Process Engineering

Training Program
MODULE 13
Kiln Volatiles
Section

Content

CETIC Volatiles Group- Final Progress Report

CETIC Sub Commission Behavior of Volatile Material in Kiln Systems

Investigation into Potential Low Temperature Volatilization

Factors Affecting Sulphate and Alkali Cycles in Rotary Kilns

Alkali Volatilization- A Review of Literature Available in 1977

A Study in the Volatile Cycles on HOPE # 2 kiln

Design and Experience with Bypasses for Chloride, Sulphate, and Alkalis

Kiln Gas Bleed Considerations

Ring Formations in Cement Kilns

10

Kiln Build-Up Meeting

11

Cement Seminar- Rings, Balls, and Build-Ups

12

Ring and Buildups in Cement Kilns

HBM Process Engineering Conference

Minimization of Volatile Cycles

Blue Circle Cement

PROCESS ENGINEERING TRAINING

PROGRAM

Module 13
Section 1

CETIC Volatiles Group- Final

Progress Report

Blue Circle Indusuies PLC

Internal memo
3

SEE BELOW:-

rot?

C P KERTON
Your re::

Date

21 December 1994

Copies

Stijec

CETIC VOLAl-IUS GROUP - FINAL PROGRESS REPORT

Herewith a copy of an English version of the final report from the CETIC GOUP
which worked from 1990-1992, initially inspired by 3lue Circles initiatives.
Copies of various French documents exchanged within the group are available
from me or from Cb.ris Hoit.
The text follows the order of the French original, intended for those who took
part in the work. For those who have not been following progress so closely, the
logical order of reading would be to stan with Appendix 2, which is a published
paper based on the initial work of the group on cycles of chlorides, sulfates and
alkalis driven from the burning zone. Next comes Appendix 3, which is
Tom Lowe-s account of some practical applications of these concepts when
burning petroleum coke. Returning then to the start of the main text, there is an
account of further discussions and activity on this topic leading finally to
Appendix 1 which looks at capture of SO:, in colder parts of a kiln system. (Those
concerned soleIy with pollution abatemerit might well start here!)
Proposals for possible further work are listed and those relating to sulfur
behaviour will be pursued by joint exchanges of experiences from 1995.
Suggestions for further distribution within Blue Circle will be welcomed.
c
I&
,/
f&L
>
c P Kerton
Patent & Information Service
Eric
To:

;P L Rover (2 copies)
C P B Turner
T M Lowes
C J Holt
L P Evans
R M Mutter
J M Lawton
P A Longman

ca: \?%+!\2112CETIC

Blue Circle Technical Centre

TC 94 049

CEl-lC SlJB-CO-ON

BEHAVlOUR

O F

VOLATILE

M A - IN KILNS

FINAL. PROGFZES REPORT

MAY 1994

This report is strictly confidential within the Blue Circle Group.

Additional copies should not be released outside Blue Circle
without reference to the Technical Cenme.

Please apply to the hformation Servicesat

tie adcks give0 below.

September 1994

SUMMARY

Together with an earlier report (ISTN 92/5) this text sets down the major findings of a
group which met from 1990 to 1993. For minor components volatilised in the burning
zone (alkalis and sulfates), there are several successes on production kilns in reducing
volatile cycles by attention to burning zone conditions, especially in relation to
chemical decomposition of C&I,.

Correct diagnosis of conditions is assisted by

improvements to permanent on-line exit gas monitors and to suitable standardisation

of

sampling procedures for dry process kiln enuy material. The full effects of dust cycles
in confusing results of sampling exercises remain to be established.

Sulfur volatilisation and absorption at lower temperatures have also been considered and
the conditions of temperature and atmosphere which aid reactions for SO2 capture by
various materials are outlined. Various permutations of actions are possible to abate
emissions and some will not succeed at all points in a production line or may require
moisture addition or increased residence time to improve their effectiveness.

Proposals for possible topics to continue this work are listed. (This is an English version
of the official French text).

CONTEXTS

Paee No

1.

Introduction

2.

Task of the Sub-Commission

3.

Cycles of minor elements generated

in the burning zone

4.

Cold cycles of minor elements

11

3.

Trace elements

14

6.

Future work

15

Figures

Appendix 1

SO? CAPTURE
Chknical IMechanisms far Sulfur Dioxide Absorption in Cement
Kilns and other Industrial Abatement Plant.

Appendix 2

TEXT FRO,M INTERNATIONAL SYMPOSIUM ON GAS

CL&WING AT HIGH TEMPERATURES.
Behaviour of Volatile Materials in Cement Kiln Systems.

Appendix 3

PAPER PRESENTED BY T M L0WE.S

100% Pet Coke - Problems and Solutions.

1.

INTRODUCTION

The group has the following memberx-

HOLDERBANK

P BORKI

3LUE

P KERTON (Animateur)/C HOLT

CIRCLE

CIMENTS LAFARGE

M DANDINE/K
30ULOT/M TOUSSSAINT/
X DUPONT-WAVRLN

CIMENTS

B POLGE/G Bw BERGERY/
G FLAMENT

FRANCAIS

ITALCEMENTI

R TACCHINI

C3R

c MEYERS/J PARlSIS/P RENIER/

M BRUYERE

CIMENTS

DOBOURG

ENCI

F LAMPROYEfR SPILLAERT
W VAN LOO/F ERE?JS

The foIlowing meetings too& place:-

1.

1990 Autumn

Maastricht

ENCI

2.

1991 Spring

Greenhithe

BLUE

3.

1991 Autumn

St Antoing

CBWSCF

4.

1992 Spring

Frangey

LAFARGE

5.

1992 Autumn

OrbY

HOLDERBANK

6.

1993 Spring

Salerno

ITALCEMENTI

7.

1993 Autumn

Obourg

OBOURG

CIRCLE

This year there was one meeting at Obourg on the 4 and 5 November 1993.

TC94049

2.

Tt=chu.id

Cease, .!%@cabm 1994

TASK OF THE SUEKO~MMISSION

This is to produce a state-of-the-art report concerning the behaviour of volatile

material in kilns.

This work originates from

1.

The increased use of ail sorts of secondary raw materials and fuels.

2.

The trend to produce more low alkali cements.

3.

Emission regulations which are becoming more and more rigorous.

Furthermore, during the work, important implications for kiln output and cement quality
have been found in connection with control of cycles. Our principal recent activities
are covered under the three following headings, 3-5.

(A copy of an associated paper by T M Lowes is included at the end of this report,

having been presented at the same meeting in May 1994).

3.

CYCLES OF MINOR ELEMENTS GENERATED IN THE BURNING ZONE

We have already produced a first text on classical knowledge for this group of cycles

(ISTN 92/5): this was distributed in 1992. Supplementary work is described here.

Given a sample of material and its chemical analysis, one might think that all would
become clear, but to calculate a chemical balance, it is also necessary to know the mass
flows which are involved in the calculation, and there are a number of methods of
deciding upon these. Each method has its own advantages and disadvantages, making
them more (or less) appropriate at different points in the burning line. It is important
to note that

1.

In the past this sophistication has not generally been adopted

2.

Various flow rates for hot raw meal entering the kiln can be calculated by
different means, each one giving a different burning zone volatility.

We have continued exchanges on these topics to improve our understanding. One

important parameter in the calculations is the flow rates of entrained solid particles
between preheater stages, in particular between the kiln and the preheater tower.
There are not many results available for this parameter (we note that there are not
always the same sets of results available to explain observations!).

As far as these classical cycles of minor elements are concerned, there is much
activity within member companies at present

There are two fields of particular

interest:

Firstly, the characterisation

modified geometry

and the behaviour of hot raw meal (including the use Of

cyclones and special linings), and secondly the influence of

combustion and heat transfer on volatalisation

- especially in the burning zone but also

in precalciners. The influence of the local atmosphere close to clinker granules in the

rc94049

TecbniGai

Came. SenM I994

burning zone seems to play an even greater role than had been thought in the past:
there are several results where volatilisation has been significantly reduced by more or
less simple means (additional oxygen) and we trust that one day there will be a
somewhat deeper understanding than indicated in our previous report.

This

understanding will also help us in applying results from the mathematical analysis of
cycles from CBR which has shown great variations in alkali and sulfate volatilities

in

the burning zone in different kilns (even when taking account of the major volatilisation
of chloride which explains a proportion of these differences).

This year we have tried to bring together knowledge on:

The most recent studies on control of cycles and characterising

hot raw meal

on the industrial scale;

Mathematical analysis of cycles and cyclone performance (from chemical

analysis of samples);

The effects of the flame and kiln atmosphere on volatility,

especially when

using petroleum coke.

Study

of

free energies indicates the compounds expected and which of them are the

most stable in the prevaiiing conditions of temperature, pressure and composition of the
solid liquid and gaseous phases. Mr Berard-Bergery distributed a summary of the talk
which he previously gave on this aspect of thermodynamics, He notes that without a
certain knowledge of this subject it is difficult to make progress, given the need to
explain the apparently contradictory results obtained from operating kilns. (We must
certainly bear in mind the fact that we need to consider dynamic equilibria, not only
static systems.) Thermodynamics allow us to determine the direction, intensity and
speed of transformations of physical systems as a function of conditions. The possibility
of a reaction is known from calculating the change in (Gibbs) free energy, if operating
at constant temperature and volume or of free enthalpy at constant temperature and

pressure. Examining the trend of this value as a function of temperature, the reactions
of formation or decomposition of a single chemical compound can be considered, and
the order of stability in a family of compounds (chlorides, fluorides, sulfates etc) can
be deduced. Hence, the reactions between one element and the compound of another
element can be foreseen. There are quite a number of diagrams of free enthalpy as a
function of temperature available (determined in the field of metallurgy) and also
information on partial vapour pressures which

influence the equilibria.

Along a burning line it can be seen that for chlorides KCI is the most stable and easily
vaporises producing a chloride cycle.

For fluorides CaF2 is the most stable at all

temperatures, and does not easily vaporise. For fluorides, phenomena are therefore
very different for those encountered with chlorides. For sulfates K2SO4 is by far the
most stable and CaSO4 the least. For C&30, stability also depends among other things
on the partial pressure of oxygen, so that the sulfate cycle brought about by the
decomposition and re-combination of &SO4 will be very different depending on the
oxidation/reduction conditions at different points along a burning line. Na20 is more
stable than f(20 and is found in clinker combined in the aluminate phase, whilst K20
tends to form &SO, if possible. It is interesting to see that sulfur, for example in the
form of SO,, can be captured by carbonates because sulfates are more stable. At low
temperatures sulfites can also be obtained.

CIMENTS FRANCAIS note that it is necessary to define the conditions in which a kiln
must operate to give the desired results, in the case of high fluxes of sulfur or alkali.
Results from CC3 confirm these ideas and they are being applied in other Works. Mr
Flament has told us that the text which he presented at the Berlin Congress gives a
good summary of his experiences at CCB.

At CCB, using 100% petroleum coke (ore-calcination with separate air at around 55-60%
kcal) a good kiln output is found when there is a higher and more constant oxygen level
at the kiln exit (1.5 - 2.0%), a greater fuel fineness (residue 3.6% compared with 5.0%),
a more stable flame shape, and a less severe burning regime - obtained by means Of an

examination and adjustment of geometry in the burner region. The geometrical aspect
helps to avoid too high a dust cycle (including cooler dust) and flushes, which can both
cause alkali capture and blockages (contrary to the ideas of certain plant suppliers).
&SO4 nodules are found in the interstitial clinker phase, and there is a need to consider
quality aspects further. There is a sulfate: alkali ratio of 3.5, which is acceptable in
the stable kiln regime: thus it is preferred to use only 50% petroleum coke for
production purposes, so as to reduce the possibility of entering into a potentially
difficult state. Coke brings 26% of the sulfur arriving in the kiln system.

The sulfate/alkali ratio is an important parameter to take into account, but it is not the
only one. It is also necessary to have a good and continuous analysis of kiln exit gases
to allow combustion conditions to be followed.

It is preferable that gas analysis is

automatically corrected for oxygen level so as to indicate other changes more clearly.
It is equally necessary to continuously monitor precalciner

combustion conditions by

means of supplementary analysis of relevant compounds in the gas phase.

The intensity of combustion in the burning zone is an important factor to understand

and use to control events. The decomposition of CaSO, in a reducing atmosphere is the
key mechanism. Each kiln has its appropriate oxygen level in the kiln exit gases, which
must be respected for a given sulfate input. The word volatile can lead some people
into error:

whilst KCl

and NaCl are present in the form of gaseous molecules,

thermodynamics indicates that K20 and Na,O decompose in the flame and re-combine
later. It can be useful to separately tabulate the calculated volatilities

of KCl NaCl,

K$S02, NaSO,, CaSO, instead of only Cl, K,O, NaZO, SO3.

We have discussed the design and operation of by-passes, certain of which take a
significant dust burden at the kiln exit, which brings about difficulties in operation. In
Ciments Francais kilns the range of dust burdens is from some ZOO-1,000 kg of dust per
tonne of clinker. This dust burden reduces the performance of the lowest cyclone: if
the value of the dust burden is not known it can be difficult to interpret results. (It

was

noted that Weber recorded low dust burdens in all the kilns which he analysed) Other

important parameters which must be known in order to diagnose a situation are the
chemical analyses of good samples of coal and of the hot material coming into the kiln
at the feed chute level.

Given the above-mentioned data, the measurement of CO? level at the kiln exit over
a certain period allows separate calculation of decarbonisation in the kiln and the
preheater/precalciner.

The mixture of dust and new raw meal which comes into the kiln

can then be estimated from the loss on ignition. It is suggested that methods using
chemical tracers to estimate the material flow, for example K,O, can be falsified if the
level of dust cycle is not known.

It seems probable that the geometry of a plant has a marked effect on dust
entrainment. In the older generations of Dopol preheater, material fell a long distance
from the bottom cyclone into the kiln.

The kiln entry material had a low loss of

ignition which could be falsely attributed to good decarbonisation.

The more recent

Dopols have a side entry for material: a bypass can then expect to encounter a lower
dust burden. The older lateral centrifuge entry of F L Smidth also produces a poor
bypass efficiency, due to dust entrainment.

High dust recycle levels also have the

inconvenience of increasing the probability of blockage

(Mr DuPont-Wavrin noted that the Berthold Company is supplying an X-ray detector to
monitor material flow rates ex-cyclone).

It is useful to calculate the effect of dust cycles on thermal performance. A heat and
mass balance for each preheater stage allows the effect of dip tube geometry changes
to be observed. Opinions vary as to the appropriate choice for different stages, not to
mention the use or removal of cyclone exit flaps.
A variety of experiences have been reported regarding Hasle Vortex Finders. At ENCI,
excellent results have been recorded for over 3 years, whilst at CBR the tubes were lost
in 3 months- Ciments Francais have observed the same range of lifetimes. There is
consensus on the advantages of dense ceramic Hasle units in the kiln feed chute

(with

a minimum of exposed refractory cement?, when they are set up with a good
arrangement of air cannons. Some peopie have doubts as to their sensitivity to thermal
shock in other regions of kilns during heating and cooling.

ITALCEMENTI

has described similar experience with a F L Smidth chloride bypass at

Picton, also used to assist the production of low alkali clinker. Here the high dust
burden (some 1,000 kg per tonne) causes
efficiency of a dust bypass.

the so-called gas bypass to have tbe

Tests are in hand at Nazareth with a purifier which

removes SO, by injection of raw meal and water (a Monsanto design). At Colaferro two
geometrically identical preheater kilns produce respectively some 1,900 and 1,150
tonnes a day. The higher output kiln is fired with a mixture of coke and coal and
produces

build-up

problems, whilst

the

other

operates

satisfactorily

with

100%

petroleum coke. The only difference that has been noted in combustion conditions is
a higher secondary air temperature due to the use of IKN plates in the Fuller cooler.

Italcementi manages to use 100% petroleum coke on the Lepol process, even with 10%

over-grate firing, with emissions of SO, - except during build-up losses - having only
pyritic material as origin. The residue is 10% as for coal. On the dry process it is
necessary to drop the residue to 4% (or 5% for a coal/coke mixture). The precalciner
gas does not contain any SO2 when coke is used in the burning zone. The sulfur leaves
with the clinker, partly during occasional flushes.

In the past coke or anthracite was introduced into long granule-fed kilns. This helped
formation of a good burning zone, but nowadays it is found that there is also a high SO2
emission.

This coke is now added to the main burner, a procedure which operates

satisfactorily if the burning zone is controlled via NO, monitoring to avoid the problems
which can be caused by the sudden arrival in the burning zone of build-ups detached
from internal cruciforms.

In the same way, LAFARGE has continued with a major programme of geometric
"centralisation"

of burners, noting oxygen levels (typically some 3%) and SO, levels ex-

73.i~ re-sort Ls soicrlr confidemial w.rhio the Blue Ctie Group.

kiln as a function of flame momentum. Several Works

keep the centralising

mechanisms

on the kiln platform so that alignment can be corrected if there are changes after some
weeks of operation. In such circumstances 100% coke can be used (3-5% S) on the dry
process with a 5-10%

residue.

Automatic kiln entry material sampling systems are generally installed with a view to
assuring safe operation (Pfaff)
la Nouvelle.

and Lafarge gave an account of an in-depth study at Port

The company was particularly

interested in the impact of sampling

techniques on results. Here, there is a kiln fed at some 100 tonnes an hour (50% of the
heat energy coming from petroleum coke) with 3% oxygen ex-kiln. The levels of all
volatiles in the collected dust go down as a function of sample suction rate, reaching
a plateau. The dust Is really a mixture of fine material (high in volatiles and easy to
collect) and coarse material. Although the nominal isokinetic aspiration rate for the
probe was some 30 litres per second, it was not aligned with the gas fIow direction and
higher suction rates were therefore needed to obtain a representative sample.
Lafarge express the hope that a standard method can be written up, suitable for use
throughout the world. They have currently only two or three competent sampling teams
in their French group. It was noted that it is a bad practice to make use of large probes
and low capacity pumps to reduce blockages as far as possible.
CBR reported on SO2 levels in the kiln system at Antoing. The level of some thousands
ppm ex-preheater drops to 600 ppm at the stack. This loss is split

20% to the crusher,

20% to the mill and 60% to inleaking air. It was reported that at Rekingen the raw mill
is run at a reduced throughput, in order to allow SO2 capture t o continue throughout the
operating day.
A few supplementary results from the simplified mathematical model have been
distributed, with its application to the analyses of balance samples from various member
companies, so that volatilisation,

entrainment and capture coefficients can be

calculated together with the performance of some cyclones.

ThLs remrr Ls snicdy codid~riaf w-if&in tie Blue CLv!e Grarw.

A significant range of values was noted, all calculated on the same basis. K20 is a good
tracer to determine raw meal

entrainment by gases. For the calculated entrainment

values in the document distributed, one must consider the position and methods used to
collect the kiln exit dust samples - with a probe in the kiln, in the riser duct or exbypass. Nevertheless we consider that it will be very useful to extend the tabIe of
results already obtained by sending further analyses to CBR to gather a common table
describing volatilities,
volatilisation)

An example of a Blue Circle kiln (poor flame with high

is the only one to have been added this year, and the model remains to

be more widely used.

At BLUE CIRCLE some thermodynamic data lead us to think that in a typical kiln gas
there is at llOO*C a sufficient reduction potential produced from 2,000 ppm of CO to
reduce CaSO, with a consequently much higher voiatility.

A separate paper from Blue

Circle is appended, giving an account of UK experience with use of petroleum coke.

A paper from Blue Circle regarding the design of a bypass for a new Works with high
chloride raw materials was discussed.

There was also an expected high content of

alkalis in the clinker, which could perhaps be reduced by the addition of even more
chloride. The performance of existing by-pass systems indicates that despite the fact
that (according to suppliers) there is a possible dust loss of some 200 to
250 g/Nm-

in the gas extracted from the system it will be best to calculate with a

nominal level of 400

In such circumstances there will be a need for a raw meal

preparation system with a significantly higher throughput than normal.

Several remarks were made: there are examples of precalciners

blocked by sulfur; the

handling of by-pass dust rich in CaCl, is much more difficult; is the fuel penalty per
percent of by-pass closer to 5% than the 1% used in this example? (This latter figure
was supplied by the company which intends constructing the proposed Works.)

The animateur was invited at short notice to give a paper during an International
Symposium of the Cleaning of Gases at High Temperatures in December 1993, which is

This remrr Ls smcr(;l cmfidemiai withiu tie Blue Cri-cle Group.

appended.

Several specialist workers in this field have encountered problems of

blockages and build-ups which provoke their interest. (Note that data on the equilibrium
CO-CaSO~-CaS is given in "Sulphur Capture in Fluidised Bed Boilers:

the Effect of

Reductive Decomposition of CS04, by A Lyngfelt and B Leckner, Chemical Engineering

Journal, Volume 40. pages 59-69, 1989.)

4.

COLD CYCLES OF MINOR ELEMENTS

LEMENTS

Information on the effect of internal cycles on emissions to the exterior has been
exchanged, avoiding (if possible) examination of equipment for capture of such emissions
which is left to the "Environment"

sub-commission, liaising with its animateur.

topic typically concerns cold cycles of SO2

This

formed at low temperatures in the

preheater, and capture of SO, emissions in long kilns and in the Lepol process involving
cycles which originated in the burning zone.

For this topic it seems that most information

Environment

sub-commission:

has already been treated by the

it remains to define more precisely the chemical

reactions which are involved and the domains in which these are the most (or the least)
effective. We are interested in establishing information about chemical efficiency of
absorption of SO2 as a function of conditions of atmosphere, humidity, temperature,
residence time, particle size, chemical composition etc.

The animateur noted the classification

of absorption mechanisms given in two USA

papers: these concern tests carried out Davenport (Steuch) and at Lone-Star (Sheth)

The document sponsored by the British Pollution Inspectorate is also available. This
reviews published information on removal of trace gases. It covers a range of both
chemical species

and reactive materials, The sections relating to SO2 capture appear

to provide a useful framework within which the cement industrys experience can be
classified, see Appendix. Other industries

are interested in the possible future use Of

sorbents with increased reactivity (cement kiln dust?) and in regenerable agents

such

as calcium disilicate.

The reports of HOLDERBANK

to the Environment Sub-Commission have indicated that

emissions of some 1,000 to 1500 mg/Nm3 of SO2 were reduced to close to 350 mg/Nm3
during the operation of the raw mill.

The same effect could be obtained in direct

operation by the addition of Ca(OH)2 to the raw meal. This provides removal of 50%

SO2 at a stoichiometric level of 5; Polysius would suggest 80% removal at a ratio of

8. Not all users seem to have taken account of the need to use superstoichiometric
quantities

of sorbents, which sometimes may react with only some 10% efficiency.

Work carried out in the UK by Lodge-Cottrell

desulphurisation,

in the field of the electric power station

has shown an efficiency of 25% for dry lime injection, rising to 50%

in the presence of moisture, and also that sodium based reagents had a genuine action
which was almost double that of calcium based sorbents (and that these could be
introduced as solutions by means of simple nozzles).

It was mentioned that Rekingen Works had modified its raw mill throughput so that it
operated 24 hours daily, thus capturing SO 3 to conform to emission regulations. The
Santa Cruz Works of Lonestar

may do the same. In this class of activity, CIMENTS

FRANCAIS works on the basis of 50 to 75% absorption.

ENCI gave an account of experience with operation at different oxygen levels to reduce
SO2 emissions from its two-stage preheater kiln. The degree of sulfation of clinker at
Maastricht is 125%. This, along with other causes, produces an emission which must be
reduced to comply with new regulations During 3 weeks the oxygen level at the kiln
exit was altered from 1.5, 2.0, 2.5, 1.3, 2.0%. The results for SO2 level in the emitted
gases and SO3 in the raw meal and clinker are in agreement, showing that emissions can
be reduced. If the effects on quality and on kiln operation are acceptable, they intend
to buy new fans to guarantee sufficient oxygen level at maximum kiln output (with
fuzzy control). The re-installation of Magotteaux stirrers in the kiln has once again
given positive results after the last stop.

At OBOURG

there is an emission problem similar to that at ENCI, which can be

resolved by a kiln exit oxygen level of 2 to 3 % but this could give a too high a chain
temperature and an unacceptable reduction in output. Oxygen has been added
(1,000 m3/h either by the primary air channel or beneath the flame) gaining 3 to 4
tonnes per hour of clinker at an acceptable chain temperature. The cost of oxygen is
some 3 Belgian Francs per cubic metre for a permanent irstailation, but this could be

This reqorr is snict(y confidential within the Blue Ckfe Group.

offset if cheaper (higher sulfur

fuels can be used. It seems that at least a quarter, and

nore usually about a half of the SO2 disappears between the kiln and the stack, no
doubt by capture on dust. Given the large volume of gas produced by this wet process
kiln and its moist fuels, a new fan would be proportionately much more costly than for
ENCI.

At Obourg it seems likely that the longer kiln can satisfy emission limits through
control of excess air levels. For the other kiln, another method of reduction of peaks
of SO, is being studied; NaHCOS injection in the exit gas duct at the upper end of the
kiln. The trial installation from Solvay (about 400 kg/h of Na.I-IC03 powder) was leased
for longer trials. Chemical efficiency is 100%.

As already noted ITALCEMENTI

is looking at a Monsanto system involving a water/meal

scrubber for SO,, with a cost expected to be only 20% for that of an Untervaz system-

LAFARGE has studied sulfur behaviour on a semi-wet Lepol grate. In the hot chamber
there is an excellent capture of sulfur coming from the kiln, but there is also
decomposition of pyritic sulfur on the grate. This starts at the transition from the cold
to the hot chamber (500-600'C)

and is completed by the middle of the chamber.

TC9-4049

5.

15

Tecfioical Centre. Senrember 1994

TRACE ELEMENTS

lTALCEMENTl has presented a summary of results from 30 kilns, seeking to determine

the amounts of 16 metals In stack dusts, and also of 5 inorganic micropollutants.
Vibo works (precalciner),

At

the raw meal is dosed with CaF2 to influence the

decomposition of strontium sulfate and so limit the undesired effects brought about by
SrO during alite formadon. On the Lepol process CaF2 also provides a less dusty
clinker and a reduced need for kiln system cleaning. No changes were noted in the
behaviour of other halogens or of alkalis.

OBOURG has provided, a list of balances over 2 years. BLUE CIRCLE has shown a
table for retention of various elements in a number of kilns as a percentage of the
quantity brought in by the kiln feed (including recycled dust). An examination of Blue
Circles conclusions regarding behaviour of trace elements considered elements as being
either non-volatile or partly volatile. 3-5 results were available for each element for
the wet, semi-wet and dry processes. The percentage of the input found in stack dust
was very low for the non-volatile elements (As, V and sometimes Cr - unless there was
enrichment from refractories), thus reflecting tie good efficiency of de-dusting and the
varied additional contributions

from fuels. The highest proportions escaping with stack

dust were for cadmium, lead and thallium in semi-wet kilns. The levels noted were
influenced by the rate of removal of intemediate

dusts produced in the kiln processes.

For the dry process there was less enrichment of cadmium and lead.

We have specifically looked at German work previously published in ZKG, where the
need to examine the chemical combination of elements is underlined. Mercury,

for

example, must always be oxidised in a kiln system, so that the vapour pressure of the
uncombined element is not to be considered.

We have also received a supplement to the bibliographical list established in Autumn

1991, which concerns cold and trace element cycles. The published literature makes it
quite evident that cement kiIns are reputed to be potential origins of emissions of SO,,

Tl, Pb - and perhaps Cd and Hg, if these latter are found in the region. There is also
a certain interest in the upgrading of kiln dust,

TIC re,vrc is suictiv wnfide!zu*al

witi tie BILE Cric!e

G~OLT.

TC94049

17

Techical

Cam-e, ?&xBder i99d

FUTURE WORK

The Plenary session has asked for a new sub-commission to be set up in 1994/1995 to
follow the topics left from the existing group, with the addition of a study of corrosion
and the wider application of the mathematical model already developed. We must avoid
repetition of work already done, and coverage of subjects which the Maintenance Group
and the new Working Party on Flames are currently examining.

(For corrosion,

depending on the activity of the Maintenance Sub-Commission, we might envisage a

search for methods of reducing its effects, as well as the identification of compounds
which have a strongly corrosive action).

Our

group

is convinced that there remains much to be gained for the companies of

CETIC in bringing together the specialists who are concerned with Chemical phenomena
within kiln systems

and avoiding the study of topics which are more or less legal.

All are invited to consider a new division of chemical and engineering work with regard
to cycles, transport, clinkering, emissions, interaction with refractories, combustion,
sampling (gas and solid), etc etc.

A review by the Sub-Committee has produced a list of ideas, (see following page), these
points generally relate to factors which have an impact on quality, cycles, transport,
clinkering of emissions, refractory attack, combustion, sampiing of gas and solids etc.
Some of them are more relevant to groups in the Technical Commission. These ideas
remain for immediate discussion.

(NOTE :

It was subsequently agreed to concentrate initially on topics related to the

behaviour of sulfur).

C P Kerton
Technical Centre
May 1994.

LIST OF TOPICS

Environment:
Factors effecting organic emissions and their control.

Process:
Water injection in planetary coolers.
CO SO2, CaS04 balances.

Control of kiln build-ups and their chemical composition.

Effects of secondary firing on volatilisation/condensation

in Lepol grates and

their control.
Methods of successful use of even higher levels of S in fuels.

IMethods

allowing the retention of SO3 in clinker.

Control of chloride volatilisation.

Applications of CO, analysers.

Determination of dust cycles in kilns and heat and mass balances for each
preheater

stage.

Optimum fineness of petroleum coke for kiln and precalciner burners.

Effects of volatile materials on the long term and short term stability of
kilns and their consequences for static and dynamic conuol strategies

(allowing early pro-active responses t o change).

Control strategies (anticipatory control) to restore operation of a disturbed

kiln (effect on throughput, environment, quality etc).

Bypass control.

Effects of geometry on cyclone operation.

Effects of volatile cycles and dust cycles on cyclone dip tubes and on various
refractories.

Control of sulfur cycles at low temperature.

Effects of heating and cooling on refractories and linings; success with rapid
regimes.

Effects of parameters other than CO on sulfur volatilisation.

Correlations of SO2 signals with other process parameters.

Effects of injecting and additional fuel at different points in the process (for
example solids at the kiln feed chute).
Distribution of trace elements throughout kiln systems, and methods of
control.

Interaction of volatiles and refractories.

Influence of SO3 on refractor+ life - both direct (chemical) and indirect

(perhaps a lower BZT).

Correlations between SOS and free lime in operating kilns.

Effects of V and Ni (coming from coke) on refractories.

Correct regulation for burners for different levels of coke fineness and coke
mixtures.

Maintenance:
Corrosion in colder zones of kilns (in relation to Cl and S)

Workplace hygiene and safety aspects from the point of view of volatiles

Product/Quality
Treatment and use of dust rich in volatiles

Methods for the internal use and/or upgrading of dusts which cannot be
dumped.

Effects of V and Ni (coming from coke).

Effects of SO3

and of sulfate/alkali ratio on clinker quality at different

levels of free-lime (is there an optimum level?)

Effects of marginally reducing conditions on quality.

Effects of halogens on cement behaviour (standards, etc).

Effect of clinker size grading on quality.

cix/tla
13.9.94
&4:x94.049

RT lo9 pot 1b.Col)

:A
-.L---

-L- -Ii---.1; -_
I. 1

Tempercrure t
0

2 0 0

coo 6 0 0

6 0 0

I0co 1209 ISGO I~0 la^OO

-/. i .I. .
zuqo 5-l. :uq,sa iForrrn.e
-157

j-2

oxides.

.-- l..,*--l---*-l.,-~l
k

; .-...--

. f ( .*--

* . . .

I ***-I---*!~i---I -i-y---I~--l--

i;:
I- J
, ,

0
L
(I

LUU

4uu

: Ill1 III I Ill1

1000 1200 -Pm 7600
Eta
BOO

Free cncrg diagram for sulfates.

-20

/
I

T2mgrz!lLii2, Y

fret energy diasram ior chiorid:s.

j
:

1
1

-no -

Tempcraturc,=C

Fret cncrgy diagram for fluorides.

\
\
..

I
i

t
\
,

450
400

350
3 0 0
zso
2 0 0
7.50
1 0 0
s 0
0

20

30

SO

40

60

7 0
Dhbor

3-90

- -43.447

8 0

90

1 0 0

R-t

- .a65

110

1 2 0

13*

<l/mn>

4.042* ~&bor <i/mn);

Km

so3

Cl

<125fr

13

5.2

7.29

>125p

4.3

1.9

1.6

Ensemble

12.6

4.95

7.03

45op

3.6

4.4

2.3

>lSOp

3.1

Ensmble

5.3

4.2

2.1

Cl:

2.1

2.3

0.7

0.8

450
400

350
3 0 0
zso
2 0 0
7.50
1 0 0
s 0
0

20

30

SO

40

60

7 0
Dhbor

3-90

- -43.447

8 0

90

1 0 0

R-t

- .a65

110

1 2 0

13*

<l/mn>

4.042* ~&bor <i/mn);

Km

so3

Cl

<125fr

13

5.2

7.29

>125p

4.3

1.9

1.6

Ensemble

12.6

4.95

7.03

45op

3.6

4.4

2.3

>lSOp

3.1

Ensmble

5.3

4.2

2.1

Cl:

2.1

2.3

0.7

0.8

ENCI
RELATION MG/M'.SO~

- %

02

1. General
Information is presented here against the background of equipment/processes encountered outside the
cement industry, where acid gas abatement may already be practised - and where ideas for transfer
to our industry may originate. Our most frequent needs are to improve SC+ capture by calcium
compounds in raw mills and/or compensate for the absence of this absorption when mills are not
running. Whilst there is a 50 to 75% reduction in SC& levels in a number of cement Works when
hydrated lime is suitably added, some sites need to understand why they record drops of only 20 to
40% and there also is an interest in better understanding the possibilities for using of alkaline
sorbents. This note aims to provide suitable background information as an aid to better understanding
when the complications and costs of the Untervaz Solution may have to be accepted.
2. Sorbent Injection
2.1 Basic Description of Technology and Principal Variations
Sorbent injection is used primarily in pollution abatement as a means of reducing emissions of sulphur
dioxide (SO-,> and other acid gases, such as HCI and HF. Material is usually injected into the gas
stream as a fine powder, where it reacts with the acid gases, generating a dry product for collection
in dust arresment equipment. Dry injection methods are particularly suitable for small boiler and
incinerator plant or retrofit applications where the capital expenditure for other systems is prohibitively expensive. The efficiency of S&- removal is 40 - 80%, depending on the sorbenc used
(most commonly calcium and sodium compounds).
The sorbent can be injected at various points in the plant, according to the temperarure and conditions
at which it is most reactive. The most common systems for boiler plant are:
Furnace injection of calcium based compounds
Heat exchanger injection of hydrated (slaked) lime (Ca(OH),,
Post furnace injection of Ca(OHk at relatively high humidity
Post furnace injection of sodium based compounds.
In a cement process, we may see:
Classic SG- capture in the lower stages of the preheater
Hydrated lime injection at top of preheater or in conditioning tower (probably as slurry)
Return of calcined meal to cooler parts of system (preferabiy with moisture)
Wafer injection at suitable points to increase possibility of reactions with raw feed
Hydrated lime addition to raw mill
Injection of alkali compounds or solutions to exhaust gas duct
SG- capture by limestone in raw mill in the presence of moisture.
(Some different possibilities for injection are envisaged for Lepol systems.)
2.2 Principle of Operation
The reaction between sulphur dioxide and dry sorbent is a heterogeneous one . (Reactions with moist
sorbents are discussed later in Section 3.2). SC& molecules diffuse through the gas stream and are
adsorbed on to the sorbent surface before diffusing into internal pores where chemical reaction occurs.

These mechanisms are exemplified in the classical kiln/preheater sulphate cycles.

Dry calcium sorbents react with sulphur dioxide as follows:
The first stage is calcination:
CaC03 + CaO +CGCa(OH)z + CaO + Hz0
(These reactions occur at temperatures greater than 760 and 570C respectively. Dolomitic limestone
starts decomposition at a lower temperature.)
The second stage is sulphation:
CaO + SG_ + Y.Q + CaSO,
With excess oxygen, complete oxidation to sulphate occurs at temperatures ahove 800C. Below this
temperature a mixture of sulphate, sulphite and sulphide is formed. (The optimum temperature range
for direct reaction with hydrated lime is listed as 130 - 180 C.)
Sodium compounds react with SO, as follows:
2NaHCO,-Na&O,

+CG- + Hz0

Na&O, + sa* + ha- -, lN&-so, +co,

Sodium bicarbonate decomposes to sodium carbonate, which then reacts with S%- to form sodium
sulphate. The these reactions are significant at temperatures above 130 - 180C; at the lower
temperatures in this range the favoured SO, reaction is directly with the carbonate, but at higher
temperatures the thermal decomposition and sulphation reactions occur simultaneously. It is thought
that the good performance of this sorbent may be explained by the fia that a shrinking core of
bicarbonate is continuously decomposed, providing moisture at a fresh reaction interface for S%arriving through a permeable outer shell of sulphate A similar scheme is seen for potassium.
23 Selection and design considerations
Dry sorbent injection is usually one of the cheapest abatement options for SG- removal, particularly
for small or retrofit plant. as the capital cost is low. The choice of sorbent is a prime consideration
and depends very much on availability. Calcium compounds used for dry injection are primarily
naturally occurring limestone or dolomite or hydrated compounds derived from these raw materials.
Reactivity is dependent on pre-treatment as well as natural properties. The sodium sorbents of
interest are sodium bicarbonate (NaHCO,) and sodium sesquicarbonate (NaHCO,.Npl_C0,.2H,0).
These occur naturally as nacholite and trona respectively. In making a choice, process economics are
highly dependent on delivered price of reagent and rate of use, despite the initial lower cost and
comparative simplicity of operation of the dry post-furnace injection processes.
There are three temperature windows for calcium sorbent injection in boiler and incinerator systems,
which also broadly apply to cement kilns. (Note that there is a temperature zone where neither group
of reactions is very effective, especially in dry conditions.)
Calcium sorbents can be injected directly into the furnace, where at temperatures of 1100 1250C the calcination and sulphation reactions can occur,
Calcium hydroxide Ca(OHX_ will react with SO, at about 550C and hence can be injected
before the heat exchanger.

At high levels of humidity calcium hydroxide will react with SO, at even lower temperatures
(5 - 15C above the saturation temperamre of the gas, as discussed in Section 3.2). water
can be injected with the calcium sorbent into the duct between the heat exchanger and
particulate abatement equipment or elsewhere
Sodium sorbents are also injected in boiler plant to the duct between the heat exchanger and
particulate abatement device, where the temperature is in or above the range 130 - 180C.
It is possible to inject sorbents at several points in the plant and to combine this technology with other
abatement options. (No accounts of multiple injection systems are known for the cement industry).
Handling problems may be encountered and not all sources of lime are equally effective in reacting
(at a given fineness) in the time available at the point in the process where introduction is feasible.
Reactivity increases as the surface area of the sorbent increases (particle size decreases), up to 40
m3/gm. The calcium:suiphur ratio is generally set at 2, but can be as high as 6, particularly for low
sulphur coals where the mass of sorbent is still comparatively low.
The reaction efficiency for limestone ranges from 40 - 50% at calcium:sulphur ratios of 2 - 4.
Dolomites give greater conversion efficiencies, and this is attributed to the more open structure of the
sorbent material which enables greater diffusion of gas into the pores of the sorbent. Conversion
efficiencies of 70 - 80% are achieved with sorbents of hydrated lime Ca(OH), at 8OOC, and sodium
compounds also yield conversion efficiencies of 70 - 80%. (Use of a fabric Nter for particulate
abatement is claimed to enhance SO, abatement efficiency by ca. 10% because the sorbent collected
on the filter bags continues to react-with SO, during particle filtering. Dry sorbent injection to the
filter bags after a cleaning cycle, is claims as an altemativelsupplementary method to enhance
reaction.)
Superstoichiometric quantities of reagent are usually needed because efficiency of reaction is low, and
recycle may also be required. Efficiency may even be insufficient for elimination of either very high
or very low SC& concenrrations at a realistic stoichiometric ratio. Use of conditioned (ca. 10%
moist) hydrated-lime has been claimed to give improvements due to (a) breakage of particles when
brought into contact with hot gases (so generating more surface area) and (b) cooling, which increases
reaction efficiency. There are no known reports from the cement industry on this point, or on
reactivity of different sources of limestone or dolomite.
Limestone is the cheapest sorbent material currently in use. Lime (calcium hydroxide) is about 5 6 times more expensive than limestone and trona/nacholite are generally ten times more expensive
than limestone. Some studies suggest N&HC03 becomes still more efficient at higher temperatures
( e.g., up to 815 deg C); it costs 2 to 4 times as much as hydrated lime. although consumption may
be lower and there may be less residue to dispose of. (There is little or nothing known about the use.
of alkali compounds at relatively high temperatures in the cement industry.)
Combustion systems using high sulphur coals yield the most promising results where SO, levels are
2000 - 4000 ppm. At SC& concentrations < 1000 ppm the reaction is diffusion limited and it may
be more difficult to achieve desired levels of efficiency. (In general, this technology is considered
to be less efficient than the wet methods described later.)
In boiler systems, the added sorbent and its interaction with the fly ash, can cause fouling of surfaces.
Also, higher particulate loadings, decreased particle size and increased electrical resistiviry of the
particles can impair the performance of collection devices. Handling and disposal of larger quantities
of solid waste with properties different from fly ash or conventional scrubber sludge can be difficult
and increase costs. For example, sodium salts are soluble in water and hence disposal requirements
are more stringent.

3.1 Future Developments

Research is continuing to enhance knowledge of-the appropriate mechanisms acting in these injection
processes, with a view to developing alternative, moreeffective sorbents. For example, alkali metal
additives in limestone enhance SO, abatement efficiency and early indications are that lime-containing
waste materials, such as carbide mud and sugar mill mud, react faster and have a greater sorption
capacity. The use of regenerable sorbents such as calcium silicates is another possibility.
3. Spray Dryers
3.1 Basic Description of Technology and Principal Variations

Spray drying is a standard chemical engineering operation used to produce dry powders of controlled
particle size, density and moisture content Spray dryers are used in pollution abatement for the
control of acidic species in a flue gas stream. Droplets of reagent are contacted with the flue gas in
a reaction chamber - probably a modified conditioning tower in a cement Works. Liquid is
continuously evaporating from the droplets in the chamber during the neutralisation reaction and the
dry reaction product can be collected at the base of the chamber or in the dust abatement plant.
A complete system consists of the spray dryer (atomiser and reaction chamber), associated slurry/
liquid handling equipment, a particulate collection devise and soiids recycling equipment. There are
three types of atomiser in general use: rotary, two fluid or spray nozzles The reaction chamber can
be a tower or dust, and the flow of the droplets and flue gas stream are usually co-current. Lime
slurries are most often used, but sodium carbonate/bicarbonate solutions are also acceptable.
32 Principle of operation
a) Lime spray driers

The atomiser generates dropiets of lime slurry which are injected into the flue gas stream in the
reaction chamber. In the capture of sulphur dioxide, the chemical reactions which occur involve
water and are believed to be:
Liquid phase:

CaC03 + Sa- + %H,O -. CaS03.%Hz0 + Ca-

Gas/liquid phase:

Ca(OH)l

+.Sa- + Hz0 - CaSO,./iHIO

+ lXH,O

Sa- is absorbed in the aqueous phase of freshly atomised droplets forming suiphurous acid,
where the reaction of SO-, with lime or limestone proceeds rapidly, forming calcium sulphite
which may later be oxidised and form gypsum in the presence of oxygen and water.
As the droplets pass through the chamber, water evaporates to yield a porous particle which
has a dry surface but a wet interior. Sa- diffuses into the wet sore of the particle and the
reaction continues.
The reaction of SO, with lime in the absence of any moisture is slow. Consequently, in order
to extend the reactivity of the lime in the unit, the temperature near the exit is maintained just
above the saturation point of the gas.
As mentioned earlier (Section 2.2), these reactions can be involved in sorbent injection in cooler parts
of a cement production line, for example when Ca(OH)z is injected to the preheater or Sa- reacts
with limestone in the raw mill. In the absence of water, however, the reaction rate will be very slow
- for example at the top of a preheater tower. Water injection to the preheater at Santa Cruz (without

adding any extra lime) was reported to allow 10-20% reduction in SO, levels. Failure to compare
humidity levels and/or use fresh lime may account for several differences in experience of SGcapture in kiln systems.
b) Spray driers using sodium salts

Dry S&- reacts with sodium carbonate/bicarbonate from low temperatures: and hence the requirement
to enhance the reactivity by stringent control of temperature and humidity is not necessary.
3.3 Selection and Design Considerations
The principal design parameters for spray dryers are droplet size and distribution, and inlet and outlet
temperature. Multiple atomisers are used in order to achieve an even distribution of droplets in the
reaction chamber, and the droplet size has to be such that the rate of evaporation is fast enough to
prevent formation of scale in the reaction chamber as droplets/particles strike and stick to the walls,
but slow enough to enable the reaction to occur. High inlet temperatures enable more water or lime
to be injected, and low outlet temperatures (slightly above the saturation point of the gas) optimise
the abatement efficiency of the spray dryer.
Fine sprays and concentrated reagents have shorter drying and reaction times. Water evaporates from
concentrated reagents rapidly and hence the neutralisation reaction occurs mainly between the acid
gas and the porous particle. Fine droplets ( < 100 pm) are used with size tailored to the residence
time of the flue gas and droplets in the chamber or duct. The residence time in a chamber is usually
in the range 5 - 10 secs, with droplet size < 100 pm. For injection of slurry into a duct, reaction
and drying times of 1 - 2 secs are typical. Residence times and evaporative heat available in an
existing cement plant conditioning tower or gas duct system may limit the amount of SO, which can
be scrubbed.
The choice of sorbent will depend on its cost and availability: sodium salts give better once through
efficiencies, but lime has a cost advantage over trone/nacholite and the calcium based reaaion product
is insoluble in water which renders disposal easier, should this be necessary.
Spray dryers have been successfully used in Europe for controlling emissions of acid gases, primarily
for combustion plant and incinerators, using a lime sorbent which is recycled to improve its utilisation
and achieving abatement efftciencies of > 99% and > 90% for HCl and SG- respectively.
Efficiency can be enhanced by increasing the stoichiomeuic ratio for specific conditions of
temperature and humidity, but the gains are limited by sorbent utiiisation, sorbent solubility and waste
disposal costs. The Ca:S stoichiometric ratio is typicaiIy in the range 1 - 1.5 and liquid:gas ratios
in the range 0.027 - 0.04 l/m3. Spray dryers offer several advantages over wet scrubbing, especially
the fact that a dry product is formed which is easier to handle and dispose of than a liquid effluent.
The capital cost, maintenance cost and energy requirements for the spray dryer system are lower than
for wet scrubbing plant although reagent costs are higher.
The particulate collection device can influence the operating conditions of the spray dryer. Acid gas
removal can continue in a fabric filter but care has to be taken to prevent blinding of the bags.
Electrostatic precipitators, however, can operate at temperatures nearer to the saturation point of the
gas, hence the spray dryer outlet temperature can be lower which improves its abatement efficiency.
The dry product from the spray dryer (hydrated &SO,) can be used for landfill, processed to yield
anhydrite or pelletised to yield synthetic aggregate.

3.4 Scrubbers
Absorption is a process which involves mass transfer between a soluble gas and a solvent in a
contacting device; chemical reaction may or may not occur In process design, both the chemistry
of the system and the physical structure of the equipment must be considered. Unfortunately, water
alone is not effective at removing SO, from a gas stream because (unlike HCI) it is not very soiubie:
an alkaline solution is needed The driving force for gas removal is the difference between the partial
pressure of the soluble gas in the mixture and the vapour pressure of the solute gas in the liquid film
in contact with the gas. Mass transfer occurs by molecular diffusion across the interface and the rate
determining step can be in either the gas or the absorbent phase. When the gas is very soluble or
reacts chemically with a reagent in the sorbent, the process is gas phase controlled
Trace gas removal systems can be categorised by the solubility of the gas and by the reactivity of the
system. SO-, is classed as moderately soluble in water (I-IF and HCI are very soluble), and sodium
sulphite and-alkaline compounds are used with some success as additional reagents. Efficiencies of
SC& removal of some 99% are attained in appropriate circumstances. There are no known accounts
of the use of sodium sulphite solution for SO-, capture in the cement industry: usage in power stations
generally appears to be associated with systems which treat the resultant chemical products to
regenerate the solution of sulphite sorbent, or systems in which the sulphite is used alongside other
reagents, providing an initial capture of sulphur as an alkaline compound for subsquent displacement
reaction to form a more readily disposeable or saleableby-product.
Gas absorbers which attain gas/liquid contact by bubbling dirty gas through a liquid are suitable for
absorption processes which are liquid phase controlled and those which involve spraying liquid
through the gas stream are suitable for processes which are gas phase controlled. As absorption
is a rate process, the concentration gradient (driving force for the reaction) and the (high) surface area
of contact between the liquid and gaseous phase are crucial design parameters. The surface area is
determined by the packing material or droplet size and this is usually achieved using packing materials
which are coated with liquid or by droplet/bubble formation. The absorber design also has to provide
a means for renewing the liquid absorbent so that a high driving force for mass transfer is maintainedGas and liquid flow rates and pressure drop across the absorber influence the driving force, the
efficiency and in some cases the surface area (droplet formation). A good gas absorber design
removes as much pollutant as possible in as small a space as possible. The choice of equipment
depends on the abatement efficiency required, the energy and reagent requirements and the properties
of the dirty gas stream.

c. P. KERTON,
Blue Circle Te&nicai Centre, Greenhithe,
&lay 1994 (Updated, July 1994).

- I
r

l-EMPERATURE
C O N T .~R O L
WATER

FLUE GAS BYPASS AROUND DRYER

I- - - - - - - - - - - - - 1
I

I
I

FLUE GAS
~no:.i POlLEn

+ C

J--J-

SPRAY DRYER
WASTE RECYCLE

4-J

FABRIC
FILTEH
CATCH
RECYCLE

E A N
FLUE GAS

TO ASId
DISPOSAL
BIN

rBALL
cc: M I L L
_STLAKER

DETEN.TlON
SLAKEFI
NOT SHOWN
a-l;.-,,
FEED PUMP

APPENDIX 2

TEXT FROM INTERNATIONAL SYMPOSIUM ON GAS CLEANING

AT HIGH TEMPERATURES
Behaviour

of Volatile Materials in

Cement Kiln Systems

GAS CLEANING AT
H I G H TEMPERATURES
Edited

by

?
q
v

snlls o f riinny clicrnicnl

(ltrc: usit ccnIc11I
iidilslry convcnliun is loltuwrtt iii cxprcssinfi I
atialyscs ill Icrrris
of cIxitlcs. c.g.. CaO, S O , , hl,O,. clc., u s u a l l y 018 a loss [ICC b a s i s . i.e.,
allcr allowing, fnr ihc Ins5 ill wciglrl cvcnlually crpcricflcut rluc l o tlcslruclion o f clrbonalcs.
clc.. during Iml lrcnlillcnl.)

\01,A~111.1~~S5

ANI) hllXTIIANIShlS

01; VOI.A1II,ISA-IION

1.11~ pritlcitd volnlil~ ~ICIIIC~IS arc K, N;I, Ct. S. I n Ihc cast of r a w nnlcrints, ccrlain sutrur
cor~~t~~ur~d~
(rulfirtcs or ort:anics) c:ul rratlily clcco~~rt~sclvolnlilisc
Indow GooC, bul IllOSl
vdalitc co~ntmunds in raw inalcriids o n l y c v a p o r a l c pnrlially and a l higtlcr Icnitxralurcs as
111~ reed pa5scs low;rrtls llrc k i l n hrrnilig xonc. Ihc rcsitluc rcmins i n lhc producl. cilllcr
i n s o l i d ~cd~~iou ill IIIC plincitd phscs ol IIIC clinker o r as cliscrcic cocrqm~ncls. Whilst
alnml all fcctl cllloritlr. w i l l cvatwmlc, lcs.scr aniounls 0r ollicr c o m p o u n d s do s o , wliilsl in
corrlr:itl. liitd vcrlalilrs arc :ilii~o;l ;ilw;iys ciilircly cvnpor;ilcd during cmiiIiIislion.
lfv;rtKrrarrd volarilc~ rravcl hack up 11rc kiln wilt, 111c comlm~lion
irmrt;;rnic comt~ountl; (lilrcr;rlirrg I:~lwl Iical):
i)

ii)

iii)

on llic Ucctl.

li)rmitlg lhc his

Or

n rccirculaliiig

eases and cordcrrsc a

inlcrnal votnlilc

load

n ;I lirrc rlrrsl r>r rcllrlc which i s fin;llty Irappcd i n lhc g a s clrxning s y s l c m o r

r a w m i l l arrtl hx~mcs p a r t or an cxlcrilal vnlalilc c y c t c . a s 111151 i s parlly o r
d$ly rllurncd ir1 tlrc sysic111
011 coltlcr strrl:rccs

ill lhc syslcrii. linming

llic Ixrsis

Or

hiltl~ups.

Ircssurrs 10 cxploil cvcr more marginal rcscrvcs

0r r a w ni3lcrinls a n d fuels give rise l o
irrcrcxirrg
I;rmili;rrily willr llrc drm o
f
vohlilc spccics
011 process pcrlormarrcc. Wtrcri
corrtlcrrsctt
volalilcs rclurrr
low;lrtls Ihc b u r n i n g zorrc, clctxriding o n Ihc overall clrcmicnl
cr~nrliliorls
and hrnint: ccrndiliorrs,
llrry form a range or votalilc cnmpunds
wlricll
tlrcrr0clvcs cvapw;llc parli;rlly nri0 llrc cyctc only lids an utuilibriuni when ltx loiat quaillily
Icarirrt: llrc syslcm (itI clirrkcr ;md mm-rclirrrrcd tlusl) equals ltral cnlcrirrc
lhc s y s l c m .
Alk;rlix

;md ~IIII~IICS ctrl~.rirrt!

llrc trrrrrrirrp
forrc irr trraclirc I:rrt:cly Irlrrri
:I \qr;lr;llc mr~llcr~
tll;lr<, iiumiscilrlu
wills Ilrc
prLicip;rl fcrrilc Ilux.
Tlrc
lrvct or volalilcs irr
rccircul;rliorr
i
s
sit;rrilic;mlly grcalzr ll1911 lhcir lol;d ralc or irrlroducliori am1 lhc s~rbsln~~ccs
it1 llrc vapour ptrasc cnri 115 irr v;rrious slalcs or dissoci:llion a n d
recomlrinalion. III gcncml,
Itrc prclcrrd clrloritlc corrrt~rmnd
i s txrl:rsrium cllloritlc and only wlicrr lhcrc i s an cxccss I I I
clrloriric lor cOcmic;rt
crmrlrimrliori
will) pol;lssium w i l l sodium ctrtoritlc lx hrmcd as : I
rccircutalirrg vohlitc strxics.
Alk;tli sulhlcs (Na2S0.,. K,SO,t cvatxxatc cony,rucnlly,
tlisnprxzlring c n l i r c l y wticri Ilr;rlctl Ihr a loiig period. Cn.S04 ttcconiposcs arid hvcs rcsidu:d
CnO (ill, trxitlising contlilior~st. s o lh:il CaO:CnS04 niclls c a n rorcll.
T y p i c a l rccirculalirrg
vrhtilr lrds rxtlrcssctl a3 XI ul llrc 1~1lal quimlily inlrotluccd a r c a s roilOws:

5llll;llc

(lrlrrridr(s)
K-0

XXKt
XX) . rLv1

Na20
SOJ

1.

IS0 - 200
2uo . R(X)

Ic!u.utr,r = 0

Jle~su[

??V
;oJluo-J

/S?(?J(U! I!2 Jl2Uy43

UO!lJJ(dS/U%!SJp SSJJOJd

JAoqe sv

UO!lZO[2S/U~!SJ,, SSJJOJd

pJJJ M2J JO UO!lJJ[Jj

S%U!llJS
Xl2~/JJUJfl~
(OyJ gz~pzpl2#ls)

pgos) [XI! JO JJ!O~~

ssauauy J~O~/(ZO-J
~s3!lsplJeJzq3
[xl,

XllU pJ! - (Opl ![e~[e/J1zjpi)

.illS!UfJ~J PJSJ m2J J O UO!UJ[JS

*(se3 p!nbg

.~JlS!UJJqJ
3J!lS!JXJUeqJ
[XI-J
SoU!llJS JJUlq/JUl2[~
%J!UJflq pJ[[OJlUO3
tunq 1~0s

+~JX![ZJJU+I iy S:X~U pg
SNOil3V

?OtIlN03 TiBlSSOd

r!leqle

UJn,Ji ;Sa

dn-p[!nq-!wz

.iJlJtJJOJa

ra3ipqos

mz;q;rJd JO UO~I!SO~UJO~
uo+?l2do
u:1s3p ::JU!J[ZJJld
:uto?o[

luy
SxZJ[U!

:!O/SOJ:Z p[OJ
(lJ2qS)
[OJilJJ.\
WJlS.\S IJI2JqJJd

U&SJp

xrll-i

lJ!f!q~

JJqlx[-~

u! uo!ysodtuoJ ![i?([v

au02 a.l!Lunq

JnCd2.\ JJiZ.I\
:J!Jt-ipJl
.i([?-O(
&J!JtTpJJ
s+?qO@
JJJqdSOulz 7uoz dU!ung

hp3d JZ!S

(*(l!suap) IU71UOJ

(.SOu SZ%) .(JUJ!JlJJJ [tULlJql

JUOZ

%!+fnq ;o J[i_rofd .dwaL

(SJJnSSJJd
Jl-dZs\)
.dmin :uoz &J~HI~

SUO!l?!L?.\
[JQ[

Ic!u.utr,r = 0

??V
;oJluo-J
r!leqle

UJn,Ji ;Sa

mz;q;rJd JO UO~I!SO~UJO~
uo+?l2do
u:1s3p ::JU!J[ZJJld
:uto?o[

ra3ipqos

.iJlJtJJOJa

luy
Jle~su[

/S?(?J(U! I!2 Jl2Uy43

SxZJ[U!

UO!lJJ(dS/U%!SJp SSJJOJd

dn-p[!nq-!wz

:!O/SOJ:Z p[OJ
(lJ2qS)
[OJilJJ.\
WJlS.\S IJI2JqJJd

U&SJp
UO!lZO[2S/U~!SJ,, SSJJOJd
JAoqe sv

(.SOu SZ%) .(JUJ!JlJJJ [tULlJql

(*(l!suap) IU71UOJ

XllU pJ! - (Opl ![e~[e/J1zjpi)

.illS!UfJ~J PJSJ m2J J O UO!UJ[JS

*(se3 p!nbg

xrll-i

hp3d JZ!S

lJ!f!q~

JJqlx[-~

JnCd2.\ JJiZ.I\
:J!Jt-ipJl
.i([?-O(
&J!JtTpJJ
s+?qO@
JJJqdSOulz 7uoz dU!ung

pgos) [XI! JO JJ!O~~

ssauauy J~O~/(ZO-J
~s3!lsplJeJzq3
[xl,

S%U!llJS
Xl2~/JJUJfl~
(OyJ gz~pzpl2#ls)
.~JlS!UJJqJ

au02 a.l!Lunq

u! uo!ysodtuoJ ![i?([v

pJJJ M2J JO UO!lJJ[Jj

3J!lS!JXJUeqJ
[XI-J
SoU!llJS JJUlq/JUl2[~
%J!UJflq pJ[[OJlUO3
tunq 1~0s

+~JX![ZJJU+I iy S:X~U pg

JUOZ

%!+fnq ;o J[i_rofd .dwaL

(SJJnSSJJd
Jl-dZs\)
.dmin :uoz &J~HI~

SUO!l?!L?.\
[JQ[

SNOil3V

?OtIlN03 TiBlSSOd

~cmpcraturc in a WCI prnccsr k i l n , a considcrablc dab-bank c l I t.-6urcrncnls

1115 been buill
up. The USC o f lnwrr a s h rucl lowcrcd c l i n k e r K,O l c v c l b y O.lS%. dcspilc 111~ inIroduclior1
of n IiIIlc mnrc p&ssium IO lhc s y s t e m ( a n cxlra 0 . I 16 o n clinker). T h e dusl - rclurncd IO
lhc kiln - had bccomc more rich in alkalis, so lhal Ihe proportion or K,O broughl in by solid
fuel fell lrom 2 4 % I O IX% while ihai brough1
i n b y dust rcIIIrn rose from 18% I O 2 9 % .
This s~rp.gcs~s
1ha1 KzO incotporaIion in clinker no1 only dcpcnds on 1hc quan1ily inlroduccd
11u1 also . a n d almvc all clsc _ o n IIIC type of malcrinl w h i c h b r i n g s i1 i n a n d perhaps o n 111~
pnsiIimr whcrc i1 is injccIctl. On screening 111~ clinker al 20 mm. chcmic~l analysis showed
a K,O coriccnlralinr~ snmc IO% Irighcr i n Ihc cn;~rsc f r a c t i o n . The clinker a l k a l i COIIICIII has
s~~cccssli~lly been rctlucccl
i n lrials b y c a l c i u m chloride atldilinri nl l h c Ilamc.

Cnsc 6: A hil:h chlnridc coal ( - 0. I5 % Cl) can only bc ~isctl as a mix will1 anolhcr coal lo
avoid Imild-ullc wilh Iyliically 2 % Cl a1 Ihc ImI1nm Or cyclone 4 (lhc lowcsl in lhc prchcalcr
~nwcr). hn;rlyTis of Inriltl~ul~s along IIIC kiln indica1ul chlnridc lc,vclr up IO 30% (al zero loss
nn ignirinn) in 111~ coaIing Iron1 lhc base of cyclone 2 and 20% a1 SO 111 inlo Ihc kiln (tlcspilc
iit hsi ih:m 5 % lrvcl i n Imllr lmi :md rold p;nIi 0r Ilic k i l n ) .

1m1 Inr some weeks will1 I h c objccIivc or reducing b u r n i n g ronc

b y - p a s s . ICSIS WL
volaIili.~~tinn b y p l a y i n g o n process paramcIcrs a n d prnducing a IighIly mincraliscd clinker
will1 higher volalilc rclcnlion. D u r i n g changcovcr ~hcrc w a s some I c n d c n c y l o rnrrn sol1
build-ups in Ihc prchca~cr, bu1 wiIh ihc new regime csIablishcd ihcsc movctl lowards lhc kiln
fcul chu~c wiIhou1 c a u s i n g m a j o r problems for k i l n opcniion. (EvidcnIly lhcrc a r c
phcrIon1cna of boll1 shorI-Icrm a n d l o n g - I c r m sIabiliIy: o n c e a slablc b u r n i n g zone
volaIili.saIion i s csIablishcd, in lakes Iimc f o r sIablc condiIions I O arrive higher u p the syslcm
and in Ihc Iargc masses or malcrial in lhc build-ups and coalings already in cxislcncc.) The
apparcn1 b u r n i n g ~nnc IcmpcraIurc w a s rcducul b y almu~ 1 2 0 C . while K,O volalilily
droplwl IrIlm 70% III 60% i n 111~ b u r n i n g ~onc a n d IhaI ol S O , Iron1 RO% Iowartls lhc ranl:c
SO% I O GO%, provitlcd lhnl k i l n cxi1 oxygen l c v c l w a s kcp1 a b o v e 2 % . Thcrc wcrc
in~provcn~cnIs in kiln outpul raIc and rucl consumpIion and lhc cxpcrimcnlal Works adoplcd
ccrlain Or Ihcsc changes during normal nl~ralinn
rOr scvcral years.
TCSIS wcrc c;Irricd 0111 invnlving variotrs N
O
, lcvcls ( t o
Cnsc I I:
IcmpcralrIrc) a s well as chcmic~lly rcrhlcing rh0C contliliruIs.
so, :

C:1w 7:

Atldinl; a sccontl ln~~hc;l~cr

c y c l n n c s~agc IO a l o n g d r y process k i l n ( a varinrIt of
1hc I:igutc I prnccss, will1 a higher k i l n IcngIh/dian~cIcr ralio a n d a single cyclone slagc
nhnvc i1) yicldcd vnrimrs build-up problems. To resnlvc ~hcsc,
IIIC kiln gas cxi1 0, lcvcl was
iricrcA5rccl
rrorll 0.5 In I .5 %, solid rd rcsiduc al 90 microns was rcduccd
lo below 25 % and
scvcral cmr~l~r~ss~d air blaslcrs wcrc inslallctl lo dislodge malcrial rrom lhe lower regions
oC 111~ prchcnlcr. T h c s c acrions improved IIIC siIuaIior1 and sul~scqI~cnIly addiIional mcasurcs
wcrc Iakcn: rcl~/l;iory slirrcrs wcrc addul l o I h c l i n i n g near the k i l n b a c k end, hurncr a i r
vcltXiIy w a s incrcawd. a n d a non-slick lining was inslallcd in 1hc kiln cxi1 gas duel and
cycl011c dip-lubes. Ihcsc Crrm mntlc bcllcr ouIliu1 ralcs pnssihlc wilhoul b u i l d - u p s .
IIc~cnIly. a Iiil;hcr snllur rd hlcntl 113s been nscd, :~ccomp;mictl b y s l a g (S - I X) :Imong
lhc r;~w niix coml~~ncnIs.
IrchcaIcr blockage problems rccurrcd,
bu1 b u i l d - u p s c;m be
avoidctl if the SO, l c v c l i n snmplcs t a k e n Irnm the k i l n cnlry m a l c r i a l i s kcp1 b e l o w 2.5%
b y lirnilirig lucl S conlcnl and sl;ig use i n lhc r a w m i x , provided lhal i n addilion l h c oxygen
lcvcl at IIIC kiln back cm1 is kcp~ consis~cn~ly a1 or above 2%. Cnmpui~r c0nir01 or ihc kiln
Iiclpr I n achicvc succcii. rctlucinl: lhc vari:rhilily ol l h c 0, s i g n a l .

cxw Ii:

I ;~li~n:ih~ry tl:11:1 on minor clcnic111s

cr;~n~plc, suITur vnl;rIiliIy i n ;I s~;~rrtlarrI
rcginrc
0, OUI I~lls in ihc prcscncc
Or 0,; r~~~~rihclcss
a l I2WC. The volalilily 0r m i n o r c l c m c n l s
powdcrrtl m a l c r i a l lhan rOr grarIulcs.

conlirm clvccls crlrxcrvr.tl

ill ln:tclicc. I:rlr
(70% N,, 3 0 % C O ) i s close I O 100% a1 0%
ihc CrrCd Or 0, is much ICSS al 1400C lhan
i n l h c laboralory i s a l s o m u c h grcalcr r0r

K,O a n d Na,O :

indicalc

llamc

Ihc r a l i o or S O , i n Slngc I V I n S O , i n r a w iri~l varied

Iypically Iron1 I.R IO 2.7 lnr 111~ higher lcvcls ol NO., and was
3 . 0 [or a IOW 0, Icvcl. The clinker S O , c~nlcnl fell.
In a parallel manner. for K,O 1hc ralin of l h c c o n l c n l i n S~ap,c
I V IO Iha i n r a w meal varictl from 3.X IO 4 . 4 a n d rnr NazO
rrom I .6 IO 2.0.

In gcncral, rultrcing cnndiIions

a lower clinker SO,.

incrcascd SO, lcvcl a1 Slagc IV by a

hClOr 0r

2, also giving

CXSC 1 2 : S O , 11~s l~ccn tmmit~rcd ;II lhc kilt1 Iwck cd IO dclcrwiw hrc:~l rules for avoitlinl:
hltKk:igc Icndcncics. Ihc S O , sil:n:rl i s ncpisy :~nd dillicull l o inlcrltrcl willroul a krtowlAl:c
or IhC hiSlWy Or IllC SySlCltl. C.G.. a rcccn~ brcak;rway of sulfalc b u i l d - u p m a l c r i a l a r r i v i n g
i n IIIC lmrning ~nnc c a n give a h i g h S O , s i g n a l a1 111~ k i l n back end dcspilc tlrc prc~cncc ol
I<ulcs l1avc lowcrccl
a good flnnrc a n d ~cccl~~;~blc lcvcls of vnl;llilcs i n k i l n cnlry malcrial.
111~ nun~l>cr
(,I k i l n >I,,,,F ,w, yc:,r c;mcctl try I~ruln.:~tcr
Irlcsk:rl;c 1reun r~vcr )I) II) Icv\ 01:rn
ill. 1~151 lirlrc ln~~rrs Il:rvinI: :II\~I I:rllr~n loom :~r~nmtl 4,511 Jkr yc:tr 111 :rlr~url IIIII. ( I
hi.11. lvcrt
:rl\o m:ijfn l::rin; i
n
\I*rlt\ 1311wtl Ily rinl:\ :III~I Irr~3L;tw:1y\ ;II lhc kiln trllry rr.:ll.)

l3TECl~S

OF

CONDENSATION

Cast Y: A prccalcincr kiln ran well will1 a Cl lcvcl in kiln cnlry maIcrial of 3 IO 4% (aboui
0 . 5 % less 1har1 111~ K,O Icvcl) provided n o I r a c c of C O w a s indicald. (lhcrc w a s aboul
I . I % S O , i n I h c k i l n cnIry mc;ll i n Ihis siIuaIior1.) IC C O was tlc~cc~cd,
Ihcrc w a s abnu1 2%
S O , and SX K,O i n IIIC 1101 k i l n inlc1 l&cl. accompaniul hy b u i l d - u p s based o n SpurriIc
(2Caz(Si0,).CaC0,) and c u b i c K C I cryslals. ( I I i s gcncrally rccogniscd lhal regular k i l n
opcralion helps IO minimisc Ihc phcnonicnon 0r ccmcnlaliori by Ihc Crcczingl Ihawing Or
cliloridc-linsul
tlclx,siIs.)

