Downloaded from orbit.dtu.

dk on: Mar 24, 2020

Co-firing of Alternative Fuels in Cement Kiln Burners

Pedersen, Morten Nedergaard

Publication date:
2018

Document Version
Publisher's PDF, also known as Version of record

Link back to DTU Orbit

Citation (APA):
Pedersen, M. N. (2018). Co-firing of Alternative Fuels in Cement Kiln Burners. Kgs. Lyngby: Technical University
of Denmark.

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright
owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

 Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
 You may not further distribute the material or use it for any profit-making activity or commercial gain
 You may freely distribute the URL identifying the publication in the public portal

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Co-Firing of Alternative Fuels
in Cement Kiln Burners
Morten Nedergaard Pedersen
8
il 201
Apr
s is|
The
D
Ph
 Co-Firing of Alternative Fuels
in Cement Kiln Burners

PhD Thesis
by
Morten Nedergaard Pedersen

Supervisors:
Kim Dam-Johansen, Technical University of Denmark
Peter Arendt Jensen, Technical University of Denmark
Sønnik Clausen, Technical University of Denmark
Lars Skaarup Jensen, FLSmidth A/S

Technical University of Denmark

Department of Chemical and Biochemical Engineering

CHEC Research Center

30th of April 2018

 I

ABSTRACT
The production of cement is an energy intensive process, where, traditionally, 30 %
of operating costs have been related to fuels. By increasing the use of alternative
fuels in the industry, the operating costs can be significantly decreased. In addition,
use of refuse derived fuels may limit the need for landfilling, and partly biogenic
fuels can reduce CO2 emissions from the industry. The utilization of alternative fuels
in the cement rotary kiln appears to be mostly based on a trial and error approach.
Fairly little systematic knowledge of the impact that these fuels have on flame
behavior and cement clinker quality is available in the literature. This thesis
attempts to give an increased fundamental understanding of these impacts. This is
done through a literature survey highlighting the known challenges of alternative
fuels firing, experimental studies conducted at full scale cement plants investigating
the effect of cofiring on the kiln flame, and laboratory characterization of alternative
fuels. Furthermore, a model is developed to describe the combustion in the cement
kiln.

Alternative fuels for the cement industry can be both solid and liquid. Some of the
most widely used are shredded tires, meat and bone meal, and solid recovered fuel
(SRF). SRF is a fuel derived from the mechanical treatment of non-hazardous
municipal or industrial waste. Compared to fossil fuels, most alternative fuels have
a larger particle size, higher volatile and moisture contents, and a lower heating
value. This makes their use in cement kilns challenging, as burnout takes longer, and
flame temperatures are reduced. An increased understanding of these issues, and
how to overcome them, are necessary to further increase the utilization of
alternative fuels in cement kilns.

An experimental study was carried out at three different cement plants. The kiln
flames were observed with a specially developed camera, which can be inserted in
the kiln hood close to the burner and allows for detailed imaging of fuel and flame
behavior. The difference between fossil fired flames and flames cofired with
alternative fuels were studied. It was found that addition of alternative fuel to the
flame would delay ignition by 1-2 meters and lower flame temperatures. At one
plant the burner was changed. The design of the axial air injection was changed from
an annular nozzle to multiple jet nozzles. The new burner decreased the ignition
II Abstract

length of the flame and increased the dispersion of alternative fuel in the kiln, which
led to a higher clinker quality.

Additional measurements of the new burner were performed to study the impact of
burner settings. Swirl air level and direction of the axial air, were found to impact
the ignition of a petcoke flame. The air could also be used to disperse the SRF into
the secondary air stream to aid ignition and burnout. The burner settings were
linked to the cement clinker quality by means of a statistical analysis tool (Partial
Least Squares Regression), which showed that the increased dispersion of SRF
would increase the alite content of the clinker, indicating a higher quality.

SRF from two of the test plants and an additional plant was collected for fuel
characterization. The fuels were classified using a wind sieve setup, which showed
a distinction between light fuels used in the kiln and heavier fuels used in the
calciner. A characterization of fuel composition, particle size distribution, and shape
was made on two SRF samples. This resulted in a simplified description of the fuel
that can be implemented in computational models.

A one-dimensional model was developed for the rotary kiln. The model describes
and links together the fuel combustion, gas mixing, heat transfer, and clinker
chemistry. The model calculates temperatures in the gas phase and clinker bed
through the kiln and the clinker composition is given as output. Thus, the impacts of
co-firing different alternative fuels can be studied. It is also possible to explore
methods to reduce the negative effects of co-firing.

The model was used to study the influence of SRF co-firing in the kiln. It was found
that increased shares of SRF, reduced flame and bed temperatures, which caused an
increased free lime content in the clinker, indicating a decrease in clinker quality.
These effects could to some extent be avoided by increasing the energy input to the
kiln. An increased dispersion of SRF near the burner was also found to be beneficial,
confirming the conclusions from the industrial tests.
 III

RESUME (ABSTRACT IN DANISH)

Produktionen af cement er en energiintensiv proces, hvor omkring 30 % af de
løbende omkostninger traditionelt set er gået til brændsler. Disse omkostninger kan
dog sænkes ved at øge brugen af alternative brændsler i cementindustrien.
Derudover kan anvendelsen af affaldsbrændsler nedbringe nødvendigheden af
lossepladser, og delvist organiske brændsler kan sænke industriens udledning af
CO2. Brugen af alternative brændsler i cement-roterovne beror mest på tidligere
forsøg udført på de enkelte fabrikker. Systematisk viden omkring, hvordan disse
brændsler påvirker flammen og cementkvaliteten, mangler i litteraturen. Denne
afhandling forsøger at øge den fundamentale viden på dette område. Dette gøres ved
at gennemgå litteraturen og belyse de kendte udfordringer ved at anvende
alternative brændsler. Eksperimentelle studier udføres på storskala
cementfabrikker, hvor den samfyrede flamme undersøges, og en karakterisering af
alternative brændsler foretages i laboratoriet. Derudover udvikles en model til at
beskrive forbrændingen i cementovnen.

De alternative brændsler, som bruges i cementindustrien, kan være både faste og

flydende. De mest udbredte brændsler er opskårne dæk, kød- og benmel, samt fast
affaldsbrændsel (Solid Recovered Fuel – SRF). SRF fremstilles ved en mekanisk
behandling af ikke-giftigt husholdnings- eller industriaffald. Sammenlignet med
faste fossile brændsler har de fleste typer af alternative brændsler en større
partikelstørrelse, højere indhold af flygtige stoffer og fugt, samt en mindre
brændværdi. Dette gør anvendelsen af alternative brændsler i cementovne
udfordrende, da udbrændingstiden øges, og flammetemperaturen sænkes. En øget
forståelse af disse begrænsninger og hvordan de kan overkommes, er nødvendig for
yderligere at øge anvendelsen af alternative brændsler i cementovne.

Et eksperimentelt studie blev udført på tre forskellige cementanlæg. Flammerne i

cementovnene blev observeret vha. et specialfremstillet kamera, som kan indføres
igennem en inspektionslem i køleren og indsættes tæt på brænderen, hvilket
muliggør detaljerede observationer af flammen og brændslets opførsel. Forskellen
på flammer fyret med fossile og alternative brændsler blev undersøgt. Det blev
konkluderet at brugen af alternative brændsler ville forskyde antændelsespunktet
1-2 meter og sænke flammetemperaturen. På et af anlæggene blev der foretaget en
IV Resumé (Abstract in Danish)

udskiftning af brænderen. Designet blev ændret fra en brænder, hvor aksialluften

indføres gennem en ringformet luftdysse, til flere stråledysser (jets). Den nye
brænder sænkede antændingslængden af flammen og øgede spredningen af
alternativt brændsel i ovnen, hvilket øgede cementklinker-kvaliteten.

Yderligere målinger på den nye brænder blev foretaget for at undersøge

indflydelsen af brænderens indstillinger. Niveauet af swirl-luft (roterende) og
retning af aksialluften kan påvirke antændingen af en petroleumkoks-flamme
(petcoke). Luften kan også bruges til at sprede SRF ud i sekundærluften, hvilket
gavner antændingen og udbrændingen. Brænderindstillingerne blev koblet til
cementkvaliteten vha. en statistisk analyse (Partial Least Squares Regression), som
viste, at den øgede spredning af SRF forhøjede indholdet af alit i klinkeren, hvilket
indikerer en bedre cementkvalitet.

SRF fra to af testanlæggene, samt et andet anlæg, blev indsamlet for at foretage en
brændselskarakterisering. Brændslerne blev klassificeret i en vindsigte, hvilket
viste en klar forskel på lette brændsler, som bruges i cementovnen, og tungere
brændsler, som bruges i calcinatoren. En yderligere karakterisering af to
brændselsprøver blev foretaget for at bestemme størrelsesfordelingen og formen af
partikler samt sammensætningen af brændslet. Dette arbejde resulterede i nogle
simplificerende beskrivelser af brændslet, som kan implementeres i
beregningsmodeller.

En endimensionel model blev udviklet for cementovnen. Modellen beskriver og

sammenkæder forbrændingen, luftopblandingen, og klinkerkemien i ovnen.
Modellen kan benyttes til at udregne temperaturprofiler i gasfasen og klinkerlejet
gennem ovnen, hvorved klinkersammensætningen gives som output. Dermed kan
effekten af samfyring med forskellige alternative brændsler evalueres. Det er også
muligt at undersøge tiltag til at mindske den negative effekt af samfyring.

Modellen blev brugt til at undersøge indflydelsen af SRF-samfyring i cementovnen.

Det blev konkluderet, at en øget andel af SRF sænkede gassens og klinkerlejets
temperaturer, hvilket førte til en øget mængde fri kalk i klinkeren, som indikerer an
lavere klinkerkvalitet. Ved at øge energiinputtet, kunne denne effekt dog delvist
undgås. En øget spredning af SRF tæt på brænderen havde også en gavnlig effekt på
indholdet af fri kalk, hvilket underbygger konklusionerne fra de industrielle
undersøgelser.
 V

PREFACE
This PhD project was carried out at the Combustion and Harmful Emissions Control
(CHEC) Research Center of the Department of Chemical and Biochemical
Engineering at the Technical University of Denmark (DTU). The project is part of the
“Minerals and Cement Technology – MiCeTech” platform, which has been funded by
the Innovation Fund Denmark, FLSmidth A/S, Hempel, and the Technical University
of Denmark.

The supervisors for the project were professor Kim Dam-Johansen, senior
researcher Peter Arendt Jensen, senior researcher Sønnik Clausen, and project
manager Lars Skaarup Jensen (FLSmidth). I would like to thank them all for their
help and guidance during this project.

The video probes used for the flame investigations were designed by Sønnik Clausen
and manufactured by Jens Henry Poulsen of the workshop at the department. Their
help in this has been vital and is much appreciated.

Mads Nielsen of FLSmidth has been a great travel buddy at our frequent visits to
cement plants for measurements. He has been a key part of managing most of the
industrial tests, for which I am very grateful.

I am also thankful to my colleagues at FLSmidth who have participated in the

industrial tests, Christian Andersen, Morten Gabel, and Igor Pipiska. Furthermore, I
appreciate the help provided by personnel at the cement plants during tests.

David Jayanth Isaac of FLSmidth has made the CFD simulations used in this project,
for which I am also very grateful.

I would also like to thank Mohammadhadi Nakhaei for the collaboration and
discussions on characterization of alternative fuels and their combustion.

Furthermore, I much appreciate the work done by students Patrick Pedersen,

Casper Svith, Imran Khan, and Akhilesh Nair, who have been involved in the project.

Morten Nedergaard Pedersen

April 2018
VI
 VII

CONTENTS

ABSTRACT ...............................................................................................................................................I
RESUMÉ (ABSTRACT IN DANISH) ..................................................................................................... III
PREFACE ................................................................................................................................................ V
CONTENTS .......................................................................................................................................... VII
LIST OF FIGURES ............................................................................................................................... XIII
LIST OF TABLES ................................................................................................................................ XXI
ABBREVIATIONS............................................................................................................................. XXIII
NOMENCLATURE ............................................................................................................................ XXIV
1 INTRODUCTION ................................................................................................................................ 1
1.1 SCOPE OF THESIS ........................................................................................................................... 2
1.2 OUTLINE OF THESIS ...................................................................................................................... 2
1.3 PUBLICATIONS ............................................................................................................................... 3
2 CEMENT PRODUCTION AND CHEMISTRY .............................................................................. 5
2.1 PRODUCTION OF PORTLAND CEMENT ......................................................................................... 5
2.2 CHEMISTRY OF PORTLAND CEMENT ........................................................................................... 7
2.3 CLINKER REACTIONS..................................................................................................................... 9
2.4 CLINKER COMPOSITIONAL PARAMETERS ................................................................................. 10
2.5 CLINKER BURNABILITY ............................................................................................................... 12
2.6 THE CEMENT ROTARY KILN ...................................................................................................... 12
2.7 COMBUSTION AIR IN THE CEMENT KILN .................................................................................. 13
2.8 THE CEMENT KILN BURNER ...................................................................................................... 14
2.9 HISTORY OF KILN BURNER DESIGN........................................................................................... 16
2.9.1 Single Channel Burners............................................................................................... 17
2.9.2 Multichannel Burners .................................................................................................. 18
2.9.3 Low Primary Air Burners ........................................................................................... 19
2.9.4 Alternative Fuels Burners .......................................................................................... 21
2.10 SUMMARY AND CONCLUSIONS ................................................................................................. 22
3 FUELS IN THE CEMENT INDUSTRY ......................................................................................... 23
3.1 CONVENTIONAL FUELS USED IN THE CEMENT INDUSTRY ...................................................... 24
3.1.1 Coal ..................................................................................................................................... 24
VIII Contents

3.1.2 Petroleum Coke ............................................................................................................. 24

3.1.3 Gas ...................................................................................................................................... 25
3.1.4 Oil ........................................................................................................................................ 25
3.2 DRIVERS FOR ALTERNATIVE FUELS USAGE ............................................................................. 26
3.2.1 Cost of Fuel ...................................................................................................................... 26
3.2.2 Carbon Dioxide Emission Reductions................................................................... 27
3.2.3 Waste Management ..................................................................................................... 28
3.3 USE OF ALTERNATIVE FUELS..................................................................................................... 28
3.4 TYPES OF ALTERNATIVE FUELS................................................................................................. 29
3.4.1 Solid Recovered Fuel ................................................................................................... 30
3.5 CHALLENGES OF ALTERNATIVE FUELS FIRING ........................................................................ 32
3.5.1 Experiences of Alternative Fuels Co-Firing ........................................................ 33
3.5.2 Fuel Composition .......................................................................................................... 35
3.5.3 Particle Size ..................................................................................................................... 36
3.5.4 Moisture Content .......................................................................................................... 37
3.5.5 Kiln Dust ........................................................................................................................... 38
3.5.6 Reducing Conditions.................................................................................................... 39
3.5.7 Circulation of Volatile Species ................................................................................. 40
3.5.8 Chlorine Bypass ............................................................................................................. 43
3.6 SUMMARY AND CONCLUSIONS ................................................................................................... 43
4 KILN PROCESS MONITORING .................................................................................................... 45
4.1 CLINKER QUALITY....................................................................................................................... 45
4.2 BURNING ZONE TEMPERATURE ................................................................................................ 46
4.3 GAS MEASUREMENTS ................................................................................................................. 47
4.4 KILN TORQUE .............................................................................................................................. 47
4.5 KILN CAMERAS ............................................................................................................................ 48
4.6 KILN SCANNERS .......................................................................................................................... 49
4.7 MEASURES AGAINST CHANGING CONDITIONS ......................................................................... 49
4.8 SUMMARY AND CONCLUSIONS ................................................................................................... 50
5 EFFECT OF ALTERNATIVE FUELS CO-FIRING ON KILN FLAME................................... 51
5.1 DESCRIPTION OF TEST CEMENT PLANTS ................................................................................. 52
5.2 KILN CAMERA PROBE ................................................................................................................. 54
5.3 DESCRIPTION OF FLAME IMAGES .............................................................................................. 55
5.4 RESULTS FROM PLANT 1 ............................................................................................................ 57
Contents IX

5.4.1 Petcoke Fired Flame..................................................................................................... 57

5.4.2 Co-Fired Flame ............................................................................................................... 58
5.4.3 SRF Fired Flame ............................................................................................................. 60
5.5 RESULTS FROM PLANT 2 ............................................................................................................ 63
5.5.1 Fossil Fuel Fired Flame ............................................................................................... 64
5.5.2 Co-Fired Flame ............................................................................................................... 65
5.6 RESULTS FROM PLANT 3 ............................................................................................................ 68
5.6.1 Co-Fired Flame ............................................................................................................... 69
5.6.2 Flame with Natural Gas............................................................................................... 70
5.7 LIMITING FACTORS FOR ALTERNATIVE FUELS FIRING............................................................ 72
5.8 CONCLUSIONS .............................................................................................................................. 76
6 EFFECT OF BURNER SETTINGS ON THE CEMENT KILN FLAME AND CLINKER
QUALITY ................................................................................................................................................ 79
6.1 THE JETFLEX BURNER SETTINGS ............................................................................................... 80
6.2 CALCULATION OF BURNER MOMENTUM AND SWIRL .............................................................. 81
6.3 FLAME MEASUREMENTS............................................................................................................. 82
6.3.1 Petcoke Flame ................................................................................................................ 83
6.3.2 Co-fired Flame with Petcoke in Annular Channel ............................................ 87
6.3.3 Co-fired Flame with Petcoke in Central Channel .............................................. 92
6.3.4 Retraction of the Central Channel .......................................................................... 98
6.4 STATISTICAL DATA ANALYSIS ................................................................................................. 100
6.4.1 Impact of Process Variables ................................................................................... 103
6.4.2 Impact of Burner Settings ....................................................................................... 105
6.5 CONCLUSIONS ........................................................................................................................... 106
7 PHYSICAL CHARACTERIZATION OF SRF ............................................................................ 107
7.1 CHARACTERIZATION OF SRF IN LITERATURE ....................................................................... 108
7.2 WIND SIEVE SETUP .................................................................................................................. 109
7.2.1 Results from Wind Sieve Tests ............................................................................. 111
7.3 MANUAL FRACTIONATION OF SRF......................................................................................... 113
7.4 PARTICLE MASS DISTRIBUTION .............................................................................................. 114
7.5 PARTICLE SHAPE DETERMINATION BY IMAGING .................................................................. 116
7.6 CONCLUSIONS ........................................................................................................................... 119
8 CEMENT KILN MODEL DEVELOPMENT .............................................................................. 121
8.1 CEMENT KILN MODELING IN LITERATURE ............................................................................ 122
X Contents

8.1.1 Engineering Models ...................................................................................................122

8.1.2 Computational Fluid Dynamic Models ...............................................................125
8.1.3 Conclusions and Next Steps ....................................................................................127
8.2 MODELING OF PROCESSES IN THE CEMENT ROTARY KILN ..................................................127
8.3 MODEL OVERVIEW ...................................................................................................................128
8.4 SOLIDS MOVEMENT IN KILN ....................................................................................................129
8.5 CLINKER REACTIONS ................................................................................................................130
8.5.1 Equations for the Extent of Clinker Reactions ................................................130
8.5.2 Simplified Equations of Clinker Reactions ........................................................132
8.5.3 Kinetics of Clinker Reactions .................................................................................133
8.5.4 Enthalpy of Clinker Reactions and Melting ......................................................134
8.6 BURNER AIR FLOW ...................................................................................................................135
8.6.1 Jet Expansion ................................................................................................................136
8.6.2 External Recirculation and the Craya-Curtet Parameter ............................137
8.6.3 Swirling Flows .............................................................................................................138
8.6.4 Air Flow in the Cement Kiln....................................................................................139
8.6.5 Model for the Secondary Air Entrainment ........................................................141
8.7 HEAT TRANSFER IN THE CEMENT KILN .................................................................................143
8.7.1 Convective Heat Transfer ........................................................................................145
8.7.2 Conductive Heat Transfer........................................................................................146
8.7.3 Radiative Heat Transfer ...........................................................................................146
8.7.4 Heat Loss ........................................................................................................................147
8.8 COMBUSTION OF SOLID FUELS ................................................................................................149
8.8.1 Heating ............................................................................................................................149
8.8.2 Drying ..............................................................................................................................152
8.8.3 Devolatilization ...........................................................................................................152
8.8.4 Char Oxidation .............................................................................................................154
8.8.5 Fuel Combustion in the Clinker Bed ....................................................................157
8.8.6 Gas Phase Reactions ..................................................................................................160
8.8.7 Combustion Model Approach ................................................................................162
8.9 GAS PROPERTIES IN THE KILN MODEL ...................................................................................164
8.10 ENERGY AND MASS BALANCES IN KILN ...............................................................................166
8.10.1 Temperature in Clinker Bed ................................................................................166
8.10.2 Temperature in Gas Phase....................................................................................167
Contents XI

8.10.3 Mass Flow in Clinker Bed ..................................................................................... 168

8.10.4 Mass Flow in Gas Phase......................................................................................... 169
8.11 MODEL SOLUTION PROCEDURE ........................................................................................... 170
8.12 SUMMARY AND CONCLUSIONS .............................................................................................. 172
9 MODEL VALIDATION AND SRF CO-FIRING........................................................................ 173
9.1 MODEL VALIDATION – PILOT SCALE...................................................................................... 173
9.1.1 The Cemflame 3 Experiments ............................................................................... 173
9.1.2 Experimental Facilities ............................................................................................ 174
9.1.3 Experimental Results................................................................................................ 174
9.1.4 Model Input Parameters.......................................................................................... 176
9.1.5 Model Results .............................................................................................................. 178
9.2 MODEL VALIDATION – INDUSTRIAL SCALE ........................................................................... 181
9.2.1 The Industrial Tests .................................................................................................. 182
9.2.2 Model Results MBM Tests ....................................................................................... 183
9.2.3 Model Results SHW Tests ....................................................................................... 187
9.3 DISCUSSION OF MODEL PREDICTIONS ................................................................................... 190
9.4 CO-FIRING WITH SRF .............................................................................................................. 191
9.4.1 SRF Properties............................................................................................................. 191
9.4.2 Effect of Co-Firing Rate ............................................................................................ 193
9.5 POSSIBILITIES TO INCREASE SRF CO-FIRING ....................................................................... 196
9.5.1 Effect of Moisture Content ...................................................................................... 196
9.5.2 Effect of Particle Size ................................................................................................ 197
9.5.3 Effect of Increased Specific Energy Input ......................................................... 199
9.5.4 Effect of Burner Settings.......................................................................................... 200
9.5.5 Discussion of Options to Increase Co-Firing.................................................... 206
9.6 CONCLUSIONS ........................................................................................................................... 207
10 FINAL CONCLUSIONS .............................................................................................................. 209
10.1 SUGGESTIONS FOR FUTURE WORK ...................................................................................... 212
11 REFERENCES .............................................................................................................................. 215
12 APPENDICES ............................................................................................................................... 233
A. FLAME ANALYSIS OF IMAGE INTENSITIES ........................................................................... 234
B. DETERMINATION OF KINETICS FOR ALITE FORMATION .................................................... 236
C. DETERMINATION OF VIEW FACTORS FOR RADIATION ...................................................... 238
D. OXYGEN REQUIREMENT FOR COMBUSTION ........................................................................ 241
XII Contents

E. VALIDATION OF BIOMASS COMBUSTION MODEL ................................................................243

Details of Experiments .................................................................................................243
Fuels Used .........................................................................................................................243
Kinetics Used in Model .................................................................................................245
Model Predictions Devolatilization .........................................................................245
Model Predictions Char Oxidation...........................................................................248
F. COMPARISON OF ISOTHERMAL AND NON-ISOTHERMAL DEVOLATILIZATION MODEL....250
Comparison of Models ..................................................................................................250
Adjusting Devolatilization Kinetics of Isothermal Model ...............................251
G. BED AND GAS PHASE ENERGY BALANCES ...........................................................................255
Energy Balance for the Bed Phase ...........................................................................255
Energy Balance for the Gas Phase ...........................................................................259
H. MODEL PARAMETERS FOR INDUSTRIAL COMBUSTION SIMULATIONS ..............................262
Standard Kiln Parameters ..........................................................................................262
Standard Gas Parameters ...........................................................................................263
Standard Parameters for Fuel...................................................................................264
Standard Parameters for Clinker .............................................................................265
Parameters for MBM Simulations ...........................................................................266
Parameters for SHW Simulations ............................................................................267
Parameters for SRF Simulations ..............................................................................268
I. DETAILS OF COMBUSTION SIMULATIONS .............................................................................269
Details of Coal Combustion Simulation...................................................................269
Details of SRF Co-Firing Simulation .........................................................................273
J. MODEL FOR THE CLINKER COOLER ......................................................................................279
Determining the Cooler Heat Transfer Coefficient .............................................281
 XIII

LIST OF FIGURES
Figure 2-1: Process flow sheet of a cement plant with preheater tower consisting of cyclones and
calciner [14]. ......................................................................................................................................................................6
Figure 2-2: Simplified phase diagram showing the cement clinker reactions and corresponding
temperatures as solids pass through the kiln system from left to right [22]. ......................................9
Figure 2-3: A burner in a grate cooler cement kiln indicating temperature zones in the kiln and
different air flows. ........................................................................................................................................................ 13
Figure 2-4: The Jetflex burner, a modern multichannel cement kiln burner from FLSmidth A/S. The
additional channel can e.g. be used for oxygen enrichment, liquid fuel firing, or a flame detector.
.............................................................................................................................................................................................. 15
Figure 2-5: Development of the cement burner technology since 1960 and the main fuels used
[30,37]. .............................................................................................................................................................................. 16
Figure 2-6: A typical axial flame from a single channel kiln burner [40]. ERZ = External Recirculation
Zone .................................................................................................................................................................................... 18
Figure 2-7: Multichannel burner coal flame [40]. ERZ = External Recirculation Zone, IRZ = internal
Recirculation Zone. ...................................................................................................................................................... 19
Figure 2-8: Flow pattern in a low NOX burner [43]. Note that the swirl channel is located inside the
coal channel. ................................................................................................................................................................... 20
Figure 3-1: Use of fossil and biomass based alternative fuels in selected countries and regions [62].
Fossil alternative fuels include e.g. waste oil, tires, plastics, refuse derived fuel; biomass include
e.g. agricultural waste, sewage sludge, charcoal. CIS = Commonwealth of Independent States.
.............................................................................................................................................................................................. 29
Figure 3-2: Distribution of alternative fuels in the German (a) and Austrian (b) cement sectors in %
of annual thermal use in 2015 [67,68]. .............................................................................................................. 30
Figure 3-3: Composition of Solid Recovered Fuel based on visual sorting [74,75]. ................................... 31
Figure 3-4: Effect of the amount of swirl and axial air on a flame co-fired with 30 % pulverized sewage
sludge and comparison to coal flame (base). The temperatures are measured as kiln wall
temperature [80]. ......................................................................................................................................................... 33
Figure 3-5: Free lime, SO3, and P2O5 content in clinker for different MBM feed rates [81]. ................... 34
Figure 3-6 Adiabatic flame temperature for various fuels from Table 3-4 calculated at different
moisture contents at an air to fuel ratio of 1. .................................................................................................. 38
Figure 3-7: Build up tendency as function of SO3 and Cl in the hot meal [112]. .......................................... 42
Figure 4-1: Image from a kiln camera showing the combustion in the kiln [130]. The hot clinker bed
can be seen on the left side. Unignited pulverized fuel is seen directly in front of the burner. The
combustion zone (bright area at center) is further removed from the burner. .............................. 48
Figure 4-2: Kiln shell temperatures when firing petcoke (a) and co-firing with 10 % sewage sludge
by energy (b). Temperatures are lower around 22-27 m when co-firing, but higher around 18-
20 m [31].......................................................................................................................................................................... 49
Figure 5-1: SRF used at plant 1 (a), Plant 2 (b), and Plant 3 (c). ........................................................................ 54
Figure 5-2: a) The camera inserted in the probe. b) Camera probe inserted through side of kiln hood.
c) Camera probe inserted next to kiln burner. ............................................................................................... 55
Figure 5-3: a) Sketch of the camera view (top view). b) Explanation of camera view. ............................ 56
Figure 5-4: Images taken 0.4 seconds apart showing the fluctuations in the kiln flame. d) Frame
averaged over 5 seconds. Images from Plant 1 during full petcoke load. ........................................... 57
Figure 5-5: Images during full petcoke firing of old burner (a+b) and new burner (c+d). Old burner
operating with 250 mbarg swirl air pressure and new burner operating with 190 mbarg swirl
air pressure. .................................................................................................................................................................... 58
XIV List of Figures

Figure 5-6: Images during co-firing of petcoke and SRF of old burner (a+b) and new burner (c+d).
The old burner operates with 80 % energy by SRF and 240 mbarg swirl air pressure. The new
burner operates with 70 % energy by SRF and 180 mbarg swirl air pressure................................ 59
Figure 5-7: Co-firing with the new burner at Plant 1 with 180 (a) and 100 (b) mbarg swirl air
pressure. Burning SRF particles are highlighted by green circles. Distances are not indicated in
the figure since no reference size is present.................................................................................................... 60
Figure 5-8: Images during 100 % SRF firing for old burner (a+b+c) and new burner (d+e+f). The
old burner operates a swirl air pressure of 240 mbarg. The new burner with a swirl air pressure
of 150 mbarg. The camera has been turned further downstream in image c, than in the other
images. .............................................................................................................................................................................. 61
Figure 5-9: a) Fuel dosing of coal/petcoke, and SRF and the kiln drive power consumption (secondary
axis) during the test day. b) Gradient of the power consumption. Data from Plant 2. ................. 64
Figure 5-10: Images of the coal/petcoke fired flame with 10.5 t/h coal/petcoke. ..................................... 65
Figure 5-11: Images of co-firing of coal/petcoke with SRF. 7.5 t/h coal/petcoke is used with 4.5 t/h
SRF (30 % SRF energy input). ................................................................................................................................ 65
Figure 5-12: Comparison of the coal/petcoke fired flame (a) with the SRF co-fired flame (b). Images
are averaged over 5 seconds................................................................................................................................... 66
Figure 5-13: View along the kiln wall and charge under the burner for coal-fired case (a) and co-fired
case (b+c). The kiln wall is seen in the left side of the images. The kiln bed can be seen on the
right side of the images having a more orange color than the wall. Burning particles are bright
spots in the images, which have been highlighted with green circles. ................................................. 67
Figure 5-14: a) Fuel dosing to the kiln burner during the test day. b) Temperatures as measured by
pyrometer, thermographic camera, kiln hood thermocouple and the kiln torque during the test
day. Data from Plant 3................................................................................................................................................ 68
Figure 5-15: Flame of petcoke co-fired with SRF and sewage sludge. a+b) 3 t/h petcoke, 1.5 t/h
sewage sludge, 6 t/h SRF (60 % AF energy input), recorded at 14:56. c) 4.3 t/h petcoke, 1.5 t/h
sewage sludge, 5 t/h SRF (45 % AF energy input), recorded at 15:45. ............................................... 69
Figure 5-16: View of the co-fired flame from the side (a) and under the burner (b). Flames in bed are
located by green circles. Fuels: 4.3 t/h petcoke, 5 t/h SRF, 1.5 t/h sewage sludge (45 % AF
energy input). ................................................................................................................................................................ 70
Figure 5-17: Different flame shapes while the gas flow is increased over 200 seconds from 0 (a+d)
to 500 (c+f) Nm3/h. Petcoke dosing is constant at 4.3 t/h. Top row shows single frames and
bottom row are images averaged over 5 seconds. ........................................................................................ 71
Figure 5-18: SRF and gas fired flame. 5 t/h SRF, 1.5 t/h sewage sludge, and 4500 Nm 3/h natural gas
(45 % AF energy input). ........................................................................................................................................... 71
Figure 5-19: Relationship between the energy from alternative fuels firing at the main burner (MB)
and the energy input in the kiln (red dots) and the calciner (black x) for Plant 1 (a) and Plant 3
(b). Data are based on hourly averages for one month of operation. ρ is Pearson correlation
coefficient. ....................................................................................................................................................................... 73
Figure 6-1: The Jetflex axial air nozzles can be turned allowing different nozzle configurations such
as: a) All nozzles pointing inwards 0°. b) All nozzles turned 30° [145]. The blue lines indicate
the entrainment of secondary air into the burner jet. ................................................................................. 80
Figure 6-2: Burner in normal operation (a) and with retracted center (b) [145]. .................................... 81
Figure 6-3: Effect of swirl air pressure on the petcoke flame with axial air nozzles at 30°. Swirl
pressure of a) 30 mbarg, b) 80 mbarg, c) 180 mbarg. ................................................................................ 83
Figure 6-4: Intensity profiles over time when swirl pressure is changed from 180 to 30 mbarg. a)
along the horizontal centerline, and b) across the image 1 m in front of the burner. .................. 84
Figure 6-5: Effect of axial air nozzle configuration on the petcoke flame with swirl pressure of 180
mbarg. Axial air nozzle position of a) 0°, b) 30°, c) 180°. .......................................................................... 86
Figure 6-6: Intensity profiles over time when nozzles are changed from 0 to 180°. a) along the
horizontal centerline, and b) across the image 1 m in front of the burner. ....................................... 86
List of Figures XV

Figure 6-7: Intensity profiles for the petcoke flame images with different swirl (S in mbarg) and
nozzle configurations (N in °) a) along the horizontal centerline, and b) across the image 1 m in
front of the burner. ...................................................................................................................................................... 87
Figure 6-8: Effect of swirl pressure on the co-fired flame with axial air nozzles at 60°. Swirl pressure
of a) 50 mbarg, b) 150 mbarg. SRF contributes 60 % energy. ................................................................. 88
Figure 6-9: Intensity profiles over time when swirl pressure is changed from 50 to 150 mbarg. a)
along the horizontal centerline, and b) across the image 1 m in front of the burner. ................... 89
Figure 6-10: Effect of swirl pressure on the co-fired flame with axial air nozzles at 30° using
alternative camera view. Swirl pressure of a) 100 mbarg, b) 180 mbarg. SRF contributes 70 %
energy. ............................................................................................................................................................................... 89
Figure 6-11: Effect of axial air nozzle configuration on the co-fired flame with swirl pressure of 150
mbarg. Axial air nozzle position of a) 180°, b) 60°. SRF contributes 60 % energy. ....................... 90
Figure 6-12: Intensity profiles over time when nozzles are changed from 180 to 60°. a) along the
horizontal centerline, and b) across the image 1 m in front of the burner. ....................................... 91
Figure 6-13: Intensity profiles for the co-fired flame images with different swirl (S in mbarg) and
nozzle configurations (N in °) a) along the horizontal centerline, and b) across the image 1 m in
front of the burner. ...................................................................................................................................................... 92
Figure 6-14: Effect of swirl pressure on the co-fired flame with axial air nozzles at 60° and petcoke in
center. Swirl pressure of a) 150 mbarg, b) 10 mbarg c) 180 mbarg. SRF contributes 65 %
energy. ............................................................................................................................................................................... 93
Figure 6-15: Intensity profiles over time when swirl pressure is changed from 150 to 10 to 180
mbarg. a) along the horizontal centerline, and b) across the image 1 m in front of the burner.
.............................................................................................................................................................................................. 94
Figure 6-16: Effect of swirl pressure on the co-fired flame with axial air nozzles at 30° and petcoke in
central channel. Swirl pressure of a) 50 mbarg, b) 180 mbarg. SRF contributes 70 % energy.
.............................................................................................................................................................................................. 95
Figure 6-17: Effect of axial air nozzle configuration on the co-fired flame with swirl pressure of 150
mbarg. Axial air nozzle position of a) 0°, b) 30° c) 60°, d) 180°. SRF contributes 65 % energy.
.............................................................................................................................................................................................. 96
Figure 6-18: Intensity profiles over time when nozzles are changed from 0 to 180°. a) along the
horizontal centerline, and b) across the image 1 m in front of the burner. ....................................... 97
Figure 6-19: Intensity profiles for the co-fired flame with petcoke in central channel images with
different swirl (S in mbarg) and nozzle configurations (N in °) a) along the horizontal
centerline, and b) across the image 1 m in front of the burner. .............................................................. 98
Figure 6-20: Effect of retracting burner center with axial air nozzles at 180° and swirl 180 mbarg. a)
Center normal, b) Center Retracted. SRF contributes 90 % energy. ..................................................... 99
Figure 6-21: Intensity profiles for the images with center forward (F) or retracted (R) a) along the
horizontal centerline, and b) across the image 0.1 m in front of the burner. ................................... 99
Figure 6-22: a) MSEP values as function of number of latent variables in PLSR. b) Comparison of
measured and modeled values of clinker alite content using 4 latent variables, in normalized
form. ................................................................................................................................................................................ 103
Figure 6-23: Regression coefficients in the PLSR model (𝒀𝑷𝑳𝑺𝑹=XPLSRBPLSR). Error bars are 95 %
confidence interval determined by 10-fold cross validation. Variables are explained in Table
6-3. ................................................................................................................................................................................... 103
Figure 7-1: TGA experiment of SRF sample showing two distinct mass loss steps [165]. ................... 109
Figure 7-2: Sketch of the wind sieve setup used to characterize alternative fuels. ................................ 110
Figure 7-3: Results of wind sieve characterization of various RDFs.............................................................. 112
Figure 7-4: Results of manual sorting and classification of RDF-A and RDF-B. Fines are particles under
2 mm as determined by sieving. ......................................................................................................................... 114
XVI List of Figures

Figure 7-5: Mass distribution fitted to Rosin-Rammler distribution of the plastic and biomass
fractions of RDF-A and RDF-B. ............................................................................................................................ 115
Figure 7-6: Camera setup with two cameras to determine particle size and shape. .............................. 116
Figure 7-7: Sphericity of biomass and plastic of RDF-A and RDF-B in relation to the particle mass.
........................................................................................................................................................................................... 118
Figure 8-1: CFD simulations showing temperature profiles in the cement kiln for different co-firing
scenarios of a) lignite, b) 50 % SRF, and c) 50 % optimized SRF [163]. Notice the high
temperatures at the lower kiln wall halfway through the kiln for case b and c. .......................... 126
Figure 8-2: Sketch of the kiln model showing the main processes accounted for in the model. ...... 128
Figure 8-3: Geometry of a rolling bed indicating the applied nomenclature. ............................................ 130
Figure 8-4: Sketch of the development of a turbulent jet in a pipe through different regions [218].
Region 1: The flow development region, Region 2: Fully developed region, Region 3: Possible
formation of recirculation zones. Region 4: Fully developed pipe flow. In this figure only, U 1
denotes co-flow velocity, U velocity difference (u-U1), D diameter, b the jet radius. Subscripts 1
denote co-flowing stream, and m the center value. r, x, and u denote radius, axial distance, and
axial velocity (as used in the remainder of the thesis). ........................................................................... 136
Figure 8-5: Examples of acid-alkali mixing used to investigate the kiln flame. The dark plume (alkali)
represents the cement kiln flame. The dark color dissipates as the alkali is mixed with acid,
representing the secondary air. Top and bottom images are single frames and middle picture is
averaged over 60 seconds [223]. ....................................................................................................................... 141
Figure 8-6: a) The jet radius as function of axial distance from burner with different burner axial
momentum and swirl. b) Zoomed version of a, with regression lines. Data derived from CFD
simulations of a petcoke flame performed by David Isaac. .................................................................... 142
Figure 8-7: Paths of heat transfer in the cement rotary kiln. ............................................................................ 144
Figure 8-8: Trajectories (a) and time in suspension (b) for particles with diameter 0.1-10 mm
injected in kiln. up0 = 30 m/s, ρp = 1000 kg/m3, ug = 5.8 m/s, Tg = 750 °C, rk = 1.95 m. ........ 158
Figure 8-9: Sketch of the landing spot and particle conversion using normal distributions for particles
with diameter 0.1-10 mm. Parameters are similar to Figure 8-8. ...................................................... 159
Figure 8-10: The eddy dissipation rate as function of distance from the burner in the kiln. a) Normal
data, b) log transformed data. Data derived from CFD simulations of a petcoke flame performed
by David Isaac. ............................................................................................................................................................ 161
Figure 8-11: The energy sources in a section of the bed used to derive the temperature equation for
the bed. .......................................................................................................................................................................... 166
Figure 8-12: The energy sources in a section of the gas phase used to derive the temperature equation
for the gas phase........................................................................................................................................................ 168
Figure 8-13: Example of the iterative procedure (shooting method) to find initial solution of the gas
and bed temperature. Solid line indicates bed temperature and dashed line indicates gas
temperature. ............................................................................................................................................................... 171
Figure 8-14: The temperature profiles in the gas and bed phase before (dashed lines) and after
convergence (solid lines) has been reached. Bed temperature without clinker reactions (dotted
line) is also shown. ................................................................................................................................................... 172
Figure 9-1: Measurements along the axial centerline of the kiln flame. a) CO concentration, burnout,
gas temperature and NOx concentration. b) oxygen concentration, hydrocarbon concentration,
and radiative heat flux upon the kiln walls [80]. ........................................................................................ 175
Figure 9-2: Kiln wall temperatures of baseline coal flame experiments and co-firing with different
amounts of plastic foils in % of the energy input [80]. ............................................................................ 175
Figure 9-3: a) Size distribution of fuel as reported in the Cemflame experiments [80]. b) Rosin
Rammler plot of the fetnuss coal size distribution. ................................................................................... 176
Figure 9-4: Comparison of measurements and model predictions of the gas temperature through the
kiln simulator of the Cemflame experiments [80]. Lines are model and markers are
List of Figures XVII

experimental points. The peak temperatures are measured at the axial centerline of the flame.
The average temperatures are the mean based on traverses in radial direction through the
flame. .............................................................................................................................................................................. 179
Figure 9-5: a) Comparison of measurements and model predictions of gas concentration of CO,
oxygen, and hydrocarbons (CH4) and average fuel burnout of the Cemflame experiments [80].
Lines are model and markers are experimental points. b) Conversion of the 10 discrete fuel
sizes according to model. ....................................................................................................................................... 180
Figure 9-6: Radial traverses of a) CO and hydrocarbon (HC) concentrations and b) gas temperatures
in the Cemflame experiments [80]. ................................................................................................................... 181
Figure 9-7: Comparison of clinker free lime content measured in experimental tests by Ariyaratne
[81] and corresponding model calculations during MBM tests. Error bars indicate standard
deviation in measurements. ................................................................................................................................. 184
Figure 9-8: Model simulation results of gas temperatures (a) and bed temperatures (b) through the
kiln for different co-firing scenarios with MBM. ......................................................................................... 185
Figure 9-9: Comparison of clinker free lime content measured in experimental tests by Ariyaratne
[82] and corresponding model calculations during SHW tests. Error bars indicate standard
deviation in measurements. ................................................................................................................................. 188
Figure 9-10: Model simulation results of gas temperatures (a) and bed temperatures (b) through the
kiln for different co-firing scenarios with SHW. .......................................................................................... 189
Figure 9-11: The clinker free lime content as the share of SRF energy input is increased.................. 194
Figure 9-12: Model simulation results of gas temperatures (a) and bed temperatures (b) for different
energy inputs of SRF. ............................................................................................................................................... 195
Figure 9-13: The clinker free lime content for different moisture content in the SRF. SRF energy input
is 30 %. ........................................................................................................................................................................... 196
Figure 9-14: Model simulation results of gas temperatures (a) and bed temperatures (b) for different
SRF moisture contents. SRF energy input is 30 %. .................................................................................... 197
Figure 9-15: The clinker free lime content for different SRF particle sizes. The standard SRF particle
diameter is multiplied by a factor in the range 0.1-1. SRF energy input is 30 %. ........................ 198
Figure 9-16: Model simulation results of gas temperatures (a) and bed temperatures (b) for different
SRF particle sizes. The indicated factor is multiplied to the original particle diameter. SRF
energy input is 30 %. ............................................................................................................................................... 198
Figure 9-17: The clinker free lime content for increasing specific energy input by 1. Increasing coal
and SRF feed rate, 2. Increasing SRF feed rate, 3. Decreasing the raw meal feed rate. SRF energy
input is 30 % in case 1 and 3, but changes in case 2. ................................................................................ 199
Figure 9-18: Model simulation results of gas temperatures (a) and bed temperatures (b) for different
increases in specific energy consumption by reduction of raw meal feed. SRF energy input is 30
%. ...................................................................................................................................................................................... 200
Figure 9-19: The clinker free lime content for different cases of burner settings. SRF energy input is
30 %. ............................................................................................................................................................................... 201
Figure 9-20: Model simulation results of gas temperatures (a) and bed temperatures (b) for different
burner setting cases. SRF energy input is 30 %. ......................................................................................... 202
Figure 9-21: The clinker free lime content for different secondary air entrainment rates. The standard
entrainment rate is multiplied by a factor in the range 0.5-1.5. SRF energy input is 30 %. .... 203
Figure 9-22: Model simulation results of gas temperatures (a) and bed temperatures (b) for different
entrainment rates. The indicated factor is multiplied to the original entrainment rate. SRF
energy input is 30 %. ............................................................................................................................................... 204
Figure 9-23: The clinker free lime content for different landing places of the SRF. The standard
landing place is multiplied by a factor in the range 0.5-1. SRF energy input is 30 %................. 205
XVIII List of Figures

Figure 9-24: Model simulation results of gas temperatures (a) and bed temperatures (b) for different
SRF particle landing factors. The indicated factor is multiplied to the original landing place. SRF
energy input is 30 %. ............................................................................................................................................... 205
Figure 12-1: a-c) Three example images in grayscale with varying image intensity. The image
intensity increases from top left to bottom right in each image. d-f) Image intensities in
horizontal direction from left to right, and in vertical direction from top to bottom. ............... 234
Figure 12-2: Image intensities in the 3 example images. a) in horizontal direction in center of image
(pixel 1 is left side of Figure 12-2) , and b) in vertical direction in right most pixel column (pixel
1 is top of Figure 12-2). .......................................................................................................................................... 235
Figure 12-3: Comparison of model predictions with laboratory burnability tests for sample 2. a)
Using standard kinetics (A = 3*108 m3/kg/s, Ea = 440 kJ/mol). b) using optimized kinetics.
........................................................................................................................................................................................... 237
Figure 12-4: Sketch to determine the radiation view factors. .......................................................................... 238
Figure 12-5: Model and experimental comparison for devolatilization. ..................................................... 247
Figure 12-6: Model and experimental comparison for char oxidation. ........................................................ 249
Figure 12-7: Comparison of devolatilization times predicted by isothermal and non-isothermal
model for different gas temperatures (°C). ................................................................................................... 250
Figure 12-8: Linearized versions of Figure 12-7 for a) dp < 1 mm and b) dp > 1 mm in a log-log plot.
........................................................................................................................................................................................... 251
Figure 12-9: Comparison of devolatilization times predicted by isothermal and non-isothermal
models after adjustment of preexponential factor (a) and activation energy (b). ...................... 252
Figure 12-10: Comparison of devolatilization times predicted by isothermal and non-isothermal
models using adjusted kinetics for a) dp < 1 mm and b) dp > 1 mm. ................................................ 254
Figure 12-11: Temperatures through the kiln of the wall, gas, bed (without clinker reactions as
dashed line), fuel average temperature, secondary air, and outer shell. ......................................... 269
Figure 12-12: The mass fraction of the different clinker phases through the kiln. ................................. 271
Figure 12-13: The mole and mass fraction of N2, O2, CO2, H2O, CH4 and CO through the kiln for a coal
fired simulation.......................................................................................................................................................... 272
Figure 12-14: Temperatures (a), extent of drying (b), extent of devolatilization (c), and extent of char
oxidation (d) for the 10 discrete particle size classes of coal particles. ........................................... 273
Figure 12-15: Temperatures through the kiln of the wall, gas, bed fuel average temperature,
secondary air, and outer shell. ............................................................................................................................ 274
Figure 12-16: The mass fraction of the different clinker phases through the kiln. ................................. 275
Figure 12-17: The mole and mass fraction of N2, O2, CO2, H2O, CH4 and CO through the kiln for a coal
fired simulation.......................................................................................................................................................... 275
Figure 12-18: Temperatures (a), extent of drying (b), extent of devolatilization (c), and extent of char
oxidation (d) for the 10 discrete particle size classes of coal particles. ........................................... 276
Figure 12-19: Temperatures (a), extent of drying (b), extent of devolatilization (c), and extent of char
oxidation (d) for the 6 discrete particle size classes of biomass particles in SRF. ‘x’-markers
indicate the distance of fuel landing in the clinker bed. .......................................................................... 277
Figure 12-20: Temperatures (a), extent of drying (b), extent of devolatilization (c), and extent of
melting (d) for the 6 discrete particle size classes of plastic particles in SRF. ‘x’-markers indicate
the distance of fuel landing in the clinker bed. ............................................................................................ 278
Figure 12-21: A grate cooler in a cement plant. The clinker is moved forward by reciprocating grates.
Cooling air is forced through the clinker bed by air fans. The heated air is used as secondary air
in the cement kiln [33]. .......................................................................................................................................... 279
Figure 12-22: Sketch of a simplified cooler model divided into Ncooler segments. ................................... 280
Figure 12-23: Absolute difference between the secondary air temperature and the average air
temperature out of the cooler as function of Hcooler. .................................................................................. 282
List of Figures XIX

Figure 12-24: The gas temperatures (solid lines) through the 5 bed segments, with bed temperatures
(dashed lines) and the average gas outlet temperature (dotted line). ............................................. 283
XX
 XXI

LIST OF TABLES
Table 2-1: Typical composition of Portland cement clinker [11]. .........................................................................8
Table 3-1: Relative cost of various fossil fuels per energy unit [39]. ................................................................ 26
Table 3-2: Indicative net CO2 emissions from various fuels used in the cement industry [59]. ........... 27
Table 3-3: Example on the quality variation of SRF. Measurements based on one SRF supplier over
4.5 years [77]. ................................................................................................................................................................ 32
Table 3-4: Proximate and ultimate analysis of select fuels used in the cement industry. ....................... 36
Table 3-5: Typical compounds found in deposits in the cement kiln and preheater [23,112]. ............ 42
Table 5-1: Properties of the fuels utilized at the cement plants. Data are on an as received basis.
Moisture and ash in coal and petcoke depends on sampling before (Plant 2 and 3) or after
milling (Plant 1). ........................................................................................................................................................... 53
Table 5-2: Amount and pressure of primary air typically used in the burners at the three cement
plants. Plant 3 uses a burner with no separate swirl air channel. .......................................................... 54
Table 5-3: Comparison of key operating parameters for the old and new kiln burners at Plant 1.
Values are averages for one month of operation. Ignition point data from Figure 5-5 and Figure
5-6. ...................................................................................................................................................................................... 63
Table 5-4: Overview of the impact of alternative fuels on the flame ignition point at the three cement
plants. C: Coal, PC: Petcoke, SRF: Solid Recovered Fuel, SS: Sewage Sludge. ..................................... 77
Table 6-1: Overview of burner pressures and corresponding velocities, axial momentum and the
approximated swirl number. .................................................................................................................................. 82
Table 6-2: Overview of burner setting tests. ............................................................................................................ 100
Table 6-3: Overview of variables used to predict the clinker alite content in the PLSR model. ........ 102
Table 7-1: Example calculations on particle shape. Terminal velocity determined at 20 °C with
assumed drag coefficient of 0.44. ....................................................................................................................... 110
Table 7-2: Details of the fuels tested in the wind sieve. ....................................................................................... 111
Table 7-3: Parameters for the Rosin-Rammler distribution of RDF-A and RDF-B samples. ............... 116
Table 7-4: Mass weighted average sphericity for biomass and plastics fractions of RDF-A and RDF-B.
........................................................................................................................................................................................... 118
Table 8-1: Reactions and kinetics for the model by Mastorakos et al. [201]. ............................................ 132
Table 8-2: Preexponential factors and activation energies to determine rate constants for the clinker
reactions. ....................................................................................................................................................................... 133
Table 8-3: Enthalpy of the clinker reactions and the melting of clinker [217]. ........................................ 135
Table 8-4: Momentum and swirl number in CFD simulations, performed by David Jayanth Isaac, used
to determine secondary air entrainment. ...................................................................................................... 142
Table 8-5: Momentum and swirl number in CFD simulations and corresponding values of K edr,1 and
Kedr,2. ................................................................................................................................................................................ 162
Table 9-1: Model parameters used to model the Cemflame experiments. .................................................. 177
Table 9-2: Estimated input parameters for to model the Cemflame experiments. Parameters marked
by * have been assumed. ........................................................................................................................................ 178
Table 9-3: Details of the meat and bone meal (MBM) co-firing tests [81]. ................................................. 182
Table 9-4: Details of the solid hazardous waste (SHW) co-firing tests [82]. ............................................. 182
Table 9-5: Details of the fuels used in the industrial tests. ................................................................................. 183
XXII List of Tables

Table 9-6: Average clinker oxide composition and calculated LSF, SR, AR, and Bogue C3S (not
corrected and corrected for free lime) during the MBM and SHW industrial tests of Ariyaratne
et al. [81,82]. ................................................................................................................................................................ 187
Table 9-7: Distribution of the SRF composition, particle mass and equivalent diameter, and sphericity
used for simulating SRF co-firing....................................................................................................................... 192
Table 9-8: Composition of the SRF and the Plastic, Inert, and Biomass fractions used for modeling SRF
co-firing. ........................................................................................................................................................................ 193
Table 9-9: The feed rate of coal, SRF, and raw meal used for simulating the effect of SRF co-firing.
........................................................................................................................................................................................... 194
Table 9-10: Axial momentum and swirl number used to determine the effect of burner settings on
clinker free lime and the corresponding values of Kent,jet and ε/k calculated at 10 and 25 m from
the clinker exit. ........................................................................................................................................................... 201
Table 12-1: Raw meal composition of test samples. ............................................................................................. 236
Table 12-2: Measured CaO, C2S, and C3S in burnability tests. ........................................................................... 236
Table 12-3: Optimized best fitting kinetics for the alite formation. ............................................................... 237
Table 12-4: The proximate and ultimate analysis of the fuels used in the model studies. Including the
calculated oxygen consumption and adiabatic temperature by simple (1) and rigorous method
(2). ................................................................................................................................................................................... 242
Table 12-5: Experimental conditions used in literature. .................................................................................... 243
Table 12-6: Properties of the fuels used in experiments. ................................................................................... 244
Table 12-7: Kinetics of biomass combustion reactions and additional physical parameters. ........... 245
Table 12-8: Values for regression parameters to determine isothermal kinetics. .................................. 253
Table 12-9: Parameters for the kiln in industrial scale validation. Parameters marked by * have been
assumed. ....................................................................................................................................................................... 262
Table 12-10: Parameters for the primary and secondary air inlet for industrial scale model validation.
........................................................................................................................................................................................... 263
Table 12-11: Standard parameters for fuels used in the simulations. .......................................................... 264
Table 12-12: Standard parameters for the clinker/kiln bed. Parameters marked by * have been
assumed. ....................................................................................................................................................................... 265
Table 12-13: Flows of fuel and secondary air with secondary air temperature and oxygen
concentration (mol%) in flue gas for the MBM simulations. ................................................................ 266
Table 12-14: Particle diameter of coal and MBM particles used in simulations. ..................................... 266
Table 12-15: Flows of fuel and secondary air with secondary air temperature and oxygen
concentration (mol%) in flue gas for the SHW simulations. ................................................................. 267
Table 12-16: Particle diameter of coal and SHW particles used in simulations. ...................................... 267
Table 12-17: Flows of fuel and secondary air with secondary air temperature and oxygen
concentration (mol%) in flue gas for the SRF simulations. ................................................................... 268
Table 12-18: Particle diameter of coal and SRF particles used in simulations. ........................................ 268
Table 12-19: Cooler conditions for the calibration calculations based on coal validation case Test 1-
1. ....................................................................................................................................................................................... 281
 XXIII

ABBREVIATIONS
Abbreviation Explanation
AF Alternative Fuel
AR Alumina Ratio
ASR Alkali Sulfate Ratio
a.r. As Received
CCAI Calculated Carbon Aromaticity Index
CFD Computational Fluid Dynamics
CIS Commonwealth of Independent States
EC Experimental Condition
ERZ External Recirculation Zone
HC HydroCarbon
HD HotDisc
ID Fan Induced Draft Fan
IFRF International Flame Research Foundation
IRZ Internal Recirculation Zone
LHV Lower Heating Value
LSF Lime Saturation Factor
MB Main Burner
MBM Meat and Bone Meal
MSEP Mean Square Error of Prediction
MSW Municipal Solid Waste
NIR Near Infrared
PLSR Partial Least Squares Regression
RDF Refuse Derived Fuel
RMSE Root Mean Square Error
RPM Revolutions Per Minute
PC PetCoke
SHW Solid Hazardous Waste
SR Silica Ratio
SRF Solid Recovered Fuel
SS Sewage Sludge
TGA ThermoGravimetric Analysis
VDZ Verein Deutscher Zementwerke (German Cement Works Association)
WBCSD World Business Council for Sustainable Development
WSGG Weighted Sum of Grey Gasses
XRD X-Ray Diffraction
XRF X-Ray Fluorescence
XXIV

NOMENCLATURE
Roman Explanation Unit
A Preexponential factor in Arrhenius Unit dependent on
expression order of reaction
A Area m2
a Weighting factor in WSGG model -
Bi Biot number -
Bp Secondary air requirement for kg/kg fuel
stoichiometric fuel combustion
Bp,exc Excess air ratio for the secondary air kg/kg secondary air
BPLSR Regression coefficients in PLSR model -
C Concentration mol/m3
C125 Content of calcite particles larger than 125 -
µm
CD Drag coefficient -
CmRR Cumulative mass in Rosin Rammler wt%
distribution
Cp Heat capacity J/(kg*K)
Ct Craya-Curtet parameter -
D Diffusion coefficient m2/s
d Diameter m
de Equivalent diameter m
dRR Characteristic dimeter in Rosin Rammler m
distribution
E Energy J
Ea Activation energy in Arrhenius expression J/mol
F Force N
Fr Rotational Froude Number -
Gr Grashof Number -
Gopt Optimum burner momentum N
GX Axial thrust N
GY Flux of angular momentum N·m
g Gravitational acceleration m/s2
H Kiln bed height m
H Enthalpy J/kg or J/mol
Hcooler Clinker cooler heat transfer coefficient W/(m K)
ΔH Heat of reaction J/kg
h Heat transfer coefficient J/(s*m2*K)
I Burner momentum N
I’ Burner momentum normalized by burner N/MW
power
K Parameter/constant in equation -
k Reaction rate constant Unit dependent on
order of reaction
L Length m
LC Chord of the sector covered by bed m
l Length or thickness m
lf,drop Landing spot of fuel in the kiln m
lf,drop2 Spread on fuel landing spot m
Nomenclature XXV

Roman Explanation Unit

lm Total path length for radiation m
M Molar mass kg/mol
MSO3 Sulfur Modulus -
m Mass kg
ṁ Mass flow rate kg/s
mRR Characteristic mass in Rosin Rammler kg
distribution
N Number -
n Kiln Rotational speed 1/s
ṅ Mole flow rate mol/s
nRR Spread parameter in the Rosin Rammler -
distribution
P Power W
Pr Prandtl number Pr
p Pressure Pa
Q Heat flow J/s
Q44 Content of quarts particles larger than 44 -
µm
q Heat flux J/(m2 s)
R Reaction rate on mass basis kg/(m3 s)
R’ Reaction rate on molar basis mol/(m3 s)
Re Reynolds number -
Reω Rotational Reynolds number -
Rgas Ideal gas constant J/(mol K)
r Radius m
S Swirl number -
Sc Schmidt number -
Sh Sherwood number -
T Temperature K
Tq Torque of kiln rotation N·m
t Time s
tp Particle thickness m
u (Axial) Velocity m/s
ud Dynamic mean velocity m/s
uk Kinematic mean velocity m/s
V Volume m3
Vdiff Diffusion volume cm3/mol
v Velocity m/s
w Tangential Velocity m/s
X Conversion -
XPLSR Matrix of predictor variables in PLSR model -
x Axial distance m
Y Mass fraction -
YPLSR Matrix of response variables in PLSR model -
̂
YPLSR Predicted values of YPLSR -
y Molar fraction -
XXVI Nomenclature

Greek Explanation Unit

α Absorption of radiation -
γ Ratio of specific heats -
Δ Change -
δij Kronecker delta -
ε Emissivity of radiation -
ε/k Turbulent dissipation rate 1/s
η Fraction of solid fill in kiln -
θ Angle subtended by solids at kiln centre Radians
κ Absorption coefficient in WSGG model 1/m or bar/m
λ Thermal conductivity J/(s*m*K)
μ Dynamic viscosity kg/(s*m)
ν Kinematic viscosity m2/s
ξ Bed angle of repose Radians
ρ Density kg/m3
σ Stefan Boltzmann constant J/(s*m2*K4)
σf,drop Standard deviation on fuel landing spot m
τ Transmissivity for radiation -
ϕ Stoichiometry for reaction (mol based) mol O2/mol fuel
ϕm Stoichiometry for reaction (mass based) kg O2/kg fuel
φ Sphericity -
ψ Kiln slope Radians
Ω View factor for radiation -
ω Rotational speed Radians /s
Nomenclature XXVII

Subscript Explanation
0 Initial
amb Ambient
app Approximate
b Bed
bu Bulk
char Char
char,gsc Char gasification
char,ox Char oxidation
coat Coating
cond Conduction
conv Convection
devol Devolatilization
dry Drying
edr Eddy Dissipation Rate
ent Entrainment
ex Exit
exc Excess
ext External
f Fuel
film Film layer
fl Flame
g Gas
ht Heat Transfer
i Index
j Index
k Kiln
max Maximum
n Nozzle
o Outer
p Particle
prim Primary air
RR Rosin Rammler
r Reaction
rad Radiation
req Requirement
res Residence
s Surface
sec Secondary air
sh Shell
vol Volatiles
w Wall
x Axial direction
y Vertical direction
1 INTRODUCTION

Concrete is the most utilized manmade material, with an annual production

exceeding 1 m3 per person worldwide. The success of concrete stems from its
relatively low cost, low energy consumption, and the fact that it can be made from
materials most abundant in the crust of the Earth [1]. Without reinforced concrete
it would not have been possible to build modern landmarks such as the Burj Khalifa
(Dubai) [2] or the Three Gorges Dam (China) [3]. Concrete consists of sand and
gravel as aggregate particles that are held together by a binder, the cement. In some
cases, steel can be added as reinforcement to further strengthen the concrete. 2000
years ago, the Romans used volcanic ash as the binder to build structures like the
Pantheon, which still stands today [4,5]. Today, the main type of cement is Portland
cement, which mainly consists of di- and tricalcium silicates. When the Portland
cement is mixed with water it hydrates and hardens to bind together the stone and
gravel in the concrete [6].

The extensive use of Portland cement clinker is, however, not without issues. The
global cement production in 2017 amounted to 4.1 billion tons [7] and the cement
industry is responsible for around 5 % of the worlds CO2 emissions [8].
Furthermore, the manufacture of cement is highly energy intensive with the global
average thermal energy usage being around 4.2 GJ/ton clinker, while the most
efficient plants use 2.9 GJ/ton clinker [9]. This results in high costs related to fuels,
which conventionally makes up for 30-40 % of the cement production cost [10].
2 1 Introduction

Thus, there is a clear cost driver for the cement manufacturers to utilize cheap
alternative fuels. However, alternative fuels also need to be readily available for the
cement manufacturer and the legislation needs to be supportive. This is e.g. possible
in the European Union where the landfilling of unprocessed waste has been banned,
and a number of treatment plants upgrade the waste so it can be used as an
alternative fuel in the cement industry [11]. Thus, the use of alternative fuels in the
cement industry can also serve as a waste management option to reduce landfill and
conserve fossil fuels. The increased use of alternative fuels and the reduction of
fossil fuels may also contribute to reducing the CO2 emissions from the industry [12].

1.1 Scope of Thesis

The objective of the work carried out in this PhD project is to gain increased
systematic and scientific knowledge on the combustion of alternative fuels in the
cement rotary kiln. With this knowledge, it will be possible to propose guidelines for
the increased use of alternative fuels in the cement industry. The primary focus of
this project was on the following two areas:

Large Scale Measurements

In connection with the development and commissioning of the FLSmidth Jetflex

burner, several measurement campaigns were carried out at three different
industrial scale cement plants.

The main aim of the measurements has been to get a better understanding of how
the cement kiln flame is influenced by co-firing with alternative fuels, and how the
co-fired flame can be optimized to give the best cement clinker quality.

Cement Kiln Modeling

In order to better understand the effect that alternative fuels have on the cement
kiln process, it is desired to model the impact. In this way, the effect of the properties
of alternative fuels, such as heating value, moisture content, and particle size, can be
evaluated directly, through a simplified kiln and combustion model.

1.2 Outline of Thesis

The thesis is divided into three parts consisting of a literature survey (chapters 2-
4), an experimental section (chapters 5-7), and a modeling section (chapters 8-9).
1.3 Publications 3

Chapter 2 will give a short introduction to the cement making process and an in-
depth description of kiln burners. In chapter 3, a description of fuels used in the
cement kiln is given, with a special focus on some of the challenges in utilizing
alternative fuels. Typical measurements obtained at a cement plant to gauge the
state of the process is covered in chapter 4.

The experimental section is covered in chapters 5-7. Chapter 5 and 6 covers the
industrial tests, which investigates the effect of co-firing on the kiln flame using a
specially developed kiln camera. Chapter 7 deals with a detailed characterization of
SRF to obtain input parameters to combustion modeling.

Chapters 8 and 9 covers the model work. Chapter 8 begins with a review of the
models for the rotary kiln found in the literature. This is followed by a review of the
different mechanisms, e.g. heat transfer and combustion, occurring in the kiln. The
knowledge is used to formulate mathematical models valid for the kiln. Chapter 9
contains model validation and results of the modeling work.

The final conclusions of the project are found in chapter 10.

1.3 Publications
During this PhD project the following journal article has been published:

M. N. Pedersen, M. Nielsen, S. Clausen, P. A. Jensen, L. S. Jensen, and K. Dam-Johansen,

“Imaging of Flames in Cement Kilns To Study the Influence of Different Fuel Types,”
Energy & Fuels, vol. 31, no. 10, pp. 11424–11438, 2017.

Significant contributions (second author) have been made to the following

conference article:

D. J. Isaac, M. N. Pedersen, D. Grévain, L. S. Jensen, and M. Nielsen, “Numerical

Evaluation of Co-Firing Solid Recovered Fuel With Petroleum Coke in a Cement
Rotary Kiln - A Case Example,” in Proceedings of The 12th International Conference
on Computational Fluid Dynamics in the Oil & Gas, Metallurgical and Process
Industries, 2017, pp. 613–620.

In addition, the following manuscript has been submitted to the journal Energy and
Fuels (second author):
4 1 Introduction

M. Nakhaei, M.N. Pedersen, H. Wu, D. Grevain, L.S. Jensen, P. Glarborg, P.A. Jensen, K.
Dam-Johansen, “Aerodynamic and Physical Characterization of Refuse Derived
Fuel,” Energy and Fuels (Manuscript).
2 CEMENT PRODUCTION AND
CHEMISTRY

The following chapter aims to give an overview of the basic principles of cement
manufacturing. The primary focus will be on understanding the processes in the
cement kiln and the impact of the kiln burner.

2.1 Production of Portland Cement

Ordinary Portland Cement (Type 1 Portland Cement in the USA) is the most
common type of cement manufactured today and will be the focus of this chapter.
Portland Cement can be used directly as is, or mixed with different additives, e.g.
slag or fly ash, to form composite cements with various properties [11]. Other types
of cement include rapid-hardening cement, manufactured with an increased alite
content or finer grinding, sulfate resistant cement with lower aluminate content,
and white cement, which have a reduced content of ferrite and is produced under
reducing conditions [13].

The main raw materials used for cement production are naturally occurring
limestone, shale, and sand, which are sources of CaCO3, SiO2, Fe2O3, and Al2O3.The
materials are normally mined in a quarry, which should ideally be placed close to
the cement factory, to reduce transportation costs. Raw materials that are relatively
dry are usually preferred since removing excess water requires large amounts of
energy [4]. Figure 2-1 provides an overview of the different processes in the cement
plant after the raw materials have been mined and crushed.
6 2 Cement Production and Chemistry

Figure 2-1: Process flow sheet of a cement plant with preheater tower consisting of
cyclones and calciner [14].

The quarried raw materials are crushed to reach a size of 20-80 mm after which they
are stored in a blending bed, which serves as storage and preliminary
homogenization [15]. The raw materials are mixed and grinded in a mill to obtain a
homogenous mixture with a particle size approximately between 5 and 125 µm. The
small particle size is required to obtain proper reaction rates of the particles in the
kiln. It is generally recommended that calcite particles are grinded finer than 125
µm and quartz is grinded finer than 45 µm [16]. Hot flue gasses from the preheater
are typically used to dry the raw materials in the mill [6]. The milled and mixed raw
materials is called raw meal. If further homogenization of the raw meal is required,
it can be stockpiled in silos designed for blending [15]. It is important to mix the raw
materials in the right proportions in order to get the proper composition required
for the cement manufacture. Furthermore, a homogenous raw meal will yield the
best and most stable product quality. In addition, a stable kiln feed reduces kiln
instabilities and thus production costs [17].
2.2 Chemistry of Portland Cement 7

After milling, the cement raw meal is led to the preheater tower, which consists of a
series of cyclones. Modern cement plants usually have five or six cyclones depending
on the drying requirements of the raw meal [18]. Older types of cement plants have
fewer cyclone stages or none at all. In the preheater, the raw meal is heated by the
hot gasses from the calciner and cement kiln and obtains temperatures up to 850 °C,
while the gas is cooled from 1000 to 300 °C [4].

As the raw meal reaches temperatures around 700 °C the calcium carbonate begins
to decompose forming calcium oxide [15]. The reaction is highly endothermic and
consumes around 60 % of the thermal energy required in the cement process [4].
The energy for calcination is provided by combustion of fuels in the calciner, where
the raw meal reaches temperatures between 800 and 900 °C and up to 95 % of the
calcium carbonate is converted [18]. Around 50 % of the CO2 emissions from cement
manufacture are due to the decomposition of calcium carbonate [8].

From the calciner the material is admitted to the cement kiln where it is heated
further and undergoes the reactions that gives the cement its characteristic
attributes. The reactions will be discussed in detail in chapter 2.3. In the kiln the
materials partly melt and form nodules known as clinker [6].

After the reactions in the kiln, the clinker is quickly cooled by ambient air in the
clinker cooler. The preheated air from the cooler reaches temperatures of 1000 °C
and is used as secondary air for the combustion process in the kiln [15]. Excess air
from the cooler is used as combustion air in the calciner, so-called tertiary air. The
cooled clinker is transported to storage until it is needed. The clinker typically has a
size of 3-25 mm [14] and needs to be grinded to smaller and more uniform particle
size in order to increase the reaction rate of the cement [4]. During the milling
different additives such as gypsum, coal fly ash, or sand can be added to the cement
in order to control the setting time of the cement [4]. The additives might replace a
substantial amount of cement clinker in the finished cement and can reduce the
energy requirement and CO2 emissions of the manufacturing process [19].

