Repositorio Institucional de Documentos

2013 95 Irene Bolea Agüero Heat transfer in oxy-fuel fluidized bed boilers Departamento Director/es Instituto Universitario de Investigación Mixto CIRCE Romeo Giménez, Luis Miguel Director/es Tesis Doctoral Autor Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA Departamento Director/es Irene Bolea Agüero HEAT TRANSFER IN OXY-FUEL FLUIDIZED BED BOILERS Director/es Instituto Universitario de Investigación Mixto CIRCE Romeo Giménez, Luis Miguel Tesis Doctoral Autor 2013 Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA Departamento Director/es Director/es Tesis Doctoral Autor Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA HEAT TRANSFER IN OXY-FUEL FLUIDIZED BED BOILERS PhD Thesis Irene Bolea Agüero Supervised by: Dr. Luis Miguel Romeo Giménez  From the 19th century, the world has witnessed one of the most spectacular changes on the way of living along the History. The industrial revolution meant a drastic development for the economy, the quality of life, and the beginning of the society as we know it now, a globalized world. The steam generation was the breakthrough that let machines do double in half time than humans. And the coal has been the element accompanying men on such travel, followed later by the oil and the gas. But this story of success and progress had also hidden sides that we did not want to see until they were almost uncovered. Far from flattening, the curve of production and consumption has been never-ending exponential. From then on, manufacturing and transportation, the two pillars of the industrial era, are completely dependent on those non-infinite fuels. Globalization and progress meant not the same for every country and inequalities accentuate among them. And, in spite of this, the consequences of the progress of some will be paid by all. The climate change will be suffered by all the countries. But it is evident that those with infrastructure will cope better with the up-coming events. The dad Development must do his job and assumes his responsibilities. We, researchers are his tools, to carry out this hard task. Each of us must look for the piece of the jigsaw that is still not placed, to accelerate the change towards a fairer world and to lessen the irremediable repercussion of the acts of the teenager Progress that wanted to be Development. v La otra particularidad a la que se enfrentará la operación de los EHE en calderas de oxicombustión, residirá en la elevada cantidad de sólidos que habrán de recorrer este equipo y enfriarse hasta temperaturas adecuadas, para cumplir los requerimientos del balance de energía en la caldera. Existe escasa información en la literatura sobre las consecuencias de la distribución no uniforme de los coeficientes de transferencia de calor entre los tubos de un EHE. Aún así, todos coinciden en que el caudal de sólidos afecta dicha distribución, apareciendo zonas de transferencia de calor más intensa, y otras con fluidización más pobre, que no permiten tanto intercambio de calor con los tubos. En el Capítulo 4 se aborda esta cuestión desde la experimentación en un modelo frío de EHE y el modelado detallado de un EHE, para integrarlo en el modelo global de CFB desarrollado al comienzo. El modelo frío de EHE estaba originariamente conectado a un CFB bidimensional. Para poder alimentar manualmente caudales de sólidos en un rango amplio, el EHE se ha desacoplado del CFB para ser operado independientemente. De esta manera, se han podido manipular las dos variables: la velocidad de fluidización y la cantidad de sólidos que atravesaban el EHE. Los resultados de los experimentos mostraron que los coeficientes de calor locales se incrementaron en general, con respecto a aquellos obtenidos cuando no había sólidos alimentados. Sin embargo, mayores caudales de sólidos promueven también más desigualdades de los coeficientes locales entre unas zonas y otras. Para poder extrapolar los resultados de los coeficientes locales, el coeficiente medio de transferencia de calor se ha evaluado con respecto al tiempo que las partículas alimentadas necesitan para atravesar el equipo hasta la tubería de recirculación. De esta manera, se ha obtenido un factor de corrección que, aplicado al coeficiente de transferencia de calor sin aporte de sólidos, cuantifica el aumento de éste debido la circulación de sólidos. A partir de estos resultados, y los obtenidos en el Capítulo 3, se pueden obtener unos coeficientes de transferencia de calor que tengan en cuenta los parámetros particulares de las calderas de oxicombustión. Así, se ha desarrollado un modelo de EHE, en base a los detalles geométricos publicados por Man et al. (2012) de un EHE real. Integrando este módulo con el modelo anterior, se han obtenido las áreas de intercambio de calor que serían necesarias para las diferentes concentraciones de O2 planteadas al comienzo de la tesis. A pesar de que el incremento de superficie vi de intercambio era evidente, dada la cantidad de calor que habría de evacuar el EHE, este aumento queda moderado por el incremento de coeficiente de transferencia de calor dado por el mayor caudal de sólidos calientes que atraviesan el equipo. Con esta tesis se ha demostrado que las superficies de transferencia de calor en los lechos fluidos de oxicombustión habrán de adaptarse a las nuevas características de operación. La relevancia del EHE será particular de los grandes CFB. La predicción de los coeficientes de transferencia de calor en este equipo diferirá de los modelos utilizados en la combustión convencional. En consecuencia, para optimizar el rendimiento de las futuras calderas de oxicombustión en CFB, el diseño de la configuración del EHE tendrá que tener en cuenta los aspectos tratados en esta tesis. CONCLUSIONES Las grandes calderas de lecho fluido en oxicombustión van a requerir mayor caudal de sólidos recirculados, para poder moderar la temperatura de manera adecuada. Esto conllevará un diseño particularizado de los intercambiadores de calor externo (EHE). Las contribuciones de esta tesis se dirigen a demostrar teórica y experimentalmente los puntos más diferenciadores en los EHE de un lecho fluido en oxicombustión de uno en combustión convencional: - El modelo unidimensional estacionario de lecho fluido circulante (CFB) a gran escala desarrollado en esta tesis permite cuantificar la evacuación de calor necesaria en los EHE para las diferentes condiciones de concentración de O2 en la corriente oxidante y diferentes tamaños de caldera. La relevancia de este equipo apunta a dos diferencias fundamentales con los EHE asociados a la combustión convencional: la composición del gas de fluidización, y el aumento de sólidos recirculados - El estudio teórico original de los modelos de transferencia de calor en lechos fluidos burbujeantes, tanto desde el punto de vista empírico como mecanístico, ayudan a predecir las posibles influencias de los cambios de vii composición en las atmósferas de oxicombustión respecto a las de combustión con aire. - Los coeficientes de transferencia de calor en el lecho fluido burbujeante de 90 kW se miden y presentan por primera vez en esta tesis durante la oxicombustión con mezclas de O2/CO2 y con O2 diluido en gases de escape, para comparar los resultados con los medidos en combustión con aire. - El tratamiento de los datos experimentales desde los números adimensionales y desde el enfoque mecanístico, permite proponer una expresión completa que predice los valores de coeficientes de calor obtenidos experimentalmente durante la oxicombustión en lecho fluido burbujeante - La influencia de los sólidos recirculados en la transferencia de calor de un EHE se analiza experimentalmente en un modelo frío a escala, desacoplado del CFB original. Los resultados muestran un incremento de la transferencia de calor con el caudal de sólidos recirculados, a la vez que las los valores de los coeficientes de calor de unos tubos a otros también aumentan. - Por primera vez se propone una expresión que modifique el coeficiente de transferencia de calor sin recirculación de sólidos, y que tenga en cuenta el tiempo que tardan las partículas en abandonar el EHE - La simulación de une EHE, integrado en el modelo de CFB previamente desarrollado, incluirá el cálculo de los coeficientes de transferencia de calor tal y como se propone en esta tesis. Este modelo incluirá además el efecto de la re-carbonatación y sus posibles consecuencias en la de-fluidización del EHE. - Finalmente se ha calculado el área de transferencia de calor requerida en el EHE para diferentes condiciones de concentración de oxígeno a la entrada del CFB. Gracias a las expresiones desarrolladas aquí, que tienen en cuenta de manera más realista las condiciones propias de la oxicombustión, las superficies de transferencia de calor resultantes aumentan a medida que se incrementa la evacuación de calor necesaria en el EHE, pero de manera más moderada de lo esperado. viii ix RESUMEN I CONCLUSIONES VI NOMENCLATURE XIII LIST OF FIGURES XIX LIST OF TABLES XXIII CHAPTER 1 CONTEXT, JUSTIFICATION AND OBJECTIVES 1 1.1THE GLOBAL SCENE 1 1.1.1MAKING THE COAL SUSTAINABLE: THE CO2 CAPTURE AND STORAGE CHAIN (CCS) 4 1.1.2STATE OF THE ART OF OXY-FUEL COMBUSTION 9 1.1.3THE ROLE OF FLUIDIZED BED ON CCS 16 1.2JUSTIFICATION 17 1.3OBJECTIVES AND SCOPE 21 CHAPTER 2 LARGE OXY-FUEL CIRCULATING FLUIDIZED BED BOILER MODELING 25 2.1INTRODUCTION 25 2.1.1LARGE CFB MODELING 26 2.1.2OXY-FUEL CFB MODELING EXPERIENCES 30 2.2FUNDAMENTALS ON CFB MODELING 32 2.2.1FLUID-DYNAMICS 35 2.2.2COMBUSTION AND POLLUTANT FORMATION 39 2.2.3HEAT TRANSFER 47 2.3THE OXY-FUEL CFB MODEL 52 2.3.1FLUID-DYNAMICS SIMULATION 55 2.3.2COMBUSTION SIMULATION 58 2.3.3ENERGY BALANCE 64 x 2.4MODEL VALIDATION 68 2.4.1AIR FIRING 68 2.4.2OXY-FIRING 70 2.5THE ROLE OF THE EXTERNAL HEAT EXCHANGER 74 2.5.1REDUCTION OF OXY-FUEL BOILER SIZE 74 2.5.2VARYING POWER INPUT 75 2.5.3VARYING O2 AT INLET 77 2.6CONCLUSIONS 79 CHAPTER 3 HEAT TRANSFER IN CIRCE EXPERIMENTAL OXY-FUEL COMBUSTION BUBBLING FLUIDIZED BED 81 3.1INTRODUCTION 81 3.2HEAT TRANSFER IN FLUIDIZED BEDS 82 3.2.1PREVIOUS FINDINGS 82 3.2.2THEORETICAL INFLUENCES OF OXY-FUEL CONDITIONS 95 3.2.3SUMMARY 103 3.3EXPERIMENTAL SET-UP 104 3.3.1PREVIOUS EXPERIENCES ON OXY-FUEL FLUIDIZED BEDS PILOT PLANTS. 104 3.3.2CIRCE OXY-FUEL BUBBLING FLUIDIZED BED PILOT PLANT 106 3.3.3EXPERIMENTS PLANNING 109 3.4HEAT TRANSFER COEFFICIENTS MEASUREMENT 113 3.4.1MEASURING PROCESS 113 3.4.2HEAT TRANSFER RESULTS 118 3.4.3UNCERTANITY ANALYSIS 120 3.5DISCUSSION ON THE RESULTS OF THE HEAT TRANSFER IN OXYFUEL BFB 122 3.5.1EMPIRICAL APPROACH ANALYSIS 122 3.5.2CONTRIBUTION OF RADIATION HEAT TRANSFER 129 3.5.3MECHANISTIC APPROACH AND PROPOSAL OF MODIFICATION FOR OXY-FUEL CONDITIONS 130 3.6CONCLUSIONS 134 xi CHAPTER 4 THE INFLUCENCE OF SOLIDS RECIRCULATION ON HEAT TRANSFER IN AN EXTERNAL HEAT EXCHANGER 137 4.1INTRODUCTION 137 4.2HEAT TRANSFER IN EXTERNAL HEAT EXCHANGER FLUIDIZED BEDS 138 4.2.1HEAT TRANSFER TO TUBES BUNDLES 138 4.2.2INFLUENCE OF SOLIDS RECIRCULATION 141 4.2.3REAL EXTERNAL HEAT EXCHANGERS 145 4.3EXPERIMENTAL SET-UP 148 4.4RESULTS OF HEAT TRANSFER IN THE COLD EHE 151 4.4.1COOLING RATES 151 4.4.2UNEVEN DISTRIBUTION OF HEAT TRANSFER COEFFICIENTS 155 4.4.3AVERAGE HEAT TRANSFER COEFFICIENT 158 4.5INTEGRATION OF THE EHE MODEL IN THE OXY-FUEL CFB MODEL 161 4.5.1FLY-ASH RE-CARBONATION 161 4.5.2EXTERNAL HEAT EXCHANGER SIMULATION 167 4.5.3INTEGRATION RESULTS 169 4.6CONCLUSIONS 173 CHAPTER 5 SYNTHESIS, CONTRIBUTIONS AND RECOMMENDATIONS 175 5.1SYNTHESIS 175 5.2CONTRIBUTIONS 179 5.3RECOMMENDATIONS AND FURTHER WORK 182 5.4PUBLICATIONS 183 REFERENCES 187 xii xiii NOMENCLATURE A area m2 A0 gas-distributor area per nozzle m2 BFB Bubbling Fluidized Bed C molar concentration of a gas compound kmol/ m3 CFB Circulating Fluidized Bed CCS Carbon Capture and Storage cp specific heat capacity J/kgK d diameter m D bed diameter m Dg diffusivity of oxygen in nitrogen m2/s Dsh horizontal dispression coefficient m2/s Dsv horizontal dispression coefficient m2/s E enthalpy J/kgK EHE External Heat Exchanger f fraction of time of a phase conctacitng a suface g gravity constant m/s2 Gs elutriated solids mass flux kg/m2s H bed height m h heat transfer coefficient W/m2K HT Heat Transfer k conductivity W/mK xiv kbm backmixing parameter Kc apparent kinetic constant for surface reaction m/s kD calcination kinetic constant m/s l characteristic length scale m L length m LHV Low Heating Value kJ/kg LS loop seal m mass flow rate kg/s M dimensionless parameter for the film thermal resistance estimation n number of p proportion of particle types - P pressure bar Pp partial pressure bar Q heat kW r radius m/s R thermal resistance m2K/W Rc calcination rate RFG Recycled Flue Gas s path length m S2 typical deviation T temperature ºC T average temperature ºC t time s tce contact time of clusters in the emulsion xxi Figure 3.19. Influence of non-dimensional parameters on Nusselt number for similar fluidizing velocities .......................................................................................................... 126 Figure 3.20. Influence of non-dimensional parameters on Nusselt number for similar bed temperature ...................................................................................................................... 127 Figure 3.21. Comparison of tests results with prediction by Molerus correlation .............. 128 Figure 3.22. Estimation of the variation of radiant heat transfer coefficient with bed temperature ...................................................................................................................... 130 Figure 3.23. Maximum and average deviation of predicted values for different values of M ........................................................................................................................................... 132 Figure 3.24. Validation of calculated heat transfer coefficients ........................................... 133 Figure 4.1. Void fraction around an horizontal tube (Umekawa et al., 1999) ..................... 139 Figure 4.2. Dispersion coefficients calculated for different fluidization velocities and different tubes arrangements, according to Eq. 4.6 and Eq. 4.7. ................................ 141 Figure 4.3. Results from the heat transfer measurement in a cold model of an External Heat Exchanger. Adapted from Wang et al. (2003) ....................................................... 143 Figure 4.4. Heat transfer rates at different locations of a loop-seal heat exchanger. Adapted from Johansson et al. (2006) ........................................................................................... 144 Figure 4.5. Heat transfer coefficients measured in different positions in a real HHE. Data taken from Wedermann and Werther (1993) ................................................................ 146 Figure 4.6. Heat transfer coefficients measured in one of the chambers of the external heat exchanger. Data taken from Man et al. (2012). ............................................................. 147 Figure 4.7. Experimental set-up. Filled circles indicate the thermocouples location ......... 149 Figure 4.8. Photo of the loop seal, the instrumented tubes bundle, and the air supply ..... 149 Figure 4.9. Example of cooling curve for a testing tube. Zoom to the first seconds of cooling to calculate the cooling rate. ............................................................................................ 150 Figure 4.10. Cooling rates at the first row, near the distributor, low velocities .................. 152 Figure 4.11. Comparison between low and higher tubes with two solid rates .................... 153 Figure 4.12. Comparison between low and higher tubes with two aeration velocities ....... 153 Figure 4.13. Comparison of four tests at different row heights ............................................ 154 Figure 4.14. Heat transfer coefficients, a) high velocity and increasing solids rate; b) low velocity and different solids rate ..................................................................................... 156 Figure 4.15. Heat transfer coefficients with two velocities and solids rate conditions ....... 157 Figure 4.16. Influence of a) solids flow, and b) velocities ratio, on the average heat transfer coefficient .......................................................................................................................... 158 Figure 4.17. Residence time influence on the average heat transfer coefficients ............... 159 Figure 4.18. Influence of residence time of recycled particles on heat transfer coefficients in the cold model and in the real EHE ................................................................................ 160 Figure 4.19. Carbonation-calcination equilibrium ................................................................ 162 xxii Figure 4.20. Partial pressure and equilibrium pressure of CO2 along the riser, 40% O2 at inlet ................................................................................................................................... 163 Figure 4.21. Partial pressure and equilibrium pressure of CO2 along the riser, 60% O2 at inlet ................................................................................................................................... 163 Figure 4.22. Modeling flow chart of the integration of an External Heat Exchanger into the global CFB model ............................................................................................................. 168 Figure 4.23. Results of recycled solids flow and heat transfer coefficients in the EHE ...... 170 Figure 4.24. Results of heat transfer coefficients with and without recycled solids, and the calculated heat transfer surface required for the proper heat exchange ..................... 171 Figure 4.25. Heat transfer coefficients estimated by different authors ............................... 172 xxiii LIST OF TABLES Table 1.1. Pilot and demonstration plants of oxy-fuel combustion (Spero and Montagner, 2007; Total, 2007; McCauley et al., 2008; Barbucci, 2009; Burchhardt, 2009; Ochs et al., 2009a; Scheffknecht, 2009; Sturgeon, 2009; Wall et al., 2009; Alvarez et al., 2011; Spero et al., 2011; Schoenfield and Menendez, 2012) ...................................................... 13 Table 1.2.. Research groups with experimental experiences on oxy-fuel fluidized bed pilot plant .................................................................................................................................... 20 Table 2.1. Models of fluidized bed combustors ......................................................................... 33 Table 2.2. Nomenclature of Table 2.1 ....................................................................................... 34 Table 2.3. Correlations for estimating bubble velocity ............................................................ 37 Table 2.4. Coal composition used in the simulations .............................................................. 59 Table 3.1. Empirical expressions for the prediction of heat transfer coefficient from fluidized bed to vertical walls ........................................................................................................... 85 Table 3.2. Effective thermal conductivity of particulate phase by different authors ............ 90 Table 3.3. Composition of fuels used in the tests ................................................................... 111 Table 3.4. Experiments planning matrix, planned and achieved ranges ............................. 