Automatica, Vol. 8, pp. 309-323. Pergamon Press, 1972. Printed in Great Britain.

A Dynamic Model of a Cement Kiln*

Un module dynamique d'un four h ciment
Ein dynamisches Modell eines Zementofens
~[HHaMHqecKa~ MoRe.rib ueMeHTHO~ neqn

H. A. SPANG, I I I t

A cement kiln is a highly non-linear distributedprocess. A partial-differential equation

model incorporating an improved model of the flame gives a dynamic response which
appears to match the qualitative behavior of an actual kiln.

Sununary--A cement kiln is a counter-flow type of process
in which a complex series of exothermic and endothermic
reactions take place. To understand its behavior and to
provide a means of improving computer control, a dynamic
partial differential model of the kiln has been developed.
In this paper, the basic reactions and assumptions of the
model will be discussed. A model of a coal]oil flame which
takes into account the surrounding temperatures is developed.
The dynamic behavior of the kiln as exhibited by this model
is then indicated. In particular, it is shown that the inter-
action between the two exothermic reactions causes the kiln
to be basically unstable with respect to regulating the
position of the burning zone.
NOTATION
In order to simplify the writing, we use the follow-
ing standard symbols for the cement components:
C=CaO, A=A12Oa, S=SiO2, F=Fe203,
= CaCO a and ~b= CO2. Thus 2 C a O . SiO2 = C2S,
4 C a O . A1203 • F e 2 O a = C 4 A F , etc. A list of all
symbols is given in Appendix B at the end of the
paper.
All components are normalized with respect to
the pounds of CaO in the burning zone. We use, in
the material balance equations, the following
abbreviations: A = # A 1 2 0 3 / # CaO, F = # Fe203/
#CaO, S= #SiO2/#CaO, C= #CaO/#CaO,
0~= # CaS / 7~CaO, fl = ~ C2S / ~ CaO, ~ = # CaA /
# CaO, 6 = # C 4 A F / # CaO, 09= # n20 / ~ CaO,
= # C O 2 / # CaO and ~k= # C a C O 3 / # CaO.
INTRODUCTION
A CEMENT kiln is a distributed parameter system
which has a highly complex dynamic behavior due
to the chemical reactions. The chemistry of making
cement has been studied in some detail [1, 2]. The
steady state temperature behavior has also been
* Received 19 April 1971; revised 29 November 1971.
The original version of this paper was accepted for presenta-
tion at the IFAC Symposium on The Control of Distributed
Parameter Systems which was held in Banff, Canada during
June 1971. It was recommended for publication in revised
form by Associate Editor H. Kwakernaak.
t GE Research and Development Center, Schenectady,
N.Y. 12309.
studied by GILBERT [3] and LYONS et al. [4]. A
transient solution is indicated by LYONS et al. [5],
but a limited amount of response data is presented.
In this paper, a dynamic model of a kiln will be
developed. In order to reduce the computational
problems, the equations are averaged with respect
to the radius of the kiln. The resulting equations
are similar to those of LYONS et al. [5] and the
author has used many of their values for the co-
efficients of the model. An important addition is a
dynamic model of the flame. Instead of Lyon's
model of the flame which assumes a rate of reaction
dependent only on flame length and not the
surrounding temperature, a more detailed model is
developed taking into account the effect of tem-
perature on the rate of reaction. The interaction
between the flame and the solid causes an unstable
behavior with respect to burning zone position
which is not evident with non-interactive flame
models. The response of the model to gas velocity
and kiln speed is examined. It is further shown
that, like the actual process, cement can be made
for a period of time after the flame has been turned
off. In general, the behavior of the model appears
reasonable, especially in the burning zone/flame
dynamics. However, due to a lack of adequate
sensors, quantitative confirmation has not been
possible.
QUALITATIVE DESCRIPTION OF
A CEMENT KILN
A highly simplified diagram of a cement kiln is
shown in Fig. 1. It basically consists of a drum
approximately 400-600 ft long and 10-20 ft in
diameter. The length of the kiln mainly depends on
how long it takes to heat up the raw materials to the
"clinkering" temperature, which is approximately
2500°F. In some plants, a cyclone preheater is used
to raise the initial temperature of the solids in which
case the length of the kiln is greatly shortened.

309
310 H . A . SPANG,III

SOLIDS

Z:O Z=

FIG. 1. Simplified geometry of a cement kiln.

The raw materials fed into the kiln contain which determine the behavior of the kiln: the pre-
calcium carbonate ( C a C O 3 ) , silica (SiO2), shale heat zone, the calcining zone, and the burning zone.
(A1203), and iron ore (Fe203). These are ground The position of the boundaries of the zones depends
to a very fine powder and mixed according to the on the temperature and chemical reactions taking
type of cement being made. Upon heating by the place in the solid. Hence, they shift in time and
hot gases, various reactions occur, as shown in length depending on the amount of thermal energy
Table 1. A typical temperature curve for the kiln obtained from the gas and wall.
is shown in Fig. 2. There are three important zones In the preheat zone, the solid material is heated
to the point where the calcining reactions can
begin. The initial rise in temperature is retarded by
TABLE 1
evaporative cooling as water is removed.
At 1620°F (2000°R) the highly endothermic cal-
Temp. (°F) Reaction Heat change
cining zone defined by the reaction of CaCO3 con-
verting into CaO begins. This is entirely a heat
212° Evaporation of free water Endothermic
1620° and Evolution of carbon dioxide Endothermic limited reaction with the solid temperature increas-
above from calcium carbonate ing very slowly. The speed of the reaction is
CaCO3-~ C.aO+ CO2 entirely determined by the amount of heat that can
2200° Start of combination of lime Exothermic
and silica be supplied by the gas and wall.
2CaO+SiO2~C2S The burning zone occurs when the calcining is
2300° Liquid formation Endothermic completed. The solid temperature rises about
4CaO +A1203 +Fe2Oa--* C4AF
3CaO +A1203-~C3A
1500°R due to the exothermic formation of C2S.
above Completion of cement Three slightly endothermic reactions also take
2300° compounds place. The first two are the formation of a molten
2CaO +SiO2-~C2S Exothermic
C2S +CaO-oC3S Endothermic
liquid consisting of C4AF and C3A. Thi~ liquid
causes the mix to cohere into small balls known as