2.2 Chemistry of Portland Cement

According to the European standard EN 197-1 [20], the requirements for Portland
cement clinker are that calcium silicates constitute 2/3 of the clinker mass, the ratio
of CaO to SiO2 shall not be less than 2, and the content of MgO shall not exceed 5
8 2 Cement Production and Chemistry

wt%. The typical composition of Portland cement clinker can be seen in Table 2-1.
Various additives and fillers can be added to substitute part of the clinker in
Portland composite cements [20].

Table 2-1: Typical composition of Portland cement clinker [11].

The main components in the cement are alite, belite, aluminate, and ferrite. Alite is
the most abundant mineral and is responsible for most of the strength development
of the concrete [15]. The CaO in the cement clinker that has not combined with other
components, is called free lime. Furthermore, the clinker can contain minor species
such as MgO, K2SO4, and Na2SO4. In Table 2-1 the clinker minerals are listed in their
pure forms, but in real clinker, smaller amounts of other oxides may be incorporated
into the crystal structure as impurities [13].

In cement chemistry, a shortened notation is used for oxides, such as C for CaO and
S for SiO2 [13]. This is convenient to shorten the chemical formulas seen in Table
2-1. It is also common practice to report the composition of the clinker or raw meal
on an oxide basis, e.g. K2O or SO3 [21], even though these components are not
necessarily present in the kiln.
2.3 Clinker Reactions 9

2.3 Clinker Reactions

Figure 2-2 is a simplified phase diagram showing how the different clinker phases
are formed as the solids pass through the kiln system. The first steps of the cement
clinker formation are taken in the preheater and calciner where the raw material is
heated to temperatures of around 900 °C. This facilitates the decomposition of the
calcium carbonate into calcium oxide:

𝑪𝒂𝑪𝑶𝟑 → 𝑪𝒂𝑶 + 𝑪𝑶𝟐 R 2.1

The decarbonation reaction is an equilibrium reaction and the extent of the reaction
depends both on the temperature and the partial pressure of CO2 in the gas phase
[6]. The reaction is also strongly endothermic, requiring 1800 kJ/kg CaCO3 (at
standard conditions) [14].

Figure 2-2: Simplified phase diagram showing the cement clinker reactions and
corresponding temperatures as solids pass through the kiln system from left to right
[22].

When the cement raw material enters the rotary kiln at temperatures around 900
°C, the remaining CaCO3 is decomposed and the formation of the clinker phases
begin [2]. Belite (C2S) begins to form at temperatures above 600 °C [15], but is
mainly formed in the kiln between 900 and 1250 °C as a solid-solid reaction between
CaO and SiO2 [2]. The reaction can in a simplified manner be represented as:
10 2 Cement Production and Chemistry

𝟐𝑪𝒂𝑶 + 𝑺𝒊𝑶𝟐 → (𝑪𝒂𝑶)𝟐 ∙ 𝑺𝒊𝑶𝟐 R 2.2

This reaction equation shows the pure form of belite. Under industrial conditions
several impurities are normally present which are incorporated in the various
clinker phases [2]. The formation of belite reaches its maximum at around 1250 °C,
after which it is consumed in the formation of alite (C3S) [2]. Between 900 and 1250
°C the crystalline phases of aluminate (C3A) and ferrite (C4AF) are formed. The
aluminate is formed as a reaction between CaO and Al2O3:

𝟑𝑪𝒂𝑶 + 𝑨𝒍𝟐 𝑶𝟑 → (𝑪𝒂𝑶)𝟑 ∙ 𝑨𝒍𝟐 𝑶𝟑 R 2.3

The ferrite is formed in a reaction between CaO, Al2O3, and Fe2O3:

𝟑𝑪𝒂𝑶 ∙ 𝑨𝒍𝟐 𝑶𝟑 + 𝑪𝒂𝑶 + 𝑭𝒆𝟐 𝑶𝟑 → (𝑪𝒂𝑶)𝟒 ∙ 𝑨𝒍𝟐 𝑶𝟑 ∙ 𝑭𝒆𝟐 𝑶𝟑 R 2.4

Alumina and iron oxide are not essential constituents of the Portland cement clinker,
but they serve as fluxing agents and lower the energy requirement of the process,
making it more economical [4]. As the temperature increases above 1250 °C, the
aluminate and ferrite phases begin to melt [2], and at 1450 °C as much as 20-30 %
of the cement mix may have melted [13]. The molten phase serves the purpose of
accelerating the clinker phase reactions by facilitating diffusion in the liquid phase,
which is significantly faster than in the solid phase. This allows additional Ca to
diffuse to and react with the belite to form alite:

𝟐𝑪𝒂𝑶 ∙ 𝑺𝒊𝑶𝟐 + 𝑪𝒂𝑶 → (𝑪𝒂𝑶)𝟑 ∙ 𝑺𝒊𝑶𝟐 R 2.5

In addition, the melt agglomerates the particles and binds them together forming
clinker nodules while also reduces the amount of dust in the kiln [2]. The hot clinker
should be swiftly cooled in the clinker cooler to prevent R 2.5 from occurring in
reverse, which would lower the amount of alite.

2.4 Clinker Compositional Parameters

The clinker or raw meal composition is often described using a few key parameters
related to the content of the different oxides.

The Lime Saturation Factor (LSF) is calculated based on the mass fraction of the 4
main clinker oxides [13]:

𝒀𝑪𝒂𝑶
𝑳𝑺𝑭 = E 2.1
𝟐. 𝟖 ∗ 𝒀𝑺𝒊𝑶𝟐 + 𝟏. 𝟐 ∗ 𝒀𝑨𝒍𝟐 𝑶𝟑 + 𝟎. 𝟔𝟓 ∗ 𝒀 𝑭𝒆𝟐 𝑶𝟑
2.4 Clinker Compositional Parameters 11

If the LSF is unity, the oxides are present in proportions where they, in theory,
combine fully yielding a clinker with zero free lime. If the LSF is above unity, then
the amount of calcium is too high to be fully incorporated in the clinker minerals,
and free lime will be present in the clinker. Typical values of LSF are in the range
0.92-0.98. The higher the LSF the more alite, C3S, can be formed in the clinker [13].

Two other important metrics are the silica ratio (SR) and the alumina ratio (AR),
also sometimes called alumina and silica modulus. They are defined as:

𝒀𝑺𝒊𝑶𝟐
𝑺𝑹 = E 2.2
𝒀𝑨𝒍𝟐 𝑶𝟑 + 𝒀𝑭𝒆𝟐 𝑶𝟑

𝒀𝑨𝒍𝟐 𝑶𝟑
𝑨𝑹 = E 2.3
𝒀𝑭𝒆𝟐 𝑶𝟑

The SR is normally in the range 2.0-3.0 and the AR 1.0-4.0. The SR governs the
proportion of silicate phases to iron and alumina in the clinker. Increased values
indicate that less melt, which is mainly caused by iron oxides and alumina, can be
formed in the clinker. This causes the clinker reactions to occur slower, making the
clinker harder to burn.

The AR describes the ratio between alumina and ferrite phases in the clinker. It
governs the temperature at which the melt phase forms, which is lowered by
increased iron content [23].

Another important equation is the Bogue formula [13,24], which may be used to
estimate the composition of the four main clinker phases.

𝒀𝑪𝟑 𝑺 = 𝟒. 𝟎𝟕𝟏𝟎 𝒀𝑪𝒂𝑶 − 𝟕. 𝟔𝟎𝟐𝟒 𝒀𝑺𝒊𝑶𝟐 − 𝟔. 𝟕𝟏𝟖𝟕 𝒀𝑨𝒍𝟐 𝑶𝟑

E 2.4
− 𝟏. 𝟒𝟐𝟗𝟕 𝒀 𝑭𝒆𝟐 𝑶𝟑

𝒀𝑪𝟐 𝑺 = −𝟑. 𝟎𝟕𝟏𝟎 𝒀𝑪𝒂𝑶 + 𝟖. 𝟔𝟎𝟐𝟒 𝒀𝑺𝒊𝑶𝟐 + 𝟓. 𝟎𝟔𝟖𝟑 𝒀𝑨𝒍𝟐 𝑶𝟑

E 2.5
+ 𝟏. 𝟎𝟕𝟖𝟓 𝒀 𝑭𝒆𝟐 𝑶𝟑

𝒀𝑪𝟑 𝑨 = 𝟐. 𝟔𝟓𝟎𝟒 𝒀𝑨𝒍𝟐 𝑶𝟑 − 𝟏. 𝟔𝟗𝟐𝟎 𝒀 𝑭𝒆𝟐 𝑶𝟑 E 2.6

𝒀𝑪𝟒 𝑨𝑭 = 𝟑. 𝟎𝟒𝟑𝟐 𝒀 𝑭𝒆𝟐 𝑶𝟑 E 2.7

In the above equations the measured free lime content should be subtracted the lime
content before being applied. The Bogue formula may differ considerably from the
actual phase composition, since pure clinker phases are assumed in the calculations,
while impurities may be incorporated into the clinker in an actual kiln.
12 2 Cement Production and Chemistry

2.5 Clinker Burnability

The ease at which the clinker oxides react in the cement clinker is termed
burnability. An easy burning clinker will form alite at relatively low temperatures,
while a hard burning clinker will require higher temperatures. Thus, burnability is
important for the energy consumption in the cement kiln. Historically, it has been
easier to measure the free lime rather than the alite content, thus free lime content
has been used to determine the completeness of the clinker reactions.

The burnability can be tested by burning raw meal in a laboratory oven. A procedure
is to form small clinker nodules which are burned for 30 minutes at e.g. 1400, 1450,
and 1500 °C [23]. The lower the free lime content after burning, the easier the
clinker is to burn.

It is also possible to estimate the burnability theoretically using equations available

in the literature. These equations attempt to estimate the free lime content, based
on the raw meal composition and particle size. One example is given in E 2.8. The
equation estimates the free lime in laboratory clinker burned for 30 minutes at 1400
°C [25].

𝒀𝑪𝒂𝑶𝒇 = 𝟎. 𝟑𝟑 ∗ 𝑳𝑺𝑭 + 𝟎. 𝟎𝟏𝟔𝟖 ∗ 𝑺𝑹 − 𝟎. 𝟑𝟓𝟑 + 𝟎. 𝟗𝟑 ∗ 𝑸𝟒𝟒 + 𝟎. 𝟓𝟔

E 2.8
∗ 𝑪𝟏𝟐𝟓

The clinker thus becomes harder to burn by increased LSF and SR, which are related
to the chemical composition. In addition, the content (mass fraction) of coarse
quartz, Q44, (SiO2) and calcite, C125, (CaCO3) with particle sizes above 44 and 125 µm,
also affects the burnability. These large particles are unable to react in the kiln due
to diffusion limitations.

If the clinker burnability in a kiln changes due to changes in the raw meal kiln feed,
it can be expected that the specific heat consumption of the kiln should be adjusted,
to maintain a constant free lime content [26].

2.6 The Cement Rotary Kiln

The formation of cement clinker occurs in the rotary kiln. Figure 2-3 shows a rotary
kiln and a grate cooler indicating the main temperature zones in the kiln. The cement
rotary kiln is essentially a long cylindrical tube that acts as a large counter-current
heat exchanger. At one end the raw materials, preheated and calcined in the cyclone
2.7 Combustion Air in the Cement Kiln 13

tower, are admitted at around 900 °C. At the other end of the kiln, fuel is being fired
through the burner, to heat the material in the kiln to a maximum temperature of
around 1450 °C. The kiln is composed of an outer steel shell, which is lined with
refractory material that acts as thermal insulator and protects the outer shell [27].
The type of refractory bricks vary along the length of the kiln depending on the
temperature of the material and gas [13]. In the hot parts of the kilns, where the
clinker melts, a coating is formed on top of the refractory, which further helps
insulate and protect the kiln walls [13].

Figure 2-3: A burner in a grate cooler cement kiln indicating temperature zones in the
kiln and different air flows.

The kiln is typically 50-100 meters long with diameters between 3 and 7 meters. Old
wet kilns are long with L/d around 30, while modern pre-calciner kilns are shorter
with L/d around 10-15 [6]. The kiln is tilted at 1-3° and rotates at 1-4 rpm to
facilitate the movement of the raw materials from one end to the other. The
residence time of the raw materials is typically 20-40 minutes with 10-15 minutes
in the burning zone, the hottest part of the kiln [2,13]. The residence time of the
gasses is around 5-10 seconds [28]. Typical production capacities are around 3,000
ton/day [11], but the largest kiln in the world has a capacity of 13,000 ton/day [29].

2.7 Combustion Air in the Cement Kiln

As indicated in Figure 2-3 several different types of air are relevant for the
combustion process in the cement kiln. Primary air is injected through the burner
and can be divided into transport air, axial air, and swirl air (also called radial or
tangential air). The transport air is used to pneumatically inject the fuel into the kiln
14 2 Cement Production and Chemistry

at velocities around 30 m/s [27]. The axial and swirl air are used to control the flame
shape [30]. The primary air constitutes around 5-15 % [31] of the air required to
combust the fuel. Older generations of burners may have used up to around 30 %
primary air [32].

The remainder of the combustion air is called secondary air. This is the air that is
used to cool the hot clinker. Heat recuperation in the cooler preheats the secondary
air to temperatures around 1000 °C before it enters the kiln [31]. The gasses are
pulled through the kiln by an induced draft (ID) fan, which creates an underpressure
in the kiln. The low pressure may cause some air, 2-10 %, to leak into the kiln
between the rotating kiln and stationary cooler [27,32]. The air required to cool the
clinker exceeds the amount of air necessary for the combustion in the kiln. Some of
the excess air is used for the combustion in the calciner, so-called tertiary air. The
remaining excess air leaves the clinker cooler and may be used elsewhere, e.g. for
drying of raw materials [33].

A correct amount of combustion air is imperative for a high thermal efficiency in the
kiln. If the amount of combustion air is too low, it results in unburned fuel and CO
emissions from the kiln. On the other hand, excessive energy is required to heat up
excess combustion air, which will cool the flame. An increase in the oxygen content
at the kiln raw material inlet from 1 to 5 vol% will increase the specific heat
consumption by approximately 10 % [32]. It is normally recommended that the
oxygen content in the kiln exhaust gasses is in the range 0.7-3.5 vol% [34].

2.8 The Cement Kiln Burner

The kiln burner provides the thermal energy that is required to heat the raw meal
from approximately 900 to 1450 °C, which facilitates the formation of a liquid phase
and the initiation of the chemical reactions that form the cement clinker. Often the
burner is referred to as the main burner, although in modern preheater and calciner
plants, the larger share of fuel is fired at the calciner.

Most modern cement kiln burners are so called multichannel burners, which contain
several channels and nozzles for injection of different kinds of fuel and air. An
example of a modern multichannel burner is the FLSmidth Jetflex burner shown in
Figure 2-4. The burner is designed with a large central channel for alternative fuels
firing. The center is surrounded by a channel for swirl air, which is injected through
2.8 The Cement Kiln Burner 15

the slanted vanes giving a tangential motion of the air. Next follows two annular
channels for pulverized fuel and gas, respectively. The burner has 20 axial air
nozzles, which can be turned individually to shape the flame. The outside of the
burner is lined with refractory to protect the steel from high temperatures. The size
of the burner is dependent on its capacity, but it will typically be around 1 m in
diameter, including 0.2 m of refractory. The length of the burner is typically 10-20
m, which is required since the burner must be long enough to reach through the kiln
hood as shown in Figure 2-3.

Figure 2-4: The Jetflex burner, a modern multichannel cement kiln burner from
FLSmidth A/S. The additional channel can e.g. be used for oxygen enrichment, liquid
fuel firing, or a flame detector.

To minimize the energy consumption in the rotary kiln it is important to achieve an

efficient combustion process. Consequently, the ideal kiln burner should be able to
[30,31,35,36]:

• Provide a high temperature flame maximizing heat transfer to the raw

material

• Control the temperature and shape of the flame

• Obtain high product quality and production capacity

• Give stable process conditions

• Minimize the specific energy requirement

16 2 Cement Production and Chemistry

• Provide adequate and efficient mixing of the hot secondary air

• Obtain quick ignition and full conversion of fuel in suspension with a

minimum of fuel dropout

• Produce a minimum of harmful emissions, e.g. NOX.

• Use a minimum of excess air and primary air

• Be fuel flexible

Some of these points might be conflicting and the focus of the kiln burner may
change between different cement plants or geographical regions, depending on e.g.
legislation. For instance, there is a high focus on utilizing alternative fuels in the
European cement sector, which may tend to increase process instability, which will
be described in chapter 3.5.

2.9 History of Kiln Burner Design

Since the early 1900’s the kiln burner has developed from being mainly coal fired to
being mainly oil fired. During the oil crisis of the 1970’s, the fuel was switched back
to coal or petcoke. Recently, since the 1990’s a demand for an increased amount of
alternative fuels firing has arisen. In the meantime, environmental legislation has
also become stricter, setting limits on emission of pollutants such as NOX [30]. To
adapt to the varying fuels and process requirements, the kiln burner has also
developed, as shown in Figure 2-5.

Figure 2-5: Development of the cement burner technology since 1960 and the main
fuels used [30,37].
2.9 History of Kiln Burner Design 17

While the burners themselves have changed there has also been a change from
direct firing to indirect firing since the 1970’s. The directly fired burner is connected
directly to the fuel mill. While this is a simple system it has the disadvantage that the
primary air is used to dry the fuel before it is admitted to the kiln. This will result in
larger amounts of primary air and unnecessary water in the kiln flue gas, which
results in lower heat efficiency. It is also more difficult to regulate the system, since
it is required to control the fuel feed to the mill and the mill speed simultaneously.
In addition, if the mill breaks down it will lead to a kiln stop. In indirect fired systems,
an intermediate storage for the pulverized fuel is installed after the mill. This
reduces the disadvantages of the direct fired system, but also requires additional
equipment. Another advantage of the indirect fired system is that one mill can feed
both the kiln and the calciner [38,39].

2.9.1 Single Channel Burners

The early kiln burners were single channel burners typically fired with pulverized
fuel, i.e. coal. Fuel and primary air are injected through a single channel and outlet
velocities of the air is typically 40-60 m/s. They use a high amount of primary air
(15-45 %) to ensure a high burner momentum. The burner momentum, I, is
calculated by the axial velocity, uprim, and mass flow, 𝑚̇𝑝𝑟𝑖𝑚 , of the primary air
according to E 2.9. A higher momentum gives a more efficient mixing of air and fuel
in the kiln and thus promotes proper combustion of fuel. The momentum is often
normalized, I’, by the burner thermal capacity and is normally in the range 4-12
N/MW [31].

𝑰 𝒖𝒑𝒓𝒊𝒎 ∗ 𝒎̇𝒑𝒓𝒊𝒎
𝑰′ = = E 2.9
𝑷𝒃𝒖𝒓𝒏𝒆𝒓 𝑷𝒃𝒖𝒓𝒏𝒆𝒓

In addition to the high amount of primary air, the disadvantages of single channel
burners are that it is difficult to change the load (limited turndown ratio),
consequently, they are difficult to manage during startup periods. Moreover, there
is a very limited possibility to change the flame pattern during operation, since the
amount of primary air and coal throughput should be proportionate. It is possible to
gain some control by adjusting the primary air velocity and amount depending on
the fuel load [35].
18 2 Cement Production and Chemistry

The flame can be sketched as in seen in Figure 2-6. The flame typically spreads with
an angle around 10° and the mixing will typically be complete at around 3 kiln
diameters downstream of the burner [40]. In the plug flow zone, after the mixing
point, oxygen will mainly be supplied through diffusion [41]. Thus, the fuel burnout
and CO reduction should preferably be completed ½ a kiln dimeter before the macro
mixing is complete. Otherwise, the mixing intensity and combustion rate is lowered
and unburnt fuel may cause reducing conditions in the clinker bed [40,41].

Figure 2-6: A typical axial flame from a single channel kiln burner [40]. ERZ = External
Recirculation Zone

2.9.2 Multichannel Burners

The multichannel burner (three channel burner) was developed in the late 1970’s.
Separate channels were made for the coal and primary air, which allowed for a
higher primary air velocity and introduction of swirl air. This created a better mixing
of the fuel and hot secondary air than what was possible for the single channel
burner and a better control of the flame [35]. The swirl air expands the flow of
pulverized fuel and stabilizes the flame by creating an internal recirculation zone
[42] as shown in Figure 2-7. The effect of swirl will be discussed in more detail in
chapter 8.6. The high degree of flame shaping is beneficial to adjust a burner to the
requirements of each single kiln system. It helps in kiln optimization, securing a
stable coating formation, kiln shell temperature, and clinker quality [35]. Another
advantage of the multichannel burner was that the high primary air velocities in the
range 70-130 m/s made it possible to keep a high momentum by reducing the
amount of primary air to 10-15 % of the combustion air, which gives a better heat
economy [35]. The higher mixing that can be obtained with a multichannel burner
will create a shorter, but higher temperature flame, which ensures better heat
transfer to the clinker bed.
2.9 History of Kiln Burner Design 19

A sketch of the multichannel burner flame is shown in Figure 2-7. Compared to the
single channel burner sketched in Figure 2-6, an internal recirculation zone is
formed due to swirl flow. In addition, the mixing is completed closer to the burner.

Figure 2-7: Multichannel burner coal flame [40]. ERZ = External Recirculation Zone,
IRZ = internal Recirculation Zone.

The construction of the burner is normally a center channel for oil or gas surrounded
by a jacket for swirl air, which gives the flame a divergent and rotating motion. The
next outer channel is the coal channel, which is enclosed in an annulus or by jets for
axial primary air. The coal is injected at around 20 m/s. The low coal velocities
decrease wear and allows for better flame shaping. The amount of coal conveying
air should be around 2-3 % of the combustion air (minimum 0.2 m3/kg coal), in
order to ensure pulse free flow [35].

2.9.3 Low Primary Air Burners

Low primary air burners, also often called low NOX burners, were developed in the
1980’s driven by stricter environmental legislation. NOX formation increases with
increasing temperature and oxygen availability. Thus the principles for lowering the
NOX are flame stabilization and recirculation of combusted gasses, concentrating the
fuel along the center axis to prevent the fuel from leaving the flame, reduced primary
air amount to 8-10 % (compared to 10-15 % in the multichannel burner), and more
even heat distribution in the flame to avoid high peak temperature regions [42].

NOX is mainly formed if the fuel is allowed to burn in oxygen rich conditions. Thus,
low NOX burners use a low amount of primary air, typically less than 10 %. This
decreases the availability of oxygen in the hot part of the flames and lower the
production of NOX [30]. The lower primary air amount can be achieved by a redesign
20 2 Cement Production and Chemistry

of nozzles that allow for higher primary air velocities, thus keeping burner
momentum at the same level [42].

Staged combustion, i.e. recirculation of combustion gasses to the burner, can also
serve to lower NOX emissions [35] and the concept is shown in Figure 2-8. The
recirculated hot gasses will quickly heat up and ignite the fuels and they contain a
low amount of oxygen, which limits the NOX formation [40].

Figure 2-8: Flow pattern in a low NOX burner [43]. Note that the swirl channel is
located inside the coal channel.

The ignition point is also quite important for the NOX formation [40,42]. Ignition
close to the burner tip is preferred as it tends to lower the NOX emissions [42]. If the
ignition occurs far downstream in the kiln, more oxygen can be entrained into the
flame leading to high NOX levels [40,42]. The mono-channel burners create very high
levels of NOX since they have a late ignition. For multichannel burners, the effect of
swirl was two-sided. The swirl creates an internal recirculation zone for fuel staging,
while on the other hand an increased swirl also entrains extra air, which increases
the oxygen availability [42]. The location of the swirl air channel compared to the
fuel channel might also have an impact on the NOX. If the swirl is placed on the inside
of the coal channel, the air might throw the fuel into the hot secondary air, where it
will burn at high oxygen concentrations being more prone to form NOX [31,35].

The fuel type also has some influence on the NOX formation. Depending on the coal
characteristics, e.g. type, fineness, and volatile content, the ignition point might be
influenced [42]. A high volatile coal has an early ignition, which can reduce NO X
formation [40]. The coal particle size may also play a role and a larger particle size
2.9 History of Kiln Burner Design 21

will delay the ignition and thus increase the NOX [44]. On the other hand finer coal
particles tend to burn at higher peak temperatures increasing the NOX [40].

The low NOX burners are typically not suitable for a high utilization of AF. The low
amount of primary air and burner momentum is not sufficient to easily ignite the AF
particles and fuel dropout of unburnt particles is likely to occur [30].

2.9.4 Alternative Fuels Burners

The current burner development aims to increase the share of alternative fuels and
provide high fuel flexibility. The first trials to introduce alternative fuels into the
rotary kiln were made through a separate pipe on top of the existing burner.
However, this resulted in the fuel being carried over the top of the flame and
dropping to the clinker bed unconverted [45]. Modern rotary kiln burners consist of
multiple channels to increase the fuel flexibility, as shown in Figure 2-4. Different
tubes and annular channels are applied with the possibility of burning both
pulverized and lumpy solid fuels as well as liquids and gasses [30]. The channels for
alternative fuels are located in the center of the burner to keep the fuel in the center
of the flame as long as possible. Lumpy and fluffy solid fuels have a tendency to block
pipes, thus a large round tube is preferred for these types of fuel [45]. The center
fuel channel is typically surrounded by an annulus for pulverized fuels, i.e. coal or
petcoke firing, and the primary air channels.

The difference between burners from different manufactures is the positioning of

the primary air channels, which can be located inside or outside the annulus for
pulverized fuel [46]. Another distinction is whether the air is injected in an annulus
or through individual jets and how the primary air can be controlled to create axial
momentum and swirl [46]. Burner manufacturers seem to prefer the jet design for
axial primary air (see Figure 2-4). The merit of the jet design is an increased
entrainment of hot secondary air between the primary air jets, which will increase
the heating rate of the hard to ignite alternative fuels [46]. Nørskov [31] and
D’Hubert [47,48] provides a survey and comparison of the state of the art burners
currently available on the market.
22 2 Cement Production and Chemistry

2.10 Summary and Conclusions

Portland cement is the most commonly produced type of cement. It mainly consists
of the calcium silicates alite and belite as well as ferrite, aluminate, and other minor
components. The formation of cement clinker takes place inside a rotary kiln, where
the materials are heated to 1400-1500 °C, to facilitate the proper reactions between
the calcium and silica.

The high temperatures required in the kiln, makes the production of Portland
cement clinker highly energy intensive. The energy is supplied by burning fuels at
one end of the rotary kiln. In order to minimize the energy usage, the combustion
process should be as efficient as possible. For this purpose, the design of the kiln
burner has continuously evolved to allow combustion of different fuels, a higher
flexibility, and to comply with stricter environmental legislation.
3 FUELS IN THE CEMENT
INDUSTRY

The selection of fuel for the cement production is an important parameter for the
cement plant, especially since the fuel often makes up a significant cost of the plant
operation. Before a fuel is selected it is important to consider the following three
parameters [39]:

• Costs, e.g. fuel purchase, availability, fuel handling, maintenance.

• Product quality, e.g. impacted by unburnt particles in the clinker
• Environmental impact, e.g. CO2, CO, and NOX emissions,

An example of how these factors may change and force plants to use new fuels has
recently been seen in Egypt [49]. The Egyptian cement sector has traditionally relied
on natural gas and oil as their main fuels. However, in 2013 the Egyptian
government decided to promote the export of oil and gas, which reduced the
availability for domestic industries. Thus, many cement manufacturers decided to
start using coal or petcoke and are heavily investing in coal mills and other
necessary equipment and process changes.

This chapter will give a short overview of different fuels widely utilized in the
cement industry. They range from conventional fossil fuels such as coal, gas, and oil,
to various alternative fuels. Furthermore, the potential negative impacts that
alternative fuels can have on the kiln process, will be discussed.
24 3 Fuels in the Cement Industry

3.1 Conventional Fuels Used in the Cement Industry

3.1.1 Coal
Coal is the most widely used fuel in the global cement industry [39]. Many plants
switched from oil and gas firing to coal firing in the late 1970’s due to a steep rise in
the price of these fuels [15].

The rate of coal combustion is highly influenced by the particle size, the combustion
time being roughly proportional to the diameter squared [39]. The smaller the
particles, the faster they are heated and react, resulting in higher peak temperatures.
The volatile content in the coal impacts the ease of ignition, and thus heavily affects
the combustion in the kiln [40]. A rule of thumb for coal combustion says that the
residue on a 90 µm sieve should not exceed 50 % of the volatile matter level [40].
E.g. a 20 % volatile matter coal would require that a maximum of 10 % of the coal is
retained on a 90 µm sieve. A finer grinding than this may, all other things being
equal, result in high peak temperatures and excessive NOX formation [40]. In
conjunction with the milling, the coal is dried to a moisture content of 0.5 to 2.0 %.
A little residual moisture is preferable since the presence of OH radicals from the
water will help ignite the coal [39].

When using coal, or other solid fuels, in the cement kiln it is important to consider
that the ash will be incorporated into the cement clinker. This needs to be accounted
for by a proper adjustment of the raw meal composition prior to the burning process
[4]. It is also imperative that the ash particles are sufficiently small and injected far
inside the kiln in order for the particles to be incorporated into the clinker phases,
otherwise it could adversely affect the alite content in the clinker [4].

The types of coal fired in the cement kiln varies broadly from region to region
depending on the coal availability. There can be quite a large variation between
different coal types such as anthracite, bituminous coals, and lignite. Bituminous
coals are most widely utilized in cement plants followed by subbituminous coals and
lignite [39].

3.1.2 Petroleum Coke

Petroleum coke (petcoke) is a by-product from oil refining cracking processes. The
oil is subjected to extreme conditions driving off the volatiles and leaving a solid
3.1 Conventional Fuels Used in the Cement Industry 25

residue that is almost pure carbon [50]. As a by-product the price of petcoke is
typically favorable compared to coals, and it has become a very widely used fuel in
the cement industry [39].

A challenge compared to most coals is that petcoke is less reactive than coals.
Petcoke has a low content of volatiles, normally 5-15 % [51]. This makes petcoke
difficult to ignite compared to medium and high volatile coals, and once it is ignited,
it only burns slowly. In order to ignite the petcoke as quickly as possible and allow
full combustion of the particles, a fast mixing of secondary air and fuel is necessary
[50], which provides oxygen to the fuel surface. Thus, the burner design should
allow for higher momentum than during coal firing and single channel burners are
not well suited for petcoke firing [50]. A fine grinding of the petcoke can also counter
the low reactivity [51] as it increases the surface area of the fuel and thus contact
with oxygen.

Another challenge in petcoke utilization is the high sulfur content, normally in the
range 2-7 % [52]. The high input of sulfur may result in operational challenges in
the cement kiln such as build-ups and blockages [53,54]. This is discussed in further
detail in chapter 3.5.

3.1.3 Gas
Natural gas is the predominant type of gaseous fuel used in cement plants. It is
typically supplied to the cement plant in pipelines at high pressure. Before the gas
can be used it is necessary to reduce the pressure to 3-10 bar.

Since gas contains no ash components it is not necessary to adjust the cement raw
meal mixture to account for the ash added as with the solid fuels [4]. A challenge
with a gas flame is that it produces low amounts of soot and thus has a low radiance.
However, the high amount of dust present in the cement kiln may alleviate this issue
[4]. Gas is easy to ignite and is often used for heating up the kiln.

3.1.4 Oil
The oil used in cement kilns is normally heavy fuel oil (No. 6 fuel oil according to
ASTM classification [55]) . Oil has a high viscosity and should be heated to around
50 °C in order to be easily pumped. In order to properly atomize the oil in the burner
a further heating to around 120 °C is necessary [15]. A challenge with using oil is
26 3 Fuels in the Cement Industry

that it, like petcoke, may contain high amounts of sulfur. For heavy fuel oil the sulfur
content can be as high as 4 % [4].

The ease of oil ignition can be estimated by the Calculated Carbon Aromaticity Index
(CCAI) [39]:

𝑪𝑪𝑨𝑰 = 𝝆 − 𝟖𝟏 − 𝟏𝟒𝟏 ∗ 𝒍𝒐𝒈(𝒍𝒐𝒈(𝝂 + 𝟎. 𝟖𝟓)) E 3.1

Here the density, ρ [kg/m3], and kinematic viscosity, ν [mm2/s], are measured at 15
and 50 °C, respectively. Fuel oil with a CCAI between 800 and 870 is easy to ignite,
while if CCAI is above 870, ignition may be difficult [39].

3.2 Drivers for Alternative Fuels Usage

During the last 20 years, an increased interest in alternative fuels for the cement
industry has appeared. The primary drivers are cost, carbon dioxide emissions, and
waste management, as further discussed in the following chapters.

3.2.1 Cost of Fuel

Energy accounts for 30-40 % of the operating costs of cement manufacturing when
using fossil fuels [10], and is thus an important factor for any cement plant. The cost
of various fossil fuels is indicated in Table 3-1.

Table 3-1: Relative cost of various fossil fuels per energy unit [39].

Fuel Relative cost

Coal 100

Petcoke 30-50

Fuel oil 150-250

Natural gas 200-300

Table 3-1 shows that significant cost reductions are possible by choosing an
appropriate fuel. Switching from coal to petcoke may cut fuel costs by up to 2/3,
explaining why many cement plants have switched from coal to petcoke, even
though it is a more troublesome fuel. Fuel oil and natural gas are around twice as
expensive as coal, and relatively few plants use these fuels as their main fuel [11].
The solid fossil fuels are more cost effective, even though pretreatment in the form
3.2 Drivers for Alternative Fuels Usage 27

of milling and drying is necessary, which results in increased investments costs for
e.g. coal mills.

Waste derived alternative fuels may be significantly cheaper than the conventional
fuels presented in Table 3-1. This is the primary driver for the increased use of
alternative fuels in the cement industry. It has been estimated that a cement plant
with an annual production of 1 million tons of cement, could save up to 2.4 million
€ annually, by replacing 30 % of fossil fuels with no-cost alternative fuels [56]. In
some cases, the cement plant can be paid to accept alternative fuels [57].

3.2.2 Carbon Dioxide Emission Reductions

The cement industry is responsible for approximately 5 % of global anthropogenic
CO2 emissions [8]. Around 40 % of the emissions are due to fuel combustion [8].
Consequently, it is important that the cement industry contributes, if carbon
emissions are to be reduced [19]. Table 3-2 contains an overview of the net CO2
emission from various fuels often used in the cement industry, although significant
differences can occur depending on the sourcing of the fuel, especially for RDF.
Petcoke and coal are the most CO2 intensive fuels, while the other types of fuel emit
significantly less CO2. Especially, RDF, animal meal, and wood result in lower CO2
emissions since the fuels are wholly or partly biogenic and CO2-neutral [58]. Thus,
the use of alternative fuel could help in reducing the CO2 emissions from the cement
industry [56].

Table 3-2: Indicative net CO2 emissions from various fuels used in the cement industry
[59].

Net CO2 emission Net CO2 emission

Fuel Fuel
(g CO2/MJ) (g CO2/MJ)

Petcoke 101 Plastic 75

Coal 96 RDF 9

Natural gas 54 Animal meal 0

Used tires 85 Waste wood 0

Waste oil 74
28 3 Fuels in the Cement Industry

3.2.3 Waste Management

A third reason to use waste fuels in the cement industry is that it might help divert
waste away from landfills [56]. Combustion in cement kilns is a good option to
recover the energy in waste and conserve fossil resources [11]. The conditions in a
cement kiln are well suited for the thermal destruction of organic molecules. The
temperatures in the cement kiln exceed 1200 °C, the residence time is more than 2
seconds, and the oxygen concentration exceeds 3 %. This allows the cement kiln to
have very few emissions of organic toxins such as dioxins and furans [60]. Thus
alternative fuels can normally be combusted without causing excess gaseous
emissions [60]. In the cement kiln, fuel ashes are also incorporated into the clinker,
which eliminates the solid waste stream that is otherwise encountered during waste
incineration [61]. This may make it more beneficial to burn waste in cement plants
than in dedicated waste incinerators [61].

3.3 Use of Alternative Fuels

The use of alternative fuels in the cement industry for selected countries and regions
is shown in Figure 3-1. The data have been taken from the World Business Council
for Sustainable Development (WBCSD) Cement Sustainability Initiative, which
collects data in the “Getting the Numbers Right” database [62]. The database covers
numbers from more than 900 cement installations worldwide, which is around 20
% of the global production. The initiative best covers Europe where 95 % of cement
producers reports to the database. In other regions the coverage is less, which may
lead to somewhat skewed results [63].

Figure 3-1 shows that the use of alternative fuels is most advanced in the EU. Here
the use began to increase during the 1990’s with 2 % of the energy being derived
from alternative fuels. By 2000 this number had increased to 9 % and in 2013 to 37
%. Germany is one of the countries where the use of alternative fuels has progressed
the most and alternative fuels provide above 60 % of the energy.

North America is the region with the second highest use of alternative fuels, totaling
15 % in 2013, which is close to the world average. China and India have only very
small substitution rates. An improvement here could have a very large potential
since they are the largest cement producers in the world with 59 and 7 %,
respectively, of world production in 2017 [7]. Consequently, there is a large
3.4 Types of Alternative Fuels 29

potential for increasing the use of alternative fuels worldwide to a level near that in
Europe. However, it does require significant infrastructure and well managed waste
collection systems, which are not necessarily present outside of Europe and North
America.

Figure 3-1: Use of fossil and biomass based alternative fuels in selected countries and
regions [62]. Fossil alternative fuels include e.g. waste oil, tires, plastics, refuse
derived fuel; biomass include e.g. agricultural waste, sewage sludge, charcoal. CIS =
Commonwealth of Independent States.

In the European Union a framework for handling of waste and co-combustion of

waste has been set up through the “Waste Framework Directive” (2008/98/EC) and
the “Waste Incineration Directive” (2000/76/EC) [64,65]. The legislation defines
when a waste can be incinerated and sets rules to minimize the environmental
impact of waste incineration. Thus, a well-developed waste handling infrastructure
is present in Europe, and waste fuel is easily accessible for cement manufacturers.
In other parts of the world the use of waste products may be controversial since
cement kilns are not subject to stringent emission control [9,66].

3.4 Types of Alternative Fuels

Alternative fuels for the cement industry spans a wide variety of different fuels that
can be of either fossil or biogenic origin. Examples of commonly used fuels of fossil
origin are: waste oil, used tires, plastics, and solvents, while biomass fuels include:
agricultural waste, waste water sludge, rice husks, and charcoal [57,66].
30 3 Fuels in the Cement Industry

Furthermore, alternative fuels can be either gaseous, liquid, or solid, where solid
fuels are most widely utilized [28]. The fuels may be further subdivided into various
categories relating to e.g. size [10].

Figure 3-2: Distribution of alternative fuels in the German (a) and Austrian (b) cement
sectors in % of annual thermal use in 2015 [67,68].

Figure 3-2 shows the distribution of various kinds of alternative fuels in the German
and Austrian cement sectors. These countries utilize some of the highest amounts of
alternative fuels. It is seen that the largest fraction of fuel is plastics and commercial
waste. Waste tires also constitute a large fraction of the fuels used in these countries.
The use of biomass derived fuels: paper, sewage sludge, meat and bone meal, wood,
and agricultural residue contribute less than 10 % of the thermal share in these two
countries. An increased use of these fuels may be beneficial for the CO2 emissions in
the cement industry.

3.4.1 Solid Recovered Fuel

The largest fraction of alternative fuels in the German and Austrian cement
industries is plastics and commercial waste. In a broader sense these fuels are
typically called Refuse Derived Fuel, RDF, or Solid Recovered Fuel, SRF.

There appears to be no clear definition of RDF in the literature. Typically, RDF is

meant as the combustible fraction of municipal solid waste, but the definition can
also be broader covering a range of different commercial wastes such as: packaging,
tires, biomass, and waste oils [69]. Within the European union SRF is defined as
“fuels prepared from nonhazardous waste to be utilized for energy recovery in
waste incineration or co-incineration plants regulated under community
environmental legislation” [70]. The quality of SRF is assured through a number of
standards within the framework of the European Committee for Standardization
Technical Committee 343 (CEN/TC 343) [71].
3.4 Types of Alternative Fuels 31

SRF is produced from municipal solid waste, industrial or commercial waste. The
waste may be processed in mechanical or mechanical-biological treatment plants,
where the combustible fractions of the waste are separated out to be utilized for co-
combustion. The treatment differs depending on the source of the waste material
and the quality requirements for the processed SRF [72]. Normally the treatment
consists of various steps of shredding and sieving to reduce and homogenize the
particle size and remove fines. The fines typically consist of non-combustibles or
organic material with a high water content. Additional steps of the process are
magnetic separators to separate out iron and use of Near Infrared (NIR)
spectroscopy to separate out chlorine rich parts. Organic material can be removed
by biological composting, which also dries the material due to the heat released by
the biological decomposition [73]. The finished product has a more uniform quality,
higher heating value, and lower moisture content than the raw waste [72].

Figure 3-3: Composition of Solid Recovered Fuel based on visual sorting [74,75].

An example of the composition of different SRF’s based on the works of Sarc and
Lorber [75] and Krüger et al. [74] is shown in Figure 3-3. The figure underlines the
differences between SRF from different sources. The largest part of the SRF is fines,
which are particles less than 16 mm, and thus difficult to define. Plastics also
contribute a large fraction of the SRF composition and so does textiles and paper.
32 3 Fuels in the Cement Industry

The SRF can also contain significant amounts of inert materials such as metals, glass,
and stones.

The wide difference in SRF composition is also mirrored in the combustion behavior
of the fuel. Fuel rich in paper and plastics tends to have a high heating value, while
fuel rich in biomass and inerts has a low heating value [75].

The composition of SRF varies widely from different manufacturers [73], but the
quality from a single supplier can also vary substantially based on the origin of the
source material [75,76]. Vainikka et al. [77] reported on the fuel quality from one
supplier of SRF where the SRF was sampled over 4.5 years. Table 3-3 shows the large
span of quality that was found in the study.

Table 3-3: Example on the quality variation of SRF. Measurements based on one SRF
supplier over 4.5 years [77].

Min. Mean Max.

Moisture
7.9 16.7 40.4
(wt% as received)
Ash
4.5 8.9 20.2
(wt% dry matter)
Lower Heating value
13.3 19.0 24.2
(MJ/kg as received)
Some cement plants have a direct discharge of SRF from trucks to firing in the kiln
without intermediate storage. This can obviously result in significant changes when
trucks from different suppliers are handled. Other cement plants have an
intermediate storage, where the SRF is mixed and homogenized prior to use. An
example of such a system is the FLSmidth Feedex reclaimer [78]. This will to some
extent lower the short-term variations in fuel quality, but not eliminate it
completely.

3.5 Challenges of Alternative Fuels Firing

The benefits of utilizing alternative fuels in the cement industry have been outlined
above in chapter 3.2. However, the utilization of alternative fuels also introduces
some challenges since most alternative fuels are vastly different from conventional
fossil fuels. This chapter will give an overview of some of the challenges typically
encountered when using alternative fuels.
3.5 Challenges of Alternative Fuels Firing 33

3.5.1 Experiences of Alternative Fuels Co-Firing

Lockwood and Oy [79] presented an early review of RDF co-firing in a cement kiln.
They noted that the RDF combustion requires more combustion air and generates
more flue gas, which reduces the thermal efficiency of the kiln. Thus, the specific
heat consumption is increased, and production may be limited, if the kiln system
cannot cope with the increased flue gas amount. They concluded that the upper limit
of RDF use would be 10-15 percent of the total fuel consumption.

The Cemflame 3 experiments [80] aimed to evaluate the effect of co-firing coal with
different alternative fuels, such as sewage sludge and plastic waste. Experiments
were made in a pilot scale cement kiln simulator of 9 m length and 0.78 m in
diameter. In total, 240 different flames were investigated. The main problems of the
alternative fuels are their particle size, which results in lower combustion and heat
release rates than the solid fossil fuels. It was recommended that the specific surface
area of particles should be at least 100 m2/kg (approximately corresponds to dp =
600 µm for ρp = 1000 kg/m3), to obtain burnout in the gas phase of an industrial
cement kiln.

Figure 3-4: Effect of the amount of swirl and axial air on a flame co-fired with 30 %
pulverized sewage sludge and comparison to coal flame (base). The temperatures are
measured as kiln wall temperature [80].

The influence of burner settings was also investigated in the Cemflame 3

experiments. The swirl could help move the ignition point of the co-fired flame
34 3 Fuels in the Cement Industry

closer to the burner and increase flame temperatures. This in turn increases the heat
flux to the kiln simulator wall and the temperature of the wall. An example is shown
in Figure 3-4 for a flame co-fired with 30 % pulverized sewage sludge. It is noticed
how the cofired flame results in lower wall temperatures than the baseline coal
flame. For the co-fired flame, the wall temperature is increased by increasing the
swirl air.

Ariyaratne et al. [81,82] have made some detailed tests of the influence of co-firing
meat and bone meal (MBM) [81] and solid hazardous waste (SHW), which consisted
of organic solvents mixed with wood chips [82]. In these tests, the co-firing rate was
increased stepwise, and the effect on clinker quality and other operational data were
measured. The data from the MBM tests are shown in Figure 3-5. The main
limitation was found to be an increased free lime content in the clinker, which was
believed to be caused by lower flame temperatures [83,84]. CaO and P2O5 in the
MBM ash could also have a negative impact on free lime content. It was found that
the upper limit for co-firing was 20 and 40 % of the energy input, for the SHW and
MBM, respectively. The SHW had a considerably larger particle size, which explains
its lower co-firing limit.

Figure 3-5: Free lime, SO3, and P2O5 content in clinker for different MBM feed rates
[81].

Nørskov [31] studied the effect of injection velocity of dried sewage sludge on the
kiln system. It was found that an injection velocity of 30 m/s was to be preferred
compared to 48.5 m/s. With the high injection velocity, it was possible to achieve up
3.5 Challenges of Alternative Fuels Firing 35

to 6 % substitution of the petcoke energy, but with the low injection velocity 10-12
% substitution was achieved. The low injection velocity results in more fuel burning
closer to the burner, which releases the fuel energy in the clinker burning zone,
where it is required. With the higher injection velocity, more fuel energy is released
downstream of the clinker burning zone. The main limitation of sewage sludge use
turned out to be a higher evaporation of sulfur, which led to blockings in the calciner.
To reduce the sulfur evaporation, it was necessary to increase the kiln inlet oxygen
concentration from 1-2 vol% to 5 vol%.

The examples presented here demonstrates common issues encountered when co-
firing alternative fuels. The different limitations will be discussed in further detail in
the following chapters.

3.5.2 Fuel Composition

The composition of alternative fuels is different from fossil fuels, as indicated in
Table 3-4. Typically, the volatile content will be higher and consequently the
char/fixed carbon content will be lower. This is most likely a positive attribute of
the alternative fuels, since fuel with a high volatility are easier to ignite. In addition
the char that is produced from biomass is also more reactive than coal char [85]. For
pine wood particles of around 3 mm in diameter combusted in 20 % oxygen, the
devolatilization and char burnout both takes approx. 2-4 seconds [86]. When the
oxygen concentration is lowered to around 10 %, the char combustion takes 4-5
times as long as the devolatilization [87]. Consequently, the char combustion is
highly dependent on the amount of oxygen available, while the devolatilization only
requires sufficient heat, and when the volatiles are released they burn quickly, if
they can be mixed with oxygen.

The minor elements contained in alternative fuels can be mixed with the clinker in
the kiln and have a significant impact on the clinker reactivity or the process stability
[88]. The most significant species are sulfur and chlorine. Sulfur can be beneficial in
low amounts in the clinker, since CaSO4 melts at low temperatures and acts as a
fluxing agent in the kiln [13]. However, sulfur in excess of 1.2 % may decrease the
alite content and stabilize belite and free lime [13,89]. Excessive chlorine in the
clinker may cause corrosion of steel reinforcement in the final concrete [13] and the
level of chlorine in Portland cement is typically limited to 0.1 % [4]. The impacts of
36 3 Fuels in the Cement Industry

circulation of these species is discussed in chapter 3.5.7. While the sulfur content of
SRF is low, the chlorine content is relatively high compared with coal or petcoke.

Table 3-4: Proximate and ultimate analysis of select fuels used in the cement industry.

Paramet Bitum. Pet- Pine Polyet-

Unit Oilc SRFe MBMf
er coala cokeb Woodd hylened
Moisture wt% a.r. 3.0 - - 8.4 0 16.7 4.0
Volatiles wt% dry 39.4 11.0 - 81.5 98.6 70.8 63.4
Ash wt% dry 14.1 1.2 - 0.1 1.4 8.9 28.2
Fixed
wt% dry 46.5 87.8 - 18.4 0.03 13.2 8.3
Carbon
C wt% dry 68.7 86.5 85.7 51.6 83.7 54.7 49.1
H wt% dry 4.9 3.6 10.9 6.1 15.5 7.7 7.2
N wt% dry 1.1 1.8 0.2 0.1 0.01 0.7 10.1
S wt% dry 1.8 5.5 2.6 0.1 0 0.18 0.5
O wt% dry 9.5 1.7 0.5 42.0 0 36.7 4.9
Cl wt% dry 0.03 0.02 - 0 0 0.41 0.20
Na wt% dry 0.08 0.05 - 0 - 0.14 0.42
K wt% dry 0.38 0.42 - 0.02 - 0.09 0.08
LHV MJ/kg dry 27.8 34.1 40.2 19.0 42.2 23.2 19.4
a: Data from [90] based on average of 11 samples. Na and K content from [91] based on 5 samples.
b: Data from [92] based on average of 22 samples. Cl, Na, and K from [93]
c: Data from [94].
d: Data from [95]. Na and K from [96]
e: Data from [77]. Volatiles and fixed carbon from [97].
f: Data from [83]; Cl, Na, and K content from [98].

Other trace elements that may influence the clinker quality are alkalis and
phosphorous. Sodium and potassium are important for binding sulfur in the clinker,
but may also inhibit alite formation or cause expansion of the concrete leading to
cracking [14]. Phosphorous is especially high in meat and bone meal, and has a
tendency to stabilize belite and lower the alite formation in the kiln [28,88].

3.5.3 Particle Size

Solid fossil fuels, like coal and petcoke are normally grinded to below 100 µm, before
being fired in the kiln. This is necessary to ensure a fast heating and ignition of the
fuels, as well as a fast burnout. However, the water content and soft paper and
plastic fractions makes milling of SRF more problematic than milling of coal, and the
metal and glass content of SRF may result in significant wear of the mills [99]. Thus,
comminution of the particles can be quite expensive, and it is preferably avoided.
SRF particle size can be up to 100 mm [100], which can be utilized in cement
calciners equipped with auxiliary combustion devices such as the FLSmidth Hotdisc
3.5 Challenges of Alternative Fuels Firing 37

[101] or the Thyssenkrupp Prepol step combustor [102]. If direct firing in the kiln
is desired, the particle size will have to be smaller, typically < 30 mm as determined
by sieving [100].

The consequence of the larger particle size is a longer ignition and combustion time
[87], which may move the combustion zone further into the kiln and reduce the
flame peak temperature. In addition, unburnt particles can, depending on the
residence time in the flame, fall unconverted into the clinker bed. Here the fuel
depletes the oxygen, resulting in local reducing conditions, which is described in
further detail in chapter 3.5.6.

Normally, 40 % of the fuel energy needed in the cement process is fired in the kiln,
while the rest is fired in the calciner [31]. Due to the longer combustion time of SRF,
the fuel energy may not be released in the clinker burning zone, but rather further
upstream in the kiln. This may lead to a higher share of fuel in the main burner of
between 50-60 % of the total thermal input. Thus, the alternative fuel is used to
substitute calciner firing, rather than fossil firing at the main burner.

3.5.4 Moisture Content

The moisture content in SRF is generally high and may vary between 10-40 %, as
seen in Table 3-3. This has primarily three negative effects on the fuel: increased
density, increased conversion time, and lower peak temperature.

The moisture content in the fuel will increase the particle density, which makes wet
particles heavier than dry ones. An increased particle mass will influence the
particle terminal velocity and inertia. Consequently, it will be more difficult to keep
wet particles in suspension in the kiln and influence the particle flow by the burner
primary air.

Increased moisture content increases particle conversion times, since additional

energy is required to evaporate the water and heat the particles. For wood particles
this results in a delayed ignition time, which may be increased by a factor 10 [87],
when the moisture content is increased from 10 to 40 wt%. The devolatilization
time may be increased by a factor 2 [87,103]. This will evidently move the main
combustion zone further into the kiln and increase the risk of non-complete particle
conversion.
38 3 Fuels in the Cement Industry

Lastly, the moisture content significantly reduces the peak temperature in the
combustion zone, since less heat is carried per unit of fuel and significant amounts
of energy is used to evaporate the added water. The effect of the moisture on the
temperature can be sketched by calculating the adiabatic flame temperature at
different moisture contents, as shown in Figure 3-6. Here it is seen that increasing
the moisture content of SRF from 0 to 50 %, may decrease the adiabatic flame
temperature by more than 500 °C. Since the heat transfer from gas to clinker bed is
mainly radiation based and highly dependent on the temperature, such a
temperature difference may result in significantly lower heat transfer rates.

Figure 3-6 Adiabatic flame temperature for various fuels from Table 3-4 calculated at
different moisture contents at an air to fuel ratio of 1.

3.5.5 Kiln Dust

A high dust load in the rotary kiln can be very problematic since it limits heat
transfer from the flame to the clinker bed. Inside the kiln, saltation is the main form
of particle entrainment [104]. Saltation is a type of particle transport where large
particles are picked up by the air, but due to their size they fall back into the bed,
rebounding or ejecting other particles [27]. If light particles are present they may
follow the gas flow out of the kiln [105]. The most influential parameter on particle
elutriation and dust formation is thus the average gas velocity [104]. Saltation is
limited in the burning zone where the clinker melts and form nodules [105].
3.5 Challenges of Alternative Fuels Firing 39

However, an insufficient temperature and melt formation limits agglomeration and

dusty clinker can be formed [25]. If under-burnt clinker enters the cooler, large
amounts of fine dust particles are carried with the secondary air into the burning
zone limiting the visibility and severely cooling the burning zone [34].

3.5.6 Reducing Conditions

As mentioned in section 3.5.3, one of the disadvantages of the large particle size of
most alternative fuels, is that fuel may drop unconverted to the clinker bed, consume
the oxygen and result in local reducing conditions.

One adverse effect of reducing conditions is that Fe3+ can be reduced to Fe2+, which
results in a loss of ferrite phase (C4AF) and an increase in aluminate content (C3A).
The formed FeO may also substitute CaO in C3S. Thus, an effective increase of LSF,
SR, and AR is observed, and the clinker becomes harder to burn. To compensate, the
temperature needs to be raised, creating larger alite crystals and a poorer cement
quality. FeO can also replace CaO in the alite, which is more prone to decompose to
belite and free lime upon cooling, and with a lower alite content the compressive
strength of the cement is lowered [106,107]. The adverse effects of reducing
conditions may to a high extent be lowered if the clinker is rapidly cooled from
above 1250 °C in air [107].

The indicator for reducing conditions in the kiln is typically a brown color of the
clinker, especially in the center, compared to its normal dark grey. It is the calcium
aluminate ferrite (C4AF) that colors the cement. Pure C4AF is brown, but when
magnesium ions (Mg2+) are incorporated in the crystalline phase, the color changes
to grey. When the iron is reduced from Fe3+ to Fe2+, under reducing conditions, it
can displace the Mg2+-ions, which affects the color of the clinker [107,108]. It is
normal for some Fe3+ to be reduced to Fe2+ in the hottest parts of the kiln, but it will
normally be re-oxidized during cooling, if oxygen can diffuse into the center of the
clinker nodules. Thus, peak temperatures and clinker nodule size and porosity also
affects the clinker color [109]. If the clinker produced in a reducing atmosphere is
rapidly cooled from above 1250 °C in oxidizing conditions, the brown color can be
avoided [107].
40 3 Fuels in the Cement Industry

3.5.7 Circulation of Volatile Species

Under normal kiln conditions, evaporation of sulfur, chlorine, and alkali occurs in
the hot parts of the flame zone. The volatility is heavily influenced by temperature,
retention time in the burning zone, particle size, and concentration of sulfur and
oxygen in the kiln atmosphere [110]. Since the escape of volatile species from the
solids is dependent on the particle size and surface area, it will be most prominent
during the early stages of clinkering, before the grains have sintered together [111].
The species that do not fully evaporate in the kiln, leaves the kiln with the clinker.

The evaporated species are transported with the kiln gas towards the inlet of the
kiln, where they may condense on or react with the colder dust from the preheater
and calciner. In this way, the volatile species are returned to the kiln, and a
circulating pattern is established, which can lead to accumulation of volatile species
in the kiln. As the concentration of volatile species increase they can condense on
surfaces in the kiln, preheater or calciner, where it can lead to blockages and ring
formation, severely affecting the process operation by a restriction of material and
gas movement. [13].

Local reducing conditions can enhance the evaporation of sulfur, which can take
place at lower temperatures than in oxidizing conditions. The sulfur release is e.g.
enhanced by the reaction with CO in R 3.1 or the shift of equilibrium in R 3.2, when
oxygen is depleted [112]:

𝑪𝒂𝑺𝑶𝟒 (𝒔) + 𝑪𝑶(𝒈) → 𝑪𝒂𝑶(𝒔) + 𝑺𝑶𝟐 (𝒈) + 𝑪𝑶𝟐 (𝒈) R 3.1

𝟏
𝑪𝒂𝑺𝑶𝟒 (𝒔) ⇌ 𝑪𝒂𝑶(𝒔) + 𝑺𝑶𝟐 (𝒈) + 𝑶𝟐 (𝒈) R 3.2
𝟐

Measurements made in a pilot rotary kiln simulating the conditions of the kiln raw
material inlet, indicate that the evaporation of sulfur mainly occurs if fuel
devolatilization occurs in the clinker bed [54]. If these results are valid for the hotter
parts of the rotary kiln, they may indicate that sulfur evaporation can be limited if
fuel devolatilization occurs in suspension, while char oxidation can be allowed in the
bed.

The sulfur can form different chemical compounds, depending on the availability of
other species in the kiln, which may influence the volatility of sulfur. The following
order of combination of the volatiles has been observed [110]:
3.5 Challenges of Alternative Fuels Firing 41

1. Chlorine has a strong affinity for K and Na and forms KCl and NaCl. The
remaining chlorine forms CaCl2.
2. The residual alkalis combine with sulfur to K2SO4 or Na2SO4. If an excess of
alkali is present they can be present as carbonates or hydroxides [112].
3. The residual sulfur combines with calcium to form CaSO4.

The volatility of chlorine is very high and normally most of the chlorine evaporates
in the burning zone of the kiln. However, the volatility of sulfur is very dependent
on the specific compound. At the high temperatures in the kiln CaSO4 is fairly
volatile, while alkali sulfates are less volatile. Thus a measure for the tendency of
sulfur evaporation is given by the sulfur modulus (MSO3), calculated based on the
mass fraction of sulfur, potassium, and sodium [23]:

𝒀𝑺𝑶𝟑
𝑴𝑺𝑶𝟑 = 𝟖𝟎 E 3.2
𝒀𝑲𝟐 𝑶 𝒀𝑵𝒂𝟐 𝑶
𝟗𝟒 + 𝟔𝟐

The coefficients of the equation are equivalent to the molar mass of the involved
species. A MSO3 value around 1 is considered sufficient to ensure a surplus of alkalis
to combine with the sulfur. In this case, the sulfur volatility (defined in E 3.3) is in
the range 0.3-0.5, while if there is a surplus of sulfur over alkalis most of the sulfur
evaporates and the volatility is 0.9-1. Generally, a value above 0.7 indicates too high
sulfur recirculation, meaning that sulfur will accumulate in the kiln system. The
volatility is defined as [110]:

𝒀𝑺𝑶𝟑 𝒄𝒍𝒊𝒏𝒌𝒆𝒓
𝑺𝑶𝟑,𝒗𝒐𝒍 = 𝟏 − E 3.3
𝒀𝑺𝑶𝟑 𝒉𝒐𝒕 𝒎𝒆𝒂𝒍

Another definition of the SO3 available to form CaSO4 is given as excess SO3. The
formula is given in E 3.4 and indicates the amount of sulfur not bound by alkali
sulfates. If there is no significant excess SO3 the sulfur volatility will be low around
0.35, but it will increase with additional excess SO3 [113].

𝑬𝒙𝒄𝒆𝒔𝒔 𝑺𝑶𝟑 = 𝒀𝑺𝑶𝟑 − 𝟎. 𝟖𝟓 𝒀𝑲𝟐𝑶 − 𝟎. 𝟔𝟓 𝒀𝑵𝒂𝟐𝑶 E 3.4

The build-ups in the kiln and preheater are primarily a mixture of K2O, SO3, CaO, and
Cl-, the exact combination differs widely depending on the location and the
operating conditions at the specific plant [114]. Common minerals found in rings
and deposits in the kiln and preheater are listed in Table 3-5.
42 3 Fuels in the Cement Industry

Table 3-5: Typical compounds found in deposits in the cement kiln and preheater
[23,112].

Name Formula Typical Location

Spurrite (2CaO·SiO2)2·CaCO3 Lower cyclones

Ca10(SiO4)3(SO4)3X2
Ellestadite Riser duct
Where X = OH, F, Cl

Sulfospurrite (2CaO·SiO2)2·CaSO4 Kiln rings

Anhydrite CaSO4 Cyclones and riser duct

While the sulfur content in alternative fuels such as SRF is significantly lower than
that of coal or petcoke, and the alkali content is similar, the chlorine content is
significantly higher, see Table 3-4. The build-up tendency in the kiln and preheater
is significantly increased when both sulfur and chlorine is present, as indicated by
Figure 3-7. This explains the potential increased deposit formation from alternative
fuels, even though the sulfur input from the fuel is significantly lowered. KCl
promotes the formation of deposits, either by condensing on lime dust to form a
sticky surface leading to agglomeration and build-ups [115] or acting as a
mineralizer in the formation of spurrite [116]. Another adverse effect of high
chlorine concentration is that it promotes corrosion in the kiln [117].

Figure 3-7: Build up tendency as function of SO3 and Cl in the hot meal [112].
3.6 Summary and Conclusions 43

A more detailed description of the circulation of volatiles, especially sulfur, is given

by Choi and Glasser [111], Nielsen et al. [14,53,118], and Cortada Mut et el.
[54,112,119].

3.5.8 Chlorine Bypass

To minimize the problems caused by recirculating species and deposits, a bypass
can be installed at the kiln inlet. This is typically needed to achieve high degrees of
alternative fuels firing, without excessive build-up problems. The bypass works by
extracting a fraction of the hot air at the kiln inlet, where the concentration of
chlorine is high. The hot air is then mixed with cold air in a quench chamber in order
to condense the chlorine [28].

A disadvantage of a kiln bypass is that hot gas is removed from the preheater causing
a considerable heat loss. Additionally, large amounts of bypass dust has to be
disposed of [120]. In order to decrease the heat and material loss, the particles can
be separated in a cyclone, since small particles have a high concentration of chlorine,
while the large particles are mainly cement raw meal and can be returned to the kiln
[120]. At plants utilizing large amounts of alternative fuels, the bypass dust amount
can reach as much as 10,000 ton per year [121]. Thus, the companies Holcim and A
TEC have developed a system called ReduDust, where the chlorine is removed from
the bypass dust. This creates a KCl rich salt that can be sold for industrial purposes
and the cleaned bypass dust can be used in the cement grinding [122].

3.6 Summary and Conclusions

There are several considerations to make when a fuel is selected for use in the
cement kiln. Fuel cost, environmental impact, and potential impact on the kiln
process are some of the factors that should be considered. Traditional use of fossil
fuels account for approximately 30 % of the operating costs at a cement plant. Thus,
a high incentive to use cheaper alternative fuels is present. Furthermore, the
utilization of waste fuels can divert waste away from landfills and partly biogenic
fuels has the potential to reduce CO2 emissions.

The most common alternative fuels include shredded tires, meat and bone meal, and
Solid Recovered Fuel (SRF). The properties of most alternative fuels are very
different from fossil fuels. The main challenges are related to a larger particle size,
44 3 Fuels in the Cement Industry

higher moisture content, and decreased combustion temperatures. Consequently,

the introduction of alternative fuels can cause different issues in the cement kiln.
Decreased temperatures can cause a lower cement quality and potentially increase
amounts of dust in the kiln. The longer combustion time can result in unconverted
fuel in the bed, which induces local reducing conditions. This can promote sulfur
recirculation, which may cause blockages in the kiln system. Another adverse effect
of reducing conditions, is the discoloration of clinker and reduced quality.

The discoloration and lower clinker quality caused by reducing conditions, may to
some extent be reverted if the clinker is rapidly cooled in an oxidizing atmosphere.
Sulfur evaporation may be limited if the fuel undergoes devolatilization before it
enters the bed.
4 KILN PROCESS MONITORING

Monitoring and control of the kiln combustion is paramount for obtaining a good
product quality as well as stable and economical production. As already discussed
in previous chapters the quality of alternative fuels can change quite drastically over
time, especially SRF as seen in Table 3-3. As an example, the moisture content and
the heating value may change, which can alter the temperature in the clinker
burning zone. Thus, monitoring of the combustion is particularly important when
dealing with high amounts of alternative fuels.

This chapter aims to give an overview of methods that can be used to monitor the
cement kiln processes. The measures discussed here can help determine how co-
firing with alternative fuels influence the kiln. As such the methods are important
for the industrial tests, described in chapters 5 and 6.

4.1 Clinker Quality

The primary objective of the kiln is to create a cement clinker of high quality. This is
achieved as a combination of residence time and temperature in the burning zone.
If the burning zone is too cold or the clinker passes through it too quickly, it will
result in under-burnt clinker. In the under-burnt clinker, the clinker reactions have
not reached a sufficient conversion, which results in reduced reactivity and strength
of the concrete. In addition, high amounts of free lime may create destructive
expansion [13]. Conversely, if the clinker is over-burnt at too high temperature or
time, the clinker mineral crystals grow too large, and the clinker will be difficult to
grind [17].
46 4 Kiln Process Monitoring

As an indicator for the completeness of the clinkering reactions the free lime content
is often used. The free lime content is the CaO in the clinker that has not combined
with SiO2, Al2O3, or Fe2O3. If the free lime content is increasing, it indicates that the
burning zone is cooling, and vice versa. The free lime content should normally be
around 1 % (Table 2-1). The simplest method for measuring if the clinker is burnt
correctly is to fill a cone of 1 litre volume and fill it with loosely packed clinker. The
mass should preferably be in the range around 1.25-1.35 kg/l. A lower density
indicates under-burning, while a higher density indicates over-burning [13]. More
precise measurements of free lime content are also available [123].

Today, X-ray fluorescence (XRF) and X-Ray diffraction (XRD) methods are used in
many cement plants. The methods give a possibility to determine the complete
elemental analysis (XRF) and quantify the crystalline phases (XRD) of the clinker.
XRD is a powerful tool that can be used to directly measure the clinker phases and
amount of free lime, and thus directly gives a measure of the clinker quality
[124,125]. The plant operator can use this information to determine if the clinker is
adequately burned and make appropriate adjustments.

A drawback of using clinker measurements as the primary means of control is that

there is a significant time lag between the time of sampling and the analysis.
Typically, it might take up to 45 minutes from a clinker is discharged at the cooler
until it is ready in the laboratory. Furthermore the analysis may take between 30
and 90 minutes [17]. Thus, it is also important for the operator to keep an eye on
other more rapid, but indirect measures, to monitor the kiln.

4.2 Burning Zone Temperature

The burning zone temperature is important to monitor closely. The temperature can
be measured in different manners through e.g. thermocouples, radiation
pyrometers or acoustic pyrometers [126]. The use of thermocouples can often be
used to measure the secondary air temperature in the kiln hood and the
temperature in the kiln inlet. To measure the temperatures in the flame or burning
zone, pyrometric methods based on infrared light are normally used [127]. Other
indicators of the burning zone temperature are NOX in the kiln inlet or kiln torque.
4.3 Gas Measurements 47

4.3 Gas Measurements

The gasses at the kiln inlet are normally measured for the contents of O2, CO, and
NOX, which gives an indication of the quality of combustion and conditions in the
kiln. The benefits of monitoring the kiln off gases are e.g. kiln performance gains,
energy conservation, emission reductions, and improved production quality [128].

The measurement of oxygen at the kiln inlet is important for controlling the
combustion in the kiln. If the oxygen concentration becomes too low there is a risk
of reducing conditions, which impacts clinker quality and sulfur evaporation as
discussed in chapters 3.5.6 and 3.5.7. In addition, the presence of CO in the kiln inlet
may be increased if insufficient oxygen is present. This can result in a heat loss and
significant pollution, if it is not oxidized in the calciner. On the other hand, a too high
oxygen content results in a diluted off gas with lower peak temperature, since the
excess air needs to be heated by the combustion. The oxygen content at the kiln inlet
should preferably be in the range 0.7-3.5 % [34].

Measuring the NOX level in the kiln inlet can give an indication of the burning zone
temperature [127]. The majority of NOX from the cement kiln is thermal NOX, while
smaller amounts are fuel and prompt NOX , thus the formation is highly dependent
on the temperature in the burning zone [129]. The NOX level from the kiln may be
reduced in the calciner where fuel is injected into oxygen deficient areas [129].

Getting reliable gas measurements at the kiln inlet can, however, be difficult. The
high temperature, high dust load, as well as corrosive and condensable species in
the kiln inlet challenge the design of gas extraction probes. The design of probes
must entail cooling, cleaning and gas extraction, but the complex design is also more
prone to failure [128].

4.4 Kiln Torque

The measurement of the kiln torque, as main drive power consumption or amps, can
give a relative estimate of the kiln burning zone temperature. As the temperature
increases in the kiln the amount of molten material in the kiln will also increase. This
causes the material in the kiln to be carried further up the side of the kiln as it
rotates, which increases the required torque to rotate the kiln. The amount of
material in the kiln also affects the torque. If e.g. the coating thickness changes, this
48 4 Kiln Process Monitoring

can also influence the kiln torque [17]. The power consumption is a function of the
torque required to turn the kiln and the rotational speed:

𝑷𝒌 = 𝟐𝝅𝒏𝑻𝒒 E 4.1

Here Pk is the power consumption of the kiln, n is the rotational speed (rot/s) and
Tq the torque.

4.5 Kiln Cameras

Cameras can be placed in the kiln hood of the cement kiln and be used to observe
the near burner region of the kiln. A typical flame image from a kiln camera is shown
in Figure 4-1. Using this information, it can be monitored how the flame ignition
point, shape, or intensity changes when e.g. changing fuels and the operators can
adjust accordingly.

