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 Technology–Bombay, 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 gas–solid 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 (solid–solid, gas–solid and homogeneous reactions in gas
phase) in rotary kiln were modeled using three sub-models coupled to each other. Solid–solid 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-flow 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 influence 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.
䉷 2007 Elsevier Ltd. All rights reserved.

Keywords: Cement; Energy consumption; Reaction engineering model

1. Introduction A schematic of typical clinker making process is shown in

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 re-
duction 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 䉷 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2007.01.063
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 re-
quired 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 sufficient 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 as-
sembly 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 Exhaust to
Raw meal atmosphere
Pre-heater Hot gases
Assembly to pre-heater
Coal
Calciner Kiln Exhaust
Pre-heated
Raw meal
Kiln
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 refrac-
tories. Counter current flow of gas entrains solid particles in
the free board region. Such entrainment enhances rates of ra-
diative heat transfer by increasing effective emissivity and con-
ductivity. The hot clinkers are discharged from kiln to clinker
cooler and hot gases from kiln exhaust are sent to the cal-
ciner. In a clinker cooler, a part of energy of solids is recov-
ered 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 pre-
heater, calciner, kiln and cooler in order to capture key char-
acteristics 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 fluid dynamics (CFD) based models to simulate
either calciner (for example, Lu et al., 2004) or kiln (for exam-
ple, 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
2591
Tertiary Air
Secondary Air Vent Air
Coal
Cooler
Cooled clinker
Air to cooler
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
reaction–engineering 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 com-
putational model could also predict the influence of kiln oper-
ating 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 in-
cluded coupling of pre-heater, calciner, kiln and clinker cooler.
This work was undertaken to fulfill 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 strat-
egy 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 findings 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 pre-
heater, 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 briefly.
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 pre-
heated 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 flow 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 flows in cyclones. Our CFD simulations (carried
out for typical values of operating conditions as gas flow rate =
60.83 kg/s; solid flow rate = 50 kg/s; inlet gas temperature =
740 K; inlet solid temperature=500 K; cyclone diameter =6 m;
Pre-heater
Gas-solid heat transfer
(gas/solids/walls), particle
laden turbulentflow,
losses
Strongly
Coupled
Rotary Kiln
Coal/gas phase comb, bed
height variation, clinker formation,
melting/solidification, coating
formation, losses
Fig. 2. Key issues in modeling cement clinker process.
Table 1
Kinetic constants for coal combustion reaction
Reactions K0
C + O2 = CO2 9.18 × 10−1 (kg/m2 s KPa)
CH4 + 2O2 = CO2 + 2H2 O 1.6 × 1010 (m3 /kg s)
height of cyclone = 11 m) indicated that the solid and gas tem-
peratures inside the cyclone were quite uniform (in range of
637–644 K) and close to exit temperatures (average tempera-
ture 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 flux 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 frame-
work the calcination reaction was assumed to occur in the last
cyclone. The framework is, however, quite general to accom-
modate 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 combus-
tion 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 res-
idence time of raw meal in the cyclone. Few empirical correla-
tions have been proposed to predict the average residence time
of particles in cyclones (Kang et al., 1989; Lede et al., 1987).
Calciner
Turbulent gas phase/
coal combustion, calcination,
heat transfer (gas/solid/
walls), losses
Clinker Cooler
Gas-solid heat transfer,
temperature gradients in
clinker bed region
E (J/mol) Reference
6.81 × 104 Hamor et al. (1973)
1.08 × 105 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 param-
eter 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 confirmed from our prior simula-
tions (Warudkar et al., 2005) that varying residence time in the
calciner by 10% had relatively small effect (∼ 2.5%) on pre-
dictions 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 cal-
cination 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 exper-
iments were close to industrial operations (bed temperature ∼
1000.1300 K). Their data was therefore used to find 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 solid–solid
and solid–liquid 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 re-
fractories. Counter current flow of gas entrains solid particles
in the freeboard region. Such entrainment enhances rates of ra-
diative heat transfer by increasing effective emissivity and con-
ductivity. In this section we discuss the key issues involved in
modeling the cement kilns very briefly. 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.
2.4. Clinker cooler
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-flow 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 cool-
ers is predicting the heat transfer coefficient between hot solids
and cross-flow 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 coefficient for
2593
wide range of temperatures (200–1000 ◦ C). Since the temper-
atures in clinker cooler are in the same range this correlation
was used to model heat transfer coefficient 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 frame-
work of computational models was developed for dry process
of clinker formation since this process is widely used in ce-
ment 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 evap-
oration 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 follow-
ing inlet streams were considered:
1. Solids from the (i − 1)th cyclone (Ms,i−1 at temperature
Ti−1 ).
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
Ms,i−1 + Mse,i+1 = Ms,i + Mse,i , (1)
Mse,i = (1 − m,p ) × Ms,i . (2)
In the above equations m,p represents the particle capture ef-
ficiency 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 writ-
ten as
Ms,i−1 · Cp,s · Tc,i−1 + 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

Mg,Mse,iTi Gas out

Ms,i-1Ti-1
Raw meal Coal in
Tiw,i Tow,i Air in
i 12
Loses
Radiation and convection
Mg,Mse,i-1Ti-1
Losses