3000

_9
o
~20~

10(~

16o B 260 38o

DISTANCE ALONGLENGTHOF KiLN (f))
FIG. 2. A typical temperature profile.
 A dynamic model of a cement kiln
"clinkers". Although the required reactions could
not occur without this liquid, its effect on the
dynamics of the kiln is felt to be second order and
was only included in the model implicitly by using
lower values of activation energy in the C2 S, C3 S
reactions. The third endothermic reaction occurs
when the C2S combines with the remaining CaO to
form CaS. This reaction has minimal effect on the
temperature profile. The position in the kiln of the
burning zone appears to depend almost entirely on
the length of the preheat and calcining region. In
all of the simulation runs, the calcining reaction is
completed before the C2S reaction beings.
SUMMARY OF ASSUMPTIONS
In developing the model, every effort is made to
make it as simple as possible while maintaining the
essential dynamics that have been observed qualita-
tively. Some of these assumptions could be relaxed
without increasing the computational complexity.
However, until more detailed and accurate measure-
ments are made, one has little justification for
changing the model.
The assumptions which are made are as follows:
(1) The inside and outside diameters of the kiln
are constant.
(2) The solids transported in the gas are not
explicitly included in the model.
(3) The clinking rate expressions do not take into
account the effect of trace compound concen-
trations.
(4) Reaction rates are determined by Arrhenius'
law.
(5) Specific heats, latent heats, and heats of
reactions are independent of temperature and
position.
(6) Coefficients of convection and emissivity are
independent of temperature and position.
(7) Conduction in the solid and gas is negligible.
The conduction in the axial direction of the
wall is also considered negligible.
(8) The velocity and mass of the solid and gas is
constant throughout the length of the kiln.
(9) The heat transfer due to the release of water
vapor and CO2 into the gas can be treated as
convective heat transfer.
THE MODEL
In this section, due to space limitations, only the
general form and assumptions will be indicated for
the thermodynamic and material balance equations
used in the model. The specific equations are given
311
in Appendix A. The thermodynamic equations for
the temperature of the solid, gas and kiln wall have
been simplified by averaging out any variation in
the angular and radial directions. Thus, these tem-
peratures will depend only on the position along the
length of the kiln and time. The resulting tempera-
tures will be fairly representative of the actual tem-
peratures in the kiln, especially for the solid and the
gas since these are well mixed. It is a somewhat
cruder approximation for the wall temperature due
to the inherent thermal mass.
The equation for heat conduction for any cylinder
of cross-section, F, is
O2T, ~2T'
kVZT'=k[ l f~l'rOT~' +-~r
Or\ ] -~1 ~02 +-~z2 1
=
Cv P -~T' _
-~- + ~ O)
where k is the thermal conductivity of the cylinder,
C~ is the specific heat, p is the density, and Q is the
net amount of heat generated within and/or moving
into a unit volume other than by conduction. We
define the average temperature per unit area as
T=lffrdrdOrT' (2)
where A is the cross-sectional area,
A =ffr rdrdO.
Multiplying equation (1) by r, integrating, and using
the divergence theorem, we obtain
t~2T ~ 8T t3T
kA.-£2-f_2 + k I ~ @ = CvpA-- + AQ (3)
Oz Jr, On Ot
where the second term on the left-hand side is the
derivative of the temperature with respect to the
outward normal integrated along the boundary of
the cross section, 1-"t. This term represents the heat
transfer across the boundary due to convection and
radiation whose effects will be assumed to be addi-
tive. It will be assumed that the heat transfer by
convection will obey Newton's law of cooling which
states that the heat transfer is proportional to the
linear diffelence in temperature. We shall further
assume that radiation is governed by the fourth
power law of black-body radiation.
Under these assumptions, the heat transferred
across the boundary for the solid, for example, can
be written in the quasi-linear form:
fl2(Ta- T~) + fl3(Tw- T~) (4)
 312
where
/,'x
f12 = 2 r l sln(2)[f2 + 1"73 × 10- 9eges(7"2
2 in the physical properties which cause the transit
+ Ts )(T o+ 7 w)]
f13 = rx(21r - P)[.[3 + 1.73 x 10- 9hewe~(T2,
+ r~)(vw + rs)].
h = i + 2ho sin(p/2)
2z~-p
and p is defined by Fig. 1. The other symbols are
given in Appendix B. The terms multiplying the
square brackets represent the length of the boun-
dary between the solid and the gas and between the
solid and the wall. The first term, f, is Newton's
proportionality constant and the remaining terms
account for the heat transfer due to radiation. The
solid receives heat by radiation not only from the
area of contact but also some through the gas from
the rest of the inner wail. The parameter ho is the
fraction of radiation not absorbed by the gas and h
is the effective increase in the radiational boundary.
The heat transfer into the gas and wall are handled
in a similar manner.
The heat loss between the wall and the outside air
must be handled slightly differently since the outside
wall temperature will be at an intermediate value
between the inside wall and the ambient air tem-
perature. This outside temperature is determined
by the conduction through the wall, and the radia-
tion and convection heat losses. Using the approxi-
mation, given in Ref. [6], that the convection loss is
proportional to the 5/4 power of the difference
between the outside wall temperature and the
ambient, the total heat loss per unit area is:
q= k (Tw- To)= 1.73 × lO-%w,[T~-T~]
F2 --/"1
+0"27(To-T,) s/4 •
where ew, is emissivity of the outside wall. The
first term is the conduction through the wall which
must be equal to the sum of the radiation and con-
vection losses. Solving for T o for various inside
wall and ambient temperatures, it can be shown
that a reasonable approximation is:
2~r2]4(T,- Tw)
where 1~. varies only from 0.6 to 0.75 which is a 25
per cent increase for a 5 fold increase in Tw from
600 ° to 3500°R.
During the heating of the solid, water vapour and
CO2 is released into the gas stream. From strictly
material balance considerations, the mass of the
H . A . SPANG,III
solid should be decreased to account for this loss.
However, there are indications that the total mass
of the solid may actually increase especially in the
burning zone. The reason for this is due to changes
time across a unit volume to vary quite widely [7]
depending on temperature and chemical composi-
tion. The modeling of this effect is quite complex
and could not be included in this study. Therefore,
the greatly simplifying assumption is made that the
material flow in both the solid and the gas is con-
stant independent of material balance considera-
tions.
The heat transfer between the solid and the gas
due to this material release can be shown to be
proportional to the difference in the solid and gas
temperatures. It can therefore be modeled by the
convection term of the previous heat transfer equa-
tions. The gas and solid both are well mixed with
the movement of heat into a region primarily
caused by the flow of the solid and gas and only
slightly by conduction. This is equivalent to:
~T
k~"T=O2oz and Q=Cppv-~z- q
where q is the heat generated by the chemical reac-
tions in the solid and by the flame in the gas. Since
the gas will reach steady state very rapidly, as com-
pared with the solid and wall, we shall assume that
OTg/~t = 0 .
Due to the thermal constants of the brick of the
kiln wall, very little conduction occurs in the z
direction which again means that
k~2Z____~w= 0.
~z z
(5)
Applying the above assumptions to equation (1)
yields the three thermodynamic equations shown in
Appendix A. It should be observed that for the co-
ordinates shown in Fig. 1, the velocity of the solid,
v~, is negative.
The solid to solid and liquid to solid reactions
which take place in the kiln are very difficult to
describe analytically because they depend upon
(6) many factors such as the amount of surface pro-
vided by the grains, the conditions of contact, the
phase of the particles, and the temperature. A
reasonable approximation to the rate constant for
the reactions can be expressed by the normal gas to
gas reaction equation:
ki = Ai exp( - Ei/R Ts) (7)
 A dynamic model of a cement kiln
where A i is a constant dependent on the texture of
the material, but relatively independent of tempera-
ture; R is the gas constant; Ts is the temperature
of the solid, Ei is the activation energy which
represents the energy necessary to bring a particle
to the state of reactivity. We shall assume that it is
independent of temperature.
The change with time in the concentration of any
compound in a given cross-section must be equal
to the amount of that compound moving into the
region plus the amount generated in that region.
Thus, the material balance equations are of the
form:
- vs +R(K)
dt oz
where K is' the concentration of a given compound
and R(K) is the amount generated. The detailed
equations are given in Appendix A. Since CO2 is
released into the gas, its concentration depends on
velocity of the gas rather than the velocity of the
solid.
LYONS et al. [5] used a simplified model of the
flame which assumed that the reaction rate of the
flame is inversely proportional to the distance from
the end of the kiln. While this greatly simplifies the
model, it effects the dynamic response of the model
by ignoring the interaction of the rate of reaction of
the flame with the surrounding solid and gas tem-
peratures. A slightly more complex model which
treats the flame as an exothermic chemical reaction
is developed in Appendix C. In this model, the
flame is treated as an exothermic chemical reaction
which transfers its heat directly to the gas. This is,
of course, an over-simplification which is reasonable
for coal or oil flames. In that case the flame is
luminous over its entire length due to the burning
of small particles or droplets of the fuel. The rate
of reaction is determined by the rate of diffusion of
the oxygen to the particles. Assuming a first order
reaction equation, we show in Appendix C that the
reaction rate is
1 V~bMC,(M,P) 2-]
-I - - IkedoCe
pgLP#~Io~(RTg)2J
where do is the diffusion coefficient. This model
takes into account that the reaction rate will be
slower at high temperatures since the density of the
oxygen is inversely proportional to temperature. A
more detailed discussion of the behavior of the
flame can be found in Ref. [7]. Again, it is reason-
able to assume that the flame is always in steady-
state.
313
The heat generated by the chemical reactions is
proportional to the rate of reaction. Using the con-
vention that the coefficient is positive for endo-
thermic reactions, the heat generated in the solids is:
P~ [_AHckl~-AH,oR,o
qc=(l + Ai+ Fi+ S~)
- AH#k#S(C) 2 - AH~k~Cp]. (10)
The constant multiplying the brackets is the weight
of CaO for a total solid mass of A,ps. The heat
generated by the flame is
q: = ( - AH:)AgM~R
(8) --G y ( - AH :)[ ~MC:( MaP)~lk~IoC~ . (11)
pgV, LPrMo,(RTg)~J
Computational approach
It can be seen that in order to solve the equations
in Appendix A one must solve a two point boundary
value problem since the initial temperature and
chemical composition of the solid are known at one
end of the kiln while the temperature of the gas and
the amount of fuel are known at the other. The
approach taken was to linearize the energy balance
equations about the previous values of temperature.
The two point boundary value problem can then be
uncoupled by using the Riccati transformation
technique which sets
Tg(t, z)=v(t, y)+k(t, z)T~(t, Z ~1 •
By substituting the above equation into the
linearized material balance equations for v(t, z) and
k(t, z) can be obtained which have boundary con-
ditions at only one end of the kiln. The solution
can then be iterated at each time increment to
obtain the true solution to the non-linear problem.
During each iteration the true non-linear material
balance equations are used.
The approach was implemented on the computer
using a first order finite difference scheme which
can be shown to be consistent and stable in the
sense of RICHTMYER [9] for all integration step sizes
in z and t. For accuracy, a variable step size was
used in the z direction while a step size of 0.07 hr
(9)
was used tbr the time increment. On a G E 635 or
IBM 360/50 computer, each time increment takes
approximately 0.18 sec with the cases shown in
Figs. 6-8 taking approximately 36 sec each.
RESPONSE OF THE MODEL
The first test of the model was made without the
flame model. With the input gas temperature set at
3000°R and with twice the velocity of the gas,
sufficient Btu was supplied by the gas to obtain a
314 H . A . SPANG, III