Figure 4-1: Image from a kiln camera showing the combustion in the kiln [130]. The
hot clinker bed can be seen on the left side. Unignited pulverized fuel is seen directly
in front of the burner. The combustion zone (bright area at center) is further
removed from the burner.

The German Cement Works Association (VDZ) has used kiln cameras to evaluate the
effect of fuels and burner settings on the kiln flame. However, only very sparse
information was published on this area [30,131]. There are also examples in the
literature of such cameras being used for modeling and control of the kiln process.
Lin et al. [130] have attempted to predict the NOx emission from a cement kiln based
on partial least squares regression (PLSR). They showed the information from a
camera in combination with regular process measurement can be used to give a
4.6 Kiln Scanners 49

better prediction of NOx than using the process measurements alone. The company
Powitec [132] also uses flame images in their automatic control for kiln burners.

4.6 Kiln Scanners

Another method to monitor the kiln is the use of infrared scanners to measure the
temperature of the outside kiln shell [133]. The kiln scanner can be used to monitor
for high temperatures on the kiln shell, so called hotspots, which may indicate
damage to the refractory or loss of coating. The kiln shell temperature will also
change depending on the temperatures inside the kiln. Nørskov [31] showed how
co-firing with dried sewage sludge can affect the kiln shell temperature profile, as
shown in Figure 4-2. Lower temperatures are seen in the area 22-27 m, and higher
temperatures are seen around 18-20 m, which may indicate spilling of the
alternative fuel into the clinker bed in this area.

Figure 4-2: Kiln shell temperatures when firing petcoke (a) and co-firing with 10 %
sewage sludge by energy (b). Temperatures are lower around 22-27 m when co-
firing, but higher around 18-20 m [31].

4.7 Measures Against Changing Conditions

If the conditions in the kiln changes, there are several control options possible for
the kiln operator, to revert the kiln back to stable conditions.
50 4 Kiln Process Monitoring

If the oxygen concentration becomes too low or CO spikes are observed in the kiln
off gasses, not enough air is available in the kiln. The air level can be increased by
increasing the draft in the kiln by adjusting the ID fan.

The burner primary air, axial and swirl air, can be used to influence the flame shape.
If the burning zone is too cold, it might be necessary to increase the flame intensity,
which can be achieved by a higher momentum or swirl air. If hot spots on the kiln
scanner are detected, the temperature may be too high, or the flame can be
impinging on the refractory. This might be remedied by lowering the swirl, which
tends to narrow the flame.

Plants that fire large amounts of alternative fuels normally also utilize a fossil fuel
such as coal or petcoke. If the burning zone temperature decreases due to e.g. too
much water in the alternative fuel, it will be common to increase the amount of fossil
fuel, to increase the temperature.

The kiln rotational speed can be used to control the residence time of the clinker in
the kiln. If the clinker is overburnt, the residence time in the burning zone can be
reduced by increasing the rotational speed of the kiln.

If the other factors fail to make an impact it is also possible to change the kiln feed.
With a lower kiln feed, it is easier to adequately burn the clinker, and the kiln will be
less sensitive to changes.

Furthermore, it is also possible to increase the firing in the calciner to increase the
calcination degree, which lowers the fuel requirement in the kiln.

4.8 Summary and Conclusions

To ensure a good and stable product quality it important to monitor the kiln process.
This is especially the case when utilizing alternative fuels, as their quality, e.g.
moisture content and heating value, can change significantly over time, which can
influence the combustion process. The quality of clinker samples can be determined
by use of X-ray fluorescence (XRF) and X-ray diffraction (XRD). However, it is also
important to monitor more indirect measures, which offer a quicker response.
These include e.g. gas measurements, burning zone temperatures, and the kiln
torque. Changes to the kiln flame, such as intensity and ignition point, can be
monitored via a camera.
5 EFFECT OF ALTERNATIVE
FUELS CO-FIRING ON KILN
FLAME

The experimental part of this PhD project constitutes two studies that were made in
industrial cement kilns. Furthermore, a characterization study of SRF was made to
obtain parameter inputs for combustion modeling purposes.

The first study, described in this chapter, investigates the effect of alternative fuels
co-firing on the cement kiln flame. This was done with a camera specially developed
to be inserted in the kiln hood, close to the flame. This allows for detailed imaging of
the flame. Measurements were performed at three different cement plants. At one
of these plants the kiln burner was changed to the FLSmidth Jetflex burner, which
allows for a comparison of two burners. This work has been published as:

M. N. Pedersen, M. Nielsen, S. Clausen, P. A. Jensen, L. S. Jensen, and K. Dam-Johansen,

“Imaging of Flames in Cement Kilns To Study the Influence of Different Fuel Types,”
Energy & Fuels, vol. 31, no. 10, pp. 11424–11438, 2017.

The second study sought further details into the possibilities of adjusting the flame
of the Jetflex burner. Furthermore, it was investigated how the burner settings can
influence co-firing. Details of these experiments can be found in chapter 6.
52 5 Effect of Alternative Fuels Co-firing on Kiln Flame

The characterization study of SRF can be found in chapter 7. The results of the study
have been written into an article manuscript that has been submitted to the journal
Energy and Fuels:

M. Nakhaei, M.N. Pedersen, H. Wu, D. Grevain, L.S. Jensen, P. Glarborg, P.A. Jensen, K.
Dam-Johansen, “Aerodynamic and Physical Characterization of Refuse Derived
Fuel,” Energy and Fuels (Manuscript).

5.1 Description of Test Cement Plants

Measurements were carried out at three different European cement kilns all
producing ordinary Portland cement clinker. Brief details of the plants are given
below. Details of the fuels used at each plant are collected in Table 5-1.

Plant 1 produces around 3,500 ton clinker per day and has a kiln with a diameter of
5 m and 77 m length. It is equipped with a 5 stage preheater, in-line calciner, and an
FLSmidth HOTDISC device, primarily burning coarse SRF, providing energy to the
calciner. Under normal conditions approx. 65 MW is fired in the kiln with 70 % of
the energy being from SRF and the remainder from petcoke. The plant maintains a
kiln inlet oxygen concentration of 4-5 %.

Plant 2 operates a semi-dry process with a capacity of 4,300 ton of mineralized

clinker per day. It is equipped with a 2-string 5 stage preheater and in-line calciner,
where the main fuel is SRF supplemented by coal. The kiln has a diameter of 4.75 m
and is 74 m long. In the kiln, the fuel energy input is 70 MW with the main fuel being
a mix of 75 % coal and 25 % petcoke. As alternative fuel SRF or granulated tires can
be used in amounts of up to approximately 10 or 30 % of the energy input,
respectively. The oxygen concentration at the kiln inlet is 5-6 %.

Plant 3 has a clinker capacity of 3,100 ton per day in a kiln of 5 m in diameter and
length of 68 m. The plant has a 5-stage preheater, an in-line calciner, and FLSmidth
HOTDISC, burning primarily whole tires and coarse SRF <120 mm. In the kiln,
petcoke is used as the main fuel. The plant uses several types of alternative fuels in
the kiln with the largest fraction being SRF and smaller amounts of dried sewage
sludge or waste oil contributing up to 50 % of the energy in the kiln. The total energy
input of fuel in the kiln is around 65 MW. Due to defunct equipment, the oxygen
concentration at the kiln inlet was not monitored during the measurement
campaign; however, it is likely around 5 %.
5.1 Description of Test Cement Plants 53

The three plants have different burners from major burner suppliers. At plant 1 the
burner was recently changed to the Jetflex burner from FLSmidth. This allowed a
comparison of two different burners at the same site. The Jetflex burner is described
in chapter 2.8.

Details of the fuels used at the three plants are shown in Table 5-1. The analysis of
the fuel is on a wet basis. The samples for coal and petcoke listed for Plant 2 and 3
are taken from the stock at the respective plants. Thus, the moisture content is
relatively high compared to that of Plant 1. Before being used in the kiln the solid
fossil fuels are grinded in mills to a particle size < 100 µm and dried to a moisture
content of approx. 2 % at Plant 2 and 0.6 % at Plant 3. An example of the SRF used
at the different plants is given in Figure 5-1.

Table 5-1: Properties of the fuels utilized at the cement plants. Data are on an as
received basis. Moisture and ash in coal and petcoke depends on sampling before
(Plant 2 and 3) or after milling (Plant 1).

Particle Typical
LHV Moist. Ash Vol. S Cl
Size use

Plant Fuel MJ/kg wt% wt% wt% wt% wt% % or t/h

3.5 % > 90 20-40 %

1 Petcoke 32.3 0.9 7.2 13.8 3.4 0.04
µm sieve energy
60-80 %
1 SRF 17.5 19.8 15.5 45.9 0.5 0.66 < 30 mm
energy
13 % > 90
2 Coal 25.2 11.9 6 34 0.44 0.02 7
µm sieve
44 13 % > 90
2 Petcoke 30.7 8.8 0.58 9.8 5.89 2
(ppm) µm sieve
2 SRF 18.7 17.3 12.5 - - - - 1.5
Granulated
2 34.4 1.5 6 - - - - 3
tires

3.5 % > 90
3 Petcoke 31.9 6.1 1.3 11.8 5.5 - 4
µm sieve
Max
3 SRF 19.3 21.5 - - 0.29 1 30x30x2 5
mm
Sewage
3 13.8 9.2 28.2 - - - - 1
Sludge
54 5 Effect of Alternative Fuels Co-firing on Kiln Flame

Figure 5-1: SRF used at plant 1 (a), Plant 2 (b), and Plant 3 (c).

Another important aspect of how the burners operate, is the use of primary air
which is used to shape the flame [30]. This is typically divided into axial air, which
adds axial momentum to the flame, and swirl (radial) air, which may create internal
recirculation zones and stabilize the flame [35,42]. Transport air is also used to
pneumatically convey the fuel through the burner. In addition air is used to cool the
burner, often called central air, and occasionally some air can be used to disperse
alternative fuels [31]. Table 5-2 contains an overview of the primary air used at the
kiln burners at the three different cement plants.

Table 5-2: Amount and pressure of primary air typically used in the burners at the
three cement plants. Plant 3 uses a burner with no separate swirl air channel.

Plant 1 Plant 2 Plant 3

Old New
Burner Burner
Primary Air Amount (m3/h)
12000 12000 6200 6600
(Excl. transport air)
Axial Air Pressure (mbarg) 230 210 300 220

Swirl Air Pressure (mbarg) 240 150 40 -

Other Air Pressure (mbarg) 50 10 70 100

Transport Air Amount (m3/h) 4000 4000 4400 4200

Fraction Primary Air
~20 ~20 ~13 ~10
(% of total combustion air)

5.2 Kiln Camera Probe

The cameras used in the cement kilns are typically installed at the back of the kiln
hood to be protected from the high temperatures in the kiln. However, the view can
be obstructed by large amounts of dust in the kiln hood [130]. For this work, a
portable kiln camera was developed. The camera used was an IDS camera (model
number UI-5240CP-C-HQ) with a resolution of 1.3 megapixels and a maximum
5.3 Description of Flame Images 55

framerate of 50 frames per second. The camera was placed in an approximately 2 m

long water cooled probe, Figure 5-2a. The probe was made of concentric stainless
steel tubes with cold (~10 °C) cooling water flowing at the center and the hot (~60
°C) return water at the outside. The camera lens was covered by a neutral-density
filter, with optical density 1, to reduce the light intensity and protect the camera
from thermal radiation. Compressed air is passed through the probe and helps keep
the filter clean from dust and provides additional cooling of the camera. The camera
probe can be inserted through the side of the kiln hood close to the burner tip as
shown in Figure 5-2b+c. This makes it possible to get a clear image of the burner
compared to conventional cameras, which are placed further away from the burner.

Figure 5-2: a) The camera inserted in the probe. b) Camera probe inserted through
side of kiln hood. c) Camera probe inserted next to kiln burner.

5.3 Description of Flame Images

The main objective of this study has been to get an increased understanding of the
effect that alternative fuels have on the flame in the cement kiln based on visual
observations. Several hours of video footage under different operating conditions
have been recorded at the cement plants included in this study. In the following
sections those hours of video recordings are condensed into a few representative
images of the flames.

The sketch in Figure 5-3a shows how the camera is inserted through the side of the
kiln hood and the approximate field of view into the kiln. The Image in Figure 5-3b
shows the typical view seen with the camera. On the right side of the image the
burner tip is seen. The fuel is injected through the burner forming a dark flame
plume, which expands and becomes wider as it moves away from the burner. At
56 5 Effect of Alternative Fuels Co-firing on Kiln Flame

some point the fuel is ignited, giving a bright high intensity region. Below the fuel
plume, the hot clinker bed at the bottom of the kiln can be seen. Above the fuel plume
the opposite kiln wall can be observed. It is slightly darker than the clinker,
indicating a lower temperature. The camera can be moved around, which will
change the view. For instance, the camera can be turned downwards under the fuel
plume, to observe if fuel drops to the clinker bed.

In the following the ignition point is defined as the distance from the burner where
a sudden change in image intensity is observed. The distance will be measured along
the center of the fuel plume. In Figure 5-3b this is seen as the point where the dark
fuel plume changes color to bright yellow/white. As seen in the image this occurs
earlier at the top and bottom of the flame plume compared to the center, but the
center value will be used as the ignition point.

Figure 5-3: a) Sketch of the camera view (top view). b) Explanation of camera view.

The cement flame is characterized by a high degree of turbulent motion, which

means that it is constantly fluctuating. In addition, there can be various amounts of
dust in the kiln, which can influence the visibility. The images in Figure 5-4a-c are
taken 0.4 seconds apart and show how the flame ignition point and plume width
change. To limit the effect of the short term variations an average frame can be
produced. The greyscale image in Figure 5-4d s has been averaged over 5 seconds.
This concept will be used through the remainder of the paper, where images in color
are single frames from the recorded videos and images in greyscale are averaged
over 5 seconds. In the images, the distance to the burner tip has been estimated, with
the numbers indicated in meters. This will help estimate how much e.g. the ignition
point is changed by different fuel types. The description of how the distances are
estimated can be found in the supporting material of the published article [134].
5.4 Results from Plant 1 57

Figure 5-4: Images taken 0.4 seconds apart showing the fluctuations in the kiln flame.
d) Frame averaged over 5 seconds. Images from Plant 1 during full petcoke load.

5.4 Results from Plant 1

5.4.1 Petcoke Fired Flame

A wholly petcoke fired flame from the old and new burners used at Plant 1 is seen
in Figure 5-5. First of all, it is noticed that the petcoke is ignited relatively far from
the burner with the earliest ignition approximately 3 meters from the burner tip.
This is generally one of the issues of using petcoke in the cement kiln and is caused
by the low volatile content of petcoke compared to coal, see Table 5-1 [50,51]. It is
seen that the new burner (Figure 5-5c+d) ignites the petcoke earlier than the old
burner (Figure 5-5a+b). As shown in Figure 5-4 the ignition point can fluctuate
more than 1 meter in a short time, but on average the new burner has an ignition
point between 3 and 4 meters from the burner, while the ignition point of the old
burner is more than 4 meters away, outside the frame of the image. The earlier
ignition is caused by a more effective entrainment of the hot secondary air into the
fuel stream, which allows for a faster heat up of the fuel. This is primarily achieved
by three measures on the new burner. The swirl channel is located inside the coal
channel on the new burner, while it was located on the outside on the old burner.
This has a tendency to push the fuel outwards, where it meets with the hot
secondary air [35]. In addition, the swirl channel vanes are angled more, which gives
58 5 Effect of Alternative Fuels Co-firing on Kiln Flame

a higher tangential velocity to induce swirling motion and increase mixing. Lastly,
the axial air jets of the new burner will allow for an increased secondary air
entrainment compared to an annular air channel [36,46]. It is generally thought to
be beneficial with a high amount of entrainment and early ignition as this gives a
short high temperature flame, where heat can easily be transferred to the clinker.
An early ignition can also help reduce NOx since less oxygen has had time to entrain
into the flame [40,42]. On the other hand, when the swirl channel is located inside
the petcoke channel, the fuel is pushed out into more oxygen rich conditions, which
may increase the NOx [35,40,42].

Figure 5-5: Images during full petcoke firing of old burner (a+b) and new burner
(c+d). Old burner operating with 250 mbarg swirl air pressure and new burner
operating with 190 mbarg swirl air pressure.

5.4.2 Co-Fired Flame

Most of the time Plant 1 operates with a co-firing scenario where 60-80 % of the
energy input is supplied by SRF. The ratio between the fuels is used to control the
burning zone temperature and will typically be influenced by quality variations in
the SRF. I.e. if the moisture content of the SRF increases and the heating value
decreases, it will be necessary to increase the amount of petcoke to keep the
temperature constant. Images of the co-firing case are seen in Figure 5-6. The flame
plume is narrower and lighter in color compared to the full petcoke cases shown in
5.4 Results from Plant 1 59

Figure 5-5 and the ignition point of the flames is moved further away from the
burner. For the old burner, Figure 5-6a+b, the ignition is still outside the image
frame, more than 5 meters from the burner tip. For the new burner, Figure 5-6c+d,
the ignition point is around 5 meters from the burner, which is 2 meters further
away than when only petcoke was fired. The difference is caused by SRF, which
delays the ignition due to a longer heating time, caused by a larger particle size and
high moisture content. The narrower fuel plume is likely caused by a lower amount
of petcoke being used than in Figure 5-5. The petcoke is added through an annular
channel close to the edge of the burner, with the swirl channel located inside. Thus,
it is fairly easy to disperse the petcoke outwards compared to the SRFæ

Figure 5-6: Images during co-firing of petcoke and SRF of old burner (a+b) and new
burner (c+d). The old burner operates with 80 % energy by SRF and 240 mbarg swirl
air pressure. The new burner operates with 70 % energy by SRF and 180 mbarg swirl
air pressure.

In the lower left corner of Figure 5-6c some burning particles can be seen. These are
SRF particles whirled out of the flame due to the swirling flow. When they enter the
hot secondary air, it is possible to ignite the particles faster than in the cold fuel
plume. The amount of SRF that is whirled out of the flame can to some extent be
controlled by the swirl, as shown in Figure 5-7. Here a flame with 180 mbarg swirl
air pressure is compared to one with 100 mbarg swirl air. With the increase in
pressure, the exit flow velocity of the tangential swirl air is increased, which
60 5 Effect of Alternative Fuels Co-firing on Kiln Flame

increases the angular momentum and the swirl intensity [135]. Burning SRF
particles can be seen as small specks of light on the darker background and have
also been highlighted by blue circles. With the high amount of swirl more SRF
particles are seen to burn outside the main flame. Almost twice as many particles
are marked in Figure 5-7a compared to Figure 5-7b. It is unfortunately not certain
whether this can be considered as a representative measure of the amount of SRF
burning outside the flame. Thus, if a specific cement plant is vulnerable to particles
burning in the clinker, i.e. local reducing conditions; it may be beneficial to lower the
swirl. However, this will also have a negative effect on the mixing and the flame
intensity, causing the ignition to take place further inside the kiln. This is also shown
in this image, where the flame intensity in the upper left corner is much higher with
the high swirl. The images in Figure 5-7 are recorded with a different view than
previous images. The camera has been turned to look further down the kiln, and the
tip of the burner is outside the image on the right side. It has not been possible to
estimate the distance in these images, since no reference size is present. In the
previous pictures the burner is used as size reference.

Figure 5-7: Co-firing with the new burner at Plant 1 with 180 (a) and 100 (b) mbarg
swirl air pressure. Burning SRF particles are highlighted by green circles. Distances
are not indicated in the figure since no reference size is present.

5.4.3 SRF Fired Flame

The petcoke that is fired in an annular channel around the SRF blocks for a clear
view of the SRF flow and combustion behavior. At plant 1 it was possible to turn off
the petcoke for a brief period to study how the SRF behaves when injected into the
kiln. In Figure 5-8 these results are presented. A small amount of residual petcoke is
present in the annular channel in the pictures of the old burner (Figure 5-8a-c). The
5.4 Results from Plant 1 61

fuel flow from the old burner is characterized by a low degree of dispersion, where
the fuel follows the initial injection trajectory. The low degree of fuel dispersion is
detrimental to the combustion of the fuel, since it creates a dense cold core in the
flame, which inhibits ignition.

Figure 5-8: Images during 100 % SRF firing for old burner (a+b+c) and new burner
(d+e+f). The old burner operates a swirl air pressure of 240 mbarg. The new burner
with a swirl air pressure of 150 mbarg. The camera has been turned further
downstream in image c, than in the other images.

When the camera view is changed, Figure 5-8c, it can be seen how the particles
continue far inside the kiln without being ignited. It can be assumed that the
particles will eventually land in the clinker bed largely unconverted. The new burner
is better at dispersing the SRF particles. After an initial ~1 m where the SRF is
densely packed, it starts to spread out. Some particles still tend to follow the
injection trajectory, which can be observed in the averaged image of Figure 5-8f.
This is primarily caused by denser lumps of particles, which are difficult to disperse.
One such lump can be tracked in Figure 5-8d and e, which are taken 0.3 seconds
apart. These lumps are most likely caused by the feeding system. In this case the SRF
is fed through a rotary feeder located ~20 meters behind the burner. The feeder
rotates, and its compartments are blown clean by the conveying air, which tends to
generate the fuel lumps, which can be observed at the burner tip. With a more
uniform feeding, it would likely be easier to disperse the particles properly.
62 5 Effect of Alternative Fuels Co-firing on Kiln Flame

The total primary air flow for the two burners is the same at approx. 12,000 m 3/h,
but the old burner operates at a significantly higher swirl pressure, 240 mbarg
compared to 150 mbarg. This should allow for a higher tangential velocity and the
possibility to obtain a higher angular momentum. However, as evident from the
images, the swirl is not utilized to affect the flow of SRF. The design of the new
burner with a higher angle of the swirl vanes and the channel located close to the
SRF seems highly beneficial for the dispersion of SRF in the kiln.

The main advantage of the higher dispersion of particles is a better mixing with the
hot secondary air, which leads to an earlier ignition. Some burning particles can be
observed in the top and bottom left corners of the images in Figure 5-8d and e, when
the fuel leaves the cold fuel core. The high degree of fuel spreading also causes some
of the fuel particles to be whirled out of the flame by the centrifugal forces generated
by the swirl. In the cement industry, this is traditionally viewed as a negative thing,
since it can lead to local reducing conditions in the kiln bed, which promotes brown
clinker, stabilization of belite, and increased sulfur evaporation [106–108,111,119].
However, this does not appear to be a specific issue at the plant, where, generally,
the quality of the clinker is good, with an alite content above 65 wt.%. Brown clinker
and build-ups have not been reported by the plant operators, which suggest that the
plant is insensitive to reducing conditions. In fact, it may be beneficial to purposely
spread the fuel near the kiln outlet, where the oxygen concentration is high and the
clinker nodules are already formed, since the evaporation of sulfur will be limited
by the smaller surface area for evaporation [111]. In addition, the energy contained
in the fuel will be released before the clinker burning zone and contribute to
increasing the temperature here, presumably yielding an increased cement quality.
Experiments carried out by Nørskov [31] have shown that injecting alternative fuels
too far into the kiln limits the possible substitution in the cement kiln, and just
substitutes calciner firing. This will be discussed in further detail later.

Table 5-3 contains measures of the key parameters for the burner performance. The
data are averages from one month of operation. Most importantly, with the new
burner it was possible to achieve an increased use of SRF while the petcoke
consumption was lowered, resulting in a higher energy share of SRF. At the same
time the clinker alite content, which is a measure of the quality, increased. Operation
of the old burner was supported by oxygen enrichment at the burner where 460
5.5 Results from Plant 2 63

Nm3/h of oxygen was used to increase the combustion quality. With the new burner
this was not necessary, and the oxygen was only used on a few test days. The clinker
production was slightly lower with the new burner. In conclusion, the changes in the
design were shown to give an earlier ignition and increased dispersion of SRF,
resulting in an improved performance.

Table 5-3: Comparison of key operating parameters for the old and new kiln burners
at Plant 1. Values are averages for one month of operation. Ignition point data from
Figure 5-5 and Figure 5-6.

SRF Clinker Oxygen

Clinker
Energy Alite enrichment Ignition Point
Production
Share Content at burner
% wt% Nm3/h t/h m
Petcoke: >4 m
Old Burner 62.7 65.3 460 147
Co-firing: >5 m
Petcoke: 3-4 m
New Burner 67.5 67.5 37 144
Co-firing: 5 m
Change in % 7.7 3.5 -91.9 -1.8 -

5.5 Results from Plant 2

The fuel dosing during the test day at Plant 2 and the kiln drive power consumption
are shown in Figure 5-9. The power consumption of rotating the kiln is often used
as a relative measure of the kiln temperature [17]. As the temperature rises, more
material will melt inside the kiln, which will make it more difficult to rotate the kiln,
increasing the kiln torque and power consumption. However, the power will also be
influenced by other factors such as the amount of material in the kiln or the
rotational speed.

Before 13:00, the kiln was fired exclusively with a mix of coal and petcoke. At 13:15
the SRF amount was increased to 1.5 t/h. This level can normally be tolerated at the
plant without issues. Just after 16:00 the SRF firing was increased to 4.5 t/h for 20
minutes, while the coal was reduced to 7.5 t/h. There are generally large fluctuations
in the kiln drive power consumption during the day even though the kiln firing is
kept constant. This is caused by changes in the kiln feed and calciner firing, which
are not shown here. However, as the SRF is increased to 4.5 t/h there is a sudden
large drop in the kiln power, which is clear when observing Figure 5-9b, where the
gradient of the power consumption is shown. A rapid increase is seen as the SRF is
reduced back to 1.5 t/h. The drop in power consumption can indicate a lower
64 5 Effect of Alternative Fuels Co-firing on Kiln Flame

temperature in the kiln, which will lower the cement quality. The kiln operator also
reported an increased amount of dust in the cooler with increased SRF. This may
indicate increased sulfur volatilization caused by local reducing conditions.

Figure 5-9: a) Fuel dosing of coal/petcoke, and SRF and the kiln drive power
consumption (secondary axis) during the test day. b) Gradient of the power
consumption. Data from Plant 2.

5.5.1 Fossil Fuel Fired Flame

The fossil fired flame from Plant 2 is shown in Figure 5-10. The plant fires a mix of
75 % of coal and 25 % petcoke. The flame is seen to ignite very close to the burner,
around 1-2 meters from the tip, compared to the 3-4 meters at Plant 1, which was
shown in Figure 5-5. This is generally due to the high volatility of the coal, compared
to petcoke, see Table 5-1. The flame is seen to ignite in an uneven pattern, which is
clear in Figure 5-10a and c. The burner is designed with a number of axial air jets
surrounding the coal inlet. Four larger gaps are made between some of the jets,
which allows for a place where additional secondary air can entrain into the fuel jet.
This heats up the fuel quickly and ignites it. The three images in Figure 5-10
5.5 Results from Plant 2 65

underline the turbulent nature of the kiln flame, which constantly fluctuates. In
Figure 5-10a, the ignition point is very close to the burner, while it is more removed
in Figure 5-10b.

Figure 5-10: Images of the coal/petcoke fired flame with 10.5 t/h coal/petcoke.

5.5.2 Co-Fired Flame

Example images from the recorded videos with SRF co-firing are seen in Figure 5-11.
The ignition point is between 2-4 meters from the burner, which is 2 meters further
away than in the coal fired flame. The flame also appears less intense, indicating a
lower flame temperature. This is a similar observation to the co-firing case seen at
Plant 1 and is likely caused by a reduction in the coal amount and a longer heating
and burning time of the larger SRF particles, which delays the ignition.

Figure 5-11: Images of co-firing of coal/petcoke with SRF. 7.5 t/h coal/petcoke is used
with 4.5 t/h SRF (30 % SRF energy input).

The flame appears to ignite earlier at the bottom. This may be due to high radiation
from the hot clinker below the flame, which gives an uneven heating of the fuel. It
may also be caused by the flow of the secondary air from the clinker cooler, which
is mainly coming from below [136]. The ignition at the side is no longer as
pronounced as for the coal fired flame, although it is still seen to some extent e.g. in
Figure 5-11c. The longer ignition time and lower flame intensity are expected to
cause a lower temperature near the burner, which also lowers the clinker burning
zone temperature. As seen in Figure 5-9, there is also a slight reduction in the kiln
66 5 Effect of Alternative Fuels Co-firing on Kiln Flame

drive power consumption, when the SRF firing is increased to 4.5 t/h. This is also an
indication of a lowered temperature in the kiln caused by the SRF.

A side by side comparison of the coal/petcoke fired flame and the flame co-fired with
SRF is shown in Figure 5-12. In this figure, the videos have been averaged over 5
seconds. This evens out the turbulent changes, which are observed in Figure 5-10
and Figure 5-11 and makes the comparison more straightforward. It becomes
evident how the ignition point is moved away from the burner, when SRF is fired
and how the intensity of the flame is also lowered, indicating a lower temperature.

Figure 5-12: Comparison of the coal/petcoke fired flame (a) with the SRF co-fired
flame (b). Images are averaged over 5 seconds.

Figure 5-13 shows a view under the flame along the wall of the kiln. This is done by
turning the camera downwards compared to the normal view used in Figure 5-10
and Figure 5-11. This view gives an opportunity to track if particles have dropped
out of the flame. During full fossil fuel firing, there is no fuel to be seen outside the
flame as indicated in Figure 5-13a. When SRF is added to the flame, some particles
fall out of the flame and they burn on the wall or charge as shown in Figure 5-13b+c.
Only a very small number of particles are observed to drop out of the flame. It can
thus be concluded that most of the SRF stays in the flame, until the particles cannot
be tracked any longer due to the limited visibility. However, there are signs that the
particles are not fully converted in the flame and will cause reducing conditions
further downstream in the kiln.
5.5 Results from Plant 2 67

Figure 5-13: View along the kiln wall and charge under the burner for coal-fired case
(a) and co-fired case (b+c). The kiln wall is seen in the left side of the images. The kiln
bed can be seen on the right side of the images having a more orange color than the
wall. Burning particles are bright spots in the images, which have been highlighted
with green circles.

The high amount of SRF firing, 4.5 t/h, was only upheld for 20 minutes before the
amount was reduced to 1.5 t/h. The kiln operator reported an increased dust load
during the testing and there were indications of a lower kiln temperature based on
the kiln drive power consumption (see Figure 5-9). The increased dust load is
presumably caused by local reducing conditions, since the SRF is not fully converted
while in suspension. The reducing conditions promote the decomposition of CaSO4
and the evaporation of SO2, which results in recirculation and accumulation of sulfur
[119]. The sulfur creates a separate melt that is immiscible with the main clinker
phases and has a low viscosity and surface tension [137]. It may thus have an
adverse effect on the clinker nodulization and increase the dust load [25], if too
much sulfate melt is present. Excessive dust in the cooler and kiln inhibit the heat
transfer and cools the burning zone [104]. Comparing the images from Plant 2 with
those of Plant 1 also show a significantly lower visibility in Plant 2, presumably due
to a high dust load. One of the reasons for this is that the kiln manufactures
mineralized clinker, where fluoride and sulfur are used as mineralizers/fluxes to
lower the burning zone temperature requirement [138,139]. Mineralized clinker
may be burnt at temperatures around 200 °C lower than normal clinker [138]. This
may first appear promising for the use of alternative fuels (AF), which tend to burn
at lower temperatures [83]. However, the mineralized clinker is also more sensitive
to process changes [140] and as observed at Plant 2, even relatively small amounts
of AF cannot be handled, due to increased sulfur volatility.
68 5 Effect of Alternative Fuels Co-firing on Kiln Flame

5.6 Results from Plant 3

The fuel dosing used during the test day at Plant 3 is shown in Figure 5-14a and
selected process measurements are shown in Figure 5-14b.

Figure 5-14: a) Fuel dosing to the kiln burner during the test day. b) Temperatures as
measured by pyrometer, thermographic camera, kiln hood thermocouple and the kiln
torque during the test day. Data from Plant 3.

Specifically, the measurements include the kiln torque and a number of temperature
measurements to indicate the kiln burning zone temperature. The plant uses an
infrared pyrometer and thermographic camera to gauge the temperature in the near
burner zone of the kiln as well as a thermocouple measuring the temperature in the
kiln hood. Initially, the SRF dosing has been 6 t/h, but due to a low temperature in
the kiln, at 15:30 the dosing was lowered to 5 t/h and the petcoke dosing increased.
This resulted in a temperature increase, which can be seen by the pyrometer or
camera temperature in Figure 5-14b. The petcoke was deliberately shut off from
16:00 to 16:30, in order to better observe the SRF flight behavior in the kiln. In the
meantime, natural gas was used instead to keep the energy input to the kiln
constant.
5.6 Results from Plant 3 69

5.6.1 Co-Fired Flame

The co-fired flame from Plant 3 is shown in Figure 5-15. It takes approximately 4
meters from the burner tip before the flame is ignited. This is somewhat similar to
what was seen in Plant 1, where petcoke and SRF is also used, while Plant 2 had a
very early ignition due to the volatile coal. The images in the figure also underline
the large changes that can occur in kiln conditions when firing alternative fuels. The
images of Figure 5-15b and c are recorded one hour apart, and the intensity in Figure
5-15c is much higher, indicating a higher burning zone temperature. This is caused
by increasing the petcoke dosing by 1 t/h and reducing the SRF dosing by 1 t/h. The
change is also consistent with the higher temperature measurements, as seen in
Figure 5-14.

Figure 5-15: Flame of petcoke co-fired with SRF and sewage sludge. a+b) 3 t/h
petcoke, 1.5 t/h sewage sludge, 6 t/h SRF (60 % AF energy input), recorded at
14:56. c) 4.3 t/h petcoke, 1.5 t/h sewage sludge, 5 t/h SRF (45 % AF energy input),
recorded at 15:45.

The flame at Plant 3 is sometimes very divergent compared to Plant 1 and 2, which
can cause the flame to impinge on the bed as shown in Figure 5-16a. This should
generally be avoided since it may overheat the refractory and contribute to local
reducing conditions [31]. The design of the burner differs from that used at the other
plants. The other burners have separate channels for axial and swirl air, while the
Plant 3 burner has only one channel, where the swirl level is adjusted by increasing
the tangential angle of the air inlets. If the angle becomes too high, it seems that the
burner lacks axial momentum to stabilize the flame, causing a very diverging flame.
A relatively small amount of SRF particles are whirled out of the flame close to the
burner as also indicated in Figure 5-16a. Further inside the kiln, just around the
point where the petcoke ignites approximately 6 meters from the burner, some
burning particles can be observed in the kiln bed, see Figure 5-16b. In the videos,
70 5 Effect of Alternative Fuels Co-firing on Kiln Flame

several burning particles can be seen around this point, which indicates that it is
where the SRF particles begin to drop out of the flame.

Figure 5-16: View of the co-fired flame from the side (a) and under the burner (b).
Flames in bed are located by green circles. Fuels: 4.3 t/h petcoke, 5 t/h SRF, 1.5 t/h
sewage sludge (45 % AF energy input).

5.6.2 Flame with Natural Gas

The SRF is fired through a pipe at the center of the burner, which is surrounded by
an annular channel for the injection of petcoke. Similarly to Plant 1 the visibility of
the SRF flow is obstructed by the petcoke. In an attempt to understand the SRF flow
better, the petcoke was turned off for a brief amount of time. To keep the heat input
to the kiln, gas was supplied instead, as shown in Figure 5-14 at around 16:15. While
the gas flow is increased, from 0 to approx. 500 Nm3/h, the ignition point moves
closer to the burner tip from approx. 6 meters to 2 meters, see Figure 5-17. The
combustion also becomes more intense, with the image becoming more
oversaturated. It would seem that this could be quite a viable method to increase
the ignitability of the co-fired flame and obtain a high temperature region close to
the burner. However, the price of natural gas in most countries will probably make
this kind of operation unfeasible. Per unit of energy, gas may be more than four times
as expensive as petcoke [39]. The pyrometer measuring the flame temperature is
seen to increase from around 1000 °C to 1050 °C, with a peak at 1100 °C, around the
time where the petcoke dosing is lowered. The kiln hood temperature increases in
the same period, which also indicates an increased temperature near the kiln outlet.
The temperature measured by the kiln camera installed at the plant, however, has a
lower reading, while the gas is being fired. The kiln torque is also seen to decrease,
which may indicate less heating further inside the kiln. Gas flames normally produce
5.6 Results from Plant 3 71

small amounts of soot compared to coal flames, which results in a lower flame
emissivity [4]. The heat transfer from flame to clinker bed can thus be inhibited. This
could explain the lower clinker temperatures measured by the camera and the lower
kiln torque.

Figure 5-17: Different flame shapes while the gas flow is increased over 200 seconds
from 0 (a+d) to 500 (c+f) Nm3/h. Petcoke dosing is constant at 4.3 t/h. Top row
shows single frames and bottom row are images averaged over 5 seconds.

Some images of the SRF and gas fired flame without petcoke are shown in Figure
5-18. The visibility of the SRF changes from frame to frame as shown in the sequence
of images in Figure 5-18, which is mainly due to the natural variations in flame
ignition. In Figure 5-18b it is seen that the SRF is quite hard to ignite, and it passes
at least 6 meters from the burner tip without being ignited, and then disappears in
the gas flame. It is very likely to continue further inside before being properly ignited
and the conversion in the flame may be low.

Figure 5-18: SRF and gas fired flame. 5 t/h SRF, 1.5 t/h sewage sludge, and 4500
Nm3/h natural gas (45 % AF energy input).
72 5 Effect of Alternative Fuels Co-firing on Kiln Flame

The SRF is not packed as densely as seen in Plant 1 (see Figure 5-8) when it enters,
which is likely due to a lower feeding rate. The dispersion of the SRF appears to be
better than for the old burner in Plant 1, and similar to what is achieved for the new
burner at Plant 1. A significant amount of small flames can be seen in the bed, in the
lower part of the images of Figure 5-18. This indicates that a large fraction of the
SRF may burn in contact with the clinker rather than in suspension.

5.7 Limiting Factors for Alternative Fuels Firing

It is highly beneficial to understand the limiting factors for alternative fuel (AF)
firing in order to eventually increase the amount of AF firing. Thus, some of the
issues encountered at the three plants when firing AF will be discussed in further
details here.

Plant 1 fires the highest amount of alternative fuels in the kiln of the three plants
studied here. The plant produces clinker with an alite content above 65 %, which is
a typical level for Portland cement [11]. Issues with brown clinker or build-ups in
the kiln or preheater have not been observed during the measurements with SRF.
The plant has a chlorine by-pass and several air blasters are installed in the
preheater tower to combat build-ups. The main limitation is the lower combustion
temperatures obtained during SRF co-firing. If the SRF firing gets too high, the kiln
temperature is lowered. Thus, petcoke is needed to create a high temperature zone
to obtain an adequate clinker quality. The petcoke is also used to adjust the burning
zone temperature when the SRF heating value or moisture content changes. The
plant has previously used oxygen enrichment in the kiln to stabilize the operation at
high SRF firing and lower the petcoke consumption. The installation of the new
burner has been beneficial for Plant 1. It has been possible to increase the alite
content of the clinker, which resulted in increased compressive strength of cement
mortar. Furthermore, the substitution with SRF was increased and oxygen
enrichment was no longer used. The main difference between the designs of the two
burners has been discussed in chapter 5.4, and is related to the mixing intensity
achieved by the burners and the ability of the new burner to spread the SRF in the
combustion zone.

Plant 2 fires the lowest amount of AF and has a low tolerance. This appears to be
mainly caused by the manufacture of mineralized clinker, which is more sensitive to
5.7 Limiting Factors for Alternative Fuels Firing 73

reducing conditions since the sulfur loading is higher than for ordinary clinker. The
limit of SRF firing appears to be around 2 t/h, while the plant can tolerate up to 3
t/h of the granulated tire. In a study by Nielsen et al. [53], tire granulate was found
to be able to release more sulfur from cement raw materials than plastic and wood,
which are the main constituents of SRF. Thus, the reason that a higher amount of
granulated tire can be tolerated, is most likely related to the smaller particle size.
This results in a faster conversion of the fuel, resulting in less fuel ending up in the
kiln bed to induce reducing conditions.

Plant 3 uses an intermediate amount of alternative fuels in the kiln. The produced
clinker has an alite content above 60 % (calculated by Bogue formulas [13,24]).
Plant 3 sometimes encounters problems with reducing conditions, when firing too
much alternative fuels. This is seen as brown cores in the cement clinker and some
deposit build-ups in the calciner and cement kiln. The plant recently installed a by-
pass to reduce the volatile circulation, which is expected to alleviate some of the
problems with build-ups, and may allow for a further increase in the use of
alternative fuels.

Figure 5-19: Relationship between the energy from alternative fuels firing at the main
burner (MB) and the energy input in the kiln (red dots) and the calciner (black x) for
Plant 1 (a) and Plant 3 (b). Data are based on hourly averages for one month of
operation. ρ is Pearson correlation coefficient.

An interesting difference between Plant 1 and Plant 3 is shown in Figure 5-19. The
figure shows the relationship between the fraction of alternative fuels firing at the
main burner and the energy input at the main burner and calciner (Hotdisc
included) for Plant 1 (Figure 5-19a) and for Plant 3 (Figure 5-19b). The data plotted
in the figure are based on hourly averages from one month of operation giving a
74 5 Effect of Alternative Fuels Co-firing on Kiln Flame

large degree of data scatter. The Pearson correlation coefficient [141] has been
calculated as a measure for the relationship between the variables.

Normally, around 60 % of the total fuel used in the cement process will be fired in
the calciner and the remainder in the kiln. This is because the calcination of
limestone is highly endothermic requiring 1800 kJ/kg [142]. This is also the balance
at both plants when no AF is fired in the kiln. At Plant 1, the heat input at the burner
and calciner is rather independent of the amount of AF fired at the main burner.
However, at Plant 3, the energy input at the kiln is increased as the amount of AF in
the kiln is increased, while the firing in the calciner is reduced. This shifts the
balance between kiln and calciner firing and when high amounts of AF are utilized
around 60 % of the energy is fired in the kiln. This indicates that the fuel energy from
the AF is not released quickly enough to contribute to increasing the clinker burning
zone temperature, and rather substitutes firing in the calciner. At Plant 1, the
correlation is small, indicating that the fuel fired in the main burner is actually
utilized in the kiln. Thus, there are indications that AF burns closer to the kiln exit at
Plant 1 than at Plant 3. Nørskov [31] made some experiments with different
injection velocities of AF and found that an injection velocity of 30 m/s was to be
preferred over a velocity of 50 m/s. The high injection velocity causes the fuel to be
injected too far into the kiln, and the energy is not released in a proper location to
contribute in rising the clinker temperature. It is possible that a lower injection
velocity of AF at plant 3 could ensure that the SRF is not injected too far into the kiln,
which allows for the energy to be released in the clinker burning zone.

Another large difference between the plants is their tolerance to reducing

conditions. At Plant 1 neither brown clinker or deposits were observed during
testing, at Plant 2 the dust load in the kiln is increased when utilizing high amounts
of AF due to sulfur evaporation, and at Plant 3 brown clinker and deposits are
sometimes encountered. The difference may largely be caused by the sulfur
evaporation in the kiln. The sulfur evaporation may be limited if there is a high
amount of alkalis in the clinker, since alkali sulfates are less volatile than calcium
sulfate. One measure is the sulfur modulus calculated as [23]:

𝒀𝑺𝑶𝟑
𝑴𝑺𝑶𝟑 = 𝟖𝟎 E 5.1
𝒀𝑲𝟐 𝑶 𝒀𝑵𝒂𝟐 𝑶
𝟗𝟒 + 𝟔𝟐
5.7 Limiting Factors for Alternative Fuels Firing 75

The sulfur modulus calculated for the clinker of the three plants is on average 0.64,
1.51, and 0.84. A value around 1 is adequate to ensure that there is sufficient alkali
to combine with the sulfur [23]. It can thus be seen that Plant 2 operates at a high
sulfur modulus, due to the high input of sulfur in the mineralized clinker, but this
also means that it is more susceptible to sulfur evaporation caused by reducing
conditions. The sulfur modulus for Plant 1 and Plant 3 are low enough that most of
the sulfur should be able to combine with alkalis. However, comparing the fuels of
Plant 1 and 3 (Table 5-1), it is seen that the sulfur content of the petcoke used in
Plant 3 is higher than that in Plant 1 and the chlorine content of the SRF is higher.
The plant will thus have a higher input of sulfur and chlorine through the fuel, which
may set a limit for the utilization of alternative fuels. In addition, Plant 3 did not have
a by-pass to lower the amount of recirculating species. Thus, Plant 3 is likely to be
more vulnerable to reducing conditions causing sulfur evaporation and deposits
than Plant 1.

Brown core clinker is also sometimes encountered at Plant 3. The brown color is
caused by a reduction of Fe3+ to Fe2+, which may substitute MgO in the ferrite phase
(C4AF), which otherwise gives cement its dark grey color. The reduced iron can also
affect the formation of alite, lowering the cement quality [106–108]. Alternative
fuels are burned in contact with the bed in both Plant 1 and Plant 3, but Plant 1
tolerates it better, with no formation of brown clinker. It may be related to where
the AF mainly burns, which was discussed above and illustrated in Figure 5-19. If
the AF burns too far inside the kiln it may drop into a bed of un-nodulized clinker,
with a large surface area which may be more prone to iron reduction [111]. If the AF
mainly burns close to the burner where the nodules have already formed, the
surface area for evaporation of sulfur and iron reduction is significantly smaller and
the oxygen concentration will be higher. The brown clinker cores may also be
related to the rate of clinker cooling. Locher [107] showed that the adverse effects
of burning under reducing conditions could be limited by a rapid cooling from 1250
°C in air. Perhaps high clinker porosity will be beneficial to counter brown cores,
since it will help oxygen to diffuse in and reoxidize the clinker during cooling.

In summary, all three plants are limited by the conversion rate of the alternative
fuels. For Plant 1 the main limitation is that the AF does not burn quickly enough to
obtain sufficiently high temperatures. Thus, some petcoke is needed to maintain a
76 5 Effect of Alternative Fuels Co-firing on Kiln Flame

high temperature and proper heat transfer to the clinker. At Plant 2 and Plant 3 the
main limitation is that the fuel is not converted quickly enough before ending in the
bed, eventually leading to localized reducing conditions in the kiln. It would be
interesting to study if the utilization could be improved by drying or milling of SRF
to obtain faster conversion of the fuel. Excess heat typically available in the off-
gasses could be used to dry alternative fuels, but milling of SRF is difficult due to the
soft paper and plastic fractions and impurities that may damage the mill [99].

5.8 Conclusions
A specially developed camera setup has been used to study the kiln flames at 3
different cement plants. The probe was designed so it could be inserted directly in
the cement kiln hood where the temperature is around 1000 °C and the dust load is
high. This allowed for a detailed study of the influence of alternative fuels on the
cement kiln flame.

An overview of the flames studied at the three different cement plants is given in
Table 5-4. Adding alternative fuel to the flame had at all three cement plants a
negative impact on the flame. At Plant 1 the ignition point was between 3-4 meters
from the burner tip when petcoke was fired alone, and when SRF was added to the
flame the ignition point was between 5-6 meters from the burner tip. At Plant 3 the
ignition point was at a similar distance while co-firing petcoke and SRF. The flame
at Plant 2 ignited within 1 meter. A mix of coal and petcoke was used, which ignites
more readily than the petcoke at Plant 1 and 3 due to a higher volatile content of the
coal. At Plant 2 the ignition point was also shifted approx. 2 meters when SRF was
added to the flame. At all three plants, the flame intensity was also lowered when
using AF, which indicates a lower combustion temperature. This is mainly due to the
high moisture content and large particle size of alternative fuels compared to
conventional fuels, which results in a lower conversion rate of the fuel. At Plant 2
and Plant 3 measurements of the kiln drive power consumption also suggested that
the temperature in the kiln was decreased when co-firing AF. The lower
temperatures in the kiln may negatively affect the clinker quality.
5.8 Conclusions 77

Table 5-4: Overview of the impact of alternative fuels on the flame ignition point at
the three cement plants. C: Coal, PC: Petcoke, SRF: Solid Recovered Fuel, SS: Sewage
Sludge.

Plant 1 Plant 2 Plant 3

Old Burner New Burner Old Burner
Co-fired Co-fired
Flame Fossil Co-fired Fossil Co-fired Fossil Co-fired
1 2
5-11+ 5-12+
Figure 5-5a+b 5-6a+b 5-5c+d 5-6c+d 5-16c 5-18c+f
5-13a 5-13b
PC: 4.3
PC: 2.1 PC: 1.5 PC: 4.3 SRF: 5
- -
Fuel (t/h) - - C: 8.4 C: 6 SRF: 5 SS: 1.5
SRF: 0 SRF: 4.5 SS: 1.5 Gas: 500
Nm3
AF Energy
0 80 0 70 0 30 45 45
Share (%)
Ignition
>4 >5 3-4 ~5 1-2 2-4 ~4 ~2
Point (m)

At all three plants, it was observed how some of the SRF drops out of the flame and
burns in contact with the cement clinker. At Plant 2 and 3 this leads to problems
with sulfur evaporation or brown cored clinker, while Plant 1 appears to be more
robust to local reducing conditions. Why this is the case is not fully understood, but
Plant 2 is sensitive to reducing conditions due to manufacturing mineralized clinker
and has a high sulfur modulus, which makes sulfur evaporation more probable.
Between Plant 1 and Plant 3 there may be a difference in where the SRF mainly
burns or there could be a difference in the cooling which can prevent reduced
clinker.

The burner design does have a significant influence on the flame. The burner design
at Plant 2 could create ignition sources, by allowing secondary air to be entrained
into the fuel stream in a few locations. A clear difference in the two burner designs
tested at Plant 1 was also observed. The change from an annular axial air channel to
axial air jets benefited the ignition when using petcoke. The design of the swirl
channel influences the flame swirl level, increasing the dispersion of SRF particles
in the flame. The old burner at Plant 1 had very little dispersion of the SRF. This
creates a cold core of SRF in the flame, which will be difficult to ignite and the SRF
will continue far into the kiln and eventually land in the clinker bed, likely largely
unconverted. Alternatively, the SRF can be spread as much as possible, which occurs
with the new burner at Plant 1. This will evidently result in more SRF dropping out
78 5 Effect of Alternative Fuels Co-firing on Kiln Flame

of the flame and ending up in the clinker bed close to the burner. On the other hand,
the energy in the fuel will be released earlier, contributing better to maintain a high
temperature in the clinker burning zone. Based on the results presented here, the
second option with a high degree of fuel dispersion appears beneficial. It may also
reduce the problems with localized reducing conditions induced by fuel in the bed,
since the fuel will be in contact with nodulized clinker, which has a relatively small
surface area, in the part of the kiln where the oxygen concentration is highest. By
changing the burner at Plant 1, it was possible to increase the alite content of the
clinker while increasing the substitution of SRF.
6 EFFECT OF BURNER SETTINGS
ON THE CEMENT KILN FLAME
AND CLINKER QUALITY

It is fairly well known how burner settings, such as axial momentum and swirl, can
influence the solid fossil fuel flame on factors like ignition, temperature, and NOX
formation. For a co-fired flame this knowledge is more limited. It is generally
understood that the burner should allow a fast mixing of secondary air and
alternative fuel in order to promote ignition [30,46]. Detailed studies of the burner
settings have been made in the Cemflame 3 experiments (see chapter 3.5.1 and
9.1.1), while one study [143] (see chapter 8.1.2) used CFD simulations to study
different operating conditions, amongst them swirl level, when co-firing in a cement
kiln.

A more detailed understanding of the influence of burner settings on the co-fired

kiln flame is thus beneficial. To this end, further experiments were made using the
Jetflex burner installed at Plant 1. The aim was to understand the impact that burner
settings, e.g. swirl, has on the co-fired flame, and investigate which approach is best
suited to optimize the flame characteristics and clinker quality. The kiln camera was
used to study the near burner changes in the flame pattern using different burner
settings, while a statistical analysis (Partial Least Squares Regression) was
80 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

performed on collected operating data to relate the burner settings to the clinker
quality.

6.1 The Jetflex Burner Settings

The FLSmidth Jetflex burner has previously been described (see chapter 2.8). The
Jetflex burner has a number of different possibilities to influence the flame shape.
The amount and velocity of axial air, swirl air, and central air (for cooling) can be
adjusted by valves controlling the flow and thus the air pressure and velocity to each
channel. The axial air provides axial momentum to the flame, while the swirl air is
introduced tangentially giving the flow a swirling motion. This is a common way of
controlling the flame shape [30,35,42,144]. The flow of central air is small and used
to cool the burner front plate.

The 20 axial air nozzles of the burner can be turned individually 360°, which can
further impact the flame. The axial air nozzle openings are flat and have a slight
offset of 10°. The nozzle position is defined as 0° when the offset is pointing towards
the burner center and 180° when the offset is pointing away from the burner center.

Figure 6-1: The Jetflex axial air nozzles can be turned allowing different nozzle
configurations such as: a) All nozzles pointing inwards 0°. b) All nozzles turned 30°
[145]. The blue lines indicate the entrainment of secondary air into the burner jet.

Nozzle configurations of 0° and 30° are shown in Figure 6-1a and b. When the
nozzles are at 0° the configuration almost resembles an annular channel for the axial
air, and the entrainment of secondary air is low. When the nozzles are at 30°, the
annular configuration is broken up, which allows for extra space between the
nozzles for entrainment of secondary air [46], as indicated by the blue lines in Figure
6-1.
6.2 Calculation of Burner Momentum and Swirl 81

Furthermore, it is possible to switch the fossil fuel from being fired through the
conventional annular channel to be mixed with the alternative fuel and be fired
through the large central pipe. This allows for a reduced amount of cold transport
air, benefiting the specific heat consumption in the kiln.

Lastly, it is possible to retract the center of the burner, including the swirl channel,
as shown in Figure 6-2. When the swirl air reaches the tip of the burner it can
suddenly expand, which creates a stronger recirculation zone in front of the burner.
In cold testing of the burner it was found that this could help disperse the alternative
fuel [145].

Figure 6-2: Burner in normal operation (a) and with retracted center (b) [145].

6.2 Calculation of Burner Momentum and Swirl

Valves at the burner air channels are used to control the air flow in the axial and
swirl channels. When both valves are fully open the pressure in both the swirl and
axial channels are around 180 mbarg. Closing the swirl valve causes the flow and
pressure of swirl air to decrease, while the axial air flow and pressure is increased,
since the primary air fan is giving a constant air flow to the burner. The exit
velocities at the swirl and axial nozzles can, for isentropic flow, be calculated by
[146]:

𝜸−𝟏
𝟐𝜸 𝑹𝒈𝒂𝒔 𝒑𝒆𝒙 𝟏
𝒗𝒆𝒙 =√ 𝑻𝒃𝒖𝒓𝒏𝒆𝒓 (𝟏 − ) E 6.1
𝜸 − 𝟏 𝑴𝒂𝒊𝒓 𝒑𝒃𝒖𝒓𝒏𝒆𝒓

In the equation vex is the exit gas velocity from the nozzle, Rgas is the gas constant,
Mair is the molar mass of air, Tburner is the temperature of air in the burner, pex is the
82 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

pressure at the nozzle exit (ambient pressure), pburner is the pressure in the burner,
and γ is the ratio of specific heats.

With additional data on the burner geometry and nozzle areas it is possible to
calculate the mass flow of swirl and axial air, and thus the burner momentum:

𝑰 = 𝒎̇𝒆𝒙 𝒖𝒆𝒙 = 𝝆𝒆𝒙 𝑨𝒆𝒙 𝒖𝟐𝒆𝒙 E 6.2

Here I is the burner momentum, ṁex is the air mass flow, ρex is the air density at the
nozzle exit, Aex is the nozzle area, and uex the axial gas velocity at the nozzle.

The swirl number in its common definition proposed by Beér and Chigier [135,147],
is difficult to compute without detailed flow measurements, which were not
performed here. Thus, a simplified swirl number is used instead. The number is
calculated based on maximum axial and tangential velocities, uex and wex,
respectively, measured at the nozzle exit [148]:
𝒘𝒆𝒙
𝟐𝒖𝒆𝒙
𝑺𝒂𝒑𝒑 = 𝒘 E 6.3
𝟏 − 𝟐𝒖𝒆𝒙
𝒆𝒙

The relevant operating pressures for the performed tests and corresponding nozzle
velocities, axial momentums and the approximated swirl number are given in Table
6-1.

Table 6-1: Overview of burner pressures and corresponding velocities, axial

momentum and the approximated swirl number.

Swirl Axial Swirl air Axial air Axial Approximated

Pressure Pressure velocity velocity Momentum Swirl Number
mbarg mbarg m/s m/s N/MW -
10 240 42 234 9.6 0.07
30 235 72 232 9.7 0.12
50 230 93 229 9.8 0.17
80 225 116 227 10.0 0.22
100 220 130 225 10.0 0.26
150 200 156 216 9.9 0.34
180 190 170 211 9.8 0.40

6.3 Flame Measurements

The data presented in this chapter are comprised of video recordings with the kiln
camera obtained at Plant 1 (described in chapter 0). The video footage has been
acquired over several visits to the test cement plant stretching over one year, in
6.3 Flame Measurements 83

which several hours of video footage have been recorded. From this data a few
representative images have been selected and presented here.

In the following chapters images shown in color are single frames from the recorded
videos. Images in gray scale are averages over 5 seconds (around 150 frames).
These are presented to reduce the frame to frame variation. Two different
approaches are used to further highlight the difference between burner settings, as
will be explained later. The following chapters will first discuss the influence of swirl
and nozzle configuration on the petcoke flame, and afterwards the attention will be
turned to the flame co-fired with SRF.

6.3.1 Petcoke Flame

6.3.1.1 Effect of Swirl Air
The effect of changing the swirl air pressure on the petcoke flame is shown in Figure
6-3.

Figure 6-3: Effect of swirl air pressure on the petcoke flame with axial air nozzles at
30°. Swirl pressure of a) 30 mbarg, b) 80 mbarg, c) 180 mbarg.

Petcoke flames with swirl air pressures of 30, 80, and 180 mbarg are shown in
Figure 6-3, with the axial air nozzles being at 30°. The swirl increases the width of
the flame plume, which causes an increased mixing of fuel with the hot secondary
air surrounding the flame plume. This causes an earlier ignition which is moved
from around 5 meters when the swirl pressure is 30 mbarg (Figure 6-3a) to between
3 and 4 meters when the swirl pressure is 180 mbarg (Figure 6-3c).
84 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

In order to further highlight the difference between the burner settings Figure 6-4
is presented. The recorded color-video frames are converted to a series of 8-bit
grayscale images, having an intensity scale from 0 to 255. An intensity value of 0 is
black, and a value of 255 is white. In each image the intensity profiles along the
horizontal centerline (green lines in Figure 6-3) and perpendicular to the centerline
1 m in front of the burner are determined. These are then plotted over time to yield
Figure 6-4, which indicates how the image intensity changes over time (x-axis) and
distance (y-axis). The color on the plots indicates the image intensity, with dark blue
being low image intensity, indicating cold areas such as the unignited fuel plume,
and yellow being high image intensity, indicating hot areas such as the ignited fuel.
A simplified example of the approach is presented in Appendix A, for additional
clarity.

Figure 6-4: Intensity profiles over time when swirl pressure is changed from 180 to
30 mbarg. a) along the horizontal centerline, and b) across the image 1 m in front of
the burner.

Figure 6-4 shows how the flame is impacted when the swirl pressure is lowered
from 180 mbarg to 30 mbarg, at 5 seconds. In comparison the images in Figure 6-3
shows how the flame appears before and after similar changes. The low intensity
region in the center of the Figure 6-4b indicates the dark petcoke plume seen in
Figure 6-3. The swirl is changed by closing a valve at the burner, which gives an
almost immediate effect on the flame. The reduction of the swirl pressure is seen to
give a contraction of the flame plume in Figure 6-4b between 5-15 seconds. The
swirl also impacts the ignition point of the flame as shown in Figure 6-4a. Here it can
be seen that the bright yellow region, which represents a high image intensity and
fuel ignition, becomes smaller after 20 seconds. Thus, the ignition point is moved
further away from the burner. The figure also indicates the constantly changing
6.3 Flame Measurements 85

nature of the flame. In Figure 6-4a, it can be seen that the flame burns back and
ignites earlier at around 40, 60, and 80 seconds. This could possibly be linked to
some pulses in the fuel flow or changes in the air flow around the burner. For
instance, the kiln rotates at approximately 3 RPM, which coincides with the
frequency of the flame pattern. In addition, a dark band is seen in the region 90-100
s, which is caused by excessive dust from the clinker cooler obstructing the flame
view. The dust is likely caused by the emptying of air blasters that are installed in
the cooler to prevent buildups, so-called ‘snowman’ formation.

6.3.1.2 Effect of Nozzle Configuration

The effect of nozzle configuration on the petcoke flame is shown in Figure 6-5 and
Figure 6-6. Nozzle configurations of 0, 30 and 180° are shown for a highly swirled
flame with swirl pressure of 180 mbarg. With the nozzles at 0° the flame is ignited
between 4 and 5 meters from the burner. When the nozzles are turned to 30° or
180°, the ignition occurs between 3 and 4 m. The flame plume before ignition has a
similar width when the nozzles are at 0 and 30°, but is widened when the nozzles
are turned to 180°. In Figure 6-6 it is seen how the image intensity changes when
the nozzles are turned from 0 to 180°. The 20 axial air nozzles are rotated one at a
time and takes longer than when the swirl is changed. The first nozzle is changed at
20 seconds, while the last is changed at 180 seconds. The flame plume is widened
(Figure 6-6b), especially between 80-120 seconds, and the ignition is moved closer
to the burner (Figure 6-6a).

It is shown how the nozzles can help influence the mixing of hot secondary air and
fuel. When the nozzles are in 0°, the configuration is relatively closed, and the jets
mimic an annular channel where the entrainment of hot secondary air into the fuel
stream is slow, which results in a later ignition of the petcoke. Turning the nozzles
outwards to 180°, gives extra space for the fuel stream to expand and widens the
fuel plume. This slows the fuel stream and gives an increased mixing of hot
secondary air, which results in an earlier ignition. It appears that a similar ignition
length can be achieved by just turning the nozzles to 30°. In this case the nozzle
configuration is more open than in 0°, see Figure 6-1, and allows increased amounts
of secondary air to be entrained into the fuel jet, moving the ignition point closer to
the burner, without expanding the fuel plume as much as when the nozzles are at
180°.
86 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

Figure 6-5: Effect of axial air nozzle configuration on the petcoke flame with swirl
pressure of 180 mbarg. Axial air nozzle position of a) 0°, b) 30°, c) 180°.

Figure 6-6: Intensity profiles over time when nozzles are changed from 0 to 180°. a)
along the horizontal centerline, and b) across the image 1 m in front of the burner.

The impact of swirl and nozzle configuration is compared in Figure 6-7. The graphs
compare the image intensity of the averaged frames in Figure 6-3 and Figure 6-5.
Figure 6-7a shows the image intensity along the horizontal centerline and thus
indicates the point of ignition, as the distance where the intensity begins to increase
sharply and approaches a value around 200. With the nozzles at 30° this is mainly
impacted by increasing the swirl to 180 mbarg. With the swirl at 180 mbarg, the
ignition point is similar with nozzles at 30 and 180°. Turning the nozzles to 0° delays
the ignition from approximately 4.5 m to 5.0 m.

Figure 6-7b shows the intensity across the flame plume 1 m in front of the burner
and indicates the width of the flame plume. The unignited flame plume is dark,
6.3 Flame Measurements 87

resulting in low image intensity. Thus, the flame plume width can be compared by
comparing the width of the intensity dip observed in Figure 6-7b. The flame plume
is widest at ~55 cm, when the swirl is high at 180 mbarg and the nozzles are turned
outwards to 180°. It is narrowest with ~30 cm, when nozzles are at 30° and swirl at
30 mbarg.

Figure 6-7: Intensity profiles for the petcoke flame images with different swirl (S in
mbarg) and nozzle configurations (N in °) a) along the horizontal centerline, and b)
across the image 1 m in front of the burner.

Based on the performed tests it is concluded that the petcoke flame can be adjusted
by both changing the swirl and the axial air nozzle configuration of the Jetflex
burner. It is possible to both influence how fast the fuel expands and the ignition
point. The optimal settings will likely wary between plants. Normally, it is of interest
to have a high temperature flame, which creates effective heat transfer to the clinker
bed. However, a too high temperature may harm the kiln refractory or lead to
excessive NOx [30,31,35,36]. Thus, for a hard to ignite fossil fuel such as petcoke it
may be necessary to have a high swirl and nozzle offset to facilitate ignition. With a
more volatile and easy to ignite coal, lower swirl and a low nozzle offset can be used.
To reduce NOx emissions from the kiln it is important to limit the oxygen availability
in the high temperature combustion zone. This can possibly be achieved by having
the nozzles at 0° where mixing of fuel and secondary air is limited.

6.3.2 Co-fired Flame with Petcoke in Annular Channel

In this chapter the effects of swirl air and nozzle configuration on the co-fired flame
with petcoke in the annular channel are investigated.
88 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

6.3.2.1 Effect of Swirl Air

An example of how the swirl influences the co-fired flame is seen in Figure 6-8 and
Figure 6-9. Here the axial air nozzles are at 60° and the swirl level is increased from
50 to 150 mbarg. The plume is widened by the increased swirl level. However, the
ignition point does not appear to be changed significantly. Image intensity is
increased slightly in the lower left corner of the image Figure 6-8b.

Figure 6-8: Effect of swirl pressure on the co-fired flame with axial air nozzles at 60°.
Swirl pressure of a) 50 mbarg, b) 150 mbarg. SRF contributes 60 % energy.

Figure 6-9 shows how the image intensities change over time as the swirl air
pressure is increased from 50 to 150 mbarg is seen. The change occurs between 10
and 15 s and widens the flame plume. The ignition point is not influenced, likely
because it is further than 4 m from the burner and thus not shown in the images
above.
6.3 Flame Measurements 89

Figure 6-9: Intensity profiles over time when swirl pressure is changed from 50 to
150 mbarg. a) along the horizontal centerline, and b) across the image 1 m in front of
the burner.

Figure 6-10: Effect of swirl pressure on the co-fired flame with axial air nozzles at 30°
using alternative camera view. Swirl pressure of a) 100 mbarg, b) 180 mbarg. SRF
contributes 70 % energy.

The images presented in Figure 6-10 shows a different view where the camera has
been turned more downstream in the kiln to better observe the region where the
flame ignites. The burner is used as a size reference to determine the length of the
90 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

flame in previous images. With the different view, the burner tip is no longer visible,
and it is not possible to determine distances in the image. In this case a swirl level of
100 and 180 mbarg with nozzles at 30° is compared. An increased high intensity
region is observed, indicating that the swirl helps boost the combustion, resulting in
increased temperatures closer to the burner. It can also be observed how additional
SRF is falling out of the main fuel plume, when the swirl is increased. The SRF
particles are seen as orange specks, indicating that they ignite as they leave the
petcoke plume and enter the hot secondary air, mainly below the petcoke plume.

6.3.2.2 Effect of Nozzle Configuration

Figure 6-11 and Figure 6-12 show an example of the effect of the nozzle
configuration on the co-fired flame.

Figure 6-11: Effect of axial air nozzle configuration on the co-fired flame with swirl
pressure of 150 mbarg. Axial air nozzle position of a) 180°, b) 60°. SRF contributes 60
% energy.

The effect of having the nozzles at 180° instead of 60° is to expand the fuel jet faster.
This is most notably seen close to the burner. When the nozzles are at 60° an
expansion of the fuel plume is seen from around 0.8 m in front of the burner. With
6.3 Flame Measurements 91

the nozzles in 180°, the fuel plume is widened almost at the burner front and keeps
a more uniform width. The width at a distance of 3 m appears similar. The ignition
point benefits slightly from the faster expansion obtained by having the nozzles at
180°.

In Figure 6-12 the effect of changing the nozzles from 180 to 60° is illustrated. The
change starts at around 30 s and continues until 270 s. Since the nozzles are changed
one at a time a brief period with an asymmetric nozzle configuration is obtained.
This initially makes the flame plume expand downwards in the period 30-60 s, as
observed in Figure 6-12b. As the bottom nozzles are then adjusted, the bottom of
the flame plume contracts. As the remaining top nozzles are turned from 180 to 60°
a contraction of the top of the flame plume is also observed between 200 and 270 s.
The narrow flame plume also affects the ignition point as shown in Figure 6-12a,
which moves further away from the burner.

Figure 6-12: Intensity profiles over time when nozzles are changed from 180 to 60°.
a) along the horizontal centerline, and b) across the image 1 m in front of the burner.

The intensity profiles in Figure 6-13 summarize and compare the impact of swirl
and nozzle configuration for the co-fired flame. The ignition point (Figure 6-13a) is
not influenced much by changing the swirl between 50 and 150 mbarg. Turning the
nozzles to 180° results in an ignition point around 4 m from the burner tip. The flame
plume width (Figure 6-13b) is increased by higher swirl and nozzle configuration
angle. The plume width is ~30 cm with nozzles at 60° and swirl at 50 mbarg, while
it is ~50 cm with nozzles at 180° and swirl at 150 mbarg. There is a difference in the
plume width and ignition point between the repeat cases with nozzles at 60° and
swirl 150 mbarg (red and green graphs). The images shown in Figure 6-8, used to
create the blue and red curves, were recorded one hour earlier than those in Figure
92 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

6-11, used to create the green and black curves. This could result is some differences
in the kiln state. For instance, a higher temperature in the kiln hood was measured
during the latter set of videos. This could impact ignition, and explain why the green
graph of Figure 6-13a obtains a higher intensity than the red graph, which are made
with similar burner settings. Thus, the state of the kiln can influence the
repeatability of experiments. However, many experiments are made within short
time intervals, as shown in e.g. Figure 6-12, where the state of kiln does not have
time to change. However, comparisons of operation on different days, should be
analyzed with care.

Figure 6-13: Intensity profiles for the co-fired flame images with different swirl (S in
mbarg) and nozzle configurations (N in °) a) along the horizontal centerline, and b)
across the image 1 m in front of the burner.

In comparison to the petcoke flame the co-fired flame is more difficult to ignite when
the nozzles are at 60° and the swirl air pressure is either 50 or 150 mbarg. Changing
from low to high swirl does not have the same impact on ignition point as for the
petcoke flame. Changing the nozzles to 180° has a larger effect in this case, which
gives an ignition point at around 4 m that is similar to the petcoke flame. However,
the intensity does not reach near 250 as it does for the petcoke flame.