1. Refractory
2. Shell

Ms,i,Ti Partially calcined raw meal

Secondary air Tertiary air Vent air

∂T
=0
∂y

L ∂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 specific heat of solids walls, which is equal to loss from shell walls due to convection
and air, respectively. Subscript g represents the air and Tc,i and radiation. The energy balance for heat transfer in cyclone
represents the temperature of solids and air in the ith cyclone. cross-section is written as
hcyc represents the heat transfer coefficient for energy exchange
2 ·  · L · kr · [Tiw,i − Tr,i ]
between particle laden gas and cyclone inner walls. hcyc was hcyc · Acyi · [Tc,i − Tiw,i ] = , (4)
evaluated from the following empirical correlation given by ln(rr /ri )
Gupta and Nag (2000) for heat transfer in cyclones: 2 ·  · L · kr · [Tiw,i − Tr,i ] 2 ·  · L · ksh · [Tr,i − Tow,i ]
  = ,
hcyc dc P ln(rr /ri ) ln(r0 /rr )
= 702.818 + 9.0287 × 10−14 u0 Re + 11.1385 (5)
kg u20
  2 ·  · L · ksh · [Tr,i − Tow,i ]
P
+ 4.50398 × 10−5 Re + Rc , ln(r0 /rr )
u0
= hconv · Acyo · [Tow,i − T0 ] +  · cy · Acyo · [Tow,i
4
− T04 ].
where
  (6)
4 −T4
Tiw g dc
Rc = Fp−w . (3b) In the above equations, Tiw,i is the internal wall temperature of
Tiw − Tg kg
the ith cyclone, Tr,i is the temperature of interface of refractory
The LHS of Eq. (3a) thus represents the total energy entering and shell, Tow,i is the temperature of external wall of the ith
the cyclone and RHS represents the energy leaving out of the cyclone and T0 is the ambient temperature. L is the total height
cyclone. At steady state the heat given to cyclone walls must the cyclone, kr is the thermal conductivity of the refractory
be same as heat conduction in through refractory and cyclone and ksh is the thermal conductivity of cyclone walls. r0 is the
 K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 – 2607 2595

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 coeffi-
cient of external wall, cy is the emissivity of the cyclone outer
wall and  is the Stefan–Boltzmann 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 Newton–Raphson
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.
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 cal-
ciner is written as
dmp,c
= −A0 · fv,0 · mpc,0 − ks,c · e−E1 /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 first 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
 
dyv,c 1 dmp,c
=− A0 · fv,0 · mpc,0 + yv,c · , (8)
d mp,c d
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 specific 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 first 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 coefficient, hc , is evaluated using the
correlation of Ranz and Marshall as
hc dp
= 2 + 0.6(Re)0.5 (Pr)0.33 . (11)
kg
The final 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
CaCO3 → CaO + CO2 . (12)
The over all mass balance for the raw meal particle is given by
dmp,L Mw CO2
= −rc · , (13)
d Mw CaCO3
where
rc =r ks · e−E2 /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 car-
bonate, respectively. rc is the rate of calcination of calcium car-
bonate 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 influence 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 reac-
tion as

rc = r ks · e−E2 /RT L · 4 ·  · rp2 · Mw CaCO3 , (15)

where yv,c is the mass fraction of volatiles in the coal particle. where
The individual species balance for rate of mass change due 
r
ks = r ks , PCO2 < 10−2 Peq , (16)
to char combustion is given by
 
  r  PCO2 − Peq
dyc,c 1 dmp,c ks = r ks , 10−2 Peq < PCO2 < Peq (17)
=− ks,c · e−E1/RT cl · Ap · pO2 + yc,c · , Peq
d mp,c d
(9) and Peq = 1.826 · 107 e(−19680/TL ) .
2596 K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 – 2607


The individual species balances for limestone and calcium dyv 1
oxide are given as = mg,in · yv,in − mg,out · yv,out
dt mg
dmCaCO3 ,L + [mp,cin · yv − mp,cout · yv ] · c Np
= −rc , (18)  
d rcombg
− · Vreact · Mw vol · Zvol
dmCaO,L rc · Mw CaO Mw vol
= . (19) 
d Mw CaCO3 dmg
−yvol · , (24)
dt
The energy balance for the raw meal particle is given as 
dyw 1
d(mp,L Cp,s TL ) = mgin · ywin − mgout · ywout
= hc,L ·L Ap · (Tg − TL ) dt mg
d   
rcombg dmg
+  · L ·L Ap · (Tg4 − TL4 ) + · Vreact · Mw w · ZH2 O − yw · ,
Mwvol dt
dmp,L
+ Cp,s · TL · , (20) (25)
d
where yO2 , yCO2 , yv , yw are the respective mass fractions of
where TL is the temperature of raw meal particle, L is the oxygen, carbon-dioxide, volatile matters and water. Mw O2 ,
emissivity of solid particle, L Ap is the area of the raw meal Mw CO2 , Mw vol and Mw w are their respective molecular
particle (which was calculated based on conversion i.e., mass of weights. Vreact is the volume of reactor. Subscripts in and out
raw meal particle reacted) and  is residence time of raw meal represent the inlet and outlet conditions and Z is the stoichio-
particle in the calciner. hc,L was estimated by using Eq. (11). metric coefficient.
The energy balance equation for the gas phase is given as
3.2.2. Continuous phase
dmg Cp,g Tg
The over all gas mass balance is given as = mgin · Cp,g · Tg,in − mgout · Cp,g · Tg
dt
dmg + Sgcomb + Sccomb + Scalc
= mgin − mgout + [mp,cin − mp,cout ] · c Np
dt − hcyc · Acyi · (Tg − Tiw ), (26)
+ [mp,Lin − mp,Lout ]·L Np , (21)
where
where mg is the mass of the air in the calciner, c Np is the
Sgcomb = rcombg · Hcombg · Vreact , (26a)
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 Sccomb = − hc · Ap · (Tg − Tcl )+ · c · Ap · (Tg4 −Tcl4 )
0
of oxygen, carbon-dioxide, volatile matters and water can be
dmp,c
written as + Cp,c · Tcl ·
d
 
dyO2 1
= mg,in · yO2 in − mg,out · yO2 out − (1 − fc ) · Hcomb · rcomb d, (26b)
dt mg
[mp,c · yc,cin − mp,c · yc,cout ] · c Np · Mw O2  
− Scalc = − hc,L ·L Ap · (Tg − TL )+ · L ·L Ap · (Tg4 − TL4 )
Mwchar
   0