The clrcc1 w h i c h rccirculaling vcd;IIilcs cxcrt o n b u i l d - u p rnrm:rIion a~ 111~ k i l n ( g a s ) cxi1

dcl~ntls ori coniposilion (governing Icmpccr;ilnrc
or liquid lirrm;rIion :mcl Ihus lhc posilion and
h;trclncss or Iiiriltl~ul1s as well as Ilic surRcc Icnsirm and viscnsily Or Ihc liquid condc~is:ilc)
Ihc p l a n 1 p$omclry and IllC
a n d o n 1hc quanlily ( w h i c h gnvcrns ralc O r CurmaIion).
ll~rougl~pul a l s o play a par1 i n m a k i n g clrccls mnrc or less prnnounccd. I n cxlrcmc cases.
Cnr qrrcnching a n d scparalc dc-tluslirr~ l o
par1 of 111~ k i l n P,aSCS ;ITC bird O F by-passed
rcmnvc volnlilcs . i n c u r r i n g linancial pcnallics i n plan1 cosl, complcxily a n d r0Ci use.

Cnsr In: T o c x n m i n c IIIC rcnsibiliry oC p r o d u c i n g a sulraic-rich clinker

I n

wiIhou1 insInlling

lhc pasl. s c v c r a l empirical

limils have been l~roposcd

rOr

conccnlralions

volnlilcs

~cmpcraturc in a WCI prnccsr k i l n , a considcrablc dab-bank c l I t.-6urcrncnls

1115 been buill
up. The USC o f lnwrr a s h rucl lowcrcd c l i n k e r K,O l c v c l b y O.lS%. dcspilc 111~ inIroduclior1
of n IiIIlc mnrc p&ssium IO lhc s y s t e m ( a n cxlra 0 . I 16 o n clinker). T h e dusl - rclurncd IO
lhc kiln - had bccomc more rich in alkalis, so lhal Ihe proportion or K,O broughl in by solid
fuel fell lrom 2 4 % I O IX% while ihai brough1
i n b y dust rcIIIrn rose from 18% I O 2 9 % .
This s~rp.gcs~s
1ha1 KzO incotporaIion in clinker no1 only dcpcnds on 1hc quan1ily inlroduccd
11u1 also . a n d almvc all clsc _ o n IIIC type of malcrinl w h i c h b r i n g s i1 i n a n d perhaps o n 111~
pnsiIimr whcrc i1 is injccIctl. On screening 111~ clinker al 20 mm. chcmic~l analysis showed
a K,O coriccnlralinr~ snmc IO% Irighcr i n Ihc cn;~rsc f r a c t i o n . The clinker a l k a l i COIIICIII has
s~~cccssli~lly been rctlucccl
i n lrials b y c a l c i u m chloride atldilinri nl l h c Ilamc.

Cnsc 6: A hil:h chlnridc coal ( - 0. I5 % Cl) can only bc ~isctl as a mix will1 anolhcr coal lo
avoid Imild-ullc wilh Iyliically 2 % Cl a1 Ihc ImI1nm Or cyclone 4 (lhc lowcsl in lhc prchcalcr
~nwcr). hn;rlyTis of Inriltl~ul~s along IIIC kiln indica1ul chlnridc lc,vclr up IO 30% (al zero loss
nn ignirinn) in 111~ coaIing Iron1 lhc base of cyclone 2 and 20% a1 SO 111 inlo Ihc kiln (tlcspilc
iit hsi ih:m 5 % lrvcl i n Imllr lmi :md rold p;nIi 0r Ilic k i l n ) .

C:1w 7:

Atldinl; a sccontl ln~~hc;l~cr

c y c l n n c s~agc IO a l o n g d r y process k i l n ( a varinrIt of
1hc I:igutc I prnccss, will1 a higher k i l n IcngIh/dian~cIcr ralio a n d a single cyclone slagc
nhnvc i1) yicldcd vnrimrs build-up problems. To resnlvc ~hcsc,
IIIC kiln gas cxi1 0, lcvcl was
iricrcA5rccl
rrorll 0.5 In I .5 %, solid rd rcsiduc al 90 microns was rcduccd
lo below 25 % and
scvcral cmr~l~r~ss~d air blaslcrs wcrc inslallctl lo dislodge malcrial rrom lhe lower regions
oC 111~ prchcnlcr. T h c s c acrions improved IIIC siIuaIior1 and sul~scqI~cnIly addiIional mcasurcs
wcrc Iakcn: rcl~/l;iory slirrcrs wcrc addul l o I h c l i n i n g near the k i l n b a c k end, hurncr a i r
vcltXiIy w a s incrcawd. a n d a non-slick lining was inslallcd in 1hc kiln cxi1 gas duel and
cycl011c dip-lubes. Ihcsc Crrm mntlc bcllcr ouIliu1 ralcs pnssihlc wilhoul b u i l d - u p s .
IIc~cnIly. a Iiil;hcr snllur rd hlcntl 113s been nscd, :~ccomp;mictl b y s l a g (S - I X) :Imong
lhc r;~w niix coml~~ncnIs.
IrchcaIcr blockage problems rccurrcd,
bu1 b u i l d - u p s c;m be
avoidctl if the SO, l c v c l i n snmplcs t a k e n Irnm the k i l n cnlry m a l c r i a l i s kcp1 b e l o w 2.5%
b y lirnilirig lucl S conlcnl and sl;ig use i n lhc r a w m i x , provided lhal i n addilion l h c oxygen
lcvcl at IIIC kiln back cm1 is kcp~ consis~cn~ly a1 or above 2%. Cnmpui~r c0nir01 or ihc kiln
Iiclpr I n achicvc succcii. rctlucinl: lhc vari:rhilily ol l h c 0, s i g n a l .

cxw Ii:

I ;~li~n:ih~ry tl:11:1 on minor clcnic111s

cr;~n~plc, suITur vnl;rIiliIy i n ;I s~;~rrtlarrI
rcginrc
0, OUI I~lls in ihc prcscncc
Or 0,; r~~~~rihclcss
a l I2WC. The volalilily 0r m i n o r c l c m c n l s
powdcrrtl m a l c r i a l lhan rOr grarIulcs.

conlirm clvccls crlrxcrvr.tl

ill ln:tclicc. I:rlr
(70% N,, 3 0 % C O ) i s close I O 100% a1 0%
ihc CrrCd Or 0, is much ICSS al 1400C lhan
i n l h c laboralory i s a l s o m u c h grcalcr r0r

Cast Y: A prccalcincr kiln ran well will1 a Cl lcvcl in kiln cnlry maIcrial of 3 IO 4% (aboui
0 . 5 % less 1har1 111~ K,O Icvcl) provided n o I r a c c of C O w a s indicald. (lhcrc w a s aboul
I . I % S O , i n I h c k i l n cnIry mc;ll i n Ihis siIuaIior1.) IC C O was tlc~cc~cd,
Ihcrc w a s abnu1 2%
S O , and SX K,O i n IIIC 1101 k i l n inlc1 l&cl. accompaniul hy b u i l d - u p s based o n SpurriIc
(2Caz(Si0,).CaC0,) and c u b i c K C I cryslals. ( I I i s gcncrally rccogniscd lhal regular k i l n
opcralion helps IO minimisc Ihc phcnonicnon 0r ccmcnlaliori by Ihc Crcczingl Ihawing Or
cliloridc-linsul
tlclx,siIs.)

1m1 Inr some weeks will1 I h c objccIivc or reducing b u r n i n g ronc

b y - p a s s . ICSIS WL
volaIili.~~tinn b y p l a y i n g o n process paramcIcrs a n d prnducing a IighIly mincraliscd clinker
will1 higher volalilc rclcnlion. D u r i n g changcovcr ~hcrc w a s some I c n d c n c y l o rnrrn sol1
build-ups in Ihc prchca~cr, bu1 wiIh ihc new regime csIablishcd ihcsc movctl lowards lhc kiln
fcul chu~c wiIhou1 c a u s i n g m a j o r problems for k i l n opcniion. (EvidcnIly lhcrc a r c
phcrIon1cna of boll1 shorI-Icrm a n d l o n g - I c r m sIabiliIy: o n c e a slablc b u r n i n g zone
volaIili.saIion i s csIablishcd, in lakes Iimc f o r sIablc condiIions I O arrive higher u p the syslcm
and in Ihc Iargc masses or malcrial in lhc build-ups and coalings already in cxislcncc.) The
apparcn1 b u r n i n g ~nnc IcmpcraIurc w a s rcducul b y almu~ 1 2 0 C . while K,O volalilily
droplwl IrIlm 70% III 60% i n 111~ b u r n i n g ~onc a n d IhaI ol S O , Iron1 RO% Iowartls lhc ranl:c
SO% I O GO%, provitlcd lhnl k i l n cxi1 oxygen l c v c l w a s kcp1 a b o v e 2 % . Thcrc wcrc
in~provcn~cnIs in kiln outpul raIc and rucl consumpIion and lhc cxpcrimcnlal Works adoplcd
ccrlain Or Ihcsc changes during normal nl~ralinn
rOr scvcral years.
TCSIS wcrc c;Irricd 0111 invnlving variotrs N
O
, lcvcls ( t o
Cnsc I I:
IcmpcralrIrc) a s well as chcmic~lly rcrhlcing rh0C contliliruIs.
so, :

K,O a n d Na,O :

indicalc

llamc

Ihc r a l i o or S O , i n Slngc I V I n S O , i n r a w iri~l varied

Iypically Iron1 I.R IO 2.7 lnr 111~ higher lcvcls ol NO., and was
3 . 0 [or a IOW 0, Icvcl. The clinker S O , c~nlcnl fell.
In a parallel manner. for K,O 1hc ralin of l h c c o n l c n l i n S~ap,c
I V IO Iha i n r a w meal varictl from 3.X IO 4 . 4 a n d rnr NazO
rrom I .6 IO 2.0.

In gcncral, rultrcing cnndiIions

a lower clinker SO,.

incrcascd SO, lcvcl a1 Slagc IV by a

hClOr 0r

2, also giving

CXSC 1 2 : S O , 11~s l~ccn tmmit~rcd ;II lhc kilt1 Iwck cd IO dclcrwiw hrc:~l rules for avoitlinl:
hltKk:igc Icndcncics. Ihc S O , sil:n:rl i s ncpisy :~nd dillicull l o inlcrltrcl willroul a krtowlAl:c
or IhC hiSlWy Or IllC SySlCltl. C.G.. a rcccn~ brcak;rway of sulfalc b u i l d - u p m a l c r i a l a r r i v i n g
i n IIIC lmrning ~nnc c a n give a h i g h S O , s i g n a l a1 111~ k i l n back end dcspilc tlrc prc~cncc ol
I<ulcs l1avc lowcrccl
a good flnnrc a n d ~cccl~~;~blc lcvcls of vnl;llilcs i n k i l n cnlry malcrial.
111~ nun~l>cr
(,I k i l n >I,,,,F ,w, yc:,r c;mcctl try I~ruln.:~tcr
Irlcsk:rl;c 1reun r~vcr )I) II) Icv\ 01:rn
ill. 1~151 lirlrc ln~~rrs Il:rvinI: :II\~I I:rllr~n loom :~r~nmtl 4,511 Jkr yc:tr 111 :rlr~url IIIII. ( I
hi.11. lvcrt
:rl\o m:ijfn l::rin; i
n
\I*rlt\ 1311wtl Ily rinl:\ :III~I Irr~3L;tw:1y\ ;II lhc kiln trllry rr.:ll.)

l3TECl~S

OF

CONDENSATION

The clrcc1 w h i c h rccirculaling vcd;IIilcs cxcrt o n b u i l d - u p rnrm:rIion a~ 111~ k i l n ( g a s ) cxi1

dcl~ntls ori coniposilion (governing Icmpccr;ilnrc
or liquid lirrm;rIion :mcl Ihus lhc posilion and
h;trclncss or Iiiriltl~ul1s as well as Ilic surRcc Icnsirm and viscnsily Or Ihc liquid condc~is:ilc)
Ihc p l a n 1 p$omclry and IllC
a n d o n 1hc quanlily ( w h i c h gnvcrns ralc O r CurmaIion).
ll~rougl~pul a l s o play a par1 i n m a k i n g clrccls mnrc or less prnnounccd. I n cxlrcmc cases.
Cnr qrrcnching a n d scparalc dc-tluslirr~ l o
par1 of 111~ k i l n P,aSCS ;ITC bird O F by-passed
rcmnvc volnlilcs . i n c u r r i n g linancial pcnallics i n plan1 cosl, complcxily a n d r0Ci use.

.c]%:,qv%3

!:J(~OS)%3)Zr,>r

alI1

JO ?Illl,Zl

01 ,~",~'"~,I
'I\l.,ll"dlllO.3

:SMfl[,oJ

I\laJ7JJ!p

Al,'

15,ll!

a\Il

\I!

II! aSl!i,d

t,',l,aA?p

W!J

UlOllOq

Wt!Al!J3~l!

II! &OR,,

uv s! 2~311.1.

JW!?l,"J',

thl-,l,l\lq

2110

Z?:~\!l~.!l,Wll

311,

I,! \\1!,,,

,31!,\1!~.)

,N,"F/S!!?

xJno3

(lllthl!

ll,!J

1117111.75

a\(lJO S,i,l?!Jl!,?

JO JJl\ll!ll

,C!lW"d XII JO JOICXpII!

\l"!l\!,SllJ.W.,

1 11
> II

l,,,l!,,,s

11

Ill! SC p;r<n

I i II

60

5'0

JO xl!&\
XC,.

A[,

'12Ill!
'Sll,!!,

L'O

-, :, A,,,,

Sq ,lK,

>

JlpIO JJC aJXI.,,)

R'I - 2'1
%

Sb - 51

s-i - SZ

in

(OZ1!,-.J
OS

t??) S!,l!J,V

p"l! ilJX3Jd ~,l!!l.,ll!lll

J;ll,lO
J" IlO!,!S"dUlO~ al,, 1,"
JO ,aAJ, J31,$, ,! \,,!M
JII!JXiO 0, T,,q!SsOd
"'1 i(,!lll ,,

X3,NI!

Oh!N IWO.,,)
$'I, .I. lllt,l!l (1~3 I'!l".,,
lililiI~~~~,~IiI~~~,~~,,~~

AIll!

,TJ\:,,

,+A!,'!,>),
Sth-,3,!lll,

Sth-p,!lN, lJ"s
'0s "0 ,,Xl!I,

X,!SC.,)

anp II! sthl-p,!nq pxcq.aiciwqlc~

273111Os3J a,,,
,'!ll,'!A!pll!
J\ll"S

ST! 'thl-pI!"',

(,,lAOlllaJ

SSa2OJd

l',!Z, XII ll! ti,ql!J2,Ol

JW?I,T)Jd JOJ Jaytl!,:,

8110!11!1111W110.7
k&,.X,C 0, J;r,IaJt,
0, k,\l.?(111.7)
11" ,3 '7A["'" '.a.? 'll,!ry l! \I! 5,q!~~!lll,w!

:(~plllS
JnJ lS.7JW! ,l?!J"dS
JO %l!Sq SW!K,XJX.,) W3lll:J
%l!AW,,OJ X,1 \I! Il.,+,
d,,I!Jlt,~l
JW!.71,2Jd l! JO Sdl!lS J7h\O, X,1 II! p.,ll!J;l,"l
X, Ill!.7 1,2!',"\
S;l,!ll!,nA
JO S,,",,\!J,,,l.),,,,r
'll0!11!~!~1!!
I! S! 71.71,1

ll[I

S&l-,l,!no '(~J\llCJX,\Wl

',,%?lS

'Slll~Il,'lJl,
"II<,.

urn111y10

'(X,JnlS

?lCJ,llF

b\O,"q SlJl!ls
se X,,C!l!ll!

3.006 pm ~8 unmmq sasl\ap\io3 auo,v ?~!Jo,I,~

SU!IIJ;~A~~
e11~~10tt~r~d m ~J~III:J~I!~ ssa, r! 3JabI.I.

p"aJ ll,!?f

"I!'(

'llllt!lll~,!~,)
a&f)8
'~~ltTJ,ns
!,l?~,I!
p!lllJ!,

II! JSW,J110.7 S;lil!J,flS

lSO,A,
3~11idm JOJ ;IJ~\ICJXIIII~I

JO

X,, J" s;llll\,3kl

JA[OAII! lIJ!I,M

(&RS9.() + Oz!N - OztN ha)

PJI!

',l!!JW!lll

ssa,
I,I!M
'paXo,d1117 s\JJa,sXs
Qu!ilfa\I-2Jd
J~,tliu!s
3\,1 c1;ln!3
'JT~~IXIJ~
s! u,!y
XW,OJ
X,\!X3 tl,!lt!,OA
JO ,W"l,!,a~!,
311, l,~llfll,l,"
_ W,.h ,'!\"S,\l!
JO SJ!l!,!C,!SSOd
S!! ,,,7"\
,Stlp
JO Sa,Jh ,,!llJ2,11
a&L!,,
"Sa,,.,.
'kJ,\.!\llaS
'l?M-!\ll;ls
',3M - llO!~,ZJCd~lt,
,W,

(TWXlJd

ap!Jo,c,n tmpos)
\lr!ql uo!les~r;lpiro:,

JX!J

e iv

13i\p

II! I,l!A\
IKllJO
S! I, '(,L(j,
301 II! Ill!l,JW

'S;lll!.Y[lS.
Sll(llJCA
J" llll!Il!lllJl!(
X,l 3ll!,l!l!
.
% , pUllOJC OI atl!JOil,I
JO \lO!lC[ll\llllJX!

~~ro!~e~rnlc~a

JO SJall\1l!ll1

1111 ~llltll!~,J
~ll!,llK&p

II,!?

alttu!1u!
71,l \!lOJJ
Sk' \1111,1J

luri!sralo,,

k,,$h,s

S,TJCMOI

a\,\

Jam01

\IO!iCJl\tJJlJO3

UPJ

- m(j 18UCJ Sl,l

X,IIWJJ Xq '0s

al,) WC,llJ,El 01 3ll!lS2Jalll!

/rOS'"N IllJlSh XJIWJ~\

h,q'!',flJd
ISOIII
"&,m,,
l.32) ,calii
MCJ p211!2,w

JO llO!I!SOdlllO.,
'OSl!3
po!yy

l,+,M ,3)( JO llO!I!ppC

SaWJXl!

," 93llCJ

alI1

?I,1 SV 'StIrI-pl!nq
arr!l!lr!
- sm!~lrls I10 sl!scxl.,p
lIl!rn Sl! _ ,wC Ufl!Sal[pC
xne.2
,fXIl
!Sl1O!IX!~J
,,!"!"'~\I~
aY,OAOJd UCJ pXtJ >,,I 110 Sl!S"h~ '(Z aJn3!:,) Dew,*
S,(1,11 p!ll,\!I JO \,"!,l!t\,JOJ
3\l!?.W,It!
'aJlll~J;lt,lll;ll
<!\I1 II! ,'!'h, Sl1O!l!SW,l\10.7
~,CJ,I,~

rrmyr3 sC t~o~wslr~pi~o~ wJ!p '9.2 '~JU~JOJ spltn~11110~

Mall pill! St,OJp ~JW!JX,lllJI ,CllJ~lll!
'SWaJJll! S3ll!Il!O.,

JO SS;rU~~!l,l
~,,'!1',,c,
UK,
,,,",a', ,,JAa

i! - "OS*>('"OSIQZ - a,!u!rtq%a,
p,,l! p?l,S!,qC,S;,
7q hl!\ll l!!Jq!,!ll,,a

~I!lI.,3IIX~

pWd>\\,>J ,"!J~ll!\\l
\I,
1," p?I.,X,Y? ,"I,

:('pos)'y!<$

ll,l?,
IIIOJ)
J!llH!llr(pWlJal,l

!alqw~~,,~

:SNl~l~~S
'SI~UllOJ3
'!lI.I,

::,(oS)ry?(

:('O>r3'('0!S)'V~~) >\!JJlltk
DJF.
~~,'hllC~:,
,Xl,J!],,3,~!
JJI! SpWlOdlll~~ W,',lllO2 Ja\IlO
pill!

.c]%:,qv%3

!:J(~OS)%3)Zr,>r

::,(oS)ry?(

:('O>r3'('0!S)'V~~) >\!JJlltk
DJF.
~~,'hllC~:,
,Xl,J!],,3,~!
JJI! SpWlOdlll~~ W,',lllO2 Ja\IlO
pill!
i! - "OS*>('"OSIQZ - a,!u!rtq%a,
p,,l! p?l,S!,qC,S;,
7q hl!\ll l!!Jq!,!ll,,a
JO SS;rU~~!l,l
~,,'!1',,c,
UK,
,,,",a', ,,JAa
," 93llCJ

alI1

\IO!iCJl\tJJlJO3

UPJ

'!lI.I,

!alqw~~,,~

:SNl~l~~S
'SI~UllOJ3

:('pos)'y!<$

ll,l?,
IIIOJ)
J!llH!llr(pWlJal,l

~I!lI.,3IIX~

pWd>\\,>J ,"!J~ll!\\l
\I,
1," p?I.,X,Y? ,"I,

rrmyr3 sC t~o~wslr~pi~o~ wJ!p '9.2 '~JU~JOJ spltn~11110~

Mall pill! St,OJp ~JW!JX,lllJI ,CllJ~lll!
'SWaJJll! S3ll!Il!O.,

?I,1 SV 'StIrI-pl!nq
arr!l!lr!
- sm!~lrls I10 sl!scxl.,p
lIl!rn Sl! _ ,wC Ufl!Sal[pC
xne.2
,fXIl
!Sl1O!IX!~J
,,!"!"'~\I~
aY,OAOJd UCJ pXtJ >,,I 110 Sl!S"h~ '(Z aJn3!:,) Dew,*
S,(1,11 p!ll,\!I JO \,"!,l!t\,JOJ
3\l!?.W,It!
'aJlll~J;lt,lll;ll
<!\I1 II! ,'!'h, Sl1O!l!SW,l\10.7
~,CJ,I,~
SaWJXl!

JO llO!I!SOdlllO.,
'OSl!3
po!yy

l,+,M ,3)( JO llO!I!ppC

al,) WC,llJ,El 01 3ll!lS2Jalll!

/rOS'"N IllJlSh XJIWJ~\

h,q'!',flJd
ISOIII
"&,m,,
l.32) ,calii
MCJ p211!2,w

S,TJCMOI

a\,\

Jam01

- m(j 18UCJ Sl,l

X,IIWJJ Xq '0s

II! I,l!A\

',l!!JW!lll

IKllJO
S! I, '(,L(j,
301 II! Ill!l,JW

JX!J

e iv

alI1

JO ?Illl,Zl

ap!Jo,c,n tmpos)
\lr!ql uo!les~r;lpiro:,

~~ro!~e~rnlc~a

Al,'

15,ll!

"I!'(

p"aJ ll,!?f

a\Il

\I!

II! aSl!i,d

'llllt!lll~,!~,)
a&f)8
'~~ltTJ,ns
!,l?~,I!
p!lllJ!,

II! JSW,J110.7 S;lil!J,flS

lSO,A,
3~11idm JOJ ;IJ~\ICJXIIII~I

'(X,JnlS

urn111y10

JO

'S;lll!.Y[lS.
Sll(llJCA
J" llll!Il!lllJl!(
X,l 3ll!,l!l!
.
% , pUllOJC OI atl!JOil,I
JO \lO!lC[ll\llllJX!
13i\p

k,,$h,s

luri!sralo,,

(TWXlJd
alttu!1u!
71,l \!lOJJ
Sk' \1111,1J
II,!?

t,',l,aA?p

'Slll~Il,'lJl,
"II<,.
W!J

2110

?lCJ,llF

ll[I

b\O,"q SlJl!ls
se X,,C!l!ll!
II! &OR,,

uv s! 2~311.1.

thl-,l,l\lq

',,%?lS

Wt!Al!J3~l!

JW!?l,"J',

UlOllOq

S&l-,l,!no '(~J\llCJX,\Wl

3.006 pm ~8 unmmq sasl\ap\io3 auo,v ?~!Jo,I,~

SU!IIJ;~A~~
e11~~10tt~r~d m ~J~III:J~I!~ ssa, r! 3JabI.I.

X,, J" s;llll\,3kl

Z?:~\!l~.!l,Wll

311,

I,! \\1!,,,

,31!,\1!~.)

,N,"F/S!!?

ssa,
I,I!M
'paXo,d1117 s\JJa,sXs
Qu!ilfa\I-2Jd
J~,tliu!s
3\,1 c1;ln!3
'JT~~IXIJ~
s! u,!y
XW,OJ
X,\!X3 tl,!lt!,OA
JO ,W"l,!,a~!,
311, l,~llfll,l,"
_ W,.h ,'!\"S,\l!
JO SJ!l!,!C,!SSOd
S!! ,,,7"\
,Stlp
JO Sa,Jh ,,!llJ2,11
a&L!,,
"Sa,,.,.
'kJ,\.!\llaS
'l?M-!\ll;ls
',3M - llO!~,ZJCd~lt,
,W,

JO SJall\1l!ll1

I\laJ7JJ!p

JA[OAII! lIJ!I,M
xJno3

SSa2OJd

ll,!J

1117111.75

a\(lJO S,i,l?!Jl!,?

X,!SC.,)

Sth-p,!lN, lJ"s

'0s "0 ,,Xl!I,

ST! 'thl-pI!"',

AIll!
(lllthl!

01 ,~",~'"~,I
'I\l.,ll"dlllO.3

JlpIO JJC aJXI.,,)

anp II! sthl-p,!nq pxcq.aiciwqlc~

(,,lAOlllaJ

:SMfl[,oJ

JO JJl\ll!ll

5'0

,+A!,'!,>),

Sth-,3,!lll,

1 11
> II

60

,TJ\:,,

273111Os3J a,,,
,'!ll,'!A!pll!
J\ll"S

>

L'O

I i II

,C!lW"d XII JO JOICXpII!

Ill! SC p;r<n

Oh!N IWO.,,)
$'I, .I. lllt,l!l (1~3 I'!l".,,
lililiI~~~~,~IiI~~~,~~,,~~
X3,NI!

1111 ~llltll!~,J
~ll!,llK&p

\l"!l\!,SllJ.W.,

Sq ,lK,

11

l,,,l!,,,s

-, :, A,,,,

p"l! ilJX3Jd ~,l!!l.,ll!lll

J;ll,lO
J" IlO!,!S"dUlO~ al,, 1,"
JO ,aAJ, J31,$, ,! \,,!M
JII!JXiO 0, T,,q!SsOd
"'1 i(,!lll ,,

(&RS9.() + Oz!N - OztN ha)

t??) S!,l!J,V

(OZ1!,-.J
OS

s-i - SZ
Sb - 51
R'I - 2'1
%

in

:(~plllS
JnJ lS.7JW! ,l?!J"dS
JO %l!Sq SW!K,XJX.,) W3lll:J
%l!AW,,OJ X,1 \I! Il.,+,
d,,I!Jlt,~l
JW!.71,2Jd l! JO Sdl!lS J7h\O, X,1 II! p.,ll!J;l,"l
X, Ill!.7 1,2!',"\
S;l,!ll!,nA
JO S,,",,\!J,,,l.),,,,r

PJI!

'll0!11!~!~1!!
I! S! 71.71,1

JO xl!&\
XC,.

A[,

'12Ill!
'Sll,!!,

l',!Z, XII ll! ti,ql!J2,Ol

JW?I,T)Jd JOJ Jaytl!,:,

8110!11!1111W110.7
k&,.X,C 0, J;r,IaJt,
0, k,\l.?(111.7)
11" ,3 '7A["'" '.a.? 'll,!ry l! \I! 5,q!~~!lll,w!

(fmm raw materials1

FIGURE 3
OXIDISED KILS SULFL.iR
CYCLE
( . . . . . . . . . vapour phase: -solid phase)

FIGURE 3
OXIDISED KILS SULFL.iR
CYCLE
( . . . . . . . . . vapour phase: -solid phase)

(fmm raw materials1

1115 cffccls o n cliukcr

iuay be sutntnariscd

as follows:

Iluxinl:

aclioii:

0
l
6
0

l o w e r lcmperalure o f firs1 l i q u i d phase f o r m a t i o n

cbangc al l i q u i d viscosily
allcraliun 0r sdxx Iciisioii OK l i q u i d
inihlilic;il~o~i ul crystal ulorlhology.

c o m b i n e d williin Ihc silic;

.md aluminalss o r as auliydrirc (CaSO,). which d~s~I~cs umrc
slowly III~II a l k a l i sulfales, whilst llle fraclions of K,O aud NazO which are s o l u b l e ill w;ucr
approach I .O ;mJ 0.S. rcspxiivsly. al a raliu OI:IIJOUI I.S. Al sitllalc:;ilk:ili
raliiJ9 ahvc I .S
Ircntls a r c wmewl~al crralic.
Iur IWSI uoru~l clinkers IIW prilicip4 rulli~~c phase w i l l bc
alhil~italilc
(Iwlassitlrri/roditr[rl s u l f a t e ) with a m a x i m u m WN;I r a t i o o f 3 . 0 . Il~is IIllasc i s
accompanied
b y m i n o r quaniiiirs o f K,SO, a n d c a l c i u m Iangbsiniic. Na$O, bciug hmd
o n l y Ibr u n u s u a l l y l o w K/Na raiios.
A s w e l l a s lhe s o l i d solulion rffecls a n d the f o r m a t i o n o f compou~ids d e s c r i b e d almvr,
v a r i o u s permulaliotis
o f volalilcs ( e s p e c i a l l y in Itie p r e s e n c e o f f l u o r i n e ) c a n influcrux rhc
s~ructurs arid belnviour o f aliic aml bcliir crysials ( M o i r & Classtx, l!J92).

Il~ass rclaliurls:

I lydraulic
l

aclivily:

lhe fL;iclivtlics 01 lhc cliukcr u~immls arc allcrcd b y

sul~d mlulion ;uldlor by
the cffccrs o f crystal syinrilctry (IQli leu~pcralure stabilization o f poly~norphs)
arid/or clfucls occurritrg during liydraliotl (e.g., coaling of cc~uenl particles by
insolubly
salis).

( I I i s d i f f i c u l t lo isolale

Ihcsc Ihrcc classes of cffccl i i i praclics.)

I II gcuc~al. incorporalioo
of a d d i t i o n a l clillkcr solC;~~e it1 a situatioll willi e~.ccss a l k a l i s y i e l d s
a inore dilficuli iiIlI>ilfCl\l
grithhbilily with adviiolagcs i n the Iua~kcl or hlmvctl
early
concrclc s1rc11g111 a n d workabilily.
IIIc cf~cc~r or utimJr ~WII~~II~IIIS WI the vircosily a n d shrlace lensioll of l i q u i d phascr c a n
be cou~plrx. Luwcr viscosiliss e n c o u r a g e alilc ( c a l c i u m lrisilicale) forinalion.
Calciuiu
SUIT;IIL flux call. I~ow~!vsr. stabilise bclilc ( c a l c i u m disilicatc) antll~r CatIsc lhr prodtrcliori o f
clinker alilc will1 lime i n c l u s i o n s . I:urlher, iu clinkers with a l o w alk;lli CWICIII, bclilc
sI;tibilis;dicm due III cxccts S O , Icxls I~I tlilliuulr cl,lnl,iii;ll,ilily.
S~rtm~ly c h e m i c a l l y fcducir$
comJilioiis I I I Ilic burriiug LOII(: cau Live a camcril will! par Ilow cliaiaclcrislics
(tluc lo Irsc
K;,O a n d N+O), p o o r workhility ( d u e IO IIIC i n c r e a s e d conlcnl o f Iricalcium aluminalc arid
ils rraclivily), poor slrcuglh (lower lricalcium silicalc conlenl) a n d Vilriill)lt colour.
A l k a l i s rclainrd i i i clinker a r e prcscm cilhrr a s slablr sulhles o r iibSlllbd i n llld silicate arid
alulnitialr siruc~urcs: ihcsc inlltrencc IIIC bchaviour o f fresh cor~crele a n d m o r t a r d u e l o Ilieir
N;r20 hi\s a I~\WC matkcd ~ende~\cy
than K,c) 10 ~OIIII wI~II~IIS ill
various solubililier.
I:or clinkers with (ruular) ratios of sulfiilc:lol;~l alkalis below 0.3, iIIItloSl
c a l c i u m alurnimilr.
a l l 111e sulh~e i s combimxl ill waltz soluble form, K$O,, beill& prcddomimmt. A pruporlion
ol the a l k a l i s ale i n s o l i d soluiiou itI IIIC c l i n k e r aluciiinalr phase and lhis has ati advorle
rlfccl ori the inirial cc~~~cnI rcaclivily illld Illus o n comxele a n d morlaf r h e o l o g y .
For ralios belwceii 0 . 5 a n d I .O, a c e r t a i n quanlily of langbrinilc
i s a l s o formed (and no1 a l l
I:or r a t i o s alrove 1.0, si~nilicanl f r a c t i o n s o f lhr Sulfales a r e
lltr a l k a l i s a r e wlul~lc).

II is gc~icf;~lly ~.up~~xd IhI (t~llicr Ihcl~~rs ljcitlg cqii;~lJ ll~c cxlcul 1,1 vt8I:tlili*:tlit~ll ~~L,I~:I\cI
a s the Iliermal efficiency a l IIIC kiln iucrcascs. Iliis i s prol~ably du1: Iu iltc lillliliug sllUci
o f vapur wluralion
b y a l k a l i c o m p o u n d s , a s c o n f i r m e d b y sludics of IIIC irca~msm o f k i l n
dusl i n a I00 IIII~ diarnslcr hidiscd bed (lclinnr rf ul, 1978) lo e x a m i n e ihc fcasibili~y o f
produciril: cliriker ~~OIII WIII~III kilt1 flue dull will1 caplure of IIIC a l k a l i s distilled f r o m 111l:
b e d for p o s s i b l e use ill 111e ferliliscr i n d u s t r y .

I I i s su~gerlcd IIuI

!II
V

v*

i;i

p*

J
M,

=
=
=

M,

= -me..llilllM,1-.-.
(1' - 1") WJ

salurhtd vapour coiicciilralioii

wluralrd vapour p r e s s u r e
OT au a l k a l i compouml
g a s pcssur~
niolrcul;ir
wcigld III valmiir
m o l e c u l a r weigh1 o f g a s .

i i i Irai~sporl pscs ( k g / k g )
1
) ~:IIIIC imils
1

Give0 malluzmalical
cxprcssiims f o r S;~lt~rillI2~l vapour prcssltrcs ilS ~m~ciicms of lc~llpcraluru
a n d krmwlcdpz of kill) systru~ Icmpcralurc p r o f i l e s , the w~ura~cd val)uur conccmraiioll
WI
be calculaled f o r exh a l k a l i compout~d a n d IIIIIS he maximurn
qu;miiiits cval~ra~cJ f r o m
llle feed lxx unit IllilSS of g a s e s . Ilieri, considcrirlg the amoums ol gas passirlg Il~rou~:h
lhc
kih al v a r i o u s temperatures,
llx Iruc llllillllily ol volaiilcs Irar~sp~rlud per uuii m a s s o f
c l i n k e r GUI be calc~tla~txl
arld f r o m his krtowlcdp, ideal* volaiilc cycles CarI lx tlsduccd.
For example, saturated vapour
KCI
K,SO,
Na$O,

pressures at IZOOC

0. I8 am
0 . 8 x lo illIll
0 . 1 3 x IO] alIll

are (for IIIE

prrc

substances):

(0.6 x IOj am with dccolnpition

suI~Imssec1)
(0.01 x IO ~IIII with d e c o m p o s i t i o n s u p p r e s s e d )

c o m b i n e d williin Ihc silic;

.md aluminalss o r as auliydrirc (CaSO,). which d~s~I~cs umrc
slowly III~II a l k a l i sulfales, whilst llle fraclions of K,O aud NazO which are s o l u b l e ill w;ucr
approach I .O ;mJ 0.S. rcspxiivsly. al a raliu OI:IIJOUI I.S. Al sitllalc:;ilk:ili
raliiJ9 ahvc I .S
Ircntls a r c wmewl~al crralic.
Iur IWSI uoru~l clinkers IIW prilicip4 rulli~~c phase w i l l bc
alhil~italilc
(Iwlassitlrri/roditr[rl s u l f a t e ) with a m a x i m u m WN;I r a t i o o f 3 . 0 . Il~is IIllasc i s
accompanied
b y m i n o r quaniiiirs o f K,SO, a n d c a l c i u m Iangbsiniic. Na$O, bciug hmd
o n l y Ibr u n u s u a l l y l o w K/Na raiios.
A s w e l l a s lhe s o l i d solulion rffecls a n d the f o r m a t i o n o f compou~ids d e s c r i b e d almvr,
v a r i o u s permulaliotis
o f volalilcs ( e s p e c i a l l y in Itie p r e s e n c e o f f l u o r i n e ) c a n influcrux rhc
s~ructurs arid belnviour o f aliic aml bcliir crysials ( M o i r & Classtx, l!J92).

;uldlor

iuay be sutntnariscd

as follows:

lcmperalure o f firs1 l i q u i d phase f o r m a t i o n

al l i q u i d viscosily
un 0r sdxx Iciisioii OK l i q u i d
l~o~i ul crystal ulorlhology.

by

II is gc~icf;~lly ~.up~~xd IhI (t~llicr Ihcl~~rs ljcitlg cqii;~lJ ll~c cxlcul 1,1 vt8I:tlili*:tlit~ll ~~L,I~:I\cI
a s the Iliermal efficiency a l IIIC kiln iucrcascs. Iliis i s prol~ably du1: Iu iltc lillliliug sllUci
o f vapur wluralion
b y a l k a l i c o m p o u n d s , a s c o n f i r m e d b y sludics of IIIC irca~msm o f k i l n
dusl i n a I00 IIII~ diarnslcr hidiscd bed (lclinnr rf ul, 1978) lo e x a m i n e ihc fcasibili~y o f
produciril: cliriker ~~OIII WIII~III kilt1 flue dull will1 caplure of IIIC a l k a l i s distilled f r o m 111l:
b e d for p o s s i b l e use ill 111e ferliliscr i n d u s t r y .

I I i s su~gerlcd IIuI

!II
V

:
Calciuiu

v*

i;i

p*

J
M,

=
=
=

vily:

clivtlics 01 lhc cliukcr u~immls arc allcrcd b y

sul~d mlulion
ccrs o f crystal syinrilctry (IQli leu~pcralure stabilization o f poly~norphs)
clfucls occurritrg during liydraliotl (e.g., coaling of cc~uenl particles by
y
salis).

ale

M,

of a d d i t i o n a l clillkcr solC;~~e it1 a situatioll willi e~.ccss a l k a l i s y i e l d s

grithhbilily with adviiolagcs i n the Iua~kcl or hlmvctl
early
d workabilily.

>ilfCl\l

nker a r e prcscm cilhrr a s slablr sulhles o r iibSlllbd i n llld silicate arid

ihcsc inlltrencc IIIC bchaviour o f fresh cor~crele a n d m o r t a r d u e l o Ilieir
N;r20 hi\s a I~\WC matkcd ~ende~\cy
than K,c) 10 ~OIIII wI~II~IIS ill
I:or clinkers with (ruular) ratios of sulfiilc:lol;~l alkalis below 0.3, iIIItloSl
combimxl ill waltz soluble form, K$O,, beill& prcddomimmt. A pruporlion
n s o l i d soluiiou itI IIIC c l i n k e r aluciiinalr phase and lhis has ati advorle

salurhtd vapour coiicciilralioii

wluralrd vapour p r e s s u r e
OT au a l k a l i compouml
g a s pcssur~
niolrcul;ir
wcigld III valmiir
m o l e c u l a r weigh1 o f g a s .

i i i Irai~sporl pscs ( k g / k g )
1
) ~:IIIIC imils
1

Give0 malluzmalical
cxprcssiims f o r S;~lt~rillI2~l vapour prcssltrcs ilS ~m~ciicms of lc~llpcraluru
a n d krmwlcdpz of kill) systru~ Icmpcralurc p r o f i l e s , the w~ura~cd val)uur conccmraiioll
WI
be calculaled f o r exh a l k a l i compout~d a n d IIIIIS he maximurn
qu;miiiits cval~ra~cJ f r o m
llle feed lxx unit IllilSS of g a s e s . Ilieri, considcrirlg the amoums ol gas passirlg Il~rou~:h
lhc
kih al v a r i o u s temperatures,
llx Iruc llllillllily ol volaiilcs Irar~sp~rlud per uuii m a s s o f
c l i n k e r GUI be calc~tla~txl
arld f r o m his krtowlcdp, ideal* volaiilc cycles CarI lx tlsduccd.

Ihcsc Ihrcc classes of cffccl i i i praclics.)

~WII~~II~IIIS WI the vircosily a n d shrlace lensioll of l i q u i d phascr c a n

iscosiliss e n c o u r a g e alilc ( c a l c i u m lrisilicale) forinalion.
~!vsr. stabilise bclilc ( c a l c i u m disilicatc) antll~r CatIsc lhr prodtrcliori o f
i n c l u s i o n s . I:urlher, iu clinkers with a l o w alk;lli CWICIII, bclilc
cxccts S O , Icxls I~I tlilliuulr cl,lnl,iii;ll,ilily.
S~rtm~ly c h e m i c a l l y fcducir$
rriiug LOII(: cau Live a camcril will! par Ilow cliaiaclcrislics
(tluc lo Irsc
r workhility ( d u e IO IIIC i n c r e a s e d conlcnl o f Iricalcium aluminalc arid
lrcuglh (lower lricalcium silicalc conlenl) a n d Vilriill)lt colour.

= -me..llilllM,1-.-.
(1' - 1") WJ

For example, saturated vapour

Sulfales

are

KCI
K,SO,
Na$O,

pressures at IZOOC

0. I8 am
0 . 8 x lo illIll
0 . 1 3 x IO] alIll

are (for IIIE

prrc

substances):

(0.6 x IOj am with dccolnpition

suI~Imssec1)
(0.01 x IO ~IIII with d e c o m p o s i t i o n s u p p r e s s e d )

Ilrc
Na

I~;III~IKJ~( capacity

14 air lo; vapours al lZ(XJC i

cqj;lcily nl I?S(lC i s ;IIJ~JIII

~SO,, < 0.S 1:/g. Ihc

II cmi IljcrcInrc hc Iurcsccn Ilral (unless llic equilibrium vapour prcssurcs difCcr grally Irom
urlur;jlrA values) lbcrc w i l l b c lilllc problem i n removing KCI from m a n y k i l n lluc dusls in
a lluitliscd IJC~ witli a g a s llow ralc o f , s a y , 2 g per grammc o f dust, idll10ugb Ibe c a p a c i t y
C0r sdhk removal m a y b c limilcd. The sa111c reasoning a p p l i e s 1 0 k i l n s , will1 wcl process
kilns typically slmwinl: n ratio rjf a lilllc less 1ljan 2 g/g gas/solids in ~ljc burning z0rjc and
pcrljapr 2 . 7 5 l;Ig ;II IIW b a c k end. will1 corrcsl~nding values C0r lljc d r y prtxzcss (wilboul
I)rcc;jlcitl;lliolj)
Or I .4 ~$1; and I .91 g/g.
I)c\pilc IIIC lacl IIWI qualilalivc diCCcrcr1ccs
arc rrllcclctl in s:jmplc calculations, Suclj
IimcT lnrjicr lljn~i tllosc cricounlcrcd i n

I~c~wcc~~ I W O k i l n s (OIJC d r y process a n d OIJC wet)

idcal calcolalctl rccirculalinl; loads arc aboul IO
praclicc. TIJC probable reasons include:

3)

In~~~w~~lclr ct~~~l:wl Iwlwccn g:iws :uul s o l i d s i n llir k i l n . wlicrc o n l y a sii1;1ll

Iraclitui 01 lltc s o l i d surracc i s cxlxjsrd a l ; I given lime. ( I I i s cxpcclcd llial
Ibcrc i% bcllcr coril;~cl
i n llic cc~lrlcr dusly rqiojis Or lbc syslct~~.)

11)

Vul:llilis:llion c:b;lr;lclcrislics of IIIC alk:lli-ccrlJl;lirliIlg

IniItcritIs ;,I a givcu IJI;IIJI.
i.c., conrljinnliorl i n m0rc co~~iplcx
silicalcs a n d aluminalcs. 1~01 o n l y sijihks.

0r co~~jp~rintls

wljicb arc condcnscd as/on solid dusl or

rtl0lc.

cl

Iraijspfjri

(1)
Cl

I~~Ictluciion

I:orrnsliotj

b)

I~rjs~al~lc opcralion or p r o d u c t i o n k i l n s : praclical c o n d i t i o n s a r c 1101 cxaclly

Il~osc cxl~clcd k)r very l o n g lcrm slabilily Or lcmpcralurc a n d malcrial Ilow.

nl vap0ur

prcssurc over scjlulions of a l k a l i c o m p o u n d s .

i~~lq~~c lrcnlmcnl d ibc Irnjjspjrl

clicikcr nodules i n Ihc kill).

Or

or

ljcal and

IIIANI<S

tbur: nil - 7(X1 g/1;: K,SO, - 1 g/1::

I W O limes bil;llcr.

Or

vnfxnjr

willrin

UJC bed

Or

Olbcr compounds. e.g., CaSO,, dcpcndin~ on alknli:sullaic ralio.

Tbc lurillcr tlcvclqjnrcn~ or a prctliclivc model w i l l Ijavc I O iakc accour~l

o f such [actors, as
well as illc crrccls Or coni~xlsilion Or kiln almospbcrc. In rcccril ycnrs llicrc has been mucll
invcslig:llirjtj ol l;jclors /:nvcrninp, IIIC block;tgc of cyclones a n d lbcir pcck~rmancc i n biGI
Icurpcralurc coal c~11iil~5lion proccsscs
(itI Ilic Iiopc or prolccling Iurbinc blatlcs i
n
tlirrcl
cycle clcclric power gcucr;tlicln syslcms). Wljcn linic (or limesi0nc) i
s
injcclcd lo nbsorb
S O , . 111~ COIJI~HJ~JI~S
a n d Il~crrnodynamic crilcria invoked a r c cxaclly 1110s~ cncounlcrcd in
IIIC CCIIKIII intluslry - particularly wl~cn rclalivcly l1igb cbloridc coals arc oscd. II is probable
Illal Iljcrc i s now sullicicnl n c a d c m i c knowlcdgc IO bclicr Irc31 o u r silualion a n d a l l o w
improvctl mrxlcllinl; a n d undcrslanding. Anolbcr aspccl 10 consider i s knowlcdgc acquired
rrnm sludy 0r ibc rcgciicraliori 0r CaO sorbcrils u.sul ror S O , s c r u b b i n g : a g a i n , dala
p0lcnlially rclcvanf in k i l n sys~cms arc protluccd, lor cxamplc, o n p r c s s u r c s o f S O , i n IIIC
sy~cm CaSO.,I C;rSI CnO i n IIIC prescncc ol v a r i o u s conccnlrations ol C O a n d C O , . II i s
Impcd Illal ibis paper promolcs cross-rcrliliulion
bctwccn IIJCSC v a r i o u s fields or w o r k a n d
lbc bcnly oC praclical knowlctlgc available i n ~ljc ccmcnl i n d u s t r y .

lltis Ical. ori~;in;llly I~lsul o n v a r i o u s irjlcrn~tl amI crlcrr~l rqK)rIs, Ita\: l~ccrj atl;~l~lcd 01t lllc
b a s i s of urchjl di.uzussion will1 Icclmical slall rrom a mrmbcr ol ccmcnl companies. 7banks
arc given lo all involved, as well as IO lbc Directors of WIIC Circle Industries PLC ror lbcir
permission I O publish Ibis paper.

Clioi. G.-S. & Glaswr, 1:. I .

(19RR).Ibc Sulplrur Cycle i n Ccmcn~ K i l n s : Vap0ur
Ircssurcs and Solid~lliasc
Slabilily O
r
llrc Sulpbalc Iljascs. Ccr~~ct~l rind Concrrlc
Ilcsc:~rcl~, IR. 367.371.
Kct~n. C . I . & M u r r a y , It. 1. (1984). Iortland CCIJIOJI Irotluclicm. 111 Slntcljtrr nntl
Icrlonnnnrr oC CWICII~S. c c l . I . Ilarncs. Appliccl Scicncc Iublisltcrs I.id. Ihrkirj~. ~IJ 705.
7 . x
M o i r . (i. K . & Gksscr, I:. I . (lYY2). hlincrnliscrs, h4otlificrs a n d Activators i n ~ljc
Clinkcring
Irr~css.
In 9111 It~lrrwtliowjl Corjgrcsi ~IJ lljc Cljcjrjislrg
or CCIJJCII~. NCW
Dribi, I n d i a , 1 9 9 2 : Congress Itcporls, Vnlumc I , National C o u n c i l r0r Ccmcnl a n d
Iluildirj~ Malcri:tls, New Dcllji, p p . I2S- 154.
Rilzmann,
338-343.

I l . (1971). C y c l i c Ibcrtomcna i n Rotary K i l n Systcrr~s. ~cjrlclll-l(nlk-Gip(,

Icllmar, n.. Kljor, I . I I . . k Gregory,

Glps, 3 I , 288-290.

S. (1979).

Iroccrsing or K i l n Dusi.

24.

APPExmX 3

PAPERPRESENTED BYT MLOWES

100% Pet Coke - Problems and Solutions

100% PET COKE - PROBLEMS AND SOLU-I-IONS

Good morning Lady and Gentlemen,
The citie of my paper is 100% Petcoke - Problems and Solutions.
It will give information on how Blue Circie Cement has moved from zero to 40%
petcoke over a 3 year period, indicating the technological problems that need to be
overcome for any works that seeks to fire 100% petcoke.
Before beginning my presentation I would like to thank Gerard Flament of CCB and
Jean Pierre Piliard of Ciments Lafarge for their help and information during this
development phase within Blue Circle Cement.
The paper mainiy deals with the dry process.
LepoI and long dry processes.

However, the conclusions apply to wet,

Slide 1
This overhead indicates BCCs approach to petcoke. Firstly, prior to 1991 there was
zero use of petcoke because ic was coo much trouble and low ash coal was much
NICER. However, in 1992 due to a significant recession, we moved up to 30%
replacement and in 1993 40% overall on 10 Works. Some Wor:ks using none, 100% usage
at times on a Lepol process, 80% on the large semi-wets at Northfleet and 65% on a
wet process at our Masons Works.
In 1993 the prices increased, consequently the financial benefit accrued to the project
was only the same in 1993 as 1992.
In 1994 there has been a slight market improvement, some Works can se11 everything
they can make and consequently we are already failing to meet the 1994 plans for use
co petcoke.

Slide 2
This slide gives some indications of why one should use petcoke. Firstly, its price can.
be up to, and sometimes even more than, 50% less than the coal price per GJ. It can
improve cement quality, if there is an excess of alkalies over sulphates. In addition, if
you are coal mill limited on cement olant output it can, with the appropriate Hardgrove
index, increase coal mill capacity.
In addition, it can be claimed to help the environment as petcoke needs to be burnt and
if it is burnt in conventional power stations the SO, emission {vi11 increase, whereas in
most Of our processes it is significantly retained. However, the real reason for using
it is that in a recession, the full kiln OUtpUt is not reouired and consequently operational
costs are at a premium and petcoke can make a significant reduction in OperatiOnal
costs.

Slide 3
This siide indicates reasons for not using petcoke. Firstly, it can increase SO?
emissions, this even applies to a certain extent within a dry process as, for example,
even in a precalciner betueen 6000 and 10000 ppm SO, is the maximum that can be
absorbed before bypass into the preheater sys=m occurs.
Due to the extra S02/S03 going into the system I dont need to celi anybody there is
significant increase in potential for ring formation and blockages. In addition the
blending operations are increased unless one is reaily firing at 100% petroleum coke.
.AIso there can be coal mill fineness problems having to meet the lower 90 micron
residue and there is more variability in sulphur and Hardgrove index. in addition, some
investment may be required.

Slide 4
This slide identifies the main problem of using petcoke which is controlling the S02/S03
cycles which leads to blockages and increased SO? emission.
This depends on the total SO3 input, the Na30 equivalent, the molar ratio of SO3 to
alkaiies in the clinker, the combinability temperature of the clinker and in addition to
this, how well the kiln is controlled and the overall flame conditions that exist within
the kiln.

Slide 5
This slide shows some facts concerning peccoke usage. Firstly, for price reasons the
main interest lies in the higher sulphur petcoke which typically varies between 44% and
S+% suiphur. For a dry process kiln with 800kcal/kg using 100% of a 5% sulphur
peccoke, this is equivalent to around 1.33% SO3 on clinker.
As most Works are limited to around 0.6 Nay0 equivalent, this corresponds to a
molecular ratio of 2 to a clinker SO3 of 1.5%. .Above a molecular ratio of I, calcium iangbeinite, which is a salt of alkali and calcium
sulphate with 2 molecules of calcium sulphate for 1 of potassium sulphate, forms in
addition to calcium sulphate.
Calcium lambonite decomposes at around 152OC leaving alkali and calcium sulphate.
Calcium sulphate decomposes at 145OC. However, these decomposition temperatures
are both in the presence of excess oxygen.
If there is in excess of 2000 PP?J of CO in contact with caicium sulphate above a
1000C it would break down to SO3 and calcium oxide and hence exacerbating any
particular problems one has with SO2 emission and preheater/kiln blockages.
For a molecular ratio greater than 2 in stage 4 raw meal, this generally is recognised
as producing hard deposits.

Slide 6

This slide shows what conditions you need to avoid and what conditions you need to have
to be able to use 10096 petcoke.
Firstly, the conditions that need to be avoid& are chemical reduction in the burning
zone, Over-burning and low Na,O equivalent. For 1009/o petcoke one needs to have a
good flame, a readily combinable raw mix and a good kiln control of free lime and back
end oxygen.

Slide 7
This slide shows what are the requirements of a good flame and uses the CEMFLAIME
1 information fairiy extensively.
One needs to ensure that the burner has adequate momentum, e.g. around 7N/MW. This
is to ensure that there is adequate air available to combust the voiatiles and carbon of
the peuoleum coke on a micro mixed basis before the petcoke comes in contact wirh
the burning zone at between 2 and 3 kiln diameters, depending on wheKher one has a
high swirl or a zero swirl burner.
It is importam Khar the ignition takes places very near the burner, a 60% bluff body will
help to promote Khis by removing the jet establishmenr. region and creating an internal
reverse flow zone. In addition, a secondary air temperature of greater than 800C is
required to achieve this early ignition.
Petroleum coke is a by-product of a process which essentially means that all of the

lower temperature volaciles have been extracted and therefore peuoleum coke needs
to reach around 900C before its first volatiles can be released.
T O cope with this lower volatile release rate and notentially a less porous stiumre of
the carbon as a result of its formation process, a90 micron residue of around 6-7% is
required, following the general rule of +90 micron residue being 50% of the volatile
matter.
In addition, it is absolutely essential that a uniform enKrainmenK of secondary air takes
place into the flame, particularly ensuring that the secondary air coming under the
burner is not SKarved. To ensure the maximum possibilities a burner cenually lined u?
the kiln is most important.
Even if all of these criteria are met, one will still operate in reducing condition
promoted by unburned carbon and CO if there is not adequate conuol of the back end
oxygen. A recommended level is around a minimum of 3.5%.

Slide 8
This slide shows results from the CE:VFLAME 1 trials which indicates impact of back
end oxygen on CO for both a medium volatile coal and flexicoke, with 90 micron
residues of 1246 and 2%.

The measurements are made at the back end of the kiln simulator which is equivalent
CO an L/D of 12. As you can see, for the flexicoke, co ensure r.har: rhere is a CO level
less than 1000 PPM, one is essentiaiiy looking for 3490 oxygen. A similar condition is
i?Ot necessary for a medium voiatiie
coal where ic does look char: in Che a-?;% region
the conditions for less ihan 1000 ?Phl can be achieved.

Slide 9
This slide shows a typical situation on our dry process Works at Hope which indicates
the impact of hard burning and CO/oxygen on stage 4 SO3. Hope Works has a clinker
Molar Ratio of 2 with a British Coal.
This was for a high volatile coal with a 90 micron residue of around 15% with the burner
lined centrally up the axis and a momentum of at least 7NI1MW.
It can be seen that hard burning increases the amount of SO3 in stage 4. However, it
also increases the amount of NaZO equivalent and consequently not until one gets above
around 1300 ppm of NOx is there a significant increase of excess SO3 over alkalies due
CO the thermal breakdown of calcium sumhate
in the burning zone.
However, the most interesting point is the impact of back end oxygen OR the level of
CO and its consequent impact on stage 4 SO3.
It can be seen that once the back end oxygen drops down beiow 1.8%, i.e. to 1.4%, the
CO increases from 500 porn to 2500 ppm and a corresponding increase of stage 4 SO3
from 3.2% to 4.5%.
Under the conditions of the experiment, around 24 times the Na30 equivalent in the
clinker was appearing the in stage 4 raw meal. Consequently, fo; a molar ratio Of 2,
this corresponded to around 3.75% S03.
It can be seen that under reducing conditions promoted solely by the lack of back end
oxygen, the SO3 in stage 4 increased to 4.5% which is a totally unacceptable level and
blockage problems occur on the Works at this ievei.
The Works normally seek to run at around 3% stage 4 SO;.

Slide 10
While a good flame is important, alone it is not enough.
Good kiln controi is required, i.e. using Linkman, to avoid continual SO3 recycling. For
exampie, if one has a blaster operadon that breaks down deposits, if once it reaches the
burning zone the conditions are SO hot that 90% of the SO3 is sent back into the
preheater system, obviously the blaster operations are not being suitably supplemented.

In addition, it is very important to have a combinability temperature essentially less

than 1500C. Also, one operates at round about I-I?XJ free lime, or else a similar
situation exists as to that I referred to previously under the need for good kiln conuol.
Otherwise a continual recycling of the SOTin the syste.m wili occur leading to
permanent problems associated with blockages and breakaway.
In addition, it is important that one continually checks what is achieved for the stage
4 material. Preferably, this is supplemented by an SO2 probe that can be used to
monitor what is happenin g in the back end of the kiln and even be linked into the high
level control. For example, 3000 ppm of SO7 equates to around an extra 1.05% SO3 on
stage 4 raw meal.

Slide 11
And finally, in conclusion, now we know how to do it, how can 100% petcoke be used
with lower NOx emission, bearing in mind this normally mea&u running into on the verge
of reducing conditions?
The answer is, I dont know at the moment. However, CEMFLAIME
take place at the end of this year, will provide the answer.
Thank you

2 which is due to

--__-._---_-

---

100%PET COKE - PROBLEMS AND SOLbTIONs]

_.. -_-_.. _--_-_.- ---------.---- .-.
-__-

o BCC's approach to Pet Coke

- Prior to 1991 zero use too much trouble, low ash
coal much NICER
- 1992 - 30%
- 1993 - 40% 100% Lepol, 80% semi wet, 65% wet.
- 1993 prices increased
- 1994 market improvement
- 1994 budget 35%

0
ti

0)
Ci

0
d
-

(0
Ci

53
m
0
s

0
cd
-

i?
0
1,

8
us
-

8
b

I
1

i
-

t
1
i
f
(
I

i
-

0
d

8
ul

0
ti

Ln

k
3

-_--..---___

____ - __...-- _-__- . . _ _.. -I__-__-_--.

1I-- 100%
PET COKX - PROBLEMS AND
SOLUTIONS /
_.------_-_--.---__-- -___--. -- 1
-.-_-. ..-._.- . ..__._ _-. -. -_.-.._ ._.. - .__... _ _-. _- ,_.___ ,.__ -___
l

Why use Pet Coke?

A Price - up to 50% < Coal/Gj
A Improve cement quality
A Increase coal mill capacity
A Help the enviroment
A Recession, fxlll kiln output not required

--- -_- _ -____. --__ ___.______ _.__ __,_____-_I__., _ -._____ _____ __._ _ __--

,
I
/ 100% PET.-.-_ COKE
PROBLEMS
AND
SOLUTIONS
--.---- -..-- -_.-.- . .- . ..-.--- . . - . I ._ _._- ~----.-__.--.- -.----.-. _-----o Why not use Pet Coke?
A Increase in SO2 emission
A Increase in problems with rings and blockages
A Blending operations increased
A Coal mill fineness problems
A S ,and Hardgrove variability
A Investment may be required

.---__-__-.__-_- -.-__ --__ ..--

. . .._.- -_,.-..-. __.._______

-__-

_---

111000/o PET COKE__--..- PROBLEMS

AND
SOLUTIONS
.__- - -__.__ ._. ._-_____. --____-__--_.--.-__-- ____.-_ --_.-. ___ .._....-... -..
o Main PROBLEM
A Controlling SO2/SO3 cycles which leads to increased
blockages and SO2 emission

0 Depends on:A Total SO3 input

A Na20 equivalent
A Clinker Molar Ratio
A Combinability Temperature
A Kiln Burning Control

A Hame Conditions

.-- -------- -_----

--

--

100% PET-COKE - PROBLEMS AND

-------SOLUTIONS]
.__
---_.--- -___--.-.

A 5% S Pet Coke - 100% usage - 800 kcals/kg - 1.33% SO3 on

clinker

A For 0.6 Na20 eq. MR = 2 for Clinker SO3 of 1.55

A Above a MR of 1, K2S04.2CaS04 - Calcium Lambinite
and CaS04 form
A Calcium Lambinite decomposes at 152OC, CaS04 -145OC
in Excess 02
A 2OOOppm CO at above 1OOOC breaks down CaS04
A MR >2 in Stage w ray meal produces
hard deposits.
I,

-------.-... ----

-._-.-

---__-_-

FT
Sx-j
_--~
---- _---_--- ____.---__._COKE
___ -_.. -_.__._. _ -,_ PROBLEMS
__-.- -.- . ..-- - - --.-- AND
l

FOR 100% Pet Coke need to AVOID:

A Chemical Reduction in Burning Zone
A Over-burning
A Low Na20

o For 10096 Pet Coke need to IXAVE:

A Good FLAME
A Combinability < 145OC
A Good Kiln Control of Free Lime and Back End
oxygen

--.----------..-.. _ --...-_ ---..... --. ..-.

------

1 100% PET
AND . . -SOLW=]
.-. -. _---_.COKE
- -..-_--..- -.__.._ -- PROBLEMS
_. . . _ -._.._-_ . __ -_----_..----.
---l

Good FLAME as per CEMFLAME 1

A Burner Momentum - 7 N/MW
A Bluff Body 60%
A +9Ou residue 50% VM
A Central Burner
A Secondary Air > 800C
A Back End Oxygen - 3.5%

I
i

<

------.- ._-.- - .-.-. . .-. - _ _._ _._ . _ _ . . ._-------P-w-- .- .-. .-. .- - - - -

100%- PET
PROBLEMS
AND
-----I
---_-- -_--.-.--SOLUTIONS
-. . - c . -.- .--~-.--.--- - -COKE
0 Good FLAME alone not enough need:
Ilr. Good Kiln Control - Linkman to avoid continual
SO3 recycling
A Raw meal residue low enough for good
combinability
A 14.5 % Free Lime.
A Continual checking of performance
A SO2 probe to monitor and control, eg 3000 ppm
= 1.05% SO3 on Stage IV raw meal

!
.I! *,

I
i
!

-.. -

-.-- _..--___ ..__ .-- .__- --_----___-

.__- _-__.,. - _ - _

-COKE - PROBLEMS AND SOLUTIONS]

I.-._- ~ .____.

__.--,. . ..: ._I

-.-...

---.

. -

-.

.I

__.__

_... _,_.

..__.____

---...

o Now we know how to do it!

0 How can 100% pet coke be used with low
Nbx emission

CEMFLAME 2 has the ANSWER

I_-_-

Blue Circle Cement

PROCESS ENGINEERING TRAINING

PROGRAM

Module 13
Section 2

CETIC Sub Commission Behavior of

Volatile Material in Kiln Systems

ISTN 9215
CETIC ~~-C~~~~IssIOh

BEHAVIOtiR OF VOLATILE lMATERL4~ J.N =N

SYSTEMS
REVIEW OF CLASSICAL ECNOWLEDGE ON &mOR ELE&fEiiS - JUNE 1992
Summary
This report is based upon various documents originating within 3lue Circle and from external
sources, supplemented by accounts of experience in Works operated by CETIC member
companies. A review is given of factors governing the behaviour of minor feed constituents
(Cl, K, Na, S) in cycles originating i!t the kiln burning zone and of the consequent effects
on process plant performance. Recirculating loads of volatile species are formed and these
are often implicated in the formation of build-ups, coatings and bloc.kages in cooler parts of
a kiln system. Some relevant practical experience is listed and implications for product
quality are summarised.
In a given plant, burning zone temperature and atmosphere are the dominant driving forces
for these cycles. A key area for action to gain control lies in the selection and preparation
of fuels, burner settings and raw materials. Process control (including dust management) can
also be important, especially as regards selection and maintenance of sensors which give
direct or indirect information on burning zone atmosphere (i.e., chemically reducing or
oxidising with respect to the volatiles). Several dry process Works have found useful results
from study of the results of sampling and analysis of kiln entry meal on a regular basis.
Differences between pIants may arise from design features or from characteristics of raw
materials and fuels. Various modifications :o details of process design and operation/ control
may be used to alleviate plant difficulties or modify clinker quality, depending on the local
permutation of inputs and temperatures which is involved. For exampie, changes may be
made to cyclone geometry or non-stick linings added, or benefits may be found from
alterations to raw meal or fuel chemistry. These factors may also be considered when
selecting new equipment.
The prospects for further improvements in understanding and modelling of process
phenomena associated with volatile cycles have improved in recent years with the completion
of relevant thermodynamic studies in other areas of technology. The addition of secondary
firing or precalcinarion can significantly alter the behaviour of some kiln systems, and
appiication of the improved academic knowledge to some relevant situations may be worthy
of encouragement.

IS-m 9215
CETIC SUB-COiMiullSSION

BEHAVIOUR OF VOLATDLE
SYSTEMS

REVIEW OF CLASSICAL JKNOWLEDGE

amTERI4L.S IN KILN

ON IMINOR ELEMENTS - JrJiuE 1992

CONTENTS
Page No.
1

1.

INTRODUCTION

2.

VOLATILES:
WHAT ARE THEY AND WHAT ARE THE
MECHANISMS OF VOLATILISATION?
1

3.

CYCLIC NATURE OF VOLATILJZS

3.1
3.2
3.3
3.4

4.

EFFECTS OF VOLATILE CYCLES

WET AND (LONG) SEMI-WET PROCESS
LEPOL PROCESS
DRY PROCESS

CHANGES N PROCESS CONDITIONS WHICH CAN INFLUENCE

7
BERAVIOUR OF VOLATILES
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9

5.

N KILN SYSTEMS

PARAMETERS NFLUENCNG VOLATTLE RECIRCULATION

ITALCEIMENTI
CBR
LAFARGE
OBOURG
CMENTS FRANCAIS
ENCI
HOLDERBAlNK
BLUE CIRCLE

EFFECTS ON CLINKER

13

Sulfate retention
Reducing conditions
Fundamental aspects
6.

TOWARDS A MODEL OF VOLATILE CYCLES

14

7.

EFFECTS OF CONDENSATION

17

CETlC SUB-COIMMISSION

BEHAVTOUR OF VOLATILE IMATERIALS

SYSTEMS

IN KILN

REVIEW OF CLASSICAL KNOWLEDGE ON mOR ELEMENTS - JUNE 1992

1.

INTRODUCTION

This text, originally based on various reviews available within BCI and on external literature,
has been adapted on the basis of discussion within the CETIC Volatiles Sub-Commission.
During an examination of the available literature, some 190 apparentIy relevant articles were
identified in Chemicai Abstracts for the period 1967-1991. It is interesting to note that a
more detailed examination showed that 60 of these articles covered not only topics relevant
to kiln operation, the environment and associated laboratory tests, but also:
l

the production of mineral&d clinkers (despite efforts to exclude this topic

from the list)

low temperature clinkering by chloride addition

the cement/sulfuric acid process

the use of waste materials in kiln feed.

65% of these articles came from. Eastern Europe, suggesting that there may be a knowledge
base in the countries formerly behind the Iron Curtain. These subjects remind us that volatile
components are nor merely bad news for cement producers, but may have useful effects on
product performance and on economics in certain circumstances.
2.

VOLATI-LES: WHAT ARE THEY AND WHAT ARE THE MECHANISIMS OF

VOLATILISATION?

The concern is with elements or compounds which partially or entirely evaporate in the kiln
and which are transported towards cooler zones in kiln gases. Some can escape via the stack
or are trapped in precipitator dust. Others condense as they leave the kiln and/or react with
or condense on the feed in the preheater system.
The principal volatile elements are : K, Na, Cl, S, Pb, TI, Cd, V, Hg, Zn and As. The
elements F, V and As can be classed as moderately volatile and the others as
significantly.
They originate from inorganic and organic compounds in raw materials and fuel. In the case
of raw materials, certain compounds can readily volatilise at a temperature below 600 deg.
C, especially mercury, thallium and sulfur (if present as sulfide or in organic combination),
but in generai volatile compounds in raw materials only evaporate pat-My as the feed passes
along the kiln system and through the burning zone.. -The residue will be retained in the
product, either in solid solution in the principal phases of the clinker or as discrete

compounds - normally alkali chiorides and sulfates. Tne typical proponions of volatiles in
raw materials which evaporate while the feed passes through the !tiIn system are as foIIows
(primary volatiliries, according to certain writers):

so3
K2O

NazO
Cl, Hg
F
Pb, n
C d , V, Zn

c/o
60 - 90
30 - 70
20 - 40
96 - 9 9
1 0 - 40
60 - 99
IO - 20

In contrast, fuei voiatiles are almost always entirely evaporated during combustion.
Evaporated volatiles travel back up the kiln in the gas phase where:
i>

they condense on the feed and form the basis of a recirculating volatile load
in the kiln system

ii)

or they condense as a fine dust or fume which is finally trapped in the

precipitator and which wiI1 then become part of an external volatile cycle if
dust is partly or wholly returned to the kiln system

iii)

or they condense on a colder surface in the kiln system and form the basis of
a build-up

iv)

or they escape through the stack either in the gas or vapour phase or as a fine
fume.