6.3.3 Co-fired Flame with Petcoke in Central Channel

As already mentioned a special feature of the Jetflex burner is that the petcoke can
be switched from being fed in the regular annular channel to the central channel
where it is mixed with the SRF. In this chapter setting changes are investigated with
the petcoke in the central channel.
6.3 Flame Measurements 93

6.3.3.1 Effect of Swirl Air

Switching the petcoke from the annular to central channel has a relatively large
impact on the flame appearance as seen when comparing Figure 6-14 with Figure
6-8. With the petcoke in the central channel, the fuel plume is initially narrow.
However, with higher degree of swirl the fuel quickly expands to be as wide as when
petcoke is injected through the annular channel. The SRF appears to be more
dispersed with burning particles above and below the main fuel plume. When
petcoke and air is in the annular channel it may act as a barrier that restricts the
flow of SRF, which is not the case with petcoke in the central channel. With the
petcoke in the central channel, the fuel plume also appears more transparent. This
could perhaps be caused by some of the petcoke sticking to the larger SRF particles.
The ignition of the petcoke seems to be easier, when it is fired through the annular
channel. In Figure 6-14 the majority of the fuel is not ignited within the image frame.
Only the SRF particles that are whirled out into the hot secondary air are ignited.
When the petcoke is fed through the annular channel, it may be easier to mix with
the secondary air, aiding ignition. Furthermore, as it encases the SRF the petcoke
will absorb most of the thermal radiation from the kiln walls, which results in a
faster heating.

Figure 6-14: Effect of swirl pressure on the co-fired flame with axial air nozzles at
60° and petcoke in center. Swirl pressure of a) 150 mbarg, b) 10 mbarg c) 180
mbarg. SRF contributes 65 % energy.
94 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

Changing the swirl air pressure when the petcoke is in the central channel is seen to
have a large effect on the flow of fuel as shown in Figure 6-14 and Figure 6-15. The
image sequence in Figure 6-14 shows an initial swirl pressure of 150 mbarg, it is
then reduced to 10 mbarg, and increased to 180 mbarg. The difference between the
150 and 180 mbarg is slight, but turning the swirl pressure to 10 mbarg has a very
large effect. In this case the fuel is hardly dispersed in the kiln, resulting in a very
narrow fuel plume. Observing the intensity plots in Figure 6-15 it is seen that the
change in swirl has no influence on the ignition point, but a large impact on how well
the fuel is dispersed in the kiln cross section. With higher swirl, the dispersion of the
fuel happens rapidly around 0.5 meters, which may be caused by a recirculation
zone that that serves to break up the fuel.

Figure 6-15: Intensity profiles over time when swirl pressure is changed from 150 to
10 to 180 mbarg. a) along the horizontal centerline, and b) across the image 1 m in
front of the burner.

The images presented in Figure 6-16 shows a different view where the camera has
been turned more downstream in the kiln to better observe how the SRF behaves
further from the burner. The swirl pressure has a large influence of how the SRF is
dispersed in the kiln. When the swirl is high, the SRF is easily whirled out of the
petcoke plume, where it begins to burn as it enters the hot secondary air. Much of
the fuel is seen to be burning close to the walls and will be in contact with the clinker.
With the lower swirl level, less of the SRF is whirled out of the petcoke plume, and
it does not ignite. It will instead travel further into the kiln, and eventually burn
further away. Comparing the images to those in Figure 6-10 more SRF is whirled out
of the central petcoke plume. Likely, this is because the SRF flow is no longer
restricted by the petcoke and transport air in the annular channel.
6.3 Flame Measurements 95

It is often reported that unconverted fuel in contact with the clinker may cause local
reducing conditions [46], which can increase volatile circulation [111,119] and
cause brown clinker with reduced quality [106–108]. However, these negative
effects were not observed during testing, even though large amounts of particles can
be seen burning at the walls and in contact with the clinker, when the swirl level is
high. With the low swirl level, the SRF particles are likely to end up in the clinker
regardless, since the temperature and conversion is low in the central part of the
fuel plume. This cannot be observed in the videos, since it occurs too far from the
burner, and the view is obstructed by the petcoke flame and clinker dust.

Figure 6-16: Effect of swirl pressure on the co-fired flame with axial air nozzles at
30° and petcoke in central channel. Swirl pressure of a) 50 mbarg, b) 180 mbarg.
SRF contributes 70 % energy.

6.3.3.2 Effect of Nozzle Configuration

The effect of changing the axial air nozzles is shown in Figure 6-17 and Figure 6-18.
An interesting behavior is observed here. The fuel plume expands as expected when
nozzles are changed from 0 over 30 to 60°. However, when the nozzles are then
turned to 180°, the plume is narrower than in 30 or 60°.
96 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

Figure 6-17: Effect of axial air nozzle configuration on the co-fired flame with swirl
pressure of 150 mbarg. Axial air nozzle position of a) 0°, b) 30° c) 60°, d) 180°. SRF
contributes 65 % energy.

When the nozzles are at 30 or 60° a sudden expansion of the fuel occurs around 0.5
m in front of the burner. When the nozzles are at 180°, this sudden expansion is less
pronounced, and instead the fuel plume is wider at the burner front. A similar
behavior was observed when the petcoke was in the annular channel as shown in
Figure 6-11. It is possible that the change of nozzle configuration changes the
6.3 Flame Measurements 97

location or size of the internal recirculation zone, which affects the dispersion of
SRF.

It is observed that the SRF does not ignite within the first 5 meters in the central
part of the fuel plume. It only ignites when it is dispersed into the hot secondary air
surrounding the fuel plume. Thus, no impact on the nozzles is observed along the
horizontal centerline in Figure 6-18a. There is a change in the spreading of the fuel
and the width of the fuel plume as observed in Figure 6-18b. As the nozzles are
changed, the plume is shifted downwards from around 70 to 110 s. The plume
stabilizes more around the center when the remaining nozzles are adjusted around
250-300 s.

Figure 6-18: Intensity profiles over time when nozzles are changed from 0 to 180°. a)
along the horizontal centerline, and b) across the image 1 m in front of the burner.

The impact of swirl and nozzles when the petcoke is fired through the central
channel is summarized and compared in Figure 6-19. In the horizontal direction the
settings have little impact on the observed intensity, see Figure 6-19a. The intensity
is generally below 50, indicating no ignition of the fuel in the central part of the fuel
plume. The width of the fuel plume and the spreading of the fuel is indicated by
Figure 6-19b. Here a significant impact is seen when the swirl is lowered to 10
mbarg (red curve), which results in a very narrow fuel plume. The nozzle
configuration of 30 or 60° (magenta and yellow curves) shows only minor
differences. The largest difference is seen between 0 and 180° (black and dark green
curves). Observing the intensity dip in Figure 6-19b, the curve for 180° is shifted to
the left compared to the curve for 0°. This indicates that the fuel is pushed more
downwards, which is also evident comparing Figure 6-18a and d.
98 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

Figure 6-19: Intensity profiles for the co-fired flame with petcoke in central channel
images with different swirl (S in mbarg) and nozzle configurations (N in °) a) along
the horizontal centerline, and b) across the image 1 m in front of the burner.

6.3.4 Retraction of the Central Channel

The last feature of the burner that will be discussed here is that the center of the
burner can be retracted, while petcoke and SRF is fired through the central channel.

Figure 6-20a shows the flame with the center in the normal position, with a swirl of
180 mbarg and nozzles at 180°. In Figure 6-20b, the center is retracted. The effect is
that the fuel is slightly more dispersed in the immediate vicinity of the burner within
20 cm. In Figure 6-21b, the image intensity is slightly lower with the center
retracted, than when it is in the normal position. Further from the burner the
difference is not discernible. The plots with intensity changes over time are not
shown here, since the difference is too small to be discerned on that kind of plot.

A scaled down version of the burner was tested in cold flow, before the industrial
tests were performed. In the cold tests, retracting the center had a large impact on
the fuel flow in front of the burner, compared to what is seen here. The reason is
likely a higher flowrate and forward momentum of the fuel and transport air in the
industrial tests, compared to the cold tests. The additional forward momentum
makes it more difficult to change the flow direction of the fuel.
6.3 Flame Measurements 99

Figure 6-20: Effect of retracting burner center with axial air nozzles at 180° and
swirl 180 mbarg. a) Center normal, b) Center Retracted. SRF contributes 90 %
energy.

Figure 6-21: Intensity profiles for the images with center forward (F) or retracted
(R) a) along the horizontal centerline, and b) across the image 0.1 m in front of the
burner.
100 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

6.4 Statistical Data Analysis

In order to determine the optimal conditions for the operation of the burner when
co-firing SRF a series of different burner settings were tested over 6 days according
to the schedule in Table 6-2. The purpose was to test the impact of the different
settings described earlier in this report and couple their impact to the cement
clinker quality, which is measured as the clinker alite content. Alite (Ca3SiO5 or C3S
in cement chemist notation) is the main phase of interest in the Portland cement
clinker [11].

Table 6-2: Overview of burner setting tests.

Test Test Fuel pos. Axial Swirl Nozzle Center

number time pressure pressure position Position
(hours) (mbarg) (mbarg) (°)
1 3.2 Center 210 200 180 Forward
2 16.7 Annular 210 200 180 Forward
3 8.1 Center 210 200 180 Forward
4 17.2 Annular 210 200 180 Forward
5 0.5 Center 210 200 180 Forward
6 2.2 Center 210 200 180 Back
7 1.4 Annular 210 200 180 Forward
8 11.6 Center 210 200 180 Forward
9 3.0 Annular 210 200 180 Forward
10 2.6 Center 210 200 180 Forward
11 1.5 Center 220 130 150 Back
12 1.2 Center 235 82 150 Back
13 1.3 Center 242 56 150 Back
14 1.8 Center 215 140 150 Back
15 2.4 Center 210 170 60 Forward
16 26.4 Annular 210 170 60 Forward
17 44.4 Annular 220 150 60 Forward

In order to determine the impact of the burner settings, and potential other process
variables, on the clinker alite content, Partial Least Squares Regression (PLSR) is
used. PLSR is a statistical tool to handle multivariate data [149–152], and can be
used for e.g. process analysis and monitoring [152,153]. Within the cement industry
PLSR has previously been used to predict quality and emissions [130,154–156].

The process variables are collected in a matrix, XPLSR, that are used to describe the
alite content, YPLSR. The PLSR model determines a number of regression coefficients,
BPLSR, that can be used to estimate YPLSR by the matrix 𝑌̂𝑃𝐿𝑆𝑅 according to [153]:
6.4 Statistical Data Analysis 101

̂ 𝑷𝑳𝑺𝑹 = 𝑿𝑷𝑳𝑺𝑹 𝑩𝑷𝑳𝑺𝑹

𝒀 E 6.4

There are a number of different methods to determine the most influencing

variables for the PLSR [157]. One of the simpler methods is to observe the
association between predictor variables and the response by observing the
regression coefficients in the PLSR, i.e. BPLSR. In this way the impact of some of the
burner settings can be determined [153].

In the PLSR model 23 measured variables are used to predict the clinker alite
content. An overview of the measured variables is given in Table 6-3. Of most
interest here are the variables related to the burner settings, such as swirl air
pressure and the nozzle configuration. In addition, some other variables that may
have an influence on the clinker quality, such as fuel dosing, are also added to the X-
matrix. Furthermore, the composition of the clinker, measured as Lime Saturation
Factor (LSF), Silica Ratio (SR), and Alumina Ratio (AR), is included. Most of the X
variables are continuous measurements, but two categorical variables are also used.
Petcoke can be fired through the central (assigned 1) or the annular channel
(assigned -1), and the burner center can be retracted (assigned 1) or in normal
position (assigned -1).

The PLSR data analysis is performed in Matlab 2015b. The raw data are firstly
reviewed, and obvious outliers are removed. For the data in question this was
mainly related to the gas concentrations at the kiln raw meal inlet. Oxygen
concentrations above 10 % are generally uncommon and may be caused by leakage
of false air, thus these data were removed together with the corresponding CO and
NOx measurements. The rotational speed of the kiln, also had some outliers, which
also caused outliers in the kiln power consumption.

The clinker is sampled every two hours (with some exceptions), and the clinker
phases are determined by X-ray diffraction [125] to determine the alite content. The
process data are logged on a minute basis, which causes relatively large scatter in
the data. Thus, to reduce the data scatter and number of process measurements, the
measurements in the time interval 60 minutes before each clinker sample were
averaged, to yield the value that is used in the regression. Afterwards the data were
mean centered and normalized to standard deviation 1. In total 57 data points were
used in the regression.
102 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

Table 6-3: Overview of variables used to predict the clinker alite content in the PLSR
model.

Variable Name Description Unit

SR Silica Ratio measured in the cement clinker -
Nozzle Configuration – Angle of the axial air nozzles on
Nozzle Conf. °
the burner
Power consumption of the kiln drive. Value will tend to
Kiln Power kW
increase at higher kiln temperatures
LSF Lime saturation factor in the cement clinker -
SRF to MB SRF fired in the main burner ton/h
Temperature in the kiln hood. Approximates the
Kiln Hood T °C
secondary air temperature
If the burner center is retracted (1) or if it is at normal
Retracted Center -
position (-1)
Swirl P Burner swirl pressure mbarg
MB Total Power Total heat input in main burner MW
NOx at Kiln Inlet NOx measured at the kiln raw material inlet ppm
If the petcoke is fired through the central channel (1) or
Petcoke in Center -
in the annular (-1)
PC to Calciner Petcoke fired in the calciner ton/h
Calciner Total
Total heat input in the calciner MW
Power
Axial P Burner axial pressure mbarg
Kiln RPM Rotational speed of the kiln in revolutions per minute RPM
PC to MB Petcoke fired in the main burner ton/h
Ratio of amount of SRF fed to main burner to the
SRF/Rot. ton/h/%
rotational speed of the SRF feeder
SRF to Calciner SRF fired in the calciner ton/h
O2 at Kiln Inlet Oxygen measured at the kiln raw material inlet %
Kiln Feed Amount of raw material fed to the kiln ton/h
HD Total Power Total heat input in the Hotdisc MW
AR Alumina ratio measured in the cement clinker -
CO at Kiln Inlet CO measured at the kiln raw material inlet %

The number of latent variables is determined by computing the Mean Square Error
of Prediction (MSEP) [158] as determined by 10-fold cross validation [159], which
is shown in Figure 6-22a. The minimum value of the MSEP is achieved for 4 PLS
components. The resulting measurement and model predictions are shown in
Figure 6-22b. The data are shown in normalized form. The model follows the overall
trend well, but seems to be unable to capture some of the outliers of lower alite
content.
6.4 Statistical Data Analysis 103

Figure 6-22: a) MSEP values as function of number of latent variables in PLSR. b)

Comparison of measured and modeled values of clinker alite content using 4 latent
variables, in normalized form.

6.4.1 Impact of Process Variables

The regression coefficients for the PLSR model, BPLSR, are shown in Figure 6-23.

̂ 𝑷𝑳𝑺𝑹 =XPLSRBPLSR). Error bars

Figure 6-23: Regression coefficients in the PLSR model (𝒀
are 95 % confidence interval determined by 10-fold cross validation. Variables are
explained in Table 6-3.

Before focusing on the effect of burner settings, the effect of other variables will be
reviewed, to determine if the regression predictions are reasonable. It is seen in
Figure 6-23 that some of the parameters that yield high alite content are related to
104 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

the kiln temperature. Kiln power, kiln hood temperature and NOx at the kiln inlet all
have a positive impact on the clinker quality. With high temperatures in the kiln the
amount of melt in the kiln is increased, which increases the required torque and
power consumption to rotate the kiln [17]. The formation of NOx will generally
increase with increased temperatures as well [41].

Other factors that have a positive impact on the alite content are the clinker lime
saturation factor (LSF) and silica ratio (SR), which are related to the composition of
the clinker. With higher values of these ratios more lime and silica are present in the
clinker, which can combine to form the alite. On the other hand, an increased LSF
and SR increases clinker burnability. At the other end of the spectrum the clinker
alumina ratio (AR) has a negative impact. With increased iron content the factor is
lowered, which favors the reaction of lime and silica to form alite [34].

It is seen that SRF fired in the main burner (SRF to MB) presumably has a positive
impact on the clinker quality, while increasing the petcoke (PC to MB) has a negative
impact. Since SRF is more difficult to combust and burns at lower temperatures, this
appears counterintuitive. However, since high amounts of SRF firing is the normal
at this plant, high amounts of petcoke indicates an upset kiln. Typically, the plant
operators will increase the amount of petcoke when the kiln temperature is
lowered, thus it would likely coincide with lower alite content. It is also observed
that the clinker alite content is increased with the total energy input to the kiln (MB
Total Power). Another factor that is related to the fuel is the SRF/Rot, this measure
is calculated as the amount of SRF fed to the kiln divided by the rotational speed of
the SRF weigh feeder. If the SRF is dense, i.e. has high moisture content, the same
mass can be fed with lower rotational speed than when the SRF is less dense, i.e. dry.
The measure thus indicates the humidity of the SRF fed to the burner, and indicates
that a higher humidity, which results in higher SRF/Rot numbers, has a negative
impact on the alite content.

The factors that can have a negative impact on the alite content are e.g. kiln feed, as
well as oxygen and CO at the kiln raw meal inlet. With an increased raw meal feed
rate to the kiln, alite content will decrease, since the temperature decreases with
more material present in the kiln. Too much oxygen in the kiln raw meal inlet
indicates an excessive air to fuel ratio. This lowers the adiabatic flame temperature
[34], which could lead to lower alite content. On the other hand, the oxygen content
6.4 Statistical Data Analysis 105

should not be too low, since that will inhibit combustion. Spikes of CO may indicate
an incomplete combustion, which will also negatively impact the clinker alite
content.

It is interesting that the model indicates that firing of SRF in the Hotdisc (HD) should
decrease the alite content. Perhaps excessive firing in the Hotdisc rather than in the
calciner can negatively impact the calcination degree of the raw meal as it enters the
kiln. The fuel burned in the Hotdisc is also of a relatively low quality, and expected
to have a high content of impurities, such as metals and glass. Large silica particles
originating from glass may have difficulties reacting with the lime in the kiln due to
diffusion limitations [25], which could lower the alite content.

Based on this review of impact of different process measurements, it appears that

the PLSR model is reasonable and that most measures behave as expected. Thus, the
attention will now be turned to the effect of the burner settings.

6.4.2 Impact of Burner Settings

Increasing the swirl level (Swirl P) or the nozzle configuration (Nozzle Conf.), i.e.
turning the nozzles outwards to a larger angle, appear to have a positive impact on
the clinker alite content. From the observations made with the camera, it is seen that
increased swirl or nozzle angle will help disperse SRF into the secondary air stream
and ignite it. For a co-fired flame with the petcoke in the annular channel, higher
swirl and nozzle configuration can also help ignite the petcoke faster. It is believed
that spreading the SRF near the burner end will cause a more rapid combustion of
the fuel, which contributes to increasing the temperature in the clinker burning
zone. With a low swirl, the fuel will burn too far inside the kiln and not contribute
energy, where it is required the most.

It appears beneficial to have the petcoke in the central channel rather than the
annular channel, although the value of the regression constant has a high deviation.
Based on the visual observations of the flame, this may be caused by the increased
dispersion of SRF in the kiln, when it is not enclosed by the petcoke. Petcoke in the
annular channel can also shield the SRF from radiation, meaning that it will take
longer to heat and ignite the SRF.

A retracted center may also increase the alite content of the clinker. Based on the
videos this appears to slightly increase the dispersion of the SRF when the petcoke
106 6 Effect of Burner Settings on the Cement Kiln Flame and Clinker Quality

is fed through the central channel. However, the burner was only operating for 8
hours with a retracted center during the test of 146 hours. This is perhaps too few
operating hours to make concrete conclusions.

6.5 Conclusions
The recently developed Jetflex cement kiln burner from FLSmidth has several
different possibilities to shape the flame. This gives a flexible burner where it is
possible to adjust the flame as e.g. fuel type or quality changes. In addition to the
adjustment of axial momentum and swirl level, it is possible to direct the axial air
using rotatable axial air nozzles and the petcoke can be switched from being fired in
a conventional annular channel, to being mixed with the SRF in a large central pipe.
It is also possible to retract the central part of the burner. Using a specially
developed camera probe it has been possible to investigate and quantify how the
different burner features impact the flame in the cement kiln.

When the kiln operates solely on petcoke it is possible to impact the ignition point
of the flame by adjusting the swirl and the nozzle configuration, changing it from
approximately 5 m to 3.5 m from the burner. The co-fired flame is harder to ignite
and the effect of swirl and nozzles on the ignition point of the co-fired flame is less
pronounced. When the petcoke is fired with the SRF through the central channel,
instead of in the annular channel, ignition of the petcoke is more difficult. However,
it becomes easier to disperse the SRF into the hot secondary air, which helps igniting
the SRF. Increasing swirl or nozzle angle can increase the dispersion of SRF, but it
will also whirl the SRF out of the main fuel plume. This causes it to burn on the kiln
walls and in contact with the cement clinker. Retraction of the central part of the
burner slightly improves the dispersion of the SRF.

The effect of burner settings were coupled to the cement clinker quality, by
analyzing recorded operating data using Partial Least Squares Regression. Based on
this analysis it can be concluded that increasing the swirl and turning the axial air
nozzles to a larger angle, has a positive effect on the clinker alite content. Firing the
petcoke through the central channel and retracting the central channel also had a
positive effect on the clinker quality. The overall effect of these four measures is to
increase the dispersion of the SRF near the burner. Presumably this will help
establish a high temperature zone where alite can be formed.
7 PHYSICAL CHARACTERIZATION
OF SRF

The classification system of the European standard CEN/TS 15359 classifies SRF
based on the heating value and content of chlorine and mercury [160]. However, this
classification is not detailed enough to determine how the fuel will burn in a cement
kiln. Normally it is also required to know the moisture content of the fuel, and the
particle size. For a cement kiln burner, a common size limitation for SRF is that
particle size should be smaller than 30 mm [100] and predominantly two-
dimensional. However, this is a vague definition and may be difficult to observe in
practice. More detailed guidelines for the properties of SRF may help cement
manufacturers assure the SRF quality and better determine limitations of co-firing.

Additionally, a thorough understanding of fuel properties on a particle level is

necessary to create a detailed input for computational models. The parameters
should include:

• Particle size/mass distribution and shape

• Composition of fuel
• Combustion kinetics

The particle size and shape are important for the combustion and flow behavior of
the fuel. Larger particles take longer to heat and convert, while a larger surface area
decreases conversion times. Furthermore, light and heavy particles are influenced
differently by air flows. The composition of the fuel, with respect to e.g. proximate
108 7 Physical Characterization of SRF

and ultimate analysis is also important for the combustion behavior. Lastly, the rate
at which the fuels react will also impact the combustion process.

Considering the large size of alternative fuels, generally above 1 mm, it is believed
that the primary limitations for combustion are heat and oxygen transfer to the
particle. Thus, it is more important to focus on an accurate description of the particle
size and shape, rather than a detailed kinetic mechanism for combustion. This is the
foremost purpose of a physical characterization method of SRF, which is covered in
this chapter. Furthermore, the overall heat and mass balances obtained by a
proximate, ultimate, and heating value analysis, are also important.

The work presented in this chapter has been carried out in collaboration with fellow
PhD student, Mohammadhadi Nakhaei. An article manuscript [161] covering this
work, with Mohammadhadi Nakhaei as first author, has been submitted to the
journal Energy and Fuels.

7.1 Characterization of SRF in Literature

SRF is highly heterogeneous, consisting of different materials, as outlined in chapter
3.4.1. The main parts of the combustible fractions can be derived into biomass,
paper & cardboard, plastics, and textiles (see Figure 3-3).

Liedmann et al. [162] attempted to characterize RDF and suggested a method for
implementing the RDF particles into a CFD model for use in a power plant. Other
articles by the authors used the same approach to model the co-firing in a cement
kiln [143,163]. The RDF was divided into five groups: 2D foils, 3D plastics, paper &
cardboard, textiles, and fines. Proximate and ultimate analysis and heating value
was determined for each fraction. The mass of particles from each fraction was
determined, and the sphericity was calculated based on images of the particles.
Aerodynamic properties, drag and lift coefficients, were determined in a drop shaft.

Thermogravimetric Analyses (TGA) of different materials from SRF are quite

abundant in the literature [164–166]. They show that different kinds of paper and
cardboard, which are based on lignocellulosic materials, have a very similar
devolatilization behavior, with the main mass loss occurring between 300-400 °C.
Different kinds of plastic (except polyvinylchloride) also behave similarly with the
main mass loss occurring between 400-500 °C. Of these plastics, polyethylene is the
most widely used plastic [167], and will also be present in the largest amount in SRF
7.2 Wind Sieve Setup 109

[164,168]. Polyethylene is almost completely volatile, while the lignocellulosic

biomass will leave a char upon devolatilization [164,166]. Furthermore, it is also
possible to see two distinct steps of mass loss in SRF samples during devolatilization,
as e.g. shown in Figure 7-1. The size of the steps vary based on the content of plastics
and biomass [165].

Figure 7-1: TGA experiment of SRF sample showing two distinct mass loss steps [165].

Consequently, to simplify the combustion model for SRF, it will be natural to split
the SRF particles into a cellulosic biomass fraction and a plastic. The textile fraction
can also be split into two, since it for the most part will consist of either natural
fibers, e.g. cotton, or artificial fibers, e.g. nylon or polyester, which also decompose
in the temperature range 400-500 °C [169]. If this approach is followed, it is only
necessary to split the SRF sample into two fractions with different
chemical/combustion properties, compared to the five different fractions proposed
by Liedmann et al. [162]. A further detail of the size distribution and shape for
particles in both fractions is still necessary.

7.2 Wind Sieve Setup

A regular sieving analysis is often inappropriate to characterize the size of SRF
[170]. Since particles are only separated by their size, differences in particle shape
may impact results significantly. If particles are instead separated by their terminal
velocity, both the particle mass and shape will affect the classification. A simple
example is presented in Table 7-1.

Assuming a sieve with an aperture size of 1 mm, a flat cuboid particle of dimensions
1x1x0.1 mm may be retained while a spherical particle of diameter 0.8 mm will pass.
110 7 Physical Characterization of SRF

However, since the spherical particle has a higher volume and lower surface area,
one would expect it to burn slower than the flat particle. If the particles are instead
separated by their terminal velocity, the flat particle has a smaller terminal velocity,
which is consistent with it burning faster.

Table 7-1: Example calculations on particle shape. Terminal velocity determined at

20 °C with assumed drag coefficient of 0.44.

Particle Particle Equivalent Volume Surface Projected Terminal

shape size diameter area area velocity
mm mm mm3 mm2 mm2 mm/s
Spherical 0.8 0.80 0.27 2.0 0.50 1.6
Cuboid 1x1x0.1 0.58 0.10 2.4 1.0 0.88

Figure 7-2: Sketch of the wind sieve setup used to characterize alternative fuels.
7.2 Wind Sieve Setup 111

A wind sieve setup has been used to characterize the SRF. A sketch of the setup is
shown in Figure 7-2. The sieve separates particles based on their terminal velocity.
Particles are fed into a tube of diameter 24 cm and height 333 cm. The air velocity
in the tube is maintained at a set-point controlled by a Pitot tube. Particles are fed at
the middle of the tube, and particles with a terminal velocity higher than the set
point will fall down, while the remaining particles are carried upwards with the air
stream. The heavy fraction of particles is collected in a tray at the bottom, and the
light fraction of particles is separated using a cyclone and collected at the top. The
heavy fraction is collected and inserted into the wind sieve again using a higher air
velocity. Thus, a particle sample can be fractioned into multiple groups based on
terminal velocity.

7.2.1 Results from Wind Sieve Tests

Six different RDF samples obtained from three different cement plants were tested
in the wind sieve. An overview of the fuels is given in Table 7-2. Industrial flame
measurements were performed at Plant 1 and Plant 2 (see chapters 5 and 6). The
samples are all SRF, except for RDF-C, which is the granulated tire used in the main
burner at Plant 2. The samples were dried at 105 °C for three hours before being
tested in the wind sieve. The high temperature was chosen to also disinfect the
samples. No significant melting of plastic particles was observed at this
temperature.

Table 7-2: Details of the fuels tested in the wind sieve.

RDF
Place of Use LHV Moisture Ash
Sample
MJ/kg wt% wt%
Plant X
A - - -
Calciner
Plant 1
B 15.4 21.4 17.6
Kiln
Plant 2
C 31.4 1.5 6.0
Kiln
Plant 2
D 18.7 17.3 12.5
kiln/calciner
Plant 2
E 17.8 18.5 13.0
calciner
Plant 1
F 14.6 29.9 19.2
Hotdisc
112 7 Physical Characterization of SRF

Results of the wind sieve characterization are shown in Figure 7-3. The fuels appear
to be divided into three main groups, light fuel (RDF-C), medium fuel (RDF-A, RDF-
B, RDF-D), and heavy fuel (RDF-E and RDF-F).

RDF-C and RDF-D were both burned in the kiln at Plant 2 during the industrial flame
imaging tests. The upper limitation of firing is around 3 t/h for RDF-C and 2 t/h for
RDF-D. Higher degrees of alternative fuels firing tends to result in reducing
conditions, which is a sign of the particles not being fully converted in the flame. The
wind sieve analysis shows that RDF-C is significantly lighter, and it is thus expected
to be converted faster than RDF-D. Thus, higher degree of co-firing can be achieved
using RDF-C. RDF-E is also used at Plant 2, but only in the calciner. This fuel is seen
to be significantly heavier than the fuels used in the kiln.

RDF-B and RDF-F are both used at Plant 1. RDF-B is used in the main burner in the
kiln, while RDF-F is used in the Hotdisc [101]. The Hotdisc is designed to burn large
particles that are converted too slowly to be burned in the kiln or calciner. Thus, it
is reasonable that RDF-F is heavier than the RDF-B.

RDF-A is used in the calciner of Plant X. The distribution is very similar to that of
RDF-B. Thus, it should be possible to use the same fuel in the kiln.

Figure 7-3: Results of wind sieve characterization of various RDFs.

A well-known rule of thumb for the particle size of coal for use in cement kilns, states
that the residue at a 90 µm sieve should be maximum half of the volatile content
[40]. It would be beneficial if a similar simple rule could be proposed for the use of
RDF.
7.3 Manual Fractionation of SRF 113

Co-firing of SRF at Plant 1 averages around 70 % of the energy input (see Table 5-3).
It is expected that a fuel should behave similarly to RDF-B in the wind sieve, to be
suitable for a high degree of co-firing. Therefore, to achieve above 50 % energy
substitution in the kiln it is suggested that the fuel should have at least 20, 60, and
80 wt% light fraction tested at 3, 5, and 7 m/s, respectively, similarly to RDF-B.

This is an early suggestion for how the wind sieve analysis can be used to gauge fuel
quality for the cement kiln. Further tests are necessary to make more concrete
conclusions.

At Plant 2 co-firing only reached 30 % energy using RDF-C and 10 % using RDF-D,
even though these fuels are lighter than RDF-B used at Plant 1. Thus, other
limitations than the particle size contribute to limiting the use of RDF. The limitation
at Plant 2 appears to be a kiln that is more sensitive to reducing conditions than
Plant 1.

7.3 Manual Fractionation of SRF

Samples of RDF-A and RDF-B were collected from the wind sieve to make an in-
depth analysis of the two samples.

In the wind sieve, samples were fractionated into five groups according to their
terminal velocities. Terminal velocity ranges for the groups were <2, 2-3, 3-5, 5-7,
and >7 m/s. Each of the five groups were first sieved to remove fines smaller than
2 mm. These particles were too small to reliably determine in which group they
belong. The remaining fuel was manually sorted and classified into four groups:
plastic, biomass, textiles, and inert. The sorting was based on a visual
characterization based on particle texture. Results of the sorting are shown in Figure
7-4.

For both RDF samples the amount of plastics in the light fractions (lower terminal
velocity range) is high. In the heavier fractions the plastic is reduced, and the amount
of biomass is increased. This is likely because a high amount of the plastic is from
packaging foil, which has a high surface to mass ratio, which means that it has low
terminal velocity. The biomass particles are mainly thicker paper and cardboard
pieces and woodchips, which are heavier than the plastic foils. The amount of inert
is also highest in the fraction 5. These are mainly metals and stone, which are quite
dense, and thus have a high terminal velocity.
114 7 Physical Characterization of SRF

Figure 7-4: Results of manual sorting and classification of RDF-A and RDF-B. Fines are
particles under 2 mm as determined by sieving.

The main distinction between the two RDF samples is that RDF-A is richer in plastics
in the light factions and the amount of fines is somewhat higher in RDF-B.

7.4 Particle Mass Distribution

The biomass and plastic particles of each fraction (20 samples in total) were
reduced to a subset with 50-100 particles by the quartering method [171]. The mass
of each particle was measured on a scale. Subsequently, the data were fitted to a
Rosin-Rammler distribution [172,173]:

𝒎𝒑 𝒏𝑹𝑹
𝑪𝒎𝑹𝑹 (𝒎𝒑 ) = 𝟏𝟎𝟎 − (𝟏𝟎𝟎 ∗ 𝒆𝒙𝒑 (− ( ) )) E 7.1
𝒎𝑹𝑹
7.4 Particle Mass Distribution 115

CmRR is the cumulative mass, mp is the particle mass, mRR the characteristic mass
(equivalent to the 63rd percentile), and nRR is the spread parameter.

The Rosin-Rammler distribution was initially developed to describe the particle

distribution of coal [172], but has also been found to be valid for shredded refuse
fuel [174].

The mass distribution of the biomass and plastic fractions of RDF-A and RDF-B are
shown in Figure 7-5. The parameters for the fitted Rosin-Rammler distributions are
given in Table 7-3.

Figure 7-5: Mass distribution fitted to Rosin-Rammler distribution of the plastic and
biomass fractions of RDF-A and RDF-B.

It is observed from the data that the particle mass increases with the terminal
velocity. The parameter mRR (63rd percentile) increases through the groups with
higher terminal velocity. The spread parameter nRR also appears to increase slightly
with the terminal velocity, but this trend is less clear and noisier. Furthermore, it is
observed that sample A is generally slightly heavier than sample B, for both the
116 7 Physical Characterization of SRF

plastic and biomass fractions. This may be explained by RDF-A having a slightly
lower particle sphericity, i.e. it is flatter, as discussed in the following chapter.

Table 7-3: Parameters for the Rosin-Rammler distribution of RDF-A and RDF-B
samples.

Biomass RDF-A Biomass RDF-B Plastic RDF-A Plastic RDF-B

Velocity mRR nRR mRR nRR mRR nRR mRR nRR
Range (mg) (mg) (mg) (mg)
(m/s)
<2 20.2 1.61 9.3 1.35 16.9 1.51 8.3 1.54
2-3 27.1 1.87 23.1 1.52 45.4 1.63 27.1 1.35
3-5 102.4 1.50 31.3 1.62 99.3 1.50 44.2 1.43
5-7 330.5 1.92 114.3 1.57 410.7 1.65 201.9 1.37
>7 874.9 1.89 844.1 1.77 785.8 1.68 565.8 1.63

7.5 Particle Shape Determination by Imaging

The shape of particles also have a significant effect on their combustion time
[86,175]. The larger the surface area of a particle, the faster heat and oxygen can be
transferred to the particle. In addition, the shape of particles also influence their
flow and terminal velocity [162].

A camera setup was developed to take pictures of single particles of SRF, to describe
their size and shape. The camera setup is shown in Figure 7-6. It consists of a back
lit platform where particles are placed on. Images are taken by two cameras placed
above and in front of the particle.

Figure 7-6: Camera setup with two cameras to determine particle size and shape.
7.5 Particle Shape Determination by Imaging 117

The particle mass was determined by weighing the particles as explained in chapter
7.4. Based on the particle mass, the particle volume, Vp, is calculated assuming a
density of 650 kg/m3 [176] for biomass and 950 kg/m3 [177] for plastics. Assuming
that the particles are cuboids, the thickness of the particle can be determined by:

𝒕𝒑 = 𝑽𝒑 /𝑨𝒑 E 7.2

The particle projected area, Ap, is derived from the top image taken in the camera
setup. With the volume and surface area, the particle sphericity can be calculated:

𝟔𝑽𝒑 𝟐/𝟑
𝝅( 𝝅 )
𝝋𝒑 = E 7.3
𝟏/𝟐
𝟐𝑨𝒑 + 𝟒𝒕𝒑 𝑨𝒑

The sphericity of the biomass and plastic samples of RDF-A and RDF-B are shown in
Figure 7-7, and the mass- weighted average values are presented in Table 7-4. It is
observed that the sphericity of the plastic particles is generally slightly lower than
for the biomass particles. The sphericity generally increases with the terminal
velocity, i.e. the particles are flatter and more sheet like at low terminal velocity. In
each group there is a slight trend that increasing mass results in a lower sphericity.
This is most visible for the fraction 5-7 m/s. This is reasonable as a heavy and a light
particle can have the same terminal velocity, if the heavy particle has a larger
projected area, i.e. it is flatter and has lower sphericity. RDF-A has a slightly lower
sphericity than RDF-B. This is also consistent with a slightly higher mass for RDF-A
in each of the terminal velocity ranges.
118 7 Physical Characterization of SRF

Figure 7-7: Sphericity of biomass and plastic of RDF-A and RDF-B in relation to the
particle mass.

Table 7-4: Mass weighted average sphericity for biomass and plastics fractions of
RDF-A and RDF-B.

Velocity Biomass Biomass Plastic Plastic

Range RDF-A RDF-B RDF-A RDF-B
(m/s)
<2 0.14 0.15 0.09 0.10
2-3 0.23 0.28 0.12 0.15
3-5 0.31 0.39 0.21 0.26
5-7 0.41 0.49 0.33 0.39
>7 0.50 0.60 0.52 0.55
7.6 Conclusions 119

7.6 Conclusions
A wind sieve setup was used to characterize the size of different refuse derived fuels
from three different cement plants. A clear distinction can be made between fine,
medium, and coarse fuels. Fuels used in the kiln are lighter than fuels used in the
calciner. The current data is limited, but it is suggested that SRF for use in a cement
kiln should have at least 20, 60, and 80 wt% light fraction tested at 3, 5 m/s, and 7
m/s, respectively.

Two samples were selected from the wind sieve and a more in-depth analysis was
made. Plastics and biomass in SRF behave very differently during combustion. Thus,
the plastic and biomass content of the SRF was determined. Plastic particles tend to
accumulate in the low terminal velocity ranges, while biomass is more prevalent in
the high terminal velocity ranges. This indicates that plastic particles are flatter,
which is also reflected in the determination of the particle sphericity.

The mass of individual particles of biomass and plastic was determined, and the
mass distribution was fitted to a Rosin-Rammler distribution. The average particle
mass increased with the wind sieve terminal velocity range. It was also found that
particle sphericity increased with the terminal velocity. Furthermore, heavy
particles in the same wind sieve fraction tended to have a lower sphericity. This
confirms that the wind sieve separation accounts for both particle mass and shape.
Both factors are important for the particle combustion. The combustion time will
generally be shorter for low mass particles with a low sphericity, which can be found
in the low terminal velocity ranges of the wind sieve.

The knowledge of SRF composition, particle size distribution, and shape factors can
be used as input to detailed models for the combustion in the kiln, such as CFD
simulations. The data derived in this chapter will be used as input for modeling the
combustion of SRF described in chapter 9.
120
8 CEMENT KILN MODEL
DEVELOPMENT

The purpose of this chapter is to present a simplified model for the cement kiln. The
model will be used to calculate temperature profiles in the gas phase and clinker
bed. This is coupled to a simplified model of the formation of clinker minerals, to
determine free lime or alite content in the clinker. The model takes into
consideration fuel properties, burner settings, and other process parameters and
can thus be used to estimate how these factors influence kiln temperature and
clinker composition. Consequently, it can also be used to propose changes that
compensates for the negative effects of alternative fuels co-firing.

It is a 1-D model to describe changes in the axial direction of the kiln. This approach
has been chosen over Computational Fluid Dynamics (CFD) simulations. It is
believed that a 1-D model will be adequate to capture the main differences in
temperature caused by variations in fuel quality. In addition, this model is faster to
converge, less complex, and easier to use.

The first part of this chapter provides an overview of the various models found in
the literature, as well as the major findings and conclusions of these studies. In later
parts of this chapter the main processes in the cement kiln will be described, and
simplified models are presented. These sub-models, for e.g. clinker flow, heat
transfer, and reactions, are combined in the final kiln model. The kiln model is
122 8 Cement Kiln Model Development

validated using literature results and used to investigate the impacts of SRF co-firing
in chapter 9.

8.1 Cement Kiln Modeling in Literature

The literature is quite rich in models attempting to describe the various processes
taking place in the cement kiln. Several models have been developed since the
1970’s based on engineering principles to understand and optimize the cement kiln.
Within the last 10-20 years advanced models aided by Computational Fluid
Dynamics (CFD) have also appeared. However, only a few recent studies have dealt
with alternative fuels firing in the cement kiln.

8.1.1 Engineering Models

Some of the first attempts to model the heat transfer in rotary kilns were made by
Imber and Paschkis in the early 60’s [178]. In their work two extreme cases were
considered for the kiln charge temperature, a well-mixed and non-mixed case. In the
well mixed case the bed was assumed to have a uniform temperature in the radial
direction and in the non-mixed case a non-uniform temperature was assumed. A
number of simplifying assumptions were made in their analysis, and a lack of
correlations for the heat transfer coefficient made the model of limited use.

A group of researchers from the University of British Columbia made a detailed

study of the heat transfer in the cement kilns publishing a number of articles in this
area, first by Brimacombe and Watkinson [179,180], Gorog et al. [181–183], and
later by Barr et al. [184,185].

They used a gas fired pilot scale rotary kiln of 0.406 m diameter and 5.5 m length to
experimentally study the heat transfer mechanics and later compare with modeling
results. The kiln was equipped with numerous thermocouples to measure
temperatures in the bed, gas, and wall, which allowed determination of the various
heat transfer mechanics in the rotary kiln. Using the kiln, it was possible to study the
effect on heat transfer of a long range of operating parameters such as rotation
speed, degree of fill, and firing rate [179].

Watkinson and Brimacombe [180] used the pilot rotary kiln to study the heat
transfer from the gas phase to kiln wall and bed. They determined that motion of the
bed has a great influence on the heat transfer. At high bed feed rates and rotation,
8.1 Cement Kiln Modeling in Literature 123

the bed motion changed from slumping to rolling, which increased the heat transfer
rate. This could result in the gas to bed heat flux being up to ten times higher than
the gas to wall heat flux. It was found that radiation only contributed some 30
percent of the heat flow. The temperature in the pilot experiments was around 1000
K. In a real cement kiln, gas temperatures are much higher, and radiation becomes
more important. It must be expected that radiation in an industrial kiln contributes
more than the 30 percent found in the pilot kiln.

Gorog et al. made a thorough contribution to model the different mechanisms of heat
transfer in the rotary kiln. The work describes radiation [181] and regenerative heat
transfer (heat transfer between covered wall and bed) [182]. In later work [183] the
radiative heat transfer from a flame was modeled to calculate an axial temperature
profile of the flame and wall. It was assumed that the flame was cylindrical in shape
and moved in plug flow through the kiln. The length of the flame and the
entrainment of secondary air were calculated by empirical relations from Beér and
Chigier [186] and Ricou and Spalding [187]. The temperature profile was calculated
for fuel oil, natural gas, and producer gas for different levels of firing rate, secondary
air temperature, primary air fraction, and oxygen enrichment. Gorog and Adams
used the developed model as a basis to optimize the design and performance of a
rotary lime kiln in a number of papers [188–192].

Barr et al. [184,185] refurbished the pilot kiln used by Watkinson and Brimacombe,
which was described above. This allowed for additional measurements and accuracy
of the heat transfer modes. In addition, heat transfer to limestone, a reactive bed,
was studied. The heat transfer models from the previous work of Gorog et al. were
combined to yield a detailed model of the cross-sectional heat flow, which covered
all the modes of heat transfer in the kiln.

Spang [193] developed a dynamic model for the axial temperature profile of the
cement kiln based on heat and mass balances. The study included a simplified
approach to model the kiln reactions by Arrhenius type expressions. The fuel
combustion was implemented by assuming an exothermic reaction that releases the
energy directly to the gas phase. The reaction rate was controlled by diffusion of
oxygen from the gas to the particle surface, which was assumed adequate for coal
and oil flames. The model was used to investigate the response to changes in e.g. gas
and solids flow.
124 8 Cement Kiln Model Development

Locher proposed models for the entire pyro-processing line of a cement plant
including the preheater, calciner, bypass, rotary kiln, and grate cooler in a series of
publications [194–196]. The kiln is modeled by dividing it into 20-50 sections in the
axial direction. In each section the mass and energy exchange are calculated. Heat
transfer accounts for gas-wall, gas-bed, and wall-bed interactions through radiation,
convection, and conduction. Compared to the other models described in this
chapter, Locher’s model also considers entrainment of dust, although the complete
details of the calculation method are not stated. Chemical reactions of the clinker are
also considered.

A rather recent model has been developed by Mujamdar and Ranade [197]. The
model accounts for the solids flow with a non-uniform bed height and the chemical
reactions in the bed. The gas temperature was assumed to follow a fixed profile,
which was derived from CFD-calculations, previously performed by the authors. In
later work [198], coal combustion was added to the model. The conversion of coal
was assumed to follow three steps: heating, devolatilization, and char combustion.

The investigation of alternative fuels combustion on the cement kiln is relatively

limited for 1-dimensional models, but was attempted by Haas and Weber [199].
Initially they assumed a simplified ideal kiln consisting of a well stirred reactor and
showed that co-firing of RDF would lower the adiabatic flame temperature, due to
the lower calorific value of the RDF compared to coal. To maintain the same heat
transfer to the bed when the adiabatic flame temperature drops, it is necessary to
increase the energy input to the kiln. The authors went on to model the kiln as a
series of well stirred reactors. The combustion of RDF was modeled considering only
the solid conversion controlled by oxygen diffusion and surface reaction. This is a
serious simplification, due to the high volatile content of the RDF. The RDF reaction
kinetics were assumed similar to that of coal. Thus, the differences in the reaction
rates are related to the large size difference between coal and RDF particles, which
was characterized by a specific surface area. The effect of different co-firing rates of
RDF were then studied, assuming that the total heat transfer to the clinker should
be constant. The effect of increased RDF was a lower peak gas temperature and a
longer flame zone. If the RDF particle size is assumed similar to that of coal, the flame
becomes shorter and the peak temperature is boosted, although the coal will still
burn at higher peak temperature.
8.1 Cement Kiln Modeling in Literature 125

The effects of alternative fuels on kiln temperatures (with no clinker bed) have been
investigated by Nørskov [31]. In this work it was also attempted to account for the
influence of burner settings by describing the entrainment of secondary air into the
primary air and fuel jet. It was concluded that co-firing with sewage sludge or RDF
would shift the peak flame temperatures further away from the burner due to the
larger particle size and slower combustion. Flame temperatures would also be
reduced.

8.1.2 Computational Fluid Dynamic Models

During the last 10-15 years, the use of computational fluid dynamics (CFD) for
numerical models of the cement kiln has increased significantly. The increasing
computing power makes it possible to solve detailed three-dimensional models of
flow, temperature, and combustion.

Some of the first work using computer aided numerical models for cement kilns was
done by Lockwood [200] in the 1990’s. In this work, they compared predictions of
the k-ε and Reynolds stress model for turbulence for a coal fired flame. The
modeling results were also compared with experimental results obtained at the
pilot scale burner of the International Flame Research Foundation. Since then,
numerous studies have been made where CFD is used to model coal fired flames,
such as [201–203].

The amount of literature dealing with the combustion of alternative fuels in the
cement kiln is limited. However, the CFD modeling attempts emphasize some of the
challenges outlined in chapter 3.5.

Ariyaratne et al. [83] used CFD to compare the combustion of coal and meat and
bone meal (MBM). The main finding was that MBM burns at 300 K lower
temperatures than coal due to a higher air demand, higher ash and moisture content,
and poor char burnout.

Liedmann et al. [163] simulated the co-firing of lignite and RDF in a cement kiln. An
example of their results is shown in Figure 8-1. Introducing RDF (Figure 8-1b)
results in a shorter and colder flame, caused by the increased conversion time of the
larger particles. Some of the RDF lands unconverted in the bed around 20 meters
from the burner, which creates a high temperature zone. The unburnt particles are
primarily textiles, paper and cardboard, and 3D plastics owing to their large size.
126 8 Cement Kiln Model Development

Thus, an optimized RDF where textiles were removed, and the size was reduced, was
also modeled (in Figure 8-1c). This resulted in a higher temperature flame and less
material burning in the bed.

Figure 8-1: CFD simulations showing temperature profiles in the cement kiln for
different co-firing scenarios of a) lignite, b) 50 % SRF, and c) 50 % optimized SRF
[163]. Notice the high temperatures at the lower kiln wall halfway through the kiln
for case b and c.

Another paper by Liedmann et al. [143] studied the effect of different operating
conditions on a co-fired kiln flame. It was found that an increased axial momentum
would increase peak temperatures near the burner, but slightly reduce SRF burnout
due to a lowered residence time. An increased swirl resulted in a wider flame with
lower peak temperatures. Light SRF particles were also whirled out of the flame,
causing more material to burn on the walls and in the clinker. An increased
secondary air temperature caused increased flame temperatures, which resulted in
faster ignition and higher burnout of SRF.

Isaac et al. [204] investigated the effect of the moisture content of SRF particles on
a co-fired cement kiln flame. A higher moisture content caused delays in flame
ignition, lower flame temperatures, and decreased burnout of the fuel.

Examples of alternative fuels combustion CFD modeling in the calciner can also be
found [205,206].
8.2 Modeling of Processes in the Cement Rotary Kiln
127

8.1.3 Conclusions and Next Steps

The literature survey has revealed different approaches to model the combustion in
cement kilns. The CFD models yield detailed information of the gas flow coupled
with the combustion reactions. However, their downside is typical significant
computational requirements and software that requires considerable experience. It
is believed that a 1-D model will be adequate to describe the main differences
between fuel types. relating to conversion time and combustion temperatures. Thus
a 1-D model will be developed here.

The models present in the literature form a good foundation for understanding the
key processes in the cement kiln e.g. combustion, heat transfer and clinker reactions.
Most 1-D models in the literature only deal with fossil fuels combustion. Thus, a
further development of these models to investigate the effects of alternative fuels
co-firing is performed in this thesis. The model will couple the gas phase
temperature to the bed temperature to determine the clinker phase composition.
This has been done before, but it has not been coupled with models for alternative
fuel combustion. The effect of burner settings has only been investigated in one CFD
study and the study by Nørskov [31]. It is deemed relevant to incorporate a model
for the mixing of secondary and primary air, to also investigate the effect of burner
settings on the kiln temperature.

8.2 Modeling of Processes in the Cement Rotary Kiln

Although the cement rotary kiln in itself is fairly simple, it is essentially a giant
rotating cylinder, many complex phenomena occur inside the kiln at the same time.
Hot calcined meal enters at one end and undergoes several different reactions to
form the cement clinker. In addition to the 4 main oxides, (CaO, SiO2, Fe2O3, Al2O3)
other components are also present in the kiln, e.g. MgO, SO3, K2O, Na2O, Cl (using
oxide notation). These combine in different ways forming numerous different
mineral phases. The counterflowing gas has a complex flow pattern with
complicated combustion reactions taking place. In addition, there is heat and mass
transfer between the different phases. It is possible to write an entire PhD thesis on
any of these subjects, such as clinker phase chemistry and kinetics [207],
recirculation of volatiles in the kiln [112], or heat transfer in the kiln [208].
128 8 Cement Kiln Model Development

Consequently, it will be necessary to make several simplifying assumptions in a

model that encompasses several different mechanisms.

A brief introduction to the clinker reactions and the kiln was given in chapter 2.3
and 2.6. The purpose of the following subchapters is to present a more detailed
description of the processes occurring in the kiln. A focus point is the mathematical
equations that describe these processes, and to propose simplified models for each
of these sub-processes. The different models are then combined into a global model
for the cement kiln.

8.3 Model Overview

A sketch of the proposed 1-D kiln model is shown in Figure 8-2, which illustrates the
major mechanisms in the cement kiln, which are accounted for by the model.

Figure 8-2: Sketch of the kiln model showing the main processes accounted for in the
model.

At one end, Boundary 1, of the kiln primary air and fuel is admitted through the kiln
burner with secondary air admitted around the burner. The secondary air will be
entrained into the primary air jet, which causes it to expand, until it fills the entire
kiln cross section. Meanwhile the fuel is heated and will undergo drying,
devolatilization and char combustion. The released gasses, assumed to be CO and
CH4, will then combust in the gas phase.

From the opposite end of the kiln, Boundary 2, the preheated and calcined meal is
admitted, which will flow through the kiln counter current to the gas. The kiln
charge (bed) is heated by the hot gasses, which will initiate and accelerate the
clinker reactions.
8.4 Solids Movement in Kiln 129

It is assumed that fossil fuels will follow the gas flow and stay in suspension in the
kiln, as indicated in the sketch. Larger alternative fuels can drop out of the flame and
burn in the bed. This is not indicated in the sketch.

8.4 Solids Movement in Kiln

The movement of the solids in the rotary kiln can be divided into an axial direction,
which determines the residence time, and a transverse motion, which primarily
governs mixing and heat transfer of the bed.

The transverse motion in a cylinder depends largely on the rotational speed. At low
speeds the motion is characterized as slipping, where the bed slips along the wall.
At very high rotational speeds the material may be centrifuged towards the wall. In
a cement kiln the flow is normally characterized as rolling motion, where there is a
steady discharge from the top of the bed to the lower parts of the kiln [27]. This
forms an active surface layer, where renewal continually occurs, and a passive layer
below the surface. In the passive layer the material is closely packed and rotates
with the kiln. Most of the mixing occurs in the active layer, and mixing is thus
increased by a larger active layer, which can be promoted by e.g. an increased kiln
speed [209].

The regime of bed motion can be estimated based on the rotational Froude number
[27]:

𝝎𝟐 𝒓𝒌
𝑭𝒓 = E 8.1
𝒈

Here ω is the rotational speed [rad/s], rk the kiln diameter, and g the acceleration
due to gravity. Rolling motion has a rotational Froude number between 0.5*10-3 and
0.2*10-1.

The typical geometry of the bed is sketched in Figure 8-3. Normally the angle of
repose, ξ, is between 30-50°, but it increases with rotational speed and is lowered
with increased particle size [14].

The axial motion of material in the kiln is dependent on the slope of the kiln, typically
1-3° [2], and the forward angle caused by the transverse flow, which is induced by
the rotation of the kiln. By assuming a uniform bed height across the kiln the average
residence time can be calculated by [210]:
130 8 Cement Kiln Model Development

𝑳𝑲 𝒔𝒊𝒏 𝝃
𝒕𝒓𝒆𝒔 = E 8.2
𝟐𝝅𝒓𝒌 𝒏𝝍

In the equation, LK is the kiln length, n is the rotational speed [rot/s], ξ the bed angle
of repose [rad], ψ the kiln slope [rad], θ is the angle between the bed and cylinder
center [rad], and rk is the kiln radius.

Figure 8-3: Geometry of a rolling bed indicating the applied nomenclature.

8.5 Clinker Reactions

The study of the chemical kinetics of the cement clinker reactions is a complex task
due to the many species and various reaction pathways in the cement kiln [4]. It will
only be attempted to describe the 5 main reactions, as also described in chapter 2.3.
The equations to describe the extent of clinker reactions will be reviewed in chapter
8.5.1. A simplified approach used in kiln model will be presented in chapter 8.5.2.
The approach for determining appropriate kinetics is presented in chapter 8.5.3.

8.5.1 Equations for the Extent of Clinker Reactions

The decomposition reaction of CaCO3 mainly takes place in the calciner, but the
residual CaCO3 normally constitutes 5-8 % [18]. The remaining decomposition will
occur in the cement kiln. The rate of CaCO3 decomposition is dependent on several
factors, amongst others it is highly dependent on both the temperature and the
partial pressure of CO2 [4].
8.5 Clinker Reactions 131

The decomposition of CaCO3 can, according to Imlach [211], best be described by a

contracting geometry equation, valid for reaction control in cylinders [212]:

𝟏 𝒌
𝟏 − (𝟏 − 𝑿)𝟐 = 𝒕 E 8.3
𝒓𝒑

Here X is the fraction of the limestone decomposed at time t, k is the rate constant
given by an Arrhenius expression, and rp is the particle radius.

The solid-state reactions in the cement rotary kiln are primarily governed by
diffusion. The reaction occurs at the interphase between the crystals, where e.g. CaO
diffuses into SiO2. The rate of formation decreases as the thickness of the product
layer increases [212]. The diffusion reaction can be described by the Jander equation
[212,213]:

𝟏 𝟐 𝒌
(𝟏 − (𝟏 − 𝑿)𝟑 ) = 𝒕 E 8.4
𝒓𝟐𝒑

The equation performs rather well for the initial and intermediate stages of the
decomposition, but fails to accurately describe the later stages, due to the formation
of reaction product around the unreacted core, which inhibits further reaction
[4,212]. The equation by Ginstling and Brounshtein may describe the latter stages
better than the Jander equation [212,214]:

𝟐 𝟐 𝒌
(𝟏 − 𝑿) − (𝟏 − 𝑿)𝟑 = 𝟐 𝒕 E 8.5
𝟑 𝒓𝒑

According to the notation used by Levenspiel [215] this is equivalent to the spherical
shrinking core model for an ash diffusion controlled reaction.

Imlach [211] presents some numbers for activation energy that can be used to
calculate the rate constants of the above equations. Telschow [207] used the Jander
equation to calculate the extent of alite formation and compared model results with
experiments performed in a lab-scale rotary kiln. The deviation to experimental
results was on average around 5 %, but at low heating rates the model overestimates
the alite formation. Chromy [216] has presented kinetic results for the formation of
alite and belite based on isothermal experiments.
132 8 Cement Kiln Model Development

8.5.2 Simplified Equations of Clinker Reactions

A somewhat simpler approach to the modeling of clinker reactions will be used here.
The approach was used by Mastorakos et al. [201] who suggested to model
limestone decomposition as well as the four clinker reactions given in Table 8-1. The
kinetics of the four clinker reactions were estimated by trial and error, until the
computational model results matched the composition at the kiln exit. Mujumdar et
al. [197,198] also used these kinetics and showed that the results compared well to
clinker compositional data from industrial kilns.

Table 8-1: Reactions and kinetics for the model by Mastorakos et al. [201].

Reaction A Unit of A Ea (kJ/mol)

1
1 𝐶𝑎𝐶𝑂3 → 𝐶𝑎𝑂 + 𝐶𝑂2 108 176
𝑠

𝑚3
2 𝐶𝑎𝑂 + 2𝑆𝑖𝑂2 → 𝐶2 𝑆 107 240
𝑘𝑔 ∗ 𝑠

𝑚3
3 𝐶𝑎𝑂 + 𝐶2 𝑆 → 𝐶3 𝑆 109 420
𝑘𝑔 ∗ 𝑠

𝑚3
4 3𝐶𝑎𝑂 + 𝐴𝑙2 𝑂3 → 𝐶3 𝐴 108 310
𝑘𝑔 ∗ 𝑠

8 𝑚6
5 3𝐶𝑎𝑂 + 𝐴𝑙2 𝑂3 + 𝐹𝑒2 𝑂3 → 𝐶4 𝐴𝐹 10 330
𝑘𝑔2 ∗ 𝑠

The kinetics of reaction are given by a pseudo-liquid rate. Instead of considering the
diffusion limitations as discussed in chapter 8.5.1, the reactions rates will be
assumed to depend on the concentration of each clinker mineral or phase.

𝑹𝟏 = 𝒌𝟏 ∗ 𝒀𝑪𝒂𝑪𝑶𝟑 ∗ 𝝆𝒃 E 8.6

𝑹𝟐 = 𝒌𝟐 ∗ 𝒀𝑪𝒂𝑶 ∗ 𝒀𝑺𝒊𝑶𝟐 ∗ 𝝆𝟐𝒃 E 8.7

𝑹𝟑 = 𝒌𝟑 ∗ 𝒀𝑪𝒂𝑶 ∗ 𝒀𝑪𝟐𝑺 ∗ 𝝆𝟐𝒃 E 8.8

𝑹𝟒 = 𝒌𝟒 ∗ 𝒀𝑪𝒂𝑶 ∗ 𝒀𝑨𝒍𝟐𝑶𝟑 ∗ 𝝆𝟐𝒃 E 8.9

𝑹𝟓 = 𝒌𝟓 ∗ 𝒀𝑪𝒂𝑶 ∗ 𝒀𝑨𝒍𝟐𝑶𝟑 ∗ 𝒀𝑭𝒆𝟐𝑶𝟑 ∗ 𝝆𝟑𝒃 E 8.10

8.5 Clinker Reactions 133

In the equations R is the reaction rate [kg/m3 s], k is the rate constant, which is
determined by a standard Arrhenius expression, Y is the mass fraction of the phase,
and ρb is the bulk bed density.

8.5.3 Kinetics of Clinker Reactions

The kinetics proposed by Mastorakos [201] that are presented in Table 8-1 are used
as the basis for the reaction kinetics. However, some modifications are made to
better fit the modeling used here.

The kinetics are adjusted so that the limestone calcination (Reaction 1) mainly
occurs at around 850-900 °C. The kinetics for the formation of belite, aluminate and
ferrite (Reactions 2, 4, and 5) are adjusted so the reactions mainly take place in the
temperature interval 900-1200 °C.

The formation of alite is the most important to adequately model, since alite is the
main phase of interest in the cement clinker. The kinetics are adjusted based on
experimental measurements performed in the laboratory. Two sets of kinetics are
found for the reaction, one relatively fast set that mimics an easy to burn clinker,
and a slower set, which mimics a hard to burn clinker. The details of the derivations
can be found in Appendix B. The final kinetics that are used in the kiln model are
given in Table 8-2.

Table 8-2: Preexponential factors and activation energies to determine rate constants
for the clinker reactions.

k3,1 easy k3,1 hard

k1 k2 k3,2 k4 k5
burnability burnability
A 105 105 3.32*108 1.1*108 2.86*1057 107 107
Unit of A 1 𝑚3 𝑚3 𝑚3 𝑚3 𝑚3 𝑚6
𝑠 𝑘𝑔 ∗ 𝑠 𝑘𝑔 ∗ 𝑠 𝑘𝑔 ∗ 𝑠 𝑘𝑔 ∗ 𝑠 𝑘𝑔 ∗ 𝑠 𝑘𝑔2 ∗ 𝑠
Ea (kJ/mol) 175 260 440 440 187 310 330

In the kiln, alite formation starts to occur as a melt phase is formed above 1200 °C.
In order to limit the reaction of alite formation, so that it mainly occurs above 1200
°C, an effective rate constant is used for reaction 3 as shown in E 8.11. The values to
determine k3,2 are selected to be highly dependent on temperature. This gives the
result that it will severely hinder reactions below 1200 °C, but have negligible effect
when the temperature is above 1200 °C.
134 8 Cement Kiln Model Development

−𝟏
𝟏 𝟏
𝒌𝟑,𝒆𝒇𝒇 =( + ) E 8.11
𝒌𝟑,𝟏 𝒌𝟑,𝟐

It should be mentioned that there are some implications with this simplified
approach that should be considered when analyzing the results.

Due to the way the mathematical model is formulated it is possible for the clinker to
react fully, i.e. obtaining a free lime content of zero. This will not be possible in an
industrial kiln, where the presence of large particles of silica and lime will impose
diffusion limitations to the extent of reaction [25].

In the clinker model, only the main oxides and clinker phases are considered. Minor
components such as MgO, Na2O, and K2O are assumed to be inerts, which do not
affect the clinker composition. The model will thus predict the Bogue composition
[24] at full conversion, which does not necessarily represent the clinker composition
from industrial kilns.

The kinetics of the alite reactions are derived from laboratory experiments where
model clinker is burned at up to 1500 °C for 30 minutes. In an industrial kiln the
residence time at such high temperatures will be much shorter. However, it is
possible to obtain a similar free lime content as in the laboratory. The slower rate of
reaction in the laboratory compared to an industrial kiln is mainly caused by a
higher porosity in the laboratory clinker [207], and a significant mixing effect in the
industrial kiln.

8.5.4 Enthalpy of Clinker Reactions and Melting

The energy consumed and released as the clinker is formed can have a significant
impact on the heat balance in the cement kiln. Most notably the calcination reaction
is very endothermic, while the formation of alite is exothermic. In addition, the
melting of clinker also requires energy. The enthalpies employed in the model are
given in Table 8-3.

Furthermore, evaporation of volatile species such as SO2 and KCl from the clinker
also requires energy, but these effects are not included in the model.
8.6 Burner Air Flow 135

Table 8-3: Enthalpy of the clinker reactions and the melting of clinker [217].

Reaction ΔH Unit of ΔH

kJ
1 𝐶𝑎𝐶𝑂3 → 𝐶𝑎𝑂 + 𝐶𝑂2 1780
𝑘𝑔 𝐶𝑎𝐶𝑂3

kJ
2 𝐶𝑎𝑂 + 2𝑆𝑖𝑂2 → 𝐶2 𝑆 -732
𝑘𝑔 𝐶2 𝑆

kJ
3 𝐶𝑎𝑂 + 𝐶2 𝑆 → 𝐶3 𝑆 59
𝑘𝑔 𝐶3 𝑆

kJ
4 3𝐶𝑎𝑂 + 𝐴𝑙2 𝑂3 → 𝐶3 𝐴 -33.5
𝑘𝑔 𝐶3 𝐴

kJ
5 3𝐶𝑎𝑂 + 𝐴𝑙2 𝑂3 + 𝐹𝑒2 𝑂3 → 𝐶4 𝐴𝐹 -33.8
𝑘𝑔 𝐶4 𝐴𝐹

kJ
Clinker Melting 600
𝑘𝑔 𝑐𝑙𝑖𝑛𝑘𝑒𝑟

8.6 Burner Air Flow

As discussed in chapter 2.8 and 2.9, the distribution of axial air and swirl air is one
of the main ways to control the flame in the cement kiln. One of the main design
parameters of the cement kiln burner is the burner momentum, I, calculated as the
mass flow of primary air multiplied by the axial velocity. The momentum is often
normalized, I’, by the thermal power, Pburner, of the burner and reported in units of
[N/MW].

𝑰 𝒖𝒑𝒓𝒊𝒎 ∗ 𝒎̇𝒑𝒓𝒊𝒎
𝑰′ = = E 8.12
𝑷𝒃𝒖𝒓𝒏𝒆𝒓 𝑷𝒃𝒖𝒓𝒏𝒆𝒓

The burner primary air is what controls the flame shape and intensity. The primary
air creates a high velocity in front of the burner, which together with the
recirculation zones, define the mixing rate with secondary air. The faster the
secondary air is entrained into the fuel stream, the faster the fuel will heat up and
ignite. For conventional fossil fuels, it is normally the mixing rate of oxygen that is
the limiting factor for combustion, rather than chemical kinetics [136]. Therefore, a
more intense mixing of oxygen creates a more efficient combustion process. In this
chapter the theory of the mixing process will be described in further detail.
136 8 Cement Kiln Model Development

8.6.1 Jet Expansion

When a fluid is injected from a nozzle into a larger body filled with a fluid, either
stagnant or co-flowing, the jet mixes with the surrounding fluid, creating turbulence,
while the jet grows outwards. Figure 8-4 shows the principle of the mixing in co-
flow through different regions. Close to the nozzle, in the flow development region
(region 1), the turbulence will grow towards the central axis and a wedge is formed
where the velocity is undiminished at the jet injection velocity. This wedge is called
the potential core. The flow enters the fully developed region (region 2) when the
turbulence has fully penetrated to the central axis and the core disappears. In the
fully developed flow region, the axial velocity obtains a profile as indicated in Figure
8-4, with the highest velocity um along the center and the co-flowing velocity close
to the walls. By the end of region 2, the secondary flow is completely entrained into
the primary jet flow. Under certain conditions a recirculating zone may form in
region 3. In region 4 the flow is fully developed pipe flow [218].

Figure 8-4: Sketch of the development of a turbulent jet in a pipe through different
regions [218]. Region 1: The flow development region, Region 2: Fully developed
region, Region 3: Possible formation of recirculation zones. Region 4: Fully developed
pipe flow. In this figure only, U1 denotes co-flow velocity, U velocity difference (u-U1),
D diameter, b the jet radius. Subscripts 1 denote co-flowing stream, and m the center
value. r, x, and u denote radius, axial distance, and axial velocity (as used in the
remainder of the thesis).

Ricou and Spalding [187] investigated the entrainment into free-flowing jets. They
injected air into a chamber with stagnant air and found that the entrainment could
be described by the equation:
8.6 Burner Air Flow 137

𝟏
𝒎̇ 𝒙 𝝆𝒔𝒆𝒄 𝟐 E 8.13
= 𝑲𝒆𝒏𝒕 ( )
𝒎̇𝟏 𝒅𝒏 𝝆𝒑𝒓𝒊𝒎

They suggested a value of 0.32 for the entrainment constant, Kent, for air mixing with
air. The equation should hold for free jets up to values of 400 for x/dn.

Under combustion conditions the entrainment will be influenced by the change in

temperature and density. It has been suggested to take this into account by adjusting
by the flame density:
𝟏 𝟏
𝒎̇ 𝒙 𝝆𝒔𝒆𝒄 𝟐 𝝆𝒇𝒍 𝟐 E 8.14
= 𝑲𝒆𝒏𝒕 ( ) ( )
𝒎̇𝟏 𝒅𝒏 𝝆𝒑𝒓𝒊𝒎 𝝆𝒑𝒓𝒊𝒎

Here, ṁ denotes the mass flow of the jet. Kent is the entrainment constant, x the
distance from the burner, dn the burner nozzle diameter, and ρ the density. The
subscripts prim, sec, and fl denotes primary flow, secondary flow, and flame,
respectively.

Other effects in combustion may also influence the entrainment, e.g. buoyancy [187].

8.6.2 External Recirculation and the Craya-Curtet Parameter

In a cement kiln the jet is confined by the kiln walls. This may influence mixing by
setting up an external recirculation zone, as sketched in Figure 8-4. A confined jet
will otherwise behave as a free jet up to the point where recirculation occurs [219].

To determine if external recirculation takes place in the kiln the Craya-Curtet

number can be calculated. The parameter describes when recirculation takes place
for confined jets, normally when lower than the range 0.75-0.98 [218]. This also
influences the flame in the cement kiln. When Ct <0.7, short and intense flames are
encountered and when Ct > 1, the flame is long and lazy [27].

The number is defined as [219]:

𝒖𝒌
𝑪𝒕 =
E 8.15
√𝒖𝟐𝒅 − 𝟏 𝒖𝟐𝒌
𝟐

With uk being the kinematic mean velocity and ud the dynamic mean velocity:

𝒖𝟏 𝒓𝟐𝟏 + 𝒖𝟐 (𝒓𝟐𝟐 − 𝒓𝟐𝟏 )

𝒖𝒌 = E 8.16
𝒓𝟐𝟐
138 8 Cement Kiln Model Development

𝒖𝟐𝟏 𝒓𝟐𝟏 + 𝒖𝟐𝟐 (𝒓𝟐𝟐 − 𝒓𝟐𝟏 ) 𝟏 𝟐

𝒖𝒅 = + 𝒖𝟐 E 8.17
𝒓𝟐𝟐 𝟐

Here, u1 and u2 are the velocities at the nozzle/burner (primary air) and of the
surrounding stream (secondary air), respectively. Likewise, r1 denotes the radius of
the nozzle/burner and r2 that of the enclosure. The kinematic velocity is the velocity
that would be obtained if primary and secondary streams were fed uniformly, i.e. an
average velocity. The dynamic velocity is the velocity that the streams would have if
they were mixed to the same momentum, minus the static pressure head of the
secondary stream.