stable steady-state profile. When the input gas practice, a similar approach is used to start a kiln,
temperature was increased to 3200 °, the burning but the depth and velocity are initially low and
zone moved down the kiln at a rate of about 7 ft/hr, slowly increased. This tends to minimize the un-
as shown in Fig. 3. This is considerably slower than productive raw material waste and the total amount
observed in actual kilns and is caused by the small of fuel consumption.
temperature difference between the gas and solid It was found necessary to reduce the radiation
under these input conditions. This test illustrates coefficients, ~, eg and e , from those used without
that the gas profile can have a substantial effect on the flame. The original values allowed a large heat
the dynamic behavior of the kiln. transfer between the burning zone and the flame.
4000 4000 . . . . . . . The inherent instability due to the positive feedback
350O T =0"0HR ! 3500 • T.0"28 HR caused extremely high temperatures in the burning
--0O0O --3000 - /" "
2~0o
zone with little heat reaching the preheat zone.
~ 20O0 Reducing the radiation coefficient reduces this
~; ,,ooI positive feedback without markedly effecting the
i
{o0OI ~'°001 ........ ..... i heat transfer characteristics in the preheat zone due
0ool 500 :" .I
%- O0 40 ~ 120 160 200 240 6EE{3~ 36O 400
to its lower temperatures.
LENGTH(FEET) LENGTH(FEET)
4OO( . . . . . 406
To improve the heat transfer from the gas to the
. . . . . t , , .
00O( T= F40 HR solid, iron chains are often hung in the preheat
~ 3 0 c T : 6 ~ " ~65C~
35(
zone. In order to match more realistically such a
GGO(
kiln, the effect of chains in the preheat zone was
150( re,u_ .."
simulated by multiplying by a constant the heat
"SOLID
transfer coefficient fl~ during the first 50 ft of the
kiln. The effect of various multiplying constants on
40 80 120 160 200 LN0 280 329 25~0400 40 a~ ,~ ,6 ~o ~ ~ ~ ~o 400
LENGTH{FEETI LENGTH{FEET) the solid temperature is shown in Fig. 4. These
4000 . . . . . . . ] 40001
35001
curves are temperature profiles at T = 1.40 hr after
350O T =4"O6I'~ T. G-40HE
L~OO° , 5O00 the effects of the chains were simulated starting
with the same initial profile.

