Aplicación Del Ciclo de Calcio Para Mitigar Emisiones de Co2
Aplicación Del Ciclo de Calcio Para Mitigar Emisiones de Co2
Aplicación Del Ciclo de Calcio Para Mitigar Emisiones de Co2
Charles Dean
A thesis submitted for the degree of Doctor of Philosophy and for the
an inherent synergy with cement production in that a by-product is predominantly CaO, which forms
nearly 70 % of the material feed to a cement works. However, integration of the two processes will
inevitably lead to a decrease in the thermal efficiency of producing 1 kg of clinker due to the energy
penalty of the Ca-loop calciner. Furthermore, experimental studies are required to determine whether
any change in the trace element inventory of the sorbent takes place due to cycling between
calcination and carbonation conditions, and whether such changes impact on the phase composition or
For the flow-sheeting part of this investigation, the thermal efficiency of producing 1 kg clinker when
integrated with Ca-looping was found to increase by between 0.15 and 1.49 MJ depending on the
configuration. For the experimental part of this investigation, it was found that cycling in the absence
of fuel did not lead to a change in trace element content of the sorbent. However, in the presence of
fuel combustion, the trace element inventory of the sorbent was seen to increase. Al increased for all
three fuels indicating that ash from the fuel was mixing with the sorbent during the calcination step.
Cycling in the presence of Lea Hall coal led to an increase in B and Na, whilst cycling in the presence
of refuse-derived-fuel led to an increase in Cu, Na and Ti. The production of cement from cycled
sorbent indicated that cycling in the absence of fuel did not lead to a change in phase composition.
This was also the case for cements produced from sorbent cycled in the presence of continuously fed
La Jagua and Lea Hall coals. However, a slight increase in the % wt. alite was observed for cements
produced from sorbent exposed to batch-fed La Jagua. Lastly, after undergoing performance tests it
was found that there was a general downward trend in the strength of the cement mortars produced
with increasing cycle number. Setting times were within the range set out in the British and American
Standards.
2
Acknowledgements
I would like to express my sincere thanks to my supervisors, Dr. Paul Fennell and Prof. Denis
Dugwell for giving me the opportunity to come to Imperial and study towards this PhD; also for their
guidance and support throughout this research. Thankyou also to Dr. Nigel Paterson who provided
I gratefully acknowledge the financial support for this work provided by the EPSRC and also by
Cemex Operations U.K. Ltd. Thankyou also to Dr. Mike Andrews at Cemex Rugby who provided
guidance, useful literature and many of the materials used in this project.
Thanks also to all other members of the Energy Engineering group who have made it a great place to
3
Contents
Abstract.............................................................................................................................................. 2
Acknowledgements ........................................................................................................................... 3
Contents ............................................................................................................................................. 4
List of Figures.................................................................................................................................... 9
List of Tables ................................................................................................................................... 13
Introduction ..................................................................................................................................... 16
Chapter 1. Background ............................................................................................................. 18
1.1. The Calcium Looping Cycle .............................................................................................. 18
1.1.1. Basics of the Cycle ............................................................................................................ 18
1.1.1.1. Sorbent deactivation ........................................................................................................ 21
1.1.1.2. Sorbent Performance ....................................................................................................... 25
1.1.1.3. Economics of the Ca-looping Cycle .................................................................................. 28
1.2. Cement Production .......................................................................................................... 29
1.2.1. Ordinary Portland Cement (OPC) ..................................................................................... 29
1.2.2. Cement Production .......................................................................................................... 30
1.2.2.1. Raw Materials ................................................................................................................... 30
1.2.2.2. Raw Meal Pre-treatment .................................................................................................. 32
1.2.2.3. Pre-heating ....................................................................................................................... 33
1.2.2.4. Calcination ........................................................................................................................ 33
1.2.2.5. Clinkerization .................................................................................................................... 34
1.2.2.6. Clinker cooling .................................................................................................................. 35
1.2.2.7. Cement Blending and Grinding ........................................................................................ 36
1.2.3. Cement Chemistry ............................................................................................................ 36
1.2.3.1. Clinker Phases................................................................................................................... 36
Alite (C3S) .......................................................................................................................................... 37
Belite (C2S) ........................................................................................................................................ 38
Tricalcium Aluminate (C3A) ............................................................................................................... 39
Calcium Aluminoferrite (C4AF) .......................................................................................................... 39
1.2.3.2. Clinker Phase related Equations ....................................................................................... 39
1.3. Integration of Ca-looping and Cement Production ........................................................ 40
Chapter 2. Flowsheeting ............................................................................................................ 48
4
2.1. Chapter Summary ............................................................................................................ 48
2.2. The Calcium Looping Process .......................................................................................... 48
2.2.1. Determining and ECO2 for a steady-state system ......................................................... 49
2.2.2. Determining the energy requirements at the calciner (ηth) ............................................. 52
2.2.3. Incorporating Carbonation Kinetics and Bed Inventory (WCaO) ........................................ 55
2.2.4. Examining system start-up ............................................................................................... 60
2.3. Cement Plant Flowsheeting............................................................................................. 62
2.1.1. Cement Plant Mass and Energy Balance .......................................................................... 62
2.1.1.1. Calculation of gas streams through the system ............................................................... 63
2.1.1.1. Calculation of solid streams through the system ............................................................. 65
2.1.1.2. Calculation of heat demands ............................................................................................ 65
2.1.1.3. Calculation for kiln hood air infiltration (‘false’ air) ......................................................... 67
2.4. Integration of Cement Plant and Calcium Looping Plant Flow-sheets .......................... 67
2.4.1. Configuration 1 ................................................................................................................. 69
2.4.2. Configuration 2 ................................................................................................................. 70
2.4.3. Configuration 3 ................................................................................................................. 72
2.4.4. Summary of Cement Plant and Ca-looping Integration ................................................... 73
2.5. Conclusions ...................................................................................................................... 74
2.6. Further Work.................................................................................................................... 75
Chapter 3. Production of spent sorbent ................................................................................... 76
3.1. Chapter Summary ............................................................................................................ 76
3.2. Vertical Tube Furnace Design and Operation ................................................................. 76
3.2.1. Basic Furnace Design for ‘No Fuel’ Experiments .............................................................. 76
3.2.2. Calibration of FTIR for CO2 ............................................................................................... 78
3.2.3. Modification of rig design for coal experiments .............................................................. 79
3.2.4. Modifications for RDF: feeder design ............................................................................... 82
3.3. Design of Experiments ..................................................................................................... 83
3.3.1. Determination of minimum fluidising velocity................................................................. 83
3.3.2. Determining coal particle size .......................................................................................... 85
3.4. Description of Cycling experiments ................................................................................ 85
3.4.1. ‘No Fuel’ Experiments ...................................................................................................... 85
3.4.2. Coal Experiments .............................................................................................................. 87
3.4.3. RDF Experiments .............................................................................................................. 88
5
3.5. Mass balance on CO2 from a single calcination (No Fuel) .............................................. 88
3.6. Conclusions ...................................................................................................................... 89
3.7. Further Work.................................................................................................................... 89
Chapter 4. Trace Element Analysis of Fuels, Limestone and Sorbent .................................. 90
4.1. Chapter Summary ............................................................................................................ 90
4.2. Description of Relevant Analytical Techniques .............................................................. 90
4.2.1. ICP-AES and ICP-MS .......................................................................................................... 91
4.2.2. IC ....................................................................................................................................... 93
4.2.3. Choice of Analytical Methods........................................................................................... 93
4.2.4. Important parameters for trace element analysis ........................................................... 95
4.3. Experimental Section ....................................................................................................... 96
4.3.1. Preparation of solid samples for ICP-AES by wet acid digestion ...................................... 96
4.3.2. ICP-AES Operation ............................................................................................................ 97
4.3.3. Converting ‘counts’ to sample concentrations ................................................................ 98
4.3.4. Determination of Spectral Wavelengths .......................................................................... 99
4.3.5. Checking Repeatability of Results .................................................................................. 107
4.3.6. Efficacy of the Digestion Method for materials of interest ............................................ 108
4.3.7. Effect of a single calcination on trace elements............................................................. 109
4.3.8. Observing S using IC ....................................................................................................... 111
4.3.9. Concentration of Elements in the Input Materials ......................................................... 112
4.3.10. Elements to be included in the analysis of sorbent ....................................................... 113
4.4. Results & Discussion: Element Inventories in Cycled Sorbent ..................................... 114
4.4.1. ‘No Fuel’ Experiments .................................................................................................... 114
4.4.2. Batch-fed Experiments ................................................................................................... 116
4.4.3. Continuously-fed Experiments ....................................................................................... 117
4.4.4. Mass Balances on Elements ........................................................................................... 120
4.5. Conclusions .................................................................................................................... 125
4.6. Further Work.................................................................................................................. 125
Chapter 5. Investigation into the effect of using repeatedly cycled sorbent on phases in
cement 124
5.1. Chapter Summary .......................................................................................................... 126
5.2. Relevant Analytical Techniques .................................................................................... 126
5.2.1. Fundamentals of X-Ray Diffraction (XRD) ...................................................................... 126
5.2.2. Quantitative X-Ray Diffraction (QXRD) ........................................................................... 128
6
5.2.2.1. The Internal Standard Technique ................................................................................... 128
5.2.2.2. The Reference Intensity Ratio Technique ...................................................................... 129
5.2.3. Fundamentals of X-Ray Fluorescence (XRF) ................................................................... 130
5.3. Experimental Section ..................................................................................................... 130
5.3.1. Developing a method for producing laboratory OPC ..................................................... 130
5.3.2. Identifying alite and alite polymorphs using XRD .......................................................... 132
5.3.3. Developing a method for measuring % alite in the clinkers........................................... 134
5.3.4. Quantification of clinker phases proportions using XRF ................................................ 136
5.4. Results and Discussion: % wt. of alite in clinkers produced from cycled sorbent ....... 136
5.4.1. No Fuel / 2 g La Jagua per cycle...................................................................................... 136
5.4.2. Continuously-fed coal and RDF ...................................................................................... 137
5.5. Conclusions .................................................................................................................... 138
5.6. Further Work.................................................................................................................. 139
Chapter 6. Performance Tests................................................................................................. 138
6.1. Chapter Summary .......................................................................................................... 138
6.2. Theory behind the tests................................................................................................. 138
6.2.1. Cement Consistence ....................................................................................................... 138
6.2.2. Cement Setting Time ...................................................................................................... 139
6.2.3. Compressive Strength of Cement................................................................................... 140
6.3. Equipment and Materials .............................................................................................. 144
6.3.1. Ordinary Portland Cement (OPC) ................................................................................... 144
6.3.2. Aggregate (Sand) ............................................................................................................ 144
6.3.3. Mixer .............................................................................................................................. 144
6.3.4. Vicat Apparatus .............................................................................................................. 145
6.3.5. Vicat mould..................................................................................................................... 146
6.3.6. Jolting Apparatus ............................................................................................................ 146
6.3.7. Compressive Strength Tests ........................................................................................... 147
6.4. Method development (Miniaturization of tests) ......................................................... 147
6.4.1. Standard Consistency ..................................................................................................... 147
6.4.2. Setting times................................................................................................................... 148
6.4.3. Compressive strength..................................................................................................... 148
6.5. Results and Discussion................................................................................................... 150
6.5.1. Standard Consistency ..................................................................................................... 150
7
6.5.2. Setting times................................................................................................................... 150
6.5.3. Compressive Strength .................................................................................................... 151
6.6. Conclusions .................................................................................................................... 155
6.7. Limitations of the experimental method ...................................................................... 156
6.8. Further Work.................................................................................................................. 157
Chapter 7. Conclusions and Further Work ........................................................................... 158
7.1 Conclusions..................................................................................................................... 158
7.2 Further Work .................................................................................................................. 160
Appendix 2-1. Matlab code for the steady-state stochastic model ............................................. 162
Appendix 2-2. Matlab code for the time-dependent stochastic model ....................................... 167
Appendix 3-1. Matlab code for fitting absorbance to concentration........................................... 173
Appendix 6-1. Experimental Procedures ....................................................................................... 174
Appendix 6-2. Stress-strain curves for OPC cement and cement from cycled sorbent ............... 177
References ...................................................................................................................................... 178
8
List of Figures
Figure 1-1. Typical example scheme for Ca-looping post-combustion CO2 capture (enthalpies
calculated using data provided by (Mcbride, Zehe et al. 2002).
Figure 1-2. Equilibrium vapour pressure of CO2 over CaO as a function of temperature (Mcbride,
Zehe et al. 2002).
Figure 1-3. Carrying capacity of CaO sorbent through 50 CO2 capture-and-release cycles represented
in terms of mass change vs. time (TGA data for Havelock limestone).
Figure 1-4. Sorbent conversion vs number of cycles. Solid line corresponds to data from eq 1-8.
Figure 1-5. Sensitivity analysis showing the impact of varying critical cost parameters by ±30% on
CO2 capture cost (MacKenzie, Granatstein et al. 2007)
Figure 1-6. Overview of a cement production plant. a) Pre-treatment of the raw materials; b) high
temperature process; c) post-treatment of clinker / cement.
Figure 1-7. An In-line calciner pre-heater system. a) the pre-heater tower, b) the pre-calciner, c) the
rotary kiln, d) the clinker cooler, e) burner and secondary air inlet, f) tertiary air pass (FLSmidth
2011).
Figure 1-9. Temperature dependence of the 7 polymorphs of pure tricalcium silicate (alite) (Taylor
1997).
Figure 1-10. Temperature dependence of the five polymorphs of pure dicalcium silicate
Figure 1-11. Mass balance on raw materials, fuels and CO2 in a typical modern dry cement
production process.
Figure 1-12. Mass balance on (a) a reference cement plant and (b) the novel CO2 mitigation strategy
proposed. Basis = 1 tonne of cement, LHV 25 MJ / kg, 10 % air excess (Rodriguez et al., 2008a).
Figure 1-13. The effect of cement product SO3 content and Silica. Modulus (SM) on major phase
formation (Bhatty et al., 2004).
Figure 2-2. Schematic diagram depicting the procedure for establishing an age distribution via the
stochastic model on each cycle.
Figure 2-3. Comparison of rN vs. cycle number for the two methods.
Figure 2-4. % Hin / (Hin + Hcomb) for a. The INCAR model; b. the stochastic model.
Figure 2-5. Degradation curves for a. Undoped Longcliffe; b. HBr-doped Longcliffe; c. Kelly Rock
samples hydrated by steam and preheated for 24 h at different temperatures. d. Precipitated Calcium
Carbonate (PCC).
9
Figure 2-6. a. Undoped Longcliffe (50 cycles) b. HBr-doped Longcliffe (50 cycles) c. Thermally pre-
treated / steam pre-activated Kelly Rock (30 cycles) d. Precipitated calcium carbonate (30 cycles).
Figure 2-7. Scheme of the kinetic model adopted by (Alonso, Rodriguez et al. 2009).
Figure 2-8. TGA data for CO2 conversion as a function of time (Havelock). Red line indicates an
approximate value for t*, whilst the black arrow indicates the direction of cycle number through the
data.
Figure 2-9. Flow-sheet of the stochastic model incorporating bed inventories into the calculation of
ECO2.
Figure 2-11. Plot of ECO2 as a function of time a. WCaO = 400, FR = 10, F0 = 0.01; b. WCaO = 400, FR =
1, F0 = 0.01
Figure 2-12. Flow-sheet of a cement rotary kiln system including cyclone preheater section,
precalciner and tertiary air duct.
Figure 3-1. Core components of the vertical tube furnace (fluidised bed design).
Figure 3-7. CO2 concentration and system temperatures for ‘no fuel’ cycling experiments.
Figure 3-8. CO2 concentration and system temperatures for cycling experiments. La Jagua coal
(batch-fed).
Figure 3-9. CO2 concentration as a function of time on a single cycle. Red lines indicate where the
start and end of the points included in the mass balance.
Figure 4-1. Schematic diagram of an ICP assembly showing the three concentric tubes composing the
torch, the RF coil, the different plasma regions, and the temperature as a function of height above the
load coil.
Figure 4-2. Comparison of ICPAES and ICPMS for NIST SRM 1633b (average of three repeat
samples, error bars shown as relative standard deviation, RSD).
10
Figure 4-3. Examples of wavelength calibration for Mn (257.610 nm) and Cr ( 267.716 nm) showing
correlation co-efficients for each line.
Figure 4-4. Example of ICP-AES drift over the course of a four hour run.
Figure 4-5. Wavelength peaks for a) Bi at 206.170 and 2) Zn at 206.200 nm (0.5 ppm).
Figure 4-6. Spectral peak at a) 407.735 nm (La) and b) 407.771 nm (Sr) (0.5 ppm).
Figure 4-8. Comparison of experimental values against certified values for NIST SRM 1d
(argillaceous limestone) for a. sample concentrations > 1000 ppm and b. sample concentrations <
1000 ppm.
Figure 4-9 (a-k). Trace element inventories of sorbent cycled without fuel. Error bars shown as RSD
values.
Figure 4-10. Trace element inventories of sorbent upon cycling in the presence of batch fed La Jagua.
Error bars shown as RSD values.
Figure 4-11. Trace element inventories of sorbent upon cycling in the presence of continuously fed
fuel (La Jagua, Lea Hall and RDF). Error bars shown as RSD values.
Figure 5-2. Temperature profile of the horizontal tube furnace (Max. temperature of 1297 ºC for a
wall set-point of 1350 ºC).
Figure 5-3. Alite and Belite Peaks for compressed but not cooled (green line), cooled but not
compressed (red line) and both (blue line). XRD settings: Scan step size: 0.0167, Time per step: 60
seconds).
Figure 5-4. XRD scan from 49 – 54º 2θ showing the largest available alite peak at 51.7 2θ.
Figure 5-5. XRD scan from 32.6 – 34º 2θ showing a singlet aluminate peak at 33.2 2θ.
Figure 5-7. % wt. alite in clinkers produced from sorbent cycled without fuel and in the presence of
La Jagua coal combustion (2 g / cycle). Error bars shown as RSD.
Figure 5-8. % wt. alite in clinkers produced from sorbent cycled in the presence of a) continuously
fed La Jagua coal, b) continuously fed Lea Hall coal and c) RDF.
Figure 6-1. Relationship between porosity and compressive strength of Portland cement
(Chindaprasirt and Rukzon 2008)
Figure 6-6. Setting times of OPC and cement produced from cycled sorbent (1 cycle, no fuel).
Figure 6-7. Results of the strength test for different shaped-samples (OPC, 3-day test)
Figure 6-8. Images showing a) uneven finishing of top side of cylindrical samples b) visible marks of
uneven hydration.
Figure 6-10. Stress-strain behaviour for OPC 3-day a) cylinders; and b) cubes.
12
List of Tables
Table 1-1. Physical properties of calcium compounds relevant to the Ca-looping cycle.
Table 1-4. Abbreviation of the chemical formulae used in the cement field .
Table 1-5. Behaviour of trace elements in the kiln and their effect on cement phases and clinker
mineralogy.
Table 2-1. Results for comparison of ̅ for the two methods as a function of F0 and FR.
Table 2-3. Optimum parameters to minimise calciner heat demands for different sorbents.
Table 2-4. Time taken for ECO2 to reach steady state (tss) for different values of WCaO, FR and F0.
Table 2-6. Comparison of key values in the mass balance against those of a reference case.
Table 2-7. Values of false air used in the model (kg / kgck).
Table 2-8. Combinations of WCaO / FR which give a ECO2 value of 92 % and the associated values of
Hin.
Table 4-4. Spectral wavelengths for elements of interest and prevalence in the inputs.
Table 4-5. ICPAES results of same single-cycled ‘no fuel’ CaO sorbent run on 3 different dates over
3 weeks.
Table 4-6. Certified Values for SRM 1d Limestone, Argillaceous and conversions from % oxides to
13
Table 4-7. Sample concentrations (ppm) and theoretical weights (mg) and the results of the
“Students” t-test comparing quantities (mg) of elements in single-cycled sorbent against the original
limestone.
Table 4-8. Example of “Students” t-test comparing concentration (ppm) of elements in single-cycled
Table 4-9. Concentration of trace elements in the inputs (mean of twenty samples, ppm).
Table 6-1. Compressive strengths of different cement types according to European classification.
14
Parts of thesis have been published in (Dean, C. C. et al. 2011) and (Dean, C. et al 2012).
John Blamey, Mohamad Al-Jeboori and Nick Florin helped with the review article upon which the
15
Introduction
All scientific bodies now concede that the evidence for a link between observed climate change and
human activity, primarily CO2 emissions, is compelling (Oreskes 2004). According to the IPCC
(Metz, Davidson et al. 2007) and IEA (IEA 2006), the cement industry is responsible for a
contribution of between 5 and 7 % to global anthropogenic CO2 emissions. These emissions come
both from chemistry inherent to the cement process (calcination of limestone) and the burning of
fossil fuels, necessary to provide the high temperatures required for the production of clinker.
A state-of-the-art cement plant will include heat integration within the process such that further
reductions in fuel use / CO2 emissions are not possible based on thermodynamic limitations;
furthermore, emissions from limestone calcination are seemingly unavoidable due to the need to
provide CaO for reaction with other compounds to produce clinker. However, the emergence of CaO-
sorbent-based CO2 capture systems (“The Ca-looping Cycle”) offers the potential to supply CaO to a
cement process, avoiding CO2 emissions from both limestone calcination and from the burning of fuel
to provide heat to drive the endothermic calcination reaction. This leaves only the portion of CO2
Calcium looping (Ca-looping) is a CO2 capture scheme using solid CaO-based sorbents to remove
CO2 from flue gases, i.e., from a power plant or other large stationary industrial process, producing a
concentrated stream of CO2 (~ 95 %) suitable for storage. The Ca-looping process uses a CaO-based
sorbent, usually derived from limestone, reacting via the reversible gas–solid reaction between CO2
CaO purged from the Ca-looping process will however have been subjected to repeated cycles of
calcination and carbonation, most likely in the presence of fuel combustion at the calcination stage
(assuming in-situ fuel combustion system). This repeated cycling means that the CaO may be
different in chemical composition to that of single-calcination CaO; it will certainly contain higher
16
levels of ash, which may be physically mixed in with the sorbent, or which may react with the CaO to
form initial clinker compounds before the sorbent has reached the kiln.
Furthermore, trace elements either contained in the ash sent with the sorbent or released from the fuel
through devolatilisation and subsequent reaction with the CaO will be higher in concentration than
that of single-calcination CaO. Most trace elements commonly present in fossil fuels are known to
impact on cement chemistry in one way or another (see: Background). The practical work detailed in
this thesis describes an investigation into 1. changes in the concentration of trace elements in the
sorbent due to repeated cycling in the presence of three different solid fuels; 2. the effect of using
repeatedly cycled CaO on the % wt. alite in clinker, and to correlate any effect to changes in sorbent
trace element concentration; 3. The effect of using repeatedly cycled CaO on the mechanical
17
Chapter 1. Background
Calcium looping (Ca-looping) is a CO2 capture scheme using solid CaO-based sorbents to remove
CO2 from flue gases, i.e., from a power plant or other large stationary industrial process, producing a
concentrated stream of CO2 (~ 95 %) suitable for storage. Reviews of the Ca-looping cycle have
recently been published (Blamey, Anthony et al. 2010; Dean, Blamey et al. 2011); the main features
of which are summarised herein. For further information, see also reviews by Stanmore and Gilot and
others (Stanmore and Gilot 2005; Anthony 2008; Harrison 2008; Li and Fan 2008; Florin and Harris
2008a), and the IEA GHG High Temperatures Solid Looping Cycles Network (IEA 2010).
The Ca-looping process uses a CaO-based sorbent, usually derived from limestone, reacting via the
reversible gas–solid reaction between CO2 and CaO(s) to form CaCO3(s) (Eq. 1-1). The forwards step is
known as calcination and is an endothermic process which readily goes to completion under a wide
Eq. 1-1
A typical Ca-looping process for CO2 capture as proposed by (Shimizu, Hirama et al. 1999) is shown
in Figure 1-1. In this example, the heat necessary for calcination is provided by oxy-combustion of
coal, however other methods have been suggested such as providing an external heat source (Alvarez
18
Figure 1-1. Typical example scheme for Ca-looping post-combustion CO2 capture (enthalpies
calculated using data provided by (Mcbride, Zehe et al. 2002).
The equilibrium vapour pressure of CO2 over CaO according to Eq. 1-1 can be calculated as a
function of temperature; partial pressures of CO2 greater than the equilibrium partial pressure at a
given temperature will favour carbonation, while those lower than the equilibrium will favour
calcination (see Figure 1-2). As a result, if a sorbent is cycled between two vessels at suitable
temperatures, carbonation of sorbent can be effected in one and calcination in the other. The
objective of CO2 capture is to obtain a pure stream of CO2 suitable for storage; one method of
achieving this is by separation of CO2 from an exhaust gas obtained from power stations or industry
19
Figure 1-2. Equilibrium vapour pressure of CO2 over
CaO as a function of temperature (Mcbride, Zehe et al.
2002).
The calciner temperature (> 850 ◦C) is dictated by the high CO2 partial pressure necessary to produce
a near pure stream (Figure 1-2), and may also be chosen as a compromise between higher
temperatures, which increase the rate of calcination at the expense of rapid degradation of the sorbent
(see: section 1.1.1.1) and milder temperatures which protect the sorbent; the carbonator temperature is
chosen as a compromise between the higher equilibrium (maximum) capture at lower temperatures
Calcium looping has a number of advantages compared to many closer-to-market capture schemes,
including: the use of circulating fluidised bed (CFB) reactors (owing to the good gas-solid contacting
and uniform bed temperature), a mature technology developed and deployed at large scale (Cuenca
M.A. and E.J. 1995; Grace J.R., Avidan A. et al. 1997) up to 460 MWe (Foster_Wheeler, 2012);
sorbent derived from cheap, abundant and environmentally benign limestone precursors; and the
relatively small efficiency penalty that it imposes on industrial power production process (i.e.,
capture). A further advantage is the synergy with cement manufacture, which potentially allows for
simultaneous mitigation of a large proportion of CO2 emissions from the cement process. Integration
20
of power generation and cement manufacture leads to a further reduction in the overall efficiency
penalty. If attributed to the power station, the efficiency penalty is reduced by 1.5 – 2 % (Blamey et
al 2010). The Ca-looping cycle can also be applied to gasification / pre-combustion techniques,
where it can be used to increase hydrogen production by removing CO2 from the products of the water
gas shift Eq. 1-2 and reforming reactions e.g. Eq. 1-3 (Harrison 2008; Florin and Harris 2008a), thus
Eq. 1-2
Eq. 1-3
Ideally, CaO sorbent would be continually cycled between carbonator and calciner reacting with and
evolving 1 mol of CO2 per mol CaO on each cycle. However, the rate of carbonation of CaO makes a
transition from a fast to a very slow diffusion-controlled rate through the product layer (see inset
Figure 1-3) (Bhatia and Perlmutter 1983; Mess, Sarofim et al. 1999; Alvarez and Abanades 2005)
Figure 1-3. Carrying capacity of CaO sorbent through 50 CO2 capture-and-release cycles
represented in terms of mass change vs. time (TGA data for Havelock limestone).
21
The conversion of CaO to CaCO3 at this transition is often defined as the maximum carbonation
conversion or ‘carrying capacity’ of the sorbent and is well known to decrease upon cycling of the
sorbent through multiple CO2 capture-and-release cycles (Curran 1967; Barker 1973; Deutsch and
Heller-Kallai 1991; Silaban and Harrison 1995; Aihara, Nagai et al. 2001; Abanades 2002; Abanades
Loss in sorbent reactivity (as observed in Figure 1-4) can be caused by a number of factors: sintering
of the porous CaO during the high temperature calcination stage (the major cause of loss in
reactivity); competing sulphation / sulphidation reactions; loss of bed material through attrition
CaO formed by calcination of CaCO3 is very porous (Barker 1973) because, as shown in Table 1-1,
there is a marked reduction in molar volume from CaCO3 to CaO, resulting in a theoretical porosity of
0.54. However, CaO used for Ca-looping is prone to sintering (i.e., changes in pore shape, pore
shrinkage and grain growth, at elevated temperatures), which causes a reduction in reactive surface
area and a drop-off in reactivity. The transition to conditions that accelerate the rate of sintering of
CaO occurs at approximately 900 °C (Borgwardt 1989a), while increasing partial pressure of CO2 and
22
steam have also been found to increase the rate of sintering (Borgwardt 1989b). Sintering of CaO can
therefore be expected to be severe in the calciner, which in the example scheme in Figure 1-2 is
expected to be operating above 900 °C at 101 kPa CO2. There is also a contribution of reactive
CaO has a very strong affinity to SO2 and will readily form CaSO4 in the oxidising conditions
expected in post-combustion systems (see Eq. 1-4 for direct sulphation and Eq. 1-5 for indirect
sulphation, i.e., sulphation of the carbonate). As a result, CaO derived from limestone has been used
as a sorbent for SO2 on an industrial scale (Smith 2007); e.g., Drax Power Station (coal-fired,
3.96GW) (Drax 2008) in the UK captured 90 % of its sulphur emissions using an independent wet
limestone–gypsum flue gas desulphurisation plant. CaO can form CaS upon reaction with H2S under
reducing conditions (see Eq. 1-6 for direct sulphidation and Eq. 1-7 for indirect sulphidation, again,
sulphidation of the carbonate). Sulphated sorbent requires very high temperatures or reducing
conditions to be regenerated, though sulphide sorbent can be regenerated more readily (Lyngfelt and
Leckner 1989; Hayhurst and Tucker 1991; Anthony and Granatstein 2001); both reactions can be
considered as resulting in reactive sites lost for carbonation. Grace and co-workers (Ryu, Grace et al.
