Chemical Engineering Science 62 (2007) 2590 2607

www.elsevier.com/locate/ces

Rotary Cement Kiln Simulator (RoCKS): Integrated modeling of

pre-heater, calciner, kiln and clinker cooler
Kaustubh S. Mujumdar a,b , K.V. Ganesh a , Sarita B. Kulkarni a , Vivek V. Ranade a,
a Industrial Flow Modeling Group, National Chemical Laboratory, Pune 411 008, India
b Department of Chemical Engineering, Indian Institute of TechnologyBombay, Powai, Mumbai 400 076, India

Received 10 January 2007; accepted 26 January 2007

Available online 14 February 2007

Abstract
This paper presents an integrated reaction engineering based mathematical model for clinker formation in cement industry. Separate models
for pre-heater, calciner, rotary kiln and cooler were initially developed and coupled together to build an integrated simulator. Appropriate models
for simulating gassolid contact and heat transfer in pre-heaters were developed. Calciner was modeled by considering simultaneous combustion
of coal particles and calcination of raw meal. Complex heat transfer and reactions (solidsolid, gassolid and homogeneous reactions in gas
phase) in rotary kiln were modeled using three sub-models coupled to each other. Solidsolid reactions in the bed region of the kiln were
modeled using pseudo-homogeneous approximation. Melting of solids in the bed and formation of coating within the kiln were accounted.
Clinker cooler was simulated by developing a two-dimensional model to capture cross-ow heat transfer between air and hot clinkers. The
individual models were coupled with each other via mass and energy communication through common boundaries. The coupled model equations
were solved iteratively. The model predictions agree well with the observations and experience from cement industry. The model was used
to gain better understanding of inuence of operating conditions on energy consumption in cement plant. Several ways for reducing energy
consumption were computationally investigated. The integrated model, the developed software RoCKS (for Rotary Cement Kiln Simulator)
and results presented here will be useful for enhancing our understanding and for enhancing the performance of clinker manufacturing.
2007 Elsevier Ltd. All rights reserved.
Keywords: Cement; Energy consumption; Reaction engineering model

1. Introduction
Cement making processes are extremely energy consuming.
Typically for producing one ton of cement, a well-equipped
plant consumes nearly 3 GJ. For each ton of clinker produced,
an equivalent amount of green house gases are emitted. The
manufacture of cement has been the focus of considerable
attention worldwide because of the high energy usage and high
environmental impact of the process. Considering the recent
impetus on reduction in emission of green house gases and reduction in energy consumption, there is a renewed emphasis
on developing computational models for cement industry and
using this understanding for performance enhancement.

Corresponding author. Tel.: +91 20 2590 2170; fax: +91 20 2590 2621.

E-mail address: vv.ranade@ncl.res.in (V.V. Ranade).

0009-2509/$ - see front matter 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2007.01.063

A schematic of typical clinker making process is shown in

Fig. 1. The raw meal consisting of predetermined quantities
of CaCO3, SiO2, Al2O3 and Fe2O3 are passed sequentially
through pre-heater, calciner, kiln and cooler to form cement
clinkers. In a pre-heater section the raw meal is pre-heated to
calcination temperature via hot gases coming from calciner. In
a calciner, raw meal is partially calcined. The energy required
for endothermic calcination reaction is provided by combusting
a suitable fuel. In most cases, coal is used to provide the required energy, especially in India. The calciner is supplied with
tertiary air from the cooler and air coming out of kiln exhaust.
The former is to supply sufcient O2 for coal combustion and
later to utilize the heat of kiln gases to enhance calcination
reaction. The hot gases from calciner are sent to pre-heater assembly for pre-heating the solids. The partially calcined solids
from the calciner are fed slowly to a rotary kiln. In the rotary
kiln, remaining calcination and other clinkerization reactions

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

Calcineous
Raw meal

2591

Exhaust to
atmosphere

Pre-heater
Assembly

Hot gases
to pre-heater
Coal
Tertiary Air
Calciner

Kiln Exhaust

Pre-heated
Raw meal

Secondary Air

Vent Air

Kiln
Coal

Cooler

Air to cooler

Cooled clinker

Fig. 1. Schematic of cement clinker process.

occur (formation of C2 S, C3A, C4AF). The energy required for

endothermic clinker reactions is provided by combusting coal
in the kiln. The pulverized coal along with the pre-heated air
(secondary air) is fed to the kiln in a counter current mode
with respect to solids. Part of the solids melts in the kiln.
The melt formation causes an internal coating on kiln refractories. Counter current ow of gas entrains solid particles in
the free board region. Such entrainment enhances rates of radiative heat transfer by increasing effective emissivity and conductivity. The hot clinkers are discharged from kiln to clinker
cooler and hot gases from kiln exhaust are sent to the calciner. In a clinker cooler, a part of energy of solids is recovered back by heat exchange with air. The pre-heated air from
the coolers is passed to kiln and calciner as secondary and
tertiary air, respectively. A small part of air may be vented if
required.
This brief overview of clinker formation clearly demonstrates
the strong coupling among pre-heater, calciner, kiln and cooler.
It is therefore essential to develop an integrated model for preheater, calciner, kiln and cooler in order to capture key characteristics of clinker manufacturing and to enable the model to
be used as simulation or optimization tool. Such an attempt is
made in this work.
Recently some attempts have been made to develop
computational uid dynamics (CFD) based models to simulate
either calciner (for example, Lu et al., 2004) or kiln (for example, Mastorakos et al., 1999; Mujumdar and Ranade , 2003).
Though such CFD models show promise in simulating details
of combustion and burner designs, it is almost impossible to

build CFD models for simultaneous and coupled simulations

of pre-heaters, calciner, kiln and cooler. The CFD models are
thus not very useful to gain understanding of coupling and
exploring ways to reduce overall energy consumption per ton
of clinker. Some attempts have also been made to develop
reactionengineering models for kiln (for example, Mujumdar
and Ranade, 2006; Spang, 1972). Such models have shown
promising capabilities in capturing the overall behavior and
providing useful clues for reducing energy consumption in
rotary cement kilns. The numerical experiments using the computational model could also predict the inuence of kiln operating parameters on net energy consumption (NEC) in kilns.
Such guidelines can provide useful hints to operating engineers
for kiln optimization. However, none of these models have included coupling of pre-heater, calciner, kiln and clinker cooler.
This work was undertaken to fulll this need. The motivation
of the present work was to develop a framework of reaction
engineering based computational model for clinker formation
in cement industry and use this framework subsequently for
exploring possible performance enhancement. The paper is
organized as follows.
The key issues in modeling individual models are discussed
in Section 2. The computational model and the modeling strategy are thereafter presented in Section 3. Section 4 reports
the results of computational simulations of model with respect
to key operating parameters. The use of the developed model
to explore possible ways of reducing energy consumption in
kiln is discussed in Section 5. Key ndings of the study are
summarized at the end.

2592

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

2. Key issues and modeling approach

Key issues governing the performance of individual units are
schematically shown in Fig. 2. We discuss the issues of preheater, calciner and clinker cooler and review the previous work
related to it to provide background for the models developed in
this work. The key issues governing rotary kiln were discussed
in our recent work (Mujumdar et al., 2006) and therefore, are
discussed here very briey.
2.1. Cyclone pre-heaters
The calcineous raw meal is passed through a set of pre-heater
cyclones (depending on pre-heater assembly) before it is sent
to the calciner. In the pre-heater section, the raw meal is preheated to around calcination temperature by hot gases coming
from the calciner. The operation of cement pre-heater is similar
to that of a conventional cyclone. Solids are fed to the cyclone
along with the gas coming from the previous pre-heater. Both
solid and gas spiral down towards the cyclone bottom where the
gas reverses its direction and leaves through the exit duct while
the solids leave from the bottom to the next unit. The ow inside
the cyclone is characterized by high swirl and turbulent motion.
This provides excellent heat transfer between gas and solids. We
used CFD based model for studying mixing and heat transfer
of gas solid ows in cyclones. Our CFD simulations (carried
out for typical values of operating conditions as gas ow rate =
60.83 kg/s; solid ow rate = 50 kg/s; inlet gas temperature =
740 K; inlet solid temperature=500 K; cyclone diameter =6 m;

height of cyclone = 11 m) indicated that the solid and gas temperatures inside the cyclone were quite uniform (in range of
637644 K) and close to exit temperatures (average temperature of 640 K). Therefore, in this work, each pre-heater cyclone
was treated as completely mixed cell for individual phases and
was, therefore, represented by a pair of temperature (one for
gas phase and the other for solid phase). The heat losses from
cyclone are controlled by the heat ux across the cyclone walls.
Therefore it is essential to predict the heat transfer between
particle laden gas and cyclone walls in the pre-heater cyclone.
The empirical relation proposed by Gupta and Nag (2000) was
used to determine these losses.
2.2. Cyclone calciner
The calcination of raw meal starts in the pre-heater unit itself
once the calcination temperature is reached. In this framework the calcination reaction was assumed to occur in the last
cyclone. The framework is, however, quite general to accommodate relaxation of this assumption. Energy required for the
calcination reaction is given by coal combustion. Shrinking
core reacting model was used for modeling coal combustion.
The char combustion kinetics and gas phase volatile combustion kinetics was used from Hamor et al. (1973) and Li et al.
(2003). The kinetic parameters are given in Table 1. One of the
main issues in modeling cyclone calciner is to estimate the residence time of raw meal in the cyclone. Few empirical correlations have been proposed to predict the average residence time
of particles in cyclones (Kang et al., 1989; Lede et al., 1987).