rcombg dmg dmp,L
− · Vreact · Mw O2 · ZO2 − yO2 · , + Cp,s · TL + rc · Hcalc d. (26c)
Mwvol dt d
(22) In the above equations, Sgcomb, Sccomb are the heat source term
 for gas-phase from volatile combustion and char combustion,
dyCO2 1 respectively. Scalc is the heat sink term from calcination. Hcombg ,
= mgin · yCO2 in − mgout · yCO2 out
dt mg Hcalc are the enthalpies of volatile combustion and calcination.
[mp,c · yc,cin − mp,c · yc,cout ] · c Np · Mw CO2 hc,L is the convective heat transfer coefficient between raw meal
+ particles and air. The steady state equations across the cyclone
Mw char
walls were written same as that of pre-heaters explained in
+ [mp,Lin − mp,Lout ]·L Np the previous section to obtain temperature of calciner internal
  walls, refractory and outer walls.
rcombg
+ · Vreact · Mw CO2 · ZCO2 The calciner model equations were solved using an iterative
Mw vol
 method. The model equations for gas phase were solved as-
dmg suming steady state. For the first iteration, source terms from
− yCO2 · , (23)
dt 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 defined to
accelerate convergence. Typically about 20 iterations were re-
quired to achieve convergence. The differential equations for
discrete phase were solved by modified Gear’s method imple-
mented in ODEPACK (Hindmarsh, 1983). The algebraic equa-
tions for continuous phase were solved using Newton–Raphson
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 trans-
fer in the freeboard region was developed. The Kramers model
(Kramers and Croockewit, 1952) which relates volumetric flow
rate of solids, v , with kiln tilt ( , radian), angle of repose (
, In Eq. (28) the first and second terms of the right-hand side
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 fluids. Melting of solids in bed re-
gion and formation of coating within the kiln were accounted.
Combustion of coal in the freeboard region was modeled by
accounting devolatilization, finite rate gas phase combustion
and char reaction. Knowing the bed and freeboard gas temper-
atures, 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 flow with constant
grate speed. Air was assumed to enter in a cross-flow 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 finally the vent air (Locher, 2002) as shown in Fig. 3c.
The amount of secondary, tertiary and vent air (in kg/s) go-
ing 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 in-
put to the model. To get the temperature profiles 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
coefficient between air and solids was calculated from empiri-
cal correlation assuming solids as packed bed as discussed pre-
viously. 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. (27)
dx
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
coefficient between solid clinker and air in the cooler,  is
the porosity, Tg is air temperature at any point in the cooler.
represents the conductive heat transfer. The last term in right-
hand side represents convective heat transfer between the air
and solids.
The mass balance for air can be written as
dma(i,j )
= 0. (29)
dy
Energy balance for air can be written as
j(g ug,x Cp,g Tg ) j(g ug,y Cp,g Tg )
+
jx 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 final
term is due to the convection between solids and air.
In Eq. (30) hc,c is convective heat transfer coefficient be-
tween solid clinkers and air. Developing accurate models for
convective heat transfer coefficient between solids and air is
important in capturing heat transfer in the cooler. In this work
the convective heat transfer coefficient 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 commer- These values are usually known or can be easily available for
cial clinker cooler are significantly higher (Re ∼ 1000.2000) any cement plant and can therefore be used to generate good ini-
as compared to the experimental conditions of Nsofor and tial guess for faster convergence of solution. The pre-processor
Adebiyi (2001). However, as discussed earlier, there are no solves mass and energy balance equations as discussed in the
other systematic experimental studies reported to predict con- following. Based on the percentage calcination in the calciner,
vective heat transfer coefficients in clinker coolers. The empiri- the mass of CO2 produced in calciner was calculated as
cal correlation (Eq. (31)) was developed for particle sizes close
Mw CO2
to those found in industrial clinker coolers and for wide range mCO2 ,C = mCaCO3 ,i · P · , (32)
of temperature conditions as observed in clinker coolers. Fortu- Mw CaCO3
nately, the empirical correlation seems to be weekly dependent
where mCaCO3 ,i is the total amount of CaCO3 in the inlet raw
on Reynolds number (Reynolds number is to power 0.2). There-
meal. The mass flow rate of solids entering the kiln was calcu-
fore possible errors associated with Eq. (31) are not expected
lated as
to change the simulation results significantly (predicted Nus-
selt number is ∼ 50.60). Hence Eq. (31) was used to predict Ms,C = Ms,P − mCO2 ,C , (33)
gas solid heat transfer in clinker coolers in the present model.
All the physical properties for determining heat transfer coef- where Ms,C is the mass flow rate of raw meal leaving the
ficient were calculated at an average temperature of solids and calciner or entering the kiln, Ms,P is the mass flow rate of the
air as Tf = (Ts + Tg )/2. The system of algebraic linear equa- solids entering the pre-heater. The corresponding mass fraction
tions formulated for above model equations was solved using of solids species leaving the calciner or entering kiln were
tri-diagonal matrix algorithm (TDMA). calculated as
mCaCO3 ,i − mCaCO3 ,i · P
3.5. Integrated model and solution strategy xCaCO3 ,C = ,
Ms,C
The individual models for pre-heater, calciner, kiln and (mCO2 ,C ) · (Mw CaO )
cooler described in the previous section were coupled with xCaO,C = ,
(Ms,C ) · (Mw CO2 )
each other to develop a simulator for the entire system. The
schematic of the simulator is shown in Fig. 4. The required in- mSiO2 ,i mAl2 O3 ,i
xSiO2 ,C = , xAl2 O3 ,C = ,
puts to the simulator are flow rates and composition of (a) raw Ms,C Ms,C
meal entering the pre-heater, (b) air entering the cooler, (c) coal mFe2 O3 ,i
entering the calciner and the kiln, and (d) the material prop- xFe2 O3 ,C = , (34)
Ms,C
erties and operating parameters of the individual equipments
(for example, kiln RPM, grate speed of cooler). However, to where x is the mass fraction of the component in the raw meal.
solve the integrated simulator, it is necessary to know the in- The amount of clinker leaving the kiln or entering the cooler
let conditions for the calciner (flow rate, mass fractions and Ms,K was calculated as
temperature of solids and air from pre-heater, kiln and cooler),
pre-heater (flow rate and temperature of air from calciner), kiln (Ms,C ) · xCaCO3 ,C
Ms,K = Ms,C − · Mw CO2 . (35)
(flow rate, mass fractions and temperature of secondary air MW CaCO3
from cooler and partially calcined raw meal from the calciner)
Based on overall material balance on kiln, the amount of air
and cooler (flow rate and temperature of solids from kiln).
leaving the kiln was calculated as
To generate these inputs a pre-processor was developed. The
function of pre-processor was two-fold. The pre-processor was Mw CO2
used to develop good initial guess for the simulator and also to Mg,K = Mg,S + Ms,C · xCaCO3 ,C ·
Mw CaCO3
check for any inconsistency of input data. The pre-processor
generated the initial guess (for mass flow-rates, composition Mw CO2
+ Mc,K · yc,K · , (36)
and temperatures of raw meal and air) for the individual mod- Mw CaCO3
els based on overall material and energy balances. Following
where Mg,S is the mass of secondary air entering the kiln, Mg,K
parameters were provided to the pre-processor to achieve this:
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
1. Percentage calcination occurring in the calciner (P ). of char entering the kiln. The amount of air leaving the pre-
2. Temperature of secondary air (Tg,S ) and tertiary air (Tg,T ) heater assembly was calculated as
leaving the cooler.
3. Temperature of air leaving the kiln (Tg,K ). Mg,P = Mg,K + Mg,T + mCO2 ,C
4. Temperature of air exiting the pre-heater to the atmosphere Mw CO2
(Tg,P ). + Mc,C · yc,c · , (37)
Mw CaCO3
5. Temperature of solids exiting the cooler (Ts,R ).
6. Heat losses (HLoss,K ) and heat of clinkerization reaction in where Mg,P is the mass of air entering the pre-heater, Mg,T
the kiln (HR,K ). is the mass tertiary air entering the calciner, mCO2 ,C is the
 K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 – 2607 2599