111 Table 3.5. Experimental operational ranges and devices ..................................................... 112 Table 3.6. Summary of AF tests stable zones for heat transfer measurements .................. 116 Table 3.7. Summary of OF tests stable zones for heat transfer measurements .................. 117 Table 3.8. Summary of OF+RFG tests stable zones for heat transfer measurements ........ 118 Table 3.9. Instantaneous and averaging relative uncertainty for heat transfer determination ................................................................................................................... 121 Table 4.1. Experiments planning matrix for the operation of the cold EHE ....................... 151 Table 4.2. Recarbonation inputs for the simulations ............................................................ 164 Table 4.3. Recarbonation results from the simulations ....................................................... 165 1 CHAPTER 1 CONTEXT, JUSTIFICATION AND OBJECTIVES 1.1 THE GLOBAL SCENE In 1992, the United Nations Framework Convention on Climate Change (UNFCCC) set an intergovernmental framework aimed to be a starting point to cope with the global climate change. The ultimate objective of the Convention was “to stabilize greenhouse gas concentrations at a level that will prevent dangerous human interference with the climate system” (UNFCC, 1992). With 194 Parties, the UNFCCC has near universal membership. Under the Convention, membership governments committed to: i) gather and share information on greenhouse gas emissions, national policies and best practices; ii) launch national strategies for addressing greenhouse gas emissions and adapting to expected impacts, including the provision of financial and technological support to developing countries; and iii) cooperate in preparing for adaptation to the impacts of climate change. This Convention represented the universal acceptation that human activities are the major cause of the increasing greenhouse gas emissions in the atmosphere and the agreement of reinforcing the countries to cope with climate change in cooperation with the international community. The Parties agreed to take precautionary measures to anticipate, prevent or minimize the causes of climate change and mitigate its adverse effects. The International Panel of Climate Change (IPCC) has been making an outstanding task of preparing a comprehensive review and recommendations with respect to the state of knowledge of the science of climate change; social and economic impact of climate change, possible response strategies and elements for inclusion in a possible future international convention on climate (IPCC, 2010). In spite of the international community commitment, facts of greenhouse gas (GHG) 1.1. The global scene 2 emissions to the atmosphere in the last two decades are not as comforting as they should. According to the Fourth Assessment report by IPCC (2007): ”… the atmospheric concentrations of long-lived GHGs CO2 and CH4 in 2005 exceed by far the natural range over the last 650,000 years. Global increases in CO2 concentrations are due primarily to fossil fuel use, with land-use change providing another significant but smaller contribution.…”. IPCC concluded that reductions of at least 50% in global CO2 emissions compared to 2000 levels will need to be achieved by 2050 to limit the long-term global average temperature rise to 2-2.4ºC (IEA, 2010a). Energy production is the human activity that represents the major source of CO2 emissions in the developed countries, because CO2 is an intrinsic waste from the combustion of fossil fuels for power generation. Figure 1.1. Shares of anthropogenic greenhouse-gas emissions in Annex I countries, 2009, adapted from IEA (2011a). The economic growth, together with the increasing world population are narrowly related to the emissions of GHGs to the atmosphere, due to the dependence on the fossil fuel combustion as primary energy all over the world, in the so-called carbonbased economy. In spite of the smoother increment of the global GDP, the international indicators remain unaltered. The small variation of advanced economies does not alter the global tendency. The rate of growth of OCDE countries is estimated on 2% in the following 30 years, and the non-OCDE will grow more than 4.5% (ExxonMobil, 2012). Estimations of the world energy demand ranged from 30-40% in the following decades (BP, 2011a; ExxonMobil, 2012), while this growth will achieve 63% in a “usual as business” scenario (EIA, 2011). CO2;92% CH 4 ;7% N 2 O;1% Waste 3% Agriculture 8% Energy 83% Industry 6% Chapter 1. Context, Justification and Objectives 3 Fossil fuels accounts for more than 60% of the electricity generation. Coal has been a key component of the electricity generation mix worldwide. It fuels more than 40% of the world’s electricity, although this share is higher in some areas. In countries like South Africa, 93% of power comes from coal, in Poland, 92%, in China, 79%, in India, 69% and in the United States, 49% (IEA, 2010b). In Europe, the growth of coal demand is being progressively replaced by the gas, but still accounts for almost 40 % of the power generation share. The reports coincide to recognize that energy efficiency continues to improve globally, and at an accelerating rate. The IEA World Energy Outlook (2011b) remarks the need to achieve an even higher pace of change, with efficiency improvements accounting for half of the additional reduction in emissions. In 2009, 43% of CO2 emissions from fuel combustion were produced from coal, 37% from oil and 20% from gas, Figure 1.2. Growth of these fuels in 2009 was quite different, reflecting varying trends that are expected to continue in the future. Global CO2 emissions actually decreased by 0.5 Gt CO2, between 2008 and 2009, representing a decline of 1.5%. However, trends varied greatly: the emissions of developed countries decreased, whereas the emissions of developing countries increased up to 54%. Figure 1.2. World CO 2 emissions by fuel. Adapted from IEA (2011a) 1.1. The global scene 4 Urgent actions are needed for changing the directions of the CO2 emissions trends, aiming for the long-term target of limiting the global average temperature increase to 2°C. Some scenarios are pessimistic about the effectiveness of policies to act quickly enough. And all of them coincide on pointing out the reduction of CO2 emissions from fossil fuel sources as a priority for a real global reduction. 1.1.1 MAKING THE COAL SUSTAINABLE: THE CO2 CAPTURE AND STORAGE CHAIN (CCS) Electricity and heat generation accounts for more than 60% of the CO2 emissions derived from coal combustion (Figure 1.3). This represents most of the stationary power plants. From a technological point of view, this should be a simplification when applying measures for emissions reduction from the source itself. This CO2 reduction is based on increasing the power plants efficiency and on deploying CO2 capture and storage technologies. According to the IEA World Energy Outlook (2011b) widespread deployment of more efficient coal-fired power plants and carbon capture and storage (CCS) technology could promote the long-term prospects for coal. Therefore, it is necessary the development of clean fossil fuels power plants. The development of zero and near zero emissions power plant technologies is gaining importance worldwide and large demonstration projects are expected in the coming decade for new plants (IPCC, 2005). But if drastic reductions are requested in the medium term, it is also necessary to support and deploy technologies that could be able to capture part of CO2 from existing power plants. Figure 1.3. CO2 emissions by fuel and sector. Data taken from IEA (2011a) Coal Oil NG 0 5000 10000 15000 20000 25000 30000 35000 Allsectors CO 2 emissionsin2009 (millionstonnes) Elect & Heat Manuf. Coal Chapter 1. Context, Justification and Objectives 5 There are three main approaches to capture CO2 from fossil-fired energy systems: pre-combustion, post-combustion, oxy-fuel. These technologies have application in large, stationary carbon emission point sources. However, capture is only one link in the CCS chain. A CCS system also requires CO2 compression, a means to transport it and a storage site, as illustrated in Figure 1.4. This means a great complexity and effort required to successfully remove CO2 from the atmosphere that goes beyond the carbon capture step. Figure 1.4. CCS chain scheme Looking at Figure 1.4, it is clear that much of the barriers found by any CCS system to be economically feasible will relay on the way that energy requirements are minimized and integration of the overall power system with CCS chain components. Following, the technologies considered for capturing CO2 from fossil fuel combustion in the near and medium term will be explained. Precombustioncapture In pre-combustion capture, carbon dioxide is separated from a gaseous fuel mixture under reducing conditions prior to its combustion. The majority of commercially available pre-combustion CO2 separation technologies rely on the use of liquid solvents for absorption of CO2 from a gas stream (Kohl and Nielsen, 1997). CO2 is usually separated by scrubbing with a physical solvent. A simplified diagram is depicted in Figure 1.5. It involves the conversion of a carbonaceous fuel, such as natural gas, coal, biomass or oil to a gaseous mixture that primarily consists of H2 and CO, called syngas. The fuel conversion step during gasification or reforming is endothermic and requires supplementary heating, typically supplied by partial Power plant CO 2 capture system CO 2 Compression CO 2 Transport Fluegas withno CO 2 Geological Storage Heatandpower Heatand power Fuel 1.1. The global scene 6 oxidation of the fuel. In CO2 capture applications, high-purity O2 obtained from an Air Separation Unit (ASU) is often used as oxidant, yielding the high temperatures required to produce H2 and CO-rich syngas. Figure 1.5. Pre-combustion capture scheme The H2 and CO in the syngas is converted to H2O and CO2 via the Water-Gas Shift (WGS) reaction. The WGS reaction is facilitated by commercially available catalysts, that are suitable for application under sweet and sour conditions, i.e. low and high sulphur (H2S and COS) concentrations respectively. The CO2 is then separated through physical absorption with a scrubbing solvent. The CO2-rich solvent is later flashed by pressure reduction which releases the CO2, while the regenerated CO2-lean solvent is reused for CO2 absorption. The recovered CO2 is subsequently dried and compressed to facilitate its transport and storage, while the hydrogen-rich stream can be combusted in a combined cycle for generation of electricity or used for synthesis of chemicals. This technology is very promising in terms of energy efficiency penalty, but it is the one in the earliest stage of development, and the capital cost is still elevated. A IGCC with capture would increase the construction costs 30% with respect to the case with no capture, whereas operation and maintenance costs would be 20% higher (Kyungtae et al., 2011). The first IGCC for CO2 capture was run in the power plant in Puertollano, Spain. The pilot separates an annual equivalent of up to 36.5 ktonne CO2 using the MDEA solvent, and 700 tonne H2 employing a pressure swing adsorption unit (Romeo et al., 2010b). Gasifier Shift CO 2 capture system Compression, transportand storage Heatandpower Power Fuel Air Gasturbine H 2 Syngas CO 2 ASU Steam turbine N 2 Stean Chapter 1. Context, Justification and Objectives 13 Project/ Institution Country Size (MWth) Status /Schedule Boiler Type Fuel CO2 capture dtails Babcock & Wilcox USA 30 Demonstration tests during 2007-2008 Pilot PC Bit. SubBit, Lig No Capture Jupiter USA 15 Start-up 2011 PC retrofit. No FGR NG, , High sulphur coal 25% flue gases to be treated. No storage Doosan Babcock UK 40 Start-up 2009 Pilot PC Vattenfall Germany 30 Start-up 2008 Pilot PC Lig. Bit. With CCS Total, Lacq France 30 Start-up 2009 Industrial. Steam for utilities NG and Liq. Fuels With CCS in gas depleted reservoir Enel Italy 48 Start-up planned 2012 Presurized PC Jupiter (Pearl Plant) USA 66 Started-up 2009 22 MWe PC Bit Side stream Callide Australia 90 Air tests in March 2009 O2/CO2 plant contract: August 2009 30 MWe Retrofit PC Bit 75 tpd to CCS. Road transport to a depleted gas field Ciuden-PC Spain 20 Start-up in 2010 Pilot PC Antr. Pet cok CiudenCFB Spain 30 Pilot CFB Antr. Pet cok Babcock & Wilcox USA 400 100 MWe PC SubBit With CCS Jamestown USA 150 Announced for 2015 CFB 78 MWe gross /44 MWe with CCS Bit, biomass With CCS to Michigan Basin Endesa Spain 1500 Announced for 2015 With CCS Vattenfall Germany 1000 Announced for 2015 250 MWe PC Lig. Bit. With CCS KEPCO (Youngdong) Korea 400 Announced for 2016 100 MWe Repowering SubBit, Bit With CCS Table 1.1. Pilot and demonstration plants of oxy-fuel combustion (Spero and Montagner, 2007; Total, 2007; McCauley et al., 2008; Barbucci, 2009; Burchhardt, 2009; Ochs et al., 2009a; Scheffknecht, 2009; Sturgeon, 2009; Wall et al., 2009; Alvarez et al., 2011; Spero et al., 2011; Schoenfield and Menendez, 2012) 1.1. The global scene 14 Barriersandfutureprospects LargeScaleOxygenproduction One of the great drawbacks of oxy-fuel combustion comparing with other CO2 capture technologies is the energy cost of oxygen production. Current technology, based on cryogenic air separation, for producing 95% purity O2, requires around 200 kWh/tonO2 (Sarofim, 2007). In a steam power plant this means energy losses of around 7%-10% of the LHV efficiency. Possibilities of achieving higher O2 production efficiency with the current cryogenic ASU are limited. An added drawback of this technology is the residual N2 that remains in the oxygen stream. It can reach up to 5% volume diminishing CO2 stream purity. However, different concepts are also investigated: Ion Transport Membrane (ITM), developed by Air Products, is a promising technology. It operates at temperatures around 800-900ºC, at which the crystalline structure incorporates oxygen ion vacancies (White et al., 2009). Oxygen Transport Membrane (OTM), patented by Praxair, is also a new technology based on droped Zirconia, being able to consume 75% less power than cryogenic ASU (Wilsoon et al., 2009). Membranes show the advantage of avoiding presence of N2 from air, and only bound N2 would be present in the gas stream. Ceramics Autothermal Recovery (CAR) is under development by Linde and BOC (Santos, 2009). CO2cleaningfordisposal Some research is needed to know the effect of recycling flue gas stream from the different points of the flue gas circuit: recirculation fans consumption, flue gas cleaning equipments sizing, or increase in accumulated pollutants in the boiler are some of the issues under study (Kather, 2007). Flue gas desulphurization tests were already carried out in Schwarze Pumpe, obtaining similar results as in airfiring cases (Yan et al., 2009) On the other hand, uncertainties in the purity requirements for CO2 geological disposal are still an issue, although some guidelines were determined from ENCAP project (Sarofim, 2007). In this respect, CO2 stream from oxy-fuel combustion has more non-condensable species concentration than from other capture technologies (Seevam et al., 2008). Properties of gasses blends are not known around the critical point and changes in the phase diagram due to the presence of non-condensable gases in the CO2 stream are difficult to determine. O2 content in the flue gas Chapter 1. Context, Justification and Objectives 15 stream is around 3%. This represents a great drawback when the storage site is a depleted hydrocarbon site, because high probability of combustion reactions. In this case, O2 should be removed before storage. Moreover, undesired air ingress into the system is the major source of impurities in oxy-firing process. These leakages could be reduced by pressurizing gas feeding into the boiler. Especial care must be taken with NOx, SO2 and O2 content, which represent around 0.25%, 2.5% and 3% vol. respectively. NO and SO2, can react in the compressors in the presence of water and oxygen, forming H2SO4 and HNO3. New concepts are also proposed, for integrating captured CO2 with compression phase (Romeo et al., 2009a). For diminishing compression costs, pressurized combustion systems would lead to a high pressure flue gas stream, for example, in the case of a pressurized fluidized bed. In the case of oxy-fuel, this would also help to avoid air leakages into the system, reaching a higher purity CO2 stream. Burnersandboilersdesign Adjustments in the burner designs need to accomplish several issues: higher oxygen feed concentration, different primary air for fuel conveying, different aerodynamics due to denser CO2 gas around the burner and changes in flame staging because of NOx feeding with RFG. With the help of computational fluiddynamics modeling, new burner designs have been proposed by several researchers from academia and industry (Becher, 2008; Tchunko et al., 2008; Lee et al., 2009; Scheffknecht et al., 2009; Seltzer et al., 2009) Regarding the boiler design, heat and energy balances are changing in the case of oxy-fuel combustion. As explained above, heat transfer is modified by the flue gases properties and flows. This leads to a re-sizing of the heat transfer exchange areas, reducing the radiant zone and increasing the area of the convective pass (Jordal et al., 2004; Jäntti et al., 2007; Ochs et al., 2009b). High ash fuels could also cause slugging and fouling problems in the radiant area (Scheffknecht, 2009). Fluidized bed combustors are a highlighted option for implementing oxy-fuel combustion in the medium term time framework. In general, fluidized bed technology presents interesting characteristics when compared to pulverized fuel boilers, mainly related to the pollutant emissions control. This has allowed to improve the combustion performance of difficult, such as low-rank fuels, or fuels 1.1. The global scene 16 that are difficult to pulverize. Additionally to these known advantages, fluidized bed present a remarkable feature that makes it suitable for burning under oxy-fuel conditions. Thanks to the high bulk density of inert particles and their movement, temperature is easier to control and uniform. For high O2 concentration in fluidizing streams higher fuel is fed for the same boiler geometry. More compact boilers lead to less boiler wall surface to exchange heat. Along next section, fluidized bed combustion fluidized beds will be explored. 