Pre-heater

Calciner

Gas-solid heat transfer

(gas/solids/walls), particle
laden turbulentflow,
losses

Turbulent gas phase/

coal combustion, calcination,
heat transfer (gas/solid/
walls), losses

Strongly
Coupled

Rotary Kiln
Coal/gas phase comb, bed
height variation, clinker formation,
melting/solidification, coating
formation, losses

Clinker Cooler
Gas-solid heat transfer,
temperature gradients in
clinker bed region

Fig. 2. Key issues in modeling cement clinker process.

Table 1
Kinetic constants for coal combustion reaction
Reactions

K0

E (J/mol)

Reference

C + O2 = CO2
CH4 + 2O2 = CO2 + 2H2 O

9.18 101 (kg/m2 s KPa)

1.6 1010 (m3 /kg s)

6.81 104
1.08 105

Hamor et al. (1973)

Li et al. (2003)

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

However, all the attempts for prediction of residence time in

cyclones were based on lab scale cyclones and none of the
studies were extended or reported for industrial scale cyclones.
This parameter was therefore treated as an adjustable parameter in the model. In the present work, we have adjusted the
residence time so as to get desired degree of calcination as per
industrial observations. It was conrmed from our prior simulations (Warudkar et al., 2005) that varying residence time in the
calciner by 10% had relatively small effect ( 2.5%) on predictions of percentage calcination. It is also essential to obtain
relevant kinetics for calcination reaction in calciner. Thermal
decomposition of limestone calcination is a complex process.
A wide discrepancy is observed in the proposed rates for calcination reaction. In our recent work (Mujumdar and Ranade,
2006) we have compared models proposed by 18 investigators
which showed wide scatter. Watkinson and Brimacombe (1982)
have reported experimental data on calcination of limestone in
experimental kiln. The experimental conditions of their experiments were close to industrial operations (bed temperature
1000.1300 K). Their data was therefore used to nd calcination
kinetics in this work.
2.3. Rotary kiln
The partially calcined raw meal is passed slowly to the rotary
kiln where the clinkerization reactions occur. In the initial part
of the kiln the remaining calcination occurs. Other solidsolid
and solidliquid clinkerization reactions take place as the solid
bed moves towards the burner. Part of the solids melts in the
kiln. The melt formation causes an internal coating on kiln refractories. Counter current ow of gas entrains solid particles
in the freeboard region. Such entrainment enhances rates of radiative heat transfer by increasing effective emissivity and conductivity. In this section we discuss the key issues involved in
modeling the cement kilns very briey. The main key issues for
modeling the rotary cement kilns are estimating the residence
time of solids in the kiln, cinkerization reaction in bed region,
coal combustion in freeboard region, heat transfer between bed
freeboard and walls, melting/coating formation around the kiln
walls. These issues are discussed in detail in our recent work
(Mujumdar et al., 2006) and therefore are not repeated here.

2593

wide range of temperatures (2001000 C). Since the temperatures in clinker cooler are in the same range this correlation
was used to model heat transfer coefcient between solids and
gas. The computational models for individual components and
the coupling strategy are discussed in the following section.
3. Computational models and solution methodology
3.1. Cyclone pre-heater model
A schematic of pre-heater unit considered for developing
computational model is shown in Fig. 3a. The present framework of computational models was developed for dry process
of clinker formation since this process is widely used in cement industry. For the dry processes, the moisture content is
generally present in very small amount (typically 0.5%, see
for example Engin and Ari, 2005; Peray, 1984). The energy
requirements for removing the moisture from the feed being
small (less than 0.5% of the total energy consumption), the feed
was considered to be free of moisture in this work. However,
the developed framework is quite general and including evaporation of moisture from the feed is straightforward. The gas
phase and solids in a cyclone was assumed to be completely
back mixed. In Fig. 3a, Ms is the mass of solids entering the
cyclone. Mg is the mass of the air entering the cyclone. Mse is
the mass of solids entrained from a cyclone. Each cyclone was
assumed to be lined with refractory of thickness tr .
Thus, for any ith cyclone in pre-heater assembly the following inlet streams were considered:
1. Solids from the (i 1)th cyclone (Ms,i1 at temperature
Ti1 ).
2. Solids that are entrained by gas from (i + 1)th cyclone
(Mse,i+1 at temperature Ti+1 ).
3. Air from (i + 1)th cyclone (Mg at temperature Ti+1 ).
The outlet streams for this cyclone are:
1. Solids going out of cyclone (Ms,i at temperature Ti ).
2. Solids that are entrained by gas (Mse,i at temperature Ti ).
3. Air going out (Mg at temperature Ti+1 ).
The steady state material balance equation for ith cyclone is
written as

2.4. Clinker cooler

Ms,i1 + Mse,i+1 = Ms,i + Mse,i ,

(1)

The hot solids from the kiln are discharged on the grate of
clinker cooler. As the grate moves with uniform speed along
the cooler length, solids lose their heat to cross-ow air. A part
of the air is generally sent to the kiln as secondary air, a part to
calciner as tertiary air and a part is vented to the surroundings
(vent air). The most important key issue in modeling grate coolers is predicting the heat transfer coefcient between hot solids
and cross-ow air. There is no information on modeling of heat
transfer in such cases. In absence of any relevant information
we have used heat transfer correlation in packed bed reactors
to estimate the heat transfer. Nsofor and Adebiyi (2001) have
carried experimental measurements and presented correlation
for forced convection gas particle heat transfer coefcient for

Mse,i = (1 m,p ) Ms,i .

(2)

In the above equations m,p represents the particle capture efciency of the ith cyclone. M represents the mass of the solids
(in kg/s) and subscripts s and se represent solids and entrained
solids, respectively, as explained earlier.
The steady state energy balance for the ith cyclone was written as
Ms,i1 Cp,s Tc,i1 + Mse,i+1 Cp,s Tc,i+1
+ Mg Cp,g Tc,i+1
= Ms,i Cp,s Tc,i + Mse,i Cp,s Tc,i
+ Mg Cp,g Tc,i + hcyc Acyi (Tc,i Tiw,i ).

(3a)

2594

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

Gas out

Mg,Mse,iTi
Ms,i-1Ti-1
Raw meal
Tiw,i
i

Tow,i

Coal in

Air in

12

Loses

Radiation and convection

Losses

Mg,Mse,i-1Ti-1

1. Refractory
2. Shell
Partially calcined raw meal

Ms,i,Ti
Secondary air

Tertiary air
T
y

Vent air

=0

T
=0
x

Hot clinker, Ts, in

Cold clinker, Ts, out

T
=0
x
T
=0
y

Cooling Air, Ta
Fig. 3. (a) Schematic of (a) cement pre-heater, (b) cement calciner, and (c) grate cooler.

In the above Cp,s and Cp,g represents the specic heat of solids
and air, respectively. Subscript g represents the air and Tc,i
represents the temperature of solids and air in the ith cyclone.
hcyc represents the heat transfer coefcient for energy exchange
between particle laden gas and cyclone inner walls. hcyc was
evaluated from the following empirical correlation given by
Gupta and Nag (2000) for heat transfer in cyclones:


hcyc dc
P
= 702.818 + 9.0287 1014 u0 Re + 11.1385
kg
u20


P
+ 4.50398 105
Re + Rc ,
u0
where
Rc = Fpw

4 T4
Tiw
g

Tiw Tg

dc
.
kg

walls, which is equal to loss from shell walls due to convection

and radiation. The energy balance for heat transfer in cyclone
cross-section is written as
hcyc Acyi [Tc,i Tiw,i ] =

2  L kr [Tiw,i Tr,i ]
,
ln(rr /ri )

(4)

2  L kr [Tiw,i Tr,i ] 2  L ksh [Tr,i Tow,i ]

=
,
ln(rr /ri )
ln(r0 /rr )
(5)
2  L ksh [Tr,i Tow,i ]
ln(r0 /rr )
4
= hconv Acyo [Tow,i T0 ] +  cy Acyo [Tow,i
T04 ].
(6)

(3b)

The LHS of Eq. (3a) thus represents the total energy entering
the cyclone and RHS represents the energy leaving out of the
cyclone. At steady state the heat given to cyclone walls must
be same as heat conduction in through refractory and cyclone

In the above equations, Tiw,i is the internal wall temperature of

the ith cyclone, Tr,i is the temperature of interface of refractory
and shell, Tow,i is the temperature of external wall of the ith
cyclone and T0 is the ambient temperature. L is the total height
the cyclone, kr is the thermal conductivity of the refractory
and ksh is the thermal conductivity of cyclone walls. r0 is the

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

external diameter, rr is the internal diameter of the shell and

ri is the cyclone internal diameter. Acyo is the external surface
area of the cyclone. hconv is the convective heat transfer coefcient of external wall, cy is the emissivity of the cyclone outer
wall and  is the StefanBoltzmann constant for radiative heat
transfer.
The above material and energy balances were written for
n cyclones. For each cyclone there are six equations and six
unknown variables. The set of nonlinear algebraic equations
comprising 6n equations was solved using NewtonRaphson
method to get mass and temperature of solids, gas and walls of
pre-heater cyclones.
3.2. Cyclone calciner model
The mathematical model of calciner was based on a
schematic shown in Fig. 3b. The assumptions made in the
model are as follows:
1. Gas phase is completely back mixed.
2. The raw meal and coal particles were treated as discrete
phases having uniform particle size. All the particles were
assumed to have the same residence time in calciner.