2006; Sun, Grace et al. 2006) investigated the effect of sulphate and sulphide formation on CO2
uptake of sorbent and found that sulphate formation is a more significant problem as it results in pore
23
blockage from the formation of CaSO4, which has a much larger molar volume (seeTable 1-1). This
Eq. 1-4
Eq. 1-5
Eq. 1-6
Eq. 1-7
There has been a large volume of research published on the attrition characteristics of limestone for
SO2 capture (Anthony and Granatstein 2001; Smith 2007; Chen, Grace et al. 2008; Saastamoinen,
Pikkarainen et al. 2008; Scala, Montagnaro et al. 2008; Blamey, Anthony et al. 2010). However,
there is less research regarding limestone for CO2 capture. (Fennell, Pacciani et al. 2007a) showed
that the most significant attrition occurred in the early cycles of repeated calcination / carbonation
cycling experiments. Pilot plant studies published by (Alonso, Rodríguez et al. 2010) observed
significant attrition of sorbent at the start of the experiment, while pilot plant studies published by
(Lu, Hughes et al. 2008) showed significant attrition throughout operation (> 60 % of material ended
up as fine material in the cyclones after 25 cycles, reducing the CO2 capture rate significantly). The
mechanical stability of the sorbent is therefore receiving considerable attention and is likely to be a
critical factor in the selection of original limestone for large-scale testing; on this basis, there is a
drive towards standardised testing of the mechanical stability of limestone to determine suitability for
use in a fluidised environment (Schueppel 2009; Zuberbuehler 2009). An important aspect of scale-
up to a typical proposed system, such as that described in Figure 1-1, is the effect of fuel combustion
in the calciner. The combustion of the fuel results in two undesirable effects: the introduction of fine
ash and chars into the cycle, which have the potential to deposit onto the surface of a particle and
result in undesirable side reactions and low temperature melt formation; and higher local temperatures
related to fuel-combustion (Joutsenoja, Heino et al. 1999; Scala, Chirone et al. 2003; Corella, Toledo
24
et al. 2006). Accordingly, (Hughes, Macchi et al. 2009) report lower reactivity of sorbent than
anticipated and the formation of a dense shell on the surface of particles when combusting high-ash
hardwood pellets in the calciner; a similar problem has been described by (Kuramoto, Shibano et al.
2003). There have been a number of attempts to model the decay rate of the maximum carbonation
conversion upon cycling, which focus on the reactivity towards CO2 disregarding sulphation, attrition
and ash fouling effects (Abanades 2002; Abanades and Alvarez 2003; Wang and Anthony 2005;
Grasa and Abanades 2006; Lysikov, Salanov et al. 2007; Gonzalez, Grasa et al. 2008). For
descriptions of these decay equations and a discussion of their individual merits and drawbacks please
refer to (Blamey, Anthony et al. 2010). It should be noted that most of the models proposed are semi-
empirical and cannot currently be used to predict the behaviour of individual limestones from their
intrinsic properties. The author’s preferred equation due to its accurate representation of the loss of
reactivity over large numbers of cycles (> 20) is Eq. 1-8, proposed by (Grasa and Abanades 2006),
where is the maximum carbonation conversion of the sorbent in cycle N, a∞ is the residual (final)
conversion of the sorbent after a very large number of cycles, and k is a decay rate constant:
( ) Eq. 1-8
It is likely that in a realistic system where sulphation, attrition and ash fouling are unavoidable, a ∞ will
tend to zero, however in small scale tests a residual capacity of ~ 60 mg CO2 / g sorbent is found,
which is competitive with other CO2 sorbents (Rodriguez 2008). A real system will also differ in that
the residence time of the sorbent in the calciner or carbonator may not be sufficient to allow full
calcination or maximum carbonation, making application of these formulas difficult; such matters are
under investigation by Abanades and co-workers (Grasa, Abanades et al. 2009; Rodríguez, Alonso et
al. 2010).
25
1.1.1.2. Sorbent Performance
There is currently a great deal of work being conducted to improve sorbent performance upon cycling.
This can be done by alteration of the process conditions or by sorbent enhancement. There are
numerous process variables which will affect sorbent performance, the most important of which is the
temperature of the calciner. However, steam partial pressure in the carbonator, the residence time of
the particles in the calciner and the carbonator, the presence of ash, sulphur and other minor species,
and particle size are likely to also play significant roles. The temperature of the calciner can be
decreased if the partial pressure of CO2 is reduced, thus minimising the extent of sorbent sintering
upon cycling; the two simplest ways of doing this are by lowering the total pressure (Ewing, Beruto et
al. 1979; Alvarez and Abanades 2005; Sakadjian, Iyer et al. 2006) or by introducing steam into the
calciner (Alvarez and Abanades 2005; Wang, Lin et al. 2008; Wang, Lin et al. 2009). The main
research areas for sorbent enhancement are doping of natural limestones with trace amounts of
organic salts, the production of synthetic sorbents, the hydration of spent sorbent, and thermal pre-
treatment; an ideal enhanced sorbent will display high mechanical strength while maintaining its
reactive surface area over repeated cycling, without being prohibitively expensive—for economic
assessments see work by Romeo and co-workers (Romeo, Lara et al. 2009a; Lisbona, Martinez et al.
2010).
There is considerable variability in the reactivity and friability of natural limestones, which should be
investigated before selecting a sorbent (Laursen, Duo et al. 2000; Ryu, Grace et al. 2006; Alvarez,
Peña et al. 2007; Fennell, Pacciani et al. 2007a). These properties depend on impurities and the
conditions / age of formation. Dolomites have shown a reduced decay rate upon cycling, despite a
lower initial reactivity owing to a lower amount of CaO per unit mass (Dobner, Sterns et al. 1977;
Silaban, Narcida et al. 1996); natural sorbents such as shells have also been studied and have been
shown to exhibit mildly increased reactivity in comparison to natural limestone in some cases (Ives,
Mundy et al. 2008; Li, Zhao et al. 2009). Doping of natural limestones has been investigated;
however the results have been variable (Salvador, Lu et al. 2003; Sun, Grace et al. 2006; Fennell,
26
Davidson et al. 2007b; Florin and Harris 2008b), even when the same dopant has been used. The
efficacy of doping is highly dependent on the concentration of dopant used; too low a concentration
will have no effect, too high may contribute to pore blocking and potentially agglomeration of
particles. ‘Synthetic sorbents’ are also being developed. Examples include: precipitated CaCO3 from
a slurry bubble column (Agnihotri, Mahuli et al. 1999; Gupta, Iyer et al. 2004; Florin and Harris
2008c) Florin and Harris, 2008c) to enhance the reactive surface area of the sorbent; sorbents derived
from different calcium precursors, e.g., calcium acetate, calcium ethanoate (Lu, Reddy et al. 2006; Lu,
Khan et al. 2008; Liu, Low et al. 2009) with particular success using a MgO support (Liu, Feng et al.
2010) to similarly enhance the reactive surface area; dispersal of CaO within an inert porous matrix
such as mayenite (Li, Cai et al. 2005; Pacciani, Müller et al. 2008a) to improve mechanical stability;
and use of cementitious binders (Manovic and Anthony 2009a; Manovic and Anthony 2009b) to
calcined sorbent, though this is often at the expense of mechanical strength of the sorbent (Hughes, Lu
et al. 2004; Manovic and Anthony 2007; Fennell, Davidson et al. 2007b; Manovic, Lu et al. 2008;
Sun, Grace et al. 2008; Zeman 2008). Thermal preactivation / pre-treatment, by treating sorbent at
high temperature under N2, has been found to improve long-term reactivity of sorbents (Manovic and
All studies of sorbent enhancement should consider that sorbent reactivity is dependent on several
variables: calcination temperature and time; carbonation temperature and time; CO2 concentration in
carbonation and calcination; and, the difficulty in defining ‘normal’ behaviour in a limestone, which
may be used as a reference. These factors are important, in that they make comparisons between
different studies difficult. For example, very different results are obtained for deactivation rates when
calcination is conducted under mild vs. severe conditions (Grasa and Abanades 2006; Manovic and
Anthony 2008a; Manovic and Anthony 2008c; Manovic and Anthony 2009a; Wang, Ramkumar et al.
2010) and also due to increases in carbonation temperature (Manovic and Anthony 2008c) (both of
which are reported to be due to the effect of sintering). Longer carbonation times can regenerate the
27
sorbent (Barker 1973; Lysikov, Salanov et al. 2007; Chen, Song et al. 2009) though the effect of this
is somewhat inconsistent elsewhere (Manovic and Anthony 2008c). A higher CO2 concentration
during carbonation has been shown to improve the uptake of some synthetic sorbents (Pacciani,
Müller et al. 2008b). In cases where the CO2 concentration is higher during calcination this
consistently leads to a more rapid drop-off in reactivity (Li, Cai et al. 2005; Lu, Hughes et al. 2009);
however there are cases where the effect of CO2 concentration in calcination is only noticeable over
the first five cycles (Manovic and Anthony 2009a). The presence of SO2 leads to a reduction in
carbonation reactivity owing to the competing sulphation reaction (see Section 1.1.1.1).
USD / tCO2 which compares competitively with estimates for amine scrubbing, e.g., 32.5 – 80 USD /
tCO2(Herzog 1998; David and Herzog 2000) (figures converted to 2006 USD from Canadian dollars,
using mean exchange rate June 2006). Their study included a sensitivity analysis and identified the
cost of limestone and the assumed Ca / C ratio as having the most significant influence on the cost
28
(Figure 1-5). The parameters shown to have the most significant effect on the economics of the
process were the cost of the limestone and the Ca / C ratio (i.e. the amount of limestone in the system
(Romeo, Lara et al. 2009b) examined the impact on the cost of CO2 avoided of varying the Ca / C
molar ratio and purge rate (as a percentage of the solids inventory); they showed that the impact of
purge rate on the capture cost was more significant for higher Ca / C ratios (Ca / C > 4) because at
higher Ca / C ratios, increasing the purge rate does not lead to an increase in CO 2 capture capacity
while the cost rises monotonically with purge rate. The study examined the effect of these process
variables purging both from the carbonator and calciner. Purging from the carbonator allows a lower
possible capture cost to be reached because material is removed before a further energy intensive
calcination step is required. However, purging from the calciner allows material to be removed as
CaO which can provide both an energy and CO2 credit to cement production. Purging from the
carbonator, the optimum configuration (i.e., resulting in the minimum capture cost) comprised a Ca /
C ratio of 4 and a purge rate of 2 % giving a minimum cost of 19.8 USD / tCO2. Purging from the
minimum cost of 20.5 USD / tCO2. The results showed consistently, regardless of purge location or
fuel type, that the minimum costs could be achieved with higher Ca / C ratios and lower purge rates.
In all cases studied the cost of CO2 capture remained competitive (< 29 USD / tCO2) (figures
converted to USD from Euros using mean exchange rate January 2008).
Cement is a hydraulic binder forming a paste with water, which sets and hardens following hydration
reactions and creates the strength and durability of concrete. Cement is either classified by its
29
However, the most common classifications are related to the European Standards for Common
Cements (e.g. EN 197-1) defined by the European Committee for Standardization (BSI 2011) and the
ASTM Standards (e.g. C150 / C150M and C595 / 595M) defined by the American Society for Testing
and Material (ASTM) (ASTM 2009). A detailed overview of the different cement classifications is
given in (Jackson 2007). Portland cement or ‘Ordinary Portland Cement’ (‘OPC’) forms by far the
The main constituents of cement production are calcium oxide (CaO), silicon dioxide (SiO 2),
aluminium oxide (Al2O3) and iron oxide (Fe2O3). Typical sources of these oxides are limestone,
chalk, marl, clays (kaolinite, illite, feldspar), shale, tuff, oil shale, bauxite and iron ore (Bye 1983).
These materials often contain alkalis, heavy metals, sulphate, sulphide, phosphate, fluoride and
chloride compositions in low or trace concentrations (Bye 1983; Taylor 1997; Glasser, Bhatty et al.
2004). Besides natural materials, waste products from other industrial processes such as lime sludge
or fly ash from coal combustion can be used. In the use of waste materials, the addition of relatively
pure limestone, sand or iron ore may also be necessary (Johansen and Kouznetsova 1992) to adjust for
absent chemical compounds and to achieve the required standards (Jackson 2007). A typical chemical
composition of a four-component raw meal is listed below (Table 1-2). The chemical composition of
the raw material entering the system is continually controlled in plant laboratories, often through
automated means.
30
Table 1-2. Typical chemical composition of a four component raw meal (Jackson 2007).
Limestone Shale Sand Iron Oxide Kiln
(% wt.) (% wt.) (% wt.) (% wt.) feed*composition
(% wt.)
Cement production involves the use of high temperatures, including the drying of raw materials, the
endothermic decarbonisation reaction of limestone and clinker formation. The energy required is
provided by the combustion of fuel, namely, coal, lignite, heavy and light fuel oil, fuel gas and / or
petroleum coke (Table 1-3). These traditional fuels are being increasingly substituted by alternative
fuels such as scrap tyres, waste oil, wood waste, plastic waste and refuse-derived fuel
(RDF)(CEMBUREAU 1997; Pizant and Gauthier 1997; Trezza and Scian 2000; Mackes and
Lightburn 2003; Mokrzycki and Uliasz-Bochenczyk 2003; Proscoandro, Mazziotti et al. 2003;
Kääntee, Yevenhoven et al. 2004; Stöppel 2004; Pipilikaki, Katsioti et al. 2005; Trezza and Scian
2005; Greco and Enfil 2006; Zabaniotou and Theofilou 2008; World_Cement_News 2011), disposing
waste and reducing the deletion of natural resources simultaneously. Raw minerals may also be
replaced, for example iron oxide can be replaced with steel carcasses from scrap tyres.
31
Table 1-3. Consumption of fuel at cement
plants in the EU (European_Commission)
Fuel type Consumption (%)
Petcoke (fossil) 38.6
Coal (fossil) 18.7
Petcoke and coal 15.9
(together)
Fuel oil 3.1
Lignite and other solid 4.8
fuels
Natural gas 1.0
Waste fuel 17.9
Figure 1-6. Overview of a cement production plant. a) Pre-treatment of the raw materials; b) high
temperature process; c) post-treatment of clinker/cement.
Raw materials are first crushed and separately homogenized to provide a final product of consistent
quality (Figure 1-6a). Further raw meal preparation was previously distinguished by dry- and wet
32
pre-treatment depending on the moisture content of the raw materials. A feedstock with a moisture
content of > 15 % wt. was often prepared under wet conditions where water was added in order to
achieve a slurry with a water content of 25 – 50 % wt. (Bye 1983; Jackson 2007). The advantage of
this method is the achievement of high raw meal uniformity. However, the use of wet kilns has
decreased rapidly in recent years due to its low thermal efficiency. The more energy-efficient dry
method is by far the most employed process in modern cement production. In the dry process, raw
materials are ground and dried simultaneously. The blending of the different materials is carried out
either by simultaneous grinding or after the grinding step by air agitation in large silos enabling
1.2.2.3. Pre-heating
Clinker raw meal is pre-heated in the cyclone pre-heater (Figure 1-6b, Figure 1-7a) which is
essentially a series of cyclones arranged in 50–120 m towers. Hot gases (~ 1000 K) from the
precalciner and kiln heat the raw meal in counter-current flow increasing the feed temperature from ~
200 K to ~ 800 K in less than 60 seconds (Taylor 1997). High contents of inorganic volatiles in the
gases(Na, K, S and Cl) are problematic when released from the raw material or fuel in the rotary kiln
as they may condense at the pre-heater walls forming solid deposits (ring build-ups) (Choi and
Glasser 1988) which places limitations on the motion of gas and material around the system (Bye
1983).
1.2.2.4. Calcination
The separation of the calcination process of calcareous materials from processes in the rotary kiln
enabled a reduction in kiln length or, with unchanged kiln dimensions, led to an increase in
production capacity thus decreasing investment and operational costs (Bye 1983). The hot raw meal
enters the calciner at about 800 K from the preheater. The energy required for the endothermic
decarbonisation reaction (calcination) is obtained from hot kiln gases and additional combustion of
33
fuel at a secondary burner taking place at temperatures of 1223 – 1273 K. In a fully-integrated
process, hot air from the clinker cooler (‘tertiary’ air) is utilised as combustion air (Figure 1-7f). The
consumption of fuel for calcination can account for up to 60% of fuel needed for cement production
(Alsop, Hung et al. 2007). The degree of decarbonisation constitutes 90 – 95% whilst solid material
Figure 1-7. An In-line calciner pre-heater system. a) the pre-heater tower, b) the
pre-calciner, c) the rotary kiln, d) the clinker cooler, e) burner and secondary air
inlet, f) tertiary air pass(FLSmidth 2011).
1.2.2.5. Clinkerization
The heart of the cement-making process is the rotary kiln which varies in length and diameter
depending on the design of the pyro-process system (i.e. from raw meal pre-heating through to clinker
cooling). Typical dimensions for rotary kiln systems with cyclone pre-heater and precalciner are 50–
100 m in length and 3–7 m in diameter (length to internal diameter ratio ~11–16) (Alsop, Hung et al.
2007). These kilns operate typically at a tilt of 1 – 3 ° from the horizontal and with a rotation velocity
34
of 2 – 4.5 rev / min resulting in material kiln retention times of ~ 20–40 min (Alsop, Hung et al. 2007)
The pre-calcined solid meal is fed into the kiln at the higher end whilst fuel, ‘secondary’ air (hot
combustion air drawn in from the cooler – Figure 1-7e) and primary air (a much smaller air stream in
comparison to secondary and tertiary air streams – as a means of controlling the flame temperature) is
blown into the kiln and combusted at the lower end creating a flame with temperatures of around
2200K (Klauss 2000) (Figure 1-7e). The calcined meal moves counter current to the hot gases toward
the hot region (sintering zone) of the kiln thereby achieving heat transfer between gases, solids and
the kiln walls. The calcined feed is heated up to ~ 1700 K leading to reactions and mineralogical
changes in the different zones of the kiln. These reactions include decomposition of the remaining
uncalcined CaCO3 and formation of the clinker phases which are alite and belite. At temperatures of
up to 1400 K solid state reactions occur enabling the belite, aluminate and ferrite phases to be
gradually formed. In addition, at these temperatures, inorganic volatiles are released from the feed
material and fuel consisting mainly of alkalis, sulphates and chlorides, which condense and form
deposits as rings inside the kiln. Melting of the aluminate, ferrite and to some extent the belite phase
occurs at higher temperatures (1573 – 1723K) forming a liquid phase which leads to solid particles
sticking, referred to as granulation / nodulisation. Lastly, ‘free’ (excess) CaO and belite react to form
Controlled clinker cooling is very important for clinker quality and for the energy efficiency of the
overall production process since it enables heat recovery as depicted in Figure 1-7. It is essential that
the hot clinker is cooled down quickly to below 1200 – 1250 °C as fast cooling causes the re-
crystallisation of the finely grained aluminate phase (Bye 1983) which results in a desirable slow,
controllable hydration reaction during the cement setting process. Conversely, a coarsely grained
aluminate phase formed due to low cooling rates causes overly rapid setting of cement (Jackson
2007). In addition, the important alite clinker phase is only thermodynamically stable at temperatures
above 1250 °C. Below 1250 °C, alite decomposes into belite and calcium oxide (Figure 1-8) (Mohan
35
and Glasser 1977). Rapid cooling to temperatures below this critical temperature results in metastable
alite.
Lastly, the higher the cooling rate, the higher the heat recovery which improves the energy efficiency
For the preparation of cements, clinker needs to be ground and blended with additives (Figure 1-6c),
the most important of which are gypsum or anhydrite which control the setting of concrete during
hydration. Other additives are for example blast-furnace slag, limestone or fly ash depending on the
desired cement type (Bye 1983; Johansen and Kouznetsova 1992; Dunstetter, de Noirfontaine et al.
Cement chemistry is a complex topic; some of the processes involved are still not yet fully
understood. However, a summary of the essential chemistry relevant to the four main clinker phases
is given here.
Cement clinker is comprised of four main phases, all containing CaO. These phases are commonly
36
Table 1-4. Abbreviation of the chemical formulae used in the cement field (Bye 1983)
Chemical compound Abbreviation Clinker phase Clinker phase
Abbreviation
CaO C Ca3SiO5 C3S
SiO2 S Ca2SiO4 C2S
Al2O3 A Ca3Al2O6 C3A
Fe2O3 F Ca4Al2Fe2O10 C4AF
Ca12Al7O33 C12A7
In OPC the four main phases appear in order of prevalence; tricalcium silicate ( C3S); dicalcium
silicate (C2S); tricalcium aluminate (C3A) and tetra-calcium alumino-ferrite (C4AF). The first two are
responsible for cement strength and commonly contain impurities; in their impure form these are
commonly known as alite and belite respectively. Tricalcium aluminate (C3A) incorporates impurities
such as Si4+, Fe3+, Na+ and K+ and is the most reactive of the clinker phases
Alite (C3S)
Alite is the most important clinker phase in cement since it controls mainly the initial and ultimate
strength of cement. OPC clinker consists of ~ 50 – 70 wt. % of alite which contains 71 – 75 % wt.
CaO, 24 – 28 % wt. SiO2 and 3 – 4 % wt. substituted ions (Taylor 1997). Ions typically incorporated
within the alite crystal lattice are Mg2+, Al3+ and Fe3+. The impurities in alite stabilize high
temperature polymorphs at low temperatures (below the related decomposition temperature) (Maki
and Goto 1982). So far, there exist seven known polymorphs between room temperature and 1343 K:
three triclinic (denoted with T), three monoclinic (M) and one rhombohedral (R) polymorph (Figure
1-9) (Bye 1983; Taylor 1997; Dunstetter, de Noirfontaine et al. 2006). Due to incorporations in the
alite crystal lattice, M1 and M3 polymorphs are present mostly in industrial clinker. Inversion of the R
polymorph to M3 and further more to M1 occurs when cooling clinker from 1723 K, forming small
crystals (M3) rich in substituents or large crystals but poor in substituted ions (M 1). Especially, high
MgO- concentrations promote high nucleation, resulting in formation of small automorphic M3-
37
crystals (Maki and Goto 1982; Taylor 1997). The different polymorphs do not show significant
Figure 1-9. Temperature dependence of the 7 polymorphs of pure tricalcium silicate (alite) (Taylor
1997).
Belite (C2S)
The second most prevalent clinker phase in OPC is belite. Its hydration products develop similar
strength in cement as alite, only much more slowly. Belite makes up between 15 and 30 % wt. of
OPC clinker and consists of 60 – 65 % wt. CaO, 29 – 35 % wt. SiO2 and 4 – 6 % wt. substituted
oxides, mainly Al2O3 and Fe2O3, but also K2O, Na2O, MgO, SO3 and P2O5. Belite crystallizes in five
polymorphs: α-belite, α’H-belite, α’L-belite, β-belite (H = “high” and L = “low” symmetry) and γ-
belite (Figure 1-10), which differ in structural and hydraulic properties. The α’- polymorphs are the
most hydraulic forms of belite, whereas γ-belite is a non-hydraulic polymorph and does not account
for the setting and hardening of cement. β-belite is also a hydraulic polymorph, the most common in
industrial OPC, but less hydraulic than the α’- polymorphs. A phenomenon that needs to be prevented
through the inclusion of trace compounds is disintegration (dusting) of clinker. This happens if β-C2S
is not stabilized during cooling and / or by inclusions affording a part β-γ-C2S inversion. γ-C2S
crystals are less dense (more voluminous) than β-C2S crystals and cause cracking of other β-C2S
crystals, forming a voluminous powder and dust (Imlach 1976; Bye 1983; Taylor 1997).
Figure 1-10. Temperature dependence of the five polymorphs of pure dicalcium silicate
38
Tricalcium Aluminate (C3A)
C3A constitutes 5 – 10 % wt. of OPC clinker and is the most reactive component. Pure C3A consists
of 62 % wt. CaO and 38 % wt. Al2O3 and does not exhibit temperature dependent polymorphs.
However, ion substitution of Ca2+ in the structure of the pure C3A causes changes in crystal structure.
Typically Ca2+ is substituted by Mg2+, 2 K+ and 2 Na+, Al3+ by Fe3+ and Si4+, but only the alkali metals
affect the structural changes (Bye 1983; Taylor 1997) from a cubic crystal structure (pure C3A) to
orthorhombic and monoclinic structures via intermediate structures of lower symmetry. In industrial
clinker products, orthorhombic and cubic structures are the most common polymorphs. The
orthorhombic form features dark, prismatic crystals, whereas the cubic polymorph forms fine grains
Calcium aluminoferrite constitutes 5 – 15 % wt. of OPC (Taylor 1997). The pure phase contains 46 %
wt. CaO, 21 % wt. Al2O3, 33 % wt. Fe2O3, but in industrial clinker up to 10 % wt. of incorporated
oxides appear (mostly MgO). This phase is composed of any solid solution composition of
Ca2(AlxFe1-x)2O5, with 0 < x < 0.7 and is typically termed brownmillerite (Mineral_News ; Mindat.org
2013) The denotation C4AF represents only a single point in this series for x = 0.5. Substitution free
C4AF is yellowish brown, while substituted (Mg2+, Zn2+), air cooled C4AF is black.
Several equations have been derived to describe the quality and quantity of OPC of a known raw
material composition based on phase relations in the four component system. In all equations the
chemical compositions are expressed in % wt. The quality of clinker is often referred to as the
amount of ‘free’ (non-reacted) CaO in the sample which is known to reduce the strength of concrete
(Bye 1983). Up to now many theoretically and empirical derived equations have been developed to
calculate the so called “Lime Saturation Factor” (LSF) (Eq. 1-9) (Taylor 1997; Macphee and
39
Lachowski 2007) which is used to quantify the amount of CaO in the raw material that can be
combined with SiO2, Al2O3 and Fe2O3 to form the four main clinker phases. For satisfactory clinker
quality the LSF value should be in the range of 92 – 98 % (Macphee and Lachowski 2007). Other
relevant parameters are the “Silica Ratio” or “Silica Modulus” (SR or SM)” (Eq. 1-10) and the
“Alumina Ratio” or Alumina Modulus” (AR or AM) (Eq. 1-11). The SR value is usually in the range
of 2 – 3 and describes the proportion of the silica phases to aluminate and ferrite phases. It also
reflects the ratio of solid phases (the silica phases) to the liquid phase formed by aluminate and ferrite
in the kiln. AR expresses the ratio between the aluminate phase and ferrite phase and indicates which
Eq. 1-9
Eq. 1-10
Eq. 1-11
A method used to determine the potential quantities of C3S, C2S, C3A or C4AF from the four major
oxides contained in cement raw meal is given by the Bogue calculation. Although this calculation is
known to over-predict alite content and is considered somewhat out-of-date, it is still a standard
method used within the cement industry. More detailed information on the Bogue calculation can be
One major advantage for the Ca-looping cycle over many competing technologies is that the
exhausted / purged CaO can be used as a raw material in cement production, presenting an
opportunity to partially decarbonise both power generation and the cement process. Cement
40
production is both a highly resource and energy intensive process. The energy consumption of
producing a tonne of clinker ranges from 3.1 to 7.5 GJ (CEMBUREAU 1999) with over 60 % of the
energy use in the process owing to heat input to the precalciner (EU 2001) to drive the endothermic
The cement production process emits between 0.6 and 1 kg of CO2 per kg of cement (ECRA 2007)
Figure 1-11. Mass balance on raw materials, fuels and CO2 in a typical
modern dry cement production process (RM = raw materials other than
CaCO3). Coal LHV 26.3 MJ/kg, 68.9% carbon content. Petroleum coke
LHV 34.2 MJ/kg, 87.2% carbon content. Data taken from (Alsop,
Hung et al. 2007).
A means to reducing the direct emissions of CO2 from the calcination process has been assessed
which involves driving the reaction using very hot CaO particles (T > 1273K) passed to the
precalciner from a separate combustor (Rodriguez, Alonso et al. 2008) (see Figure 1-12). This is
intended to replace the conventional method of driving the calcination reaction in the cement works
(i.e. in situ fuel combustion). Such a process would increase the energy requirement by 0.8 GJ / tonne
cement, but allows up to 60 % of the CO2 emissions to leave the system in a form suitable for storage
41
(following compression and purification) at an estimated cost of 19 USD / tCO2. This is shown to be
highly competitive with other figures for cost of CO2 avoided as determined in an economic study of
CO2 capture in the cement process using amines (144 USD / tCO2) and using an oxy-fired kiln
Figure 1-12. Mass balance on (a) a reference cement plant and (b) the novel CO2 mitigation strategy
proposed. Basis = 1 tonne of cement, LHV 25 MJ / kg, 10 % air excess (Rodriguez et al., 2008a).