1.1.3 THE ROLE OF FLUIDIZED BED ON CCS The particular characteristics of fluidized bed (FB) technology are caused by the fluid-like behavior of certain types of solids particles when a liquid or gas suspends them from below. Such a basic phenomenon provides an intense contact and interaction between the fluidization substance and particles and among the particles themselves. This allows FB to be susceptible to be applied to numerous industrial processes. Energy conversion in all its forms is one of the major applications for fluidized beds. Initially, fluidized beds were conceived by Fritz Winkler in 1912 for coke gasification. It was not until the 60’s when Douglas Elliot recognized the possibility of burning coal in fluidized beds to generate steam (Basu, 2006). The use of fluidized bed boilers has grown drastically in the last years. One of the main reasons is that environmental regulations have become stricter regarding pollutant emissions from combustion sources. Fluidized beds have the capability of capturing SO2, thanks to the addition of calcium-based sorbents directly to the furnace. One of the main problems that fluidized bed combustion must cope with is the risk of inert matter agglomeration. This phenomenon is due to the melting of solids particles due to peaks of temperature. Melted particles become sticky and form agglomerates extremely quickly. This drives to a bed defluidization and consequent un-desirable operation turn off. For avoiding this problem, bed temperature must be always kept below ashes melting temperature. This is actually not a real handicap, having into account that optimum temperature for SO2 capture under conventional air-firing is around 850ºC. Additionally, this temperature is sufficiently low for avoiding thermal mechanisms of NOx formation, lowering the emissions of this compound as well. Chapter 1. Context, Justification and Objectives 17 For these reasons, fluidized bed is a very promising candidate to apply oxy-fuel combustion as a CCS technology. The support given by the European Commission to the deployment of oxy-fuel combustion to large circulating fluidized bed boilers, has been translated into two highly relevant projects within the Seventh Framework Program. First, the Flexi-Burn project, funded with 6.4 million Euros, started in 2009 and ended in 2012. Its aim was fully utilizing all the new CFB design and advancements for merging a CFB boiler with supercritical one through (OT) steam cycle and air separation unit, together with the CO2 capture unit. The air/oxy flexible concept aim to have the capability of substituting 20% of the coal input with biomass fuels. The demonstration phase in this project took place, in parallel, in the CIUDEN 30 MWth CFB plant, for the oxy-fuel tests, and in Lagisza plant, the 460 MWe OT CFB boiler. The second project, called O2GEN, within EU Seventh Framework Program was approved in the beginning of 2012, with a funding of 6 million euros. It will be coordinated by CIRCE. The focus is now to optimize and integrate the main components of the oxy-fuel CO2 capture system, the ASU, the CFB boiler and the compression unit, aiming to reduce the overall energy penalty down to half of the current state of the art. The demonstration stage will be mainly developed in the 30 MWth CFB unit in CIUDEN and the results will be deployed in the OXY-CFB300 Compostilla Project (Table 1.1). The Compostilla Project was selected by the European Energy Program for Recovery (EPR) to receive funding of 180 milion Euros to develop the technology. It will be the first demonstration scale oxy-fuel power plant, based on supercritical CFB boiler, involving the whole CCS chain. 1.2 JUSTIFICATION From a global point of view, the research community has an urgent task to develop instruments for avoiding CO2 emissions to the atmosphere as soon as possible, as has been internationally recognized. The consequences of the climate change are imminent, and the necessity of mitigate CO2 emissions from anthropogenic origin is an issue of absolute priority. One of the most relevant sources of CO2 is the large fossil fuel combustor used in the energetic and the steel and concrete sectors. It is clear that avoiding emissions from large stationary sources, such as said boilers and combustors, the essential contribution CO2 capture and storage 1.2. Justification 18 technologies, has been recognized by the international authorities, as revealed by the Working group III report for the UNFCC on CO2 mitigation (Sims et al., 2007). Still, it must be beard in mind, that CCS must be a transition technology towards the decarbonization of energy sources (IEA, 2011c). The CCS chain technology acquires especial relevance for the appliance to delocalize, low-cost and carbon intensive fuels, such as coal. Indeed, coal is the fuel with the largest R/P ratio (reserves-to-production) and the world proved reserves in 2010 were sufficient to meet 118 years of global production. Moreover, in 2010 the world coal production grew around 6% over the historical average, and Asia Pacific region accounted for 89% of global growth. Solely China is planning to build 110 GWe of supercritical power plants by 2015 (BP, 2011b). Additionally given the successful commissioning of the first supercritical CFBC at Lagisza (Poland), both China and Russia are initiating programs for constructing supercritical CFBC units (Burnard and Bhattacharya, 2011). It is then clear that fluidized bed technology is gaining importance because its unique features for burning broad range of local solid fuels. Particularly relevant is its capacity of avoiding SO2 emissions by the addition of limestone based sorbents inside the boiler, contributing this way to the use of low-rank local lignite, indigenous of many regions around the world. The possibility of keeping temperatures, also below the point of NOx thermal generation, drives fluidized bed combustion process to the concept of clean coal technology approach. It appears then as an outstanding opportunity for applying oxy-fuel combustion into this type of technology. This will be the first step towards developing a realistic concept of zero-emissions coal power plants. From this perspective, oxy-fuel stands as one of the three feasible technologies to be implemented in the near term and, in fact, already at the demonstrations stage. Although the ratio of CFB combustors over the PC combustors for conventional airfiring was only around one fourth in North America and Europe was in 2007 (Koornneef et al., 2007), the oxy-fuel pilot and demonstration roadmap shows that both technologies are considered equally suitable. In fact, the two options are already working at a demonstration scale, 30 MW. Oxy-fuel PC was commissioned in 2008 in Schwarze Pumpe (Germany) and oxy-fuel CFB began operation in CIUDEN (Spain) in 2011 (Alvarez et al., 2011). Chapter 1. Context, Justification and Objectives 19 With this global picture in mind, research must be focused on overcoming the barriers to make oxy-fuel (OF) technology a reliable option for the new power plants generation. Expected differences with the well known air-firing (AF) technology have been early recognized. The main research groups that have carried out experiments on oxy-fuel pilots plants are summarized in Table 1.2. As seen in Table 1.2, one of the main concerns of scientific community is coping with the pollutant emissions inherent in solid fuels combustion, particularly coal. This is the reason why most of the studies focus on looking deeper inside the SO2 capture and NOx destruction during OF mode. Many of the groups, in fact, had experiences on the research on these topics and retrofitted existing plants to adapt them to OF, like in CANMET, Utah or VTT. From the table it is noteworthy that little attention has been paid to the behavior of the fluidizing bed itself. It has been widely recognized that the distinguishing particularities inherent to the fluidized bed technology are related to the movement of particles, strongly influencing heat transfer inside and thus, providing the outstanding characteristic of uniform and controlled temperature. Heat transfer during OF in FB has been only measured by ALSTOM (Nsakala et al., 2004), finding not remarkable differences between AF and OF heat transfer coefficients, although no numerical values were published. However, as stated by Eddings et al. (2009), the control of bed temperature when fixing exit O2 will be critical when varying oxygen levels. At great scale, some modeling results have been published, highlighting the role of a proper recycled solids rate and temperature for a proper control of bed heat balances (Nsakala et al., 2004; Saastamoinen et al., 2006; Seddighi et al., 2010). This temperature control will be an issue narrowly related with the SO2 capture process. On the one hand, as stated by García Labiano et al. (2011) optimum temperature for SO2 capture will differ from AF conditions, increasing for OF mode. Also, as indicated by Varonen (2011) or Nsakala et al. (2004), there are high probability of recarbonation of fly-ash, giving place to problems in the heat exchanger tubes, or meaning a risk of de-fluidization in the large CFB external heat exchanger. Additionally, temperature in fluidized beds must be strictly controlled under the limits of agglomeration and sintering risk. 1.2. Justification 20 Group Experimental facility Modeling Issues Refs CANMET CFB ID 100 mm H= 5 m No  Sulfure Capture  NOx emissions (Jia et al., 2007) VTT CFB ID167 mm H=8 m Yes, dynamic simulation  Sulfure Capture  NOx emissions (Pikkarainen, 2007) Czestochowa University of Technology CFB ID 250 mm H=5 m Yes, at large scale (not validated)  Fuel conversión (S,N,C) (Czakiert et al., 2010; Krzywanski et al., 2010a) ALSTOM CFB ID lower = 0.66 m ID upper = 1 m H = 18 m Yes, at large scale (not validated)  Sulfure Capture  NOx emissions  Combustion efficiency  Heat transfer (Nsakala et al., 2004) CSIC-ICB BFB ID 100 mm H = 0.6 m No  Sulfure Capture (de Diego et al., 2011) University of Utah CFB ID 250 mm H = 6.4 m No  Sulfure Capture  NOx emissions (Eddings et al., 2009) METSO CFB 1 x 1 m2 H = 13 m Yes  Sulfure Capture  NOx emissions (Seddigh et al., 2011; Varonen, 2011) Wien Technical University CFB ID 150 mm H =5 m No  Sulfure Capture  NOx emissions  CO emissions (Tondl et al., 2011) CNR BFB ID 40 mm H = 1 m Yes  Sulfure Capture  Attrition (Scala et al., 2011) CIRCE BFB ID 207 mm H = 2.7 m Yes  Sulfure Capture  NOx emissions  CO emissions  Combustion efficiency  Fluid-dynamics  Heat transfer (Romeo et al., 2010a; Guedea et al., 2011; Lupiáñez et al., 2011) Table 1.2.. Research groups with experimental experiences on oxy-fuel fluidized bed pilot plant For this reasons the current research is going to stress the attention on the heat transfer during the oxy-fuel fluidized bed operation. Results will have to be extrapolated towards the implications that these results will have on the future Chapter 1. Context, Justification and Objectives 21 large oxy-fuel boilers. Second generation designs of oxy-fuel boilers working with higher concentration of O2 in the comburent, will need to handle greater heat removal with less available surface for this exchange. 1.3 OBJECTIVES AND SCOPE In spite of the existence of several experimental pilot plants worldwide, running oxy-fuel experiments, none of them have measured heat transfer during oxy-fuel combustion in the bubbling zone of a fluidized bed. The relevance of this issue, resides not only in the quantification of heat transfer during the combustion in an oxy-fuel boiler. Large oxy-fuel boilers will have to handle greater solids rates and the overall heat balance of plant will change. The external heat exchangers attached to the CFB boilers in the solids recycled loop will gain an essential relevance. These devices work, in most of the cases, in a bubbling regime, to assure a proper conveying of solids back to the boiler. Heat from recycled hot particles will be removed partly on this device. It is assumed that the external heat exchanger will be fluidized with recycled flue gas, consisting mostly on CO2. The conditions in an external heat exchanger of an oxy-fuel boiler, will be similar to those existing on our bubbling oxy-fuel fluidized bed. Hence, the study of the heat transfer in the bubbling fluidized bed under oxy-firing conditions will be essential to correctly make proper design assumptions in future external heat exchanger devices. This thesis aims to demonstrate which will be the key aspects that will modify the heat transfer surfaces design in the future commercial scale oxy-fuel CFB boilers. With this purpose, this thesis will be addressed to:  establish how the heat balance of a large CFB will change in oxy-fuel combustion  quantify the heat transfer during oxy-fuel combustion in a bubbling fluidized bed, under several operating conditions  assess which will be the practical consequences of these changes within the plant configuration  evaluate the influence of larger solids recirculation on the performance of the external heat exchanger fluidized beds By means of a mathematical model of a large-scale oxy-fuel boiler, results will highlight the relevance that extra heat transfer surfaces to acquire a proper heat 1.3 Objectives and scope 22 balance of the plant when applying oxy-fuel combustion to the boiler. The external heat exchanger (EHE) will be then an essential device in future oxy-fuel CFB boilers. A new sub-model of generic EHE will be developed and integrated in the CFB loop previously developed. The appropriate estimation of the heat transfer coefficients in the EHE during fluidization with recycled flue gases lead to great uncertainties to quantify the heat transfer surface needed in the EHE for certain cooling requirements. Since the conventional frame of operation of EHE are inside the bubbling regime, measurements of heat transfer in the oxy-fuel bubbling fluidized bed pilot plant in CIRCE will give useful information. Additionally the contribution of recycled solids through the EHE on the heat transfer coefficient will be also quantified, by means of experiments on a EHE cold model. Experimental derived expression will be then included in the EHE sub-model, to reach the heat transfer surface area required under several oxy-fuel conditions. This work is divided into three different sections: First, a comprehensive one-dimensional model of a large-scale circulating fluidized bed has been developed. Oxy-fuel conditions have been run in the model. Results of the energy balance and the main difference of the operation conditions are explained in Chapter 2. The reader will find a review of the main modeling approaches and the explanation of the model itself, dividing the overall program in three main sub-models: fluid-dynamics sub-model, combustion sub-model and heat transfer sub-model. Validation of the model in air-firing conditions is made against literature results of large-CFB boilers. The energy balance validation under oxyfuel conditions is also compared with three published reports by different authors. The results are shown, analyzing the consequences of changing O2 content at inlet, the temperature of recycled solids and the effect of the boiler geometry on the overall plant balance. An essential role of the external heat exchanger device is drawn from this study. This will lead to the next section, when heat transfer in an oxy-fuel bubbling fluidized bed will be studied. Chapter 3 explores the main previous findings on heat transfer in fluidized beds, and the expected influence of oxy-fuel conditions, according to the known approaches. Heat transfer coefficient is then measured during the oxy-firing operation in the bubbling fluidized bed pilot plant in CIRCE. The bubbling fluidized bed pilot plant is described, and the way of cooling the bed during operation is explained. Influences of oxy-fuel conditions on the coefficients are Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 29 residual carbon content in particles that were underestimated by the model for coarse particles. However, this deviation was considered neglected because of the small portion of these type of particles in the dilute zone. Predictions of gaseous species were presented and well fitted by the simulation. More recently a similar approach with a particle population model was developed by Redermann and co-workers (2009) and applied to the 105 MW Stadtwerke Neumünster combustion plant (Germany), which operates with refuse-derived fuel. The model aimed to calculate the solids mass flows as well as the corresponding particle size distributions at any point inside the combustion system. The hydrodynamics considered were quite simplified, but they included detailed attrition expressions, with fitted attrition constants from experiments in the same plant. They also added the cyclone sub-model and external heat exchanger (EHE) as a stirred tank reactor. They could compare measurements of pressure drop and solids volume concentration profiles with data measured in the power plant. Measurements in large scale units are complicated and most of the times it is not possible to obtain data from the points that would be necessary for proper model validations. Because of this, Werther et al. (2009) proposed the so-called simulation-assisted measurement, consisting in introducing the measured values as inputs into the model for deducing information about the process at another location inside the combustion chamber. With this purpose they developed a 3D semi-empirical model. The horizontal velocity profile at every height was calculated as a potential flow field. In the dense zone, vertical and horizontal solids dispersion was considered but gas flowed in plug flow. In the upper part, splash, dilute and exit zones involved horizontal and vertical gas and solids dispersion. In the combustion sub-model, drying and devolatilization of fuel particles, primary fragmentation was assumed before volatile and char combustion. The aim was to determine secondary air penetration into the combustion chamber and its effect on the local combustion processes. The model was successfully developed for the 252 MWth CFB boiler of Stadtwerke Duisburg AG in Duisburg, Germany. Most of the data published along the years from a semi-industrial scale CFB combustor were obtained in the 12 MW CFB boiler installed in the Chalmers University of Technology, in Sweden. In spite of being a commercial size boiler, available for its use during the winter, this facility was equipped with a comprehensive set of instrumentation taps for acquiring detailed and realistic 2.1. Introduction 30 results on the plant performance during the operation. Long runs were possible, due to the commercial used availability, so realistic results were obtained continuously. Regarding fluid-dynamics, developments were gathered in the review paper by Pallarès and Johnsson (2006). Detailed heat transfer findings were also reported by Andersson (Andersson, 1988; Andersson, 1996). 