2595

where yc,c is the mass fraction of char in coal particle. The

energy balance for a coal particle is given by
dmp,c Cp,c Tcl
= hc Ap (Tg Tcl ) +  c Ap (Tg4 Tcl4 )
d
dmp,c
+ fc Hcomb rcomb + Cp,c Tcl
.
d
(10)
In the above equation, Cp,c is the specic heat of coal and Tcl
is the coal temperature. Tg is the gas temperature. Hcomb is
the heat released due to coal combustion and fc is the fraction
of energy released due to coal combustion, which is absorbed
by the particle. The rst and second terms in RHS of above
equations are the energy absorbed by the coal particle due to
convection and radiation from the gas phase during the heat
up. The heat transfer coefcient, hc , is evaluated using the
correlation of Ranz and Marshall as
hc dp
= 2 + 0.6(Re)0.5 (Pr)0.33 .
kg

(11)

The nal term in Eq. (10) represents the loss sensible heat from
coal particle due to loss of mass.
3.2.1.2. Raw meal particle The calcination reaction is given as

The mass and energy balance equations are presented below.

3.2.1. Discrete phase
3.2.1.1. Coal particles Mass balance for coal particles in calciner is written as
dmp,c
= A0 fv,0 mpc,0 ks,c eE1 /RT cl Ap pO2 . (7)
d
In the above equation, A0 is the devolatilization constant, fv,0
is the initial mass fraction of volatiles in coal particle and the
mpc,0 is the initial mass of coal particle, ks,c is the rate constant
of char, Ap is the available surface area of particle (which was
calculated as function of conversion i.e., mass of char reacted),
E1 is the energy of activation and pO2 is the partial pressure of
O2 .  is time spent by a coal particle in calciner. The rst term
in the RHS of above equation represents the loss in discrete
phase mass due to devolatilization of coal particles and the
second term in the RHS of above equation represents the loss
in discrete phase due to char reaction.
The individual species balance for rate of mass change due
to devolatilization is given by


dmp,c
1
dyv,c
A0 fv,0 mpc,0 + yv,c
=
,
(8)
d
mp,c
d
where yv,c is the mass fraction of volatiles in the coal particle.
The individual species balance for rate of mass change due
to char combustion is given by


dmp,c
1
dyc,c
=
ks,c eE1/RT cl Ap pO2 + yc,c
,
d
mp,c
d
(9)

CaCO3 CaO + CO2 .

(12)

The over all mass balance for the raw meal particle is given by
dmp,L
Mw CO2
,
= rc
Mw CaCO3
d

(13)

where
rc =r ks eE2 /RT L 4  rp2 Mw CaCO3 ,

(14)

where the mp,L , MwCO2 , MwCaCO3 are mass of raw meal

particle, molecular weights of carbon-dioxide and calcium carbonate, respectively. rc is the rate of calcination of calcium carbonate and rp radius of shrinking raw meal particle. The term
r k is the rate constant of calcination of calcium carbonate of
s
calcination. It has been established that presence of CO2 in
gas phase inhibits calcination rate (Stanmore and Gilot, 2005).
However, there is no consensus on representing inuence of
CO2 partial pressure on calcination reaction. In this work we
use a simple linear form of relationship (Hu and Scaroni, 1996)
to model effect of presence of CO2 on rate of calcination reaction as


rc = r ks eE2 /RT L 4  rp2 Mw CaCO3 ,

(15)

where


ks = r ks , PCO2 < 102 Peq ,



PCO2 Peq
r 
, 102 Peq < PCO2 < Peq
ks = r ks
Peq
r

and Peq = 1.826 107 e(19680/TL ) .

(16)
(17)

2596

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

The individual species balances for limestone and calcium

oxide are given as

1
dyv
=
dt
mg

dmCaCO3 ,L
= rc ,
d

(18)

dmCaO,L
rc Mw CaO
=
.
d
Mw CaCO3

(19)

+ [mp,cin yv mp,cout yv ] c Np


rcombg
Vreact Mw vol Zvol

Mw vol

dmg
,
(24)
yvol
dt

dyw
1
mgin ywin mgout ywout
=
mg
dt



rcombg
dmg
Vreact Mw w ZH2 O yw
,
+
Mwvol
dt

The energy balance for the raw meal particle is given as

d(mp,L Cp,s TL )
= hc,L L Ap (Tg TL )
d
+  L L Ap (Tg4 TL4 )
dmp,L
+ Cp,s TL
,
d

mg,in yv,in mg,out yv,out

(20)

(25)

where TL is the temperature of raw meal particle, L is the

emissivity of solid particle, L Ap is the area of the raw meal
particle (which was calculated based on conversion i.e., mass of
raw meal particle reacted) and  is residence time of raw meal
particle in the calciner. hc,L was estimated by using Eq. (11).

where yO2 , yCO2 , yv , yw are the respective mass fractions of

oxygen, carbon-dioxide, volatile matters and water. Mw O2 ,
Mw CO2 , Mw vol and Mw w are their respective molecular
weights. Vreact is the volume of reactor. Subscripts in and out
represent the inlet and outlet conditions and Z is the stoichiometric coefcient.
The energy balance equation for the gas phase is given as

3.2.2. Continuous phase

The over all gas mass balance is given as

dmg Cp,g Tg
= mgin Cp,g Tg,in mgout Cp,g Tg
dt
+ Sgcomb + Sccomb + Scalc

dmg
= mgin mgout + [mp,cin mp,cout ] c Np
dt
+ [mp,Lin mp,Lout ]L Np ,

(21)

where mg is the mass of the air in the calciner, c Np is the

number of particles of coal coming in per unit time and L Np
is the number of particles of raw meal coming in per unit time.
The individual species mass balance for rate of change of mass
of oxygen, carbon-dioxide, volatile matters and water can be
written as

dyO2
1
mg,in yO2 in mg,out yO2 out
=
dt
mg

[mp,c yc,cin mp,c yc,cout ] c Np Mw O2



rcombg
Mwvol

Mwchar
Vreact Mw O2 ZO2 yO2

dmg
dt

Sgcomb = rcombg Hcombg Vreact ,

(26a)
 
hc Ap (Tg Tcl )+ c Ap (Tg4 Tcl4 )
Sccomb =
0

+ Cp,c Tcl

dmp,c
d

(1 fc ) Hcomb rcomb


d,

(26b)

 

hc,L L Ap (Tg TL )+ L L Ap (Tg4 TL4 )


dmp,L
+ Cp,s TL
(26c)
+ rc Hcalc d.
d
0


mgin yCO2 in mgout yCO2 out
[mp,c yc,cin mp,c yc,cout ] c Np Mw CO2
Mw char

+ [mp,Lin mp,Lout ]L Np


rcombg
Vreact Mw CO2 ZCO2
+
Mw vol

dmg
,
yCO2
dt

(26)

where

Scalc =

(22)
dyCO2
1
=
dt
mg

hcyc Acyi (Tg Tiw ),

(23)

In the above equations, Sgcomb, Sccomb are the heat source term
for gas-phase from volatile combustion and char combustion,
respectively. Scalc is the heat sink term from calcination. Hcombg ,
Hcalc are the enthalpies of volatile combustion and calcination.
hc,L is the convective heat transfer coefcient between raw meal
particles and air. The steady state equations across the cyclone
walls were written same as that of pre-heaters explained in
the previous section to obtain temperature of calciner internal
walls, refractory and outer walls.
The calciner model equations were solved using an iterative
method. The model equations for gas phase were solved assuming steady state. For the rst iteration, source terms from
discrete phase were assumed to be zero. The temperature and

