Mmkkhlu
Mmkkhlu
Mmkkhlu
1. INTRODUCTION
Mathematical models have been developed in the past decades to describe calcination
processes in rotary kilns (Gordon, 1969; Sood et al., 1972; Li & Friday, 1974; Brooks, 1989;
Perron et al., 1992; Davis, 1996). The models were usually built upon the basic principles of
conservation of heat and mass, and took into account chemical reactions. No models were
found in the literature for the calcination of limestone in rotary kilns to produce lime. The
objective of this work was to modify the model proposed by Perron et al. (1992) for
petroleum coke calcination in order to describe calcination of limestone in rotary kilns.
Modifications were introduced to the sub-models to account for variations in the model
parameters and thermophysical properties as a function of the state variables. Also, some of
the sub-models were replaced by more suitable ones.
2. MATHEMATICAL MODELLING
As
x x
Control volume for Control volume for
the gas phase the solid phase
Figure 1 – Control volumes for the gas and the solid phases.
The differential control volumes (Fig.1) are defined as dVg = Ag dx for the gas phase and
dVs = As dx for the solid phase. Each dependent variable is represented by an average in the x
direction, for each control volume.
Assuming steady state conditions, the mass conservation equations can be written as
d
(u s .As .ρi ) = mi + R i
dx
(1)
d
dx
( )
u g .A g .ρi = m i + R i
where us and ug are the velocities of the solid and the gas phase (m⋅s-1), respectively, As and
Ag are the cross-sectional areas of the solid and the gas phases (m2), respectively, ρi is the
density of species i (kg⋅m-3), mi is the mass flow rate of species i per unit length of kiln
(kg⋅m-1⋅s-1) and Ri is the rate of consumption of species i (kg⋅m-1⋅s-1). The energy equations
were developed assuming negligible viscous dissipation and that the variations in kinetic and
potential energy are not significant (Perron et al., 1992):
m n o
d Ts
G s Cp s = ∑ Q i + ∑ H j m j + ∑ ∆H k R k
dx i =1 j=1 k =1
(2)
d Tg m n o
G g Cp g = ∑ Q i + ∑ H j m j + ∑ ∆H k R k
dx i =1 j=1 k =1
where Gg and Gs are the mass flowrates of the gas and the solid phases (kg⋅s-1), respectively,
Cpg and Cps are the heat capacities of the gas and the solid phases (kJ⋅kg-1⋅K-1), respectively,
Tg and Ts are the temperatures of the gas and the solid phases (K), respectively, Qi is heat
flowrate per unit length by the radiation, the conduction and the convection (kJ⋅m-1⋅s-1), Hi is
the sensible heat due to the mass transfer of species j, between the phases (kJ⋅kg-1), ∆Hk is the
heat of the reaction k (kJ⋅kg-1) and Rk is the the rate of consumption per unit of length of
species i due to the reaction k (kJ⋅m-1⋅s-1).
The sub-models related to the terms in Eqs. 1 and 2 are described as follows.
Modelling solids flow in a rotary kiln is quite complex, since physical, chemical and
rheological properties vary along the bed. The major factors affecting bed motion in the axial
direction are bed height variation, particles fluidization due to flow of volatile components
released from the bed, and properties dependence on temperature. There is an extensive
amount of work on solids flow in rotary kilns and the model presented here for bed axial
velocity (us) is based on the studies developed by Perron & Bui (1990):
νr 2
us = K
(
h 2 cos(φ) + 2 h 2h r − h 2 sin( φ)) (3)
where K is a characteristic constant, ν is the kiln inclination angle (rad), r is the kiln internal
radius (m), φ is the dynamic angle of repose (rad) and h is the bed height (m).