2.1.2 OXY-FUEL CFB MODELING EXPERIENCES The interest of applying oxy-fuel combustion into fluidized bed technology, has been growing in the last years. Some groups have developed their own models for their experimental pilot plants, as previously summarized in Table 1.2. For the moment oxy-fuel CFB is not deployed at demonstration scale. Maximum thermal power size of pilot boilers currently running are less than 30 MW. Hence, modeling of oxy-fuel CFB is either a model of a small-rig and it is validated in the rig itself, or an existing and validated large-scale CFB model for air combustion adapted to oxy-fuel combustion conditions. Large scale oxy-fuel modeling has been scarcely published in the literature. Four main group’s works are available in the open sources. Three of them have experiences in oxy-fuel experimental fluidized bed facilities, but their modeling in large CFB cannot be validated. Nsakala and co-workers (2004) published a comprehensive report about OF in FB. They showed some results from their experimental Multi-Use-Fuel facility and modelled a large-scale OF CFB. They did not find remarkable differences between AF and OF heat transfer coefficients inside the boiler or inside the external heat exchanger, arguing that heat transfer in both zones are dominated by particles. When comparing the heat balance between large-scale AF and OF CFB models, they encountered a significant importance of an external way of transferring heat from particles. This is due to a boiler size reduction for the OF case, resulting in less available area inside the boiler. However they stated an additional handicap of using an external fluidized bed as heat exchanger: the temperature of limestone to be calcined for high CO2 partial pressures (around 900ºC) is higher than usual AF conditions (around 850ºC). Below this temperature, recarbonation of the formed calcium oxide from limestone would consume CO2. In the case of an external heat exchanger fluidized with flue gas, they calculated that almost all CO2 present in Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 31 the gas would be recarbonated at typical particles recirculation temperatures (around 650ºC). Thus, fluidization process would fail. For avoiding this, they proposed a moving fluidized bed, where no fluidizing gas was needed. Conventional AF temperature in oxy-fuel combustion were tested in the mini-CFB located in CANMET Energy facilities (Wu et al., 2011). They found CaCO3 presence in ashes supporting the belief that calcination was avoided. Saastamoinen and co-workers (2006) carried out experiments in a bench scale CFB-BFB facility, for different O2/CO2 ratios. They attained stable operation conditions and presented sulphur capture equally or more effective in oxy-fuel CFB than in air CFB combustion conditions. They also simulated a large scale CFB boiler, reporting some important re-design remarks for new oxy-fuel CFB boilers configuration. They calculated an available surface in a OF boiler being 38% smaller than the one for AF when feeding 60% O2 at inlet. They estimate that higher heat flux surfaces would lead to significant reduction in the boiler size. They suggested that attention must be paid in arranging the heat transfer surfaces so that no blocking of gas or solids may occur and in locating surfaces not so close that suspension in between could cool down excessively (Jäntti et al., 2007). Related to oxy-fuel efficiency penalty, it has been proposed a fire-more concept for producing more steam with the same size boiler, based on higher heat transfer fluxes in oxyfuel combustion (Hack and Shah, 2008). Seddighi et al. (2010) modelled a large scale fluidized bed and predicted particle recirculation rates higher than 30 kg/m2s for controlling bed temperature at O2 concentration as high as 90% at inlet (Johnsson, 2010; Seddighi et al., 2010). Krzywanski and co-workers (2010b) modelled an oxy-fuel boiler based on the 670 t/h lignite CFB boiler in Turow power station in Poland and simulated mixtures of O2/CO2 and O2/N2 atmospheres, confirming higher heat transfer to furnace walls when increasing O2 concentration at inlet. They considered a limit of 60% of O2 at inlet, stating that bed temperature over this concentration will exceed 1050ºC and so ash melting would take place. We will now review the fundamentals of one-dimensional modeling air-firing CFB boilers. From them, the strategies selected to model the oxy-fuel CFB boiler will be selected, and further explained in section 2.3. 2.2. Fundamentals of CFB modeling 32 2.2 FUNDAMENTALS ON CFB MODELING Process taking place inside a fluidized bed combustor can be grouped in three interdependent groups of phenomena: particles fluid-dynamics, combustion and heat transfer. In the Figure 2.2 interactions among said groups of phenomena are represented, as explained by Pallarès and Johnsson (2006): Figure 2.2. Interaction between the three great groups of phenomena occurring in a CFB combustor. Adapted from Pallarès and Johnsson (2006) Double arrows highlight the sensitivity of heat transfer and fuel conversion with fluid-dynamics modeling. On the other hand, fluid-dynamics could be reasonably well modeled assuming certain typical values for combustion or heat transfer parameters. This is the feature that makes fluidized bed exhibit such unique characteristics, versus other combustion technologies. Movement of particles along the riser determines where and how well are the fuel particle burnt and also, how intense is the heat transferred to the boiler walls. FLUID-DYNAMICS PHENOMENA FUEL CONVERSION HEAT TRANSFER Fuel, inert and sorbentdistribution Gaseous compounds External parameters External parameters External parameters Ref Type Scope Hydrodinamics Fragm. Combustion SOx Capture NOx HT Validation /Scale Level Zones Phases Devolatilization Coal combustion Shrinking Gen Rad (Adánez and Diego, 1995) CFB A II A B A C Y B C C D N (Aibéo and Pinho, 2003) CFB A II B B ? A B N (Arena et al., 1995) BFB & CFB A I A B Y A C ? C D N (Krzywanski et al., 2010b) OXY-CFB B II B C A D N B B B B N (Seddighi et al., 2010) OXY-CFB A II B A A C N D C A B N (Chen and Saxena, 1977) BFB? A I A A N F A N B C D C Y Literature (Sotudeh-Gharebaagh et al., 1998) CFB A I A A N A D N B B D D Y 800 kW (Gungor, 2009) CFB A II B B A C ? B D Y 50 kW (Gayan et al., 2004) CFB A II B Y D C Y D C C D Y 300 kW 100 kW (Adanez et al., 2001) Turbulent A II A B A C Y B C C D Y 300 kWth (Alagöz, 2006) BFB A I A B A C ? B C A C Y 300 kWth 16 MW (Chen and Xiaolong, 2006) CFB DYN B II Y C C ? B B A B Y 410 t/h Pyroflow CFB (Steady state) (de Souza-Santos, 2007) BFB I B ? Y 1.6 MW (Werner, 2001) CFB A II A B N A E Y 0.25m2 (Gungor and Eskin, 2008) CFB A III B B Y A D Y D D B D? Y 50 kW 160 MW (literature) (Wang et al., 1999) CFB C II A B Y A D N B B A A Y 12 MW (Huilin et al., 2000) CFB A II B B Y A D Y B C A A Y 30 MW (Romeo and Cortés, 1998) PFB A II B B N A A N C C B B Y 80 MW (Hannaes et al., 1995) CFB B III B B Y E C Y B B B B Y 120 MW (Lee and Kim, 1999) CFB A II B B Y A D Y B C A A Y 200 MWe (Werther et al., 2009) CFB B III A B Y C C Y D C D D Y 252 MWt (Vepsalainen et al., 2009) CFB A IV E D B B Y 370 MWt Table 2.1. Models of fluidized bed combustors 2.2. Fundamentals of CFB modeling 34 HYDRODYNAMICS: Coal combustion Scope: A: Difusionally controlled A: Boiler B: Rate of CO oxidation comparable to other chemical reactions B: Boiler + Cyclone + Standpipe + Loopseal (+EHE) C: Difusionally and kinetically controlled C: Steam cycle included D: Kinetically controlled Level: Shrinking particle model: I: 1D plug flow and stirred tank Y: yes II: 1.5D core-annulus structure N: no III: 2D numerically solved SO2 Capture: IV: 3D A: Desulfurization model Zones: B: Semi-empirical SO2 kinetics A: Two zones: dense and dilute C: Uniform in dense bed B: Three or more zones in the reactor D: Not considered Phases: NOx A: No distinctions between bubble and emulsion phases A: Formation and reduction model B: Two phase model: bubble and emulsion phases B: Semi-empirical formation and reduction kinetics C: Three phase model: bubble, cloudwake and emulsion phase C: Not considered Fragmentation included HEAT TRANSFER Y: yes A: Semi-empirical renewal model N: no B: Empirical correlations COMBUSION: C: Temperature is design variable Devlolatilization: D: Not included A: Uniform in dense bed Radiation: B: Instantaneous at feed point A: Zone Method C: Rate related to solids mixing rate B: Particle to gas heat exchange through StefanBoltzmann law D: Devolatilization kinetics C: Discrete Ordinates Method E: Particle movement model D: Not considered independently F : Not considered E: Two flux model (Schuster-Schwarzschildapproximation) VALIDATION Y: yes N: no Table 2.2. Nomenclature of Table 2.1 According to this scheme, numerous authors have developed their boiler models along last decades, with more or less level of complexity, depending of the aim on which the model was focused. In Table 2.1 some of these authors are shown, Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 35 attending at the strategies of modeling followed by each of the aforementioned fields. 2.2.1 FLUID-DYNAMICS A very extensively used and experimentally confirmed way of modeling fluidized beds divides the combustor into several zones in which particles movement is different and so they are the processes taking place in each region. In the bottom part of a fluidized bed, a dense zone is clearly distinguished. Above this zone, concentration of solids decays dramatically. Some researchers include in their models an intermediate zone called splash zone. At the top of the riser, a fourth region can be considered, the exit zone, very much dependant of exit geometry. In Table 2.1 this division is shown in the corresponding column, distinguishing between those authors that consider dense and dilute zones (Adánez and Diego, 1995; Arena et al., 1995; Sotudeh-Gharebaagh et al., 1998; Wang et al., 1999; Adanez et al., 2001; Werner, 2001; Alagöz, 2006; Werther et al., 2009), and those that also include spares and exit zones (Hannaes et al., 1995; Romeo and Cortés, 1998; Lee and Kim, 1999; Huilin et al., 2000; Aibéo and Pinho, 2003; Gayan et al., 2004; Gungor and Eskin, 2008; Gungor, 2009; Krzywanski et al., 2010b; Seddighi et al., 2010). Thebottomzone This region occupies a small volume in the overall CFB combustor, above the distributor, up to a height of around half a meter in large boilers. It is in this region, where particle density is hundred times higher than in the rest of the boiler and thus, most of combustion and chemical reactions occur here. The way of modeling the bottom zone can be as simple as considering it a well mixed stirred tank or assuming the two-phase model or even including the cloudwake phase and the mass transfer between the phases. The common point among these treatments is assuming that the bottom zone behavior is similar to that found in a bubbling bed regime. This was confirmed for example, with the experiments made by Svensson et al. (1996), that found that increasing fluidization velocities makes an increase in gas flow through the bubbles and the bubble dynamics, showing fluctuation frequencies similar to those in bubbling beds, around 1 Hz. In spite of these fluctuations generated by bubbles, dense bottom bed 2.2. Fundamentals of CFB modeling 36 has a quite uniform time-average void fraction. One of the main characteristics of the bubbling bed regime, observed in the dense part of a CFB, is the linear timeaveraged vertical pressure drop profile above the distributor, as confirmed by the authors. If an homogeneous mixing is assumed, i.e. the effect of the particle side distribution (PSD) on jetsam/flotsam segregation can be neglected, void fraction along the dense bottom zone is also constant. It can be expressed as:   emulsionbbb εδ1δε    Eq. 2.4 According to the two-phase model, bottom zone consists of a dense or emulsion phase, formed by bed particles and the interstitial flow between the particles; and a bubble phase, made of uprising gas bubbles, considered to be free of solids (Kunii and Levenspiel, 1991). First approaches assumed that all the gas in excess of umf flowed through the bed as bubbles, while the emulsion stayed at minimum fluidizing conditions. The expression for a single bubble rise velocity yields then:   2 1 bbr gd0.711u  Eq. 2.5 However, there have been experimental evidences that the flow of bubbles is overestimated by this assumption (Shen et al., 2004). This is because gas crosses a certain section of the bed, not only by translation of bubble voids, but also by the flow through the bubbles relative to them and the interstitial flow relative to the particles in the dense phase. So, the overall gas balance would fulfill the following expression: vismftf uuuu    Eq. 2.6 where the visible flow, uvis, would yield: bbvis uδu  Eq. 2.7 Or it can be also expressed as the dimensionless visible bubble flow, Ψ: mf vis uu u ψ  Eq. 2.8 Several empirical expressions have been proposed and used by researchers for estimating the bubbling velocity. Some of them are summarized in Table 2.3. Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 37 Reference Correlation Farrokhalaee (1979) in (Davidson et al., 1985)   D 0.24x expHu0.44uuu 0.230.32 mf 0.62 b   Eq. 2.9 Peters et al. (1982) in (Davidson et al., 1985)   mfb u-uYu      2 mfmf uu5uu0.130.785Y  Eq. 2.10 Werther (1983) in (Kunii and Levenspiel, 1991)     br 1.35 t 0.5 bmfb ud1.13duu1.6u  Eq. 2.11 Werther and Wein (1994) in (Pallarès and Johnsson, 2006) 0.18 1.45Arψ  Eq. 2.12 Johnsson et al. (1991) in (Pallarès and Johnsson, 2006)   4 0CFB A4hfψ ref 5 b 1 CFB ΔP102.6116.6d0.129u0.3121f   ref ΔPbeing the pressure drop between riser height interval from 0.135 to 1.635 m above gas distributor Eq. 2.13 Eq. 2.14 Table 2.3. Correlations for estimating bubble velocity Thetransportzone When bubbles reach the limit of the dense bed they explode in the surface. Aggregates of particles are thrown into the splash zone. Due to the differences of particles physical properties, drag forces per unit weight also differs from one type of particle to another. A high drag force per unit weight makes the particles move upwards, whereas particles with low drag per unit weight will tend to sink to the bottom. During this instantaneous explosion of bubbles in the bottom bed surface, competition between both mechanisms, mixing and segregation, occur simultaneously. A strong back-mixing is caused in this region by the decay in the solids concentration. The transport region comprises the height between the dense bottom zone limit and the top of the riser. For covering both, the upper diluting and the lower dense regions, the following expression was proposed by Li and Kwauk in 1980 (Horio, 1997):   i sb sp zzaexp εε εε    Eq. 2.15 This way of modeling back-mixing and decay of solids concentration in the freeboard has been extensively used by most of models in Table 2.1 2.2. Fundamentals of CFB modeling 38 The decay constant for these models, a, is inversely proportional to the gas velocity. It represents fairly well the effect of the back-mixing of solids above the dense bed, dominated by clustered solids thrown up into the freeboard. For this reason, Johansson and Leckner (1995) proposed a two-solid-phase model, consisting of a cluster and disperse phases. The first, describing a ballistic type of solids back-mixing, similarly to Eq. 2.15 and the second, dominating the contribution of the wall layer falling solids effect. The formed clusters in the dense bottom zone are immersed in the disperse phase but there are few that reach the top of the riser. So, particles are elutriated from the bottom rising upward along the central region of the riser. In this region, the flow is turbulent due to high slip velocities (the difference between the mean interstitial gas velocity and the mean upward particle velocity) and particles are thrown towards the wall. A falling film is then formed. This falling layer captures particles close to it and those particles are entrained back to the core less frequently. This makes the concentration of solids along the riser height adopting an exponential shape (Davidson, 2000). The expression derived by Johansson and Leckner (1995) including the mentioned back-mixing processes yields:         zHKexpρHzaexpρρρ exitexitxhdisp,bb     Eq. 2.16 Where a corresponds, as in Eq. 2.15, to the back-mixing from homogeneous clustering flow in the splash zone. K is the decay coefficient for the back-mixing in the wall-layer flow (Johansson et al., 2007). Zijerveld et al. (1997) calculated the decay constant using Eq. 2.15 and Eq. 2.16 in three boilers, the 12MWth CFB, a 1.2m x 0.8m and a 0.083 m ID CFB. They found that Eq. 2.15 fails to predict satisfactorily the concentration in the larger risers, so they suggest to avoid that expression for those commercial CFBs. The exponential behavior of solids along the riser is then due to the lateral transport of solids from the core to the down-flow wall layer of solids. According to Horio (1997) there are two other ways of modeling the lateral solids transfer. One approach explains the core-annulus flow by the radial velocity distribution and radial density profile. The other explanation is based on time-average mass and momentum balances. The wall thickness, tδ, is the distance between the riser walls to the point where net flux of solids is zero in the vertical direction. This magnitude has been Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 45 CONchar NO 2    R. 2.7 22 char CON 2 1 CO NO  R. 2.8 Nitrous oxides, N2O, are formed similarly than NOx. It is formed from N-fuel and it requires the same precursors: HCN and NH3. So, reactions yielding these intermediate compounds are common for all nitrogen oxides: COONNO NCO 2    R. 2.9 HONNO NH 2    R. 2.10 Bed temperature is an important parameter. According to de Diego et al. (1996) bed temperature makes NOx emissions decrease but N2O increase. This influence is even greater for higher air excess. Also, when incrementing air excess, NOx and N2O increase. This is explained because higher O2 concentration increases volatile and chars combustion and also, at the same time, concentration of char and CO is lower, so heterogeneous reduction of NO is decrease in the char surface. N2O is also increase by higher volatile (NH3 and HCN) conversion via homogeneous reactions as: HNCOO HCN    R. 2.11 COONNO NCO 2    R. 2.12 NOx and N2O is reduced by increasing secondary air injection. This is an extensively used method for control of pollutants. This is because combustion occurs in oxygen-limited conditions, favoring volatile conversion from N to N2, instead to NOx. Moreover, char is accumulated and higher CO concentration allows decomposition of NO via CO (R. 2.8) in the char surface. Finally, residence time of gas in the lower part increases. Thus, NOx decreases because NOx decomposition is faster than NOx formation along the riser. Mechanisms of formation and destruction for NO and N2O during fluidized bed combustion have been thoroughly studied in the literature. A clarifying way of understanding main conversion paths of fuel nitrogen during fluidized bed combustion was explained by Leckner (1997) and it is schematically present in Figure 2.4 2.2. Fundamentals of CFB modeling 46 Figure 2.4. Main conversion paths of fuel nitrogen. Adapted from Leckner (1997) In oxy-fuel combustion there have been numerous investigations of NOx formation specially in pulverized coal combustion. Experiments from the Czestochowa CFB pilot plant (Czakiert et al., 2006) show a significant limitation of fuel-nitrogen conversion to NOx during oxy-fuel combustion. In the 50kW CFB in Southeast University in China the effect of operation parameters of oxy-fuel combustion on NO emissions were analyzed by Duan et al. (2011a). NO formation decreased in 21%/79% O2/CO2 atmosphere. They explained it as due to the less O and OH radicals formed under lower bed temperature. In addition, low temperature and high CO2 concentration bring possibility of higher CO concentration in O2/CO2 atmosphere. As the O2 concentration at inlet is increased gas velocity diminishes and this means longer residence time of fuel particles in the combustor, which would promote fuel-N conversion into NOx precursors. Increasing bed temperature or oxygen stoichiometric ratio brings higher NO emission in O2/CO2 atmosphere, which is consistent with the results in air-fired CFB combustion. Stream staging is more efficient for controlling NO emission in oxy-CFB combustion than that in air combustion. Another important issue in oxy-fuel is the positive effect of RFG on the decrease of NO emissions. Okazaki and Ando (1997) examined the effects of CO2 concentration, together with the less recycled-NO in the flame, and the interaction of fuel-N and Nfuel Nchar Nvol NH 3 HCN NH 2 N 2 N 2 N 2 NO N 2 ONO N 2 ONO N 2 O N 2 N 2 NO Nchar NH 3 HCN N 2 O Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 47 recycled NOx on the decrease of the final NOx exhausted from the coal combustion system with the recycled CO2. They concluded that the conversion ratio from the fuel-N to exhausted NOx is reduced to less than one-fourth of that with air combustion, and the effect of a reduction of recycled-NO in the furnace was dominant and amount to 50%-80%. In the current model, NOx formation and destruction reactions are not included, since their energy interactions can be neglected, compared with combustion and sulfation reactions. 