Alternatively, the purged material from the Ca-looping system (for example if applied to power
generation) (Figure 1-1) which would consist predominantly of CaO (if removed at the calciner) with
a proportion of ash and calcium sulphate (the proportion of which would depend on Ca-looping
operating parameters) can be used in place of the CaCO3 which is usually the main constituent of the
cement feed. The ash content could also fulfil some or all of the cement aluminosilicate requirements
usually supplied in the clay or in additives. This means that the direct emissions of CO2 from the
calcination stage are avoided in the cement works, as is a substantial amount of the energy required to
effect the calcination. Basic calculations (Blamey, Anthony et al. 2010) indicate that a reasonable
purge rate of ~ 48.9 kg/s (Romeo, Ballesteros et al. 2008) would otherwise require 144MWth, simply
to calcine, thus theoretically reducing the energy required within the cement works for clinker
manufacture of 3.1 GJ / t by approximately 50%. As a result, both of the industrial processes undergo
significant decarbonisation, the energy required to produce clinker in the cement plant is almost
42
halved, the raw material costs of the cement process are vastly reduced and the issue of waste disposal
The inherent advantage of integrating cement manufacture with the Ca-looping process is clear.
However, the potential to use waste material from the cycle to fulfil the requirement for calcareous
material in the cement process may be limited by its chemical compatibility in relation to the demands
The factors potentially limiting or having implications for the use of the deactivated CaO are
discussed here. Firstly, due to the strong affinity of CaO to SO2 (Eq. 1-4), a proportion of the CaO in
the waste material will be in the form of CaSO4. The proportion will depend upon the Ca / C ratio
used in the Ca-loop, the composition of the fuel and the recycle / purge rate. The maximum sulphur
content of OPC in the U.S. is 2.5 % as specified by ASTM C150 (ASTM 2009); above this level,
sulphur leads to expansion and cracking of the cement paste upon hydration. (This expansion is
dependent upon the alumina content; therefore this value increases to 3 % for cements with higher
Figure 1-13. The effect of cement product SO3 content and Silica
Modulus (SM) on major phase formation (Bhatty et al., 2004).
43
The proportion of sulphur contained in the final cement product is also known to impact upon the
formation of the four major cement compounds which determine the strength properties of the
cement. An increase in sulphur content within the 0 – 3 % range has been shown to lead to a decrease
in alite (the cement compound which provides long term strength development) and an increase in
belite (the compound providing short-term strength development) irrespective of the Silica Modulus.
A desirable level of SO3 in OPC clinker is 0.6 % (Alsop, Hung et al. 2007). It should also be noted
that excess sulphur entering the cement works can lead to operational problems arising from cycles of
However, in modern cement plants reasonable sulphur levels in the system and the resulting
circulation can be achieved using a kiln by-pass line to remove unwanted vaporised components from
the kiln (in particular K2O, Cl and Na2O3 as well as SO3). Another potential issue is that of trace
elements released from fuel use in the Ca-looping calciner. The in-situ combustion of solid fuels is
likely to result in partial or total vaporisation of trace elements contained in the fuel. Repeated
exposure of the solids to the combustion environment over a long series of cycles could lead to a
build-up of certain trace elements in the solids, either through physical or chemical attachment to the
calcium oxide, calcium sulphate or the ash. These elements would therefore be present in the waste
A key concern for the cement industry when considering the use of a marginal raw material is its trace
element content, the incorporation of the elements into the clinker and their effects on the production
process and on cement performance (Bhatty 1995). Highly volatile elements such as cadmium,
mercury, selenium and thallium tend to leave the cement process either in the cement kiln dust (CKD)
or in the emissions and pose less of a concern for operation of the plant or for the final cement
product. Less volatile elements are likely to incorporate into the clinker causing a variety of desired
or undesired effects. For example, the presence of beryllium in quantities as low as 0.25 % has been
shown to severely affect the setting and strength properties of cement (Bhatty 2003) due to its impact
on alite crystals.
44
A summary of the behaviour of trace elements and their effects on alite productionis given in Table 1-
5. The issue of trace elements becomes particularly relevant when the use of modified sorbents in the
Ca-looping cycle are considered for integration with cement manufacture as they may contain high
In addition to chemical compatibility, both attrition and agglomeration of the exhausted material
could impact on its use as a raw feed to the cement process. Typically the cement feed is made up of
particles with the following size distribution: 88 μ (83%), 88–300 μ (15%) and >300 μ (2%) (Alsop,
Hung et al. 2007). If the material is too fine due to attrition the particles would become entrained in
the exhaust gas and lost in the CKD. However, if the feed is too coarse the load on the kiln would be
increased due to a reduction in material burnability (i.e., the ease by which the raw materials form the
clinker phases, measured in terms of the fuel consumption required to achieve these phases). This can
be overcome by subjecting the material to grinding, however a grinding stage would result in a loss of
the sensible heat provided by the purged CaO stream which could otherwise be used to reduce the
45
Table 1-5. Behaviour of trace elements in the kiln and their effect on cement phases and clinker
mineralogy, adapted from (Glasser, Bhatty et al., 2004)
Element Behaviour Impact on alite
Volatile and therefore
leaves in Cement Kiln
Dust. Can also
Arsenic (As) Reduces formation
incorporate in clinker as
low-volatile calcium
arsenates
Causes decomposition
Boron (B)
Stabilises
Replaces Ca in all
Reduces melt temperature and improves clinker
Barium (Ba) clinker phases except
mineralogy
ferrite
Beryllium (Be) Decomposes alite
Bismuth (Bi) Goes into alite
Forms volatile halides /
Reduces melt temperature
Cadmium (Cd) sulphates and leaves in
Improves burnability
CKD.
Produces dendritic belite
Chromium (Cr) Decomposes alite
Reduces melt viscosity
Can adversely affect alite and belite formation
Copper (Cu) Goes to ferrite
Lowers melt temperature
Potassium (K) Lowers melt temperature
Replaces Ca in alite and belite, forms solid solution
Lanthanum (La)
with alite, enhances clinkering
Lithium (Li)
Magnesium (Mg) Improves burnability
Manganese (Mn) Can replace Si and Ca in alite
Sodium (Na) Lowers melt temperature
Nickel (Ni) Goes to ferrite Replaces Ca in alite and stabilises monoclinic form
Incorporates in clinker
as calcium antimonates
Antimony (Sb) under oxidizing Reduces size of alite and belite crystals.
conditions and high
temperatures
Decomposes alite
Tin (Sn) Stays in clinker Produces dendritic belite
No effect in traces
Small amounts favour alite formation
Strontium (Sr) Large amounts cause belite formation
Promotes free lime formation
Decomposes alite to belite
Titanium (Ti) Goes to ferrite
Reduces melt temperature
Forms larger alite crystals
Vanadium (V) Goes into alite Produces ragged belite
Reduces melt viscosity
Tungsten (W) Reduces melt viscosity Forms large round alite and Type III Belite crystals
Yttrium (Y) Substitutes Ca in alite and belite
Modifies alite crystals, reduces free lime, improves
Zinc (Zn) Enters belite and alite
clinkering
Zirconium (Zr) Modifies alite and belite crystals
46
Chapter 2. Flowsheeting
The aim of this chapter is to find the change in thermal efficiency of 1 kg clinker (ηck) after integration
of a generic dry cement process with the Ca-looping process such that some of the CO2 from the
cement process can be captured. This has been achieved by first producing flow-sheets of the two
independent processes to then enable them to be integrated into a single process under different
realistic configurations.
Flowsheeting of the Ca-looping process has been carried out to enable calculation of the thermal
efficiency of the process (Hin / FCO2 MJ where Hin = the heat demands at the calciner and FCO2 = the
flow of CO2 to the carbonator in kmols,) under different sets of conditions. The most fundamental
parameters used as variables in this investigation are the sizes of the purge / replenishment streams
(F0) and solids recycle rate (FR) relative to the amount of CO2 coming into the carbonator (FCO2)
(Figure 2-1). CO2 capture efficiency of the carbonator (ECO2) and the average reactivity of the sorbent
( ̅ ) are also both important parameters that need to be calculated from F0 and FR; ̅ is also required
47
Figure 2-1. Basic flow-sheet of the Ca-looping System. (all flows in moles) Other nomenclature not
mentioned in the text: FN2 = N2 to the carbonator, ̅ = maximum average conversion of solids in the
system, μ=split of fuel to the calciner. Industrial flue gas CO2 concentration (fCO2)=15 % v/v.
Other flows of interest include the vent stream which is primarily comprised of the N2 which passes
through the carbonator without reaction plus the CO2 coming from the combustor which is not reacted
with the sorbent in the carbonator (FCO2*[1-ECO2]). The total amount of high purity CO2 leaving the
calciner is equal to the number of moles in the recycle reacted in the carbonator (FR* ̅ ) plus the
number of moles of CO2 entering in the replenishment stream (F0). The split of fuel use between the
combustor and the calciner is represented by μ where μ is the split of fuel going to the calciner.
Abanades obtained an equation for ECO2 for a steady-state system via a mass balance on the
carbonator, essentially as the amount of CO2 captured in the carbonator ([FR+F0]* ̅ ) divided by the
total amount of CO2 coming into the carbonator (FCO2+F0) (Eq. 2-1) (Abanades 2002).
̅
Eq. 2-1
48
Eq. 2-1 requires estimation of ̅ to determine how much of the CO2 is leaving in the recycle and
purge after conversion to CaCO3. ̅ is estimated through summation of the product of conversion as
a function of cycle number ( – See: Eq. 1.8, Background) and the mass fraction of particles for each
cycle (rk):
̅ ∑ Eq. 2-2
The mass fraction of particles for each cycle is calculated by performing a succession of cycle mass
balances to find the mass fraction of particles that have circulated ‘N’ times (otherwise referred to as
Eq. 2-3
Eq. 2-4
Eq. 2-5
As described, ECO2 is dependent on the age distribution rN. This value can also be calculated by
another means. A stochastic population balance model has been developed as part of this PhD to
calculate the age distribution rN using an iterative procedure. The model essentially utilises a vector
whose values represent the number of times a given quantity of CaO (a ‘packet’, e.g. 100 moles) has
been cycled through the Ca-looping system. On each cycle of the model, the existing packets have
their ages increased by one, whilst a fraction of the total CaO in the system is purged by removing a
given number of cells from the vector randomly. These cells are replaced with ‘fresh packets’ whose
49
Figure 2-2. Schematic diagram depicting the procedure for establishing an age distribution via the
stochastic model on each cycle.
As in the INCAR method described above, this method allows the age distribution to be calculated as
a function of FR and F0. However, the stochastic method has an inherent advantage in that it enables
the degradation in reactivity to be calculated on a time basis (as opposed to steady-state) and therefore
allows real data for carbonation kinetics to be used as opposed to simplified representations as will be
detailed below (see: section 2.2.3). The match between the results for rN up to fifteen cycles can be
Figure 2-3. Comparison of rN vs. cycle number for the two methods (analytical and stochastic).
As indicated in Table 2-1, the results for ̅ calculated using the stochastic method also match those of
the results reported by INCAR to 3 d.p. This is to be expected given that ̅ is dependent on rN. (NB.
The error between the values of ̅ is considered acceptable for the purposes of this project.)
50
Table 2-1. Results for comparison
of ̅ for the two methods as a
function of F0 and FR.
̅
F0 FR INCAR SPBM
0.1 3.55 0.1964 0.1965
0.2 2.29 0.3058 0.3055
0.3 1.83 0.3823 0.3824
0.4 1.61 0.4354 0.4351
This demonstrates that the stochastic model is also capable of calculating ECO2 to an acceptable degree
of accuracy from Eq. 2-1 given that the equation is dependent on ̅ . For the annonated Matlab code
According to (Rodriguez 2008) the total energy requirement at the calciner (MJ / moles CO2, Hin) is
composed of three terms and is represented by Eq. 2-6; the energy required to drive the endothermic
made up of the fresh limestone stream (F0) and the carbonated material returning from the carbonator
( ̅ ) (first term); the energy required to raise the temperature of the solids recycle coming from the
carbonator from 923 K to 1223 K ( ), where the recycle stream is made up of carbonated material
and, the energy required to raise the temperature of the fresh limestone from ambient temperature
( ̅ )
[ ̅ ̅ Eq. 2-6
51
Eq. 2-6 has been incorporated into the stochastic model. In (Rodriguez 2008), the results are reported
as the heat requirement at the calciner (Hin) as a percentage of the total system heat requirements (Hin
+Hcomb, where Hcomb is the energy supplied to the combustor). Essentially, the lower the fraction of the
energy being added to the system which goes to the calciner, the higher the efficiency of the overall
process (though it should be remembered that there would be recuperation of some of the heat added
in the calciner into the steam cycle). The plot also indicates the effect of reactivation lower than the
maximum conversion potential (possibly due to partial conversion in the carbonator) on the % heat
requirements at the calciner. Comparisons between the two methods indicate that the stochastic
model is capable of generating the same results as that of the INCAR method (Figure 2-4 a+b). This
is an expected result given Hin is essentially dependent on ̅ and that the values for rN and ̅ matched
a. b.
Figure 2-4. % Hin / (Hin + Hcomb) for a. The INCAR model; b. the stochastic model.
The next step was to use the stochastic model to find Hin for different modified sorbents, namely;
HBr-doped sorbent and steam-activated sorbent (data for both from Al-Jeboori, Fennell laboratory);
thermally pre-treated sorbent (Manovic and Anthony, 2008a); and precipitated calcium carbonate
(PCC) (Florin, Fennell laboratory). These all have degradation curves similar to that produced from
Eq. 1-8 (see: Chapter 1 section 1.1.1.1) except for thermally pre-treated sorbent which has a
52
a. b.
d.
c.
Figure 2-5. Degradation curves for a. Undoped Longcliffe, (15 % CO2, calcination at 900 °C,
carbonation at 700 °C); b. HBr-doped Longcliffe, (15 % CO2, calcination at 900 °C, carbonation at
700 °C) c. Kelly Rock samples hydrated by steam and preheated for 24 h at different temperatures. d.
Precipitated Calcium Carbonate (PCC) (15 % CO2 during carbonation, 100 N2 during carbonation).
All measurements except those for preheated Kelly rock limestone were fitted using equation 1-8. The
experimental measurements of carbonation extent were used directly for the Kelly rock. The results of
using Eq. 2-6 for the modified sorbents is given in Table 2-2 and Figure 2-6a-d. It can be seen that
the thermally pre-treated sorbent results in the lowest calciner heat demand of 24.50 %. However, the
energy required to thermally pre-treat the sorbent should also be factored in for a true reflection of the
53
Table 2-2. Values of Hin as a function of F0 for different sorbents.
Hin (% total)
0.1 217 (30.7) 262 (33.6) 207 (29.7) 177 (26.5) 205 (29.5)
0.2 224 (31.4) 248 (33.6) 222 (31.2) 203 (29.3) 227 (31.6)
0.3 244 (33.2) 259 (34.5) 244 (33.2) 231 (32.00) 251 (33.9)
0.4 268 (35.3) 279 (36.3) 269 (35.4) 258 (34.5) 277 (36.0)
a. b.
c. d.
Figure 2-6. a. Undoped Longcliffe (50 cycles) b. HBr-doped Longcliffe (50 cycles) c. Thermally pre-
treated / steam pre-activated Kelly Rock (30 cycles) (pretreated for 24 hrs at 900 °C) d. Precipitated
calcium carbonate (30 cycles).
54
The undoped longcliffe has the highest minimum calciner heat demand of 33.54 %; a direct
comparison can be made with Figure 2-6b which shows the results for the HBr-doped Longcliffe. It
can be seen that doping the sorbent with HBr brings the heat demand down from ~ 33.54 % to ~ 29.51
%. This can be attributed to the fact that doping the sorbent results in a higher value for the final
reactivity ( ) and therefore a smaller purge / replenishment stream (F0 / FCO2) can be used (0.064
rather than 0.18) whilst achieving the same value of ̅ . Both the thermally pre-treated and the
precipitated calcium carbonate show low heat demands of 159.55 and 192.22 MJ respectively (Table
2-3). This too can be attributed to the fact that they enable the size of the purge / replenishment
stream to be reduced to a near negligible amount (0.0092) whilst maintaining the same level of
reactivity. However, it is important to note that such negligible purge rates may not be achievable in
Table 2-3. Optimum parameters to minimise calciner heat demands for different sorbents.
Thermally / steam pre-treated Kelly Rock N/A 0.0092 0.19 159.55 24.50
These results are of interest to cement manufacture because they will enable the overall energy
penalty of the CO2 capture process to be reduced upon integration. It is also pertinent to mention that
the use of doping agents such as Br will have implications on the both the final cement product (i.e. as
a trace or minor element) and also as a potential source of increased corrosion on steel components.
However, the doping concentrations used are very small (Al-Jeboori, M. J. et al. 2013).
55
2.2.3. Incorporating Carbonation Kinetics and Bed Inventory (WCaO)
The next step in the work by INCAR was to incorporate bed inventories and carbonation rate kinetics
into the calculation of ECO2. In their work the following assumptions were made: instantaneous and
perfect mixing of solids in both the carbonator and calciner; plug flow for the gas phase in the
carbonator; and, instantaneous and complete calcination of particles in the calciner. It is assumed that
the bed contains NCa moles of CaO but only a fraction (fa) reacts in the bed at a rate rave (average
reaction rate of the active material, s-1). The remaining fraction of bed particles (1- fa) are considered
inactive. Simultaneously, there is a continuous arrival of newly calcined particles (FR) that are
To facilitate integration of the reaction rate and the residence time distributions it was assumed that
the CaO particles attain the maximum conversion XN at a constant rate in a characteristic time, t*;
after that the reaction rate becomes zero (Figure 2-7 and Eq. 2-7) in line with theory regarding a fast
reaction rate followed by a slow reaction regime controlled by CO2 diffusion through the product
layer formed on the free CaO surfaces as described in Chapter 1 (see: Chapter 1, section 1.1.1.1):
56
if t < t*
{
Eq. 2-7
if t > t *
This equation implies that particles with a residence time above t* reach their maximum average
conversion XN, and that their reaction rate is zero from that point onwards. It is argued here however
that this is an over-simplification of the actual carbonation rate and extent of carbonation (maximum
conversion) as derived from experimental data. The stochastic model described previously has been
adapted to enable integration of real experimental data for carbonation rate (Figure 2-8) with the
residence time distribution by picking out values from vectors of carbonation conversion over time
depending on how long the particle or packet has spent in the carbonator.
t*=~80
This required two separate vectors to be created; one for the carbonator and one for the calciner. In
each ‘second’ of the simulation, packets are passed around the system (i.e. between the vectors) using
three stochastic functions; two between the carbonator and calciner enabling packets to be passed
between the reactors at random, and a third function purging packets at random (as in the previous
57
Figure 2-9. Flow-sheet of the stochastic model incorporating bed inventories into the calculation of
ECO2.
During the course of the simulation the residence times of packets at the carbonator are recorded
depending on the number of seconds they have spent in the reactor without being passed to the
calciner for regeneration. Packets which are passed to the calciner and returned to the carbonator
have their ages increased by one to effect the degradation of reactivity over repeated cycles. This
enables the model to ‘pick out’ values from the carbonation rate (e.g. Figure 2-7) based on the
residence time at the carbonator and the number of cycles. ECO2 was then calculated as the number of
moles of CO2 reacting in each second (i.e. by summing all active packets in the carbonator based on
their age and residence time) divided by the number of moles of CO2 coming in from the combustor.
These results were then compared against those of the INCAR method. It was not possible to acquire
exact values for ECO2 calculated via the INCAR method for reasons of confidentiality (Alonso 2013).
However approximate values have been derived based on plots published in (Alonso, Rodriguez et al.
2009). The next step was to incorporate carbonation conversion from experimental data (Figure 2-
10), where F0 = the purge / replenishment rate, FR = the solids recycle rate and WCaO = the bed
inventory.
58
59
Figure 2-10. Comparison of ECO2 values for the two methods. Blue bars refer to the results of the
published INCAR work; green lines to the stochastic method using the INCAR kinetic model; and,
red lines to the stochastic model using experimental data.
The results comparing the stochastic model with the INCAR method indicate that the model yields
results similar to that from INCAR; at higher purge rates (F0 = 0.1) the stochastic model outputs
values for ECO2 very similar to those in the INCAR results, whilst at the lower purge rates (F0 = 0.01)
the results for ECO2 are somewhat higher. However these differences can be attributed to the fact that
a different modeling method is being used; one is based on an arithmetic series whilst the other is
based on the random movement of data between two vectors. The results for the stochastic model
using experimental data are consistently lower than that of both the INCAR results and the results of
the stochastic model using the INCAR model. This is an expected result given that the equivalent
value of t* in the experimental data is close to 80 (marked with a red line on Figure 2-8) as opposed to
the 20 seconds used in the INCAR model. This means that it takes around 4 times as long for a
packet to reach the maximum conversion resulting in lower carbonator CO2 capture efficiencies. For
the annotated Matlab code for this stage of the model please see Appendix 2-2.
60
2.2.4. Examining system start-up
As well as enabling real carbonation data to be used, the stochastic model also enables study of the
start-up of the system, given that it operates on a time basis rather than being based on steady-state
calculations. For example, Figure 2-10a+b show the CO2 capture efficiency of the carbonator as a
function of time.
a.
b.
The red lines in Figure 2-11a+b indicate the steady-state values of ECO2 calculated by the stochastic
model for fixed values of WCaO and F0 and two FR values (FR = 10 and 1 respectively). These plots
show that not only does a lower recycle rate reduce the carbonator efficiency from 0.44 to 0.15, it also
61
increases the time taken for the system to reach steady-state (defined as less than 1 % change from
one second to the next) from ~ 20,000 seconds (~ 5.5 hours) to ~ 100,000 seconds (over 1 day) (Table
2-4).
Finally, an added advantage of using the stochastic method is the ease by which extra units can be
The main purpose of this chapter is to find the thermal efficiency of producing 1 kg clinker (nck) for a
cement plant integrated with the Ca-looping cycle under different configurations. To do this it was
first necessary to construct a mass and energy balance for a generic dry cement process. This was
developed using Microsoft Excel on a basis of 1 kg cement clinker leaving the clinker cooler before
the grinding stage. A description of how the mass and energy balance has been constructed is
provided here.
62
2.1.1. Cement Plant Mass and Energy Balance
The main flows of a modern heat-integrated cement production plant are described in the schematic
diagram of Figure 2-12 which shows an integrated dry cement process flow-sheet consisting of four
main stages; preheater, precalciner, rotary kiln and clinker cooler. Air entering via the clinker cooler
is heated by the hot clinker leaving the kiln and is passed to the kiln and calciner as preheated
combustion air. O2-depleted air leaving the kiln is then used to preheat calcined solids entering the
kiln. Lastly, O2-depleted air leaving the calciner is used to heat the solids in the cyclone preheater
section. Solids enter at the preheater and are then calcined at the precalciner before being passed to
the kiln. The model described here ignores the bypass gas and bypass dust lines as these do not
The total air input into the system is comprised of air supplied to the cooler (FAIR,CO) (2.81 kg / kgck),
primary air supplied to the kiln (FAIR,P) (0.175 kg / kgck) (both taken from (Alsop, Hung et al. 2007))
and false air infiltrating at the preheater and combustion stages (FAIR,(PH), FAIR,(PC), FAIR,(KN)). The
values for false air were taken from (Kiln_Performance_Tests_Task_Force 1992); it is possible to
calculate false air using a modified Bernoulli equation (see below, section 2.1.1.3) however some of
the values required for this equation were unknown and were estimated. The division of FAIR,CO
between kiln and calciner is dependent on the fuel / air requirements at the respective combustion
stages. The calculation of heat demands is given below (2.1.1.2). The split of heat demands between
the kiln and precalciner therefore dictates the split of air leaving the cooler. The value for primary air
is subtracted from that of the secondary air to ensure the amount of total air supplied to the kiln is
sufficient but not in excess; this would otherwise lead to a lower thermal efficiency at the kiln. The
remaining air from the cooler after the secondary and tertiary air is subtracted to give the value for the
amount of cooler vented to atmosphere (FAIR,V). The flow of gas from the kiln to calciner and from
63
the calciner to the preheater consists of O2-depleted air (N2 and CO2 from combustion of fuel at the
kiln).
Figure 2-12. Flow-sheet of a cement rotary kiln system including cyclone preheater
section, precalciner and tertiary air duct.
64
2.1.1.1. Calculation of solid streams through the system
The size of raw material solid streams through the system required to produce 1 kgck (limestone,
sand, iron ore) were calculated using the Bogue equation (see: Background) with a target alite and
belite composition of 70 % and 20 % respectively, and key parameters set accordingly (S / R = 2.5; A
/ F = 3). The change in the composition of the solids stream between precalciner and kiln was
The energy demand at the kiln ( , Eq. 2-8) is composed of 5 terms; the energy required to
raise the temperature of the calcined raw meal stream ( ) from calciner temperature (1223 K) to
sintering temperature (2073 K) (see below for calculations of the size of raw meal flows); the energy
required to raise the temperature of incoming air streams from ambient (298 K) to kiln temperature
(primary air; false air) (see below for calculation of raw meal flow); the energy required to raise the
temperature of the secondary air from the temperature at which it leaves the cooler (assuming heat
losses of 30 % (Alsop, Hung et al. 2007)); the energy required to raise the temperature of the fuel
from ambient (298 K) to sintering temperature; energy losses at the kiln (radiation and convection,
from (Alsop, Hung et al. 2007)); and, the exothermic clinkering reaction (-355 kJ / kg):
( ) ( )
Eq. 2-8
( )
As such, the calculation for Eq. 2-8 required an iterative calculation given that the fuel and air
requirements are dependent on each other. Based on this calculation the kiln energy demands were
65
Next, the calciner energy demands ( ) were estimated using Eq. 2-9 comprised of eight terms;
the energy required to raise the temperature of the incoming uncalcined raw meal; the energy required
to raise the temperature of the incoming O2-depleted air; the energy required to raise the temperature
of the CO2 coming in from the kiln; the energy required to raise the temperature of the tertiary air for
combustion; the energy required to raise the temperature of the false air; the energy required to raise
the temperature of the fuel from ambient to calciner temperature; energy losses (radiation and
convection) and the energy required to drive the endothermic calcination reaction:
( ) ( )
( ) ( )
Eq. 2-9
( ) ( )
Eq. 2-9 also required an iterative calculation and gave a result for the calciner heat demands of 2.96
MJ / kgck. Based on these calculations the thermal efficiency of producing 1 kg of clinker (ηck) was
estimated to be 4.34 MJ, with almost 70 % of the total heat demand at the calciner. Notably, this is a
higher total value than that of the most efficient processes, residing in the top 90 % percentile of kilns
globally (WBCSD 2006). Comparison of key values in the mass balance against those of a reference
cement plant (Kiln_Performance_Tests_Task_Force 1992) indicate that this high final value is mainly
due to the size of the tertiary air stream entering at the calciner which is nearly double that of the
The fact that the calciner heat demands comprise 70 % of the total energy demands (as opposed to the
usual 60 %) suggests it is this value which causes the thermal efficiency to be lower than that of the
66
Table 2-6. Comparison of key values in the mass balance against those of a reference case.
The volume flow of infiltrated air at the kiln hood (V,FA,KN) can be estimated using a modified
Bernoulli equation which includes a correction factor for the assumptions of frictionless flow, Eq. 2-
̇ √ Eq. 2-10
However some of these values are unknown (e.g. the cross-sectional area of openings at each of the
respective pyro-processing stages) and so the values for false air have been taken directly from the
Precalciner 0.22
Kiln 0.04
67
2.4. Integration of Cement Plant and Calcium Looping Plant Flow-sheets
The aim of this section was to begin to investigate the best configuration for the integration of a
cement plant with Ca looping. For all configurations the following changes to the cement plant / Ca-
Loop flow-sheets apply. The precalciner unit at the cement plant is replaced with a dual-CFB calciner
and carbonator setup. Given that the calciner must be oxy-fired, the enthalpy associated with the hot
tertiary air previously supplied to the precalciner is available elsewhere (e.g. for preheating of raw
materials, fluidising of carbonator solids or for driving a steam cycle). There is also no longer false
air associated with the calciner. In all configurations the kiln flue gas is passed to the carbonator
rather than the calciner to enable capture of CO2 arising from fuel use at the kiln and to fluidise
carbonator solids. Limestone is added to the dual-CFB either via the carbonator or calciner, separate
from the raw meal which is added to the system either via the preheater or directly to the kiln. (NB.
In this model it is assumed that fresh limestone is added to the calciner according to Eq. 2-6). Lastly,
CaO sorbent purged from the calciner is passed to the kiln separate from raw meal stream. This last
point however requires some clarification; current Ca-looping pilot-scale demonstrations have used
sorbent particle sizes larger than the 90 μ required for formation of calcium silicates (Rodríguez,
Alonso et al. 2011); CanMET, 400 – 800 μ; INCAR, 130 – 180 μ; IFK, 350 μ. However particle sizes
as low as 50 μ are permissible in CFBCs (Sun, Grace et al. 2008). It is therefore worth pointing out
that a grinding stage between the CFB calciner and kiln may be required which would cause
complications in terms of the energy supplied to the kiln via the hot raw meal stream. For the
purposes of this investigation it will be assumed that the size of CaO sorbent particles purged from the
calciner to the kiln match that of the 90 μ required for formation of calcium silicates in the kiln. It is
important to note that though the energy demand of the base-case scenario (i.e. a basic cement plant)
is somewhat higher than that of the most modern and efficient kilns, it is the relative differences in
energy demand between different potential configurations with Ca-looping that are most important,
68
For all configurations the cement plant mass balance was scaled up from 1 kgck to 46.29 kgck /
second; this is equivalent to a 4000 tonne clinker / day cement plant. This is to ensure that the correct
amount of CO2 was entering the carbonator from the kiln (0.13 kg / kgck equating to 0.14 kmols -1).