Heat fluxes are calculated using empirical correlations, specific for rotary kilns. Heat
transfer by radiation between the gas phase and the bed and between the gas phase and the
kiln internal wall can be evaluated by the equations developed by Hottel & Sarofim (1967),
valid for ε ≥ 0.8:
ε g Tg4 − α gTk4
Q rgk = σA rgk (ε k + 1) k = s, w (4)
2
where Qrgk is the radiative heat flux per unit length of kiln, between the gas and the solid
phases, and the gas and the internal wall (kJ⋅m-1⋅s-1), σ is Stefan-Boltzmann constant
(5,67×10-8 J⋅m-2⋅K-4⋅s-1), Argk is the area of heat transfer per unit length of kiln (m), ε and
α are the emissivity and absorptivity.
Heat fluxes by convection between the gas phase and the bed and between the gas phase
and the kiln internal wall are evaluated as
(
Qcgk = h cgk A cgk Tg − Tk ) k = s, w (5)
where Qcgk is the convective heat flux per unit length of kiln, between the gas and the solid
phases, and the gas phase and the internal wall (kJ⋅m-1⋅s-1), Acgk is the area of heat transfer per
unit length of kiln (m), and hcgk are the heat transfer coefficients given by (Tscheng &
Watkinson, 1979)
kg kg
h cgs = 0.46 Re 0D.535 Re 0ω.104 η−0.341 and h cgw = 1.54 Re 0D.575 Re −ω0.292 (6)
De De
where De is the hydraulic diameter (m), ReD and Reω are the axial and the angular Reynolds
numbers, kg is the thermal conductivity of the gas phase (kJ⋅m-1⋅K-1⋅s-1) and η is the degree of
fill of the kiln. The hydraulic diameter De, ReD and Reω are evaluated as follows
Ag ρg u g De ρ g ωD 2e
De = 4 , Re D = , Re ω = (7)
A cgs + A cws µg µg
Heat flux by radiation between the coke bed and the kiln internal wall is evaluated
according to the relationship given by Ketlakh & Tsibin (1978):
(
Q rws = σ A rws ε w εs (1 − εg ) Ω Tw4 − Ts4 ) (8)
where Qrws is the radiative heat flux per unit length of kiln, between the solid phase and the
internal wall (kJ⋅m-1⋅s-1), Arws is the area of heat transfer per unit length of kiln (m), and Ω is
form factor based on the internal wall and the bed.
Heat transfer by conduction between the coke bed and the kiln internal wall is given by
where Qcws is the conductive heat flux per unit length of kiln, between the solid phase and the
internal wall (kJ⋅m-1⋅s-1), Acws is the area of heat transfer per unit length of kiln (m), with the
heat transfer coeficient (hcws) evaluated using the empirical correlation developed by Tscheng
& Watkinson (1979):
0.3
k ω R i2 Γ
h cws = 11.6 b (10)
A cws α b
where kb is the thermal conductivity of the bed (kJ⋅m-1⋅K-1⋅s-1), Acws is the heat transfer area
per unit length of kiln (m), ω is the kiln rotational speed (rad⋅s-1), Γ is the angle of fill of kiln
(rad).The heat consumption rate during limestone calcination can be evaluated as Qr = R∆H.
The temperature at the kiln internal wall is evaluated by solving the equations that
describe heat transfer through the wall. Heat fluxes due to convection, conduction and
radiation penetrate the internal wall comprised of refractory bricks. The heat is transmitted in
the radial direction, through the wall’s refractory and steel layers. The heat is then released at
the kiln external wall. A schematic representation of the heat fluxes is presented in Fig. 2.
Ri er ea
Gas Q rgw
Phase Qcgw
Qrws
Tg
Qw →i
Limestone Qcws
Particles Ts Qi → e
Refractory Tw Qce0
Qre0
Steel Te
Ti T0
The heat flux by conduction in the angular direction (regenerative flux) was considered
negligible (Barr et al., 1989). An energy balance in radial direction leads to the following
equations:
Qi → e = Q ce 0 + Q re 0 (13)
Heat fluxes due to conduction are evaluated as (Pitts & Sissom, 1991)
2π
Q w →i = (Tw − Ti ) (14)
k r ln[(R i + e r ) R i ]
2π
Q i→e = (Ti − Te ) (15)
k a ln[(R i + e r + e a ) (R i + e r )]
The heat losses due to radiation and convection at the kiln external wall are given by
(Özisik, 1985):
(
Q re 0 = σA e εe Te4 − T04 ) (16)
Heat tranfer by convection between the kiln external wall and the surroundings can be
forced or natural. In the case of forced convection (ReD2 > GrD ), the heat transfer coefficient
is evaluated as (Özisik, 1985)
b
0.62 Re1 2 Pr 1 3 Re
a
D
Nu m = h ce 0 = 0.3 + 14
1+
k0 0,4 2 3 28200
1 +
Pr
where
ρDu ∞ gD 3 ρ − ρ D µCp
Re D = , GrD = , Pr = (19)
µ ν2 ρ k0
where Num is the average Nusselt number, GrD is Grashoff number for the gas phase.