The remainder of this text concentrates on the cycles of minor elements (Cl, K, Na, S)
originating in the burning zone. The cycles of trace eiements and/or cy
Iess hot parts of the system are not examined in detail.
When volatiles condense on raw meal and are returned towards the burning zone, depending
on the overall chemical conditions and burning conditions, they form a range of volatile
compounds which themselves evaporate partially and are partly retained in the clinker - either
as discrete compounds or in solid solution in the clinker phases.
In general, chloride preferentially forms potassium chloride (KCl) and only when there is an
excess of chloride over the needs for chemical combination with the available potassium will
sodium chloride (NaCl) be formed as a recirculating volatile species.
A large part of the alkalis and sulfate entering the burning zone is in practice present in the
form of molten sulfates, forming a separate Iiquid phase immiscibIe with the principal ferrite
flux.

The typical proportions of voiatiie compounds which evaporate in the burning zone (in the
absence of reducing conditions) are as follows:
%
40 - 60
40 - 60
40 - 100
60 - 90
20 - 40
97 - 99
96 - 99
80-100

K,SOq
NaiSO?
2CaSO,.K$O,
K,O in solid solution
NaZO in solid solution
KC1
NaCl
CaSO,

In consequence, a cycle of recirculating volatiles develops, which only finds an equilibrium

when the total quantity leaving the system (in clinker, dust and stack Iosses) equals that
entering the system (in raw materiais and fuel). Various tables and figures are appended,
showing melting and boiling points and vapour pressures.
At this point the level of volatiles in recirculation can be significantly greater than the total
mte of introduction of volatiles. The volatiles in the vapour phase can be in various states
of dissociation and recombination. Alkali sulfates evaporate congruently, that is to say that
they disappear entirely when heated for a long period, whilst &SO, decomposes and leaves
residuaI CaO (in oxidising conditions). CaO:CaS03 melts can form. Typical recirculating
volatile loads expressed as % of the total quantity introduced are as follows:
%

Chloride

5ooiJ
200 - 650
150 - 200
200-800

K2O

Na,O
so3

The effect which these recirculating volatiles exert on the formation of build-ups at the kiln
(gas) exit depends on the composition (which governs the temperature of liquid formation and
thus the position and hardness of the build-up) and on the quantity (which governs the rate
of formation). The dependence on vapour pressure on temperature differs for various species
and the relative volatility (or rank order of volatility) can vary in the cold and hot zones
of the kiln system.
In the past, several empirical limits have been proposed for the concentrations of volatiles
admissible in a kiln, e.g., 0.03% Cl on clinker for preheater kilns. Today there is a
tendency to prefer to specify concentrations tolerable at the kiln inlet (in hot kiln feed at zero
loss on ignition). By way of indication, the concentrations of volatiles which can be tolerated
in the lower stages of a preheater are typically given in the following ranges (exceptions
being of special interest for study):
5%
Cl
1.2 - 1.8
2.5
- 4.5
so3
2.5.3.5
Alkalis (eq. Na,O)
3

The FL.5 encrustation index

R =

Total molar S inout

Total K,O input + 0.5 (Total NazO input)

can be used as an indicator of the potential nature of any build-up, as follows:

R31
0.7 < R < 0.9
R < 0.5

Hard build-ups based on SO,

Relatively soft build-ups (easily removed)
Carbonate-based build-ups in due course.

Most suIfates condense in the range 9000 - 1100 deg. C. The presence of fluorine can
aggravate build-up problems due to the formation of fluoride compounds and. their aid to the
formation of various silicates.
Potassium chloride alone condenses between 800 and 900 deg. C (and sodium chloride at a
slightly lower temperature). Build-ups can develop in the kiln feed chute or in the riser duct
towards the bottom stage of the preheater.
There is an optimum temperature for the capture of SO, by a freshly cakined raw meal
(e.g., 880 deg. C in one study). Primary condensation is expected to be in the form of liquid
alkali sulfates. It is often interesting to calculate the composition of the sulfate phase in kiln
inlet material, with its addition of KCl. Fusion in the system Na,SO,/ K$O,/ CaSOJ KC1
begins at below 700 deg. C. Liquid films on dust particles are The origin of build-ups and
as the thickness increases, internal temperature drops and new equilibria may be established
and compounds formed.
There is less literature on the phenomena governing condensation than on evaporation. The
effects of atmosphere and the implications of precalciner operation are perhaps worthy of
study, especially for sulfate compounds which can be present in various states of oxidation
and which can react with water vapour to form bisultites and bisulfates. The characteristics
are as follows:

Chemical
State

Fuel and raw material

Clinker

Vapour

Oxidised
s4+, 9

Sulfares, e.g.,
gypsum, anhydrite,
alum, etc.

Alkali sulfates
and aBali/alkali
metal sulfites

SO2 and (at

low temp.)

Elemental S

Neutral, S

Reduced,
S2-

so3

Pyrites, marcasite
organic sulfides

Oldhamite, CaS,
complex suiftdes of
calcium, aluminium, etc

Non-volatiie

The oxidised sulfur cycle is illustrated among the appended figures.

3.

CYCLIC NATURE OF VOLATILES N JSXN SYSTE-MS

3.1

EFFECTS OF VOLATILE CYCLES

The volatilisation and condensation of volatiles in a kiln produce two undesirable effects on
the kiln process:
l

the formation of build-ups - possibly blockages and the possible emission of

SO, from the system

transfer of heat away from the burning zone towards cooler parts of the
system.

There are also effects on the clinker, which may be summarised as fofollows:
Fluxing action:
a
lower temperature of first liquid phase formation
l
change of liquid viscosity
l
alteration of surface tension of liquid
0
modification of crystal morphology.
5

Phase re!arions:
0
the relative thermodynamic stabilities of the clinker minerals can be altered by
solid solution effects.
Hydraulic activity:
0
the reactivities of the clinker minerals are altered by solid solution and/or by
the effects of crystal symmetry (high temperature stabilisation of polymorphs)
and/or effects occurring during hydration (e.g., coating of cement particles by
insolubie salts).
(It is difficult to isolate these three classes of effect in practice.)
3.2

WET AND (LONG) SEMI-WET PROCESS

There is an obvious external cycie (dust return) as well as an internal cycle which has,
perhaps, received less attention in the literature.
In an example involving a coal with a sulfur content of 4.5% (not very differ&t from the
level for a blend of coal and sulfur-rich petroleum coke), there was a high internal cycle and
significant stack losses. In one BCI kiln these represented 42% of the quantity introduced
to the system and were 5 times higher than the proposed BATNEEC concentration.
Typically, FLS expect 30% or more of the sulfur entering a wet process kiln to escape via
the stack. The external cycle can be influenced by the proportions of dust returned to the
system by various routes.
3.3 LEPOL PROCESS
The volatile cycle is more complicated on account of the additional external cycles due to
dust from riddlings and cyclones, even though the basic situation seems similar to that for
wet process kilns. However, there is a major opportunity for a volatile bleed (especially
SO,) by disposing of riddlings and cyclone dust as well as that from precipitators.
3.4 DRY PROCESS
As indicated above there are both internal and external volatiIe cycles.
In general, sulfur escaping the preheater is found in the form of dust rather than SQ and
forms part of the external cycle involving the raw mill, conditioning tower and precipitator.
Further, it is expected that SO2 in the gas phase will be absorbed in the lower stages of the
preheater, at least if there are no adverse conditions (e.g., chemically reducing conditions).
When sulfide is present in the feed, a significant fraction will be lost from the preheater as
SO, (e.g., 30 - 50%) and partly absorbed in the raw mill and the precipitator system.
6

4.

CHAtYGES N P R O C E S S
BEfIAVIOLJR OF VOL;ITILES

4.1

PAJiUMETERS

CO?iDITIONS rvHICH C&Y INFLUEYCE

NFLUENCNG VOLATILE RECIRCULATION

The principal parameters which influence volatile behaviour are as follows:

temperature
time
type/composition
concentration
diffusion towards the solid surface (as controlled by clinker nodule size and
flux content and composition)
gas/solid contact
other reactions
kiln atmosphere (laboratory tests suggest that water vapour plays a role, as
well as the traditional kiln gas components).
(For summary, see table among appended figures.)
Once a kiln is in operation, the main parameters available for the kiln operator to
alter/control are th.e temperature and the atmosphere in the kiln. The rate of gas flow seems
to be of secondary importance. Process design and local chemical conditions also play a
part, determining the total quantities introduced to the system, chloride content, relative
concentrations, combinability, fuel and to the firing system parameters and mixing within
the clinker bed (with exposure of nodules to gas flow at the surface).
It is often difficult to distinguish chicken from egg in the industrial operation of a kiln. For
example, when CBR changed from gas (or indeed, to 92s) as a fuel in Canada in the past,
there were effects on volatile cycles. This is to be expected due to the presence of a higher
level of water vapour in the combustion products (e.g., coke < 5 % , lignite > 10%) and the
higher vapour pressures of alkali hydroxides (see attached figures), but such changes are
always accompanied by other alterations to the quantity and composition of volatiles
introduced and/or the raw meal chemistry. The effect of nodule size on volatile behaviour
seems equally. impossible to separate from that of the effects of liquid from alkali melts on
clinker size grading (as discussed in past CETIC work).
It is important to note that, in general, no-one measures typical volatile balances (save
globally): available detailed balances are usually gained when there is a need to investigate
a non-typical kiln system due to some problem or another. A number of illustrations follow,
taken from the experiences of participants in the CETIC Group.
4 . 2 ITALCEIHENTI
When CaC12 was introduced via the flame (0.3% on clinker) to volatilise K,O in a Gepol
preheater kiln of 2000 t/d capacity (Stockertown, USA), it was noted that operation could
continue for many
_- hours with kiln entry levels of Cl of,around 3% with somewhat reduced
7

loss on ignition (heat pump effect). This indicates the capacity of this type of preheater to
tolerate a little more Cl than cyclone preheaters (although long term results from Stockertown
are still awaited).
4.3 CBR
With chloride injection on a precalciner system (with bypass), there was a marked effect on
Cl and alkali levels in kiIn inlet material but not on sulfate. (The same was true at
Stockertown.) Full system equilibrium was not reached for three days. At another Works
(about 60% heat energy input at the precalciner and no bypass), an upper sulfate limit on kiln
inlet material of 3% was established in order to avoid any build-up problems (although
perhaps not corrosion problems!). In this latter case the sulfate level was lower in the
presence of chlorine and of potassium (see later data); at the same time there was also a
certain reduction in SO, loss from pyritic material in the preheater. It is known that this
precalciner rapidly forms build-ups above about 91.5 deg. C. In a wet process kiln, the
formation of rings can be followed with the she&scanner and the flame setting altered to
eliminate them (provided that action is taken within 2 to 3 hours).
At Lixhe, various phenomena were noted when returning to coal firing after the use of gas,
including a reduction in sulfate cycles and in the level of decarbonation at the preheater tower
exit. This latter phenomenon results from a lower heat release within the preheater, notably
from CaS04 recombination. The thermal effect is of the order of 75 kcal/kg with the inverse
effect in the burning zone. Blue Circle have calculated the magnitude of this effect as 109
kg/kcal at Hope Works in the past. At Lixhe (dry process) the sulfate level in the kiln entry
material can be reduced (and that of chloride raised) by increased burner momentum: wear
of the burner tip allows the level to rise again in due course. A non-stick kiln feed chute
lining from Hasle also gives good results.
Also at Lixhe, a number of interesting relationships have been established from the results
of three years operation (with analysis for volatiles in kiln inlet material twice per shift).
For example:
No. of kiln srops per monsh = -5.3 f 2.7 x %S03 (in hot meal)
for cyclone blockages)

I?* = 0.88

%S03 (in hor meal). = 4.8 - 0.36 x % K,O (in hor meal)

R = 0.68

It was noted that the input of used tyres at the kiln back end gave good results in mechanical
removal of build-ups.
4.4 LAFARGE
A compilation of volatilisation levels was made in 1985 for all company kilns. The

companys calculation of volatility is somewhat different from that often encountered

elsewhere.
The following results (which have changed little since that time) were
obtained:
8

so3

K2O

Na,O

I-w

56

34

14

Dry

80

69

26

Precalciner

55

49

55

(Retention levels are close to 100 % for preheater kilns and at lower levels for the other
processes, being related to the level of dust disposal.) If is noted that Lepol (and wet)
process kilns give Iower volatilisation levels than preheater kilns and that precalcinafion
seems to influence these phenomena. Nevertheless, due to the small quantities of dust
involved, the enrichment of volatiles in precipitator dust is highest for the Lepol process.
(Italcementi note that for balances determined on 23 kilns the enrichment factor for dust
compared with raw material is close to one for SO,, K,O and NazO in the dry process, but
between 9 and 11 for the Lepol; in the case of chloride the comparison is between a factor
of 8 for the dry process and 123 for the Lepol; the values for long, granule-fed semi-dry
kilns are intermediate.)
In examining the performance of a new low-NO, burner with a higher momentum than the
previous one, it was noted that the CO signal could be made to disappear (for the same C+
level) with less decarbonation at the preheater exit and a higher level of kiln drive power
(Amps). These effects were accompanied by an apparently higher BZT (with less clinker
free-lime), less gaseous SO2 at the kiln exit, and kiln inlet material with less SO, and a
higher loss on ignition at 1000 deg. C.
4.5 OBOURG
Here the clinker KzO level is used as a control parameter for the wet process tiln and a
considerable data-bank of measurements has beeri built up. At the end of 1990 the use of
a lower ash fuel began and the clinker K,O level dropped from 0.69% (1990 mean) to 0.55%
(mean for January/February 1991), despite the introduction of a little more potassium to the
system (equivalent to. 1.43 % on clinker instead of 1.34%). Evidently the dust - returned to
the kiln - had become more rich in alkalis. It can be shown that the proportion of $0
brought in by the solid fuel fell from 24% to 17.5% while at the same time that brought in
by the dust return rose from 18.5% to 29 %. This observation leads to the conclusion that
$0 incorporation in clinker does not only depend on the quantity introduced but also - and
above ail else - on the type of material which brings it in and perhaps on the position where
it is injected. On screening the clinker at 20 mm the same chemical analysis was found for
each fraction with the exception of K,O, where the concentration was about 10% higher in
the coarse fraction.
Here the clinker alkali content has successfully been reduced by chloride addition at the
flame. Ciments Franc$s have also demonstrated this effect with CaC12 addition to slurry
(3.5% K,O in the dust in comparison with 2.7%). Blue Circle also once succeeded in
demonstrating the efficacity of this method (accompanied by non-return of part of the dust)
9

in lowering clinker K,O ieveis by a half in a long kiln, despite the fact that some questions
remained about the most appropriate place to introduce the chloride; it must be noted that
there can be problems with the flow behaviour of precipitator dust above a certain chioride
level.
4 . 6 CIivIENTS

FRANCAIS

On adding precalcination to a Lepol grate with (stoichiometrically) more sulfur than alkalis
in the system, an increase was noted in sulfate and alkali levels at the grate and especially
in the dust and nodules beneath the grate. More potassium sulfate was found in the clinker
with precalcination; CaSO, was found in the cyclone dust and somewhat less in the
precipitator dust, arising from kiln dust which had travelled across the layer of nodules on
the grate. There was a little more SO2 emission with precalcination. On another Lepol kiln
no such effects had been found on adding precalcination; at Frangey, Lafarge had noted a
somewhat higher recycle of potassium and sulfates with Lepol precalcination. (At this latter
Works, with use of chloride-rich substitute fuels, it is found necessary to carry out sampling
over periods of at least a week to determine consistent volatile balances.)
To use a high chloride coal (-0.15% Cl) on the dry process it has been found necessary to
prepare a mix with another coal to avoid build-ups with typically 2% Cl at the bottom of
cyclone 4. Analysis of build-ups along the kiln indicated that chloride levels reached 30%
(at zero loss on ignition) in the coating from the base of cycione 2 and 20% at 50m into the
kiln (despite its less than 5% level in both hot and cold parts of the kiln); run-out of material
from a stopped kiln gave chloride levels rising up to 0.7%.
During a past experimental campaign of burning chlorinated wastes, it was noted that the
simultaneous presence of chloride and sulfate at high levels could give rise to emissions of
HCI and SOz. Apart from this, the volatiles were all assimilated in the c!inker (burned at
low temperature with free lime levels of up to some 13% and formation of CaCI,.C,S) or
in the dust (and probably also trapped in refractories).
4.7 ENCI
After the addition of a second stage to the preheater, various build-up problems were
encountered. In 1985/6 the kiln exit 0, level was increased from 0.5 to 1.5%, the solid fuel
residue at 90 microns was reduced to below 25 % and several Cardox units were installed.
These actions improved the situation and in 1987 20m of Magotteaux stirrers were
installed, the burner air velocity was increased to 100 m/s and a Hasle non-stick lining was
installed in the duct and cyclone dip-tubes. These efforts made an output rate of some 110
t/h possible without build-up problems.
More recently petroleum coke (3.5% S) has been fired (first at 4.5 t/h and afterwards 6.5
t/h), accompanied by oxygen addition at the flame and, finally, by the use of slag (S - 1%)
as a raw mix component (5%, and then 10%) with yet more coke (8.0 to 8.5 t/h, i.e.,
- 60% of fossil fuel energy). At the start of 1991 the production rate was 120 t/h
accompanied by preheater blockage problems. Duringr ,199l it was deduced that problems
10

with build-ups could be avoided if the SO, level in the kiln entry material was kept beiow
2.5%. This was possible with a coke input rate of about 8 t/h and a slag level of 10%
provided the oxygen level at the kiln back end was kept consistently at 2%. If these
conditions are not met, then kiln operation rapidly runs into problems. The Fuzzy Logic
control system helps to achieve success, as the 0, signal standard deviation has fallen from
0.45% to 0.25% and that for the SO, in hot kiln-feed from 0.6% to 0.2%.
4 . 8 HOLDERBAhX
In the past, Iaboratory data have been gathered on minor elements and these confirm the
effects that are now more widely known. For example, suIfur volatility in a standard regime
(70% Nz, 30% CO) is close to 100% at 0% 0, but falls in the presence of 0,; nevertheless
the effect of O2 is much less at 1400 deg. C than at 1200 deg. C. The volatility of minor
elements in the laboratory is also much greater for powdered material than for granules.
(See appended illustrations, taken from external literature.) Currently there is interest in raw
meal morphology and in the distribution of volatiles in the meal at the start of clinkering.
A precalciner kiln system in Spain ran for many years with a Cl level in the kiln entry
material between 3 and 4 % (that is to say about 0.5 % less than K20) and with no trace of
CO. In this case there was about 1.1% SO, in the kiln entry meai and 4% K,O and no
problems, but if CO was present there was about 2% SO, and 5% I(,0 in the hot kiln feed
accompanied by Spur&e-based build-ups and cubic KC1 crystals. (It is also recognised in
Ciments Franc@ that regular kiln operation helps to minimise the phenomenon of
cementation by the freezing/thawing of chloride-based deposits.)
While burning wastes at Clarkesville (wet process), it is found necessary to ensure that the
clinker Cl content is always kept below 0.3 %, otherwise the kiln becomes unstable. At
Origny there have been an enormous number of kiln stops caused by preheater blockages
which, when sampled, do not contain many volatiles. This effect seems to have its origin
in a liquid phase formed by calcite arising from a chalk with an extreme level of fineness
which can decarbonate and recarbonate very rapidly. The problems have been much reduced
by altered cyclone geometry and helped a little by the use of a mechanical cleaning device.
4.9 BLUE CIRCLE
On one precalciner kiln it is difficult to find typical volatilities. For every determination
there are almost always different values (40% for SO, and 50% for K,O changed to 25%
and 4O%, for example). One can imagine that this is caused by variations in the nature of
the raw materials and the content in the kiln feed of sulfate (and sulfite) captured in the meal
after initial low temperature voiatilisation in the preheater.
In order to examine the possibility of producing a sulfate-rich clinker (2% SC&) using certain
available resemes of material and without installation of a by-pass on a new kiln, rests were
carried out for about three weeks on a dry process kiln at another site (-35 t/h). The
objective was to reduce the volatilisation in the burning zone by playing on process
parameters and producing a lightly mineral&d clinker (-- 1% K,O, -0.15% NazO, and
11

- 0.15 % F, as usual at that Works, but with double the usual level of sulfate) and with a
silica ratio a litT.ie lower than usual (-2.7 instead of -3.2), accompanied by a change in
alumina ratio from 2.8 to 2.2. During the changeover there was some tendency to form soft
build-ups in the preheater, but with the new regime established these moved towards the kiln
feed chute without causing any major problems for kiln operation. The apparent burning
zone temperature was reduced from about 1500 deg. C to 1380 deg. C, while K20 volatility
dropped from 70% to 60% in the burning zone and that of Sq from 80% towards the range
50% to 60% provided that kiln exit oxygen level was kept above 2%. There were
improvements in the output and fuel consumption of the kiln and, in fact, the experimental
Works adopted certain of these changes during its normal operation for several years, until
the asrival of demand for low alkali clinker.
At Hope Works (dry process), tests were carried out involving various NO, levels as well
as reducing conditions.

so, :

The ratio of SO3 in Stage IV to Scl, in raw meal varied

typically from 1.8 to 2.7 for the higher levels of NO, and was
3.0 for a low 0, level. The clinker SO, content fell.

KzO and NazO :

In a parallel manner, for K,O the ratio of the content in Stage

IV to that in raw meal varied from 3.8 to 4.4 and for NazO
from 1.6 to 2.0.

In general, reducing conditions increase SO, level. at Stage IV by a factor of 2, giving a

lower clinker SO,.
As already described in a paper to the CETIC Technical Commission, at Cauldon Works
(and later in other dry process Works) SO, has been monitored at the kiln back end to
determine the local rules for avoiding blockage tendencies. The SO, signal is noisy and
difficult to interpret without a knowledge of the history of the system, e.g., a reCent fall of
sulfate build-up material arriving in the burning zone can give rise to a high SO, signal at
the kiln back end despite the presence of a good flame and acceptable levels of volatiles in
the kiln entry material. At Dunbar there has been success in reducing the number of kiln
stops per year caused by preheater blockage from over 90 (1987) to less than 10 (1989), lost
time hours having also fallen from around 450 per year to about a hundred. (There were
also major gains in stops caused by rings and breakaways at the kiln entry seal.) There is
no need to keep such monitoring equipment in permanent operation once the rules are
established, but it expected that renewed investigations will be needed each time the
conditions of operation change. In practice it is now found that with this know-how sulfurrich petroleum coke can be used (to a certain level) even on dry process kilns which in the
past have given problems with just coal firing - but in several cases the build-ups seem to
have moved from the preheater into the kiln (where they are destroyed). In a general manner
it can be supposed that there are problems of both short-term and long-term and stability:
once a stable burning zone volatilisation is established, one must wait for stable conditions
to arrive higher up the system and in the large masses of material which form the build-ups
and coatings already in existence.

12

During tests of SO, monitors at the exit of a long kiln with filter cake feed, it was noted that

the signal usually remained stable (below 100 vpm). But, when the oxygen level fell there
was an inverse correlation between the 0, and SO, signals. In this case levels of around
1000 vpm SO, were reached, with considerable variations; it is supposed that cycles moved
further up the kiln (nearer to the analyser) during low oxygen periods.
At Plymstock Works (dry process), when changing the BZT from 1390 to 1500 deg. C the
ratio of SO, in the fourth stage to that in the raw feed rose from 1.2 to over 4.0. Similar
results were obtained at Lichtenburg Works (South Africa).
It is noted that a better understanding of volatile recirculation is useful for kiln operation
because back end oxygen indicators can sometimes be misleading.
At Masons Works (wet process), raising the 0, level gave higher levels of SO, and K20
retention in clinker (concentrations rising from 0.18 to 0.54% SO, and from 0.33 to 0.80%
K20). Also for the wet process (Westbury Works):
Kiln 1, low momentum flame,
2 % 0, - 2500 mg/Nm3 at kiln exit.
Kiln 2, good momentum flame,
2 % 0, - 2 5 0 mg/Nm3 SO, at kiln exit
1 % 0 , - 1350 mg/Nm3 SO, at kiln exit.
In practice alkalis are controlled on the wet process (by means of the LINKman system):
Masons:

% alkali target

NO, set-point
@pm at precip.)
550
4OO-500
300

< 0.55
0.55 - 0.7
> 0.7

Ravena (USA):
- alkalis controlled by NO, set-point
- sulfate/alkali ratio controlled by 0, set-point.
5.

EFFECTS ON CXANKER

Sulfate retention: In general, changin,0 from a situation with excess alkalis offers
advantages.
In general, a higher SO, content:
improves early strength
improves workability
produces a more difficult apparent grindability.
Increased ciinker alkali levels can also be associated with sulfate retention for Lepoi and wet

process kicilns, especially if there is already an excess alkali content.

13

The effects of minor components on the viscosity and surface tension of liquid phases can
be complex. Lower viscosities encourage alire formation. Calcium sulfate flux can,
however, stabilise belite and/or cause the production of clinker alite with lime inclusions.
(Sulfate liquid systems are capable of influencing ionic transport and chemical combination
despite the limited solubility of the principal clinker compounds.)
In clinkers with a low al*ka.li content, there is the possibihty of belite stabilisation (difficuli
combinability) due to excess SO,.
Reducing conditions: Reducing conditions in the burning zone can give a cement with poor
flow characteristics (due to free K,O and NazO), poor workability (due to the increased
content of C3A and its reactivity), poor strength (lower C,S content) and variable colour.
Fundamental aspects: At some future date, the production of lower LSF and/or mineral&d
clinkers may be of interest.
Alkalis retained in clinker are present either as stable sulfates or absorbed in the silicate and
aluminate structures. NazO has a more marked tendency than K,O to form solutions in CIA.
For ciinkers with (molar) ratios of sulfate:total alkalis below 0.5, almost all the sulfate is
combined in water soluble form, K$O, being predominant. A proportion of the alkalis are
in solid solution in the clinker C,A and this has an adverse effect on the initial cement
reactivity and thus on concrete and mortar rheology.
For ratios between 0.5 and 1.0, a certain quantity of langbeinite (2CaSO,.K$OJ is also
formed (and not all the alkalis are soluble). For ratios above 1.0, significant fractions of the
sulfates are combined within the sihcates and aluminates or as anhydrite (CaSOJ, which
dissolves more slowly than alkali sulfates, whilst the fractions of KzO and Na,O which are
soluble in water approach 1.0 and OS, respectively, at a ratio of about 1.5. At sulfate:alkaii
ratios above 1.5 trends are somewhat erratic. For most normal clinkers the principal sulfate
phase will be aphitalite with a maximum WNa ratio of 3.0. This phase is accompanied by
minor quantities of K-$0, and calcium langbeinite, Na,S04 being found only for unusually
low WNa ratios.
As well as the solid soiution effects and the formation of compounds described above,
various permutationsof volatiies (especially in the presence of fluorine) can influence the
stmcture of alite and belite crystals. (A comprehensive review of recent work forms part of
the text of G K Moir and F P Glasser at the 1992 International Congress on Cement
Chemistry in New Delhi.)
6.

TOWARDS A MODEL OF VOLATILE CYCLES

Various empirical volatility factors have been proposed and used with a certain measure of
success. This section considers the possible approaches to a more fundamentally based
model.
It is generally supposed that (other factors being equal) the extent of voiatilisation decreases
as the thermal efficiency of the kiln increases. An explanation may lie in the fact that this
14

is due to the limiting effect of vapour saturation by al,kaIi compounds. Studies by 3lue CircIe
of the treatment of kiln dust in a 100 mm diameter fluidised bed tend to confirm this
hypothesis.
This study examined the feasibility of producing a low quality clinker from flue dust with
capture of the alkalis distilled from the bed (for possible use in the fertiliser industry).
Saturated vapour pressures at 1200 deg. C are (for the pure substances):
KC1 0.18 atm
0.8 x 1O-3 am
K2SO4
(0.6 x 10m3 atm with decomposition suppressed)
Na,SO, 0.13 x lo3 atm
(0.01 x 10s3 atm with decomposition suppressed)
The transport capacity of air for vapcur at 1200 deg. C is thus
KC1

700 g/g
4 g/g
x0.5 g/g

K2S04

Na2S04

(The capacity at 1250 deg. C is about two times higher).

It can therefore be foreseen that (unless the equilibrium vapour pressures differ greatly from
saturated values) there will be little problem in removing KC1 from many kiln flue dusts in
a fluidised bed with a gas flow rate of, say, 2 g per gramme of dust, although the capacity
for sulfate removal may be limited.
(A wet process kiln typically operates with a ratio of a little less than 2 g/g gas/soiids in the
burning zone and perhaps 2.75 g/g at the back end; the corresponding values for the dry
process are 1.4 g/g and 1.94 g/g).
It is suggested that
V
v*
p*

=
=

P
M,
M,

=
=
=

(I*) (M,L_
(P - p*) c-q

saturated vapour concentration in transport gases (kg/kg)

saturated vapour pressure
)
) same units
of an alkali compound
gas pressure
>
moIecuIar weight of vapour
molecular weight of gas.

Given mathematical expressions for saturated vapour pressure as a function of temperature

and knowledge of the temperature profile in the kiln system, the saturated vapour
concentration can be calculated for each alkali compound and thus the maximum quantities
evaporated from the feed per unit mass of gases. In considering the amounts of gas passing
15

through the kiln at vaiou~ temperatures, the true quantity of volatiles transported per unit
mass of clinker can thus be calculated and from this knowledge, ideal volatile cycles can
be deduced. (It is to be noted that, paradoxically, when alkali addition allows BZT to be
reduced then blockage probIems can be lessened due to the dominance of temperature in the
evaporation mechanism).
Despite the fact that qualitative differences between two kilns (one dry process and one wet)
are reflected in sample calculations, such ideal calculated recirculating loads are about 10
times larger than those encountered in practice. The probable reaSons are:
Incomplete contact between gases and solids in the kiln, where only a small
fraction of the solid surface is exposed at a given time. (It is expected that
there is better contact in the colder dusty regions of the system.)

b)

Volatilisation characteristics of the alkali-containing minerals at a given plant.

(Alkalis seem to be lost more e&y from silicates and aluminates than when
present as sulfates).

cl

TEuISpOIt

d)

Reduction of vapour pressure over solutions of alkali compounds.

e>

Inadequate treatment of the transport of heat and of vapour within the bed of
clinker nodules in the kiln.

Formation of other compounds, e.g., &SO,, depending on the alkaksulfate

ratio.

Of compounds which are condensed ~/on solid dust or fume.

Unstable operation of production kilns, so that practical conditions are not

exactly those expected for very long term stability of temperature and material
flow.
The further development of a predictive model will have to take account of such factors, as
well as the effects of.composition of kiln atmosphere. In recent years there has been much
investigation of factors governing the blockage of cyclones and their performance in high
temperature coal combustion processes (in the hope of protecting turbine blades in direct
cycle electric power generation systems).
When lime (or limestone) is injected to absorb SO,, the compounds and thermodynamic
criteria of interest in combustion systems are exactly those encountered in the cement
industry - particularly when relatively high chloride coals are used. It is probable that there
is now sufficient academic knowledge to better treat our situation and allow an improved
modelIing and understanding. Another aspect to consider is knowledge acquired from study
of the regeneration of CaO sorbents used for SO, scrubbing: again, data potentially relevant
to kiln systems are produced, for example, on pressures of SO, in the system CaSO,/ CaS/
CaO in the presence of various concentrations of CO and CO, (see appended figures).
16

(It is interesting to note that volatile condensation has an effect on the kiln power signal used
for process control. Any perturbation of the chemical composition of the kiln feed which
raises volatile content will increase kiln Amps; a control strategy seeking a constant Amp&re
signal would have the effect of reducing burning zone temperature, yielding under-burned
clinker with a relatively high volatile content. A strategy using a constant fuel feed rate
would be equally inadequate, due to the depression of burning zone temperature produced
by the increased volatile load. The best control of product quality should result from a
system based on observation of the peak clinker temperature, that is to say, indirectiy by kiln
exit NO, control.)

7.

EFFECTS OF CONTXXSATION

The most probable primary condensation is in the form of liquid alkali sulfates, Melting in
the ternary system Na$SO,/ K,SO,/ CaSO, starts at below 800 deg. C. Addition of KC1
increases the range of suIfate compositions which is liquid of this temperature and allows
formation of liquid melts even below 700 deg. C. Deposits on the feed can provoke
chemical reactions; they can equally cause adhesion and - as with deposits on surfaces initiate build-ups. (Direct condensation as calcium langbeinite is not expected on
therrrtodynamic grounds.)
While the literature tends to agree (although not totally) on vapour pressures of pure a&li
compounds, information for the more complex species of interest in cement kilns is more
rare. Studies at Aberdeen University have produced self-consistent results for sulfosilicate,
sulfoaluminate and langbeinite. The order of volatility alters with temperature. As indicated
earlier, liquid alkali sulfate systems have a poor dissolving power for most of the principal
oxides of cement clinker. However, they have low viscosities and a low surface tension
against silicates and thus cover and englobe these particles very effectively. It seems likely
that the small quantity of silicate which is dissolved has a high mobility, so that the liquids
are effective at producing a reaction (for preference towards C2S at 700 to 800 deg. C.).
Stabiiisation of carbonates has been suggested (CaCO, can dissolve in the liquid phase in the
presence of alkali sulfates in the range 880 - 900 deg. C., forming a liquid rich in CQ and
the presence of fluoride can cause further complications - but equally (in combination with
certain concentrations of other compounds) certain advantages as far as clinker quality is
concerned.

C P KERTON
June 1992 (English version, December 1992).

)jats~c;ld

w D, Trans. hwhy Svc.. 66 (81, I966 - 1973. (197W.

rcrry R 11 & Chil!on C II, Chemical Engineers tlandlmk,

New York (S\h

cJi\ion

McGraw-Ililt,

1973).

Temperature (deg. C

~r3pt-1 to confirm that proportion volatilised

hating regime is dm-xteristic of raw mix

for oiven
5

(i.e., (fiat volatilisation reacfion is first order with rap+% to alkali content).

ha from Palmer K & Bayik 0, pc~

R
. epon M-117(1952] ud from

W&J

H. Rock ProdUCu 45 (2j.M - 68 (19413

/
I
I
i
II
f
i

I350

Temperature (deg. Cl

MeI tinq

point

ea -c 8 0 0
ca 440-#SO
edso-904
m9ou-954
CID

7 954

Meltinn &ae of (a) the CaSO ic SO -NazS04 System

4-w
and (b) the Effect of 6 KC1

(C)

(From

---c------ ,wpp- ------4~p-.

fue .air

Oxidized
Portion of the S Cycle Relative to the Kiln.
Dotted paths represent circulation in the vapour phase,
solid l i n e s i n t h e s o l i d p h a s e ( s ) .

SLJAlAl,~l~Y 01: 1~ACIO1ci

\\IIICII INi~LUi:NC~

Il~l~VIOU1~O1~~lLNOI~VOLA'I'IL~COMI~UNDS1N
1WaN SYSTEMS
1.

Burning zone temperalure

(alkali vapour pressures)

- level
- variations

2.

Temperature profile of burning zone

3.

Comp&ition of alkali liquid systems in the

burning zone

4.

Atmosphere in burning zone

globally reducing
locally reducing
water vapour

IIll!I

MAJOR
INFLUENCE

IOSSll3LE CONTICOL
ACTIONS

SECONDAItY
INFLUENCE

0
0

Soft burning (including fluxes and mineralisers).

Controlled burning.

Flame/burner settings.
Fuel Characteristics.

Selection of raw feed chemistry

(sulfate/alkali ratio).

0
0
0
0

Burner/flame settings.
Fuel characteristics.
Coal/coke fineness.
Choice of fuel (solid, liquid, gas).

5.

Clinker size grading

Selection of raw feed chemistry

(sulfate/alkali ratio) - including flux.

6.

Clinker llux content (density)

As above.

7.

Thermal efficiency (gas flow rates)

Process design/seleclion,

8.

Preheater system design

vertical
cold areas/air inleaks
anti-build-up lining
geometry
solid/gas loading

9.

Precalciner:

design
operation

LO.

Composition of alkali phase in prehealer

11.

Dust return

12,

By-pass system

0
0
:
0
0

Process design/selection.
Elimination of air inleaks / Insulation.
Design.
Type of precalciner.
Throughput.

Selection of precalciner.
Control of precalciner.

0
0

Selection of raw mix components.

Selective quarryinglbcneficiation,
0

ConHol.
Add.
0 = rrrrcertoin

X
0

Drnper

1, PCA Rcjwt MRS.68 (19.54).

Gtrr C B Kcil I:, IIZ (4). 7 _ 9 (I9Mj).

DUNBAR-HOURSLOST
HOURS LOST

DUNBAR-KILNSTOPS
NO. OF STOPS

!xo

100

400

80

._

300

60

200

40

100

20

0
PREHEATER BLOCKAGES
1987

RINGS/KIN INLET BUILD-UP

1988

1989

PREHEATER .BLOCKAGES

RINGSMILN INLET BUILD-UP

C2K53

lC

Log plot of the total pressure, in atmospheres,

of rhe decomposition products of various
sulphates occurring in the cement kiln.
Abbreviations:
C5S,S = 2Ca Si0&aS04
C A s= 3Cail 0 .CaSO
&$33 = 2CaS&k2S044
Choi G-S & Ckser F P. Cement & Ccmxete Research 18. 367 - 374 (1988).

Phase diagram at atmospheric pressure for the system

containing the solids CaSO,, CaO and CaS, as well us
the gases SO,, CO and CO,. The reducing potenti equals
p(CO)lp(COJ. Region A is where Cu# and CuSO,
are the only solids; region B Fras &SO, repkced by CaS
(see text). Contours of equd p(SOJ are
c shown.
ayhurst

A N & Tder R F, I Inst. Energy 64. 212 - 229 (fgg1).

Blue Circle Cement

PROCESS ENGINEERING TRAINING

PROGRAM

Module 13
Section 3

Investigation into Potential Low

Temperature Volatilization

INVSSTICATION IN'@3 PV!ZNTIAL LOW TFXPEXATUEE VOUTILISATIOH

WITEIBT?BP~O&B3D

OxFo3DH!.3wOKgSKIL9,

PBSCALCINEB

AHDPREHUTERS-fSPERS

Objectives
1) To identify potential lov temperature volatile compounds in the raw
materials of the'new Oxford Works,
2) To assess the quantities of volatile8 present ia the gas and Baterial
streams at partiizular positions vitbin the system, in particular vithin
the preheater and precalciner sreas.

3) To establish whether these calculated values will affect the conolusions

on the size of bypass required for the nev Oxford Works.

The rav materials for the propose ,xford Works are fairly rich in alkalis,
snd sulphates (up to 1.546 So3 and % 820). A bypass has been proposed l ti
bleed off a proportion of the kiln'& to prevent these alkalis and sulphates
causing problems in clinker quality and build up vithin the preheater and
precalciner, The design of this bypass has been discussed extensively in the
technical note STH 41/73. This note covered almost all possible condition8
which could affect the design of the bypass. flowever one area of uncertainty
highlighted by SlY gl/lj vns the possibility of significant raw material
volatility (;,3C%) at low temperatures (<loooOC). Engineering B & D vere
contacted as the problem was of a similar nature to the work carried out in
the investigation idto the processing of cement flue dusts.
This note applies'the theoretical prediqtions of the previous flue dnst
work together with other infonrration available fmm pubUshed literature to
the potential problem of low temperature volatilisation vithin the Oxford
preheater/precalcin#r system.
Conclusions
1) Of the potential problem causing corn unds investigated (K2SO4, Ba$O4,
KC1 and sulphur'contaiuin& compound8 p" l?a$O4 could be effectively dis
cussed a8 a low temperature volatile (see Table 1).
2) If the precalciher ia operated at normal temperstures (-SW'C) then
the amount of Kj$SO4 in the vapour phase is insignificant. flovever at

higher tea eratures (*- 95OOC) a significant amount (.-low4 Xgs per
clinker P of K2SO4 can be in the vapour phase. Tnns a high temperature in thi precalciner should be avoided as it vould have a
detrimental effect on kiln operation.

Kg

3)

The vast Gjoritp of K2SO4 and rJa$Oq will be in the condensed phase
on the suspended solids at the bypass position. This would mean that
bleeding gas alone at the bypass position vi11 not reduce the quantity
of alkalis in the kiln system. The most effective uay of reducing
alkalis would be to bleed off the dust on which the alkalis condense.

4)

The quantity of KC1 in the precalciner for a precalciner operating

temperature of 850 - 9ooOC is high (1-a on clinker) but this-is
consistent with the values predicted in SE7 81/13 and;
F&RlIfS
of normal vorks operating experience. Thik is taken into acconnt in
the bleed requiment.

S)

The effect of sulphur is summarised Fn Table 2. This table demonetratee

that the source of sulphur, the kiln exhaust covsitlon and the presences
of other minerals determines whether the sulphur will be bled off by the
bypass, recycled within the preheater, or exhausted from the stack.

6)

If sulphur is present as calcium sulphate, or calcium sulphide tha t?o

controlling parameters on the volatility are the gas temperature and gas
composition. Houever the effect of gas composition reduces with decreasing temperatare and the evolution of SO2 will not be large (7% IBX) at
temperatures experienced &the kiln back end, and will notie sicpificantly higher with reducing conditions than with oxidisiq conditions.

7)

Low temperature volatilisation due to the presence of certain mineral

organic natter or iron pyrites will result in an increased SO2 emission
from the stack but should not have any detrimental effect on the wlphur
cycle in the precalciner/preheater system.

8)

Sulphur from the kiln fired fuel will tend to form a recirculating load
betueen the kiln and preheater unless bled from the system via the bwss.

9)

Sulphur from the precalciner fuel will form a recirculating load in the
same manner as sulphur from the kiln fuel. Houever this wxwt be bled
from the system until the sulphnr has been recycled.

10)

Approximately 9546 of the So2 in kiln and preheater gas streams will be
absorbed by the CaO in the kiln riser duct providing that the gas ad
materials stream are in contact for sufficient time for an equilibrium
to be reached. This time is very short due to large surface area of CaO.

Recommendationa
1) Further work should be carried out on the Oxford raw materials to establish the form in which the sulphur is present in the kiln feed.
2) As the wunt of lov temperature volatilisation is dependant on the
form of sulphur in the raw meal a standard method for determining raw
material volatility should be developed. Lf such a method were available
it could be carried out as a normal laboratory test on the raw mater!-'
The information could then be used either for design information for
future works, or as inforznation to the kiln controller to enable mar
efficient operation of the bypaas and kiln.

3) When the form of sulphw in the raw mterial has been established,
a simulation of tjhe Oxford Works raw meal could be tried out at an
existing XI W works. (possibly Plymstock). %upling throughout
the system would then show where, and at vhat temperature, volatilisation
and condensation of alkali and sulphur containing compwnds occuza.

CONTEXTS

Objective
Conclusions
Recomendations
1.

RiTRODUCTION

2.

THROREXCAL CONSIDEXATIONS
2.1 Volatiles Present in the Rack end Gas
2.2 Calculation of the Kaxidmii3

Amount of Katerial in

the Vapour Phase

3. SJLPZUR

COFiAlfJING COXFOUXDS

3.1 Reaction of CaO and SO2

3.2 Effect of Reducing Conditions on the CaS/CaS04
Equilibrium
3.3 Effect of Iron Sulphides

3.4 Effect of Xinerals

3.5 Effect of the presence of organic sulphur
C' i
contaminous compounds

7
8

3.6 Effect of sulphur in the fuel

4.

CONCLUSIONS

5.

RECOXNENDATIONS

10
11
12

7) Estimated Alkali Vapour in the Kiln/Preheater

13

Gas per Kg of Clinker (no bypass)

2) Effect of Sulphur on Raw Material

Volatility

74-15
16-23

a. CRCULATION

LIST

24

INVESTI~TION

IF!$ POTENTIAL LOW TdI!XPERATURR VOIATILISATION

WITRIN T3Ei PROP$SXD OXFORD NEWWORXSKIE?, PFECALCIHER

ANDPREHEmmsYsTEt?s

Objectives
1) To identify potential low temperature volatile compounds in the rav
materials of the hew Oxford Worka.
2) To assess the qua&ities of volatile8 present in the ~8 and material
streams at particular positions within the system, in particular VithiL
the preheater and precalciner areas.

3) To establish whether these calculated values vi11 affect the ccnclnsions

on the size of bypass required for the new Oxford Works.

1.

INTRODUCTION
The Services Department of Research Division have produced a comprehensive
assessment of the siie bypass that would be necessary tc prevent process
problems due to volatile alkalis/sulphates within the pmposed Oxford new
Works kiln syste (STN 81/13).

The report highlights problem sreaa micularly with reference to the effect of possible volatflisation vithin the
prscalciner or preheater system.

The chemistry of the compounds concerned

is of a similar nature to tbat of materials studied by Eagineeriug Research

and Development of the processinc of cement flue dusts. This note apples
the theoretical predjctions of the previous flue dust work, together with
other infomation awilable from published literature, to the potential
problem of volatilisation in the Oxford preheater/precalciner system.
The assumptions nsed'in SIN St/I3 acknovledge that a certain amount of lov
temperature volatilisation may take place and allows for up to 3096 lov teaperature volatilisation with a bypass of up to 30% Rovever if low tenpersturc
volatilisation vere to exceed 3cq6, then the following problems could occur.

il

The build up of a rec$rculating load within the preheater system

causing the kiln feed to have a concentration of volatile cornpnents which could exceed the design capacity of the bypass.

ii)

The pssibility of material build up uithin the preheater system

causing blockage problems.

iii) The release of vapour particularly So2 at a temperature below the

temperature of the formation of free lime. This would mean that
the bulk of the SO2 would not be recaptured by lime but released
from the system in the stack exhaust gas, causing the bypass and
preheater design to be oversized.
2.

THEORETICAL COEJDEFUTIONS
2.1 Volatiles Present in the Back end Gas
.
The first stage of the investigation was to identify all the volatile
compounds present in the kiln exhaust gas; then from themdynamic
IL -.J. c
information on>heir ;zD ehavio.ar with temperature, compounds which could
cause problems could be identified. The standard volatile com_r>onentT
of X20, Na20, So

and CL were considered.

The form in which these

components will recondense is well documented, and is listed below.

1) Chloride will recondense as KCl,
2) K20 remaining and Na20 ,will recondense as K2sD4 and Na2S04.

3) If there'is excess sulphate, as in the Oxford case, Cam4 will

be formed which preferentially forms the double salt (2 CaSO

g2s04) *
2.2 Calculation of the Maxfmum Amount of Material in the Tapour Phase
The solid gas phase equilibria of El, K2S04 and Ba2sD4 has been
/
extensively studied at Barnstone by Khor?.
Prom curves of saturation vapour pressure tempera-e for KCl, %s04
and Na SO (see Fig. 1) the saturation mass of vapour in the exhaust

gas can be calculated using the equation:

Where Ys equals the saturation mass of aUralA vamur in the exhaust

e-9

Pa

is equal to the saturation vapour pressure (obtainable from Pig. 1)

is the total pressure (atzn)

Mv

i&-the molecular weight of the vapour component

M
g

is the molecular weight of the gas.

Taking the total pressure to be one atmosphere and the molecular weight
of the gas to be approximately 30 we obtain the curves shown in Pig. 2.
If we assume that 4@ of the fuel is burnt in the kiln and 6% in the

precalciner, if there is no 'bleed, the mass of gas to clinker can be

approximated to 0.55 Kg gas per Kg of clinker in the kiln and 1.7 X3
gas per Kg clinker in the precalciner.

Using this information we can

obtain the CuIves shown in figure 3. which give the narimum vapouz
carrying capacity of the gases.

The czurve at a higfier level in the

precalciner indicates the higher gas quantities in the precalciner.

By assuming the burning zone temperature to be 1450C,
temperature to be 1050C,

the back errd

and the precalciner to have a maximum tezp-

erature of 950C Table 1, can be constructed. This shows that we

would expect all the available KC1 to be vapouris-A

hofnr- -a-h&

the kiln,? small amount of K.$04 to be'vapourised.&d

quantity of Na2SC4.
erature (85O'C

a negligible

If the precalciner is operated at a lower tv

is more ty-pioal.)

then the amount of K2SC4 vapourised

also becomes negligible.

This analysis recognizes KCI. as low temperature volatile, hoverer
this is also recognized in Sl!S 81/13, and the quantities of KC1 predieted in .the precalciner region by STB 81/13

(1.6% on clinker) com-

pares well the quantity predicted by this analysis over a temperature

range of 850 - 900C (1-246 on clinker). The results of this

fundamental approach show that the assumptions made in Sn 81113
concerning chloride recirculation are reasonable.

As the assumptions made in STM 81/l-J ~~ncernina alkali recimlatian are stx-mn
to be reasonable

bv this inventiaatlon. anv wtemlal aaerstinz

procuemz voula nave w ari8e rrom an cute-rnatlve

source, prooablg

from sulphur contsining compounds. This is discussed in the

following section.
3. SUUHUR

CONTldNISG

COMEWXES

Sulphur can be present in the kiln system from several different sources.

1)

Calcium sulphate and calcium sulphide in the rair materials.

ii)

Iron sulphides in the raw materials.

iii) Sulphur present in organic matter in the raw material.

iv)

Sul?hur present in the fuel.

Ro similar partial pressure data to that used in the previous section wa

available for CaSO

CaS and iron sulphide. In this case a fundamental

4'
approach considering the thermodynamics of the reaction of SO2 and CaO MS
considered.
This investigation was complicated by the fact that the equilibria and
rate of reaction are affected by the gas environment in which the reaction
takes place e.g. reducing or oxidising, and the presence of certain minerals
in the rav meal e.g. Si02, MgQ and FeO,
3.1 Reaction of CaO and SO2
Some of the many possible reactions of CaO and SO2 are listed belou3.

I)

caso4(s)

CaS04(S)

yEI)
'O(g)

+
+

=2(s)

ca'o(8)

+
+

O2(g)

=2(g)

co2(S)

3)

cs4(*)

4)

C=s4(s)

5)

+ @(*) + cs(*) + 4 co(g)

+

4co(g)

cas(8)

3 -4(9) + C"s(s) + 4 ys>

4c02(g)

+ 4 =2(g)

Beactions 2, 3 and 4 will only take place under reducing conditions

and these will be discussed later. Of the two remaining reactions
laboratory tests 4 on the adsorption of SO2 on cement raw meal have
shorn that considerable adsorption of SO2 occurs in the temperatare
range 600 to

gooc according to the reverse of equstion (5).

6) 4 CaO(s)

+ 4

a2fg) 9 3 C=s4(s) + C=Scs)

From thermodynamic data we can obtain values of enthalpy, entmpy

and Gibbs free energy for the reaction at different temperatures.
Then using the Qan't Hoff isotherm the value of Kp (the equilbrium
constant) can be found.
DG

-2.303

%T Log

K -

(2)

F'rom the stoichiometry and assuming the actitivities equal to unity,

it can be seen that Kp depends solely upon the partial pressure of
&J ;,J/*L;
4.) & t h e
From
this
knowledge
we
can
obtain
a
curve
(Figure
=2*
extent to which #e reactior6progreases to the right hand sic with
_~._. --.---increasing temperature. Th2.s curve show that, at a kiln back end
temperature of 1050C all the SO2 would be in the gas phase vhere as
over the precalciner temperature range (85O'C - 95O'C)

there would be

between 2% to 1446 dissociation to fern S02. This is within the 3096

low temperature volatillsatlon allowed for in the report STH 81/13.
This equilibrium is for a static sitcation and does not consider the
removal of SO2 from the system or the rate at which the reaction
proceeds.

In the kiln/preheater system the compounds are constantly

being removed and replenished and so the equilibrium is also dependant on reaction rates and residence time of the compounds. This

analysis does shou however that the bulk of the SO2 will be generated
from the decomposition of CaSo4 at a temperature greater than the
typical temperature of the precalciner and so a bypaes at this Mgher
temperature would reduce much of the So2 available for reforming CaSO

CaS in the precalciner.

3.2 Effect of Reducing Conditions on the CaS/CaSO, Equilibrium
-t
If sulphur containing limestone is roasted in air all the salphur is
converted to C&O4 by the reverse of reaction (1)4.
7)

CaO

So2

02

+ Cas4

Reduction in the level of oxygen causes reaction (6) to be favoured

which has bee,n found to be the major reaction in cement kiln exhaust
gases. Further reduction of the level of oxygen causes reactions (2)
and (4) to be favoured. The extreme case of reducing conditions is
given by equation (3).
If CaS is present a further reaction can occur with carbon dioxide5
according to the equation.
8) CaS

+ 3 co2

= cao

-+ so2

3co

!I'he temperature dependance of these reactions have been studied by

Turkdogen and Olsson5 who obtained expressions for the equilibrium
constants

10gpm2

(p~o/pco2)3

lo~W2

(pC02/pcO)

= - (20,000/T) + 9.27
=

(9617/T)

8.021

From these expressions the salphate/sulphide

(3)
(4)

equilibrium diagrams can

be dravn vhich are shown in Figure 4, 5 and 6. These curves show that
adjustment of the ol;ygen potential of the kiln atmosphere will adjust
the level of sulphur retentio:n
at higher temperatures.

and that this effect is more sensitive

At the relatively lov temperatures, 95O'C and

ii) the presence of potassium containing minerals very much enhances

the decomposition of CaYO

when compared to the effect of analogz

non potassium co&ainiry~ minerals.

These obseroatione are valid

for temperatures below 12OO'C and at atmospheric pressure.

The presence of minerals which enhance evolution of s02, by the
same mechanism will obviously retard its recapture. This could
mean that SO2 is lost from the system via the stack in greater
amounts than that predicted by STN 61/13.

3*5

Effect of the presence of orwic sulphur contaminoua..compou.nds

Another form in which sulphux can enter the kiln syetem is in the form
of vegetable matter. !&is vi11 evolve SO2 at very much lover temperatures
than mineral based

. If this was the mjor form of sulphnr in the rav

%
materials, this uould mean that almost all the raw material sulphnr would
be lost via the stack causing the bypass tc be very much oversized. The
effects illustrated in sections 3.4 and j.5 can be investigated by eqerimerit and would ahow up as enhanced volatility at lair temperatures.
Experiments carried out on the Oxford raw material have not shown this fn
be the case.

3.6 Effect of sulphur in the fuel:

Sulphur in the fuel will be present in the kiln and precalciner gases a8
From the discussions of their reaction of CaO and SO2 (section j.1)

=2it was stated that considera't)le amonnts of sulphur vould be absorbed at

600 to 900'~.

In fact the peak rate of absorption for SO2 is at 850C5,

Other factors affecting the amount of sulphur absorption are the concentration of CaO and S02, the surface area of the CaO and the
. exposure time.
In a kiln and preheater system the time for vhich CaO i8 exposed to So2
at a temperature where absorption can take place 18 very short. HoweV8r
the fineness of the CaO particles means that an enormous surface area of
CaO is presented to the gas stream.
This is generally the overiding factor and the reaction vi11 tend to
reach equilibrium very quickly.

Hence in the stage IV preheater cyclone8

there is almost 100% absorption of SO2 by the feed.

With a precalciner the reaction is complicated by two streams of
gas and material, at different temperatures combining. The precalciner gas stream will contain So2 from the fuel. The precalciner
_..
material stream carriesbffsolid of increasing CaO content (see ?ig.
8). Eouever there will be little recombination of CaO with SO2 in
this stream, due to the gas temperature of approxinrately llOOC.
(see Figure 4).
This will combine with gas and material streams from the kiln in the
riser duct from the kiln.

(Por the RSP ONOQA system shorn in Figure

8, a mixing chamber is used),,

This material stream will also contain

CaO and So2 from the kiln fuel and rav materials. The gas temperatare
is also around llOOC and so there &ill also be very little combination
CaO with S02.

As the gas and material progress up the riser duct the

endothermic production of CaO from CaCO

streams.

coolstthe gas and material

This makes absorption of SC2 by the time more favourable.

The absorption should have reached a maximum at the point uhere the
gas and materials streams separate.
be close to the equilibrium <absorption
at about 9% (see Fig. 4).

The absorption at this point till

for this temperature (-. 900C)

This is borne out by operating experience

with suspension preheater kilns.

The presence of a bypass will mean that kiln produced SO2 will be bled
off immediately whilst SO2 from the precalciner will be recycled before
being bled from the system.

4.

CORCLUSIOBS
1) Of the potential problem causing compounds investigated &So,, Sa2s04,
KC1 and sulphur containing compounds),Na2S04 could be effectively die+
cussed as a lov temperatie

volatile (see Table 1).

2) If the precalciner is operated at normal temperatures &850C)

then

the amount of K2sD4 in the vaponr phase is insignificant. Sowever at

higher temperatures (- 950C) a significant amount (u10W4 Kgs ger
Kg clinker) of K2S04 can be in the vapour phase. Thus a high temperature in the precalciner should be avoided as it would have a
detrimental effect on kiln operation.

3) The vast majority of K SO and Na So will be in the condensed phase

2 4
2 4
on the suspended solids at the bypass position.

This would mean that

bleeding gas alone at the bypass position will not reduce the quantity
of alkalis in tha kil,~l system.

The most effective way of reducing

alkalis would be to bleed off the dust on which the alkalis condense.

4) The quantity of KC1 in the precalciner for a precalciner operating

temperature of 850 - TCO'C is high (1-a on clinker) but this is
-4%
consistent with the values predicted in ST3 01/13 and&rem results
of normal vorks operating experience.

This is taken into account in

the bleed requirement.

5) The effect of sulphur is wbrised in Table 2, This table demonstrates
that the source of sulphur, the kiln exhaust composition and the presence
of other minerals determines whether the sulphur vi11 be bled off by the
bypass, recycled within the preheater, or exhausted from the stack.
6) If sulphur is present as calcium sulphate, or calcium sulphide the two
control&g parameters on the volatility are the gas temperature and gas
composition.

3ouever the effect of gas composition reduces with decreas-

ing temperature and the ev$ution of SO2 will not be large (1% mar) at
temperatures experienced annthe kiln back end, and will not be significantly higher with reducing conditions than with oxidising conditions,

7) Low temperature volatilisation due to the presence of certain minerals,

organic matter or iron Fyrites will result in an increased SO 2 emission
from the stack but should not have any detrimental effect on the sulghur
cycle in the precalciner/preheater
8)

system.

Sulphur from the kiln fired fuel will tend to form a recirculating load
between the kiln and preheater unless bled from the system via the bmsa.

9) Sulphur from the precalciner fuel will form a recirculating load in the
same manner as wlphur from the kiln fuel.

Houever this cannot be bled

from the system until the sulphur has been recycled.

10) Approximately

9546 of the SO2 in kiln and preheater gas streams will be

absorbed by tne CaO in the kiln riser duct providing that the gas and
materials stream are in contact for '&fficient time for an equilibrium
to be reached.

5.

This time is very short due to large 8urface area of CaO.

RECOI+2EDATIOBS
1) Further work should be carried out on the Oxford raw materials to establish the form in which the sulphur is present in the kiln feed.
2) As the amount of low temperature volatilisation is dependant on the

form of sulphur in the raw meal a standard method for determining rav
material volatility should be developed.

If wch a method were available

it could be carried out as a normal laboratory test on the rau mterial.

The information could then be used either for design information for
future works, or as information to the kiln controller to enable zare
efficient operation of the bypass and kiln.

3) When the form of sulphur in the raw material hae been established,
a simulation of the Oxford Works raw meal could be tried out at an
existing BCI dry works, (poselibly

Plymstock). Sampling throu&out

the system would then show where, and at what temperature, volatilisation
and condensation of alkali and sulphur containing compounds occura*

1)

oxford Works - Modern Dry process with sulp;lur Rypass - an

assessment of the technical risk.

STN M/13.

PA Loognarl,

D.S. Svift, March 1981.

2)

Processing of Cement Flue dust.. PhD Thesis, Jaw Huei Wor,

December 1979.

3)

Sulphur pollution from coal combustion. Effect of mineral

component on the thermal stabilities of sulphate ash dsd
calcium sulphate.

Baker D.C. and Altar A. Ewironmental

Science and Technology. P!!ch 1981.

4)

Recirculation problems in Rotary Kiln Systems.

H, Xitznann, Neubechun.
338-343.

5)

Translation of Zement-Kalk-Gips

(8)

1971.

Desulphurisation of hot reducing gases vith calcined Dolomite

ET Turkdo6an & R.G, Olsson.

Irorrcuz&ing & Steel %king 1978 Ho. 4.

Estiloated Alkali Vawur in the Kiln/Freheater

Gas per Kg of Clinker (no bypass

Position in
System

WQ4 &3s/
Kg Clinker

Na2w4 %s/
Kg Clinker

4.8 x 1O-2

1 . 4 x 10 -2

Burning zone
at 145OOC

0.55 Kgs gas/

Kg Clinker

IxL QdQT
Clinker
Complete
Volatilisation
Expected

I
At 3ack End
at 1050C

0.55 &P @S/

8 x 1O-4

lo-5

-2
5 . 6 x 10

per Kg Clinker

Precalciner
high temp. 95oOc
1.7 Kg3 es/

-5
5 I 10

Megligible

Negligible

Negligible

4 . 6 I 10 -2

Kg Clinker
Precalciner

normal temp.850C
1.7 &3s gas/
Kg Clinker

1 x 70

-2

2.
TAB=

Effect of Sulphur on Rw Fkterial Volatility

Source of Sulphur

Conditions aff'ecting
reactions

Likely effect
Cam4 will be recycled being

CaS and CaSC

Oxidising

vapourised in the kiln and

condensing to form CaSO

4 in
the precalciner unless tha SO2
level is reduced. CaS will
tend to form CaSC

CaS and CaSO

Reducing

and act in

Similar to oxidising conditions except CaS will form

in the precalciner and the
So2 vi11 evolve at a sli&.tly
lower temperature.

FeS2

The bulk of the sulphur will

be converted to CaS but an
increased quantity of .sulphur
vi11 be lost from 'he system
via the stack.

CaS and CaSO

CaS and CaS0

Presence of
so* Fe203

Increased losses of So2 from

sodium and calcium

the stack when corn-red with

montmorillonite

pure CaS04

Presences of
Potassium containing
minerals

Y&e SO2 in the stack exhaust

when compared vith the effect
of SiO 2 etc.

Cod/.

l . . .

TABLE 2. Cont'd.

Source of Sulphur

Organic Sulphur

Conditions effecting
reactions

LikeQ effect

A11 the sulphur vi11 be lost

via the stack exhaust.

Sulphur the fuel

Temperature

("C)

Figure It Saturation Yapour Pressure of El, K SO ank Na SO

2 4
2 4

2FIGIJEE

Temperature @)
Figure 2: Saturation Mass of KCl, K SO and NapS04.Yapom
2 4

in Air

-:18-

__- .- -_--

_.-

-.-

- --z

Blue Circle Cement

PROCESS ENGINEERING TRAINING

PROGRAM

Module 13
Section 4

Factors Affecting Sulphate and Alkali

Cycles in Rotary Kilns and the
Implications

ANDTHE IMPLIC?YXONS OFTZESE EZFFECTS KITH RESPECT ?o P-S CDNTROL

SYNOPSIS

i)

To identify the fundmental

and empirical relationships governing

sulphate and alkali cycles in rotarykilns.

ii) Using these relationships; to explain the kiln performance with
reference to clinker sulphate ark3 alkali retentions at bth
Northfleet and Hope M&s.
iii) In the long term, to learn kx4 best to control the rotary kiln process
with respect to sulphate and alkali levels, in order to qtimise the
prciluction

of clinker that ineets current rmrket requirements.

Anexamii-la tion of alkali/sulphate cycles has 'keen carried out using

data frantrials

carriedcutatNxthf1ee-t andF@e b&As. The fundamental

an3 e-rpiricxl relationships qzvernirig these cycles are discussed along with
deviations fran those relationships.

The magnitude of the suggested fat-

tors ki-iich cause deviations fran the fundamental relationship is estimated

using data cbtained fran +mAs trials and ,&lished literature.
is then used to mre closely mcdel the real situation.

This data

The resultant r&e1

when refinedmaybe used to estimate the recirculating load of

alkali/sulphate

(fran hereon described as wlatiles) within the rotary kiln

system for wxks where insufficient empirical data is available to

calculate the recirculating load using the traditional average crass balance
method.

Using the fundamental relationships discussed in this report, the

requirements for a control stratq for rreintaining

volatiles in the kiln systxn is proposed.

a'constant level of

ll-Ls strategy is based a2 the

kiln X& signal.

i)

A simple idealised wlatiles cycle within a rotary kiln can be devloped frcmkncwledge

of the temperature profile and gas flows within

the kiln system.

ii) The rregnitude

of the recirculating load of wlatiles in this simple

rwcdel is principally a functionof tb.e peak feed temperature. The

quantity of gas pr unit of clinker is another impxtant factor since
it determines the rraximum volatile recycling capacity of the kiln
gases: i.e. the wet process, through its higher energy demand, results
inrcrxe gas being available to carry a larger quantity of volatiles.
Conversely, a precalciner with its smaller gas quantity within thekiln,
nwinff

tm

90%

dec~rlnnatinn

&

theburning of 60% of the a& cutside

the kiln in the preheater, is unable tc suppxt such ahighlevelof

volatiles as a suspension preheater !ciln or wet process kiln.
iii) The idealised cycle does rzot fully explain the cbserved situation.
However, through studying actual -rating data, the various effects
that influence recirculation canbe estimated and amxe realistic
model developed.

iv)

Tne four principal factors that alter volatiles recirculation b&aviour fran the ideal cycle are the gas/solid mixing, kiln atmsphere
cqsition, dust insufflation and feed -ition.
The level of gas/solid mixing in the burning zone is a function of the
kilnvolume loading and the aznbined effect ofkiln speed and feed
residence time.

The effect of pr mixingmuld

appear tobe to

decrease the actual quantity of mlatiles recirculation to around

l/Sth to l/lOth that of the level predicted assuming an ideal cycle.
The gas/solid mixing tith reference to the recapture of volatiles
material is a function of the type of process e.g. suspension preheaters are highly efficient gas/solid mixers at the kiln back end
whereas wet process kilns are rmch less efficient.
VI

The effect of a reducing atrrosphere

is to reduce thequantityof SO3

retained in the clinker and increase the So3 lost as SO2 via the
stack exhaust.
vi)

Dust insufflation will typically cause a large increase in the

volatile recirculating load.

vii) Volatile recirculation will tend to induce a cyclic pattern of behaviour for the quantity of mlatiles retained in the clinker *en
clinkeris~ttoaconstant~~limemntent.
The cyclic behaviour of the recirculating mlatiles causes changes in
the apparent burning characteristics of the clinker bynodifying the
quantity of flux in the pre-buming and burning mnes and also induces
a~geinfhequantityof~trequiredto~~aconstant

-iv-

recirculating load, i.e. the heat requirement for mlatilisation

the burning zc)ne will increase *xith increasing mlatile

in

content of the

feed.

This cyclic behaviour makes kiln axtroltc

a&t-ant

free lime

based on kiln amps exceedingly difficult. A amtrol strategy to mintain a am&ant

level of volatile recirculation till rerrove

cyclic behaviour.

this

Ihis stability can be achieved by maintaining a

constant burning mne teqerature

with a constant feed input. Mrk

carried out at I-b&pe Wrks has shown that the best relationship between
an observed kiln pa.rar&er and the level of recycle tas found to be
N&.

This wouldman

tain a ax&ant

thataaxtrolstratqybased

onK& shouldmin-

level of reci2miLation and allow the free lime wntent

to float inlinewith changes in the feed chemistry.

i)

F'urtherwrk

shouldbe

carried cut to identify the mlue of E, the gas/

solid mixing factor, particularly wi"Lh reference to the kiln ,param+

ters of mL.xne loading, kiln speed and kiln angle. This factor, along
with thepeak

feed temperature, wuld allowamre

of the level of bxning

mne volatilisation to ba predicted.

ii) Ihe relationship between clinker m&hate

studied to enanpss

accurate estimate

and q should be further

the effect of ,parameters

position, coal consmrpticn and kiln cutput.

such as B.E.02,

feed ax-

iii) The effect of oxidising or reducing atrmsphere cm mlatile cycles

should 'ZE studied further.

Current theory and practical results indi-

cate an inverse relationship between the &antities of SO2 and 02 in

the gas stream andhence a direct relationshipbatween

So3 in the

clinker and 02 in the gas stream.

iv)

The effect of raw meal mineralogy shculd ke examined in order to

attempt to qmntify its effect m the ideal cycle.

VI

The net effect of the thermal requirement for empczating and oHIdensing circulated mlatiles r&is to be determined mre specifically.
This is of particular relevance where preheater cleaning causes sudden
surges of mterial with a high mlatile content to enter the 'kiln.

vi)

Control strategies based on N& to mintain a azmstant mlatile

recirculation load should be develo,Ded and evaluated in preference to
those based mkiln amps and a amstant free Lime cmtent.