The CaO sorbent purge size was fixed at the rate of CaO required to supply the amount of CaO
needed for a cement plant of this size (0.45 kg / kgck or 0.55 kmols-1) giving a F0 / FCO2 value of ~ 4.
Different values of FR / FCO2 and WCaO were tested to achieve a minimum heat demand at the calciner
whilst capturing the maximum amount of CO2 (~ 0.12 kg CO2 / kgck or 126 mols-1) assuming an ECO2
value of 92 %. The calciner heat demands under this configuration for different bed inventories
200 1 166.05
10 2 167.43
2.45 3 167.43
1.75 4 167.65
1.5 5 169.45
Values for Hin under this configuration range from 166.05 – 169.45 MW as a function of WCaO and FR.
Next, for the three configurations described below it was first necessary to calculate the enthalpy of
69
2.4.1. Configuration 1
In the first configuration, all waste heat streams (the high concentration CO2 stream, FCO2; the CO2-
depleted carbonator flue gas, FFGCB; waste cooler air, FTA+FVA) are used to drive a steam cycle. In this
configuration, the cooler vent air normally released to atmosphere (constituting an unused waste heat
stream) can also be used; this is not the case in a standard cement plant as the vent air on its own
would not provide the energy needed to drive a steam cycle. In this configuration, raw meal is passed
It was found that introducing the raw meal to the kiln from ambient increased the energy demands at
the kiln by 0.86 MJ / kg from 1.38 to 2.24. This equates to a kiln energy demand of 103.69 MJ / s for
a 4000 tonne / day cement plant. Therefore the system as a whole including the precalciner (operating
with the optimum values indicated in Table 2-8 operates at 269.74 MW, with ~ 2 / 3 of the total
However this configuration makes available four streams for the powering of a steam cycle.
Assuming a waste heat boiler efficiency of 22 % (Babcock and Wilcox 1978), the enthalpy of these
streams (Table 2-9) could supply 56 MW of electricity. This is an excess above the ~ 20 MW (Alsop,
70
Hung et al. 2007) required for a cement plant. Therefore ~ 36 MW of electricity could be supplied
back to the grid or used elsewhere for other nearby industrial processes.
2.4.2. Configuration 2
In this configuration, all waste heat streams are used to preheat the raw meal (Figure 2-14).
It was not possible to calculate the temperature of the solids and gases passing through four preheaters
due to there being too many unknown values (NB. Previously these values had been taken from
literature (Ahn, Ozcan et al. 2011)). Instead, by using Eq. 2-11 the heat transfer was modelled as a
single unit making it possible to calculate the temperature of the raw meal coming into the kiln, where
Qo = the enthalpy of the outgoing hot gas streams, Cp(RM) = the specific heat capacity of the raw meal
(kJ / kg / K), mRM = the mass of the raw meal entering the kiln (kg) and ΔT = the change in temperature
of the raw meal from ambient (K). Assuming a heat transfer efficiency of 70 % (Alsop, Hung et al.
71
2007) the final temperature of the raw meal before entering the kiln was calculated to be 1063 K, 513
̇ Eq. 2-11
This brought the kiln energy demand down from 1.38 by to 0.9 MJ / kgck or 41.66 MW. This
equated to a total energy demand of 207.71 MW with ~ 80 % of the energy demands at the Ca-loop
calciner.
2.4.3. Configuration 3
In configuration 3 the aim is to produce enough electricity to power the cement plant with other
streams being used to preheat the raw meal to bring down the kiln energy demand as much as
possible. Again, assuming a waste heat boiler efficiency of 22 %, this would suggest that the enthalpy
required to drive a steam cycle to produce the 20 MW of electricity required at the cement plant
would be at ~ 80 MJ / s. This would suggest the vent air (70.29 MJ / s) plus the carbonator off-gas
(17.16 MJ / s) would be the best candidates with a combined enthalpy of 87.45 MJ / s. As such these
two streams would provide 19.29 MW of electricity. (NB. Although this efficiency is low it would
not be realistic to add in a highly efficient cycle as it would only be one turbine).
72
Figure 2-15. Integration of cement manufacture and Ca-looping: Configuration 3.
This leaves the remaining tertiary air and calciner off-gas (high purity CO2) to preheat the raw meal.
In this configuration the raw meal is preheated to 957.47 K, whilst bringing the kiln energy demand
down from 1.38 MJ / kg to 1.03 MJ / kg. (NB. It was not possible to calculate the temperature of all
streams through the preheater section due to there being too many unknowns; as such the preheater
A summary of the results of the integration of the two processes is given in Table 2-10 below:
73
Table 2-10. Summary of cement plant and Ca-looping integration results.
Total energy Elec.
demands demand
Kiln energy demands
from the
Notes
grid
MJ / kg ck
Config. (MJ / kg ck) MW MW (MW)
(ηck)
Provides most amount
of excess electricity and
1 1.382.24 103.69 4.345.83 269.74 ~ – 36
removes need for the
preheater section.
Reduces the kiln energy
demand by the largest
2 1.380.9 41.66 4.344.49 207.71 ~ 20
amount but provides no
electricity for the plant.
Preheats the raw meal to
~ 1000 K bringing the
kiln energy demand
3 1.381.03 47.68 4.344.62 213.73 ~ 0
down by 0.35 MJ / kg
and supplies enough
electricity for the plant.
All three configurations capture 92 % of the CO2 arising from the burning of fuel at the kiln which
equates to ~ 0.12 kg CO2 / kgck (or 126 mols-1) whilst 0.01 kg CO2 / kgck (or 11 mols-1) are lost to
atmosphere.
2.5. Conclusions
The Ca-Looping system has been modelled using a stochastic population balance model, a
method which is able to garner similar results for the age distribution (rN), the maximum
average reactivity of the sorbent ( ̅ ) and the CO2 capture efficiency of the carbonator (ECO2)
This method has been furthered to run on a time basis as opposed to per cycle and to
incorporate the carbonation rate and extent of carbonation from experimental data rather than
74
A generic dry-process cement plant mass and energy balance has been developed which
resulted in a figure of 4.34 MJ for the thermal efficiency of producing 1 kg of clinker. This is
within the top 90th percentile of kilns globally; it is thought that this lower thermal efficiency
Integration of the Ca-looping process with the cement plant indicates that the thermal
efficiency of producing 1 kg of clinker can range from 4.49 – 5.83 MJ depending on whether
the system supplies itself with electricity through waste heat recovery or whether electricity is
supplied from the grid. Up to ~ 36 MW electricity may also be passed back to the grid,
however this results in the lowest thermal efficiency of producing 1 kg of clinker of 5.83 MJ.
For all configurations 92 % of all CO2 produced at the kiln is captured which equates to ~
75
Chapter 3. Production of spent sorbent
Chapter 2.
Chapter 3.
3.1. Chapter Summary
Sorbent purged from the Ca-loop is likely to differ in chemical composition to that of CaO passed
directly to a cement kiln due to an increase in exposure to a combustion environment over a given
number of calcination / carbonation cycles. Spent sorbent has been produced in the laboratory using a
The core components of the furnace consist of a 508mm ID Incoloy 800HT pipe, heated electrically
by a high-output transformer (1600 A, 2.5 V) via copper electrodes (dimensions 300 x 50 x 5 mm)
clamped at the top and bottom of the heating zone, and two sets of ten copper cables (75mm2 cross-
section, 800mm length) (see Figure 3-1). The electrodes are positioned to ensure that the centre of the
bed of limestone (70 mm in height when fluidised) sits at the hottest part of the reaction vessel (see
arrows on Figure 3-1). The power to the electrodes is controlled by a hardware-based power control
unit, operated in turn by a purpose-written program, written in Agilent VEE™ Pro 7.0 software. The
temperature of both the bed and the outer reactor wall are measured at points along the bed and wall
using type ‘K’ thermocouples and is monitored / controlled by this program via PID control. The tube
and the electrodes are insulated with three layers of Superwool® 607 Max blanket to minimise heat
losses from the system. The heat shrink used on the copper cables and the rubber sheath closest to
the electrodes are water-cooled by a series of soft copper tubing coils (1 / 8” o.d.), tightly bound
around the cable for approximately 50mm along the length of the cable. The quartz reaction vessel
sits inside the Incoloy tube as a sliding fit, supported from underneath by various means depending on
the configuration. The dimensions of the reactor were chosen to allow 100 g limestone to be calcined
76
in a single run to enable the maximum amount of clinker to be produced as limited by the size of the
higher temperature horizontal furnace (see Chapter 5, section 5.3.1). Threads were also made at the
top and bottom of the Incoloy pipe to enable the furnace to be used as a closed system if required.
Figure 3-1. Core components of the vertical tube furnace (fluidised bed design).
Gases are supplied to the base of the bed by cylinders via calibrated rotameters. The off-gas is drawn
Given that the required flow-rate to the spectrometer is only 1 l /m, a bleed valve is positioned
between the spectrometer and pump to enable the flow-rate to be finely controlled, as measured by a
77
calibrated rotameter. Gases are drawn through quartz wool and CaCl filters to minimise reactor fines
and moisture passing to the spectrometer gas cell. Initially, a piece of ¼” pipe bent to a right angle
was positioned at the top of the reactor serving as the feed to the off-gas line towards the gas analysis
apparatus. However, this was later replaced with a quartz piece to enable the reactor fines to be
The CO2 concentration at the off-gas is measured using a Perkin Elmer Spectrum 100 spectrometer
and is controlled by Perkin Elmer Timebase software. To calibrate for CO2 concentrations of CO2
ranging from 5 – 50 % are fed into the rig using N2 as the balance and the corresponding absorbance
recorded. The wavelength range used for measuring CO2 was 3619 ± 50 cm-1. (The larger peak at
2350 cm-1 cannot be used; though offering greater sensitivity, clipping occurs at the top of the peak
Absorbance vs. concentrations is plotted and a fit of the line used to convert absorbance to
78
Figure 3-3. Calibration Curve for CO2.
experiments (for the code please see Appendix 3-1). The FTIR is also calibrated for carbon monoxide
(CO) at the 2025 cm-1 peak to enable some understanding of the combustion characteristics of the
Initially, coal experiments were performed using a spouted-bed design to allow coal to be fed into the
bed batch-wise from underneath via a compressed air “bullet” feeder whereby coal granules are
injected into the base of the bed by opening and closing two valves (Figure 3-4).
79
Figure 3-4. Schematic of batch ‘bullet’ coal feeder.
This was later replaced with a “rotary hopper” feeder to enable continuous feeding of coal over a
given number of minutes. This feeder consists of a quartz hopper made from a modified titration
tube. A plastic rotary cup feeder sits at the base of the hopper, rotated by a motor. Coal granules fall
into a quartz feed line, swept in by air at a flow-rate of 1 l / m. This flow-rate is used as a
compromise between that needed to sweep the granules into the bed against the incoming fluidising
gas, whilst minimising the effect of the sweep gas impacting on the heat requirements of the system.
The fuel feed line is encased inside a second quartz tube to allow cooling water to be passed along the
length of the feed line. The vertical position of the fuel feed line is again set as a compromise such
that the granules drop deep inside and not on top of the bed, whilst minimising the effect of the
cooling water (3 l / m) on the heat requirements of the system. This design enabled coal granules to
be fed continuously without concerns regarding contamination from metal components (Figure 3-5).
80
Figure 3-5. Schematic of continuous coal feeder.
Given that the incoming fluidising gas mainly enters through the central part of the bed (from an ariel
view) the fuel feed line is positioned off-centre to ensure that the fluidising gas does not prevent the
sweep-gas from entraining the granules into the bed. The fuel hopper is attached to the feed line using
electrical heat-shrink insulation material. During the cycling experiment, fuel is fed from the point at
which calcination is detected based on the initial rise in CO2 concentration. Coal is fed at a rate of 0.5
g / minute over a period of 16 minutes (8 g of coal in total for each cycle) based on flowsheeting work
which predicted the ratio of ash to limestone in each packet of the stochastic model where ash was
converted to fuel based on the ash content of the fuel being used (for an explanation of ‘packets’see:
81
3.2.4. Modifications for Refuse-derived Fuel (RDF): feeder design
The feeding of RDF at the laboratory-scale is problematic due to its heterogeneity and the size of the
however this produces a fluffy / dusty sample which cannot be fed by either of the two feeders
previously described. By pressing the homogenised sample into a pellet, it is possible to produce a
solid sample. A novel feeding apparatus was developed as part of this PhD (Figure 3-6) which
enabled the solid sample to be fed from underneath by operating the system as a spouted bed.
The spout consists of a 3/8” OD tube contained within a ¾” OD tube. The outside tube has a 1/4” tube
welded at the base; the fluidising gas passes through this tube and through the gap between the 3/8”
and 3/8” tube and into the base of the quartz liner. This way, the incoming gas serves two vital
82
functions. Firstly, it allows the fluidising gas to reach the bed whilst leaving the 3/8” tube free for the
pellet to be inserted without the issue of the fuel blocking the incoming gas. Secondly, by having the
fluidising gas travel up the outside of the fuel feeding tube, the gas serves as cooling air, hindering
any potential for combustion / pyrolysis of the fuel before reaching the bed. A 1/8” water-cooling coil
is also included in the design between the spout and the inside of the reactor tube. A distributor plate
at the top of the spout secures the fuel and gas lines, whilst preventing limestone particles from falling
through the gas line. A quartz rod with a bevelled end, secured by the screw cap at the top of the
reactor in the centre of the bed, is positioned such that it sits snugly inside the fuel line, preventing
As the fuel is pushed into the bed from underneath, the quartz rod lifts just enough to allow the fuel to
enter the bed at the desired rate, to be replaced by pushing down on the rod from above once all the
fuel has been administered. The valve at the base of the spout is bored through, allowing the fuel
“cartridges” (which require replacement on each cycle) to be changed whilst keeping the fuel feed line
flush along the inside to prevent blockages. Lastly, the RDF is pressed into pellets 6mm wide (2mm
narrower than the fuel feed tube ID) and 60mm long using a mould and press. The fuel can be
administered in a semi-continuous mode by using a lab jack to raise the rod and pellet manually.
It was found however that pellets needed to be used on the same day that they were made; leaving
them overnight led to deformation such that they wouldn’t fit into the feeder. As such, a limitation
was placed on how much fuel could be fed in a cycling experiment based on how many pellets could
be produced in a morning enabling the cycling experiment to be carried out in the afternoon.
83
3.3. Design of Experiments
The size of limestone particles used in these experiments was 500 – 710 μm, chosen such that the
mean diameter of bed particles was within the range of that used in commercial fluidized-bed
combustors (0.5 – 1 mm) (Lanauze 1985). It was necessary to choose a flow rate which fluidised this
size of particle without introducing unnecessary heat demands to the system. The minimum fluidising
velocity was determined both through the use of an equation from the literature and empirically. To
determine this value through theory, first the Reynolds Number at minimum fluidising conditions
(Remf) was calculated using Eq. 3-1 proposed by Wen and Yu (Wen and Yu 1966).
[ ] Eq. 3-1
where:
( ) Eq. 3-2
Where: dp is the particle diameter, ρf is the fluid (gas) density, ρs is the particle density, g is the
The superficial velocity at minimum fluidising conditions (Umf) can then be calculated from Eq. 3-3:
Eq. 3-3
Fluid viscosities for N2 were taken from Nist Webbook (NIST 2011) with a density of limestone of
2600 kg / m3. Eq. 3-3 indicated a value of 0.02 m / s which, given the cross-sectional area of the
84
Next, was determined experimentally by gradually increasing the flow rate to the bed at 1 l / m
intervals and observing the point at which the bed started to fluidise. This method indicated that at 2 l
/ m the bed started to rise and particles started to circulate. This equated to a value of 0.012 m / s
for the size of reactor, nearly half that of the value calculated via Eq. 3-3. It was decided that the
value of used would be taken from the empirical results and that a U / Umf value of ~ 2 would be
used to ensure bed fluidisation whilst minimising the heat demands of the system which would
increase to ~ 3 U / Umf at the higher calcination temperatures. Therefore the total flow of gas into the
The U / Umfwas kept the same as that of the ‘No Fuel experiments (U / Umf = 2) and was deemed high
enough to ensure rapid mixing (< 1s) of two media of differing size and density (Fennell, Davidson et
al. 2005). Assuming a coal density of 1300 kg / m3, this would indicate a particle size of 0.71 – 0.85
mm according to the correlation by Wen and Yu detailed in the previous section to match the U / Umf
value of the coal with that of the limestone. Experiments which observed the mixing of the two media
under cold conditions indicated that the particles mixed rapidly at these two particle sizes.
The quartz liner is first rinsed with deionised water and dried overnight. The supporting tube is put
into position at the base of the reactor. A type K thermocouple is secured such that the tip is located
at the centre of where the bed will be in both horizontal and vertical planes (35 mm from the top of
the sintered disc for a bed height of 70 mm). The bed temperature set-point is set to 900 ºC. The off-
gas line is secured in place at the top of the quartz liner such that the top of the line sits 50 mm into
the top of the quartz liner. Both sets of software (Agilent VEE and Timebase) are started
85
simultaneously to allow control of the system and to enable plots to be made at a later stage. A 15 %
CO2 calibration is passed into the system via a rotameter before then passing 15 % CO2 from a 100 %
CO2 cylinder with N2 as the balance. A 100 g batch of 500 – 710 μ Longcal P25 is dropped into the
liner to form the bed which causes the bed temperature to drop to below 400 °C (see Figure 3-7).
Calcination starts to take place which is inferred from the increasing CO2 concentration in the off-gas.
Figure 3-7. CO2 concentration and system temperatures for ‘no fuel’ cycling experiments.
Notably, for the majority of calcination in the first 5 cycles, the temperature of the bed rose gradually
from the temperature at which calcination started (~ 810 °C) towards 850 °C. At this point the
temperature rose quickly above the set-point of 900 °C. Some degree of temperature control was
required to prevent the wall temperature from going above 1000 °C (the temperature at which Incoloy
becomes vulnerable to shape change) as indicated with an asterisk on cycle three of Figure 3-7. This
was achieved by adjusting the voltage on the front of the power controller. On the fifth calcination
this was achieved by reducing the set-point of the bed to 875 °C (marked with a hash on Figure 3-8.
It can also be seen that on the third cycle by reaching the maximum wall temperature of 1000 °C that
the safety alarm on the power controller was triggered causing the power to cut out leading to a
temporary drop in CO2 concentration (marked with a double asterisk on Figure 3-7). For some
unknown this did not occur on the first calcination despite the wall temperature reaching 1000 °C.
86
End of calcination is defined as the time at which the CO2 concentration is seen to return to the 15 %
baseline. After 60 s at the baseline, the temperature set point is set to 650 ºC to affect the carbonation
step. This process is repeated for a given number of cycles. After the last calcination, the CO 2 is
switched off and the bed allowed to return to ambient temperature. The quartz liner is then removed
For the batch coal experiments, coal is fired into the base of the bed as described in section 3.2.3. As
can be seen in Figure 3-8, firing the 2 g coal charge into the bed causes the CO2 concentration to
increase from that coming from calcination to between 30 and 35 %. This also causes the bed
temperature to increase to that of the temperature set point (900 ºC), which in turn causes the
temperature to drop rapidly due to a loss of power given that the set point has been reached.
Figure 3-8. CO2 concentration and system temperatures for cycling experiments. La Jagua coal
(batch-fed).
In the case of the continuously fed coal experiment, coal is fed in gradually by switching on the
motor. The sweep gas (air) and water cooling are also both switched on simultaneously. The
87
introduction of fuel can be seen to increase the bed temperature but an increase in CO2 concentration
is not observed given that the fuel is being fed in gradually. However once calcination is complete the
steady addition of the fuel is observed until coal is no longer fed. At this point the CO2 concentration
returns to the 15 % baseline. An rapid increase in CO2 observed at the start of calcination indicates
however that the fuel is being fed from the start of calcination. The low CO (CO not shown in Figure
3-8) concentration also indicates that near-complete combustion of the fuel is taking place.
In the RDF experiments, the feeding of fuel is only semi-continuous. The effect of adding the fuel in
this way was somewhat erratic due to the heterogeneous nature of the fuel, leading to surges in bed
Figure 3-9. CO2 concentration as a function of time on a single cycle. Red lines indicate where the
start and end of the points included in the mass balance.
the CO2 on the first calcination and comparing it to the known amount of CO2 going in from a 100 g
batch of limestone. The number of moles of CO2 going in is given by the amount of limestone
multiplied by the known purity of the limestone (in this case for Longcliffe, 98.25 %) divided by the
88
relative atomic mass of CaCO3 (100.77). Figure 3-9 shows CO2 concentration in the off-gas on a
single cycle, with an indication of the start and end of the mass balance.
The number of moles of CO2 leaving in the off-gas is given by Eq. 3-4, where = number
of moles of CO2 leaving the system arising from calcination, ̇ = the molar flow rate of gas going in
(air + CO2), = the concentration of CO2 in the incoming gas (15 % v / v) and = the
̇ Eq. 3-4
∫ ( )
The result of these two calculations gave values for CO2 going in from the limestone and CO2 leaving
the system through calcination of 0.974 and 0.956 respectively. This indicated an error of less than 2
3.6. Conclusions
A laboratory-scale system has been designed, built and commissioned capable of producing
A range of fuel feeding systems have been tested and included in the commissioning process.
The mass balance on the first cycle indicates that the limestone is fully calcined.
Further development of the heating system would ensure the entire calcination step takes
Further work would include producing sorbent with a wider range of solid fuels.
89
Chapter 4. Trace Element Analysis of Fuels,
Limestone and Sorbent
This chapter details the methods used to enable quantification of the trace element inventory of
materials entering and leaving the fluidised bed system. To ensure that the loss of elements was due
to repeated cycling and not a single cycle, a comparison was made between the original limestone and
the sorbent homogenised after a single cycle (‘0’ cycles). Though these results are less relevant to
Ca-Looping (than those for repeated cycling), a Ca-looping calciner will have a larger quantity of
fresh limestone entering than a standard cement calciner due to the need to continually replenish and
purge bed material. Therefore the release of volatile elements as a result of a single calcination
becomes relevant from the point of view of calciner operation (due to potential build-up of problem
Three analytical methods capable of performing trace element analysis are available for use at
Spectrometers (ICP-AES) and an Ion Chromatography (IC) unit in the Department of Chemical
Life Sciences. Other methods such as atomic absorption spectroscopy are not available and are
therefore omitted from the following discussion. A more detailed discussion for ICP-AES is given as
this is the technique employed for the majority of the analyses in this project (see Section 4.2.3 for the
90
4.2.1. ICP-AES and ICP-MS
Atomic emission spectroscopy (AES) involves excitation of a liquid sample of low acidic
excitation within the plasma imparts additional energy to the atoms promoting them to excited states.
Both atomic and ionic excited state species then relax to the ground state via the emission of a photon
producing visible light line spectra characteristic of the element present. The total number of photons
is directly proportional to the concentration of the related element in the sample and recorded as
Concentration of elements present below 100 μg / l can be determined by comparing the intensities of
the line spectra obtained against the spectra from calibration solutions of known concentrations. The
detection limits typically down to 0.0001 – 0.1 parts per million (ppm). ICP-AES lacks the sensitivity
to accurately determine trace element content in very dilute solutions below these concentrations, for
which ICP-MS would be a more suitable choice. A further drawback with this method is the potential
for interference between the different light wavelengths used to infer the concentration of a given
91
element (see section 4.3.4 for a detailed discussion). Radial view is the classical operation mode,
where the plasma is operated in a vertical orientation with the analytical zone observed from the side
of the plasma. This limits the effect of potential spectral and background interferences. In axial view,
the plasma is rotated to a horizontal position and the ICP is observed from the end of the plasma. The
axial view can provide a 5- to 10- fold improvement in the limits of detection (LOD; see Section 4.2.4
for a definition). The list of elements capable of being observed by ICP-AES is given below (Table 4-
1).
Table 4-1. List of elements that can be determined by ICP-AES (Hou 2000).
Li, Na, K, Rb, Ce, Pr, Nd, Sm, Eu, Gd, Sc, V, Ti, Cr, Mn, Fe, B, C, N, Al, Si, P, S, Cl,
Cs, Be, Mg, Tb, Dy, Ho, Er, Tm, Yb, Co, Ni, Cu, Zn, Y, Nb, Ga, Ge, As, Se, Br, In,
Ca, Sr, Ba Lu, Th, U Zr, Mo, Ru, Th, Pd, Ag, Sn, Sb, Te, I, Tl, Pb, Bi
Cd, La, Hf, Ta, W, Re,
Os, Ir, Pt, Au, Hg
In contrast ICP-MS is extremely sensitive due to the highly sensitive detection of the mass
spectrometer resulting in part per trillion (ppt) detection limits (< 1 - 100 ppt), ~ 2 to 3 orders of
magnitude lower than ICP-AES. As with ICP-AES, the quantification of trace elements is obtained
by external calibration curves using multi-element standard aqueous solutions. As with AES,
limitations of ICP-MS include interpretation of the results. The cations produced in the plasma are
classification has a major limitation since it is possible for two or more species to be detected at the
same m/z values. For example, interferences occur when residual Cl- anions are present in solution
when HC1 or HC1O4 are used during the sample preparation. In the case of As, polyatomic ion
species can be formed between argon and chlorine ions resulting in interfering peak in the mass
spectrum at a mass equal to their combined atomic mass, (40Ar35Cl); As will have the same m/z ratio
of 75. ICP-MS is also relatively intolerant to high levels of total dissolved solids which should
generally be no higher than 0.02 % in the final solution. ICP-AES is far less disturbed by the
92
dissolved solid content of the sample. Therefore, the possibility of introducing more highly
concentrated solutions in the ICP-AES instrument (generally 50 to 100 times more concentrated)
The IC operates by the release of anions and cations from a chromatograph column based on their
interaction with a resin. Solutions pass through a pressurized chromatographic column where ions are
absorbed by the column, made of the resin. As an eluent runs through the column, the absorbed ions
begin to separate. The retention time of different species determines their concentration in the
solution and as such are characterised by their release over time. Elements that can be observed
The decision over which method to use was made on the basis of an estimation of the element
in by the dilution factor (DF) (see below, section 4.2.4) which for a 20 mg sample in 10 ml of solution
is 500. This indicates that ICP-AES would serve as the most appropriate analytical technique, given
the detection limits described above (section 4.2.1) (i.e. parts per million rather than parts per trillion)
and the limited number of elements observable by IC. Furthermore, the high financial cost of running
all samples through an ICP-MS from a different department would render the project financially
unviable (~ £1000 / day as opposed to £50 / day for in-house (Jarvis 2013)). An initial comparison of
the two methods actually indicate that ICP-AES actually performed better in an analysis of NIST
standard reference material (SRM) 1633b (fly ash) for three elements chosen at random (Figure 4-2)
where all values for the ICP-AES were within the range of the experimental limits imposed by the
standard.
Figure 4-2. Comparison of ICPAES and ICPMS for NIST SRM 1633b (average of three repeat
samples, error bars shown as relative standard deviation, RSD).
However, neither of the in-house ICP-AES machines are capable of observing S. This is because the
only observable wavelength for this element is at the bottom of the light spectrum at ~ 189 nm below
94
the minimum wavelength detectable by the machine (190 nm). According to the machine operating
instructions it was possible to observe the lower range of the light spectrum through attachment of a
N2 purge line. However this was found not to be the case on either machine. Therefore attempts have
The accurate measurement of the trace element content in a solid is dependent on two parameters; the
dilution factor (DF) and the limit of detection (LOD), defined here. The DF is the correction factor
applied to the concentration in the liquid solution to obtain the concentration in the solid. The dilution
factor results from the sample preparation (digestion) procedure and the dilution required prior to
analysis such that the total dissolved solid satisfy the ICP requirement. It is calculated as the volume
of the solution (Vs) divided by the sample weight (Ms) and can therefore be defined as follows:
Eq. 4-1
The LOD is the lowest value at which one can confidently say the element exists in the sample
(Charpentau 2009). Taking into account both instrumental sensitivity and sample preparation, the
LOD is calculated as the concentration equal to three times the standard deviation of the background
signal (3σbk) (or the ‘procedural blank’ – see section 4.2.3) multiplied by the known concentration in a
calibration solution (γcb, e.g. 1ppm) divided by the mean of the raw counts of the blank ( ̅ )
subtracted from the counts of the calibration solution ̅ . These are quoted as concentration in the
solid for each preparation procedure. The LOD may therefore be defined by Eq. 4-2
95
( ) Eq. 4-2
̅ ̅
The relative standard deviation (RSD) of the results for the unknown samples provides a measure of
precision and is the standard deviation (SD) multiplied by 100 and divided by the mean ( ̅ ) Eq. 4-3:
Eq. 4-3
̅
The quantification of trace elements by ICP-AES or ICP-MS requires that the sample be dissolved in
acidic solution. This was carried out using strong acids which also dissolve the trace elements
(discussed below). Upon cycling, CaO sorbent produced from the cycling experiments (Chapter 3)
was ground to – 90 µ by hand using a pestle and mortar (to avoid contamination from more
convenient grinding methods) before undergoing a homogenisation step. The resulting CaOH2
aliquots are accurately weighed to ~ 20 mg and the exact weights recorded to enable the DF to be
calculated. The aliquots are placed into PTFE vessels of volume 7 ml and 1 ml of 69 % HNO3 added.