User Input Consistence Checks

Dimensions, MOC, Call Preprocessor &
Mass Flow Rate, Generate initial guess
Mass Fractions, N Y
Temperature

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
(Ms,R · Cp,s · Ts,R + Mg,T · Cp,g · Tg,T + Mg,S · Cp,g · Tg,S ) − (Mg,in · Cp,g · Tg,in )
Ts,K =
(Ms,K · Cp,s )
In the above equation, Mg,in and Tg,in are the mass flow rate
and temperature of air entering the cooler and Ts,R is the tem-
perature of solids exiting the cooler. The temperature of solids
entering the kiln or exiting calciner (Ts,C ) is calculated as
the temperature of solids and gases at exit of individual com-
ponents were within error of ±1%. Suitable under-relaxation
parameters were used. Typically 10–20 iterations were required
. (38)
for solution to converge. We have also carried out several test
simulations of limiting cases to verify that implemented nu-
merical techniques and computer programs are correctly solv-
ing 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 )
Ts,C =
(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
(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 )
Ts,P =
(Ms,P · Cp,s )
In the above equation, Hc,C is the heat released due to coal com-
bustion in the calciner and Hcalc is the heat required by calci-
nation 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
. (39)
were solved by switching off the calcination and clinkeriza-
tion reactions in the calciner and kiln, respectively. For these
simulations the material and energy balances converged to an
error of ±1% giving verification that numerical calculations
are correctly solving the model equations. It was also verified
that the converged solution is not a function of initial guess or
. (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, cal-
ciner, 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 Table 3