Under combustion conditions, the density of the primary and secondary air streams
will be different due to temperature differences, and the Craya-Curtet parameter
must be adapted to take this into account [219].

8.6.3 Swirling Flows

The above analysis only considers axial flow, but cement kiln burners typically take
advantage of swirl air to improve the flame characteristics. Swirl is induced by
injecting air in a tangential manner, which causes a rotating motion. Swirl can have
a large impact on the flow field, jet expansion, flame size and stability, etc. [144,148].

The intensity of swirl can be characterized by the swirl number [27,135]:

𝟐𝑮𝑿
𝑺= E 8.18
𝑮𝒀 𝟐𝒓𝒏

The flux of angular momentum is defined as:

𝒓𝒏
𝑮𝒀 = 𝟐𝝅 ∫ (𝝆𝒖𝒘)𝒓𝟐 𝒅𝒓 E 8.19
𝟎

The axial thrust is defined as:

𝒓𝒏
𝑮𝑿 = 𝟐𝝅 ∫ (𝒑 + 𝝆𝒖𝟐 )𝒓 𝒅𝒓 E 8.20
𝟎

In the equations rn is the radius of the nozzle/burner, p is the static pressure, u and
w are axial and tangential velocities. In practice it may be difficult to measure the
velocity and pressures needed to calculate the swirl number [144]. Thus, Gupta
[148] proposes an approximation of the swirl flow, which is simpler to compute. It
8.6 Burner Air Flow 139

is calculated based on the maximum axial and tangential velocities at exit of the
burner nozzles:

𝟏 𝒘𝒆𝒙
𝟐 𝒖𝒆𝒙
𝑺𝒂𝒑𝒑𝒓𝒐𝒙 = E 8.21
𝟏𝒘
𝟏 − 𝟐 𝒖 𝒆𝒙
𝒆𝒙

At low swirl intensities (S < ~0.4) the main effect of adding swirl is that the primary
jet expands faster due to increased entrainment. At higher degrees of swirl (S >
~0.6), strong axial and radial pressure gradients are set up near the burner. This
creates a zone of internal recirculation in front of the burner, which quickly expands
the jet, but also slows it. Other effects that may affect the recirculation zone is the
design of nozzles and enclosure size [148]. The recirculation zone mixes hot
combustion gasses back to fresh fuel, which heats it up leading to earlier ignition. It
also slows down the forward movement of fuel, which keeps the flame stabilized
close to the burner. Both factors contribute to reduce flame length and increase
intensity. The swirl can thus be used to influence the temperature and gas
composition, which also affects the formation of pollutants, e.g. NOx [148].

According to Nørskov [31] the entrainment constant for swirling flows can be
described as:

𝑲𝒆𝒏𝒕 = 𝟎. 𝟑𝟐 + 𝟎. 𝟖𝑺 E 8.22

Analyzing the results published by Park and Shin [220] (regression of their
published data) the entrainment rate can be determined as:

𝑲𝒆𝒏𝒕 = 𝟎. 𝟐𝟐 + 𝟎. 𝟑𝟖𝑺 E 8.23

8.6.4 Air Flow in the Cement Kiln

It is inherently difficult to measure inside a large scale industrial rotary kiln as
already outlined in chapter 4. Traditionally, the mixing in the cement kiln has been
studied using down-scaled cold models. This can be compared to large scale
operation, if the model is scaled appropriately to e.g. keep Reynolds number equal.
It is not straightforward to translate the model results to information of the full scale
system, since e.g. density differences are introduced by the combustion [221,222].
More recently advanced CFD models are used to a higher degree. Some interesting
studies on the mixing in cement kilns are outlined in this section.
140 8 Cement Kiln Model Development

Moles et al. [136] made a study of a number of wet kilns to determine typical ranges
of excess air, gas velocities and the Craya-Curtet number. They proceeded to study
the flow in a scaled down kiln model using water and air flow. In addition, large scale
measurements were attempted to validate the flow models. It was found that the gas
flow through the cooler and cooler hood had an impact on the gas flow in the kiln,
which creates a vortex under the burner and tends to push the flame downwards.

Ruhland [223] used acid-alkali modeling to study the flame in a cement kiln model,
as shown in Figure 8-5. An alkaline solution containing a colored indicator is injected
as primary air into a kiln filled with acid that acts as secondary air. When the two
liquids mix, the pH is neutralized, and the color removed. This gives an estimate of
the mixing length in the model. Ruhland found a formula for describing the mixing
length produced by a smooth non-swirled nozzle. The mixing length is equivalent to
the flame length for fossil fuels, where the combustion is mainly limited by the
oxygen availability.
𝟎.𝟒𝟒𝟐
𝑳𝒇𝒍 𝟐 𝟏
= √𝑲 [𝟑. 𝟐𝟏 ( + 𝑩𝒑 ) + 𝟑. 𝟖𝟔𝟐 ( ) ∗ 𝒆𝒙𝒑(𝒂 + 𝒃)] E 8.24
𝒅𝒏 𝟑 𝑩𝒑,𝒆𝒙𝒄 − 𝟏

With the following definitions used:

𝒎̇𝒑𝒓𝒊𝒎 + 𝒎̇𝒔𝒆𝒄
𝑲=
𝒎̇𝒑𝒓𝒊𝒎 𝒎̇𝑠𝑒𝑐 E 8.25
(𝝆 + 𝝆 ) 𝝆𝒇𝒍
𝒑𝒓𝒊𝒎 𝑠𝑒𝑐

𝟏.𝟐𝟒𝟓
𝒎̇𝒔𝒆𝒄 𝒅𝒏
𝒂 = 𝟐. 𝟏𝟐 ( ) E 8.26
𝒎̇𝒑𝒓𝒊𝒎 𝒅𝒌 − 𝒅𝒏

𝒅𝒌 − 𝒅𝒏
𝒃 = 𝟎. 𝟏𝟎𝟓𝟐 E 8.27
𝒅𝒏

In the equations, the following definitions are used: Lfl flame length, dn nozzle
diameter, Bp secondary air requirement for fuel combustion [kg/kg fuel], Bp,exc
excess air ratio for the secondary air, ṁ mass flow, ρ density, dk kiln diameter. The
subscripts prim, sec, and fl, indicate primary air, secondary air, and flame,
respectively.
8.6 Burner Air Flow 141

Figure 8-5: Examples of acid-alkali mixing used to investigate the kiln flame. The dark
plume (alkali) represents the cement kiln flame. The dark color dissipates as the alkali
is mixed with acid, representing the secondary air. Top and bottom images are single
frames and middle picture is averaged over 60 seconds [223].

Bhad [224] modeled a coal fired cement kiln using CFD. The study focused on
different levels of swirl. With a high swirl the flame length reduces and ignites
earlier. The flame also becomes wider, which can overheat the refractory.

Orfanoudakis [225] studied a scaled down rotary kiln burner in the laboratory. The
effect of swirl was studied for coal and gas fired flames with both experiments and
CFD. The results showed that an internal recirculation zone was formed at swirl
numbers above 0.65. Increasing the swirl number further also increased the width
of the recirculation zone. Especially at high swirl numbers, coal particles can be
centrifuged out of the internal recirculation zone (IRZ), which may increase NOX
emissions.

8.6.5 Model for the Secondary Air Entrainment

In order to formulate some simple rules for the mixing of secondary and primary air
that can be used in the 1-D model, a number of CFD simulations were performed.
These simulations have been carried out by David Jayanth Isaac of FLSmidth. The
simulations are made on a petcoke flame from the FLSmidth Jetflex burner in a full-
scale cement plant. The burner with a diameter of 0.64 m and the kiln with a
diameter of 5 m. A thorough description of similar simulations for co-firing petcoke
and SRF using the same kiln geometry and burner can be found in [204].
142 8 Cement Kiln Model Development

Six different simulations with varying swirl and axial momentum were made
according to Table 8-4. The axial momentum and swirl number are increased by
increasing the flow rate of axial and swirl air. The swirl air flow rate is set at two
levels 50 and 100 %, which results in swirl numbers of ~0.2 and ~0.35. The swirl
number and the simplified approximate swirl number were calculated using
equations E 8.18 and E 8.21.

Table 8-4: Momentum and swirl number in CFD simulations, performed by David
Jayanth Isaac, used to determine secondary air entrainment.

Swirl Approximated
Momentum Swirl Level Kent,jet by
Number Swirl Number Kent,jet
(N/MW) (%) E 8.30
(E 8.18) (E 8.21)
5.3 50 0.18 0.25 0.186 0.188
6.0 100 0.35 0.48 0.237 0.235
7.0 50 0.20 0.26 0.203 0.200
8.0 100 0.36 0.47 0.243 0.247
9.0 50 0.20 0.25 0.208 0.209
10.0 100 0.35 0.45 0.255 0.254

The jet radius was found in the simulations as the radial distance where the axial
velocity of the jet becomes equal to the inlet secondary air velocity. The jet radius is
plotted for the 6 simulation cases in Figure 8-6.

Figure 8-6: a) The jet radius as function of axial distance from burner with different
burner axial momentum and swirl. b) Zoomed version of a, with regression lines. Data
derived from CFD simulations of a petcoke flame performed by David Isaac.

The jet in Figure 8-6 is seen to initially, between 1 and 5 meters, expand linearly.
The rate at which this occurs is increased by both axial momentum and swirl. At
around 7 meters the location of an external recirculation zone inhibits the further
8.7 Heat Transfer in the Cement Kiln 143

expansion of the jet. At a distance of 20 meters, the kiln diameter is expanded, which
explains the odd shape of the curves around this point. The kiln diameter is
constricted near the burner, since it has been assumed that a kiln coating is present.

For the model developed in this study, it will be assumed that the jet expands
linearly as seen in Figure 8-6b. No recirculating zone is assumed to be present, thus
the jet will expand until it reaches the kiln wall following the equation:

𝒓𝒋𝒆𝒕 = 𝒓𝒏 + 𝑲𝒆𝒏𝒕,𝒋𝒆𝒕 ∗ 𝒙 E 8.28

And the mass fraction of secondary air entrained into the primary air jet will be
proportional to the jet radius:
𝒓𝒋𝒆𝒕 − 𝒓𝒏
𝒀𝒆𝒏𝒕,𝒔𝒆𝒄 = E 8.29
𝒓𝒌 − 𝒓𝒏

The entrainment constant, Kent,jet, is found based on the jet expansion rate found in
Figure 8-6b. The slope of each data set is found between 1-5 m, where the graphs
are linear. The slopes, i.e. entrainment constants, for the six different simulation
cases can be found in Table 8-4. A multilinear regression is then employed to
determine the relationship between burner momentum, swirl, and the entrainment
constant:

𝑲𝒆𝒏𝒕,𝒋𝒆𝒕 = 𝟎. 𝟏𝟏𝟔 + 𝟒. 𝟔𝟗 ∗ 𝟏𝟎−𝟑 ∗ 𝑰′ + 𝟎. 𝟐𝟓𝟗 ∗ 𝑺 E 8.30

Here I’ [N/MW] is the burner momentum and S is the swirl number. Both swirl and
axial momentum is according to this regression model serving to increase the
entrainment of secondary air into the jet, which expands the jet radius. The numbers
of the entrainment constant calculated using this equation can be found in Table 8-4.

8.7 Heat Transfer in the Cement Kiln

The heat transfer in the cement kiln is of very high importance as the main goal of
the kiln is to heat the cement raw materials to a sufficiently high temperature for the
clinker reactions to occur. Some attempts to model the heat transfer was already
highlighted in chapter 8.1, but a more thorough evaluation of the different
mechanisms will be covered here.

In the cement kiln, several modes of heat transfer occur, as sketched in Figure 8-7.
The hot gasses exchange heat with the clinker bed and exposed wall through
radiation and convection. The wall radiates heat to the bed, and where the wall is in
144 8 Cement Kiln Model Development

direct contact with the clinker bed, conduction also occurs [196]. In addition, there
is a heat loss to the surroundings by conduction through the kiln wall. The exact loss
is difficult to estimate due to the buildup of the kiln coating of varying thickness on
the inside of the kiln [31]. Finally there is also heat transfer within the clinker bed
itself [27], governed by both conduction, radiation and gas convection between the
particles and motion of the particle bed [185], where the particle motion will
typically dominate the heat transfer [27]. There are indications that the bed may be
close to isothermal (at least in pilot scale) due to the good mixing of the bed [185].
Lehmberg et al. [226] report that mixing is typically complete within 4 rotations. If
segregation occurs in the bed, due to significant differences in particle sizes. Boateng
and Barr [227] modeled the bed core to be around 200 °C colder than the bed surface
(with freeboard temperatures of 800 °C). Under the same operating conditions, the
bed was largely isothermal for a more homogenous particle size distribution.

Figure 8-7: Paths of heat transfer in the cement rotary kiln.

In a cement rotary kiln two main heat transfer zones are present; the flame zone
near the burner flame and the non-flame zone further downstream in the kiln. In the
flame zone the radiating gases are confined to the visible flame at high temperatures,
while they further downstream expand to occupy the entire freeboard, but also
cools down. Thus, in the flame region, heat is primarily transferred by radiation from
the flame, while convection plays a minor role. In the non-flame zone convection
plays a larger role [183]. Due to the high temperatures in the kiln, radiative heat
transfer is the dominating form of heat transfer [181,196], constituting around 90
% of the total heat transferred from the gas to bed [27]. Heat transfer in the radial
8.7 Heat Transfer in the Cement Kiln 145

direction is predominant, while radiation in the axial direction of the kiln is

negligible, even though large temperature differences can occur [181,183,185]. This
is caused by a low transparency of the gas to its own radiation.

8.7.1 Convective Heat Transfer

The convective heat transfer between the gas and the bed or the wall can be
described by the following relation:

𝒒𝒄𝒐𝒏𝒗,𝒊𝒋 = 𝒉𝒄𝒐𝒏𝒗,𝒊𝒋 ∗ (𝑻𝒊 − 𝑻𝒋 ) E 8.31

Here the convective heat transfer coefficient between gas and bed can be described
by an empirical relation found by Tscheng and Watkinson, based on experiments in
a pilot scale rotary kiln [228]:

𝝀𝒈 𝝀𝒈
𝒉𝒄𝒐𝒏𝒗,𝒈𝒃 = ∗ 𝑵𝒖𝒄𝒐𝒏𝒗,𝒈𝒃 = ∗ 𝟎. 𝟒𝟔 ∗ 𝑹𝒆𝟎.𝟓𝟑𝟓 ∗ 𝑹𝒆𝟎.𝟏𝟎𝟒
𝝎 ∗ 𝜼−𝟎.𝟑𝟒𝟏 E 8.32
𝒅𝒆 𝒅𝒆

Here the Reynolds number and the rotational Reynolds number are defined as:

𝒖𝒈 ∗ 𝒅𝒆
𝑹𝒆 = E 8.33
𝝂𝒈

𝒅𝟐𝒆 ∗ 𝝎
𝑹𝒆𝝎 = E 8.34
𝝂𝒈

ug is the axial velocity of gas, ω is the kiln angular velocity [rad/s], νg is the kinematic
viscosity of gas, λg is the gas thermal conductivity, and η is the kiln fill degree. de is
the equivalent diameter determined as [228]:

𝟐 ∗ 𝝅 − 𝜽 ∗ 𝒔𝒊𝒏 𝜽
𝒅𝒆 = 𝒓𝒌 ∗ E 8.35
𝜽 𝜽
𝝅 − 𝟐 + 𝒔𝒊𝒏 𝟐

With rk being the kiln radius and θ the central angle subtended by the kiln bed (see
Figure 8-3).

The convective heat transfer coefficient between the gas and exposed walls is
determined by [228]:

𝝀𝒈 𝝀𝒈
𝒉𝒄𝒐𝒏𝒗,𝒈𝒘 = 𝑵𝒖𝒄𝒐𝒏𝒗,𝒈𝒘 = ∗ 𝟏. 𝟓𝟒 ∗ 𝑹𝒆𝟎.𝟓𝟕𝟓 ∗ 𝑹𝒆−𝟎.𝟐𝟗𝟐
𝝎 E 8.36
𝑫𝒆 𝑫𝒆

The porous layer of rolling particles at the bed surface is commonly believed to
increase the convective heat transfer to the bed over that of the exposed wall.
146 8 Cement Kiln Model Development

However, in one study the effect was not very pronounced, and the convective heat
transfer coefficients from gas to bed were in the range 1-2 times the value from gas
to exposed wall [184].

8.7.2 Conductive Heat Transfer

Where there is contact between the kiln bed and the wall, heat is transferred
through conduction according to:

𝒒𝒄𝒐𝒏𝒗,𝒘𝒃 = 𝒉𝒄𝒐𝒏𝒅,𝒘𝒃 ∗ (𝑻𝒘 − 𝑻𝒃 ) E 8.37

The heat transfer coefficient for conduction has also been investigated by Tscheng
and Watkinson [228] and can be determined by:

𝝀𝒃 𝟐 𝝆𝒃 ∗ 𝑪𝒑𝒃 𝟎.𝟑
𝒉𝒄𝒐𝒏𝒅,𝒘𝒃 = 𝟏𝟏. 𝟔 ∗ ∗ (𝒏 ∗ 𝒓𝒌 ∗ 𝜽 ∗ ) E 8.38
𝒍𝒘𝒃 𝝀𝒃

Here n is the kiln rotational speed [revolutions/s], lwb is the covered wall
circumference [m] where the bed touches the wall. λ, Cp, and ρ are the heat
conductivity, heat capacity, and density, respectively.

Pilot scale measurements in a gas fired rotary kiln have shown that the regenerative
heat transfer from the covered wall to the bed can account for up to 50 % of the net
energy to the bed, although it normally is lower (around 10-20 %) [184]. The
amount is dependent on the location in the kiln and tends to be highest near the
material inlet. In the flame zone a larger part of the energy is transferred from the
flame to the exposed kiln bed, and there tends to be a net heat transfer from the bed
to the covered wall.

8.7.3 Radiative Heat Transfer

It will be assumed that the radiative heat transfer takes place between the three
surfaces of flame jet, kiln bed, and kiln walls and only occurs in the radial direction.
This assumption significantly simplifies the model calculations, since the local heat
transfer will not be dependent on upstream or downstream conditions. The
radiation in the axial direction is only around 10-20 % of the radiation in radial
direction in the hottest part of the flame [31,183].

For radiation between N diffuse grey surfaces the radiation exchange can be written
as N equations of the form [229]:
8.7 Heat Transfer in the Cement Kiln 147

𝑵 𝑵
𝜹𝒊𝒋 𝟏 − 𝝐𝒋
∑ ( − 𝜴𝒌𝒋 ∗ ) 𝒒𝒋 = ∑(𝜹𝒊𝒋 − 𝜴𝒊𝒋 ) ∗ 𝝈𝑻𝟒𝒋
𝝐𝒋 𝝐𝒋
𝒋=𝟏 𝒋=𝟏
E 8.39
𝑵

= ∑ 𝜴𝒌𝒋 𝝈 ∗ (𝑻𝟒𝒊 − 𝑻𝟒𝒋 )

𝒋=𝟏

In the equation δij is the Kronecker delta, ε the emissivity, q the heat transfer flux, T
the temperature, σ the Stefan-Boltzmann constant, and Ω is the view factor between
two surfaces.

The view factor for radiation can be determined by Hottel’s crossed strings method
[229] described in Appendix C. In addition, there will be radiative heat transfer to
the fuel, which will be described in chapter 8.8.1.

8.7.4 Heat Loss

In the kiln there will be a significant heat loss of typically 10-20 % of the total energy
input. Heat is transferred through the kiln walls and to the colder ambient
environment outside the kiln. To minimize the heat loss, the kiln is lined with
refractory bricks. A kiln coating, formed of molten and then solidified clinker, is also
typically present in the clinker burning zone, which further limits the heat loss. The
thickness of the coating will have a great effect on the heat loss but it is difficult to
determine the exact thickness [31]. The outside temperature of the kiln shell is
typically monitored by a kiln scanner, as discussed in chapter 4.6 [133].

For model purposes it will be assumed that a kiln coating is present near the burner
end of the kiln. The thickness and extent of the coating can be adjusted in the model.
The default values are a thickness of 25 cm, during the first 15 meters of the kiln
(measured from the burner end), which afterwards linearly decreases to 0 cm for
the next 5 meters.

For the model it will be assumed that there is thermal equilibrium between the heat
transferred to the wall and the heat transferred through the wall as a loss:
𝑵𝒉𝒕

𝒒𝒍𝒐𝒔𝒔 = ∑ 𝒒𝒊 = 𝒒𝒓𝒂𝒅,𝒈𝒘 + 𝒒𝒓𝒂𝒅,𝒃𝒘 + 𝒒𝒄𝒐𝒏𝒗,𝒈𝒘 + 𝒒𝒄𝒐𝒏𝒅,𝒃𝒘 E 8.40

𝒊

The heat transfer to the wall is given by radiation from the gas and bed, convection
from the gas, and conduction from the bed.
148 8 Cement Kiln Model Development

It will be assumed that heat is first transferred through the coating (if it is present),
then the kiln refractory, and at the outer surface it is transferred to the surroundings
due to convection and radiation. The heat conduction through the kiln steel shell
will be neglected here, since the resistance will likely be small compared to that of
the thicker kiln coating and refractory.

The heat conduction through the coating:

𝑻𝒘 − 𝑻𝒄𝒐𝒂𝒕
𝑸𝒄𝒐𝒂𝒕 = 𝝀𝒄𝒐𝒂𝒕 ( 𝒓𝒌,𝒊 ) ∗ 𝟐𝝅𝜟𝒙 E 8.41
𝒍𝒏 ( )
𝒓𝒌,𝒊 − 𝒍𝒄𝒐𝒂𝒕

The heat conduction through the kiln refractory:

𝑻𝒄𝒐𝒂𝒕 − 𝑻𝒔𝒉
𝑸𝒘 = 𝝀𝒘 ( 𝒓𝒌,𝒐 ) ∗ 𝟐𝝅𝜟𝒙 E 8.42
𝒍𝒏 ( 𝒓 )
𝒌,𝒊

And at the kiln shell surface energy is lost by convection and radiation:

𝑸𝒔𝒉 = (𝒒𝒓𝒂𝒅,𝒍𝒐𝒔𝒔 + 𝒒𝒄𝒐𝒏𝒗,𝒍𝒐𝒔𝒔 ) ∗ 𝟐𝝅𝒓𝒌,𝒐 𝜟𝒙 E 8.43

The radiative loss and convective losses are given as:

𝒒𝒓𝒂𝒅,𝒍𝒐𝒔𝒔 = 𝝈𝝐𝒔𝒉 ∗ (𝑻𝟒𝒔𝒉 − 𝑻𝟒𝒂𝒎𝒃 ) E 8.44

𝒒𝒄𝒐𝒏𝒗,𝒍𝒐𝒔𝒔 = 𝒉𝒄𝒐𝒏𝒗,𝒔𝒉 ∗ (𝑻𝒔𝒉 − 𝑻𝒂𝒎𝒃 ) E 8.45

It is possible to determine the convective heat transfer coefficient based on the

Nusselt number and the outer kiln diameter:

𝒅𝒌,𝒐
𝒉𝒄𝒐𝒏𝒗,𝒔𝒉 = 𝑵𝒖𝒄𝒐𝒏𝒗,𝒔𝒉 ∗ E 8.46
𝝀𝒂𝒎𝒃

Where the Nusselt number is determined based on the rotational Reynolds number,
the Grashof number, and the Prandtl number as shown in E 8.47 [230]. The equation
is valid in quiescent media, while the potential effect of wind speed was neglected.
𝟎.𝟑𝟓
𝑵𝒖𝒄𝒐𝒏𝒗,𝒔𝒉 = 𝟎. 𝟏𝟏 ∗ ((𝟎. 𝟓 ∗ 𝑹𝒆𝟐𝒘 + 𝑮𝒓) ∗ 𝑷𝒓) E 8.47

With the following definitions of the dimensionless numbers:

𝝎
𝑹𝒆𝒘 = 𝒅𝟐𝒌,𝒐 E 8.48
𝟐𝝂𝒈,𝒂𝒎𝒃
8.8 Combustion of Solid Fuels 149

𝟏 𝟏
𝑮𝒓 = 𝒈 ∗ 𝒅𝟑𝒌,𝒐 ∗ ∗ (𝑻𝒔𝒉 − 𝑻𝒂𝒎𝒃 ) ∗ E 8.49
𝑻𝒂𝒎𝒃 𝝂𝟐𝒈,𝒂𝒎𝒃

𝝁𝒈,𝒂𝒎𝒃 𝑪𝒑𝒈,𝒂𝒎𝒃
𝑷𝒓 = E 8.50
𝝀𝒈,𝒂𝒎𝒃

In the calculation of the Grashof number it has been assumed that the coefficient of
thermal expansion is equal to 1/Tamb, which is true for ideal gasses.

Since the heat transfer through the coating and refractory must be equal, it is
possible to eliminate the unknown temperature Tcoat from the equations E 8.41 and
E 8.42, eventually yielding:

𝒓𝒌,𝒊 𝒓𝒌,𝒐 𝑵𝒉𝒕

𝒍𝒏 ( ) 𝒍𝒏 (
𝟏 𝒓𝒌,𝒊 − 𝒍𝒄𝒐𝒂𝒕 𝒓𝒌,𝒊 )
𝑻𝒘 = ( ∗( + ) ∗ ∑ 𝒒𝒊 ∗ 𝒍𝒊 ) + 𝑻𝒔𝒉 E 8.51
𝟐𝝅 𝝀𝒄𝒐𝒂𝒕 𝝀𝒘
𝒊

The shell temperature, Tsh, can be eliminated from the equation by use of the
equations E 8.43-E 8.45. This is not trivial to solve due to the 4th power exponent in
the radiation term, but it can be done numerically.

8.8 Combustion of Solid Fuels

Combustion of solid fuels conceptually follow the four steps heating, drying,
devolatilization, and char oxidation. Subsequently the gasses released during the
devolatilization will burn in the gas phase. A brief review of the different
mechanisms will be presented along with the equations applied in the model.

8.8.1 Heating
When the fuel particles are injected into the kiln, they are heated as they are exposed
to the hot secondary air and radiation from the environment. Small particles such as
pulverized petcoke or coal, which generally have a particle diameter below 200 µm,
can normally be considered isothermal. In larger particles, such as from alternative
fuels with particle sizes in the range 1-10 mm, heat transfer inside the particle may
be lower than the heat transfer to the particle, which creates significant temperature
gradients inside the particle. To evaluate if the particle can be considered
isothermal, the Biot number can be evaluated [231]. This number is the ratio
between external heat transfer and internal heat transfer in the fuel particle.
150 8 Cement Kiln Model Development

𝒉𝒆𝒙𝒕 ∗ 𝒓𝒑
𝑩𝒊 = E 8.52
𝝀𝒑

In the equation, hext is the external heat transfer coefficient, rp the particle radius,
and λp is the particle heat conductivity. Biot numbers lower than 0.2 indicates that
the particle can be assumed isothermal [231].

The particle temperature for small isothermal particles can be determined by the
following equation:

𝒅𝑻𝒑 𝟏 𝟏 𝟏 𝒅𝑿𝒅𝒓𝒚
= ∗ ( 𝑨𝒑 ∗ 𝒒𝒆𝒙𝒕 − 𝒎𝒑,𝟎 ∗ 𝒀𝑯𝟐𝑶,𝟎 ∗ 𝜟𝑯𝒆𝒗𝒂𝒑 ) E 8.53
𝒅𝒕 𝒎𝒑 𝑪𝒑𝒑 𝝋𝒑 𝒅𝒕

The first term on the right-hand side of the equations accounts for the heat transfer
to the particle, while the second term accounts for the loss of energy due to
evaporation of water from the particle. It is assumed that the devolatilization has a
net energy consumption of 0, and thus does not contribute to the particle heating.
Furthermore, it is assumed that any energy released during volatile and char
combustion is released solely to the gas phase.

For plastic particles, which are present in SRF, it is assumed that melting and
decomposition of the polymer requires some energy, as proposed by Nakhaei et al.
[232], which is added to the temperature equation.

𝒅𝑻𝒑 𝟏 𝟏 𝟏 𝒅𝑿𝒅𝒓𝒚
= ∗ ( 𝑨𝒑 ∗ 𝒒𝒆𝒙𝒕 − 𝒎𝒑,𝟎 ∗ (𝒀𝑯𝟐𝑶,𝟎 ∗ 𝜟𝑯𝒆𝒗𝒂𝒑
𝒅𝒕 𝒎𝒑 𝑪𝒑𝒑 𝝋𝒑 𝒅𝒕
E 8.54
𝒅𝑿𝒅𝒆𝒗𝒐𝒍 𝒅𝑿𝒎𝒆𝒍𝒕
+ 𝒀𝒗𝒐𝒍,𝟎 ∗ 𝜟𝑯𝒅𝒆𝒄𝒐𝒎𝒑 + 𝜟𝑯𝒎𝒆𝒍𝒕 )
𝒅𝒕 𝒅𝒕

According to calculations performed by Nakhaei et al. [232], an isothermal model

for polyethylene predicts devolatilization times within a 20 % error compared to a
more rigorous 1D-model. Since the polymer evaporates as a droplet, temperature
gradients in the particle are limited. Thus, the isothermal particle model will be used
here.

For other particle types, that form a char, non-isothermal effects are more
important. For non-isothermal particles, a 1-D model will be applied to take into
consideration the temperature gradients [233]:
8.8 Combustion of Solid Fuels 151

𝝏𝑻𝒑 𝝀𝒑 𝟏 𝟐 𝝏𝑻𝒑 𝝏𝟐 𝑻𝒑 𝝆𝒑,𝟎 𝜟𝑯𝒆𝒗𝒂𝒑 ∗ 𝒀𝑯𝟐𝑶,𝟎 𝝏𝑿𝒅𝒓𝒚

= ( + 𝟐
)− E 8.55
𝝏𝒕 𝝆𝒑 ∗ 𝑪𝒑𝒑 𝝋𝒑 𝒓 𝝏𝒓 𝝏𝒓 𝝆𝒑 𝑪𝒑𝒑 𝝏𝒕

The boundary conditions for the partial differential equation for heat transfer in
large particles are given in E 8.56 and E 8.57. They indicate the heat transfer to the
particle and that the temperature gradient at the center is 0, due to symmetry.

𝝏𝑻𝒑 (𝒕, 𝒓 = 𝒓𝒑 )
𝝀𝒑 = 𝒒𝒆𝒙𝒕 E 8.56
𝝏𝒓
𝝏𝑻𝒑 (𝒕, 𝒓 = 𝟎)
=𝟎 E 8.57
𝝏𝒓

The partial differential equation is solved by discretizing into a number of coupled

ordinary differential equations using the Method of Lines [234].

The external heat transfer to the particles is divided into convection and radiation.
The convective heat transfer between gas and particle is given by [235]:

𝒒𝒄𝒐𝒏𝒗,𝒈𝒑 = 𝒉𝒄𝒐𝒏𝒗,𝒈𝒑 ∗ (𝑻𝒈 − 𝑻𝒑,𝒔 ) E 8.58

The heat transfer coefficient is determined based on the Nusselt number which is
found using the Ranz Marshall correlation [235]:

𝒉𝒄𝒐𝒏𝒗,𝒈𝒑 ∗ 𝟐𝒓𝒑 𝟏 𝟏
𝑵𝒖 = = 𝟐 + 𝟎. 𝟔𝑹𝒆𝟐 𝑷𝒓𝟑 E 8.59
𝝀𝒈

The radiative heat transfer to the particle from the gas is determined by the equation
[31]:

𝝐𝒑 + 𝟏
𝒒𝒓𝒂𝒅,𝒈𝒇 = 𝝈 (𝝐𝒈 𝑻𝟒𝒈 − 𝜶𝒈 𝑻𝟒𝒑,𝒔 ) E 8.60
𝟐

In the equation, σ is the Stefan-Boltzmann constant, ε the emissivity, α the

absorptivity, and, T the temperature.

It is also assumed that there is radiation between the bed and the fuel particles
according to:

𝟏
𝒒𝒓𝒂𝒅,𝒃𝒇 = ∗ 𝝉 𝝈(𝝐𝒇 𝑻𝟒𝒇 − 𝝐𝒃 𝑻4𝒃 ) E 8.61
𝑲𝒓𝒂𝒅 𝒈

The radiative exchange is limited by the factor τg, which is the transmissivity of the
gas phase. Furthermore, the factor Krad limits radiation to the fuel particles. Since
radiation will only occur to the fuel particles at the surface of the flame, it was found
152 8 Cement Kiln Model Development

to be necessary to further limit the radiation. A factor of 2 was found to be

appropriate for the model validation in pilot scale, while a factor 100 was used for
the industrial kiln as described in chapter 9. In the industrial kiln the fuel mass flow
is larger relative to the kiln cross section, which leaves less particles exposed,
justifying the larger correction factor.

8.8.2 Drying
When the fuel particle approaches a temperature around 100 °C moisture in the
particle will begin to evaporate. Evaporation of water requires a lot of energy [236],
and the fuel heating and ignition can thus be delayed by a high moisture content.
This is especially of interest in alternative fuels where e.g. the moisture content in
SRF can range from 10-40 % (see Table 3-3). In the cement industry pulverized
fossil fuel normally has a low moisture content, since the fuel is dried during milling.
This makes the drying step less important in the combustion sequence for fossil
fuels.

Different models are used in the literature to model the evaporation of water in
combustion [237]. Here a simple kinetic model based on a first order Arrhenius
expression will be used for the evaporation [236]:

𝒅𝑿𝒅𝒓𝒚 −𝑬𝑨,𝒅𝒓𝒚
= 𝑨𝒅𝒓𝒚 ∗ 𝒆𝒙𝒑 ( ) (𝟏 − 𝑿𝒅𝒓𝒚 ) E 8.62
𝒅𝒕 𝑹𝒈𝒂𝒔 𝑻𝑷

The conversion, Xdry, is the mass-based fraction of initial water content that has
evaporated. Thus, Xdry is 0 when the particle is wet and 1 when it is dry. Adry is the
pre-exponential factor for drying, Ea,dry the activation energy, Rgas the ideal gas
constant and Tp the particle temperature.

8.8.3 Devolatilization
As the fuel is heated further to temperatures exceeding 200 °C, the solid begins to
decompose into smaller volatile molecules that leave the solid matrix. For coal the
process is divided into three stages. Loosely bound H2O and CO2 is released below
350 °C. Primary devolatilization takes place due to rupture of chemical bonds in the
temperature region 350-550 °C. Secondary devolatilization occurs in the
temperature range 600-800 °C, breaking the strongest covalent bonds, forming
primarily H2 and CH4 [31,238]. Biomass primarily devolatilizes at temperatures
8.8 Combustion of Solid Fuels 153

around 200-500 °C, depending on the content of cellulose, hemicellulose, and lignin
[239]. For SRF the paper fractions primarily behaves as biomass and devolatilizes at
300-400 °C, and plastics at 400-500 °C [164].

The products formed from the devolatilization are normally divided into 3 groups
being gasses, tars (i.e. condensable species), and char [85]. A major difference in the
combustion behavior of fossil fuels and SRF is the amount of volatiles formed (see
Table 3-4). SRF for instance has a volatile content around 70 %, while coal has a
volatile content of approximately half that, and petcoke has an even lower volatile
content. In SRF, plastics, e.g. polyethylene and polypropylene, are almost fully
volatile, while the biomass fraction, e.g. paper and cardboard, has a volatile content
around 80 %.

In addition to the type of fuel and its chemical composition, physical conditions
during the devolatilization also has an impact. The rate of devolatilization and the
products formed during it, depend on e.g. temperature, heating rate, particle size,
and catalytic activity of ash components [85,238].

In this work a single global first order reaction will be assumed to account for the
devolatilization of the fuel.

𝒅𝑿𝒅𝒆𝒗𝒐𝒍 −𝑬𝑨,𝒅𝒆𝒗𝒐𝒍
= 𝑨𝒅𝒆𝒗𝒐𝒍 ∗ 𝒆𝒙𝒑 ( ) (𝟏 − 𝑿𝒅𝒆𝒗𝒐𝒍 ) E 8.63
𝒅𝒕 𝑹𝒈𝒂𝒔 𝑻𝑷

More advanced models can be found in the literature as e.g. discussed in [237,239].

8.8.3.1 Products of Devolatilization

The major products from biomass and coal devolatilization are CO2, CO, and CH4,
while smaller quantities of H2 and C2 hydrocarbons can also be formed [85,240]. As
a simplification, here it will be assumed that CO and CH4 are the only products of
devolatilization. This approach involves less species in the model, while it is still
possible to adjust the heating value of the volatiles by adjusting the ratio between
formed CO and CH4.

The split of the two are found based on a mass and energy balance based on the
proximate analysis and the heating value of the fuel. The proximate analysis of a fuel
states the moisture, ash, fixed carbon (char), and volatile content. Thus, the mass
fractions sum to 1:
154 8 Cement Kiln Model Development

𝒀𝑯𝟐𝑶 + 𝒀𝒂𝒔𝒉 + 𝒀𝒄𝒉𝒂𝒓 + 𝒀𝒗𝒐𝒍 = 𝟏 E 8.64

It is assumed that the fuel char content from the proximate analysis (fixed carbon)
is pure carbon and has an energy content of ΔHchar = 32.8 MJ/kg [241], which is the
reaction enthalpy for forming CO2 from C. The energy content of the volatiles can
then be calculated based on the fuel heating value subtracted the char energy:

𝜟𝑯𝒗𝒐𝒍 = 𝑳𝑯𝑽 − 𝒀𝒄𝒉𝒂𝒓 ∗ 𝜟𝑯𝒄𝒉𝒂𝒓 E 8.65

The fraction of CH4 and CO formed from the volatiles can then be calculated as:

𝜟𝑯𝒗𝒐𝒍
− 𝜟𝑯𝑪𝑶
𝒀 E 8.66
𝒀𝒗𝒐𝒍,𝑪𝑯𝟒 = 𝒗𝒐𝒍
𝜟𝑯𝑪𝑯𝟒 − 𝜟𝑯𝑪𝑶

𝒀𝒗𝒐𝒍,𝑪𝑶 = 𝟏 − 𝒀𝒗𝒐𝒍,𝑪𝑯𝟒 E 8.67

While this assumption fulfills the mass and energy balance for the fuel, the elemental
composition (on C, O, H) may not balance, since the ultimate analysis of the fuel is
disregarded. This can lead to some issues when calculating the oxygen requirement
of the combustion. To account for possible differences, additional nitrogen can be
added to the secondary air, to keep a realistic air consumption. Some example
calculations illustrating the difference are presented in Appendix D.

8.8.4 Char Oxidation

After the volatile species have been driven off, a solid residue called char remains. It
mostly consists of carbon, potentially with some ashes and minor amounts of
hydrogen, oxygen, nitrogen, and sulfur [242]. The amount and composition of the
formed char depends on parameters such as fuel type, temperature, and heating rate
during the devolatilization [85,242].

The solid char can be oxidized forming CO or CO2, or it can be gasified by water or
CO2 [85,242] according to the overall reactions:

𝑪 + 𝑶𝟐 → 𝑪𝑶𝟐 R 8.1

𝟏
𝑪 + 𝑶𝟐 → 𝑪𝑶 R 8.2
𝟐

𝑪 + 𝑯𝟐 𝑶 → 𝑪𝑶 + 𝑯𝟐 R 8.3

𝑪 + 𝑪𝑶𝟐 → 𝟐𝑪𝑶 R 8.4

8.8 Combustion of Solid Fuels 155

The rate of each reaction is governed by the gas phase composition, but also highly
dependent on temperature. Under combustion conditions the oxidation reaction
normally dominates [242] and the formation of CO is favored over that of CO2 [242].
The rate of steam gasification is 2-5 times higher than that of the CO2 gasification for
wood chars [85].

Char burnout is normally slower than the devolatilization. In pulverized coal flames,
the devolatilization process is completed within 0.1 s, whereas char burnout takes
around 1 s [243]. For alternative fuels such as biomass [31,86,87] or sewage sludge
[31], the char oxidation also tends to take longer, although only by a factor 2-3. This
is caused by a higher volatile and lower char content in most alternative fuels
compared to coal.

The char reactions are governed by several different steps [85]:

1. Film diffusion of oxidizing/gasifying agent to char surface

2. Diffusion through ash layer and particle
3. Adsorption onto the reaction site
4. Chemical reaction
5. Desorption of gaseous products
6. Diffusion of products through particle and ash layer
7. Film diffusion into the ambient gas.

Thus, it is evident that the char reaction rate will be dependent on several factors
relating to both the chemical and physical properties of the char. The chemical
structure of the char determines the chemical reaction rate. Meanwhile, the pore
structure of the char affects the total surface area and the diffusion rate in the char
and presence of inorganic constituents may catalyze the reaction [242]. Factors
relating to the devolatilization may also affect the char reactivity. For example high
heating rates during devolatilization will typically result in a more reactive char
[85].

The combustion can be divided into three regimes depending mainly on particle size
and temperature and the gaseous diffusion rates [85]:

1. Chemical control: At low temperatures and for small particles, the diffusion
rate is high compared to the reaction rate. The oxidizing/gasifying agents can
156 8 Cement Kiln Model Development

penetrate far into the particle and the conversion occurs throughout the
particle
2. Intra-particle diffusion control: For larger particle sizes or higher
temperatures, the diffusion into the particle is low, and most of the reaction
occurs in a thin layer close to the particle surface
3. External diffusion control: At high temperatures the reactions will be fast and
the diffusion of the oxidizing/gasifying agent through the film layer is
limiting and the reaction occurs at the particle surface

For alternative fuels, such as SRF, external diffusion is the main limitation at
temperatures above 1000 °C. For the less reactive and smaller particles of coal and
petcoke, chemical kinetics is limiting when the temperature does not exceed 2000
°C [31].

For modeling of the char oxidation, a shrinking core model [215,242] will be
employed here. It is thus assumed that the char oxidation takes place at the particle
surface. Diffusion to the particle surface as well as reaction kinetics will be
considered, and no ash layer is assumed. It is assumed that the char can react with
either O2 or H2O according to R 8.2 and R 8.3. Since the gasification with CO2 is
slower than H2O gasification, CO2 will be neglected.
𝟐
𝒅𝑿𝒄𝒉𝒂𝒓 𝟑 (𝟏 − 𝑿𝒄𝒉𝒂𝒓 )𝟑 𝑴𝒄𝒉𝒂𝒓
= ∗ ∗
𝒅𝒕 𝝋𝒑 𝒓𝒑 𝝆𝒄𝒉𝒂𝒓
E 8.68
𝟏 𝑪𝑶𝟐,𝒃𝒖 𝟏 𝑪𝑯𝟐𝑶,𝒃𝒖
∗( ∗ + )
𝝓𝑶𝟐 𝟏 𝝓𝑯𝟐𝑶 𝟏
𝒌𝒆𝒇𝒇,𝑶𝟐 𝒌𝒆𝒇𝒇,𝑯𝟐𝑶

In the equation, Xchar is the conversion with respect to char, φp the particle
sphericity, Mchar and ρchar are the molecular mass and density of the char, Cbu is the
bulk concentration, and ϕ is the stoichiometry for the reaction [mol O2 or H2O/mol
fuel]. The effective rate of the char reaction, keff is combined by the rate of reaction,
kchar, and the rate of film diffusion, kfilm:

𝟏 𝟏 𝟏
= + E 8.69
𝒌𝒆𝒇𝒇,𝒊 𝒌𝒄𝒉𝒂𝒓,𝒊 𝒌𝒇𝒊𝒍𝒎,𝒊

The rate constant for the chemical reaction is found based on an Arrhenius
expression:
8.8 Combustion of Solid Fuels 157

−𝑬𝑨,𝒄𝒉𝒂𝒓
𝒌𝒄𝒉𝒂𝒓 = 𝑨𝒄𝒉𝒂𝒓 ∗ 𝒆𝒙𝒑 ( ) E 8.70
𝑹𝒈𝒂𝒔 𝑻

The rate of film diffusion is determined based on the diffusivity of oxygen or water:

𝑫𝒊 ∗ 𝑺𝒉
𝒌𝒇𝒊𝒍𝒎,𝒊 = E 8.71
𝟐𝒓𝒑

The Sherwood number is determined based on the Reynolds and Schmidt numbers
[215]:
𝟏 𝟏
𝑺𝒉 = 𝟐 + 𝟎. 𝟔 ∗ 𝑹𝒆𝟐 ∗ 𝑺𝒄𝟑 E 8.72

8.8.5 Fuel Combustion in the Clinker Bed

It is evident from the performed industrial tests and camera measurements that a
substantial part of the SRF does not fully combust while it is in suspension in the
cement kiln. Large fractions of the fuel may only be combusted as it drops out of the
flame and into the clinker bed. This is consistent with the literature modeling results
described previously [143,163,204]. Thus, it will be necessary to account for fuel
combustion in the bed.

8.8.5.1 Fuel Trajectories

It will be assumed that fossil fuel particles are small enough to be carried with the
gas stream and not drop out of the flame. Only the larger alternative fuel particles
can drop out of the flame.

Approximate fuel trajectories can be calculated based on some simple force

balances. In the vertical direction the fuel is impacted by gravity as well as drag, from
an assumed quiescent gas:

𝟏 𝟐
𝑭𝒚 = 𝑨𝒑 𝝆𝒈 𝑪𝑫 ∗ (𝒗𝒑,𝒚 ) − 𝒈 ∗ 𝒎𝒑 E 8.73
𝟐

In the axial direction the particle will be slowed by drag from the slower flowing air.

𝟏 𝟐
𝑭𝒙 = − 𝑨𝒑 𝝆𝒈 𝑪𝑫 ∗ (𝒖𝒑,𝒙 − 𝒖𝒈 ) E 8.74
𝟐

For simplicity, the gas can be assumed to flow at a velocity equal to the inlet velocity
of the secondary air, and the gas properties can also be evaluated at the secondary
air temperature. This eliminates the effect of changing velocity due to an expanding
158 8 Cement Kiln Model Development

jet and temperature differences. The drag coefficient, CD, may be assumed a value of
0.44. Example trajectories are shown in Figure 8-8. The larger the particle is, the
further it can travel inside the kiln, as it is less influenced by drag. However, it may
also have a lower residence time since it drops faster.

Figure 8-8: Trajectories (a) and time in suspension (b) for particles with diameter
0.1-10 mm injected in kiln. up0 = 30 m/s, ρp = 1000 kg/m3, ug = 5.8 m/s, Tg = 750 °C,
rk = 1.95 m.

It will be assumed that particles on average travel as far as calculated by this simple
approach. The landing spot will be calculated as the location where the particles hit
the wall, i.e. have travelled a vertical distance equal to the kiln radius. A normal
distribution around the landing spot will be assumed for each particle size. The fuel
trajectories are based on constant parameters, but in reality they will be influenced
by changing conditions as e.g. the kiln gas is heated.

8.8.5.2 Fuel Conversion

When the particles hit the bed or kiln wall it will be assumed that they are converted
instantly. As the particles hit the bed, they are likely to be heated very fast, which
will quickly release the volatiles to the gas phase. The volatile content may also
increase (char yield reduced) due to the higher heating rate. On the other hand, if
the particles are covered in the bed, oxygen diffusion to the particles will be severely
hindered and char oxidation will be slow. Consequently, this simplification will be
more accurate for fuels with a relatively low char content, which is the case for many
alternative fuels
8.8 Combustion of Solid Fuels 159

Since it is assumed particles drop around the landing spot, with a normal
distribution, and the particles are instantly converted, the conversion can be
described by the probability density of the normal distribution:
𝟐
𝒅𝑿 𝟏 (𝒙 − 𝒍𝒇,𝒅𝒓𝒐𝒑 )
= ∗ 𝒆𝒙𝒑 (− ) E 8.75
𝒅𝒙 𝝈𝒇,𝒅𝒓𝒐𝒑√𝟐𝝅 𝟐𝝈𝟐𝒇,𝒅𝒓𝒐𝒑

The normal distribution will integrate to 1, which is convenient as the conversion is

bounded between 0 and 1. This equation will be applied both for drying,
devolatilization, and char oxidation when fuels enter the bed.

It is assumed that particles drop within a distance of lf,drop2 of the average drop spot
lf,drop. Then the standard deviation can be calculated by E 8.76, and 99.7 % of
particles will drop within the interval [lf,drop- lf,drop2; lf,drop+lf,drop2]

𝒍𝒇,𝒅𝒓𝒐𝒑𝟐
𝝈𝒇,𝒅𝒓𝒐𝒑 = E 8.76
3

Using these assumptions, the fuel landing and conversion in the bed can be sketched
as shown in Figure 8-9. The probability density function indicates the area around
the landing spot, lf,drop, where the fuel falls into the bed. The cumulative distribution
function describes the fuel conversion, assuming it is instant when fuel enters the
bed.

Figure 8-9: Sketch of the landing spot and particle conversion using normal
distributions for particles with diameter 0.1-10 mm. Parameters are similar to Figure
8-8.
160 8 Cement Kiln Model Development

8.8.6 Gas Phase Reactions

During the devolatilization of the fuel and the char oxidation combustible gasses are
released to the gas phase. The combustible volatiles in the model are assumed to be
CO and CH4, which react according to:

𝟏
𝑪𝑶 + 𝑶𝟐 → 𝑪𝑶𝟐 R 8.5
𝟐

𝑪𝑯𝟒 + 𝟐𝑶𝟐 → 𝑪𝑶𝟐 + 𝟐𝑯𝟐 𝑶 R 8.6

As the temperature in the kiln is high, the gas phase reactions are primarily limited
by turbulent mixing of the fuel and oxygen eddies. Accordingly, the eddy dissipation
concept proposed by Magnussen and Hjertager [244] will be used to describe the
combustion.

The model assumes that the combustion rate can be limited by either oxygen
availability or fuel availability. The rate of fuel limited combustion is given by:
𝝐
𝑹𝒄𝒐𝒎𝒃,𝟏,𝒊 = 𝑲𝟏 𝑪𝒊 𝑴𝒊 E 8.77
𝒌

Here i indicates the species that can be either CO or CH4. Ci is the gas concentration,
Mi is the molar mass, and the ratio ε/k is the turbulent eddy dissipation rate.

If oxygen availability is limiting the reaction, then the rate is given as:

𝑪𝑶𝟐 𝑴𝑶𝟐 𝝐 𝑪𝒊 𝑴𝒊 ∗ 𝝓𝒎,𝑶𝟐,𝒊

𝑹𝐜𝐨𝐦𝐛,𝟐,𝒊 = 𝑲𝟏 ∗ ( 𝑵𝒓𝒆𝒂𝒄𝒕𝒂𝒏𝒕𝒔 ) E 8.78
𝝓𝒎,𝑶𝟐,𝒊 𝒌 ∑𝒋=𝟏 𝑪𝒋 𝑴𝒋 ∗ 𝝓𝒎,𝑶𝟐,𝒋

The reaction rates are expressed in units [kg/(m3s)]. K1 is a constant with an

assumed value of 4 and ϕm is the stoichiometric coefficient for the reaction [kg
oxygen/kg fuel].

The reaction rate of fuel combustion is determined by the lowest of the above rates:

𝑹𝒄𝒐𝒎𝒃,𝒊 = 𝒎𝒊𝒏(𝑹𝒄𝒐𝒎𝒃,𝟏,𝒊 , 𝑹𝒄𝒐𝒎𝒃,𝟐,𝒊 ) E 8.79

8.8.6.1 The Eddy Dissipation Rate

The turbulent dissipation rate is determined based on the CFD simulations of the
full-scale rotary kiln, which were also used to derive parameters for the jet
expansion (see chapter 8.6.5). The average eddy dissipation rate inside the jet is
calculated and shown in Figure 8-10. The rate is very high close to the burner where
8.8 Combustion of Solid Fuels 161

the gas velocities are high, while it decreases further from the burner as the gas
velocities are decreased.

Figure 8-10: The eddy dissipation rate as function of distance from the burner in the
kiln. a) Normal data, b) log transformed data. Data derived from CFD simulations of a
petcoke flame performed by David Isaac.

In a log-log plot, Figure 8-10b, the data are approximately linear as long as the jet
spreads linearly (compare with Figure 8-6), up to around 7 m. Thus, the eddy
dissipation rate can be described as:
𝝐
= 𝑲𝒆𝒅𝒓,𝟏 ∗ 𝒙𝑲𝒆𝒅𝒓,𝟐 E 8.80
𝒌
162 8 Cement Kiln Model Development

The values of Kedr,1 and Kedr,2 can be found in Table 8-5. It is seen that the eddy
dissipation rate increases with the axial momentum of the burner and decreases
when the swirl is increased. This is likely caused by a lower axial velocity for the
higher level of swirl.

Table 8-5: Momentum and swirl number in CFD simulations and corresponding
values of Kedr,1 and Kedr,2.

Swirl
Momentum Swirl Kedr,1 by Kedr,2 by
Level Kedr,1 Kedr,2
(N/MW) Number E 8.81 E 8.82
(%)
5.3 50 0.18 102 -1.46 104 -1.47
6 100 0.35 97.7 -1.54 96.2 -1.53
7 50 0.20 116 -1.47 115 -1.47
8 100 0.36 110 -1.53 110 -1.54
9 50 0.20 131 -1.48 130 -1.48
10 100 0.35 124 -1.54 126 -1.54

The values of the constants Kedr,1 and Kedr,2 can be determined by a multilinear
regression based on the axial momentum and the swirl number:

𝑲𝒆𝒅𝒓,𝟏 = 𝟕𝟖. 𝟗 + 𝟕. 𝟑𝟗 ∗ 𝑰′ − 𝟕𝟕. 𝟏 ∗ 𝑺 E 8.81

𝑲𝒆𝒅𝒓,𝟐 = −𝟏. 𝟑𝟗 − 𝟏. 𝟒𝟔 ∗ 𝟏𝟎−𝟑 ∗ 𝑰′ − 𝟎. 𝟑𝟕𝟐 ∗ 𝑺 E 8.82

According to this regression model the eddy dissipation rate, Kedr,1, increases with
the axial momentum and decreases with the swirl number. The rate at which the
eddy dissipation decreases further from the burner, Kedr,2, is decreased by both
momentum and swirl. The values of the two constants calculated using E 8.81 and E
8.82 are shown in Table 8-5.

8.8.7 Combustion Model Approach

Solid fossil fuels such as coal or petcoke are fairly homogenous. Thus, in the model
the only difference between two coal particles will be the size difference. The model
is equipped to handle different particle sizes, which can be designated as a discrete
distribution. This has been done based on a Rosin-Rammler distribution [172,173],
divided into 10 discrete particle sizes.

In comparison SRF is highly heterogenous, consisting of different materials, as

outlined in chapter 3.4.1. The main parts of the combustible fractions can be derived
into biomass, paper & cardboard, plastics, and textiles (see Figure 3-3).
8.8 Combustion of Solid Fuels 163

In the experimental section, covered in chapter 7, it was discussed how accounting

for only biomass and plastic combustion is sufficient to describe the combustion of
SRF. A detailed physical characterization was made of two SRF samples, where the
size distribution of biomass and plastics have been determined. These data will be
used as input to the model for the size characterization of the SRF.

For the combustion reactions the plastic will be assumed to behave as polyethylene
and the model proposed recently by Nakhaei et al. [232] will be implemented here.
For the polyethylene an isothermal model predicts devolatilization times within a
20 % error compared to a more rigorous 1D-model. Thus, the isothermal particle
model will be used here.

The model for the biomass combustion is as described in the previous subchapters
8.8. The model is validated against experimental data collected from the literature
in Appendix E. The results indicate that the same conversion kinetics can be used
for a large range of different wood types and sewage sludge. The reason being that
heat transfer and oxygen diffusion limits the devolatilization and char oxidation,
rather than kinetics. As long as particles have diameter in the range of millimeters,
it seems justified to apply the same kinetics to different alternative fuels, as long as
they follow the general steps of drying, devolatilization, and char oxidation.

A comparison of using an isothermal and a 1-D model for predicting biomass

devolatilization times of particles in the range 0.1-20 mm can be found in Appendix
F. Significant errors can be made by assuming isothermal particles. The difference
between the models increases as the particle diameter and the gas phase
temperature surrounding the particle is increased. For particle above 10 mm, the
isothermal devolatilization time may be only 10 % of the non-isothermal time.
Nevertheless, for the kiln model it was found that an isothermal model could be
applied without significant error. It is assumed that alternative fuels can drop into
the bed and burn there (see chapter 8.8.5). The largest particles will mainly burn in
the bed rather than in suspension, making the heating rate negligible. Accounting
for non-isothermal particles increases model convergence time by approximately a
factor of three and does not change the results significantly. Consequently,
isothermal particles were assumed in the model results presented later in chapter
9.
164 8 Cement Kiln Model Development

The effects of particle shape are included by accounting for the sphericity in the
particle temperature (E 8.53-E 8.55) and char oxidation equations (E 8.68). The
equations have been derived for spherical particles. In the case where the particles
are not spherical, the particle equivalent diameter is used instead of the actual
diameter. The inclusion of sphericity in the heat transfer and char oxidation
equations accounts for the higher surface area of non-spherical particles compared
to their spherical counterparts. The result is that the effective particle surface area
for heat transfer and char oxidation is increased by a factor 1/φ.

8.9 Gas Properties in the Kiln Model

The estimation of physical properties is an important part of the model. They e.g.
influence the convective heat transfer in the kiln as outlined in chapter 8.7.1 . This
is especially true for the gas phase where large differences in temperature and
composition are expected.

The heat capacity of the gas will change both according to temperature and the gas
composition. However, it has been elected to keep it constant for simplicity. The
value applied in the model will be assumed equal to that of the initial heat capacity
of the secondary air. The initial heat capacity will be calculated based on data from
the National Institute of Standards and Technology [241], using the following
correlation:
𝑵
𝟏 𝑻𝒈 𝑻𝒈 𝟐
𝑪𝒑𝒈 = ∑ 𝒚𝒊 (𝑲𝟏,𝒊 + 𝑲𝟐,𝒊 ∗ + 𝑲𝟑,𝒊 ∗ ( ) + 𝑲𝟒,𝒊
𝑴𝒈 𝟏𝟎𝟎𝟎 𝟏𝟎𝟎𝟎
𝒊=𝟏
E 8.83
𝟐 −𝟐
𝑻𝒈 𝑻𝒈
∗( ) + 𝑲𝟓,𝒊 ∗ ( ) )
𝟏𝟎𝟎𝟎 𝟏𝟎𝟎𝟎

The density will be calculated based on the ideal gas law:

𝑵𝒈
𝒑
𝝆𝒈 = ∑ 𝑴𝒊 ∗ 𝒚𝒊 E 8.84
𝑹𝑻𝒈
𝟏

The viscosity is determined based on the following expression used by Guo et al.
[245].

𝝁𝒈 = 𝟎. 𝟏𝟔𝟕𝟐 ∗ 𝟏𝟎−𝟓 ∗ √𝑻𝒈 − 𝟏. 𝟎𝟓𝟖 ∗ 𝟏𝟎−𝟓 E 8.85

8.9 Gas Properties in the Kiln Model 165

The thermal conductivity is based on the following polynomial proposed by

Thunman and Leckner [246]. The data are valid for atmospheric air in the range
250-3000 K.

𝝀𝒈 = −𝟕. 𝟒𝟗𝟒 ∗ 𝟏𝟎−𝟑 + 𝟏. 𝟕𝟎𝟗 ∗ 𝟏𝟎−𝟒 ∗ 𝑻𝒈 − 𝟐. 𝟑𝟕𝟕 ∗ 𝟏𝟎−𝟕 ∗ 𝑻𝟐𝒈

+ 𝟐. 𝟐𝟎𝟐 ∗ 𝟏𝟎−𝟏𝟎 ∗ 𝑻𝟑𝒈 − 𝟗. 𝟒𝟔𝟑 ∗ 𝟏𝟎−𝟏𝟒 ∗ 𝑻𝟒𝒈 + 𝟏. 𝟓𝟖𝟏 E 8.86
∗ 𝟏𝟎−𝟏𝟕 ∗ 𝑻𝟓𝒈

For the char oxidation it is important to determine the diffusivity of oxygen and
water (see E 8.71). The constants can be calculated by the Fuller-Schettler-Giddings
approach based on diffusion volumes [247]:

𝟏 𝟏
√( + 𝑴 ∗ 𝟏𝟎𝟎𝟎)
𝑴 𝒊 ∗ 𝟏𝟎𝟎𝟎 𝒂𝒊𝒓
𝑫𝒊 = 𝟏 ∗ 𝟏𝟎−𝟕 ∗ 𝑻𝟏.𝟕𝟓
𝒈 ∗ E 8.87
𝒑 𝟐
𝟏/𝟑 𝟏/𝟑
( 𝟓 ∗ 𝑽𝒅𝒊𝒇𝒇,𝒊 ∗ 𝑽𝒅𝒊𝒇𝒇,𝒂𝒊𝒓 )
𝟏. 𝟎𝟏𝟑𝟐𝟓 ∗ 𝟏𝟎

D is the diffusion coefficient [m2/s], Tg is the gas temperature, M the molar mass, p
the pressure, and Vdiff is the diffusion volume [cm3/mol]. The coefficients on the
molar mass and the pressure are to recalculate into [g/mol] and [atm] units, which
are the units used by Fuller, Schettler, and Giddings [247].

The values of viscosity, heat conductivity, and diffusion coefficients are determined
for air. Some error can thus be expected, since the gas composition in the kiln is
different following combustion.

The emissivity of the gas for radiation is determined by the Weighted Sum of Grey
Gasses (WSGG) model. The method described by Johansson et al. [248] will be used
here. It is assumed that only H2O and CO2 contribute to the emissivity of the gas.
𝑵𝒈𝒂𝒔

𝝐𝒈 = ∑ 𝒂𝒋 (𝟏 − 𝒆𝒙𝒑 (−𝜿𝒋 𝒍𝒎 (𝒑 ∗ 𝟏𝟎−𝟓 )(𝒚𝑯𝟐 𝑶 + 𝒚𝑪𝑶𝟐 )) E 8.88

𝒋=𝟎

Ngas = 4 components are used here. In the equation a is a weighting factor in the
WSGG model and κ is an absorption coefficient, p is the pressure. Details of how
these are calculated will not be presented here, but can be looked up in the source
[248]. lm is the total path length for radiation. It will be assumed that the value for
an infinite cylinder can be used here [229]:

𝒍𝒎 = 𝟎. 𝟗𝟓𝒅𝒌 E 8.89
166 8 Cement Kiln Model Development

For the radiation model proposed in chapter 8.7.3 it will be assumed that the
secondary air does not absorb radiation, since it does not contain H2O or CO2.
However, it will contain entrained clinker dust that could affect radiation [196], but
this effect will be neglected here.

8.10 Energy and Mass Balances in Kiln

The following chapter will describe the final equations to determine gas and bed
temperatures in the kiln, based on energy and mass balance considerations. The
equations are derived for changes in the axial direction of the kiln. Detailed
derivations can be found in Appendix G.

8.10.1 Temperature in Clinker Bed

A sketch of the different heat sources to a section of the bed is shown in Figure 8-11.
This approach is used to derive the energy balance and the temperature equation
for the bed. At the left boundary, clinker enters at one temperature and a given
fraction of melt, and it exits at the right boundary at a second temperature and melt
fraction. Inside the section, energy is lost as CO2 leaves the bed (1), heat is released
or consumed by the clinker reactions (2), and heat is transferred to or from the bed
(3).

Figure 8-11: The energy sources in a section of the bed used to derive the temperature
equation for the bed.

In Figure 8-11 and E 8.90 E is energy, Tb is the bed temperature, x is the axial
distance in the kiln, Cpb the bed heat capacity, 𝑚̇𝑏 the mass flow of clinker, R the
reaction rate of the i’th reaction [kg/m3s], ΔH the enthalpy for reactions or melting,
Ab the cross sectional area of the bed, q the heat flux to the bed, l the contact area of
the bed (e.g. between gas and bed), and M is the molar mass.
8.10 Energy and Mass Balances in Kiln 167

The temperature in the bed can be described by the following equation:

𝑵𝒓
𝒅 𝟏 𝟏 ∑ 𝑹 ∗ 𝜟𝑯 ∗ 𝑨
𝑻 = ( 𝒊 𝒓,𝒊 𝒃
𝒅𝒙 𝒃 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕,𝒎𝒂𝒙 𝒎̇𝒃 𝒊
(𝑪𝒑𝒃 + 𝑻 )
𝒎𝒆𝒍𝒕,𝟐 − 𝑻𝒎𝒆𝒍𝒕,𝟏 1
E 8.90
𝑵𝒉𝒕
𝑴𝑪𝑶𝟐
(𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 ) ∗ 𝑹𝟏 ∗ 𝑨 ∑ 𝒒𝒊 ∗ 𝒍𝒊 )
+ 𝑴𝑪𝒂𝑪𝑶𝟑 𝑏 +
𝒊
2
3

On the right-hand side of the equation term 1 describes the heat from the
endothermic and exothermic clinker reactions. Term 2 accounts for enthalpy
changes of melting where CO2 leaves the bed, and term 3 describes the heat transfer
to the bed.

It is assumed that the bed will begin to melt as it reaches a certain temperature,
Tmelt,1 = 1280 °C [217]. Upon further heating the bed will continue to melt, with the
melt fraction increasing linearly to an upper temperature, Tmelt,2 = 1450 °C.

𝑻𝒃 − 𝑻𝒎𝒆𝒍𝒕,𝟏
𝑿𝒎𝒆𝒍𝒕 = 𝑿𝒎𝒆𝒍𝒕,𝒎𝒂𝒙 ∗ E 8.91
𝑻𝒎𝒆𝒍𝒕,𝟐 − 𝑻𝒎𝒆𝒍𝒕,𝟏

The C3A and C4AF phases form most of the melt phase in the kiln. Typically, around
20 % of the clinker will melt in the kiln [249]. For bed temperature above Tmelt,2 the
melt fraction is capped to the value of Xmelt,max.

8.10.2 Temperature in Gas Phase

A section of the gas phase is shown in Figure 8-12. The sketch shows the different
energy source terms used to derive the heat balance in the gas phase. In the gas
phase, energy is entrained with the secondary air (1), CO2 is released from the bed
(2), gasses are released from the fuel (3), heat is transferred to and from the gas (4),
and heat is released from the combustion reactions (5).

In Figure 8-12 and E 8.92 E is energy, Tg is the gas temperature, x is the axial distance
in the kiln, Cpg the gas heat capacity, 𝑚̇𝑔 the mass flow of gas, and R [kg/m3s]
denotes rate of entrainment, CO2 release, gas release from the fuel, and reactions. H
indicates enthalpy and ΔH the enthalpy for reactions, Ag and Ab the cross sectional
area of the gas and bed, q the heat flux to the gas, and l the contact area of the gas
(e.g. between gas and bed).
168 8 Cement Kiln Model Development

Figure 8-12: The energy sources in a section of the gas phase used to derive the
temperature equation for the gas phase.

The gas phase temperature is determined based on the following equation.

𝒅 𝟏 𝑹 ∗ 𝑯𝒔𝒆𝒄 ∗ 𝑨𝒈 𝑹𝑪𝑶𝟐 ∗ 𝑯𝑪𝑶𝟐 ∗ 𝑨𝑏

𝑻𝒈 = [ 𝒆𝒏𝒕 +
𝒅𝒙 𝒎̇𝒈 ∗ 𝑪𝒑𝒈 1 2

𝑵𝒉𝒕 𝑵𝒓
𝑹𝒇𝒈 ∗ 𝑯𝒇𝒈 ∗ 𝑨𝒈 ∑ 𝒒𝒊 ∗ 𝒍𝒊 ∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝒈
+ + + E 8.92
3 𝒊 𝒊
4 5

𝒅
− 𝒅𝒙 (𝒎̇𝒈 )𝑪𝒑𝒈 ∗ 𝑻𝒈 ]
6

On the right-hand side of the equation term 1 describes the energy obtained as
secondary air is entrained into the jet, term 2 as CO2 is released from the bed to the
gas phase. Term 3 accounts for the release of gasses from the fuel. Term 4 describes
the heat transfer to and from the gas and term 5 the heat released in the combustion
reactions. Term 6 account for changes in the energy as the mass flow rate is changed.

8.10.3 Mass Flow in Clinker Bed

It will be assumed that the mass flow of clinker is constant throughout the kiln,
except for the mass loss due to calcination. This loss is given as:

𝒅𝒎̇𝒃 𝒅𝝆𝒃 𝑴𝑪𝑶𝟐

= 𝒗𝒃 ∗ 𝑨𝒃 = −𝑹𝟏 ∗ ∗ 𝑨𝒃 E 8.93
𝒅𝒙 𝒅𝒙 𝑴𝑪𝒂𝑪𝑶𝟑
8.10 Energy and Mass Balances in Kiln 169

8.10.4 Mass Flow in Gas Phase

The mass flow in the gas phase changes due to entrainment of secondary air, release
of volatiles during drying, devolatilization, and char oxidation, and release of CO2 as
residual limestone in the clinker is calcined. The overall balance can be written as:

𝒅𝒎̇𝒈 𝑴𝑪𝑶𝟐 𝑨𝒃
= 𝑨𝒈 (𝑹𝒆𝒏𝒕 + 𝑹𝒅𝒓𝒚 + 𝑹𝒅𝒆𝒗𝒐𝒍 + 𝑹𝒄𝒉𝒂𝒓 + 𝑹𝟏 ∗ ∗ ) E 8.94
𝒅𝒙 𝑴𝑪𝒂𝑪𝑶𝟑 𝑨𝒈

In addition to the overall mass balance, the gas components N2, O2, CO2, H2O, CO, and
CH4 will be tracked in the kiln. The combustion air is assumed to be atmospheric air
consisting of N2 and O2. During combustion, O2 will be consumed while the different
combustion products are released. Minor species in the fuel and gas phase such as
N, S, and Cl, will not be tracked. The following balances are written on a molar basis.