2.2.3 HEAT TRANSFER For decades, there have been numerous investigations on heat transfer in fluidized bed, specially carried out in laboratory scale facilities. Although not completely independent, there is a generalized treatment of heat transfer from bed to surfaces as the contribution of three mechanisms: the transient conduction of particles when contacting the surface, also called particles convection; the gas convection, and the radiation heat transfer. In fast beds such as circulating fluidized bed, heat transfer is strongly dependent on particles fluid-dynamics. Parameters like rate of solids circulation, superficial velocity through the column, and particle size, will define the dominant heat transfer mechanism. Particles travel upwards along the core in the form of strands. When strands approach the wall layer, they change direction. Therefore, energy is transferred from particle strands or agglomerates at bed temperature that “enters” into the wall and exchange heat with the surface during certain residence time. This is called renewal packet theory and it is further explained in the next chapter. Radiation heat transfer turns out the dominant heat transfer mechanism for the low density boiler zones. Normally, gas-particle suspension can be considered isothermal. Thus, the expression for the radiative heat transfer coefficient related to a plane grey heat transfer surface follows the expression:     wb 2 w 2 bbwr TTTTσεh Eq. 2.25 Where 2.2. Fundamentals of CFB modeling 48          1 ε 1 ε 1 ε wef bw 1 Eq. 2.26 As stated by Baskakov and Leckner (1997) there is not a unique methodology for accurately estimate the value of emissivity of an isothermal suspension. All those methodologies coincide in indicating that particle emissivity is the main factor for the suspension emissivity, so as particle porosity. For estimating values of the radiative heat transfer, these authors separated the effect of gas phase convection heat transfer from the measures values, calculating the gas contribution by the expression: 0.430.8 gc Pr0.003ReNu  Eq. 2.27 The heat transfer coefficient in the membrane wall of the 12MWth CFB at Chalmers was 104 W/m2K. Convection and radiation heat transfer coefficients were estimated around 20 and 78 W/m2K respectively, confirming good fitting of aforementioned assumptions. Basu and Nag (1996) collected heat transfer data from several commercial CFB units. Average heat transfer coefficients ranged from 80-220 W/m2K and shows great dependency on suspension density. He correlated data against suspension density and obtained the following expression: n s aρh Eq. 2.28 with 40 and 0.5, for 5 25/m  and 750    850ºC. Similar expressions were proposed by Divillo and Boyd (1994), where 23.2 and  0.55, although this was proven by Xie et al. (2003) to overestimate data for densities greater than 200 kg/m3. Basu and Nag (1996) explained the dependence of heat transfer in large units with temperature. There is an obvious increase of heat transfer coefficient with increasing temperature, due to higher gas thermal conductivity and higher radiation. At a relatively high suspension density of 20 kg/m3 the increase in heat transfer coefficient with temperature was linear. This may not be the case at a more dilute bed where the radiation becomes dominant. Regarding fluidization velocity there were not much influence on heat transfer in large units. It was reported that change in secondary air rate did not affect the heat transfer Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 49 coefficient in the upper part of the furnace, but an increase in the primary air velocity did increase it (Andersson and Leckner, 1992). This happened because the increased primary air transported more solids to the upper section increasing the suspension density in that region of the bed. Particle size had a strong effect in laboratory units, but influenced much less heat transfer in large units. This is not the case for bubbling fluidized beds where the particle size is the most important factor influencing heat transfer irrespective of the plant size. Breitholtz et al. (2001) reviewed heat transfer coefficients from laboratory and industrial scale fluidized bed boilers and proposed an exponential simplified expression for particle convection heat transfer that fit data of all the cases studied. Considering the different mechanisms to transfer heat as independent contributions, the overall bed-to-wall heat transfer could be expressed as: maxr,r b 0rconv hηραhhh  Eq. 2.29 Where hconv comprises particle and gas convection mechanisms and hrad is radiation heat transfer coefficient. Parameters for the equation yields α0 = 25 and b = 0.58. ηr refers to the radiation efficiency term, that gives an idea about the shadow effect of the boundary layer. In the extremes cases, this term would becomes 1 for a very thin optical wall layer, and 0.65 when optical layer is thick. ηr takes the form:       1.6 2.6 ρ arctan 0.140.86 r ηs Eq. 2.30 This expression shows an asymptotical shape on the value 0.65, i.e., for dense wall layer only 65% of the energy emitted by radiation will be seen by the wall. Cheng and co-workers (2007) reviewed numerous measured heat transfer coefficients in large fluidized beds. They measured heat transfer in a 12 MWe, 50 MWe and 135 MWe CFB boilers and observed again the high dependence of these values with the solids concentration in the riser, and in different zones of the boiler, such as the cyclone, and the economizer. They confirmed that it is mainly influenced by solid suspension density and furnace temperature. They stated following a correlation in the form of: β s α sTKρh Eq. 2.31 Being K, α and β regression coefficients, which are not specified in the paper. They proposed this expression because in the review of large CFB most of expressions for 2.2. Fundamentals of CFB modeling 50 prediction heat transfer coefficients were in the form of Eq. 2.28, and they stated that temperature had a remarkable influence and must be included in the correlation. Recently, Krzywanski and co-workers (2010b; 2010a) applied the model developed for the 670 t/h CFB boiler operated in Turow Power Station in Poland. They calculated the heat transfer coefficients in the dilute zone, by the expression: 1f 1 hε1 h h  Eq. 2.32   rconv1 hhh    Eq. 2.33 Where εf is a fouling coefficient ranging from 0.001-0.003 and  is around 0.85, representing the non-uniform flow near the heat transfer surface.                                   0.67 mw eq 0.4 0.8 eq eq conv H D 1Pr100 ν uD D k 0.214h Eq. 2.34 g w 4 g w 3 gs w r T T 1 T T 1 Th 2 1h σh             Eq. 2.35 s α the emissivity of the gas-solid phase was calculated as the absorptivity of a path s through the gas-particle suspension expressed by Bouguer's law: κs se1h   Eq. 2.36 Where pp B dρ 1.5 κ   Eq. 2.37 w furnace A 3.6V s Eq. 2.38 Comparison of both, the method by Breitholtz et al. and Krzywanski et al. is represented in Figure 2.5. In Figure 2.5a) it can be observed that estimated values of total heat transfer coefficient from the two approaches are closer for higher solids density in the riser. The model used by Krywanski gives very low values for small solids densities. Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 51 Looking at the estimated values for radiative heat transfer, Figure 2.5b), in the model of Krzywanski et al. they tend to zero for very low solids concentration, due to the form of Eq. 2.13. The influence of particle densities shows opposite behavior in the approach by Breitholtz et al. Denser riser leads to denser wall layer and the shadow effect to the heat transfer surfaces increases, and thus, heat transfer by radiation decreases. This is not the case of the expression used by Krzywanski et al. since the term for suspension absorptivity makes denser medium leading more intense radiation. The influence of the radiation path thickness is shown to influence remarkably on the estimation of the radiation. This term is not considered by Breitholtz et al. but the empirical term of radiation efficiency allows taking density of wall layer into account. Figure 2.5. Dependence of the a) total and b) radiative heat transfer coefficient with the bed density according to Beritholtz and Leckner (1997) and Krzywansky et al. (2010b) for different radiating layer thickness, s 0246810 0 45 90 135 180 225 Average solids concentration (kg/m 3 ) Total heat transfer coefficient (W/m 2 K) Krzywanski s=0.05m Krzywanski s=0.1m Krzywanski s=0.1m Krzywanski s=0.2m Krzywanski s=0.2m Krzywanski s=0.3 m Krzywanski s=0.3 m Breitholtz Breitholtz 0246810 40 60 80 100 120 140 160 180 200 Average solids concentration (kg/m 3 ) Radiation heat transfer coefficient (W/m 2 K) Krzywanski s=0.05m Krzywanski s=0.1m Krzywanski s=0.1m Krzywanski s=0.2m Krzywanski s=0.2m Krzywanski s=0.3 m Krzywanski s=0.3 m Breitholtz Breitholtz a) b) 2.3. The oxy-fuel CFB model 52 In none of the aforementioned models, radiative gas properties influence appreciably the radiative heat transfer coefficient. This is not the case for heat transfer in pulverized coal boilers, where radiation from the flame through the gas is the dominant heat transfer mechanism. It is generally assumed that heat transfer under oxy-fuel combustion conditions behaves similarly to air-firing conditions, i.e. main parameters influencing heat transfer such as bed temperature, particles concentration, particles diameter or gas velocity affects heat transfer the same way as it does under oxy-fuel conditions. However, there are no experimental published data about this issue and models available in the literature takes the same models used for air-combustion. 2.3 THE OXY-FUEL CFB MODEL The overall model for the OF in CFB has been sub-divided into three nonindependent modules that interact with each other as shown in Figure 2.2: i) solids fluid-dynamics, ii) combustion and iii) energy balance. The sequence of sub-models arranged for CFB modelling is represented in Figure 2.6. Main initial considerations are following explained, for understanding the significant operational factors that would determine the overall CFB behaviour. ‐ Fuel input and fuel characteristics: Thermal power input is determined by fuel rate and by its composition and heating value. These parameters allow calculation of stoichiometric O2 needed plus certain excess for assuring proper combustion. In the case of air combustion, an air-excess around 20% is generally accepted to reach around 3-6 % O2 in flue gases. Similar oxygen contents in flue gas have been considered in the OF simulations. In any case, the fuel thermal input imposes the amount of oxidant flow. ‐ Inert matter properties: Main fluidization parameters such as fluidization velocity or terminal velocity are determined by the particle density of the bed material and particles size distribution, among other factors. A solids inventory consisting of particles with a diamter of 200 μm and a density of 2500 kg/m3 (Geldart B type) has been assumed. ‐ Boiler geometry. The amount of gas entering the boiler and the boiler cross sectional area will settle the fluidization velocity and thus, how solids will Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 53 be distributed along the riser. Secondary gas inlet is located at 5 m height from the distributor. ‐ Pressure drop along the combustor: This parameter is easily measured with conventional pressure taps and bottom bed height can be estimated. Values around 9000-12000 Pa have been considered as average range on conventional air firing boilers. Thus, the pressure drop limits the height of the lower dense region (Yang et al., 2005; Redemann et al., 2009; Lu et al., 2010) The first sub-module in the programme aims to estimate the solids concentration profile for certain fuel input, boiler geometry and pressure drop across the boiler. It calculates fluid-dynamics in the dense and in the transport zones, such as void fraction, gas velocity and wall layer thickness at every height. For estimating gas properties, temperature profile must be an input that will be calculated in the energy balance sub-model. Figure 2.6. CFB model flow-chart 2.3. The oxy-fuel CFB model 54 The combustion module calculates the main formed species based on disperse and cluster phases present at each cell. Temperature must be also an input for estimating if calcination of limestone takes place before sulfation, according to Eq. 2.22. Finally, the energy balance sub-module will carry out the energy balance cell by cell, according to the combustion, heating of gases, heat carried by the solids and cooling at each height. In Figure 2.7 it has been graphically represented the processes taking place in each part of the control volume considered in this model. Black arrows represent the flux of solids, modeled by the fluid-dynamic module, where primary and secondary gas are the inlets, so as the boiler geometry and pressure drop. Red circle arrows represent the heat of reactions. As can be seen, combustion in the loop-seal was not considered. During air firing combustion, unburnt fine particles are likely to be burnt by the O2 present in the fluidizing gas. However, O2 in the fluidizing gas of the EHE will be avoided, for economical reasons. Figure 2.7. CFB boiler mass and energy balance scheme Primary gas Fuel Secondary gas Recycled particles Flue gas Water walls External Heat Exchange Convective pass Particles Gas Combustion Energy Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 61 C2H6, C2H4 and C2H2. In none of the runs convergence on the volatile composition was reached. In the Figure 2.10, some of the simulations are shown: In Figure 2.10a), the basics components and in Figure 2.10b), C6H6 is added. Figure 2.10. Simulations results of non-convergence of applying volatile composition methodology, considering. a) CO2, CO, H2, H2O and CH4; b) CO2, CO, H2, H2O, CH4 and C6H6 There is no point in which all parameters are above zero and below 100%, so no feasible solution is possible considering these compounds. The same occurred when adding other compounds to the iterations of Eq. 2.51. Possible causes of this lack of convergence could be that this methodology was developed for biomass fuel, with a higher volatile content than coal. An alternate possibility of estimating volatile 020 40 60 80 100 -10 0 10 20 30 40 50 60 70 CH 4 (%) CO, CO 2 , H 2 , H 2 O (%) COCO CO 2 CO 2 H 2 H 2 H 2 OH 2 O 020 40 60 80 100 -200 -150 -100 -50 0 50 100 150 200 250 CH 4 (%) C 2 H 6 , CO, CO 2 , H 2 , H 2 O (%) C 2 H 6 COCO CO 2 CO 2 H 2 H 2 H 2 OH 2 O a) b) 2.3. The oxy-fuel CFB model 62 composition is by looking at the literature coal composition analysis. Hence, energy balance is compromised, by not fulfilling the heating value of the volatile part. This leaded to adopt a simpler way of considering volatile composition that fulfilled the energy balance, by compromising accuracy in the mass balance, since the simulation was focused on looking at the overall system energy balance and minimum deviation in species formation could be neglected. Thus, all volatile carbon would yield CO2. To complete the volatile heating value LHVvolatile, part of hydrogen would produce H2 and the rest would be released in the form of H2O and NH3. Once volatile composition is estimated, these compounds are released and burnt quickly, before fixed carbon. Volatile and moisture are released proportionally to the accumulated solids in the dense phase and to the cluster phase solids in the riser. This is in accordance with the assumption made by numerous authors of considering volatile combustion taking place completely in the dense zone of the boiler (Basu, 1999; Gungor, 2009) . In every cell volatile matter is oxidized before char, if oxygen is available. This means that volatile combustion is instantaneous and controlled by the devolatilization rate, as explained by Oka (2003). Char is released proportionally to the accumulated solids in the dense phase and to the cluster phase plus the disperse phase solids in the riser. This agrees with the concept that elutriated fine char particles can burn along the riser. These fuel particles find an oxygen enriched zone after the secondary air injection. Un-burnt char particles that found reducing atmosphere in the dense phase during volatile combustion will have time to be in contact with oxidant before the exit zone. Below secondary gas is injection, a reducing conditions zone can appear. This is because volatile combustion can consume the primary oxygen before the secondary gas is fed. In practice, the good back-mixing of gas and solids in the fluidized bed, allows oxygen from the secondary injection point encountering un-burnt carbon. For taking into account this good back-mixing of gas and solids, that ideally avoids un-burnt matter, a new parameter is considered. The back-mixing parameter, kbm is applied to char and volatile matter in the oxidizing conditions areas, so that the result is a complete conversion of char. This is graphically represented in the Figure 2.11. Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 63 Figure 2.11. Scheme of applying back-mixing parameter, kbm. Z being the char in each cell In the left hand side of the Figure 2.11 O2 from primary air is consumed before the secondary air is fed. Thus, it appears a region of reducing conditions. In the right hand side, the secondary air contact also with the unburned fuel below the feeding point, thanks to the application of the back-mixing parameter, kbm. Using the temperature profile input, from the energy-balance module, allows to estimate the equilibrium CO2 partial pressure of the calcination-carbonation reaction, as expressed in Eq. 2.22. This value is compared with the CO2 partial pressure at each height. Thus, it is possible to estimate if calcination is taking place or limestone is sulfated directly through the one-step reaction. Values of CO2 partial pressure over this value, lead to calcination conditions. If Peq values are higher than CO2 partial pressure, direct sulfation would take place. Together with the efficiency values for each case, SO2 emissions and CaO formation is calculated along the riser. S conversion to SO2 occurs after volatile released and combustion. Efficiency values of 80% and 95% have been applied for direct and indirect desulfurization respectively. These values were experimentally obtained in Bolea et al. (Bolea et al., 2012a) 2.3. The oxy-fuel CFB model 64 2.3.3 ENERGY BALANCE The CFB system must fulfill the overall energy balance as shown in Figure 2.12. Fuel input and gas, together with the recycled solids from the loop-seal are the global entrances of the energy balance system. Heat transfer to walls, to the convective pass and to the external heat exchanger, must ideally complete the overall energy balance. Figure 2.12. Sankey diagram of the CFB Programming sequence begins then from bottom dense bed. Inputs are the temperature of this zone, primary gas and fuel entering the furnace. Also solids recirculation rate from the fluid-dynamical sub-model is needed. In this zone, combustion takes place as calculated in the combustion sub-model. An initial temperature for recycled solids is assumed. Another required initial value is a temperature value for solids falling down by the walls. Thus, temperature of elutriated particles is the input for energy balance of first cell in dilute zone. Energy balance of each cell in the core makes a first temperature profile for core cells. From the last core cell, energy balance to the annulus cell starts. When energy balance of wall cells takes place in the first wall cell, energy balance in the dense zone is actualized. Bed temperature profile is obtained by the energy balance applied to every cell. The programming sequence is represented in Figure 2.13: Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 65 Figure 2.13. CFB energy balance modeling flow-chart In Figure 2.13 represents how the temperature profile from this sub-model is introduced again in the combustion sub-model, and reaches the convergence by iterative process. This temperature is used for calculating equilibrium CO2 partial pressure profile and so, estimating if direct or indirect sulfation conditions are taking place. Dense zone and dilute zone are treated differently. The schemes of energy balance in both zones are represented in Figure 2.14 and Figure 2.15. 