This was to leach elements from, in the case of limestone or CaO, the resulting calcium nitrate salts
according to Eq. 4-4due to the subsequent heat and pressure applied at the microwave stage. The lids
are secured and the vessels microwaved according to a heating regime (Table 4-3) to affect the hot
Eq. 4-4
96
Table 4-3. Microwave heating regime.
Stage 1 2 3 4 5 6 (cooling)
Time (mins) 1 2 5 5 5 20
After 45 mins cooling at – 7 ºC (to ensure volatile elements are not released upon opening), the inside
of the vessels are rinsed into 10 ml volumetric flasks via a small plastic funnel using 5 % HNO 3. The
solutions are topped up to 10 ml to enable the DF to be calculated. The solution was then shaken and
poured into a 50 ml centrifuge tube and drawn into a 20 ml plastic syringe. This allowed the solution
to be filtered via a 0.45 μ Whatman filter into a 10 ml test tube ready for insertion into the ICP-AES.
The instrument was first warmed up for 75 minutes with argon flowing to the plasma to enable the
plasma temperature to equilibrate, as advised the instrument user guide. A test sample was then
inserted containing all elements in 1 ppm concentration. The results of this test sample allow the peak
of each wavelength included in the analysis to be examined for interferences. This also allows for the
correct positioning of peak measurement readings and baseline points i.e. on the baseline rather than
Calibration of the instrument at the start of each run was performed by analysing three multi-element
solutions at concentrations of 0.125, 0.25 and 0.5 ppm respectively. The counts recorded for the
‘standard blank’ (i.e. the matrix solution, 5 % HNO3) provides a baseline value for the calibration
line. The x-axis shows the known concentration of the calibration solutions whilst the y-axis shows
97
Figure 4-3.Examples of wavelength calibration for Mn (257.610 nm) and Cr ( 267.716 nm) showing
correlation co-efficients for each line.
The unknown samples are then run after calibration. Each sample lasts ~ 6 minutes to allow all
elements of interest to be registered including 25 seconds uptake time (for the solution to reach the
plasma) and 60 seconds of washing using a 2 % HNO3 solution to prevent contamination between
samples. A 1 ppm ‘drift’ solution was analysed every 5 samples to correct for instrument sensitivity
which varies as a function of time due to changes in ambient temperature and small fluctuations to the
temperature of the plasma itself. The extent of the drift for one of the ICP-AES in-house machines
used for this project was shown to be + 10 % over the course of an eight hour run (Figure 4-4).
98
Figure 4-4. Example of ICP-AES drift over the course of a four hour run.
The element counts for each element in the unknown sample ( ̅ ) first has the counts for the
procedural blank ( ̅ ) subtracted to account for any contamination introduced during the preparation
procedure. The procedural blank consists of a solution containing all reagents used in the digestion of
samples without containing the material undergoing analysis. This blank undergoes all of the
preparation stages of a normal sample. Its purpose is to provide a background for contamination
picked up during sample preparation. The blank-corrected counts are then multiplied by the gradient
of the calibration line (β, e.g. Figure 4-3) to give the solution concentrations. This multiplied by the
DF for each sample gives the concentration of each element in the original solid sample .
( ̅ ̅ ) Eq. 4-5
99
4.3.4. Determination of Spectral Wavelengths
The light wavelength selection was based upon the following factors (Boss and Fredeen 1997) :
Analyte concentration vs. Emission line sensitivity (or detection limit, D.L.)
Initially, wavelengths were run for all elements of interest. However the ICP-AES requires ~ 1 ml per
element per sample which places a limitation on the number of wavelengths / elements that can be
observed from a 10 ml sample to only ten wavelengths. Therefore it was necessary to determine the
most suitable wavelength for each element based on the factors listed above. This section deals with
this issue for the list of elements detailed in the background section of this thesis, which are listed
again here:
As, B, Ba, Be, Bi, Cd, Cr, Cu, K, La, Li, Mg, Mn, Na, Ni, Sb, Sr, Ti, V, W, Y, V, Zn, Z..
The information used for this section has been taken from the Inorganic Ventures on-line interactive
periodic table (Inorganic_Ventures 2013). According to (Gaines 2012), the data on this website is
referenced from (Barnes 1981; Thompson and Walsh 1983; Montaser and Golighty 1992).
Aluminium (Al): Al has wavelengths at 167.078, 394.401 and 396.152 nm. The first wavelength is
below the limit detectable by the machine whilst the third wavelength has a major interference with
Arsenic (As): As has three spectral lines at, nm. The line with the lowest estimated D.L (189.042
nm) was below that detectable by the machine (i.e. below 190 nm), whilst the 228.812 line is known
to have a strong interference with Cd. Therefore the 193.696 nm line was chosen. The solution
100
concentrations of all input materials were below the LOD value for this line, indicating that As was
Barium (Ba): Barium has spectral lines at 230.424, 233.527 and 455.403 nm. Although the
wavelength with the best detection limit is 455.403, this is known to have an interference with Zr,
whilst the 230.424 has an interference with Mo. The only line not to have an interference is 233.527
Beryllium (Be): Beryllium has wavelengths at 234.861, 313.042 and 313.107. The first two both
have interferences with elements likely to be present in the unknown samples (Fe and V respectively).
The 313.107 nm wavelength has interferences with elements unlikely to be present in the samples
(Ce, Th, Tm), and still has a useable estimated detection limit of 0.0007 ppm. However, the solution
concentrations of all input materials were below the LOD value for this line, indicating that Be was
Boron (B): Boron has spectral lines at 249.773, 249.678 and 208.959 nm. The 208.959 line is
reported to have the least number of minor interferences (Mo) and no major interferences, whilst the
other two lines have eight and five respectively. Therefore the 208.959 line was chosen.
Bismuth (Bi): Bi has wavelengths at 190.171, 206.170,. The 190.171 line showed poor sensitivity
due to it being below 200 nm. Therefore it was necessary to use the 206.170 line. According to the
peak produced from the 0.5 ppm solution, no interference was reported. However its close proximity
to the 206.200 nm Zn line (Figure 4-5) indicates that care should be taken using this line for Bi if the
101
Figure 4-5. Wavelength peaks for a) Bi at 206.170 and 2) Zn at 206.200 nm (0.5 ppm).
However, the solution concentrations of all input materials were below the LOD value for this line,
Cadmium (Cd): The 228.802 line is known to have a strong interference with As line 228.812,
eliminating the use of this line. The other two lines have experimental LOD’s of 0.0105360059 and
0.0051136520 respectively. Given the low expected concentration of Cd in all inputs, the line with
the lowest LOD value (226.502 nm) should be used. However, the solution concentrations of all input
materials were below the LOD value for this line, indicating that Cd was not to be included in any
further analysis.
Chromium (Cr): Cr has wavelengths at 205.560, 267.716 nm and 283.563 nm. The solution
concentrations reported by the 205.560 nm line were an order of magnitude lower than the LOD
values calculated, whilst the R2 value on the 283.563 nm was very low at 0.2257. Therefore the
Copper (Cu): Cu has wavelengths at 219.958, 224.700 and 324.754 nm. A strong interference is
known to exist with the second line. The other two lines have minor interferences with elements
unlikely to be present in any of the samples to be analysed in this investigation (Nb, U, Th, Hf, and
Ta). The 324.754 nm line has the lowest reported detection limit therefore this line was used.
102
Lanthanum (La): La was supposedly detected in all three fuels (La Jagua, 708.14; Lea Hall, 702.56;
RDF, 680.88 ppm respectively) at the 407.735 nm line. However this is due to its close proximity
with the 407.771 nm Sr line causing interference; the similarity between the shapes of the peaks
corresponding to these two readings can be observed (Figure 4-6). This was confirmed by the
408.672 nm La line which reported solution concentrations of all input materials below the LOD
value for this line.. Therefore the 408.672 nm La line would be used, however La was not detected in
Figure 4-6 Spectral peak at a) 407.735 nm (La) and b) 407.771 nm (Sr) (0.5 ppm).
Lithium (Li): Li has wavelengths at 323.261, 460.286 and 670.784 nm. The first two wavelengths
both report interferences with elements which may be present in the samples ( Sn and Zr respectively)
therefore the highest wavelength of 670.784 nm was used. This was also the wavelength with the
lowest detection limit. However, the solution concentrations of all input materials were below the
LOD value for this line, indicating that Li was not to be included in any further analysis.
Potassium (K): K has wavelengths at 404.721, 766.490 and 771.531 nm. It was found that no counts
registered at either 404.721 or 771.531, potentially due to the high detection limit of these two
wavelengths (1.1 and 1.0 ppm respectively). Therefore the 766.490 nm line was used which reported
lowest detection limit of 0.0002 ppm whilst the 285.213 line has a small interference with Cr.
However according to (Welz and Sperling 1997) the 279.553 nm line becomes non-linear even at very
low concentrations. This was confirmed when standard reference material samples were run
containing known concentrations of Mg. The results for the 279.553 were consistently lower than the
certified value whilst the 285.213 garnered concentrations which matched that of the certified values.
This suggested the small Cr interference with the 285.213 line was not significant enough to affect the
Manganese (Mn): Mn has wavelengths at 257.610, 259.373 and 260.569 nm. The 257.610 line
offered the lowest detection limit with no interference from other elements therefore this wavelength
was used.
Molybdenum (Mo): Molybdenum has spectral wavelengths at 202.031, 203.844 and 204.598 nm
These lines reported experimental LOD’s of 0.0360, 0.146 and 0.0630 ppm respectively. The 202.031
line provided a LOD value ~ 10 times that of the solution concentration in the fuel samples (~ 0.2
ppm) indicating that this was the line that should be used. However, the solution concentrations of all
input materials were below the LOD value for this line, indicating that Mo was not to be included in
Sodium (Na): Na has wavelengths at 330.237, 588.995 and 589.595 nm. The first wavelength has
known significant interferences with Znwhilst the 588.995 line registered no counts. Therefore the
589.595 line was used which also provided the lowest detection limit of 0.07 ppm
Nickel (Ni): Nickel has wavelengths at 221.647, 231.604 and 232.003 nm. The 232.003 line has a
strong interference with Cr whilst the other two have small interferences with Si and Sb respectively.
However the 231.604 line was used given that Sb is less likely to be present in the samples than Si
(Figure 4-7). The 231.604 wavelength also offered the lowest detection limit.
104
Figure 4-7. Strong Ni peak at 231.604 nm.
Phosphorous (P): P has wavelengths at 177.495, 178.287 and 213.618 nm. The first two
wavelengths are below the minimum wavelength detectable by the ICP machine (i.e. below 190 nm).
The 213.618 nm line has a strong interference with both Cu and Mo. However, the results of the
limestone standard reference material suggested that this interference did not affect the results. As
Lead (Pb): Pb has wavelengths at 168.215, 217.000 and 220.353 nm. The first wavelength is below
190 nm and therefore could not be used. The 217.000 nm reported interferences with Sb whilst the
220.353 line reported an interference with Sb, both elements potentially present in the samples.
However the 220.353 offered the lowest detection limit therefore this wavelength was used.
The solution concentrations of all input materials were below the LOD value for this line, indicating
Antimony (Sb): Sb has wavelengths at 206.833, 217.581 and 231. 147 nm. All three report minor
interferences with elements likely to be present in the samples (Cr, Fe and Ni respectively). However
the solution concentrations in all inputs were below the limit of detection for all wavelengths therefore
105
Strontium (Sr): Sr has wavelengths at 407.771, 421.552 and 460.773 nm. No interferences were
reported with any of these wavelengths therefore the 407.771 wavelength was used which offered the
Titanium (Ti): Ti has wavelengths at 323.452, 334.941 and 336.121 nm. All three wavelengths
reported small interferences with elements which could be present in the samples; Cr, Mo / Co and Ni
respectively. However the 323.452 wavelength offered the lowest detection limit of 0.0038 ppm
Vanadium (V): V has wavelengths at 290.882, 292.402 and 309.311 nm. The 309.311 line though
offering the lowest detection limit of 0.005 ppm reported an interference with Mg. The other two
wavelengths reported no relevant interferences therefore the 292.402 line was used which offered the
Tungsten (W): W has wavelengths at 207.911, 209.475 and 224.875 nm. The 209.475 line reported
an interference with Mo. The 207.911 line reported no interferences and offered the lowest detection
limit therefore this line was used. However, the solution concentrations of all input materials were
below the LOD value for this line, indicating that W was not to be included in any further analysis.
Yttrium (Y): Y has wavelengths at 360.073, 371.030 and 377.433 nm. None of these wavelengths
reported any interference therefore the 360.073 line was used which offered the lowest detection limit
of 0.005 ppm. However, the solution concentrations of all input materials were below the LOD value
for this line, indicating that Y was not to be included in any further analysis.
Zinc (Zn): Zn has wavelengths at 202.548, 206.200 and 213.856 nm. All three wavelengths reported
interferences with elements which may be present in samples, namely; Ni / Cu, Cu and Sb / Bi
respectively. The 213.856 line offered the lowest detection limit therefore this line was used.
106
Zirconium (Zr): Zr has wavelengths at 272.261, 193.198 and 343.823nm. The 272.261 nm line
reported a strong interference with Cr. The other two wavelengths reported no relevant interferences
therefore the 343.823 line was used which offered the lowest detection limit of 0.007 ppm. However,
the solution concentrations of all input materials were below the LOD value for this line, indicating
The wavelength choice for each element is given in Table 4-4 following the discussion above:
Table 4-4. Spectral wavelengths for elements of interest and prevalence in the inputs. ‘X’ indicates the
presence of an element in an input; ‘-‘ indicates the absence. (Elements not highlighted in bold were
removed from subsequent analysis).
Al X X X X
As 193.696 - - - -
B 208.959 X X X -
Ba 233.527 X X X X
Be 313.107 - - - -
Bi 206.170 - - - -
Cd 226.502 - - - -
Cr 267.716 X X X X
Cu 324.754 X X - X
La 408.672 - - - -
Li 670.784 - - - -
K 766.490 X X X X
Mg 285.213 X X X X
Mn 257.610 X - - -
Mo 202.031 - - - -
Na 589.555 X X X X
107
Ni 231.604 - X X X
P 213.618 X - - -
Pb 220.353 - - - -
Sb - - - - -
Sr 407.771 X X X X
Ti 323.452 - X - X
V 292.402 - - - -
W 207.911 - - - -
Y 360.073 - - - -
Zn 213.856 - X - X
Zr 343.823 - - - -
Due to the sensitivity of the ICP-AES to external factors such as ambient room temperature, it was
necessary to perform a check on machine integrity to enable reliable comparisons to be made between
sorbents of different cycle numbers run on different dates (Table 4-5). The same sample was run on
three different days over the course of three weeks for a range of elements.
108
Comparing these concentrations, the results show that for all elements except Na it was possible to
run samples on different dates without affecting the results, even for elements of very low
The deviation of the experimental values from the certified values provides a measure of the accuracy
of the results for that digestion method and for the trace element procedure as a whole. A significant
error between these two values suggests an issue in the procedure, and likely to be as a result of poor
sample digestion. Figure 4-8 compares the experimental values for digesting NIST SRM 1d,
(argillaceous limestone) against the certified values, whilst Table 4-6 shows the conversion of the
concentration from % oxides given in the certified standard to elemental ppm concentrations as
Figure 4-8. Comparison of experimental values against certified values for NIST SRM 1d
(argillaceous limestone) for a. sample concentrations > 1000 ppm and b. sample concentrations <
1000 ppm. Average of six results. Error bars shown as RSD.
109
Table 4-6. Certified Values for SRM 1d Limestone, Argillaceous and conversions from % oxides to
element ppm concentrations.
Conc. (%, 3 ± (%, 2
Oxide Element Conc. (ppm, 3 s.f.) ± (ppm, 2 s.f.)
s.f.) s.f.)
Na2O 0.0109 0.0016 Na 80.9 0.74
MgO 0.301 0.010 Mg 1820 0.60
Al2O3 0.526 0.013 Al 2780 0.52
P2O5 0.0413 0.0025 P 180.2 0.43
S 0.1028 0.0062 S 1030
K2O 0.1358 0.0046 K 1130 0.83
Mn 0.00209 0.0005 Mn 209 5.0
Fe2O3 0.3191 0.0068 Fe 2230 0.69
ZnO 0.0022 0.0003 Zn 17.7
SrO 0.0303 0.0010 Sr 256
TiO2 0.0306 0.0065 Ti 183 0.60
Cr2O3 0.0012 0.0002 Cr 8.21 0.69
BaO 0.0033 0.0011 Ba 29.6
The results for Mg and Cu are excellent with experimental values close to the certified values. For
most elements however the experimental values are slightly lower than the certified values. This
suggests that not all of the material was being digested and that some of the elements are being lost,
possibly during the filtration process. The experimental results for Na and K are both much higher
than the certified values. The reason for this is currently unknown (see: Further Work).
NB. This was early work during the PhD process therefore not all elements were analysed.
To ensure that loss of elements was due to repeated cycling and not a single cycle, a comparison was
made between the original limestone and the sorbent homogenised after a single cycle (‘0’ cycles).
Though these results are less relevant to Ca-Looping (than those for repeated cycling), a Ca-looping
calciner will have a larger quantity of fresh limestone entering than a standard cement calciner due to
the need to continually replenish and purge bed material. Therefore the release of volatile elements as
a result of a single calcination becomes relevant from the point of view of calciner operation (due to
potential build-up of problem elements in calciner components, e.g. ring-build-ups / filters etc.)
110
Due to the different form of the two samples (original limestone as CaCO3 and calcined material as
Ca(OH)2 following homogenisation) it was not possible to directly compare concentrations; the
concentrations in the calcined material would appear higher due to the fact the sample has a smaller
mass following calcination. For a direct comparison, concentrations of each element (ppm) in the
samples were converted to theoretical quantities contained in the original 100 g batch (mg) based on
the relative atomic masses of CaCO3 and Ca(OH)2 respectively. The “student’s t-test” was used to
identify whether a change in the theoretical weight has taken place. The results indicate that the
Table 4-7.Sample concentrations (ppm) and theoretical weights (mg) and the results of the “Students”
t-test comparing quantities (mg) of elements in single-cycled sorbent against the original limestone
(x=mean, s=variance, DoF=degrees of freedom, Ho=null hypothesis). Results in green indicate
rejection of the null hypothesis (Ho) due to a statistically significant increase in concentration
following a single calcination; red marks indicate a decrease. Blank = no significant change.
Longcal (Homog.) 0 cycle (CaOH2) t-test parameters
x2 ppm (mg) s2 x1 ppm (mg) s1 DoF p-value Ho
Al 124.28 (12.06) 9.25108 299.50 (21.77) 0.88256 13 3.6E-4
Ba 12.68 (1.49) 0.00824 40.89 (3.64) 0.08280 13 1E-9
Cr 2.16 (0.21) 0.00006 4.18 (0.30) 0.00338 13 0.001
Cu 124.28 (0.72) 0.00903 7.20 (0.52) 0.00662 13 0.002
K 83.99 (8.15) 2.69523 117.21 (8.52) 1.22182 13 0.637
Mg 1217.54 (118.10) 6.94377 1628.23 (118.37) 31.45362 13 0.902
Mn 44.74 (4.34) 0.00582 65.07 (4.73) 0.05269 13 7E-4
Na 49.37 (4.79) 0.07712 65.48 (4.76) 0.17122 13 0.876
P 16.51 (1.60) 0.00332 21.19 (1.54) 0.01294 13 0.195
Sr 114.80 (11.14) 0.03985 72.11 (5.24) 0.74468 13 7.3E-9
The results show that Cu and Sr decreased whilst Al, Ba, Cr and Mn all saw an increase. The loss of
elements would suggest devolatilisation during calcination. However , both Sr and Cu have boiling
points above the maximum system temperature. An increase would suggest either contamination
from the reactor / laboratory procedure, the liberation of elements due to heating and fluidisation or as
a result of high variability of element concentrations within the limestone itself. However an
alternative explanation may be presented to account for these findings. The differences observed may
be attributable to the efficacy of the digestion procedure in digesting different types of sample. This
111
followed by a single carbonation allowing the sorbent to be removed as CaCO3. This enabled direct
comparison of the cycled material against the original limestone. These results are shown in (Table 4-
8).
Table 4-8 indicates that the single-cycled CaCO3 contained a higher concentration of Al and Cr than
the original limestone and a lower concentration of K. The decrease in K could be attributed to
devolatilisation. The fact that the single-cycled Ca(OH)2 samples contained less Sr than the original
limestone whilst the single-cycled CaCO3 did not suggests that Sr was easier to extract from CaCO3
than Ca(OH)2. Similarly, the fact that the single-cycled Ca(OH)2 samples contained more Ba and Mn
than the original limestone whilst the single-cycled CaCO3 did not suggest that these two elements are
easier to extract from Ca(OH)2 than from CaCO3. Al and Cr on the other hand saw an increase in
both the Ca(OH)2 and CaCO3 samples. This could be attributed to the liberation of these elements
from the limestone due to heating and fluidisation or due to contamination. The high concentration of
Cr in the residue removed from one of the bed thermocouples (> 3500 ppm) suggests that Cr
contamination from this source is a strong possibility. This contamination would have to take place in
the first cycle given that Cr contamination was not detected in the cycling experiments. However the
112
4.3.8. Observing S using IC
S is one of the elements of interest that cannot be observed on either of the ICP instruments available
(see section 4.2.3) but can be observed using IC. The acidic solutions resulting from digestion of the
solid samples are run through the machine similar to that of an ICP. It is found however that the high
number of ions in the 5 % acid solutions prepared for ICP leads to a shift in the release of release time
of the elements from the chromatograph column rendering the results unusable.This was found when
using both NHO3 and HCl as the digestion medium. The use of KOH crystals as to neutralize the
solution before insertion was not found to remedy this issue. As such, S was omitted from analysis in
Analysis of the input materials was carried out to determine which elements should be included in the
sorbent analysis and to enable mass balances on elements to be performed. Twenty of each sample
type was digested and the mean values recorded to enable consistent comparison between the
Elements shown to have high variability tend to be those of low concentrations; this was an expected
finding. However some elements of higher concentrations show high variability in more than one
instance despite sample concentrations above 100 ppm, e.g. Sr. Lea Hall coal showed the greatest
The list of elements to be included in the sorbent analysis was based on the lists of elements known to
impact on alite production (as detailed in the Background section) and elements present in the inputs
(see section 4.9 above). On this basis, the list of elements to be included in the analysis of sorbent
cycled in the absence and presence of fuel respectively was given below. Elements shown to remain
113
Table 4-9. Concentration of trace elements in the inputs (mean of twenty samples, ppm).Cases where
the RSD values are greater than or comparable to the mean value markedwith an asterisk (*).
Mn 107.14 22.71 - - - - - -
P 35.79 12.60 - - - - - -
4.3.11.
the same upon cycling and not present in the fuel (e.g. Mn) are not included in the analysis of sorbent
cycled in the presence of fuel, as follows (NB. Though not a trace element, Al was included in the
In the absence of fuel: Al, B, Ba, Cr, Cu, K, Mg, Mn, Na, P, Sr.
In the presence of La Jagua combustion: Al, B, Ba, Cr*, Cu, K, Mg, Na, Ni*, Sr, Ti, Zn.
114
In the presence of Lea Hall combustion: Al, B, Cr*, Cu, K, Mg, Na, Ni*, Sr.
In the presence of RDF combustion: Al, Ba, Cr*, Cu, K, Mg, Na, Ni*, Sr, Ti, Zn.
* Cr and Ni are omitted from the results for sorbent cycled in the presence of fuel due to low
Sorbent cycled without and with fuel are digested and analysed for their trace element inventories.
This was to determine whether cycling has an impact on the element inventory.
a. b.
c. d.
115
e. f.
g. h.
i. j.
116
k.
Figure 4-9 (a-k). Trace element inventories of sorbent cycled without fuel. Error bars shown as RSD
values.
Given that all results remain the same within the experimental limits (indicated in Figure 4-9 as the
RSD); these results indicate that subjecting the sorbent to up to 15 calcination / carbonation cycles in
the absence of fuel does not affect its trace element inventory. This was an expected result given that
the dissociation temperature for the elements included in the analysis is above that of the maximum
system temperature.
In the batch experiments coal was fed into the base of the spouted bed in 2 g charges on each cycle.
These experiments served as an initial test as to whether inserting coal at these sorts of quantities does
impact on the trace element inventory of the cycled sorbent, before proceeding with the continuously-
117
Figure 4-10. Trace element inventories of sorbent upon cycling in the presence of
batch fed La Jagua. Error bars shown as RSD values.
The results show that in the initial experiments an increase in the trace element concentration was
observed for Ba, Cr, K, Mn, Sr, and Ti. B, Na and K remain the same, whilst Cu sees a small
decrease. Cr saw a particularly large increase. This suggests that the batch-fed coal system was
In the continuously-fed coal experiments, sorbent was cycled five and eight times for the coals and
three and five times for the RDF respectively. The number of cycles for the coal experiments is based
on the number of cycles experienced by the sorbent for a Ca-loop integrated with a cement kiln as
derived by the stochastic model (see: Flowsheeting Chapter section 2.2.1). The number of cycles for
the RDF experiments is limited by the number of pellets that can be produced in a morning before
carrying out the experiments in the afternoon (see: Production of Spent Sorbent Chapter section
3.1.5). The trace element inventories of these sorbents are given in (Figure 4-11).
118
No B analysis undertaken
119
No Ti Analysis undertaken
No Zn Analysis undertaken
Figure 4-11. Trace element inventories of sorbent upon cycling in the presence of continuously fed
fuel (La Jagua, Lea Hall and RDF). Error bars shown as RSD values.
120
Figure 4-11 indicates that based on the increase in Al content, some ash did mix in with the sorbent
for all sorbents produced in the presence of fuel combustion. For the experiments with fuel, the trace
concentration of Ba, K, Sr and Zn remained the same where present in the original fuels. For Lea
Hall, B sees an increase over eight cycles from ~ 20 to 40 ppm, exceeding the experimental limits
(RSD); thought does not for La Jagua. This was also the case for the La Jagua batch experiments
where the concentration of B remained the same despite being present in the original fuel. For Cu, the
low level concentrations in the sorbent on the first cycle (~ 15 ppm) remains the same over eight
cycles for the two coals, though sees an increase over five cycles for the sorbent cycled in the
presence of RDF from ~ 20 ppm to ~ 180 ppm. Na remains the same for the sorbent cycled in the
presence of La Jagua, though sees an increase in the results for Lea Hall and RDF from ~ 250 to ~ 550
ppm and from ~ 200 to 400 ppm respectively. This was an expected result given that the Na content
of the La Jagua coal is an order of magnitude lower than that of the other two fuels. Ti remains the
same for La Jagua over eight cycles, though sees an increase in the sorbent cycled in the presence of
Some of these results are expected whilst others are not. An expected result would be an increase in
the trace element inventory over repeated cycles for every case where non-volatile elements are
present in the original fuel. However it was found that this is not always the case.
Mass balances have been calculated with upper and lower bounds using the RSD values of the
incoming fuel and the results for cycled sorbent for elements shown to have increased upon cycling
121
Table 4-10. Results for % retention of Al in cycled
sorbent.
Fuel / Cycle No. Lower Absolute Upper
La Lagua
1 104.40 133.79 183.59
5 75.17 84.97 97.23
8 65.86 75.20 86.24
Lea Hall
1 97.54 112.50 137.43
5 30.47 32.19 35.66
8 26.70 28.82 32.56
RDF
1 75.41 85.09 98.00
3 33.72 35.13 36.95
5 33.09 34.82 36.94
The mass balances for Al (Table 4-10) indicate the % retention for this element decreases with cycle
number for all elements. This suggests that the amount of ash entering and leaving the system is
reaching equilibrium as cycle number increases. Given that the amount of ash leaving will increase
over a given period of time, after a certain point the amount of ash entering and leaving will remain
constant. The difference in the results for Al for the three fuels is likely to be down to differences in
the nature of the fuels. For example, the retention of Al for Lea Hall on the fifth and eighth cycles
was much lower than that of La Jagua, suggesting an increased friability of this fuel. The retention of
Al for RDF was also lower suggesting that the nature of this fuel, the difference in the feeding system
and its combustion characteristics in relation to the coal granules played a role in the results observed.
The same trend was also shown for all results relating to B and Na (Table 4-11, Table 4-13),
suggesting that these elements are being retained in the sorbent via the ash and not through
dissociation from the solid fuels and reaction with the sorbent. The other elements (Cu and Ti)
showed considerable errors in the mass balances; this could be accounted for by the high variability in
the concentration of these elements in the original limestone and in the fuel (RDF) (Table 4-9). The
presence of Ti in the stainless steel RDF feeder could also account for the mass balances being
The increase in B suggests that this element would have implications in terms of the cement chemistry
for cement produced from sorbent cycled with Lea Hall coal given the known effect of B on alite
production (see: Background) suggesting that this element requires further observation for any future
123
Both the Na retained in the sorbent and that lost from the bed has implications in terms of volatile
cycles. Na lost from the bed indicates that volatile cycles could develop in the Ca-looping calciner as
Na was introduced in the fuel but not retained in the sorbent. Na retained in the sorbent will be passed
to the kiln; the high kiln temperatures will cause this element to dissociate causing volatile cycles to
form within the kiln and between the kiln and Ca-looping calciner. This could limit the amount of
4.5. Conclusions
The results of the single cycle experiments indicate that Sr is easier to extract from CaCO3
than from Ca(OH)2, whilst Ba and Mn were both easier to extract from Ca(OH)2 than from
CaCO3. The results of these experiments also suggest that Cr contamination from the
thermocouple is a strong possibility though well below any limit likely to impact on clinker
composition.