The N2 flow in the primary jet is changed due to entrainment of secondary air:

𝒅𝒏̇ 𝑵𝟐
= 𝑨𝒈 ∗ 𝑹′ 𝒆𝒏𝒕 ∗ 𝒚𝑵𝟐,𝒔𝒆𝒄 E 8.95
𝒅𝒙

Oxygen is entrained into the flame and used up during combustion of char, CO, and
CH4:

𝒅𝒏̇ 𝑶𝟐
= 𝑨𝒈 ∗ (𝑹′𝒆𝒏𝒕 ∗ 𝒚𝑶𝟐,𝒔𝒆𝒄 − 𝑹′ 𝒄𝒉𝒂𝒓,𝒐𝒙 𝝓𝒄𝒉𝒂𝒓 − 𝑹′ 𝑪𝑶,𝒐𝒙 𝝓𝑪𝑶
𝒅𝒙 E 8.96
− 𝑹′𝑪𝑯𝟒,𝒐𝒙 𝝓𝑪𝑯𝟒 )

CO2 is formed from the oxidation of CO and CH4, but can also be released from the
clinker during calcination:

𝒅𝒏̇ 𝑪𝑶𝟐 𝟏 𝑨𝒃
= 𝑨𝒈 ∗ (𝑹′𝑪𝑶,𝒐𝒙 + 𝑹′𝑪𝑯𝟒,𝒐𝒙 + 𝑹𝟏 ∗ ) E 8.97
𝒅𝒙 𝑴𝑪𝒂𝑪𝑶𝟐 𝑨𝒈

H2O is released during the drying of the fuel and the combustion of CH4:

𝒅𝒏̇ 𝑯𝟐𝑶 𝑹𝒅𝒓𝒚

= 𝑨𝒈 ∗ ( + 𝟐𝑹′𝑪𝑯𝟒,𝒐𝒙 ) E 8.98
𝒅𝒙 𝑴𝑯𝟐𝑶

CH4 is released during devolatilization and removed through combustion:

𝒅𝒏̇ 𝑪𝑯𝟒 𝒀𝐯𝐨𝐥,𝑪𝑯𝟒

= 𝑨𝒈 ∗ ( 𝑹 − 𝑹′𝑪𝑯𝟒,𝒐𝒙 ) E 8.99
𝒅𝒙 𝑴𝑪𝑯𝟒 𝒅𝒆𝒗𝒐𝒍

CO is released during devolatilization and char oxidation, and then removed as it is

oxidized into CO2 in the gas phase reactions.
170 8 Cement Kiln Model Development

𝒅𝒏̇ 𝑪𝑶 𝒀𝒗𝒐𝒍,𝑪𝑶 E
= 𝑨𝒈 ∗ ( 𝑹 + 𝑹′𝒄𝒉𝒂𝒓,𝒐𝒙 − 𝑹′𝑪𝑶,𝒐𝒙 )
𝒅𝒙 𝑴𝑪𝑶 𝒅𝒆𝒗𝒐𝒍 8.100

In the above equations reaction rates denoted R are on a mass basis [kg/m3/s] and
rates denoted R’ are on a molar basis [mol/m3/s]. ϕ is the stoichiometry for the
combustion reactions [mol O2/mol fuel]. Y is mass fraction, y is mole fraction, Ag and
Ab are the cross-sectional area of the gas and bed, M is molar mass, ṁ is mass flow,
and ṅ is molar flow.

8.11 Model Solution Procedure

The model is solved using Matlab R2015b, where the differential equations are
solved using the stiff ODE solver ode15s [250].

Due to the counter current flow of the gas and bed, the model is divided into two
parts one for the gas phase and one for the bed phase. This is necessary since it is
not feasible to calculate the extent of clinker reactions backwards with respect to
time. The model for the gas phase includes the equations for the primary air,
secondary air, fuel, and shell and wall temperatures. The model for the bed phase
includes bed temperatures and clinker reactions.

Initially, the model part for the gas phase is solved assuming no reactions in the bed
phase. The initial conditions, e.g. air and fuel flow rates and temperatures, are
known at the burner end (kiln clinker outlet), while the temperature of the pre-
calcined raw meal is known at the other end of the kiln (raw material inlet). Thus,
initially an outlet temperature of the clinker is guessed, and a number of iterations
are performed until the inlet temperature matches the initial condition. This is done
by utilizing the shooting method [251].

An example is shown in Figure 8-13. In this case the inlet raw material temperature
is known to be 830 °C, and the clinker outlet temperature is guessed to be 1327 °C.
This temperature, however, results in an inlet temperature of 769 °C, which is lower
than the actual temperature of 830 °C. In the second iteration the guess on the outlet
temperature is increased. This then results in an inlet temperature that is too high.
Additional iterations are performed until the guessed clinker outlet temperature
yields an appropriate raw meal inlet temperature, within 10 °C of the actual inlet
value.
8.11 Model Solution Procedure 171

Figure 8-13: Example of the iterative procedure (shooting method) to find initial
solution of the gas and bed temperature. Solid line indicates bed temperature and
dashed line indicates gas temperature.

After the initial temperature profiles have been found, the clinker reactions are
added to the bed. This changes the bed temperature due to heat release and
consumption in the clinker reactions. This also has an effect on the gas temperature,
since the heat transfer is dependent on the bed temperature. Furthermore, the
calcination of residual limestone can change the gas flow, as CO2 is released from the
bed into the gas phase.

The solution is performed in an iterative procedure, where the coupling between

the bed and gas phase is handled by successively solving the model for the gas phase
followed by the model for bed phase. The iterations are assumed to have reached
convergence after a minimum of 4 iterations and when the average gas and bed
temperature changes less than 3 K upon successive iterations.

An example of the different solutions is shown in Figure 8-14. The initial gas
temperature is found assuming no reactions in the clinker phase. When clinker
reactions are added (dashed blue line), the temperature of the bed rapidly increases
around 30 meters as belite (C2S) is formed. As the bed temperature is now higher,
the heat transfer between gas and bed is changed, and the gas temperature also
becomes higher, which in turn affects the bed temperature, etc. After convergence
the temperatures stabilize at levels that are slightly higher than the initial
temperatures.
172 8 Cement Kiln Model Development

Figure 8-14: The temperature profiles in the gas and bed phase before (dashed lines)
and after convergence (solid lines) has been reached. Bed temperature without
clinker reactions (dotted line) is also shown.

8.12 Summary and Conclusions

The review of the literature revealed that many research groups have made a
concerted effort to model the different processes in the cement kiln. However, only
a few works have dealt with the combustion of alternative fuels in the cement kiln,
and none of this work has attempted to couple the combustion models to the clinker
chemistry and quality. Consequently, the kiln model developed here includes a
simple model for the clinker formation. The burner settings may also have a
significant impact on the kiln flame, which was also accounted for in the model.

Many processes occur in the cement rotary kiln at the same time, these include: solid
movement, clinker reactions, air flow and mixing, heat transfer, and combustion. A
brief review of these mechanisms has been given, and simplified models have been
proposed to account for the different processes. These sub-models were combined
to yield a model for the cement kiln. The kiln model is able to calculate temperature
profiles in the kiln and estimate the clinker phase composition. The model will be
validated and tested in chapter 9.
9 MODEL VALIDATION AND SRF
CO-FIRING

The purpose of this chapter is to present model simulation results and behavior and
to test the developed model. The model will be validated against literature results
obtained in a pilot scale kiln as well as results obtained in an industrial kiln.
Afterwards, the model will be used to investigate the impact of co-firing SRF.

9.1 Model Validation – Pilot Scale

The model is first validated against measurements from the Cemflame 3
experiments carried out at the IFRF (International Flame Research Foundation),
which were also briefly described in chapter 3.5.1. These are to the author’s
knowledge the most detailed measurements available in a pilot-scale rotary kiln.
While the pilot scale measurements do not include a clinker bed, they contain more
detailed measurements of gas phase temperatures and composition, than what can
be obtained in a full-scale plant. Thus, they provide a basis for validating the
combustion part of the kiln model.

9.1.1 The Cemflame 3 Experiments

The Cemflame 3 experiments are described by Haas et al. [80]. The goal was to
evaluate the effect of co-firing coal with different alternative fuels, such as sewage
sludge and plastic waste. The 4 main objectives were detailed as follows:
174 9 Model Validation and SRF Co-Firing

1. To assess different co-firing fuel types for use in cement and lime kilns and
to give recommendations for their optimum application.
2. To maximize the co-firing ratio of alternative fuels in a kiln whilst
maintaining acceptable combustion (heat transfer) performance.
3. To identify flame characteristics for cement kiln flames when co-firing
alternative fuels.
4. To evaluate the effect of oxygen boosting in a cement kiln.

9.1.2 Experimental Facilities

Experiments were made in a pilot scale cement kiln simulator of 9 m length and 0.78
m in diameter. The kiln was refractory lined to maintain high temperatures and was
equipped with a burner with a thermal input of 2 MW. A pre-combustor unit
supplied hot combustion air (900 °C) as secondary air to the kiln.

A series of detailed measurements were made on a coal flame as reference, while

the measurements on alternative fuels co-firing are less detailed. For the coal flame,
temperatures, gas composition, and burnout were measured along the centerline of
the flame, and at various cross sections.

The kiln does not contain a clinker bed, and the experiments can thus only be used
to validate the temperature and reactions in the suspended gas phase. The kiln
model is thus slightly adjusted to not contain the bed.

9.1.3 Experimental Results

Selected results of the coal flame measurements are shown in Figure 9-1 and Figure
9-2. Figure 9-1 shows measurements in the gas phase taken along the axial
centerline of the flame, while Figure 9-2 shows wall temperatures for the baseline
coal flame as well as several co-firing scenarios with varying amounts of plastic foils.
The wall temperature derived from Figure 9-2 is used as input to the model, instead
of calculating the wall temperature based on a heat balance through the wall.
9.1 Model Validation – Pilot Scale 175

Figure 9-1: Measurements along the axial centerline of the kiln flame. a) CO
concentration, burnout, gas temperature and NOx concentration. b) oxygen
concentration, hydrocarbon concentration, and radiative heat flux upon the kiln walls
[80].

Figure 9-2: Kiln wall temperatures of baseline coal flame experiments and co-firing
with different amounts of plastic foils in % of the energy input [80].
176 9 Model Validation and SRF Co-Firing

The size of the fuel is detailed in Figure 9-3a. The data for the fetnuss coal was
extracted from the figure and fitted to a Rosin-Rammler plot in Figure 9-3b.

Figure 9-3: a) Size distribution of fuel as reported in the Cemflame experiments [80].
b) Rosin Rammler plot of the fetnuss coal size distribution.

The Rosin Rammler distribution is given in equation E 9.1. The cumulative mass,
CmRR, for a given particle size, dp, can be described by the 63rd percentile, dRR, and
the spread, nRR. The parameters for the size distribution were fitted as dRR = 41.6
µm and nRR = 1.22. In the model 10 discrete particle sizes will be used, based on the
fitted distribution.

𝒏𝑹𝑹
𝒅𝒑
𝑪𝒎𝑹𝑹 (𝒅𝒑 ) = 𝟏𝟎𝟎 − (𝟏𝟎𝟎 ∗ 𝒆𝒙𝒑 (− ( ) )) E 9.1
𝒅𝑹𝑹

9.1.4 Model Input Parameters

The operating parameters that are used to model the Cemflame coal flame are listed
in Table 9-1. The coal flame measurements are based on a medium volatile fetnuss
coal with the proximate composition as stated in Table 9-1. Additional 0.06 MW of
natural gas is used to support the flame, but this will be neglected in the model, as
the energy input is very low.

Not all necessary model input parameters are stated in the Cemflame report. Thus,
Table 9-2 defines some additional parameters that are used in the model. These
numbers are estimated based on information available in the Cemflame report, or
taken from elsewhere, if necessary. They include for instance the kinetics of drying
and devolatilization of the coal. In addition, some assumptions are made on inlet
primary air and fuel temperatures, which are not stated in the Cemflame report.
9.1 Model Validation – Pilot Scale 177

Table 9-1: Model parameters used to model the Cemflame experiments.

Unit Value
Fuel Composition
Volatiles wt% dry 24.25
Char wt% dry 72.84
Ash wt% dry 2.79
Moisture wt% a.r. 0.64
Fuel Properties
Density kg/m3 1336
Lower Heating Value MJ/kg dry 33.97
Preexponential factor for char kg/(m2 s
0.01
oxidation Pa)
Activation energy for char
J/mol 105*103
oxidation
Kiln Parameters
Inner Diameter m 0.78
Length m 9
Burner Diameter m 0.22
Secondary air flow kg/h 2720
Secondary air temperature °C 900
Initial oxygen concentration of gas mol% 20.9
Burner Firing and Air
Energy Input MW 1.94
Coal Feed kg/h 210
Axial Air kg/h 62.5
Swirl Air kg/h 62.5
Transport Air kg/h 75

It is stated in the report that the simulated secondary air is generated by combustion
of natural gas in a pre-combustion chamber. The oxygen content is then adjusted to
20.9 mol%. The remaining composition is not stated, and it will be assumed in the
simulations that the remaining gas is nitrogen. In reality, significant amounts of CO2
and H2O will be present in the gas due to the combustion in the pre-combustion
chamber.

The burner momentum and swirl number during the baseline experiments are not
stated. Elsewhere in the report it is stated that experiments were performed at 3.1
N/MW (low momentum) and 4.1 N/MW (high momentum) and swirl numbers up
to 0.55. It will be assumed that the momentum was 3.1 N/MW and the swirl number
0.24. For the sake of simplicity, the heat capacity of the two gas phases were
assumed constant in the calculations. The heat capacity in Table 9-2 for the
secondary air is calculated for air at 900 °C, while the value for the primary air was
evaluated as the average heat capacity of air and flue gas at 1500 °C.
178 9 Model Validation and SRF Co-Firing

Table 9-2: Estimated input parameters for to model the Cemflame experiments.
Parameters marked by * have been assumed.

Parameter Unit Value Reference

Fuel parameters
Fuel particle emissivity - 0.9 [31]
Fuel heat capacity J/kg/K 1400 [31]
Initial temperature of fuel °C 50 *
Preexponential factor for drying of coal 1/s 5.13*106 [31]
Activation energy for drying of coal J/mol 87.9*103 [31]
Preexponential factor for devolatilization
1/s 9.59*104 [31]
of coal
Activation energy for devolatilization of
J/mol 82.6*103 [31]
coal
Primary and secondary air parameters
Temperature of primary air °C 50 *
Initial nitrogen concentration of gas mol % 79.1 *
Heat capacity of primary air J/kg/k 1290 *
Heat capacity of secondary air J/kg/k 1179 *
Burner Momentum N/MW 3.1 *
Burner Swirl Number - 0.24 *
Other parameters
Wall Emissivity - 0.8 [31]

9.1.5 Model Results

The model and experimental results from the Cemflame 3 experiments are
compared in Figure 9-4 and Figure 9-5. It should be noted that the experimental
points have been shifted by x/dk = 0.23, to set the burner tip at x/dk = 0 (compared
to Figure 9-1 and Figure 9-2). As input to the model the wall temperature was
derived from Figure 9-2. The experimental data were smoothed using a moving
average filter with a window size of 10 and local regression using a 2nd degree
polynomial (Matlab function ‘smooth’ with method ‘loess’). This significantly
reduces the data scatter giving the red line in Figure 9-4a.

Figure 9-4a shows the temperatures of the gas and wall. Figure 9-4a shows ‘peak’
temperatures, which are measured at the axial centerline of the flame, while the
‘average’ temperatures are arithmetic means based on traverses through the flame.
The model shows a rapid increase in the flame temperature as secondary air is
entrained into the primary air jet and as the volatiles are burned. This gives a peak
temperature of close to 2100 °C, which is somewhat above the measured values. As
combustion takes place the flame temperature exceeds that of the secondary air, and
the flame will be cooled as the remaining secondary air is entrained into the flame.
9.1 Model Validation – Pilot Scale 179

At x/dk = 2, the secondary air is fully entrained, which gives a bend in the
temperature profile, where it starts to increase again as combustion of the char
takes place. From x/dk = 2.5, most of the combustion has already occurred and the
temperature begins to decrease as heat is lost to the wall. The temperature
predicted by the model decreases slightly faster than what has been measured, but
the temperature is generally in the correct range.

Figure 9-4: Comparison of measurements and model predictions of the gas

temperature through the kiln simulator of the Cemflame experiments [80]. Lines are
model and markers are experimental points. The peak temperatures are measured at
the axial centerline of the flame. The average temperatures are the mean based on
traverses in radial direction through the flame.

Figure 9-5 shows the measurements and model predictions of selected gas phase
components and the fuel burnout. In the model the only hydrocarbon is CH 4, while
others (although not specifically stated) are measured in the Cemflame
experiments. The fuel burnout is in the model defined as the total conversion, which
is the weighted average of drying, devolatilization, and char oxidation:

𝒀𝑯𝟐𝑶 ∗ 𝑿𝒅𝒓𝒚 + 𝒀𝒗𝒐𝒍 ∗ 𝑿𝒅𝒆𝒗𝒐𝒍 + 𝒀𝒄𝒉𝒂𝒓 ∗ 𝑿𝒄𝒉𝒂𝒓

𝑿𝒕𝒐𝒕 = E 9.2
𝒀𝑯𝟐𝑶 + 𝒀𝒗𝒐𝒍 + 𝒀𝒄𝒉𝒂𝒓

The total conversion is calculated for each particle size and averaged over the 10
different particle sizes used in the model to give the burnout plotted in Figure 9-5a.
The conversion of the individual particles is shown in Figure 9-5b.
180 9 Model Validation and SRF Co-Firing

Based on the model it can be seen that volatiles are released, and small particles
undergo char oxidation within 1-2 kiln diameters. This results in a rapid
consumption of oxygen, resulting in a low oxygen content at x/dk = 1. A small peak
in CO and CH4 concentration is also observed. In the experiments a higher CO peak
and lower oxygen concentrations were measured, than what the model predicts.
This may be because concentrations are measured at the center of the flame, where
oxygen depletion can occur, while the model predicts an average for the cross
section. The oxygen concentration increases and the CO and CH4 is converted as the
remaining secondary air is entrained into the flame. The model mixing is predicted
to x/dk ~ 2, which corresponds well to the measured local maximum in oxygen
concentration. The oxygen concentration then decreases through the rest of the kiln
as char oxidation of the larger coal particles take place.

Figure 9-5: a) Comparison of measurements and model predictions of gas

concentration of CO, oxygen, and hydrocarbons (CH4) and average fuel burnout of the
Cemflame experiments [80]. Lines are model and markers are experimental points.
b) Conversion of the 10 discrete fuel sizes according to model.

The model predicts an average burnout that appears to match the experimental
trend well as seen in Figure 9-5a. The 10 discrete fuel particle sizes show that the
size does not influence drying and devolatilization particularly, which is completed
within 1 kiln diameter yielding a conversion of 0.25. For the larger particle char
oxidation takes much longer, and the two largest particle size groups (dp = 70 and
110 µm) are not completely converted within the kiln as seen in Figure 9-5b.

It should be noted that the measurements are taken along the axial centerline of the
flame. In comparison, the model calculates average temperatures and
concentrations in the cross section of the jet. This means that the local
9.2 Model Validation – Industrial Scale 181

measurements in the center of the flame could be significantly different than what
the model predicts, if large radial gradients are present during the experiments.
Radial traverses of the gas temperature and CO concentrations indicate that this
might be the case, especially close to the burner, as shown in Figure 9-6. For
instance, it is seen that the gas temperature is at least 100 °C higher in the center of
the kiln than close to the walls. The gas temperature could thus be somewhat
underestimated by the model compared to the measurements.

The CO concentration is also highly dependent on the radial position. Thus, the very
high peak CO concentration and low oxygen seen in Figure 9-5 may be due to oxygen
depletion in the center of the flame.

Figure 9-6: Radial traverses of a) CO and hydrocarbon (HC) concentrations and b) gas
temperatures in the Cemflame experiments [80].

9.2 Model Validation – Industrial Scale

The validation of the model in industrial scale will be performed against
measurements described by Ariyaratne et al. Two sets of industrial tests were
performed to investigate the effect of co-firing rate on the clinker quality. The first
set of tests investigated the effect of co-firing meat and bone meal (MBM) with coal
[81] and the second set of tests dealt with solid hazardous waste (SHW) [82]. Results
of the tests were already briefly discussed in chapter 3.5.1. Additional details of the
kiln system, such as amount of primary air and secondary air, can be found in
subsequent papers, where the same kiln system was modeled [83,84].
182 9 Model Validation and SRF Co-Firing

9.2.1 The Industrial Tests

The industrial tests were made at a Norwegian cement plant. The kiln system
consisted of a 4-stage, 2-string preheater with calciner, with a kiln length of 68 m
and diameter of 4.4 m. The kiln had a production capacity of 3,400 ton clinker per
day.

Tests were performed with coal as the baseline fuel, and the feed rate of MBM or
SHW was incremented in steps lasting two hours (as previously shown in Figure
3-5). The coal firing rate was adjusted to keep the heat input constant. The clinker
quality was reported by measuring the free lime content. Details of the tests can be
found in Table 9-3 and Table 9-4. During the SHW tests, 1.5 t/h liquid hazardous
waste was also co-fired in the kiln. To simplify the calculations, this was replaced by
0.7 t/h of coal, which has the same energy content.

Table 9-3: Details of the meat and bone meal (MBM) co-firing tests [81].

Coal feed MBM feed Raw meal Clinker Specific

Test Number
rate rate feed rate free lime Energy Input
ton/h ton/h ton/h wt% MJ/kg clinker
1-1 7.8 0.0 220 1.27 1.55
1-2 6.5 2.0 220 1.69 1.56
1-3 5.3 4.0 220 1.42 1.57
1-4 4.5 5.0 220 1.51 1.55
1-5 3.9 6.0 220 2.29 1.56
1-6 3.2 7.0 220 3.03 1.56
1-7 3.2 7.0 210 2.92 1.63

Table 9-4: Details of the solid hazardous waste (SHW) co-firing tests [82].

Coal feed SHW feed Raw meal Clinker Specific

Test Number
rate rate feed rate free lime Energy Input
ton/h ton/h ton/h wt% MJ/kg clinker
2-1 7.6 0.0 218 0.99 1.48
2-2 7.0 0.5 227 1.68 1.36
2-3 6.7 1.0 226 2.30 1.36
2-4 6.4 1.5 226 1.99 1.35
2-5 5.7 2.0 226 2.49 1.27
2-6 5.6 2.5 223 3.19 1.33
2-7 5.9 2.5 216 2.23 1.43
2-8 5.9 2.5 221 1.91 1.40
2-9 5.7 3.0 221 1.83 1.41
2-10 7.4 0.0 220 1.05 1.43

The raw meal feed rate was also reported. This was converted to a clinker
production rate by using a factor 0.64. The assumption is that an average raw meal
9.2 Model Validation – Industrial Scale 183

feed rate of 220 t/h (5,280 t/day) results in 3,400 ton clinker per day. The standard
deviation on the free lime measurements were on average 0.16 wt%. This was
determined based on the steps in the MBM feed rate where two measurements of
free lime were performed [81].

The data from the tests indicate that an increased co-firing rate increases the free
lime content of the cement clinker. This is most likely caused by reduced flame
temperatures in the kiln as discussed in chapter 3.5.1.

Details of the fuels can be found in Table 9-5. Parameters for the Rosin-Rammler
particle size distribution were derived from the data presented in the articles. The
shape of the fuel particles has not been detailed in the articles. Thus, all particles
have been assumed spherical. This is perhaps a fair assumption for the coal and
MBM, but the wood chips from the SHW are more irregularly shaped. This affects
the sieving analysis of size, and the combustion characteristics of the fuel.

Additional details of parameters used in the model can be found in Appendix H.

Table 9-5: Details of the fuels used in the industrial tests.

Coal for Coal for

Property Unit MBM SHW
MBM tests SHW tests
Moisture wt. % 1.0 4.0 17.6 30.0
Volatiles wt. % 23.0 60.9 27.5 51.4
Char wt. % 62.4 8.0 51.4 1.0
Ash wt. % 13.6 27.1 19.4 17.6
LHV MJ/kg 28.0 18.5 27.1 15.0
Density kg/m3 1287 1354 1336 640
dRR µm 42.2 726.4 40.7 2663.1
nRR - 0.88 1.00 0.90 1.61

9.2.2 Model Results MBM Tests

The focus of this chapter will be on the differences between the free lime content
measured in the experiments of Ariyaratne et al. and the predictions of the
developed model. Example temperature profiles of the gas and bed are shown to
further understand the differences in the model predictions. In Appendix I results of
the coal case simulation are discussed in further detail. Temperatures and fuel
conversion through the kiln are presented in this appendix, among other things.

A comparison of the experimental and model results of free lime content in the
clinker during the MBM tests is shown in Figure 9-7. According to the experiments
184 9 Model Validation and SRF Co-Firing

the co-firing with MBM does not influence the free lime content significantly through
Tests 1-1 to 1-4 where the MBM feed rate is 5 t/h. It is only when the feed rate is
increased to 6 t/h and above during Test 1-5 to 1-7 that the free lime content
significantly increases. In comparison the model predictions show a steady increase
in free lime content as the MBM feed rate is increased through Tests 1-1 to 1-6. In
Test 1-7, the model predicts a sharp decrease in the free lime content as the raw
meal feed rate is decreased from 220 to 210 ton/h, while the experiments only show
a minor decrease.

Figure 9-7: Comparison of clinker free lime content measured in experimental tests
by Ariyaratne [81] and corresponding model calculations during MBM tests. Error
bars indicate standard deviation in measurements.

The model generally overpredicts the effect of co-firing MBM on the free lime
content. The mean absolute deviation is 40 % (RMSE = 0.94 wt%). Especially the
values of Test 1-3 and 1-4 are overpredicted by close to 100 %. The model predicts
a much larger effect of reducing the raw meal feed rate from 220 to 210 t/h during
Test 1-7, than measured in the experiments. During experiments, the lower feed rate
was only upheld for 1 hour, which may be too little time to see the full effects on the
free lime content.

The MBM contains 27 % ash with close to 50 % of the ash being calcium and the
other 50 % phosphorous [81]. The calcium introduced with the fuel may not have
sufficient time to combine with the clinker, which would increase the free lime
content, and the clinker. Moreover, phosphorous can stabilize belite and lower alite
9.2 Model Validation – Industrial Scale 185

formation and thus increase the free lime content [81,88]. According to Taylor [13]
higher contents than 0.2 % P2O5 decrease alite, and during Test 2-6 the clinker
phosphorous content is 0.5 wt% (see Figure 3-5). Neither of these effects are
included in the model. If they were to be included, they would increase the free lime
content, further increasing the overprediction of the free lime content.

Examples of the calculated gas and bed temperature profiles in the kiln are shown
in Figure 9-8. Observing the temperature profile for the coal case, Test 1-1, the gas
temperature is quickly increased as entrainment of the secondary air heats the
primary air and the fuel. This causes the coal to rapidly devolatilize, and the volatiles
are combusted to increase the temperature. The maximum temperature is reached
at around 8 m from the kiln clinker exit (3 m from the burner). The flame core is
now hotter than the surrounding secondary air, and entrainment of the remaining
secondary air cools the flame. The coal char is combusted, which for a while keeps
the gas temperature around 1900-2000 °C and results in a second temperature peak
at around 15 m. Afterwards the gas temperature gradually decreases as most of the
coal has been converted and heat is transferred to the bed. Further details of the
temperature profile for the coal case can be found in Appendix I.

Figure 9-8: Model simulation results of gas temperatures (a) and bed temperatures
(b) through the kiln for different co-firing scenarios with MBM.

As more MBM is added to flame, the gas temperature increases less rapidly. The
amount of coal is reduced, and the larger MBM particles take longer to heat and
devolatilize. Furthermore, the amount of combustion air is increased. The peak gas
temperature of 1850 °C is reached at around 23-25 m, which is where the largest
MBM particles drop into the bed. In comparison, the peak temperature in the coal
186 9 Model Validation and SRF Co-Firing

case is 2200 °C. The major difference is caused by a much lower adiabatic
temperature, caused by a higher air consumption of the MBM (see Appendix D). For
the remaining length of the kiln, the temperatures during MBM co-firing are
generally higher than for the coal case. This also leads to higher gas temperatures at
the raw meal inlet.

The bed temperatures initially increase slowly from the raw meal inlet side. The
higher gas temperatures at the raw meal inlet in the co-firing cases contribute to
heating the bed faster. Around 30-40 m from the clinker outlet, the temperature
increases rapidly. This is caused by the exothermal formation of belite (C2S). As this
reaction is complete at around 20-30 m, the temperature increase of the bed flattens
out. At this point the bed temperatures are higher for the co-fired cases, due to the
late combustion of MBM. However, as the clinker continues through the kiln, the
high peak temperatures of the coal combustion, results in significant increases to
the bed temperature, which accelerates the formation of alite and lowers free lime
(see Figure 9-7). As the clinker moves to the exit of the kiln, it is cooled by the
incoming secondary air and heat loss to the surroundings.

Comparing the cases Test 1-6 and 1-7, it is seen how the gas temperatures are almost
the same. However, the bed temperature increases faster during Test 1-7. The lower
flow of raw meal and clinker is easier to heat and the bed peak temperature is raised
by approximately 20 °C.

The outlet temperature of the clinker is lowered from 1500 °C in the coal case to
1400 °C when co-firing in Test 1-6 and 1-7. This lower discharge temperature of the
clinker, will result in lower temperatures in the clinker cooler, which will also
reduce the secondary air temperatures [252]. A simple model was formulated for
the clinker cooler in order to describe the effect of the clinker outlet temperature on
the secondary air temperature. The model is described in Appendix J. The result is
that the secondary air temperature is decreased from 750 °C in the coal case to 640
°C in Test 1-6 and 7. The lower secondary air temperature also contributes to
decreased combustion temperatures, and thus bed temperatures, during MBM co-
firing.
9.2 Model Validation – Industrial Scale 187

9.2.3 Model Results SHW Tests

A comparison of the experimental and model results of free lime content in the
clinker during the SHW tests is shown in Figure 9-9.

The free lime content in the coal firing scenario Test 2-1, is well described by the
model. The specific energy input during Test 2-1 is 1.48 MJ/kg clinker, compared to
1.55 MJ/kg clinker, during Test 1-1. However, the free lime content during Test 2-1
is lower at 0.99 wt% compared to 1.27 wt%. The kiln model accurately predicts a
free lime content of 0.98 wt% and 1.29 wt% for the two cases. The main difference
between the cases appears to be a difference in the clinker composition shown in
Table 9-6. The Lime Saturation Factor (LSF) is lower during the SHW tests, which
suggests, that a lower free lime content can be expected during the SHW tests. The
calculated C3S content by the Bogue formula, shows a higher C3S content can be
expected during the MBM tests. The model predicts a C3S content of 55.3 and 57.6
wt% during the SHW and MBM tests, respectively, which is similar to the Bogue
calculations, if corrected for the clinker free lime.

As seen in Table 9-5 there is also a slight difference in the composition and heating
value of the used coal. However, this difference is not sufficient to describe the
differences in the free lime content. All other model parameters related to the kiln
were kept constant in the two model cases.

Table 9-6: Average clinker oxide composition and calculated LSF, SR, AR, and Bogue
C3S (not corrected and corrected for free lime) during the MBM and SHW industrial
tests of Ariyaratne et al. [81,82].

Bogue Bogue
CaO SiO2 Al2O3 Fe2O3 Inerts LSF SR AR
C 3S C 3S
wt% wt% wt% wt% wt% - - - wt% wt%
MBM
63.53 20.37 5.37 3.45 7.28 0.97 2.31 1.56 62.8 57.6
Test
SHW
63.63 21.00 5.24 3.48 6.65 0.94 2.41 1.51 59.2 55.2
Tests

In Figure 9-9, the experiments show a steady increase in the free lime content during
Test 2-2 to 2-6, where the SHW is added to the flame in increments of 0.5 ton/h. In
Test 2-7 and 2-8 the SHW feed rate is kept constant at 2.5 ton/h, similar to Test 2-6.
The decrease in the free lime content compared to Test 2-6 is caused by an increase
of the coal feed by 0.3 ton/h and a reduction of the raw meal feed rate to 216 ton/h
188 9 Model Validation and SRF Co-Firing

in Test 2-7 and 221 ton/h in Test 2-8, compared to 223 ton/h in Test 2-6. In Test 2-
9 the SHW feed rate is increased to 3 ton/h, and a slight reduction is seen in the
experimentally determined free lime content. In Test 2-10 the SHW co-firing is
switched off, and a reduction in free lime content is seen. The free lime content is at
1.05 wt% slightly higher than the 0.99 wt% in Test 2-1, presumably caused by a
lower specific energy input.

Figure 9-9: Comparison of clinker free lime content measured in experimental tests
by Ariyaratne [82] and corresponding model calculations during SHW tests. Error
bars indicate standard deviation in measurements.

The overall trend of the model results follows the experimental results well. On
average the absolute deviation between model and experimental free lime is 17 %
(RMSE = 0.45 wt%).

The largest deviation is seen for Test 2-5 and Test 2-7. The model predicts the
largest free lime content for Test 2-5, where the specific energy input is the lowest
at just 1.27 MJ/kg clinker, compared to 1.32 MJ/kg during Test 2-6. In Test 2-7 the
raw meal feed rate is decreased to just 216 ton/h, which has a greater effect on
modeled free lime content, than the experimentally determined. In the MBM tests it
was also experienced that the model was more sensitive to changes in the raw meal
feed rate than the experiments.

In fact, there seems to be some transient effects on the experimental free lime
content. When settings are changed it will take some time for an industrial size kiln
to stabilize around new settings. For instance, the clinker residence time in the kiln
9.2 Model Validation – Industrial Scale 189

is typically around 30 minutes, which gives a considerable time lag for changes in
the raw meal feed rate. There may also be some thermal inertia in the kiln when
fuels are changed and the temperature profile shifts. This may be one of the reasons
the model predicts a local minimum for the free lime content during Test 2-7, while
the experiments show a steady decrease in the free lime content.

Model predictions for selected cases of the gas and bed temperature profiles
through the kiln are shown in Figure 9-10. The trends of the curves are similar to
that shown for the MBM cases. The peak gas temperature is lowered by adding SHW
to the flame, while for Test 2-7 a second temperature peak between 20-25 m is
created by the SHW dropping out of the flame and burning in the bed at this distance.
The lower gas temperatures are also reflected in lower bed temperatures, which
serve to decrease the free lime content. Test 2-7 has the lowest raw meal feed rate
and it can be seen how the bed temperature is increased more readily. It is only in
the last 15 m of the kiln that the bed temperature of the coal case becomes higher.

Figure 9-10: Model simulation results of gas temperatures (a) and bed temperatures
(b) through the kiln for different co-firing scenarios with SHW.

Comparing the graphs in Figure 9-8 and Figure 9-10 the temperatures during SHW
co-firing are les influenced than during MBM co-firing. This can be explained by the
lower degree of co-firing. A maximum of 7 ton/h MBM is co-fired, while only 3 t/h
SHW is co-fired, corresponding to 60 and 30 % of the energy input, respectively.
Furthermore, the air consumption per MJ of fuel is higher for the MBM than for SHW,
which increases the gas flow in the kiln, and reduces peak temperatures for MBM
co-firing.
190 9 Model Validation and SRF Co-Firing

9.3 Discussion of Model Predictions

The model has been compared to experimental results obtained in both pilot and
industrial scale. The ignition, entrainment, temperatures, and combustion rate of a
coal flame are described adequately when compared to the measurements from the
pilot scale kiln. The free lime content measured for the coal fired flame in the two
industrial test scenarios is also well determined by the model. The model responds
well to changes in fuel type when co-firing, energy input, and raw meal feed,
although the model tends to be more sensitive to these changes than what was
measured in the industrial tests. Overall, the model has shown good agreement with
the validation cases.

Normally it is expected that gas temperatures in the kiln are a maximum of around
2000 °C, while the clinker experiences temperatures up to 1450-1500 °C [31]. The
model predicts somewhat higher temperatures for both the gas phase and the bed,
peaking at 2200-2300 °C and almost 1600 °C, respectively. This can be explained by
some of the simplifications applied in the model

Sources in the literature determined that neglecting axial radiation would only lead
to errors of around 10 % of the total heat transfer [31,183]. Based on this it was
decided to neglect it in the model. If axial radiation was included it could serve to
reduce peak flame temperatures and flatten the temperature profile as more heat
would radiate away from the very hot areas.

As temperatures increase above 2000 °C the major combustion products CO2 and
H2O may start to dissociate into smaller products, which absorbs heat [253]. If this
occurs, it would lower the peak temperature, and release the heat later, when the
temperature has been lowered enough that H2O and CO2 become stable. Some
calculations were made separately from the model that indicate that the peak
temperature during coal combustion could be reduced by approx. 250-300 °C due
to dissociation.

It has been assumed that the gas phase of the kiln is dust free. In an industrial kiln
significant amounts of dust are entrained from the clinker cooler and the kiln.
Entrainment of around 0.03 kg dust per kg clinker, carried with the secondary air
from the cooler to the kiln is not uncommon [196]. Close to the burner entrained
9.4 Co-firing with SRF 191

dust in the secondary air would decrease the radiation from the flame to the bed,
and thus heat transfer and bed temperatures.

If the bed was to be assumed non-isothermal the surface could be much hotter than
the interior, perhaps up to 200 °C or more [227]. This hot surface would decrease
the temperature gradient between the gas and bed and lower heat transfer,
resulting in lower bed temperatures on average.

Even though the bed temperatures may be overestimated to some extent, the effect
of this is compensated by the relatively slow kinetics derived from laboratory
experiments. The result is that the free lime content calculated by the model agrees
well with industrial measurements.

9.4 Co-firing with SRF

SRF is one of the most commonly used alternative fuels in the cement industry. It
was also the main alternative fuel used at the three cement plants studied in the
large-scale experiments (see chapter 5 and 6). Therefore, it is desired to study the
effect of co-firing SRF in further detail using the developed model. As seen when co-
firing MBM or SHW, it is expected that co-firing with SRF will lower kiln
temperatures and increase the clinker free lime content, which was also observed
in the industrial tests of this PhD study. Consequently, it will also be investigated if
the firing can be altered to minimize the negative consequences of co-firing SRF.

9.4.1 SRF Properties

Two samples of SRF were characterized in detail in chapter 7. The properties of SRF-
B, will be used as input to the mathematical model.

The SRF sample has been divided into 5 size groups based on the wind sieve
fractionation. In each of these groups the content of plastics, biomass, textiles, inerts,
and fines was determined. For the model, the textiles will be assumed to be 50 %
biomass and 50 % plastic. Furthermore, the fines will be collected in one group,
assuming 30 % ash, 35 % biomass, and 35 % plastic content. It is expected that the
fine fraction has a low heating value, since moisture and ash tends to accumulate
amongst the smallest particles [72,100,162]. Thus, a high ash content is assumed in
the fine fraction. The inerts from the five wind sieve fractions and the ash from the
192 9 Model Validation and SRF Co-Firing

fine fraction will be collected in one group as well. This creates a total of 13 different
size groups for the model – 6 biomass, 6 plastic, and 1 inert group.

The fines were sieved to be < 2 mm, and it will be assumed that the average particle
diameter is 1 mm. The particle mass distribution was determined for the remaining
biomass and plastic fractions.

During the model validation (see Chapter 9.2) coal and alternative fuel was divided
into 10 discrete particle size categories, based on the given particle size distribution.
However, doing this for both the biomass and plastic fraction for each of the 5 wind
sieve subgroups would result in a system of 2x5x10 equations for the particles,
taking a considerable amount of time to solve. To reduce model calculation times,
only the mean particle mass in each subgroup is used.

The mean particle mass is taken as the mass of the particle at which 50 % of the
cumulative mass is achieved. This can be calculated based on the characteristic size
of the Rosin-Rammler distribution:
𝟏
𝒏𝑹𝑹
𝟏𝟎𝟎 − 𝟓𝟎 E 9.3
𝑪𝒎𝑹𝑹,𝟓𝟎 = (−𝒍𝒏 ( )) ∗ 𝒎𝑹𝑹
𝟏𝟎𝟎

The details of the composition and size of the SRF sample as implemented in the
model is given in Table 9-7.

Table 9-7: Distribution of the SRF composition, particle mass and equivalent
diameter, and sphericity used for simulating SRF co-firing.

Mean
Mean particle
Fraction Content (wt%) equivalent Sphericity
Mass (mg)
diameter (mm)
(Terminal
velocity Plastic Biomass Plastic Biomass Plastic Biomass Plastic Biomass
(m/s))
1 (<2) 2.22 0.53 6.5 7.1 2.4 2.8 0.10 0.15
2 (2-3) 5.04 3.23 20.7 18.2 3.5 3.9 0.15 0.28
3 (3-5) 15.69 22.53 34.2 25.0 4.1 4.3 0.26 0.39
4 (5-7) 9.28 11.51 154.5 90.5 6.8 6.6 0.39 0.49
5 (>7) 3.35 8.40 451.9 686.2 9.7 13.0 0.55 0.60
Fines 4.69 4.69 0.5 0.3 1.0 1.0 0.10 0.15
Inerts 8.84 - - - - - - -

The SRF heating value and proximate analysis of the SRF sample has been
determined and shown in Table 9-8. Factors such as heating rate and final
9.4 Co-firing with SRF 193

temperature has an influence on the char yield of biomass and SRF [254], thus the
proximate analysis typically overestimates the char yield [96]. Here it will be
assumed that the actual char yield is 70 % of that measured in the proximate
analysis, and the volatile content is correspondingly greater.

It is assumed that the plastic and ash fractions have fixed properties, and the
biomass properties are adjusted so the average composition matches that
determined for the SRF. The plastic fraction is assumed to form no char and have a
volatile content of 97.5 %, close to that of polyethylene [164] and an ash content of
2.5 %. The lower heating value is assumed as 35 MJ/kg, which is somewhat lower
than for pure polyethylene [95]. Since polyethylene and other plastics used for
packaging absorb only little water [177], it is assumed that the water stays on the
particle surface and the content is small, at 5 %. Thus, most of the SRF water content
is held by the biomass particles. The assumed dry and as fired composition is shown
in Table 9-8.

Table 9-8: Composition of the SRF and the Plastic, Inert, and Biomass fractions used
for modeling SRF co-firing.

Dry Basis Wet Basis

Property Unit SRF Plastic Inert Biomass SRF Plastic Inert Biomass
LHV MJ/kg 20.2 35 0 12.0 15.4 33.3 0 7.3
Moisture wt% 0 0 0 0 21.4 5.0 0 33.0
Ash wt% 22.4 2.5 100 24.7 17.6 2.4 100 16.5
Volatiles wt% 69.8 97.5 0 60.1 54.9 92.6 0 40.2
Fixed Carbon wt% 7.8 0 0 15.2 6.1 0 0 10.2
Adj. Volatiles wt% 72.2 97.5 0 64.7 56.7 92.6 0 43.3
Adj. Fixed
wt% 5.4 0 0 10.7 4.3 0 0 7.2
Carbon
Amount in
wt% - 40.3 8.8 50.9 - 33.3 7.0 59.7
sample

9.4.2 Effect of Co-Firing Rate

The effects of co-firing SRF ranging from 0 % to 100 % of the energy input has been
investigated using the developed kiln model. The employed coal, SRF, and raw meal
feed rates are given in Table 9-9. The same conditions and model inputs employed
during the MBM validation are used here. The only difference is the SRF fuel
properties. The parameters used in the model calculations are detailed in Appendix
H.
194 9 Model Validation and SRF Co-Firing

Table 9-9: The feed rate of coal, SRF, and raw meal used for simulating the effect of
SRF co-firing.

SRF Energy Coal SRF Raw Meal

Input [%] [t/h] [t/h] [t/h]
0 7.80 0.00 220
10 7.02 1.42 220
20 6.24 2.84 220
30 5.46 4.26 220
40 4.68 5.68 220
50 3.90 7.10 220
60 3.12 8.52 220
70 2.34 9.94 220
80 1.56 11.4 220
90 0.78 12.8 220
100 0.00 14.2 220

Figure 9-11: The clinker free lime content as the share of SRF energy input is
increased.

The free lime content from the simulations is shown in Figure 9-11. The free lime
content steadily increases as the SRF energy input is increased with a plateau
present around an SRF energy input of 30-50 %. The upper limit for the free lime
concentration is 2.5 % [81]. This limit is reached at 60 % SRF energy input. In
addition to lowering the alite content and cement strength [2], excessive free lime
may also cause concrete expansion [13]

Temperature profiles in the gas and bed for selected degrees of co-firing are shown
in Figure 9-12. The temperature trends are similar to those explained for the MBM
and SHW co-firing in chapters 9.2.2 and 9.2.3. As more SRF is added to the flame the
9.4 Co-firing with SRF 195

initial peak temperature around 10 m is lowered. This is caused by a lower flow rate
of coal and a higher flow rate of air, since the SRF combustion requires more air per
energy content (see Appendix D). Furthermore, the colder clinker leaving the kiln,
results in reduced secondary air temperatures.

Figure 9-12: Model simulation results of gas temperatures (a) and bed temperatures
(b) for different energy inputs of SRF.

The residence time in suspension for the SRF particles is around 1.5 seconds. This is
only sufficient to devolatilize the smallest of the SRF particles, with a terminal
velocity lower than 2 m/s (see additional details in Appendix I). The larger particles
are only converted when they land in the bed at around 20 m. Thus, for co-firing, a
second temperature peak in the gas is present around 25 m. This is especially clear
for the case with 100 % SRF firing, where the temperature plateaus at around 500
°C until 18 m, where the SRF is converted as it enters the bed.

The lower gas temperature obtained during co-firing also influence the bed
temperatures, which are significantly reduced, which serves to increase the free
lime content. The peak bed temperature is also shifted further into the kiln, caused
by the delayed combustion of SRF. This results in lower clinker temperatures at the
kiln outlet, which cause lower secondary air temperatures.

According to the model, the free lime content does not change as the SRF co-firing
rate is increased from 30 to 40 %. In Figure 9-12b it observed that the temperature
profiles of the two cases are quite similar, with the same maximum temperature.
The temperature for the 40 % co-firing scenario is just shifted further into the kiln.

The 30 % SRF firing case is discussed in further detail in Appendix I.

196 9 Model Validation and SRF Co-Firing

9.5 Possibilities to Increase SRF Co-Firing

It is evident that co-firing with SRF lowers the temperatures in the kiln, which has a
negative impact on the free lime content. In the following chapters, it will be
investigated if the impact of co-firing can be lessened, by changing the SRF quality
or other process parameters. The 30 % co-firing case is selected as the base scenario
to study improvements.

9.5.1 Effect of Moisture Content

The SRF contains 21.4 wt% water, a high amount compared to the 1.3 wt% of the
coal. The moisture content in SRF may also vary considerably over time.
Furthermore, it was determined during the industrial tests, that the moisture
content would have a negative effect on the clinker alite content (see chapter 6.4.1).
Consequently, investigating the effect of moisture content is relevant.

The moisture content of the SRF was set to vary between 0 and 30 wt%, while the
energy input was kept constant. Accordingly, the SRF feed was increased from 3.3
to 4.9 ton/h, to compensate for the lower heating value, when the moisture content
increased.

Figure 9-4 shows the free lime content in clinker caused by changes in the moisture
content of SRF. The free lime content changes from 1.8 to 2.2 wt% when the SRF
moisture content is increased from 0 to 30 wt%.

Figure 9-13: The clinker free lime content for different moisture content in the SRF.
SRF energy input is 30 %.
9.5 Possibilities to Increase SRF Co-Firing 197

Figure 9-14 shows the gas and bed temperature profiles through the kiln for
moisture contents of 0, 10, 20, and 30 wt%. The gas temperatures are seen to be
decreased by added moisture. The peak gas temperature at 26 m is decreased by
approximately 40 °C when the moisture content is increased from 0 to 30 wt%. The
added moisture increases the flue gas amount in the kiln for the same energy input
and will lead to reduced flame temperatures. In addition, the devolatilization of the
SRF will be delayed due to the energy intensive water evaporation. The lower gas
temperatures also cause lower temperatures in the bed.

Figure 9-14: Model simulation results of gas temperatures (a) and bed temperatures
(b) for different SRF moisture contents. SRF energy input is 30 %.

9.5.2 Effect of Particle Size

Another factor that contributes to lower combustion temperatures during co-firing
with SRF is the large particle size. The largest coal particles in the simulations are
170 µm in diameter, while the SRF particle diameter ranges from 1 to 13 mm. This
increases the heating time of the particles and delays their combustion.

To investigate the effect of particle size, a factor ranging from 0.1 to 1 is multiplied
to the original particle diameter of the SRF, which decreases the particle size. Figure
9-15 shows the effect of lowering the SRF particle size. The free lime content
decreases from 2.0 wt% in the base scenario to 1.3 wt% with SRF particles 1/10 the
size of the base scenario. This is a free lime content that is comparable to the one
obtained when firing only coal.
198 9 Model Validation and SRF Co-Firing

Figure 9-15: The clinker free lime content for different SRF particle sizes. The
standard SRF particle diameter is multiplied by a factor in the range 0.1-1. SRF energy
input is 30 %.

The temperature profiles in the gas and bed are shown in Figure 9-16 for select SRF
particle sizes. As the particle diameter is reduced, the SRF particles will burn closer
to the burner and the peak gas temperature is increased and moved closer to the
burner. This also affects the bed with increased temperatures.

Figure 9-16: Model simulation results of gas temperatures (a) and bed temperatures
(b) for different SRF particle sizes. The indicated factor is multiplied to the original
particle diameter. SRF energy input is 30 %.
9.5 Possibilities to Increase SRF Co-Firing 199

9.5.3 Effect of Increased Specific Energy Input

The specific energy input is as standard 1.55 MJ/kg clinker. To reduce the negative
effects that co-firing of SRF has on the free lime content, the energy input to the kiln
can be increased. Figure 9-17 shows the effects of increasing the specific energy
input by up to 10 %. This is done in three different ways:

1. The coal and SRF feed rate is increased to keep the energy input of SRF at 30
% of the total.
2. The feed rate of SRF is increased to match the desired increase in specific
energy input, while the coal input is kept constant.
3. The raw meal feed rate and thus clinker production rate is decreased.

All three methods can be used to lower the free lime content in the clinker during
co-firing. Reducing the raw meal feed rate is more efficient than increasing the fuel
input. Increasing both the coal and SRF feed rate is slightly more effective than only
raising the SRF feed rate. To obtain a free lime content of 1.3 wt%, which is
comparable to coal firing, the energy input should be increased by slightly more than
10 %, or the raw meal feed rate should be decreased by 5 %.

Figure 9-17: The clinker free lime content for increasing specific energy input by 1.
Increasing coal and SRF feed rate, 2. Increasing SRF feed rate, 3. Decreasing the raw
meal feed rate. SRF energy input is 30 % in case 1 and 3, but changes in case 2.

Changes to the temperature profiles of the gas and bed caused by increases in the
specific energy input by reducing the raw meal feed is shown in Figure 9-18.
200 9 Model Validation and SRF Co-Firing

Figure 9-18: Model simulation results of gas temperatures (a) and bed temperatures
(b) for different increases in specific energy consumption by reduction of raw meal
feed. SRF energy input is 30 %.

The gas temperature is only increased slightly by lowering the raw meal feed. For
instance, the peak gas temperature increased by 20 °C, when the specific energy
input is increased by 10 %. The effects on the bed temperature are somewhat
greater. The bed temperature increases more rapidly, and the peak temperature is
increased by 40 °C.

9.5.4 Effect of Burner Settings

Based on the experimental results of the tests on the Jetflex burner (see chapter 6),
it was concluded that the burner settings could be adjusted to increase the clinker
quality when co-firing. Increased swirl and dispersion of SRF was found to be
beneficial for co-firing SRF.

The model was developed with the goal that it should be able to help understand
how burner settings can be used to increase the co-firing rate of alternative fuels.
For this purpose, the mixing and eddy dissipation rates were derived based on CFD
modeling, considering the burner axial momentum and swirl number (see chapters
8.6.5 and 8.8.6.1).

The burner settings were tested on the co-firing scenario using 30 % SRF. The axial
momentum and swirl number were changed 50 % around their original values of
5.8 N/MW and 0.26, in four new cases as shown in Table 9-10. The corresponding
effect on the entrainment rate of the jet and the turbulent eddy dissipation rate is
also shown.
9.5 Possibilities to Increase SRF Co-Firing 201

Table 9-10: Axial momentum and swirl number used to determine the effect of burner
settings on clinker free lime and the corresponding values of Kent,jet and ε/k calculated
at 10 and 25 m from the clinker exit.

Case 1 Case 2 Case 3 Case 4 Case 5

Axial Momentum
5.8 8.7 8.7 2.9 2.9
(N/MW)
Swirl Number 0.26 0.39 0.13 0.39 0.13
Kent,jet 0.21 0.26 0.19 0.23 0.16
ε/k at 10 m (1/s) 9.2 9.4 12.9 5.9 8.9
ε/k at 25 m (1/s) 0.8 0.8 1.2 0.5 0.9

The model predicted free lime content for the five burner setting scenarios is shown
in Figure 9-19. The free lime content is only decreased slightly in case 2 and 3
employing a 50 % higher axial momentum. Employing a 50 % lower axial
momentum in case 4 and 5 has a more significant effect on increasing the free lime
content. The swirl has less of an impact, but a higher swirl tends to increase the free
lime.

Figure 9-19: The clinker free lime content for different cases of burner settings. SRF
energy input is 30 %.

Gas and bed temperature profiles through the kiln for the different burner setting
cases are shown in Figure 9-20. The burner settings influence the peak gas
temperatures around 10 and 25 m. Case 5 has the lowest entrainment rate, which
results in the highest peak temperature during the coal devolatilization around 10
m. Case 2 and 4 have the highest entrainment rates and the lowest peak
temperatures around 10 m.
202 9 Model Validation and SRF Co-Firing

Figure 9-20: Model simulation results of gas temperatures (a) and bed temperatures
(b) for different burner setting cases. SRF energy input is 30 %.

The highest peak temperature at 25 m is obtained by case 3 and the lowest by case
4. This difference is related to the eddy dissipation rate, which control the rate of
volatile combustion. The higher rate of volatile combustion benefits the gas
temperature and increases the bed temperature causing the lowest clinker free lime
content for case 3. Conversely, case 4 settings result in the highest free lime content.

According to the model, both axial momentum and swirl influence the entrainment
rate and the eddy dissipation rate. The gas temperatures are more sensitive to
changes in the eddy dissipation rate, which control the volatile combustion. A high
combustion rate is best achieved by increasing the axial momentum and lowering
the swirl.

9.5.4.1 Effect of the Entrainment Rate

Changing the burner momentum and swirl influences both the entrainment rate and
the volatile combustion rate. In the following it is desired to only change the
entrainment rate. The standard value, of Kent,jet = 0.21, is used and factors ranging
from 0.5 to 1.5 are employed to vary the value of Kent,jet between 0.11 and 0.32.

Figure 9-21 shows the effect of changing the entrainment rate on the free lime
content. Lower entrainment rates result in a higher free lime content. As standard
the free lime content is 2.0 wt%, which can be increased to 2.5 wt% or decreased to
1.8 wt% by adjusting the entrainment rate.
9.5 Possibilities to Increase SRF Co-Firing 203

Figure 9-21: The clinker free lime content for different secondary air entrainment
rates. The standard entrainment rate is multiplied by a factor in the range 0.5-1.5. SRF
energy input is 30 %.

Figure 9-22 shows gas and bed temperature profiles in the kiln for the different
values of the entrainment factor. With a low entrainment rate (factor 0.5), a gas peak
temperature of 2050 °C is reached at 10 m. With a higher entrainment rate the gas
temperature at this location is reduced to 1650 and 1500 °C, for a factor 1 and 1.5,
respectively. The higher flow of gas caused by a fast entrainment of the secondary
air reduces the peak temperature. However, after the mixing is complete, the case
with the highest entrainment rate maintains the highest gas temperature. The
higher entrainment initially also causes a higher heating rate of the primary gas, as
the incoming secondary air heats and devolatilizes the coal.

Even though gas temperatures are significantly higher with a low entrainment rate,
the heat transfer to the bed is impeded. The enveloping secondary air prevents the
hot flame gasses from convective heat transfer to the bed. When the jet diameter
expands faster, the larger surface area also increases the radiative heat transfer.
Consequently, a higher entrainment rate is beneficial.
204 9 Model Validation and SRF Co-Firing

Figure 9-22: Model simulation results of gas temperatures (a) and bed temperatures
(b) for different entrainment rates. The indicated factor is multiplied to the original
entrainment rate. SRF energy input is 30 %.

9.5.4.2 Effect of Fuel Landing Place

In the model the SRF particles are assumed to be well dispersed in the primary air
jet, until they drop into the bed, where they are instantly converted. The effect of
burner settings on the particle trajectories have not been considered in the model.
However, the industrial tests on the Jetflex burner (see chapter 6) showed that
increased swirl was beneficial for the co-firing of SRF. The swirl particularly
influenced the dispersion of SRF near the burner. The increased dispersion will also
affect the distance at which the particles land in the bed. This has been confirmed by
the CFD model study by Liedmann et al. [143], as discussed in chapter 8.1.2.

The effect of the particle landing was studied in the model by calculating the particle
landing place, lf,drop, as described in chapter 8.8.5. A factor between 0.5 and 1 was
then multiplied to the calculated value, which forces the particles into the bed
earlier. The effect of the landing place is shown in Figure 9-23. With the normally
calculated value of the landing place (Fuel Landing Factor = 1), the free lime content
is 2.0 %. As the particles drop into the bed earlier the free lime content is decreased,
to a minimum value of 1.4 wt%, using a factor of 0.65. This is only 0.1 wt% higher
than the coal fired case. Decreasing the factor further leads to a slight increase in the
free lime content, which is 1.5 wt% using a factor of 0.5.
9.5 Possibilities to Increase SRF Co-Firing 205

Figure 9-23: The clinker free lime content for different landing places of the SRF. The
standard landing place is multiplied by a factor in the range 0.5-1. SRF energy input is
30 %.

The temperature profiles of the gas and bed of select cases are shown in Figure 9-24.
The main effect of the particles dropping into the bed closer to the burner is a faster
conversion and heat release from the fuel. This results in faster temperature
increase in the gas phase and the peak related to SRF firing originally around 25 m
is moved closer to the burner by 10 m and increased by 100 °C.

Figure 9-24: Model simulation results of gas temperatures (a) and bed temperatures
(b) for different SRF particle landing factors. The indicated factor is multiplied to the
original landing place. SRF energy input is 30 %.

The changes in the gas phase temperature also increase the bed temperature,
resulting in higher peaks as the fuel burns closer to the burner. The free lime content
206 9 Model Validation and SRF Co-Firing

is lowest when the fuel landing factor is 0.65, even though higher bed temperatures
are reached when it is 0.5. When the factor is 0.65, the bed is heated faster, and the
maximum bed temperature is almost constant for 3-4 m, while for the factor 0.5, the
bed temperature is more peaked. The heating rate of the bed is slower, which results
in shorter residence time of the clinker at high temperature.

9.5.5 Discussion of Options to Increase Co-Firing

In the above sections different methods to limit the negative effects of co-firing have
been reviewed. Based on the findings, some suggestions for increasing co-firing of
SRF are proposed. The conclusions are likely also valid for other types of alternative
fuels.

For the investigated simulation case it can be expected that 30-40 % SRF can be used
in the kiln without making significant changes to the kiln operation. This results in
an increase in the free lime content from 1.3 wt% when only firing coal to 2 wt%
when co-firing 30 % energy SRF. The kiln system, ID fan etc., should however be able
to cope with the increased amount of flue gas formed. At 30 % co-firing, the flue gas
mass flow rate is increased by 4 %.

If the increase in free lime, and consequently lower alite content, cannot be
tolerated, or higher SRF co-firing is desired, the first step should be to increase the
specific energy consumption in the kiln by increasing the SRF feed rate. If SRF can
be obtained cheaply at the cement plant, increased firing will not significantly
increase the production costs of the cement. Lowering the production rate has been
shown to be more effective on the free lime content, but it is likely to be a more
expensive option.

The optimal burner settings for co-firing should also be investigated. The model
shows that increased burner momentum, increased entrainment rate of secondary
air, or increased dispersion of the SRF could help decrease the clinker free lime
content when co-firing SRF. If the burner at the plant does not allow any significant
flexibility, installing a new burner should be considered.

After the above possibilities have been exhausted, changes to the SRF can be
considered. This approach will require new equipment for SRF drying and
comminution. Drying of SRF could possibly be achieved by utilizing hot waste air
9.6 Conclusions 207

from the process. Drying of the SRF from 20 to 0 % moisture could decrease the
clinker free lime content from 2.0 to 1.8 wt%, when utilizing 30 % SRF by energy.

A size reduction of the fuel can also be considered. A factor 10 reduction of the SRF
particle size resulted in a decrease in free lime content from 2.0 to 1.3 wt%, when
utilizing 30 % SRF by energy. However, comminution of SRF is difficult, and size
reduction should be one of the last options to be pursued.

Other options to increase SRF co-firing that have not been investigated here are e.g.
oxygen enrichment.

It should be emphasized that an increased clinker production may be pursued

instead of alternative fuels firing, if there is enough margin on the free lime content.
The choice comes down to a focus on maximizing production contra minimizing
production costs, and the best choice for any given cement plant would depend on
its market conditions.

9.6 Conclusions
A simplified 1-D model for calculating gas and bed temperatures in the kiln has been
developed. The model couples the calculated temperature profiles to the clinker
quality, measured as the free lime content. To the author’s knowledge, this is the
first model that attempts to describe the effects of co-firing alternative fuels by
coupling the combustion process and temperatures to the clinker quality.

The model was validated against pilot experiments showing appropriate

temperature profiles, gas species concentration, and fuel burnout, for a coal flame
without any clinker. Industrial tests described in the literature examined the effect
of co-firing meat and bone meal (MBM) and solid hazardous waste (SHW) on the
clinker free lime content. The model adequately described the trends of co-firing
and specific energy input on the free lime content.

The model was used to investigate co-firing of solid recovered fuel (SRF). Co-firing
with SRF reduced gas and bed temperatures in the kiln and resulted in an increased
clinker free lime content compared to pure coal firing. This is caused by lower
combustion temperatures of the SRF due to a higher air requirement per energy
content. The larger particle size and high moisture content of the SRF also lead to a
slower combustion rate, which moves the peak temperatures further downstream
208 9 Model Validation and SRF Co-Firing

in the kiln. This results in reduced temperatures of the clinker that exits the kiln.
Consequently, the secondary air temperature is lowered, which also contributes
negatively to the temperatures in the kiln.

Different approaches to limit the negative impact of co-firing were investigated

using the model. The most promising is to increase the specific energy input of the
kiln, by increasing the feed rate of SRF. As an example, when co-firing 30 % SRF by
energy, the clinker free lime content is increased from 1.3 to 2.0 wt%. Increasing the
SRF feed rate by 10 %, reduces the free lime to 1.4 wt%. Another option is to
optimize the burner operation. This may be done by increasing the entrainment rate
of secondary air or increase the dispersion of SRF to cause it to burn closer to the
burner. This generally increases the bed temperatures and lower clinker free lime.
The industrial experiments conducted during this PhD study, confirmed that burner
settings, especially increased dispersion of SRF, could increase the clinker quality
when co-firing SRF.
10 FINAL CONCLUSIONS

Through the last 20-30 years the cement industry has increased the use of
alternative fuels, substituting conventional fossil fuels such as coal and oil. The
development has been driven by the possibility to lower operating costs and CO2
emissions. Furthermore, the use of waste derived fuels in the cement industry
reduces the need for landfilling. Alternative fuels can be introduced both in the
cement calciner and rotary kiln. The focus of this thesis has been on co-firing in the
kiln.

Tests were carried out at three different full-scale cement plants where fossil fired
and solid recovered fuel (SRF) co-fired flames were compared, using a specially
developed water cooled camera probe. The probe can be inserted into the cement
kiln hood, which makes it possible to observe the flames in detail, investigating e.g.
ignition point and fuel flow. It was found that co-firing between 30 and 70 % SRF by
energy, would delay ignition by approximately 2 meters. The intensity and
temperatures of the flames were also lowered. This is caused by the large particle
size and high moisture content of the SRF, which takes longer to ignite than fossil
fuel. A decrease in kiln drive power consumption was also observed, which further
suggests lower temperatures in the kiln. The lower temperature may negatively
affect the clinker quality by decreasing the formation of alite.

Furthermore, it was observed how some of the SRF drops out of the flame and burns
in contact with the clinker bed. At two of the plants this led to problems with sulfur
evaporation or reduced brown clinker. The last plant appeared to be more robust to
210 10 Final Conclusions

these issues related to local reducing conditions. The difference in robustness is not
fully understood at the present, but it may be related to sulfur modulus or oxygen
availability near the clinker bed.

Recently, FLSmidth A/S developed the Jetflex burner, which was specifically
developed for burning alternative fuels. At one of the test plants, the burner was
changed to the Jetflex burner. The main difference in burner design was a change
from an annular axial air channel to axial air jets. In addition, a more powerful swirl
channel was installed, which was placed inside the coal annular channel, thus being
closer to the central alternative fuel channel. The design changes benefited the
ignition of petcoke and made it possible to disperse SRF in the near burner zone.
Measurements of the clinker quality showed a 4 % increase in the alite content when
operating with the new burner, which was achieved with an 8 % higher use of SRF.

Additional tests were carried out with the Jetflex burner. The burner was designed
with several different possibilities to influence the flame shape. In addition to the
conventional changes to burner momentum and swirl intensity, it is also possible to
change the direction of the axial air jets and mix pulverized fossil fuel and SRF in a
common central channel. The effect of the different changes was studied using the
kiln camera probe. For a petcoke flame the ignition point could be changed from
approximately 5 to 3.5 meters by adjusting the swirl and axial air jets. SRF co-fired
flames were more difficult to ignite, but dispersion of the SRF into the hot secondary
air by a high swirl level was possible, which helped ignite the SRF particles.

A statistical analysis of collected operating data using partial least squares

regression (PLSR) revealed that higher swirl and directing the axial air nozzles
outwards, contribute to increase the clinker alite content when co-firing SRF. Firing
petcoke through the central channel with SRF and retracting this channel was also
found to have a positive effect. Overall these measures contribute to disperse the
SRF near the burner, which presumably increases combustion temperatures in the
near burner field during co-firing.

Different SRF samples were collected from cement plants and a physical
characterization was performed to study the SRF composition, particle size and
shape. The SRF samples were first classified into different terminal velocity ranges
using a wind sieve setup. The setup may provide an improved distinction between
particle sizes, compared to a normal sieving analysis, which may not be suited for
10 Final Conclusions 211

irregular shaped particles such as SRF. It was shown how light fuels, suited for use
in the cement kiln, could be distinguished from heavier fuels, suited for use in the
calciner. Based on the initial measurements it is recommended that SRF for use in
the main burner should have at least 20 wt% of the sample with a terminal velocity
lower than 3 m/s, 60 wt% lower than 5 m/s, and 80 wt% lower than 7 m/s.

Two SRF samples separated in the wind sieve were selected for further detailed
analysis. The fraction of biomass and plastic in the samples was determined.
Furthermore, the distribution of particle mass and the particle shape, defined as the
sphericity, were determined. The average particle mass of samples with a terminal
velocity below 2 m/s was around 10 mg, while the mass for particles with a terminal
velocity greater than 7 m/s was 600 mg. This corresponds roughly to equivalent
diameters of 3 and 12 mm. The shape was also found to change with the terminal
velocity. Lighter particles had an average sphericity of 0.1, while the heavier
particles had a sphericity of 0.5. With lower mass and higher surface area, it is
expected that the particles with low terminal velocity burn significantly faster than
those with high terminal velocity.

A one-dimensional model was developed to help understand the impact of co-firing

on the temperatures in the kiln. The model describes temperature changes in the
gas and clinker bed in the axial direction of the kiln. A simplified clinker model was
developed to couple the bed temperature to a clinker quality, measured as the
residual free lime content. The model was validated using results obtained from the
literature. The coal flame model was validated against experiments performed in a
pilot scale rotary kiln simulator, without clinker. Co-firing of meat and bone meal
(MBM) and solid hazardous waste (SHW) were also tested with the model, with
predictions of clinker free lime content agreeing well with results from industrial
tests. The absolute deviation between the free lime content calculated by the model
and the experiments was on average 40 % and 20 %, for the MBM and SHW cases,
respectively.

Furthermore, the model was used to examine the effects of SRF co-firing on
temperatures in the kiln and the clinker free lime content. It was found that co-firing
reduced gas and bed temperatures in the kiln, while the peak temperatures were
moved further away from the burner. The reduced temperatures resulted in an
increased free lime content of the clinker, indicating a reduced clinker quality. The
212 10 Final Conclusions

lower combustion temperatures can be explained by a higher air consumption for

the same energy input of SRF compared to coal. In addition, the large and wet SRF
particles take longer to convert than coal particles, shifting the temperature peak
further into the kiln. Much of the SRF was found to not be converted while in
suspension, having a residence time around one second. Only particles with a
terminal velocity lower than 2 m/s (as determined in the wind sieve), can be
expected to be devolatilized while in suspension in the kiln. Thus, especially larger
particles end up burning in the bed. The camera observations of the industrial co-
fired flames confirm these findings.

Different methods to decrease the negative impacts of co-firing were investigated.

Increasing the energy input of SRF may be the simplest method to mitigate the
effects of reduced temperatures. As an example, co-firing of 30 % SRF by energy,
was found to increase the clinker free lime to 2.1 wt% compared to 1.3 wt% when
firing only coal. By increasing the specific fuel energy input to the kiln by 10 %, it
was possible to reduce the free lime content to 1.4 wt%.

Alternatively, increased entrainment of secondary air and dispersion of SRF closer

to the burner can also help increase clinker temperatures and decrease clinker free
lime. This finding agrees with the conclusions of the performed industrial tests.
Pretreatment of the SRF by drying and particle comminution can decrease the
conversion time and reduce the clinker free lime content. However, this may be
expensive options that require investments in new equipment.

10.1 Suggestions for Future Work

In the industrial tests it was experienced that SRF would burn in contact with the
clinker at all three cement plants where tests were performed. At one plant this was
found to be very severe, while it was less of an issue at the two other plants. The
cause appears to be a difference in the sensitivity towards local reducing conditions,
in the form of sulfur evaporation and brown cores in the clinker. A further
understanding of what imposes these limitations and how they can be avoided,
would serve to increase the possibility for alternative fuels firing.

The characterization of SRF using the wind sieve setup has shown that it is possible
to separate particles with different particle size and shape. These factors heavily
influence the combustion time. Only a limited number of samples have been
10 Final Conclusions 213

analyzed. A more rigorous analysis of samples from different cement plants, coupled
with a knowledge of the limitations and specific challenges experienced at those
plants, would be highly beneficial. This could serve as a basis to uncover the
relationship between SRF quality and corresponding limitations to co-firing.

The developed model relies on many simplifying assumptions. However, its

predictions are still reasonable. It may be hard to significantly increase the accuracy
of the model, keeping the model one-dimensional. Especially, the flow of gas and
effect of burner settings, which were derived from CFD simulations, and the fuel flow
is difficult to resolve in a simple one-dimensional model. Thus, moving forward it is
recommended to focus mainly on the use of 3-dimensional CFD simulations.
However, the use of a more advanced model is only justified if it can be validated. In
addition to the flame images provided here, measurements of temperatures and gas
concentrations in a cement kiln, would be desirable.
214
11 REFERENCES

[1] K. L. Scrivener and R. J. Kirkpatrick, “Innovation in use and research on cementitious

material,” Cement and Concrete Research, vol. 38, no. 2, pp. 128–136, 2008.