FIGURE t
------we

RECIRCULATING
HOPE WORKS

Stack

VOLATILE

1977

BALANCE

Gas Stream

coal

0.123

K2

2.624

K2

0.019

Na20

0.033

Na20

0.180

Na20

0.005

S O3

0.381

so 3

4.930

so

0,480

C l

0,018

C l

1.120

Cl

0.000

0.019

K2

Precipitator
K2

0.044

Na20

0.088

SO3

0.040

Cl

0.010

0.004

I Duet

1R e c i r c u l a t i n g
K2
Na20
so
C l

-Is

Preheater

82 jXj

2.457

K2

2.605

O.lt9

Na20

4.503

so 3

0.175
4.450

77.11

1.092

Cl

1,108

99.7

0.015

0.015

8.6

83.4
46.7

3.125
0.375

Feed

$-Retenlion
-

K2

0.624

K2

0.668

Na20

0.218

Na20

0.226

Na20 0 . 2 0 0

so.
5

1.201

SO

1.2'11

so

Cl

0.009

Cl

0.019

C l

0.156

0.160

0.012
0.004

Volatilised i

K2
Nu20

Raw Feed

q6.6

1.300

53.3
%%,C

0.003

0.3

O.lGO

31.4

HOPE

WORKS

KILN : R E L A T I O N S H I P 6

YCH
NX%

ZCGC

ICC%

21 I?102

m/84/13

MD'IHE iMPLICATICNSOF'Z%SE EFFECTS WI'H RESPECT T0 PXCESS GXIWL

Page lb.
SXNOPSIS
OBJE~IVES

1.0 lzYmxum1m
Figure1

Recirculating Volatile Balance t30

Figure2

raOpe Wrks No 2 Kiln Relationship v

beween Kiln
Exit W, and Clinker SO3 cbnt
Sam Tine Base

ks 1977

2.0 FUNDAMENTAL RELATIONSHIPS ?G'FZTIG bQ

LECYCLLES
%
Figure 3 Idealised K2SO4 cycle for a Northfleet Kiln Feed
LSF 98%
Figure 4 Idealised Cycle for a

pe Wrks Kiln Feed

3.0 DEXL..IONS FFXX THE IDEAL CYCIE

%
3.1 Gas-solid Mixing
3.2 The effect of
cooler ,oart of

5
7
8
9
9

mixing and reaction rate in the

10
11
12

ical Trer& of So2 amxntratim in

of Kiln Feed Ccqosition

13
14
15

Page PD.
4.0 IKPLIC~TICNS OF WUTILE
OF KYI'ARY KIU?S

RECIRCUUVION

To 'IWZ pKCl5S.S

CCWjX)L

16

4.1 Burning Zone Temprature

17

4.2 Gas Qm-ktity

18

4.3 Xixing

18

Efficiency

Figure 7 S02, NO2 and 02 axxentrations

Exhaust (dry basis)

in the Kiln

19

4.4 Gas solid mixing in the cooler ,mrt of the kiln

rate
4.5 Gas Clarqpsitions

20

4.6 -Kiln Feed CDnpsition

21

4.7 Ixlst

21

Return

5 . 0 coNcus1ms

21

6.0

22

RECXMQXDATIONS

24

7.0 REFERENCES

APPENDIX 1
THEDRIES OF
CORREcrED IDEAL CYCLES

Q3NSTRUC!XoNOF

IDEAL CYCE.S RM)

APPFxxx2
EEFECTOFREDJC
APPEHDIx3
HEN PIPE

CD!JDITIONS

CN THE vOLU?ILIT'Y

OF ALKALI SULE'HATES

%
F KWTILE

PECIRCUATIoN

AT WPE KXKS

ANDTHE

DIPLICATIONS OF'MESE

EFEECTS KtTH RESPECf'T3 PKCESS CrXTWL

OE!JEmIvE
i)

To identify the fundamental and empirical relationships governing

sulphate and alkali cycles in rotary kilns.

ii) Using these relationships; to explain the kiln

ce with

reference to clinker sulphate and alkali re entions at b&h

No-fleet and I-Qpe Parks.
iii) In the long term,

b
the rotary kiln process

with respect to sulphate and plka

prcduction of&clinker that

in order to optinise the

nt mrket requiremnts.

1.0 1mmLUcr1oN
Kiln based su &ate and alkali cycles ("volatile cycles") have
almys

been of i.qx3rta.n
\ to the cement raker, since the level of
4 ficant effect co the kiln characteristics, 'the
recyclehasa
%
clinkerch
' try and the associated quality of the cement prcduced.
Howaver, w1
e advent of dry process kilns these cycles have been
3
under r
xtensive study due to build-up problems in the kiln ,preheater Q
an relatively high alkali levels in the clinker. i%re
r Q * enviromental
to

pressures have added impetus to the study, due

increasing interest in the level of q and s4, &ssiom fram the

kiln exhaust stack.

ANDTHE IMPLIC?YXONS OFTZESE EZFFECTS KITH RESPECT ?o P-S CDNTROL

SYNOPSIS

i)

To identify the fundmental

and empirical relationships governing

sulphate and alkali cycles in rotarykilns.

ii) Using these relationships; to explain the kiln performance with
reference to clinker sulphate ark3 alkali retentions at bth
Northfleet and Hope M&s.
iii) In the long term, to learn kx4 best to control the rotary kiln process
with respect to sulphate and alkali levels, in order to qtimise the
prciluction

of clinker that ineets current rmrket requirements.

Anexamii-la tion of alkali/sulphate cycles has 'keen carried out using

data frantrials

carriedcutatNxthf1ee-t a&F&e b&As. The fundamental

an3 e-rpiricxl relationships qzvernirig these cycles are discussed along with
deviations fran those relationships.

The magnitude of the suggested fat-

tors ki-iich cause deviations fran the fundamental relationship is estimated

using data cbtained fran +mAs trials and ,&lished literature.
is then used to mre closely mcdel the real situation.

This data

The resultant r&e1

when refinedmaybe used to estimate the recirculating load of

This re,mrt piqwses an alternative .meticd of estirwiting

t3e

recirculating volatiles based on the fundamental relationships

governing the process of recirculation.

An ideal volatile cycle 'based

on these relationships is pro,oosed, along with ,mtulates for

deviations frcax this ideal cycle.

Ran this rrcdel,

a new estimate for

the internal cycle is developed for mrious kiln amditions.

These relationships are then used to support
controlstrategybased onmintaining clinker

es levels, espe-

cially the sulphate level, at a constant va ue rather than to allcw

them to vary over a large range of values. b
2.0 FUNDAEEVTAL,

fELATICNSHIPS AFFEcrrwG V3& C Y C L E S

The tm fmdamentalthe

'c ,parameters which affect the

4
volatilization of any rmteria
e the temperature and its imle fraction in the gas stream (i.e. 3@al pressure).
system these are iranife
tity of gas atilabl

Inarotarykiln

the temperature of the feed and quan-

Q
to carry the volatiles. The relationships

between temperature, \
m
fraction and recirculation are discussed in
more detail in
ideal cycle.

ndix 1, along with details of 'mw to construct an

e
Ran these relationships we mn axstruct a diagram to

erial is wlatilised and recondensed. Fbr the ,p.xpcse

9
rt the recirculation of mly one wlatile - K2SO4 - is

ShGdhCWth

of thi

discuss9
, dlthough the same technique myba used for any mlatile.
shows an ideal cycle diagram for Wrthfleet Mrks i%. 4 kiln.
Fi e
Q
In studying this figure, cry rmst mncieve the zaw feed being fed to
the kiln and rmving along the Xaxis until the feed temperature has

reached a pint at 5-6 'kiln diameters frm the nose ring where the
K2SO4 in the kiln feed is able to volatilise. The K2SO4 then begins
to vaporise into the gas stream where it attains the gas temperature
and is trans~rtedback

along thekiln to apointtiere

perature falls to a level tiere the K2SO4 will reco

the gas teme (and rem

bine in the event of dissociation), at about 13 to 15

-zfiiiL
'
ters fran
the rose ring in the case of the first cycle.

les will con*

tinue to build up the levels of volatile in the kiln bed until the
maximm

quantity of K2SO4 tich can be

is reached. This.

will occur at the &peak feed temperatur

around three kiln diameters

frcm the nose ring.

any excess K2SO4 in

with the clinker. Fbr

example, consider

approximately 25 cycles
reached and K2SO4
culating load is 25% K2SO4 QI clinker.
etical kiln feed history is illustrated in figure
Theiqmrtantfhingtomtecnthiscycleisthe
ity of K2SO4 mlatilised.

This is indicative of the

ility temperature of Hope tbrks kiln feed and the 1-r

1.92 kgs of gas per kg clinker in the burning zone and 2.75 kg of gas'
per kg clinker at the back end.

E?qp Parks 'has 1.40 kg of gas par kg

clinker in tie burning zone and 1.94 kg of gas per kg clinker at the
back end.

FiKurd
100

- Iticaliscd KSO!, c y c l e f o r n NorthIleet

90

80

pdircction

70

4-edircctiorl

o f

gas

ol feed

80

50

b0

JO

20
solid
gas

10

to

stream

t
0

K I L N DIflHETERS

F R O M HOOD

K i l n heed

LSF 985\:

L
0

C
0

&MN113 NO X O3SIlI1UlOA tOSZW SSW

3.0 DJZVZ~TIONS

FFCM TEE IDEAI;

CicLE

The volatile rrms balance in figure 1 gives details of the maan

long tern recirculating ILoad at Wpe Wrks.

If ma consider K3SO4,

the equivalent average quantity of K2SO4 recirculating is 1.94% which

is only l/lOth of that predicted by the thecretical

It is

these deviations fran the ideal cycle which are

section, and, tiere pxsible,
these deviations

in this

quantitative est

the ideal cycle axe given.

The factors which are amsidered

discussed in mre detail belcw. ?he

f the effect of
"

\
be of major iiqm2ance

are

ors are:-

i)

The effect of gas/solid '

the burning zone.
+

ii)

The effect of gas/s&i

g and reaction rate in the oc;oler

,"rts of the kiln.

iii)

The effect of kiln atmsphere.

zone of a rotary kiln, vqmrisation

till take

volatilised is in mntactwith the

unsaturated

hut

this wntact,

gas..

The rotary kiln is very inefficient at ~rcducing

as the bulk of therraterial is contained in the feed bed

which presents cxiLy a szmllsurface

to the gas strfxrnatany

time.

For this reason a mixing factor (E) with dimensionless values between
0 andlneeds

to be intrcxduced.

The mlue of E will approach 1 tien

the gas/solid mixing is very efficient and allows all the feed +a acne
into cc&act

with the gas.

Thus the value of E represents the frac-

tion of the total solid surface ewsed to the gas stream. Using this
interpretation be can say that E will be a function of the volume
loading of the kiln: (a small~l.umeloading.presents

w ger surface
per unitmss of feed than ahigh volume loading) and the ccmbjned
u
function of kiln speed and kiln angle. A high Q
.
factor wuld be
expected with alowkiln

slope and ahighki

Work is currently being carried cut has

m aza-quter

and vice versa.

ncdelling as to

the precise value of E; the 'nest value ZL

r hasedcnpresent
information is 0.1 to 0.2.
+
.l to 0.2 implies a decrease in the

The use of values of E

maximum arw3unt

of volatilisation
% the Iwrning zone by a factor of 5

to 10.

0
0

3.2 TIE EFFECT OF GAS SOL

"

ANDKEACI'ICNRATEIN'-!XEC0LERPARl'OF

THEKIlxsi!Ls~
%
This e ect is very mch a function of the type of process.
For example,
a dry process kiln &i&has intimate mixing of gas
+
solids at temperatures tiere reactions 6311 cccur very
ally all the volatiles till be captured. The mixing
of this part of the process can therefore be said to
apprwch

unity.

For a wt or semi-wet process, the tilk of feed

material is retained within the feed bed and therefore less intimate
mixing is likely resulting in a much 1-r level of Mlatiles
captured (circa 50% efficiency).

being

This later case is illustrated in da'~a obtained fran an HES

bum carried out at Pamstone ibrksc21 see figure 3 here the
quantities of K20 and SO3 retained in porthole samples 1 taken from
the chain section outlet, and porthole 3 taken fra a point just prior
to the burning zone, show v-q little difference (i.e. change in K20
and only 0.5% change in SO3 content).. The stirring ction in the
chain section of a long wet 'kiln does increase the mixin
efficiency,
9
hence there is a sharp decrease in So3 content

rtholelto

the

stack of around 70% and in K20 mntent of around v

906. It is possible
that the actual mixing efficiency within a lon
%t kiln is likely to
be prirrarily a function of the chain

This section refers i

'tally to the oxidising potential of

the kiln atrrosphere and its -ifi?

eff
upon the volatilisation of sO3.
The fact that a reducing

In atxrxphere as indicatedby alowbck

Q
end oxygen content (generally < 1.0%) gives rise to lu48er clinker SO3
andhigher exhaustSO2 \n thehigher oxidistig conditionhas

well documented

myle and Ferd41,

been

B.-KG&~] and ~la~urne~~~.

e
Eetails of th possible rre&ani'sins mncemedwith this process stay be
found in
values

' 2 tile the effect is illustrated in figure 6. The

&
in figure 6 will be dependent upon the type of process

(e.g. wet
Q), feed and fuel -ition and burner design. The
feature is the shape of the curve i.e. a very rapid change
in SO2 content in the exhaust gas (andhence an equivalent change in
the SO3 in the clinker) for a sznall&ange

in oxidisingpotential

FIQURE 5

K2

0.13

SO 3

0.32

BARNSTONE

HES

TRIAL

NO. 6 AUGUST 1980 -

VOLATILE RECIRCULATION

Feed
K2
s o3

I.77
5.+3

I
I
I
1
1
A

End
of Kiln
Back

Temp. 2-300 C

Temp.

800~

Temu

Ti?ICAL

2500

TRZXDS

OF SO2 CO?;CE?XX4TION

IY KIL!J EXYAUST

NZARLY ALL SULFUR

?XTTED AS SO2
Y7
v
2000

1
I

1500

.?ROBABLE TREYD WL"r!

LOU ALKALI FEED
1000

----__

500

I.0

2.0

02 CO?lCEXTMTION

Extracted

A.0

3.0

(I)

from :

Doyle and r"er& A?$icatiocs of Flue Gas halysis to Cement i<lln

Cperation 9ock Products ::ov 1923

5.0

of the 'kiln gases from reducing to oxidisirg cmditions.

pattern has teen observed at LXorthfleet

A similar

Wr'ks during tests cm the So2

monitor where SO2 is practically undetectable at hack end oxygen

levels of qeater than 1.0% tit very quickly rises to levels of &me
500 ppn when back end oxygen falls below 1.0&[71.

A change in SO2

concentration in the exlmust gas at Sbrthfleet of 0

the equimlent of approximtely

500 ppn So2 is

0.4% SO3 a2 clinker. b
b
v

3.4 TE-E EFFFCI OF KILN FEED (XXKSITION

This can be categorised into tin0

i)
ii)

The effect of the rawrreal chemis

&
The effect of the rawrieal

try after being rrcdified

&

by

recirculation.
@Y
mly differ in the absolute quantities

cementrawrEa1s

of volatiles, but also tt*o

Tit meals -tith the same mlatiles

amtent

may have very differe

' eralcgy. Mrk by Baker and r~tmarrC81' for
%Y
exarqle, has
that certain minerals have a greater propensity to

inlpartaninde
encourage the

ationof

SO3 in particular.

These minerals

' ualcfiaracter to the particular rawmaal. and there is

no way as yet
tern ex

estimating their effect upon the recirculation pat-

empirical
4%

techniques.

irculation of the mlatiles themselves causes the content of

@
volatiles entering the burning zone to increase. This has tm affects
upn the burning characteristics of the raw meal:-

i)

Both the armuntof flux within the burr&g zone and the
apparent length of the burning zone will tend 'io T. in
increase in kiln amps due to increased volatile load could, for
example, be mistaken for an indication of overburning and any
resultant reduction inburning zone terrrperature axld lead to
an increase in clinker free lime.

ii)

An increase in fines within

the systemwillalso tend to encourage clinker

ctions
w
which in turn should decrease the free Ii,
axtent.
T7
77
The evapration or disocciation of the wlatlles in the burning
zone will de-r-and a significant quantity
XT high grade heat fran
the flame.

ver, is recovered further

?%is high grade heat
k
dcxn the kiln as lw grade h t when the wlatiles recondense.
'
effect" is shown in appendix
+
tiat the recirculating wlatiles make

Pa estimation of this 'heat

3.

The overall effect *

a significant demand (up

%100 kcals/kg clinker) upon the heat
atilable in the

zone andhence they tend to increase

the totalheatinputreqirementto

the kiln.

\
f&s could aggrevate each other and with the net
system, raw
mealch

'stry, and the oontrolstrategy

smployed.

3.5 THE EFFECT OF RETURNED lxiZ?I

A rmjor differenc:e

between the idealised cycle and the real

situation is, that although much of the recirculated mterial

condenses cm solids derived frcan the kiln feed, these solids are
generally held in suspension and therefore swept frcm the kiln in the
gas stream.

If the dust is returned, especially by

lation, then

the observed cycles should bqin to approach th

mixing factor (E) will begin to approach unity.
The greater the armmtof dust return
, the greater the
Y
recirculating load and hence the clos

clinker mlatiles till

to the system (except for
at t-ratures

burning mne is considerably

within the

ter than cne at-sphere and hence, ir

theory, should have an k-finite

%pacity for volatilisation).

4.C

0
e
IMPLICATIONS OFKKx4T&E FECIRCUIATION~ PIE Pm C?N'I'floLOF
mARY IiIL~S
This s
discusses the implications that mlatile
4
recirculatio has onkiln mntrol. ijherever possible the individual
factors dis
ed in the previous section are related to kiln cofkrol
42
parame
s uch as burning zone temperature and kiln feed rate. SOme
fo
of the factors such as recapture of mlatiles will tend to be related
to

c&-

ss design rather than any &aarameter

part of a control strategy.

tNhich can be altered as

These should be rmted, as it is highly.

probable that any control strategy considering mlatiles recirculation

muldbe dependent to a certain extents the type of,orocess.

The quations given in table lcgl shm the iqxtance of

temperature as a factor amtrolling

mlatile recirculation, because

the amunt of recirculation has a logarithmic relationship to

temperature.

In a dxy process kiln, the usual strategy is to control

the kiln to produce clinker to a selected range of fr

by mnitoring the kiln amps.

' e cmntent
w
This signal is a function of, amngst

Y7
other things, the amount of flux present in the )u . However, in
v
section 3.4 it was pstulated that volatile
the armuntof flux in thekilnwith litt
inputratetothekiln.

Thekilnmps

sane degree cn the rragnitude of the

strategybasedmthe

irculation will mxlify

T
or m change in the fuel

'
therefore be dependent to
&
tile cycle.

Akiln control

ining a steady level of kiln

amceptofiii?
'

amps is likely to disturb the

'les cycle when recovering fran a

%
process disturbance khich '
in turn affect the ammtofkrning
zone flux and hence kiln
-Q For example, a perturbation in kiln
feedchenistrythat'
Zip.

amps set pint

eases mlatiles till tend to,increase 'kiln

%
react by reducing the BZT to maintain the
the clinker produced would be both underburnt
in mlatiles.

makltain

tant coal feed strategy is similarly insufficient to

le kiln control,

as any increase in recirculating

Q will tend to depress the temperatures in the burning zone

vo1ti
andhence theheat inwtto the kiln will heed tobe wied in order
to control the solids temperature andhence the clinker free lime
content.

In order to mintain a specific temperature in the burning zone,

the best ,Darameter

to consider is therefore the peak feed teqerature.

Unfortunately present day instrumentation is unable to supply a

reliable direct indication of ,oeak feed temperature. An alternative
method of establishing a peak feed temperature is to measure the
parameter qon which the peak feed temperature depe
gas teqxxature.

fhe W

This maybe carried out indirectlyby&asuring

parameter dependent upon the peak gas temperat

This will then be a function of the peak
that the feed rate is n-aintained

at a

constant level. The

relationship between NO, and clinker So

as alreadybeen discussed
4
(see figure 2). Further evidence fo the relationship between wlatiles content and N& is show in fi
concentration is seen to foil

. Here the kiln exit So2

&
exy step change in Q, highlighting

the particularly strong correla

%
'
between these tko ,oarameters.Clll
4.2 GAS CXJANTI'IY

0
0

This, as mention
%? previously, is principally a process design
variable. Any

on the mlatiles cycle due to gas quantity

changes assoc' ted with fuel changes are liable to be masked by the
greater s
*
ty to changes in the burning imne temperature.
4
The appr
te strategy txxards gas quantity control should therefore
Qmintdin constant gas flew through the kiln together with
be-try
a

t feed rate.

4.3 MEQNG EFFICIENCY

This factor is essentially determined by the process design.
As discussed in section 3.1, the efficiency (E) is determined by a Ccnr

bined function of the kiln speed and 'kiln slo,pe which determines the
amount of effective surface area of feed within the burning zone. As
a consequence of the gecmetry of the circle, a significant change in
volume loading will only have a small change in the effective surface
area of the feed within the typical limits of kiln volume loading. As
thekiln

angle is fixed, thekiln

speed is the cnly

meter tich can have any influence ctl the wlatile

postulated that this influence should not be lar

cess' le paraFITIt is
cycle.
rasmallrange

F
of kiln speeds as there are tw opposing effects. An increase in kiln
speed would increase the rate at which effecti
32 surface of feed is
exposed for volatilisation

to take pla
but decrease the time
+L
available for evapration.
I t is p tulated that the likely effect
is for them to cancel each other
unlikely to be critical in

'In speed is therefore

a
a constant level of

volatile recirculation.

\a function of process design. Changes due to

This is similarly
this effect mul
design but

causetl in bet process kiln by altering the chain

e
ges during operation of the kiln are unlikely to affect

(Gas wition tich affects this will be

volatiles r
ure.
a
discuss
t h e next section).
a
4-5

=fFF=
As was stated in section 3.3 a &ange in oxidising~tentialof
the kiln gases can have a large effect cn the recirculation of SO3.
If a strategy of maintaining a constant wlatiles
achieved, a constant back end oqgen

cycle is to be

level will be necessary.

4.6 K!ZN FEED CMFOSITION

The effect a2 kiln mntrol of the kiln feed rrcdified by mlatiles has already keen discussed with reference to temperature in section 4.1.

If%wever, even if a constant level of recycle is rmintained,

there stillmyke changes in the feed chemistry entering the'azming

zone due to changes in feed &en&try of the raw f
qUate blending.

cau edby inade-

Pius, as with the strategyofhrning

-9 a target

free line ken &anges in LSF of the feed muld

instability, a

v
control strategy based cm mlatiles also r*es mw feed of
consistent chmistry.
4.7 DUST KEI'UEN
A dust return system tends

rate 01 a discrete level (ie.

either on or off).

either turning cn or off a dust

return systan till dis

cycles as the change in iciln wla-

tiles input will affect

culating load is

Ps the size of the recirdependent, the mjor factor

thatwuldbe

affected

- the mixing factor. 'Ihis muld

take cm anew

and a mew volatiles recirculation equilibrium

%
bymaintaining a steady Wrning zone tmperature.

lised volatiles cycle within a rotary kiln can be devekncwledge of the temperature profile and gas flows within
the kiln system.

ii) The magnitude of the recirculating load of mlatiles in this simple

rmdel is principally a function of the peak feed temperature. Another
factor, the quantity of gas per unit of clinker, is iqxxtant since
it determines tie mxirmm mlatile recycling capacity of the kiln
gases: i.e. the wet process through its higher
in more gas being available to otrry a larger
Conversely, a precalciner with its smaller gas

y within the kiln,

owing to 90% decarbonation and the burning of

-1 outside

the kiln in the preheater, is unable to supp%kuch a high level of

volatiles as a suspension preheater ki
iii) The idealised cycle does not

wet L;rocess kiln.

the observed situation.

Ffowaver, through studying actual cpe

ing data the various effects
n
that influence recirculation
estimated and a mre realistic
model developed.
iv) The four principal fa(;tors tit alter mlatiles recirculation bahavia= fra the ideal \
cyc .e are the gas/solid mixing, kiln atmqhere
fflation and feed amposition.
The level of

solidmixing

in the bnxning zone is a function of the

kiln mime4!ti
1
' g and the ambined effect of kiln speed and feed
residen

') e.
Q

l3e effect ofpr mixing wuld appear tobe to

the actual quantity of wlatiles recirculation, to around

l/lOth that of the level predicted assuming an ideal
cycle.
The gas/solid mixing with reference to the recapture of volatiles
material is a function of the type of process e.g. suspension pr*

heaters are highly efficient gas/solid mixers at the kiln 'back end
whereas wet process kilns are mch less efficient.

VI

The effect of a reducing atmosphere is to reduce the quantity of so3

contained in the clinker and increase the So3 lost as SO-2 via the

stack exhaust.
vi)

IXlst insufflation till typically cause a large

in the

volatile recirculating load.

vii) Volatile recirculation will tend to indu

a cyclic pattern of beha-

viour for the quantity of volatiles retain

& in the clinker hhen
clinker is burnt to a amstantfree
The cyclic behaviour of the re

wntent.

'

e?
ating volatiles muses changes in

the apparent burning characteris

@%zs of the clinker byrzdifying
quartity of flux in the

the

uming zones and also induces

a change in-the

tomintain

a constant

recirculating load.

basedcxlki e

exceedingly difficult.

cyclic 43OUT.

This stability can be achieved by titaining

This cyclic b&ml

Q mkeskilnmntroltoconstantfreelime
Aamtrolstrategy

to n-&n-

tain a 03 tan level of volatile recirculation will rezmve this

burning zonetmperaturewith

aoonstantfeed

input. Wrk

out at Hope Works has st-hown that the best relationship betm
an observed kiln parameter and the level of recycle was found to be
PI%.

This muldman that a aontrolstrategybesed QIQ shouldmin-

tain a asnstant level of recirculation and allow the free lime content

6.

FBXMGBDATIONS

i)

E'urther

mrk should be carried cut to identify the mlue of E, the gas/

solid mixing factor, particularly with reference tc the kiln parameters of volume loading, kiln speed and kiln angie. This factor, along
with the peak feed temperature, kould allowarmre

acwe estimate
of the level of burning zone tolatilisation to
redicted.
Y7
T?
ii) The relationship bst*en clinker sulphate an F.Dx should be further
studied to envss the effect of ,mrameters\
s
as 3.E.02,
position, ~1 consmption

and kiln ou

iii) The effect of oxidising or reducin

should be studied further.

feed am-

.
6
sphere cc mlatile

cycles

Current%zry and practical results indi-

cate an inverse relationship

the gas stream and hence

the quantities of SO2 and 02 in

4
ect relationship between SO3 in the

clinker and 02 in the gasQ

s
.
iv)

The effort nf raw

ek
,
1

ergiccv &c&d be examined~~im-order

to

f;
atterqt to quan
VI

The net eff

its effect m the ideal cycle.

%
the thermal requirement for evaprating

and cm-

densing c' e
cul ed vclatiles reeds to ke detemiined mre specifically.
This is & icular relemnce where preheater cleaning causes sudden
SIX
f mterial with a high volatile cmntent to enter the kiln.
vi)

c3
Control strategies based on Mx to maintain a constant mlatile
recirculation load should be developed and emluated

in preference to

those based on kiln amps and a constant free lime ontent.

7.

F33ERExcEs

1)

Weber P.

Heat Transfer in Fbtary Kilns with due Regard to Cyclic

Processes and Phase Formation. ZKG tiglish Special Edition 1963.

2)

Longman P.A. Swift LG. Oxford Mrks -~twem Dry Process with sulphur
By-Pass - An asseswnent of the technical Risk. SI'N 81/13.

3)

Rogers A.R.

Assessment of the potential to q-tin&sew performance by

application of improved prczess sensors : tinit

Kiln September 20-24th 1982. TN 82/28.
4)

Coyle B.W. and Fenk F.W. Applications of

works operation.

5)

Brmm A.W.

FwkPrcducts,

Nov19

Chadbume J.F.
Kilns.

.
FkSXCh

, SR-65/7/M&3.

Continuous i% 'tar'
9
g of Gaseous Emissions an Cement

Paper presented to

Control Association, Jun

7)

gas analysis to wnent

%

Retention of Alkali and SuLph

4.2 in Clinker.

Division Repx~tis 1 and 2 SP-65/33

6)

*
of I-bps wxks
Y7
v

d Meeting of the fir EMlution

4
9.

Lorimer A.D.J.

cklline 802
Qnitoring : Assessment of LXXJV and

Electrcchemical

cell

: May to October1984.
%terns at Northfleet

m/84/21.
EBkr D.C. and k A.
Effect

of '

Ashand

8ul.phu.r mllution &an coal ambustion.

Cxnponents cn the Them&l. Stabilities of sulphated

cium S&hate.

EZlvironmental

Sciences and Technology,

March @@
.

9)

The processing of Cement Flue Ibst, PhD Thesis, ?he

m
University of Aston in Birmingham, December 1979.

10)

Haspel D.W.

Le@ Grate Mt Transfer bechanisns. TM. IxJH.005,

Weardale bbrks,

1973.

11)

Analyser 2.3nk at the ?A &3vre lbrks. CETIC mviromt

Gq2pelle M.
Sub-m-mission

12)

Turkdqan

September 1984.

E.T. and Olsson R.G.

with Calcined l%lmite.

13)

Myers J.C.

Sul&ur

Desulphurisation of Wt Reducing Gases

Iron Making and Steel ?&king No. 4, 1978.

Balances in @m%t Kilns and

Thesis, University of Texas,

Austin, 1977.

&
D.

stenson

De

r 1984

@Y

ilers.

!4Sc

(i)
APPENDLyl
THM)RIES OF FECIFCULATICN, 3-E CTINSI'X~ION OF IDEAL CYCLES

The volatile aanponents of aarnent raw mals and fuels are

vaporised ken the feed temperature is raised to a level tvhere the
vapour pressure of the volatiles bscunes significant.

The wlatile

caqments will then impart a partial pressure to the gas stream. Cn

cooling, the vqmur pressure of the arnponents will decrease causing
condensation.

E?y measuring the kapour pressure of certain cmpments

at different temperatures, enpirical relationships for the saturation

vapour pressure and temperature have been established.

Equations

relating saturation vap0u.r pressure to temperature are given in

table 1, equations 2 to 7 and shown graphically in figures (i) to
(iii) C91.
The saturation repour pressure can bs used to calculate the
saturation vapcur mncentratiun
vf

= P*
PT

by applyirlg Raoults Law:

Mv
Gj--

(1)

where vf is the saturation concentration of bap3ur in the

carry%

gas

(kg/kg

9as)

P* is the saturation vapour pressure (any pressure unit)

P is the pressure of the gas (same units)
Mv and m are rm1ecula.r lHeights of the vapour and gas reqpectively.

An example for the saturation axcentrations

of I@1 and K2S04

in air is shown in figure iv.

Examples of bw to construct a volatile cycle using equations
ltc 7 is shcmlbs!low.

(ii)
The first stage in oonstnxting

an ideal cycle is to obtain a

feed and gas temperature profile. For this, kxwledge

of the ox-

binability of the particular feed for a particular targetted free lime

must be available.

This can be cbtained fran empirical data. This

will give the pfaak feed temperature.

In the exarrples

shown below the

peak feed temperature for the wet process is lSOOC and 1490C for the
dry process.

Azonedheatbalance

rrcdel for the kiln can then be used

to derive a temperature profile.

Fbr cur examples the temperature

profiles are sho+m in figures v and vi (the oscillations in figure (v)

are due to graph interpretation; thecurveshouldbemth).

By

use of equations 2-7, the partial pressure of the volatile QsmFonents

maybe calculated. Equation lwill
concentration.
ccqosition.

The mlecular

then give the saturation vapour

weight of the gas is estimated frcm its

The calculations are sumnar ised in tables 2 and 3.

By mking

assumptions for the gas quantities at various &mints

in the kiln, the actual amount of mlatile &per !kilc.grarn of clinker can
be found.

2-ii.s is s-ised in table 4.

Table 4 can be used to amstruct

an idealised cycle as seen in

figures (vii) and (viii).

Ekplamtion

of the idealised q&e

To explain the idealised cycle, an example of akiln

to&ich

is fed the equivalent of 2.0% K2SO4 in the feed and fuel is 0311sidered.

Cm the first cycle the feed will be tranqmrted

by the

turning action of kiln until, in the case of the dry process in figure
(vii), a pk.nt4.5

diameters frm the rrxe ring is reached.

feed reaches a teqerature

where significant empration

of

Here the
K2SO4 Qn

(iii)

occur as K20 and sO3.

The m:.atile afnpments are taken up 'by the gas

stream &ere they rapidly achieve gas tmperature

and are tranqmrted

away fran the burning zone to the back end by the kiln draught. At
twelve 'kiln diameters the gas tenperature has fallen to a level tiere
significant condensation as K2SO4 on cccur.

This is then inmx-

porated txck into the feed and transported fomds along with the
fresh K2SO4 fran the kiln feed.
On the second ideal cycle there is an quimlent of 4.0% ~2~04
in the kiln feed approaching the burning zone.

The cycles will con-

tinue until a +.ntattw kiln diameters tiere the riexi~um allckable

evaporation determined by feed temperature is reached. Ebr the dry
process curve this is 16% m clinker after eight idealised cycles. Cn
the ninth cycle the K2SO4is raroved fran the kiln in the clinker as
the equivalent

of

2.0% K2SO4.

The corrected cycle

The idealised cycles shown in figures (vii ) and (viii ) are rx3t
followzd in practice.
The factors of 'mixing efficiency in the burning zone and
backerd, dust loss and reducing conditions all account for deviations
fran the idealised cycle.

Tables 5 and 6 give results of calculations

based m assumptions of suitable Mlues for these factors. Fran fhese

tables, figures (ix) and (x) are cbtained. 'Ihe solid lines cn figures
(ix) and (x) indicate the 1evels of mlatile corrected for the mixing
efficiency.

The broken lines show further axmctions

to decrease the

amount of material recaptured in the cooler parts of the kiln due to

dust loss and reducing conditions.

Saturated Vapour Pressure of KCl,

TABLE
- - ^ - - I -I-

K SO and Na SO

3--4- Z-4

Reference

LoglO

(p,/atm)

= - '47oo

5 9oo

Decomposition Suppreoscd
Loglo

( ps/atm)

Loglo

(po/ntm)

Hart and Laxton, 19671

llalntead,
1970
(cf. Koaugi, 1967;
Duboia et al, 19613)

P '- ' "O" 5 'O"

6.23

Cubicdfotti
Kuneahea,

,
DccomnPnit.lPn
I

Loglo

(pa/Atm)

-y

-t

7*oe1

1400 - 1625

and
1972

g
t:u
h

0.30

5
g,

0.25-

0
Lla
5
2

0.20 -

0.15 -

0.10 -

1200

1250

Temperature (Cl

Saturated Vauour
6 omce:

JANAZ',

Pressure of SC1

1960 or J.snz, 1969)

TJ,zzz ( ii)
______---s-.-

A-

Decomposition

unsupprrsscd

DecomposXon

suppfms*d

3-

.5-

.o-

S-

10 -

OS-

1150

1200

12so

1
1350

13M)

Temperature ["Cl

Saturated Vawur and Decomuosition

ISource:

?ressures of K,sO.

Eart and Lax-ton, 1967; Salstead,

19701

1.c
?ICc?z (iii)
------------

0.'

0.f

0.1

ri)
T
z
;
3
::
v

0.6

0.5

&

P
v
1
1
",

0.4

;;:

0.3

0.2

0.1

0.0
Temperature

Saturated

Vaoour and Deconuosition

(Source:

Cubicciotti

("Cl

Pressures of Ha SO

and Xeneshea,

1972)

2-4

.C
u

1V

C
0
.-

Temper abre ( C )
l

Saturation

Capacity of XC1 ad IT 2SO

- - j -Vau0u.r
------in Air from llOo-ljOoC

FXCURL
(v)
_..--------

:wL. t PI~oce:is
F
e e d uh _..--__
ps tempeniturc
------..__
--^----_ p-oSi1.e
-----..------------.--

6 - F E E D TENP
6 - GM TEW

EIGUNL
(vi)
..--------c2500

proTilL?
:I)ry rOCCS9
- - - -temperature
- - ^^-CI________------L.------- -Feeh
- L - - dand
- - gas

- FEELI TEHP

2000

SD0

- .-. .-.~-.-7*-i --i+---- --T-

K I L N OIFlMETERS

.__ -,-,-__-_
.-+-- -+..-- -.
14
-2 o *-.. 10
11
+s-.-;s-12

FROM NOSE RING

-.. -I-. _. _ _-,

15

IE

Gas
t

Sabrated Concentration
i.3 qas (!&kg p.5.)
Kcl

WC

K2So4

K2SO4

-2504

?ia2SO4

500

1250

.2.$6

3.07x10-3

7.02~10-3

600

1350

infinity

0.031

0.032

*,

0.099

0.123

0.169

0.232

0.031

0.032

2.93

3.94x10-3

3.24x10-3

0.474

1.45xlO'3

3.64xlo-3

1150

1450

0.601

1500

1500

knit*

1870

1350

2om

1260

1.99x10- 3

0.169
9.95

.25x10-3
0.232
I.

infinity
0

I,

zcoo

1130

1950

10(X

1870

950

9.95

1710

910

1.45

4.29

0.023

10

1550

850

0.281

0:429

9.65X10-3

11

14.60

810

0.111

0.140

4.71x10-3

12

1370

axI

0.040

0.042

3.9SXlo-3

13

1330

730

45.27

0.0246

0.0239

9.59x1&4

14

1290

690

5.62

0.0148

0.0131

15

1250

650

2.46

3.67x10-3

.02~10-3

16

1203

600

1.16

~27~10-3

-06x10-3

0.091
0.046

T?%aLz 3.

DE ?5cGss

Satc.zit+d
so
Of

Feed.
tcqY

Ciln

WC

Concemraticn

ia 9 (kq/kg 9s)
KC1

K2So4

-2So4

XL

K2SO4

.*2SO4

liam
eers

0
1

605
aao

1370

0.0400

0.0423

1417

I.

0.0688

O.oaol

4.

0.111

0.140

*a

0.0681

0.079

6,

3.78

0.0118

0.0101

950

84

0.046

4,

0.031

L-lfinhy

1594

1460

0.434

2110

1416

infhity

2274

1273

,.

2227

0.737
infinity

2106

925

1940

8%

430.0

a*

0.0204

1799

a74

3.465

159.4

0.0144

1678

a56

0.983

2.208

0.0107

10

157s

a42

0.360

0.583

a.4xo-3

11

14a5

a30

0.144

0.192

6.a2x10-3

12

1407

a21

o.ffi14

0.0700

5.81x10-3

13

1338

al4

0.0272

0.026a

S.llxLO-3

14

1277

ma

0.0125

15

1224

803

16

1177

798

0.0108

4.57x10-3

;.03x10-3

t.59x10'3

4.17x10-3

3.02x10-3

2.05x10-3

3.79x10-3

(xiii)

TAaLz 4.

Sal 33LriJ Tl,ss

KC1

.%2SO4

K2g4

in5.nity

5.72

6.25

..

9.04

11.45

LS.87

20.02

9.74

lo.a7

1.69

1.44

61s
-.
1n:sr+y
I.

2.17

1.7S

Dry mess

7.75

a.00

124 scc/x;$ :Ks a

up w

24.7

30.1

5 diamzter

$2.23

33.0

1.43 kg/k; clizkz

7.75

a.0

Aker 3

732

2.4.a

2.06

Diemeters

6.1

14s

0.44

0.264

1.99

4.65

27.9

2.07

14.1

2.12

7.65

10

1.67

2.9s

wet prccess
213 sx/lZi :s ?d
Up to S

11

1.37

1.G

Oiaetezs

12

1.16

1.21

2.499 icg/kq di-k

13

1.02

0.29

After 5

14

0.96

&meters

1s

0.03

3.059

1 6

0.75

514.0

6.6

Sas c3 solid
11
12

infinity
1.

13

28.4
12.2

0.382
13.9

infizity

33.9

42.3

,a

12.2

12.3

5.45

5.33

13800

7.32

7.31

14

817

2.49

2.14

1719

4.53

4.01

15

324

1.20

0.913

752

2.65

2.1s

16

170

0.61

0.408

354

1.31

0.94

20

0.44

kg/k;/cli*e

t:

+.

FIGURE
(viii)
--1----1-----

cycle for
-----_----------^---__^___
Idcnliscd

NA2SOlt

so
6 - D R Y PROCES
f.!l - Y E T PROCES

50

40

so

20

10

KILN DIRMETERS FROH NOSE RING

(xvi)
Table 5

Corrected Cycle for Mixing Efficiency

No. of
kiln
diameter

Concentration (% &per Kg
Clinker Dry E2m !SS

Concentration (S pr Kg
Clinker IGet EYOI ss

KC1

K2SO4

Na2SO4

KC1

K2SOq

Na2SO4

Infinit

1.14

1.21

123

0.434

0.35

1.97

2.29

Infinit:

1.55

1.6

3.17

4.04

4.94

6.02

8,

2
3

I,

1.95

2.17

0.8

0.34

0.29

1,

.32

1.6

146

0.496

0.412

.22

29

0.09

0.053

.8

5.50

0.57

2.82

0.42

1.53

10

0.33

0.59

11

0.27

0.29

12

0.23

0.24

13

0.20

0.06

14

0.18

15

0.17

16

0.16

11

lilfinit

I,

12
13

tions

%2 = 0.2

30.3

40.1

Infinity

39.0

49.2

EBE m

12.89

14.6

I,

14.0

14.7

= 0.95

5.72

5.60

15870

9.02

8.41

14

856

2.61

2.25

1975

5.21

4.61

15

340

1.26

0.96

865

3.65

2.47

16

1 179

USlimp

11.60
1.55

0.64

0,43

= 0.85

hii)

bxclusicn (1 per kg Clinker)

cry prccess

CmclL;sim (3 per kg Clhkr)

Xet Praess
KU

11
12

infinity
4.

13

36.4

48.1

15.4

17.5

infinity
I.

6.86

6.72

19KXl

?ss.ui&ons

QSO4

Xa2SO4

46.8

1 59.0

!x5; lass

16.8

17.6

= 20% cn

10.8

10.1

Clinker

14

1027

3.13

2.70

2370

6.25

5.53

15

408

1.51

1.15

l@iO

4.38

2.46

16

215

0.77

0.52

4aa
*

1.81

1.30

11

121

160

156

197

Pfhdng

12

51.2

58.4

56

58.8

Cmditioixi

13

22.9

22.4

36.1

33.6

SQ3

14

10.4

20.a

18.4

RecoroirationS

14.6

15

S.04

3.a4

9.88

decxe.asedby

16

i.56

1.72

6.04

4.32

75%

11

146

192

la7

236

ca7biii

12

61.6

70

67.2

70.4

13

27.4

26.9

43.2

40.4

14

12.5

10.8

25

22.1

17.5

11.8

15

6.04

4.60

16

3.08

2.08

7.24

5.2

ZffeG

~IcilJlikb
( ix
..---------.w

Corrc?cLcd Cplc
i-'or KL'CO'I
_____---.-----^------.--.---

20

<

@ - D R Y HIXINC FRCTOR
CJ- YET HfXfNG FACTOR
A - DRY DUST LOSS
0 - YET DUST 1
x - D R Y REDUCI
x - Y E T REDUCI
+ - D R Y CBH8II
y - Y E T CbHBIt

16

16

i.i;oLlt,

3.

&if

'ITJ GA
'I'0 SOLIIJ

--w-e --I--- l
I
I
\
----I
I
6

\x , \

14

\
\

-I-

12

f
0

KILN DIRMETERS

FROM NOSE RING

y\

\
\ -\

10

\ >(
\I I

\1

\
\
\G
\

FIGUHC
(x)
------.e---

C o r r e c t e d Cplc
l*or IJA2L;ch
_-__..-------^I-------c----

6 - D R Y tlfXINC FRCTCJR
2 - Y E T tiIX1Nf.i FRCTOR
- D R Y D U S T LOSS
@ - YET DUST LOSS
x - DRY REDUCING
x - YET REDUCING
-t- - D R Y COMElINED
f - YET COMBINED

<

K I L N DIRMETERS

SULID

I0 cxs

cA!i TO SOLID

FROM NOSE RING

Effect of Reducing Conditions cm the Volatility of Alkali Sulphates

The precise mechanisins

by which reducing conditions within the kiln

remove SO2 frm alkali sulphates is not fully understood.

real effect observed and documented by my other mxkers

Eiowever,

it is a

in this field, e.g.

Chatiume ,4 and by Brcx& fran within Blue Circle;

A similar chernicalprccess to the removal of SO2 is the desul@rising
of steel using lime.

This has been extensively studied by Turkdogan and

01ss0n. 021
The reactions described by Tuxkdogan and Olsson involve only calcium
sulphate.

However, for cur plrFoses we riay irake the reasonable assumption

that the alkali sulphates till undergo similar reactions, as is suggested by

their relative psitions

in the periodic table of elfments.

I

Calcium sulphate will deccqcse

reaction

under the effect of heat alone by the

CaSOq(s)

= CSO(~)

+ SO2(g)

4 h$2(g)

Wwever under reducing amditions

&Gs = 408 W ml-l

the follwing

(1)

reactions will be

preferred:
Ea.504

cqg) =

cao(s)

e(s) =

+ S02(g)

+ ~203~)

AGs

=(s) + co(g) 4@

15

Id

ml-l

=04(s)

C&04(S)

-+ 4CO(g) = CELLS +

where&?

is the c%ange in Gibbs F'ree E&qy at 25C and L atmosphere

4C02(g)

Reabsorption of SO2 till corer bythe

$CaOo)

AGS

+ 297 kJ m31-1
183

kJ

ml-l

reactions:

+ 4SO2(g) = 3CaSO4(,) f c~s(~)

;lGs - 787 W x11-l

(2)

(3)
(4)

or in the presence of oxygen by the reaction:

-o(s)

+ SZ(g)

+ $02(g)

= -Q4(s)

.GS

(6)

- 409 kiJ id-1

The wlue:GS is an indicator of the pssibility of reactions

occurring under standard conditions (289K and 1 amsphere).

If the mlue

ofSGs is negative, then therrm$mmically the reaction should proceed qmntaneous1y.

Generally the mre Ixksitive,UY, the rmre heat is required in

order to m&e the reaction take place.

Thus cne would expect reaction 4 to

be favoured rather than reaction 1.

This analysis takes m amount of the kinetics of the reaction.
Howaver, sane idea of the reaction kinetics under oxidising and reducing
conditions my be c&An& by mnsidering the relative reactivities of SO2
ami so3.

?hermlecule SCl3is ilDre reactive than sO2, both in terms ofkine-

tics and therrr&ynamics, so the preferential formation of So3 rather than

SO2 wuld lead tomre sulphurbeing

in a reactive formtiere it muld

react with tie available GO, KS, t&20 etc. 'Ihe equilibriabetween
and SO3 have been studied by LMyers.C13]

So2

Fran the eqirical equations

obtained byMyers, the graph in figure 2 (i) can be axstru~ed.

This

shms that the higher the percentage 02 present, the more likely is the formationof SO3 andhence the formtionof sul+ates whichtill. be mptured
bythemcuningfeed.

The lcwoqqen curves shas thatsulphurwill be

present as SO2 which can escape in the'kiln exhaust. Ektreme reducing

conditions lead to the formation of H2S reaqnisable

by its bad egg"

odour: inthis case therewuldbe virtuallymrecapture

on-feed.

of sulphurbythe

0
d

St

(xx-ii)

ua

NLlIS1J3hN83

20s

PD
EQSQI

::

0
4

(xxiii)
APPENDIX3
Heat pipe effect of volatile circulation at F!op? ibrks

i)

Latent Heats of Vapourisation/Disscciation

KC1

670

Kcals/Kg

N.B. V = Vqxmrisation
D = Dissociation

(v)

K2S04

+ 563 Kcals/Kg
+ 1805 Kcal.s/Kg

Na2SO4

+ 733 Kcals/Kg
+ 2096 Kcd.s/Kg

(A?

t 675 Kc&/Kg

(D)

Obtained fran Khor

caso4

c91

Recirculating volatiles
K20

2.457 % on clinker

Na20

0.149 % on clinker

SO3

4.509 % on clinker

Cl

1.092 % on clinker

This is quivaledto
KC1

2.29 % on clinker

K2SO4 1.94 % on clinker

Na2SO4 0.34 % on clinker
CaS04 5.46 % on clinker
Therefore Heat pipe effect, assuming all mqxmznts
KC1

16.46 Kcals/Kg

disscciate fully, is:

clinker

K2SO4 10.92 Kcds/Kg clinker

35.02 Kcals/Kg

clinker

Na2SO4 2.49 Kcds/Kg clinker

7.13 Kcals/Kg

clinker

CaSO4 36.85 Kcals/Kg clinker

Total 108.87 Kcals/Kg clinker
Therefore rraxinnrm

total heat pipe effect for I-bps is 109 Kc&s/Kg clinker.

Blue Circle Cement

PROCESS ENGINEERING TRAINING

PROGRAM

Module 13
Section 5

Alkali Volatilization- A Review of

Literature Available in 1977

ALKALI
A iU3VIEFw

VOLATILISATION

OP LITERATURE AVAILBBcg IN 1977

Information from 119 published references has been collated and

critically reviewed. The effect of alkalis on clinker properties and kiln
operation are briefly discussed, followed by an outline of the origins of
alkalis in raw materials and fuels and of the published physical
properties of relevant pure compounds.
Known physical and chemical factors affecting volatilisation are assessed
- a key objective of this review. Finally, information on the
distribution of alkalis in clinker is presented, together with an outline
of initial attempts at theoretically estimating recirculating loads and of
practically reducing their magnitude.
From the review it is concluded that there is general agreement on the
factors affecting the volatilisation of alkalis. Volatilisation is
increased by smaller nodule size, lower flux content in the clinker,
higher levels of chloride, lower levels of sulphur, higher temperatures,
longer heating periods and by higher contents of water vapour in the kiln
gases. It Is not clear what the effects of other factors are (over the
ranges studied in the available literature), and it may be inferred that
these are relatively small: gas flow rate and composition, alkali content,
presence of fluoride, LSF and S/R and oxidising conditions in the kiln..
Several of these factors appear worthy of further study.
There is, nevertheless, insufficient quantitative evidence to predict the
degree of volatilisation of alkalis from a given raw mix under particular
conditions. The interplay of phase equilibrium and transport phenomena is
probably too complicated for it to be worth while attempting to simply
calculate volatilisations from known thermochemical data: more complex
modelling is needed.
In practical terms, the circulation of alkalis is affected as much by
their condensation and deposition as by their volatilisation. It would be
useful to devote some attention to this hitherto somewhat neglected aspect
of the phenomenon.

ALKALI

VOLATILISATION

ISTN S7/4

A REVIEW OF LITERATURE AVAILABLE IN 1977

C O N T E N T S
Page

Nr.

1.

INTRODUCTION

2.

EFFECTS OF ALKALIS

2.1

Effects of high alkali levels in clinker

2.1.1 Alkali-aggregate reaction
2.1.2 Air-setting of cement
2.1.3 Effect on strength

2.2

Effects of high alkali levels on the manufacturing process

2.2.1 Effect on kiln fuel consumption
2.2.2 Corrosion of refractories
2.2.3 Build-up and preheater blockages.

3.

SOURCES OF ALKALIS
3.1

Alkalis in limestones

3.2

Alkalis in clays and shales

3.3

Alkalis in fuels

3.4

Methods of reducing alkali ,zontents

of materials

4.

PHYSICAL PROPERTIES OF ALKALI METAL COMPOUNDS

11

5.

FACTORS

12

5.1

AFFECTING

VOLATILISATION

Physical factors
5.1.1 Nodule and particle size
5.1.2 Effect of the liquid content of the clinker
5.1.3 Gas flow rates
5.1.4 Duration and temperature of heating
Cont./...

-iiN
Page r

5.2 - Chemical factors

5.2.1 K20 and Na20 contents
5.2.2 Effect of chloride and fluoride
5.2.3 Effect of SO2 and SO3
5.2.4 Effect of water vapour
5.2.5 Alkali containing minerals
5.2.6 Effect of LSF and SR
5.2.7 Effect of kiln atmosphere

6.

DISTRIBUTION OF ALKALIS IN CLINKER

28

7.

ALKALI CIRCDLATION IN THE KILN SYSTEM

31

7.1

Estimation of alkali circulation

7.2

Reducing alkali circulation

7.2.1 Removal and treatment of flue dust
7.2.2 Bypasses on preheater kilns
7.2.3 Other processes for alkali reduction

8.

CONCLUSIONS

9.

REFERENCES

10.

NAME INDEX TO REFERENCES

11.

TABLE 1

40

ISTN 87/4

ALKALI VOLATILXSATION
1.

- A REVIEW OF LITERATURE AVAILABLE IN 1977

INTRODUCTION
The following report summarises published information available
in 1977 on the subject of the behaviour of volatile alkalis in
cement clinker manufacture.

.These

materials are distinguished from other clinker-forming

materials by being solid at the lower temperatures encountered in the

kiln

system

but vaporising at higher temperature - for instance in the

kiln burning zone.

The effect of this is that they circulate within

the kiln and preheater system and can, under.certain

attain high concentrations.

circumstances,

The quantitive effect on alkali levels in

the clinker on removing some of this recirculating material by a

by-pass may not be immediately predictable.

Together with the compounds of the alkali metals sodium and

potassium, it is also convenient to consider other compounds involving
sulphur, chlorine and (less commonly) fluorine, which have boiling
points within the same range and which can exhibit similar behaviour.

Interest in the behaviour of alkalis in cement manufacture first

arose in connection with the possibility of obtaining potash for use
as a fertiliser from clays, feldspars and other unconventional
materials

(I).

A considerable amount of work was done on this topic

in the United States during the First World War, as at that time most
of that country's supply of potash was imported from Germany.
application,

In this

it was necessary to ensure that as high a proportion of

the potash in the raw feed as possible was volatilised and recovered

in the flue dust, and that this potash was water-soluble: this latter
feature could be achieved by heating in an oxidising atmsphere (2).
!
This m&hod of producing @ash was uneconmic when ample supplies of
potashwereavailable

fromother sources andsomrkwas discontinued

after the early 1920's.

The problem of alkali levels in CempJls was once mre brought
into pmninencewheh

it was suggested (3) thatcemants

withhigh

alkali levels reactedwith certainaggregates, suchas

scxw chertsr

shales and limestones, to produce destructive -ion-

This led

various national specifying authorities to limit the alkali content of.

cementS, ark3 so it was necessaryto
levels present.

investigatein5ms

of r&uciqg

In the 1950's, the use of the suspension preheater

for dry process kilns intr&ucedfurther

problems as alkalis are

recirculated within the kiln/prehsater system quite efficiently, so

t&&clinker

alkali levels cxnnotba

flue dust (which rarely &ibits

rich in alkalis).

reduced simplybydiscardingthe

a separately-collectable

FurLher, the high levels of recirculating alkali

salts tend to condense at critical pints

build-ups.

fraction

in the system and create

This high level of recirculation can be reduced (with soma

increase in fuel qmsuqtion)

by bleeding off sune of the gas from the

kilnbackendanddiscar~ngtheassociatedltustburden.
of bleedrequiredto

Thedegree

produce given results cannot always be predicted

accurately.
Research has been undertak~~atvaxious

establishuents, with the

aim of constructing a theoretical model for alkali behaviour in

kiln/preheater system and devising laboratory tests which can be

applied to raw materials to predict their behaviour in large scale
kiln systems under various circumstances. Possible areas for such
research are identified in the course of this review.
2.

EE7?Ems OF ALKALIS

2.1 Effects of high alkali levels in clinker

2.1.1 AUcdli-aqqreqate reaction
It was found by Stanton in 1940 (3) that cracking in concrete
structures

couldbeattributedto

reaction

betweenalkalisinthe

cemant and certain minerals in the aggregate, This phexxxmmn

wds

investigated by a number of organisiations, notably the US Wrreau of

Reclamation (41, frcm which it became apparent that the problem had
occurred with aggregates containing amorphous or microcrystalline
silica.

Examples of such aggregates are opaline cherts, siliceous

shales and limestones and scma types of andesite, rhyolite and

obsidian (5).
theUSAto

As a result of this work it becama standard practice in

specify a rfaximm value of 0.6% by weight for equivalent

Na20 content (Na20

aggregates.

+ 0.658 K-20) for the cement to be used with such

This limit was intrcxluced

as a tentative revision to the

ASIM standard for cemant in June 1959 and in subsequent versions

(e.g.

6) it is included as a characteristic which can be specified by

the purchaser if he considers it desirable.

The mst serious alkali-aggregate reaction problem were found
in the USA- mainly on the West Coast and in Nebraska and Kansas.

Reactive aggregates were also found in North Germany

(Schleswig-Holstein) and Denmark (71, but flint gravels found in
Germany and the UK have not generally been considered to be reactive.
Alkali-aggregate reaction is also known to occur occasionally in other
countries, e.g. Portugal and New Zealand (8).
Alkali limits are included in the standard specifications of
some countries.

For example Brazil (91, Peru (lO).and Venezuela (11)

followed the ASTM in setting an optional limit of 0.6% for (Na20 +

0.658 K20) where a reactive alkali is expected, and Mexico (12) has
set limits of 1.2, 0.9, 1.2, 0.8 and 0.9% for Types I, II, III, IV and
V cements reapectivsly, with an optional lower limit of 0.6% for use
in concrete with reactive aggregate.
It has, however, been suggested (13) that alkali contents below
0.6X, perhaps as low as 0.35X,

can still lead to reaction in some

circumstances.
Potash and soda generally have an equivalent effect in causing
expansion (7) but under some circumstances (14) it appears that more
potash than soda can be tolerated without expansion.
The form of the alkali does not seem to make a great difference
(13) in the long term, although water-soluble alkali is liberated more
rapidly from crystalline than from glassy alkali-containing phases in
early stages of hydration.

It is suggested that the significant

factor appears to be the concentration of hydroxide ions in the pore

solution (15).

2.1.2 Air setting of cement

Rapid air-setting of c-t in storage is due $ the' formation
of syngenite (X2SO4.CaS04.H20) (16, 17) and thus it is generally
desirable to keep the K20 level in the clinker as low as possible
where other factors indicate that air-setting-may be a problem.
2.1.3 Effect on strenqth
It appears to be generally accepted in the literature that, at
levels typically found in cm-e&s , increasing the aUcali.contentof
the clinker tends to increaseearlystrengths
strengths.

anddepress ultimate

Forexample,AS?Mtestson2in.~~cubesgavethe

following results for cemnts with different soluble K20 contents (and
similar levels of potential C3S, C3A and fineness) (18):

Cartpressive
strength at

Cexnt with
0.03% soluble K20
(MPa)

Carwtwith
0.61% soluble K$
@Pa)

1 &Y

8.41

9.79

3days

17.58

18.34

7day-5

27.72

24.41

28 days

45.09

32.41

The sama investigation found, however, that clinker fran which

the alkali (0.57% Na20 equivalent, minly as potassium salts) had.been
remved

by reburning with ammniumchloride

had lowx strengths atall

tinkas _.thahthe
addition.

samcemantreburned

DatxlreportedbyMussgnug

ina similarmannerwithoutNH4Cl
(21) ix-ldicatedthestrengthof

high-alkali clinker as being 100% greater than that of low-alksli

clinker atlday,
subsequently.

2O%greater

at3 days and similar at7days

These findings were basedon

clinkers withalkalisulphatecontents

and

tests co alargentznber

of

beW Oand4.0%

2.2 Effects of high alkali levels on the manufacturing process

2.2.1 Effect on kiln fuel consm@ion
The evaporation of alkalis in the burning zone consums
high mature,
Mature.

heat at

which is subsequently liberated at a lmer

Additionalfuelisrequiredtomaintainthesamburning

zone temperature under these conditions, and Weber (22) estimates thad
this can be up to 31 kcal/kg of clinker for a specific suspension
preheater kiln,12 kc&/kg

for aIqolkiln

and 5 kcal/kg for awt

process kiln, the difference being explained by the opportunity for

effective use of the low-grade heat liberated in each mse.
EIowever,

this fuel penalty my not be realised in practice,

because the presence of alkalis will tend to pramte

clinker minerals atalowar

the fornation of

temperature, thus permitting the burning

zone temperature to be reduced, other things being equal (19, 20).

2.2.2 Corrosion of refractories

Thamstserious
said not to be cWcal

adverse effect of alkalis on kiln linings is

but machanical (23, 24, 25). AlJ6i.i

salts -

notably

K2SC4 and Kc1 - condense on the brick at teqeratures

of

700-10bO"C and fill the pores, thus increasing the risk of cracking
and spalling with temperature changes.

In a kiln used part of the

time for the nanufacture of white cemaht under reducing conditions, it

was found (261 that sulphides, e.g. KE'eS2,
brick.

me deposited on the

Although the sulphides did not damage the brick and possibly

even tended to strengthen it, under oxiding conditions theywere

converted to sulphates with highly deleterious results.

2.2.3 E?uild-up

and preheater blockaqes

The volatile ca-qounds evaporated in the burning zone tend to

recondense on dust particles and in the cooler parts of the kiln
system.

There my ba a considerable teqerature

range in tiich these

canpounds are in the liquid state (see Section 2) and in parts of the
kiln system in this temperature range, severe build-up problems can
OCCUT.

Places where this happens axe the inletendof the kiln and

the preheater cyclones (271, in which coatings can lead to much

impaired flow of materials and ultimately complete blockage (28).
"Rule of thumb" limits have been quoted (29) of 0.015% for chloride
and 1.0 for the mlecular

ratio of sulphate to alkalis in the raw mix.

Although clogging problems are widespread with suspension preheater

kilns, it has boer~
them (30).

claimed thatshaftpreheaters are notsubjectto

3.

souw=Es OF ALKALIS
3.1

Alkalis in limestones
Innormal cases,publisheddata

suggests thatthelimastone

the raw mix is only a minor source of volatile carpounds.

in

Average

values for 345 American limastones are (31):

K20

0.33%

Na$

0.05%

=3

0.05%

Cl

0.22%

Anumber

of Lkrbyshirelimestones

had higher SO3 contents~ in

the range 0.1 to 4.8% (32).

One~uld~alkaliproblems
calcareousmai%rials
mud.

Suchmaterials

(33) for ewmple,

tobecanali.kelywhenusing

of marine origin, such as coral andaragonite

are used in Hawaii and the southern UnitedStates
but as far as is known no particular problems due to

high sodiumchloride contents are encountered.

Inecpdments

carried outatawet-prccess
xorks where sea
_
water was used to-h oyster shell rawmaterial and tom&e uq the
raw slurry (341, it was found that substitution of fresh water
decreased the Na$ levels in the clinker to scene &xx-k

but also

increased the K20 content, so that only a sr&l reduction of total

alkalicontentcould be obtained in this way.

3.2 Alkalis in clays and shales

These mterials
raw mix.

normally provide the bulk of the alkalis in the

Minerals with high alkali contents include feldspars, for

example orthcclasewhich
can contain up to12%

can contain I+ to17% X2Oand

Na2O,andclayminerals,

albitewhich

such as micas and

illite (35).

Analysis of 33 illites (36) showed a range of Na2C

contents.frcxn

0.05 to 1.05% with a maan of 0.27% and X2C contents fran

4.6 to 11.0% with a mean of 6.7%.

3.3 Alkalis in fuels
Coals contain alkalis both in themineralmtter,~~andX~
together forming 1 to 6% of a typical British coal ash, and a.s sodium
and potassium chlorides, either in the free stateor
coal stitances.

adsorbed on the

Chlorine levels can be up to 1% ;37).

are typicallymlow in sulphur, as arecoals

British coals

franthewastem

United

States, and Ruhr coal contains about 2.8% So3 (35). Eastern LE coals
are higher in sulphur as the following analysis (%'by
coal) confirms (38):
Illinois coal
w

0.15%

Nazo

0.12%

=3

10.48%

Cl

0.22%

(Ash

9.58%)

weight of dry

It has been suggested (39) that coal my contain iodine which is

found in flue dust, but this muld not ha expected to have a
significant effect.
Heavy fuel oils, for ewrrple No. 6 Fuel Oil Bunker

C) include

about 1% of water containing dissolved salts,mainly sodium chloride,

so that atypical ashmycontain
of up to about 0.1% of the oil.

32%Na20,

ckesponding

tocontents

Sulphur levels can be high, ranging

frcxn about 2% (SO31 for a low sulphur oil with 0.04% ash to 10% GO31
for a high sulphur oil with 0.02% ash (40).
Natural gas contains no alkalis.

Sulphurlevels

(mainly in tbe

form of H2.S) can ba quite high in the raw gas, but pipeline
specifications generally limit H2S contents to 4 ppxn by voluma to
avoid corrosion, so that rmst of this is removed (41).

(4~ by

volume of H2S corresponds to about 18.6 ppn by weight of S03, or about

0.0015 g SC3 per 1000 kcal).
Wastemterials

usedas

fuelcouldcontain

appreciable

volatiles, depending 'on the sour&. . Addition of chlorinated

hydrccarbon waste to fuel has keen proposed (42) as amans of
reduciiq clinker alkali levels.
3.4

Methods of reducinq alkali contents of materials

It does not appear to be possible to reduce the contents of

alkalis in clays and shales by any practicable nmns.

experiments by the Portland Cement Association (43)

A series of
investigated

leaching, flotation and heating methods and showed that the only

mtl-&.5 to reduce the alkalicontentof

the mterialappreciably

ware

(a1 heating at a temperature and LSF approaching those used in cemant

manufacture and (b) heating with concentrated mineral.
4.

Pl-lYSICALPROPElRTIES

acid.

OFALSAJLI MILTALCOMPOUNDS

Melting and boil& teqeratures

(at 1 amphere)

for scma

potassium; sodium and calcium cq&nds a.re given in&ble 1.

SOUrCeS

for these are (40) and (441, except where otherwise indicated.

It is

not'clear how significant these repours are under practical kiln

co&iitioti, since various eutectics form;
K2SO4/aSOq/I@1systemthelmest

for example in the

fusion teaiperatures

are in the range

650-700C (26).
VaqoU.pressures

have been investigated fran an early date (47).

Fig. 1 shows saturated repour pressures of KCH,

andNaF

&OH, KCl, &Cl, KF

(40) together with approximate values for K2S04 (48) and

Na2sOq (49);

these latter should be treatedwith scma caution as they

are ex%rapolated beyond the range of validity claimad for the

equations but should indicate general trends. For further details of
sadim chloride see (51, for potassium fluoride SW [SlJ and for
potassium and sodium sulphates (52). A considerable amunt

nf wxk on

the behaviour of potassium sulphate at high temperatures has baen

carried out in connection with its possible use as a seed in
magnetohydrodynamic power generation (53, 54).
the partial presssures of the VazTious

&an calculation of

species present at teqeratures

in the range 1500-2000'K and 1 amphere pressure when

sulphur-ccntaining
ctxnpcunds

fuel oil is burnt, it appears that the potassium

present in significant quantities muld be K-$04,, K$O3

andKOH.
Althouqn me vapour pressures \Flg.ll

agree v1 general. terms

with the order of volatility observed in practice, one muld

anticipate considerable difficulty in using them to predict absolute
volatilisation levels at given temperatures. Plbst of the species
involved do not cccur as pure canpounds in the solid state and
pressures andteqeratures

inside nodules of clinker, for instance,

are not wall known.

5.

FxxxxsAFFE13TINGvoLATILIs~IoN
By "volatilisation" we m3n the departure fran heat-treated
arterial of a proportion of a stitance
llliX.

This heattreatmtzntmayba

case volatilised stitances

originally present in a raw

under labratoryccnditions,

are generally r-v&l fran the system, or

in

tendto

recycled,'giving a lowar overall %xJ%e

obtain~under

in which

condense and ba

of vclatilisation".

Values

different circumstances must therefore be carpared with

care.
5.1 Physical factors
5.1.1 Nodule and particle size
It muld be expected that under certain conditions the rate'of
loss of volatile suketances from nodules could be limited by their

diffusion frcm the interior of the nodule to the surface. Experiments

by the-FCA with various sizes of nodules of raw mix and with rebmning
clinker nodules of various sizes tend to confirm this' idea (551,
althouqh the author suggests elsewhere (56) that in practice the
effect of nodule size my be small in cmparison
temperature and residence~tima

with factors such as

at-mature. This

view is supported

bv a series of statistically-controlled experiments at NIIITsenrent

(57).

Results of Goes Ad Keil (58) and Draper (55) have been

replotted on a lag/log scale (Fig.21 and agree (with the exception of

one value for lmn pellet diameter), indicating a relationship of the
form:
volatilisationd

.3'd
JAnalysis of pieces of clinker of different s&as

(59) showA

K20

levels to be higher & the larger pieces.

Although it appears that the effect of nodule size on
volatilization is not very great in practice (for example, a variation
in pill size of f 25% would alter volatilisation by C 7.5% - lo%),

it

could nevertheless be useful to investiuate it more thoroughly, both

to verify the relationship, which is based on very limited data, and
as a guide to the factors limiting alkali volatilisation under various
ccxnbinations

of burning tirre and temperature.

Scma mrk has been &ne with beds of poxdered raw mterials

of

various depths (55, 591, the results of which broadly parallel those
for nodulised raw mterials.

No publications dealing'with the effect

of the size of individual particles on volatilisation have been found;

it seems reasonable to propose that below a certain size this muld
have little effect but it muld be useful for stamlardising
-imental

results to put this ptulate

to the test.

5.1.2 Effect of the liquid content of the clinker

Since the effect of increasing the liquid content is to reduce
the porosity of the clinker, this factor would be expected to lower
the rate of alkali volatilisation (58, 60, 61). Mussgzmg, howaver,
found no correlation baMaen

total alkali (KS + Na$) and Fe203

contents of a number of clinkers (211 anda separate series of

experim31.ts

showed that the A/F rt'

a 10 was the least significant -of the

five factors studied (571, being overshadoWed by temperature, time,

sulphate content and pellet size.