typical clinker manufacturing configuration was selected as a (a) The physical properties of solids and air, (b) particle size and composition
of coal and (c) particle size and composition of raw meal
base case. Some assumptions were made to fill in the gaps in the
available data. The details of selected configuration are given in Description Air Raw meal Coal
Tables 2a–c. Though the developed mathematical framework is
(a)
general enough to accommodate temperature dependent phys- Thermal conductivity, W/m K 0.116 0.5 0.5
ical properties like heat capacity, at this stage, these properties Emmisivity 0.4 0.9 0.8
were treated as constants. The physical properties of solids and Heat capacity, J/kg K 1000 1000 1000
air used in this work are specified in Table 3a. Our prior simu- Viscoscity, kg/m s 1e − 05 – –
Density, kg/m3 1.3 1500 1000
lations of kiln and calciner (Mujumdar et al., 2006; Warudkar
Char calorific value, kcal/kg – – 5600
et al., 2005) indicated that the errors in overall energy Volatile calorific value, kcal/kg – – 11 900
consumption associated with the assumption of temperature
(b)
independent values of specific heat were within 1%. The oper-
Coal particle size, 50 m
ation of the base case (described in Tables 2–4) was computa- Volatile (CH4 )a , 27%
tionally studied to understand the various processes occurring Char, 58%
in individual units in clinker formation. On obtaining satisfac- Ash, 15%
tory results from the base case, several numerical experiments (c)
were performed using the model for understanding interac- Raw meal particle size, 50 m
tions among different processes and for possible optimization CaCO3 , 80%
of clinker manufacturing process. CaO, 0%
SiO2 , 14%
Fe2 O3 , 3%
4.1. Base case simulation AL2 O3 , 3%
a Mujumdar et al. (2006).
The predicted results from the simulation of the base case
are summarized in Table 4. The mass fractions and tempera-
tures of solids and air in pre-heaters, calciner, kiln and cooler
obtained from the simulation are plotted in Figs. 5 and 6, re-
pre-heater assembly. Abscissas 4 and 5 denote the calciner in
spectively. It is important to note that the flow of air is counter
the system. Abscissas 5–15 denote the rotary kiln and 15–18
current with respect to the flow of solids in the system. The ab-
denote the cooler section. Fig. 5 shows a plot of mass frac-
scissas of Figs. 5 and 6 denote particular equipment in clinker
tions in pre-heater, calciner, kiln and cooler (only CaO, C2 S
formation as discussed below. Abscissas 1–4 corresponds to
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. How-
ever, in the pre-heater section the raw meal gets heated from
Table 2 300 to 1069 K and hot gases from calciner get cooled (from
The dimensions of (a) pre-heater unit, (b) kiln and (c) cooler 1224 to 539 K) as can be seen from Fig. 6. As the raw meal
S/No. Description Units Values passes through the calciner, it gets partially calcined. There-
fore, CaO concentration increases in the calciner section as
(a) can be seen from Fig. 5. Similarly since CO2 is formed due
1 No. of pre-heaters 4
2 Height of cylindrical section m 5
to calcination and coal combustion, the mass fraction of CO2
3 Height of conical section m 3 increases in the calciner. Coal combustion in the calciner ac-
4 Diameter of cyclone m 3 counts for rise in temperature of both solids and gas in the
5 Diameter of cone tip m 1 calciner (see Fig. 6). Remaining clinkerization reactions occur
6 Refractory thickness m 0.13 in kiln. The mass fraction and temperature profiles obtained in
7 Shell thickness m 0.03
8 Inlet duct height m 1
kiln (as shown in Figs. 5 and 6) are similar to previously pub-
9 Inlet duct width m 1 lished results (Mujumdar and Ranade, 2006; Mastorakos et al.,
10 Diameter of outlet pipe m 1 1999). Since there is no reaction occurring in the cooler, mass
(b)
fraction of solids in the clinker cooler do not vary. However,
1 Length m 50 air entering the cooler gets pre-heated (from 300 to 1200 K)
2 Inner diameter m 3.4 and solids get cooled (from 1632 to 476 K) in the cooler sec-
3 Coating thickness m 0.136 tion. The predicted energy requirements of individual processes
4 Refractory thickness m 0.2 like clinkerization reactions, losses, melting predicted by the
5 Shell thickness m 0.025
model are listed in Table 4. The obtained results are qualita-
(c) tively similar to previously published results (Engin and Ari,
1 Length m 11 2005). The performance of the overall system was characterized
2 Width m 1
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 Solid inlet temperature, K 300 1069.2 1214.8 1622.4

2 Mass flow rate, kg/s 50 50 37.74 32.3
3 Air inlet temperature, K 1214.8 1114.5 1229.9 300
4 Air flow rate, kg/s 60.8 46.7 16.2 45
5 Coal flow rate, kg/s – 2.15 0.9 –
6 Coal inlet temperature, K – 350 350 –
7 Heat with solids in, kJ/kg clinker 463.0 1650.2 1415.3 1622.4
8 Heat with air in, kJ/kg clinker 2297.9 1603.6 615.0 416.7
9 Heat with coal in, kJ/kg clinker – 23.2 9.7 –
10 Combustion of coal, kJ/kg clinker – 1876.7 747.1 –
11 Heat of reaction, kJ/kg clinker – 1384.5 219.0 –
12 Heat of melting, kJ/kg clinker – 44.2 – –
13 Heat of solids leaving, kJ/kg clinker 1650.2 1415.3 1622.4 463.0
14 Heat of air leaving, kJ/kg clinker 1014.4 2297.9 1603.6 1415.3
15 Heat of vent air in cooler, kJ/kg clinker – – – 109.4
16 Heat with ash, kJ/kg clinker – 3.22 2.24 –
17 Loses, kJ/kg clinker 98.9 43.5 140.7 –

0.7 reasonable when compared with industrial observations. Over-

CaO -Solid mass fraction all the integrated simulator was able to predict the clinker man-
0.6 C2S-Solid mass fraction ufacturing process in cement industry reasonably well.
CO2-Gas mass fraction

0.5 4.2. Influence of key design and operating parameters on NEC

Mass fraction

0.4 On obtaining a reasonable agreement, the model was used to

explore space of design and operating parameters to understand
0.3 influence of these parameters on the performance of clinker
manufacturing. All these simulations were carried for a fixed
0.2 product composition (C3S mass fraction 0.48 in the product).
Calciner Cooler This was achieved by altering coal flow rate either to calciner
0.1 Pre-Heaters or kiln. This analysis is presented in the section below.
Kiln

0 4.2.1. Effect of number of pre-heaters

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 ).
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 flow 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
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 composi-
tion 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 pre-
heaters in pre-heater assembly increases, solids get pre-heated
to a higher temperature before they enter the calciner (see sec-
ondary axis in Fig. 7). Therefore the coal requirement for a
fixed product composition decreases. Thus the NEC decreases
(41) as number of pre-heaters increases. However, the overall cap-
ital cost increases by increasing number of pre-heaters in the
system. The developed model will be useful to carry out cost
to benefit 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 per-
centage calcination in calciner on NEC. To vary the percent-
age calcination in calciner the coal feed rate to the calciner
and kiln was altered for the same clinker composition. The
2602 K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 – 2607

2000 Solid temperature 2000

Gas temperature

Solid temperature, K
1500 Gases entering 3rd preheater 1500
Gas temperature, K

1000 1000

Solids leaving 3rd preheater

500 Calciner Cooler 500
Pre-heaters Kiln

0 0
0 1 2 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 profile across pre-heaters, calciner, kiln and cooler in a cement clinker process.