For natural convection (ReD2 < GrD), the heat transfer coefficient is evaluated as (Özisik,
1985):
where the subscripts n and n-1 correspond to the present and previous iterations, respectively.
This results in a linearization procedure, since temperature values at the previous iterations
are known.
The mathematical models available in the literature for the limestone calcination rate are
based on different types of assumptions and present different levels of complexity. In all the
relevant works, the kinetics models were developed for a single limestone particle and took
into account its interaction with the gas phase. In the present work, the model presented by
Hills (1968) was adopted due to the fact that it is applicable to particle diameters up to 23mm.
It also explicitly takes into account the effects of the system equilibrium pressure, the CO2
partial pressure and gas velocity over the particles on the thermal decomposition rate. The
particles thermal decomposition is treated as a moving boundary problem. The equation for
the decomposition of a limestone composed of 98% of calcium carbonate is
where ∆Hc is the reaction heat (4×104 J/mol). The equations for the thermal decomposition
specific rate of limestone used in this work are exactly as presented by Hills (1968).
According to Watkinson & Brimacombe (1982), the limestone bed can be described as a
rolling bed. The rolling bed is comprised of a rolling and a stationary layer. In the rolling
layer, the particles are mixed to the gas phase, whereas in the stationary layer they are not.
Martins & Oliveira (1998) presented a set of equations to evaluate the petroleum coke bed
calcination rate as a function of the specific combustion rate of the particles and their
residence time in each layer of the rolling bed. These equations can be used to evaluate the
thermal decomposition rate of limestone if the specific combustion rate is replaced by the
specific decomposition rate (Rsc).
3. IMPLEMENTATION
The model for the limestone calcination kiln is comprised of 4 ordinary differential
equations, corresponding to the mass and energy conservation equations 1 and 2, and a large
set of nonlinear algebraic equations. The dependent variables are calcium carbonate mass
fraction in the solid bed, carbon dioxide mass fraction in the gas phase, and solid and gas
temperatures. The system of equations must be solved iteratively. A fourth-order Runge-Kutta
method was employed for the solution of the differential equations. The rotary kiln for
limestone calcination modeled and simulated in the present work is equivalent to that
described by Henein et al. (1983). The kiln design data are presented in Table 1. The kiln
operational variables used in the simulation runs (Watkinson & Brimacombe, 1982) are
presented in Table 2. The feed composition was considered to be 98.6% calcium carbonate.
Table 1. Design data for the limestone calcination kiln
For the numeric solution of the heat and mass transfer equations 100 integration points
were used, distributed evenly in the axial length of the kiln. The effect of the number of
integration points (np) in the convergence of the solution was made in a previous work
(Martins, 1999). The discreet values of 50, 100, 200, 450 and 900 integration points were
tested and the convergence of the solution was verified except for np=50. Through that
analysis it was verified that np=100 is appropriate for this work. The simulation results for the
limestone bed, gas phase and kiln internal wall temperature profiles are presented in Figs. 3
and 4 respectively, where they are compared to the experimental measurements obtained by
Watkinson & Brimacombe (1982).