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

mass of species obtained by solving continuous phase were used

in discrete phase equations to get the new source terms from
the discrete phase. The sources from the discrete phases were
passed to continuous phase to get the new mass and temperature
terms for discrete phase. This procedure was continued till the
subsequent changes in temperature of gas phase were within
0.1%. Suitable under-relaxation parameters were dened to
accelerate convergence. Typically about 20 iterations were required to achieve convergence. The differential equations for
discrete phase were solved by modied Gears method implemented in ODEPACK (Hindmarsh, 1983). The algebraic equations for continuous phase were solved using NewtonRaphson
method.
3.3. Kiln model
A comprehensive one-dimensional model was developed to
simulate complex processes occurring in rotary cement kilns.
A modeling strategy comprising three sub-models viz. model
for simulating variation of bed height in the kiln, model for
simulating clinkerization reactions and heat transfer in the bed
region and model for simulating coal combustion and heat transfer in the freeboard region was developed. The Kramers model
(Kramers and Croockewit, 1952) which relates volumetric ow
rate of solids, v , with kiln tilt ( , radian), angle of repose (
,
radian), radius of kiln (R, m), rotational speed of kiln (n) and
height of solids (h) was used to model bed height variation in
the kiln. The clinkerization reactions in solid bed were modeled
assuming solids as pseudo uids. Melting of solids in bed region and formation of coating within the kiln were accounted.
Combustion of coal in the freeboard region was modeled by
accounting devolatilization, nite rate gas phase combustion
and char reaction. Knowing the bed and freeboard gas temperatures, the temperatures of kiln inner wall, refractory and shell
were obtained by solving steady state energy balance across
the kiln walls. The details of the models and model equation
are discussed in detail in our recent publication (Mujumdar
et al., 2006) and are not repeated here for the sake of brevity.
3.4. Cooler model
The mathematical model of cooler was based on a schematic
shown in Fig. 3c. Solids of uniform particle size and constant
porosity were assumed to move in a plug ow with constant
grate speed. Air was assumed to enter in a cross-ow mode
with respect to solids in y direction. The amount of air fed to
the cooler was distributed as secondary air (to kiln) from the
front section of cooler, followed by the tertiary air (to calciner)
and nally the vent air (Locher, 2002) as shown in Fig. 3c.
The amount of secondary, tertiary and vent air (in kg/s) going to kiln, calciner and exhaust, respectively, were assumed
to be proportional to the fraction of length of each section in
the cooler. The fractional length of each section was user input to the model. To get the temperature proles of solid bed
and air, the clinker cooler was divided into n segments along
the length of the cooler and m segments along the height of
the cooler. Mass and energy balances were solved for these

2597

segments. Conductive heat transfer was considered for solids in

both horizontal and vertical directions. Convective heat transfer
coefcient between air and solids was calculated from empirical correlation assuming solids as packed bed as discussed previously. The boundary conditions used in the model are shown
in Fig. 3b. The model equations are presented in the following.
Mass balance for solids can be written as
dms(i,j )
= 0.
dx

(27)

Assuming steady state operation, the energy balance equation

can be written as
j(s (1 )us,x Cp,s Ts ) j(s (1 )us,y Cp,s Ts )
+
jx
jy
j{(1 )ks jTs /jx} j{(1 )ks jTs /jy}
=
+
jx
jy
a hc,c (Ts Tg ).

(28)

In this equation s is the cement clinker density, Cp,s is clinker

heat capacity, us is grate speed, and Ts is clinker temperature
of solid at any point, ks is clinker thermal conductivity, a is the
surface area per unit volume, hc,c is convective heat transfer
coefcient between solid clinker and air in the cooler,  is
the porosity, Tg is air temperature at any point in the cooler.
In Eq. (28) the rst and second terms of the right-hand side
represents the conductive heat transfer. The last term in righthand side represents convective heat transfer between the air
and solids.
The mass balance for air can be written as
dma(i,j )
= 0.
dy

(29)

Energy balance for air can be written as

j(g ug,x Cp,g Tg )
jx
=

j(g ug,y Cp,g Tg )
jy

j{kg jTg /jx} j{kg jTg /jy}

+
+ a hc,c (Ts Tg ).
jx
jy
(30)

In this equation g is the density of the air, ug,y is inlet speed

of cooling air, and Tg is air temperature at any point, k is air
thermal conductivity, Ts is solid temperature at any point in the
cooler. In Eq. (30) the left-hand side terms represents the net
energy input by the air. First two terms in the right-hand side
represent the conduction between the air layers and the nal
term is due to the convection between solids and air.
In Eq. (30) hc,c is convective heat transfer coefcient between solid clinkers and air. Developing accurate models for
convective heat transfer coefcient between solids and air is
important in capturing heat transfer in the cooler. In this work
the convective heat transfer coefcient was calculated based on
empirical expression given by Nsofor and Adebiyi (2001). The
empirical expression is given as
Nu = 8.74 + 9.34 [6(1 )]0.2 Re0.2 Pr0.33
(30 < Nu < 60; 50 < Re < 120).

(31)

2598

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

It is important to note that the Reynolds numbers for commercial clinker cooler are signicantly higher (Re 1000.2000)
as compared to the experimental conditions of Nsofor and
Adebiyi (2001). However, as discussed earlier, there are no
other systematic experimental studies reported to predict convective heat transfer coefcients in clinker coolers. The empirical correlation (Eq. (31)) was developed for particle sizes close
to those found in industrial clinker coolers and for wide range
of temperature conditions as observed in clinker coolers. Fortunately, the empirical correlation seems to be weekly dependent
on Reynolds number (Reynolds number is to power 0.2). Therefore possible errors associated with Eq. (31) are not expected
to change the simulation results signicantly (predicted Nusselt number is 50.60). Hence Eq. (31) was used to predict
gas solid heat transfer in clinker coolers in the present model.
All the physical properties for determining heat transfer coefcient were calculated at an average temperature of solids and
air as Tf = (Ts + Tg )/2. The system of algebraic linear equations formulated for above model equations was solved using
tri-diagonal matrix algorithm (TDMA).

These values are usually known or can be easily available for

any cement plant and can therefore be used to generate good initial guess for faster convergence of solution. The pre-processor
solves mass and energy balance equations as discussed in the
following. Based on the percentage calcination in the calciner,
the mass of CO2 produced in calciner was calculated as

3.5. Integrated model and solution strategy

xCaCO3 ,C =

The individual models for pre-heater, calciner, kiln and

cooler described in the previous section were coupled with
each other to develop a simulator for the entire system. The
schematic of the simulator is shown in Fig. 4. The required inputs to the simulator are ow rates and composition of (a) raw
meal entering the pre-heater, (b) air entering the cooler, (c) coal
entering the calciner and the kiln, and (d) the material properties and operating parameters of the individual equipments
(for example, kiln RPM, grate speed of cooler). However, to
solve the integrated simulator, it is necessary to know the inlet conditions for the calciner (ow rate, mass fractions and
temperature of solids and air from pre-heater, kiln and cooler),
pre-heater (ow rate and temperature of air from calciner), kiln
(ow rate, mass fractions and temperature of secondary air
from cooler and partially calcined raw meal from the calciner)
and cooler (ow rate and temperature of solids from kiln).
To generate these inputs a pre-processor was developed. The
function of pre-processor was two-fold. The pre-processor was
used to develop good initial guess for the simulator and also to
check for any inconsistency of input data. The pre-processor
generated the initial guess (for mass ow-rates, composition
and temperatures of raw meal and air) for the individual models based on overall material and energy balances. Following
parameters were provided to the pre-processor to achieve this:
1. Percentage calcination occurring in the calciner (P ).
2. Temperature of secondary air (Tg,S ) and tertiary air (Tg,T )
leaving the cooler.
3. Temperature of air leaving the kiln (Tg,K ).
4. Temperature of air exiting the pre-heater to the atmosphere
(Tg,P ).
5. Temperature of solids exiting the cooler (Ts,R ).
6. Heat losses (HLoss,K ) and heat of clinkerization reaction in
the kiln (HR,K ).

Mw CO2
,
Mw CaCO3

mCO2 ,C = mCaCO3 ,i P

(32)

where mCaCO3 ,i is the total amount of CaCO3 in the inlet raw

meal. The mass ow rate of solids entering the kiln was calculated as
Ms,C = Ms,P mCO2 ,C ,

(33)

where Ms,C is the mass ow rate of raw meal leaving the

calciner or entering the kiln, Ms,P is the mass ow rate of the
solids entering the pre-heater. The corresponding mass fraction
of solids species leaving the calciner or entering kiln were
calculated as

xCaO,C =

mCaCO3 ,i mCaCO3 ,i P
,
Ms,C

(mCO2 ,C ) (Mw CaO )

,
(Ms,C ) (Mw CO2 )

mSiO2 ,i
mAl2 O3 ,i
, xAl2 O3 ,C =
,
Ms,C
Ms,C
mFe2 O3 ,i
=
,
Ms,C

xSiO2 ,C =
xFe2 O3 ,C

(34)

where x is the mass fraction of the component in the raw meal.

