25 % to References
40 %, depending on the value of the other parameters. It
approaches 100 %, as the Damköhler number approaches 0. [1] Henein, H.; Brimacombe, J. K.; Watkinson, A. P., Experimental Study of
Transverse Bed Motion in Rotary Kilns, Metall. Trans. B, 14 B (1983)
Apart from its effect within the Damköhler number, there is
No. 3, pp. 191±220.
only a small additional influence of H. The effectiveness factor [2] Mellmann, J.; Specht, E., Mathematische Modellierung des Übergangs-
decreases if the volume of the gas increases during reaction, verhaltens zwischen den Formen der transversalen Schüttgutbewegung
in Drehrohren, Teil 1, ZKG International, Edition B 54 (2001) 6,
and vice versa. pp. 281±296.
The conclusion is that the reaction at the particle surface [3] Elgeti, K.; Wohlfarth, A., Materialtransport in Drehrohröfen mit
within the bed becomes quite important when the Damköhler taschenförmigen Einbauten, Aufbereitungstechnik 9 (1969) pp. 477±485.
[4] Woodle, G. R.; Munro, J. M., Particle Motion and Mixing in a Rotary
number is smaller than 2. It should be noted that the Kiln, Powder Technol. 76 (1993) pp. 241±245.
Damköhler number used in this paper is inversely propor- [5] Blumberg, W., Selektive Konvektions- und Kontakttrocknung im
Drehrohr, VDI-Verlag, Düsseldorf 1995.
tional to the void fraction. Thus, for very small void fractions
[6] Perron, J.; Bui, R. T., Rotary Cylinders: Transverse Bed Motion
we have large Damköhler numbers and, consequently, only a Prediction by Rheological Analysis, The Can. J. of Chem. Eng. 70
small contribution to the overall reaction from the reaction (1992) pp. 223±231.
[7] Ferron, J. R.; Singh, D. K., Rotary Kiln Transport Process, AIChE J. 37
within the particle bed. For normal void fractions, however, (1991) No. 2, pp. 747±758.
this inner reaction may play an important role. [8] Geisler, O.; Wilbrand, K., Energetische Optimierung von Drehrohr-
Verbrennungsanlagen durch Simulation, VDI-Ber. Band 1385, VDI-
Received: September 17, 2001 [CET 1487] Verlag, Düsseldorf 1998, pp. 383±397.
[9] Wilbrand, K., Thermische Bodenbehandlung in Drehrohr-Verbrennung-
sanlagen ± Simulation und Messung, Dissertation, TU Hamburg-
Harburg 1996.
[10] Baker, C. G. J., Air-Solids Drag in Cascading Rotary Dryers, Drying
Symbols used Technol. 10 (1992) pp. 365±393.
[11] Mellmann, J., Zonales Feststoffmassestrommodell für flammenbeheizte
Drehrohrreaktoren, Dissertation, Technische Universität Magdeburg
b constant referring to volume change, Eq. (10) 1989.
c molar concentration of the educt in the gaseous phase [12] Blumberg, W.; Schlünder, E.-U., Transversale Schüttgutbewegung und
kc
m 1 konvektiver Stoffübergang in Drehrohren, Teil 1: Ohne Hubschaufeln,
H
Da = i
e 2pn, Damköhler number Chem. Eng. Process. 35 (1966) pp. 395±401.
[13] Spurling, R. J.; Davidson, J. F.; Scott, D. M., The Transient Response of
k reaction constant for a reaction order m Granular Flows in an Inclined Rotating Cylinder, Trans IChemE 79
m reaction order (2001) Part A, pp. 51±9±61.

Chem. Eng. Technol. 25 (2002) 6, Ó WILEY-VCH Verlag GmbH, D-69469 Weinheim, 2002 0930-7516/02/0606-0655 $ 17.50+.50/0 655

You might also like