Sorbent cycled in the absence of fuel does not undergo a measurable change in its trace
The fuel systems designed and built as part of this project are effective in introducing the fuel
into the system and in enabling the ash to mix with the sorbent during combustion as
The decrease in the % retention for Al and other elements with reasonable RSD values
suggest that an equilibrium is reached for the amount of ash entering the system after a certain
number of cycles
The results show that B would need further observation given its observed increase over eight
124
The results for Na indicate that this element could have implications in terms of volatile
cycles developing both in the Ca-looping calciner and the cement kiln. Such features of the
system could limit the amount of sorbent that can be passed directly to a cement kiln.
Further work would include producing and testing sorbent with a wider range of solid fuels to
see which other elements may be of interest in terms of their retention in the sorbent and their
Further work would include to ascertain why when digesting NIST SRM 1d (argillaceous
limestone) experimental values for Na and K were much higher than the certified values.
125
Chapter 5. Investigation into the effect of using
repeatedly cycled sorbent on phases in cement
The aim of this chapter is to determine the wt % alite in the clinkers produced from cycled sorbent. It
was first necessary to develop a method for producing clinker in the laboratory which is described
below. Both X-ray Diffraction (XRD) and X-ray Fluorescence (XRF) analytical techniques were then
used. It was found that XRD was the more useful of the techniques as it enabled both the
identification of alite polymorphs and determination of % wt. in the clinkers through the production of
a calibration curve.
A crystal is any material whose constituent atoms, molecules, or ions are arranged in an ordered
pattern extending in all three spatial dimensions. The distance between these atomic planes or layers
can be referred to as the lattice parameters or lattice plain spacing (dhkl, Ǻ), i.e. the distances between
the layers. The type of crystal can be identified by measuring these distances; X-ray diffraction
An X-ray is an electromagnetic wave with wavelengths (λ) between 0.1 and 10 Ǻ. These wavelengths
are comparable to the interatomic distances within crystalline materials. When X-rays encounter such
material, they are scattered by the electrons. The field of the X-rays causes the electrons within the
material to oscillate and these electron oscillations are the starting point of secondary waves of the
same frequency and wavelength as the X-ray “primary” waves. These secondary waves are not
physical “reflections” of X-rays, though considering them as such is correct in a geometric sense and
126
provides a means of measuring the lattice plane spacing from the X-ray wavelength, the order of
reflection (n) and the angle of incidence / reflection (θ) (Figure 5-1). This is known as Bragg’s law
(Eq. 5-1):
Eq. 5-1
Two X-ray beams of the same phase and wavelength approach the crystal and are scattered from two
different atoms within it. The lower beam travels an extra length of 2dsinθ. These waves superimpose
when this length is equal to or an integer multiple of the wavelength of the X-rays and constructive
interference occurs. Constructive interference gives rise to the different diffraction phenomena seen
in XRD. The smallest repeating structure in the lattice of a crystalline material is referred to as a ‘unit
cell’. It is these repeating unit cells that give rise to the constructive interference defined by Bragg’s
law.
The intensities (Ii) of X-rays diffracted by phase component ‘i’ were shown by Klug and Alexander
(Klug and Alexander 1954) to be related to weight fraction (xi), density (ρi) and the mass absorption
co-efficient (µi) by Eq. 5-2 (where Ki is a constant which depends upon the geometry of the
127
( )
Eq. 5-2
∑
As such peak intensity is roughly proportional to the abundance of that phase. However this
relationship is usually not linear due to phase and sample absorption. The mass absorption co-
efficient (µ/ρ) is a measure of how strongly a material absorbs or reflects the X-rays. This becomes
important when the material under examination contains more than one phase, and for quantitative
analysis. A list of mass absorption co-efficients relevant to cement manufacture are given in (Table 5-
1).
XRD can be used to quantify as well as identify phases in a composite material. Cement clinker is a
material of at least four phases, each with a different chemical composition and therefore a different
level of X-ray absorption. Klug and Alexander (Klug and Alexander 1974) showed that an absorption
correction can be made by the use of an internal standard and calibration curves which effectively
128
5.2.2.1. The Internal Standard Technique
Because of differences in the mass absorption co-efficient (μ / ρ) of each phase present in the material,
Eq. 5-3
[ ( ) ]
Where, xJand ρJ are, respectively, the weight fraction and density of the analyte, µJ and µMare the
mass absorption coefficients of component J and the matrix, and KiJ is a constant whose value
depends on the diffracting power of component J and the geometry of the apparatus.
The Reference Intensity Ratio (or ‘RIR’) is defined as the ratio of the strongest peak of the unknown
phase to that of an internal standard in a 1:1 mixture. In the case of clinker, RIR’s were determined
for the strongest non-overlapping peaks. Calibration using this method requires calculation of the
Calculation of peak or pattern intensities and calculation of the RIR constant (Eq. 5-4):
Eq. 5-4
Where, Ireljs is the relative peak intensity of diffraction peak j of internal standard s, Ireliα is the peak
intensity of diffraction peak i of phase α, Xα is the mass fraction of phase α and Xs the mass fraction of
internal standard s.
129
However, it is possible to generate semi-quantitative results using RIR values published in the ICDD
database according to Eq. 5-5, where X’α is the mass fraction of phase α after correction for the
Eq. 5-5
Both the use of RIR values from the ICCD database and the generation of experimental RIR values
have been tried in the course of this investigation, as described below (section 5.3.3)
X-ray fluorescence (XRF) is the emission of characteristic "secondary" (or fluorescent) X-rays from a
material that has been excited by bombarding with X-rays. Ionization of the atoms take place when
the material is exposed to short-wavelength X-rays or (‘gamma’ rays), and as such results in the
ejection of one or more electrons from the atom. The removal of an electron in this way renders the
electronic structure of the atom unstable, causing electrons in the higher orbital to fall into a lower
orbital to fill the hole left behind. In falling, energy is released in the form of a photon, the energy of
which is equal to the energy difference of the two orbitals involved. Thus, the material emits radiation
which has energy characteristic of the atoms, enabling the elements present in the substance to be
identified.
The raw materials were acquired or prepared according to the following size fractions to enable the
formation of clinker phases; CaO, Al2O3 and Fe2O3 90% to pass – 90 μ; 100 % SiO2 to pass a – 45 μ
sieve. Al2O3 and Fe2O3 were supplied in powder form from Fisher Scientific. – 45 μ SiO2 was
acquired from Sibelco Sands (product code HPF5). CaO was produced in the fluidised bed reactor
130
through calcination of Longcal P25 limestone (in the absence of fuel), supplied by Longcliffe
quarries, and ground to the required size fraction in a pestle and mortar. The four materials were then
proportioned as follows to produce a synthetic raw meal; 68.3 % CaO; 22.7 % SiO2; 0.05 % Al2O3;
0.04 % Fe2O3. Enough material was produced to fill a 100 mm alumina boat (short enough to fill but
not exceed the 120 mm high temperature region of the horizontal tube furnace, see Figure 5-2) and
fired at 1450° C (furnace wall temperature). (NB. It was estimated that at this wall temperature the
inside of the furnace was estimated to be ~ 50 ° lower at 1400 °C based on the temperature profile at
1300 ° C, due to the maximum temperature of the thermocouple inside the furnace being 1350 °C).
XRD was then used to identify the presence of alite based on peak position (see section 5.3.2)
The results show that mixing the materials by hand in this fashion results in a high belite clinker
(green line on Figure 5-3). Compressing the clinker to 100 bar increases the alite content (red line).
Lastly, rapidly cooling the clinker after firing shows a further increase in the size of the alite peak
(blue line). (NB. At first an alumina boat was used to hold the clinker brick whilst firing. However
on occasions this boat cracked during the rapid cooling stage due to thermal shock. Therefore a
131
Intensity (counts)
4900
3600
2500
1600
900
400
100
0
23 24 25 26 27 28 29 30 31
2Theta (°)
Figure 5-3. Alite and Belite Peaks for compressed but not cooled (green line), cooled but not
compressed (red line) and both (blue line). XRD settings: Scan step size: 0.0167, Time per step: 60
seconds).
After pressing the brick, the maximum amount of material that could be fitted into the boat after
pressing was 50 g; this is important to mention as it dictated how much sorbent was used from the
cycling experiments.
Guidance for the identification of alite and belite polymorphs has largely been taken from (Stutzman
1996). This section describes the identification of alite and alite polymorphs by peak position and
shape for three types of clinker / cement; an industrial cement, a NIST cement standard reference
The main peaks for alite and belite (~ 31.5 – 45 º 2θ) overlap rendering this range unusable for
identification of either phase. Therefore the polymorphs of alite thought to commonly occur in
industrial clinkers based on their geometry (M1, M3 and T1) are distinguished by the diffraction lines
132
well-defined doublet at 51.7; and the triclinic T1 polymorph exhibiting a triplet between 51 and 52 º
2θ. Figure 5-4 compares this peak for the three types of clinker/cement. For the industrial clinker, a
slight doublet is visible indicating that the alite polymorph in this clinker was M3. The other two
clinkers exhibited a slight shoulder, suggesting these were also a doublet. This was confirmed by
observing the same peak for the pure alite phase produced for the calibration line (see section 5.3.3)
(Figure 5-4). This indicates that the peak is a doublet, and as such both this and therefore the
laboratory clinker were of the M3 polymorph (provided the selective extraction procedure described in
Figure 5-4. XRD scan from 49 – 54º 2θ showing the largest available alite
peak at 51.7 2θ.
Quantitative analysis of β-C2S is possible at the 31º 2θ peak, with a much weaker but still resolvable
peak residing at 35.2 º. For identification of the interstitials, the C3A cubic form exhibits strong
singlet peaks at 33.3, 47.7 and 59.4, while the orthorhombic form exhibits splitting of the strong peak
at 33.3 into a strong singlet at 33.2 and a weaker close doublet at 32.9 – 33. The XRD scan in Figure
5-5 exhibits a singlet peak at 33.3 suggesting that the C3A in the laboratory clinkers is cubic; singlet
133
Intensity (counts)
10000
6400
3600
1600
400
0
31.2 31.4 31.6 31.8 32.0 32.2 32.4 32.6 32.8 33.0 33.2 33.4 33.6 33.8
2Theta (°)
The largest available peak for alite is at 30.4 ° (the larger peak at 51.7 ° cannot be used as the alite
polymorph produced in the laboratory has been shown to be a doublet peak). Any changes in the
proportion of alite and belite can be determined by calculating the ratio of the two background-
subtracted peak area count intensities given that dividing through in this way negates any of the
matrix absorption effects (as described in section 5.2.1). However, to determine the proportion of the
two phases, single phase corundum (α-Al2O3) can be mixed with the clinker samples in a 1:1 ratio.
The ratio of the alite and belite peak area intensities to the corundum peak indicate the abundance of
that phase using the RIR value taken from the ICCD database However, using the RIR values from
the PDF cards can only be deemed semi-quantitative given that the values have been determined on a
different machine.
RIR values for alite can be found experimentally by fabricating alite as a single phase and producing a
calibration curve for the phase of interest. Single phase alite was produced by a selective extraction
procedure (Stutzman 1996). The first stage was to homogenise, press and fire commercial cement
with CaCO3 powder to convert all belite in the clinker into alite. The amount of CaCO3 to be added
needed to be such that all belite was converted to alite without any CaCO3 remaining. This was
difficult given that the proportion of belite in the commercial clinker was unknown. An approximate
134
figure for commercial clinkers of 20 % was used indicating that 4 g of CaCO3 was needed; XRD
analysis confirmed that very little CaCO3 was present in the resulting alite sample. The remaining
phases (C3A, C4AF and CaSO4) were then removed using a potassium hydroxide (KOH) / sugar
selective extraction procedure. 30 g KOH pellets and 30 g sucrose were dissolved in 300 ml DI water
at a temperature of 95 °C. The solution was filtered using a 0.45 μ Whatman filter and Buchner
funnel via a vacuum pump. The remaining residue was washed with 50 ml of DI water, followed by
The alite was then combined with a matrix material under different proportions (20, 50 and 70 %
respectively) to produce a 1 g sample which was then mixed with 0.13 g of an internal standard
(CeO2) (Table 5-2). A calibration curve was produced by plotting the ratio of the alite peak intensity
over the CeO2 peak intensity plotted against the known phase proportions of alite (Figure 5-6).
Various matrix materials tested (MgO, CaCO3 and SiO2) indicated that interference between peaks
was an issue in the production of the calibration curve. It was found that Fe2O3 was able to be used as
a matrix material without the issue of interference with the alite or CeO2 peak.
135
5.3.4. Quantification of clinker phases proportions using XRF
An attempt was made to quantify the proportion of phases in the clinker using the Bogue Equation
(see: Background) by identifying the proportion of the four major compounds (CaO, SiO 2, Al2O3 and
Fe2O3) via XRF. It was found that the equation outputted the % wt alite at over 100 %, which is
clearly not possible. Although the Bogue equation is known to overpredict % wt alite in clinker,
discussion with other users of XRF for similar purposes in the department suggested that this is due to
the sample being in powder rather than pressed pellet form which is known to introduce error and
prevent quantitative analysis. Therefore it was decided that XRD would remain the analytical method
5.4. Results and Discussion: % wt. of alite in clinkers produced from cycled
sorbent
The production of spent sorbent described in Chapter 3 was then used in the production of clinker to
determine whether cycling and the associated changes in trace element inventory of the sorbent
Figure 5-7 shows the results for the % alite in clinkers produced from cycled sorbent without fuel and
in the presence of 2 g La Jagua being combusted per cycle. These results were generated by mixing
the samples with corundum in a 50:50 mixture. For the clinker produced from sorbent cycled without
fuel no change was detected outside the experiment limits up to fifteen cycles. However, clinker
produced from sorbent cycled in the presence of La Jagua batch fed coal indicated an increase over
136
Figure 5-7. % wt. alite in clinkers produced from sorbent cycled without fuel and in the presence of
La Jagua coal combustion (2 g / cycle). Error bars shown as RSD.
This suggests that the elements reported to have increased (see: Chapter 4, section 4.4.2) may be
impacting on the % alite in the resulting clinkers. However, the high variability in the clinker
produced from sorbent cycled without fuel suggests that increase observed in the clinker produced
from batch-fed sorbent may be due to experimental error. The variations in the size of the error bars
also suggest that this could be the case. This is further discussed below.
Figure 5-10 shows the results for the % alite in clinkers produced from cycled sorbent in the presence
of continuously fed fuel; La Jagua coal, Lea Hall coal and RDF respectively. These results were
generated by mixing the samples with 0.13 g of CeO2 and using the RIR value calculated from Figure
5-6 to determine the % wt of alite. The results (Figure 5-8) show that no change was detected outside
the experiment limits up to eight cycles for the coals and up to five cycles for the RDF respectively.
137
Figure 5-8. % wt. alite in clinkers produced from sorbent cycled in the presence of a) continuously
fed La Jagua coal, b) continuously fed Lea Hall coal and c) RDF.
As such, the increase in % alite seen in the batch-coal experiments described in the previous section
may be due to experimental error rather than due to the introduction of trace elements. This would be
an expected result given that the change in the trace element inventories reported in Chapter 4 (section
4.4.2) are insignificant in comparison to the changes required to affect a change in % alite detailed in
5.5. Conclusions
Methods have been developed to enable clinker to be produced in the laboratory with a % wt
Methods have been developed to enable the intensity of X-ray diffraction peaks to be
converted to wt % alite.
XRF was not useful in accurately calculating the wt % alite in the clinkers using the Bogue
equation; any further analysis using XRF would require pelletisation of the samples.
Clinkers produced from sorbent repeatedly cycled in the absence of fuel did not lead to a
change in the % wt alite, nor did cycling in the presence of continuously fed coal nor semi-
138
However, repeated cycling in the presence of 2 g of La Jagua did appear to effect a % wt
change in alite over the course of fifteen cycles. This is either due to the incorporation of
Larger numbers of clinker samples are required to enable definitive conclusions to be drawn
with regards to the change in % wt. alite, in particular for the result with 2 g La Jagua batch-
139
Chapter 6. Performance Tests
The aim of this chapter is to determine whether cement produced from cycled sorbent differs in
mechanical performance to that of standard OPC. Due to the limited amount of sorbent that can be
produced from the cycling experiments (~ 56 grams), the first stage is to miniaturise the tests by
reducing the quantity of cement required whilst ensuring the test still complies with parameters set out
in the standards. The tests used in this investigation are standard consistency (to determine the water /
cement (w / c) ratio, cement setting time and compressive strength. The procedures for these tests are
covered in British / European standards BS-EN 197-3 (standard consistency and setting time) (BSI
2008) and BS-EN 197-1 (compressive strength) (BSI 2005) respectively. These tests are also covered
in the equivalent American standards ASTM C187 (standard consistency) (ASTM 2006), C191
(setting time) (ASTM 2008) and ASTM C109 (compressive strength) (ASTM) and respectively.
Consistence describes the state of fresh concrete, which is the firmness of form of a substance or the
ease with which it will flow (Neville 2010). The consistence or workability of concrete is important
as it determines whether proper compaction can be achieved, and also the ease with which concrete
can be transported, placed and finished without segregation. These can ultimately affect the strength,
stability and durability of the concrete mix. Workability also allows for practicality in planning for
construction works where contractors are able to plan and complete work on time before setting
occurs.
140
6.2.2. Cement Setting Time
Setting time refers to the stiffening or change of phase of cement paste from a fluid to rigid state. In
determining the setting time of cement, a cement paste of a standard consistence is immersed in water.
The change of phase of cement is caused by selective hydration of C3A and C3S where the silicates
and aluminate in cement form products of hydration or hydrates. These reactions are accompanied by
a temperature rise of the cement paste. Setting times are defined by initial and final setting time. The
initial setting time corresponds to a rapid rise in temperature whilst final setting time refers to the peak
of the temperature rise (Neville 1991). Both tests measure the time it takes for the cement paste to
The initial setting time of the cement paste is determined when the 1 mm diameter Vicat needle (see:
section 6.3.4 below) under prescribed weight penetrates the paste and measures 6 mm from the base
plate. There is a tolerance of ± 3 mm allowable by the British Standards. The initial setting time is
recorded to the nearest 5 minutes from the time of cement first mixing with water. A minimum time
of 45 minutes is prescribed by BS EN 197-1 for cement of strength classes 52.N and 62.5 N whilst 60
minutes is prescribed for strength classes of 32.5 N and R and 42.5 N and R (BSI 2005). The final
setting time of the cement paste is determined by using a needle with ring attachment of diameter ~ 5
mm. The final setting time is when the needle penetration does not mark the specimen surface with a
complete circular impression, i.e. the needle does not penetrate the cement paste for more than 5 mm.
ASTM C191 (ASTM) provides a predictive method to calculate the final and initial setting time of the
cement paste (Eq. 6-1) where E = time in minutes of last penetration > 25 mm; H = time in minutes of
first penetration < 25 mm; C = penetration reading at time E; and D = penetration reading at time H:
(( ) ) Eq. 6-1
There are several factors that can affect the setting time of cement paste. These include cement
cement paste which in turn effects the setting times. For example, low C3S and C3A contents are
found to reduce excessive heat during hydration giving longer initial and final setting times. Lower
fineness cement also produces slow hydration and reduces temperature rise (Moser). The British
Standard specifies that the allowable tolerance for temperature is ± 2 °C and ± 1 °C for mixing and
storing cement paste respectively with relative humidity in the range of 60 % - 90 %. To comply with
the ASTM standard, two tests on similar pastes should not a have Vicat initial time difference of more
than 34 minutes and a Vicat final time difference of more than 56 minutes (ASTM).
Strength of concrete is widely considered to be its most valuable property as it usually gives an
overall picture of concrete quality because it is directly related to the structure of cement paste
(Neville 1991). Compressive strength is defined as the maximum resistance of a concrete sample to
an applied pressure and is the most common measure of performance when determining the suitability
of concrete in the design of buildings and other structures. It can vary from of lows of 17 MPa and
can exceed 70 MPa for certain applications. Results from compressive strength tests are used to
determine the concrete mixture as delivered to meet the specified strength as required of any job.
Concrete mixture is designed such that the average strength is sufficiently higher than the specified
strength. To comply with the strength requirements of a job specification, the average of 3
consecutive tests must equal or exceed the specified strength, and no single test should fall below the
specified strength by more than 3.45 MPa or > 10 % when the specified strength is greater than 35
MPa. The compressive strength of a sample is calculated by dividing the failure load by the cross-
sectional area resisting the load. Table 6-1 shows the mean compressive strengths of different cement
142
Table 6-1. Compressive strengths of different cement types according to European
classification(Neville 1991).
Compressive I IA II IIA III IIIA IV V
strength,
MPa (psi)
1 day - - - - 12.0(1 10.0 - -
740) (1450)
3 days 12.0 10.0 10.0 8.0 24.0 19.0 - 8.0
(1740) (1450) (1450) (1160) (3480) (2760) (1160)
7 days 19.0 16.0 17.0 14.0 - - 7.0 15.0
(2760 (2320) (2470) (2030) (1020) (2180)
28 days - - - - - - 17.0 21.0
(2470) (3050)
Strength, as well as durability and volume changes of hardened cement paste, depends not so much on
the chemical composition as on the physical structure of the products of hydration of cement and on
their relative volumetric proportions. A primary factor is porosity, i.e. the relative volumes of pores
or voids in the cement paste, which can be viewed as sources of weakness. There is a direct inverse
Other sources of weakness arise from the presence of aggregate which itself may contain flaws as
well as causing micro-cracking at the interface with the hydrated cement paste. As aggregates
represent three quarter of the total volume of concrete mortar, its properties such as size, shape and
porosity can seriously influence the overall strength of cement. Aggregate particles contain pores that
143
influence the strength of bonds formed at the interface of the aggregate and hydrated cement paste. It
is understood that rougher textures or more angular shaped aggregates can result in a stronger bond
leading to less micro-cracking at the interface. Larger aggregate particles are also known to require
less water than finer aggregates during wetting when hydration occurs. This demand alters the
effective water / cement ratio of the cement paste. A finer mix of aggregate can then decrease this
The OPC used in tests is Type I and has the strength class 32.5 R.
Pre-packed 1.35 kg BS EN 196-1 sand packs manufactured by David Ball Specialist Sands were used
in these experiments. Each pack consists of more than 98% rounded silica particles with size ranging
6.3.3. Mixer
A Kenwood Chef mixer was used for cement paste mixing. A plastic bowl of width 20 cm and height
18 cm and an approximate capacity of 5 litres are used. The stainless steel blade revolves about its
own axis and is driven in a planetary movement around the axis of the bowl at various speeds (Table
6-2).
144
6.3.4. Vicat Apparatus
There are two apparatus that that can be used to determine the
(ASTM 2008) and ASTM C266 - 08e1 (ASTM 2009b) for Vicat
experiments the Vicat needle (Figure 6-2) has been designated as the
Figure 6-2. Vicat
apparatus of choice. apparatus(CMT_Equipment
2012).
consists of a weight attached to moving parts with interchangeable ends. The total mass of the
145
Figure 6-3. Schematic of the Vicat apparatus.
Schematics of the three interchangeable ends that can be attached to the moving parts are given in
Figure 6-4. The plunger is made from steel in form of a right cylinder with 45 mm length and 10 mm
diameter. The needle is also made from steel in the form of a right cylinder with 45 mm length and
The Vicat mould is used to contain the cement paste during the standard consistence and setting time
tests. It is made from plastic and shaped as a truncated cone with an internal diameter of 75 ± 10 mm
and 40.0 ± 0.2 mm deep. The mould is paired with a base-plate, made from metal with a thickness of
2.5 mm.
The jolting apparatus used has been custom-built to comply with the specifications of BS EN 196-1
(BSI 2005).
146
6.3.7. Compressive Strength Tests
The EZ-50 Universal Materials Testing Machine from Lloyd Instruments was used to test the
compressive strengths of the concrete samples. The EZ-50 machine is capable of testing applications
Prior to testing, the compressive strength testing gear is put in place. The de-moulded concrete
sample is placed onto the machine and hard cardboard or plastic sheets are placed surrounding the test
area to avoid release of test samples. As the test is running, data is gathered by the NEXYGENPlus
Data Analysis software. Here, further data analysis on the stress, strain and load of the sample can be
Miniaturization of standard consistency, setting times and compressive strength tests were carried out
owing to the limited amount of sorbent produced in the cycling experiments. For descriptions of the
The standard consistency test was initially carried out using standard mould sizes according to the
standard; the results indicated a required w / c ratio of 2.5. The sizes of moulds were then reduced
and the standard consistency test repeated. The test utilized a plunger apparatus of diameter (10 ±
0.05) mm to penetrate the mould sample. Hence, it was decided that the minimum diameter of the
miniaturized mould should not be any less than 26 mm to provide a minimum 8 mm of space from the
plunger to the walls of the mould. The minimum mould diameter size available above 26 mm was
147
found to be 31 mm. Upon miniaturization of the standard mould to a 31 mm diameter cylinder with a
height of 40 mm, the w / c ratio required to pass the standard consistence test was found to have
Setting times tests were then carried out with OPC using sample sizes set out in BS EN 196-3 (BSI
2008) to test the standard method. Next, moulds of smaller diameters were sourced or produced and
each size was tested to see whether the same results could be achieved. The heights of each mould
were kept at 40 mm to accommodate the operating specifications of the Vicat apparatus. Steel or
plastic pipes were chosen as appropriate moulds given their availability and low cost. However, given
that only standardized pipe sizes are manufactured, the range of pipe diameters available for use was
limited. At this size, the setting times test using the Vicat apparatus matched the results of the
standard tests. The final size of 31 mm still garnered initial and final setting times within the limits of
the standard.
In the miniaturization process, mixing by hand was also attempted but the resulting cement paste was
found to be greatly inconsistent and difficult to reproduce. Hence, the sizes of moulds were also
limited to the minimum amount of cement paste that could be mixed with the standard mixing
OPC concrete prisms were first made according to BS EN 196-1 (BSI 2005) and tested for their
compressive strength. However, it was later concluded that the prisms were too long to produce
accurate testing results due to the need to miss out a flexural strength test. The flexural strength test
was important to reduce the length of prisms to produce a prism with an appropriate length to width
ratio suitable for compressive strength testing. The ratio most commonly used is 2:1. However, with
a limited supply of spent sorbent available for making concrete, it was decided that a ratio of 1:1
148
would be more suitable as it required less material. This led to a redesign of the BS EN 196-1
standard mould to produce cubic concrete blocks. This was done in accordance with the guidelines
40 mm.
The concrete blocks were then tested for their compressive strength and compared to the minimum
required strengths as listed in BS EN 197. Further miniaturization of cubic moulds was attempted but
due to limitations in design and time constraints, it was concluded that cylindrical moulds would be
more suitable for miniaturization and testing. This also allowed a comparison between the
Cylindrical moulds were produced using both steel and plastic drain pipes due to their availability and
easy manipulation of length. The dimensions were decided in accordance with BS EN 12390-1:2000
and a length to diameter ratio of 2:1 was used. Cylindrical concrete blocks were made using several
moulds of different diameter. The moulds were then attached to the jolting table’s base plate with
duct tape to keep the samples upright (Figure 6-5). This ensured a more consistent production of
Each concrete sample was tested for its compressive strength. The strength of each block was shown
to match or exceed the minimum strength required as listed in ASTM C150-04. Miniaturization of
149
mould sizes was continued until a minimum size was found. The minimum size was determined
when further miniaturization of mould size failed to match the required minimum strength. This was
found to be 31mm. This miniaturization effectively decreased the amount of cement required for the
OPC cement paste tested for standard consistence, in standard mould sizes as according to BS EN
196-3 was found to utilize a w / c (water / cement) ratio of 0.25. This ratio was obtained by
increasing the water content in 0.05 increments from a ratio of 0.2, while performing standard
consistence tests on each specimen. As previously stated upon miniaturization of the standard mould
to a 31 mm diameter cylinder with a height of 40 mm, the w / c ratio required to pass the standard
consistence test was found to have increased from 0.25 to 0.275. Standard consistency testing was not
carried out for cement produced with spent sorbent due to the limited amount of sorbent available.
Instead, a w / c ratio of 0.275 was assumed for the setting times test.
Utilizing a w / c ratio of 0.275, the average initial setting time for OPC cement over 7 samples was
found to be 134 minutes using the miniaturised moulds. The final setting time for the same number of
samples was found to be 179 minutes. The initial and final setting times for cement produced with
spent sorbent was found to be 75 and 240 minutes respectively. The cement used was made from
150
single-cycle sorbent cycled in the absence of fuel. The setting times results are given in Figure 6-6
below.
These results fall within the standard range where initial setting time is required to be above 60
minutes and final setting has to occur before 600 minutes as reported in BS EN 197-1.
The standard consistence and setting times of cycled-cement sorbent is found to fall within the
required standards. This means that the concrete mix is sufficient in terms of workability and allows
for full compaction which is vital in achieving the maximum possible density. As the strength of
concrete is highly influenced by the presence of voids / porosity, a high density is preferred to
First the samples produced from OPC were tested for their compressive strength. Three 40 mm cubic
samples were produced for both 3-day and 7-day testing along with four 31 mm cylindrical samples.