2.3. The oxy-fuel CFB model 66 Figure 2.14. Energy balance scheme in the dense zone of the CFB The dense zone is considered isothermal and no heat transfer surfaces are present in this zone, i.e. a constant temperature along bottom bed zone is assumed. This assumption is common in numerous models in the literature (Table 2.1). The intense fuel mixing of this area, together with high solids density and their high thermal capacity make this assumption fair enough. Thus, the dense zone energy balance is treated as a whole, following the expression:        bb CHARcomb, bb VOLcomb, bb bb Outg,ing,Ing,bbCore,1Wall,1Wall,1RSRSS QQ TEmTEmTmTmTmCp   ... ... Eq. 2.52 Right hand side of Eq. 2.52 is the combustion energy from the char and volatile reactions. This energy is invested in heating up the gaseous oxidant stream mg,in, the solids from the recycled pipe, mRS and the solids falling from the wall layer of particles, mWall. The output gaseous and solid flows are the elutriated solids to the transport zone, mCore, and the resulting flue gas with the remaining oxidant, mg,Out, respectively. The energy balance in the dilute zone is divided into the core cells and the wall layer cells. The energy balance for each core cell is:         wallcore HT i CHARcomb, i VOLcomb, i i out, gas,1-i i in, gas,i i lat,i i out, core,1i i in, Core,S QQQ TEmTEmTmTmTmCp     Eq. 2.53 Volatile and char combustion heats up the solid and gaseous streams arriving at the cell, mCore,in, mgas,in, and leaving the cell, mcore,out, mlat and mgas,out and heat is also transferred to the wall layer, QHT|core  wall. The energy balance of a wall layer cell is then: Solids recirculation (T recirc ) Wall layer downward solids (T wall i+1 ) Eentrained solids (T out 1 ) FUEL CONVERSION Fed fuel (T in ) Fed gas (T in ) Flue gas + remaining fed gas (T out 1 ) Dense bed mass/energy balance Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 67   surfacewall HT wallcore HTi i lat,i i out, Wall,1i i in, Wall,S QQTmTmTmCp   Eq. 2.54 In the wall layer no combustion reaction is considered, and energy entering with the falling solids, mWall,in, and solids from the core, mlat, together with the heat from the core cell, QHT|core  wall, is transfer to the heat exchange surface QHT|wall  surface and to the falling solids, mWall,out. The model by Breitholtz et al. (2001) shown in Eq. 2.29 was implemented to estimate heat transfer to the walls. Energy is transferred with the solids to the wall layer, according to the mass balance and the core temperature. The radiation from the core to the wall is calculated with Eq. 2.25. Heat transferred by radiation is not straight received by the walls. Instead, the core heats up the falling solids. The radiation efficiency of Eq. 2.30 is included, since high solids density is expected in oxy-fuel combustion, and thus, denser wall layer will increase the shadow effect to the radiation from the core. This could be clearer understood in Figure 2.15 Figure 2.15. Energy balance scheme in a cell of the dilute zone of the CF Once the integration of the three sub-models, fluidynamics, combustion and energy balance, converges, we obtain the temperature, pressure, gas composition and solids density at every height along the boiler. The results drawn from the model must be now compared to real values, in order to confirm the model validity and so, to enable to be used as a reliable tool to establish the implications of oxy-fuel combustion in large oxy-fuel CFB boilers. B FUEL CONVERSION Solids upward (Tout i-1) Solids upward (Tout i) Gas (Tout i-1) Gas (Tout i) Wall layer downward solids (Twall, i) Wall layer downward solids (Twall, i+1) Lateral solids (Touti i) HEAT TRANSFER TO WALLS Core cell mass/energy balance Wall cell mass/energy balance core particles radiation 2.4. Model validation 68 2.4 MODEL VALIDATION 2.4.1 AIR FIRING In spite of the scarce published data on commercial and large-scale fluidized bed boilers, some published measurements could be useful for confirming the model feasibility and reliability: Published data Author Pressure drop profile Yang et al. (2005) Flue gas composition Lee and Kim (1999) Temperature profile Hannaes et al. (1995) The measurements published by Yang et al. (2005) has been taken as a reference for comparing pressure drop along the boiler. The 135We CFB boiler, located at Zibo power station was 38 m height, and the cross section was 6.6 x 13.1 m2. They specified the coal used in the plant, that contained 45.39% of fixed carbon, 4.44% moisture and 16.73% of volatile matter. The heating value was 21060 kJ/kg. Some of the operation conditions at which measurements were taken were indicated. A temperature of 885ºC in one case and 896ºC in the other was average along the riser. Elutriated solids were 5.79 and 6.50 kg/m2s respectively. With these inputs, comparison of measurements and model are shown in Figure 2.16.  Figure 2.16. Comparison of model pressure profile prediction with the one reported in Yang et al. (2005) Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 69 From the Figure 2.16 it can be confirmed that realistic values of pressure drop profile are drawn from the model. For main gas species concentration in the boiler, emissions of CO2 and O2, are reported by Lee and Kim (1999) and Hannaes et al. (1995), for a 200 and 120 MW boiler respectively. The former was a 32 m height CFB with a cross section of 19 m 7m. The coal used is a Korean Anthracite with 53.7% fixed carbon and only 4% of volatile matter and 3.3% moisture. Average temperature was 900ºC. Entering said parameters, comparison results are shown in Figure 2.17.  Figure 2.17. Comparison of flue gas composition prediction with the one reported in Lee and Kim (1999) Figure 2.17 cannot be considered a proper validation, since measurements of the gas composition along the height of the riser were not available and the deviation of CO2 and O2 results is usually small, even for the simplest modelling approach. Even though, the common available data from real boilers are taken at the flue gas outlet. Figure 2.17 is also interesting for showing the estimated CO2 and O2 profile by the model. It shows the reducing zone commented in the section 2.3.2 and it can be also distinguished that most of the combustion reactions take place in the lower zone of the boiler. Temperature profile measurements in a 120 MW boiler was reported by Hannaes et al. (1995), for validating the IEA-CFB model. The boiler is 30 m height and 38 m2 in cross section. However, they do not specify the coal used in the tests, so a bituminous coal was assumed, with a fix coal fraction of 63%, 5.7% moisture and 24% volatile, and 24450 kJ/kg as heating value. 0% 5% 10% 15% 20% 25% 0 5 10 15 20 25 30 35 Volumefraction(%) Height(m) O2Exp(Lee1999) CO2Exp(Lee1999) O2model CO2model 2.4. Model validation 70 Figure 2.18. Comparison of the model temperature profile result with the one reported in Hannaes et al. (1995) In Figure 2.18 the two highest thermocouples indicated a drastic decrease of temperature. The reason was the additional heat exchanger tubes located at this point. Solids at this location were then further cooled. In general, the model developed in this work gives reliable results. It can be adapted to different input conditions, satisfactorily estimating the measurements from large scale boilers. More accurate predictions could be achieved, particularly with a more detailed combustion modelling. However, for the aim of this work, the model is more useful if it can be flexible for different configurations and inputs parameters. 2.4.2 OXY-FIRING Since there is no available data from commercial CFB installations operating under oxy-firing conditions, the validation for these cases is based on a quantitative comparison of published simulations (Nsakala et al., 2004; Saastamoinen et al., 2006; Seddighi et al., 2010) In order to validate the present model, heat transfer ratio per fuel input in the different heat transfer locations will be compared. Three main zones are distinguished in the the CFB system: furnace walls, including water walls and wing walls; external heat exchanger (EHE) in the loop-seals, for cooling down recycled particles before re-injected to the boiler; and the energy contained in the flue gas stream. Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 77 20-43%. In spite of keeping temperature difference between the bed and the recycled solids similar in all cases, the increase of Gs makes the contribution of the EHE increments from 42.3 to 52.8%. 2.5.3 VARYING O2 AT INLET The distinguishing characteristic of OF combustion versus AF combustion lies on the possibility of modifying the O2 content in the fluidizing gas. While in AF, the only way of changing fluidizing velocity is by incrementing the air excess, in OF case, it also can be varied by manipulating the O2 concentration at inlet. In future concepts of OF CFB boilers it will be essential to know what flexibility is possible to achieve for a certain boiler arrangement similar to AF case. In the following runs an area of 16 x 9.4 m was considered, as the adequate size for air firing boilers. Reaching AF temperature conditions is possible with an O2 concentration around a 30% v. This statement has been extensively assumed for pulverized fuel boilers (Scheffknecht, 2009) but fluidized bed do not present such a great restriction in feeding wider range of O2 concentration. A proper temperature profile along the furnace is possible for higher O2 concentrations, when fuel input is also increased. Therefore, higher volume of gases must be introduced into the boiler and so, higher velocity entrains more solids to the particles separator device (cyclone). Figure 2.23 shows the distribution of heat transfer along the system, for keeping similar temperature profile in the boiler, but different conditions of the oxidant stream.  2.5. The role of the external heat exchanger 78  Figure 2.23. Heat transfer share with air design boiler geometry and different O2 concentrations at inlet When O2 at inlet is incremented up to 40%, it is possible to raise entrainment of solids, up to 20.7 kg/m2s. More particles are managed by the EHE and so 36.7% of the total fuel input is removed in this part. But it is also possible to maintain the fluid-dynamical conditions by increasing the temperature difference of solids that comes out from the boiler. In the last column of Figure 2.23, Gs has been kept similar than in the 30% O2 case, but solids have been cooled down to 620ºC. Hence, the contribution of EHE represents 35% of the overall fuel input, as high as the case of high Gs. In these runs, geometry has been kept constant, but Gs changes by varying fuel input, up to 730 MW. From the graph it is found that the possibilities of proper regulation of the fluidized bed operation fall on the external device for cooling down the particles that are coming back into the boiler. It is possible to recycle larger amount of solids, or to cool them down to lower recycled temperatures. In both cases, an important effort must be dedicated to a proper external heat exchanger design. Conventionally, EHE devices for CFB are bubbling fluidized beds or moving fluidized beds. Air Oxy30% Oxy40% Oxy40% RateLoopSeal 21.83% 24.61% 37.73% 40.80% RateFlueGas 35.82% 31.94% 25.39% 23.98% RateWalls 42.36% 43.46% 36.68% 35.22% Gs(kg/m2s) 11.0 11.2 20.7 10.6 Fuelinput(MW) 570 580 850 730 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Heattransfershareperzone Trecirc=720ºC Trecirc=620ºC Chapter 2. Large oxy-fuel circulating fluidized bed boiler modeling 79 2.6 CONCLUSIONS Two of the models of large-scale oxy-fuel CFB available in open literature have been developed by boiler companies, with long experience on air-firing large CFB. These models are presumably complex, and based on computational fluid-dynamics modeling, although modeling data are not public. In spite of this lack of details, the one-dimensional modeling developed up to here is able to predict the oxy-fuel boiler heat balance with similar results as those previously mentioned. The heat balance has been then compared using the heat ratios, dividing the overall energy balance into three shares: the energy removed by the walls, the flue gas sensible heat through the convective pass, and the heat removed in the external heat exchanger, the three of them, represented by unit of fuel input. The values obtained from the comparison made the model fairly reliable to be used for further analysis. The effect of changing O2 concentration on the design of certain boiler geometry was evaluated. Fixing the fuel input and the temperature at which solids are recycled, three different O2 concentrations were run. The results indicated that a much compact boiler is required for the case of 60% O2. For this case, the boiler cross sectional area was 33% less than the case of 30% O2. This implied a reduction in the available surface for heat removal in the water walls. Consequently, the heat transferred in the EHE increased twice from the 30% O2, to the 60% O2 cases. When boiler geometry was not variable, fuel input did increase for higher O2 percentages, to reach the proper fluidizing velocity. Again, this led to higher heat removal requirements in the EHE up to 53% from the overall combustion heat. Oxy-fuel applied in the CFB boiler presents outstanding operating flexibility, for a range of O2 at inlet, analogous to the conventional case in which boiler load must change for adapting the energy market requirements. On the other hand, the EHE is now an essential device for acquiring the mentioned flexibility and it will not be possible to disregard it in future oxy-fuel CFB boilers designs. There will be then two essential differences in the operation between EHE attached to an oxy-fuel and an air-fired CFB. On the one hand, the fluidizing gas composition cannot be air, but it will consist of flue gases bleeding, in order to preserve the purity of the CO2 concentrated flue gas, within acceptable limits for 2.6. Conclusions 80 transport and storage. On the other hand, higher recycled flow of solids will lead to non-uniformities along the tubes in the EHE. Consequently, the average heat transfer coefficient will be affected. Following, both issues will be experimentally assessed and discussed. 81 CHAPTER 3 HEAT TRANSFER IN CIRCE EXPERIMENTAL OXY-FUEL COMBUSTION BUBBLING FLUIDIZED BED 3.1 INTRODUCTION The particular relevance of higher heat removal in the EHE of an oxy-fuel CFB leads the attention towards an adequate design of the heat transfer areas and the allocation of this equipment. EHEs work usually under bubbling regime conditions. This allows the proper displacement of solids from the bottom of the cyclone leg, to the recycling pipe. At the same time, low velocity fluidization regime prevents from excess of gas flow through the dip-leg solids column (Grace et al., 1997). The bubbling regime exhibits an intense heat transfer from the particles to the heat tubes, due to the dominant contribution of the particle convection mechanism. Unlike in conventional EHE, the oxy-fuel CFB’s EHE will not be fluidized with air, but with RFG. Only one previous experience evaluated heat transfer coefficients under oxy-fuel fluidized bed conditions (Nsakala et al., 2004). They measured the heat transfer coefficients in the riser of a CFB, finding no significant differences with the ones measured under air-firing. In the CFB riser, the radiation contribution and gas convection prevails over the particle convection mechanism, unlike in dense beds. Thus, presumably, most researchers assume that the change on the gaseous atmosphere would not mean a significant difference on heat transfer. However, the change on gas composition would affect the conductivity and heat capacity of the gas layer adjacent to the wall. This gas layer is, in fact, the limiting actor to the heat flow. The fluid-dynamics is also assumed to behave similarly in OF than in AF mode by most of researchers, although the bubbles were 3.2. Heat transfer in fluidized beds 82 found to be slightly smaller and with more vigorous bubbling than in air-firing case (Guedea et al., 2011). This affects the fraction of time during which particles are in contact with the surface. Heat transfer in an oxy-fuel bubbling fluidized bed (BFB) is measured in this work. This will allow evaluating the differences with the known air-combustion operation. The two main approaches used for the prediction of heat transfer coefficients in fluidized beds, the empirical and the mechanistic approaches, are examined to forecast the differences between air and oxy-fuel operation. The heat transfer coefficients measured during the pilot plant operation will allow proposing a complete expression for the heat transfer coefficient during oxy-fuel combustion in fluidized beds. 3.2 HEAT TRANSFER IN FLUIDIZED BEDS 3.2.1 PREVIOUS FINDINGS Heat transfer between a surface and a gas flowing through a bed of particles is many times higher than expected for the same flow in an empty column. Thus, the heat transferred through particles in movement is rather more intense than the gas convection alone. Heat from hot particles to the cold inlet fluidization gas is transferred intensively. Thus, bed and gas temperatures are generally considered equal to the average bed temperature, even in the vicinity of the distributor. Researchers confirmed this phenomenon and it was mostly represented by means of empirical expressions. Compendium of correlations of heat transfer between gas and solid particles in fluidized beds can be found in general literature of fluidized beds like (Botterill, 1989; Howard, 1989; Kunii and Levenspiel, 1991; Oka, 2003; Basu, 2006). Although these quantification is particularly relevant for the modeling of fuel particle conversion, in this work the heat transfer from the particles to the gas will be considered high enough to be neglected. The mechanisms involved in the heat transfer between the bed and the surface include: ‐ The solid conduction that takes place while solid particles contact the surface. The emulsion phase contains the majority of solid particles, but Chapter 3. Heat transfer in CIRCE experimental oxy-fuel combustion bubbling fluidized bed 83 there is also a small proportion in the bubble phase. Although it is based on solids conduction, this mechanism is known as particle convection, because the movement of solids greatly enhances the energy transfer, hpc. ‐ The gas convection from both, the bubble phase and the gas found in the emulsion, hgc. ‐ The radiation heat transfer from the layer adjacent to the wall, hrad. As an engineering approximation the contribution of these mechanisms are regarded as independent, from each other, and the heat transfer coefficient is then written as: rpcgc hhhh    Eq. 3.1 The particle convection heat transfer mechanism prevails in fluidized bed region for bubbling fluidized beds and in the freeboard, radiation mechanism dominates. Thus, heat transfer coefficients to immersed surfaces in the fluidized region can range from 250 to 700 W/m2K and in the freeboard these coefficients are less than 100 W/m2K (Breitholtz et al., 2001; Oka, 2003). The transition in the heat transfer values in the splash zone was measured by Pidwerbecki and Wely (1995). Moreover, radiation heat transfer is often neglected when the bed temperature is lower than 600ºC. Numerous authors analyzed the parameters influencing heat transfer at low temperature, isolating the radiation contribution. In most of cases, the cooling rate of a heated surface (van Heerden et al., 1953; Ziegler et al., 1964; Schmidt and Renz, 2005; Patil et al., 2006) or sample probe (Grewal and Saxena, 1981; Mathur et al., 1986; Sunderesan and Clark, 1995) is used to measure the energy transfer. Since the earliest studies it was stated that the greatest influence on heat transfer intensity is that of the fluidization velocity and particle size: On the one hand, fluidizing velocity influence is usually described in three stages: ‐ At fixed bed regime, heat is transferred by conduction of particles in contact with the surface and by convective gas flow between bed particles. With onset of fluidization, particle convection heat transfer starts. At minimum fluidization velocity, heat transfer increases sharply. ‐ With further increase in fluidization velocity, heat transfer coefficient continues to increase, and it reaches a maximum at an optimal fluidization velocity. Below this limit, the major contributor to heat transfer is the particle convection, increasing with particle mixing. 