Itwmld'te

of interest to

investigate the magnitude of this effect in store detail, as there

appears to be scrne disagreeman tas to its iqortance.
5.1.3 Gas flow rates
The effect of gas flow rates on the volatilisation of alkalis
has been studied to a limited extent only, for example by Jackson and
Morgan (621

who detected ho significant variation. I&oratory

vimants

on volatilisation rates scmatimas incorporate means for

passing a gas canposition approximating to kiln gas over the raw mix
(551, for instance 77% N2, 20% CO2 and 3% 02 (56).

So2 (see Section

4.2.3.) and water vapour (Section 4.2.4.) have been added to this.gas
in varying quantities, but no results of altering its flow rate have

been found.
It is possible that this could be significant in saz
.cases, as significant variations in volatilisation with stirrins a bed
of powder have been reported (59).
5.1.4 Euration and temperature of heating
There is general agreemant that the degree of volatilisation
increa.%es

with the tim andwith

the Mrperature

of heat treatment,

although it has been noted (60) that after heating fran 700C to
14OO"C,

little additional volatilisation takes place on subsequent

heating at 1400C for a further 30 minutes. The effects of duration

and teqerature
in rnagni.tGde

are thought to be highly significant and about equal

(57).

The rate of volatilisation frcxn a particular raw mix under gives

temperature andother
investigated, buts-

definedexternal

&hditions

has not been

interesting conclusions can be drawn fran

additional analysis of reported data.

An assumption which is widely made, albeit usually implicitly,
is that the proportion of a species volatilised on heating for a given
time and at a particular temperature is characteristic of the raw mix,
i.e.

that the volatilisation rate is proportional to the amunt

present (a). Thus:

-da=ka
dt:

- a0 r 1h = orkdf
atJ

a-

tI

In (at)
(a,>

-kt

where at is the amount of alkali present at.time t and a0 the amount

present at the start of heating.
is, if the volatilisation

If this assumption is correct, that

reaction is first order with respect to K20

or Na20 content, plotting the log of the amount of alkali remaining

against time should give a straight line.

Data from Palmer and

Baylees (34) and Woods (63) were analysed in this way and results for
selected mixes are shown in Fig. 3.

It will be noted that although

the fit is very good in most cases, the line of best fit does not
always pass through the point corresponding to 100% original alkali
content at time t=O.

To explain this, it is necessary to consider t

heating conditions which were applied.

In (34)

the samples were

heated for 20 minutes at 1800F (980C) to decarbonate kthem and

immediately transferred into a high-temperature furnace maintained at
2600F (1427"C),
time.

in which they were keept for the appropriate length oi

The heating programme in (63) is not stated.

predicted initial alkali level substantially below

In the case of a
100X,

a possible

explanation is that part of the alkali is driven out while the sample
is being heated up to the experimental temperature.

This appears to

be reasonable, although it is surprising that, if this is

SO,

no

correlation was found between the rate constant k and the initial
alkali level ao.

This could be due to the presence in the mix of

chloride levels appreciably less than those needed to react with all
the K20 (the addition of chloride is known to increase volatilisatic

significantly - see Section 4.2.2.).

(34)

The line shown for Mix 12 of

with added CaC12 shows a very low level of K20 remaining at the

start of heating at 1427"C,

and a rate constant which, although the

highest of those observed, is not much higher than that of the same
mix without CaC12 (not shown in Fig. 3), - 0.0820 and - 0.0771
respectively.

In the latter case, however, no volatilisation appeared

to have taken place at lover temperatures.

Another problem is presented by results such as those for Mix 5

of (34)

and Mix 3564, 125O"C,

of (63).

In these cases, it appears

that volatilisation does not begin until some time after the start of
heating at the final temperature.
K20 volatilisation takes plz&e

after 60 minutes, at 125O'C

In the case of Mix 3565 of (631,

at llOO"C,

no

at 12OO'C it begins only

it begins after 30 minutes and at 1300C

and higher temperatures it starts immediately.

This delay is perhaps

due to the alkali-containing materials initially requiring to react

with other components before the alkalis are released.

The correlation coefficients were recalculated in selected cases

including the point a = 100X;, t = 0.

As would be expected, the

correlation was usually slightly improved where the original line of

best fit passed near this point, but where significant changes in k
and/or a0 resulted, the effect on the correlation was unpredictable.
It vould therefore be desirable to carry out experiments on the rate
of volatilisation, in which the actual alkali contents at t = 0 are
measured.

If the relation suggested by the data already examined was

,.
confirmed, this would mean that the proportion of alkali volatilised

under given conditions was independent of the absolute amunt

present

(until; of course, all the alkali was lost).

An alternative way of indicating the propensity of a raw mix to
lose alkali on heating has been proposed by Palmar and Rayless (34).
This is the "resistance factor which is the sum of the percentages OI
alkali remaining after heating for 10, 30, 50 and 70 minutes at 2600'F
(1470C).

This empirical mathod appears to be suitable for cuqar'i3-g

raw mixes with each other, but does not permit the prediction of
volatilisations

under other conditions.

Its use has not been taken up

by other mrkers.
Maintaining clinker atan elevated temperature has been used as
a n&hod of reducing its alkali content (64, 65). A tE Bureau of
Reclamation specification for low heat, low alkali cement was ntet by
reheating the clinker frcm the kiln to 1427C in a converted rotary
cooler-

Na$ equivalents decreased frcm 0.889 to 1.19% in the

(loss-free) raw mix to 0.411 to 0.890% in the clinker and to 0.315 to

0.716% in the treated clinker.
was abxt550

The heat input to the clinker treater

kcal/lq clinker,which

renders themathodunattractive

in nomal circmnstances.
Data obtained by the Portland Cement Association (56) for a
number of kilns show that total alkali volatilisation tends to
decrease as the residence time of the mterials

in the burning zone

decreases, but the effect of other factors is tcogreatto

quantitative analysis.

permit a

The effect of mrature

magnitude to that of tine (51).

on volatilisation is canparable in
It has been stated that "the burning

of cement clinker takes place within a temperature range where a

canparatively small rise in the temperature will cause a noticeable
increase in the amount of alkalis volatilised" (66).

It is therefore

said to be necessary to control the burning zone temperature

accuratfAy.

Itcertainlyseems

to be the case that there is a big

increase bet- ll.50 and 13OO'C (particularly bat- 1200 and

1250C) in the percentage of the original K20 volatilised in one hour
(591, although in normal conditions one would expect rmst clinker to
reach these teqeratures

in a kiln. Kate constants have been

calculated fran the data of Wads (63) at a range of matures

for

K20 and Na20 volatilisation frcxn one raw mix and these are shown in
Fig. 4.

The rate constant increases roughlylinearlywith

temperature, the effect of tgmperature being much more significant for

K2Othan

for Na20.

Itmightbeexpe&ed

fromtheorythatlqeK~uld

be proportional to l/T ("K), but the limited data available do not

appear to confirm this and further exp~imznts

muld be desirable.

The increase in K$ volatilisation in a given tima with temperature

.follows a logistic curve (591, since the absolute rate of
volatilisation after a given tima initially increases with temperature
;%
(as the effect of the increase in kwith temperature wthe daninant
factor) and then declines again (as the amount of alkali remaining
becomes the dxninant factor).
'i>- ature at which volatilisation begins can be as low as..
700-800C for El (21) but is rrore typically in the region of 1100C.

The volatilities of the various species are discussed in mre detail

below.
5 . 2 Chemical factors
5.2.1 KS and Nag contents
It has already been suggested (in Section 4.1.4.

above) that the

kinetics of alkali volatilisation are first order with respect to RIO

and Na$, and if this is the case the percentage volatilisation is
Hcmever,dataobtainedbytheFa

independent of the absolute levels.

(67) suggest that reduction in alkalis in fullsize

with alkalicontentof

the rawmix.

kilns increases

These findings are not

necessarilycontradictory,as

clays andshales withhigher K$and

Na$ my, for ample ,contain

thein inaless firmlyminbined

form,

lost at loam temperatures, but this is certainly an area which xmld

repay investigation.
As far as the relative volatilisation of Na$) and K20 is
concerned, it appears (56) that volatilisation of Na20 is negligible
where Kg volatilisation is less than 30%, and is at about half the
level of additional K20 volatilisation above the 30% "threshold".
5.2.2 Effect of chloride and fluoride
The early studies on the recovery of potash fran cement flue
dust and other minerals (62) showad that the amount of KF volatilised
under given conditions could be increased by addition of calcium or
sodiumchloride.

Itwas found to be~necessary to add just over the

quantity needed to combine with the K20 to give KCl.

more effective in promoting volatilisation than

NaCl

CaCi2.

was rather

With the

interest in reducing total alkali in the clinker, the effects of NaCl

and CaCl2
-(63)

on Na20 volatilisation also were investigated, and Woods

found that although NaCl

K20 than was CaC12,

increased by NaCl

was indeed more errective in liberating

the Na20 content of the clinker was noticeably

addition.

The use of calcium chloride was studied

~
in full size kilns by Holden (68), who found that the removal of
alkali by addition of calcium chloride could be estimated for CaC12
additions up to 1.5% by the empirical equation:

Ar f

0.559 K, [CaClZI
100 - L

where A, * total alkali removal expressed as Na20,

efficiency of utilisation of CaCl2,
material and [CaClz]

K, = apparent

L = loss on ignition of raw

= % of CaC12 added (dry basis). The efficiencies

of CaCl2 utilisation observed were 27% to 34% (average 32%) for wet
process kilns and 45% to 56% (average 48%) for long dry process kilns.
Addition of 35% commercial grade hydrocholoric acid has been used as a
cheaper alternative to CaC12 where this was not readily available
locally (69).

No particular problems were encountered in adding the

acid to the slurry tanks and control of the quantity added was better
than when flake calcium chloride was used.

Problems have arisen with low volatilisation and build-up

problems when CaC12 is used in long dry process kilns(70). Although
no published data are to hand, it would be expected that levels of

chloride addition would cause severe problems in preheater kilns and

muld probably have little effect on alkali reduction.
Contrary to the above results, laboratory tests on smrrles

man

six works in Central Asia (71) showed that addition of NaCl did not
appreciably increase and inmstcases

reduced the Na2C1contentof

the

clinker, but that CaC12 was in rmst cases more effective than NaCl.
Although it is known that addition of chlorides decreases alkal
contents, further mrk needs to be Qne before the precise effect at
various temperatures can be predicted.
Eata on the effect of fluorides on alkali volatilisation are

spar=.

woods (63) states that where calcium fluoride was added to a

raw mix to produce a high early strength cerrwt, the clinker mntaine
less potash than usual butahoutthe

same amount of soda. He suggest

that addition of CaF2 &uld probably only be of value in pramting

volatilisation where the raw mix was high in KS but relatively ion i.
Na20.

According to Sprung and van Seebach (721, theremdymmic

considerations favour the formation of calcium fluoride and alkali

matal

sulphates

The amunts
cases.

rather

thanalkali

fluorides

of fluoride in rawmaterials

andcalciumsulphate.

are, however, small in mst

Cost and possible toxicity muld probably make fluoride

addition unattractive as a means of increasing alkali volatilisatior

even if it were mre effective,
5.2.3 Effect of So2 and So3
The water-soluble alkali salts remaining in clinker are almst

entirely sulphates (16), chlorides having been lost through

volatilisation.

The effect of sulphur compounds 0; the alkali content

of the clinker is therefore considerable, the sulphur deriving both

from the fuel and from sulphur compounds naturally occurring in (or
added to) the raw materials.

The sulphur content of the clinker is

typically 50 to 65% of the total sulphur input from all these sources,
so that it is possible for the sulphur content of the clinker to be
higher than that of the raw feed (22).

As the vapour pressure of

alkali sulphates is considerably lower than that of chlorides (Fig.1)

it would be expected that higher levels of sulphur in the clinker
would tend to prevent alkali volatilisation and this is in fact
observed (54, 56).
greater

than

on

The effect on K20 volatilisation is appreciably

Na20

volatilisation

(62).

K20,

Na20

and

chloride

contents of flue dust are reduced even in the presence of chloride in

the feed (73) - which can be advantageous in improving the efficiency
of electrostatic precipitation - so, it would seem, demonstrating that
alkali

sulphates

are

formed

or

volatile chlorides and oxides.

retained

in

Statistical

preference

to

examination

the
shows

more
that

SO3 content is second in importance only to time and temperature in

affecting alkali volatilisation (57). Operating data (56) from ITS
kilns

show

relation

between

volatilisation

of

SO3

and K20,

although

it is not clear what is the significance of this and there is some

evidence (56) that the effect of SO3 is greater at higher
temperatures.

The equilibrium between calcium sulphate and CaO, SO2

and

been

SO3

lOOOC,

has

extensively

investigated.

At temperatures above

combustion of sulphur compounds in the fuel results

predanhantly

in SC2 (271, which can react with lime in the

temperature range 600-900C.

3

4 CaO + 4 SO2. . 3 CaSO4 + CaS

The xraximum

absorption of SC2 by raw meal was obtained at 880C;

this is significant since a high proportion of the sulphur in the kiln

exit gas can be absorbed by raw meal in the preheater (74). Snaller
amounts of sulphur can also be absorbed byrawmterials,

where these

are dried by kiln exit gases in the rawmill.

The decaqosition

of pure C&O4 begins at about 900-lOOO'C, but

is initially very slow.

Sax5CaSC4remainsunconvertedatabout

1385C, which is the melting point of the CaO-CaSO4 entectic (75).

The decaqosition

rate is much increased by the addition of other

materials, eg silica, alumina, kaolin and iron odde, and by the

presence of water vapour.

-sition

atmosphere, when CaS is formed.

for mmfacturing

cen?ent

is also faster in a reducing

This fact has been used in processes

and sulphuric acid, by heating a mix of

anhydrite, coke and argillaceous materials in a rotary kiln (76).

-

3 - 4

+x

+m2+cas

+c&-+4c.&+4=2

The lime reacts with the argillaceous ccqxxents

minerals and the SC2 is separated fran the exit gas.

to give clinker
It is necessary

to maintain oxidising conditions in the exit gas to avoid the

formation of elemental sulphur, for instance by reaction of the H2S
andSO2.

It has been claimed that low-alkali ceinentcanbeproduced

in this prccess by adding calcium chloride in the usual way (77).

This 'is perhaps unexpected in view of the effect of even a small
percentage of added calcium sulphate on alkali volatilisation, and no
doubt results from dissociation of the CaC12, allowing alkalis to
evolve as chlorides.
Further mrk on this aspect should investigate in greater detail
the effect of varying levels of sulphate addition at various
temperatures and in the presence of other materials; also, the
relative effects of So2 in the kiln atxmsphere and calcium sulphate in
thefeed.
5.2.4 Effect of water vapour
Fran the relatively high vqxmr pressures of KOH and NaOH it
might be exqected that water vqour

+muld pramte

volatilisation and

this is confirmed by laboratory expariments by Goes and Keil (581,

which gave the following results

H2OiIlkilngas
% byvol.

K20 volatilisation with 30 min at

1330c
%

14OOY
%

64

83

76

100

10

100

n.d.

Wcods (70) reports that satka works have claimad that spraying
small Zn-ounts

of water (approx.

2ml/kg clinker) into the burning zone

directly above the flame my reduce the clinker alk$i

-0.1 to 0.2% (on clinke?z).

content by

This effect might have sme relevance to

the observed tendency for gas (andperhaps oil-fired) kilns to have

higher alJcali

volatilisations than coal-fired kilns (18, 54, 65) since

more water vapour muld ba produced by ccanbustion of hydrocarbons.

(Spiers

reports moisture levels in canbustion products as:

2 - 5% coke
10 - 15% Lignite
7 - 9% coking coass
5 - 7% Anthracite
1 0 % - H.F.O. 1

5.2.5 Alkali-containinq minerals

Scdiumandpotassiumin

rawmterials

can occur ina number of

different minerals (see Section 2) and these can vary in their

tendency to lose alkalis (57).

Elpertits

with rawmaals

containing

mica, illite and orthaclase feldspar (58) showad that alkali w lost
mre readily fran mica and illite than from feldspar at temperatures
of 900-1250C.

It is suggested that this my be due to difference in

bond energies for the various minerals. Another factor may explain
differences between minerals; for example in gr eensand (glauconite)
the silicate is hydrated, (62) leading to formation of K20 in the
presence of H$ when heated, allowing the formation of Koti, tiich has
a higher vapour pressure than

KS.

(see 4.2.4.)

It muld be useful to investigate the response to heat treatmant

of various potash and scda containing minerals in mre detail, and
this muld require sama a-cans of allowing for the effects of other
factors and also devising a quantitative msasure of the "availability"
of K2OandNa20

in theminerals.

5.2.6 Effect of LSF and SR

Although little reliable infomtion

on this subject has been

found, such as there is indicates thatitwxld

further Czq&nmts.

be useful tocarryout

Ekperirrwts on remving alkalis frcmraw

materials by heating (43) indicated that there was no mlatilisation

on heating mixtures of clay or shale with 10% or less of limastone to
fusion and that alkalis wsre not released in significant quantity
until the amuntof

lime approached that of a normal cenwt raw mix.

The data are shown in Fig. 5 where percentage loss of alkali is shown
in relation to the proportion of CaC03 in the mix, and in Fig. 6,
which shows alkali loss in relation to the TLSF - which has been
estirrated

since the caqxxsition of the shale was not reported. It

will be seen fran these figures that the proportions investigaM

do

not permit us to say whether ease of volatilisation varies with LSF

within the range normlly

found, or whether aminimumlima

required, above which there is little variation.

content is

Earlier experiments

(62) had shown that KS could be volatilised frm greensand

(glauconite)/l~stone/cdlcium

chloride mixes containing as little as

one third limestone, if these were heated to temperatures just below

that of fusion.

The effect of silica ratio on alkali volatilisation has not been

investigated. Tbare is scma evidence (68) of a higher, calculated c3S
content leading to lowar alkali levels in clinker, and it ms found
(61) that clinker alkali levels could ba reduced by replacing part of
the argillaceous mterial

with sand.

Howevex, in this case the effect

of increased S.R. was not distinguished fran the effects of lmer

alkali levels in the sand reducing &-se in the raw feed and the
necessary higher burning matures, which one wuld

expect to ba

more significant.
5.2.7 Effect of kiln a-here
In normal centant
cment/sulphuric

cperations, as distinct fran the

acid process (see 5.2.3 above), the limited

information available frm laboratory tests (59) suggests that

volatilisation is not greatly affected by the kiln amphere.
results indicate that a reducing amphere pramtes

The

volatilisation at

1200C and 1250C and perhaps slightly depresses it at 13OO"C, but the
data do notreallyperm.ituseful

conclusions to be drawn.

FundamantaJ.ly,a reducingatmxiphere

is expectedtoprcm&e

volatilisation of alkali sulphates in the burning zone.

6.

DISTRIEUTIOJJ7OF~SINCLIN?GZR
The distribution of alkalis is not revie++ed

in detail, as the

literature on this subject is very extensive.

E'rcm the efforts to utilise flue dust during the First World
War, it had been noted (78) that potash in flue dust existed in

several different forms with different degrees of solubility in water

or acid, but no substantial progress k~s made until the 1930's and
1940's, when ah extensive programe
(US) Natioml

of mrk GELS carried out at the

Bureau of Standards, in cooperation with the Portland

Cement Association (79 - 84).

Newkirk (85, 86) concluded that the

alkali-containing compounds M&A3 andKC2jSl2ere

presence of SC3 andCaSO4
which xas imiscible
<I,

and reacted to give an alkalisulphate

with other phases. For R$/SO3

the SC3 reactedpreferehtiallywith the KS.

phase is imzdiately

unstable in the

mlecular

phase

ratios

The alkalisulphati

soluble in mter (87) although alkali fran other

phases is not.
Segnit (881 suggested that at least some, and possibly a
considerable part, of the soda in Portland cemnt

occurs in the C$.

On the other hand, investigations at the Bulgarian Building Research

Institute appeared to suggest that both KS and Na-$ are predminantly
in the form of aluminates (89).

Toropov and Dobrovolksky (60) support

Newkirk's opinion that most of the KS is in belit&;

they excluded

clinkers with free alkali sulphate from consideration.

Pollitt and Brown (16) generally mnfirm

Newkirk's findings on

alkali sulphates, noting that equilibrium conditions are not attained

in a kiln, so that works clinkers tend to contain mre calcium
sulphate thah do latxlratory

clihkers.

They also suggestthatscme

potash could cxcu~ in the C3A and that t&h potash and soda could
exist in alite in small quantities.

Azelitskaya and co-workers (90, 91)

found that addition of

gypsum to raw mixes containing alkalis diminished t?ne effect of

alkalis, which (at a raw mix SO3 content of 0.6% or less) was to
decrease C3S

content.

With the addition of.5 to 10% gypsum to the raw

mix (which also raised the LSF and hence the C3S content), alkalis
were in the form of their sulphates.

When investigating dust return,

Luginina and Shaposhnikova (92) found that the presence of high levels
of alkali sulphate tended to reduce alite formation and to give alite
crystals-with inclusions of belite and alkali glass and eroded edges.
Other work (93) confirmed that the addition of gypsum to
alkali-containing mixes tends to convert the alkalis to soluble
sulphates;

in this work the optimum SO3 level in the clinker appears

to be about 4%.

Subsequent work by Abbassi (94) confirms the general trend of

previous

investigations.

Alkalis can exist in clinker separately, as

sulphates or carbonates, or in other phases:

preferentially.

C3A dissolves soda

However, all the phases can contain at least some

potash and soda, the amounts in the flux phase being below 0.2%.

The

ratio of the K20 content of the C2S to that of the C3S is usually
about IO, the maximum K20 content found for C2S
1.9%.

in this work being

The Na20 ratio varies, but is about 10 for max. 0.6% for Na20

recorded in the C~S.

More Na20 was found (relatively) in the C3A at a

max. of 1.7% (cp, 3.2% max. for K20).

7.

ALK?GICIRCULATION

INTKEKIINSYSTEM

7.1 Estimation of alkali circulation

Equations to give the concentration of a volatile cqnent i at
various points in the system have been derived by Ritmann

for

Polysius AG (28, 95). According to these, in a system without a

bypass, the attmntof

the ith caqmentrecycled

to the incaning mterials

frcxnthe kiln gases

(referred to as the adsorbed phase, since

this is believed to consist rtkainly of alkali carpounds condensed on

raw material particles) is given by
(Eli + Bri*) .ai.Ci
Xi =
l- c2i.ai
ci

represents the amount of the.cmponent

Bri* =Bri/Ci

in the raw maal

where Bri is the amount of the cmponent

in the fuel

ffl i

represents the volatilisation of i from the raw ma1

52i

represents the volatilisation of i from the recycled naterial,

and

ai

represents theextenttowhich the component is reabsorbed from

the gases in the preheater.
Note that Xi, Ci and Bri can be in any consistent units of ItaSS,

such as (but not necessarily) g/kg of clinker or per cent of clinker.

$li, f2i ad ai are fractions not percentages.
It is ~instructive
this equation.

to consider the assurrptions

mde in deriving

If the various value& are shown in a diagram Ri is the

content of the ccmponent i in the clinker:

Xi
T
ar

Then if Xin and Xi(n-1) are the values of Xi after n and (n-l)
:ycles

respectively:

Xin = Bri + Ci Eli + Xi(n-1) 2i

But in equilibrium conditions the amount of each component

circulating is constant,so that

xi
-

= Bri + Ci Eli + Xi E2i

ai

Xi s ai (Bri + cilli)
(1-a c2i >
or Xi 31 ai Ci (Eli

+ Bri*)

(1-a C2i)

.This

assumes that the whole of the volatile component in the

fuel is volatilised and that no raw feed is lost to the stack (or
dumped) from the kiln/preheater system.

Using the same symbols and assumptions, it is clear that Ri (1

content of component i in the clinker) will be the amount not
volatilised, i.e.

Ri = (IL-&li)

Ci + Cl-f2i)

Xi

. i

hence
Ri = (l-<li)

Ci + (l-[*i)

= Ci
(l-aiE2i )

E(l-Eli)

+ ai Bri" (1-{2;)

II

f ai (<li-E2i

If ws can neglect the volatile content of the fuel, for exaqle

for K20 and Nag if oil or gas is used, Bri* = 0 and
[ (l-Eli)

Ri = ci

+ ai (Eli -czi)

(l-aic2i)
= Ci (l-Eli (l-ail 1
(
7l-aiE2il )
If a fraction V of the kiln gases is by-passed, then the
volatile ccqonent

i passing on to the preheater will be reduced to a

level Xi Cl-V)/ai,

and the axrount

recirculating will be Xi (1-V).

We

can therefore represent the effect of a by-pass in the above equation

by replacing ai by a'i where
a'i = ai (1-V)
These assm@ions

are cleiuly open to criticism. The concept of

differing volatilisations for raw mterial

and recycled mterial

is

swrted by soma qrtitalevidence (28) which suggests that

alkalis are rtore readily volatilised frcm recycled mterial
raw mterials,

than from

although other mrk (55, 59) indicates that alkalis in

flue dust are less readily volatilised than those in raw materials.
However, the values for any single ccqqent

do not autamtically

take

into account the amount of other components present. For example, the
volatilisation of KS will be. increased by the presence of chloride
and&creased by the presence of SO3.

As these ccaqxments are lost

fran the system to different extents, the proportions in the

recirculating material willvaryas the system tends towards
equilibrium.

Although the volatile content of coal ash is relatively

small, it is not necessarily all lost (96) and sma alkalis my reach
the clinker from this source.

It is also possible that SO3 fran oil

fuel could be re-absorbed at relatively high ixqeratures.

Ibere

seems no reason to assume that dust lost frcm the prekzker

will

contain alkalifrcmnthekilngases

only and thathoalkaliwill

be

lost frcm the raw material (55). It appears that in practice the
factorsE1

a.udc2 are estimatedempirically

in a particular case, so

that while these equations may be useful for predicting the effect of
varying the level of ah existing by-pass, for example, they muld
probably ba of little value in predicting the bahaviour of a raw
mterial

in aproposed

hew system.

Teoreanu and Puri (94) adopt different ratios which seem to be

less gene-rally applicable than those of Ritzmann. They &, hcmver,
intrcduce the possibility of non-equilibrium conditions in the kiln,
for instance where alkali-containing coatings build up over a period
and then break free.

It is clear that in addition to the factors affecting

volatilisation which have been discussed at scnte length abve, the
condensation and recirculation of alkalis are also important. One
mans which has been adopted for following alkali circulation is the
use of radioisotopes such as K42.

Inanexperimantonawetprccess

kiln (981 it was found that the man number of cycles was about five,
the time for each being 110-120 minutes. This value is on the high
side, caqaredwith
(99, 100).

those derivedbycalculaticm

frmalkalibalances

The efficiency of capture of dust by the preheatar system

varies according to the type. Ritmann

quctes typical values for dust

collection efficiency for cyclone preheabars of 60-90% for stages

3 and 4 (1011.

The total capture of alkali-ccntaining dust will

therefore be very high, and hence the alkali levels in'the flue dust
are very low (102) and the circulating load very high.
kilns a proportim

& wet process

of the alkali, say 40% of the KS (221, is

precipitated, and alkali levels in the flue dust are mderate

(102).

The typical Le@ kiln (22) emits little dust and apparently traps
aln-ostno alkali in the grate:

very high values have been reported

for the alkali content of the flue dust (22).

7.2 Reducinq alkali circulation
7.2.1 Rezmval and treatnwt

of flue dust

Over a significant period of tima,

the kiln systmmustbe

zero.

the net gain of alkalis in

Therefore, apart frm the small amunt

of alkali in the stack gas which is not precipitated in the dust

collectors, all the alkali put into the systemmustleave either in

the clinker or in the flue dust. Thus, it is only possible to reduce

the alkali level in the clinker to a level below that arriving in the
rawmeal

if the flue dust is either dmped

or treated to reduce its

alkali content before being returned to the process (55).

In a wet

process kiln, discarding the dust reduced the equivalentNa20

clinker fran 0.68% to 0.54% (68). As dust rermml
amour&of

circulating alkali, however, the rduction

in the

will reduce the

in clinker alkali

that can be achieved by discarding dust will be less than the alkali
content of the reamed
Anumber

dust.

of ways of treating the dust to reduce its alkali

contenthavebeenproposed.

As the finer fraction contains a higher

proportion of alkalis (103,104) it should be possible to reduce alkali

circulation by returning only the coarser fraction of the dust,
although it is doubtful whether the differential muld often be large
enough to mke this procedure mrthwhile.
For reasons discussed above, there is a considerable variation
in water solubility of K20 between flue dusts, ranging in one study
(103) fran 2 to 96%.

Because of this the efficiency of leaching

vaxies, but it has frequently baen used or proposed (105 - 108).

is obviously best suited for use with uet process kilns.

Problerrs

It
may

arise in disposal of the salt solution produced, as it is rarely

econanic to recover these salts as solids.
Another mans of reducing the alkali content of flue dust is to
heat it to, say, 1200C.
bed reactor (109).

This can be done conveniently in a fluidised

7.2.2- Bypasses on preheater kilns

In preheater kilns - especially 4stage

cyclone preheater kilns

-virtually all the alkalis in the gas franthe

kiln are absorbed by

raw feed in the preheater.

in the flue dust is

Thealkalicontent

therefore very low and little reduction in the alkali content of the
clinker or in the circulatingloadcan
treating the dust.

be obtainedbyremving

or

One way in which this could can be achieved is by

removing part of the gas and/or dust before it enters the preheater,
cooling it, collecting the dust and discharging the gas to the
a?mosphere (21).
of alkali.

Problems tend to arise in practice with the txlild-w

Although it muld be preferable to cool the gases by uater

spray, as the gas volumes needing cleaning are nmch less in this case,
there is so far little practical experience of this, ccoling by mixing
with cold air having been used in rmst cases (ll.0).

It appea.rs

that

the effect of a bypass is greatest up to about 20% bleed (991, above

which the increased fuel consumption, which is about 40 kcal/kg of
clinker at 10% bleed (211, becomes excessive in relation to the alkali
reduction obtained.

Suggested bypass levels (ill) are 2-4% to reduce

operational problem, eg blockages, and S-15% to reduce alkali levels

intheclinke.r.

A possible modification proposed to the bypass (112)

is to pass the bled-off gas through an initial cyclone and return the
coarse dust (with a low alkali content) direct to the kiln. Although
this means thatmre
of alkaliremval,

gas has to be bled off tomintain

the same level

the quantity of dust to be collected is much less.

Incorporation of a bypass enables the maximum permissible

chloride content of the raw fe& (see Section 1.2.3.) to be increased
frcxn 0.015% to about 0.075% at 10% bypass (29). As wxld be -ted
the effect of a bypass in reducing K2Ois
Na$.

It has baen sugges&

itnaybecaae

greater thau the effect on

that if high levels of bypass are needed

desirable to use a precalciner system (ll3,

volIxrrs of gas passing through the kiln is thenmuch

increase in fuel consuqtion

ll4)

as the

reducedandthe

at high bypass levels (eg 50 to 100%) is

less important.
Although alkali problems are n-c&often

eucounteredwith

cyclone

preheaterkilns,bypassescanbeusedwithgratepreheaterkilnsdlso
(IS, IX). In this case, the cleaned bypass gas can be used for

7.2.3 Other processes for alkali reduction

Same alternative n&hods
beenproposed.

for reducing alkali circulation have

One of these (ll7,

118) involves using excess hot air

fran the cooler for drying pellets on a grate preheater. The kiln
gases~~dbepassedthro~htfiecdlciningpartof
passed toa dust-collecting systemtogether
hadbeen

thegrateandthen

with the cooler air which

passed through the other part of the grate.

An alternative mathod of lowaring alkali circulation in

suspension preheater kilns which has been proposed (although not yet
used in practice) is to pass some or all of the kiln exit gases over a
cooled surface (ll9)

to produce a deposit of condensed alkalis and

high-i

dust.

Possible modifications to this process include the

use of~water-cooledtubes

or of bdies

system and a cooling/alkali remval

exchanged

plant.

betwzn

the kiln

8. CONCLUSIONS
From the review it is concluded that there is general agreement on the
factors affecting the volatilisation of alkalis. Volatilisation is
increased by-smaller nodule size, lower flux content in the clinker,
higher levels of chloride, lower levels of sulphur, higher temperatures,
longer heating periods and by higher contents of water vapqur in'the kiln
gases. It is not clear what the effects of other factors are (over the
ranges studied in the available literature), and it may be inferred that
these are relatively small: gas flow rate and composition, alkali
content, presence of fluoride, LSF and S/R and oxidising conditions in
the kiln. Several of these factors appear worthy of further study.
There is, nevertheless, insufficient quantitative evidence to predict the
degree of volatilisation of alkalis from a given raw mix under particular
conditions. The interplay of phase equilibrium and transport phenomena is
probably too complicated for it to be worth while attempting to simply
calculate volatilisations from knoun thermochemical data: more complex
modelling is needed.
In practical terms, the circulation of alkalis is affected as much by
their condensation and deposition as by their volatilisation. It would be
useful to devote some attention to this hitherto somewhat neglected
aspect of the phenomenon.

T. G. BURNHAM.

RRC/TGB/JLMC/D31:CPK/LEX32
11.03.87:12.02.R8.

1.

BUCK, E-C.:
Bibliography on the extraction of potash frm complex
mineral silicates, such as feldspar, leucite and glauconite (greeeand
marl).
Metallurgical and Chemical Engineering, 1918, 18, (11, 33-37 and (21,
90-95

2.

MEG, A.R.: The directheattreatmntof

cemntmill
dust to increase
its water-soluble potash content.
Journal of Industrial and mgineering Che&stry, 1918, 10, (21,
106-109

3.

STANKlN, T.E.: Expansion of concrete through reaction betwaen cemnt

and aggregate.
Proceedings of the Amarican Society of Civil wincers, 1940, 66-,
(121, 1781-1811

4.

MCCONNELL, D., MIELENZ, R.C., HOLUND, W.Y. and GREENE, K.T.:

&rent-aggregate reaction in concrete.
Journal of the American Concrete Institute, 1947, 19, (21, 93-128

5.

TRFMp&, B. and DREW, E.D.: California's concrete aggregate

resouxes.
Rock Products, 1953, -56, (11, 147-149, 200-201

6.

AMERIcANSCCIETYFoRTESTIKANDKU%UALS:
Portland cement.
ASIM c150-74

7.

LLCBER, F.W. and SPRUNG, S.: Causes andmechanism

aggregate reaction.
Beton, 1973, 2, (71, 303-306 and (81, 349-353

8.

EWEXRFF, P., IDDRN, G.M., KJAER, A., PLUM, N.M. and POUISEN, E.:
Chemical reactions involving aggregate.
Proceedings of the Fourth titernational Symposium on the Chemistry of
Cement, Washington, 1960. Vol. II, Paper VI-l, m-749-806

9.

ASSCC~ BRASIL~?ADE~RMAS
EspecificaCao.
AE%?T/CB-18
(19th January 1973)

10.

INSTITU'IO DE INVESTIGACION TECNCIKXA INDUSIXAL

TECNICAS: C&auto Portland Tipo I, Normal.
IIT= 334.009 &@ember 1971)

Specification for

of alkali-

TECNICAS: Cimntomrtland:

Y DE NORMAS

11. COMISION VENEZOLANA NOEUQS INDUSIRIALES: Especificaciones para

cwanto Portland,
CCCA: Ce 100 (1972)
12.

DIRECCIONGENERAL
lx% Cl-1955

DENORM?lS:

Norm Oficial para Cmanto

Portland.

13. DAVIS, C.E.S.: Studies in -t-aggregate

reaction. XVIII. The
effect of soda content and of cooling rate of Portland cement clinker
on its reaction with opal in mxtar.
AustraJian Journal of Applied Science, 1951, 2, (11, 123-131
14.

DAVIS, C.E.S.: Studies in cemant-aggregate reaction.: XXVI.

Caxpaxison of the effect of soda and potash on expansion.
Australian Journal of Applied Science, 1958, 2, (11, 52-62

15. DIAMMD, S.: A review of alkali-silica reaction and expansion

mechanisms. 1. Alkalies in cemsnts and in concrete pore solutions.
Cenwt and Concrete Research, 1975, 5, (41, 329-346
16. ~PQLLI'IT, H.W.W. and mCX@, A.W.: Tk distribution of alkalis in
Portland cement clinker.
Proceedings of the Fifth International Symposium on the Chemistry s?
Carwt, Tolqo, 1968. Part I, Vol. I, Paper I-126, pp. 322-333
17. MAIIOUXHEX, F.: An explanation oflq formation in cemant
Zment-Kalk-Gips, 1972, 2, (81, 395-396
18. MCCOY, W.J. and ESHEXUR, O.L.: Significance of total and
water-soluble alkali contents of Portland-t.
Journal of Materials, 1969, 2, (31, 684-694
19. CHDWZH, V.F., AZELITSKAYA, R.D., m. I.F. and
MMXIKIN, Yu.1.: Effect of alkalis on mineralisation and hydration
processes of calcium silicates.
Tsentant, 1963, (51, 7-9
20.

'JDROFW, N.A. and IxIBRovoL'SKIY,

K.A.: The effect of salts of the
mat&s sodium and potassium on the formation of clinker minerals.
Tserrwt, 1965, (11, 6-7

2l.

m, G,: Contribution to the alkali problem in suspension

preheater kilns.
Zemnt-Kalk-Gips, 1962, l5-, (51, 197-204

22, WEBER, P.: Heat transfer in rotary kilns with due regard to cyclic
processes and phase formation.
Wiesbaden, Rauverlag, 1963
23.

BPJITBEL, H.: Service conditions and wear of mgnesite

materials inrotaqcemntkilns.
Zemant-Kalk-Gips, 1976, 2, (71, 308-312

lining

24. ROCITSCHKA,
G. and MAJDIC, A.: Refractories for the cenent industry
a review.
Zmant-Kalk-Gips, 1974, 27, (101, 469-485
25. -G, W., OPITZ, D. and SIMMMANN,
J.: The hsring
doltits bricks in cemntkilns.
World Cement Technology; 1977, 8, (21, 39-46

of burnt

26.

TRXER, F.: Deterioration of cement rotary kiln linings by alkali

sulphides and sulphates.
Am&can Ceramic Society Bulletin, 1968, 47, (71, 630-636 (including a
discussion note by W.S. Treffner)

27.

LCXZHER, F.W., SPRW, S. and OPITZ, D.: Reactions associated with

kiln gases. Cyclic processes of volatile substances, coatings,
removal of rings.
VDZ Congress oh Process Technology of Cement Manufacture, Diisseldorf,
1971

28.

RITZMANN, H.: Recirculation problem in rota&

Zement-Kalk-Gips, 1971, 24, (81, 338-343

29.

-, H.R.: Wet or dry process kiln for your new installation.

Reek Products, 1974, 77, (51, 92-98, 100

30.

DANckJsKI, W. and STROBEL, U.: Investigations of alkali load tolerance

in dry-process kiln plants.
Zexnt-Kalk-Gips, 1976, & (101, 458-462

31.

NORTH, F.J.: Limastones, their origins, distribution and uses.

London, Murby, 1930, p.439
(quoting CLARKE, F.W.: Data of geochemistry. US Geological Survey
Bulletin, 1924, No.770, p.564)

kiln systems.

32. RUNDLE, L.M.: A canbustion m&hod for the determination of total

sulphur in limestones.
The Analyst, 1974, 99, (March), 163-165
33.

General Portland's use of aragonite at Tampa plant.

Pit and Quarry, 1973, 65, (91, 60-64

34.

PAUGR, K. and ENLESS, 0.: A study of alkali volatilization in

Portlandcemntrawmixes.
Portland Cm-eat Association report M-127, 1952

35.

DUDA, W.H.: Cmtdatabmk:

the cenmtindustry.
Wiesbaden, Bauverlag, 1976.

36.

WEAVER, C.E. and POW, L.D.:

Amterdam, Elsevier, 1973

international process engineering in

The chemistry of clay minerals.

37. CEtOSSLEy,
H.E.: A special study of ash and clinker in industry.
Paper 4: external boiler deposits.
Journal of the Institute of Fuel, 1952, 25, (Sept.), 221-225
38.

FURLONG, L.E., EFERCN, E., VERNON, L.W., WILSON, E.L.: The Exxon
donor solvent process.
Chemical Engineering Progress, 1976, 72, (81, 69-75

39. KEYSSNER, E.: The occurrence of iodine in cemnt

Chemische-Zeitung, 1925, 2, 821
.

flue dust.

40.

PERRY, R.H. and CHILITON, C.H.: Chemical engineer's handbook.

New York, McGraw-Hill, 5th edition, 1973.

41. CINES,-M-R., HAS=, D.M. and HOUSER, C.G.: ti1ecula.r

removing H2.S frannatural gas.
Chemical Engineering Progress, 1976, 72, (81, 89-93

sieves for

42. f4t.zCORD, A.T., WAGNER, L.E. and REESE, T.J.: Process for rtaking
alkalicemautclinker.
Us Patent 4,001,031 (4th January 19771, assigned to Chem+'rol
Pollution -ices, Inc.

low

43.

PORTER, E.S. and ToI;ER, H.J.: Remving alkalies fran raw materials.
Portland Carwt Association report MP-99, 1961

44.

KAYE, G.W.C. and LABY, T.H.: Tables of physical and chemical

constants.
Imdon, Ion-, 14th edition, 1973

45.

SEE, J-B.: Fluorspar and fluorine cmpounds in high-terqerature

smelting and refining of matals,
Minerals Science and Ezqineering, 1976, 2, (41, 217-241

46.

IWAKYID, N., SUITQ, M., HAM%%TSU, S. and SA'IOH,

of inorganic fluorides.
Transactions of the JWRI, 1973, 2, (21, 204-207

47.

J?CKSON, D.D. and M3RGAN, J.J.: Measurmsntof

Mporpressures
of
certainpotassiumcunpounds.
Journal of Industrial andEngineering Chexnistry,192l,~,
(21,
llO-ll8

48.

HALSTEAD, W.D.: Saturatedvapour pressure of @assiumsulphate.

Transactions of the Faraday Scciety, 1970, 66, (81, 1966-1973

49.

CUBICCICYITI,
D. and KECJESHEA, F.J.: Thermdynamics
sodium sulfate.
High Tmparature Science, 1972, 4, 32:40

50.

BA-, R-J., IEFEVER, R.A. and WILCOX, W.R.: Evaporation of sodium

chloridemalts.
Journal of Crystal Growth, 1971, 2, 317-323

51.

TmoR, L.: The vaporization tiermdynamics of fused lithium and

potassium fluorides.
Journal of Chemical vcs, 1976, 5, (81, 777-783

52.

FICAILIRA, P.J., UY, O.M., MUENW, D.W. and MARGRAVE, J.L.: Mass
spectramatric studies at high teqeratures:
XXIX, Thend.
decaqmsition and sublimation of alkali rmatal sulfates.
Journal of the Amarican Ceramic Society, 1968, 51, (101, 574-577

53.

HART, A-B., QGLNER, G.C., HAISTERD, W-D., IFXTON, J.W. and TIDY, D.:
Scme factors in seed recovery.
'
International SyqmsiumonMagnetchydrodynamic
Electrical Power
Generation, Paris, 1964, Session 7, Paper 89.

I.: Malting points

of vaporization of

54.

chemistryof
HART, A.B. and LAxroN, J.W.: Samaspectsofthe
m.h.d. seed.
Philosophical Transactions of the Royal Scciety of London, A, 1967,
261, 541-557.
.

55.

DRAPER, J.: Iaboratorystudyof

alkalimmmlin the cement burning
prOCSs - a progress report.
Portland Cement Association report MRS-68, 1954.

56.

DRA?TR, J.: Alkaliremval

in the cement burning process -a study of
operating data.
Portland Cemant Association report MRs-69, 1954.

57.

EWTIN, Z.B., FRIDMAN, I.A. and mKBICV, V.K.: Study of the

volatilisation of K20 using mthmatical
vimantal design
Tsemnt, 1966, (41, 7-9.

58.

GOFS,C andKEn, F.: The behaviour of alkalis in cmentburning

Tonindustrie-Zeitmg, 1960, 84, (81, 125-133.

59.

ANDERSON, E and NESTELL, R.J.: The volatilisation of potash fran

cenentmterials.
Journal of Industrial and Engineering Chemistry, 1917, 2, (31,
253-261. _

60.

T0ROK)V, NA. and cOBROVOL'SKIY, K.A.: Volatilisation of sodium and

potassiumoxides during clinker burning andtheir distribution in the
minerals.
Tsement, 1965, (31, 6-7.

61.

GRAZHDANSKIY, S.A., KINSTLW,

reducing the alkalicontentof
Tsemnt, 1965, (31, 18-19.

62.

J?CKSCN, D.D. and MlRGAN, J.J.: An application of the vapor pressures

of potassium cmpounds to the study of the recovery of potash by
volatilisation.
Journal of Industrial and E&nearing
Chemistry, 1921, 2; (41,
292-295.

63.

KOIX, H.: Remving alkalis by heating with admixtures.

Rock Products, 1942, 5, (21, 66-68.
(a reprint of FCA report M-74, 1941.1

64.

LMWUR, H. M&.: ~ethodof reducing alkali content in Portland

cerrPntclinker.
Portland Cement Association report M-74, 1941.

65.

TXWXJR, H. McC., MCMASTER, E.T. andJACQUES,

alkali Portlandcemntclinker.
Rook Products, 1943, 46, (71, 59, 67, 68, 74.

and ABUIA, S.Ya.: m of

clinker.

K.M.

W.: Manufacturing low

66.

CARLSEN,
H.: The behaviour of alkalis in cement raw materials during
the burning process.
Rock Products Annual Cement Industry Operations Sem,inar;1965.
(reprinted in Rock Products, 1966, 2, (51, 87, 88,' 157)

67.

LOVELAND, R.A.: Reduction of alkalies in cement kilns (study by

questionnaire)
Portland Cement Association report MRS-61, 1948.

68.

HOLDEN,
E.R.: Reduction of alkalies in Portland cement - use of
calcium chloride
Industrial and Engineering Chemistry, 1950, 42, (21, 337-341.

69.

GILLILAND, J.L.: Removal of alkalies by use of hydrochloric acid.

Symposium on alkali removal and problems, Milwaukee, Wis., 1959.
Portland Cement Association report M-158, 1960.

70.

WOODS, II.: Reduction of alkalies in cement manufacture.

Portland Cement Association report M-149, 1956.

71.

NDDEL'MAN, V.I. and UVAROVA, 1.T.: Method of reducing alkali contents

of clinkers from cement works in Central Asia.
Tsement, 1968, (41, 12-13.

72.

SPRUNG, S. and SEEBACH, H.M. von: Fluorine balance and fluorine

emission of cement kilns.
Zement-Kalk-Gips, 1968, z, (11, l-8.

73.

BANIT, F.G. and VASILIK, A.V.:

on exit gas dust removal.
Tsement, 1972, (61, 17-18.

74.

HATANO,H.: The behaviour of sulphur in the suspension preheater kiln

system.
Zement-Kalk-Gips, 1972, 25, (11, 18-19

75.

COLUSSI, I and LONGU, V.:

11 Cemento, 1974, 71, (21,

76.

Gypsum finds new role in easing sulfur shortage

REMIREZ, R.:
Chemical Engineering, 1968, 75, (241, 112-114.

77.

Production of cement and sulphuric acid from gypsum.

Cement and Lime Manufacture, 1968, 41, (41, 53-60.

78.

MERZ, A.R. and ROSS, W.H.: The nature of recombined potash in cement
mill dust.
Journal of Industrial and Engineering Chemistry, 1919, 2, (11, 39-45.

79.

BROWNMILLER, L.T. and EOGUE, R.H.: The sys tern CaO-NaZO-AlZO3.

Portland Cement Association Fellowship at the National Bureau of
Standards, Paper No. 25, 1932.

Influence of gypsum-bearing additives

Thermal decomposition of calcium sulphate.

75-98.

80.

BROWNMILLER, L.T.: A study of the system lime-potash-alumina.

Portland Cement Association Fellowship at the National Bureau of
Standards, Paper No. 30, March 1933.

81.

TAYLOR, W.C.: Phase equilibria studies on mixtures of the compounds

4CaO.Al2O3~Fe203-2CaO.Fe203-K2O.A12O3.
Portland Cement Association Fellowship at the National Bureau of
Standards, Paper No. 37, September 1938.

82.

TAYLOR, W.C.: The system 2.CaO.SiO2,K 20.CaO.Si02 and other phase

.equilibrium studies involving potash.
Portland Cement Association Fellowship at the National Bureau of
Standards, Paper No. 40, September 1941.

83.

TAYLOR, W.C.: Further phase-equilibrium studies involving the potash

compounds of Portland cement.
Portland Cement Association Fellowship at the National Bureau of
Standards, Paper No. 43, December 1942.

84.

EUBANK, W.R. and BOGDE, R.H.: Preliminary study on portions of the

systems Na20;CaO-A1203-Fe203 and Na20-CaO-Fe203-Si02.
Portland Cement Association Fellowship at the National Bureau of
Standards, Paper No. 50, March 1948.

85.

NEWKIRK, T.F.: Effect of SO3 on the alkali compounds of Portland

cement clinker.
Journal of Research of the National Bureau of Standards, 1951, 47,
(51, 349-356.

86.

NEWKIRK, T.F.: The alkali phases in Portland cement clinker.

Proceedings of the 3rd International Symposium on the Chemistry of
Cement, London, 1952, pp.151-171.

87.

ELLINGSON, O.A., GILL-AM, J.L. and KOPANDA, J.E.:

"water-soluble alkali" in cements.
ASTM Bulletin, 1956, (Feb), 63-66.

88.

SEGNIT, E.R.: Further data on the system Na20-CaO-Si02.

American Journal of Science, 1953, 251, 586-601.

89.

BA3ATSCHEV, G.N. and RADEVA, K.K.: Alkalis in Portland cement

clinker.
Silikattechnik, 1961, 12, (11, 33-35.

90.

BUDNIKOV, P.P., AZELITSKAYA, R.D. and LOKOT', A.A.: Effect of

additions of gypsum on mineral formation in alkali-containing cement
clinker.
Zhurnal Prikladnoi Khimii, 1968, (5), 953-957.

91.

AZELITSKAYA, R.D., PONOMAREV, I.F., BLONSKAYA, V.M., KARBYSHEV, M.G.,

LOKOT', A.A., and STEPANOV, V.M.: .The effect of gypsum on the phase
composition of alkali-containing clinker.
Tsement, 1969, (2).

Determination of

92,

LIx;ININA, I.G. and SHAPCSHNIKUVA, M.A.: Characteristics of clinker

mineral formation in thepresenceof alkalisulphates.
Tsemant, 1970, (81, 18-20.

93.

EUIT, Yu.M., KADS~SKII,

V-E., TURETSKII, A.M. and iANN, N.S.:
Effect of gypsum on the properties of an alkali-containing cmarit.
Tsemant, 1971, (41, 14-16.

94.

AHHASSI, G.: A contribution to the study of the influence of alkali

metals an the canposition and hydration of clinker phases.
Revue des Materiaux de Construction, 1974, (6911, 315-322.

95.

RITZMANN, H.: Hmtokeepalkalies

frcxnstealingprebeater
efficiency.
Rock Products, 1974, 77, (21, 6669.
(originally presented as a paper entitled "Raw maal preheater and
alkali problems" attheRcxkProducts
8th International Cemant
Industry Sminar, 1972).

96.

PCYITIZ, N.S. and CHEESMAN, R-D.: Effect of coal ash on the liberation
andnature of cmtantmi1lpotash.
Journal of Industrial and Engineering Chmistry 1918, 10, (21,
109-Ill.
(See also correspondence from E. Anderson and R.J. Nestell and fran
E.O. Rhodes and J.J. Porter in (121, 1030-1033).

97.

-, I. and PURI, A.: Cycle of volatile substances in rotary

kilns for Portlahdcemntclinker.
Z-t-Kalk-Gips,
1975, 25, (91, 377-379.

98.

IEHMANN, W.S. and PIASSMANN, E.: Determination of the alkali

circulation with the aid of the radio-isotope K42 in a long wat
process rotary kiln.
Zmant-Kalksips,
1957, 10, (31, 89-93.

99. WEBER, P.: Alkali problms and alkali elimination in heat ecoimnisiq
dry process kilns.
Zearant-Kalk-Gips,
1964, 17, (81, 335-344.
100. VlRlloRENKcrv,
V.I. and MLKONSKY,
preheaterkilns.
Tsemxit, 1965, (61, 12-14.

B.V.:

Alkali circulation in cyclone

101. RITZMANN, H.: The effect of dust cycles on the heat consuqtion
rotary kiln plants with raw maal preheaters.
Zemant-Kalk-Gips, 1971, 24, (121, 576-580.
102. SPRLm, s.: The chemical andmineralogical cmposition
dust.
Tanindustrie-Zeitung, 1966, 90, (lo), 441-449,

of cemnt

of

kiln

103. LITYNSKI, T. and GODEK, J.: K2C and CaO detennination in foreign and
Polish cemant dusts.
Zmant-Kalk-Gips, 1965, g, (101, 534-535.
104. HEIIbIANN, T.: Treatmantof dust frcmcemntkilns.
British patent 1,145,827, published 19th March 1969, assigned to
-F.L. Smidth & Co. A/S.
105. KESTER, BE.: Alkali reduction by kiln dust leaching.
Symposium on alkali remval and problems, Milwaukee, Wis., 1959.
Portland Cement Association report M-158, 1960.
106. STEVENS, H.A.: Wetrecoveryof kiln dustandsubsquentalkali
reduction.
Symposium on alkali removal. and problems, Milwaukee, Wis., 1959.
Portland &meat Association report M-158, 1960.
107. BADE, E.: Process for reducing the alkali cycle in clinker burning.
Zanent-Kalk-Gips, 1962, 15, (91, 403-408.
108. YURGANUV, N.N., SAFONOV, N.A. and BRODKINA, E-R,: Method for reducing
alkali circulation when returning dust to rotary kilns.
Tseiwnt, 1966, (11, 10-n.
109. WATSCN, D and ERW, A.W.: Treatment of waste products frcm Portland
cement manufacture.
US patent 4,001,030 (4th January 19771, assigned to Associated
PortlandCerrwt Manufacturers Limited.
experience andconsiderations
110. EGNN, Wand-HARD, U.: Cementwxks
relating to the design of bypass system.
Zerrwt-Kalk-Gips, 1972, 25, (61, 281-282.
111. HAEMms, P.: Alkali behaviour in a suspension preheater using a gas
by-pass - practical experience at the Rillito installation.
IEEE Cement Industry Technical Conference, Tucson, Ariz. 1976.
ll.2. SCHLUJ!ER, H.: Process for reducing the alkaliandchlorine
suspension preheater kilns.
Zerwt-Kalk-Gips, 1972, 25, (11, 20-22.

cycles in

The development and application of

113. CHRISTIANSEN, S. and MADSEN, G-M.:
precalciners for cement kilns.
Reek Products, 1975, 78, (51, 85-88, 126.
114. SMIMH, F.L. & CO. A/S: ~ethodof burning cement.
British patent 1,023,910, published 30th March, 1966.
115. lIED&DT,
H.: Measures for reducing the alkali cycle in the Le@
kiln.
VDZ Congress on Process Technology of Cmznt Manufacture, Dusseldorf,
1971, 164-167.

116. -T, E.A. and HEIAN, G.A.: Alkali remval

SyStem,
Rmk Products, 1973, 7fi, (51, 60-64, 179.

via grate - kiln

117. CLAUSEN, C.F.: U+mlkali cemmt fran high-efficiency kilns?

Rock Products, 1960, 63, (11, 148, 149, 152, 154, 164.
118. POLYSIUS GMEi: Process andapparatus for producing -toflow
alkalicontent.
British patent 874,818, published 10th August 1961.
119. VEIGEZ, J.F.: Problems of alkali reduction in the Htildt system.
Sympsosium on alkali remval and problem, Milwaukee, Wis., 1959.
Portland Cetrent Asscciation report M-158, 1960,

RRC/TGB/JMLC/b31
11.03.87.

TABLE 1
Melting andimiling
c-unds.

pints of selected calcium, potassiumand

Camp&d

CaCO3
--cam2

Mslting point
("Cl

Eoiling Point
("Cl

C!ecoqases 899 (calcite)

772 - 782

> 1600 - 2000

c*2

1386 - 1423 (45)

Ca(OHI2

deccmpses 580

Cd0

2570 - 2600

sodium

2500

2850 - 3000

1450

KP3

891

KC1

776 -

790

1407 - 1500

ET

846 -

858

1505

IUJH

360 -

380

1320 - 1322

KiQ
K2Si03

ckcqoses

ckarposes

350

976

K2Si$5

1015 * 10

K2s04

1069

*a2C03

851

dL3XiVpSeS

N&l

801

1413

NaF

988 - 997 (46)

1695

NaOH

318

1390

*aZo

sublines 1275

Na2=2o5

874

Na2SiO3

1088

Na4SiO4

1018

*a2=4
RRC/+IGB/JMLC/b31
11.03.87.

1689

890

;.

fi
,...I

t:.

!.:
L .

--.-

. . .
.:..

:
--..--+
:__:
i_

.:
..
!
---CL...
.:
8.

-:I

I
.:1.
,
: ..:I: ; I: :
2-..--

I-

[.

,
..!..Y

) I._!

-:

\-

:\.

L-i -Ii-t-r:

:. :
l----:L.L-.-i .- . : .

: .;. - ;I-!!,,f;;<

1:

1..

:-:

r.

: :: 1:

:..[:I.:
I :
-..I:

. . ..__
. -___. .. . . . . . .. ., . .
:

.-y--pq
,:
.:

- _ _ .
, _
.
.::y:: I I. ,, 1-1 J--+y-p~
. . _. _. . : .
:.

:-

~-.~:--::j..-~:~:

_ ,-.I
:

7: :I20 vol2tilisation (Log

-scel.

- .- -/__ -.. :.---/ : F&pure 3

---_,-.-i
i
: ..- i----_-e
i--q. -_. ,
I
1-. _-!------i --.-.._. . .._
! .-- LLJ--L .q ..____ T:. ..i -Lp:;
j _
j ic;.
.c
;

t I .
TW\\.
.

3
7

p-1
(
.bh

--y-y--.,

i-i
I
4
i-. . .
I --.,-.I -.I ______
&- -i- _---t-~--f
.-..---i .-.1 .I .-1.... j __.. i ______ I-. _.__ /r---- ;- i- - --- --.---L
! : ____
i l..- i
:
I ---.\p.I
1
t
!
j.
I
I
j
I
./
mien ]
.\

: i /_i\i:-+-i-i-; - _- ..i.- -.*-:.&-- - c---. -.I.. . . i. ._ 1..--i _._. .___ i .____. i___._ i. . I T-77

_I

--

:--

.
:-/

---i

..j

Figure 4
3

[
:

I
..--..!--L--L..
i
.:
!

_-

:-...--,.:
!

._. . )
-l;-..- i.--.2-L;.: . ;-.
aii ;----;i : IL
- .-- i1. _ _ _ _.;
I

____

1..

! :
. .

i _

I
._.

.:.

. T -I

i.

:
;

. - : .y,I

: -::...;.
: : .. _.
-u
L. -2 .-L-L-.
8.
,..
i-

._

.-

: : i

:.:.:
.: /.-.-. 4
._-.- .---L .._.__ I.-.-.!.-

F&me
5
. .._~ _.,_ -. __._- j___..--_
_
.__-T.--.-. - .3 t7-------,
5;
: _-_1
:
._
_
.;
r9/
.____ _ ..____.. - -- ._-- -- --_. .
1
/
2 f- -7
3;
..i .--.. --...
;. : :__ _ _
*- I
;-- :
;--,-. c-,..---. -_-_i --.. -- _ _ - _-.. _._- _.
.__._.
._..
____
! --[i -.-i *-j -. -.- : i ;.. . :...
i . ..-) .._.
:
j
--.
/
I
_ __-_ &.__ __-----_ --.
---_-___---:-.+--;--;.
I-- 7 . i yy-y1
I
; __.. ;T j . j _ ;
.
8 .
A------.:-.-- .-_---.. --,-I--.- :_-__-___ 1-m ._- -.---+--i
:
T:
1
I
t
I

.--._--_--- -__/
._ __
- ..-- I

!
.. /
&,-A--;-::

_____
-LL :.
i
,;
:

---i..--.
.-!

;
~
:
_-. I
- -...--- -:.-.-...-.

1.

.I.:

:.
! - ----: - ._,.-- --.:..--.

;.:
1. ..
_-.-- w-1-.
!
.

50;

/
L
-_ .--

_,

--..-------

._.-:

4-Q

:
,
:
L
t

.
.
:_
.
.
.-..1...
--._...
I
:
--!
!.
:
.
~+:~.:z-i;,-~.7;~~;
:
.
:
:
_---- -z-._-_
. . - :
_
;f-T. i -..--.
.-:I
.:.
F -:--:-:
-_.-_. ,
:- ._i_ :::.~ I
.--;
e ---.- / -. . -L8
-.
i-.:---f
., . . .:,r l-r..
-,
-.
;
-.-i---.- -..-:
~1
:
:.

:
-.:
:

:
I.
/..

_.

.:
,

:
_..-

:.

:
.___

:.

._- ;

_ i

;
--

--

i
rL.-.+ ._.- .;.
-;.. :
:

:
--.: . ... ___._ i

!
L.---L.--i__
i

-.--.-

-..

----

.-. _

--._---_-___-
:- ..---~

- -_- --.

__-. -..--..-

Figure 6

~-y----;
_:. \ _. .:-I:.
_
!
.:
: :.
1::. j .:..i.
.I..
.
.
.-;~+L+L.+
; :, :, ..!::
I
. . . . L..::/.
.
: I :: : I I I -- -. ) . .
:.... .:: ,:: _.
----~~Y---y-

Blue Circle Cement

PROCESS ENGINEERING TRAINING

PROGRAM

Module 13
Section 6

A Study in the Volatile Cycles on HOPE

#2 Kiln

A series of seven tests have been conducted at Hope Works during which the
No. 2 kiln was operated at selected, controlled levels of kiln back end
NOx (equivalent to burning zone temperature) and oxygen in order to study
the effects on volatile cycles of these different regimes.

Each detailed

study period lasted at least twelve hours and samples were collected from
various points around the kiln and preheater system for analysis.

CONCLUSIONS

The test results indicate that the chemical proportions of the minor
constituent volatile components at Hope lead to the following cycle
levels in stage IV meal.

The typical fluoride cycle level is 1.2 to 1.4 times the input to
the system and does not change substantially as the kiln temperature
or atmosphere is modified, over the range studied.

The chloride cycle is 20 to 30 times the total chloride input and

increases slightly with increasing temperature, although this may be
due to improved kiln stability at higher temperatures. The chloride
cycle is not modified by vide changes to kiln atmosphere.

3.

The chloride cycle is a low temperature cycle and cannot be

significantly modified by variation of kiln conditions.

4.

Chloride in the cycle combines with potassium where available.

However,

at low temperatures - equivalent to NOx levels of below

1,000 ppm - some sodium is also involved.

5.

The total potassium cycle is approximately 3 times the input level,

however about two thirds of this is present in Stage

IV in conjunction

with chloride and so cannot be controlled except by incorporation of a

bleed sys tern.

6.

About one third of the potassium in Stage IV is either feed based on

its first passage through the preheater or derived from a potassium
sulphate

based

cycle.

This

cycle ratio varies from 1.05 (minimal

recycle) at 900 ppm kiln NOx level to 1 .25 a t 1 , 4 0 0 p p m

NOx (25%

This portion can be controlled by operation at low

r e c y c l e 1.

temperatures.

7.

The sodium cycle level varies between 1

.2 and 2.0 times that in the

rarnneal over the temperature range examined.

8.

At low temperatures some sodium becomes involved in the low

temperature chloride cycle and this boosts the sodium cycle by
about

9.

10%.

The majority of the sodium in the system is involved in a sodium

sulphate

based

cycle.

This is strongly temperature dependent, and

the rate of increase also appears

to be increasing with temperature.

Over the temperature range investigated the cycle rose from 1.35 times
t h e l e v e l i n t h e feed at low temperatures to 2.0 times at high X0x
levels.

10. The total sulphur cycle is temperature dependent and rises from 1

.5

to 2.0 times that of the input over the temperature range investigated.

11. The alkali sulphate cycle has already been summarised in Points 6 and
9 and makes up about 35% of the total sulphate in Stage IV material.
These cyclic levels can be seen to be lower than the total sulphur
cycle

levels.

12. The calcium sulphate recycle is strongly temperature dependent and

rises from about 2.5 times the level in the feed at lov NOx levels to
4 times at high (1,450 ppn)

NOx levels.

13. The calcium sulphate cycle is also increased by a move into a

reducing kiln atmosphere.

At about 1,200 ppm NOx the cycle

increased from 3.1 under oxidising conditions to 5.5 times the feed
level under reducing conditions.

14. For potassium and sodium there are indications of a slov but steady
increase in losses of these components from the preheater system
as temperatures increase.

This may become part of an external

cycle or may be lost to atmosphere.

This could be established by

a longer term study of the levels of these components in the

precipitator and stack dusts.

15. Sulphur is in overall balance within the system belov NOX levels
of 1,200 to 1,300 ppm, but above this level the loss increases
sharply with rising temperature.

Again this may become part of

an external cycle or may be lost from the systen,

hovever in this

exercise no precipitator or stock dust samples were collected so this

cannot be confirmed.

16. When the kiln a:mosphere moves into reducing conditions the losses
of alkalies and sulphur from the kiln system increases sharply.

In viev of these conclusions it is recommended that, at

1.

Hope:

Kiln back-end oxygen level is maintained at 2.0 to 2.52

(as measured).

2.

Kiln back-end NOx levels are maintained below 1200

ppm,

although

the most appropriate NOX level will depend on detailed raw mix
chemistry.

3.

Communication between kiln burner and personnel cleaning the

preheater be improved.

4.

During cleaning operations short term reductions in raw meal feed

vhich

avoid

the

inception

of

reducing

conditions

will

be

beneficial in the longer term.

NOTE :

Cycle =

quan:ity i n S:age I V - o n a l o s s f r e e b a s i s
quantity in feed on a loss free basis

F u e l i n p u t l e v e l s f o r a l k a l i e s , fluorides and sulphates have not been

taken into account as the levels are relatively

lou and constant, as

indicated from the small number of fuel samples that were analysed.
The chloride input from coal has been taken into account.

1.

INTRODIJCTIOfi
Since the development of the suspension preheater based dry process for
cement manufacture the study of the inherent cycling effects of the
potentially

volatile

components,

which

are present as minor constituents

of the raw materials and fuels, has become increasingly important.

In

the less thermally efficient plants a natural loss of a portion of these

volatile components occurred in the waste gases so automatically
controlling

the

recirculation

levels

level in the clinker product.

within

the

burning

system

and

the

Previously this loss of volatile

components has also been higher than would be found today due to the
capabilities
volatilised

of

the

available

components

zones of the kiln.

pollution

condense

on

control

fine

I n the suspension

dust

equipment,
particles

preheater,

with

as

the

in

the

cooler

its

greatly

increased surface contact between gas and particle and repeated

separation

of

gas

and

particle, the recovery and retention of

volatilised components will be almost complete.

This leads to

increased proportions of these components in clinker and to the

development of operational problems in the pyro-processing stage due to
the

quantities

through

of the potentially sticky components that build-up

cyclic

processes.

These effects limited the immediate impact

of the suspension preheater technology in the industry as a whole for a

number of years.

Nevertheless increasing fuel prices and the

requirement for even larger process units provided the incentive for
identifying the causes of and solving the problems associated with
volatile
targetted

component
at

cycles 1~3.

controlling

alkali

the recirculation levels wi:hin

kiln

In general this work has been

levels

such

in

clinker

boundaries

and/or

that

maintaining

allow

reasonable

operation, whilst the solutions have been limited to indirect

methods

of

control;

modification of the raw mix, change of fuel,

installation of a bleed system, manual or automatic cleaning of areas

where excessive bui Id-up occurs .

During the developaent phase of the high level control project at Hope
Works it became evident

volatile

cycle

were produced as the combustion chamber conditions were modified.

The

more

consistent

control

that
and

considerable
greater

changes

understanding

in
of

the

process

conditions

vithin the kiln that vas developed through the reliable measurement and
interpretation of kiln back end
conventional

instrumentation

NOx level in association vith the more

offers

the

potential

for

deliberate

modi-

fication and control of the volatile cycles by variation of process

conditions
of

on

troublesome

plant

scale.

volatile

cycles

This would then permit the minimisation

and

hence

lead

to

improved

kiln

stability

and operating times.

Consequently,

it was decided to investigate the exient of variations in

the volatile cycles vhich could develop by operating a kiln under a

number of temperature and gas compositional (oxidising/reducing)
Over the course of the last tvo years, as

operational

restraints

regimes.
and

manpower availability have permitted, a series of seven studies have

taken place on No. 2 kiln system at Hope Works.

In these studies three

different kiln back end NOx levels (800 to 1,400 ppm)

atmosphere

conditions

as shovn in Table 1 .
relationship

betveen

(reducing

to

high

excess

oxygen)

and three kiln

were

targetted

Previous vork 4j5y6 has shovn the general

kiln

hOx level and burning zone temperature.

Although it is not possible to

equate a particular NOx level vith an

actual definitive material or flame temperature at the fron: end of

the

k i l n , t h e t y p i c a l r e l a t i o n s h i p f o r N o . 2 k i l n a t H o p e i s shovn i n F i g . 1 .

For control purposes the advantages of NOx m e a s u r e m e n t a r e t h a t a b o v e a

base

level NOx generation is proportional to the temperature of the

burning zone - v i t h i n s p e c i f i c l i m i t a t i o n s , - t h e r e i s a v e r y f a s t
reaciion

time, the signal suffers relatively little process noise, and

the true reading

is

unaffected

by

front

end

dust

cycles.