[2] S. Telschow, F. Frandsen, K. Theisen, and K. Dam-Johansen, “Cement Formation—A Success

Story in a Black Box: High Temperature Phase Formation of Portland Cement Clinker,”
Industrial & Engineering Chemistry Research, vol. 51, no. 34, pp. 10983–11004, 2012.

[3] R. R. Wertz, “Three Gorges Dam,” 2011. [Online]. Available:

http://www.ibiblio.org/chinesehistory/contents/07spe/specrep01.html. [Accessed: 18-
Apr-2018].

[4] P. C. Hewlett, Lea’s Chemistry of Cement and Concrete, 4th ed. Butterworth-Heinemann,
1998.

[5] J. M. Crow, “The concrete conundrum,” Chemistry World, pp. 62–66, 2008.

[6] G. C. Bye, Portland Cement, 2nd ed. Thomas Telford, 1999.

[7] US Geological Survey, “Mineral Commodities Summaries 2018,” Reston, Virginia, USA, 2018.

[8] K. A. Baumert, T. Herzog, and J. Pershing, “Navigating the numbers,” World Resources
Institute, Washington, DC, USA, 2005.

[9] International Energy Agency (IEA), “Energy Technology Transitions for industry,” Paris,
France, 2009.

[10] E. Mokrzycki and A. Uliasz- Bochenczyk, “Alternative fuels for the cement industry,” Applied
Energy, vol. 74, no. 1–2, pp. 95–100, 2003.

[11] F. Schorct, I. Kourti, B. M. Scalet, S. Roudier, and L. D. Sancho, “Best Available Techniques
(BAT) Reference Document for the Production of Cement, Lime and Magnesium Oxide,”
Sevilla, Spain, 2013.

[12] International Energy Agency (IEA), Energy Technology Perspectives: Scenarios & Strategies
216 11 References

To 2050. Paris, France, 2010.

[13] H. F. W. Taylor, Cement Chemistry, 2nd ed. London, United Kingdom: Thomas Telford, 1997.

[14] A. R. Nielsen, “Combustion of Large Solid Fuels in Cement Rotary Kilns,” PhD Thesis, Technical
University of Denmark, 2012.

[15] O. Labahn and B. Kohlhaas, Cement Engineers’ Handbook, 4th ed. Bauverlag GMBH,
Wiesbaden, Germany, 1983.

[16] F. M. Miller, “Raw Mix Design Considerations,” in Innovations in Portland Cement

Manufacturing, 1st ed., J. I. Bhatty, F. M. Miller, S. H. Kosmatka, and R. P. Bohan, Eds. Skokie,
Illinois, USA: Portland Cement Association, 2011, pp. 193–207.

[17] G. L. Young and F. M. Miller, “Kiln System Operations in Cement Manufacturing,” in

Innovations in Portland Cement Manufacturing, 1st ed., J. I. Bhatty, F. M. Miller, S. H. Kosmatka,
and R. P. Bohan, Eds. Skokie, Illinois, USA: Portland Cement Association, 2011, pp. 311–342.

[18] FLSmidth A/S, “Preheater Calciner Systems,” Copenhagen, Denmark, 2013.

[19] IEA, “Cement Technology Roadmap: Carbon Emissions Reductions up to 2050,” Paris, France,
2009.

[20] CEN (European Committee for Standardization), “EN 197-1:2011; Cement – Part 1:
Composition, specifications and conformity criteria for common cements.” 2011.

[21] J. Thomas and H. Jennings, “Mineral and Oxide Composition of Portland Cement.” [Online].
Available: http://iti.northwestern.edu/cement/monograph/Monograph3_6.html. [Accessed:
05-Apr-2018].

[22] K. Theisen, “The International Cement Production Seminar 2007, Volume II.” FLSmidth A/S,
Valby, Denmark, 2007.

[23] K. Theisen, E. Jøns, and B. Osbæck, “Cement Clinkering: Chemical and Physical Aspects,” in
Innovations in Portland Cement Manufacturing, 1st ed., J. I. Bhatty, F. M. Miller, S. H. Kosmatka,
and R. P. Bohan, Eds. Skokie, Illinois, USA: Portland Cement Association, 2011, pp. 365–400.

[24] R. H. Bogue, “Calculation of the Compounds in Portland Cement,” Industrial and Engineering
Chemistry Analytical Edition, vol. 1, no. 4, pp. 192–197, 1929.

[25] I. F. Petersen and V. Johansen, “Burnability and clinker nodule formation from a statistical
point of view,” Cement and Concrete Research, vol. 9, no. 5, pp. 631–639, 1979.

[26] L. M. Hills, V. Johansen, and F. M. Miller, “Solving Raw Material Challenges,” in IEEE-IAS/PCA
44th Cement Industry Technical Conference, 2002, pp. 139–151.

[27] A. A. Boateng, Rotary Kilns Transport Phenomena and Transport Process, 1st ed. Oxford, UK:
Butterworth-Heinemann, 2008.

[28] M. B. Larsen, “Alternative Fuels in Cement Production,” PhD Thesis, Technical University of
Denmark, Kgs. Lyngby, Denmark, 2007.
11 References 217

[29] FLSmidth A/S, “ACC Wadi’s expanded plant sets world records,” 2011. .

[30] R. Wirthwein and B. Emberger, “Burners for alternative fuels utilisation - Optimization of kiln
firing systems for advanced alternative fuel co-firing,” Cement International, vol. 8, no. 4, pp.
42–46, 2010.

[31] L. K. Nørskov, “Combustion of Solid Alternative Fuels in the Cement Kiln Burner,” PhD Thesis,
Technical University of Denmark, 2012.

[32] P. J. Mullinger, “Burner design for coal fired cement kilns,” World Cement, pp. 348–353, Dec-
1984.

[33] D. Moore, “Grate Coolers,” 2017. [Online]. Available:

http://www.cementkilns.co.uk/cooler_grate.html. [Accessed: 01-Apr-2018].

[34] K. E. Peray and J. J. Waddell, The Rotary Cement Kiln. New York: Chemical Publishing Co., Inc.,
1972.

[35] R. H. Nobis, “Latest rotary kiln burner technology: Possibilities and experiences,” IEEE
Transactions on Industry Applications, vol. 27, no. 5, pp. 798–806, 1991.

[36] A. Wagner, “Aerodynamic features of a rotary kiln burner with multijet outflow of the primary
air,” Cement International, vol. 2, pp. 88–97, 2004.

[37] A. Schall, “Burner operation and flame characteristics of kiln firing systems,” in ECRA
Seminar, 2017.

[38] D. Moore, “Firing Systems,” 2014. [Online]. Available:

http://www.cementkilns.co.uk/ck_firing.html. [Accessed: 01-Jun-2017].

[39] C. Greco, G. Picciotti, R. M. Greco, and G. M. Ferreira, “Fuel Selection and Use,” in Innovations
in Portland Cement Manufacturing, 1st ed., J. I. Bhatty, F. M. Miller, S. H. Kosmatka, and R. P.
Bohan, Eds. Skokie, Illinois, USA: Portland Cement Association, 2011, pp. 239–308.

[40] T. M. Lowes and L. P. Evans, “The effect of burner design and operating parameters on flame
shape, heat transfer, NOx and SO3 cycles,” Zement-Kalk-Gips, no. 12, pp. 761–768, 1993.

[41] T. M. Lowes and L. P. Evans, “Optimisation of the design and operation of coal flames in
cement kilns,” Journal of the Institute of Energy, pp. 220–228, Dec-1989.

[42] M. H. Vaccaro, “Low NOx rotary kiln burner technology: design principles & case study,” in
2002 IEEE Cement Industry Technical Conference, 2002, pp. 265–270.

[43] P. B. Nielsen and O. L. Jepsen, “An overview of the formation of SOx and NOx in various
pyroprocessing systems,” IEEE Transactions on Industry Applications, vol. 27, no. 3, pp. 431–
439, 1991.

[44] S. Gajewski and V. Hoenig, “Influence of pulverized coal fineness on the formation of NOx in
rotary kilns in the cement industry,” ZKG International, no. 1, pp. 44–53, 1999.

[45] M. Vaccaro, “Burning alternative fuels in rotary cement kilns,” in 2006 IEEE Cement Industry
218 11 References

Technical Conference, 2006, pp. 127–136.

[46] B. Emberger and V. Hoenig, “Rotary kiln burner technology for alternative fuel co-firing,”
Cement International, vol. 9, no. 5, pp. 48–60, 2011.

[47] X. D’Hubert, “There’s no such thing as a ‘Perfect’ burner,” Global Cement Magazine, pp. 12–23,
Feb-2017.

[48] X. D’Hubert, “Latest Burner Profiles,” Global Cement Magazine, pp. 10–18, Mar-2017.

[49] O. A. Ahmed and K. R. Osman, “The Rise of Coal Mills in the Egyptian Cement Sector,” Global
Cement Magazine, pp. 61–63, Jul-2017.

[50] P. J. Mullinger and B. G. Jenkins, “Petroleum coke firing of rotary kilns,” World Cement, no. 2,
pp. 48–56, 1987.

[51] G. Roy, “Petcoke combustion characteristics,” World cement, vol. 33, no. 4, 2002.

[52] J. Chen and X. Lu, “Progress of petroleum coke combusting in circulating fluidized bed boilers-
A review and future perspectives,” Resources, Conservation and Recycling, vol. 49, no. 3, pp.
203–216, 2007.

[53] A. R. Nielsen, M. B. Larsen, P. Glarborg, and K. Dam-Johansen, “Sulfur Release from Cement
Raw Materials during Solid Fuel Combustion,” Energy & Fuels, vol. 25, no. 9, pp. 3917–3924,
2011.

[54] M. D. M. Cortada Mut, L. K. Nørskov, P. Glarborg, and K. Dam-Johansen, “SO2 release as a

consequence of alternative fuel combustion in cement rotary kiln inlets,” Energy and Fuels,
vol. 29, no. 4, pp. 2729–2737, 2015.

[55] ASTM International, “ASTM D396-16 Standard Specification for Fuel Oils,” West
Conshohocken, Pennsylvania, USA, 2016.

[56] A. A. Usón, A. M. López-Sabirón, G. Ferreira, and E. Llera Sastresa, “Uses of alternative fuels
and raw materials in the cement industry as sustainable waste management options,”
Renewable and Sustainable Energy Reviews, vol. 23, pp. 242–260, 2013.

[57] A. Murray and L. Price, “Use of Alternative Fuels in Cement Manufacture: Analysis of Fuel
Characteristics and Feasibility for Use in the Chinese Cement Sector,” Berkeley, California,
USA, 2008.

[58] J. Fellner and H. Rechberger, “Abundance of 14C in biomass fractions of wastes and solid
recovered fuels,” Waste Management, vol. 29, no. 5, pp. 1495–1503, 2009.

[59] G. Habert, C. Billard, P. Rossi, C. Chen, and N. Roussel, “Cement production technology
improvement compared to factor 4 objectives,” Cement and Concrete Research, vol. 40, no. 5,
pp. 820–826, 2010.

[60] K. H. Karstensen, “Formation, release and control of dioxins in cement kilns,” Chemosphere,
vol. 70, no. 4, pp. 543–560, 2008.
11 References 219

[61] Cembureau, “Environmental Benefits of Using Alternative Fuels in Cement Production,”

Brussels, Belgium, 1999.

[62] WBCSD, “Getting the Numbers Right Database,” 2015. [Online]. Available:
https://www.wbcsdcement.org/GNR-2013/index.html. [Accessed: 01-Jun-2016].

[63] WBCSD, “Global Cement Database on CO2 and Energy Information,” 2015. [Online]. Available:
http://www.wbcsdcement.org/index.php/key-issues/climate-protection/GNR-database.
[Accessed: 01-Jun-2016].

[64] European Parliament and Council, Directive 2008/98/EC on Waste and Repealing Certain
Directives. 2008.

[65] European Parliament and Council, Directive 2000/76/EC on the Incineration of Waste. 2000.

[66] World Business Council for Sustainable Development (WBCSD), “Getting the Numbers Right:
Cement Industry Energy and CO2 Performance,” Geneva, Switzerland, 2011.

[67] Verein Deutscher Zementwerke (VDZ), “Environmental Data of the German Cement Industry
2015,” Düsseldorf, 2016.

[68] G. Mauschitz, “Emissionen aus Anlagen der Österreichischen Zementindustrie 2015,” Vienna,
2016.

[69] A. Gendebien, A. Leavens, K. Blackmore, et al., “Refuse Derived Fuel, Current Practice and
Perspectives,” Swindon, UK, 2003.

[70] European Commission, M/325 Mandate to CEN on Solid Recovered Fuels (SRF) . Brussels,
Belgium, 2002.

[71] CEN (European Committee for Standardization), “CEN/TC 343 - Solid Recovered Fuels,”
Brussels, Belgium, 2012.

[72] L. F. Diaz, G. M. Savage, and L. L. Eggerth, “Production of Refuse-Derived Fuel (RDF),” in Solid
Waste Management, United Nations Environment Program (UNEP), 2005, pp. 295–302.

[73] C. A. Velis, P. J. Longhurst, G. H. Drew, R. Smith, and S. J. T. Pollard, “Production and quality
assurance of solid recovered fuels using Mechanical- Biological Treatment (MBT) of waste: A
comprehensive assessment,” Critical Reviews in Environmental Science and Technology, vol.
40, no. 12, pp. 979–1105, 2010.

[74] B. Krüger, A. Mrotzek, and S. Wirtz, “Separation of harmful impurities from refuse derived
fuels (RDF) by a fluidized bed,” Waste Management, vol. 34, no. 2, pp. 390–401, 2014.

[75] R. Sarc and K. E. Lorber, “Production, quality and quality assurance of Refuse Derived Fuels
(RDFs),” Waste Management, vol. 33, no. 9, pp. 1825–1834, 2013.

[76] V. S. Rotter, T. Kost, J. Winkler, and B. Bilitewski, “Material flow analysis of RDF-production
processes,” Waste Management, vol. 24, no. 10, pp. 1005–1021, 2004.

[77] P. Vainikka, D. Bankiewicz, A. Frantsi, et al., “High temperature corrosion of boiler waterwalls
220 11 References

induced by chlorides and bromides. Part 1: Occurrence of the corrosive ash forming elements
in a fluidised bed boiler co-firing solid recovered fuel,” Fuel, vol. 90, no. 5, pp. 2055–2063,
2011.

[78] FLSmidth A/S, “Feedex overhead reclaimer,” 2014. [Online]. Available:

http://www.flsmidth.com/en-
US/Industries/Categories/Operation/Alternative+fuels+solutions/Feedex+overhead+recl
aimer. [Accessed: 01-Jan-2017].

[79] F. C. Lockwood and J. J. Ou, “Review: burning refuse derived fuel in a rotary cement kiln,”
Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and Energy,
vol. 207, pp. 65–70, 1993.

[80] J. Haas, A. Agostini, C. Martens, and E. Carrea, “The combustion of pulverized coal and
alternative fuels in cement kilns. Results on the CEMFLAME-3 Experiments,” IJmuiden,
Netherlands, 1999.

[81] W. K. H. Ariyaratne, M. C. Melaaen, K. Eine, and L. a Tokheim, “Meat and Bone Meal as a
Renewable Energy Source in Cement Kilns : Investigation of Optimum Feeding Rate,” in
International Conference on Renewable Energies and Power Quality (ICREPQ’11), 2010.

[82] W. K. H. Ariyaratne, M. C. Melaaen, and L. A. Tokheim, “Optimum feeding rate of solid

hazardous waste in a cement kiln burner,” International Journal of Energy and Environment ,
vol. 4, no. 5, pp. 777–786, 2013.

[83] W. K. H. Ariyaratne, A. Malagalage, M. C. Melaaen, and L.-A. Tokheim, “CFD Modeling of Meat
and Bone Meal Combustion in a Rotary Cement Kiln,” International Journal of Modeling and
Optimization, vol. 4, no. 4, pp. 263–272, 2014.

[84] W. K. H. Ariyaratne, E. V. P. J. Manjula, M. C. Melaaen, and L.-A. Tokheim, “Mathematical Model

for Alternative Fuel Combustion in a Rotary Cement Kiln Burner,” International Journal of
Modeling and Optimization, vol. 4, no. 1, pp. 56–61, 2014.

[85] C. Di Blasi, “Combustion and gasification rates of lignocellulosic chars,” Progress in Energy
and Combustion Science, vol. 35, no. 2, pp. 121–140, 2009.

[86] M. Momeni, C. Yin, S. K. Kær, T. B. Hansen, P. A. Jensen, and P. Glarborg, “Experimental Study
on Effects of Particle Shape and Operating Conditions on Combustion Characteristics of Single
Biomass Particles,” Energy & Fuels, vol. 27, no. 1, pp. 507–514, 2013.

[87] P. E. Mason, L. I. Darvell, J. M. Jones, M. Pourkashanian, and A. Williams, “Single particle flame-
combustion studies on solid biomass fuels,” Fuel, vol. 151, pp. 21–30, 2015.

[88] R. Klaska, S. Baetzner, H. Moller, M. Paul, and T. Roppelt, “Effects of secondary fuels on clinker
mineralogy,” Cement International, vol. 1, no. 4, pp. 88–98, 2003.

[89] S. Uda, E. Asakura, and M. Nagashima, “Influence of SO3 on the Phase Relationship in the
System CaO-SiO2-Al2O3-Fe2O3,” Journal of the American Ceramic Society, vol. 81, no. 3, pp.
11 References 221

725–729, 2005.

[90] H. E. Belkin, S. J. Tewalt, J. C. Hower, et al., “Petrography and geochemistry of Oligocene

bituminous coal from the Jiu Valley, Petrosani basin (southern Carpathian Mountains),
Romania,” International Journal of Coal Geology, vol. 82, no. 1–2, pp. 68–80, 2010.

[91] O. G. Duliu, O. A. Culicov, I. Rãdulescu, C. Cristea, and T. Vasiu, “Major, trace, and natural
radioactive elements in bituminous coal from Australia, Romania, Russia, South Africa and
Ukraine: A comparative study,” Journal of Radioanalytical and Nuclear Chemistry , vol. 264,
no. 3, pp. 525–534, 2005.

[92] J. M. Commandré and S. Salvador, “Lack of correlation between the properties of a petroleum
coke and its behaviour during combustion,” Fuel Processing Technology, vol. 86, no. 7, pp.
795–808, 2005.

[93] M. Bläsing, K. Nazeri, and M. Müller, “Fate of sulfur, chlorine, alkali metal, and vanadium
species during high-temperature gasification of Canadian tar sand products,” Energy and
Fuels, vol. 28, no. 10, pp. 6345–6350, 2014.

[94] D. Moore, “Oil Data,” 2011. [Online]. Available: http://www.cementkilns.co.uk/data_oil.html.

[Accessed: 01-Jan-2017].

[95] J. M. Heikkinen, J. C. Hordijk, W. De Jong, and H. Spliethoff, “Thermogravimetry as a tool to

classify waste components to be used for energy generation,” Journal of Analytical and
Applied Pyrolysis, vol. 71, no. 2, pp. 883–900, 2004.

[96] A. Trubetskaya, P. A. Jensen, A. D. Jensen, A. D. Garcia Llamas, K. Umeki, and P. Glarborg, “Effect
of fast pyrolysis conditions on biomass solid residues at high temperatures,” Fuel Processing
Technology, vol. 143, pp. 118–129, 2016.

[97] J. M. Lee, D. W. Kim, J. S. Kim, J. G. Na, and S. H. Lee, “Co-combustion of refuse derived fuel with
Korean anthracite in a commercial circulating fluidized bed boiler,” Energy, vol. 35, no. 7, pp.
2814–2818, 2010.

[98] U. Kääntee, R. Zevenhoven, R. Backman, and M. Hupa, “Cement manufacturing using

alternative fuels and the advantages of process modelling,” Fuel Processing Technology, vol.
85, no. 4, pp. 293–301, 2004.

[99] J. Maier, A. Gerhardt, and G. Dunnu, “Experiences on Co-firing Solid Recovered Fuels in the
Coal Power Sector,” in Solid Biofuels for Energy, 1st ed., P. Grammelis, Ed. London: Springer,
2011, pp. 75–94.

[100] R. Sarc, K. Lorber, R. Pomberger, M. Rogetzer, and E. Sipple, “Design, quality, and quality
assurance of solid recovered fuels for the substitution of fossil feedstock in the cement
industry,” Waste management & research, vol. 32, no. 7, pp. 565–585, 2014.

[101] FLSmidth A/S and FLSmidth, “HOTDISC combustion device,” Copenhagen, Denmark, 2011.

[102] Thyssenkrupp AG, “Prepol-SC,” Beckum, Germany.

222 11 References

[103] L. F. De Diego, F. Garcia-Labiano, A. Abad, P. Gaya, and J. Adanez, “Coupled drying and
devolatilisation of non-spherical wet pine wood particles in fluidised beds,” vol. 65, pp. 173–
184, 2002.

[104] E. N. Tackie, A. P. Watkinson, and J. K. Brimacombe, “Experimental Study of the Elutriation of

Particles from Rotary Kilns,” The Canadian Journal of Chemical Engineering, vol. 67, pp. 806–
817, 1989.

[105] M. Fardadi, “Modeling Dust Formation in Lime Kilns,” PhD Thesis, University of Toronto,
2010.

[106] G. R. Long, “Microstructure and Chemistry of Unhydrated Cements,” Philosphical

Transactions of the Royal Society A, vol. 310, pp. 43–51, 1983.

[107] F. W. Locher, “Influence of burning conditions on clinker characteristics,” World Cement

Technology, vol. 11, no. 2, pp. 67–73, 1980.

[108] M. Ichikawa and Y. Komukai, “Effect of burning conditions and minor components on the color
of portland cement clinker,” Cement and Concrete Research, vol. 23, no. 4, pp. 933–938, 1993.

[109] M. Clark, “Brown cores,” International Cement Review, pp. 46–47, Dec-2006.

[110] Holderbank Cement, “Holderbank Cement Seminar - Circulation Phenomena in the

Clinkerization Process,” Holderbank Cement, Jona, Switzerland, 2000.

[111] G.-S. Choi and F. Glasser, “The sulphur cycle in cement kilns: Vapour pressures and solid-
phase stability of the sulphate phases,” Cement and Concrete Research, vol. 18, no. 3, pp. 367–
374, 1988.

[112] M. D. M. Cortada Mut, “Interactions between solid fuels and raw materials in cement rotary
kilns,” PhD Thesis, Technical University of Denmark, 2013.

[113] A. H. Mortensen, A. Hintsteiner, and P. Rosholm, “Converting two kiln lines to 100% high
sulphur petroleum coke firing,” ZKG International, vol. 51, no. 4, pp. 184–196, 1998.

[114] I. R. Dominguez, J. Gómez-Millán, M. Alvarez, S. De Aza, L. Contreras, and A. H. De Aza, “Build-

up formation and corrosion of monolithic refractories in cement kiln preheaters,” Journal of
the European Ceramic Society, vol. 30, no. 9, pp. 1879–1885, 2010.

[115] W. Kurdowski and M. Soboń, “Mineral composition of build-up in cement kiln preheater,”
Journal of Thermal Analysis and Calorimetry, vol. 55. pp. 1021–1029, 1999.

[116] H. Bolio-Arceo and F. P. Glasser, “Formation of spurrite, Ca5(SiO4)2CO3,” Cement and

Concrete Research, vol. 20, no. 2, pp. 301–307, 1990.

[117] E. Jøns and M. J. L. Østergard, “Kiln Shell Corrosion,” in IEEE-IAS/PCA Cement Industry
Technical Conference, 2001, pp. 343–359.

[118] A. R. Nielsen, M. B. Larsen, P. Glarborg, and K. Dam-Johansen, “High-temperature release of

SO2 from calcined cement raw materials,” Energy and Fuels, vol. 25, no. 7, pp. 2917–2926,
11 References 223

2011.

[119] M. D. M. Cortada Mut, L. K. Nørskov, F. J. Frandsen, P. Glarborg, and K. Dam-Johansen, “Review:

Circulation of Inorganic Elements in Combustion of Alternative Fuels in Cement Plants,”
Energy and Fuels, vol. 29, no. 7, pp. 4076–4099, 2015.

[120] K. Sutoh, M. Murata, and N. Ueno, “Stable kiln operation achived by chlorine bypass system,”
World cement, vol. 29, no. 2, pp. 47–51, 1998.

[121] E. M. Sipple, “The ReduDust process – production of salt in the cement plant,” ZKG
International, 2015. [Online]. Available:
http://www.zkg.de/en/artikel/zkg_The_ReduDust_process_production_of_salt_in_the_ceme
nt_plant_2452179.html. [Accessed: 01-Jan-2017].

[122] E. M. Sipple and J. Mülnner, “A green building in a grey cement plant – Transforming bypass
dust into industrial salt with the ReduDust Process,” ZKG International, no. 11, 2015.

[123] M. P. Javellana and I. Jawed, “Extraction of free lime in portland cement and clinker by
ethylene glycol,” Cement and Concrete Research, vol. 12, no. 3, pp. 399–403, 1982.

[124] D. Broton and J. I. Bhatty, “Analytical Techniques in Cement Materials Characterization,” in

Innovations in Portland Cement Manufacturing, 1st ed., J. I. Bhatty, F. M. Miller, S. H. Kosmatka,
and R. P. Bohan, Eds. Skokie, Illinois, USA: Portland Cement Association, 2011, pp. 1373–1424.

[125] L. J. Struble, L. A. Graf, and J. I. Bhatty, “X-Ray Diffraction Analysis,” in Innovations in Portland
Cement Manufacturing, 1st ed., J. I. Bhatty, F. M. Miller, S. H. Kosmatka, and R. P. Bohan, Eds.
Skokie, Illinois, USA: Portland Cement Association, 2011, pp. 1425–1482.

[126] G. Kychakoff and A. F. Hollingshead, “Acoustic Pyrometry in Cement Manufacturing,” in

Innovations in Portland Cement Manufacturing, 1st ed., J. I. Bhatty, F. M. Miller, S. H. Kosmatka,
and R. P. Bohan, Eds. Skokie, Illinois, USA: Portland Cement Association, 2011, pp. 557–570.

[127] M. C. Mound and C. Colbert, “Automation and Optimization in the Cement Plant,” in
Innovations in Portland Cement Manufacturing, 1st ed., J. I. Bhatty, F. M. Miller, S. H. Kosmatka,
and R. P. Bohan, Eds. Skokie, Illinois, USA: Portland Cement Association, 2011, pp. 823–868.

[128] K. B. Floor and F. M. Miller, “Inlet Gas Analysis in Cement Operations,” in Innovations in
Portland Cement Manufacturing, 1st ed., J. I. Bhatty, F. M. Miller, S. H. Kosmatka, and R. P.
Bohan, Eds. Skokie, Illinois, USA: Portland Cement Association, 2011, pp. 571–588.

[129] R. P. Brower, J. P. S. Seinfield, and S. Akhtar, “NOx control and SNCR,” in Innovations in
Portland Cement Manufacturing, 1st ed., J. I. Bhatty, F. M. Miller, S. H. Kosmatka, and R. P.
Bohan, Eds. Skokie, Illinois, USA: Portland Cement Association, 2011, pp. 1017–1038.

[130] B. Lin and S. B. Jørgensen, “Soft sensor design by multivariate fusion of image features and
process measurements,” Journal of Process Control, vol. 21, no. 4, pp. 547–553, 2011.

[131] Verein Deutscher Zementwerke (VDZ), “VDZ Activity Report 2005-2007,” Düsseldorf,
Germany, 2008.
224 11 References

[132] D. Schmidt, “Highly efficient burning of clinker using flame analysis and NMPC,” in 2007 IEEE
Cement Industry Technical Conference, 2007, pp. 140–146.

[133] J. P. Kemmerer, “Refractory management with a high speed kiln shell scanner,” in IEEE
Technical Conference on cement Industry, 1990, pp. 237–251.

[134] M. N. Pedersen, M. Nielsen, S. Clausen, P. A. Jensen, L. S. Jensen, and K. Dam-Johansen, “Imaging

of Flames in Cement Kilns To Study the Influence of Different Fuel Types,” Energy & Fuels,
vol. 31, no. 10, pp. 11424–11438, 2017.

[135] N. A. Chigier, “Gasdynamics of swirling flow in combustion systems,” Astronautica Acta, vol.
17, pp. 387–395, 1972.

[136] F. D. Moles, D. Watson, and P. B. Lain, “The aerodynamics of the rotary cement kiln,” Journal
of the Institute of Fuel, pp. 353–362, 1973.

[137] H. E. Borgholm, D. Herfort, and S. Rasmussen, “A New Blended Cement Based on Mineralised
Clinker,” World Cement, no. 8, pp. 27–33, 1995.

[138] G. K. Moir, “Improvements in the Early Strength Properties of Portland Cement,”

Philosophical Transactions of the Royal Society A: Mathematical and Physical, vol. 310, no.
1511, pp. 127–138, 1983.

[139] A. Emanuelson, S. Hansen, and E. Viggh, “A comparative study of ordinary and mineralised
Portland cement clinker from two different production units: Part I: Composition and
hydration of the clinkers,” Cement and Concrete Research, vol. 33, no. 10, pp. 1613–1621,
2003.

[140] J. I. Bhatty, “Use of Fluxes and Mineralizers in the Cement Industry: A Survey,” Portland
Cement Association, R&D Serial 2045; Skokie, Illinois, USA, 1996.

[141] J. Benesty, J. Chen, Y. Huang, and I. Cohen, “Pearson Correlation Coefficient,” in Noise
Reduction in Speech Processing, vol. 2, Berlin: Springer, 2009, pp. 37–40.

[142] W. M. Haynes, Ed., CRC Handbook of Chemistry and Physics, 97th ed. Boca Raton, FL, USA:
CRC Press/Taylor & Francis, 2016.

[143] B. Liedmann, S. Wirtz, V. Scherer, and B. Krüger, “Numerical Study on the Influence of
Operational Settings on Refuse Derived Fuel Co-firing in Cement Rotary Kilns,” Energy
Procedia, vol. 120, pp. 254–261, 2017.

[144] N. Syred and J. M. Beér, “Combustion in Swirling Flows: a Review,” Combustion and Flame,
vol. 23, no. 2, 1974.

[145] FLSmidth A/S, “JETFLEX Burner,” Copenhagen, Denmark, 2017.

[146] A. J. Chapman and W. F. Walker, Introductory Gas Dynamics, 1st ed. Holt, Rinehart and
Winston, Inc, 1971.

[147] J. M. Beér and N. A. Chigier, “Stability and Combustion Intensity of Pulverized-Coal Flames-
11 References 225

Effect of Swirl and Impingement,” Journal of the Institute of Fuel, vol. 42, no. 347, pp. 443–
450, 1969.

[148] A. K. Gupta, D. G. Lilley, and N. Syred, Swirl Flows. Tunbridge Wells, UK: Abacus Press, 1984.

[149] P. Geladi and B. R. Kowalski, “Partial least-squares regression: a tutorial,” Analytica Chimica
Acta, vol. 185, no. C, pp. 1–17, 1986.

[150] H. Abdi, “Partial least squares regression and projection on latent structure regression,” Wiley
Interdisciplinary Reviews: Computational Statistics, vol. 2, pp. 97–106, 2010.

[151] S. Wold, M. Sjöström, and L. Eriksson, “PLS-regression: A basic tool of chemometrics,”

Chemometrics and Intelligent Laboratory Systems, vol. 58, no. 2, pp. 109–130, 2001.

[152] T. Kourti and J. F. MacGregor, “Process analysis, monitoring and diagnosis, using multivariate
projection methods,” Chemometrics and intelligent laboratory systems , vol. 28, pp. 3–21,
1995.

[153] L. Eriksson, P. Hagberg, E. Johansson, et al., “Multivariate process monitoring of a newsprint

mill. Application to modelling and predicting COD load resulting from de-inking of recycled
paper,” Journal of Chemometrics, vol. 15, no. 4, pp. 337–352, 2001.

[154] B. Lin, B. Recke, J. K. H. Knudsen, and S. B. Jørgensen, “A systematic approach for soft sensor
development,” Computers & Chemical Engineering, vol. 31, no. 5–6, pp. 419–425, 2007.

[155] B. Lin, B. Recke, T. M. Schmidt, J. K. H. Knudsen, and S. B. Jørgensen, “Data-Driven Soft Sensor
Design with Multiple-Rate Sampled Data: A Comparative Study,” Industrial & Engineering
Chemistry Research, vol. 48, no. 11, pp. 5379–5387, 2009.

[156] E. Marengo, M. Bobba, E. Robotti, and M. C. Liparota, “Modeling of the polluting emissions
from a cement production plant by partial least-squares, principal component regression, and
artificial neural networks,” Environmental Science and Technology, vol. 40, no. 1, pp. 272–
280, 2006.

[157] T. Mehmood, K. H. Liland, L. Snipen, and S. Sæbø, “A review of variable selection methods in
Partial Least Squares Regression,” Chemometrics and Intelligent Laboratory Systems , vol.
118, pp. 62–69, 2012.

[158] J. Josse and F. Husson, “Selecting the number of components in principal component analysis
using cross-validation approximations,” Computational Statistics and Data Analysis, vol. 56,
no. 6, pp. 1869–1879, 2012.

[159] T. Hastie, R. Tibshirani, and J. Friedman, The Elements of Statistical Learning, 2nd ed. New
York, USA: Springer, 2009.

[160] European Comittee for Standardization (CEN), “EN 15359 - Solid recovered fuels.
Specifications and classes,” Brussels, Belgium, 2011.

[161] M. Nakhaei, M. N. Pedersen, H. Wu, et al., “Aerodynamic and Physical Characterization of

Refuse Derived Fuel,” Energy and Fuels (Manuscript).
226 11 References

[162] B. Liedmann, W. Arnold, B. Krüger, et al., “An approach to model the thermal conversion and
flight behaviour of Refuse Derived Fuel,” Fuel, vol. 200, pp. 252–271, 2017.

[163] B. Liedmann, W. Arnold, S. Wirtz, et al., “Refuse Derived Fuel Co-Firing in Cement Rotary Kilns
– A methodology to specify a customized fuel by numerical simulation and fuel
characterisation,” in 10th European Conference on Industrial Furnaces and Boilers, 2015, pp.
1–11.

[164] P. Grammelis, P. Basinas, A. Malliopoulou, and G. Sakellaropoulos, “Pyrolysis kinetics and

combustion characteristics of waste recovered fuels,” Fuel, vol. 88, no. 1, pp. 195–205, 2009.

[165] D. Montané, S. Abelló, X. Farriol, and C. Berrueco, “Volatilization characteristics of solid

recovered fuels (SRFs),” Fuel Processing Technology, vol. 113, pp. 90–96, 2013.

[166] L. Sørum, M. G. Gronli, and J. E. Hustad, “Pyrolysis characteristics and kinetics of municipal
solid wastes,” Fuel, vol. 80, no. 9, pp. 1217–1227, 2001.

[167] J. Brandsch and O. G. Piringer, “Characteristics of Plastic Materials,” in Plastic Packaging:

Interactions with Food and Pharmaceuticals, 2nd ed., O. G. Piringer and A. L. Baner, Eds. Wiley,
2008, pp. 15–61.

[168] S. M. Al-Salem, P. Lettieri, and J. Baeyens, “The valorization of plastic solid waste (PSW) by
primary to quaternary routes: From re-use to energy and chemicals,” Progress in Energy and
Combustion Science, vol. 36, no. 1, pp. 103–129, 2010.

[169] Q. Chen and T. Zhao, “The thermal decomposition and heat release properties of the
nylon/cotton, polyester/cotton and Nomex/cotton blend fabrics,” Textile Research Journal,
vol. 86, no. 17, pp. 1859–1868, 2016.

[170] G. Dunnu, T. Hilber, and U. Schnell, “Advanced Size Measurements and Aerodynamic
Classification of Solid Recovered Fuel Particles,” Energy & Fuels, vol. 20, no. 4, pp. 1685–1690,
2006.

[171] E. Alakangas, “Quality Guidelines for Wood Fuels in Finland VTT-M-04712-15,” Jyvaskyla,
Finland, 2015.

[172] P. Rosin and E. Rammler, “Laws governing the fineness of powdered coal,” Journal of the
Institute of Fuel, vol. 7, pp. 29–36, 1933.

[173] P. A. Vesilind, “The Rosin-Rammler particle size distribution,” Resource Recovery and
Conservation, vol. 5, no. 3, pp. 275–277, 1980.

[174] G. M. Savage and G. J. Trezek, “Significance of Size Reduction in Solid Waste Management,”
Washington, DC, 1980.

[175] H. Lu, E. Ip, J. Scott, P. Foster, M. Vickers, and L. L. Baxter, “Effects of particle shape and size on
devolatilization of biomass particle,” Fuel, vol. 89, no. 5, pp. 1156–1168, 2010.

[176] Z. Lu, J. Jian, P. A. Jensen, H. Wu, and P. Glarborg, “Influence of Torrefaction on Single Particle
Combustion of Wood,” Energy and Fuels, vol. 30, no. 7, pp. 5772–5778, 2016.
11 References 227

[177] UL Prospector, “Polyethylene,” 2018. [Online]. Available:

https://plastics.ulprospector.com/generics/27/polyethylene-pe. [Accessed: 11-Mar-2018].

[178] M. Imber and V. Paschkis, “A new theory for a rotary-kiln heat exchanger,” International
Journal of Heat and Mass Transfer, vol. 5, no. 7, pp. 623–638, 1962.

[179] J. K. Brimacombe and A. P. Watkinson, “Heat transfer in a direct-fired rotary kiln: I. Pilot plant
and experimentation,” Metallurgical Transactions B, vol. 9, no. 3, pp. 201–208, 1978.

[180] A. P. Watkinson and J. K. Brimacombe, “Heat transfer in a direct-fired rotary kiln: II. Heat flow
results and their interpretation,” Metallurgical Transactions B, vol. 9, no. 3, pp. 209–219,
1978.

[181] J. P. Gorog, J. K. Brimacombe, and T. N. Adams, “Radiative heat transfer in rotary kilns,”
Metallurgical Transactions B, vol. 12, no. 1, pp. 55–70, 1981.

[182] J. P. Gorog, T. N. Adams, and J. K. Brimacombe, “Regenerative Heat Transfer in Rotary Kilns,”
Metallurgical Transactions B, vol. 13, no. 2, pp. 153–163, 1982.

[183] J. P. Gorog, T. N. Adams, and J. K. Brimacombe, “Heat transfer from flames in a rotary kiln,”
Metallurgical Transactions B, vol. 14, no. 3, pp. 411–424, 1983.

[184] P. V. Barr, J. K. Brimacombe, and A. P. Watkinson, “A Heat-Transfer Model for the Rotary Kiln:
Part I. Pilot Kiln Trials,” Metallurgical Transactions B, vol. 20B, no. 3, pp. 391–402, 1989.

[185] P. V. Barr, J. K. Brimacombe, and A. P. Watkinson, “A Heat-Transfer Model for the Rotary Kiln:
Part II. Development of Cross Sectional Model,” Metallurgical Transactions B, vol. 20B, no. 3,
pp. 403–419, 1989.

[186] J. M. Beér and N. A. Chigier, Combustion Aerodynamics. London, UK: Applied Science
Publishers, 1972.

[187] F. P. Ricou and D. B. Spalding, “Measurements of entrainment by axisymmetrical turbulent

jets,” Journal of Fluid Mechanics, vol. 11, no. 1, pp. 21–32, 1961.

[188] J. P. Gorog and T. N. Adams, “Design and Performance of Rotary Lime Kilns in the Pulp and
Paper Industry: Part 1 - A Predictive Model for a Rotary Lime Reburning Kiln,” in Kraft
Recovery Operations Seminar, 1987, pp. 41–47.

[189] J. P. Gorog and T. N. Adams, “Design and Performance of Rotary Lime Kilns in the Pulp and
Paper Industry: Part 2 - The Effect of Chain System Design and the Refractory Lining on Lime
Kiln Performance,” in Kraft Recovery Operations Seminar, 1986, pp. 45–51.

[190] J. P. Gorog and T. N. Adams, “Design and Performance of Rotary Lime Kilns in the Pulp and
Paper Industry: Part 3 - How Flame Characteristics and Product Coolers Affect Lime Kiln
Performance,” in Kraft Recovery Operations Seminar, 1986, pp. 53–60.

[191] J. P. Gorog and T. N. Adams, “Design and Performance of Rotary Lime Kilns in the Pulp and
Paper Industry: Part 4 - The Effect of Operating Variables and O2 Enrichment on Lime Kiln
Performance,” in Kraft Recovery Operations Seminar, 1987, pp. 63–68.
228 11 References

[192] J. P. Gorog and T. N. Adams, “Design and Performance of Rotary Lime Kilns in the Pulp and
Paper Industry: Part 5 - Heat Transfer, Kiln Geometry, and Lime Kiln Performance Rating,” in
Kraft Recovery Operations Seminar, 1987, pp. 69–73.

[193] H. A. Spang, “A Dynamic Model of a Cement Kiln,” Automatica, vol. 8, no. 3, pp. 309–323, 1972.

[194] G. Locher, “Mathematical models for the cement clinker burning process. Part 1: Reactions
and unit operations,” ZKG International, vol. 55, no. 1, pp. 29–38, 2002.

[195] G. Locher, “Mathematical models for the cement clinker burning process. Part 2: Preheater,
calciner and bypass,” ZKG International, vol. 55, no. 1, pp. 39–50, 2002.

[196] G. Locher, “Mathematical models for the cement clinker burning process. Part 3: Rotary Kiln,”
ZKG International, vol. 55, no. 6, pp. 25–37, 2002.

[197] K. S. Mujumdar and V. V. Ranade, “Simulation of Rotary Cement Kilns Using a One-
Dimensional Model,” Chemical Engineering Research and Design, vol. 84, no. 3, pp. 165–177,
2006.

[198] K. S. Mujumdar, A. Arora, and V. V. Ranade, “Modeling of rotary cement kilns: Applications to
reduction in energy consumption,” Industrial and Engineering Chemistry Research , vol. 45,
no. 7, pp. 2315–2330, 2006.

[199] J. Haas and R. Weber, “Co-firing of refuse derived fuels with coals in cement kilns: Combustion
conditions for stable sintering,” Journal of the Energy Institute, vol. 83, no. 4, pp. 225–234,
2010.

[200] F. C. Lockwood and B. Shen, “Performance predictions of pulverised-coal flames of power

station furnace and cement kiln types,” in Twenty-Fifth Symposium (International) on
Combustion, 1994, pp. 503–509.

[201] E. Mastorakos, A. Massias, C. D. Tsakiroglou, D. A. Goussis, V. N. Burganos, and A. C. Payatakes,

“CFD predictions for cement kilns including flame modelling, heat transfer and clinker
chemistry,” Applied Mathematical Modelling, vol. 23, no. 1, pp. 55–76, 1999.

[202] S. Wang, J. Lu, W. Li, J. Li, and Z. Hu, “Modeling of pulverized coal combustion in cement rotary
kiln,” Energy and Fuels, vol. 20, no. 10, pp. 2350–2356, 2006.

[203] K. S. Mujumdar and V. V. Ranade, “CFD modeling of rotary cement kilns,” Asia-Pacific Journal
of Chemical Engineering, vol. 3, no. 2, pp. 106–118, 2008.

[204] D. J. Isaac, M. N. Pedersen, D. Grévain, L. S. Jensen, and M. Nielsen, “Numerical Evaluation of

Co-Firing Solid Recovered Fuel With Petroleum Coke in a Cement Rotary Kiln - A Case
Example,” in Proceedings of The 12th International Conference on Computational Fluid
Dynamics in the Oil & Gas, Metallurgical and Process Industries , 2017, pp. 613–620.

[205] H. Mikulčić, E. von Berg, M. Vujanović, and N. Duić, “Numerical study of co-firing pulverized
coal and biomass inside a cement calciner.,” Waste Management & Research, vol. 32, no. 7, pp.
661–669, 2014.
11 References 229

[206] M. Nakhaei, H. Wu, P. Glarborg, K. Dam-johansen, D. Grévain, and L. S. Jensen, “CFD simulation
of a full-scale RDF-fired calciner,” in Proceedings of the 2nd International Workshop on CFD
and Biomass Thermochemical Conversion, 2016.

[207] S. Telschow, “Clinker Burning Kinetics and Mechanism,” PhD Thesis, Technical University of
Denmark, 2012.

[208] P. V. Barr, “Heat Transfer Processes in Rotary Kilns,” PhD Thesis, University of British
Columbia, 1986.

[209] H. Henein, J. K. Brimacombe, and A. P. Watkinson, “Experimental study of transverse bed

motion in rotary kilns,” Metallurgical Transactions B, vol. 14, no. 2, pp. 191–205, 1983.

[210] W. C. Saeman, “Passage of Solids Through Rotary Kilns,” Chemical Engineering Progress, vol.
47, no. 10, pp. 508–514, 1951.

[211] J. A. Imlach, “Non-isothermal Investigation of the Clinker Kinetics of Reactions Occuring

During Clinker Formation,” vol. 6, pp. 797–802, 1976.

[212] A. Khawam and D. R. Flanagan, “Solid-State Kinetic Models: Basics and Mathematical
Fundamentals,” The Journal of Physical Chemistry B, vol. 110, no. 35, pp. 17315–17328, 2006.

[213] W. Jander, “Reaktionen im festen Zustande bei höheren Temperaturen.

Reaktionsgeschwindigkeiten endotherm verlaufender Umsetzungen,” Zeitschrift für
anorganische und allgemeine Chemie, vol. 163, no. 1, pp. 1–30, 1927.

[214] B. I. Ginstling, A. M.; Brounshtein, “Concerning the Diffusion Kinetics of Reactions in Spherical
Particles,” Journal of Applied Chemistry of the USSR, vol. 23, pp. 1327–1338, 1950.

[215] O. Levenspiel, The Chemical Reactor Omnibook. Corvallis, Oregon, USA: Oregon State
University, 2002.

[216] S. Chromy and Z. Hrabe, “Process of Portland clinker formation, reactivity and burnability of
cement raw materials - Part 2 Kinetics of clinker formation,” Zement-Kalk-Gips, no. 7, pp.
368–373, 1982.

[217] G. Locher and M. Schneider, “Modeling in Cement Kiln Operations,” in Innovations in Portland
Cement Manufacturing, 1st ed., J. I. Bhatty, F. M. Miller, S. H. Kosmatka, and R. P. Bohan, Eds.
Skokie, Illinois, USA: Portland Cement Association, 2011, pp. 935–972.

[218] N. Rajaratnam, Turbulent Jets. Amsterdam: Elsevier, 1976.

[219] A. G. Guruz, H. K. Guruz, S. Osuwan, and F. R. Steward, “Aerodynamics of a Confined Burning

Jet,” Combustion Science and Technology, vol. 9, no. 3–4, pp. 103–110, 1974.

[220] S. H. Park and H. D. Shin, “Measurements of entrainment characteristics of swirling jets,”

International Journal of Heat and Mass Transfer, vol. 36, no. 16, pp. 4009–4018, 1993.

[221] P. J. Mullinger, “Flame control in rotary lime kilns,” World Cement, vol. 24, no. 6, pp. 20–23,
1993.
230 11 References

[222] I. Akman, U. Durgut, C. G. Manias, and S. T. Hill, “Kiln Burner Design,” World Cement, no. 12,
pp. 53–59, Dec. 2004.

[223] W. Ruhland, “Investigation of flames in cement rotary kilns,” Journal of the Institute of Fuel,
vol. 40, pp. 69–75, 1967.

[224] T. P. Bhad, S. Sarkar, A. Kaushik, and S. V. Herwadkar, “CFD Modeling of a Cement Kiln with
Multi Channel Burner for Optimization of Flame Profile,” in Seventh International Conference
on CFD in the Minerals and Process Industries, 2009, pp. 1–7.

[225] N. Orfanoudakis, A. Hatziapostolou, K. Krallis, et al., “Experimental and numerical

investigations in the near-burner region of a versatile multi-fuel burner,” Journal of
Computational Methods in Sciences and Engineering, vol. 7, pp. 3–19, 2007.

[226] J. Lehmberg, M. Hehl, and K. Schügerl, “Transverse Mixing and Heat Transfer in Horizontal
Rotary Drum Reactors,” Powder Technology, vol. 18, no. 2, pp. 149–163, 1977.

[227] A. A. Boateng and P. V. Barr, “A thermal model for the rotary kiln including heat transfer
within the bed,” International Journal of Heat and Mass Transfer , vol. 39, no. 10, pp. 2131–
2147, 1996.

[228] S. H. Tscheng and A. P. Watkinson, “Convective Heat Transfer in a Rotary Kiln,” The Canadian
Journal of Chemical Engineering, vol. 57, pp. 433–443, 1979.

[229] R. Siegel and J. R. Howell, Thermal Radiation Heat Transfer, 3rd ed. Washington, DC, USA:
Taylor & Francis, 1992.

[230] K. M. Becker, “Measurement of Convective Heat Transfer from a Horizontal Cylinder Rotating
in a Tank of Water,” Stockholm, Sweden, 1963.

[231] K. M. Bryden, K. W. Ragland, and C. J. Rutland, “Modeling thermally thick pyrolysis of wood,”
Biomass and Bioenergy, vol. 22, pp. 41–53, 2002.

[232] M. Nakhaei, H. Wu, D. Grevain, et al., “Experiments and Modelling of Single Plastic Particle
Conversion in Suspension,” Fuel Processing Technology (Manuscript).

[233] C. A. Koufopanos, N. Papayannakos, G. Maschio, and A. Luchessi, “Modelling of the Pyrolysis

of Biomass Particles - Studies on Kinetics , Thermal and Heat Transfer Effects,” The Canadian
Journal of Chemical Engineering, vol. 69, no. 4, pp. 907–915, 1991.

[234] S. Hamdi, W. E. Schiesser, and G. W. Griffiths, “Method of lines,” 2007. [Online]. Available:
http://www.scholarpedia.org/article/Method_of_lines. [Accessed: 01-Oct-2017].

[235] W. E. Ranz and W. R. Marshall, “Evaporation from drops Part 1,” Chemical Engineering
Progress, vol. 48, no. 3, pp. 141–146, 1952.

[236] K. M. Bryden and M. J. Hagge, “Modeling the combined impact of moisture and char shrinkage
on the pyrolysis of a biomass particle,” Fuel, vol. 82, no. 13, pp. 1633–1644, 2003.

[237] J. M. Johansen, “Power Plant Burners for Bio-Dust Combustion Thesis,” PhD Thesis, Technical
11 References 231

University of Denmark, 2015.

[238] A. N. Matzakos, “Fundamental Mechanisms of Coal Pyrolysis and Char Combustion,” PhD
Thesis, Rice University, 1992.

[239] C. Di Blasi, “Modeling chemical and physical processes of wood and biomass pyrolysis,”
Progress in Energy and Combustion Science, vol. 34, no. 1, pp. 47–90, 2008.

[240] G. Borghi, A. F. Sarofim, and J. M. Beér, “A model of coal devolatilization and combustion in
fluidized beds,” Combustion and Flame, vol. 61, no. 1, pp. 1–16, 1985.

[241] D. R. Burgess, “Thermochemical Data,” in NIST Chemistry WebBook, P. J. Linstrom and G. W.

Mallard, Eds. Gaithersburg, Maryland, USA: National Institute of Standards and Technology,
2018.

[242] N. M. Laurendeau, “Heterogeneous kinetics of coal char gasification and combustion,”

Progress in Energy and Combustion Science, vol. 4, no. 4, pp. 221–270, 1978.

[243] I. W. Smith, “The combustion rates of coal chars: A review,” Symposium (International) on
Combustion, vol. 19, no. 1, pp. 1045–1065, 1982.

[244] B. F. Magnussen and B. H. Hjertager, “On mathematical modeling of turbulent combustion

with special emphasis on soot formation and combustion,” Symposium (International) on
Combustion, vol. 16, no. 1, pp. 719–729, 1977.

[245] Y. C. Guo, C. K. Chan, and K. S. Lau, “Numerical studies of pulverized coal combustion in a
tubular coal combustor with slanted oxygen jet,” Fuel, vol. 82, no. 8, pp. 893–907, 2003.

[246] H. Thunman and B. Leckner, “Thermal conductivity of wood - Models for different stages of
combustion,” Biomass and Bioenergy, vol. 23, no. 1, pp. 47–54, 2002.

[247] E. N. Fuller, P. D. Schettler, and J. C. Giddings, “A New Method for Prediction of Binary Gas-
Phase Diffusion Coefficients,” Industrial and Engineering Chemistry, vol. 58, no. 5, pp. 19–27,
1966.

[248] R. Johansson, B. Leckner, K. Andersson, and F. Johnsson, “Account for variations in the H2O to
CO2 molar ratio when modelling gaseous radiative heat transfer with the weighted-sum-of-
grey-gases model,” Combustion and Flame, vol. 158, no. 5, pp. 893–901, 2011.

[249] V. Johansen and J. I. Bhatty, “Fluxes and Mineralizers in Clinkering Process,” in Innovations in
Portland Cement Manufacturing, 1st ed., J. I. Bhatty, F. M. Miller, S. H. Kosmatka, and R. P.
Bohan, Eds. Skokie, Illinois, USA: Portland Cement Association, 2011, pp. 401–438.

[250] Mathworks, “ode15s,” 2018. [Online]. Available:

https://se.mathworks.com/help/matlab/ref/ode15s.html. [Accessed: 03-Apr-2018].

[251] A. Kaw, “Shooting Method for Ordinary Differential Equations,” nm.mathforcollege.com,

2009. [Online]. Available: http://nm.mathforcollege.com/topics/shooting_method.html.
[Accessed: 03-Apr-2018].
232 11 References

[252] “Everything You Need to Know About Operation and Control of Clinker Cooler.” [Online].
Available: http://www.cementequipment.org/. [Accessed: 13-Apr-2018].

[253] S. R. Turns, An Introduction to Combustion, 3rd ed. McGraw-Hill, 2012.

[254] R. B. Silva, S. Martins-Dias, C. Arnal, M. U. Alzueta, and M. Costa, “Pyrolysis and char
characterization of refuse-derived fuel components,” Energy and Fuels, vol. 29, no. 3, pp.
1997–2005, 2015.

[255] J. M. Johansen, P. A. Jensen, P. Glarborg, M. Mancini, R. Weber, and R. E. Mitchell, “Extension of

apparent devolatilization kinetics from thermally thin to thermally thick particles in zero
dimensions for woody biomass,” Energy, vol. 95, pp. 279–290, 2016.

[256] D. D. Evans and H. W. Emmons, “Combustion of wood charcoal,” Fire Safety Journal, vol. 1, no.
1, pp. 57–66, 1977.

[257] C. Di Blasi, F. Buonanno, and C. Branca, “Reactivities of some biomass chars in air,” vol. 37, pp.
1227–1238, 1999.

[258] H. Lu, W. Robert, G. Peirce, B. Ripa, and L. L. Baxter, “Comprehensive Study of Biomass Particle
Combustion,” no. 4, pp. 2826–2839, 2008.

[259] Y. Kim and W. Parker, “A technical and economic evaluation of the pyrolysis of sewage sludge
for the production of bio-oil,” Bioresource Technology, vol. 99, pp. 1409–1416, 2008.

[260] A. Demirbas, “Effects of moisture and hydrogen content on the heating value of fuels,” Energy
Sources, Part A: Recovery, Utilization and Environmental Effects , vol. 29, no. 7, pp. 649–655,
2007.

[261] R. Ahmad, T. A. Khan, and V. Agarwal, “Mass and Energy Balance in Grate Cooler of Cement
Plant,” International Journal of Scientific Engineering and Technology , vol. 2, no. 7, pp. 631–
637, 2013.

[262] K. S. Mujumdar, K. V. Ganesh, S. B. Kulkarni, and V. V. Ranade, “Rotary Cement Kiln Simulator
(RoCKS): Integrated modeling of pre-heater, calciner, kiln and clinker cooler,” Chemical
Engineering Science, vol. 62, no. 9, pp. 2590–2607, 2007.

[263] D. Touil, H. F. Belabed, C. Frances, and S. Belaadi, “Heat Exchange Modeling of a Grate Clinker
Cooler and Entropy Production Analysis,” International Journal of Heat and Technology, vol.
23, pp. 61–68, 2005.

[264] U. Akay, “The Fons Delta Level Transmitter,” International Cement Review, 2013.
12 APPENDICES

The following appendices are included to provide additional clarification and details
on certain subjects covered in the thesis.

A. Flame Analysis of Image Intensities ................................................................................................................. 234

B. Determination of Kinetics for Alite Formation ............................................................................................ 236
C. Determination of View Factors for Radiation .............................................................................................. 238
D. Oxygen Requirement for Combustion ............................................................................................................. 241
E. Validation of Biomass Combustion Model ..................................................................................................... 243
F. Comparison of Isothermal and Non-Isothermal Devolatilization Model ......................................... 250
G. Bed and Gas Phase Energy Balances ................................................................................................................ 255
H. Model Parameters for Industrial Combustion Simulations .................................................................... 262
I. Details of Combustion Simulations ................................................................................................................... 269
J. Model for the Clinker Cooler ................................................................................................................................ 279
 234 12 Appendices

A. Flame Analysis of Image Intensities

An example of the analysis of changing image intensities is shown here. Three
example images consisting of 11x11 pixel are shown in Figure 12-1a-c. In the images
the intensity is increased from the top left corner to the bottom right corner. The
upper left pixel in Figure 12-1a has a value of 0 (i.e. black), and the bottom right
pixel in Figure 12-1c has a value of 255 (i.e. white).

The graphs presented in Figure 12-1d-f are the image intensities in the horizontal
and vertical direction of each image in Figure 12-1a-c. The horizontal image
intensity profile is taken across the image from left to right in row 6, and the vertical
profile is taken from the top to bottom in column 11. The graphs thus indicate how
the image intensity changes from the left to the right, and from the top to the bottom.
Furthermore it is seen that the intensity is generally higher in Figure 12-1c than in
Figure 12-1a.

Figure 12-1: a-c) Three example images in grayscale with varying image intensity. The
image intensity increases from top left to bottom right in each image. d-f) Image
intensities in horizontal direction from left to right, and in vertical direction from top
to bottom.

In Figure 12-2 the intensity profiles from Figure 12-1d-f are combined to show how
the intensity changes from pixel to pixel in each frame (y-axis), and how it changes
A Flame Analysis of Image Intensities 235

between the frames (x-axis). The color indicates the image intensity from Figure
12-1a-c. Thus it is seen in Figure 12-2a+b how the image intensity increases as the
pixel number is increased from 1 to 11, and it can be seen how the general image
intensity increases over the three frames.

Figure 12-2: Image intensities in the 3 example images. a) in horizontal direction in

center of image (pixel 1 is left side of Figure 12-2) , and b) in vertical direction in
right most pixel column (pixel 1 is top of Figure 12-2).

The same approach is used for the recorded videos. They just consist of a greater
number of frames with a higher resolution than what is exemplified above.
 236 12 Appendices

B. Determination of Kinetics for Alite Formation

The burnability of raw meal samples from two different cement plants were
determined at the FLSmidth laboratories. The raw meal composition is given in
Table 12-1. The minor oxides (MgO, K2O, etc.) have been grouped as inerts. The loss
on ignition is the mass loss due to calcination when the raw meal sample is heated.

Table 12-1: Raw meal composition of test samples.

Loss on
Composition SiO2 Al2O3 Fe2O3 CaO Inerts Total
ignition
Sample 1 wt. % 13.90 3.80 2.56 43.10 1.24 35.20 99.80
Sample 2 wt. % 13.70 3.09 1.76 43.10 2.74 35.50 99.89

The raw meal burnability is tested according to an internal FLSmidth Standard. The
samples are burned for 30 minutes at temperatures of 1400, 1450, and 1500 °C.
Afterwards the free lime, C2S, and C3S content were determined, with the results
shown in Table 12-2. It is observed that sample 1 obtains a lower free lime content
than sample 2, thus it is easy to burn compared to sample 2.

Table 12-2: Measured CaO, C2S, and C3S in burnability tests.

Temperature (°C) 1400 1450 1500

Composition CaO C2S C3S CaO C2S C3S CaO C2S C3S
Sample 1 wt. % 1.8 22 60 1.4 21 60 0.9 20 61
Sample 2 wt. % 4.8 30 52 3.8 23 56 2.5 19 63

The kinetics of Mastorakos [201] were adjusted using a trial and error procedure.
The initial adjustment resulted in using the following values for the alite formation,
A = 3*108 m3/kg/s and Ea = 440 kJ/mol. Using these kinetics, the model of alite was
compared with the burnability tests for sample 2, as shown in Figure 12-3a. The
reaction is too slow at 1400, while it is too fast at 1450 and 1500 °C.

The rate constant for reaction at each of the three temperatures were then adjusted
to minimize the difference between the model predictions and the laboratory
determined alite content, as shown in Figure 12-3b. An excellent fit is seen between
the model and lab measurements of alite content. A deviation is seen in the belite
and free lime contents, since the real samples do not behave according to the
idealized simplified clinker composition.
B Determination of Kinetics for Alite Formation 237

Figure 12-3: Comparison of model predictions with laboratory burnability tests for
sample 2. a) Using standard kinetics (A = 3*108 m3/kg/s, Ea = 440 kJ/mol). b) using
optimized kinetics.

The optimal rate constants at each temperature have now been determined. Using a
fixed activation energy of 440 kJ/mol, the appropriate preexponential factor is
determined.

The same procedure can be followed for sample 1, to find the kinetics for an easy
burning cement clinker. The optimized best fitting kinetics are given in Table 12-3.
A factor 3 difference is seen in the preexponential factor for the two samples. Thus,
sample 1 will more easily react, which is consistent with its low burnability.

Table 12-3: Optimized best fitting kinetics for the alite formation.

A Ea
(m3/kg/s) (kJ/mol)
Sample 1 3.32*108 440
Sample 2 1.10*108 440
 238 12 Appendices

C. Determination of View Factors for Radiation

The radiative problem can be sketched as seen in Figure 12-4, where the three
surfaces that radiation occurs between are the jet (green circle at center), bed (red
line between A1 and A2), and kiln wall (blue outer circle).

Figure 12-4: Sketch to determine the radiation view factors.

Since it is an enclosure the sum of view factors must be 1. We have for the three
surfaces:

𝜴𝟏,𝟏 + 𝜴𝟏,𝟐 + 𝜴𝟏,𝟑 = 𝟏 E 12.1

𝜴𝟐,𝟏 + 𝜴𝟐,𝟐 + 𝜴𝟐,𝟑 = 𝟏 E 12.2

𝜴𝟑,𝟏 + 𝜴𝟑,𝟐 + 𝜴𝟑,𝟑 = 𝟏 E 12.3

Surface 1 (the bed) is flat and surface 3 (the jet) is convex so they can’t see
themselves and the view factor is 0. In addition the reciprocity relation is given as
[229]:

𝑨𝒋
𝑨𝒊 𝜴𝒊,𝒋 = 𝑨𝒋 𝜴𝒋,𝒊 ⇒ 𝜴𝒊,𝒋 = = 𝜴𝒋,𝒊 E 12.4
𝑨𝒊

This allows the following rewriting of the equations E 12.1-E 12.3:

𝜴𝟏,𝟐 + 𝜴𝟏,𝟑 = 𝟏 E 12.5

C Determination of View Factors for Radiation 239

𝑨𝟏
𝜴 + 𝜴𝟐,𝟐 + 𝜴𝟐,𝟑 = 𝟏 E 12.6
𝑨𝟐 𝟏,𝟐

𝑨𝟏 𝑨𝟐
𝜴𝟏,𝟑 + 𝜴 =𝟏 E 12.7
𝑨𝟑 𝑨𝟑 𝟐,𝟑

This results in three equations with four unknown view factors. I.e. it is needed to
determine one before the equations can be solved.

The view factor Ω1,3 (between the bed and jet) will be determined by use of Hottel’s
crossed-string method [229]. From Figure 12-4 the result of the method yields as
the problem is symmetrical:

𝒍𝑨𝟏𝑩𝟐 + 𝒍𝑨𝟐𝑩𝟏 − 𝒍𝑨𝟏𝑩𝟏 − 𝒍𝑨𝟐𝑩𝟐 𝟐𝒍𝑨𝟐𝑩𝟏 − 𝟐𝒍𝑨𝟏𝑩𝟏

𝑨𝟏 ∗ 𝜴𝟏,𝟑 = =
𝟐 𝟐 E 12.8
= 𝒍𝑨𝟐𝑩𝟏 − 𝒍𝑨𝟏𝑩𝟏

Since the problem is considered in 2-D only, as only radial heat transfer is
considered, the surface area reduces to just the length of the chord between points
A1 and A2:

𝜽
𝑨𝟏 = 𝒍𝑨𝟏𝑨𝟐 = 𝟐 ∗ 𝒔𝒊𝒏 ∗𝒓 E 12.9
𝟐 𝒌

Thus equation E 12.8 can be rewritten to:

𝒍𝑨𝟐𝑩𝟏 − 𝒍𝑨𝟏𝑩𝟏 E
𝜴𝟏,𝟑 =
𝜽
𝟐𝒓𝒌 ∗ 𝒔𝒊𝒏 𝟐 12.10

The distance lA1B1 can be determined as:

E
𝒍𝑨𝟏𝑩𝟏 = √𝒍𝟐𝑨𝟏𝑪 + 𝒍𝟐𝑩𝟏𝑪 = √𝒓𝟐𝒌 + 𝒓𝟐𝒋𝒆𝒕
12.11

This is the Pythagorean theorem as the triangle connecting A1, B1, and the center of
the jet forms a right-angled triangle, due to the line A1B1 being a tangent to the jet
circle.

The distance lA1D1B2 = lA1D1+lD1B2 also needs to be determined. Distance lA1D1= lA1B1,
which is known from E 12.11 is already determined. The distance lD1B2 lies on the
circle arch and is determined by the angle ∠B2CD1. The following relation is true for
the angle:
240 12 Appendices

E
∠𝑩𝟐 𝑪𝑫𝟏 = ∠𝑨𝟏 𝑪𝑨𝟐 = 𝜽
12.12

Thus, the last length can be found as:

E
𝒍𝑫𝟏𝑩𝟐 = 𝒓𝒋𝒆𝒕 ∗ 𝜽
12.13

And finally, the view factor between bed and jet can be found as:

√𝒓𝟐𝒌 + 𝒓𝟐𝒋𝒆𝒕 + 𝒓𝒋𝒆𝒕 ∗ 𝜽 − √𝒓𝟐𝒌 + 𝒓𝟐𝒋𝒆𝒕 𝒓𝒋𝒆𝒕 ∗ 𝜽 E

𝜴𝟏,𝟑 = =
𝜽 𝜽 12.14
𝟐𝒓𝒌 ∗ 𝒔𝒊𝒏 𝟐 𝟐𝒓𝒌 ∗ 𝒔𝒊𝒏 𝟐

With this solved the remaining three view factors can be deduced:

E
𝜴𝟏,𝟐 = 𝟏 − 𝜴𝟏,𝟑
12.15

𝜽
𝑨𝟑 𝑨𝟏 𝟐𝝅𝒓𝒋𝒆𝒕 𝟐𝒓𝒌 ∗ 𝒔𝒊𝒏 𝟐 E
𝜴𝟐,𝟑 = (𝟏 − 𝜴 )= (𝟏 − 𝜴𝟏,𝟑 )
𝑨𝟐 𝑨𝟑 𝟏,𝟑 𝟐𝝅𝒓𝒌 − 𝒓𝒌 𝜽 𝟐𝝅𝒓𝒋𝒆𝒕 12.16

𝜽 E
𝑨𝟏 𝟐𝒓𝒌 ∗ 𝒔𝒊𝒏
𝜴𝟐,𝟐 =𝟏− 𝜴 − 𝜴𝟐,𝟑 = 𝟏 − 𝟐 𝜴 −𝜴
𝑨𝟐 𝟏,𝟐 𝟐𝝅𝒓𝒌 − 𝒓𝒌 𝜽 𝟏,𝟐 𝟐,𝟑 12.17

The following definitions of areas, that reduces to length chords in 2-D, have been
used:

𝜽 E
𝑨𝟏 = 𝟐𝒓𝒌 ∗ 𝒔𝒊𝒏
𝟐 12.18

E
𝑨𝟐 = 𝟐𝝅𝒓𝒌 − 𝒓𝒌 𝜽
12.19

E
𝑨𝟑 = 𝟐𝝅𝒓𝒋𝒆𝒕
12.20
 D Oxygen Requirement for Combustion 241

D. Oxygen Requirement for Combustion

The simplified method of calculating the volatile yield includes a heat and mass
balance on the char and volatile yield given by the proximate analysis and the
heating value.

This results in three products from the devolatilization: carbon monoxide (CO),
methane (CH4), as well as remaining char (C).

The oxygen consumption for the combustion, per kg fuel, can then be calculated as:

𝒀𝒄𝒉𝒂𝒓 𝒀𝒗𝒐𝒍 𝒀𝒗𝒐𝒍 E

𝒏𝑶𝟐,𝒓𝒆𝒒 = + ∗ 𝒀𝒗𝒐𝒍,𝑪𝑶 + ∗ 𝒀𝒗𝒐𝒍,𝑪𝑯𝟒 ∗ 𝟐
𝑴𝑪 𝑴𝑪𝑶 𝑴𝑪𝑯𝟒 12.21

Where Ychar and Yvol indicate the mass fraction of char and volatiles determined by
the proximate analysis, Yvol,CO and Yvol,CH4 are the fraction of volatiles present as CO
and CH4.

A more rigorous method is to consider the fuel ultimate analysis where the oxygen
consumption can be calculated as:

𝒀𝑪 𝒀𝑯 𝟏 𝒀𝑺 𝟏 𝒀𝑶 E
𝒏𝑶𝟐,𝒓𝒆𝒈 = +𝟐 + −
𝑴𝑪 𝑴𝑯 𝟐 𝑴𝑺 𝟐 𝑴𝑶 12.22

Where YC, YH, YS, and YO indicate the mass fraction of carbon, hydrogen, sulfur, and
oxygen in the fuel. The nitrogen in the fuel will be assumed released as N2.

A comparison of the oxygen consumption of the two methods is shown in Table 12-4,
for the different fuels used in this study. It is seen that for some fuels there is a
considerable difference in the calculated oxygen consumption. This difference in air
consumption of the two methods also leads to significant differences in the adiabatic
temperature. However, for the coal the difference between the two methods is
insignificant. For coal 2 and the SHW the ultimate analysis is not reported in the
source material.

In order to compensate for the difference in air consumption between the two
methods, extra nitrogen is added to the secondary air. Compensating with nitrogen
instead of air, keeps the total flow of oxygen into the kiln constant. The addition of
extra nitrogen, will limit the combustion temperatures, which may otherwise be too
high, if too little air is admitted to the kiln.
242 12 Appendices

Table 12-4: The proximate and ultimate analysis of the fuels used in the model studies.
Including the calculated oxygen consumption and adiabatic temperature by simple
(1) and rigorous method (2).