10O0 /,""!{OLIO I0001 /,.'SOLID 4000 .... : - ; - T ~ - - - F --- ; ---. i. . . .

5O0 5ool 3500"

O~ 4~3 80 JGO I~0 ~0
. .Z40. ~0
. 320 360 40O
LENGTH(FEET) LENGTH(FEET)
:3000 GF = 5500 #/HR
Fzo. 3. Transient effect of a 200° increase in initial gas
T = 1"40 HR
temperature. -- 2500-

2OOO
When the equations for the flame model were I0 XCHAIN~ =.-.,..7.-/
included into the cement kiln model serious numeri-
: /.,~•../ ~ ' i 5 x CHAIN
cal difficulties were encountered in establishing an
I000 - / /
initial temperature profile. These difficulties were j ~ / / ~ NO CHAIN
caused by the two exothermic reactions of the burn- 500 f--
]
ing flame and of the CGS reaction in burning zone
V

of the solid. These reactions cause an essentially 0 40 80 120 160 200 240 280 320 360 400
positive feedback situation which allow high tem- LENGTH (FEET)
peratures to build up in the calcining and burning FIG. 4. Effect of chains in first 50 ft on the temperature
of the solid•
zones of the kiln. In fact, no truly steady-state
solution was obtained though the final profile, given
in Fig. 5, is close. The true implications of this The higher the multiplying factor the faster the
with respect to an actual kiln is unknown since it is initial heating of the solid. However, the rate of
impossible to get accurate enough measurements to increase is less after the chain section. This results
determine its true dynamic state. However, it would in the position of the calcining and burning zones
seem to indicate that possibly the actual kiln is un- remaining fixed. The higher multiplying factor
stable with respect to holding a constant position causes the gas to give up more heat in the chain
for the burning zone. section but less in the remainder of the preheat zone
The procedure used to initialize the complete because there is less temperature differential be-
model was to start with the solid, gas, and wall at tween the gas and the solid. One would expect that,
room temperature and to heat it to the desired since the initial calcining temperature is arrived at
operating condition. During this process, as a earlier, the position of the burning zone would start
matter of convenience, the bed depth and the to be affected if these runs had been continued for a
velocity of solid were held constant. In actual longer time.
 A dynamic model of a cement kiln 31 5

The final profile obtained with the flame is shown We will now examine the step function response
in Fig. 5. Although not completely steady-state, to changes in gas velocity, kiln speed, and the turn-
especially the wall temperature, this profile is used ing off of the flame. Each of these runs were started
as the starting point for the transient studies. Com- with the profile given in Fig. 5. The step change in
paring with the no-flame case, in Fig. 3, the preheat all cases is large in order to show the effect more
zone is longer due to the lower gas temperatures in easily. A smaller change would to a lesser degree
that region. However, the hotter burning zone provide the same effect.
temperatures allow a much higher gas temperature In Fig. 6, the gas velocity has been reduced from
in the calcining zone. This results in a much shorter 40,000 to 30,000 ft/hr. This instantaneously
shortens the flame which concentrates the burning

/ ~i.~,~-I.
calcining zone.
to a shorter area and creates a higher flame tem-
4500 1 I 1 I l I I i / , ,,~%i perature. The flow of heat out of any cross section
4000. is also reduced. The net effect is to increase greatly
/I i ~" the temperature of the gas prior to and over the
3500 ~ ,. peak of the burning zone, but to decrease the gas
• j /
.=
E3000 ,9, temperature within the calcining and preheat zones.
Since the effective thermal time constant of the
~2500
<[
solid and wall at the burning zone temperature is
~2ooo short, the temperature of the solid builds up rapidly
i.o ~ in the flame region. The increase of the peak solid
1500
k-- temperature in the burning zone further increases
IOOO - / / the peak gas temperature causing a continual
I i FUEL/ 05 ''~
500 / increase in the gas, wall and solid temperatures in
the last 40 ft of the kiln. The flattening of the peak
0 I I 1 I I I I I J/i '0
40 80 120 160 200 240 280 320 360 400 gas temperatures at T=0.91 and 1.47 hr is due to
LENGTH (FEET} off-scale plotting and not the simulation.
FIG. 5. Final (nonsteady-state) profile used as start for
rest of studies:
(a) Temperature. ,=1- /A, i ~L_ ,,,~,. i

% .... ~'"k y-..s-q"., :\\ !

~'~... 1 f ~ / " " ,Jb.ot
Co CO3
,~r: /i
,.51 LF~TH FEET) LI~eTH (FEn

,.oo ~ c~s i. ll,=l_ . ...... ~ it\ i

0.5 . . . . . . . . . . . . . . _s!o~. . . . . . . . . . .-,-'-"L\~ ,oo~.........