645 1100 666 1500

Solid temperature entering calciner, K
Energy consumption, kcal/kg clinker

Energy Consumption, kcal/kg clinker

661
640
1080

Kiln exit gas Temperature, K

1400
656
635
1060 651
1300
630
646
1040
625 1200
641

1020 636
620
1100
631
615 1000
2 3 4 5 6 626 1000
Number of pre-heaters 40 50 60 70 80 90 100
% Calcination
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 Table 5
then it increases with further increase in percentage of calcina- Heat of reaction in calciner and kiln
tion. The secondary axis of Fig. 8 shows that the kiln exit gas S/No. Heat of reaction
temperature also shows a similar trend. Table 5 shows a com-
1 Calcination, % 50 60 70 80 90
plete comparison of heat of reaction occurring in kiln and cal-
2 Heat of reaction 1038.5 1176.6 1384 1618.5 1802.5
ciner in this process. The heat of reaction in kiln decreases as in calciner, kJ/kg
the percentage calcination increases in calciner. The total heat clinker
of reaction in kiln is the summation of heat of calcination (en- 3 Heat of reaction in 582.9 433.2 219 −20.2 −178.4
dothermic reaction) and the heat of clinker formation (exother- kiln, kJ/kg clinker
mic 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 flue gas
temperature and increase in losses from kiln shell. Therefore RoCKS software were thus able to provide valuable clues for
the NEC and kiln flue gas temperature increases if more than determining the optimum percentage calcinations desired for
70% calcination occurs in the calciner. The model and the minimizing NEC.
 K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 – 2607 2603

650 1350

Energy consumption, kcal/ kg of clinker

640 2500
645 1300

Secondary air temperature, K

Energy Consumption, kcal/kg clinker

Residence time of solids in kiln, s

635
2000 640 1250

630 635 1200

1500
630 1150
625

625 1100
1000
620
620 1050
0.06 0.08 0.1 0.12 0.14 0.16 0.18
615 500 Grate speed, m/s
2 3 4 5 6 7 8 9
Kiln Rpm Fig. 10. Effect of cooler grate speed on overall energy consumption.

640 2000 seen from Fig. 10 that the NEC increases with increasing grate
speed. The increase in grate speed reduces residence time of
Energy consumption, kcal/ kg clinker

635
solids in the cooler. This results in less convective heat transfer
Residence time of solids, s

1600 between solids and air as clearly indicated by temperature of

secondary air plotted in Fig. 10. Therefore the simulation results
630 indicate that it is better to operate grates in the cooler at lower
1200 speed. The simulations presented here provide useful trends of
energy consumption as a function of key operating parameters
625
in cement clinker process. This result also gives us a scope to
understand the importance of design parameters (kiln tilt) on
800
620 plant performance and can be very useful to plant engineers.

4.2.4. Effect of solid loading

615
0 0.5 1 1.5 2 2.5
Kiln tilt, Degree
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 flow 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 beneficial 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
influence of varying grate speed on overall energy consumption.
The results for these simulations are shown in Fig. 10. It can be
400 The predicted results in the form of NEC and corresponding
3 3.5
overall losses for different solids flow rates are shown in Fig. 11.
It can be seen that the NEC per unit weight of product decreases
as solids flow rate increases. This is because the net energy loss
from the entire system decreases as the solid flow rate increases
(see Fig. 11). Thus, it is beneficial from the point of view
of energy consumption to operate the units with higher solids
flow rate. Other operational concerns like increase in dusting
and mixing, however, need to be considered while identifying
maximum solids flow rate specifically 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 significantly with
changing coal composition (ash content 9%, 15% and 40%).
For these simulations the coal flow rate to the kiln was ad-
justed so that the same amount of energy is supplied to the
kiln. Therefore the insignificant change in overall energy con-
sumption does not seem to be surprising. However, as the
coal composition changes, the flame characteristics in the kiln
vary. The predicted dimensionless flame length by the simu-
lator for varying coal composition is shown in Fig. 12. The
flame length was calculated by tracking the region in freeboard
where char and volatiles composition in coal go to zero. The
2604 K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 – 2607

650 350 600 1120

Total losses in clinker process, kJ/kg clinker

Energy comsumption, kcal/kg of clinker

Energy consumption, kcal/kg clinker

1115
640 330 598

Kiln flue gas temperature, K

1110
630 310 596

1105
620 290
594
1100
610 270
592
1095

600 250
44 45 46 47 48 49 50 51 590 1090
Raw meal flow rate, kg/s 15 20 25 30 35 40 45
Air flow rate through secondary shell, kg/s
Fig. 11. Effect of raw meal flow rate on overall energy consumption.
Fig. 13. Effect of secondary shell, on overall energy consumption.

650 0.56 emissivity and thermal conductivity (Engin and Ari, 2005).
Energy comsumption, kcal/kg of clinker

However, merely covering kiln shell with metallic shell and

insulating it can lead to enormously high shell temperatures.
Flame length, dimensionless

640
630
620
610
600
30 40 50
Char Percentage, %
Fig. 12. Effect of coal composition on overall energy consumption.
dimensionless flame length was calculated as the ratio of pre-
dicted flame length to length of the kiln. It can be seen that
coal with higher ash content tends to have a longer flame as
compared to coal with lower ash content. The flame 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 flames. Such simulations can
therefore provide useful information to kiln operators to pre-
dict the flame 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 radia-
tion and convection are a significant 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
0.52 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 realis-
0.48 tic 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
0.44 of clinker to 1.4 kJ/kg of clinker on applying a secondary
shell and insulation of dimensions and operating conditions
specified in Mujumdar et al. (2006). The NEC reduces from
0.4 ∼ 2635 kJ/kg clinker to 2493 kJ/kg clinker (i.e., 630 kcal/kg
60 70 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 fly 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 flow. The inte-
grated 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 unfortu-
nately not possible since adequate industrial data could not be
obtained. However, the model predictions agreed reasonably
with industrial observations. RoCKS was used to understand
influence of various design and operating parameters on overall
 K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 – 2607 2605