The amount of clinker leaving the kiln or entering the cooler
Ms,K was calculated as
Ms,K = Ms,C

(Ms,C ) xCaCO3 ,C
Mw CO2 .
MW CaCO3

(35)

Based on overall material balance on kiln, the amount of air

leaving the kiln was calculated as
Mg,K = Mg,S + Ms,C xCaCO3 ,C
+ Mc,K yc,K

Mw CO2
Mw CaCO3

Mw CO2
,
Mw CaCO3

(36)

where Mg,S is the mass of secondary air entering the kiln, Mg,K
is the air leaving the kiln or entering the calciner, Mc,K is the
amount of coal entering the kiln and yc,K is the mass fraction
of char entering the kiln. The amount of air leaving the preheater assembly was calculated as
Mg,P = Mg,K + Mg,T + mCO2 ,C
+ Mc,C yc,c

Mw CO2
,
Mw CaCO3

(37)

where Mg,P is the mass of air entering the pre-heater, Mg,T

is the mass tertiary air entering the calciner, mCO2 ,C is the

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

User Input
Dimensions, MOC,
Mass Flow Rate,
Mass Fractions,
Temperature

Call Preprocessor
N

2599

Consistence Checks
&
Generate initial guess

Call Sub-models

Update variables

No

Converged
Yes
Post Processing

Fig. 4. Solution methodology of the simulator.

CO2 produced in calciner due to calcination reaction and Mc,C

is the amount of coal entering the calciner and yc,c is the mass
fraction of char entering the calciner. The temperature of solids
leaving the kiln was calculated as
Ts,K =

(Ms,R Cp,s Ts,R + Mg,T Cp,g Tg,T + Mg,S Cp,g Tg,S ) (Mg,in Cp,g Tg,in )
.
(Ms,K Cp,s )

In the above equation, Mg,in and Tg,in are the mass ow rate
and temperature of air entering the cooler and Ts,R is the temperature of solids exiting the cooler. The temperature of solids
entering the kiln or exiting calciner (Ts,C ) is calculated as
Ts,C =

(39)

were solved by switching off the calcination and clinkerization reactions in the calciner and kiln, respectively. For these
simulations the material and energy balances converged to an
error of 1% giving verication that numerical calculations
are correctly solving the model equations. It was also veried
that the converged solution is not a function of initial guess or

(Ms,C Cp,s Ts,C + Mg,C Cp,g Tg,C + Hcalc Mg,K Cp,g Tg,K Mg,T Cp,g Tg,T Hc,C )
.
(Ms,P Cp,s )

In the above equation, Hc,C is the heat released due to coal combustion in the calciner and Hcalc is the heat required by calcination reaction. This was easily calculated based on percentage
calcination occurring in the calciner. The heat losses in calciner
are negligible as compared to total heat supplied to the calciner
(< 5% of total energy input) and therefore was not considered
in pre-processor calculations. In this way the input conditions
(mass, mass fractions and temperature) for pre-heater, calciner,
kiln and cooler were calculated using pre-processor. The values
calculated by pre-processor were passed as input conditions to
the individual models. The individual models were then solved
iteratively as shown in Fig. 4. The iterations were continued till

(38)

for solution to converge. We have also carried out several test

simulations of limiting cases to verify that implemented numerical techniques and computer programs are correctly solving the model equations. For example, the calciner and kiln

(Ms,K Cp,s Ts,K + Mg,K Cp,g Tg,K + HR,K + HLoss,K Mg,S Cp,g Tg,S Hc,K )
.
(Ms,C Cp,s )

In the above equation, Hc,K is the heat released due to coal

combustion in kiln, HR,K is heat required for clinker reactions
and HLoss,K is the loss from the kiln. The temperature of solids
entering the kiln is essentially same as temperature of gases
leaving the calciner (Tg,C ). Finally, the temperature of solids
entering the calciner or leaving the pre-heater assembly (Ts,P )
was calculated as
Ts,P =

the temperature of solids and gases at exit of individual components were within error of 1%. Suitable under-relaxation
parameters were used. Typically 1020 iterations were required

(40)

under-relaxation parameters. An easy to use, graphical user

interface (GUI) based software called RoCKS (Rotary Cement
Kiln Simulator) was developed based on the integrated modules
of pre-heater, calciner, kiln and cooler.

4. Results and discussion

The integrated model (RoCKS) presented in the previous
section was used to simulate performances of pre-heater, calciner, rotary kiln and cooler in clinker manufacturing. Based on
the available data on rotary kilns (Mujumdar et al., 2006) and

2600

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

available information from some of the cement industries, a

typical clinker manufacturing conguration was selected as a
base case. Some assumptions were made to ll in the gaps in the
available data. The details of selected conguration are given in
Tables 2ac. Though the developed mathematical framework is
general enough to accommodate temperature dependent physical properties like heat capacity, at this stage, these properties
were treated as constants. The physical properties of solids and
air used in this work are specied in Table 3a. Our prior simulations of kiln and calciner (Mujumdar et al., 2006; Warudkar
et al., 2005) indicated that the errors in overall energy
consumption associated with the assumption of temperature
independent values of specic heat were within 1%. The operation of the base case (described in Tables 24) was computationally studied to understand the various processes occurring
in individual units in clinker formation. On obtaining satisfactory results from the base case, several numerical experiments
were performed using the model for understanding interactions among different processes and for possible optimization
of clinker manufacturing process.
4.1. Base case simulation

Table 3
(a) The physical properties of solids and air, (b) particle size and composition
of coal and (c) particle size and composition of raw meal
Description

Air

Raw meal

Coal

(a)
Thermal conductivity, W/m K
Emmisivity
Heat capacity, J/kg K
Viscoscity, kg/m s
Density, kg/m3
Char caloric value, kcal/kg
Volatile caloric value, kcal/kg

0.116
0.4
1000
1e 05
1.3

0.5
0.9
1000

1500

0.5
0.8
1000

1000
5600
11 900

(b)
Coal particle size, 50 m
Volatile (CH4 )a , 27%
Char, 58%
Ash, 15%
(c)
Raw meal particle size, 50 m
CaCO3 , 80%
CaO, 0%
SiO2 , 14%
Fe2 O3 , 3%
AL2 O3 , 3%
a Mujumdar

The predicted results from the simulation of the base case

are summarized in Table 4. The mass fractions and temperatures of solids and air in pre-heaters, calciner, kiln and cooler
obtained from the simulation are plotted in Figs. 5 and 6, respectively. It is important to note that the ow of air is counter
current with respect to the ow of solids in the system. The abscissas of Figs. 5 and 6 denote particular equipment in clinker
formation as discussed below. Abscissas 14 corresponds to

Table 2
The dimensions of (a) pre-heater unit, (b) kiln and (c) cooler
S/No.

Description

Units

Values

(a)
1
2
3
4
5
6
7
8
9
10

No. of pre-heaters
Height of cylindrical section
Height of conical section
Diameter of cyclone
Diameter of cone tip
Refractory thickness
Shell thickness
Inlet duct height
Inlet duct width
Diameter of outlet pipe

m
m
m
m
m
m
m
m
m

4
5
3
3
1
0.13
0.03
1
1
1

(b)
1
2
3
4
5

Length
Inner diameter
Coating thickness
Refractory thickness
Shell thickness

m
m
m
m
m

50
3.4
0.136
0.2
0.025

(c)
1
2

Length
Width

m
m

11
1

et al. (2006).

pre-heater assembly. Abscissas 4 and 5 denote the calciner in

the system. Abscissas 515 denote the rotary kiln and 1518
denote the cooler section. Fig. 5 shows a plot of mass fractions in pre-heater, calciner, kiln and cooler (only CaO, C2 S
and CO2 mass fractions are plotted for the sake of brevity).
Since there is no reaction occurring in pre-heater section, the
composition of CaO and CO2 in this section do not vary. However, in the pre-heater section the raw meal gets heated from
300 to 1069 K and hot gases from calciner get cooled (from
1224 to 539 K) as can be seen from Fig. 6. As the raw meal
passes through the calciner, it gets partially calcined. Therefore, CaO concentration increases in the calciner section as
can be seen from Fig. 5. Similarly since CO2 is formed due
to calcination and coal combustion, the mass fraction of CO2
increases in the calciner. Coal combustion in the calciner accounts for rise in temperature of both solids and gas in the
calciner (see Fig. 6). Remaining clinkerization reactions occur
in kiln. The mass fraction and temperature proles obtained in
kiln (as shown in Figs. 5 and 6) are similar to previously published results (Mujumdar and Ranade, 2006; Mastorakos et al.,
1999). Since there is no reaction occurring in the cooler, mass
fraction of solids in the clinker cooler do not vary. However,
air entering the cooler gets pre-heated (from 300 to 1200 K)
and solids get cooled (from 1632 to 476 K) in the cooler section. The predicted energy requirements of individual processes
like clinkerization reactions, losses, melting predicted by the
model are listed in Table 4. The obtained results are qualitatively similar to previously published results (Engin and Ari,
2005). The performance of the overall system was characterized
in terms of NEC per unit weight of product (clinker coming out

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

2601

Table 4
Complete energy balance of the system
S/No.