[_~ . . . . . . . . )
.~!.:,
, i',
:.L...... '~ °,
o.~ ~%=., 4~00l
Lrd~'r H I r E ~
: T r r- • 7 'ISO0, r ~
LIENGTH(FEll
r ~ 7 • ,
,ICOOr 4~¢0
, I ' ' ~ ~ ~ \ ~4A ,=t . . . . . . . //-,~ ,,,. . . . . . . . . .¢ !~
0 40 80 120 160 200 240 280 320 360 400
LENGTH (FEET)
F~o. 5. Final (nonsteady-state) profile used as start for ~,,~y.;..J ..- i,.o~ ~,,,~ ...... ),o:~
rest of studies: ,U:............ "
,~U
,~ I'~ .........
= .., ,~i'
(b) Chemical composition.
LEMfTH (FEET/ LEM6TH ~FEET)

Comparing the two cases it can be shown that the FIG. 6. Effect of a 25 per cent reduction in gas velocity:
(a) Temperatures plotted as a function of distance,
total input/output efficiency of the kiln with the
flame is twice as good as that without the flame.
The reason for this is the more efficient use of the Even though the gas, wall and solid temperatures
thermal energy in the calcining and burning zones. are increasing, the burning zone is moving toward
This increased efficiency is obtained even though the flame end of the kiln at a rate of 14 ft/hr. This
more Btu are used in the preheat and calcining apparent contradiction is caused by the reduced
regions since the wall temperatures are lower. amount of heat being transported down the kiln.
316 H.A. SPANG, II I

Since less thermal energy is available, both the at the rate of 14 ft/hr due to less heating of the solid
preheat and calcining zones become longer moving in a given length. The lengthening of the zone
the burning zone down the kiln. occurs slowly because of the higher time constants
In actual practice, measurements of the solid and at that lower temperature. The calcining zone also
gas temperatures cannot be made as a function of is lengthened due to less energy transfer from the
length, but only at a few fixed points along the kiln. gas, causing the burning zone to move out at a rate
As illustrated in Fig. 6(b), this type of measurement of 28 ft/hr.
can be very misleading. If one were looking at the
solid temperatures at 380 or 400 ft, one might feel 45OOi /. 45o~
that the burning zone moves back toward the flame
increasing the solid temperature and then forward
with a resulting decrease. Actually, as Fig. 6(a)
shows, the burning zone moves in only one direc-
tion. The change in temperature is due to the io:o '/ ~ iooo '
O5 :
narrowing of the zone as more heat is applied. The
reading at 350 ft gives the "true" picture of the zone t £NGTH(FLEET) LENGTH(FEET}

movement, though it can only be interpreted

T,O~3 HR
properly because we know the state of the entire 3500 3~0~ ,~

:. j~
kiln.
4500 i :
450012-"---~G-...., '
4000 4000. ---- TW
rUE / 05 m FUEL' C,5 ~
3500 Z = mooFT 5500 " "'" ' • TS 5OO / 500 /
/
L~5o® o LENGrH (FEET) LEITH ~FEETI

a~oo 4000 / ' 4o00 ]'~\ ;

~.~ooo . . . . l~____

1500 .............................. FO
TS
I000 IOOO i-
O~
500 5001 -
FUEL
0 a-'-~ m I -- 0 c!. . . . . F2EL..... O 500: ~00

4~r0
OOT 057 i"07 1"57 ~07 C~T 057 ~07 1'57 2'OT
T)ME (HR) TIME {HR] Oo i
4o ~, :
~o i
leo i
2oo ~: ~l ;o
~o: :(~o 4oo Oo ~ ~ ,
iz,3 ~ ~ z~ :
~ ~ ~/ .~J
LEN~TII[FEET) LENGTHIFEET)
4500 ~ 1~-S
.................... ... ......
4000 . TG 40OO FIG. 7. Effect of a 26.5 per cent increase in the flow rate
3 5 0 0 ~ 3500 • ........... TS
of the solid:
(a) Temperatures plotted as a function of distance.
g,5oo- ,.=. =~
~, EOOO~ zo~r ! 4500 I ~ ]
!
45¢0 : t ~ 4500 : T "
~- ,~oo ..... ~L ._ ,o § 4000 • ] 4OOO- 4 0 0 0 L ~

3500 Z,IOO FT I 3500 ~_ 35OO

I000 L FUEL 0~5 iO00~!ii.~
..... -T-G - " o ~

500 l ............. 500 - Z ~400 FT 05 2. 3oco

~: 2500
7 207 . . . . . T_G
.... i
TIUE 1~1 TIME (HRI 2000
r- T*
FIG. 6. Effect of a 25 per cent reduction in gas velocity:
Co) Temperatures plotted as a function of time.
1500-...,

~000
.,..TS ........... dI
Ii 1500 L

I000[
Z =500 FT ~ ~500v

I000-
Z ,~50 FT ~1'0~

-.0"5
500 500
~0- ! J _ i ! ...... -%L__
01
Increasing the rotary kiln speed is modeled as an O',37 057 1-07 1"57 2"07
TIME {HRI
007 (;'57 ~
TIME (ItRi
1"57 207 007 ff57 I'07 1.5: 2'OT
TIME (HRI
increase in the linear velocity of the solid from 150 4500- : ;

to 190 ft/hr, and this has an entirely different effect 4000 4000 Z ,400 ET
Z,ESO rf
on the temperature profiles as shown in Fig. 7. The 35OO~
3~00; I
increase in velocity means that the solid material 3000~ ~ TW

resides in the kiln for a shorter period of time which ~soo i

=
--
I~ 2soo :
~ L~TW
....... ..TS
•
i
=
] t
causes less heating of the material. Any hot material ; ~ zoool ~ i
that is initially there moves out of the kiln faster
than before. The net effect is an immediate decrease
"!
moo . . . . Lug L . . . . . 05 IOOO!
! 0"5

in the burning zone peak temperature. The peak 5OO

I ! 90, i
gas temperature is also reduced due to less thermal ~7 ~ ~ ~57 ~ ~T 057 107 F57 2,07
TiME {Hill TIME (HRt
feedback from the solid. Though the temperature
of the gas remains essentially constant throughout FIG. 7. Effect of a 26.5 per cent increase in the flow rate
the preheat zone, the length of that zone increases (b) Temperatures plotted as a function of time.
 A dynamic model of a cement kiln 317

4500 -- [
For this increase in kiln speed, the measurements 45O('[ : 7~-- T. . . . . --~,--T-

of the temperature at fixed points give an accurate 4 o o 0 i-

picture of the behavior of the kiln as shown in Fig. 3500 - Z=lO0 FT 3500 - Z * 500 FT ~ 3500 TW I

7(b). However, there is an immediate slight increase ~ 3000

in the exit solid temperature as more of the initially 2500[

... TS i Q ....... Ts
hot solids move out of the kiln. 2000 . . . . ~ TG -~

It is well known that a kiln will make cement long

after the flame has been turned off due to the ~oooI. "........ T~....
I
storage of heat in the walls. This effect is shown in 500~

Fig. 8 with the velocity of gas reduced to 10,000 OOT 057 1'07 1"57 2'OT
TME (HR TIME (HRI TIME (HRI
ft/hr. The temperature of the gas in the burning
zone reduces immediately to the wall temperature.
Since the gas is still being heated in the burning
zone, it continues to supply heat to the preheat and
calcining zone. This internal feedback greatly ]
extends the time cement is made. From the plot of 2000~ " TS = - "
the chemical composition at the flame end of the
kiln, shown in Fig. 8, we see that good cement can ,ooo_
be made for at least ½ hr, if not longer. Of course, ~! 1 ~or- '
the temperature of the initial profile will influence
this time. The time is also inversely proportional to TiME (HRI TIME (HRI

the velocity of gas and the velocity of solid since FIG. 8. Transient response after turning off the flame:
these determine how fast the heat moves out of the
(b) Temperatures plotted as a function of time.
kiln.