The 3-day samples fractured and ultimately failed during testing. However, all 7-day test samples
were able to withstand the maximum load of 40 kN as safely allowable by the compressive strength
machine, and hence resisted total failure. As such there are no results to display for the 7-day test.
151
Figure 6-7. Results of the strength test for different shaped-samples
(OPC, 3-day test)
The stress-strain behaviour of all samples were very similar (see Figure 6-7 and Figure 6-9,
Appendix) and all samples were found to perform above the minimum required strength of 12 MPa as
stated in the British Standard. (Figure 6-7). Notably, the cubic samples were found to have
compressive strengths ~ 25 % higher than the cylindrical samples. However they also required 30 %
more cement. Therefore the cylindrical moulds were chosen for testing of the cements produced from
For cylindrical samples at a height to diameter ratio of 2, the de-moulded samples were found to have
uneven surface areas at the axial planes, with visible marks of uneven hydration, as shown in Figure
6-8. However this does not appear to have impacted on the strength of the cylinders produced, at least
not to the extent that the cylinders failed before the minimum 12 MPa.
152
a. b.
Next, the cement samples produced from cycled sorbent were tested. Due to the limited quantity of
sorbent, compressive strength tests were only conducted on 5-cycled sorbent cement with fuel, whilst
it was possible to perform the test on no fuel sorbent-cements up to 15 cycles. Physical inspection
showed that the produced cement samples were also found to have uneven surface area due to poor
Figure 6-9. Average compressive strengths for different cycled-sorbent cement. OPC=Ordinary
Portland Cement; 1C NF= 1 cycle, no fuel; 5C NF=5 cycle, no fuel; 10C NF. Average of 3 samples
with error bars shown as the standard deviation. Red dotted line enables comparison with OPC.
153
Figure 6-9 illustrates the average compressive strengths of each cycled-cement in comparison with the
OPC. It shows that all samples tested clearly passed the minimum strength requirement of 12 MPa
for the 3-day test as stated in ASTM C150-04a. It can also be observed that all cycled-sorbent cement
performed better than OPC with the exception of 10-cycled sorbent. On average, cement from cycled
sorbent result in a compressive strength value 4.7 % higher than OPC. The standard deviation for
each cement type was calculated, and was found to vary between 1.0 and 5.8.
The cement produced from 5-cycle sorbent with fuel was found to differ from 5-cycle cement
produced from sorbent without fuel by only 0.2 MPa. This represents a difference of only 1%. As
such it is reasonable to assume that the introduction of additional fuel impurities into the cement did
not have an impact on the strength of the cement. However, further testing would be required to draw
Figure 6-10 and Figure 6-11 (Appendix 6-1) allow comparison of the deformation and strength
characteristics of OPC and cement from cycled sorbent to be made. It shows that OPC and cycled-
sorbent cement have similar non-linear stress-strain behaviours. All specimen curves start with an
upwards slope that is followed by a linear ramp-up which is more steep. A loss of linearity is then
observed before the peak or maximum load is reached. This loss of linearity can be understood to be
an indication of the initiation of the fracture process within the test sample.
Generally speaking, it can be seen that there exists a downward trend in sample strength as the
number of cycles increase. A study by Blamey et al. (Blamey, Anthony et al. 2010) discovered that
repeated cycling of sorbent causes higher degrees of CaO particle sintering, or changes in pore shape
and shrinkage. This is thought to decrease the rate of hydration of samples as water encounters more
resistance while filling up the pores. This is turn would hamper the hydration reactions and build-up
of solid hydration products. Subsequently, this could affect the degree of hydration which is a main
indicator of porosity and overall strength. As such, this could explain why higher numbers of cycles
154
6.6. Conclusions
Successful new methods involving miniaturised test specimens have been developed,
allowing comparative studies of setting times and compressive strength tests to be conducted
for OPC and sorbent-cycled cements. Only 75 g of cement are now required to produce one
sample for setting time and compressive strength tests using a mould size of diameter 31 mm.
Hand mixing of this small amount of mortar was unsuccessful therefore the procedure still
Standard water / cement ratio of the miniaturised OPC sample was found to be 0.275.
The initial and final setting times of 1-cycled sorbent sample was found to be 75 and 240
minutes respectively which fell within the standard range recorded in BS EN 197-1.
Cubic samples have been shown to be 25 % stronger than their cylindrical equivalent
following miniaturisation.
Compressive strength conducted on all sample specimens passed the minimum required
strength as stated in the ASTM C150-04a. Cements made from cycled sorbent were found on
Due to the small number of test sample size conducted, these initial results provide
155
6.7. Limitations of the experimental method
standard methods and procedures to be made. The tests were carried out using rudimentary
equipment, as deemed reasonable for a preliminary study. Further work would enable the
the factors leading to the inconsistent quality of samples. In the compaction of specimens
during moulding, the influence of uneven and non-uniform compaction across the entire
sample was apparent once demoulding of the sample was carried out. This irregular
compaction could have been caused by a design flaw in which the base plate of the jolting
table was not horizontal when rested on the stop during the jolting process. Although this flaw
is minimal in practical sense, the quality of compaction that could be provided is clearly
affected.
Miniaturisation leading to wall effects. Another factor that has contributed to the quality of
samples is the uneven finishing at the top surface. It was observed that as the size of the
moulds decreased, wall effects became more apparent and finishing of samples was less
perfect. In certain cases, the top surface was elliptical when demoulded.
Limited number of specimens. Production of test specimens was severely limited due to the
availability of cycled sorbent. In most cases, only 3 samples of each specimen batch were
available for testing. This small sample size, although able to provide some initial trends,
156
6.8. Further Work
The number of test samples should be increased to enable results that would be of statistical
significance.
Compressive stress tests should be carried out for 28 days samples to utilize available
samples.
Further studies in water / cement ratio, flexural strength and degree of hydration should be
performed.
157
Chapter 7. Conclusions and Further Work
7.1 Conclusions
The Ca-Looping system has been modeled using a stochastic population balance model, a method
which is able to garner similar results for the age distribution (rN), the maximum average reactivity of
the sorbent ( ̅ ) and the CO2 capture efficiency of the carbonator (ECO2) as those published in
literature by INCAR. This method has been furthered to run on a time basis as opposed to per cycle
and to incorporate the carbonation rate and extent of carbonation from experimental data rather than a
simplified kinetic model. However this part of the model requires further work to bring the results for
ECO2 closer to those published in literature. A generic dry-process cement plant mass and energy
balance has been developed which resulted in a figure of 4.34 MJ for the thermal efficiency of
producing 1 kg of clinker. This is within the top 90th percentile of kilns globally; it is thought that this
lower thermal efficiency is due to the size the tertiary air stream. Integration of the Ca-looping
process with the cement plant indicates that the thermal efficiency of producing 1 kg of clinker can
range from 4.49 – 5.83 MJ depending on whether the system supplies itself with electricity through
waste heat recovery or whether electricity is supplied from the grid. Up to ~ 36 MW electricity may
also be passed back to the grid, however this results in the lowest thermal efficiency of producing 1 kg
of clinker of 5.83 MJ. For all configurations 92 % of all CO2 produced at the kiln is captured which
A laboratory-scale system has been designed, built and commissioned capable of producing spent
sorbent following a given number of cycles of calcination and carbonation. A range of fuel feeding
systems have been tested and included in the commissioning process. The mass balance on the first
158
The results of the single cycle experiments indicate that Sr is easier to extract from CaCO3 than from
Ca(OH)2, whilst Ba and Mn were both easier to extract from Ca(OH) 2 than from CaCO3. The results
of these experiments also suggest that Cr contamination from the thermocouple is a strong possibility
though well below any limit likely to impact on clinker composition. Sorbent cycled in the absence of
fuel does not undergo a measurable change in its trace element inventory up to fifteen cycles. The
fuel systems designed and built as part of this project are effective in introducing the fuel into the
system and in enabling the ash to mix with the sorbent during combustion as indicated by the
concentrations for Al in the sorbent cycled with fuel. The decrease in the % retention for Al and other
elements with reasonable RSD values suggest that an equilibrium is reached for the amount of ash
entering the system after a certain number of cycles. The results show that B would need further
observation given its observed increase over eight cycles and its known role in the formation of alite
crystals. The results for Na indicate that this element could have implications in terms of volatile
cycles developing both in the Ca-looping calciner and the cement kiln. Such features of the system
could limit the amount of sorbent that can be passed directly to a cement kiln.
Methods have been developed to enable clinker to be produced in the laboratory with a % wt of alite
comparable to commercial clinkers. Methods have been developed to enable the intensity of X-ray
diffraction peaks to be converted to wt % alite. Clinkers produced from sorbent repeatedly cycled in
the absence of fuel did not lead to a change in the % wt alite, nor did cycling in the presence of
continuously fed coal nor semi-continuously fed RDF. However, repeated cycling in the presence of
2 g of La Jagua did appear to effect a % wt change in alite over the course of fiftten cycles. This is
potentially due to the incorporation of impurities from the fuel in the sorbent.
Successful new methods involving miniaturised test specimens have been developed, allowing
comparative studies of setting times and compressive strength tests to be conducted for OPC and
sorbent-cycled cements. Only 75 of cement are now required to produce one sample for setting time
and compressive strength tests using a mould size of diameter 31 mm. Hand mixing of this small
amount of mortar was unsuccessful therefore the procedure still depends on the use of the automated
159
mixer. Standard water / cement ratio of the miniaturised OPC sample was found to be 0.275. The
initial and final setting times of 1-cycled sorbent sample was found to be 75 and 240 minutes
respectively which fell within the standard range recorded in BS EN 197-1. Cubic samples have been
strength conducted on all sample specimens passed the minimum required strength as stated in the
ASTM C150-04a. Cements made from cycled sorbent were found on average to be 4.7% higher in
compressive strength than OPC. Due to the small number of test sample size conducted, these initial
results provide indications; future work should include a higher number of test samples.
The stochastic model requires further work to bring the results for ECO2 closer to those
published in literature.
Incorporation of the effects of sulphation and ash need to be included into the stochastic
model.
Further development of the fluidised bed heating system would ensure the entire calcination
Further work would include to ascertain why when digesting NIST SRM 1d (argillaceous
limestone) experimental values for Na and K were much higher than the certified values.
Further work would include producing and testing sorbent with a wider range of solid fuels to
see which other elements may be of interest in terms of their retention in the sorbent and their
160
Larger numbers of clinker samples are required to enable definitive conclusions to be drawn
with regards to the change in % wt. alite, in particular for the result with 2 g La Jagua per
cycle.
The number of performance test samples should be increased to enable results that would be
of statistical significance.
Compressive stress tests should be carried out for 28 days samples to utilize available
Performance test specimens should be capped or grinded to provide greater consistency in the
shape of samples.
For the performance tests, further studies in water / cement ratio, flexural strength and degree
161
Appendix 2-1. Matlab code for the steady-state stochastic model
CcHrat_x = [];
CcHrat_s = [];
CcHin_s = [];
xave = [];
xcarb = [];
XCyc = [];
FRs = [];
%% MODEL PARAMETERS
no_cyc = 10000;
Eff = 0.7;
FCO2 = 1;
CaO_pp = 1/1000;
no_p_init = 1/CaO_pp;
% HBr-doped 13 cycles
% Xr=0.1264;
% a=1.6238;
% Grasa
% Xr=0.2182;
% a=0.8574;
% Grasa
Xr=0.075
a= 0.52
% Undoped 50 cycles
% Xr=0.0454;
% a=0.7212;
% HBr-doped 50 cycles
% Xr=0.1264;
% a=0.6956;
%% HEAT VALUES
% enth_Cp = NASA_Glenn;
% enth = enth_Cp(:,:,1) %kJ/mole
% Cp = enth_Cp(:,:,2) %kJ/mole/K
% For_Heat_Reaction
% Rev_Heat_Reaction
deltaHf_CaCO3_1223=-1100.89;
deltaHf_CaCO3_923=-1138.92;
deltaHf_CaCO3_298=-1206.61;
deltaHf_CaO_1223=-588.805;
deltaHf_CaO_923=-604.21;
163
% DC_NF3 =
[[[[0.428515625000000,0.363501439000000,0.330254898000000,0.313230989000000
,0.304289844000000,0.298452259000000,0.293200762000000,0.287717424000000,0.
281930435000000,0.275723404000000,0.269637282000000,0.263434758000000,0.257
539380000000,0.251919856000000,0.246416143000000,0.241304319000000,0.236380
211000000,0.231539449000000,0.227094017000000,0.222995748000000,0.218904741
000000,0.215104537000000,0.211526487000000,0.208085461000000,0.204758872000
000,0.201642910000000,0.198480362000000,0.195349262000000,0.192430211000000
,0.189567573000000]]],[linspace(0.189567573000000,0.15,no_cyc-30)]];
%
% % Decay Curve NF4 - Precipitated CaCO3 (PCC16 'Average Data', 50% CO2
constant)
% DC_NF4 =
[[0.410432210000000,0.397836621000000,0.382362341000000,0.367062136000000,0
.353302005000000,0.340903570000000,0.329768645000000,0.319751711000000,0.31
0593807000000,0.302342266000000,0.294884819000000,0.287793125000000,0.28133
1231000000,0.275150986000000,0.269473015000000,0.264138108000000,0.25909893
4000000,0.254341408000000,0.249762652000000,0.245503098000000,0.24148803100
0000,0.237674422000000,0.234025499000000,0.230559649000000,0.22721115200000
0,0.224139610000000,0.221185814000000,0.218551220000000,0.215808660000000,0
.213155289000000],[linspace(0.213155289000000,0.2,no_cyc-30)]];
%
% % Decay Curve NF5 - Precipitated CaCO3 (PCC16 'Average Data', 50% CO2
constant)
% DC_NF5 =
[[[0.472195292000000,0.446153579000000,0.429463322000000,0.416460146000000,
0.405331780000000,0.395367066000000,0.386290971000000,0.377784588000000,0.3
69710727000000,0.362190862000000,0.354918854000000,0.348016503000000,0.3413
88859000000,0.334989753000000,0.328786773000000,0.322780572000000,0.3169943
98000000,0.311385031000000,0.305858174000000,0.300555600000000,0.2953312790
00000,0.290295880000000,0.285433686000000,0.280661533000000,0.2760979470000
00,0.271734742000000,0.267591564000000,0.263208714000000,0.259195194000000,
0.255256653000000]],[linspace(0.255256653000000,0.22,no_cyc-30)]];
%
% % Decay Curve NF6 - Precipitated CaCO3 (PCC16 'Average Data', 15% CO2
carb, 100% N2 calc)
DC_NF6 =
[[0.482700781000000,0.470601391000000,0.453828698000000,0.438135703000000,0
.424121977000000,0.412051138000000,0.401179913000000,0.391262287000000,0.38
2218794000000,0.374055620000000,0.366040435000000,0.358356916000000,0.35151
2792000000,0.344657725000000,0.338059139000000,0.331552393000000,0.32530831
4000000,0.319401136000000,0.313620057000000,0.307844213000000,0.30242858600
0000,0.296921596000000,0.291774347000000,0.286701331000000,0.28192571900000
0,0.277218154000000,0.272773733000000,0.268517747000000,0.264467329000000,0
.260582505000000],linspace(0.260582505000000,0.23,no_cyc-30)];
%
% % Decay Curve NF7 - Precipitated CaCO3 (PCC16 'Average Data', 50% CO2
carb, 100% N2 calc)
% DC_NF7 =
[[0.472195292000000,0.446153579000000,0.429463322000000,0.416460146000000,0
.405331780000000,0.395367066000000,0.386290971000000,0.377784588000000,0.36
9710727000000,0.362190862000000,0.354918854000000,0.348016503000000,0.34138
8859000000,0.334989753000000,0.328786773000000,0.322780572000000,0.31699439
8000000,0.311385031000000,0.305858174000000,0.300555600000000,0.29533127900
0000,0.290295880000000,0.285433686000000,0.280661533000000,0.27609794700000
0,0.271734742000000,0.267591564000000,0.263208714000000,0.259195194000000,0
.255256653000000],linspace(0.255256653000000,0.23,no_cyc-30)];
% % Decay Curve VM1 - Pre-sintered Sorbent @ 900C - Manovic & Anthony 2008
- Figure 3
164
% DC_VM1 = [[0.65 0.71 0.738 0.754 0.762 0.762 0.75 0.741 0.7238 0.7512 0.7
0.6848 0.6349 0.6463 0.6377 0.6197 0.605 0.5186 0.5673 0.553 0.5337 0.5194
0.509 0.49 0.481 0.475 0.463 0.452 0.444
0.4358],linspace(0.4358,0.35,no_cyc-30)];
%
% % Decay Curve VM2 - Pre-sintered Sorbent @ 900C - Manovic & Anthony 2008
- Figure 3
% DC_VM2 = [[0.65 0.71 0.738 0.754 0.762 0.762 0.75 0.741 0.7238 0.7512 0.7
0.6848 0.6349 0.6463 0.6377 0.6197 0.605 0.5186 0.5673 0.553 0.5337 0.5194
0.509 0.49 0.481 0.475 0.463 0.452 0.444
0.4358],linspace(0.4358,0.35,no_cyc-30)];
for F0set=1:length(F0_mfr)
for p=1:length(F0_mfr{F0set}(1,1:end))
F0 = F0_mfr{F0set}(p);
age = ones(1,no_p_init);
[CcHrat_s, xave, XCyc, age, FRs]=stoch(F0, CcHrat_s, xave,
XCyc, age, FRs);
if F0set==1
[CcHrat_x, xcarb]=Heat_Req_xcarb(CcHrat_x, F0, xcarb);
end
end
end
%% FUNCTIONS
function [CcHrat_s, xave, XCyc, age, FRs] = stoch(F0, CcHrat_s, xave, XCyc,
age, FRs)
for c=1:no_cyc
clear p_react
% p_react = ((1*(((1/(1-
Xr))+(a*(age))))).^-1)+Xr;
p_react = DC_NF6(age);
XCyc{F0set}(c,p) = mean(p_react);
FRCyc{F0set}(c,p) = Eff/XCyc{F0set}(c,p);
no_p_sys = FRCyc{F0set}(c,p)/CaO_pp;
age(1,:) = age(1,:)+1;
[chosen] = purge_func(age,F0,c,FRCyc);
age =
[[age],[ones(1,round(((F0/CaO_pp)+(no_p_sys-length(age)))))]];
age = age(setdiff(1:length(age),chosen));
end
xave{F0set}(p) = mean(XCyc{F0set}(no_cyc-100:no_cyc,p));
FRs{F0set}(p) = Eff/xave{F0set}(p);
CcHin_s{F0set}(p) =
(Rev_Hr*((FRs{F0set}(p)*xave{F0set}(p))+F0))+(((FRs{F0set}(p)*(1-
xave{F0set}(p))*(deltaHf_CaO_1223-
165
deltaHf_CaO_923))+(FRs{F0set}(p)*(xave{F0set}(p))*(deltaHf_CaCO3_1223-
deltaHf_CaCO3_923))))+(F0*(deltaHf_CaCO3_1223-deltaHf_CaCO3_298));
CcHrat_s{F0set}(p) =
(CcHin_s{F0set}(p)/(CcHin_s{F0set}(p)+Hcomb))*100;
end
kk=CcHrat_s(ECRA);
find(kk==min(kk));
save 20130418_results_grasa
xcarb(p,:) = linspace(1E-
10,xave{F0set}(p),50);
for x=1:length(xcarb)
x_carb = xcarb(p,x);
FRx = (FCO2*Eff)/x_carb;
CcHin_x(p,x) =
(Rev_Hr*((FRx*x_carb)+F0))+(((FRx*(1-x_carb)*(deltaHf_CaO_1223-
deltaHf_CaO_923))+(FRx*(x_carb)*(deltaHf_CaCO3_1223-
deltaHf_CaCO3_923))))+(F0*(deltaHf_CaCO3_1223-deltaHf_CaCO3_298));
CcHrat_x(p,x) =
(CcHin_x(p,x)/(CcHin_x(p,x)+Hcomb))*100;
end
end
end
166
Appendix 2-2. Matlab code for the time-dependent stochastic model
close all
clear all
MCaCO3 = 100.04;
%% Loading FB Data
% FB DATA
% convdatFB = load('LC_30cycles_nosteam_r_per_second.mat');
% [dconv1 dconv2] = size(convdatFB.convdatFBpersecond)
% convdatFBextent = load('LC_30cycles_nosteam.mat');
% convdatFB = load('LC_30cycles_nosteam.mat','conversion');
% convdatFB.conversion(400:501,23)=convdatFB.conversion(399,23)
% convdatFB.conversion(399:501,24)=convdatFB.conversion(398,24)
% convdatFB.conversion(500:535,:)=[];
tb=1;
% TGA DATA
ext_conv = xlsread('G:\backup 20120910\Work\4 Lab Work\2 Cycling
Experiments\Others\NF\TGA_NF_CONVDATA_HAVELOCK','Sheet1');
ext_conv(701:730,:)=[];
% ext_conv = convdatFBextent;
% conv_per_s = convdatFB.conversion;
% t_star=[20];
% k=0.52;
% Xr=0.075;
%
% for N=1:100
% conversion_grasa(N)=(1/((1/(1-Xr))+(k*N)))+Xr;
% ext_convrev(N,:)=linspace(conversion_grasa(N),0,t_star)
% end
%
% ext_conv=fliplr(ext_convrev)
%
% for N=1:100
% for t=t_star:(730-t_star)
% ext_conv(N,t)=conversion_grasa(N);
% end
% end
% ext_conv=transpose(ext_conv);
%
%
[dconv1 dconv2] = size(ext_conv);
pos=floor(linspace(1,dconv1,(dconv1/tb)));
for l=1:dconv2
ext_conv_10s(:,l) = ext_conv(pos,l);
end
[dconv3 dconv4] = size(ext_conv_10s);
for i=2:dconv3
conv_per_s(i,:) = ext_conv_10s(i,:)-ext_conv_10s((i-1),:);
end
conv_per_s(1,:) = ext_conv_10s(1,:);
167
for k=1:dconv4
for j=1:dconv3
if conv_per_s(j,k)<0
conv_per_s(j,k)=0;
end
end
end
% plot(
%% Particles
dp = 500/1e6; % Particle Diameter in Metres
ro = 2500; % Particle Density (kg/m3)
%% MODEL PARAMETERS
deltaHf_CaCO3_1223=-1100.89;
deltaHf_CaCO3_923=-1138.92;
deltaHf_CaCO3_298=-1206.61;
deltaHf_CaO_1223=-588.805;
deltaHf_CaO_923=-604.21;
% Cement Plant
CaORequiredpersecond = 4000*1000*0.65/24/60/60/56;
% Model Parameters
no_seconds = 36000; % simulation length
(e.g. run for one hour)
168
FR = FCO2*[1]; % Ca entering leaving
Cb (kmols-1)
F0 = FCO2*[0.01]; % Ca entering Cc fresh
(kmols-1) % for the cement plant use: 0.55;
CaO_pp = 0.1; % CaO per packet(kmol)
WCaO = (50*MW); % Solids inventory (each
unit), CaO (kg)
pkts_pu = ((WCaO/56)/CaO_pp); % Packets per unit
(moles)
ash_perpkt = Fash/pkts_pu;
S_perpkt = FS/pkts_pu;
for r=1:length(FR)
clear pdatCbpdatCc
FR_=FR(r);
packetsFR = round((FR_)/CaO_pp); % No. of particles to pass to
calciner
for p=1:length(F0)
clear pdatCbpdatCc
pdatCb = zeros(6,pkts_pu);
pdatCc = zeros(6,pkts_pu);
pdatCb(1,:) = 1;
pdatCc(1,:) = 1;
pdatCb(2,:) = 1;
pdatCc(2,:) = 1;
F0_=F0(p);
packetsF0 = ceil((F0_)/CaO_pp); % No. of particles to purge from
calciner
freshstream = zeros(6,packetsF0);
freshstream(1,:)= ones;
freshstream(2,:)= ones;
for s=1:no_seconds
169
count=count+1
pdatCb(3,:)=0;
% Length of time in current unit (s)
pdatCb(1,:) = pdatCb(1,:)+1;
pdatCc(1,:) = pdatCc(1,:)+1;
% Ecb
CO2reacting(s)=sum(pdatCb(3,:));
Ecarb(s,r,p) = CO2reacting(s)/FCO2run(s);
if Ecarb(s,r,p) > 0.92
Ecarb(s,r,p) = 0.92;
end
[chosen1] = stoch_func1(pkts_pu,packetsFR); %
function: pass packets from Cb to Cc
pdatCb(1,chosen1) = 0; %
reset residence time to zero
pdatCc =[[pdatCc],[pdatCb(:,chosen1)]]; %
append Cb matrix with Cc packets
FRdata =[pdatCb(:,chosen1)];
pdatCb(:,chosen1) =[]; %
delete packets from Cb matrix
[chosen2] = stoch_func2(pkts_pu,packetsFR); %
function: add packets from Cc to Cb
pdatCc(1,chosen2) = 0; %
reset residence time to zero
pdatCc(2,chosen2) = pdatCc(2,chosen2)+1; %
increase age of calcined particles by one
pdatCb = [[pdatCb],[pdatCc(:,chosen2)]]; %
append Cc matrix with Cb packets
pdatCc(:,chosen2) = []; %
delete packets from Cc matrix
pdatCc = [[pdatCc],[freshstream]]; %
append Cc matrix with fresh packets
170
[chosen3] = stoch_func3(pkts_pu,packetsF0); %
function: purge old packets from Cc
kg_ash_N_ratio(s,:) = pdatCc(5,chosen3)./pdatCc(2,chosen3);
avekg_ash_N_ratio(s) = mean(kg_ash_N_ratio(s,:));
pdatCc(:,chosen3) = []; %
Remove purge packets from Cc matrix
CaCO3inrecycle(s)=sum(FRdata(4,:))*CaO_pp;
CaOinrecycle(s)=(CaO_pp*length(FRdata))-CaCO3inrecycle(s);
% CcHrat(r,p) = (CcHin(r,p)/(CcHin(r,p)+Hcomb))*100;
end
Ecarbss(r,p)=mean(Ecarb((no_seconds-100):no_seconds,r,p));
end
end
LongcliffeBatch=100; % g
molesCaO=LongcliffeBatch/MCaCO3;
%% PLOTS
F0_FCO2= F0./FCO2;
% figure(1)
% Ecarbss=transpose(Ecarbss)
% plot(F0_FCO2,Ecarbss)
% legend('FR=1','FR=2','FR=3','FR=4','FR=5')
% xlabel('F0/FCO2')
% ylabel('Steady State Carbonator Efficiency (Ecarb)')
% axis([0 0.5 0 1])
%
% figure(2)
% CcHrat=transpose(CcHrat)
% plot(F0_FCO2,CcHrat)
% legend('FR=1','FR=2','FR=3','FR=4','FR=5')
% xlabel('F0/FCO2')
% ylabel('% Hin/Hin+Hcomb')
% axis([0 0.5 0 100])
save 20130502_results_50_1_001_SM_exp
171
% figure(3)
% bar(gramsfuelpercycle_experiments)
%% Stochastic Functions
chosen1=[];
counter=0;
if packetsFR>0
while length(chosen1) ~= packetsFR
chosen1=unique([chosen1;ceil(rand(counter,1)*((pkts_pu-1)))]);
counter=packetsFR-length(chosen1);
end
end
end
chosen2=[];
counter=0;
if packetsFR>0
while length(chosen2) ~= packetsFR
chosen2=unique([chosen2;ceil(rand(counter,1)*((pkts_pu-1)))]);
counter=packetsFR-length(chosen2);
end
end
end
chosen3=[];
counter=0;
if packetsF0>0
while length(chosen3) ~= packetsF0
chosen3=unique([chosen3;ceil(rand(counter,1)*((pkts_pu-1)))]);
counter=packetsF0-length(chosen3);
end
end
end
end
172
Appendix 3-1. Matlab code for fitting absorbance to concentration
clear all;clc;
filename = input('Filename? ', 's');
data = importdata(filename, ',');
abs = data.data(:,1);
conc = data.data(:,3);
clear data
a0=[2 2 0.1];
options=optimset('lsqnonlin');
a=lsqnonlin(@calibration,a0,[],[],options,abs,conc);
x=0:0.05:25;
y=a(1).*(exp(-x)-1)+a(2).*x+a(3).*x.^2;
% y=y*100;
plot(x,y);
hold on
plot(abs,conc,'x')
axis tight
xlabel('absorbance')
ylabel('calculated concentration(%)')
% axis([0 5 0 100]);
format long
sprintf('The calibration curve equation is y=%d*(exp(-x)-1)+%d*x+%d*x^2',a)
173
Appendix 6-1. Experimental Procedures
General Requirements
Standard Consistence
1. Lightly oil the cylindrical or truncated form mould and the base-plate by using mineral oil.
2. By using the plunger attachment for the Vicat apparatus, lower the plunger to rest on the base-
plate. Manually adjust the pointer or scale to read zero.
3. Raise the plunger to the stand-by position.
4. Weigh (500 ±1) g of cement with a mass balance.
5. Measure out a quantity of water (starting with 125 ml) by means of graduated cylinder or
burette.
6. Place the water and cement into the mixing bowl taking care to avoid loss of water.
7. Within 10 seconds of cement in contact with water, mix the mixture at low speed.
8. Record the time to the nearest minute as ‘zero time’.
9. After 90 seconds of mixing, stop the mixer for 30 seconds, place the paste adhering to the wall
and bottom part of the bowl to the middle of the bowl by using a plastic scrapper or a spoon.