The temperature data were measured by thermocouples inserted throughout the kiln. In
Figure 3a, it is clearly seen that, from the limestone feed position (0m) up to 3.5m in the kiln,
the increase in temperature is solely due to heating of the bed. At approximately 3.5m,
calcination of the bed begins for a limestone feed rate of 47.6 kg/h. Temperature remains
nearly constant due to the endothermicity of the calcination reaction. Notice that for a feed
rate of 61.7 kg/h the onset of calcination is delayed due to the fact that a larger amount of
mass needs more time to heat up to reaction temperatures. The rapid increase in temperature
at the end of the kiln indicates that most of the bed have already been calcined before that
portion of the kiln. The predicted temperature profiles for the bed presented good agreement
with experimental data by Watkinson & Brimacombe (1982).
1600 1800
1400 1600
1200
1200
1000
1000
800
800
600
600
400 This w ork (47.6 kg/h) This w ork (47.6 kg/h)
This w ork (61.7 kg/h) 400 This w ork (61.7 kg/h)
200 Watkinson & Brimacombe (1982) (47.6 kg/h) 200 Watkinson & Brimacombe (1982) (47.6 kg/h)
Watkinson & Brimacombe (1982) (61.7 kg/h) Watkinson & Brimacombe (1982) (61.7 kg/h)
0 0
0 1 2 3 4 5 0 1 2 3 4 5
Axial position (m) Axial position (m)
(a) (b)
Figure 3 – Temperature profile (a) limestone bed and (b) gas phase.
1400
1200
Internal wall temperature (K)
1000
800
600
This work (47.6 kg/h)
400
This work (61.7 kg/h)
200 Watkinson & Brimacombe (1982) (47.6 kg/h)
Watkinson & Brimacombe (1982) (61.7 kg/h)
0
0 1 2 3 4 5
Axial position (m)
The predicted temperature for the gas phase (Fig. 3b) decreases in a nearly linear fashion
from the gas feed end (5.5m) to its exit end of the kiln (0m) for both limestone feed rates used
in the simulation runs. The predicted results also agreed well with the experimental data. The
predicted temperature profiles for the kiln internal wall (Fig 4) presented similar behavior to
the measured data. In the experimental work by Watkinson & Brimacombe (1982) the
thermocouples were embedded at 12.7mm within the refractory bricks, and the predicted
values were calculated in this same position. Once the temperatures of the internal wall and
the refractory-stell junction were already calculated, the temperature at 12.7mm within the
refractory shell was calculated by eq. 14.
Watkinson & Brimacombe (1982) took samples from the kiln during calcination, at
positions between 2 and 5.5m, for the experiments with limestone feed rate of 47.6 kg/h.
These samples were analyzed and for each of the samples the percentage conversion of
CaCO3 to CaO (percentage calcination) was calculated. Figure 5 presents the curve for the
calculated percentage calcination as a function of the bed temperature utilizing the proposed
model.
100
60 Watkinson &
Brimacombe (1982)
40
20
0
1000 1100 1200 1300 1400
Bed temperature (K)
The data measured by Watkinson & Brimacombe (1982) are also shown. The predicted
values were in good agreement with the measured ones, with a maximum percentage
difference smaller than 10%. The average percentage calcination measured by Watkinson &
Brimacombe (1982) at the discharge end of the kiln was 97.6%. In the present work, the
predicted value was 99.02% (relative difference of –1.45%). Watkinson & Brimacombe
(1982) observed that the bed height varied in the axial direction from 47mm at the limestone
charge end to 92mm at the discharge end of the kiln. This variation in bed height corresponds
to a variation in the local degree of fill of 6.5 to 17%, respectively, and it causes the bed axial
velocity to decrease from the charge to the discharge end of the kiln. The predicted axial
velocity profile for the bed is presented in Fig. 6.
4,5
Axial velocity of the bed (m/h)
4,0
3,5
3,0
2,5
2,0
1,5
1,0 Axial velocity (G = 47.6 kg/h)
0,5 Axial velocity (G = 61.7 kg/h)
0,0
0 1 2 3 4 5
Axial position (m)
These velocity profiles were used in the calculation of the particles residence time in the
kiln. The calculated value was 2.77h for the limestone feed rate of 47.6 kg/h. This predicted
value is within the range measured by Watkinson & Brimacombe (1982): 3.0h for a feed rate
of 48.6 kg/h, and 2.2h for a feed rate of 46.7 kg/h.