2.

VOLATILZ

2.1

CYCLES

General

Considerations

7,8,9,10

The minor components that are generally considered to be involved

in

major

volatile

cycles

are

the

chloride,

alkali,

and

sulphur

s p e c i e s , with some attention also being paid to fluorides.

Prime

control of the volatile component cycles is performed in the

design stage of a Works project through the selection of rav
materials and fuels in order to optimise the relative and absolute
levels

of

the

potentially

volatile

components

the inclusion of a bleed system.

defined

the

volatiles

during

the

pyro-processing

and,

if

necessary,

Once these factors have been

will develop internal and external cycles

an internal cycle being

stages;

totally within the kiln and preheater system, vhilst an external

cycle will leave the system but be returned after a time lag (for
instance

with

the

precipitator

are kept steady, the cycles

equilibria

are

reached,

dust).

will

Where processing condition2

continue

to

develop uniil

at which time the total amounts of

volatile entering the system will be balanced by the quantities

leaving the system.

The degree of volatilisation

and the rates at which the

equilibrium are established will depend on:

(i)
(ii)
(iii)
(iv)
(VI

(vi)
(vii)
(viii)
(ix)
(xl

The

species, their

chemical

forms,

and

concentrations.

The volume of gases.

The intimacy of contact between gas and solid.
The vapour pressures of the salts.
Possibility of dissociation or further reaction.
Rate

of

diffusion

to

and

from

solid/gas

interfaces.

Degree of saturation of gas.

K i l n a:mosphere.
Kiln

temperatures.

Time/temperature profile of material within the kiln.

lost

of these factors are to some degree inter-related, and so in

norma 1 operation the only methods available to control the degree

of volatilisation

will be the kiln internal atmosphere and

temperature.
The major cycles will be internal cycles between the burning zone
and preheater where an individual cycle time of twenty to thirty
minutes would be expected.

Smaller

quantities

of

volatile

components may be involved in the longer term external cycle which

develops
raw

via

the

precipitator

dusts

and

volatile

collection

in

the

These would be expected to have much longer cycle

mill.

times of up to twenty four hou:s.

2.2

Fluorides

l1

Fluorides can be found naturally in raw materials or be

deliberately added in small quantities to the raw mix.
proportions
burning

its

mineralising

process.

action

has

beneficial

In low

effect

on

the

Generally it has a low volatility and causes

few operational problems, although a level of above 0.25Z

in

clinker may lead to setting problems particularly in winter.

However,

cases have been cited vhere a hard dense build-up has

developed

in

preheaters,

where the build-up contains high (over II)

proportions of fluoride.

2.3

Chlorides

l2

Chlorides are derived from the raw ma:erials

and the kiln fuel.

Either source will only have very small quantities of inherent

chloride,
with

the

but
high

the

high

volatilities

collection

efficiency

of
of

these
the

compounds
cyclone

together

preheater

systems will lead to the development of a greatly enhanced cycle.

The chlorides have a high affinity for the alkalies in general and
potassium

in

particular.

This property - together with the high

volatility
K20

- has been used in wet process kilns to control clinker

levels by addition of CaC12 to the rav mix o: fuel, which

leads to loss of KC1 with the exhaust gas from the kiln. In the
suspension

preheater,

however,

the

volatilised

material

is

recaptured within the system unless a bleed is utilised between

the

kiln

and

riser duct.

It is generally considered that no more

than 3X of the chloride passing from the preheater to the kiln

vi11 l e a v e the system with the clinker.

Although

considerably

higher levels have been noted in individual samples of clinker

this is probably due to a push of kiln feed, or a semi flush
situation as thermodynamic consideration indicate that no chloride
should pass through the burning zone.

On many SP kiln systems

some degree of preheater cleaning is necessary on a regular basis

and this may help to control the chloride cycle by forcing the
kiln conditions into a situation which permits a brief increase in
clinker chloride level (i .e s reduced material temperature,
increased

material

loading

and

flux

level).

Small amounts of chlorides will also leave the preheater system

with the waste gas stream.

Results of detailed balances at Rope

and Plymstock suggest that the preheater loss can be of a similar

order to that in clinker.

Taking the total loss of chloride from

the system as between 2 and 5% of the feed to the burning zone it would then be expected that a circulating load of 2 0 to 5 0
times

the rota1

2 . 4 A l k a l i e s 13p

chloride input vould develop.

l4

The major source of alkalies will be the raw

mix;

notably the

clay component although minor quantities can arise from :he fuels.
The initial free alkalies will behave in one of three ways:

1.

Remain in the material being processed and become

incorporated
formed .

2.

3e

in

the

clinker

constituents

This happens to Na20

converted

carbonates,

into

different

hydroxides - by

that

are

being

to a greater degree than K20.

compounds

chlorides,

reaction

with

other

sulphates,

constituents

of the raw mix.

3.

Diffuse to the surface of the process material and volatilise.

I n i t s ini:ial s t a t e K20 b e g i n s t o v o l a t i l i s e o v e r a w i d e r a n g e o f
temperature, depending on the form of clay in which it was
incorporated,

but irrespective of source it would be expected to

have volatilised almost completely at burning zone temperatures

although some may have been at least partially stabilised by
conversion to the less volatile
bed.

Once

volatilised

it

will

sulphate form within the material

react

sulphates at the rear of the kiln.

dust particles.

to

form

chlorides

and

These will then deposit on

I n i t i a l l y , Na20 i s l e s s v o l a t i l e t h a n K20 d u e t o

its higher bond energy and so a greater proportion of the kiln

feed Na20

would be expected to pass through the burning zone in

clinker without participating in the volatile cycles.

Na20

will react with SO2 and SO3

to form sulphates towards the

r e a r o f t h e k i l n , and with chloride vhere

i n e x c e s s o f K20.
and sulphate,

Volatilised

this species is present

Where alkali is present in excess of chloride

alkali carbonates will be formed.

Each of these

alkali compounds will deposit in a liquid state on the surface of

dust particles in the cooler zones of the kiln and lower preheater
stages and will enter the volatile cycles as the dust is separated
out in the cyclones.

Direct contact onto kiln or preheater

The compounds

surfaces may lead to the development of build-up.

will

then

re-enter

the

kiln

where

the

degree of

volatilisation

will depend on the species and the kiln conditions.

Volatility

decreases

from

chloride to carbonate to sulphate, and hence

sulphates are more likely. to pass through the burning zone.

Nevertheless the likely range o f burning zone temperatures cover
the

thermal

increase

area

in

significantly

which

alkali

volatilities

are

with

rising

temperature.

likely

to

2 . 5 S u l p h u r 15s16~17

Sulphur can e n t e r the system in a number of forms from either fuels

or raw materials.

A limited amount may evaporate in the upper

preheater stages and escape from the system in the exhaust gases.
In general SO2 and SO3 can form in the high temperature areas and
be transferred to the gas phase. In the cooler areas of the kiln
back end and preheater system, sulphates will form and re-enter
the material stream.

Preferentially

alkali

sulphates

will

be

produced with excess sulphate combining with free lime or calcium

carbonate that is available in these areas.

khere r e d u c i n g

conditions exist the equilibrium will favour the existence of SO2

a n d SO3,

so

restricting

the

formation

of

sulphates.

In this

situation loss of sulphur oxides by way of the stack may increase,

but where the gas stream passes through the raw mill the majority
of

the sulphur oxides would be expected to react with the high

active

surface calcium compounds which are produced in the milling

This will :hen return to the kiln in the raw mix as part

process.

of an external cycle.

The high boiling points of the alkali sulphates would indicate

that

relatively

Bowever,

low

dissociation

levels
may

of

volatilisation

occur, particularly

would
under

be

expected.

the

reducing

conditions which can exist to some degree within the burning zone.
Calcium sulphate also has a high boiling point, but is even more
susceptible

to

dissociation,

so

higher

from this compound would be expected.

recirculation

A s CaSC&

of

sulphate

cannot recycle as

a compound the lime from this compound remains in clinker as free

lime

- making burning more difficult - whilst the SO3 is carried

in the gas stream CO

the kiln back end where it reacts to form

a l k a l i o r c a l c i u m sulphate.

3.

PROCEDURZS

The intention was to operate the kiln under controlled back end oxygen
and kiln temperature regimes, w i t h k i l n backend NOx l e v e l b e i n g u s e d a s
an indication of change in temperature.
were 800, 1,200 and 1,400 ppm.

The

target

&OX

levels

For each of these temperature conditions

two back end oxygen levels were targetted;

(norma 1 oxygen 1 .

selected

1.8% (moderate) and 2.5%

At Hope Works there is a positive offset of 0.6 to

0.8% for the measured value of kiln back end oxygen as compared to the
real value due to the type of sampling system.

In addition to these

s i x t e s t s , a further run was conducted at 1,200 ppn~ NOx with CO evident

at the kiln back end - reducing conditions in the burning zone (Table 1).
Each test run lasted at least twelve hours, over which period samples of
preheater feed, kiln feed (ex Stage IV cyclone) and clinker were taken
at hourly intervals.

Coal samples were also taken periodically. Each

sample vas analysed for major oxides, fluoride, chloride, tra20,

SO3 using an XRF analyser.

K20 and

Ultimate and calorific value analyses were

also performed on :he c o a l s a m p l e s .

Throughout each test period and, in most cases, for a period before the
test began, kiln

operating

conditions

were

monitored.

Aver age kiln

conditions, and the range of conditions for each test are detailed in
T a b l e 2 . Tne results of chemical analyses of Stage IV material are
presented graphically for each test to show the variation with time
together with selec:ed

process data in Appendix I.

During each test a number of plant measurements

and material temperatures

- air

flovs, shell, gas

- vere completed, and a kiln operation log

maintained in order to build up sufficient information for heat and mass

balances.
of

the

This raw data is summarised

alkali

cycle

studies

this

in Appendix 2, but in the context

information

is

relatively

meaningless

as the dominant factor in the overall heat balance vas the raw meal feed
rate

vhich varied significantly from test to test, notably as a function

of whether the raw mill was in operation.

period

Additionaliy,

over the time

covered, modifications were made to the top stage cyclones which

changed the external dust cycle for the system.

4.

EX?ERI.%ESTAL
4.1

F!XSULTS

Fluorides

Average fluoride levels on a loss free basis for each test are
given in Table 3 together with the ratios of Stage 4 material and
clinker fluoride levels to the fluoride level of the preheater
feed

material.

The level of fluoride in Stage

is generally

IV

1.15 to 1.45 times that in the preheater feed indicating low

levels of recirculation.

At such low levels of recirculation and

initial input it is difficult to be definite about trends, but

plots of this data - Figure 2 - i n d i c a t e i n i t i a l l y a s l i g h t
decrease

in

average

recirculation

with

increasing

temperature,

folloved by a climb back to the original level. This pattern has

been seen before in experimental burning of mineralised
on the dry process.
volatilisation

is

Experience

more

here

dependent

on

suggested
the

that

potential

the mix than on material temperatures (Table 4).

clinkers

fluoride

flux

level

in

It is known that

an increase in flux level vi11 reduce component volatilities due

to changes in diffusion rate and reduced surface area, and so it
is

possible that in the longer term over the lower temperature

range the effect of increasing flux level vith temperature has a

greater

negative

temperature

effect

than

considerations

the

for

straight

this

forvard

volatility/

component.

At higher temperatures the recirculation rises back up to its

previous

level, as

temperature

effects

take

precedence.

These

trends are confirmed by the way in vhich fluoride levels in

clinker and preheater feed are similar at low temperatures, but
clinker levels lower than those of the preheater feed at higher
temperatures.
X0x

Plots of Stage IV material fluoride level against

level for each individual test (Figure 3) also suggest a

slight

increase

time s c a l e .

in

recirculation

with

temperature

over

this

short

There

is

no

indication

of

any

change

in

fluoride

recirculation

level being brought about by variation of the kiln back end oxygen
level.

These observations cannot cover the situation where there is an

excess

of

alkalies

over

sulphate.

In

this

instance

alkali

fluorides can form from CaF2 and so a different trend would be

anticipated.

4.2 Chlorides

Initial comparison of the proportions of chloride in the Stage IV

material

with

X0x

level for individual tests suggests a stable or

slightly falling level of chloride in Stage IV with increasing

temperature (Figure 4) .

However,

higher

X0x

levels

were

generally

associated with higher raw meal input levels and a lower coal to
clinker ratio over these test periods.

This gives a lower total

chloride input to the system and so the lower Stage IV levels would
seem to reflect the lower input quantity.

AS

chloride is volatile

at relatively low temperatures it can evaporate before the burning

tone,

and so has a short cycle time.

Figur? 5 compares the test

average da:a for the ratio of Stage IV to total input chloride

quantities and shows a slowly increasing chloride cycle vith
temperature,

but

this

cycle

is

basically

ratio falling between 20 and 30.

at the higher NOx

large

cycle

with

this

There are also indications that

levels the chloride cycle had not reached

equilibrium as Stage IV chloride levels were still rising with

t i m e a t r e l a t i v e l y s t e a d y NOx l e v e l s .

During the time period

in

which these tests took place the Works

were still maintaining a regular preheater cleaning campaign to

control

preheater

build

up, and

the

regular

seni-flush

situations

vhich this incorporates may have helped to control the S:age

chloride

levels

in

the

lower

half

of

the

predicted

range.

IV

Certainly it has been noted at Elope t h a t o c c a s i o n a l l y c l i n k e r

s a m p l e s c o n t a i n h i g h (+ 0.1X) l e v e l s o f c h l o r i d e .

Furthermore,

t h e h i g h e s t t e s t a v e r a g e c l i n k e r c h l o r i d e l e v e l (0.162) vas n o t e d
i n T e s t 1 , during vhich period the kiln was unstable.
This is no evidence that modification of the kiln atmosphere by
moving into reducing conditions causes any change in chloride cycle
as the Stage IV to total input chloride ratio for Test 5 is very
similar to those from Tests 1 and 2 which had similar NOX levels.

4 -3 Potassium

The potassium (K20) cycle measured in the Stage IV cyclone during

these tests vas betueen 3.5 and 6.5 times the input level from the
rav meal.

A S the proportion in coal is relatively small and vi11

not vary very much it has been ignored in these studies in order to
simplify the exercise.

Figure 6a presents average data for each

test for the ratio of K20 in Stage IV material to that in the

preheater
oxygen

feed,

plotted

against

average

NOx

level.

For

the

normal

tests, this K20 ratio can be seen to be increasing (from

3.85 to 4.4) as NOx rises from 900 to 1,400 ppm.

For the moderate

o x y g e n t e s t s t h e p i c t u r e i s l e s s c l e a r v i r h t h e r a t i o s f o r t h e lov
(Test 3) and high NOx (Test 6) :ests

being similar vith a sharp dip

i n T e s t 1 , at the medium NQx l e v e l .

It has previously been

suggested that K20 vi11 pre ferentially

react

lov

temperature

cycle.

vith

chloride

to

form

It is also noticeable that the test

which does not fit in with the general picture - T e s t 3

- also has

a very high chloride level in Stage IV.

the

preferential reaction of

K20

Assuming

that

in Stage IV is vith all the available

chloride it can be seen that 65 to 752 of the potassium in Stage IV

will be bound to chloride.

This cycle is a low temperature cycle

and can only be controlled by use of a bleed system.

K20 vhich has not reacted with chloride will nov react to form
sulphates.

Table 6 also includes the ratios of K20 a s s u l p h a t e i n

Stage IV to K20 in the preheater feed.

In this case the level of

K20 as sulphate in Stage IV rises from 1.05 times the K20

level in

the preheater feed at low #Ox levels to 1.24 times the preheater
feed level at high NOx levels.

At both moderate and normal oxygen

levels this ratio increases as kiln NOX level rises. At the lower
NOx levels (850

to 900 ppm)

the indications are that potassium

sulphate passes through the kiln without becoming involved in an

internal cycle.

In fact at these temperatures, it is likely that

some Na20 becomes involved in the chloride cycle, as insufficient

potassium was measured in Stage IV to totally account for the
measured chloride level.
and the trend shovs

These ratios are plotted in Figure 6b,

up clearly.

conditions test (tes:

The K2S04 ratio for the reducing

5) is slightly higher than those of the other

tests at 1,200 X0x level, but this increase is too slight to be

convincing proof of an increased cycle under reducing conditions.

Figure 6c plots the average ratios for each test between the K20
levels in clinker and preheater feed.

This clearly shows that at

low NOx levels the system is in equilibrium in terms of K20,

however as the NOx level rises the output level of K20 falls belo=
that of the input.
is approaching 202.

By an EiOx level of 1,400 ppm this inbalance

Bowever, with the onset of reducing

conditions this inbalance approaches 50f.

There art two possible

explanations for such an offset betveen output and input levels.

Firstly,

it is possible that a build up of concentrated K2SO4

KCL is developing;

and

this material sets on the surfaces vi:hin the

preheater and so prevents the material re-entering the cycle.

Secondly, there is a loss of K20 from the system.

This may pass

to atmosphere through the stack, or be collec:ed within the raw

mill and precipitators and be returned to system as part of a long
term external cycle.

4.4

Sodium

Test average data for Xa20

levels

- on a loss free basis - around

the kiln system are presented in Table 7, showing that the Na20
level in Stage IV is between 1.6 and 2 times that in the preheater
feed.

Figure 7a plots the ratio of Stage IV Na20 to preheater

feed Na20 against NOx level.

This

suggests

that

initially

r a t i o f a l l s s l i g h t l y a s NOx r i s e s b e f o r e c l i m b i n g a g a i n .
previous

the
In the

section it was suggested that at low NOx levels some Na20

reacts with the chlorides so becoming involved in the low

temperature
Ka20

cycle.

Whilst only low proportions of the available

- p o s s i b l e 1 0 % - become involved in this way, it can still

s i g n i f i c a n t l y i n c r e a s e t h e t o t a l Na20
r a t i o o f Na2S04

cycle.

I n F i g u r e 7b t h e

i n S t a g ? I V - as calculated from a chloride,

potassium and sulphate balance - to preheater feed Na20 is

plotted.
as

NOx

This shows an increase in ihe cycle derived from

increases.

The data suggests that the Stage IV Na20

Na2SO4
level

- as SO4 - rises from about 1.35 times the level in the preheater
f e e d a t l o w NOx t o a b o u t 2 . 0 t i m e s a t t h e h i g h NOx l e v e l . T h e
results of moderate and normal oxygen level tests suggest that the
recycle can be reduced slightly by operation at higher oxygen
leve 1s , although this is not born out by the low oxygen test.
AS

with K20, clinker output and preheater input Na20

Salanced

a t l o w NOx l e v e l s , b u t a s t h e NOx l e v e l r i s e s a b o v e 1 , 3 0 0

the level in clinker falls quickly;

being betveen 20 and 30% lower

than the input level at 1,400 ppm NOx.

certain

temperature

l e v e l Na2SO4

temperature dependant.
Ira2SO4,

levels are

As Na20

This suggests that above a

recirculation levels are strongly

is almost completely present as

t h i s m a k e s i t p o s s i b l e to control and in some cases signifi-

c a n t l y r e d u c e t h e fia20 r e c y c l e b y conirol

of kiln conditions.

4.5 Sulphur

The level of sulphur - calculated as SO3 - measured in Stage IV

material and in clinker was very variable during the course of
each test as the levels react strongly to changes in both kiln
temperature and

atmosphere.

However,

average values for the

samples taken around the kiln system for each test

T a b l e 8 - show a more consistent picture.

- given in

The ratio between

sulphur as SO3 in Stage IV and in raw meal is plotted against kiln

NOx level in Figure 8a and shows that over the temperature range
investigated the total sulphur cycle rises from 1.6 to 2.6 tonnes
the sulphur input from the raw meal, rising as the NOx level
rises.

There is also a substantial increase in cycle once

reducing conditions are encountered - from 2.0 to 3.0 at

1,200 ppm NOx.

In the previous two sections the alkali sulphate cycles have been
discussed.

A S these cycles increase only slowly with temperature,

these effects have been removed from the sulphate cycle by

calculating the amount of sulphur likely to be present as alkali
in raw meal and Stage IV and subtracting these from the total
quantities.

The remaining figure can be considered to be the

sulphur which takes part in a calcium sulphate derived cycle; at

ilope this makes up approximately two thirds of the sulphur in
Stage IV.

In this case the ratio rises from about 2.5 at low NOx

levels to around 4 at an NOx level of 1,450 as decomposition of

CaSO4
moves

increases.
into

This cycle also increases quickly as the kiln

reducing

conditions; from about 3.1 to 5.5 under the

test conditions.

As

in the previous two sections at low NOx levels the sulphur

output

in

clinker is approximately in equilibrium with that in the

raw meal, whiist a s NOx rises above 1 ,300 ppm clinker sulphur
levels are lower than the inputs.

Study of the data from

individual tests shows that after a sharp rise in temperature the

sulphur

cycle

initially

hours to stabilise.
at

higher

responds

very

quickly

but

takes

several

This suggests that if the kiln is held stable

temperatures

the

recycle

ratio

is

likely

to

rise

even

above the levels measured to date, and an external cycle will

develop. In reducing conditions the clinker to raw meal sulphur
ratio

also

drops

substantially,

from equilibrium to 0.55 when

operating at a 1,200 ppm NOx level at the kiln back end.

5.

GZERAL

DISCUSSION

The results indicate that with improved control of the kiln and
operation at lower temperatures and under steady oxidising conditions
it would be possible to reduce the alkali levels in Stage IV by a
moderate amount, and the sulphate level significantly.

Over the

temperature range investigated the K20 level in Stage IV material was

reduced by about 10% at the lowest temperatures, whilst Na20 and SO3
levels in Stage IV material both fell by over 302. It must also be
recalled that the highest kiln back end NOx level that was targetted
was 1,400 ppm.

Shortly before this test work began the kiln was

commonly run at an NOx level of about 1,800 ppm, and when an NOx
monitor was first installed

NOx levels of about 2,500 ppm

were common.

At that latter NOx value SO3 levels of up to 7.0% were recorded as

compared to the figures of 2.4 to 2.7 obtained during the low
temperature test work.

A similar effect on Stage IV SO3 level was

observed during the new Oxford Works Simulation Trials at Plymstock in

1980 lg,

as shown in Figure 9, although in this case extra iron was

added to the raw mix to permit the large reduction in burning zone
temperature. Overall the general trends and size of cycle for each
component fit in with theoretical studies for the vapour pressure v.
temperature relationships (Figure 10).

During the early days of high level control test work it was noted that
raising the measured kiln back end oxygen level from its previous normal
level of about 1.5% to around 2.3 to 2.5% resulted in considerably more
stable kiln operation.

This was presumably a two tier effect, as a

normal level of 1.5% gave the potential for the kiln to dip into
reducing conditions when any disturbance occurred.

The first effect

of this would be to modify the flame and hence temperatures around the
system.

Secondly the reducing conditions vould strip volatiles -

notably sulphate - from the system, so upsetting the equilibrium and

modifying the quantities of low temperature flux and increasing the
likelihood of high levels of build up around the kiln system. This

e f f e c t o f o x y g e n o n s u l p h u r l e v e l s in clinker has also been noted at

Northfleet Works 2o d u r i n g t e s t i n g o f an SO2 continuous measurement
probe,

and at Westbury.

deliberately
the

raised

consistency

in

of

In the first case oxygen levels were

order

cement

to

stabilise

clinker

SO3

levels

to

improve

quality, whilst in the latter the oxygen

level was reduced in order to increase the sulphate and alkali cycles,
so

allowing

the

Stabilization

control

and

of

reduction

alkali
of

levels

the

on

volatile

these

wet

process

cycles

will

lead

kilns.

to

more

c o n s i s t e n t c l i n k e r q u a l i t y , and there will be a lower and more

consistent heat sink on the front end of the kiln resulting from the
e v a p o r a t i v e l o a d 21 3 22.

this may be partly offset by the

However,

reduction in low temperature flux level at the rear of the kiln.

Further study would be necessary

In

the

past

most

methods

for

to

the

investigate

prediction

of

this.

volatile

recirculation

have been based on the oxide input levels, without taking real account
of their inter-dependancy and the forms in which the recirculation
develops.

The results of this exercise clearly show the need for a

more detailed breakdown into the components of the volatile cycles.

kiiilst
The

early studies by Research Division did take this approach, 23.

installation

increased
cycles .

kiln

of

Linkman high

stability

level

control

systems

will

give

and permit improved studies of volatile

A further exercise on a dry process site with excess alkali

would be of benefit in expa riding t h i s s t u d y t o c o v e r t h e compl?te

of

possible

volatile

components.

range

6.

PR..C?IC.AL

I.QLICATIOWS FOR ROPE WORKS

Tne da:a shows that the major significant cycle at 'dope is calcium
sulphate based.

This cycle increases markedly with increasing

temperature and also increases greatly with the on-set of reducing

conditions.

Consequently it can be minimised by operation at lov kiln

temperatures and in an oxidising environment.

It is not, however,

simply enough to avoid CO at the kiln back-end, as significant reducing

conditions can occur within the burning zone before an obvious increase
in CO is noted at the kiln back-end.

Previous work has shovn that a

water wash sampling system tends to give an offset in oxygen analysis

and that this offset will depend on the water flow rate through the
sampling system.

In general during the test periods this offset fell

betveen 0.6 and 0.8;

a kiln back-end oxygen reading of 2.5% vas

equivalent to a true reading of 1.9% (offset 0.6).

In order to

minimise recycle it is advisable to maintain the kiln back-end oxygen

reading - from the existing kiln instrumentation - between 2.0 and
2.5%.

Kiln front end temperatures should be at the lowest level

compatible with steady kiln operation and will depend on the rav mix
chemistry and condition of the firing system, but is likely to
generally fall in the back-end

NOx range

of 800 to 1000ppm.

It is significant that the periods when kiln operation is most likely

to fall outside these ranges are during the routine cleaning
operations, vhich is also the time when the concentrations of volatile
components within the kiln are likely to be the highest. Extra
attention to kiln operation during these periods should result in
minimisation of volatile retention, vhilst lack of attention is likely
to cause the majority of the potentially volatile components cleaned
from the preheater to be driven back and redeposited within the system.
In order to fully control kiln operation at these times it is necessary
that the kiln burner be notified before cleaning commences.

7.

COSCLCSIONS

6 KECOM?fENDATIONS

The test results indicate that the chemical proportions of the minor
constituent volatile components at Hope lead to the following cycle
levels in stage IV meal.

1.

The typical fluoride cycle level is 1 .2 t o 1 .4 times the input to

the system and does not change substantially as the kiln temperature
or atmosphere is modified, over the range studied.

2.

The chloride cycle is 20 to 30 times the to:al

chloride input and

increases slightly with increasing temperature, although this may be

due to improved kiln stability at higher temperatures.

The ch lor ide

cycle is not modified by wide changes to kiln atmosphere.

3.

Chloride in the cycle combines with potassium where available.

Eowever , at low temperatures - equivalent

to

NOx levels of below

1,000 ppm - some sodium is also involved.

4.

The chloride cycle is a low temperature cycle and cannot be

significantly

5.

modified

by

variation

of

kiln

conditions.

Tne total potassium cycle is approximat ely 3 times the input level,
however about two thirds of this is present in Stage IV in
conjunction with chloride and so cannot be controlled except by
incorporation of a bleed system.

r
0 .

About one third of the potassium in Stage IV is either feed based on

its first passage through the preheater or derived from a potassium
sulphate

based

cycle.

recycle) at 900 ppm kiln

recycle 1.

This cycle ratio varies from 1.05 (minimal

NOx level to 1.25 at 1,400 ppm NOx (25%

This portion can be controlled by operation at low

temperatures,

7.

The sodium cycle level

varies between 1 .2 and 2.0 times that in the

rawmeal over the temperature range examined.

8.

A t lou temperatures some sodium becomes involved in the low

temperature chloride cycle and this boosts the sodium cycle by
about 10%.

9.

The majority of the sodium in the system is involved in a sodium

sulphate

based

cycle.

This is strongly temperature dependent, and

the rate of increase also appears to be increasing with temperature.

Over the temperature range investigated the recycle rose from 1.35
times the level in the feed at low temperatures to 2.0 times at
h i g h NOx l e v e l s .

10.

T'ne total sulphur cycle is temperature dependent and rises from 1.6
to 2.0 times that of the input over the temperature range
investigated

11.

The alkali sulphate cycle has already been summarised in Points 6

and 9 and makes up about 35% of the total sulphate in Stage IV
material.

These cyclic levels can be seen to be lover than the

total sulphur cycle levels.

12

Tne

calcium

sulphate

recycle

is strongly temperature dependent and

rises from about 2.5 times the level in the feed at low NOx levels
t o 4 t i m e s a t h i g h ( 1 , 4 5 0 ppm)

13

NOx l e v e l s .

The calcium sulphate cycle is also increased by a move into a

reducing

kiln

atmosphere.

At about 1,200 ppm NOx the cycle

increased from 3.1 under oxidising conditions to 5.5 times the

feed

14.

level

under

reducing

conditions.

For potassium and sodium there are indications of a slow

but steady

increase in losses of these components from the preheater system

as

temperatures

increase.

This may become part of an external

cycle or may be lost to atmosphere.

This could be established by

a longer term study of the levels of these components in the

precipitator

15.

and

stack

dusts.

Sulphur is in overall balance within the system below NOx levels

of 1,200 to 1,300 ppm, but above this level the loss increases
sharply

with

rising

temperature.

Again this may become part of

an external cycle or may be lost from the system, however in this

exercise no precipitator or stack dust samples were collected so

16.

this

cannot

be

confirmed.

When

the kiln atmosphere moves into reducing conditions the losses

of alkalies and sulphur from the kiln system increases sharply.

In vieu of these conclusions it is recommended that, at hope:

1.

Kiln back-end oxygen level is maintained at 2.0 to 2.52

(as measured 1.

2.

Kiln back-end NOx levels are mainta ined b e low 1200 ppm, although
the most appropriate NOx level will depend

on detailed raw mix

chemistry.

3.

Communication between kiln burner and personnel cleaning the

preheater be improved.

4.

During cleaning operations short term reductions in raw meal feed

which avoid the inception of reducing conditions will be
beneficial

in

the

longer

term.

REFERENCES

1.

G. Mussnung 'Contribution to the Alkali Problem in Suspension

Preheater Kilns (Bumboldt
Energy

Committee of

8 December

2 .

the

1961,

German

Kiln).

Cement

Annual Meeting of the Heat and

Makers

Association,

Dusseldorf.

'Coal and the US Cement Industry, Parts 1 and 2.'

World Cement

Technology, Jan/Feb. 1982 and March 1982.

3 .

Weber 'Heat Transfer in Rotary Cement Kilns' Section 4.5.

4 .

3aspe1, Lorimer, Southern, Taylor 'Blue Circle High Level Kiln

Control' IEEE 1987.

5 .

A. Lorimer 'Fuel Savings Arising from the Operation of High Level

Control at Elope Works', Research Division, Process Development Dept.,
F N

6 .

86/68.

Haspel, Taylor, Kerton 'High Level Kiln Control based on NOx Monitoring'.
Paper to Cetic

7 .

Chemical Commission Mee:ing,

Locker, Spring and Opitz

'Reactions

Liege, June 1985 (ETN 85/21).

associated uith

kiln gases.

Cyclic Processes of Volatile Substances, Coatings, Removal of Rings'.

VDZ Congression on Process Technology of Cement Manufac:ure, 1971.
a .

Davis and Longman 'Design and Experiences with Bypasses for Chloride,
Sulphate and Alkalies'.

9.

Bra-, A.W. 'Chemical Consequences of High Efficiency Systems' Group

Technical Conference 1976.

10. Danoe and Steuck 'Behaviour of Volatile Matter in Cement Kiln Systems'
FLS

Publication.

11. Hawkins and Wilson 'Fluorine Containing Phase causing Blockages

Suspension

Preheaters'.

in

American Ceramic Society, Pacific Coast

Regional Meeting, Los Angeles, Oct. 31, 1977.

12. Polysius Review NO. 58 'Chlorine and its Behaviour in Preheater Kilns'.

13. Goes and Keil 'The Behaviour of Alkalies in Cement Burning'.

Tonindustrie-Zeitung

84(8)

125-133, 1960.

14. H. Carlson 'The Behaviour of Alkalies in Cement Raw Materials during

the Burning Process'.

Rock Products Cement Industry Operations

Seminar 1965.

15. Goldman, Kreft and Schutte 'Cyclic Phenomena of Sulphur in Cement

Kilns' World Cement Technology, November 1981.

16. Hatano 'The Behaviour of Sulphur in Suspension Preheater Kiln Systems'

Zement Kalk Gips, l/1972,

p. 18-19.

17. Etoc P 'The Cement Kiln - An efficient Trap for Sulphur'

Ciment, Beton,

Platzer, Chaux (701) 210-213, 1976.

18. Ritzemann 'Recirculation Problems in Rotary Kiln Systems' ZKG(8),

338-343, 1971.

19. -Lxperimental Prodquction

of a Minor HESC

at Plymstock Works, in order to

Investigate The Potential froblems Associated with The Production of a

Similar Material at a New Oxford Works' Eng.RD TN 81/13.

20. A Lorimer 'On-line SO2

Monitoring:

Cell Systems at Northflee::

Assessmen: of RDUV and Electrochemical

May to October 1984'. Research Dept.

Em 84/21.

21. D W Haspel 'Cement Kiln Alkali/Sulphate Cycles under Equilibrium

Conditions',

Research Division, Process Development Dept., Fh'/'84/93.

22. D G Stenson 'Factors affecting Sulphate and Alkali Cycles in Rotary Kilns
and the implications of these Effects with Respect to Process Control'
Research Dept. ETN/84/13.

TABLE 1

- EST TARGET CONDITIONS

co
-

NOx level

BE02 level
-

1,200

1.8

Nil

1,200

2.5

Nil

800

1.8

Nil

800

2.5

Nil

1,200

AS required

Positive

1,400

1.8

Nil

1,400

2.5

Nil

Test No.

TASL.E

No.

Test

TEST AVERGE

Average

Range of

NOx l e v e l

NOx values

CONDITIONS

Average

BE02

Range

of

level

-BE02 value-

1,244

573

t0

1,938

1.67

1.20 to 2.05

1,214

840

to 1,587

2.25

1.94 to 2.67

995

5 2 7 to 1,472

2.11

I .50 to 2.80

a68

543 to 1,330

2.47

.37 to 3.08

1,223

9 0 9 to 1,418

1.57

0.9 to 1.76

1,462

1,008 to 1,780

1.37

0 . 9 to 2 . 2 8

1,365

1,100 to 1,608

2.72

1.88 to 3.14

Ave CO

level

TABLE 3 -

Test No.
I, Fluoride in

AtERAGE FLUORIDE LEVELS FOR EACH TFiST (LOSS FREE BASIS)

-1

0.109

0.109

0.109

0.109

0.155

0.186

0.185

0.105

0.105

0.13

0.11

0.16

0.16

0.16

0.126

0.128

0.154

0.142

0.179

0.235

0.218

1.16

1.17

1.41

1.30

1.15

1.26

1.18

0.963

0 -963

1.193

1.009

1.032

0.870

0.845

raw meal

Z fluoride in
clinker

2 fluoride in
Stage IV

Ratio o f fluoride in
Stage IV to raw meal

Ratio o f fiuoride in
clinker to raw meal

TABLE 5 -

Test

TEST AVE.RXE CHLORIDE LEVELS (LOSS FREE BASIS)

No.

1.

% Chloride input

0.060

0.060

0.063

0.064

0 -059

0.059

o.oc+

2.

X Chloride clinker

0.16

0.14

0.009

0.017

0.008

0.008

0.010

3.

% Chloride in Stage I V

1.295

1.405

1.702

1.419

1.346

1.519

1.627

4.

Ratio 3:l

21.6

23.4

27.0

22.2

22.8

25.7

25.4

TABLE 6 -

Test

EST AVER;IGE KpO LEVELS (LOSS FREE BASIS)

1.

% K20 in raw meal

0.65

0.622

0.623

0.636

0.651

0.646

0.678

2.

% K20 in c linker

0.565

0.551

0.64

0 -615

0 -465

0.53

0.56

3.

% K20 in Stage IV

2.505

2.618

2 -749

2.452

2.387

2.811

2 -976

4.

% K20 in SCage IV

1.752

1.851

2.089

1.781

1.585

2,028

2 -133

0.753

0.761

0.66

0.671

0 -802

0 -783

0.843

Ratio 3:1

3.85

4.21

4.41

3 -85 '5

3 -667

4.351

4.389

Ratio 5:l

1.158

1.223

1.059

1.054

1.232

1.212

1.243

Ratio 2:l

0.86

0 -89

1.03

0.97

0.68

0 -82

0.83

as chloride
c % K20 in Srage IV
2.
as sulphate

TAELE 7

Test

LEVELS (LOSS FREE BASIS)

0.218

0.20

0.202

0.202

0.233

0.231

0 -277

1.

% Na20

2.

2; Na20 in clinker

0.211

0.207

0.221

0.227

0.172

0.195

0.196

3.

% fia20 S t a g e I V

0.329

0.294

0.327

0.323

0 -330

0.471

0.499

4.

% Na20

IV

0.005

0.006

0.033

0.052

0.00

0.011

0.00

IV

0 -324

0.288

0 -294

0.271

0.330

0.460

OTFFJ

R a t i o o f 3:1

1.509

1.470

1.619

1.599

1.416

2.039

1.801

R a t i o o f 5:1

1.488

1 .44

1.455

1.342

1.416

1.99

1.801

R a t i o o f 2:1

0.97

1.03

1.09

1.12

0.74

0 -84

0.71

as
5.

in raw meal

- TEST AVERAGE NA20

i n Stage

chloride

% N a 2 0 i n Stage
a s sulphate

TABE

Test

TEST AVEUGE

1
-

SO3 LEVELS (LOSS FREE BASIS)

1.419

1.473

1.668

1.696

1.

2, SO3 in raw meal

1.53

1.369

1.417

2.

% SO3 in rm (CaO)

0.696

0 -582

0.627

3.

% SO3 in clinker

1.507

1.417

1.449

1.485

0.838

1.400

1.425

4.

% SO3 in Stage IV

3.176

3.193

2.356

2.710

4.433

4.426

4.005

5.

% SO3 as CaS04 as

2.118

1.862

1.515

1.812

3 -326

3.166

2.644

R a t i o 4:l

2.076

2.142

1.663

1.910

3 .OlO

2 -653

2 -361

R a t i o 5:2

3.043

3.199

2.416

2.932

5.373

3 -856

3 -470

R a t i o 3:l

0 -985

1.035

1.023

1.047

0.569

0.84

0.84

in Stage IV

% a s

CaSO4

66.7

58 .3

64.3

66.9

75 .o

71.5

66 .o

RELATIONSHIP -BE;TwEEN NOx AND LX&T

HOPE WORKS No 2 KILN:

hourly ATerAg data

from 3/5/85

x
0

daily
August

730
1350
Burning Zone Temperature ( "cl

1400

(av.Worka pyru rending)

1450

AVel%gt

7985

dAtA

--

-.
;
.-.-:
_
:

.
_

._
.

-_.

-.
.-.----L-~--.--

I
I
I

_._.

j
i..-+
I

1
I

I
/

_.

--_- -- ---

.,-

-....-_..

I...

..-.i-.

..i

-.

;
1
!
I

..-A
----!
-- ___!

i
o---.
-CD_
-._-_I_
----

----.--

I
i
I
fi
I
I

E-p----j

_
-... _----+____-. . - __. - .
----.
-.
..: -_~
-L-.----d-h. .
. -- -/ i-. -.-. -.. _. __ -.~~~p---~a :_---- --.- i ..-- ~-.. _ . ..- :.-_&-- L- ~---. _ -:..-~------ 7---_ _.. ; -.- ---I

- ..---- ---. -----r-----..-------------:

_[
-kII+x*y-----/
_ _-_--_ _-.-._I
-.. _---__--._-__ _ . ; _____. r( __. * __ - --+-~
n.
I
c
I
I
----+--. -__ - -i- -______;__

_--.-._ _- ----+- +--+-:

-I:,. C--_-.----L
__.___ ------_i
/
- ---. -_-- _._- _ . __
-- --- -- .---- - . -. .-.- __ /
-QL
- ,L,. . --.-.--i-_-- iL--

..

1 :-.: - .._ _ -1 _.. _. _ _ _. ) . .

----- i
_ _ ._ - , - . . _.-. .I . .._ ..-.. -.-. ! _ _ _ . . .
:: ._ _. . _-_
-
_ . -- . ..- -_-.-- - ..- _.-I
_ -. r _ _ _!___
-__.
-.i--,
~ -._
I . . . .._- I..--. -. _ . . . i-1 --.
__.
,I
_ .
_i .
-

-. . . ----+.---z_i~:~~z+:~
;
:
11
:
-..-.- - . ___.- .: . .- . ______
_ . - - f _ _. __ _ ----;
..----_d
_-----.
_ _ . _ -.l.x__- x.1:
. -:-.. : : ..: 1:
, _ _ - .-. _ - _.-- . . ..-.- - _----_ _. .:_. __ _ __._ _-_ .iI---:..
: _. . . .

_ __-_- --. - -._

-----.
. . ___
- - -- ____
_ -._ _

----__
__...._ -.!.T.- -_---_- . ..-- - -_- -. - -.. -!A-- .-.. +.--L-. ..- - -- .UiLX~ hd-u**,cI) j

t-

L-

.2

P -

-.

Ii..

.i

-.

_----_-__-.---j j i .: .

~:

.:

-:

-;.-L--

---;-

I
t
I

__.__. ---_.-

.-.-----*y-T--yf-y

I
\

_ - - ._ ._ .- .
!i.c-j .-i-L_.-- : ._ - i. _. -. . ._ _ - _-._ -.
d
_-_
__.
_
--..
_
__
- m
.A
:-----.-I - If-.- _-.-.--:.

qi
1..
-4
-:
--

.._

y.--_~.;:+ -------

:
:

:
.

Lb.
ri

I .:.:
;

..:
,.::
:_
j
;

:
--.--

f
;

i.;
:

-i

:j

-2

-:

-:; -:....:.
,:. i:: ;., ..-.:
: : : ; ..:
_
:

i
.

--I

..:

-.

.I
..:

.-:

: .~
::

:.;;

_:

_.

.-

~.

::.

_.

-1..-

:.

i..:

-.=:

y
i.I

1
:

;..:
:_:
.

.,

i..

634

PLY?WJ!OCK

WORKS.. VARIATION OF SDLPHATE CYCLE WITH BZT

2c.b
so3

300

ex ntnge IV a8

of total SO, in feede

I
400
(meal

I
500

+ coal)

X?PEE;DIS

GUPHICAL REPRESESTATIOTS OF VARiATI04S

STAGE IV MATERIAL CklEYISiKV

IK

0% A?i IYDIVIDOAL TEST EASIS

E: :. -.:::.

i .-:
a-.-.
.

:
:

I
:

.
i

-1

:-

,:

.-

----...
-.7--.-

---.-

---..A--+...--.-

-.

---i

~.~:

----j

- ..-- - . .._.._ .._ .

-_

--;.

-..A----..+-.L-

_-.-

I
:

--i-_..---_7_

------

--+1

:
:
I ::
--~
:rc-r

:f

:
\I

-- :-

--L-1:
:
:

:
:/-

:
:

!
;

::

:
::
1 ::
:

.:
:.:
::j

;
:
IiL.--:_:
1
1.
----+,:L. ..I ..i:-_ ,.
v
E-. : .Y.-.
t
:.
i
;
iw
t
:

.:_:

+ .
&&
g.L
iz?---+
::-_ :
L;:

;--.
++j
t
L-;.
!
+--t
:
E--L

._-

---2

Le.

._

_.

_
..:_

r;

-:

.:

.--:
.:

-I.

;:-...__.

7
-_- . -._ _:

: . . . . _-..

.-

._I..

--

-.---

---___

.,

--..-

::
L

:::

-_
:
:

.;
:

I-,.:
- :Y.
i - ..f --z
.
.
.:

._

k:A.
----.-. -.-.-1
-

:
T----y

:.1
:
:

:.:

:
:;..:
::..

:
. .
.:
:
::,
;
.:
:

:I
:
:

:::
:

::
:
_:
:--A.-L
.:
i

i.
.:

.:

:7

,.::

24,

Ty

A
4:..
/.
\
:/ :
:
:

.,

:
j
:.
:
i::
:

j
:
::

__:

-.:
i...;.i
:

:.:

i
1

..i

.;.

..i

..:

(-.

:
.-_;
.-:
:
:
:j
.
.._..._
_....
.;
_;:
:.j
r1--.
..:. 2;. .,:
-.--;: ..: . . : -: : : . . .: .: ..: - .=

?-..---A

_-

.-i

~
..~
4.1:

-.

--.

:.

lr.

.~

:.

_.:z;
-_

.._...

..:.:_.. : .\

:..::
: _
.-.-:.
.
--

APPENDIX 2

PLANT .?EASUREiHEATS

.9/?/37
3

3-27
--

l220

lGl3

..-

23-o

I-P?

~0,7
.-

lG7
20.3
.-

q I-?

Kc1

I so

-+a-

133

QO

536

33t

I BLC,

S-11
---I822 I

I
2

3
4
s
6
31

APPENDIX1
EOPE h7XK.S ALKALI CYW
BuRNI%

ZONE c.kimIcAL

TADER 4UILIBRIUd

Material

%Feed

% B.E.

LOAD

COhDITIONS

% Prcduct 1982 clinker

Average

Absolute
% Recycled

Na20

O-2

0.4 (x2)

0.2

0.25

0.2% (xl)

K20

0.6

6.0 (x10)

0.6

0.55

5.4% (x9)

SO3

1.3

6.5 (x5)

1.3

1.3

5.2% (x4)

cao (caSO4 If

0.37

0.74 W)

0.37

0.37%

0.37% (xl

Cl

MJOI

2.0%

0.19

13.17%
* Eetemined

fran sulphate balance.

Assuming the average heat requirement to vacourise - deccxqmse alkalis

averages 800 kcals/kg (600 to 1000 kcals/Ig[l] 1 then average aBali
vapourisation load on th,n burning zone is 100 kcah/kg.

[ll

Vapourisation only - No decmpos ition of alkali salts.

m 84/93

Over a period of years, various extimtes

drive off the alkali/sulphates

of the heat required to

in the burning zone have been made. The

majority are in the order of 100 kcals/kg.

However ETIU 83/13 has sho+.n that

this can easily tax-y on a short term basis by i 50 kcals/lcg

disturbance in-the Burning Zone Tarcjerature.

due to a minor

This note references scxne of

the various source& on this subject.

1.

"Heat Transfer in Rotary Kilns" Weber. P.

ZIGS Spcial

Publication X0 9.

Alkali cycle equivalent to 100 Kcals/kg clinker.

2.

Lepol Grate Heat Transfer Mechanism

D-W.

Haspel June, 1973

ikxaxA 005
At least 70 kcals/kg clinker of heat are recovered as a resu+lUor the
condensation of alkalis/recrmbination
3.

Hope Works Alkali Cycles - See

of SO3 on the Lepl grate.

Appendix

I.

The mean heat required to vapourise the alkali/sul?hates at Eope is in

the order of 100 kcals/kg.

A further 10 kcals/kg is involved in the

melting/fusion of the salts.

The exothermic clinkering rcuctions cccuring in the burning zone are
equivalent to -97 kcals/kg (typical 1982).

APPENDIX II
HOPE 5XRKS CLINERING
~thermic

REACTIONS

Reactions:Heat of Form&ion

Hope % (1982 Av)

Clinker Canposition

k&b/kg clinker

C3S

-126.2 kcal/kg

62.7

-79.1

C2S

-174 kcal/kg

9.2

-16.0

C3A

- 3.7 kcal/kg

10.4

- 0.38

C4M

-20.1 kcal/kg

7.8

- 1.57

Total

-97.05 km&/kg clinker

90.1%

Note heat required to prcduce C3S fran C2S is +2.8 kcals/kg reactant
Endothermic Reactions:tlO0 kcals/kg

Alkali Recycling
Latent heat of Z&ion

(circa 10% of vapourising load) + 10 kcals/kg

ll0 kcals/kg
Net (Ekothermic/~doth~c~

+ 13 kcals/kg

Useful Heat in Gases and Clinker Dust

Useful Heat in Gases (Sp Heat xA T) circa 0.17 x

T = 17 kcals/kg/lOOC

Useful Heat in Clinker (SP Heat x A T) "

T=

0.30x

30 kcals/kg/lOO"C

APPENDIX Iv
Z-IBM!

DISmmCE

Area to b= cledned.

?!SSCC3ATED

WItTYI KILB CLEYNFG

(HOPE)

8m (circumference) x 5m (length)
= 4Om*

AssumS thickness circa l/4 - l/3 m

Total rr&erial to be cleaned - circa 10m3
Assuming S.g of 0.8 - 1.0, Material hocked

down 8 to 10

totes,

which circa 20% - 30% are alkalis.

With additional heat requirenent to m&e clinker 300 kcaI.s/kg
This material is equivalent to 3 'Lonnes of clinker
(distrilxlt& over two hours?)
or5

tOMeS

of feed

D. W. Has@
November 1984

of

APPENDIX III
PHYSICALC0NsmNE

ASSCCIATEDWITE! ALKALI CYCLES

Latent Heat of Vapxrisation/Eemqmsition

Kcl

+670 kcaldkg

W)

K2s04

+563 kcals/kg

W)

+1805 kcals/kg (D)

Na2SO4

t-733 kcals/kg W)
+2096 kcals/kg (D)

CasOi

+675 kcaldkg

(D)

v = Vapxr
D = Dissociate
Later&He&of

Fusion

KC1

+96.6 kcals/kg (M-P.

K2SO4

+ 57.6 kcals/'Kg

Na2SO4

+ 42-S kcals/kg (M.P.

722C)

(M-P. 1069C)
884C)

Blue Circle Cement

PROCESS ENGINEERING TRAINING

PROGRAM

Module 13
Section 7

Design and Experience with Bypasses

for Chloride, Sulphate, and Alkalis

DESIGB AND EXPERIENCE WITB BYPASSES FOR CBLORIDE, SULFATE .tvl, ALKALIS
By Xessrs P.. Davis and P-A. Longman
Blue Circle Industries, United Kingdom
1.

INTRODUCTION
During the production of Portland cement clinker, some of the
minor components $n the raw materials and fuel, notably the alkalis,
sulphur and chloride compounds are volatilised

in the kiln. These com-

ponents are then swept with the kiln gases to the cooler parts of the
kiln system uhere they condense.

In the less thermally efficient pro-

cesses, such as the wet and long dry process, a significant proportion
of these volatiles can *escape' from the system either in the prccipitator/5ag house dust or up the stack.

In the Lepol or XL process,

which we have successfully operated for more than 20 years, fairly

high levels of volatiles can also be tolerated. This is partly
because the nodule grate is not a particularly effective scrubber and
partly.because

precipitator and

intermediate cyclone dust can be

discarded thereby lowering the volatile recirculating load.

Similarly, in two stage preheater processes moderate volatile inputa
can be tolerated.
However, in the more thermally efficient processes, such as the
four stage suspension preheater process, greater amounts are "trapped"
in the feed as It passes down the preheater and are carried back into
the kiln, where a proportion again evaporates.

This repeated evapora-

tion and condensation leads to the gradual formation of an internal

volatile recirculating load until a steady state is reached(l).

-2When the first suspens'ioti.preheater

processes were:-built

1950's the importance of recirculatingvolatlles~was

In-tt.

m.t fully appr

ciated and- a- number bf plants simply could- not run.&ecausc.of build

and blotikages at" thi Wla inlet and in. the--louer, stages: of the prcheater.

I,n some instances-the clinker produced uaa of unacceotable

qualitv because af its hinher alkali content in comoarlson ulth that
produced in the simpler, but leas thermally efficient processes.
Over the yeears, as a result ef studies made by both the plant
svooliers and the cement manufacturers (see references) an
understand$ng has emerged of the factors affecting the formation of
the_ volatile recirculatinn load in a four stage ausuension oreheater
process, and of the effect such a recircnlatlon has on plant ~4
mance and cement quality.
As a reaul.t, It is possible, when employing the four stage
suspension preheater process for cement manufacture to predict the
maxImum levels of chloride. alkalis and sdohate that can be tolerate<
in the raw materials and fuel.

Where these levels are just exceeded

the situation can often be recovered, ad kiln operation and cement

qiiaLitp,ptob1en!s.

avoided, by extr$cti-ng

'befbre. tA#! preheatar

a proportion of the kiln gases

such-.tI& a .sufficient

proportion of the unwanted

volatile! components- are "bled" qut the system flowever, operation of

such a."bJeed" or Wla bypass,also depletes. the preheater of some of
Its he,- uith,the.reault. that -on-a susoension preheater process
the kiln bypass has to be llmited$to_about 25% otherwise there is
insufficient heat for effective preheating(2).

-3Bith the introduction of precalciners in the mid 1970's it vas

generally recognised that this limitation could now be overcome and
that by burning a substantial proportion of the total fuel in the precalciner it would-be possible to bleed all of the kiln gases, if
necessary, and still achieve sufficient decarbonation and calcinatioa

in the preheater(3r4,S)*

Moreover, whereas in a suspension preheater

process a fuel penalty of about 4 kcal kg'1 of clinker is incurred for

each 1% bypass, in the precalciner process this is roughly half at
around 2 kcal kg-l.
At that time, therefore, the precalciner pt'ocesi looked particularly attractive in this respect and partly for this reason'gained
videspread

adoption.

Since then operational'experience and results

have been obtained vhich have cast doubt on the effectiveness of the
kiln bypass on this type of process.
In Blue Circle Industries, in addition to operating a number of
conventioaal

suspension preheater processes with no bypasses, UC have

experience in running one suspension preheater process fitted ulth a

small bypass to remove cnlorlde and two precalciner processes fitted
vith bypasses to remove sulphur and alkalis respectively.
In this paper, we outline the chemical factors which determine.
the volatile recirculating load in a four stage preheater process and
the effects this can have on kiln operation and cement quality. Our
experience in operating both four stage preheater and precalclner
kilns with and vithout by$asses and how the bypass performance compares vith design predictions is-also discussed.

-4CFEMICZLL,
2.1

PACTORS

Volatile

Components

The two main factors which vi11 dictate uhether a kiln bppasz
necessary, and if so hov large, are:-

i)

The total quantities of alkalis (sodiumGiAd potassium), color

and eulphur present in the rav materials aad fuel which vi11
determine the overall composition of the clinker.

ii)..,The

volatility of these components in the kiln system which WI

determine-the nature of the volatile recirculating load.

The main source of alkalis is the clay and shale phases in the
secondary rau materials and in general the presence of al'alis in t1
primary calcarcous raw material or the fuel arises from coataf"~

-'ic

wtth argillaceous materials.

Chloride is normally present as alkali chloride in the raw
materials and fuel but the presence of organic chloride compounds in

coal or other fuels is possible.

Sulphur can occur as a variety of organfc and inorganic compounti
in both fuels and raw materials and in a variety of oxidation states.
The volatility of these components in the kiln system Is complex
and is dependent on the follouing interrelating factors:-

i>

Whether the volatile components occur in the fuel or the raw

materials.

ii) The relative amounts of the total alkalis, sulphur and chloride
entering the system.
iii) The burnability of the mix, uhich can be influenced by the
overall chemistry of the clinker., the fineness of the raw feea
'and whether mineralisers such as fluorine are present.
iv) The burning environment in the kiln system, which vi11 be
influenced by (iii) above, the level of free lime required, the
nature of the fuel and firing system employed and whether carbon
and/or sulphlde is present in the raw materials.
2.2

Volatile

Cycle

In practice, the volatile recirculating load in a kiln system

vi.11 gradually build up to equilibrium and it is customary for
researchers and workers in this field to estimate the final balance of
volatlles recirculating by regarding the build up in cycles.

A number

of papers have been published on this subject (6-9) and this is an

approach which we have been using for some 20 years, uith the
assistance of computer programmes for the past 10 years.
In our current computer model the first cycle is regarded as
where fuel and feed first enter the kiln system.

In general the vola-

tiles In the fuel are assumed to evaporate completely as the fuel comSusts, while of the volatiles in the raw materials, only a proportion
are assumed to evaporate - the remainder being retained in the
clinker.

-6 Of the volatiles that have been evaporated a certain propor:

(generally most) are assumed to condense on the feed in the prehe
such that In the secood cycle the material entering the burning zt
contains both recirculating volatile compounds and volatile compel
present in the raw materials.
For the second cycle a proportion of the volatlles in the raw
materials will again evaporate in the burning zone and the rentaind
stay in the clinker.

Similarly, a proportion of the recirculating

compounds will also be evaporated uhile the remainder are retained

ihe clinker.

With each ensuing cycle, therefore, the quantity of

volatile compounds recirculating and the quantity retained in the

clinker gradually builda up until the total quantity entering the
system via the feed and fuel equals the total quantity leaving

system 'via the clinker and stack.

A simple illustration of this is shoun in Figure 1. Rouever,

tsoastruct

such a balance it is necessary to have a knowledge of the

recirculating compounds being formed and the volatllities

of both

these compounds and those present in the raw materials.

Below, ue

.summarisf

the information we have acquired from both laboratory

investigations and studies of full scale plant of the behaviour of

chlorine, sulphur and alkalis in suspension preheater processes. It
was on these data that we based the design of our bypasses for our
precalciner plants, and uhich in the light of our more recent
experiences now require modifying.

-7

2.3 Chloride
Virtually all the chloride input to a kiln system vi11 be volatilised at burning zone temperatures and in general less than 3% of that
entering the burning zone will leave via the clinkerc2).
During the process of evaporation and being swept down the kiln
the chloride will preferentially react with potassfrrm

to form

potassium chloride and in general sodium chloride vi11 only form if

there is an excess of chlorine over potassium.
A very small percentage of this alkali chloride may escape up the
stack as fume or fine particles of dust but the bulk will condense out
on the feed.
By itself, potassium chloride till largely condense out in the
temperature range of 800 - 900C and as a result it can often cause
blockages and form hard deposits in the back end chute or in the riser
pipe to the lower preheater stage.

If sodium chloride is also present

the temperature range over which the chlorides vi11 condense till be
extended generally dounwards i.e. 700 - 900C and in terms of kiln
operation this may present fever problems.

A similar effect occurs in

the presence of sulphates.

2.4 Sulphur
The behaoiour of sulphur in the kiln is more complex than that of
chloride and is strongly influenced by the form of the sulphur and the
burning conditions.

-aIn general, sulphides present in the raw materials wfllread~

oxidise and evqorate such.that most if not all are driven off by
time the feed reaches the burning zone.

Calcium sulphate, whether

present in the raw materials or. fotmed in the kiln cycle, although
not particularly'oolatil,e, readily dissociates at burning zone tur
peratures particularly under:prevailing

reducing conditions, and y

only a small proportion of this sulphate- is retained in the clinke:

Alkali sulphatea similarly dissociqte

id's reducing cvirooment.

However, in an oxidising enviromaent they are'nore stable with the

result that up to 40%. of the total alLd.1 sdphates entering the
burning zone map be retained In the clinker,
It follows that the bulk of the sulphate retained in the clinks
is alkali based - either as alkali sulphate - @i?la)2S04 or as,
alkali calcium sulphates such as calcium langbeinite
This is illustrated

It

2CaSOf,.K2suq-

19 Figure 2. . In contrast the sulphate evaporate

may be alkali sulphates, dissociated sulphates or oxidised sulphide.

This mixture vi11 condense and/or react over a tide range of texperatures as it passes to the cooler parts of the kiln. SO2 gas vfl
readily react uith calcium oxide in the decarbonated feed at the bat
end of the kiln but any uhich is not trapped in this ny say well pa:
through all of the preheater and escape up the stack(lO*ll).
Most sulphate compounds will condense at temperatures of
900-1100C.

In our expeience if the alkalis and sulphates are more

or less in balance this condensation forms a light loosely bound depa

sit in the fourth stage and kiln inlet which can be scoured away by

-9passing feed or if necessary easily removed. Aowever, if there is an

excess of sulphate over alkalis the deposit can be more sticky and
harder to remove.

Densified layers of calcium sulphate and sulphosi-

licates 2(2CaO.S102)CaS04

(Figure 3) may form which vi11 adversely

affect kiln performance and. require special cleaning

facilities(12r13).

If fluorfne is present In the rav materials the

situation may be exacerbated by it promoting the formation of intermediary compounds such as the fluorinated
(CllA7.CaF2),

calcium dumim:e

spurrite 2(2. CaO .SiO2 )CaC03 and various calcium sulpho-

silicates.
2.5

Alkalis

In general a substantial proportion of the recirculating alkalis

will be associated uith either the chloride or sulphate as discussed
above.
Houever, there ulll always be a tendency for some alkalis, par
ticularlp Na20, to be retained in solid solution in the main clinker
phases(14).

The proportion of alkalis retained in this uap in the

clinker will increase if there is an excess of alkalis over sulphate

entering the system or if reducing conditions prevail in the burning
zone.

Reductioa

will lead to dissociation and preferential loss of

SO2 although with severe reductioa, alkali iron sulphide compounds

such as ?ZeS2 may form in the clinker which results in a greater
retention of alkali and sulphur in the clinker than is otherwise
expected.
In general up to 50% of allalis entering the burning zone as
alkalis oxides may be retained in the clinker.

- 10 If there is an excess of al.kali over sulphate uhere condensation

occurs at the back end of the kiln hard alkali carbonate based deposits may form.
PRACTICAL IWLICATIONS
3.1 Kiln Operatioa
Published data, coupled vith our OM experience, suggests that
the formation of blockages and hard'deposits in the fourth stage and
kiln chute are likely to start presenting problems when the level of
chloride or sulphate in the feed material leaving stage 4 exceeds 1.5%
or 3.5% respectively.

In addition such levels of sulphate may also

enhance ring formation in the kiln itself thereby restricting output.

The quantity and nature of the volatile recirculating load will
obviously vary from Works to Works, but in general practical terms in
order to avoid exceeding the above limits it will be necessary to
restrict the total volatile input to the system, when &pressed as a
percentage

of

the clinker output, to 0.025 - 0.030% chloride and

around 1.5% sulphate.

Normally,

the alkalis vi11 be bound up ulth the chloride and

sulphate and the levels that can be tolerated in the material leaving
the fourth stage till be dictated by these components.

If alkali is

predominantly present as alkali carbonate, however, we vould expect to

be able to tolerate up to at least 3% alkalis in material leaving
stage 4.

Thus in terms of kiln operation it may be possible to

- 11 -

tolerate a total alkali

iIIQUt

to the SysteZI expressed as a percentage

of the clinker output of 1.2% equivalent xa20 without running into any
serious problems.
'3.2 Cement Quality
In terms of cement quality the obvious factor which may dictate
limiting the quantity of alkalis retained in the,clinkar vi11 be the
necessity to manufacture a low alkali cement in order to avoid expansive alkali aggregate reactions occurring in the concrete.

In this

respect the ASTM limit of'O.6% equivalent Xa20 is generally recognised

worldtide.

Aowever,

even if the productioa of a low alkali cement is

not required, it may still be necessary to limit the level of al!calis

retained in the clinker.

For example alkali sulphates vi11 shorten

setting times, enhance the early strength of the cement at the expense
of its late strength, and will render the cement more prone to air
setting during etorage.

These effects vi11 normally become 'noticeable

-Jhen around 1% alkali sulphate is present in the clinker and will be

significant at the 2% level.
If there is an excess of alkalis over sulphate then the alkalis
will enter into solid solution in the main clinker phases and in
general late strength will be depressed. In this situation If the
clinker is also subjected to reducing conditions then the resultant
cement will be prone to adsorbing moisture from the atmosphere and its
flowability

properties uill noticeably deteriorate. This effect is

illustrated in Figure 4.

- 12 If there is an excess of sulphate over alkalis then the formation

of calcium langbeinite

(2CaSQ.K2S04)

will enhance strengths at all

ages.

If high levels of sulphate are retained in the clinker this till

limit the level of gypsum addition that can be made at the c&en&
mill.

This in turn will lead to an increase in the milling energy

required to grind to a given surface area or cement strength.

Although chloride is not usually retained in the clinker in
appreciable

quantities, there is always a risk where the chloride

input is high that unstable kiln operating conditions will lead to

periods of production where high levels are retained in the clinker.
This could present problems if reinforced and/or prestressed concrete
is prepared from this cement since the chloride may migrate through
the concrete and lead to corrosion of the reinforcement.
Iu the U.K. it is recommended that prestressed concrete contains
less than 0.06% chloride when expressed as a percentage of the cement
component.

In practice, after allowing for the contribution from the

aggregates and water, this means limiting the amount of chloride in

the cement to less than 0.03%.
3.3. hyuass Design
Should a kiln bypass be required to control the quantities of
volatiles recirculating in the kiln or retained in the clinker, then
this Is nornally located at the kiln inlet.

The objective is to bleed

off the kiln gases containing volatiles without removing excessive

quantities of dust.

- 13 -

Some plant suppliers have found the best location for such a take
off is from the riser immediately above the kiln and along the line of
its axis. Houever, promising results have also been obtaihed with
bypasses located on the side of the riser and fitted with a deflector
which keeps the hot gases from the kiln apart from the incoming feed.
The hot gases bled from the system ulll, of course, need to be
quenched to freeze the volatilised material, and then cooled and
filtered.

As close to the take-off point as possible, the..gas should

either be quenched tith cold air to a temperature of 250C and

dedusted in a glass bag filter, or air quenched to around 4OO"C, water
cooled to 150C in a conditioning tower and dedusted in an electrostatic precipitator.

Variable speed fans are preferred for the quench

air and filter fans to give a reduced electrical power consumptioa

should it be found possible to operate at less than the design bypass
percentage.

Tfius the capital kost of the bypass and its ancillary

equipment, whilst only a small proportion of the total cost of the

plant, is nevertheless significant.
In sizing the bypass, the procedure we have adopted is to
construct a similar volatile recirculating load to that illustrated in
Figure 1 but with a proportion of the volatiles now bled from the
system.

This is illustrated for the suspension preheater and pre-

calciner process in Figures 5 and 6 .respectlvely.

For these calcula-

tions it is assumed that the bypass is 100% efficient i.e. a 50%

bypass vill bleed off 50% of all of the volatiles approaching it in
the gas stream.

By running our computer programme for different

- 14 bypass levels it is then possible to plot the stage 4 and- clinker

volatile contents against X bypass and from these plots read off the
level of. bleed required.
3.4

Bypass

Operation

In addition to the capital cost, a bypass facility vi11 also

incur running and maintenance coats.

As mentioned previously there

vi11 be a fuel penalty and this is illustrated for both the suspension
preheater akd the precalciner processes in Figure 7.

Bypass dust will

need to be handled and disposed of and inevitably there vi11 be a

depletion of rav material reserves.
At this stage, therefore, it is uorth considering some of the
alternative

eolutions.

In terms of kiln operation a slight excess of

volatiles above the limits discussed in section 3.1 can often be dealt
with by installing poking facilities to remove hard deposits in the
riser pipe and in the lower stages of the preheater.

-An imbalance

between sulphate and alkalis can be restored by the deliberate addition of an appropriate source of volatile compound to the feed.

Uh

chloride inputs can often be avoided, for example, by purchasing a

special lov chloride coal which although generally more expensive than
the normal fuel employed we have found that in some cases its use is
cheaper overall than operating a bypass.
In tens of cement quality modifications to the setting and
strength characteristics of the cement caused by the presence of
alkali compounds can be countered by adjusting the main chemical parameters.

?or example it say be possible to offset the effects of

alkali sulphate by raising the silica ratio.

- 15 If

Works is only required to manufacture low alkali cement for

some of the time then It is conceivable that alternative raw materials

sufficiently low in alkalis could be acquired for this purpose or it
may be more economic to use selective quarrying. For example, Blue
Circle Industries are presently converting an old Lepol process plant
to a precalciner.

The level of sulphur in the kiln feed uas such as

to require a sulphur bypass in the updated process.

Aowever,

detailed

geological investigation showed that a large proportion of the sulphur

was contained vithin a well defined, approximately lm thick, band in
the shale quarry.

In terms of both capital and operating cost it was

shown that it would be-more economic to discard this high sulphur

material by selective quarrying:

therefore, no bypass has been

installed.
In some instances rather than operating a bypass to remove volatiles it may be more convenient to make further additions to the feed
in order to produce a fully mineralised
propertiee.

clinker with enhanced cement

This is a technology which ue have developed and suc-

cessfully applied in Blue Circle Industries (15-17).

4.

PRACTICAL EWERIENCE
4.1 Suspension Preheater Plants without Bypass Facilities
Blue Circle Industries currently operate a number of suspension
preheater plants without bypass facilities. Samples of kiln feed,
fuel and clinker are regularly analysed on the Works and on several
occasions samples have been taken from the stages of the preheaters ia
order to construct a volatile balance.

For the purposes of this paper

-de have selected four plants to illustrate the range of our

experience.

- 16 The total volatile input at these plants derived from both the
raw materials and fuels and expressed on a clinker basis is as
follows:-

Plant

lA

1B

4A

4B

Fuel

Oil

coal

coal

coal

Gas

Oil

Total S as SO3

,0.90

0.95

1.70

0.45

0.1s

1.05

K20

1.30

1.30

0.65

0.35

0.60

0.60

NaZO

0.20

0.15

0.20

0.10

0.50

0.50

Cl

0.02

0,ozs

0.02

0.04

0.00s

0.00s

Further details of these plants, the typical levels of volatiles

found in the stage 4 feed and the typical equivalent ??a20 content of
the clinker are shown in Figure 8.
Plant 1 was originally oil fired but in 1981 it was converted to
coal firing.

By necessity a low chloride coal is used 1:; order to

avoid having to install and operate a chloride bypass.