Description

Pre-heater

Calciner

Kiln

Cooler

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

Solid inlet temperature, K

Mass ow rate, kg/s
Air inlet temperature, K
Air ow rate, kg/s
Coal ow rate, kg/s
Coal inlet temperature, K
Heat with solids in, kJ/kg clinker
Heat with air in, kJ/kg clinker
Heat with coal in, kJ/kg clinker
Combustion of coal, kJ/kg clinker
Heat of reaction, kJ/kg clinker
Heat of melting, kJ/kg clinker
Heat of solids leaving, kJ/kg clinker
Heat of air leaving, kJ/kg clinker
Heat of vent air in cooler, kJ/kg clinker
Heat with ash, kJ/kg clinker
Loses, kJ/kg clinker

300
50
1214.8
60.8

463.0
2297.9

1650.2
1014.4

98.9

1069.2
50
1114.5
46.7
2.15
350
1650.2
1603.6
23.2
1876.7
1384.5
44.2
1415.3
2297.9

3.22
43.5

1214.8
37.74
1229.9
16.2
0.9
350
1415.3
615.0
9.7
747.1
219.0

1622.4
1603.6

2.24
140.7

1622.4
32.3
300
45

1622.4
416.7

463.0
1415.3
109.4

reasonable when compared with industrial observations. Overall the integrated simulator was able to predict the clinker manufacturing process in cement industry reasonably well.

0.7
CaO -Solid mass fraction
C2S-Solid mass fraction

0.6

Mass fraction

CO2-Gas mass fraction

0.5

4.2. Inuence of key design and operating parameters on NEC

0.4

On obtaining a reasonable agreement, the model was used to

explore space of design and operating parameters to understand
inuence of these parameters on the performance of clinker
manufacturing. All these simulations were carried for a xed
product composition (C3S mass fraction 0.48 in the product).
This was achieved by altering coal ow rate either to calciner
or kiln. This analysis is presented in the section below.

0.3
0.2
Calciner

0.1

Cooler

Pre-Heaters
Kiln

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Divisions across the cement clinker process
Fig. 5. Solid and gas mass fractions in pre-heaters, calciner, kiln and cooler
in a cement clinker process.

of the kiln). The NEC is calculated as

NEC = (ERXN,C + ERXN,K + EMELT,K ) + ELOSS
+ (EG,OUT + ES,OUT EG,IN ES,IN ).

(41)

In the above equation, ERXN denotes the energy required for

clinkertization reactions and subscripts C and K denotes the
calciner and the kiln, respectively. The term EMELT,K denotes
the energy required for melting in the kiln. ELOSS denotes the
summation of energy losses from pre-heater assembly, calciner
and kiln. The other terms denote energy ow rates (subscripts
IN or OUT) for the gas and solid streams (subscripts G or
S) which denote the energy required to raise the sensible heat
of the solids. Based on above calculations, the NEC predicted
by the integrated simulator, for these operating conditions was
2635 kJ/kg clinker (630 kcal/kg clinker) which seems to be

4.2.1. Effect of number of pre-heaters

The effect of changing number of pre-heaters in pre-heater
assembly (from 3 to 5) on NEC was studied. For this simulation
the coal in the kiln was adjusted to get same product composition at the kiln exit. The results for this simulation are shown
in Fig. 7. It can be seen from Fig. 7 that as number of preheaters in pre-heater assembly increases, solids get pre-heated
to a higher temperature before they enter the calciner (see secondary axis in Fig. 7). Therefore the coal requirement for a
xed product composition decreases. Thus the NEC decreases
as number of pre-heaters increases. However, the overall capital cost increases by increasing number of pre-heaters in the
system. The developed model will be useful to carry out cost
to benet analysis for introducing additional pre-heater in the
pre-heater assembly.
4.2.2. Effect of percentage calcination in the calciner
The pre-calcination of raw meal in calciner is an important
process in cement process. We have studied the effect of percentage calcination in calciner on NEC. To vary the percentage calcination in calciner the coal feed rate to the calciner
and kiln was altered for the same clinker composition. The

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

Solid temperature
Gas temperature

Gas temperature, K

2000

1500

2000

1500

Gases entering 3rd preheater

1000

1000

Solid temperature, K

2602

Solids leaving 3rd preheater

500

500

Cooler

Calciner
Pre-heaters

Kiln

0
0

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Divisons accross the cement clinker process

Fig. 6. Temperature prole across pre-heaters, calciner, kiln and cooler in a cement clinker process.

1080

635
1060
630
1040
625
1020

620
615

1000
2

4
5
Number of pre-heaters

Energy Consumption, kcal/kg clinker

640

666

1500

661
1400
656
651

1300

646
1200

641
636

1100

Kiln exit gas Temperature, K

1100
Solid temperature entering calciner, K

Energy consumption, kcal/kg clinker

645

631
626
40

50

60

70
80
% Calcination

90

1000
100

Fig. 7. Effect of pre-heater number on overall energy consumption.

Fig. 8. Effect of percentage calcination on overall energy consumption.

simulation results are shown in Fig. 8. As can be seen from

Fig. 8, the NEC was found to decrease till 70% calcination and
then it increases with further increase in percentage of calcination. The secondary axis of Fig. 8 shows that the kiln exit gas
temperature also shows a similar trend. Table 5 shows a complete comparison of heat of reaction occurring in kiln and calciner in this process. The heat of reaction in kiln decreases as
the percentage calcination increases in calciner. The total heat
of reaction in kiln is the summation of heat of calcination (endothermic reaction) and the heat of clinker formation (exothermic reactions). When calcination occurs pre-dominantly in the
pre-calciner (> 70%), the energy requirements for reactions in
kiln reduce drastically. This causes increase in kiln ue gas
temperature and increase in losses from kiln shell. Therefore
the NEC and kiln ue gas temperature increases if more than
70% calcination occurs in the calciner. The model and the

Table 5
Heat of reaction in calciner and kiln
S/No.

Heat of reaction

1
2

Calcination, %
Heat of reaction
in calciner, kJ/kg
clinker
Heat of reaction in
kiln, kJ/kg clinker

50
60
1038.5 1176.6

582.9

433.2

70
1384

219

80
1618.5

90
1802.5

20.2 178.4

RoCKS software were thus able to provide valuable clues for

determining the optimum percentage calcinations desired for
minimizing NEC.

630
1500
625
1000
620

615

500
2

5
6
Kiln Rpm

Energy consumption, kcal/ kg clinker

635
1600
630
1200
625
800
620

615

400
1

1.5
2
2.5
Kiln tilt, Degree

1350

645

1300

640

1250

635

1200

630

1150

625

1100

620
0.06

0.08

0.1
0.12
0.14
Grate speed, m/s

0.16

1050
0.18

Fig. 10. Effect of cooler grate speed on overall energy consumption.

2000

0.5

650

640

2603

Secondary air temperature, K

2000

Residence time of solids, s

Energy Consumption, kcal/kg clinker

635

Residence time of solids in kiln, s

2500

640

Energy consumption, kcal/ kg of clinker

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

3.5

Fig. 9. Effect of (a) kiln RPM and (b) kiln tilt on overall energy consumption.

4.2.3. Effect of kiln RPM, kiln tilt and grate speed of clinker
cooler
The effect of kiln rotational speed and kiln tilt on the overall
performance is shown in Fig. 9a and b. For these simulations
the coal ow rate to the kiln was varied to maintain constant
product composition. It can be seen from Fig. 9a as kiln RPM
decreases, the NEC decreases. Changes in kiln RPM changes
the bed height and the residence time of solids in the kiln as
can be seen from Fig. 9a and b (2002.4 s for 3 rpm; 1058.2 s for
5.5 rpm and 703.4 s for 8 rpm). Our simulation results indicate
that it seems to be benecial to operate kilns at lower rpm as
long as adequate mixing of solids is occurring. From Fig. 9b it
can be seen that energy consumptions in kilns operated at lower
tilt is less as compared to kilns at higher tilt. The grate of clinker
cooler is the important parameter that controls the residence
time of solids and subsequently the heat exchange between hot
solids and counter current air in the cooler. We have studied the
inuence of varying grate speed on overall energy consumption.
The results for these simulations are shown in Fig. 10. It can be

seen from Fig. 10 that the NEC increases with increasing grate
speed. The increase in grate speed reduces residence time of
solids in the cooler. This results in less convective heat transfer
between solids and air as clearly indicated by temperature of
secondary air plotted in Fig. 10. Therefore the simulation results
indicate that it is better to operate grates in the cooler at lower
speed. The simulations presented here provide useful trends of
energy consumption as a function of key operating parameters
in cement clinker process. This result also gives us a scope to
understand the importance of design parameters (kiln tilt) on
plant performance and can be very useful to plant engineers.
4.2.4. Effect of solid loading
The predicted results in the form of NEC and corresponding
overall losses for different solids ow rates are shown in Fig. 11.
It can be seen that the NEC per unit weight of product decreases
as solids ow rate increases. This is because the net energy loss
from the entire system decreases as the solid ow rate increases
(see Fig. 11). Thus, it is benecial from the point of view
of energy consumption to operate the units with higher solids
ow rate. Other operational concerns like increase in dusting
and mixing, however, need to be considered while identifying
maximum solids ow rate specically for cement kilns.
4.2.5. Effect of coal composition
The effect of varying coal composition to the kiln on NEC
is shown in Fig. 12. From Fig. 12, it can be seen that the
overall energy consumption does not change signicantly with
changing coal composition (ash content 9%, 15% and 40%).
For these simulations the coal ow rate to the kiln was adjusted so that the same amount of energy is supplied to the
kiln. Therefore the insignicant change in overall energy consumption does not seem to be surprising. However, as the
coal composition changes, the ame characteristics in the kiln
vary. The predicted dimensionless ame length by the simulator for varying coal composition is shown in Fig. 12. The
ame length was calculated by tracking the region in freeboard
where char and volatiles composition in coal go to zero. The