3.2. Heat transfer in fluidized beds 84 ‐ With velocities higher than the optimal velocity, the key mechanism of heat transfer is gas convection. Since bed density is lower, the influence of particle convection decreases. Particle size, on the other side, determines the relative influence of each mechanism. With increasing particle diameter, optimal fluidization velocity increases and the maximum heat transfer coefficient is significantly smaller. In fluidized beds of small particles less than 0.1 mm, particle convection makes for more than 90% of heat transfer, while in beds with particles greater than 1 mm, it makes only 20% of particle heat transfer. A detailed review of the influence of particle diameter on heat transfer can be found in Decker and Glicksman (1983). The role of gas convection becomes significant when the gas flow occurs at particle diameters greater than about 800 μm. For particles of this size and larger, the convective component of heat transfer by gas becomes significant and total heat transfer coefficient does not depend much on the particle size. Additionally, the solids density is closely related to both, the fluidizing velocity and the particle size. Breitholtz et al. (2001) gathered experimental data from pilot and large plants, and established a simple correlation that directly related the global heat transfer coefficient with the solids suspension density. Their study focused on circulating fluidized bed boilers, but they highlighted the adequacy of bed density for satisfactorily predict the heat transfer coefficients in a generalized way. Empiricalapproaches One of the most widely recommended expressions for estimating heat transfer in fluidized beds is the one proposed by Zabrodsky (1966). Due to simplicity and also, the reasonable predictions given, it has been extensively used along decades (Botterill, 1989; Howard, 1989; Basu, 2006). 0.36 p 0.6 g 0.2 g s max dk ρ ρ 37.6h           Eq. 3.2 Note that the energy units for this expression were not S.I. but kcal/h. Botterill et al. (1984) recommended that a value of about 70% of that predicted by Eq. 3.1 represents a conservative estimate for hmax. Earlier common approaches for modeling heat transfer to surfaces in fluidized beds assumed similarity to gaseous convection and assigned thermal resistance to a Chapter 3. Heat transfer in CIRCE experimental oxy-fuel combustion bubbling fluidized bed 85 boundary layer at the heat transfer surface. The enhancement found at gas velocities greater than umf is attributed to the decrease of effectiveness of the film thickness (Wagialla et al., 1990). Models following this approach attempted to correlate a Nusselt number with the Prandtl number, a modified Reynolds number and the Archimedes number, using either the particle diameter or the tube diameter and other non-dimensional parameters. Hence, the influence of factors related to operation conditions, bed geometry or solids and gas properties were correlated. Table 3.1 gathers some of these empirical expressions for the heat transfer from bed to vertical surfaces. Heat transfer to the walls 25.0 65.0 03.025.0 8.0 1 Re55.                                   f f t f p t gg pp d H d d C C 0Nu     Dow and Jakob (1951)    0.18 g smf g p 0.5 ρ ρ1 C C B0.58PrNu                   36.0 45.0 Re van Heerden et al. (1953) 3.0 6. PrRe0Nu  Levenspiel and Walton (1954) 36.0 2.0 2 4.0 76.0                           f mf p f gg pp 0.4 H H gd u C C Re0.16PrNu    Wen and Leva (1956) Table 3.1. Empirical expressions for the prediction of heat transfer coefficient from fluidized bed to vertical walls The Dow and Jacob correlation (1951) was deduced under the assumption that there were not only a boundary layer of gas participating in the heat transfer, but also a boundary layer of particles moving along the wall and hindering the heat transfer. Van Heerden et al. (1953) compared the wall with the film of laminar flow along a vessel wall containing well-stirred liquid. Levenspiel and Walton (1954) supposed an artificial structure for the fluidized bed and assumed that particles were arranged in horizontal layers. So, the boundary layer was disturbed in its plane, forming intervals between layers. Wen and Leva (1956) proposed to include 3.2. Heat transfer in fluidized beds 86 particle velocity in the heat transfer relationships, since they observed that particles had scoring action in the boundary layer. Although authors agreed in the importance of fluid-dynamics on heat transfer, the expressions failed to predict the tendency with Reynolds number as shown in Figure 3.1 Figure 3.1. Effect of Reynolds on Nusselt number prediction by different authors The increase of Nusselt number with Reynolds is not so for the whole range of Reynolds numbers. In fact, the behavior of heat transfer with Re, follows analogous curve as the heat transfer coefficient with fluidizing gas velocity. An example of experimental results obtained by Mathur et al. (1986) is shown in Figure 3.2. The well known optimum heat transfer point marked the difference between a fully dominant heat transfer by particle convection, and an increasing relevance of gas convection, due to increasing on excess gas velocity. In the secondary axle, the solids volumetric fraction is also represented. Figure 3.2. Dependence of Nusselt number with Reynolds number for vertical tubes immersed in 500 μm silica sand fluidized bed (Mathur et al., 1986) 12345678910 0 2 4 6 8 10 12 14 16 Re Nu Dow & Jacob (1951)Dow & Jacob (1951) Levenspiel & Walton (1954)Levenspiel & Walton (1954) van Heerden (1951)van Heerden (1951) Wen & Leva (1956)Wen & Leva (1956) 0.6 0.55 0.5 1‐ε 6.5 5.5 4.5 Nu Re Chapter 3. Heat transfer in CIRCE experimental oxy-fuel combustion bubbling fluidized bed 93 Figure 3.4. Predicted residence time dependence on fluidizing velocity, by different authors The bath shape of Zarghami et al. approach appears a realistic prediction of the effect of velocity in the residence time of particles. This would agree with the velocity influence on heat transfer, commented above, in which there would be an optimal velocity at which residence time were lowest and the packet presence at the wall had the major importance. Radiationheattransfer Radiation heat transfer mechanism begins to contribute appreciably to the overall heat transfer in bubbling fluidized beds at temperature greater than 500oC. Measurements of radiant heat transfer in dense beds are scarce. Yamada et al. (2001) affirmed that the reason was that conclusion from models are contradictory among authors. In their experiments they showed that the real contribution to this radiant heat transfer lies on the particles adjacent to the surface, and hinders the radiation heat transfer between the surface and the higher temperature particles in the depth of the bed. They measured the influence of particle diameter and fluidizing velocity on the radiant energy transfer to the surface. In the transport zone of circulating fluidized beds radiation models are proposed, and extensive research is available (Luan et al., 2000; Eriksson and Golriz, 2005; Glicksman, 2007). Basu and Konuche (1988) reported that the radiative component is 70%-90% of the total heat flux transferred from the suspension to the wall at 600ºC-885ºC. 0,5 11,5 22,5 0 0,1 0,2 0,3 u - umf (m/s)  (s) Chen (2003) Lu et al. (1993)Lu et al. (1993) Zarghami et al. (2007)Zarghami et al. (2007) 3.2. Heat transfer in fluidized beds 94 The high percentage of radiation to total heat transfer in CFBs compared with bubbling fluidized beds is attributed to relatively low convective heat transfer due to low particle concentrations (Luan et al., 1999). A simple approach to estimate radiation, used by many authors (Botterill, 1989; Wirth, 1994; Molerus et al., 1995b; Molerus et al., 1995a; Baskakov and Leckner, 1997; Basu, 2006), is to treat the radiation transfer as the exchange between opaque gray bodies separated by a nonparticipating medium. Then, the expression for radiation exchange yields as previously stated in Eq. 2.25. Where, for two parallel planes much larger than the distance between them, Eq. 2.26 showed the calculation of the bed-to-wall emissivity. In that expression, the term of bed emissivity εb is difficult to accurately estimate. It appears that it exceeds the value of particle emissivity, εp, because of reflections from particles near the bed surface. This was attributed to the same reason that an irregular surface has a higher emissivity than a smooth surface of the same material. Effective emissivity of fluidized bed has been correlated against different bed and wall temperatures and shown in Botterill (1989). A more simple expression was, however proposed by Baskakov and Leckner (1997) yielding: 0.64 pb εε  Eq. 3.19 The heat transfer of the layer of particles near the wall is controlled by a thermal resistance, Rw. This resistance is determined by the thickness of the gas layer between the heat transfer surface and the layer of particles. If the value of Rw is small, particles close to the surface at surface temperature influence is low. The increase of heat transfer when increasing bed temperature is not only attributable to radiation heat transfer (Glicksman and Decker, 1982). Gas conductivity also augments, causing the corresponding increase on conduction mechanisms. They actually proposed to use an additional radiative conductivity for including the radiation heat transfer for particles adjacent to the surface. They added this conductivity term to the effective thermal conductivity and yielded: 3 9 8Tdk pr   Eq. 3.20 Chapter 3. Heat transfer in CIRCE experimental oxy-fuel combustion bubbling fluidized bed 95 3.2.2 THEORETICAL INFLUENCES OF OXY-FUEL CONDITIONS The immediate difference between the air-firing and oxy-firing modes resides on the atmosphere composition in which combustion and fluidization takes place. High concentration of CO2, instead of N2 is present in the gases. For acquiring an impression of how the gas properties would vary in both cases Figure 3.5 shows the theoretical properties of a mixture of O2/N2 (Air) and a mixture of O2/CO2 at different O2 proportions, at 850ºC. Density and heat capacity were evaluated as proportional with gas composition, while conductivity and viscosity follow the recommendation by Lindsay and Bromley (1950) and Wilke (1950), respectively: Figure 3.5. Gas properties variation for different O2 concentration at inlet: a) density; b) viscosity; c) conductivity; d) specific heat capacity In the four properties observed here, for the same O2 proportion, O2/CO2 mixture shows higher values. The greatest difference turns out in the gas density, which is 45% higher in the mixture with low O2 compared with O2/N2 mixture. While increasing O2 in the mixture, the gas density decreases 18%. Heat capacity diminishes 9 percentage points from the lowest O2 content case, to the highest, 1,12 1,14 1,16 1,18 1,20 1,22 1,24 1,26 0% 20% 40% 60% 80% 100% c p (kJ/kgK) O 2 (%w.) O 2 /N 2 0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40 0,45 0,50 0% 20% 40% 60% 80% 100% (kg/m 3 ) O 2 (%w.) O 2 /N 2 O 2 /CO 2 7,2E‐05 7,4E‐05 7,6E‐05 7,8E‐05 8,0E‐05 8,2E‐05 8,4E‐05 0% 20% 40% 60% 80% 100% k g (kW/mK) O 2 (%w.) O 2 /N 2 O 2 /CO 2 4,4E‐05 4,5E‐05 4,6E‐05 4,7E‐05 4,8E‐05 4,9E‐05 5,0E‐05 5,1E‐05 5,2E‐05 0% 20% 40% 60% 80% 100% (kg/ms) O 2 (%w.) O 2 /N 2 O 2 /CO 2 3.2. Heat transfer in fluidized beds 96 approaching the values in O2/N2 case. The gas conductivity value for low O2 percentages is only 2% higher than the air case, but it increases up to 13% with respect to the O2/N2 mixture. The similar ascendant tendency is found in the gas viscosity, increasing 11% from the lowest O2 ratio case to the highest. The higher density of O2/CO2 means an increase on fluidizing gas mass flow rate, for certain fluidizing velocity. The fluidizing velocity is settled by the desired fluidization regime, the boiler geometry and, particularly, by the stoichiometric conditions. This means that for a certain fuel input in a fluidized bed, fluidizing gas velocity decreases. Apart from the increase in gas density, this occurs predominantly because the higher the O2 concentration at inlet, the less flow rate of gas feeds the combustion. This variation of fluidizing velocity is represented in Figure 3.6. Figure 3.6. a) Variation of fluidizing velocity, uf, with O2 content in the gas mixture (constant molar rate of O2) and b) the variation of the total gas molar rate The potential shape in Figure 3.6a) illustrates the inverse relationship between the O2 percentage and the fluidizing velocity, due to the stoichiometry, as shown in Figure 3.6b). The influence of gas viscosity is closely related to fluidization, since it represents the resistance to the movement of particles in the fluid. These terms are included in the Reynolds number (ufdpg/), the ratio between inertial and viscous forces. This non-dimensional variable is the key parameter when correlating the heat transfer by fluid convection, since it was for conceiving the laminar, turbulent or transitional regime, at which the stream is flowing. Slightly modified Reynolds 0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0 8,0 0% 20% 40% 60% 80% 100% n g 10 ‐4 (kmol/s) O 2 (%w.) O 2 /N 2 O 2 /CO 2 0,00 0,50 1,00 1,50 2,00 2,50 0% 20% 40% 60% 80% 100% u f (m/s) O 2 (%w.) O 2 /N 2 O 2 /CO 2 a) b) Chapter 3. Heat transfer in CIRCE experimental oxy-fuel combustion bubbling fluidized bed 97 number, is used, referred to the particle diameter instead to any longitudinal parameter. Archimedes number (dp3g(p-g)/2) also comprises the effect of particle diameter and solids and gas densities on the fluidization regime (Rabinovich and Kalman, 2011; Shaul et al., 2012). The term of Archimedes number has been usually related to the contribution of gas convection heat transfer component, like the expressions used by Kunii and Levenspiel (1991), Xavier and Davidson (1985) or by Botterill et al. (1989). In circulating fluidized beds the influence of Archimedes number is more evident. Wirth (1995) established a direct relation between the Archimedes number and the influence of the gas conductivity in the component of heat transfer by gas. He stated that, for high Archimedes numbers, the conductivity of gas did not influence the heat transfer coefficients, unlike for low Archimedes numbers that based the gas heat transfer on the gas conduction. Archimedes number is also related to the void fraction of the bed. Void fraction at minimum fluidization velocity, together with the Archimedes number, are the most influencing parameters for determining the bed porosity, according to Di Natale et al. (2008). Thermal properties of gas, such as conductivity and specific heat capacity, are higher for O2/CO2 mixtures, diminishing with the O2 content in the case of cp. Prandtl number (cp/kg) includes both variables and reflects the predominant influence of either convection or conduction heat transfer. This is also a common parameter participating in forced internal convection correlations. Prandtl varies slightly for certain gas composition and its value is usually around 0.7 for gaseous phase. The influence of gas composition at inlet is now reviewed, by means of nondimensional parameters included in Table 3.1. With this aim, Figure 3.7 shows the influence of gas composition and temperature on the densities ratio, (1-εmf)ρs/ρg, the Archimedes and Prandtl numbers, respectively, at three different temperatures 750ºC, 850ºC and 950ºC. These three temperature values approach the lower, average and upper limit of temperature in a fluidized bed boiler. 3.2. Heat transfer in fluidized beds 98 Figure 3.7. Variation of non-dimensional parameters with O2 content in the gas mixture a) densities ratio, b) Archimedes number, c) Prandtl number 020 40 60 80 100 2500 2950 3400 3850 4300 4750 O 2 (%w.) (1- mf )  p / g 950ºC 850ºC850ºC 750ºC750ºC O 2 /N 2 O 2 /CO 2 020 40 60 80 100 300 400 500 600 700 800 900 O 2 (%w.) Ar 950ºC 850ºC850ºC 750ºC750ºC O 2 /N 2 O 2 /CO 2 020 40 60 80 100 0,70 0,71 0,72 0,73 0,74 0,75 O2 (%w.) Pr 950ºC 850ºC850ºC 750ºC750ºC O 2 /N 2 O 2 /CO 2 a) b) c) Chapter 3. Heat transfer in CIRCE experimental oxy-fuel combustion bubbling fluidized bed 99 Regarding Figure 3.7a) the term of minimum fluidization voidage, εmf, was calculated using Ergun and Orning (1949) and Wen and Yu (1966) correlations. The densities ratio is 35% lower in O2/CO2 mixture for O2 content similar to air. While increasing O2, this ratio reaches values still 13% lower than the O2/N2 mixture ones. This parameter diminishes with temperature. This tendency is particular of this parameter, unlike the other non-dimensional parameters, that increases with O2 content. Archimedes number is 33% higher in O2/CO2 mixture than O2/N2 mixture, for the same O2 percentage (Figure 3.7b)). It diminishes with the O2 content, down to values close to those of air. This decrease on Archimedes number, given a nonchanging particle diameter, is dominated by descendant viscosity of the mixture. Values of Archimedes number are higher for lower temperatures, which is again the added effect of higher gas density and lower viscosity when temperature decreases. The temperature influence on Prandtl number for O2/CO2 mixture is negligible, compared to O2/N2 mixture, Figure 3.7c). The decrement of this parameter with O2 proportion coincides both, with an increment of gas conductivity and a decrement of gas specific thermal capacity (Figure 3.5). Still, this decrement represents only 2% of the value in O2/N2 mixture. Knowing the differences on gas velocity when fluidizing with O2/CO2 or air (Figure 3.6), Reynolds number is following represented in two ways. In Figure 3.8 a) gas velocity varies with O2 content, but fuel input is constant, which means that molar rate of O2 in the stream must be constant. Figure 3.8b) represents constant fluidizing velocity, regardless of the fuel that should be fed for each O2 percentage. 