With both

fuels the alkali and sulphate input is high but more or less in
balance uith the result that although build ups do form in the riser
and fourth stage they are relatively easy to remove.
The resultant clinker, however, has a high equivalent Na20 coutent, the bulk of which is present as alkali sulphate.

Although the

productioo of a low alkali cement is aot necessary to counter the

adverse effect the alkali sulphates have on the late strength of the
cement the Works have traditionally incorporated some sand in their
mix to lover the total alkali input and raise the silica ratio.

Hore

recently, mineralisers have been successfully used at this Works to

- 17 achieve both a further Improvement in late strengths and a reduction

in the level of alkali retained in the clinker.
Plant 2 is also fired uith a specially purchased low chloride
coal in order to avoid a high chloride recirculating load.
The raw materials "are relatively low in alkalis with the result
that a low alkali clinket,uith an equivalent Na20 of less than 0.60%
can be easily produced.

Unfortunately, these rau materials also con-

tain fluorine and appreciable quantities of sulphur and uhile both

these components can be limited by selective quarrylng their presence
leads to the formatioa of hard back end deposits uhich require regular
removal.

!4'hile such removals do create a certain amount of kiln

instability this is not excessive.

Overall

this

plant

demonstrates

that vith comprehensive cleaning facilities high volatile recir

culating loads can be tolerated, and the formation of hard back end
deposits can be successfully controlled, vlthout having to resort to
the operation of a kiln bypass.
A further consequence of the presence of fluorine, hovever, is
that the early activity of the oement is depressed with the result
that, in viatet in particular, setting times are extended and any
placed concrete has a tendency to "bleed". Various solutions to this
problem have been tried but have either been impractical or uneconomic.

More recently, however, this problem has been overcome by the

use of mineralisers.
Plant 3 is fired ulth a locally available coal.

As a consequence

it3 chloride Input is high and would normally require the operation of

- 18 a small chloride bypass.

Houever, the sulphate and alkali inputs are

both relatively low, and more or less in balance, vith the result that
by operating facilities to regularly clear away deposit? which form in
the riser and stage 4 the need to bleed any of the kilns' gases has
been avoided.

The resultant cement has a very low alkali coatent well

belov 0.60% equivalent Xa20.

Plant 4 contains bath an oil fired kiln and a gas fired kiln.
30th kilns have a lov chloride input but a high alkali input which
results in the equivalent Xa20 in the clinker exceeding 0.60%.
However, as extended cements are mainly produced at this Works this
does not present any problem. . In term of kiln operation, the gas
fired kiln has an excess

of alkalis over sulphate uhile the oil fired

kiln has an excess of sulphate over alkalis. Both kilns form deposits
in the riser and fourth stage which are easily removed.
Construction of the volatile recirculating load for each of the
above 'kilns has indicated that the proportion of volatiles evaporated
in the burning zone of each is similar.

In all cases (Figure 9) vlr

tually all the chloride, approximately 85% of any excess sulphate over
alkalis, around 65% of alkali sdphate and about 60% of excess alkalis
over sulphate was evaporated.
4.2 Chloride Bypass - Plant 5
In addltioo to the above suspension preheater processes, a
further one is operated in the Blue Circle group *with
to remove chloride.

a small bypass

Up to 1975 this plant uas oil fired, thereafter

it has been fired with natural gas.

- 19 The raw materials employed on this 53orks comprise a high grade

limestone, a marl and a sand.

Both the limestone and marl contain

chloride with the result that the total chloride input to the kiln
system, as a percentage of the clinker output, is around 0.10 - 0.15%.
In designing the bypass for this plant it uas assumed that both retention of chloride in the clinker and loss up the stack uould be negligible.

As a consequence it was predicted (Figure 10) that a kiln

bypass of some 8 - 10% would be required to lower the concentration of

chloride in the stage 4 feed to an acceptable level.

In practice,

however, up to 0.01% chloride can be retained in the clinker vhile a

somewhat smaller amount leaks up the stack uith the result that the
actual bypass operated is in the range 6-8X.
Volatile and mass balances have been constructed for this plant
with both fuels.

i>

These have indicated that:-

the proportions of volatlles evaporated in the buhling zoue are

similar to the other suspension preheater plants operated in the
Blue Circle Group

ii) the bypass is also effective at bleeding sulphur and al!calis from
the system
iii) changing from oil to natural gas did not affect the performance
of the bypass although some change in clinker chemistry and

cement quality and grindability as a result of the lover sulphur

input was noted.
4.3

Alkali Sypass - Plant 6

This is a 2500 tonnes per day oil fired precalciner process uhich

- 20 -

came on stream in December 1981.

The tav materials consist of a high

grade limestone, a siliceous limestone, a clay and iron oxide. The

total volatile input, as expressed as a percentage of the clinker output, i s : -

%
Total S as SO3

0.75

K20

0.75

Nat0
Cl

.0.3s
0.01

Use of these raw materials produces a mix with a relatively high

silica ratio of around 3.5 and a difficult combinability.
Consequently it was assumed when designing this plant in the late
1970's that the degree of evaporatioo which would be achieved in the
burning zone uould be at least the same as that found in our suspensiou preheater processes.
Against this background and assuming losses up the stack would be
negligible it was predicted that the bypass required to produce a lov
alkali cement would be 30% (Figure 11). At the time all the main
plant suppliers agreed with thi's prediction.
However,

since the plant has been in operation difficulties have

been encountered in producing a low al'kali cement. Close monitoring

of the process, during a series of trials carried out In conjunction
with the plant supplier, has established that while the bypass is
currently limited to only about 25% the main reason for this failure

- 21 -

to produce a low al'kali ceaent is that the degree of evaporation

a'chleved in the burning tone is lower than *was expected.

Further

investigation has indicated that the main reason for this is the very
much higher solid to gas ratio achieved ia. a precalciner'kiln

which

restricts the sueeping action of the gase&*).

On this basis (Figure 11) it is estimated that the bypass would
have to be increased to around 60% in order to produce a suitably low
al!=11 clinker.
.AI.temative

solutfoas have been considered.. Attempts to increase

the volatility in the burning zone by raising the lime saturation and
silica ratio of the feed, or by burning more fuel in the rotary kiln
rather than in the precalciner were only partially successful. Adding
chloride to the system to enhance the volatility of the alkalis or
selective quarrying of the raw materials to lower the alkali input
have been considered but are expensive. At present the Works
have found the least expensive solution is to buy In an alternative
low alkali clay and produce low al'kali cement on a campaign basis
pending modification to the bypass.
4.4 Sulphur Bypass - Plant 7
This is a 1200 tonnes per day coal fired precalciner process
which was first lit up in September 1982. The raw materials consist
of a high grade and a low grade limestone.

Both these and the fuel

employed contain appreciable quantities of sulphur *with

the result

that there Is a gross excess of total sulphate over al:kalis entering

- 22 the system.

The total volatile input, as expressed as a percentage of

clinker output is:I

Total S as SO3

2.30

K2O

1.05

Na20

0.65

Cl

0.01

These raw materials produce a mix uith a relatively lou silica

ratio and an easy combinability and it was anticipated that volatility
in the burning zone could be low.

In addition, from the form of the

sulphur in the rau materials it vas recognised during the design stage
that this might be volatillsed in the upper stages of the preheater
and escape up the stack,

Comercial decfsioas dictated that the plant

had to be ordered before the rav materials could be comprehensively

tested and consequently when sizing the bypass we decided to take the
possible conservative view that only a limited quantity of SO2 may
escape up the stack and that the volatility of the excess sulphate
over alkalis entering the burning zone as calcium sulphate could well
be dissociated to the same degree in a precalciner process as in the
suspension preheater process.

Accordingly, ue estimated (Figure 12)

that the bypass would have to be at least 30X and that cleaning facilities similar to those employed at our plant 2 may also be required.
At the same time our calculations showed that while the clinker
sulphate content would be sufficiently lowered (Figure 13) to pemit
an acceptable

gypsum addition, the allcall content of the clinker

- 23 (Figure 14) uould remain high.

Fortunately, the production of a low

al'kali cement is not required at this Works'and, although the alkalis

are mainly present as alkali sulphatcs, their effect on the cement
strength is also not critical (indeed it is beneficial) as the
majority of the cement produced at this Works is extended with an
addition of natural pozzolan.

However, the presence of alkali sulpha-

tes uas expected to lead to air setting problems necessitating

appropriate counter measures.
In general the plant supplier agreed ulth these conclusions.
Their calculations shoved that while a lover bypass would probably
suffice, a 25% bypass should be installed to cater for all eventualities and therefore they raised no objection to our proposal to
install a 30% bypass.
Ho-vet, uhen the plant was fully commissioned in November 1982
it became clear that the Win could be successfully operated wlthout
having to bleed any of the kiln's gases. Since at that time the fuel
cousumption guaranteed by the plant supplier still needed to be
checked opportunity was taken while performing this exe&se to take a
range of balance samples at different bypass settings.

The results of

this investigation are shovn by the dotted lines on Figures 12-140

The main points to emerge from this evaluation were:i>

around 50% of the total sulphide present in the rav materials

evaporated in the upper stages of the preheater and escaped up
the stack

MS

- 24 ii) the proportions of volatiles evaporated in the burning zone were

substantially lower than in any of our other plants.
As a consequence, acceptable lev.$ls of total sulphate in both the
stage 4 material and the clinker are obtained without having to resort
to bleeding any of the kiln's gases* Hovever, the level of alkali
sulphates in the clinker Is high and as expected the Works have been

forced to counter airsetting problems.

Follouing

these observations we have developed a laboratory test

method for assessing the potential for sulphur to burn off over the
temperature range likely to be encountered in the upper stages of a
preheater. 'By subjecting raw materials to this test irom a number of
sites ue have shown that this cannot be readily predicted from the
chemistry alone but is dependent on a number of factors including the
form the sulphur occurs in the raw naterials, the temperature at which
the calcareous component in the feed starts decarbonating and the
atmosphere in the preheater, which may itself be influenced by the
composition of the feed.
DISCUSSION XND CONCLUSIONS
1.

In modern, high thermal efficiency dry process kiln systems, continuity of kiln operatioa and clinker cement quality are more
susceptible to adverse effects of volatile components in the kiln
feed and fuel.

Wet, long dry and Lepol or ACL process plants are

not so sensitive.
2.

In some circumstances it has been found that such effects can be

more economically countered by selective quarrying, or by the use

- 25 of alternative raw materials or fuel, or by adjusting or mineraUsing the chemistry of the clinker, or by adopting intensive and
automated techniques for clearing build ups from the riser pipe,
rather than by resorting to a bypass.
3.

The adoption of precalciner processes has made it practicable to

employ bypasses of up to 100% of the rotary kiln exhaust gas,
compared tith a practicable limit of about 25% vith a simple
suspension preheater system.

4.

However, with a precalciner process only about 40% of the fuel

is burned in the rotary kiln itself and recent experience shovs
that the volatilisation

of alkalis and sulphates in the kiln are

lower than in normal preheater kilns. This reduces the effectiveness of the bypass where the objective is to limit the al'kali
or the sulphate content of the, clinkers.
5.

While siting a bypass to remove chloride is relatively easy, that

required to remove alkalis and sulphur Is more complicated and is
dependent on the interaction of these compounds and their volatility in both the preheater and the rotary kiln and is therefore
more difficult to size.

6.

-Although

plant data provides a number of useful reference points

from which future predictions can be based, there is also a need

for the properties of the raw materials and fuel to be e!UplOyed

to be thoroughly evaluated in the laboratory.

- 26 -

For our part in Blue Circle we have both a comprehensive modern

laboratory which can carry out such investigations and coaslderable
background experience and data from the operatioa of plants.
6.

ACKXOWLEDGEHENTS
The authors wish to thank Blue Circle Industries PLC for their
permission to publish this paper and to colleagues and Associate cornpanics vithin the Blue Circle Group for their contributions.

7.

REFERENCES
1.

Reactions associated vith the kiln gases: Volatile cycles, buildups, ring removal.
F.W. Lecher, S. Spmng C D. Opitz.
Zement-Kalk-Gips,

2.

1972, 25, (l), l-12.

Bypass systems for preheater and flash calciner kilnsN.U. Biege h L.J, Parsons .

'

Pit and Quarry, 1978, 2 (L), 91-97.

3.

Second generation precalclnlng

with bypass alternatives for

alkali control.
F.I. Kohanowski h J.L. Shy.
Proceedings of the 13th International Cement Seminar,
Chicago, 1977, 98-108.
4.

Eeductioo of alkali and sulfur content of clinker by kiln bypass

in flash calciner systei.
J. Tjarshawsky

& E.S. Porter

In Process technology of cement sanufacturing,

VDZ Kongress L977, 652-659.

- 27 5.

New cement plant in Abu-Dhabi uith precalciner 'kiln process with

100% kiln gas bypass system.
N. Xakamura

& 2. Tojo.

Clments, BiGtons, Platres, Cbaux, 1982, (7381, 268-274.

6.

Redustioo of alkali and chlor%e cycles in the suspensioa preheater

kiln.

3. Schliiter

Zement-Kalk-Gips,
7.

1972, 25, (l), 20-22

Raw meal preheater and alkali problans.

B. RItzmann.
Proceedings of the Eighth International Cement Industry Seminar,
Chicago, 1972, 39-47.

a.

Cyclic behaviour of volatile components in dry process plants for

burning cement clinker.
W. Danowski h U. Strobel.
Silikattechnik, 1977, 28, (2), 40-43.

9.

Hethod for predicting cyclic behaviour of deterious substances in

cement

'kilns.

W. Kreft.
Zement-Kalk-Gips,
10.

1982, 35, (9), 456-459.

The behaviour of sulphur in cement clinker burning.

S. Sprung.
Tonindustrle-Zeitung,

1965, 89, (5/6),

124-130.

- 23 11.

The behaviour of sulphur in the suspension preheater kiln.

H. Eatano.
Zement-Kallc-Cips, 1972, 2, (l), 18-19.

12.

Investigatioos

of the formation of rings in rotary cement kilns.

3.93. Sylla,
Zement-Kalk-Gips,
13.

1974, 27, (LO), 499-508.

Xing Formation in Rotary Cement Kilns.

D. Opitz.
Schriftenreche der Zementundustrie

14.

No. 41, 1974.

The distribution of al'kalts in Portland cenent clinker.

H.W.U. ?ollitt h A.W. Broun.
Proceedings of the Fifth Internatioaal

Symposium on the Chemistry

of Cement, Tokyo, 1968, Vol.1, Part I, 322-333.

15.

U.K. Patent No. 1498057.

16.

Use of mineralisers

to produce high strength cement.

G.R. Long.
Proceedings of the Fifth International Conference on cement
rnicroscapy.
17.

Nashville, 1983, 86-98.

Improvements in the early properties of Portland cement.

G.K. Yoir.
Proceedings of the Royal Society discussion meeting,
London, 1983, 127-138.

- 29 -

3.8.

Processing of 'kiln dust

B. Tettmar, S.R. Khor and S. Gregory
Process Technology of Cement Manufacture
WZ Koogress 1977, 658-663.

FIG 2. Calcium Langbeinite (arrowed blue) with associated alkali

sulphate (arroued red) in production clinker.

FT.G 3. Laths of calcium sulphosilicate (arrowed yellow) within dense

cal clum sulphate rich deposit from sintering zone in kiln.

FIG 4. Dark chain like droplets (arroued white) of atmospheric

moisture occurring on the surface of a polished section of a clinker
manufactured in a reducing environment.

,*i-b
- !4T.i =i
!I G
!-a.
1, cr,
+L 2
:..: iy
LLiL

u
. ,y
t

.Ti
-

:i
27
a

ail .._....
c

.; &
x.-

I-

I.

i--.-l.-.: ,.!
., . . . . . _If
II-_.
. .L.
~
I l-l. . . . . . . . . . .._.__..

!Z
-

-....... . . . . . . . . .

I-
z.- :ri
&.,_

Figure 7-f&l Consumption vs % Bypass

I_1100
73
s
3

* 1000
iii
2

/
/

.Ig

;/

/Suspension

P
51 g00

0
0
0

8
-43
A.2
3
.g 8 0 0

Precalciner Kiln
Air separate)
High Dust Loss - - - Low Dust Loss

%
cd
i-L
10

20

30

4 0 50
60
% Bypass

70

80

90

100

000000
LnLnOu3LnLo
a3a3cnm-cn
4
- 4

Figure
Kiln Volatilities of E3CI Suspension Preheater Process (No Bypass)
-$-Typical
-..
100
90

Chloride
Sulphate
(other than Alkali Sulphate)

80
n

70

QS-

Alkali
SMphate

6s 3,

Alkali S
(other than Alkali Sulphate
co .

Llr

(u
kI

20
10
0

Fiqure IO-Chloride Level 4 Stage 4 vs % Bypass in PLd 5

11
10
9
8
7
Predicted Level

- A - - - Actual Level

5
4
3
ommende
Bypass

2
1

6
10 E&pass

10

12

Figurell, -E%quivaleht NazO in Clinker (Ye) vs lo Bypass in Plant 6.

---I-

Predicted levels for suspensionprahealer

kiln volati(ities
Revised prediction using Bypass trial data

Actual levels from Bypass trial

mendat ion

40 50 .60
z!!LBBy~

80

Predicted lev& for different kiln volatibties

w - - L --- Actual iwels from Bypass trial

&commended

10

15

20

30
25
-lo Bypass
-

35

40

45

50

55

FJgure14--E4uivaient

NgzO in Clinker(%) vs % Bypass in Plant 7

2.0 f-#+zdicted IQVQ~ for different kiln volatilitibs

_c-m-- Actual IQVQIS. from Bypass trial.

1.8 I.6 -

ecommended Bypass

0.6 o-4 o-2 0

0

I
5

I
10

I
15

I
20

I
25

1
30

B
lo y p a s s

I
35

1
40

I
45

1
50

1
55

I
I
I
1
I

I. _L..
.F

In
,C
i

Blue Circle Cement

PROCESS ENGINEERING TRAINING

PROGRAM

Module 13
Section 8

Kiln Gas Bleed Considerations

KILN GAS SLEED CONSIDERATION!fj

The process under consideration wiil need a chloride bleed in order that
regular kiln operation can be maintained. With the present raw materials,
the required bleed level will be 30 to 50% of kiln gases. At this bleed
level, clinker equivalent soda levels will be very high, at 0.9 to 0.95%, but
could be reduced by 0.1% by use of a bleed of up to 75%.
Other options for clinker alkali reduction are the use of alternative raw
materials, which is also likely to reduce the bleed requirement for kiln
operation, or addition of CaCl, which will have a major effect on alkali
level but will require a larger gas bleed.
Dust loss with the bleed is likely to be 200 to 250 gnns/Nm of gas
extracted from the system but design should assume a level of 400
grms/Nma.

2.

INTRODUJCTION
Since the development of the suspension preheater based dry process for
cement manufacture, the study of the inherent cycling effects of the
potentially volatile components - which are present as minor constituents
of the raw materials and fuels - have become increasingly Important. In
the older, less thermally efficient processeS,
a natural loss of a portion
of these volatile components occurred in the waste gases, so
automatically controlling the recirculation levels within the burning
system and the level in the clinker product.
In the suspension preheater, with its greatly increased surface contact
between gas and particle and repeated separation of gas and particle, the
recovery and retention of volatilised components will be almost complete.
This leads to Increased proportions of these components in clinker and
within the kiln system; which can lead to the development of operational
problems in the pyre-processing stage due to the quantities of potentially
Where
sticky components that build-up through cyclic processes.
necessary, a portion of the kiln gas is bled off in order to remove a
portion of the volatile components and so control the levels of these
components either in clinker because of cement quality requirements or
in the kiln system because of the potential to cause blockages within the
preheater.
The minor components that are generally considered to be involved in
major volatile cycles are the fluoride, chloride, alkali and sulphur species
- although other elements do also become involved in cycles to a much
lesser degree (V, As, Pb, TL, Cd, Hg, Zn), these are not important for this
study.
1

Prime control of the volatile component cycles is performed in the design

stage of a works project through the selection of raw materials and fuels
in order to optlmfse the relative and absolute levels of the potentially
volatile components and, where necessary, the inclusion of a bleed
system. Once these factors have been defined, the voiatiles will develop
internal and external cycles during the pyre-processing stages; an internal
cycle being totally within the kiln and preheater system, whilst an
external cycle will leave the system but be returned after a time lag (for
instance with the precipitator dust). Where processing conditions are kept
steady, the cycles will continue to develop until equilibria are reached, at
which time the total amounts of volatile entering the system will be
balanced by the quantities leaving the system.
The degree of volatilisation and the ratez
established will depend on:

at which the equilibrium are

(a)

The species, their chemical forms and concentrations.

b)

The volume of gases.

(cl

The intimacy of contact between gas and solid.

(4

The vapour pressures of the salts.

(e)

Possibility of dissociation or further reaction.

(0

Rate of diffusion to and from solid/gas interfaces.

kc)

Degree of saturation of gas.

(h)

Kiln atmosphere.

(i)

Kiln

(j)

Time/temperature profile of material within the kiln.

temperatures.

Most of these Factors are to some degree inter-related and so in normal

operation the only methods available to control the degree of
volatilisation and eventual concentration in clinker and kiln system will
be the kiln internal atmosphere and temperature, and the proportion of
gas bleed from the kiln exit.
The major cycles will be internal cycles between the burning zone and
preheater where an individual cycle time of twenty to thirty minutes
would be expected. Smaller quantities of volatile components may be
involved in the longer term external cycle which develops via the
precipitator dusts and volatile colIection in the raw mill.. These would be
expected to have much longer cycle times of up to 24 hours.
Any kiln gas bleed from a suspension preheater system obviously has a
2

fuel penalty associated with it. For a standard preheater, it is generally

considered that a 30% bleed approximately represents the maximum
economically justifiable level, whilst in a precalciner the equivalent kiln
exit gas volume can be between 30 and 45% of that of a simple suspension
preheater which makes bleeds of up to 100% justifiable under certain
circumstances. Because of this factor, control of the cycles for improved
process operation is easier on a precalciner than on a suspension
preheater. Conversely control of the proportions of the volatiles in
clinker is more difficult due to the reduced volatilisation that occurs in
precalciner kilns - a consequence of factors (b) (c) (f) (i) and (j).

3.

GENERAL FUZVIEW OF PROPERTIES OF VOLATILE COMKNENTS

3.1

General

Comments

In practice, the proportion of each potentially volatile compound which

evaporates within the kiln can vary significantly depending on the species
and the type of process. Typical ranges that have been reported in the
literature are set out in Table 1, whilst Table 2 details melting and boiling
point data for the major components. In the past, empirical limits have
been proposed for the total concentrations of volatile input to a kiln
system but modern practice is to specify the concentrations that can be
- tolerated in the lower stages of a preheater. The maximum reported
ranges that are generally accepted at this point as being unlikely to cause
any operating problems are:
chloride
so,
alkalies

1.0 to 1.5%
2.5 to 4.5%
2.5 to 3.5%

It is possible to operate successfully with significantly higher levels of

individual components, however, the overall effect on kiln operation would
depend on the relative proportions of individual compounds and the effort
put to cleaning the preheater interior. The individual components are
discussed in the following sub-sections.
3.2

Fluorides
Fluorides can be found naturally in raw materials or be deliberately added
in small quantities to the raw mix. In low proportions its mineralising
action has a beneficial effect on the burning process. Generally, it has
low volatility and causeS few operational problems, although a level of
above 0.25% in clinker may lead to setting problems, particularly in
winter. Cases have been cited, however, where a hard dense build-up has
developed in preheaters where the build-up contains a high (over 1%)
proportion of fluoride.

3.3

Chlorldq
Chlorides are derived from the raw materials and the kiln fuel. The high
volatilities
of these compounds, together with the high collection
efficiency of the cyclone preheater systems, will lead to the development
of a greatly enhanced cycle. The chlorides have a high affinity for the
alkalies in general and potassium in particular. This property, together
with the high volatility, has been used in kilns (commonly on the wet
process, occasionally on the SP process) to control clinker K,O levels by
addition of CaCl, to the raw mix or fuel, which leads to loss of KC1 with
the kiln bleed or the exhaust gas from the kiln system. In the suspensfon
preheater, the volatilised material is recaptured within the system unless
a bleed is utilised between the kiln and riser duct. It is generally
considered that no more than 3% of the chloride passing from the
preheater to the kiln will leave the system with the clinker. Al*though
considerably higher levels have been noted in individual samples of
clinker, this is probably due to a push of kiln feed, or a semi-flush
situation as thermodynamic considerations indicate that no chlorideshould
pass through the burning zone. On many SP kiln systems some degree of
preheater cleaning is necessary on a regular basis and this may help to
control the chloride cycle by forcing the kiln conditions into a situation
which permits a brief increase in clinker chloride level (i.e. reduced
material temperature, increased material loading and flux level).
Small amounts of chlorides will also leave the preheater system with the
waste gas stream. Taking the total loss of chloride from the system as
between 2 and 5% of the feed to the burning zone, it would then be
expected that a circulating load of 20 to 50 times the total chloride input
could develop in a system without a kiln gas bleed.
No reports of low temperature chloride volatilisation within the preheater

have been identified.

3.4

Alkalies
The major source of alkalies will be the raw mix; notably the clay
component, although minor quantities can arise from the fuels. The
initial free alkalies will behave in one of three ways:

(1)

Remain in the material being processed and become incorporated

in the clinker constituents that are being formed. This happens to
Na,O to a greater degree than K,O.

(2)

Be converted into different compounds - chlorides, sulphates,

carbonates, hydroxides - by reaction with other constituents of the
raw mix.

(3)

Diffuse to the surface of the process material and volatilise.

In its initial state, K,O begins to volatilise over a wide range of

temperature, depending on the form of clay In which It was incorporated
but irrespective of source, it would be expected to have volatilised almost
completely at burning zone temperatures, although some may have been
at least partially stabilised by conversion to the less volatile sulphate
form within the material bed. Once volatilised it will react to form
chlorides and suiphates - chlorides preferentially - at the rear of the kiln.
These will then deposit on dust particles. Initially Na,O is less volatile
than K,O due to its higher bond energy and so a greater proportion of the
kiln feed Na,O would be expected to pass through the burning zone in
clinker without participating in the volatile cycles. Volatflised Na,O will
react with SO, and SO, to form sulphates towards the rear of the kiln and
with chloride where this species is present in excess of K,O. Where alkali
is present in excess of chloride and sulphate, alkali carbonates will be
formed. Each of these alkali compounds will deposit in a liquid state on
the surface of dust particles in the cooler zone of the kiln and lower
preheater stages and will enter the volatile cycles as the dust is separated
out in the cyclones. Direct contact onto kiln or preheater surfaces may
lead to the development of build-up. The compounds will then re-enter
the kiln where the degree of volatllfsation will depend on the species and
the kiln conditions. Volatility decreases from chloride to carbonate to
sulphate and, hence, sulphates are more likely to pass through the burning
zone. Nevertheless, the likely range of burning zone temperatures cover
the thermal area in which alkali voiatilities are likely to increase
significantly with rising temperature. In general, precalciner kilns have
significantly lower burning zone temperature than are common in other
processes and, hence, alkali sulphate volatilisation in particular is lower
in precalciners than in other processes.
3.5

SulDhur
Sulphur can enter the system in a number of forms from either fuels or
raw materials. A limited amount may evaporate in the upper preheater
stages and escape from the system in the exhaust gases. In general, SO1
and SO, can form in the high temperature areas and be transferred to the
gas phase. In the cooler areas of the kiln back-end and preheater system,
sulphates will form and re-enter the material stream. Preferentially
alkali sulphates will be produced with excess sulphate combining with free
lime or calcium carbonate and hydroxide that Is available in these areas.
Where reducing conditions exist, the equilibrium will favour the existence
In this
of SO, and SO,, so restricting the formation of sulphates.
situation loss of suiphur oxides by way of the stack may increase but
where the gas stream passes through the raw mill, the mafority of the
sulphur oxides would be expected to react with the high active surface
calcium compounds which are produced in the milling process. This will
then return to the kiln in the raw mix as part of an external cycle.
The high boIIing points of the alkali sulphates would Indicate that
relatively low levels of volatilisation would be expected. However,
dissocfation may occur, particularly under the reducing conditions which
5

can exist to some degree within the burning zone. Calcium sulphate also
has a high boiling point but is even more susceptible to dissociation, so a
higher recirculation of sulphate from this compound would be expected.
As CaSO, cannot recycle as a compound, the lime from this compound
remains in clinker as free lime - making burning more difficult - whilst
SOS is carried in the gas stream to the kiln back-end where it reacts to
form alkali or calcium sulphate.

4.

BLEED REOUIREMENT
The proposed new line is based on a precalciner process. The raw mix
contains such high levels of chloride and alkali that the system would be
inoperable without a kiln gas bleed. In addition, the alkali level in the
clinker could potentially restrict the markets open to this material.
A series of raw mix designs have been considered and associated with this
was a series of calculated material analyses based on the i3CI
volatilisation model, showing the effect of gas bleed levels between zero
to 100%. The results of these calculations are presented agafn in Table
3. ,Mixes 6 and 8 represent the proposed range of clinker chemistry and
from the material analyses these would have a chloride content in the raw
meal of approximately 0.22%. Figure 1 plots the effect of kiln gas bypass
on the chloride level in the material passing from preheater to kiln and
indicates that a bypass requirement of 33% is anticipated in order to
restrict the chloride level at this point to 1%. Figure 2 indicates the
effect of increased chloride content in the raw meal, with chloride levels
of 0.25% and 0.30% requiring 38% and 48% bypasses respectively. The
levels of chloride bleed would also control the levels of alkali and sulphate
in the kiln inlet material to acceptable levels for kiln operation. With the
level of bleed necessary for the control of chloride in order to maintain
satisfactory process operation and with the proposed raw mix, alkali
levels in clinker are likely to be about 0.95% Na, equivalent, as indicated
in Figure 3.
Such an alkali level is likely to be acceptable in many areas but may not
be acceptable in parts of the world where ASR is a significant
consideration. Lower clinker alkali levels can be achieved in one of three
ways:

(1)

Increased bypass level. Increasing the bypass level to 75% would

be expected to reduced the clinker equivalent soda level to
between 0.8 and 0.85% depending on the mfx (Figure 3). In this
case the majority of the reduction would be in the K,O
concentration due to its higher volatility.

(2)

Use of alternative raw materials. The use of an alternative raw

material to clay could significantly reduce the input of alkalies and
chloride to the process. As an example, mix 11 in Table 3
considers the effect of using 2% of a European bauxite in the mix.
6

This would reduce the cIay usage by about WY% and increase the
sandstone requirement but the overall effect on the volatile
components is to reduce the concentrations by between 25 and
40%. The immediate result of this is to reduce the kiln gas bleed
requirement for chloride control from 33% to 22%. At this lower
bleed level a clinker equivalent soda level of about 0.7% would be
anticipated, whilst high bleed levels would permit further reduction
in a similar manner to case 1 (e.g. 50% bleed 0.6% equivalent soda
- the exact effects would obviously depend on the full chemical
analysis of the material used).
(3)

Addition of CaCl when lower alkali clinker is required. This could

be used to make a separate quality clinker as needed. The chloride
would preferentially react with alkali and thus increase the
volati$ty of this species. However, this route would require that
the bleed system has the potential for operation at higher levels possibly up to 100% - during such periods of manufacture.

OTHER EFFECTS OF KILN GAS BLEEDS

As the kiln gas bleed removes gas and material at high temperature, this
obviously has a significant heat penalty on the process. A first order
estimate of the fuel penalty can be obtained from Fig. 4. The dust that
is extracted from the process has a high volatile content and so cannot be
re-used within the kiln system. Consequently any dust drawn out with the
gas bleed will effectively increase the raw meal to clinker factor and so
require increased capacity in all the stone processing and meal
preparation stages through to the kiln feed point. Recently the major
manufacturers have all been making significant efforts to reduce the
amount of dust extracted for a given gas bleed percenta e. Currently,
Q
most will design for a dust loss of 150 to 250 grms/Nm of gas bleed,
whilst users are commonly reporting figures of 150 to 400 gnns/Nm with
extremes of up to 600 grms/Nm. The effect of this range on dust loss
from the system and consequent extra raw meal requirement for the two
proposed plant sizes is shown in Table 4. It is assumed that in normal
operation the dust loss will be approximately 200 to 250 grrns/Nma but
the design parameters for raw meal preparation should allow for up to 400
grms/Nm.

AD JL/JAS

TABLE 1: VOLATWY FLANGES

(a) Primary Volatilitks

01) Specific Volatflkies

56

SO,

60-90

KS4

40-60

K,O

30-70

NaS4

40-60

Na,O

20-40

2CaSO,.Ks,

40-100

Cl, l-b

96-99

K,O In solid
solution

60-90

IO-40

Nai in solid
solution

20-40

Pb, Tl
I

Cd, V, Zn

SO-99

KC1

I O-20

NaCI

96-99

CASO,

80-100

97-99

Primary volatilities indicate reported range of volatilities irrespective of

full chemical form.
Specific volatilities indicate the reported range of volatllities
more common compounds.

for the

TABtE 2 : MELTING 6 l3OLlNG POWI- DATA

T
Melt&g Point
Boiling Point
D&XX&l
OC
Oc
OC
SoDiUM
Chloride

800

Sulphate

884

over 1200

1440 - 1465

Carbonate

850 - 854

dissociates

Hydroxide

319 - 328

1390
I

fOTA9XJ-M
Chloride

700 - 768

1407 - 1411

Sulphate

1075

1689

Zarbonate

894 - 897

bIy&oxide

320

Calcium Sulphate

1297

over 1200

disszlates

1320

1350

bJ.9r
1113
1.93

LO1

14.49
3.34
2.23
42.w
3.73

13.94
3.49
2.X
4.92
3.35

0.13
0. $1

0.14
0.5.3
0.56
0.227

22:
93.0
2.6
1

:::
43.21
35.05
0.09
0.33
cz9
96.0
2.4
1.5

w.0
0

0
0
0.49
0.7a
0.81
0.331

20
2
0.51
0.80
0.33
0.138

54
5
0.57
0.82
0.86
0.W

0.41

0.42

0.81
15.16
1.27
11.036

0.66
2.4a
1.16
1.91

m
10

0.59
0.87
0.90
0.368

0
0
0.9
O.Bc
0.36
0.34

0.43

0.45

0.39

0.40

0.47
1.33
1.W
0.67

0.43
0.87

O.Br
6.00
1.34
1.63

::2

22.2
5.12
3.42
65.60

1.23
1.38

0.41

0.42

0.Y

0.49
1.43
1.10
0.71

0.44
0.9
O.%
0.387

3.71
3.4
3. w
7.64

10
0.60

20

Y,
5

2
0.4s

0.Y

0.41
0.234

65
13
8
11

0.63

0.54

0.54

0.57

0-n

0.33

0.37
0.57
0.67
0.29

1.67
0.86
1.034

96.0

23.6

0.47
0.9

0:67
O.ZS!

f-43
L3l
j-61
L 076
Lrn
2.4

co
10

0.241

21.63
3 .

21.39
s. 35
3.56

0.25
0.59
0.U
0.75
33.0
2.6
_
1.5

z
o:a7

0
0
0.43
0.51
0.60
0.229

9
5
0.55
0.39
0.91
0.367

0.26

54
19
8
10

14.05

1-S
23.8

%
13
8
I2

8.:
0147

_--. -7
TABLE 4 : WI%CT OF DLJZED LJZVEL AND DUST LOADING
Plant output
Bleed %
I

Dust Load
grm/Nm3 .

5500 tpd
Actual Dust Los

7500 tpd

Raw Meal Equivalent

Actual Dust Loss

Raw Men1 Iquivolent

Av tpd

Max tpd

Av tpd

Max tpd

Av tpd

Max tpd

Av W

Max tpd

30

100
200
300
400
500
600

78.4
156.8
235.1
313.5
391.9
470.3

82.5
165
247.5
330
412.5
495

117.4
234.8
352.2
469.7
587.1
704.5

123.6
247.2
370.8
494.4
618
741.6

106.9
213.8
320.6
427.5
534.4
641.3

112.5
225
337.5
450
562.5
675

160.1
320.2
480.3
640.4
800.6
960.7

168.5
337.1
505.G
674.2
842.7
1011.2

50

100
200
300
400
500
600

130.6
261.3
391.9
522.5
653.1
783.8

137.5
275
412.5
550
687.5
825

195.7
391.4
587.1
782.8
978.5
1174.2

206
412
618
824
1030
1236

178.1
356.3
534.4
712.5
890.6
1068.8

187.5
375
562.5
750
937.5
1125

266.9
533.7
800.6
1067.4
1334.3
1601.1

280.9
561.8
842.7
1123.6
1404.5
1685.4

70

100
200
300
400
500
600

182.9
365.8
548.6
731.5
914.4
1097.3

192.5
385
577.5
770
962.5
1155

274
547.9
821.9
1095.9
587. I
1643.8

288.4
576.8
865.2
1153.6
618
1730.3

249.4
498.8
748.1
997.5
1246.9
1496.3

262.5
525
787.5
1050
1312.5
1575

373.6
747.2
1120.8
1494.4
1868
2241.6 J

393.3
786.5
1179.8
1573
1966.3
2359.6
-__-_

Three potential raw mixes were selected for the bypm calculation (on the basis of
kiln operation and quality).

(1)

For the situation where a bypass was in operation, the fuel consumption
wa5 assumed to rise by PKcallKg per 1% of kiln gzs bleed from the
system and 1% dust was assumed to be lost by 10% bypass in operation
(precalciner process). The fuel consurnptions and dust losses on clinker
basis were estimated as follows:

(2)

The estimated clinker analysis has been calculated for the situation where
a bleed of 0, 10, 20 50% and 100% of kfln gasa fs in operation.
When a bleed is In operation, the levels of volatiles leaving the system in
the clinker was calculated on the basis of the following assurnptiom

0)

All the recirculating voIatiIes are potentially available to leave the

system via a bleed.

(ii)

The following volatillsation

NaCl
KC1
Kfl.4
KG04

(3)

99
99
40
40

rata were assumed:

ExcessK,O
E x c e s sNa,O
Ca!Xl,
tCasOq.K&04

35
35
40
40

The total volatile input is the sum of the volatlles from the raw feed and
from the fuel expraed on a clinker basis.

WV,
aa
rr$
OA
1 .a
1..
0 .:.
. :. . .._ .
..
.:

__

cy

H
k

. .
..
._

. .
. . . ..
. . . .. . . .
.
. . . . .. . . . .. . . .. .

0
04

0
-

to

e-4

to
d

+x
6

I / i i i

,
I
I

I
i

,
4. .I

.I
;I1

,
I

I
!
i

1
1

I , I
i
:
; *
!

i
L

/ ; j / i . i ;ii i i i i ; i i i

1
r

i
.III

t
I

, ! ; I :!
t i i I; i-i
1;. ::

, 1:
11-i a

t i i I i i i
!I;

0
co

0
-3

Fig. rt.

E S T I M A T E D H E A T C O N S U M P T I O N v s *lo G A S B L E E D O F F
AT THE KILN BACK END

HEAT CONSLJt.4P1ON
()I cd/kg NETT)
950

925

9 0 0

- - -

;i;.

r
875

--

.__..

,ib l/

850

-.. -- -- ..-

---

--_--.-- -.-. --

..--. --

-.--

DUST LOSS

/=
.0
/

;./
64

SUSPENSION PREHEATF~~ K I L N
(100% F U E L IN)ECTION
I N T O T H E Kl

825

.--.e

..--

-. - P R E C A L C I N E R K I L N
--._
(LO/. FUEL INJECTION INrO THE

.fl

/
- .

IUST

*.
8 0 0 liiiclc
0
10

To

30
1.

LO
G A S BLEED

50
OFF

A T ItIE

LOSS EX KILN

*I. O N

CLINHtmH

GO

70

no

90

KILN

DACK EN0

KILN]

Blue Circle Cement

PROCESS ENGINEERING TRAINING

PROGRAM

Module 13
Section 9

Ring Formations in Cement Kilns

Ring formations in cement kilns

Greg Palmer
Plant

Superintendant,

Queensland

Cement

Ltd,

Introduction
Ring formations in cement rotary kilns have been
known to induce operational instability and restrict
gas flow to the point where production must be
stopped. Rings that can cause severe operational
problems in long wet kilns have been found in the
following areas:

Australia
process instability are caused by mechanisms 2.3
and 4.

0 electrostatic attraction

Freeze-thaw mechanisms
The freeze/thaw mechanism is used to advantage
in the burning zone of cement rotary kilns, in this
zone coating protects the refractory lining from
aggressive chemical attack and reduced the
thermal energy loss from the kiln. In fact, in the
burning zone, coating will build-up to a steady
state level of some 20 to 30cm in thickness. This is
achieved as the rotary action of a cement kiln is an
ideal environment where temperature fluctations
and tumbling feed will expose a cooler surface to
freeze any liquid.
The freeze/thawing mechanism is totally
dependent on the composition present and the
temperature in the vicinity of the solid. If the
temperature increases or the composition
changes, for example by chemical reaction, then it
is possible that the mineral system may reach the
eutectic point or congruent melting point at a
different position along the kiln axis.
The composition of the mineral system provides
valuable information as historical interpretation of
the concentration levels can be linked to various
types of rings. At a fixed temperature, the mineral
concentration determines whether or not the
eutectic or peritectic points will be reached. In an
industrial environment, it is not possible to
quantify the exact mineral system that may exist at
any one moment in a dynamic- kiln environment. Mainly because the concentrations of
fluxing and mineralising elements such as
alkalies, sulphates and chlorides are changing due
to volatilisation or condensation. Thus, as the
concentrations vary, the position of liquid
formation will also vary along the kiln axis. An
important source of alkali, sulphate and chloride is
from the electrofilter dust being returned to the
kiln. This dust can contain concentration levels
many times that of kiln feed which, and if not
monitored, can cause rapid concentration buildup.

The source of material can also be categorised as

there is a number of streams available, as shown in
Figure 1, that can effect the composition.
Thus, it can be seen that ring formation is a
dynamic process with a number of streams
affecting the composition at any point and a
number of different mechanisms which can bind
the material together. This makes any
investigation extremely difficult, as it is not
possible to hold all but one parameter constant so
that its effect can be evaluated. To overcome this
problem investigations rely upon an empirical
approach, past history and industrial observations.
In rotary kilns most ring formations that cause

The role of alkali salts and alkali aluminates

Many investigations into the role of alkalies,
sulphate and chlorides in ring formations by freeze
thawing have been undertaken and are well
documented(4).
(5). (6). (71. Choi and Glassern discuss
the role of sulphur in clinker production by
synthesising, calcium langbeinite (CaL), calcium
a l u m i n o s u l p h a t e (&A&) a n d c a l c i u m silicosulphate (C&,S) from chemical reagents. The
presence of sulphur has both beneficial and
adverse affects: it improves burnability by acting
as a fluxing agent.
Choi and Glasse@)
found that at low temperatures, below l,lOOC, the evaporation process is

Transition zone
Chain area
Kiln discbarge area
To prevent rings forming it is necessary to understand the mechanisms that assist in formation.
There are a number of mechanisms, detailed by
Rumpfc2), that can lead to ring formations, however,
there are many factors. that contribute or aid a
mechanism. Due to this complexity it is not
possible to study each factor in isolation. To
overcome this problem investigations rely upon an
empirical approach, past history and industrial
observations.
Using a combination of both XRF chemical
analysis and petrographic analysis, it is possible
to determine the mechanism most likely to cause a
build-up. This paper discusses transition zone and
kiln discharge rings that have caused operational
problems over the last few years. The investigation
found that the rings were formed by a combination
of combustion thermodynamics and freeze/thaw
mechanisms. There was no evidence of any alkali
sulphate or sulphate spurrite induced rings.
Mechanism of ring formation
The main mechanisms for rotary cement kiln rings
were categorised by Rumpfm into the following
groups:
0 melting or softening of the surface due to
friction
0 melting or freezing from the loss or input of heat
0 interlocking of particles that have been built-up
by particles held together by surface forces
0 interlockirg of fine needle like particles

WORLD CEMENT DECEMBER

1590

dominated by alkali sulphates. However, this is in

direct contradiction to Holderbank investigations
where the low temperature evaporation process is
dominated by alkali chlorides. However, the latter
process is more in line with what is observed in an
industrial process.
Alkali sulphates and chlorides have possibly the
lowest eutectic point in the clinkering system:
K2S04-Na2S04
binary system has a eutectic point
of less than 900%. The source of alkali and
sulphates is either from volatilisation from the feed
entering the burning zone or a sulphur rich fuel. As
previously mentioned, from 1 ,lOO% upwards, the
volatilisation process is dominated by alkali
sulphates and chlorides. Since the only
mechanism of bleeding the system in a long wet
kiln of alkali sulphate or chlorides is by dust loss or
substition in the clinker phases, it is easy to see
how concentration can quickly reach critical
levels.
As the alkali salts are transported, via the gas
stream towards the feed end of the kiln the dew
point temperature will be reached allowing
condensation of alkali sulphates. In the zone
where alkali sulphates can remain liquid its low
surface tension will ensure that it wets the clinker
crystals present. The identification of alkali
sulphates can be done both chemically and microscopically. If sulphate is a major mechanism, then
it would be expected that the percentage SOS,
Na,O or K20 found in a rin,g would be several times
the amount found in clinker. Also, if a sample is
prepared for microscopical observation, then alkali
sulphates can be high-lighted by potassium
hydroxide etching, which makes the identification

reasonably straightforward.
The ferrite and aluminate phases, C4AF and &A,
and alkali aluminates must be considered in ring
formation as they are a significant component in
the clinker matrix, accounting for around 17-20 per
cent of the total composition. With an alumina
modulus of 1.38 in a pure &A-C4AF mix, the
eutectic point will be reached at 1338OC. However,
substitution of Na+l and K+l ions can occur in the
ferrite and aluminate phases, but this does not
necessarily mean that the eutectic point will
decrease. In fact Bogue(16j reports for the ternary
system K20-CaO-A1203
the eutectic point is above
1300%. The KA-CdAF-C2F
system was similar
showing a eutectic point, again above 1300C. A
study of the different systems in which Na+l could
be susbtituted into &A or C4AF similarly showed
that the eutectic point was about the same level as
if K-7 w a s s u b s t i t u t e d . S o e v e n t h o u g h t h e
alumina and the ferrite components will determine
the potential liquid phase, it is the presence of
sulphate and chlorides which will determine the
eutectic point or congruent melting point at the
lower temperature range in a cement kiln.
Spurrite formation
Spurrite formation should also be considered at
this stage as it has long been associated with ring
formations. Spurrite or sulphate spurrite has been
recognized as a contributor in ring formation by the
intergrowth or matting of the distinct prism shaped
crystals, The formation of spurrite Ca; (SiO&
C03, has been extensively studied;(l).
(lo). (l) and
recently Bolio-Arced and Glasser@b investigated
the role of mineralisers
H20, F a n d C l i n t h e

Table I The Chemical analysis of the different rings

Kiln No

Location (m)
Sample
Date

21m

Om

30m

Ring
2611189

Ring
717189

Nose Ring
1 o/5/89

3Om-36m
Clinker
2116l89

Ring
27111185

Na,O

0.7

0.70

1.5

0.7

0.53

K20

0.11

0.14

0.5

0.11

0.10

so3

0.53

0.55

2.60

0.57

0.77

p205

0.06

0.11

0.10

0.12

0.12

LOI

0.32

0.71

3.22

0.56

0.32

SR

2.67

2.63

2.29

3.52

2.56

AR

1.65

0.86

1.21

1.46

1.41

LSF

94.4

83.3

% Liquid (Lea)
1338C AR> 1.38

21.3%

98.4

r.

95.7

81.2

18.1%

26.4%

19.1

27.1

% Liquid (Lea)
1338=X AR < 1.38
14cvc

24.3

I+;

1::

Fly Ash

Flaw Meal

Coal Ash
lnsutflated

oust

Alng
Clinker Oust

e
Combustion Gas
VolatIlizeed Elements

Figure 7. The source of material can also be catagorised as

there is a number of streams available that can effect the
composition.

formation of spurrite. A number of reaction

mechanisms have been proposed for the formation
of spurrite with the reaction frequently beginning
with either C&S or C$S.
It is generally accepted that spurrite is not
- stable above 900%. Thus, if the alite mechanism is
responsible for spurrite formation, then the alite
must be transported to the cooler region in the gas
stream. For alite crystals transported this way, it is
only a matter of time before the C3S will react with
CO2 or SO3 to form spurrite. Bolio-Arceo a n d
Glassero)
also studied the mechanism of spurrite
f o r m a t i o n f r o m C2S with CaF2 a n d CaClz as
mineralisers in the reaction species. It was found
that at least 0.2% CaC12 or CaFl was required. The
role of spurrite in ring formation can easily be
misunderstood, particularly if CaFz is available, for
with these compounds present, a liquid phase
would be produced, raising the question, Which
comes first - the interlocking of spurrite crystals

or the adhesion by freeze/thawing? Investigations

conducted at Holderbank suggest that ring
formation is a combination of solidification
binding the spurrite crystals which further assists
with crystals growth.
Regardless if spurrite is suspected of being the
binding mechanism, then microscopic examination will easily identify these crystals by their long
prismatic, needle-like shape. However, it must be
pointed out that investigations(l) have shown that.
the concentration of spurrite can vary radially
through a ring and, at or near the hot surface the
concentration of spurrite can be very small. It
would therefore be recommended that a radical
cross-section of a ring be sampled when
examining for spurrite.
Ash rings
Ash rings are worth mentioning briefly as they have
been historically associated with ring formations.
In coal fired kilns ash rings have been reported to
form in the transition zone and cooler inlet areas.
These rings are formed as liquified ash, from the
flame, settles on the surface of clinker crystals
c r e a t i n g a r e a s w i t h h i g h SiO;, a n d l o w CaO
concentrations. In these areas only belite can be
formed. Holderbank has reported that the chemical
composition of these rings is similar to ordinary
clinker. Microscopically, these rings are very
porous with both ash and clinker particles
aggerating to form particles in the range of 100pm
to 350pm. Cooling by rotation of the kiln or from
colder secondary air will ensure solidification of
the liquid phase. The presence of belite streaks
caused by the welding together of the crystals by
ash drops is commonly associated with these
rings.

SF7 Clinker.

c.5 CIinker

Figure 2. SiO,-Al,O,-Fe,O,

Ternary diagram.

industrial observations and interpretations

Rings forming in Kiln 6 have, on occasions, caused

severe operating problems to the point where it
has been necessary to stop production. The rings
were forming in the upper transition zone and at
the kiln discharge (nose ring). It was initially felt
that rings forming in the transition zone were
sulphate rings caused by excessive SOs being
returned in the electrofilter dust. While the nose
ring formations were caused by dusting, as the
clinker cooled by secondary air at approximately
500C, it would transform belite from the beta to
gamma type. The dilatation of the belite crystal as
it undergoes this transformation will cause
disintegration of the crystal. However, the gamma
belite transformation is uncommon as a high LSF
will cause a slight excess of lime to remain in
belite which acts as stabiliser.
Extensive investigations sing both XRF
analysis and petrographic examination, were
carried out on the rings and the following
discussion shows that a completely different
mechanism was responsible.
The chemical analysis of the different rings is
shown in Table 1, while Figure 3 and Figure 10 form
part of the petrographic examination.

Figure 3. K6 Ring 3Q36m.

Transition zone rings

The chemical analysis of the K6 ring at the 30-36m

mark on 26 January, 1989, shows no abnormal
levels of alakali, SO3 or Cl, in fact the concentrations are similar to normal clinker. The silica
and alumina ratio are 2.63 and 0.86, respectively.
While the silica modulus is normal for ordinary
clinker the alumina modulus is very low. Plotting
the alumina and silica modulii on the ternary
diagram, Figure 2, shows that the composition has
a tendency to form rings; however, this tendency is
only slightly more than normal SR clinker. While
SR clinker is known to form a thicker coating in the
burning zone, it could not be said that unwanted
ring formation was associated with this type of
clinker production. The percentage liquid phase
present in the sample, as determined by Lea(ln* is
low and calculated at 12.4 per cent. Even though
the sample has a low percentage liquid the high
iron content means that the matrix viscosity will be
low and effectively wet any surrounding particles.
Petrographic examination is shown in Figures 3
and 4; two different sections of the ring at different
magnifications. In this section the belite has been
etched blue, alite is etched brown and the closed
pores can be seen as black areas. The very dense
nature of the ring is evident by the close packing
structure of the different phases and a very low
open porosity. Figure 4 shows that the matrix is
ferrite rich with the ferrite phase being identified
by its unetched light grey appearance. There is no
sign of any alkali sulphate and as the ferrite phase
has a low viscosity it is the most likely binding
agent in this case. The freeze/thawing of the kiln
feed by the tumbling motion is further enhanced in
this area by the increases or decreases in the heat

Figure 4. K6 Ring 3036m.

Figure 5. K6 Ring 27m.

flux profile due to combustion thermodynamics.

Mathematical modelling of the heat flux profile
along the axial length of the kiln shows an area
from 20m to 35m where, given the right chemical
composition, the eutectic point could easily be
reached as a result of a changing temperature
profile. The heat flux profile is that of a long lazy
flame with the flame temperature dropping from
1460C, at 20m to 129OOC at 26m then increasing
again-to 13OOOC at 29m. Thus, the most likely
mechanism for ring formation is an increasing
flame temperature combined with a higher
percentage of liquid ferrite phase. The liquid phase
would be solidified by the tumbling kiln charge.

Figure 6. K6 Ring

Figure 7. K5

21m.

Ring 30m.

Figure 8. K5 Ring 30m.

The chemical composition of the K6 ring at the

21m mark on 7 July, 1989, has a silica and alumina
ratio of 2.29 and 1.21 respectively. Plotting these
modulii on the ternary diagram, figure 2, shows
that its composition has the greatest tendency to
form a ring. The percentage liquid phase, approximately 23.4 per cent, is higher than the ring formed
at the 30-36m mark, but only marginally higher than
normal clinker at 1,338OC. The alkali content,
KzO(0.5%), NazO(1.5%) and S03(2.60%) are higher
than typically found in clinker. liowever, even at
these concentrations it would be unlikely that
alkali sulphates could be the binding mechanism.
Petrographic examination shows that the ring is

very dense with almost no porosity. Also, as would

be expected for material in this section there was
almost no alite with belite being the major phase
present. The belite crystals varied in size from 10 to
50 pm; the large crystals would suggest that a
temperature around 1500C was reached. The
belite structure was internally disorganised, with
what appears to be alkali sulphate on or near the
surface. Figure 5 shows a section of the ring with a
band of idiomorphic periclase, round greyish
crystals, which is indicative of the ring being in
contact with a dolomite lining. Associated with the
periclase is a large amount of free lime. The matrix
is predominately ferrite with a small amount of
coarse aluminate. Figure 6 is a higher magnification view of the sample further away from the
periclase band. The shape and internal
disorganisation of the belite make it impossible to
interpret; however, it is again most likely that the
binding phase is the ferrite rich matrix. The heat
flux profile reduces significantly from 20m to 25m,
thus any small changes in the axial position of the
feed would expose the feed to large changes in the
surface temperature. As mentioned earlier the
flame temperature changes from 1460C at 20m to
1290C at 26m, but it must be remembered that
these predictations are based on a mathematical
model and in the real environment the combustion
thermodynamics could shift the heat flux profile in
any direction along the kiln axis. Again, it would
appear that a varying heat flux profile in the region
of 20m to 27m is inducing ring formation by a
freeze/thaw mechanism of a ferrite rich matrix.
The chemical analysis of the K5 ring gives an
alumina and silica modulii of 1.41 and 2.56,
respectively. Plotting these values on the ternary
diagram, Figure 2, shows that the ring formation
tendency is not all that dissimilar to ordinary
clinker. Petrographic examination of the ring, as
shown in Figures 7 and 8, reveal a very dense
structure with some closed pores (black areas) and
even less open pores (mat grey in colour). The most
evident feature in Figure 7 is the nesting of belite,
blue etched crystals, which could be attributed to
an inhomogeneous feed. Figure 8 shows a higher
magnified view of the ring. The alite crystals,
etched brown, have a maximum size of
approximately 50pm but many small alite crystals
are also present. The matrix is predominately
ferrite rich, particulariy in areas of belite clusters.
Some aluminate can be seen as a slightly darker
mat grey colour. Similarly, the mathematical
modelling of the heat flux profile in this kiln is very
similar to Kiln 6 with an increase in the flame
temperature around the 30m mark. Thus, it would
appear that the same mechanism is responsible
for the rings. That is varying flame temperature
assisting a high liquid content which is being
freezed by a tumbling kiln charge.
Nose ringsample
The sample labelled nose ring, 10 May 89, has a
chemical composition which is very similar to
ordinary clinker. The percentage sulphate is the
only minor constitutent that could be considered
slightly high. The silica and alumina modulii are
3.52 and 1.46, respectively and plotting these

proposed mechanism in this case would be a ver)

hot burning zone combined with a hot cooling zone
so that as the clinker leaves the kiln, any clinker
dust re-entering the kiln will adhere to a mobile
matrix phase and solidify on contact or as it comes
into contact with coler secondary air.

Figure 9. K6 Nose Ring.

Conclusions
It is concluded that ring formation, in Kiln 6, is
caused by freeze/thawing brought on by a
combination of combustion thermodynamics and
a low alumina ratio slurry. Mathematical modelling
of the kiln heat flux has shown that the flame
appears to be long and lazy with an increase in the
flame temperature around the 30m mark. The
chemical composition has also shown that the
alumina modulus is consistently low allowing for a
greater percentage of liquid phase to be present at
low temperatures and that it is very near the
eutectic point of the &A-C4AF system. It is also
possible that an inhomogeneous feed may further
exacerbate the problem.
There was no evidence to suggest that alkali
sulphates or chlorides formed part of the binding
mechanism.
References

figure 10. K6 Nose Ring.

values on the ternary diagram, Figure 2, shows that

the ring composition has the least tendency of all
the samples to form a ring. Petrographic
examination of this sample, is perhaps the most
interesting, and typical views are shown in Figures
9 and 10. The sample has a very open porous
framework of crystals cemented together by an
interstitial matrix of fine aluminate and ferrite
phases, as shown in Figure 9. The major phase
present is an idiomorphic alite with a size ranging
from 240pm to 100pm. The larger aiite shows signs
of decomposition with cracking and secondary
belite formation, see Figure 10. The belite crystals
on the other hand appear mainly as clusters with
crystal size ranging from 10 to 50pm. The etching
of belite shows areas of different reactivity which
is believed to be caused by alkali sulphate on or
near the crystal surface(15,.
Both the size of the alite and belite crystals
would suggest that a sintering temperature of at
least 1500% was reached. In addition, the alite
decomposition would suggest that the cooling
rate, between the burning zone and the kiln
discharge, was too slow. Thus, if the kiln cooling
zone was relatively long and hot then hot clinker
with a mobile matrix will crystallize as it meets the
cooler secondary air as it leaves the kiln. The fact
that solidification take:s ulace as the material
discharges from the kiln is evident by the rapid
cooling of the matrix. Petrographic examination of
the matrix shows it is composed of evenly
distributed fine ferrite and aluminate crystals. The
WORLD CEMENT DECEMBER 1990

(1) OPITZ, D - Ring and Coating Formation in Cement Kilns

S c h r i f f e n r e i h e d e r Z e m e n t i n d u s t r i e , H:41/1974. V e r e i n d
Deutscher Zementwerks EV Dusseldorf.
(2) RUMPF. H - Prooerties.
Bindino. Mechanisms and Strenoth
of Agglomerates Aufberei?ungs-Tecknik. No 3. 1970.

(3) LONG, G R - Deleterious Raw Materials and Their Effects on

Clinker Burnability. Proceedings of the fifth fnternationaf
Conference on Cement Microscopy March 14.17,1983. Nashville
Tennessee USA.
(4) SPRUNG, S - influence of Process Technology on Cement
Properties. Translation ZKG No 10 ~577 1986.
(5) CHOI GANG-SOON and GLASSER. F P -The Sulphur Cycle
in Cement Kilns: Vapour Pressures and Solid-Phase Stability of
the Suiphur
Phases. Cement and Concrete Research V78 ~367
1988.
(6) SYLLA, H M - Untersuchungen zur Bildung
von Ansatzringer
in Zementdrehofen. Zemenf-Kalk-Gips No 10 ~499 1974.
(7) STRUNG, J. KNOFEL. 0. DREIZLER. S, DREIZLER. I, and
BERGISCHGLADBACK. Influence of Alkalies and Sulphur on
the Properties of Cement Parts I. II and Ill. Zemenl-Kafk-Gips No
5 ~130 1985.
(8) MACGREGOR MILLER. F - Dustv Clinker and Grindabilitv
Problems. Rock Products ~152 April i98O.
(9) BOLIO-ARCEO. H and GLASSER. F P - Formation of
Spurrite
Cae(SiO&COJ. Cemenf and Concrete Research V20(2)
p301 1990.
(10) SYLtA, H M - Investigations on the Formation of Rings in
Rotary Cement Kilns. Zement-Kalk-Gips (10) p499 1974.
(11) BECKER. F and SCHRAMLI. W - Build-up of Rings Caused
by Spurrite
Formation. Cement and lime Manufacrure V42(9)
p91 1969.
(12) HOFMANNER, F - Microstructure of Portland Cement
C l i n k e r . H o l d e r b a n k Managemenf a n d C o n s u l t i n g L t d .
Holderbank
Switzerland
1973.
(13) IMLACH. J A - Determination of the Cause of Ring
Formation in Kiln No 5. Darra. Holderbank Management and
Consultina Ltd. Reoort No MA8613355iE.
(14) IMLACH. J A 1 Analysis of Cause of Ring formation in Kiln
6 at 98-121 Feet from Kiln Outlet. Holderbank Management and
Consultino Ltd. Reoort No MA8913613lE.
(15) IMLACH: J A and MISTELI. 8 - Texture Evaluation of Ring
and Slab Samples Taken from Kilns 5 and 6 at Darra.
#Holderbank M a n a g e m e n t a n d C o n s u l t i n g Lfd. Report No
MA89l758lE.
(16) BOGUE. R H - The Chemistry of Portland Cement.
Reihhold Publishing Corp. 1965.
(17) LEA, F M - The Chemistry of Cement and Concrete 3rd
edition. Edward Arnold (Publishiers) Ltd. 1983.

Blue Circle Cement

PROCESS ENGINEERING TRAINING

PROGRAM

Module 13
Section 10

Kiln Build-Up Meeting

KILN BUILD-UP MEETING

5 May 1998 - Fairfield House, Stanhope

Present:

P.Greeno
J. Collinson
D.Belemy
A.Bainbridge
K.Atkinson

M.Mutter
L.Evans
A.Edwards

J.Wilson
S.Dodd

A meeting was organised to review the causes of kiln build-ups at Weardale, Cookstown and Dunbar
works. Each works has specific build-ups and the aim of the meeting was to review any commonality
in the build-ups and methods for reducing the frequency of stops. A review of each works past and
present problems was the basis of the meeting.

Weardale Works

Three areas of build-up appear at Weardale

(iii)

Lepol roof and kiln inlet chute, with material often falling off the roof and into the kiln.
Kiln back-end, often a long build-up of around 20 feet long and up to 2 foot thick. Feed can
often be held up in this area causing problems with overloading of the riddlings system.
Back of the burning zone - this. material can often be reached and removed by water jetting.

The works are currently using open cast coal with a sulphur level of 0.9 - 1 %, and around 30 %
petcoke with a sulphur level of around 5 %. Recently water has been injected into the flame at around 5
litres per minute, which has had the benefit of tightening the flame; The water injection has also had
the effect of softening and reducing some of the build-up in the burning zone. It is proposed to increase
the water addition rate progressively to attempt to remove build-up further back in the burning zone.
The back-end build-up is thought to be alkali-sulphate based. Sulphate to alkali molar ratio is running
around 1.5 - 1.7 on OPC and 1.7 to 2 on SRC. Slag is also being used in the raw mix, which is having a
number of effects:

(i)

(ii)

Kiln outputs have increased as the level of slag used has increased.
The slag tends not to create as much build-up.

The slag addition is reducing the overall sulphate input as it is replacing higher sulphate shale. The
works is also attempting to select the best shale to minimise sulphate input.
Additional problems are experienced with the production of SRC, where nodule breakdown on the
grate is an issue leading to a dusty kiln and increased bed blinding. On occasions the kiln can build up
in 48 hours on an SRC run.

Cookstown Works

Cookstown reported that whilst in the past many similar areas of build-up occurred, steps have been
taken to reduce their formation. Since 1992 the cyclone dust has been removed from the system and
mixed with the CKD. This material is then transferred to the cement mills and incorporated into the
final product. It was commented that if the cyclone dust is returned to the process then the build-ups reoccur very rapidly. The benefit of this is two-fold, the first effect is to reduce bed blinding and the
second is to reduce the sticky alkali-sulphate material entering the kiln and forming a build-up.
Grate/chute build-ups still occur at Cookstown, although their effect has been minimised by the
installation of silicon carbide refractory on the roof (around half of the length of the above calciner)

and sides of the Lepol grate, and also in the chute. There are no reports of any difficulties in keeping
the silicon carbide in these areas. The grate and chute have around 14 air cannons installed although
their effectiveness has not been quantified. The chute and hearth were also modified around the same
period to allow an increased area for gas flow.
1997 saw a significant number of stops due to long, tapered build-ups through the burning zone. These
were often hard build-ups, and due to hardness and the shape of the taper, were difficult to remove by
any other method apart from stopping the kiln and digging out the build-up. These build-ups were
associated with the burner itself (a new FCT pipe) and the position of the burner. Since the reinstallation of the burner and its alignment, only two burning zone build-ups have occurred, both being
rings as opposed to tapers. The result of this is that the rings can be shot away and a kiln stop avoided.

Dunbar Works

Dunbar works manages to achieve a good burning zone coating (although sometimes a little deficient at
the nose ring). The build-up problem would appear to be around the 35 and 45-metre length from the
burner, with distinct rings being formed at these points. There is no real build-up in the riser, and there
is no riser-cleaning or air cannons. The effect of the rings is to create a dam in the kiln causing raw
meal to be held back and spill out of the back end seal.
Amongst recent solutions are the speeding of the kiln to 4 rpm to move the material away from the
seal, modification of the seal to reduce spillage and a rubbing ring to be installed later in the year. The
works have been trying to identify the cause of the rings. The stage 2 inlet oxygen is controlled at 4 %
in an attempt to reduce sulphate volatilisation. Observations are that there is a link between NOx and
spillage, and that less build-up is present if the fuel split between kiln and precalciner is kept around the
40/60 % (worse spillage at 45/55 % split). The build-up is sometimes claimed to be clinkered.
Discussion

Having identified the types and areas of build-ups in each of the kilns, it is important to classify the
types of build-up being formed. It would appear that the experiences of Dunbar could be a different
problem to those at the back of the kiln at Weardale (and those previously experienced at Cookstown).
Dunbar need to evaluate the composition of the build-up to identify whether it is due to volatile
recirculation or whether the rings are due to clinker being carried to the back of the kiln and depositing
on the sticky melt in the kiln. Such build-ups can be characterised as a dense yellow/brown material,
made up of material with fine particles. The principle of analysing the building blocks - the size of
particles in the build-ups - will be pursued by Technical Centre on samples of build-up received in the
future.
On a kiln where cooler control is poor, and fine, dusty clinker is periodically produced due to overburning, such build-ups can be expected.
In terms of sulphate control for Dunbar, much of the raw material sulphate is lost in the preheater tower
as it is present in the raw meal as sulphite. This acts as a bleed for sulphur in the system, although some
sulphate is returned to the system via the GCT dust. If the build-ups turn out to be sulphate based then
further examination of the volatile cycles is required.
Cookstown build-up is currently in the burning zone, and as previously stated, tend to be rings as
opposed to tapers. It is likely that the rings are originating from the return of the dust from the cooler
when the air cannons are fired on the IKN. The solution here would be to attempt a redesign of the inlet
around the IKN to reduce the use of the cannons and therefore the amount of dust returned to the
burning zone.
This leaves a comparison of the three Lepol kilns to review how to reduce the back-end build-ups at
Weardale. The obvious difference is that the cyclone dust at Cookstown is completely removed from
the system whereas this is not carried out at Weardale. When this method was tialled at Weardale, the

cyclone dust was removed from one kiln and returned to the other kiln. The kiln with cyclone dust, not
suprisingly, performed poorly. The kiln without cyclone dust resulted in a reduction in output with a
hotter kiln. Whilst the trial was possibly not conclusive, the removal of dust from the cyclones and
addition to the cement mill at Weardale will potentially increase the cement alkali level above the 0.6
soda equivalent level.
All of the riddlings are returned to all three kilns. Riddlings are often screened on other Lepol plants so
that the smaller fractions, which have the higher volatile contents, can be removed from the kiln system
and further reduce volatile cycles. It is proposed to trial this at Cookstown with the results being fed on
to Weardale.
Cookstown comment that removal of the CKD and cyclone dust will reduce the soda equivalent in the
cement by around 0.02. It is also commented that the amount of cyclone dust and the level of alkalis in
the clinker could reduce by bleeding off this material. In the trial carried out at Wear-dale in removing
the cyclone dust, around 2 tph material was collected, with around two thirds of the material having to
be dumped in the quarry.
The quantities and qualities of the materials around the system need to be reviewed and compared with
Cookstown, to examine whether the dust could be returned to the cement mills. This will also allow a
sulphate balance to be carried out to examine the extent of circulation around the system. The
recirculation of the volatile materials - alkalis and sulphates - is related to effects in the burning zone
such as flame impingement and reducing conditions in the flame, as well as the high sulphate to alkali
ratio. Calcium sulphate will decompose in the burning zone due to contact of the flame with the
material bed in the presence of CO, and also excessive temperatures in the burning zone. Therefore it is
important to ensure that a tight flame can be produced to avoid the flame touching the bed.
This, in principle, can be achieved by the use of a bluff body in the burner pipe. It was agreed to trial
such a device when an appropriate design is provided by the Technical Centre. The position and angle
of the burner will also be review. Both burners are in different positions at present but this has not
shown any significant differences in operation. If the burner is pushed into the kiln the secondary air
will be better entrained and produce a narrower flame.
Coal mill and oxygen control at Weardale will also be reviewed so a more stable operation in the
burning zone can be achieved. NOx control is currently used as a control parameter and variations in
this parameter, along with the SO2 at the back of the kiln can also give indicators as to the extent of the
recirculation in the system.
An action plan of areas to review has been compiled as a result of these discussions.