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

640

330

630

310

620

290

610

270

600

250
44

45

46
47
48
49
Raw meal flow rate, kg/s

50

1120

600

1115

598

1110
596
1105
594
1100
592

1095

590

51

Kiln flue gas temperature, K

350

Energy comsumption, kcal/kg of clinker

650

Total losses in clinker process, kJ/kg clinker

Energy consumption, kcal/kg clinker

2604

1090
15

20
25
30
35
40
45
Air flow rate through secondary shell, kg/s

Fig. 11. Effect of raw meal ow rate on overall energy consumption.

0.56

650

640

0.52

630
0.48
620
0.44

610

600

Flame length, dimensionless

Energy comsumption, kcal/kg of clinker

Fig. 13. Effect of secondary shell, on overall energy consumption.

0.4
30

40

50
60
Char Percentage, %

70

Fig. 12. Effect of coal composition on overall energy consumption.

dimensionless ame length was calculated as the ratio of predicted ame length to length of the kiln. It can be seen that
coal with higher ash content tends to have a longer ame as
compared to coal with lower ash content. The ame length is a
complicated function of amount of oxygen, amount of char and
temperature of gas and particle in the freeboard region. Coals
with higher ash content tends to consume oxygen at a slower
rate and therefore result in longer ames. Such simulations can
therefore provide useful information to kiln operators to predict the ame characteristics for wide variety of coal available
in the market.
4.2.6. Effect of secondary shell
Heat losses to the surrounding from the kiln shell by radiation and convection are a signicant source of energy loss in
cement kilns and therefore the overall process. These losses
can be reduced by using a secondary shell. The idea is to cover
the kiln shell with another metallic shell having low surface

emissivity and thermal conductivity (Engin and Ari, 2005).

However, merely covering kiln shell with metallic shell and
insulating it can lead to enormously high shell temperatures.
Hence a practical approach to use secondary shell would be to
feed air through the interstitial space of shell and secondary
shell to recover the energy and still operate kilns under realistic conditions (Mujumdar et al., 2006). The developed RoCKS
frame work was used to explore the possibility of using such
a secondary shell. The losses in kiln reduced from 140 kJ/kg
of clinker to 1.4 kJ/kg of clinker on applying a secondary
shell and insulation of dimensions and operating conditions
specied in Mujumdar et al. (2006). The NEC reduces from
2635 kJ/kg clinker to 2493 kJ/kg clinker (i.e., 630 kcal/kg
clinker to 596 kcal/kg clinker) by using secondary shell and
passing air of about 30 kg/s through the interstitial space
(Fig. 13). If the air coming out of annular space at 496 K
can be utilized within the cement plant (refrigeration, drying
of y ash and so on), the use of secondary shell appears to
be promising for reducing NEC in the clinker manufacturing
process.
5. Conclusions
A comprehensive model was developed to simulate complex
processes occurring in pre-heater, calciner, kiln and cooler for
clinker formation in cement industry. The models for pre-heater
and calciner were developed assuming solids and gas to be
completely back mixed. The computational model for the kiln
was developed assuming gas and solids as plug ow. The integrated simulator was converted into simple to use GUI based
software for cement industry, named as RoCKS. RoCKS was
used to simulate performance of pre-heater, calciner, kiln and
cooler for clinker formation. Detailed validation was unfortunately not possible since adequate industrial data could not be
obtained. However, the model predictions agreed reasonably
with industrial observations. RoCKS was used to understand
inuence of various design and operating parameters on overall

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

performance. Specic conclusions based on this computational

study are:
Including an additional pre-heater reduces NEC. The developed model can be used to evaluate relative benets of energy savings by additional pre-heater and required additional
capital expenses.
There is an optimum value for percentage of calcination carried out in calciner with respect to overall energy consumption in clinker manufacture. With the parameters selected in
this work, this optimum value of percentage calcination in
calciner is about 70.
The simulation results indicated that operating kiln with
higher solid loading, lower rpm, lower tilt and lower grate
speed reduces energy consumption per unit production. The
upper limit on solid loading (bed height) and lower limits on
rpm and tilt (mixing and heat transfer) need to be identied
based on other practical issues.
The use of secondary shell appears to be a promising method
to reduce overall energy consumption, if the hot air generates
in such secondary shell ( 200 C) can be utilized in some
other processes in cement plants.

hcyc
Hc,C
Hc,K
Hcalc
Hcomb
Hcombg
HLoss,K
HR,K
kg
kr
ks
ksh
ks,c
rk
s
r k
s

L
ma
mAl2 O3 ,i
mCO2 ,C

The model was also able to predict kiln characteristics like

maximum ame temperature and overall ame length for coals
with different compositions. The models and results presented
here will help in developing a better understanding of clinker
manufacturing process and may provide clues for possible
optimization.

Notation
a
A0
Acyi
Acyo
Ap
LA
p
Cp,c
Cp,g
Cp,s
dc
dp
E1
E2
fc
fv,0
Fpw
hc
hc,c
hc,L

surface area per unit volume, m2 /m3

devolatilization constant
internal surface area of cyclone, K
external surface area of cyclone, K
surface area of coal particle, m2
surface area of solid particle, m2
specic heat capacity of coal particle, J/kg K
specic heat capacity of air, J/kg K
specic heat capacity of solids, J/kg K
inner diameter of cyclone, m
radius of particle, m
energy of activation for char combustion, J/mol
energy of activation for calcination, J/mol
fraction of heat given to coal particle released
due to coal combustion
initial mass fraction of volatiles in coal particle
view factor
heat transfer coefcient between coal particle
and gas, W/m2 K
heat transfer coefcient between clinker and gas,
W/m2 K
heat transfer coefcient between solid particle
and gas, W/m2 K

mCaCO3 ,i
mFe2 O3 ,i
mSiO2 ,i
mg
mgin
mgout
mg,K
mp,c
mp,cin
mp,cout
mpc,0
mp,L
mp,Lin
mp,Lout
ms
Mc,C
Mc,K
Mg
Mg,K
Mg,P
Mg,S
Mg,T
Ms
Mse
Ms,C
Ms,K

2605

heat transfer coefcient between particle laden

gas and cyclone inner wall, W/m2 K
heat of coal combustion in calciner, J/kg
heat of coal combustion in kiln, J/kg
heat of calcination reaction in calciner, J/kg
heat of char combustion, J/kg
heat of volatile gas phase combustion, J/kg
heat losses in the kiln, J/kg
heat required for clinker reactions, J/kg
thermal conductivity of air/gas, W/m K
thermal conductivity of refractory, W/m K
thermal conductivity of clinker, W/m K
thermal conductivity of shell, W/m K
rate constant of char combustion, kg/m2 s kPa
rate constant of calcination of calcium carbonate, mol/m2 s1
rate constant of calcination of calcium carbonate, mol/m2 s1
total height of cyclone, m
mass of air in cooler, kg/s
mass of total aluminum oxide in solids in calciner, kg/s
mass of carbon-dioxide produced in calciner due
to calcination, kg/s
mass of total calcium carbonate in solids in calciner, kg/s
mass of total ferrous oxide in solids in calciner, kg/s
mass of total silicon dioxide in solids in calciner, kg/s
mass of gas in calciner, kg
mass of air entering in calciner,kg/s
mass of air leaving calciner, kg/s
mass of air leaving the kiln calciner, kg
mass of coal particle, kg
mass of coal particle entering calciner, kg
mass of coal particle leaving calciner, kg
initial mass of coal particle, kg/s
mass of solid particle, kg
mass of solids entering calciner, kg
mass of solids leaving calciner, kg
mass of solids/clinker in cooler, kg
mass of coal entering the calciner, kg/s
mass of coal entering the kiln, kg/s
mass of gas in cyclones, kg/s
mass ow rate of secondary air entering the
kiln, kg/s
mass ow rate of gas entering the pre-heater,
kg/s
mass ow rate of secondary air entering the
kiln, kg/s
mass ow rate of tertiary air entering the calciner, kg/s
mass of solids in cyclones, kg/s
mass of solids entrained by gas in cyclones, kg/s
mass ow rate of solids leaving the calciner, kg/s
mass ow rate of clinker leaving the kiln, kg/s