5. CONCLUSIONS
REFERENCES
Barr, P.V., Brimacombe, J.K. & Watkinson, A.P. ,1989, A Heat Transfer Model for the
Rotary Kiln: Part I. Pilot Kiln Trials. Metall. Trans. B, vol. 20B, pp. 391-402.
Brooks, D.G. ,1989, Mathematical Simulation of a Rotatory Coke Calciner. Light Met. Proc.
Tech. Sess. AIME 118th Annual Meeting, pp. 461-469.
Davis, R.A. ,1996, Mathematical Model of Magnetite Oxidation in a Rotary Kiln Furnace,
Can. J. Chem. Eng., vol. 34, pp. 1004-1009.
Gordon, R.A. ,1969, Steady-State Model of a Carbon Kiln Developed for Alcan, Canadian
General Eletric Report, Arvida, Canada.
Henein, H., Brimacombe, J.K. and Watkinson, A.P. ,1983, Experimental Study of Transverse
Bed Motion in Rotary Kilns, Metall. Trans. B, vol. 14B, 191-206.
Hills, A.W.D. ,1968, The Mechanism of the Thermal Decomposition of Calcium Carbonate,
Chem. Eng. Sci., vol. 23, pp. 297-320.
Hottel, H.C. & Sarofim, A.F. ,1967,. Radiative Transfer ,1st Edition, McGraw-Hill Inc., New
York, USA.
Ketlakh, G.A. & Tsibin, I.P. ,1978, Heat Transfer by radiation in Rotary Kiln, translated from
Ogneupory, vol. 1, pp. 17-19.
Li, K.W. & Friday, J.R. ,1974, Simulation of Coke Calciners, Carbon, vol. 12, pp. 225-231.
Martins, M.A. and Oliveira, L.S. ,1998, Modelagem e Simulação de Fornos Rotativos para
Calcinação de Coque de Petróleo, Proc. XVI Congresso Brasileiro de Sistemas
Particulados, vol. 2, pp. 509-516, Teresópolis, Brazil (Portuguese).
Martins, M.A. ,1999, Modelagem e Simulação de Sistemas de Calcinação, M.Sc. Thesis,
Universidade Federal de Minas Gerais, Belo Horizonte, Brazil. (Portuguese).
Özisik, M.N. ,1985, Heat Transfer ,2nd ed., McGraw-Hill Inc., Montreal, CA.
Perron, J. & Bui, R.T. ,1990, Rotary Cylinder: Solid Transport Prediction by Dimensional and
Rheological Analysis, Can. J. Chem. Eng., vol. 68, pp. 61-68.
Perron, J. Bui, R.T. & Nguyen, H.T. ,1992, Modélisation d’un Four de Calcination du Coke
de Pétrole: I. Le Modèle, Can. J. Chem. Eng., vol. 70, pp. 1108-1119.
Pitts, D.R. & Sissom, L.E. ,1991, 1000 Solved Problems in Heat Transfer - Schaum’s Solved
Problems Series ,1st ed., McGraw-Hill Inc., New York, USA.
Reddy, J.N. & Gartling, D.K. ,1994, The Finite Element Method in Heat Transfer and Fluid
Dynamics ,1st ed., CRC Press Inc., Boca Raton, USA.
Sood, R.R., Clark, R. & Stokes, D.M. ,1972, Computerized Model in Determine and
Operating Parameter for Rotatory Kilns for Petroleum Coke Calcination, Light Met. Proc.
Tech. Sess. AIME 101st Annual Meeting, pp. 151-161.
Tscheng, S.H. & Watkinson, A.P. ,1979, Convective Heat Transfer in a Rotary Kiln, Can. J.
Chem. Eng., vol. 57, pp. 433-443.
Watkinson, A.P. and Brimacombe, J.K. ,1982, Limestone Calcination in a Rotary Kiln,
Metallurgical Transactions B, vol. 13B, pp. 369-378.