Process Parameters

02, NO,, S02, CO, (stack and back end).

2.

Kiln feed, temperature profile and suction profile.

3.

Secondary air temperature, coal feedrate.

4.

Cement quality
- sulphate - max 3.5%
- alkalis - max 0.6%
- slump
- early strength/reactivity
- 28 day properties

5.

Cement quality

6.

Effect on product properties of any considered

change.

7.

Alkali bleed via auxiliary stack.

8.

Kiln/cooler exhaust/dust to burning zone.

9.

Lepol best practice for dust return/build-up.

10. Secondary firing.

1.

Cookstown max speed 1.8

Weardale max speed ?
Does this affect build-up?

2.

Silicon carbide on grate areas/kiln areas bricks.

3.

Grate chute design at CKN - compare with

Weardale and hearth design.

4.

Grate

SC

Operation
- bed depth and variation (fixed speed)
- doors open/closed
- pan angles

5.

Recirculation fan - use/benefits/effects.

6.

Microphone on grate - information on bed blinding.

Raw Meal
(either for SR or OPC)
1.

Fluorspar addition
- benefits?
- where in process?

2.

Secondary r.m.s
- Shale Z usage
- Slag
- PFA (up to 3% shale replacement nodule problems) .
+ nodule strengthener to increase use
of pfa
- increased low stone usage

3.

Raw meal residue

- now
- with secondary r.m.s

4.

Chemistry of particle range in raw meal.

5.

Raw meal variability

effect
- improve
-

1.

Coal/petcoke mix - impact of fuel sulphur - water

Injection

2.

Flame

3.

Coal Residue

4.

Coal ash composition and % in coal.

5.

Fuel blending in coal store.

6.

Fuel composition - frequency of sampling.

7.

PF moisture < 1%.

8.

Coal mill control.

9.

bluff body
firing pipe position in relation to nose ring
firing pipe angle
secondary air temperature

NOx/kiln burning strategy (N.B. control of kiln with

low NOx burner?)

10.

Kiln camera - Quadtech?

11.

Shell temperature for build-up.

12.

Sulphate balance/SO2
sample.

13.

Inleaking air through system.

14.

Temperature profile through grate - comparison

between works.

15.

Removal of coal mill cyclone off cooler.

recirc/grate material SO4

1.

Quantity of riddlings.

2.

Quality and psd of riddlings.

3.

Screening - possible without breaking up nodules?

4.

Position of riddlings return - back onto grate vs kiln

chute.

5.

1 Kiln no riddlings return, other riddlings from both

kilns - high alkali clinker.

6.

Nodule strength improvement.

Cyclone Dust
1.

Quantity of dust - OPC and SR.

2.

Quality of dust
- chemistry
- particle size
- hot/cold cyclones
- efficiency of cyclones between kilns compare with CKN. Vortex finders in
cyclones.

3.

Return of dust to cement mills

- how much

4.

Cost of tipping dust? Can it be sold?

5.

Return to other kiln.

6.

Return all/proportion to noddy pans/elsewhere.

7.

Predict effects of all/partial removal.

8.

Pelletizing of dust.

9.

Intermediate fan efficiency.

Blue Circle Cement

PROCESS ENGINEERING TRAINING

PROGRAM

Module 13
Section 11

Cement Seminar- Rings, Balls, and

Build-Ups

Table of Contents
Page
1.

INTRODUCTION

2.

LOCATION OF RINGS

2.1

Classification

3.

THEORETICAL ASPECTS OF RING AND DEPOSIT

FORMATION

CHARACTERISTICS OF VARIOUS RING AND

DEPOSIT TYPES

4.
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.0
4.9

5.

Exhaust Fan Deposits

Slurry Rings
Cyclone and Grate Preheater Deposits
* Meal Ring (Calcininq Ring) in Long Kilns
Middle Rings in Large Preheater Kilns
Sinter Rings (excl. coal-ash rings)
Coal Ash Sinter Rings
Clinker Rings / Cooler Inlet Deposit
Kiln Charge Balls

METHODS

OF

REMOVAL/ELIMINATION

TABLES 1 - 7
APPENDICES I - III

5
6
7
8
9
10
10
12
12

13

RINGS, BALLS AND BUILD-UPS

1.

INTRODUCTION

Rings and deposits are accumulations of solid

materials (from the powdery kiln charge) in the
rotary or static sections of clinker production
lines. They have been encountered since the earliest
days of rotary kiln production, with each development in process technology, e.g. grate and cyclone
preheaters, grate cooler, bringing with them their
own specific type of deposit.
Rather than being of academic interest, ring and
deposit formation has an appreciable influence on
personnel
plant operations, frustrating-operations
by their impairing or even impeding production, and
annoying the company management by lowering production (and sales) and increasing production costs.
As a direct consequence of rings and deposits, the
gas and material flow through the kiln is restricted,
resulting in a reduced kiln output. Especially in
the sinter zone, the presence of rings can interfere
with combustion of the fuel and can result in
improper combustion. From time to time unstable
rings and deposits can break away leading to
blockage or mechanical damage in the cooler, or in
cyclone blockages. The partial shedding of coating
from the exhaust fan blades results in severe vibration which mostly requires a short shutdown for
complete removal. The breaking of a ring almost
always causes a flush of material into the burning
zone and a temporary loss of stable kiln operations.
The formation of deposits in cyclones results in
extra costs for the labour needed to remove the
deposits by poking. The introduction of air canons
(big blasters) provides a method for their regular
automatic removal and has been installed in Group
plants with persistant preheater blockages.
High pressure water jets may also be employed.
In the worst cases, a complete shutdown is necessary
allow entrance to the affected area and mechanical removal of the blockage with compressed air
drills. This shutdown invariably weakens the sinter
zone refractories, and accelerates the next shutdown
for rebricking.

to

2. LOCATION OF RINGS

2.1 Classification
Unwanted build-ups may be classified with regard
to the type of material from which they are
formed, either sintered or unsintered. Within
these two groups the various types can be
classified as follows:
unsintered:
---------. exhaust fan deposits
. cyclone and grate preheater deposits
. slurry or mud rings
. meal rings
sintered:
-------. middle ring
. sinter ring
. clinker ring
. "snowman" in grate cooler
. kiln charge ball
Process technological characteristics of such
build-ups e.g. kiln type, location, temperature of
gas and kiln charge can be seen in Table 1.
Material technological characteristics e.g. state
of kiln charge, enrichment in various elements,
and type of texture are summarized in Table 2.
The location of the various types of the above
rings and deposits can be seen in Fig. 1.

RINGS AND BUILD- UPS IN

DIFFERENT KILN SYSTEMS

CYCLONE
PREHEATER KILN

ASH RINGS
L/D - 5-7
/

I GRATE
PREHEATER

- KILN
-

WET KILN

COATING ON
INLET TO
COOLER

MIDDLE RINGS LARGE DRY KILNS

MEAL RINGS (LONG KILNS)
L/D - 7-13
SINTER RINGS
L/D - 2-7
CLINKER RINGS
L/D- O-2

MUD BALLS
SLURRY RING

3. THEORETICAL ASPECTS OF RING AND DEPOSIT FORMATION

Although of much practical significance, little
quantitatively based, fundamental knowledge is
available on the formation of deposits from solids
suspended in gas streams. In a qualitative way,
however, the more important features of such processes are known.
The formation of a deposit is always a dynamic process in which the factors responsible for formation
outweigh the forces of degradation. In general, the
stronger the forces of destruction, the more unlikely the chance of deposit formation, but when this
does occur, a strong, hard to remove agglomeration
is the result.
After the transport of material to the area of deposition, a definite force is required to make it
adhere to the wall. This can range in magnitude from
that caused by turbulence within the gas stream,
increasing to centrifugal forces when the stream
changes direction, to that due to mechanical
pressure. Whereas preheater deposits involve the
first two, mechanical pressure certainly plays a part
in ring formation within the rotating kiln.
The forces according to Rumpf considered to cause
deposit formation can be grouped as follows:
A

melting or softening of surface due to friction

or collision

melting or freezing due to addition or removal

of heat

interlocking of aggregates built up of finer

particles held together by surface forces

interlocking of long fibrous particles

electrostatic

attraction

The mechanisms B, C and D are the ones encountered

in kiln operations. In general, the finer the
powder, the greater the tendency towards agglomeration, and in many cases the absence of particles
under a critical size (e.g. 5 urn) ensures freedom
from deposition.

4.

CHARACTERISTICS OF VARIOUS RING AND DEPOSIT TYPES

Tables 3 and 4 contain a list of typical properties of rings and build-ups encountered within the
"Holderbank" Group plants (with full chemical analysis being provided in Appendices I, II and III).
Included are such factors as volatile element concentrations and moduli of the deposited materials.
As an indication of the texture, the size of the
pores and the particles or aggregates of particles,
from which the materials were built up, is given. In
many cases the mineralogical composition is also
given.

4.1 Exhaust Fan Deposits

In the case of kilns with pressure filter systems,
in which unfiltered dust-laden gas passes through
the exhaust gas fan, deposit formation causes
problems. These arise when the deposit falls off
one blade, and brings the rotating fan out of
balance. Deposits of up to 3.5 kg/blade are known
to occur.

Characteristic

properties:

Exhaust fan deposits, composed of the finest raw

meal particles are usually red-brown, hard and
quite brittle. They exhibit a compact layered
8%.
structure and have a very low porosity of
Their chemical and mineralogical composition is
basically that of the raw meal but often the
plate-shaped clay particles are preferentially
deposited parallel to the blade surfaces. Due to
their long stay in the system, fan deposits are
enriched in the volatile components K20, Na20,
so3. Typical values include the following:
LSF
SR
K20 + Na20
SO3

20
1.0
2.1
4.7

100
1.5
3.0 %
6.0 %

The SO3 is usually present as anhydrite (CaS04).

Bindinq mechanism:
In this case, the temperature is such that liquid
phase involvement - aqueous or molten salts - can
be ruled out.
The dust particles, because of the fan rotation,
strike the blade surfaces with a high velocity and
are so compacted. As the texture of the surface,
after even a short time in operation, possesses
undulations in the order 0.5 - 20 urn, the smaller
dust particles can be mechanically "locked-on".
Subsequent development of the deposit follows by
an identical mechanism.

4.2 Slurry Rings (including mud balls)

Characteristic

properties:

These occur in long wet kilns and are composed of

the partially dried kiln charge somewhat enriched
in alkalis and S03. They are soft and can usually
be broken - and hence prevented - by heavier
chains. The H20 content lies between 20 and 30%, a
range in which clay materials exhibit a sticky,
plastic consistency. The content of the alkalis
which greatly increases the tendency to adhesion
(influence on rheological properties) can be up to
10% K20 + Na20, and about the same level of S03.
In many cases, balls form (in addition) on the
chain links by the same mechanism. A typical
example of a mud ball is plant I in Table 4.
Bindinq mechanism:
The binding mechanism is the well-known ability of
clays to form a sticky, plastic mass when containing the correct quantity of H20, and to harden
on the further water loss. To this mechanism must
also be added the crystallization of K2SO4 solution and the further strengthening of the structure by formation of CaS04. Photo la gives an
example of such a ring.

4.3 Cyclone and Grate Preheater Deposits

Characteristics:
These deposits form on the roofs, walls, outlet
and riser pipes of cyclone preheaters, in the hot
chamber of grate preheaters, and vary considerably
in appearance and homogeneity. In general, they
have a light colour varying from cream to brown to
pink, indicating that the component particles had
not been heated higher than 1200C. In some cases,
darker zones of harder burnt material can be
observed. Depending on their place of deposition,
they range from a dense, compact, definitely
layered structure, hard to break to a porous (30%)
material with only moderate strength with less
obvious layering. The former type is typical of
cyclone cones and discharge pipes while the latter
is to be found in the transition and swirl chambers. Soft deposits can, however, also be found in
the cyclones.
From a chemical viewpoint, this deposit type
usually is characterized by a concentration of the
volatile elements in the following range:
K20 = 1 - 30%
Na20 = 0 - 2%

so3 = 1 - 35%
Cl = 1 - 25%

In some cases, therefore, deposits can occur with

no appreciable increase in concentration. Typical
analyses found for deposits are given in Table 3
and Appendix I.
The mineralogical composition of preheater deposits differs as would be expected from that of the
raw meal in that the clays are essentially decomposed, and a reaction to form intermediate
minerals has taken place. Minerals containing only
the volatile elements can also be found.
Amongst the minerals found in preheater deposits
are the following:
raw meal:

calcite
quartz

CaC03
SiO2

normal intermediate phases:

free lime
periclase
mayenite

CaOf
MgO
C12A7

typical deposit phases with low melting point:

sylvite
halite
langbeinite
arcanite

KC1
NaCl
2CaS04 . K2S04
K2S04

typical deposit phases without melt involvement:

carbonate spurrite
sulfate spurrite
anhydrite

Formation

2C2S . CaC03
2C2S . CaS04
CaS04

mechanism:

The binding substance in this deposit type is the

low melting point Na20, K20, SO3, Cl based compounds. These are molten in the kiln gas and are
deposited on the cyclone walls and pipes, or first
on dust particles which then themselves are deposited out of the gas stream in these areas.
Cooling on contact or with increasing thickness
results in an appreciable strengthening of the
originally sticky deposit. Because of the extensive duration of stay in the kiln system, a reaction takes place with gaseous CO2 and SO3,
resulting in the formation of lath shaped spurrite
and sulfate spurrite which additionally strengthen
the texture. Typical textures for unsintered, preheater and kiln inlet deposits can be seen in
photos lb - 1d.

4.4 Meal Ring (Calcining Ring) in Long Kilns

Characteristics and formation:

The meal rings, often called "calcining rings" in
long kilns, are in their properties and mechanism
of formation very similar to those of preheater
deposits in heat exchanger kilns. This is perhaps
not surprising in that both build-ups occur in the
same temperature zone. Meal rings are mostly less
troublesome than preheater deposits because often,
due to their relatively poor strength, thermal
fluctuations, kiln deformation and the action Of
the material stream, they fall off periodically
under their own weiqht. A typical example of a

4.5 Middle Rings in Large Preheater Kilns

Characteristics:
Unlike meal rings, middle "rings" are dense (fine
grained) of low porosity, very hard and seldom
fall off during operations. Although termed as a
ring they are rather more elongated, like a band,
being often some 15 - 20 m long extending from 7
to 11 diameters from the outlet, e.g. 35 - 55 m
for a 5 m Qj kiln. Unlike previous types, this
deposit is clinker-like in colour indicating it
being composed of well burnt kiln charge. Perpendicular to the direction of deposition, the fine
layered structure can be seen showing the curvature of the kiln shell.
The chemical composition of middle rings is very
similar to that of clinker. This is surprising
because considering the long duration of the stay
in the kiln, no increase in concentration of the
alkalis or SO3 takes place, and often the ring
shows lower volatile element values than for
clinker. Typical analyses of a middle
ring are given in Table 5.
The minerals found in middle rings are the clinker
minerals alite, belite, aluminate, ferrite and
free CaO, the alite having often decomposed into
microscopically mixed belite and free CaO,
resulting from the temperature at the site of the
ring being under the lower stability temperature
of alite (i.e. 1260C).

Formation

Mechanism:

The mechanism of bonding is the freezing of the

clinker alumino-ferrite melt. Due to a long cool
flame, the clinker has a tendency to be fine, and
the smallest clinker particles of 150 - 450 urn are
carried back by the flame and deposited onto the
kiln wall in a zone where temperatures of below
1250C exist. The particles immediately freeze in
place, and because the kiln charge is still fine,
it does not possess sufficient abrasive action to
remove the growing ring. The typical compact
structure of a middle ring can be seen in photo le.

4.6 Sinter Rings (excluding coal-ash rings)

Characteristics:
These occur at the beginning of the sinter zone
some 4 - 5 D from the kiln outlet. They are
greyish-black in colour, strong and are (usually)
agglomerations of small clinker pellets and
clinker dust. No layer structure is obvious
because of the presence of large pores and voids.
In general, the chemical composition is that of
the clinker with no appreciable concentration of
volatile elements.
From a mineralogical viewpoint, the normal clinker
minerals alite, belite, aluminate, ferrite and
to form
free CaO are observed, with reactions
belite and CaOfree, spurrite and
belite being
found with increasing depth in the ring, i.e.
decreasing temperature, similar to the case of
middle rings.

Bonding mechanism:
The bonding is created by the freezing of the alumino ferrite clinker liquid in the case of pure
sinter rings. This phenomenon occurs especially at
the beginning of the sinter zone, where the liquid
phase is just starting to form (approx. 1280C).
Due to the rotation of the kiln, the charge in
this zone freezes with each kiln revolution: a new
wet layer sticks on, and with time a thick deposit
builds up consisting of particles of less than 1 mm.

4.7 Coal Ash Sinter Rings

Characteristics:
In kilns fired with a high ash content coal,
sinter/coal ash rings can form at 7 - 8.5 D from
the kiln outlet. They are dense, often layered and
sometimes glassy in appearance and built up from
particles some 150 - 250 urn in size. They are
rather less dense and have larger pores and voids
than middle rings. Photo 1f gives an example of
the microstructure of such material, showing the
coal ash layers.

From the viewpoint of their chemical and mineralogical composition they are essentially similar
to clinker, exhibiting the minerals alite, belite,
aluminate, ferrite and free CaO. With decreasing
temperature (increasing ring depth) reactions to
form spurrite and calcite take place, and also the
transformation of alite -> belite + CaOf and /3beliteerbelite. Details of the chemistry and
mineralogy are given in Table 6 . No enrichment
of the volatile phases can be observed. Because of
the enrichment in coal ash, the belite content is
higher than that of the clinker, and tends to be
found in layers.
Formation

mechanism:

The bonding medium here is the sticky molten coal

ash particles and perhaps to a slight extent, the
alumino ferrite clinker liquid phase occurring by
a mechanism such as in Fig. 2 showing the
typical build up during kiln rotation.
Fig. 2:
ash layer
(sticky)

kiln charge
a)

Mechanism for ring formation

ash layer
+ sticking kiln
charge

kiln

charge

b)

ash layer /
kiln charge /
ash layer

kiln charge

c)

The molten coal ash droplets adhere to the exposed

kiln lining at a point and temperature at which
they are still partially fluid and sticky. When
this sticky layer passes under the kiln charge on
each rotation, it is assumed that a single layer
of the still very fine kiln charge adheres to it.
Because of the presence of fine crystalline alite
and the overall occurrence of liquid phase, it
must be assumed that the material temperature at
the position of the ring lies above 126O'C.

The alite crystals are very small and certainly

much smaller than those of the clinker. Because of
this, it can definitely be said that the ring is
not formed from clinker dust blown back down the
kiln.

4.8 Clinker Rings / Cooler Inlet Deposit (snowman)

Such rings and deposits are formed from normal
size clinker granules and have a high porosity
containing many voids. They are usually not
troublesome to kiln operations as they can easily
be removed. Their composition and mineralogy is
identical to clinker, but in some cases, rings of
up to 3.5% K20 and 3.0% SO.3 have been observed.
The mechanism of bonding is the freezing of the
clinker liquid phase as the clinker passes
through the cooling zone (ring) or on falling down
the chute into a grate cooler, grate kilns being
usually operated so as to have no cooling zone
within the kiln itself.

4.9 Kiln Charge Balls

Kiln balls occur in cases where a tendency to meal
or sinter ring already exists and can be up to 1 m
in diameter. The chemical composition is, thus, an
important factor. They are usually found upstream
of meal or sinter rings.
They are usually made up of already calcined
material and can have a porosity of up to 558,
consisting of many fine pores. Often they consist
of a hard dense porous core, surrounded by the
majority of porous material. The core usually is a
piece of coating from say the lower heat exchangers
or the transition chamber, and often has a composition different from the kiln charge in the
area of formation. Differences in composition can
be seen in Table 7.

The mechanism of meal ball formation can be due to

either, or a combination of the following factors:
- stripping and subsequent "balling" of old,
excess coating
- agglomeration enhanced by available clinker
and/or salt melt
- ring section acts as a dam, retaining "pieces"
of material for long periods. Radial growth of
the pieces occurs by compaction and adherence
of fresh surface due to continual rolling
action of the pieces/balls over the charge.
In most cases, no liquid phase participation in
sufficient quantities is possible so that the
balls behave like a snowball and by their own
pressure material sticks to the surface. This
mechanism is similar to that of deposition on the
exhaust fan blades.

5.

METHODS

OF

REMOVAL/ELIMINATION

An important prerequisite for minimizing the tendency to form objectionable coatings and rings, is
stable kiln operation. This applies to the composition, fineness and feed rate of the raw material
and fuel, and burning zone heat control.
The tendency to form coatings in the kiln is reduced
by lowering the dust load of the kiln gas.
Objectionable coatings and rings which are formed as
a consequence of high concentrations of various circulating elements can be obviated by appropriate
reduction of the cycles in question.
This can be achieved by:
- Employing different raw materials and/or fuel
with lower concentration of the offending
element. This is generally not practicable.
- Control of the raw meal milling so as to reduce
the concentration of the very fine particles of
sizes under 20 urn.

- Intervention into cyclic

discarding dust in which
have become concentrated,
bypass installation which
of the kiln gas.

process by either
the circulating elements
or by means of a
extracts a portion

The penetration of false air into the preheater and

kiln inlet chambers should be avoided, as such cold
areas will act as sites for preferential build-up.
In order to reduce the tendency to form sinter
rings, it is in the first place necessary to reduce
the proportion of fusible matter in the clinker,
i.e. the lime standard and silica modulus should be
increased. "Coating-inactive" bricks have also
proved successful in certain cases, in reducing the
tendency toward sinter ring formation.
In coal-fired kilns a coal with a normal ash content
should be employed as coals having ash contents of
40% are characterized by a very high tendency to
ring formation. No general approach can be given to
the effectiveness of other measures, e.g. alteration
to firing conditions, as these represent variables
which are peculiar to the particular installation.
Clinker rings can be avoided by shifting the flame
further back, thus increasing the clinker temperature at the kiln outlet. As a result of this,
however, the "stickiness" range of the clinker is
shifted towards the cooler inlet. Coatings can then
be formed on the cooler inlet chute. This is particularly problematic with satellite coolers. In
instances where this occurs with grate coolers,
these coatings can be eliminated by the use of water
cooled plates on the inlet chute.

Table 1:

r' AND LOCATION OF RINGS AND COATING

(according to Opftr, 1974)

kiln

type

coatfng
slurry

type(s)

all
rlng

long

wet

location

temperature ('C)
charge
gas

exhaust
blades

fan

128/180

drying

zone

150/300

< 100

700/1100

700/800

coatfng

dry
(preheater)

stage 3 & 4

coating

Lepol

walls/roof of
hot chamber

grate

coating

long dry

Inlet

meal ring
middle ring

preheater
long dry
wet

calcining

kiln
ball

charge

slnter

chamber
zone
II
#I

calclnfng
zone

100

lOOO/llOO

1100/1400
41400
41400
<1400

---

"("%8""
<1200
<1200

all

beglnnlng of
sinter zone

1400/1600

1250/1350

slnter zone
coatfng

all

slnter

1600/1800

1350/l450

clinker

all

end of slnter
zone

coating

ring

all

1)

1000/l

rfng

grate cooler
entrance

zone

800/1600

600/800

1200/1400

1200

Table

Type

2:

GKNPRAL

MATERIAL CHARACTERISTICS

Location

Chbractar
Base Material

OF VARIOUS RING TYPES

Enriched in

Texture

-----

:
n

Raw
Heal

Coating

all

EXh8U8t
Blade8

fan

Slurry ring

NL

Drying

Lone

Int.
Med.

Clinker

Ne,K

Packing

'P8rtlcle'
8ire

den8e

fine

denee

fine

(du8t

Coating

DS

S t a g e Ji4

w/
P

Halls/roof of

Coating

DL

Inlet

chamber

Heal ring
Hiddle
ring

WL
DS

Calclnfng
.

Coating

hot chamber

DL

zone

t
t

pOr0ll*

fin8

poKOU8

fine

t
-

porou8

fine

denee
I
.

flne

porou8

fine

t
+

l
.

-Kiln charge
ball

ALL

Calcining zone

Sinter ring

ALL

Beginning of
8inter xone

poroun

large

Sinter xone
coating

ALL

Sinter

I pcwOU8

large

Cl lnker

ALL

End of sinter

prwOU8

large

PorOU8

large

t*

tone

ring

Lone

Coeting

Grat8

COOl cr

t
l

coating center

lrblt 3 :

alrnl

Volrtllt Eltatnts

.ocrtla

Alktll
Sulphul
RJtio

K20

PROPERTIES

ModlIlI
SR

An

ff

BUILD-UPS

EXAMINED

Gtnerrl tppttrrnct
LSF

Strtngtl

Colour

Ttrturt

Colour

Htntrtlogy

,f strtrks

Ltytr

SwpltI

sire (jIltI

Pore

Bld.lJnlts
fpd

I - 10

.I - 10

0.16

1.76

2.71

62

stron9

Yellw

rtdlbrwn

0.36

2.05

2.77

66

strong

ytllw

rtd/brwn

0.85

2.25

3.44

124

WJk

yellow

none

0.02

2.8

2.5

149

strong

allw/
irwn

dtrL brown

0.67

2.4

1.69

339

strong

yellow

none

0.19

0.38

3.33

1.70

180

strong

*td/brom

none

3.6

0.07

1.46

1.45

196

28.0

1164.91

3.01

2.26

103

mtdlum

brown

blttlgrttn

ports
fllltd

KCI-200
chrvgt 1-5

23.6

2.31

3.2

2.6

104

wrk

brown

none

ports
fllltd

KC1 cont.
chrrgt 2-3

4.66

2.62

2.36

69

tllwl

none

ports
filled

KC1 cont.
chrrgt O-5

0.08

1.61

1.59

157

1st cyc

7.6

Ind cyc

11.4

0.16

lrd cyc
" .

4.6

0.09

1.8

1.3

0.65

1.09

Ith c c
'Pjpt r

0.8

' (Cone
.

1.6
24.0
2.56
36.0

dtnse
porous

dtnst
porous

User
'lpt

0.61

31.5

0.16

3.6

dense
portus

strong

i rwn

VI- 5
I - 10

Cl- 3
2- 6

10 - 20

10 - 20

1: : 2:

2: : 5:

ma or: crlcltt,
-i-K
su p at spurrite, KC1
minor: s urTltc,Cr Ii t

"

4.0

:fln
nltt
.

6.4

0.19

1.6

1.1

10.2

0.27

2.5

1.3

196

strong

none

1.3

0.69

3.2

2.13

114

WtJk

none

4.6

1.65

2.09

5.69

62

weak

none

porars

6tnsr

porous
.

6.9

0.34

17.6

1.48

0.68

75

strong

,rd/broua

blrck

dtnst
porous

IO - 1s

2: : 3:
>l
100

5 - 10

20 - 30
Iu.KCl cant
20 - 30

4
I E0

Table 4 :

Plant

(*let,
(bitt)

Ring
TYPL

K20

ma20

MUd

110 a

4.1

0.40

Heal

100 a

2.18

0.16

Ball

hlkrll
Sulphur
Ratio

Volrtlla Elements

.ocrtlon

SO3

0.14

Middle

45

0.16

0.06

0.01

0.13

1.56

Middle

37 a

0.12

0.07

0.05

0.71

0.84

1.01

0.24

1.01

1.14

0.34

0.03

0.21

1.21

Ash/
Slnter
Ash/
Slnter
N

Ash/
Slnter

32.5 8

18.5 a

0.95

0.20

nodull
SR

0.44
14.2

0.20

1.56

0.56

PROPERTIES ff RIMiS EXAMINED

2.36

2.13

Generrl appearance

AH

LSF

Strength

1.40

110

strong

1.46

10

strong

Colour

1.6

92

strong

yellow/
brown

2.32

00

rtrong

yellow/
brown I

01

2.4

2.88

79

a0

strong

strong

Hlnerrlogy

brow/
9w

Texture
Pore
She (pm)

Bld.Unlts
fpd

2- 3

2- 5

He or - Sulphrte
d-purrlte

100

50 - 100

#(nor - C2S, C4AF,

mite. vrrlous
sulphrto

100

50 - 100

g;; c25, C3A. C4AF

150 - 250

150 - 450

50

50 - 100

Et;; CZS, C3A* C4AF

200

100 - 200

C3S. C S, C3A, C&Of

spurrl t e

100 - 150

200

Et&

200 - 500

150 - 300

$5, C3A, C4Af

I EP

TABLE 5 :

ring
internal

sample

Loss

Typical analyses of
at various depths

on iqn.

0.47

ring
middle
1.82

middle

ring
external
0.94

ring

clinker

0.37

Si02

23.0

17.8

21.9

21.9

Al203

4.9

5.1

5.5

5.6

Fe203

3.0

2.9 *-.

3.1

3.2

66.8

66.1

66.4

70.7

1.1

1.0

0.95

0.95

so3

0.13

0.07

0.27

0.48

K20

0.16

0.09

0.30

0.35

Na20

0.06

0.06

0.12

0.14

TiO2

0.25

0.23

0.25

0.25

m203

0.02

0.02

0.02

0.02

p205

0.24

0.21

0.20

0.20

Cl'

0.01

0.01

0.01

0.01

Total

99.74

100.36

100.01

99.57

LS

92

122

CaO
M90

96

93

SR

2.91

2.23

2.55

2.49

AR

1.63

1.76

1.77

1.75

TABLE

Chemistry and.Mineraloqy

6:

ciln point
depth in ring

29.5

of a (coal ash) ring

32.5

under hot
surf ace

middle

near shell f
I

chemical
analysis:
si02
u203
-203
CaO
M90
K20
Na20

so3
Loss on ign.

24.65
5.68
4.23
61.27
0.49
0.58
0.36
0.15
2.61

24.61
5.70
4.29
61.92
--0.79
0.34
0.03
0.27
1.63

21.77
5.48
3.53
65.31
0.64
0.17
0.20
0.34
2.48

Total

100.0

LS
SR
AR

78.1
2.49
1.34
1.0

free lime

ninerals
{wt. %)
alite
p<

99.68

100.0

79.0
2.46
1.33
4.0

93.7
2.42
1.55
2.7

det.

and pbelit

nJ

30

ry35

d
-

40

15

55

60

40

rJ

30

liquid phase
(aluminate +
ferrite)

ni 15

15

15

carbonate
spurrite+CaOf

Id15

h/ 10

&

15

< 5

45

E\belite
texture:
average pore
size (estimated) p

400

800

400

800

I
I

100 - 300 1

TABLE

Chemical composition of core and rim of meal

ball and the kiln charge composition at the
correspondrnq zone

7:

k i l n

b a l l

rim

Loss on iqn.
I

20.9

A1203

5.6

Fe203

'

Ca0
s03

I
I

kiln

core

charge

calcininq
zone
I

3.9

Si02

4.3

'-17.2
8.0

1
I
I

1.9

21.2
5.7

2.8

4.4

2.7

63.0

62.5

62.1

1.7

2.2

1.9

LSF

94

104

91

SR

2.4

1.4

1
I

2.5

Cl'

0.1

I
I

0.1 I
I

K20

2.1

1.5

1
I

0.1

2.3'

APPENDIX I

APPENDIX II: CHEMICAL ANALVSIS

MA-Mat.

Nr.

Plant
Locatlon
Loss on Ign.
sio2

A1203
Fe 03
Ca 6
MgO
SO3
K2D
Na20
TiO
Mn2 6 3

I
I
I

50'393
D

53'550
G

I
Rlser Plpe

11.8
9.2

:*2"
49:3
1.2

15.9

3.6
0.98
0.09
0.07

;:-05

0.11 2.7

Total

100.65

:"R
LSF

I
I

OF RISER PIPE AND KILN INLET DEPOSITS

1.61 1.59
157

54155 1 54118

I3

Kiln Inlet (OS)

10.1

7.4
f*$

1;*:

46:l

5:::

2:::

E
8.4

z5
0:15

0.01
0.09

252:;
'10.6

3:4

0.19
0.15

4:*;
0:83
4.6
7.6

it*:1

53956
H
Kiln
Inlet (DG)
0.57
12.4
2:
39:8
2.3
27.6
:*:r

0:os
0.38
0.27

0:19
0.16
0.08
0.09

5.0

0.03
0.05
2.1

99.7

102.82

101.23

98.83

3.12
2.13
114

2.09
5.89
52

1.48
0.68
95

2.47
1.31

196

APPENDIX

MA-Mat.

Nr.

Plant
Location
Loss on ign.
SfO2
Al203
;;a03
I? t.3
so3
K2D
Na20
ii02
Mn2D3
p

Total
CaOf
SR .
EF

III:

53952
I

CHEMICAL ANALYSIS OF RINGS FROM,ROTARY

52'040

52'037

100-l 10m

45m

37m

10.7
14.7

t::
0.40
0.18
0.06
0.23
0.23

0.47
23.0
4.9
3.0
66.4
1.1
0.13
0.16
0.06
0.25
0.02
0.24
0.01

98.89

2.19
1.48
110

ii:;
52.3
0.79

52178
M

SECTION OF KILN

49643
1

50'524

35m

18.3m

1.63
24.6

0.12
22.84

FL!
63:l
2.1
0.77
0.72
0.07
0.31
0.04
0.08
0.05

45::
61.9
0.79
0.27
0.34
0.03

35:;
63.9
0.92
1.07
1.07
' 0.24
0.36
0.15
0.24

8.11
21.3
4.5

99.74

100.44

99.68

100.25

27-h
1:63
92

2.73
2.32
88

2.46
1.33
79

2:::

32.5m

45061

El
1:84
88

1OOm

2:::

:.:6
0:95
0.28
0.28
0.02
0.08
0.28

:*:
50:8
1.13
14.2
2.18
0.16
0.22
0.04
0.24
0.01

100.56

99.38

2.88
1.55
88

0.62
2.36
1.46
78

5;::

PHOTOTABLE

1 a)

l0J.u

1:

REM 84/108

Mud ball in chains

IOU/
u

lc)
KC1

1 el

REM 841545

S.E.M. MICROGRAPHS OF DEPOSITS

1 b)

REM 84/506

Compact cyclone 2 deposit

ld)

crystals

REM 84/79

eY

1 f)

,xm

Spurrite

crystals

REM 84/69

REM 84/72

Blue Circle Cement

PROCESS ENGINEERING TRAINING

PROGRAM

Module 13
Section 12

Rings and Buildups in Cement Kilns

BURNING

ISSUES

address those most frequently encountered. inches w.g., -0.01, -0.1, -0.15, etc. Quite posThere are problems associated with the sibly the most serious effect on hood presburning of waste fuels which can be attributed to flame position, alkalies, chlorides
and sulphur.
Figure 2 shows a basic understanding of
air volume changes attributable to changes
in temperature. Our experience indicates
that many people tend to forget this relation. They comments I didnt increase the
air flow, the flame looks like it is on the
load. It wasnt yesterday; someone must
have moved the burner.
Figure 2 shows the increase in volume
caused by temperature changes. One cubic
foot of air at 100F weighs about
0.071 pounds. The weight of one
MAJOR CAUSES
fig.1
cubic foot of air seems insignifi- sure sampling over the years has been our
1 . Overheating
cant, but at each of these plotting attempt to bum on the nose. All changes
2. Slow clinker quench
the weight of air is the same in fuel ignition are immediately detected by
3. Fuel impingement on the burning zone
(0.071 pounds), only the volume hood pressure changes-there is no dampload
has changed. 4.4 cubic feet of air ening effect as there is when the flame is
4 . Long flame
at 2000F still weigh 0.071 away from the nose.
5 . Chlorides
pounds.
Figure 3 shows the different pressure
6 . Suiphur
We often hear the question, conditions found in the kiln discharge hood.
7 . Potassium
Where is the best place to sam- This is why more than one sample point is
8 . Mechanical restrictions
ple the kiln discharge hood pres- needed, with all manifolded together to
sure? But the real question is serve as one sample source. The result is the
EVALUATION PROGRAMME.
where on the hood does the measurement of an average pressure.
1. Sampling
sample point (or points) give a
It is interesting to note the effect higher
2. Care of sample (temperature, air and
pressure reading that permits rel- secondary air temperatures have on the kiln
moisture considerations)
ative control? That is, where are discharge hood pressure. The increase of
3. Samples which are consistent and
the conditions today similar to secondary air temperature increases the volrepresentative
what they were yesterday? The ume of air as well as the velocity. This
4. Documentation of conditions before and
next question is what is the cor- increase of velocity tends to drive the secduring time of the problem
rect kiln discharge hood pres- ondary air and dust toward the top of the
sure? The kiln discharge hood hood. This condition always creates a dustSOLUTIONS
should be at a slightly negative ing and puffing at the top of the hood over
1. Alter raw materials and fuel
pressure
to permit observation the kiln, whereas the bottom side of the kiln
2. Control internal alkali, sulphur and chloride
by
instruments
or persons with- may be at slightly negative pressure.
cycle
out
undue
overheating
and dusty
A change in secondary air temperature
a. Install a kiln gas by-pass for preheater
conditions.
From
the
standpoints
can
move the flame position up or down.
and calciner kilns
of
good
housekeeping
and
mainCertainly,
a change of secondary air temperb. Do not return as much total dust,
tenance,
the
hood
pressure
ature
wilt
alter the fuel ignition rate, but the
especially where precipitator fields discharge
should
be
slightly
negative.
This
concern
in
this example is the positioning of
to individual conveyors
value
should
be
c. Determine time cycle for build-up
determined by trial r
Adjust kiln burner (permit clinker quench,
and error for each
fig.3
shorten burning zone length, eliminate fuel
POSITIVE
system. It is always
impingement on the load, locate burner on
advisable periodikiln center line and slope)
POSITIVE PRESSUR
tally to review the
4. Adjust kiln material and gas temperature
selected set point to
profile
determine if condi5. increase kiln rotational speed
NEGATIVE
tions have changed.
6. Install kiln internal restrictions such as
Once the desired
dams or orifice rings
kiln discharge hood
7. Maintain the secondary air temperature
pressure is selected,
consistently
that is the target,
whether it is 0.05
he problems caused by rings and
build-ups in a kiln system always create turmoil and frequently a loss of
production. Most of these problems can be
controlled if not eliminated.
Figure 1 indicates the major causes of
rings and build-ups. There must be a good
evaluation programme, which includes a
review of the literature. When this is accomplished, there are definitely solutions to the
problems of rings and build-ups in the kiln.
Frequently the solution requires forgetting
some preconceived ideas.
This section will not cover all of the rings
and build-ups that can occur, but will

BURNING

ISSUES

secondary air temperature

is
increased.
Velocity
through the cooler throat
increases to 1275 feet per
minute. This increase of
velocity tends to raise the
flame path, which usually
causes the burning zone to
cool off and the calcined
material to flush into the
burning zone.
These examples show
why it is more important
to maintain a constant secondary air temperature than to attempt to
reach the highest possible temperature.

clinker and past a thermocouple sensor.

Methods for aspirating air from the clinker
cooler have proved to be impractical primarily because of wear created by the
clinker dust. Quite possibly the calciner kiln
system permits the most accurate measurement of combustion air temperature. The
calciner kiln system aspirates combustion
air from the clinker cooler as tertiary air for
the calciner.
In spite of its inaccuracy, the thermocouple placed in the clinker cooler throat has
been accepted as indicating a usable relative
secondary air temperature for day-to-day
kiln operation. This method of detecting secondary air temperature is fine if we remember that
m It is a relative temperature and
may read much higher because of
radiated heat from the clinker.
n It may increase or decrease,
depending upon changes in the
clinker cooler bed, without much
real change in air temperature.
Fluctuation of secondary air temperature is one of the major causes of
rings and build-ups. The kiln flame
and location must be controlled to
maintain a stable operation. A stable
kiln operation should create the pattern of coating and a ring formation
shown in figure 7. This drawing shows only
a small amount or no coating from the burning zone to the kiln discharge end. The ring

the flame path.

Figure 4 shows the burner positioned on
the kiln centre line and slope. This
position has been adjusted during
fig.6
operation to compensate for the secr - - - - - x ondary airs tendency to lift the
flame path. In this example the intent
is to direct the flame tip on the kiln
centre line and slope. Figure 4 indicates that the average secondary air
temperature is 1000F. The volume of
secondary air passing though the
fixed throat area has a velocity of 900
feet per minute.
The system in figure 5 is identical
to that in figure 4 except that the secondary air temperature has been
reduced to 700F. The velocity of the secNormally, attempts to achieve the maxiondary air through the clinker cooler throat mum secondary air temperature produce
has now been reduced to 715 feet per cyclical operation of the kiln. This promotes
minute. The flame path has been lowered the production of clinker
and the tip is no longer on the kiln centre burned in a reducing atmosline and slope, resulting in fuel impinging phere, slow quench of the
on the load. The problem of fuel impingeclinker minerals, dusting in the
C ,I
ment on the load is definitely more prokiln discharge hood and kiln
.,
, .,
nounced when the burner is adjusted to turn ring formation.
the flame toward the load. Microscopic
It is also important to recog:
analyses often indicate that the clinker was nise that the secondary air temproduced in a reducing atmosphere on this perature recorded by most
date, whereas the day before this was not plants is a relative temperature.
the case.
The secondary air temperature
Figure 6 shows what happens when the is usually detected by placing a
thermocouple
the that forms 80-115 feet from the kiln dissomewhere
near
clinker cooler throat area. charge end is in the area where calcination is
The value indicated by this complete and the liquids begin to form. The
method of sensing not only location of this ring depends upon the burnmeasures the air temperaing zone length. It is formed because of the
ture, but it also detects coexistence of calcined material, a small
radiated heat from the amount of liquid, and material still in the
clinker and the flame. A solid phase. This creates prime conditions
true secondary air temper- for build-up. The ring does not adhere to the
ature is measured by aspirefractory, is not dense and is, very fragile. It
rating a portion of the set- breaks up and falls out when the kiln temondary air away from the perature is changed by alterations to the cal-

B U R N I N G I S S U E S
cining zone and material preparation. It
may fall out when flame length and location
change.
This ring is regarded as an asset because
it serves as an orifice that increases the gas

such as a decrease in secSLOW

ondary air temperature, may
QUENCH
create the condition where the
flame tip is projected through
the load (fig 9). This lengthening of the flame
causes
fuel
I
I
impingement
on
LONG FLAME
the load, but also
causes the conical long flame ring
build-up shown in figure 8.
When this ring is detected, it
can be broken up and dropped
out by shortening the flame.
_ _ ----.-IThis type of ring can also be
prevented with a short flame
velocity at its location. This tends to hold
with its tip directed on the kiln centre line
back and mix aerated material. While the
and slope.
ring is present kiln operation tends to be staAnother example of a long flame is
ble, with less material flushing into the
burning zone. If all conditions remain stable, the ring remains and assists operation.
It does not grow substantially as the stable
operating time increases. For several days
often the ring falls out, kiln operation may
be cyclic and it is difficult to keep the raw
load out of the burning zone. We have often
reviewed kiln operators logs and found this
NOSE RING
KILN
to be a common scenario.
DISCHARGE
SLOW QUENCH
--I_--Figure 8 shows the ring caused by a long
flame. This ring may also be formed when
the flame path is directed into the load. In shown in fig 10. In this case the flame tip is
the latter case, the kiln may have experi- at least directed on the kiln slope and paralenced stable operation with the flame lel to the kiln centre line. There are apparently sufficient liquids available to produce a sticky environment which promotes the
development of a material ball.
Balls which are 6 to 12 inches
in diameter have been found
in the middle of the calcining
zone. A few of these balls
grow to diameters of 6-8 feet.
The larger balls look alarming
directed toward the load, and this location
when they are first seen passing through the
may be satisfactory as long as the flame tip
burning zone. Burning with a shorter flame
is not on the load. However, when the flame
length prevents additional balls forming
tip is directed into the load, any change, unless they are caused by a high concentration of alkalies, sulphur, and
chlorine.
The nose ring (fig 11) has
fig.10
been described as an ash
ring. Some kilns operate with
a nose ring most of the time.
This tends to restrict clinker
i
P---discharge
from the kiln.
KILN
DISCHARGE
Microscopic evaluations of
clinker produced during the

I- t i e . 1 2

presence of a nose ring indicate the presence

of slow quench. The nose ring permits a
very slow quench of the clinker because the
material is pooled when it passes out of the
burning zone. Quick quench of the clinker
minerals must be completed within the kiln
or it will not be achieved.
Slowly quenched clinker causes the C?S
to revert back to C,S and free lime. Further
slow cooling causes the C,S fin
the beta state or high-temperature form) to change to the
gamma state of C,S (a lowtemperature
form).
The
gamma form of C2S is a dust
and no longer forms a nodule.
This dust is picked up by the
flow of air and carried back
into the kiln where it enters the
burning cycle again. The slow
quench cycle continues as long
as the nose ring persists to act as a dam. The
suspended particles returning with combustion air are easily preheated because the surface area is maximised. The liquid available
at the kiln nose permits adherence of the
dust particles, and the building of the nose
ring continues.
Figure 12 shows an example of a snowman on the clinker cooler back wall. Some
snowmen grow tall enough to reach the
burner pipe. Generally, the larger the kiln,
the larger the snowman. Depending upon
the installation procedure of refractory over
dead grates, some kiln systems form snowmen on the clinker cooler side wall near the
throat. The snowman build-up is caused by
the same problem that promotes the nose
ring build-up - that is, slow quench of the
clinker.
Microscopic evaluation of clinker shows
whether the material was slowly or quickly
quenched and whether C,S changed from
the beta to the gamma state. Both the nose
ring problem and the snowman build-up
can be eliminated by adjusting the kiln
burning operation so that the clinker is
quickly quenched within the kiln.

B U R N I N G I S S U E S
We have learned to live with a dusty kiln temperature increased from 2600F to
discharge hood, especially in larger kilns. 2750F and NOx fell from 750 ppm to 350
The old small wet-process kilns were sel- ppm. The clinker went from slowly
dom dusty because the fuel consumption quenched to quickly quenched. The clinker
cooler snowmen
were eliminated,
the kiln discharge
HIGH SULPHUR RING
hood cleared and
THICKNESS CAN BE
1
we could see the
flame and burning zone. In addition the 28-day
compressive
s t r e n g t h s
increased by 600
psi over a 90 day
KICN
period
without
DISCHARGE
any increase of
fineness.
Figure 13 displays a ring formation
was high and we could not gain quick which occurs in the calcining zone or the
enough ignition to burn on the nose. This area where the
promoted the quick quench of clinker gas temperature is
within the kiln. We also found that the old sufficiently low to
wet kiln produced the most reactive clinker, permit condensawhich permitted a lower fineness for similar tion of sulphur
compressive strength levels.
and chloride comOur most recent experience of putting pounds. This ring
KILN GAS
this flame technology into practice was with is a part of the
a large wet kiln. It was necessary to remove alkali,
sulphur,
BYPASS
large snowmen from the clinker back wall. a n d
chloride
These snowmen were giants, lo-12 feet high cycle. All kilns
and 6-8 feet in diameter at the bottom. The have a variety of
kiln discharge hood was so dusty that we rings in this area:
could not see the nose of the kiln. The nose some consist of a
refractory had to br replaced every six small amount of
months and the nose castings every 12 PUkY
coating.
months. The kiln burner was adjusted to With larger rings
shorten the flame: this reduced the burning (fig 13) the kiln has to be shutdown to physzone by about 45 per cent. The burning zone ically remove the build-up. The elimination
of the cause normally requires a
fuel change, such
fig.14
as a lower sulphur
fuel, a n d the
return of less kiln
VELOCITY
dust. If a microscopic evaluation
of the clinker indicates production
in a reducing
HATES ANi
atmosphere, the
TEMPERATURE
burner should be
BYPASS QUENCH Al R
adjusted to elimiIN RISER
nated
fuel
impingement on
the load. This will
permit a higher
clinker
sulphur

level which removes a similar amount from

the cycle. If the long wet and dry kilns use
an electrostatic precipitator, the dust collected in the final fields can be wasted as
high alkali, sulphur, and chloride material.
The electrostatic precipitator works well
as a kiln gas bypass system for the long wet
and dry kiln systems. Since the solidified
alkali, sulphur, and chloride particles are
very small, they are concentrated in the final
field of the precipitator, and are easily separated and removed from the system.
Figure 14 shows some areas in the suspension preheater where problem build-ups
often occur. As we proceed up the preheater
in the direction of the kiln gas flow, the first
problem area is at the kiln feed shelf. This
problem on a preheater kiln is either caused
by leakage of ambient air into the system or
by operating with a high level of carbon
monoxide in the exit gas. Ambient air leak-

fig.15

19OOF T O 21OOF

age causes a localised condensation of alkali,

sulphur and chloride compounds. These
chemicals are vaporised in the burning zone
and exit as a kiln gas until temperature conditions are sufficiently low (about 1800F) to
cause condensation to the liquid state.
Normally, the preheater kiln exit gas temperature is above the condensation point.
When ambient air leaks into the kiln feed
end housing there is a localised cooling of
the kiln gas at the leakage source that results
in build-up at that point. A different type of
calciner kiln system build-up at the feed
shelf and feed end housing walls can also be
caused by leakage air. This build-up is
caused when the kiln feed is nearly calcined
and there are C,AF liquids present.
However, if the gas dust concentration is
sufficiently high, the liquid will adhere to
the dust particle rather than to the surface of
the wall, thereby preventing a build-up.

B U R N I N G I S S U E S
also ensure that the

quench air exits to

bypass
the kiln
AP = 2.0'
KI LN
induced draught fan.
75O*F T O ilOOF
I i - - + - Alkali,
sulphur
GAS
e
and chloride comBYPASS
-0.5. TO -1.5
pounds create no
-2.5" TO -3.5 o
build-up problems if
they exist in either
QUENCH CHAMBER m
!
7
em -1900F T O 2100F
the gaseous state or
I
.
7-J
the
solid
state.
QUENCH AIR
However, if they
/
exist in the liquid
state, they behave
like
water on dust.
fig.16
The secret to efficient
i
kiln gas bypass sysThis situation can be artificially duplitern operation is taking a portion of the kiln
cated by the introduction of dust from the exit gas at plus 1900F and instantaneously
Stage III cyclone material discharge and/or quenching it to about 750F. This permits
creating a rough feed shelf surface which the alkali, sulphur and chloride compounds
causes the a splashing of the feed out into
the gas stream. Dust re-entrained in the kiln
exit gas by a rough feed shelf surface will
increase the dust lost through a kiln gas
bypass system, so the feed shelf must have a
smooth surface when running a kiln gas bypass system.
Figure 15 shows a build-up above the
kiln gas bypass take-off and within the
quench chamber. The build-up in the kiln
riser above the kiln gas bypass take-off is
caused by the leakage of quench air from
GAS AND DUST
the quench chamber. Proper sizing of the
bypass quench chamber inlet can ensure
that quench air does not enter the riser duct.
DJST The example in fig 16 shows the parameGAS ---c
ters used for design and adjustment of the
quench chamber inlet. A two-inch pressure
loss through the quench chamber inlet will

-I

to pass from the gaseous state directly to the

solid state without passing through the liquid state. Some designers and operators
quench to higher temperature levels, i.e.
900F to 1100F. Our experience has found
more potential for build-ups in the quench
chamber at these higher temperatures.
The kiln gas bypass system appears to work
best when the quench chamber and kiln riser
duct take-off are placed above the kiln. As the
gas and dust exit the kiln, the dust is thrown
against the feed shelf while the gas is turned
upward. This separates dust particles from the
kiln exit gas stream. The cleaned gas tends to
pass on the kiln side of the riser duct for a short
time. Figure 17 shows the desired quench
chamber position and fig 18 indicates the
desired operating parameters for a kiln gas
bypass quench chamber. In our experience a
quench chamber operated with these parame
ters will not product any build-ups, and will

NCH

CHAMBER

operate with no dust in the bottom of the

chamber.
Kiln gas bypass dust collector material contains 0.520 per cent of clinker, 20-25 per cent
So,, and 4.5-5.0 per cent K20. If the percentage
of sulphur as SO3 is less, for example 16 per
cent, the bypass system is taking too much
dust from the kiln riser, etc.
There are always answers to problems
with rings and build-ups. The solution is
usually found when the attitude of the
operator is that we cannot continue to live
with this problem. JI
This paper was first presented by the author,
Floyd C Hamilton, of Hamilton Technical Services
lnc, Roanoke, Virginia, for the National Lime
Association meeting, St Louis, Missouri, United
States, October 1997.

Blue Circle Cement

PROCESS ENGINEERING TRAINING

PROGRAM

HBM PROCESS ENGINEERS

CONFERENCE

Minimization of Volatile Cycles

MINIMISATION OF VOLATILE CYCLES

1.

SUMMARY

Concentration of minor components within kiln systems due to volatile cycles can lead to
significant operational problems on all types of process, with consequent loss of output. A
portion of some of these volatile materials is also lost to the atmosphere, and minimisation of this
leakage is also increasingly becoming an environmental concern. understanding of the factors
that effect the magnitude of these cycles is an essential part of improved kiln control leading to
effective control of the cycles. Successful implementation requires co-operation between chemist,
process engineer, mechanical engineer and operators.
2.

INTRODUCTION

A limited number of minor components in the raw mix and/or fuel can become highly
concentrated within the cement kiln system and then create operational problems. The minor
components that are generally considered to be involved in major volatile cycles are the chloride.
alkali (Sodium and Potassium) and sulphur species - although other elements do also become
involved in cycles to a much lesser degree (such as Fl V, As, Pb, Tl Cd, Hg and Zn) they have
not been identified as causing operational problems and so will not be considered at this time.
These substances are present in the raw materials in low proportions in a variety of forms, but are
likely to evaporate or decompose under the temperature regimes found in the burning zone. Once
this happens they become associated with the gas stream and cool as this losses its heat to the
material bed, until they either condense or react to form compounds that will condense. At this
point they are present in the gas stream as potentially sticky liquids and will adhere to any surface
with which contact is made; this can be a particle surface or a vessel side wall. Once the liquid
has condensed on to a particle this can still stick to any wail with which contact is made until the
temperature drops to a sufficiently low level for the liquid to solidify. In the suspension preheater
and Lepoi processes the temperatures at which the potential compounds are liquid coincide with
those found close to the kiln hearth and lower preheater stages (SP kiln) or above calciner (Lepol
kiln), and this leads to material building up on the walls in these areas, or to ring formation at the
very back of the kiln. In the older long chained kilns, excessive volatile cycles can contribute to
ring formation. In either situation, at best this reduces the duct or kiln dimensions and so causes
increased pressure drop and - probably - extra dust generation, whilst at worst it reduces output,
increases kiln instability, and causes blockages to develop in the preheater system: the end result
being significant kiln down time.
In general, in the more thermally efficient processes (precalciner, suspension preheater and Lepol)
most of the volatiles will condense in the preheater and only a small fraction will condense on the
precipitator dust or escape up the stack. In contrast in the less thermally efficient processes (wet
and long dry) a higher proportion of the volatiles will pass through the kiln system and condense
on the precipitator - or bag house - dust or escape up the stack. Commonly where volatiles
condense onto a dust the finer dust fraction will develop a higher concentration of the volatile
component, due to the higher surface area of the finer dust.

Where a component is partially volatilised in the burning zone and then partially recombines into
the material stream within the kiln/preheater system it is possibie for a large amount of the
component to continuously recycle around the kiln system. This is called an internal cycle.
Where some of the component is collected with a dust stream externally to the kiln system and
then returned to the kiln, this is referred to as an external cycle.
3.

GENERAL REVIEW OF THE PROPERTIES OF THE MAJOR VOLATILE

COMPONENTS

3.1 Chlorides

Chlorides are derived from the raw materials and the kiln fuel. The high voiatilities of these
compounds, together with the high collection efficiency of the cyclone preheater systems, will lead
to the development of a greatly enhanced cycle. The chlorides have a high affinity for the alkalies
in general and potassium in particular. This property together with the high volatility has been
used in kilns (commonly on rhe wet process, occasionally on the SP process) to control clinker
K2O levels by addition of CaCl2 to the raw mix or fuel, which leads to loss of KCl with the kiln
bleed or the exhaust gas from the kiln system. In the suspension preheater, the volatilised material
is recaptured within the system unless a bleed is utilised between the kiln and riser duct. It is
generally considered that no more than 3% of the chloride passing- from the preheater to the kiln
will leave the system with the clinker. Although considerably higher levels have been noted in
individual samples of clinker, this is probably due to a push of kiln feed, or a semi-flush situation
as thermodynamic considerations indicate that no chloride should pass through the burning zone.
On many SP kiln systems some degree of preheater cleaning is necessary on a regular basis and
this may help to control the chloride cycle by forcing the kiln conditions into a situation which
permits a brief increase in clinker chloride level (i.e. reduced material temperature, increased
material loading and flux level).
Small amounts of chlorides will also leave the preheater system with the waste gas stream.
Taking the total loss of chloride from the system as between 2 and 5 % of the feed to the burning
zone, it would then be expected that a circulating load of 20 to 50 times the total chloride input
could develop in a system without a kiln gas bleed.
No reports of low temperature chloride volatilisation within the preheater have been identified.
3.2

Alkalies

The major source of alkalies will be the raw mix; notably the clay component, although minor
quantities can arise from the fuels The initial free alkalies will behave in one of three ways:

1)

Remain in the material being processed and become incorporated in the clinker
constituents that are being formed. This happens to Na2O to a greater degree than K2O

2)

Be converted into different compounds - chlorides, sulphates,

by reaction with ofner constituents of the raw mix.

3)

Diffuse to the surface of the process material and volatilise.

carbonates, hydroxides

In its initial state, K2O begins to volatilise over a wide range of temperature, depending on the
form of clay in which it was incorporated but irrespective of source, it would be expected to have
volatilised almost completely at burning zone temperatures, although some may have been at least
partially stabilised by conversion to the less volatile sulphate form within the material bed. Once
volatilised it will react to form chlorides and sulphate - chlorides preferentially - at the rear of the
kiln. These will then deposit on dust particles. Initiallv Na2O is less volatile than K2O due to its
higher bond energy and so a greater proportion of the kiln feed Na2O would be expected to pass
through the burning zone in clinker without participating in the volatile cycles. Volatilised Na2O
will react with SO2 and SO, to form sulphates towards the rear of the kiln and with chloride where
this species is present in excess of K2O Where alkali is present in excess of chloride and sulphate
alkali carbonates will be formed Each of these al
-=smm
the surface of dust particles in the cooler zone of the kiln ad lower preheater stages and will enter
the volatile cycles as the dust is separated out in the cyclones. Direct contact onto kiln or
preheater surfaces may lead to the development of build-up, The compounds will then re-enter
the kiln where the degree of volatilisation will depend on the species and the kiln conditions.
Volatility decreases from chloride to carbonate to sulphate and, hence, sulphates are more likely
to pass through the burning zone. Nevertheless, the likely range of burning zone temperatures
cover the thermal area in which alkali volatilities are likely to increase significantly with rising
temperature. In general, precalciner kilns have significantly lower burning zone temperature than
are common in other processes and, hence, alkali sulphate volatilisation in particular is lower in
precalciners than in other processes.
3 . 3

Sulphur

Sulphur can enter the system in a number of forms from either fuels or raw materials. A limited
amount may evaporate in the upper preheater stages and escape from the system in the exhaust
gases. In general, SO, and SO, can form in the high temperature areas and be transferred to the
gas phase. In the cooler areas of the kiln back-end and preheater system sulphates will form and
re-enter the material stream. preferentially alkali sulphates will be produced with excess sulphate
combining with free lime or calcium carbonate an$li@ru&&5li~-i5Xailablein~e~~
.--~~~r-l-----t:----~~a~~~~~~~~~~~~~,
so
restricting the formation of sulphates. In this situation loss of sulphur oxides by way of the stack
may increase but where the gas stream passes through the raw mill the majority of the sulphur
oxides would be expected to react with the high active surface calcium compounds which are
produced in the milling process. This will then return to the kiln in the raw mix as part of an
external cycle.
The high boiling points of the alkali sulphates would indicate that relatively low levels of
volatilisation would be expected. However, dissociation may occur, particularly under the
reducing conditions which can exist to some degree within the burning zone. Calcium sulphate
also has a high boiling point but is even more susceptible to dissociation, so a higher recirculation
of sulphate from this compound would be expected. As CaSO4 cannot recycle as a compound,
the lime from this compound remains in clinker as free lime - making burning more difficult whilst SO, is carried in the gas stream towards the kiln back-end where it reacts to form alkali or
calcium sulphate.

4.

CONTROL OF VOLATILE CYCLES

As indicated earlier the major species involved in volatile cycles are chlorides, alkali and sulphur
based components. As the development of a volatile cycle depends on the evaporation or
dissociation and condensation of a range of compounds the meiting boiling and dissociation
temperatures of these compounds will be of major significance, whilst volatility characteristics will
also be of importance. Boiling and melting point data is presented in figure 1, and volatility data
in figure 2. As chemical analysis gives details of the volatile components in the feed materials of
in material from within the system in terms of the primary oxides or elemental forms - IWO, K20,
SO3 and Cl - the volatilities are normally expressed in the same terms. and the ranges of volatility
for each that are commonly found within the cement kiln are given in figure 3. These can be
considered to be the primary volatilities for these components, and in each case the tabulated
figures show a wide range of volatilities. This doesnt help to extend our potential understanding
of what is happening within the kiln.
The volatiles however will be present in the material bed in the kiln as compounds for each of
which the volatility will depend on physical characteristics such as those set out in figures 1 and
2. As indicated in section 3 there will be an order of priority (ease) with which these compounds
will form. Therefore it is possible to study the development of a volatile cycle in terms of the
physical properties of these compounds. The specific volatilities that have been found for each
of the compounds that are likely to form are set out in figure 4. This shows for instance, that
alkali in combination with chloride is more likely to develop a cycle than that combined with
sulphate. In the same way, sulphate combined with calcium is more likely to volatilise than that
present in an alkali form.
Where volatiles are present in the raw materials some degree of cycle is bound to develop.
However, the overall cycles can be governed to a degree by optimising the relative proportions
of each volatile component to maximise the potential for the formation of those compounds with
the lower specific volatilities. The prime control of the volatile component cycle is performed in
the design stage of a works project through the selection of raw materials and fuels in order to
optimise the relative and absolute levels of the potentially volatiie components and, where
necessary, the inclusion of a bleed system. However, whenever changes are made to a sourcing
of raw materials (including fuels) the potential affection in the volatile balance needs to be
considered.
Where processing conditions are kept steady, the cycles will continue to develop until equilibria
are reached, at which time the total amounts of volatile entering the system will be balances by
the quantities leaving the system.
The degree of volatiisation and the rates at which the equilibrium are established will depend on:

(a)

The species, their chemical forms and concentrations.

(b)

The volume of gases.

(c)

The intimacy of contact between gas and solid.

(d)

The vapour pressures of the salts.

(e)

Possibility of dissociation or further reaction.

(f)

Rate of diffusion to and from solid/gas interfaces.

(g)

Degree of saturation of gas.

(h)

Kiln atmosphere.

(i)

Kiln temperatures.

(j)

Time/temperature profile of material within the kiln.

Most of these factors are to some degree inter-related and so in normal operation the only
methods available to control the degree of volatilisation and eventual concentrations in clinker and
within the kiln system will be the kiln internal atmosphere and temperature, and the proportion
of gas bleed from the kiln exit or material bleed from the system.
As indicated above, it should be possible to modify the volatile cycle by variation of the kiln
internal atmosphere (oxygen level) and burning zone temperature. In the late 1980s a series of
kiln trials were conducted at Hope Works to investigate the effect of these kiln conditions (BZT
and BEO,) on the volatile cycles. This study covers the situations of significant chloride, alkali,
and sulphate inputs. The major conclusions were:1.

The chloride cycle is 20 to 30 times the total chloride input and increases slightly with
increasing temperature, although this may be due to improved kiln stability at higher
temperatures. The chloride cycle is not modified by wide changes to kiln atmosphere
(Figure 5).

2.

Chloride in the cycle combines with aetassiu~availahle, However, at low

temperatures - equivalent to NOx Eve1.s m
o
p$ii - some sodium is also
involved.

3.

The chloride cycle is a low temperature cycle and cannot be significantly modified by
vat-ration or rum conanions.

4.

The total potassium cycle is approximately 3 times the input level (Figure 6), however
about two thirds of this is present in Stage IV in conjunction with chloride and so cannot
be controlled except by incorporation of a bleed system.

5.

About one third of the potassium in Stage IV is either f& based on its first passage
through the preheater or derived from a potassium sulphate based cycle. This cycle ratio
varies from 1.05 (minimal recycle) at 800 ppm kiln NOx level to 1.25 at 1,400 ppm NOx
(25% recycle). This portion can be controlled by operation at low temperatures (Figure
7).

6.

The sodium cycle level varies between 1.2 and 2.0 times that in the raw meal over the
temperature range examined.

7.

At low temperatures some sodium becomes involved in the low temperature chloride cycle
and this boosts the sodium cycle by about 10%.

8.

The majority of the sodium in the system is involved in a sodium sulphate based cycle.
This is strongiy temperature dependent, and the rate of increase also appears to be
increasmg with temperature. Over the temperature range investigated the recycle rose
from 1.35 times the level in the feed at low temperatures to 2.0 times at high NOx levels
(Figure 8).

9.

The total sulphur cycle is temperature dependent and rises from 1.6 to 2.6 times that of
the input over the temperature range investigated (Figure 9). Low oxygen levels increase
the suiphur cycle.

10.

The alkali sulphate cycle has already been summarised in points 6 and 9 and makes up
about 35% of the total sulphate in Stage IV material. These cyclic levels can be seen to
be lower than the total suiphur cycle levels (Figures 7, 8 and 9).

11.

The calcium sulphate recycle is strongiv temoerature dependent. The quantity of SO, as
%&(ii$L~V rises from^-aoout 2.5 times tne level in the feed at low NOx levels to 4
times at high (1,450 ppm) NOx levels (Figure 10).

12.

The calcium suiphate cycle is also increased by a move into a reducing kiln atmosphere.
At about 1,200 ppm NOx the level of SO, as CaSO4 increased from 3.1 under oxidising
conditions to 5.5 times the feed levei under reducing conditions (Figure 10).

13.

For potassium and sodium there are indications of a slow but steady increase in losses of
these components from the preheater system as temperatures increase. This may become
part of an external cycle or may be lost to atmosphere. This could be established by a
longer term study of the levels of these components in the precipitator and stack dusts.

14.

Sulphur is in overall balance within the system below NOx levels of 1,200 to 1,300 ppm,
but above this level the loss increases sharply with firing temperature. Again this may
become part of an external cycle or may be lost from the system to the atmosphere,
however in this exercise no precipitator or stack dust samples were collected or SO2
emission measurements made so this cannot be confirmed.

15.

When the kiln atmosphere moves into reducing conditions the losses of alkalies and
sulphur from the kiln system to the atmosphere increase sharply.

Overall the results show that the chloride cycle and its associated alkali cycle cannot be
controlled by process conditions. Cycles of alkali present in the form of alkali sulphate can be
controlled and minimised by careful control of the burning zone to the lowest practical
temperature.
Calcium sulphate based cycles can also be minimised by burning to low
temperatures, but in this case careful control of the kiln atmosphere {back end 02 level) is also
necessary to ensure that the on-set off reducing conditions is not possible.
While these tests were conducted on a suspension preheater kiln, the critical area is the burning
zone and so the general results will be equally applicable to other processes.
The susceptibility of calcium sulphate to increased volatilisation under even borderline reducing
conditions also emphasises the potential for the sulphate cycle to be affected by the condition of
the firing system, by the type of fuel, and by cooler operation. Low secondary air temperature,
a low volatile fuel, poor coal drying, or low momentum in the firing pipe are likely to promote
slow initial combustion, in which case significant combustion is likely to continue to occur once
the jet has fully expanded. This may give reducing conditions immediately above the material bed
and encourage sulphate volatilisation. Where this is possible it is recommended that the firing
pipe is aiigned along the axis of the kiln in order to ensure the maximum kiln length is available
for jet expansion. A cool flame is also likely to produce a long burning zone which will hold
the potentially volatile materials for a longer time period within the temperature range at which
volatilisation could occur; this will also promote higher degrees of volatilisation.
5.

CONCLUSIONS

From a Process Engineers point of view, the easiest way to minimise kiln volatile cycles is to
push the problem to the Works Chemist and expect the raw materials to be selected to give the
minimum practical inputs of potentially volatile materials. This also requires the levels that do
exist in the raw materials to be balanced to give the maximum opportunity for the formation of
compounds with relatively low specific volatilities.
The quantities of material recirculating can also be controlled by the selective removal of a dust
fraction from the process. Examples of this are the selective dumping of the finest precipitator
dust fraction on the wet process at Ravena, and the dumping of Lepoi cyclone dust at Cookstown.
The magnitude of the volatile cycles can also be reduced by the careful control of kiln conditions.
In general volatile cycles will increase slowly with increasing burning zone temperature or length.
There will also be a major increase in sulphate cycle if reducing conditions develop close to the
material bed. This increase will start to occur well before a significant increase in kiln backend
CO level becomes apparent. This emphasises the need to possess optimised fuel preparation and
firing systems and an efficient clinker cooler.