Fetnuss
Property Unit Coal Coal 1 MBM Coal 2 SHW SRF
LHV MJ/kg 34.0 28.0 18.5 27.1 15.0 15.4
Moisture wt% 0.6 1.0 4.0 1.7 30.0 21.4
Ash wt% 2.8 13.6 27.1 19.4 17.6 17.6
Char wt% 72.3 62.4 8.0 51.4 1.0 6.1
Volatiles wt% 24.3 23.0 60.9 27.5 51.4 54.9
Carbon wt% 85.2 72.9 47.1 - - 38.5
Hydrogen wt% 4.6 3.9 6.9 - - 5.5
Nitrogen wt% 1.4 1.7 9.7 - - 0.8
Sulfur wt% 0.9 1.4 0.5 - - 0.5
Oxygen wt% 4.4 5.5 4.7 - - 15.7
Source [80] [81,83] [81,83] [82] [82]
Oxygen Req. 1 mol/kg 85.5 70.0 43.8 67.8 37.2 37.2
Oxygen Req. 2 mol/kg 81.1 68.9 54.9 - - 40.9
Adiabatic Temp. 1 K 2038 2038 2071 2040 1920 1971
Adiabatic Temp, 2 K 2122 2060 1726 - - 1835
 E Validation of Biomass Combustion Model 243

E. Validation of Biomass Combustion Model

The biomass combustion model is validated against experiments found in the
literature [31,86,87,176]. The results form a good basis for validation since they
contain data on devolatilization and char oxidation times for different fuels over a
wide range of conditions with varying temperatures and oxygen concentrations.

Details of Experiments
The experiments from Momeni [86], Lu [176], and Nørskov [31] are performed at
the Technical University of Denmark, using the same single particle combustor
setup. The fuel particle is inserted into a furnace heated by a gas flame, and the
combustion process is monitored by a video camera. The experiments of Mason [87]
were performed in the flame of a Meker burner.

The experimental conditions from the different experiments are given in Table 12-5.
Five difference conditions are used by Nørskov, while the other sources use only one
set of conditions. It is assumed that for the experiments carried out in the single
particle combustor (Momeni, Lu, Nørskov), the gas temperature (convective heat
transfer) and wall temperature (radiative heat transfer) are equal. The experiments
by Mason are performed in an open flame, here it will be assumed that radiation
from the particle is governed by the ambient temperature outside the flame.

Table 12-5: Experimental conditions used in literature.

Experimental
Tg Tw ug yO2
Source condition
°C °C m/s vol%
Momeni 1 1473 1473 3.4 20
Lu 2 1504 1504 2.76 3
Nørskov EC1 3 1200 1200 2 5.8
Nørskov EC2 4 1450 1450 2 2.8
Nørskov EC3 5 1450 1450 2 6.6
Nørskov EC4 6 1475 1475 2 12
Nørskov EC5 7 1550 1550 2 5.9
Mason 8 1550 27 3 10.8

Fuels Used
The fuels studied include pine wood (Nørskov [31], Momeni [86], Mason [87]),
schima wood (Lu [176]), eucalyptus and willow wood (Mason [87]), and sewage
sludge (SS) (Nørskov [31]). Details of the fuels are found in Table 12-6. Since the
244 12 Appendices

char content is dependent on the heating rate of the biomass, it will typically be
lower than measured in the proximate analysis [96,176]. Here it will be assumed
that the char content during experiments is 70 % of the proximate char, on a dry
basis.

Table 12-6: Properties of the fuels used in experiments.

Momeni Lu Nørskov Nørskov Nørskov Mason Mason Mason

Pine Schima Pine SSA SSB Pine Eucalyp. Willow
Fuel Type Unit 1 2 3 4 5 6 7 8
Proximate wt%
85.3 80.0 85.3 73.7 68.2 80.9 82.6 82.0
Volatiles dry
Proximate wt%
14.3 18.7 14.4 7.1 10.2 17.2 16.5 16.1
Char dry
Proximate wt%
0.3 1.3 0.3 19.1 21.6 2.0 0.9 1.9
Ash dry
Moisture wt% 9.3 5.6 9.3 6.0 8.0 8.3 8.1 2.8
Volatiles wt% 81.3 80.8 81.3 71.3 65.5 78.9 80.4 84.4
Char wt% 9.1 12.4 9.1 4.7 6.6 11.0 10.6 10.9
Ash wt% 0.3 1.2 0.3 18.0 19.9 1.8 0.8 1.8
Assumed
6.4 8.7 6.4 3.3 4.6 7.7 7.5 7.7
volatiles wt%
Assumed
84.0 84.5 84.0 72.7 67.5 82.2 83.6 87.7
char wt%
Density kg/
600 658 497 696 1136 440 616 505
dry m3
Density kg/
662 697 548 740 1235 480 670 520
wet m3
Density kg/
42 60 35 24 57 37 50 40
char m3
Heat J/kg
2296 2284 2296 2066 2041 2276 2289 2276
Capacity /K

It is assumed that the initial heat capacity is influenced by the ash content of the
sample according to E 12.23, and that it is constant during combustion.

E
𝑪𝒑 = 𝑪𝒑𝟏 ∗ (𝟏 − 𝒘𝒂𝒔𝒉 ) + 𝑪𝒑𝟐 ∗ 𝒘𝒂𝒔𝒉
12.23

With Cp1 = 2300 J/kg/K and Cp2 = 1000 J/kg/K.

The particle shape is explicitly stated in the experiments by Momeni, which allows
for calculating the particle sphericity. This is not the case for the other experiments,
where sphericity will be assumed 1.
E Validation of Biomass Combustion Model 245

Kinetics Used in Model

The kinetics of the biomass combustion reactions are given in Table 12-7. Some
additional physical parameters are also available. The values are based on different
literature references.

Table 12-7: Kinetics of biomass combustion reactions and additional physical

parameters.

Parameter Symbol Unit Value Source

Preexponential
5.13*1010 [236]
factor drying Adry 1/s
Activation energy
kJ/mol 88 [236]
char drying Eadry
Preexponential
Low temperature kinetics
factor 1/s 1.11*1011
[255]
devolatilization Adevol
Activation energy
Low temperature kinetics
char kJ/mol 177
[255]
devolatilization Eadevol
Preexponential
kg/
factor char 2.54*10-3 [256]
(m2*s*Pa)
oxidation Achar,ox
Activation energy
kJ/mol 81.55 Average values from [257]
char oxidation Eachar,ox
Preexponential
factor char m/s 3.42*Tchar [258]
gasification Achar,gsc
Activation energy
kJ/mol 130 [258]
char gasification Achar, gsc
Particle emissivity εp - 0.9 Coal particle [31]
Approximate value based
Particle heat
W/(m*K) 0.12 on fuel with 10 % char
conductivity
λp [255].
Heat capacity of organic
Heat capacity of
J/(Kg K) 2300 material in range 2.1-2.5
organic fraction
CpO kJ/(kg K) [259]
Heat capacity of inorganic
Heat capacity of
J/(Kg K) 1000 material in range 0.8-1.3
ash fraction
CpA kJ/(kg K) [259]

Model Predictions Devolatilization

A comparison of predicted devolatilization times with the experimental results are
shown in Figure 12-5. The fit is generally good with average deviations for each fuel
type being maximum 15 %. Only the results of Momeni have a larger deviation. This
is due to the difficulty of describing the cylinder-shaped particles as an equivalent
sphere. Some different approaches were tested to account for the particle shape.
246 12 Appendices

Eventually, it was chosen to increase the particle surface area (for heat transfer and
char oxidation) by a factor 1/φ, since this gave the overall best fit.

For the experimental conditions the largest difference is seen for the results of
Mason and Nørskov EC2, which are at low oxygen concentration. The oxygen
concentration in the experiments have some influence, most likely because a flame
is formed above the particle at high O2 concentrations, which increases the heat
transfer. This is not considered in the model.

For the results of Mason, the experiments are made in an open flame, which means
the particle radiates heat to cold surroundings outside the flame. This might be
difficult to describe exactly with a simple model.

It should also be noted that there is significant scatter in the experimental results,
especially seen by Nørskov. The results of Mason have been derived based on
regression of several particle conversion times. Results from Momeni and Lu are
averages of multiple experiments.

The reason the model appears to work well for very different fuel types is likely the
large particles sizes. In this case heat transfer in the particle is more important than
the kinetics of devolatilization for the given fuel, which are assumed to be the same
in all cases.
E Validation of Biomass Combustion Model 247

Figure 12-5: Model and experimental comparison for devolatilization.

248 12 Appendices

Model Predictions Char Oxidation

The comparison of model predictions with experimental results of the char
oxidation times are shown in Figure 12-6. The maximum average deviation is 30 %
for the pine wood char oxidation of Nørskov. The deviation on the other results are
on average below 20 %.

For the experimental conditions the largest problems are seen for Nørskov at low
temperature (EC1) or low oxygen concentration (EC2) and for the Mason results,
which are also at low temperature. With low temperature the kinetics of the reaction
are more important, while at high temperature mass transfer of oxygen to the
particle surface is rate limiting. For the low oxygen concentration, gasification with
water is also more important. In the cement kiln the combustion temperatures are
above 1500 °C and the oxygen concentration is above 5 %, thus the errors seen here
should be on the order of maximum 20 %.

For the Mason results it is noticed that the experimental results show a curvature
with dp. For the shrinking core model for kinetically limited reaction, the
relationship between particle diameter and conversion time should be linear, as
predicted by the model. Perhaps some kind of ash diffusion could be added to get a
better fit with the experimental results. This would especially influence the largest
particles, where the largest underestimation of char oxidation time is seen.
E Validation of Biomass Combustion Model 249

Figure 12-6: Model and experimental comparison for char oxidation.

 250 12 Appendices

F. Comparison of Isothermal and Non-Isothermal

Devolatilization Model
Since SRF particles are large, internal temperature gradients are present during
devolatilization. Thus, it is necessary to somehow account for this in the model. In
this chapter a comparison of a 1D model (using the partial differential equation E
8.55) is compared to results using the isothermal particle model (E 8.53).

The partial differential equation used in the 1D model is solved by discretization

using the Method of Lines [234]. However, the discretization means that additional
equations need to be solved. This may set an additional computational strain on the
model; especially as multiple particle sizes are added to the model. Thus, a simple
approach to account for the difference between an isothermal and non-isothermal
(1D) model, may be of interest.

Comparison of Models
A comparison of the devolatilization times calculated by the isothermal and non-
isothermal model is shown in Figure 12-7. The devolatilization time is here defined
as the time it takes for the isothermal particle to reach 99.9 devolatilization, while it
for the non-isothermal particles is the center of the particle that should reach 99.9
% conversion.

Figure 12-7: Comparison of devolatilization times predicted by isothermal and non-

isothermal model for different gas temperatures (°C).
F Comparison of Isothermal and Non-Isothermal Devolatilization Model 251

The devolatilization is calculated for particles sizes in the range 0.1-20 mm and
temperatures 1000-2000 °C. The devolatilization times predicted by the isothermal
model are significantly lower than what is predicted by the non-isothermal model.
The error becomes larger as the particle size or gas temperature is increased and
temperature gradients in the particle become more pronounced.

A linear relationship between dp and tiso/tnon-iso is found for particles with dp < 1 mm,
for particles with dp > 1 mm a log-log plot reveals a linear relation, as shown in
Figure 12-8..

Figure 12-8: Linearized versions of Figure 12-7 for a) dp < 1 mm and b) dp > 1 mm in
a log-log plot.

Adjusting Devolatilization Kinetics of Isothermal Model

Johansen et al. [255] proposed to reduce devolatilization kinetics for larger
isothermal particles in order to account for their faster conversion than non-
isothermal particles. This can shift the devolatilization to occur at higher
temperatures, which will cause a delay in devolatilization. A similar approach is
used here, although a more detailed fitting of kinetic parameters is presented.

The basis for the devolatilization kinetics is a preexponential factor of 1.11*1011 1/s
and an activation energy of 177 kJ/mol [255]. The preexponential factor is first
adjusted so that tiso/tnon-iso is independent of particle size, see Figure 12-9a. Next the
activation energy is adjusted to make tiso/tnon-iso = 1, for all particles and at all
temperatures, see Figure 12-9b. One set of parameters is found for particles with dp
between 0.5 and 1 mm, and a second set is found for particles with dp above 1 mm.
For particles below 0.5 mm, the difference between the two models is lower than
0.01 s, and thus adjustments are not deemed necessary.
252 12 Appendices

Figure 12-9: Comparison of devolatilization times predicted by isothermal and non-

isothermal models after adjustment of preexponential factor (a) and activation
energy (b).

The adjusted values of preexponential factor (Adevol) and activation energy (Eadevol)
are then fitted across particle diameter and gas temperature. The following fits are
derived. In the equations dp should be inserted in mm and Tg in K.

For particles smaller than 1 mm:

The preexponential factor is found by:

E
𝒍𝒐𝒈(𝑨𝟏 ) = 𝒂𝑨𝟏 ∗ 𝒍𝒐𝒈(𝒅𝒑 ) + 𝒃𝑨𝟏
12.24

E
𝒂𝑨𝟏 = 𝒂𝑨𝟏,𝟏 ∗ 𝒍𝒐𝒈(𝑻𝒈 ) + 𝒂𝑨𝟏,𝟐
12.25

E
𝒃𝑨𝟏 = 𝒃𝑨𝟏,𝟏 ∗ 𝒍𝒐𝒈(𝑻𝒈 ) + 𝒃𝑨𝟏,𝟐
12.26

The activation energy is found by:

E
𝑬𝟏 = 𝒂𝑬𝟏 ∗ 𝒍𝒐𝒈(𝒅𝒑) + 𝒃𝑬𝟏
12.27

E
𝒍𝒐𝒈(−𝒂𝑬𝟏 ) = 𝒂𝑬𝟏,𝟏 ∗ 𝑻 + 𝒂𝑬𝟏,𝟐
12.28

E
𝒍𝒐𝒈(𝒃𝑬𝟏 ) = 𝒃𝑬𝟏,𝟏 ∗ 𝑻 + 𝒃𝑬𝟏,𝟐
12.29

For particles larger than 1 mm:

F Comparison of Isothermal and Non-Isothermal Devolatilization Model 253

The preexponential factor is found by:

E
𝒍𝒐𝒈(𝑨𝟐 ) = 𝒂𝑨𝟐 ∗ 𝒍𝒐𝒈(𝒅𝒑) + 𝒃𝑨𝟐
12.30

E
𝒂𝑨𝟐 = 𝒂𝑨𝟐,𝟏 ∗ 𝒍𝒐𝒈(𝑻)𝟐 + 𝒂𝑨𝟐,𝟐 ∗ 𝒍𝒐𝒈(𝑻) + 𝒂𝑨𝟐,𝟑
12.31

E
𝒃𝑨𝟐 = 𝒃𝑨𝟐,𝟏 ∗ 𝒍𝒐𝒈(𝑻) + 𝒃𝑨𝟐,𝟐
12.32

The activation energy is found by:

E
𝒍𝒐𝒈(𝑬𝟐 ) = 𝒂𝑬𝟐 ∗ 𝒆𝒙𝒑(𝒃𝑬𝟐 ∗ 𝒅𝒑 ) + 𝒄𝑬𝟐
12.33

E
𝒂𝑬𝟐 = 𝒂𝑬𝟐,𝟏 ∗ 𝒍𝒐𝒈(𝑻)𝟐 + 𝒂𝑬𝟐,𝟐 ∗ 𝒍𝒐𝒈(𝑻) + 𝒂𝑬𝟐,𝟑
12.34

E
𝒃𝑬𝟐 = 𝒃𝑬𝟐,𝟏 ∗ 𝒍𝒐𝒈(𝑻)𝟐 + 𝒃𝑬𝟐,𝟐 ∗ 𝒍𝒐𝒈(𝑻) + 𝒃𝑬𝟐,𝟑
12.35

E
𝒄𝑬𝟐 = 𝒄𝑬𝟐,𝟏 ∗ 𝒍𝒐𝒈(𝑻)𝟐 + 𝒄𝑬𝟐,𝟐 ∗ 𝒍𝒐𝒈(𝑻) + 𝒄𝑬𝟐,𝟑
12.36

The values of the regression coefficients are found in Table 12-8.

Table 12-8: Values for regression parameters to determine isothermal kinetics.

1 2 3
For dp < 1 mm
aA1 -17.39 52.01 -
b A1 -5.36 27.10 -
aE1 1.19E-03 2.14 -
bE1 1.64E-04 5.10 -
For dp > 1 mm
aA2 13.68 -90.95 147.24
bA2 -8.60 36.94 -
aE2 -1.44 9.85 -16.66
bE2 -1.55 9.99 -16.09
cE2 2.40 -15.56 30.45

Using the fitted kinetics for the isothermal model, isothermal and non-isothermal
devolatilization times are compared in Figure 12-10. The deviation is generally
lower than 10 %. For particles smaller than 0.5 mm, the error is larger, but they were
not included in the optimization.
254 12 Appendices

Figure 12-10: Comparison of devolatilization times predicted by isothermal and

non-isothermal models using adjusted kinetics for a) dp < 1 mm and b) dp > 1 mm.

It should be noted that this fitting approach only shifts the devolatilization for
isothermal particles to occur at higher temperatures. In this way the devolatilization
time of isothermal particles is delayed to coincide with that of non-isothermal
particles. It is purely a mathematic manipulation of the kinetics to make the
devolatilization times match. There is no physical basis for the shape of the
regression equations. In a non-isothermal particle outer layers may start
devolatilization before the inner layers. This could e.g. affect ignition of the fuel. This
has not been considered in this data fitting. Thus, using this approach could result
in a changed onset of devolatilization between isothermal and non-isothermal
particles.
 G Bed and Gas Phase Energy Balances 255

G. Bed and Gas Phase Energy Balances

Energy Balance for the Bed Phase

The equation for the bed temperature is derived based on an energy balance. The
balance is performed observing a control volume of length Δx in the axial direction
of the kiln, as shown in Figure 8-11.

The energy entering the control volume is given by the clinker flow plus the energy
contained in the clinker melt:

𝑬𝒊𝒏 = 𝒖𝑏 (𝒙) ∗ 𝑨𝒃 (𝒙) ∗ 𝝆𝒃 (𝒙) ∗ 𝜟𝒕 E

∗ [𝑪𝒑𝒃 ∗ 𝑻𝒃 (𝒙) + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 (𝒙)] 12.37

The energy leaving the control volume is similarly given by the clinker flow:

𝑬𝒐𝒖𝒕 = 𝒖𝑏 (𝒙 + 𝜟𝒙) ∗ 𝑨𝑏 (𝒙 + 𝜟𝒙) ∗ 𝝆𝑏 (𝒙 + 𝜟𝒙) ∗ 𝜟𝒕 E

∗ [𝑪𝒑𝑏 ∗ 𝑻𝒃 (𝒙 + 𝜟𝒙) + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 (𝒙 + 𝜟𝒙)] 12.38

In the equations Xmelt is the fraction of melt in the kiln, which is assumed to change
linearly, being 0 at a certain temperature and Xmelt,max at another temperature, i.e.:

𝑻𝒃 − 𝑻𝒎𝒆𝒍𝒕,𝟏 E
𝑿𝒎𝒆𝒍𝒕 = 𝑿𝒎𝒆𝒍𝒕,𝒎𝒂𝒙 ∗
𝑻𝒎𝒆𝒍𝒕,𝟐 − 𝑻𝒎𝒆𝒍𝒕,𝟏 12.39

And the rate of melt formation for later use is:

𝒅 𝑿𝒎𝒆𝒍𝒕,𝒎𝒂𝒙 𝒅 E
𝑿𝒎𝒆𝒍𝒕 = ∗ 𝑻
𝒅𝒙 𝑻𝒎𝒆𝒍𝒕,𝟐 − 𝑻𝒎𝒆𝒍𝒕,𝟏 𝒅𝒙 𝒃 12.40

With the following limitations:

𝑿𝒎𝒆𝒍𝒕 = 𝟎 𝒇𝒐𝒓 𝑻𝒃 < 𝑻𝒎𝒆𝒍𝒕,𝟏 E

𝑿𝒎𝒆𝒍𝒕 = 𝑿𝒎𝒆𝒍𝒕,𝒎𝒂𝒙 𝒇𝒐𝒓 𝑻𝒃 > 𝑻𝒎𝒆𝒍𝒕,𝟐 12.41

The energy formed and consumed by reactions in the control volume is:
𝑵𝒓
E
𝑬𝒓 = ∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝑏 ∗ 𝜟𝒙 ∗ 𝜟𝒕
12.42
𝒊=𝟏

The energy transferred to and away from the clinker bed:

𝑵𝒉𝒕
E
𝑬𝒉𝒕 = ∑ 𝒒𝒊 ∗ 𝒍𝒊 ∗ 𝜟𝒙 ∗ 𝜟𝒕
12.43
𝒊=𝟏
256 12 Appendices

The energy lost from the clinker bed due to CO2 release from the calcination
reaction:

E
𝑬𝒍𝒐𝒔𝒔 = 𝑹𝑪𝑶𝟐 ∗ 𝑯𝑪𝑶𝟐 ∗ 𝑨𝒃 ∗ 𝜟𝒙 ∗ 𝜟𝒕
12.44

The energy balance can now be derived by considering the balance equation:

E
𝑶𝒖𝒕 − 𝑰𝒏 = 𝑷𝒓𝒐𝒅𝒖𝒄𝒆𝒅
12.45

The equations E 12.37, E 12.38, and E 12.42-E 12.44 combine to:

𝒖𝑏 (𝒙 + 𝜟𝒙) ∗ 𝑨𝑏 (𝒙 + 𝜟𝒙) ∗ 𝝆𝑏 (𝒙 + 𝜟𝒙) ∗ 𝜟𝒕

∗ [𝑪𝒑𝑏 ∗ 𝑻𝒃 (𝒙 + 𝜟𝒙) + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 (𝒙 + 𝜟𝒙)]
− 𝒖𝑏 (𝒙) ∗ 𝑨𝒃 (𝒙) ∗ 𝝆𝒃 (𝒙) ∗ 𝜟𝒕
E
∗ [𝑪𝒑𝒃 ∗ 𝑻𝒃 (𝒙) + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 (𝒙)]
𝑵𝒓 𝑵𝒉𝒕 12.46
= ∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝑏 ∗ 𝜟𝒙 ∗ 𝜟𝒕 + ∑ 𝒒𝒊 ∗ 𝒍𝒊 ∗ 𝜟𝒙 ∗ 𝜟𝒕
𝒊 𝒊

− 𝑹𝑪𝑶𝟐 ∗ 𝑯𝑪𝑶𝟐 ∗ 𝑨𝒃 ∗ 𝜟𝒙 ∗ 𝜟𝒕

The Δt’s cancel out and the limit of the Δx is taken to yield:

𝒅
(𝒖 ∗ 𝑨𝒃 ∗ 𝑪𝒑𝒃 ∗ 𝝆𝒃 ∗ 𝑻𝒃 + 𝝆𝒃 ∗ 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 )
𝒅𝒙 𝒃 E
𝑵𝒓 𝑵𝒉𝒕
12.47
= ∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝑏 + ∑ 𝒒𝒊 ∗ 𝒍𝒊 − 𝑹𝑪𝑶𝟐 ∗ 𝑯𝑪𝑶𝟐 ∗ 𝑨𝒃
𝒊=𝟏 𝒊=𝟏

Assuming that the bed velocity, cross sectional area, and heat capacity are constant
the equation simplifies to:

𝒅 𝒅
𝒖𝒃 ∗ 𝑨𝒃 ∗ [𝑪𝒑𝑏 ∗ (𝝆𝒃 ∗ 𝑻𝒃 ) + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ (𝝆 ∗ 𝑿𝒎𝒆𝒍𝒕 )]
𝒅𝒙 𝒅𝒙 𝒃 E
𝑵𝒓 𝑵𝒉𝒕
12.48
= ∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝑏 + ∑ 𝒒𝒊 ∗ 𝒍𝒊 − 𝑹𝑪𝑶𝟐 ∗ 𝑯𝑪𝑶𝟐 ∗ 𝑨𝒃
𝒊=𝟏 𝒊=𝟏

The following rewriting is performed on the left-hand side of the equation to

simplify the differential expression:

𝒅 𝒅 E
𝑪𝒑𝑏 ∗ (𝝆𝒃 ∗ 𝑻𝒃 ) + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ (𝝆 ∗ 𝑿𝒎𝒆𝒍𝒕 )
𝒅𝒙 𝒅𝒙 𝒃 12.49
G Bed and Gas Phase Energy Balances 257

This can be rewritten using the product rule

𝒅 𝒅 𝒅
𝑪𝒑𝒃 ∗ 𝝆𝒃 (𝑻𝒃 ) + 𝑪𝒑𝒃 ∗ 𝑻𝒃 (𝝆𝒃 ) + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝝆𝒃 (𝑿 )
𝒅𝒙 𝒅𝒙 𝒅𝒙 𝒎𝒆𝒍𝒕 E
𝒅 12.50
+ 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 (𝝆 )
𝒅𝒙 𝑏

Inserting the expression for the rate of melt change from E 12.40 and collecting the
derivatives gives:

𝑿𝒎𝒆𝒍𝒕,𝒎𝒂𝒙 𝒅
(𝑪𝒑𝒃 ∗ 𝝆𝒃 + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝝆𝒃 ∗ ) (𝑻 )
𝑻𝒎𝒆𝒍𝒕,𝟐 − 𝑻𝒎𝒆𝒍𝒕,𝟏 𝒅𝒙 𝒃 E
𝒅 12.51
+ (𝑪𝒑𝒃 ∗ 𝑻𝒃 + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 ) (𝝆 )
𝒅𝒙 𝒃

Equation E 12.51 is now inserted into E 12.48, which yields the expression:

𝑿𝒎𝒆𝒍𝒕,𝒎𝒂𝒙 𝒅
(𝑪𝒑𝒃 ∗ 𝝆𝒃 + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝝆𝒃 ∗ ) (𝑻 )
𝑻𝒎𝒆𝒍𝒕,𝟐 − 𝑻𝒎𝒆𝒍𝒕,𝟏 𝒅𝒙 𝒃
𝒅
+ (𝑪𝒑𝒃 ∗ 𝑻𝒃 + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 ) (𝝆 )
𝒅𝒙 𝒃
𝑵𝒓 𝑵𝒉𝒕 E
𝟏
= (∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝑏 + ∑ 𝒒𝒊 ∗ 𝒍𝒊 − 𝑹𝑪𝑶𝟐 12.52
𝒖𝒃 ∗ 𝑨𝒃
𝒊=𝟏 𝒊=𝟏

∗ 𝑯𝑪𝑶𝟐 ∗ 𝑨𝒃 )

The rate of density change is caused by decomposition of CaCO3 according to:

𝒅𝝆𝒄𝒍 𝟏 𝑴𝑪𝑶𝟐 𝒌𝒈 E
=− ∗ 𝑹𝟏 ∗ [ 𝟑 ]
𝒅𝒙 𝒖𝒃 𝑴𝑪𝒂𝑪𝑶𝟑 𝒎 𝒎 12.53

And the rate of CO2 release is given as:

𝑴𝑪𝑶𝟐 𝒌𝒈 E
𝑹𝑪𝑶𝟐 = 𝑹𝟏 ∗ [ ]
𝑴𝑪𝒂𝑪𝑶𝟑 𝒎𝟑 𝒔 12.54

The definitions of E 12.53 and E 12.54 are used in E 12.52 yielding:

258 12 Appendices

𝑿𝒎𝒆𝒍𝒕,𝒎𝒂𝒙 𝒅
(𝑪𝒑𝒃 ∗ 𝝆𝒃 + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝝆𝒃 ∗ ) (𝑻 )
𝑻𝒎𝒆𝒍𝒕,𝟐 − 𝑻𝒎𝒆𝒍𝒕,𝟏 𝒅𝒙 𝒃
𝟏 𝑴𝑪𝑶𝟐
− (𝑪𝒑𝒃 ∗ 𝑻𝒃 + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 ) ∗ ∗ 𝑹𝟏 ∗
𝒖𝒃 𝑴𝑪𝒂𝑪𝑶𝟑
𝑵𝒓 𝑵𝒉𝒕 E
𝟏
= (∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝑏 + ∑ 𝒒𝒊 ∗ 𝒍𝒊 − 𝑹𝟏 12.55
𝒖𝒃 ∗ 𝑨𝒃
𝒊=𝟏 𝒊=𝟏

𝑴𝑪𝑶𝟐
∗ ∗ 𝑯𝑪𝑶𝟐 ∗ 𝑨𝒃 )
𝑴𝑪𝒂𝑪𝑶𝟑

Assuming that the CO2 has a similar constant heat capacity to that of the clinker the
enthalpy for CO2 can be written as:

E
𝑯𝑪𝑶𝟐 = 𝑪𝒑𝒃 ∗ 𝑻𝒃
12.56

Inserting this knowledge leads to:

𝑿𝒎𝒆𝒍𝒕,𝒎𝒂𝒙 𝒅
(𝑪𝒑𝒃 ∗ 𝝆𝒃 + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝝆𝒃 ∗ ) (𝑻 )
𝑻𝒎𝒆𝒍𝒕,𝟐 − 𝑻𝒎𝒆𝒍𝒕,𝟏 𝒅𝒙 𝒃
𝑵𝒓 𝑵𝒉𝒕
𝟏
= (∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝑏 + ∑ 𝒒𝒊 ∗ 𝒍𝒊 )
𝒖𝒃 ∗ 𝑨𝒃 E
𝒊=𝟏 𝒊=𝟏
12.57
𝟏 𝑴𝑪𝑶𝟐
− ( 𝑹𝟏 ∗ ∗ 𝑪𝒑𝒃 ∗ 𝑻𝒃 )
𝒖𝒃 𝑴𝑪𝒂𝑪𝑶𝟑
𝟏 𝑴𝑪𝑶𝟐
+ (𝑪𝒑𝒃 ∗ 𝑻𝒃 + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 ) ∗ ∗ 𝑹𝟏 ∗
𝒖𝒃 𝑴𝑪𝒂𝑪𝑶𝟑

And the right-hand side of the equation can be simplified to:

𝑿𝒎𝒆𝒍𝒕,𝒎𝒂𝒙 𝒅
(𝑪𝒑𝒃 ∗ 𝝆𝒃 + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝝆𝒃 ∗ ) (𝑻 )
𝑻𝒎𝒆𝒍𝒕,𝟐 − 𝑻𝒎𝒆𝒍𝒕,𝟏 𝒅𝒙 𝒃
𝑵𝒓 𝑵𝒉𝒕
𝟏 E
= (∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝑏 + ∑ 𝒒𝒊 ∗ 𝒍𝒊 )
𝒖𝒃 ∗ 𝑨𝒃 12.58
𝒊=𝟏 𝒊=𝟏

𝟏 𝑴𝑪𝑶𝟐
+ (𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 ) ∗ ∗ 𝑹𝟏 ∗
𝒖𝒃 𝑴𝑪𝒂𝑪𝑶𝟑

The temperature derivative can now be isolated:

G Bed and Gas Phase Energy Balances 259

𝒅
𝑻
𝒅𝒙 𝒃
𝑵𝒓
𝟏 𝟏
= (∑ 𝑹𝒊 E
𝑿𝒎𝒆𝒍𝒕,𝒎𝒂𝒙 𝒖 𝒃 ∗ 𝑨𝒃
(𝑪𝒑𝒃 ∗ 𝝆𝒃 + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝝆𝒃 ∗ 𝑻 ) 𝒊=𝟏
𝒎𝒆𝒍𝒕,𝟐 − 𝑻𝒎𝒆𝒍𝒕,𝟏 12.59
𝑵𝒉𝒕
𝑴𝑪𝑶𝟐
∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝑏 + ∑ 𝒒𝒊 ∗ 𝒍𝒊 + (𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 ) ∗ 𝑨𝒃 ∗ 𝑹𝟏 ∗ )
𝑴𝑪𝒂𝑪𝑶𝟑
𝒊=𝟏

The equation is further simplified to:

𝑵𝒓
𝒅 𝟏
𝑻 = (∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝑏
𝒅𝒙 𝒃 𝑿𝒎𝒆𝒍𝒕,𝒎𝒂𝒙
𝒎̇𝒃 (𝑪𝒑𝒃 + 𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑻 ) 𝒊=𝟏 E
𝒎𝒆𝒍𝒕,𝟐 − 𝑻𝒎𝒆𝒍𝒕,𝟏
𝑵𝒉𝒕 12.60
𝑴𝑪𝑶𝟐
+ ∑ 𝒒𝒊 ∗ 𝒍𝒊 + (𝜟𝑯𝒎𝒆𝒍𝒕 ∗ 𝑿𝒎𝒆𝒍𝒕 ) ∗ 𝑨𝒃 ∗ 𝑹𝟏 ∗ )
𝑴𝑪𝒂𝑪𝑶𝟑
𝒊=𝟏

If melting does not occur the equation simplifies to:

𝑵𝒓 𝑵𝒉𝒕
𝒅 𝟏 E
𝑻𝒃 = (∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝑏 + ∑ 𝒒𝒊 ∗ 𝒍𝒊 )
𝒅𝒙 𝒎̇𝒃 𝑪𝒑𝒃 12.61
𝒊=𝟏 𝒊=𝟏

Energy Balance for the Gas Phase

The energy balance for the gas phase is made for the primary air jet flow. The gas
phase energy balance is derived in a similar manner to that of the solids, based on
Figure 8-12.

The energy entering the control volume by the gas flow:

E
𝑬𝒊𝒏 = 𝒖𝒈 (𝒙) ∗ 𝑨𝒈 (𝒙) ∗ 𝑪𝒑𝒈 (𝒙) ∗ 𝝆𝒈 (𝒙) ∗ 𝑻𝒈 (𝒙) ∗ 𝜟𝒕
12.62

The energy leaving the control volume by the gas flow:

𝑬𝒐𝒖𝒕 = 𝒖𝒈 (𝒙 + 𝜟𝒙) ∗ 𝑨𝒈 (𝒙 + 𝜟𝒙) ∗ 𝑪𝒑𝒈 (𝒙 + 𝜟𝒙) ∗ 𝝆𝒈 (𝒙 + 𝜟𝒙) E

∗ 𝑻𝒈 (𝒙 + 𝜟𝒙) ∗ 𝜟𝒕 12.63

The energy released by the combustion reactions:

260 12 Appendices

𝑵𝒓
E
𝑬𝑟 = ∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝒈 ∗ 𝜟𝒙 ∗ 𝜟𝒕
12.64
𝒊=𝟏

The energy transferred to and away from the gas phase:

𝑵𝒉𝒕
E
𝑬𝒉𝒕 = ∑ 𝒒𝒊 ∗ 𝒍𝒊 ∗ 𝜟𝒙 ∗ 𝜟𝒕
12.65
𝒊=𝟏

The energy obtained from the clinker bed due to CO2 formed in the calcination
reaction:

𝑬𝑪𝑂2 = 𝑹𝑪𝑶𝟐 ∗ 𝑯𝑪𝑶𝟐 ∗ 𝑨𝑏 ∗ 𝜟𝒙 ∗ 𝜟𝒕

E
𝑴𝑪𝑶𝟐
= 𝑹𝟏 ∗ ∗ 𝑪𝒑𝒃 ∗ 𝑻𝒃 ∗ 𝑨𝑏 ∗ 𝜟𝒙 ∗ 𝜟𝒕 12.66
𝑴𝑪𝒂𝑪𝑶𝟑

The energy obtained from the fuel due to gas release from the fuel

E
𝑬𝒇𝒈 = 𝑹𝒇𝒈 ∗ 𝑯𝒇𝒈 ∗ 𝑨𝒈 ∗ 𝜟𝒙 ∗ 𝜟𝒕
12.67

The energy obtained from the entrained secondary air:

E
𝑬𝒆𝒏𝒕 = 𝑹𝒆𝒏𝒕 ∗ 𝑪𝒑𝒔𝒆𝒄 ∗ 𝑻𝒔𝒆𝒄 ∗ 𝑨𝒈 ∗ 𝜟𝒙 ∗ 𝜟𝒕
12.68

The energy balance can now be derived by considering the balance equation:

E
𝑶𝒖𝒕 − 𝑰𝒏 = 𝑷𝒓𝒐𝒅𝒖𝒄𝒆𝒅
12.69

The equations can be written as:

𝒖𝒈 (𝒙 + 𝜟𝒙) ∗ 𝑨𝒈 (𝒙 + 𝜟𝒙) ∗ 𝑪𝒑𝒈 (𝒙 + 𝜟𝒙) ∗ 𝝆𝒈 (𝒙 + 𝜟𝒙)

∗ 𝑻𝒈 (𝒙 + 𝜟𝒙) ∗ 𝜟𝒕 − 𝒖𝒈 (𝒙) ∗ 𝑨𝒈 (𝒙) ∗ 𝑪𝒑𝒈 (𝒙) ∗ 𝝆𝒈 (𝒙)
∗ 𝑻𝒈 (𝒙) ∗ 𝜟𝒕
𝑵𝒓 𝑵𝒉𝒕 E
= ∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝒈 ∗ 𝜟𝒙 ∗ 𝜟𝒕 + ∑ 𝒒𝒊 ∗ 𝒍𝒊 ∗ 𝜟𝒙 ∗ 𝜟𝒕 12.70
𝒊=𝟏 𝒊=𝟏
𝑴𝑪𝑶𝟐
+ 𝑹𝟏 ∗ ∗ 𝑪𝒑𝒃 ∗ 𝑻𝒃 ∗ 𝑨𝑏 ∗ 𝜟𝒙 ∗ 𝜟𝒕 + 𝑹𝒇𝒈 ∗ 𝑯𝒇𝒈
𝑴𝑪𝒂𝑪𝑶𝟑
∗ 𝑨𝒈 ∗ 𝜟𝒙 ∗ 𝜟𝒕 + 𝑹𝒆𝒏𝒕 ∗ 𝑪𝒑𝒔𝒆𝒄 ∗ 𝑻𝒔𝒆𝒄 ∗ 𝑨𝒈 ∗ 𝜟𝒙 ∗ 𝜟𝒕

The Δt’s cancel out and the limit of the Δx is taken to yield:
G Bed and Gas Phase Energy Balances 261

𝒅
(𝒖 ∗ 𝑨𝒈 ∗ 𝑪𝒑𝒈 ∗ 𝝆𝒈 ∗ 𝑻𝒈 )
𝒅𝒙 𝒈
𝑵𝒓 𝑵𝒉𝒕
𝑴𝑪𝑶𝟐 E
= ∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝒈 + ∑ 𝒒𝒊 ∗ 𝒍𝒊 + 𝑹𝟏 ∗
𝑴𝑪𝒂𝑪𝑶𝟑
𝒊=𝟏 𝒊=𝟏 12.71
∗ 𝑪𝒑𝒃 ∗ 𝑻𝒃 ∗ 𝑨𝑏 + 𝑹𝒇𝒈 ∗ 𝑯𝒇𝒈 ∗ 𝑨𝒈 + 𝑹𝒆𝒏𝒕 ∗ 𝑪𝒑𝒔𝒆𝒄
∗ 𝑻𝒔𝒆𝒄 ∗ 𝑨𝒈

Assuming a constant heat capacity, the derivative on the left-hand side of E 12.71
can be rewritten as:

𝒅 𝒅
(𝑨𝒈 ∗ 𝒗𝒈 ∗ 𝑪𝒑𝒈 ∗ 𝝆𝒈 ∗ 𝑻𝒈 ) = 𝑪𝒑𝒈 (𝑻 ∗ 𝒎̇𝒈 )
𝒅𝒙 𝒅𝒙 𝒈 E
𝒅 𝒅 12.72
= 𝒎̇𝒈 ∗ 𝑪𝒑𝒈 𝑻𝒈 + 𝑪𝒑𝒈 ∗ 𝑻𝒈 𝒎̇
𝒅𝒙 𝒅𝒙 𝒈

Thus, the expression for the gas temperature becomes:

𝑵𝒓 𝑵𝒉𝒕
𝒅 𝟏 𝑴𝑪𝑶𝟐
𝑻𝒈 = [∑ 𝑹𝒊 ∗ 𝜟𝑯𝒓,𝒊 ∗ 𝑨𝒈 + ∑ 𝒒𝒊 ∗ 𝒍𝒊 ∗ + 𝑹𝟏 ∗
𝒅𝒙 𝒎̇𝒈 ∗ 𝑪𝒑𝒈 𝑴𝑪𝒂𝑪𝑶𝟑
𝒊=𝟏 𝒊=𝟏
E
∗ 𝑪𝒑𝒃 ∗ 𝑻𝒃 ∗ 𝑨𝑏 + 𝑹𝒇𝒈 ∗ 𝑯𝒇𝒈 ∗ 𝑨𝒈 + 𝑹𝒆𝒏𝒕 ∗ 𝑪𝒑𝒔𝒆𝒄
12.73
𝒅
∗ 𝑻𝒔𝒆𝒄 ∗ 𝑨𝒈 − 𝑪𝒑𝒈 ∗ 𝑻𝒈 𝒎̇ ]
𝒅𝒙 𝒈
 262 12 Appendices

H. Model Parameters for Industrial Combustion Simulations

This appendix contains information of the parameter values used for the modeling
of MBM, SHW, and SRF co-firing.

Standard Kiln Parameters

Data concerning the kiln size and various other properties used in the industrial
scale model validation can be found in Table 12-9.

Table 12-9: Parameters for the kiln in industrial scale validation. Parameters marked
by * have been assumed.

Parameter Symbol Unit Value Ref.

Dimensions
Inner diameter of kiln dk m 3.9 [83]
Outer diameter of kiln dk,o m 4.4 [83]
Length of kiln Lk m 68 [83]
Pressure P Pa 101325 [83]
Radius of burner rburner m 0.22 [83]
Location of burner tip
relative to kiln clinker xburner m 5 [83]
exit
Thermal Parameters
Ambient air
Tamb K 293 *
temperature
Emissivity of kiln walls εw - 0.8 [31]
Heat conductivity of
λcoat W/(m*K) 2.7 [31]
clinker coating
Heat conductivity of
λw W/(m*K) 2.7 [31]
kiln walls
Kiln Residence Time
Rotational speed of kiln n 1/s 0.05 *
Kiln angle ψ ° 2.5 *
Bed angle of repose ξ ° 46 *
Residence time of
τ s 1838 Calc.
clinker
Coating
Distance to which kiln
coating has full Lcoat,1 m 15 *
thickness
Distance where kiln
Lcoat,2 m 20 *
coating has 0 thickness
Kiln coating thickness lcoat m 0.25 *

Data pertaining to the kiln residence time, such as kiln inclination and rotational
speed were not reported by Ariyaratne [83], thus appropriate values were assumed.
H Model Parameters for Industrial Combustion Simulations 263

A coating with a thickness of 0.25 m is assumed to be present in the first 1 meters of

the kiln. A linear decrease in thickness is assumed between 15 and 20 m, until the
coating thickness is 0 m.

Standard Gas Parameters

Parameters for the inlet gas flow conditions can be found in Table 12-10.

Table 12-10: Parameters for the primary and secondary air inlet for industrial scale
model validation.

Parameter Symbol Unit Value Ref.

Primary and
Secondary Air
Flowrate of primary air ṁg,prim kg/s 2.08 [83]
Primary air K
323 [83]
temperature Tg,prim,0
Heat capacity of J/(kg K)
1290 Calc.
primary air Cpg,prim
Flowrate of secondary kg/s
19.4-26.1 Calc.
air ṁg,sec
Standard temperature K
1023 [83]
for secondary air Tg,sec,0
Heat capacity of J/(kg K)
1154 Calc.
secondary air Cpg,sec
Oxygen gas -
concentration in flue 0.035 [84]
gas yO2
Burner Parameters
Burner momentum I N/MW 5.8 [83]
Burner swirl number S - 0.261 [83]
Entrainment rate Kent,jet - 0.195 Calc.
Constant for ε/k 1/s
101.6 Calc.
calculation Kedr,1
Constant for ε/k -
-1.50 Calc.
calculation Kedr,2

The parameters are primarily based on data as reported by Ariyaratne et al. [83].
Based on the reported axial momentum and swirl number, the entrainment constant
rate and constants for the eddy dissipation rate can be calculated. The oxygen
concentration at the kiln inlet is attempted maintained at 3.5 %, and the secondary
airflow is changed to match this, based on the oxygen requirements of the fuel (see
appendix D). The secondary air flows and oxygen concentrations for the MBM, SHW,
and SRF simulation cases can be found in Appendices H.5-H.7.
264 12 Appendices

The heat capacities of the two gas streams are assumed constant. The value for
secondary air is calculated for atmospheric air at the inlet temperature 1023 K, and
the heat capacity for the primary air is calculated at 1200 K.

Standard Parameters for Fuel

The standard values for various fuel properties can be found in Table 12-11. MBM
and SHW are assumed to have the same properties as the biomass, except for the
density, which is changed to 1354 kg/m3 for MBM and 620 kg/m3 for SHW.

Table 12-11: Standard parameters for fuels used in the simulations.

Plastic
Coal Biomass (solid/
liquid)
Parameter Unit Value Ref. Value Ref. Value Ref.
Kinetics
Adry 1/s 5.13*106 [31] 5.13*1010 [236] 5.13*1010 [236]
Eadry kJ/mol 87.9 [31] 88 [236] 88 [236]
ΔHevap kJ/kg 2257 [260] 2257 [260] 2257 [260]
Adevol 1/s 9.59*104 [31] 1.11*1011 [255] 4.8*1022 [232]
Eadevol kJ/mol 82.6 [31] 177 [255] 3.49*105 [232]
ΔHdevol kJ/kg 0 0 365 [232]
kg/(m2*P
Achar,ox 1*10-2 [31] 2.54*10-3 [256] -
a*s*)
Eachar,ox kJ/mol 105 [31] 81.6 [257] -
Achar,gsc m/s - 3.42*Tchar [258] -
Eachar,gsc kJ/mol - 130 [258] -
Amelt 1/s - - 2.17*1044 [232]
Eamelt kJ/mol - - 353 [232]
ΔHmelt kJ/kg - - 207 [232]
Others
λf W/(m*K) - 0.12 [255] -
Depends
εf - 0.9 [31] 0.9 [31] on [232]
particle
ρf kg/m3 1336 [31] 600 [86] 950 [232]
1900/22
Cp J/(Kg K) 1400 [31] [232]
00
Cp1 J/(Kg K) - 2300 [259] -
Cp2 J/(Kg K) - 1000 [259] -

For biomass, MBM, and SHW the fuel heat capacity is calculated based on the fuel
ash content:
H Model Parameters for Industrial Combustion Simulations 265

E
𝑪𝒑𝒇 = (𝟏 − 𝒘𝒂𝒔𝒉 )𝑪𝒑𝟏 + 𝒘𝒂𝒔𝒉 𝑪𝒑𝟐
12.74

Standard Parameters for Clinker

The standard parameters for the clinker bed used in the kiln model can be found in
Table 12-12.

Table 12-12: Standard parameters for the clinker/kiln bed. Parameters marked by *
have been assumed.

Parameter Symbol Unit Value Ref.

Heats of reactions
Heat of reaction for
ΔHR1 kJ/(kg CaCO3) 1780 [217]
limestone calcination
Heat of reaction for
ΔHR2 kJ/(kg C2S) -732 [217]
alite formation
Heat of reaction for
ΔHR3 kJ/(kg C3S) 59.0 [217]
belite formation
Heat of reaction for
ΔHR4 kJ/(kg C3A) -33.5 [217]
aluminate formation
Heat of reaction for
ΔHR5 kJ/(kg C4AF) -38.8 [217]
ferrite formation
Melting Parameters
Temperature where
Tmelt,1 K 1553 [217]
clinker starts to melt
Temperature where
Tmelt,2 K 1723 [217]
clinker is fully molten
kJ/
Heat of clinker melt ΔHmelt 600 [217]
(kg clinker)
maximum fraction of
Xmelt,max - 0.2 *
clinker melt
Physical Parameters
Heat capacity of
Cpb J/(kg*K) 1500 [201]
clinker bed
Emissivity of clinker
εb - 0.8 [183]
bed
Thermal conductivity
λb W/(m*K) 0.5 [198]
of clinker bed
Bulk density of clinker
ρb kg/m3 1200 [201]
bed
Calcination degree at
- - 0.965 [82]
kiln raw material inlet
266 12 Appendices

Parameters for MBM Simulations

Additional parameters used in the MBM co-firing simulations are shown in Table
12-13. These show the calculated secondary air flows and temperatures, and the
oxygen mole fraction at the kiln clinker inlet. The secondary air temperatures are
calculated based on the simplified cooler model described in Appendix J.

Table 12-13: Flows of fuel and secondary air with secondary air temperature and
oxygen concentration (mol%) in flue gas for the MBM simulations.

yO2 kiln
Coal MBM Secondary Secondary Air
Case clinker
Flow Flow Air Flow Temperature
inlet
kg/s kg/s kg/s °C %
1 2.2 0.0 23.1 750 3.5
2 1.8 0.6 24.0 717 3.4
3 1.5 1.1 25.1 680 3.3
4 1.3 1.4 25.2 671 3.2
5 1.1 1.7 25.8 654 3.2
6 0.9 1.9 26.1 642 3.1
7 0.9 1.9 26.1 642 3.1

The applied particle sizes are shown in and Table 12-14.

Table 12-14: Particle diameter of coal and MBM particles used in simulations.

Particle
Coal MBM
Group
µm µm
1 2 38
2 5 118
3 10 208
4 16 311
5 23 431
6 32 576
7 44 756
8 61 999
9 87 1,375
10 172 2,478
H Model Parameters for Industrial Combustion Simulations 267

Parameters for SHW Simulations

Additional parameters used in the SHW co-firing simulations are shown in Table
12-15.

Table 12-15: Flows of fuel and secondary air with secondary air temperature and
oxygen concentration (mol%) in flue gas for the SHW simulations.

yO2 kiln
Coal SHW Secondary Secondary Air
Case clinker
Flow Flow Air Flow Temperature
inlet
kg/s kg/s kg/s °C %
1 2.1 0.0 21.8 750 3.5
2 1.9 0.1 20.8 763 3.5
3 1.9 0.3 20.7 763 3.5
4 1.8 0.4 20.7 762 3.5
5 1.6 0.6 19.4 786 3.5
6 1.6 0.7 19.9 774 3.5
7 1.6 0.7 20.9 759 3.5
8 1.6 0.7 20.9 756 3.5
9 1.6 0.8 21.1 749 3.5
10 2.1 0.0 21.1 762 3.5

The applied particle sizes are shown in and Table 12-16.

Table 12-16: Particle diameter of coal and SHW particles used in simulations.

Particle
Coal SHW
Group
µm µm
1 2 327
2 5 848
3 10 1,218
4 16 1,567
5 23 1,921
6 31 2,299
7 43 2,725
8 58 3,239
9 83 3,945
10 160 5,621
268 12 Appendices

Parameters for SRF Simulations

Additional parameters used in the SRF co-firing simulations are shown in Table
12-17.

Table 12-17: Flows of fuel and secondary air with secondary air temperature and
oxygen concentration (mol%) in flue gas for the SRF simulations.

yO2 kiln
Coal SRF Biomass Plastic Secondary Secondary Air
Case clinker
Flow Flow Flow Flow Air Flow Temperature
inlet
kg/s kg/s kg/s kg/s kg/s °C %
1 2.2 0.0 0.0 0.0 23.1 750 3.5
2 2.0 0.4 0.2 0.1 23.4 738 3.5
3 1.7 0.8 0.5 0.2 23.6 723 3.5
4 1.5 1.2 0.7 0.2 23.8 708 3.4
5 1.3 1.6 0.9 0.3 24.0 697 3.4
6 1.1 2.0 1.2 0.4 24.2 687 3.4
7 0.9 2.4 1.4 0.5 24.4 678 3.4
8 0.7 2.8 1.6 0.5 24.6 671 3.4
9 0.4 3.2 1.9 0.6 24.8 665 3.3
10 0.2 3.5 2.1 0.7 25.0 659 3.3
11 0.0 3.9 2.4 0.8 25.3 655 3.3

The applied particle sizes are shown in and Table 12-18.

Table 12-18: Particle diameter of coal and SRF particles used in simulations.

Particle
Coal Biomass Plastic
Group
µm µm µm
1 2 1.0E+03 1.0E+03
2 5 2.8E+03 2.4E+03
3 10 3.9E+03 3.5E+03
4 16 4.3E+03 4.1E+03
5 23 6.6E+03 6.8E+03
6 32 1.3E+04 9.7E+03
7 44 - -
8 61 - -
9 87 - -
10 172 - -
 I Details of Combustion Simulations 269

I. Details of Combustion Simulations

In chapter 9, the free lime content has been shown as the result of simulations. In
addition, select temperature profiles of the gas and bed have been shown, to further
explain the differences observed in the various simulation cases. However, many
additional parameters are calculated by the model. This appendix contains a more
in-depth discussion of select parameters calculated by the model. As an example,
further discussions are made for the coal case simulations from the MBM and SRF
test scenario and the 30 % SRF energy input co-firing scenario.

Details of Coal Combustion Simulation

Figure 12-11 shows the calculated temperatures through the kiln for the coal
simulations case of Test 1-1. The kiln burner is protruding 5 meters into the kiln. As
soon as the air and fuel leave the burner it is quickly heated by radiation and
entrainment of the secondary air, which initiates the devolatilization of the coal. The
combustion of volatiles increases the temperature to around 2200 °C.

Figure 12-11: Temperatures through the kiln of the wall, gas, bed (without clinker
reactions as dashed line), fuel average temperature, secondary air, and outer shell.

After the initial peak, the temperature decreases as remaining secondary air, which
is now colder than the flame, is entrained into the flame. The entrainment is
complete at around 11 m, where the temperature begins to increase again. A second
270 12 Appendices

peak is seen as the coal char is combusted at around 15 m from the kiln burner end.
Afterwards, the temperature decreases as heat is transferred to the clinker bed and
lost to the surroundings. The gas and average fuel temperatures are very similar,
due to the small size of the coal particles.

The secondary air temperature increases slightly during the first 12 meters of the
kiln. The secondary air is assumed to be radiatively clear, as it only consists of
oxygen and nitrogen. Thus, it is assumed to only be heated by convection which is a
slower mode of heat transfer than radiation. At 12 meters, all the secondary air is
entrained. The temperature change is no longer tracked, and the temperature is thus
shown as constant in Figure 12-11.

The pre-calcined raw meal is admitted from the other end of the kiln. The
temperature is initially only increasing slowly. 30 meters from the burner end the
bed temperature increases rapidly, due to the exothermic formation of belite (C2S).
The temperature continues to increase reaching a maximum around 1600 °C at 7-8
m from the clinker outlet. The bed temperature then decreases as heat is transferred
to the secondary air, to the walls and to the burner tip. For the sake of comparison,
the dashed green line shows the bed temperature without clinker reactions. It is
seen that the exothermic formation of belite has a high impact on the temperature
in the bed.

The wall temperature is seen to initially change very drastically. Since the wall
temperature is only determined based on a heat balance based on the heat transfer
to the wall and heat transfer through the wall, it has no thermal inertia.
Consequently, the temperature can change drastically in the axial direction. For the
5 initial meters of the kiln, the burner tip is exposed. The burner is assumed to have
the same temperature as the secondary air, and an emissivity similar to that of the
kiln walls (0.9). At this point the temperature of the wall is between that of the bed
and secondary air. After the 5 m, the wall temperature quickly approaches that of
the bed. The primary air initially has an emissivity and absorptivity of 0, since no
water or CO2 is present. Thus, the heat transfer between the wall and bed will
dominate the wall temperature, and it obtains a temperature close to that of the bed.
As water and CO2 is released during the fuel devolatilization and combustion,
radiation between the wall and primary air jet becomes more influencing, and the
wall temperature ends up in between that of the bed and the gas.
I Details of Combustion Simulations 271

Lastly, Figure 12-11 also shows the temperature of the outer kiln shell. It is seen to
generally follow the trend of the kiln wall temperature. Between 15 and 20 m the
shell temperature increases rapidly. A kiln coating of 25 cm thickness is assumed to
be present in the initial 15 m of the kiln, which reduces the heat loss and shell
temperature. At 15 m, the kiln coating thickness begins to decrease reaching a
thickness of 0 m at 20 m. As the coating thickness decreases, the heat transfer
through the wall increases, and the shell temperature is increased.

Figure 12-12 shows the different clinker phases through the kiln. The position is
made to be consistent with Figure 12-11, thus the figure has the raw meal inlet on
the right-hand side, and the clinker travels left through the kiln.

Figure 12-12: The mass fraction of the different clinker phases through the kiln.

The pre-calcined raw meal enters the kiln at a temperature of 830 °C. Some residual
CaCO3 is left. In the kiln the limestone slowly decomposes, and is fully converted at
30 m, where the temperature of the bed is around 1100 °C. The formation of C3A and
C4AF starts almost immediately, but the formation rate increases between 30-40 m,
where the temperature is between 1000 and 1100 °C. This consumes all the Fe 2O3
and Al2O3 and some of the CaO.

C2S is rapidly formed from the 30 m mark, where the temperature is around 1100
°C. The SiO2 is consumed, and the CaO content also decreases rapidly. The formation
of C2S is exothermic, and the reaction quickly increases the bed temperature. At
272 12 Appendices

around 20 m from the clinker outlet, the concentration of C2S reaches its maximum,
as C3S begins to form. The formation of C3S occurs rapidly from 20 m, where the
temperature is above 1400 °C and plenty of CaO is available. In the last 5 m of the
kiln, the formation rate decreases as the most of the CaO has been used, and the
temperature is lowered.

The concentration of the gasses N2, O2, CO2, H2O, CH4, and CO through the kiln are
shown in Figure 12-13. The oxygen concentration is seen to decrease sharply at
around 7 m, where the coal devolatilization and volatile combustion occurs. The
oxygen concentration then slightly increases from 8 to 12 m, as the secondary air is
entrained into the flame. From 12 m, the concentration slowly decreases as char
oxidation of the larger coal particles take place.

Figure 12-13: The mole and mass fraction of N2, O2, CO2, H2O, CH4 and CO through the
kiln for a coal fired simulation.

The CO and CH4 concentration are seen to have a small spike at around 8 m due to
the rapid devolatilization. However, the gasses are quickly combusted, and the
concentration decrease to 0. H2O is seen to be mainly formed at around 8 m, due to
the combustion of CH4. CO2 is also increased rapidly as CO and CH4 is oxidized. The
CO2 concentration keeps increasing after 12 m due to oxidation of the remaining
char, but also caused by the decomposition of CaCO3 in the bed.
I Details of Combustion Simulations 273

Figure 12-14 shows the temperatures and conversion of each of the 10 discrete
particle size classes. The smallest particles have a diameter of 1.6 µm and the largest
170 µm. All particles, except the largest, are heated very quickly and almost follow
the gas temperature.

The drying and the conversion of the particles occur rapidly within 10 m, due to the
fast heating rate of the small particles. The size does not have a large impact on this.
The char oxidation is seen to occur rapidly for the 5 smallest particle groups, where
they are fully converted within 10 m. The largest particle group takes much longer
to be converted, and full conversion is not reached in the kiln.

Figure 12-14: Temperatures (a), extent of drying (b), extent of devolatilization (c),
and extent of char oxidation (d) for the 10 discrete particle size classes of coal
particles.

Details of SRF Co-Firing Simulation

In this chapter additional details of the SRF co-firing scenario, with 30 % SRF by
energy, are discussed. The trends of the graphs are similar to those explained in
Appendix I.1, and only the major differences will be described.
274 12 Appendices

Figure 12-15 shows the different temperatures through the kiln. It is noticed how
the peak temperature is decreased compared to the coal case. Instead the peak gas
temperature occurs at around 25 m, which is after the largest particles of SRF have
been converted.

Figure 12-15: Temperatures through the kiln of the wall, gas, bed fuel average
temperature, secondary air, and outer shell.

The average fuel temperature in Figure 12-15 is also significantly different. It only
approaches the gas temperature after 25 m, where only the ash particles are
remaining, which results in a rapid heating. Initially, an increase is seen in the
average fuel temperature followed by a decrease. The temperature is a mass-based
average. Initially, the coal particles are quickly heated, which increases the average
temperature. However, as the coal particles lose mass during the combustion, the
average temperature becomes dominated by the SRF particles, which are only
heated slowly.

The clinker phase composition is shown in Figure 12-16. Compared to the coal case,
belite is formed earlier, from around the 35 m mark. This is caused by a more rapid
heating of the bed, due to higher gas temperatures at the clinker inlet.
I Details of Combustion Simulations 275

Figure 12-16: The mass fraction of the different clinker phases through the kiln.

Figure 12-17 shows the various gas concentrations accounted for in the kiln model.
The main difference to the coal case is a slower consumption of oxygen due to the
slower combustion rate of the alternative fuels. Furthermore, the water
concentration is higher at 7.5 mol% compared to 4.0 mol% in the coal case. This is
caused by the high water content in the SRF of 20 wt% compared to 1 wt% in coal.

Figure 12-17: The mole and mass fraction of N2, O2, CO2, H2O, CH4 and CO through the
kiln for a coal fired simulation.
276 12 Appendices

The fuel temperatures and conversion of the coal particles, biomass particles, and
plastic particles are shown in Figure 12-18, Figure 12-19, and Figure 12-20. The
temperatures of the coal particles follow the gas temperatures. The drying and
devolatilization occurs rapidly, and char oxidation is only slow for the larger coal
particles.

Figure 12-18: Temperatures (a), extent of drying (b), extent of devolatilization (c),
and extent of char oxidation (d) for the 10 discrete particle size classes of coal
particles.

The temperatures and conversion of the biomass particles in the SRF are shown in
Figure 12-19. The markers in the graphs indicate the location where the fuel
particles drop into the bed. The biomass particles are initially heated slowly, due to
their high water content. It is only when the water has been evaporated that the
temperatures increase more rapidly, which leads to devolatilization of the two
smallest particle size groups. It is only the two smallest particle groups that undergo
devolatilization, while they are in suspension, and only the smallest group, where
the char is also oxidized to some extent. The inflection point on the char conversion
graph, indicates when the particles enter the bed. These particles groups
I Details of Combustion Simulations 277

correspond to the fines and the particles with a terminal velocity lower than 2 m/s,
as determined in the wind sieve. The larger particles are converted, as they enter
the clinker bed.

Figure 12-19: Temperatures (a), extent of drying (b), extent of devolatilization (c),
and extent of char oxidation (d) for the 6 discrete particle size classes of biomass
particles in SRF. ‘x’-markers indicate the distance of fuel landing in the clinker bed.

The temperatures and conversion of the plastic particles of the SRF are shown in
Figure 12-20. The curves are somewhat similar to those seen for the biomass
temperatures and conversion. The drying of the plastic particles take slightly longer,
as they undergo melting while they are drying, which also requires energy. Only the
smallest particle group is fully devolatilized while in suspension. These correspond
to the fines (<2 mm) separated from the wind sieve. The other particle groups are
devolatilized as they enter the clinker bed, as indicated by the location of the
markers.
278 12 Appendices

Figure 12-20: Temperatures (a), extent of drying (b), extent of devolatilization (c),
and extent of melting (d) for the 6 discrete particle size classes of plastic particles in
SRF. ‘x’-markers indicate the distance of fuel landing in the clinker bed.
 J Model for the Clinker Cooler 279

J. Model for the Clinker Cooler

The temperature and flow of the clinker that leaves the kiln has a direct impact on
the heat recuperation in the clinker cooler, and the secondary air temperature. The
secondary air temperature is used as an inlet condition to the model. Thus, one of
the outlet conditions (the bed temperature) is coupled to one of the inlet conditions
(the secondary gas temperature). To describe the impact, a simple model for the
clinker cooler is developed.

Modern clinker coolers are grate coolers, see Figure 12-21, which function as a form
of cross-flow cooler. The clinker is moved along the cooler by reciprocating grates.
Meanwhile cooling air is forced through the bed, cooling the clinker and heating the
air, which is used as secondary air in the kiln [33,261].

Figure 12-21: A grate cooler in a cement plant. The clinker is moved forward by
reciprocating grates. Cooling air is forced through the clinker bed by air fans. The
heated air is used as secondary air in the cement kiln [33].

Modeling of the clinker cooler is discussed by e.g. Ahmad et al. [261], Mujumdar et
al. [262], and Touil et al. [263], but a simpler approach is pursued here.

A simplified model of the cooler can be obtained by dividing the cooler into a
number, Ncooler, of segments as shown in Figure 12-22. In each segment the clinker
moves in the axial x-direction, while gas moves in the vertical y-direction. The
280 12 Appendices

clinker enters the cooler at a temperature of Tb,in and is cooled in each segment by
the gas flow, and exits the cooler at a temperature of Tb,out.

Figure 12-22: Sketch of a simplified cooler model divided into Ncooler segments.

As a simplification it will be assumed that the clinker temperature is constant in each

segment, while only the gas temperature increases in the y-direction. Thus, the
following temperature equation can be formulated for the gas temperature in a
given segment (assuming constant heat capacity and flow).

𝒅𝑻𝒈,𝒊 (𝒚) 𝟏 E
= ∗ 𝒉𝒄𝒐𝒏𝒗 ∗ (𝑻𝒃,𝒊 − 𝑻𝒈,𝒊 (𝒚)) ∗ 𝒍𝒈𝒃
𝒅𝒚 𝑭𝒈,𝒊 ∗ 𝑪𝒑𝒈 12.75

Fg,i is the gas mass flow in the i’th segment, Cpg the gas heat capacity, hconv the
convective heat transfer coefficient, and lgb is the contact area between gas and bed
per unit height of the bed [m2/m].

Since neither the value of lgb or hconv is known, they are combined into a single
constant Hcooler [W/m/K], which is the cooler heat transfer coefficient.

𝒅𝑻𝒈,𝒊 (𝒚) 𝟏 E
= ∗ 𝑯𝒄𝒐𝒐𝒍𝒆𝒓 ∗ (𝑻𝒃,𝒊 − 𝑻𝒈,𝒊 (𝒚))
𝒅𝒚 𝑭𝒈,𝒊 ∗ 𝑪𝒑𝒈 12.76

If the segments are small enough the bed temperature in each segment can be
assumed constant. The bed temperature in one segment will be assumed equal to
the outlet temperature of the previous segment. The temperature difference
between the inlet and outlet of a section can be found from a heat balance
considering the heat transfer in the segment.

𝑸𝒉𝒕,𝒊−𝟏 E
𝑻𝒃,𝒊 = 𝑻𝒃,𝒐𝒖𝒕,𝒊−𝟏 = 𝑻𝒃,𝒊−𝟏 −
𝑭𝒃 ∗ 𝑪𝒑𝒃 12.77

The heat transfer in the segment can be calculated based on the gas flow and
temperature of the segment:
J Model for the Clinker Cooler 281

E
𝑸𝒉𝒕,𝒊 = (𝑻𝒈,𝒐𝒖𝒕,𝒊 − 𝑻𝒈,𝒊𝒏 ) ∗ 𝑭𝒈,𝒊 ∗ 𝑪𝒑𝒈
12.78

The differential equation in E 12.76 can be solved directly to yield the gas
temperature through the y-direction of the segment:

−𝑯𝒄𝒐𝒐𝒍𝒆𝒓 E
𝑻𝒈,𝒊 (𝒚) = 𝑻𝒃,𝒊 − 𝒆𝒙𝒑 ( ∗ 𝒚) (𝑻𝒃,𝒊 − 𝑻𝒈,𝒊𝒏 )
𝑭𝒈,𝒊 ∗ 𝑪𝒑𝒈 12.79

It is assumed that the clinker bed height is constant through the cooler with a value
of 0.8 m [264]. If this value is inserted into E 12.79, then it is possible to determine
the gas outlet temperature of the i’th segment.

Determining the Cooler Heat Transfer Coefficient

The main form of heat transfer in the cooler is convection when the gas is forced
through the bed. The convective heat transfer coefficient is a function of Reynolds
and Prandtl number [262,263]. However, here the cooler heat transfer coefficient,
Hcooler, will be assumed constant in all segments of the cooler. Furthermore, it is only
desired to model the first part of the cooler, in order to determine the secondary air
temperature.

To determine the value of the heat transfer coefficient, a base case is needed to
calibrate the model. For this is it is needed to know corresponding values of the inlet
and outlet temperatures. The coal fired model validation case Test 1-1 (see chapter
9.2) is used as the calibration case. The conditions for this case are given in Table
12-19. For this case the bed inlet temperature to the cooler is calculated in the kiln
model to 1506 °C. The inlet temperature of the gas is assumed to be 25°, and the
secondary air temperature is given as 750 °C. The outlet temperature of the bed can
be calculated using an overall heat balance via equations E 12.77 and E 12.78,
assuming 1 stage in the cooler.

Table 12-19: Cooler conditions for the calibration calculations based on coal
validation case Test 1-1.

Inlet Outlet Heat

Mass Flow
Temperature Temperature Capacity
kg/s °C °C J/kg/K
Bed 39.1 1506 1176 1500
Gas 23.2 25 750 1154
282 12 Appendices

In the following an example using 5 segments of the clinker cooler is shown. The
secondary air temperature is equal to the average of outlet temperatures from each
segment. Thus, the optimal value of Hcooler can be found by a trial and error method
to minimize the value of the equation:
𝑵𝒄𝒐𝒐𝒍𝒆𝒓
𝟏 E
𝒂𝒃𝒔 (𝑻𝒔𝒆𝒄 − ∗ ∑ 𝑻𝒈,𝒐𝒖𝒕,𝒊 )
𝑵𝒄𝒐𝒐𝒍𝒆𝒓 12.80
𝒊

Figure 12-23 shows the value of E 12.80 as a function of Hcooler and it is determined
that a value of around 5.2*103 results in the correct temperature of the gas leaving
the cooler.

Figure 12-23: Absolute difference between the secondary air temperature and the
average air temperature out of the cooler as function of Hcooler.

Using the optimal value of clinker cooler heat transfer coefficient, Hcooler, the
temperature through each segment of the cooler can be calculated. Figure 12-24
shows the gas and bed temperatures through each segment of a clinker cooler
divided into 5 segments. The gas temperature is seen to increase through each
segment, while the bed temperature in each segment is fixed. The average outlet
temperature of the five gas streams is 750 °C.
J Model for the Clinker Cooler 283

Figure 12-24: The gas temperatures (solid lines) through the 5 bed segments, with
bed temperatures (dashed lines) and the average gas outlet temperature (dotted
line).

Using the calibrated value of the heat transfer coefficient, it is possible to calculate
different values of the secondary air temperature if there are changes in the
temperature of the clinker from the kiln, the flow of the clinker, or the flow of
secondary air.

In this example the clinker cooler was divided into 5 segments. In the actual
simulations, 1000 segments are used. With this number of segments, the
temperature of the clinker decreases by less than 1 °C in each segment, and the
assumption of constant temperature is acceptable.
Department of Chemical and Biochemical Engineering - CHEC
Technical University of Denmark
Søltofts Plads, Building 229
2800 Kgs. Lyngby
Denmark
Phone: +45 45 25 28 00
Web: www.kt.dtu.dk