performance. Specific conclusions based on this computational hcyc heat transfer coefficient between particle laden
study are: gas and cyclone inner wall, W/m2 K
Hc,C heat of coal combustion in calciner, J/kg
• Including an additional pre-heater reduces NEC. The devel- Hc,K heat of coal combustion in kiln, J/kg
oped model can be used to evaluate relative benefits of en- Hcalc heat of calcination reaction in calciner, J/kg
ergy savings by additional pre-heater and required additional Hcomb heat of char combustion, J/kg
capital expenses. Hcombg heat of volatile gas phase combustion, J/kg
• There is an optimum value for percentage of calcination car- HLoss,K heat losses in the kiln, J/kg
ried out in calciner with respect to overall energy consump- HR,K heat required for clinker reactions, J/kg
tion in clinker manufacture. With the parameters selected in kg thermal conductivity of air/gas, W/m K
this work, this optimum value of percentage calcination in kr thermal conductivity of refractory, W/m K
calciner is about 70. ks thermal conductivity of clinker, W/m K
• The simulation results indicated that operating kiln with ksh thermal conductivity of shell, W/m K
higher solid loading, lower rpm, lower tilt and lower grate ks,c rate constant of char combustion, kg/m2 s kPa
rk rate constant of calcination of calcium carbon-
speed reduces energy consumption per unit production. The s
upper limit on solid loading (bed height) and lower limits on ate, mol/m2 s−1
rpm and tilt (mixing and heat transfer) need to be identified r k rate constant of calcination of calcium carbon-
s
based on other practical issues. ate, mol/m2 s−1
• The use of secondary shell appears to be a promising method L total height of cyclone, m
to reduce overall energy consumption, if the hot air generates ma mass of air in cooler, kg/s
in such secondary shell (∼ 200 ◦ C) can be utilized in some mAl2 O3 ,i mass of total aluminum oxide in solids in cal-
other processes in cement plants. ciner, kg/s
mCO2 ,C mass of carbon-dioxide produced in calciner due
The model was also able to predict kiln characteristics like to calcination, kg/s
maximum flame temperature and overall flame length for coals mCaCO3 ,i mass of total calcium carbonate in solids in cal-
with different compositions. The models and results presented ciner, kg/s
here will help in developing a better understanding of clinker mFe2 O3 ,i mass of total ferrous oxide in solids in cal-
manufacturing process and may provide clues for possible ciner, kg/s
optimization. mSiO2 ,i mass of total silicon dioxide in solids in cal-
ciner, kg/s
mg mass of gas in calciner, kg
mgin mass of air entering in calciner,kg/s
Notation
mgout mass of air leaving calciner, kg/s
mg,K mass of air leaving the kiln calciner, kg
a surface area per unit volume, m2 /m3 mp,c mass of coal particle, kg
A0 devolatilization constant mp,cin mass of coal particle entering calciner, kg
Acyi internal surface area of cyclone, K mp,cout mass of coal particle leaving calciner, kg
Acyo external surface area of cyclone, K mpc,0 initial mass of coal particle, kg/s
Ap surface area of coal particle, m2 mp,L mass of solid particle, kg
LA
p surface area of solid particle, m2 mp,Lin mass of solids entering calciner, kg
Cp,c specific heat capacity of coal particle, J/kg K mp,Lout mass of solids leaving calciner, kg
Cp,g specific heat capacity of air, J/kg K ms mass of solids/clinker in cooler, kg
Cp,s specific heat capacity of solids, J/kg K Mc,C mass of coal entering the calciner, kg/s
dc inner diameter of cyclone, m Mc,K mass of coal entering the kiln, kg/s
dp radius of particle, m Mg mass of gas in cyclones, kg/s
E1 energy of activation for char combustion, J/mol Mg,K mass flow rate of secondary air entering the
E2 energy of activation for calcination, J/mol kiln, kg/s
fc fraction of heat given to coal particle released Mg,P mass flow rate of gas entering the pre-heater,
due to coal combustion kg/s
fv,0 initial mass fraction of volatiles in coal particle Mg,S mass flow rate of secondary air entering the
Fp−w view factor kiln, kg/s
hc heat transfer coefficient between coal particle Mg,T mass flow rate of tertiary air entering the cal-
and gas, W/m2 K ciner, kg/s
hc,c heat transfer coefficient between clinker and gas, Ms mass of solids in cyclones, kg/s
W/m2 K Mse mass of solids entrained by gas in cyclones, kg/s
hc,L heat transfer coefficient between solid particle Ms,C mass flow rate of solids leaving the calciner, kg/s
and gas, W/m2 K Ms,K mass flow rate of clinker leaving the kiln, kg/s
2606 K.S. Mujumdar et al. / Chemical Engineering Science 62 (2007) 2590 – 2607