3~F T'C4)T HR "~

3500 •
1'8 i ~ I

I~¢ /
i;f
ic~r . /
500~
1"6

1'4
k~P,t (~.rrl LENGTH(FEETI

1"2
T.O'~ HR

I'0 Z =400 FT .

0"8
500 ...... ,'o
C3S C°C03

450¢
( ~ Lr.~'mFrzn 0"6
4C¢~ 40GOF-
T,O'gl HR
, 3500 ~OOh" T'1"47 HR
0'4 ,..\~ 0

0"2 I-- C3A /\I ',,

@0¢ 0
o, ,~ ~, ~o ,~0~o(FEET]
',,o~,o'~o",oo 0"07 0"57 1'07 1'57 2"07
Lt k ~ ' 1 1 / 1 ~ TIME (HR)
FIG. 8. Transient response after turning off the flame:
FIG. 8. Transient response after turning off the flame:
(c) Chemical composition plotted as a function of
(a) Temperatures plotted as a function of distance. time.
318 H . A . SPANG, I11

CONCLUSIONS APPENDIX A

The model given in this paper appears to m a t c h Thermodynamic energy balance equations
qualitatively the behavior o f actual kilns fairly well. Gas:
This is especially true at the burning zone end o f the
kiln with the all important flame-solid interactions. ~T ,
AaCmPaVo2-~ = ,[J1( "Fw- To) + fl2(T~ - To) + q f
There are, however, two areas which require refine-
ment. The first is indicated by the peak tempera-
T o(L, t) = Tg i
tures in the burning zone which seems somewhat Solid :
high. This would indicate too high emissivity co-
efficients in the burning zone requiring the use o f AsCpspsVs~z s = f l 2 ( T g - 7'~)+ fl3( T w - Ts)
coefficients for each zone. Since there are large
physical changes in the material in each zone,
C v , p ~ A ~ 2 + Asq ,
changes in emissivity are physically reasonable.
The second refinement is needed because tem- T~(O, T)= Ts~
perature o f the gas in the preheat zone o f our model Wall:
is less sensitive to changes in the burning zone than
has been observed. This is primarily caused by the ~?Tw ,
A wCpwpw-'~- = fi, (rg - T,,,) + fl a(T, - Tw)
wall dynamics in the preheat zone indicating a
better approximation to the wall should be de-
+/~,(T. - T~).
veloped.
As in all models, it is highly desirable to obtain a
quantitative comparison with the actual kiln. This Material balance equations
has not been possible primarily due to the lack o f Water:
adequate sensors. There are usually only two points
o f gas temperature measurement and one highly &o &o R fko,~o co~0-1
questionable measurement o f the burning zone ~ot = - R , o - v , - ~ z ~' = / k , , o9>0"1
solid temperature. A specially instrumented kiln
appears to be necessary.
~o(o, 0 =~,
CO 2 :

Acknowledgements---The author acknowledges the many O(L, t)=o

helpful exchanges and discussions that he has had with at Ag p~ Mc Oz
John Friedly during the development of this model. The
author is indebted to the people in the GE Industrial Sales
Division and Process Computer Business Department for CaCO 3:
their help in understanding the physical behavior of a kiln.

~(o, t)=~
REFERENCES

[1] R. H. BOOUE: The Chemistry of Portland Cement, 2nd C3S:

ed. Reinhold, New York (1955).
[2] F. M. LEA: The Chemistry of Cement and Concrete, rev.
exl. Edward Arnold, London (1956). ~(o, 0 = 0
[3] W. GmaERT: Heat transmission in rotary kilns: Cement ~t -~k~( C)~- ~'Tz
and Cement Manufacture, Part I, 5, 1932; Part II-VI, 6;
1933; Part VII-X, 7, 1934, Part XI, 8, 1935; Part XII-
XIII, 9, 1936. C2S:
[4] J. W. LYONS,H. S. MIN, P. E. PAmSOTand J. F. PAUL:
Experimentation with a wet-process rotary cement kiln
via analog computer. Monsanto Chemical, St. Louis Ofl- Mp,kp(C)2s_Mpk~(C)fl v~Off
(unpublished). ~, t 2M , " M ,. - -~z
[5] H. S. MrN, P. E. PAmsor, J. F. PAULand J. W. LYONS:
Computer-simulation of a wet-process cement kiln fl(0, t) = 0
operation. Preprint No. 202-LA61, Instrument Society
of America, Fall Instrument-Automation Conf., Los C3A:
Angeles, California, September 1961.
[6] J. H. PERRY (Editor): Chemical Engineers' Handbook,
3rd ed. McGraw-Hill, New York (1956). d y _ M~,krCaA_ v,f7 r(o, 0 = o
17] Join RtrlXS: Investigationof Material Transport in Wet- ~t 3M c ~z
Process Rotary Kilns by Radio Isotopes, pp. 120-133,
Pit and Quarry, July, 1955. C4AF:
[8] H. A. SPANO, III: A Model of a Coal/Oil Flame for
Rotary Kilns, to be published.
[9] R. D. RlCnTMYIm: Difference Methods for Initial Value &5_ M a k a C * A F - v~~--~ a(o, 0 = o
Problems. Interscience, New York, 1957. at 4Me ~z
 A dynamic model of a cement kiln 319

Fe20 3 : available fuel:

dF My 4 OF kr
. -=-= - _ k~C AF-vs=--- F(O, t)=F, do = 1
6t 4M~ Oz (Do3/r 2) + kr

A1203: Diffusion constants:

8___AA=_ M a k ~ C ' A F - Ma k , , C 3 A - v f A Do=aoTa 3/2

dt 4Mc 3M~ " Oz
Reaction rate coefficients:
X(O, t)=A i
SiO 2 : k,=A, e x p ( - E i ' ~ i=1, a, fl, y, 6, ~o
\ RT,,]
~S M s k#S(C)2 dS S(0, t)= S~
at = -ZM~ Vs~z
k r = 3 A r e x p ( - Ev ~
re \ RTJ
CaO:
Coefficients of heat transfer:
0C
t?t - k t ~ - k~Cf l - kpS( C) 2 - k.~C3A
fl, = r l v [ f 1 + 1.73 x 10-9(1 - ho)eoew(T2 + T 2)
(T o+ Tw)]
- k~C*AF- vs~z C(O, t) =0
f12 = 2 q sin/2)[f2 + 1.73 x 10-%oss(T 2
Fuel:
+ T~)(Tg + T,)]

~z pavoLPvMo~(RTa) J f13= r1(2~ - P)[f3 + 1"73 x 10- 9hewss(T2

Ce(L, t) = CFi + T2)(T~, + rs)]

Heat of reaction: fl,=2nf4r2

h = 1 d 2ho sin(p/2)
Ps 27t-p
%= . _ [-AHek~q-AH,oRo,
(1 + Ai+ l"~+ S~)
Area coefficients:
- anpkaS(C) 2 - AB~k,C~] /.2
Ao= 2 ( p - sin p)
Heat from flame:
/.2
As = 2 (2re - p + sin p)
qj A..)I. e Mo)qk.doC
.
Po% LPFMo,(RTo)~I A w = 2 n ( r 2 - r 2)

APPENDIX B
Nomenclature
Notation Value used Units
a n
Proportionality constant for D 0.0538
A # Aluminium oxide (A12Oa) / # CaO
Aa Area of gas at given cross section ft 2
At Initial value of # A 1 2 O a / # C a O at input 0.0469 # / # CaO
Av Pre-exponential factor--fuel 5-6 x 101° 1/hr
As Area of solid at given cross section ft 2
Aw Area of wall at given cross section ft 2
AI Pre-exponential factor--CaCOa 1.64 x 1035 1/hr
A~ Pre-exponential factor--CaS 4.8 x 108 1]hr
Ap Pre-exponential factor--C2S 1.48 x 109 1/hr
Ar Pre-exponential factor--CaA 3.0 x 10 l° 1/hr
320 H.A. SPAN6, IlI

Notation Value used Units

Pre-exponential factor--C4AF 3.0 x 1012 l/hr

Ao, Pre-exponential factor--water 7.08 x 107 1/hr
C Calcium oxide (CaO)
Normalize amount of available fuel
% Specific heat of gas 0.28 Btu/# °F
Specific heat of solid 0.26 Btu/# °F
c,w Specific heat of wall 0.26 Btu/# °F
C2S 2CaO • SiO 2
CaA 3CaO • A12Oa
C4AF 4CaO • A120 3 • Fe2Oa
do available oxygen
Do Diffusion coefficient for oxygen
Activation constant--fuel 2.45 x 104 Btu/lb mole
Btu
E1 Activation constant--CaCOa 3.46 x 105
lb mole
Btu
E, Activation constant--CaS 1.10x 105
lb mole
Btu
Activation constant--CzS 8.3 x 10"
lb mole
Btu
E~ Activation constant--CaA 8.33 x 104
lb mole
Btu
Activation constant--C4AF 7.95 x 104
lb mole
Btu
E,o Activation constant--H20 1.807 x 10 4
lb mole
Btu
A Coefficient of conduction--gas to wall 4
hr (ft)2°F
Btu
A Coefficient of conduction--solid to gas 4
hr (ft)2°F
Btu
A Coefficient of conduction--wall to solid 4
hr (ft)2°F
Btu
f, Coefficient of conduction--wall to outside air 0.7
hr (ft)2°F
F Iron oxide # Fe203/# CaO
Fi Initial value of # Fe2Oa/# CaO 0.0469 #/#CaO
Gr Amount of fuel per hour 5300 #/hr
ho Fraction of radiation 0.0758
Btu
k Thermal conductivity of wall 0.9
hr (ft)°F
k~ Reaction rate--fuel
kl Reaction rate--CaCO2
Reaction rate--CaS
ks Reaction rate--C2S
kr Reaction rate--C3A
Reaction rate--C4AF
k ,~ Reaction rate--water
L Total length of kiln 400 ft
M Depending on the subscript molecular weight of
the chemicals
[see BOGOE(1) for values]

P Angle subtended by surface of solid (Fig. 1) radians

2
Btu
qc Heat generated by chemical reactions
ft3hr
 A dynamic model of a cement kiln 321

Notation Value used Units

Btu
qF Heat generated by fuel
ft hr
Q Heat generated or moving into a region
rF Particle size of fuel 10 -2 cm
rl Inside radius of kiln 5 ft
]'2 Outside radius of kiln 6.0 ft
r3 Ratio of heat transfer in chain section 5
Reaction rate of water
S # Si02/# CaO
St Initial value of # SiO2/# CaO 0.3398 #/#CaO
7. Temperature outside kiln 561.7 oR

rg Temperature of gas oR
7g, Input temperature of gas 1700 oR

7~ Temperature of solid oR

7~ Input temperature of solid 562 oR

7w Temperature of wall oR

1)g Velocity of gas 40,000 ft/hr

I)s Velocity of solid - 150 ft/hr
z Distance along kiln ft
# C3S/ # CaO
8 # C2S/# CaO
Btu
81, 82, 83, 84 Heat transfer coefficients hr °F
# C3A/# CaO
5 # C4AF/# CaO
Heat of reaction--CaCO3 1275 Btu/#CaCO 3
AH~ Heat of reaction fuel - 14,000 Btu/#
AH, Heat of reaction--CaS 11 Btu/#CsS
AH, Heat of reaction--C2S -381 Btu/#C2 S
AH~, Heat of reaction--water 970 Btu/# water
'go Radiation coefficient--gas 0.273
~s Radiation coefficient--solid 0.500
~w Radiation coefficient--wall 0.751
# CaCO3/# CaO
Initial value of # CaCO3 / # CaO at input 1.784 #/#CaO
P9 Density of gas 0.05 #/fP
P~ Density of solid 56 #/ft 3
Pw Density of wall 112 #/ft 3
# C02/# CaO
03 # Water/# CaO
031 Initial value of # Water/# CaO 0.0649 #/#CaO
Changes in constants for no flame case
vo Velocity of gas 80,000 ft/hr
% Radiation coefficient--gas 0.447
~ Radiation coefficient--solid 1.0
8w Radiation coefficient--wall 1.0