10. Restart the mixer at low speed for a further 90 seconds.
11. Vary the amount of water in step 2 until standard consistence cement has been achieved.
1. Transfer the paste immediately into the mould placed on the base-plate.
2. Fill the mould to excess without unnecessary compaction or vibration. Gently tap the mould
against the hand to remove any voids.
3. Remove the excess with a straight-edged implement, e.g. a ruler, to leave a smooth upper
surface.
4. Position the centre of the mould and base under the plunger of the Vicat apparatus.
5. Lower the plunger until it is in contact with the paste. Pause in that position in order to avoid
initial velocity or forced acceleration of the moving parts.
6. Release the moving parts quickly and allow the plunger to penetrate vertically into the centre of
the paste. The release of the plunger shall occur 4 minutes ± 10 seconds after zero time.
7. Record the scale reading, the distance between the bottom face of the plunger and the base-
plate, 30 seconds after the release of the plunger.
8. Check whether the scale reading is (6 ± 2) mm.
9. Clean the plunger immediately after each penetration.
10. Repeat the test with pastes containing different water contents until one is found to produce a
distance between plunger and base-plate of (6 ± 2) mm.
11. Record the water content as a percentage by mass of the cement to the nearest 0.5% as the
water for standard consistence.
174
Setting Times
1. Remove the plunger from the Vicat apparatus and replace it with the needle in Figure 7.
2. Lightly oil the base of the container and Vicat mould with mineral-based oil and place the Vicat
mould into the container.
3. The needle should be calibrated by lowering it to rest on the base of the container. Manually
adjust the scale or pointer to read zero.
4. Raised the needle to the stand-by position.
5. Mix the cement paste with water and cement ratio in accordance to standard consistence as
mentioned above.
6. Fill the mould similarly as according to standard consistence test procedures.
7. Add distilled or drinking water to the container so that the surface of the paste is submerged to
a depth of at least 5 mm.
8. After 1 hour from zero time, position the centre of the mould and container under the needle of
the Vicat apparatus.
9. Lower the needle until it is in contact with the paste. Pause in that position in order to avoid
initial velocity or forced acceleration of the moving parts.
10. Release the moving parts quickly and allow the needle to penetrate vertically into the paste.
11. Record the scale reading, the distance between the end of the needle and the base of the
container, 30 seconds after the release of the needle.
12. Clean the Vicat needle and repeat the penetration for every 10 minutes with at least 10 mm
separations from the last penetrating position.
13. Report the time to the nearest 5 minutes at which the distance between the needle and the base
of the container is (6 ± 3) mm as the initial setting time of the cement.
Compressive Strength
175
6. After 30 seconds of mixing, add the sand steadily over the next 30 seconds.
7. Switch the mixer to high speed and continue mixing for an additional 30 seconds.
8. Stop the mixer for 90 seconds. During the first 30 seconds, remove by means of a plastic
scrapper or spoon the paste adhering to the wall and bottom part of the bowl and place in the
middle of the bowl.
9. Continue the mixing at high speed for 60 seconds.
1. Mould the specimen immediately after the preparation of the mortar using a suitable scoop.
2. Introduce in one or more increments, the first of two layers of mortar into each of the mould
compartments, directly from the mixing bowl.
3. Spread the layer uniformly and then compact using 60 jolts of the jolting apparatus.
4. Introduce the second layer of mortar, and compact the layer with a further 60 jolts.
5. Strike off the excess mortar with a metal straightedge, drawn forwards and backwards along
each mould compartment.
6. Place a steel plate or plastic sheet on the mould and place the covered mould in a horizontal
position in a moist chamber.
1. Demould specimens, for 24 hour tests, not more than 20 mins before the specimens are tested.
For tests at ages greater than 24 hours, carry out demoulding between 20 and 24 hours after
moulding.
2. Suitably mark specimens selected for curing in water.
3. Place specimens in water at (20 ± 1) oC in curing container.
4. Ensure that specimens are kept apart and that water has free access to all six faces of the
specimens.
5. Ensure that at no time during storage the depth of water above the upper faces of the specimens
be less than 5mm.
1. Carry out strength tests at the different ages ie. 3 and 7 days.
2. Remove specimens required for testing from the water not more than 15 minutes before the test
is carried out.
3. Remove any deposit on the test faces and cover specimens with a damp cloth until tested.
4. Centre the specimen laterally to the platens of the machine.
5. Increase the stress smoothly at the rate of 0.6 MPa/s until fracture occurs.
176
Appendix 6-2. Stress-strain curves for OPC cement and cement from cycled sorbent
a.
b.
25
20
20
16
Stress (MPa)
15
Stress (MPa)
12
Cube1
Cylinder1 10
8 Cube2
Cylinder2
5
Cube3
4 Average
Cylinder 0 AverageCube
0 0 0.0125 0.025 0.0375 0.05
0 0.01 Strain 0.02 0.03 Strain
Figure 6-10. Stress-strain behaviour for OPC 3-day a) cylinders; and b) cubes.
20
15 OPC
Stress (MPa)
1C NF
10 5C NF
15C NF
5 10C NF
5C F
0
0 0.005 0.01 0.015 0.02
Strain
177
References
Abanades, J. C. and D. Alvarez (2003). "Conversion limits in the reaction of CO2 with lime." Energy
& Fuels17(2): 308-315.
Abanades, J. C., G. Grasa, et al. (2007). Cost Structure of a Postcombustion CO2 Capture System
Using CaO. 41: 5523-5527.
Agnihotri, R., S. K. Mahuli, et al. (1999). "Influence of Surface Modifiers on the Structure of
Precipitated Calcium Carbonate." Industrial & Engineering Chemistry Research38(6): 2283-
2291.
Ahn, H., D. C. Ozcan, et al. (2011). Process Integration of Ca-Looping Process with Cement
Manufacturing Plant. 3rd IEAGHG HTSLC. Vienna, Austria.
Aihara, M., T. Nagai, et al. (2001). "Development of porous solid reactant for thermal-energy storage
and temperature upgrade using carbonation/decarbonation reaction." Applied Energy69(3):
225-238.
Al-jeboori, M. J., M. Nguyen et al. (2013). " Improvement of Limestone-Based CO2 Sorbents for Ca
Looping by HBr and Other Mineral Acids" Ind. Eng. Chem. Res. 52(4) 1426-1433.
Alonso, M., N. Rodríguez, et al. (2010). "Carbon dioxide capture from combustion flue gases with a
calcium oxide chemical loop. Experimental results and process development." International
Journal of Greenhouse Gas Control4(2): 167-173.
Alonso, M., N. Rodriguez, et al. (2009). "Modelling of a fluidized bed carbonator reactor to capture
CO2 from a combustion flue gas." Chemical Engineering Science64(5): 883-891.
Alsop, P. A., C. Hung, et al. (2007). Cement plant operations handbook Dorking, Surrey, UK.,
Tradeship Publications, ©2007.
Alvarez, D. and J. C. Abanades (2005). "Determination of the critical product layer thickness in the
reaction of CaO with CO2." Industrial & Engineering Chemistry Research44(15): 5608-5615.
Anthony, E. J. (2008). "Solid Looping Cycles: A New Technology for Coal Conversion." Industrial &
Engineering Chemistry Research47(6): 1747-1754.
178
ASTM (2006). ASTM C187 11e1: Standard Test Method for Amount of Water Required for Normal
Consistency of Hydraulic Cement Paste. West Conshohocken, PA., American Society for the
Testing of Materials.
ASTM (2008). C191-08: Standard Test Methods for Time of Setting of Hydraulic Cement by Vicat
Needle. West Conshohocken, PA., American Society for the Testing of Materials.
ASTM (2009). ASTM C150 / C150M - 12 Standard Specification for Portland Cement.
ASTM (2009). ASTM Standard C109/C109M - 11b: Standard Test Methods for Time of Setting of
Hydraulic Cement by Vicat Needle. West Conshohocken, PA., American Society for the
Testing of Materials.
ASTM (2009b). ASTM 266-08e1: Standard Test Method for Time of Setting of Hydraulic-Cement
Paste by Gillmore Needles. West Conshohocken, PA.
Babcock and Wilcox (1978). Steam: Its generation and use. New York, Babcock and Wilcox.
Barker, D. J., S. A. Turner, et al. (2009). "CO2 capture in the cement industry. ." GHTGT-9. Energy
Procedia 1: 87 - 94.
Barker, R. (1973). "Reversibility of Reaction CaCO3 = CaO + CO2." Journal of Applied Chemistry
and Biotechnology23(10): 733-742.
Bhatia, S. K. and D. D. Perlmutter (1983). "Effect of the Product Layer on the Kinetics of the CO2-
Lime Reaction." Aiche Journal29(1): 79-86.
Bhatty, J. I. (1995). Role of Minor Elements in Cement Manufacture and Use. Skokie, Illinois, U.S.A,
Portland Cement Association.
Bhatty, J. I. (2003). "Effect of Minor Elements on Clinker and Cement Manufacture: A Laboratory
Analysis." Portland Cement Association Internal Report, Illinois, USASerial no. 2147.
Blamey, J., E. J. Anthony, et al. (2010). "The calcium looping cycle for large-scale CO2 capture."
Progress in Energy and Combustion Science36(2): 260-279.
Borgwardt, R. H. (1989b). "Calcium oxide sintering in atmospheres containing water and carbon
dioxide." Industrial & Engineering Chemistry Research28(4): 493-500.
BSI (2005). BS-EN 196-1: Methods of testing cement - Part 1: Determination of Strength. 1. London,
British Standards Institute.
BSI (2011). Cement. Composition, specifications and conformity criteria for common cements.
179
Bye, G. C. (1983). Portland cement Composition, production and preparation. Oxford, Pergamon
Press Ltd:.
CEMBUREAU (1997). Alternative fuels in cement manufacture. . Brussels, The European Cement
Association.
CEMBUREAU (1999). "CEMBUREAU Best Available Techniques for the Cement Industry.".
Chen, Z., J. R. Grace, et al. (2008). "Limestone particle attrition and size distribution in a small
circulating fluidized bed." Fuel 87(7): 1360-1371.
Chen, Z., H. S. Song, et al. (2009). "Long-Term Calcination/Carbonation Cycling and Thermal
Pretreatment for CO2 Capture by Limestone and Dolomite." Energy & Fuels 23(3): 1437-
1444.
Chindaprasirt, P. and S. Rukzon (2008). ""Strength, porosity and corrosion resistance of ternary blend
Portland cement, rice husk ash and fly ash mortar."." Construction and Building Materials
22(8): 1601-1606.
Choi, G. S. and F. P. Glasser (1988). " The sulphur cycle in cement kilns: vapor pressures and solid-
phase stability of the sulfate phases. ." Cem. Concr. Res.18(367).
Corella, J., J. M. Toledo, et al. (2006). "Steam gasification of coal at low-medium (600–800 degrees
C) temperature with simultaneous CO2 capture in fluidized bed at atmospheric pressure: the
effect of inorganic species. 1. Literature review and comments." Industrial & Engineering
Chemistry Research45(18): 6137 - 6146.
Cuenca M.A. and A. E.J. (1995). Pressurized fluidized beds. , Blackie Academic & Professional.
David, J. and H. Herzog (2000). The cost of carbon capture. MIT Carbon Sequestration Forum,
Cambridge, MA.
Dean, C., Hills, T., Florin, N. and Fennell, P. (2012) "Integrating Calcium Looping CO2 Capture with
the Manufacture of Cement." GHGT-11. 18th - 22nd November 2012, Kyoto, Japan.
Dean, C. C., J. Blamey, et al. (2011). "The calcium looping cycle for CO2 capture from power
generation, cement manufacture and hydrogen production." Chemical Engineering Research
and DesignIn Press, Corrected Proof.
Dean, C. C, Dugwell, D. and Fennell, P. (2011). "Investigation into potential synergy between power
generation, cement manufacture and CO2 abatement using the calcium looping cycle." Energy
and Environmental Science 4(6): 2050-2053
Deutsch, Y. and L. Heller-Kallai (1991). "Decarbonation and recarbonation of calcites heated in CO2 :
Part 1. Effect of the thermal regime." Thermochimica Acta182(1): 77-89.
180
Dobner, S., L. Sterns, et al. (1977). "Cyclic Calcination and Recarbonation of Calcined Dolomite."
Industrial & Engineering Chemistry Process Design and Development 16(4): 479-486.
Dunstetter, F., M. N. de Noirfontaine, et al. (2006). "Polymorphism of tricalcium silicate, the major
compound of Portland cement clinker. 1. Structural data: review and unified analysis." Cem.
Concr. Res.36: 39.
ECRA (2007). Carbon capture technology—options and potentials for the cement industry.
EU (2001). Integrated Pollution Prevention and Control: Reference Document on Best Available
Techniques in the Cement and Lime Manufacturing Industries., Commission of the European
Union.
Ewing, J. A. Y., D. Beruto, et al. (1979). "The Nature of CaO Produced by Calcite Powder
Decomposition in Vacuum and in CO2." Journal of the American Ceramic Society 62(11-12):
580-584.
Fennell, P. S., J. F. Davidson, et al. (2005). "A study of the mixing of solids in gas-fluidized beds,
using ultra-fast MRI." Chemical Engineering Science 60(7): 2085-2088.
Fennell, P. S., J. F. Davidson, et al. (2007b). "Regeneration of sintered limestone sorbents for the
sequestration of CO2 from combustion and other systems." Journal of the Energy Institute
80(2): 116-119.
Fennell, P. S., R. Pacciani, et al. (2007a). The Effects of Repeated Cycles of Calcination and
Carbonation on a Variety of Different Limestones, as Measured in a Hot Fluidized Bed of
Sand. Energy and Fuels. 21: 2072-2081.
Florin, N. H. and A. T. Harris (2008a). "Enhanced hydrogen production from biomass with in situ
carbon dioxide capture using calcium oxide sorbents." Chemical Engineering Science 63(2):
287-316.
Florin, N. H. and A. T. Harris (2008b). "Screening CaO-Based Sorbents for CO2 Capture in Biomass
Gasifiers." Energy & Fuels22(4): 2734-2742.
Florin, N. H. and A. T. Harris (2008c). "Preparation and characterization of a tailored carbon dioxide
sorbent for enhanced hydrogen synthesis in biomass gasifiers." Industrial & Engineering
Chemistry Research47(7): 2191-2202.
181
Glasser, F. P., J. I. Bhatty, et al. (2004). Advances in cement clinkering. In: Portland Cement
Association. Innovations in Portland cement manufacture. . Skokie/Illinois.
Gonzalez, B., G. S. Grasa, et al. (2008). Modeling of the Deactivation of CaO in a Carbonate Loop at
High Temperatures of Calcination. 47: 9256-9262.
Grace J.R., Avidan A., et al. (1997). Circulating fluidized beds., Blackie Academic & Professional.
Grasa, G., J. C. Abanades, et al. (2009). "Effect of Partial Carbonation on the Cyclic CaO Carbonation
Reaction." Industrial & Engineering Chemistry Research48(20): 9090-9096.
Grasa, G. S. and J. C. Abanades (2006). CO2 Capture Capacity of CaO in Long Series of
Carbonation/Calcination Cycles. 45: 8846-8851.
Greco, R. and G. Enfil (2006). "Combustion of unusual fuels." World Cement37: 33.
Hayhurst, A. N. and R. F. Tucker (1991). "The reductive regeneration of sulphated limestone for flue-
gas desulphurization—thermodynamic considerations of converting calcium-sulphate to
calcium-oxide." Journal of the Institute of Energy 64(461): 212-229.
Herzog, H. (1998). The economics of CO2 capture. the 4th International Conference on Greenhouse
Gas Control Technologies, Interlaken, Switzerland.
Hughes, R. W., A. Macchi, et al. (2009). "Changes in Limestone Sorbent Morphology during CaO-
CaCO3 Looping at Pilot Scale." Chemical Engineering & Technology32(3): 425-434.
IEA (2006). Energy Technology Perspectives 2006: Scenarios & Strategies to 2050. Paris, IEA
(International Energy Agency).
Ives, M., R. C. Mundy, et al. (2008). "Comparison of Different Natural Sorbents for Removing CO2
from Combustion Gases, as Studied in a Bench-Scale Fluidized Bed." Energy & Fuels22(6):
3852-3857.
182
Jackson, J. P. (2007). Portland Cement: Classification and Manufacture. Hewlett, PC. Lea’s chemistry
of cement and concrete. Oxford, Burlington, Butterworth-Heinemann: 25.
Johansen, V. and T. V. Kouznetsova (1992). Clinker formation and new processes. 9th International
Congress on the Chemistry of Cement.
Joutsenoja, T., P. Heino, et al. (1999). "Pyrometric temperature and size measurements of burning
coal particles in a fluidized bed combustion reactor." Combustion and Flame118(4): 707-717.
Kääntee, U., R. Yevenhoven, et al. (2004). "Cement manufactoring using alternative fuels and the
advantages of process modeling." Fuel. Process Technol.85: 293.
Klauss, J. (2000). "Burning cement clinker under reducing conditions in a rotary kiln – one way of
lowering the NOx emission?" ZKG-international.53: 132.
Klug, H. P. and L. E. Alexander (1954). X-Ray Diffraction Procedures For Polycrystalline and
Amorphous Materials, Wiley.
Laursen, K., W. Duo, et al. (2000). "Sulfation and reactivation characteristics of nine limestones."
Fuel79(2): 153-163.
Li, F. and L.-S. Fan (2008). "Clean coal conversion processes - progress and challenges." Energy &
Environmental Science 1(2): 248-267.
Li, Y. J., C. S. Zhao, et al. (2009). "CO2 Capture Behavior of Shell during Calcination/Carbonation
Cycles." Chemical Engineering & Technology 32(8): 1176-1182.
Li, Z.-s., N.-s. Cai, et al. (2005). "Synthesis, Experimental Studies, and Analysis of a New Calcium-
Based Carbon Dioxide Absorbent." Energy & Fuels 19(4): 1447-1452.
Lisbona, P., A. Martinez, et al. (2010). "Integration of Carbonate CO(2) Capture Cycle and Coal-
Fired Power Plants. A Comparative Study for Different Sorbents." Energy & Fuels 24: 728-
736.
Liu, W., B. Feng, et al. (2010). "Synthesis of Sintering-Resistant Sorbents for CO2 Capture."
Environmental Science & Technology 44(8): 3093-3097.
Liu, W., N. W. L. Low, et al. (2009). "Calcium Precursors for the Production of CaO Sorbents for
Multicycle CO2 Capture." Environmental Science & Technology 44(2): 841-847.
183
Lu, D., R. Hughes, et al. (2009). "Sintering and Reactivity of -Based Sorbents for In Situ Capture in
Fluidized Beds under Realistic Calcination Conditions." Journal of Environmental
Engineering135(6): 404-410.
Lu, D. Y., R. W. Hughes, et al. (2008). "Ca-based sorbent looping combustion for CO2 capture in
pilot-scale dual fluidized beds." Fuel Processing Technology89(12): 1386-1395.
Lu, H., A. Khan, et al. (2008). "Relationship between Structural Properties and CO2 Capture
Performance of CaO−Based Sorbents Obtained from Different Organometallic Precursors."
Industrial & Engineering Chemistry Research47(16): 6216-6220.
Lu, H., E. P. Reddy, et al. (2006). "Calcium Oxide Based Sorbents for Capture of Carbon Dioxide at
High Temperatures." Industrial & Engineering Chemistry Research45(11): 3944-3949.
Lyngfelt, A. and B. Leckner (1989). "Sulphur capture in fluidized bed boilers: The effect of reductive
decomposition of CaSO4." The Chemical Engineering Journal40(2): 59-69.
Lysikov, A. I., A. N. Salanov, et al. (2007). "Change of CO2 carrying capacity of CaO in isothermal
recarbonation-decomposition cycles." Industrial & Engineering Chemistry Research46(13):
4633-4638.
MacKenzie, A., D. L. Granatstein, et al. (2007). "Economics of CO2 capture using the calcium cycle
with a pressurized fluidized bed combustor." Energy & Fuels 21(2): 920-926.
Mackes, K. H. and C. R. Lightburn (2003). "Evaluating the use of green wood chips processed from
small-diameter trees as an alternative fuel for making cement." Forest Product Journal 53: 42.
Macphee, D. E. and E. E. Lachowski (2007). Cement components and their phase relations. Hewlett
PC. Lea’s chemistry of cement and concrete. Oxford, Burlington, Butterworth-Heinemann:
95.
Maki, I. and K. Goto (1982). "Factors influencing the phase constitution of alite in Portland cement
clinker." Cem. Concr. Res.12: 301.
Manovic, V. and E. J. Anthony (2007). "Steam reactivation of spent CaO-based sorbent for multiple
CO2 capture cycles." Environmental Science & Technology 41(4): 1420-1425.
Manovic, V. and E. J. Anthony (2008a). "Thermal activation of CaO-based sorbent and self-
reactivation during CO2 capture looping cycles." Environmental Science & Technology
42(11): 4170-4174.
Manovic, V. and E. J. Anthony (2008c). Parametric Study on the CO2 Capture Capacity of CaO-
Based Sorbents in Looping Cycles. 22: 1851-1857.
Manovic, V., E. J. Anthony, et al. (2008b). "CO2 looping cycle performance of a high-purity
limestone after thermal activation/doping." Energy & Fuels 22(5): 3258-3264.
184
Manovic, V., D. Lu, et al. (2008). "Steam hydration of sorbents from a dual fluidized bed CO2
looping cycle reactor." Fuel 87(15-16): 3344-3352.
Mcbride, B. J., M. J. Zehe, et al. (2002). NASA Glenn Coefficients for Calculating Thermodynamic
Properties of Individual Species. N. A. a. S. Administration. Cleveland, OH, USA.
Mess, D., A. F. Sarofim, et al. (1999). "Product layer diffusion during the reaction of calcium oxide
with carbon dioxide." Energy & Fuels 13(5): 999-1005.
Metz, B., O. Davidson, et al. (2007). Mitigation, in Intergovernmental Panel on Climate Change,
Fourth Assessment Report, Climate Change 2007.
Metz, B., O. Davidson, et al. (2005). Special Report on Carbon Dioxide Capture and Storage,
Intergovernmental Panel on Climate Change. Cambridge, UK, Cambridge University Press.
Mohan, K. and F. P. Glasser (1977). "The thermal decomposition of C3SiO5 at temperatures below
1250 °C." Cem. Conr. Res.Part 1-3. : 1.
Mokrzycki, E. and A. Uliasz-Bochenczyk (2003). "Alternative fuels for the cement industry." Appl.
Energy 74: 95.
Neville, A. M. (1991). Concrete Technology. Essex, England, Longman Scientific and Technical:
p.95.
Oreskes, N. (2004). "Beyond the ivory tower - The scientific consensus on climate change."
Science306(5702): 1686-1686.
Pacciani, R., C. R. Müller, et al. (2008a). "Synthetic Ca-based solid sorbents suitable for capturing
CO2 in a fluidized bed." The Canadian Journal of Chemical Engineering 86(3): 356-366.
Pacciani, R., C. R. Müller, et al. (2008b). "How does the concentration of CO2 affect its uptake by a
synthetic Ca-based solid sorbent?" Aiche Journal 54(12): 3308-3311.
185
Pipilikaki, P., M. Katsioti, et al. (2005). "Use of tire derived fuel in clinker burning. ." Cem. Concr.
Res.27: 843.
Pizant, J. and J. C. Gauthier (1997). "Burning alternative fuels in rotary kiln." World Cement
September: 64.
Proscoandro, M., G. Mazziotti, et al. (2003). "Effect of burning supplementary waste fuels on the
pollutant emission by cement plants: a statistical analysis of process data." Resour. Conserv.
Recycl.39: 161.
Rodriguez, N. (2008). "Heat requirements in a calciner of CaCO3 integrated in a CO2 capture system
using CaO." Chemical Engineering Journal 138(1-3): 148-154.
Rodríguez, N., M. Alonso, et al. (2010). "Average activity of CaO particles in a calcium looping
system." Chemical Engineering Journal 156(2): 388-394.
Rodríguez, N., M. Alonso, et al. (2011). "Comparison of experimental results from three dual
fluidized bed test facilities capturing CO2 with CaO." Energy Procedia4(0): 393-401.
Rodriguez, N., M. Alonso, et al. (2008). "Process for capturing CO2 arising from the calcination of the
CaCO3 used in cement manufacture." Environmental Science & Technology42(18): 6980-
6984.
Romeo, L. M., J. C. Ballesteros, et al. (2008). "Oxyfuel carbonation/calcination cycle for low cost
CO2 capture in existing power plants."
Romeo, L. M., Y. Lara, et al. (2009b). "Optimizing make-up flow in a CO2 capture system using
CaO." Chemical Engineering Journal 147(2-3): 252-258.
Romeo, L. M., Y. Lara, et al. (2009a). "Economical assessment of competitive enhanced limestones
for CO2 capture cycles in power plants." Fuel Processing Technology 90(6): 803-811.
Ryu, H.-J., J. R. Grace, et al. (2006). "Simultaneous CO2/SO2 Capture Characteristics of Three
Limestones in a Fluidized-Bed Reactor." Energy & Fuels 20(4): 1621-1628.
Saastamoinen, J., T. Pikkarainen, et al. (2008). "Model of fragmentation of limestone particles during
thermal shock and calcination in fluidised beds." Powder Technology 187(3): 244-251.
Sakadjian, B. B., M. V. Iyer, et al. (2006). "Kinetics and Structural Characterization of Calcium-
Based Sorbents Calcined under Subatmospheric Conditions for the High-Temperature CO2
Capture Process." Industrial & Engineering Chemistry Research 46(1): 35-42.
Salvador, C., D. Lu, et al. (2003). "Enhancement of CaO for CO2 capture in an FBC environment."
Chemical Engineering Journal 96(1-3): 187-195.
Scala, F., R. Chirone, et al. (2003). "The influence of fine char particles burnout on bed agglomeration
during the fluidized bed combustion of a biomass fuel." Fuel Processing Technology 84(1–3):
229-241.
Scala, F., F. Montagnaro, et al. (2008). "Sulphation of limestones in a fluidized bed combustor: The
relationship between particle attrition and microstructure." The Canadian Journal of Chemical
Engineering 86(3): 347-355.
Schueppel, B. (2009). Attrition test for calcium-based sorbent materials. 1st IEA GHG Solid Looping
Cycles Network Meeting, INCAR-CSIC, Oviedo, Spain.
186
Shimizu, T., T. Hirama, et al. (1999). "A Twin Fluid-Bed Reactor for Removal of CO2 from
Combustion Processes." Chemical Engineering Research and Design 77(1): 62-68.
Silaban, A. and D. P. Harrison (1995). "High temperature capture of carbon dioxide: Characteristics
of the reversible reaction between CaO(s) and CO2(g)." Chemical Engineering
Communications 137: 177-190.
Silaban, A., M. Narcida, et al. (1996). Characteristics of the reversible reaction between CO2(g) and
calcined dolomite. Elmont, NY, ETATS-UNIS, Taylor & Francis.
Smith, I. M. (2007). Properties and Behaviour of SO2 Adsorbents for CFBC, IEA Clean Coal Centre,
London.
Stutzman, P. E. (1996). Guide for X-Ray Powder Diffraction Analysis of Portland Cement and
Clinker, National Institute of Standards and Technology.
Sun, P., J. R. Grace, et al. (2006). "Removal of CO2 by Calcium-Based Sorbents in the Presence of
SO2." Energy & Fuels21(1): 163-170.
Sun, P., J. R. Grace, et al. (2008). "Investigation of attempts to improve cyclic CO2 capture by sorbent
hydration and modification." Industrial & Engineering Chemistry Research 47(6): 2024-2032.
Trezza, M. A. and A. N. Scian (2000). "Burning waste as an industrial resource. Their effect on
Portland cement clinker." Cem. Concr. Res. 30: 137.
Trezza, M. A. and A. N. Scian (2005). "Waste fuels: their effect on Portland cement clinker." Cem.
Concr. Res. 35: 438.
Wang, J. and E. J. Anthony (2005). On the Decay Behavior of the CO2 Absorption Capacity of CaO-
Based Sorbents. 44: 627-629.
Wang, J., S. Lin, et al. (2009). "Limestone calcination with CO2 capture (III): characteristics of coal
combustion during limestone decomposition." Energy & Fuels23: 2804-2809.
Wang, Y., S. Lin, et al. (2008). "Limestone calcination with CO2 capture (II): decomposition in
CO2/steam and CO2/N2 atmospheres." Energy & Fuels22(4): 2326-2331.
187
Wen, C. Y. and Y. H. Yu (1966). A generalized method for predicting the minimum fluidization
velocity. 12: 610-612.
World_Cement_News. (2011). "Cemex UK’s increased alternative fuels results in substantial CO2
saving." Retrieved 07/04, 2012, from
http://www.worldcement.com/sectors/environmental/articles/Cemex_UK’s_increased_alterna
tive_fuels_results_in_substantial_CO2_saving.aspx.
Zabaniotou, A. and C. Theofilou (2008). "Green energy at cement kiln in Cyprus-Use of sewage sluge
as a conventional fuel substitute." Renewable Sustainable Energy Rev.12: 531.
Zeman, F. (2008). "Effect of steam hydration on performance of lime sorbent for CO2 capture."
International Journal of Greenhouse Gas Control 2(2): 203-209.
188