Ms,P mass flow rate of solids entering the pre-heater, ug,y air velocity in y direction, m/s
kg/s Vreact volume of reactor, m3
MwCaCO3 molecular weight of calcium carbonate, kg/kmol xAl2 O3 ,C mass fraction of aluminum oxide entering kiln
MwCaO molecular weight of calcium oxide, kg/kmol xCaCO3 ,C mass fraction of calcium carbonate entering kiln
Mwchar molecular weight of carbon, kg/kmol xCaO,C mass fraction of calcium oxide entering kiln
MwCO2 molecular weight of carbon-dioxide, kg/kmol xFe2 O3 ,C mass fraction of ferrous oxide entering kiln
MwO2 molecular weight of oxygen, kg/kmol xSiO2 ,C mass fraction of silicon dioxide entering kiln
Mwvol molecular weight of volatile, kg/kmol yc,c mass fraction of char in coal particle in calciner
Mww molecular weight of water, kg/kmol yc,cin mass fraction of char entering in coal particle
Nu Nusselt number yc,cout mass fraction of char leaving in coal particle
cN number of coal particles entering calciner per
p yc,K mass fraction of char in coal particle entering
second the calciner
LN number of solid particles entering calciner per yv,c mass fraction of volatiles in coal particle
p
second yO2 mass fraction of oxygen in gas
pO2 partial pressure of oxygen in gas, kPa yO2 ,in mass fraction of oxygen entering calciner in gas
pCO2 partial pressure of carbon-dioxide in gas, kPa yO2 ,out mass fraction of oxygen leaving calciner in gas
peq equilibrium partial pressure for carbon-dioxide yCO2 mass fraction of carbon-dioxide in gas
in gas, kPa yCO2 ,in mass fraction of carbon-dioxide entering cal-
P the percentage calcination occurring inside the ciner in gas
calciner yCO2 ,out mass fraction of carbon-dioxide leaving calciner
P pressure drop across the cyclone, mm of H2 O in gas
Pr Prandtl number yv mass fraction of volatiles in gas
rc rate of calcination, kg/s yw mass fraction of water in gas
rcomb rate of combustion of char particles, kg/s Z stoichiometric component
rcombg rate of combustion of volatiles, kg/s
ri internal diameter of cyclone, m
Greek letters
r0 external diameter of cyclone, m
rp radius of solid particle, m
rr internal diameter of cyclone shell, m  porosity of clinker bed in cooler
R gas constant c emissivity of coal particle
Rc non-dimensional form of radiative heat transfer cy emissivity of cyclone outer wall
coefficient L emissivity of solid particle
Re Reynolds number m,p mass efficiency of the cyclone
T0 ambient air temperature, K  residence time of coal particle in calciner, s
Tc,i temperature of solids and gas in cyclone, K g density of air/gas, kg/m3
Tcl temperature of coal particle, K s density of solids, kg/m3
Tf average temperature of solids and air in  Stephan–Boltzmann constant (W/m2 K 4 )
cooler, K  residence time of raw meal particle in calciner, s
Tg temperature of gas, K
Tg,in temperature of gas entering calciner, K
Tg,out temperature of gas exiting calciner, K Chemical species
Tg,K temperature of gas leaving the kiln, K
Tg,S temperature of secondary air, K C2 S (2CaO·SiO2 )
Tg,T temperature of tertiary air, K C3 S (3CaO·SiO2 )
Tg,P temperature of gas leavingthe pre-heater, K C3A (3CaO·Al2 O3 )
Tiw,i the internal wall temperature of the cyclone, K C4AF (4CaO·Al2O3 ·Fe2 O3 )
Tow,i the external wall temperature of the cyclone, K
TL temperature of solid particle in calciner, K
Tr,i the temperature of interface of refractory and Acknowledgments
shell in cyclone, K
Ts temperature of solids/clinker in cooler, K The authors wish to acknowledge financial support provided
Ts,C temperature of solids entering the kiln, K by CSIR (under the NMITLI scheme) for this study. The au-
Ts,R temperature of solids exiting the cooler, K thors would also like to acknowledge many helpful discussions
u0 inlet gas velocity in cyclone, m/s with Professor Anurag Mehra during the course of this work.
us,x grate speed in x direction, m/s One of the authors, K.S.M is grateful to Council of Scientific
us,y grate speed in y direction, m/s and Industrial Research (CSIR), India for providing financial
ug,x air velocity in x direction, m/s 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 systems—a case study. Energy Conversion and Management
46 (4), 551–562.
Gupta, A.V.S.S.K.S., Nag, P.K., 2000. Prediction of heat transfer coefficient in
the cyclone separator of a CFB. International Journal of Energy Research
24, 1064–1079.
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.), Scientific Computing. IMACS
Transactions on Scientific Computation, vol. 1. North-Holland, Amsterdam,
pp. 55–64.
Hu, N., Scaroni, A.W., 1996. Calcination of pulverised limestone particles
under furnace injection conditions. Fuel 75, 177–186.
Kang, S.K., Kwon, T.W., Kim, S.D., 1989. Hydrodynamic characteristics of
cyclone reactors. Powder Technology 58, 211–220.
Kramers, H., Croockewit, P., 1952. The passage of granular solids through
inclined rotarykilns. Chemical Engineering Science 1, 259–265.
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, 215–222.
Li, J.Q., Wei, F., Jin, Y., 2003. Numerical simulation of pulverized
coal combustion and NO formation. Chemical Engineering Science 58,
5161–5171.
Locher, G., 2002. Mathematical models for the cement clinker burning
process, part 4: grate cooler. ZKG International 55 (6), 46–57.
2607
Lu, J., Huang, L., Hu, Z., Wang, S., 2004. Simulation of gas–solid, two-phase
flow, coal combustion and raw meal calcination in a pre-calciner. ZKG
International 57 (2), 55–63.
Mastorakos, E., Massias, A., Tsakiroglou, C.D., Goussis, D.A., Burganos,
V.N., 1999. CFD predictions for cement kiln including flame modeling, heat
transfer and clinker chemistry. Applied Mathematical Modeling 23, 55–76.
Mujumdar, K.S., Ranade, V.V., 2003. Coupling of CFD models with different
time scales—a 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), 165–177.
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), 2315–2330.
Nsofor, E.C., Adebiyi, G.A., 2001. Measurements of the gas–particle
convective heat transfer coefficient in a packed bed for high-temperature
energy storage. Experimental Thermal and Fluid Sciences 24 (1–2), 1–9.
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,
309–323.
Stanmore, B.R., Gilot, P., 2005. Review–Calcination and carbonation of
limestone during thermal cycling for CO2 sequestration. Fuel Processing
Technology 86, 1707–1743.
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, 369–378.