3.2. Heat transfer in fluidized beds 100 Figure 3.8. Variation of Reynolds number with O2 content in the gas mixture: a) for a molar O2 rate b) for constant uf = 1 m/s Looking at the constant velocity graph, Figure 3.8b) the decrease of Reynolds with O2 is then due to the gas density and viscosity variation. Still, it remains higher than O2/N2 mixture, but approaches the air values at higher O2 contents. The Figure 3.8a) however, remarks the importance of gas composition on the fluiddynamics parameters, dominating over gas properties variation, neither by gas composition nor gas temperature. The differences in gas composition influence, predicted by the expressions in Table 3.1 are depicted in Figure 3.9. In spite of the differences on the tendencies of non dimensional parameters, the air-firing coefficients are to some degree over the ones in oxy-firing, following a descendant tendency with increasing O2. 020 40 60 80 100 1 3 5 7 9 11 O2 (%w.) Rep 950ºC 850ºC850ºC 750ºC750ºC O 2 /N 2 O 2 /CO 2 020 40 60 80 100 1 2 3 4 5 6 7 O 2 (%w.) Re p 950ºC 850ºC850ºC 750ºC750ºC O 2 /N 2 O 2 /CO 2 a) b) Chapter 3. Heat transfer in CIRCE experimental oxy-fuel combustion bubbling fluidized bed 101 Figure 3.9. Effect of O2 content in the gas mixture on Nusselt number predicted by different authors The Figure 3.10 shows the prediction by Zabrodsky (1966), Eq. 3.2, for two gas mixtures, O2/CO2 and O2/N2, at three temperatures. Since the gas density and gas conductivity varied with opposite tendencies, the influence of O2 content at inlet is almost offset. According to this, the heat transfer would be higher under oxy-firing conditions, but the influence of O2 concentration would not be relevant. The heat transfer coefficient decreases only 5% while increasing O2 in the mixture. According to this expression, the effect of temperature would decrease the heat transfer coefficient down to 8%, due to the changes on gas properties (since radiation is not included in this expression). Figure 3.10. Prediction of heat transfer coefficient by the Zabrodsky expression (Eq. 3.2) 20 40 60 80 0 2 4 6 8 10 12 14 16 O 2 (%w.) Nu Dow & Jacob (1951) Dow & Jacob (1951) Wen & Leva (1956)Wen & Leva (1956) van Heerden (1951)van Heerden (1951) Levenspiel & Walton (1954)Levenspiel & Walton (1954) O 2 /N 2 O 2 /CO 2 020 40 60 80 100 400 410 420 430 440 450 460 470 480 O 2 (%w.) h max (W/m 2 K) O 2 / N 2 950ºC950ºC 850ºC850ºC 750ºC750ºC O 2 / CO 2 3.2. Heat transfer in fluidized beds 102 Radiation contribution from the gas phase is, in many cases, neglected in dense bed. However, it must be carefully considered for the dilute zone of oxy-fuel fluidized beds, since participative gas CO2 partial pressure is now higher. The expression in Eq. 3.21 (Baukal, 2000) can be used to calculate the emissivity of a mixture of gases: εεεε OHCOg    22 Eq. 3.21 Being εf the emissivity of the mixture, and εCO2 and εH2O, the emissivity of the two compounds at the operating temperature. Δε considers the overlapping of the emission bands of H2O and CO2. Figure 3.11 shows the influence of the CO2 content in the gas for three different cases of H2O content, considering a freeboard temperature of 700ºC. The presence of H2O inhibits the mixture emissivity. The value of the gas emissivity differs up to 15% for high CO2 content in the gas. At lower gas temperature εH2O dominates over εCO2 for both air and oxy-fuel modes, at 27% O2 at inlet (Andersson and Johnsson, 2006). Figure 3.11. Emissivity of a gas mixture of CO2 and H2O From the variation of emissivity values in Figure 3.11 the contribution of radiation mechanism will be treated as in air-firing case, including only the wall and particles emissivities, and not the gas phase radiative properties. 010 20 30 40 50 60 70 8 0 0,32 0,33 0,34 0,35 0,36 0,37 0,38 CO 2 (%)  g 15% H 2 O 5% H 2 O No H 2 O Chapter 3. Heat transfer in CIRCE experimental oxy-fuel combustion bubbling fluidized bed 109 bed temperature, four cooling jackets are surrounding the bed (Figure 3.14). The lower jacket is 50 mm height and the other three, 100 mm. Water can flow through either of the jackets independently, or through none of them. Figure 3.14. Lower section of the bubbling fluidized bed. Constructive scheme. Total water flow is measured more accurately by an electromagnetic flow-meter. Additionally, every jacket includes a turbine flow-meter and a butterfly valve, allowing or not the water circulation. The purge valve opens only in the first instants of cooling, to evacuate the steam bubbles formed before stable cooling is reached. 3.3.3 EXPERIMENTS PLANNING It was previously stated that the most important parameters influencing heat transfer in fluidized beds are the fluidizing velocity, the bed temperature and particles size. This last parameter was not varied in the plant operation, and thus, it was no included in the experimental planning. The value of fluidizing velocity and bed temperature are no directly manipulated, but by means of other operational parameters, such as oxidant stream flow rate, temperature and composition; fuel input and type; and cooling. Table 3.5 intends to summarize these 3.3. Experimental set-up 110 variables limits and monitoring devices, aiming at operating inside a proper range of values for fluidizing velocity and bed temperature. The first column in Table 3.5 shows the parameters of the plant and inert matter that keep constant during the tests. These are the boiler geometry, the distributor plate arrangement, sand characteristics and its particle size distribution. The central section of Table 3.5 indicates the variables that can be manipulated during the plant operation. uf and T are then indirectly varied: ‐ The fluidizing velocity is determined by the stoichiometric oxygen related to the fuel fed, the proportion of diluents: CO2, recycled flue gas or air-N2 and the bed temperature. The temperature influences the density of gases and thus, the velocity inside the reactor. For fitting into the desired bubbling regime, velocities were kept between 0.8 and 1.6 m/s ‐ The bed temperature is uniform in the dense part of the bed. Oxygen concentration affects the temperature profile. Particularly in this installation, temperature in the freeboard will be remarkably higher under oxy-fuel that under air-firing. Keeping temperature below 1000ºC will be essential for avoiding agglomeration problems. Although fuels tested in the plant ranged from anthracite to lignite and blends with biomass, the experiments considered for the heat transfer coefficient assessment were the ones using anthracite type of coal. The other fuels exhibited lower heating values and so, lower thermal output during operation. As a consequence, the heat of combustion was not enough to keep stable cooling and heat transfer coefficients were not considered. The two types of Spanish anthracite used in the tests are shown in Table 3.3. The ranges of experiments planned and finally achieved are summarized in Table 3.4. The range of oxidant composition was varied in oxy-fuel mode up to 65% O2 concentration. Gas velocities were lower under oxy-fuel combustion, compared with air-mode. Chapter 3. Heat transfer in CIRCE experimental oxy-fuel combustion bubbling fluidized bed 111 Fuel: Anthracite 1 Anthracite 2 HHV( Kcal/kg) 5095 5350 INMEDIATE ANALYSIS (%w.) moisture 2,05 1,00 volatile matter 10,84 7,55 fixed carbon 61,11 59,90 ashes 26,00 31,55 PROXIMATE ANALYSIS (%w.) C 64,46 60,66 S 1,04 1,33 H 2,32 2,16 N 1,20 0,87 O 4,98 3,43 Table 3.3. Composition of fuels used in the tests MODE O2 inlet (% v.) T bed (C) uf (m/s) Planned Achieved Planned Achieved Planned Achieved AF N/A N/A 800-850 785-850 0,8-1 0.95 OF 25%-35% 31%-32% 800-850 - 0,8-1 - 35%-45% 36%-44% 800-850 827-855 0,8-1 0.74-0.94 45%-55% 50%-54% 800-850 - 0,8-1 - 55%-65% 57%-60% 800-850 833 0,8-1 0.86-0.98 OF-RFG 30%-35% 36-37% 800-850 796-830 0,8-1 0.84-0.85 AF N/A N/A 850-900 859-958 0,8-1 0.95 OF 25%-35% 31%-32% 850-900 877-895 0,8-1 0.84-0.98 35%-45% 36%-44% 850-900 858-964 0,8-1 0.78-0.94 45%-55% 50%-54% 850-900 889-943 0,8-1 0.82-0.98 55%-65% 57%-60% 850-900 885-918 0,8-1 0.86 OF-RFG 30%-35% 36-37% 850-900 859-887 0,8-1 0.85 AF N/A N/A 800-850 785-850 1-1,2 1.3-1.6 OF 25%-35% 31%-32% 800-850 - 1-1,2 - 35%-45% 36%-44% 800-850 827-855 1-1,2 - 45%-55% 50%-54% 800-850 - 1-1,2 - 55%-65% 57%-60% 800-850 833 1-1,2 - OF-RFG 30%-35% 36-37% 800-850 796-830 1-1,2 1.05 AF N/A N/A 850-900 859-958 1-1,2 1.3-1.6 OF 25%-35% 31%-32% 850-900 - 1-1,2 1.02-1.23 35%-45% 36%-44% 850-900 858-964 1-1,2 1.04 45%-55% 50%-54% 850-900 889-943 1-1,2 - 55%-65% 57%-60% 850-900 885-918 1-1,2 1.05 OF-RFG 30%-35% 36-37% 850-900 859-887 1-1,2 - Table 3.4. Experiments planning matrix, planned and achieved ranges Fixed Parameters Parameters that can be modified during operation Parameters that defined the plant operation Parameter Range Parameter Control device Monitoring Range Parameter Monitoring Range Bed geometry -207 mm diameter -Lower 1 m no refractory lined -Upper 2 m additional refractory lined Gas inlet flow rate -Forced-draft fan with variable-frequency drives (AF) -Valves (OF) Flow-meters 30-150 m3/h Fluidization velocity Flow-meters 0.7-1.2 m/s Inert matter Sand Comburent composition -Gas bottles mixer (OF) Gas analyzer Air, 25%- 65%O2, FGR Bed temperature Thermocouples 750-980ºC Particle size distribution 500-800 microns Thermal power Fuel endless screw with variable-frequency drives. 30-100 kW Distributor plate Flat plate with 210 orifices Fuel Type Proximate and ultimate analysis Coals, biomass, coke… Cooling -Valves opening to the cooling jackets -Aero-cooler Water thermocouples Water flow-meter 560 kW Gas inlet temperature Proportional valve bypassing the heat exchanger 35-180ºC Table 3.5. Experimental operational ranges and devices Chapter 3. Heat transfer in CIRCE experimental oxy-fuel combustion bubbling fluidized bed 113 3.4 HEAT TRANSFER COEFFICIENTS MEASUREMENT 3.4.1 MEASURING PROCESS Heat transfer coefficients are measured by an indirect way. When cooling is required for keeping a proper bed temperature, one of the cooling jackets valve is opened, at inlet, allowing water to pass through it. In the first instant of water crossing the jacket, the purge valve is opened. Water is around 40ºC when it enters the jacket, and the bed wall is near bed temperature, at more than 850ºC. Thus, during cooling start a great overpressure takes place, because of steam bubbles generation. Opening the purge valve releases this initial steam, protecting also the welding of the jackets. When water is not vaporized anymore, the purge valve is closed and the valve at outlet is opened, to close the water circuit. Heat transfer coefficient is measured at this instant. Water mass-flow measurements and water temperature are averaged for every cooling period to calculate the energy removed by the water is approximated with Eq. 3.22:   INwater;OUTwater;cooling TTcmQ   Eq. 3.22 Being m the water mass flow, c the average water thermal capacity and Twater;OUT and Twater;IN temperatures at outlet and inlet of the cooling jacket respectively. This expression assumes the approximation of considering the water calorific value, c, constant at the water average temperature. The energy transferred to the cooling jackets is analogous to the Newton Cooling Law, calculated by means of the global heat transfer coefficient UA, yielding: ΔTUAQcooling   Eq. 3.23 ΔT is the logarithmic temperature difference evaluated as: OUTwater;bed INwater;bed INwater;OUTwater; T-T T-T ln T-T ΔTln  Eq. 3.24 3.4 Heat transfer coefficients measurement 114 Bed temperature is measured at different heights in the bed. Particularly the temperature measured at the lowest position is located at 99 mm from the distributor. Then, bed temperature of the first jacket is approximated to the average between the oxidant stream temperature and the value from the first thermocouple. Neglecting the conduction through the bed wall, UA involves two main contributions, the water convection coefficient, hwater, and the fluidized bed coefficient, hbed: bedbedwaterwater Ah 1 Ah 1 UA 1 Eq. 3.25 Water circulates through the jacket at high velocity, since the jacket is only 6 mm width. Water convection heat transfer coefficient is estimated using the formula by Gnielinsky, recommended by Mills (1995), for calculating heat transfer coefficient in the water side. The water heat transfer coefficient reaches values around 2500 W/m2K. This is around an order of magnitude higher than the heat transfer coefficient in fluidized beds and, it can be neglected from Eq. 3.25. Following criteria have been established for selecting an appropriate timeframework of measurements: ‐ At least 1 minute of stable operation on water parameters: mass flow rate and temperature at outlet and inlet (avoiding bubbling phase) ‐ During this time, fed fuel must be also constant ‐ During this time, fan drivers frequency is not varied. ‐ Registered emissions delays at least 3 minutes after any change inside the combustor. With this premise, 1 minute of stability in emissions measurements is also considered after the moment of stability in the water parameters. ‐ Values for calculations are averaging over the stability range selected ‐ The number of values increases when the time framework is longer, and this will allow reducing uncertainty, as it will be later explained. A usual cooling curve is shown in the Figure 3.15. Here they are represented the water inlet and outlet temperature, the water flow rate and, in the secondary axle, the bed temperature. The frame remarks the time of operation at which values of Chapter 3. Heat transfer in CIRCE experimental oxy-fuel combustion bubbling fluidized bed 115 measurements are considered for calculations. First seconds of cooling are instable, due to bubbles formation inside the jacket Figure 3.15. Example of the beginning of the bed cooling (17/02/2011) In Table 3.6 stable measuring AF stages are gathered. Fuel input up to 60 kW was fed. Higher thermal power was not possible under AF conditions, due to limitation of inlet velocity. A wide range of bed temperatures were reached with the high calorific value fuels (Table 3.3). Every test carried out in the pilot plant begins with an AF phase. In Table 3.7 the stable points for heat transfer calculation during oxy-fuel combustion tests are collected. A wider range of fuel inputs was possible, up to 100 kW. Last column of Table 3.7 shows the proportion of O2 at inlet, from the commercial canisters. The high proportion of O2 at inlet has been scarcely experimented before. One of the reasons could be precisely, the low cooling capacity of retrofitting fluidized beds. In these experiments we reach O2 concentrations at inlet as high as 59.6%, with a fuel thermal input up to 100kW. Cooling was possible by opening the two central cooling jackets. Oxy-fuel tests with recycled flue gas (RFG) mode were also carried out, as shown in Table 3.8. The column indication the CO2 concentration from the mixture from the bottles is lower than in OF tests. The PLC had a control routine implemented to 3.4 Heat transfer coefficients measurement 116 regulate this concentration. Data from the gas analyzers indicated O2 and CO2 concentration present in flue gas that was recycled. The flow concentration of both compounds from the bottles was programmed for keeping a constant O2 concentration at combustor inlet, around 36%. Fuel MODE Q fuel (kW) T bed (C) Vin (m3/h) Exhaust gas composition (d.b.) CO2 (%v.) CO (ppm) O2 (%v.) Anthracite1 AF 60.5 785 59.2 14.2 1748 4.8 Anthracite1 AF 60.5 790 59.9 14 1245 5.2 Anthracite2 AF 43.4 817 54.0 13.5 431 3.5 Anthracite1 AF 35.6 819 57.3 16.3 2056 5.2 Anthracite2 AF 43.4 824 56.5 12.8 378 5.1 Anthracite1 AF 34.9 831 57.1 18.3 1747 4.8 Anthracite1 AF 54.8 832 61.5 14.4 696 5.1 Anthracite1 AF 54.8 832 61.5 14.4 695 5.1 Anthracite2 AF 32.2 834 53.8 14.6 1650 3.6 Anthracite2 AF 50.9 840 54.9 19.1 3564 5.4 Anthracite1 AF 30.5 840 56.9 18.5 1378 4.8 Anthracite2 AF 32.2 843 37.0 11.6 255 5.6 Anthracite2 AF 32.2 845 59.8 11.1 183 6.5 Anthracite2 AF 46.2 846 64.5 13 608 4.9 Anthracite2 AF 32.2 859 42.9 13.4 581 4.2 Anthracite2 AF 32.2 868 56.5 11.6 246 6.7 Anthracite2 AF 32.2 870 58.7 13.5 399 4.4 Anthracite2 AF 32.2 871 56.6 15.1 4029 2.6 Anthracite2 AF 49.0 907 66.9 11.7 225 6.3 Anthracite1 AF 54.8 954 54.7 14.6 490 5.0 Anthracite1 AF 54.8 958 58.9 16.1 2223 3.5 RANGE MIN 30.5 785 37.0 11.1 183 2.6 MAX 60.5 958 66.9 19.1 4029 6.7 Table 3.6. Summary of AF tests stable zones for heat transfer measurements Chapter 3. Heat transfer in CIRCE experimental oxy-fuel combustion bubbling fluidized bed 117 Fuel MODE Q fuel (kW) T bed (C) Vin (m3/h) Exhaust gas composition (d.b.) O2 inlet (%v.) CO2 (%v.) CO (ppm) O2 (%v.) Anthracite1 OF 30.5 888 50.2 72.8 246 17.4 30.9 Anthracite1 OF 30.5 889 50.1 84.6 348 7.7 31.1 Anthracite1 OF 30.7 894 52.1 87.7 2635 4.7 31.2 Anthracite2 OF 71.4 877 43.7 79.7 166 8.9 31.2 Anthracite1 OF 50.3 896 62.3 90.8 1830 4.6 31.3 Anthracite1 OF 66.1 884 57.6 90.2 3527 3.3 31.4 Anthracite2 OF 54.6 882 45.2 80.2 241 9.1 32.0 Anthracite2 OF 32.5 847 39.0 79.9 2016 3.5 36.3 Anthracite2 OF 50.3 861 48.5 78.6 288 5.5 37.0 Anthracite2 OF 51.8 831 48.5 76.7 223 6.9 37.2 Anthracite2 OF 49.0 858 48.5 76.8 224 6.8 37.3 Anthracite2 OF 64.4 827 47.7 76.5 224 7.2 37.3 Anthracite2 OF 32.2 964 42.6 79.2 5120 1.9 40.8 Anthracite1 OF 50.9 891 51.4 81.0 1310 7.2 41.2 Anthracite1 OF 50.9 893 44.2 85.0 2756 6.2 41.3 Anthracite2 OF 32.2 944 37.1 79.2 5120 2.5 41.8 Anthracite2 OF 32.6 898 39.8 78.0 306 8.2 43.0 Anthracite2 OF 54.6 884 41.6 76.8 250 9.3 43.9 Anthracite2 OF 68.3 902 44.2 80.2 380 9.2 49.9 Anthracite2 OF 72.3 943 44.8 79.6 200 7.0 50.4 Anthracite2 OF 74.4 895 45.3 79.7 375 8.5 50.5 Anthracite2 OF 77.0 889 43.4 69.9 300 13.4 50.7 Anthracite2 OF 77.0 918 38.4 67.1 319 15.8 50.9 Anthracite2 OF 78.0 894 43.6 72.4 283 13.1 53.5 Anthracite2 OF 77.0 918 48.3 77.9 227 6.2 57.5 Anthracite2 OF 99.4 833 48.4 65.5 434 18.0 57.8 Anthracite2 OF 99.4 885 40.4 75.6 5045 6.9 59.6 RANGE MIN 30.5 827 37.1 65.5 166 1.9 30.9 MAX 99.4 964 62.3 90.8 5120 18.6 59.6 Table 3.7. Summary of OF tests stable zones for heat transfer measurements 3.4 Heat transfer coefficients measurement 118 Fuel MODE Q fuel (kW) T bed (C) Vin (m3/h) Exhaust gas composition (d.b.) CO2 bottles (%v.) CO2 (%v.) CO (ppm) O2 (%v.) Anthracite2 OF+RFG 65.8 887 49.2 55.6 1332 5.7 34.7 Anthracite2 OF+RFG 65.8 864 49.3 49.2 284 7.1 30.3 Anthracite2 OF+RFG 77.0 796 46.8 48.9 218 9.0 30.4 Anthracite2 OF+RFG 77.0 830 45.6 47.9 267 9.8 30.5 RANGE MIN 50.7 796.5 26.9 41.1 218.2 5.7 30.3 MAX 77.0 886.6 49.3 55.6 2977.3 23.8 44.4 Table 3.8. Summary of OF+RFG tests stable zones for heat transfer measurements 3.4.2 HEAT TRANSFER RESULTS The results of the heat transfer coefficient estimations for the bed to the wall in every mode are then calculated and represented in Figure 3.16. Figure 3.16. Fluidized bed heat transfer coefficients measured in each jacket 0 100 200 300 400 500 600 700 h (W/m 2 K) Air-firing Oxy-firing Oxy-firing with RFG Jacket1Jacket2Jacket3Jacket4