2606

Ms,P
MwCaCO3
MwCaO
Mwchar
MwCO2
MwO2
Mwvol
Mww
Nu
cN
p
LN

pO2
pCO2
peq
P
P
Pr
rc
rcomb
rcombg
ri
r0
rp
rr
R
Rc
Re
T0
Tc,i
Tcl
Tf
Tg
Tg,in
Tg,out
Tg,K
Tg,S
Tg,T
Tg,P
Tiw,i
Tow,i
TL
Tr,i
Ts
Ts,C
Ts,R
u0
us,x
us,y
ug,x

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

mass ow rate of solids entering the pre-heater,

kg/s
molecular weight of calcium carbonate, kg/kmol
molecular weight of calcium oxide, kg/kmol
molecular weight of carbon, kg/kmol
molecular weight of carbon-dioxide, kg/kmol
molecular weight of oxygen, kg/kmol
molecular weight of volatile, kg/kmol
molecular weight of water, kg/kmol
Nusselt number
number of coal particles entering calciner per
second
number of solid particles entering calciner per
second
partial pressure of oxygen in gas, kPa
partial pressure of carbon-dioxide in gas, kPa
equilibrium partial pressure for carbon-dioxide
in gas, kPa
the percentage calcination occurring inside the
calciner
pressure drop across the cyclone, mm of H2 O
Prandtl number
rate of calcination, kg/s
rate of combustion of char particles, kg/s
rate of combustion of volatiles, kg/s
internal diameter of cyclone, m
external diameter of cyclone, m
radius of solid particle, m
internal diameter of cyclone shell, m
gas constant
non-dimensional form of radiative heat transfer
coefcient
Reynolds number
ambient air temperature, K
temperature of solids and gas in cyclone, K
temperature of coal particle, K
average temperature of solids and air in
cooler, K
temperature of gas, K
temperature of gas entering calciner, K
temperature of gas exiting calciner, K
temperature of gas leaving the kiln, K
temperature of secondary air, K
temperature of tertiary air, K
temperature of gas leavingthe pre-heater, K
the internal wall temperature of the cyclone, K
the external wall temperature of the cyclone, K
temperature of solid particle in calciner, K
the temperature of interface of refractory and
shell in cyclone, K
temperature of solids/clinker in cooler, K
temperature of solids entering the kiln, K
temperature of solids exiting the cooler, K
inlet gas velocity in cyclone, m/s
grate speed in x direction, m/s
grate speed in y direction, m/s
air velocity in x direction, m/s

ug,y
Vreact
xAl2 O3 ,C
xCaCO3 ,C
xCaO,C
xFe2 O3 ,C
xSiO2 ,C
yc,c
yc,cin
yc,cout
yc,K
yv,c
yO2
yO2 ,in
yO2 ,out
yCO2
yCO2 ,in
yCO2 ,out
yv
yw
Z

air velocity in y direction, m/s

volume of reactor, m3
mass fraction of aluminum oxide entering kiln
mass fraction of calcium carbonate entering kiln
mass fraction of calcium oxide entering kiln
mass fraction of ferrous oxide entering kiln
mass fraction of silicon dioxide entering kiln
mass fraction of char in coal particle in calciner
mass fraction of char entering in coal particle
mass fraction of char leaving in coal particle
mass fraction of char in coal particle entering
the calciner
mass fraction of volatiles in coal particle
mass fraction of oxygen in gas
mass fraction of oxygen entering calciner in gas
mass fraction of oxygen leaving calciner in gas
mass fraction of carbon-dioxide in gas
mass fraction of carbon-dioxide entering calciner in gas
mass fraction of carbon-dioxide leaving calciner
in gas
mass fraction of volatiles in gas
mass fraction of water in gas
stoichiometric component

Greek letters

c
cy
L
m,p

g
s



porosity of clinker bed in cooler

emissivity of coal particle
emissivity of cyclone outer wall
emissivity of solid particle
mass efciency of the cyclone
residence time of coal particle in calciner, s
density of air/gas, kg/m3
density of solids, kg/m3
StephanBoltzmann constant (W/m2 K 4 )
residence time of raw meal particle in calciner, s

Chemical species
C2 S
C3 S
C3A
C4AF

(2CaOSiO2 )
(3CaOSiO2 )
(3CaOAl2 O3 )
(4CaOAl2O3 Fe2 O3 )

Acknowledgments
The authors wish to acknowledge nancial support provided
by CSIR (under the NMITLI scheme) for this study. The authors would also like to acknowledge many helpful discussions
with Professor Anurag Mehra during the course of this work.
One of the authors, K.S.M is grateful to Council of Scientic
and Industrial Research (CSIR), India for providing nancial
support.

K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 2607

References
Engin, T., Ari, V., 2005. Energy auditing and recovery for dry type cement
rotary kiln systemsa case study. Energy Conversion and Management
46 (4), 551562.
Gupta, A.V.S.S.K.S., Nag, P.K., 2000. Prediction of heat transfer coefcient in
the cyclone separator of a CFB. International Journal of Energy Research
24, 10641079.
Hamor, R.J., Smith, I.W., Tyler, R.J., 1973. Kinetics of combustion of
pulverised brown coal char between 630 and 2200 K. Combustion and
Flame 21, 153.
Hindmarsh, A.C., 1983. ODEPACK, a systematized collection of ODE
solvers. In: Stepleman, R.S., et al. (Eds.), Scientic Computing. IMACS
Transactions on Scientic Computation, vol. 1. North-Holland, Amsterdam,
pp. 5564.
Hu, N., Scaroni, A.W., 1996. Calcination of pulverised limestone particles
under furnace injection conditions. Fuel 75, 177186.
Kang, S.K., Kwon, T.W., Kim, S.D., 1989. Hydrodynamic characteristics of
cyclone reactors. Powder Technology 58, 211220.
Kramers, H., Croockewit, P., 1952. The passage of granular solids through
inclined rotarykilns. Chemical Engineering Science 1, 259265.
Lede, J., Li, H.Z., Soulalignac, F., Villermaux, J., 1987. Measurment of solid
particle residence time in cyclone reactor: a comparison of four methods.
Chemical Engineering Process 22, 215222.
Li, J.Q., Wei, F., Jin, Y., 2003. Numerical simulation of pulverized
coal combustion and NO formation. Chemical Engineering Science 58,
51615171.
Locher, G., 2002. Mathematical models for the cement clinker burning
process, part 4: grate cooler. ZKG International 55 (6), 4657.

2607

Lu, J., Huang, L., Hu, Z., Wang, S., 2004. Simulation of gassolid, two-phase
ow, coal combustion and raw meal calcination in a pre-calciner. ZKG
International 57 (2), 5563.
Mastorakos, E., Massias, A., Tsakiroglou, C.D., Goussis, D.A., Burganos,
V.N., 1999. CFD predictions for cement kiln including ame modeling, heat
transfer and clinker chemistry. Applied Mathematical Modeling 23, 5576.
Mujumdar, K.S., Ranade, V.V., 2003. Coupling of CFD models with different
time scalesa case study of rotary cement kiln. In FLUENT Users
Group Meeting held at Pune.
Mujumdar, K.S., Ranade, V.V., 2006. Simulation of rotary cement kilns using
a one-dimensional model. Transactions of IChemE, Part A, Chemical
Engineering Research and Design 84 (A3), 165177.
Mujumdar, K.S., Arora, A., Ranade, V.V., 2006. Modeling of rotary cement
kilns: applications to reduction in energy consumption. Industrial and
Engineering Chemistry Research 45 (7), 23152330.
Nsofor, E.C., Adebiyi, G.A., 2001. Measurements of the gasparticle
convective heat transfer coefcient in a packed bed for high-temperature
energy storage. Experimental Thermal and Fluid Sciences 24 (12), 19.
Peray, K.E., 1984. Rotary Cement Kilns. second ed. Chemical Publishing
Company Ltd., New York.
Spang, H.A., 1972. A dynamic model of a cement kiln. Automatica 8,
309323.
Stanmore, B.R., Gilot, P., 2005. ReviewCalcination and carbonation of
limestone during thermal cycling for CO2 sequestration. Fuel Processing
Technology 86, 17071743.
Warudkar, S., Kumar, G., Mujumdar, K.S., Ranade, V.V., 2005. Modeling of
cement plant calciners. In: CHEMCON 2005, New Delhi.
Watkinson, A.P., Brimacombe, J.K., 1982. Limestone calcination in a rotary
kiln. Metallurgical Transactions B 13B, 369378.