APPENDIX C
Development of the flame model V2x_°2x l_2 ax_0. (Cl)
The combination of fuel particles with the oxygen -7,2 re,.
is a diffusion process. In this appendix, we shall
first obtain a simple diffusion model and then We shall assume that far from the sphere x is
determine the total reaction rate for the flame. Let equal to its initial concentration, Xo. At the surface
x be the oxygen reacting with a sphere of the fuel of the sphere, we shall assume that both a first- and
which has radius r o. In spherical coordinates the second-order reaction takes place. Therefore, the
steady-state diffusion equation is boundary conditions for equation ((21) are
322 H . A . SPANG, I11

X=X 0 r~ ro (C2) where Gv is the total amount of fuel burned and

CF is the percent of fuel left at any given point.
The concentration of oxygen is given by the equa-
D S d x _ kpx r = ro (C3)
o "dr- tion of state for a gas.

where D is the diffusion constant, S, is the ratio of P M,

the surface area to the volume of the sphere and
Co~= ~ o Mo~ "
k~, k2 are reaction rate coefficients.
The solution of equation (C1) can be shown to be Similarly, the density of the gas will be
of the form
P
rpkexo Pa = MORT a •
X=Xo=
kr+(DoS,/ro}"
In the previous equations, P is the pressure, ~ the
We define do as the percentage of x available for percent oxygen in the air and M,, Mo2 the mole-
the reaction at the surface of the sphere. cular weight of air and oxygen. We have assumed
in deriving the gas energy balance equation that
the total mass flow of gas was independent of tem-
d°=~oot,o (C4) perature. Thus, the total rate per unit volume is

Assuming that the particles are nearly spherical, R = GF r~b(PMa)2Mc1.]k#loCr. (c10)

S~ = 3/r r and F,AgLPFMo~(RTg)~ ]

3Do Assuming that the fuel particles move at the

same velocity as the gas, the concentration of fuel is
do- r~kF (c5)
3Do
r2kv + 1 K v = FF . (CI1)
vgAg
where r v is the particle size, D O is the diffusion
It is again reasonable to assume that the reaction
constant for oxygen given by
of the fuel will always be in a steady-state. There-
D o= aoTa 3/2 (C6) fore, we obtain the material balance equation for
the fuel given in Appendix A.
and k r is the rate constant for the reaction of the
fuel. R&amar----Unfour ~tciment est un type de prochd6 ~tcontre-
courant dam lequel a lieu une srfie complexe de rractions
kr=3A%xg-Evl. (C7) exothermiques et endothermiques. Un modrle dynarnique,
rv "LRTa.] ~tdrrivOespartielles, du four a 6t6 6tablie g fin de comprendre
son comportement et fournir tm moyen d'amrliorer la com-
mande par calculateur. Dam le prrsent article, les rractions
We have written k r in a slightly different form in et les hypotheses fondamentales du modrle seront discutres.
order to reflect its dependence on ratio of surface Un modrle d'une flamme charbon/mazout tenant compte
des temp&atures ambiantes est 6tabli. Le comportement
area to volume of the particle. dynanuque du four illustr6 par ce modrle est ensuite indiqur.
The reaction rate per particle of fuel will be I1 est montr~, en particulier, que l'interaction entre les deux
rractions exothermiques conduit le four ~t~tre imtable en ce
R' = krdoCo2 (C8) qui concerne le rrglage de la position de la zrne de combus-
tion.
where Cos is the total concentration of oxygen. The
total rate of reaction in any given volume of air will Zusammenfassnng--Ein Zementofen stellt einen ProzeB vom
depend on the fuel-to-air ratio and is given by GegenfluBtyp dar, in dem eine komplexe Folge von exother-
men und endothermen Reaktionen stattfindet. Um sein
Verhalten zu verstehen, und Mittel zur Verbesserung der
R v - F r p ~ C t krdoCo2 . (C9) Steuerung dutch Rechner zu finden, wurde mater Benutzung
FaPF partieller Differentialgleichungen ein dynamjsches Modell
des Ofens entwiekelt. Die Hauptreaktion und -Annahme
mC 1 is the molecular weight of carbon. F r, PF, Fa des Modells werden er6rtert. Ein Modell einer Kohle-Ol-
Flamme, die die Umgebungstemperaturen in Rechnung
and Pa are the pounds per hour and the density of zieht, wird entwickelt. Das dynamische Verhalten des Ofens,
the fuel and gas, respectively. wie es durch dieses Modell dargelegt ist, wird dann angege-
To put this rate in a more usable form, let ben. Speziell wird gezeigt, dab die Wechselwirkung zwischen
den zwei exothermen Reaktionen verursacht, dab tier Ofen
im Hinblick auf die Regelung der Lage der Brennzone
Fr=GeCr grtmdsitzlich imtabli ist.
 A dynamic model of a cement kiln 323

Pe3mMe---I~eMeHTHaS n e ~ s npe~cTaa~seT c060fl iIpOTHBO- Pa3pa6oTaHa MO~Ca~ yroYmHO-MaayTHOtt II3IaMCHH, npHH-

TO~ n p o u e c c B KOTOpOM IIpOHCXO~IiT p ~ 3K3OTepMH-
qeCKHX H 3H~OTepMHqeCKIiX p e a x H l ~ . Pazpa6oTaHa RUHaM-
H~a~ napLtKa.~Has ~ Y [ ~ e p c m m a ~ H a ~ MO~CYH, C ~IcYn~o
~TO6bl HOH~ITh ee HOBeHeHHC H y ~ T b yHpaBHCHHe
IIPOHeCCa BbI~HCHHTCYIbHO~ MamHHO]]. B CTaTbC paCCMOT-
peHs~ OCHOBH/~Ie pCaKHHH H rIpC~rlOHO~KCHII~ Mo~eHH.
HMa~ nO VHHMmme o x p y z a ~ m H e TeMnepaTyp~. 3aTeM
y x a 3 a u o ~HaMU~ecxO¢ noBe~eHHe Helm x a x iloKa3ano 3TOi~
MO~ICJ~O. ]3 LlaCTHOCTH lIolta3aHo ~TO B ~ O ~ I C ~ C T B H e
MCX.B,y ~ByMH 3K3OTCpMH~eL'KHMH peammsMH BM3MBaeT
HeCTa~HJIbHOCTI~ HCLIHOTHOCHTeHbHO peryJIHpOBaHH~[ MeCTO-
HOJIO~KeHH~I 3OHLI ropeHH~.