Repositorio Institucional de Documentos

2014 20 Carlos Herce Fuente A Novel Method for Precombustion CO2 Capture in Fluidized Bed Director/es Departamento Instituto Universitario de Investigación Mixto CIRCE Cumo, Maurizio Cortés Gracia, Cristóbal Director/es Tesis Doctoral Autor Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA Departamento Director/es Autor Carlos Herce Fuente A NOVEL METHOD FOR PRECOMBUSTION CO2 CAPTURE IN FLUIDIZED BED Director/es Instituto Universitario de Investigación Mixto CIRCE Cumo, Maurizio Cortés Gracia, Cristóbal Tesis Doctoral Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA 2014 Departamento Director/es Director/es Tesis Doctoral Autor Repositorio de la Universidad de Zaragoza – Zaguan http://zaguan.unizar.es UNIVERSIDAD DE ZARAGOZA A NOVEL METHOD FOR PRE-COMBUSTION CO2 CAPTURE IN FLUIDIZED BED Carlos Herce Fuente Ph.D. Thesis January 2014 Advisors: Cristóbal Cortés Gracia Maurizio Cumo Doctoral School on Energetics XXV Cycle A NOVEL METHOD FOR PRE-COMBUSTION CO2 CAPTURE IN FLUIDIZED BED Carlos Herce Fuente Thesis submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Sapienza University of Rome, Italy University of Zaragoza, Spain ABSTRACT Enormous efforts have been realized by the international community in order to mitigate the effects of greenhouse gas (GHG) emissions in climate change. Approximately the 25% of global GHG (mainly CO2) are generated by the combustion of fossil fuels in the energy sector. Carbon capture and storage has been proposed to reduce CO2 emissions from large-scale fossil fuel power plants. Several technologies for the capture of CO2 have been developed in the last years, in three main lines: post-combustion, oxy-fuel combustion and pre-combustion. This thesis presents a novel method for pre-combustion CO2 capture in order to obtain hydrogen from coal, without GHG emissions. The main aim of this work is to develop a computational fluid dynamic (CFD) model of the steam reforming of a rich-methane syngas combined with CO2 capture by means of solids sorbents. This process is known as Sorption-Enhanced Steam Methane Reforming (SE-SMR). ABSTRACT ii SE-SMR represents a novel, energy-efficient hydrogen production route with in situ CO2 capture. This process has been studied in bubbling fluidized bed using CaO-based solid sorbents as CO2 acceptor. Two solid sorbent have been tested in laboratory, a natural sorbent (Dolomite) and synthetic sorbent (CaOCa12Al14O33). Moreover, some treatments have been developed in order to increase the carrying capacity of these sorbents. A comprehensive CFD model of the SE-SMR has been implemented. An Eulerian-Eulerian approach, combined with the Kinetic Theory of Granular Flow, has been adopted in order to simulate the hydrodynamic behavior of the bubbling fluidized bed. Kinetic models for steam methane reforming and CO2 capture have been developed, in order to include different sorbents and pre-treatments. In addition, a particle drag model has been developed in order to reduce the computational cost of the simulations at semi-industrial scale. An extensive simulation campaign has been carried out in order to validate the model at laboratory and semi-industrial scales. CFD simulations have been combined with a Taguchi Robust Design of Experiments, in order to evaluate the effect of several operational factors in the SE-SMR response. A NOVEL METHOD FOR PRE-COMBUSTION CO2 CAPTURE IN FLUIDIZED BED Carlos Herce Fuente Tesis realizada para cumplir con los requisitos del grado de Doctor Universidad de Roma “La Sapienza”, Italia Universidad de Zaragoza, España RESUMEN La comunidad internacional está realizando enormes esfuerzos para mitigar los efectos de las emisiones de gases de efecto invernadero (GEI) en el cambio climático. Aproximadamente le 25% de las emisiones globales de GEI (fundamentalmente CO2) son generados por la combustión de combustibles fósiles en el sector eléctrico. La captura y almacenamiento de CO2 se ha propuesto como una alternativa para reducir las emisiones de GEI en centrales térmicas. Numerosas tecnologías para la captura de CO2 se han desarrollado en los últimos años, fundamentalmente en tres líneas tecnológicas: postcombustión, oxicombustión y precombustión. Esta tesis presenta un nuevo método para la captura de CO2 en precombustión, produciendo hidrógeno a partir de carbón, sin emisiones de GEI. El objetivo principal de este trabajo ha sido desarrollar un modelo completo, mediante herramientas de fluido dinámica computacional (CFD), del proceso de reformado de un gas de síntesis con alto contenido en metano combinado con la captura de CO2 mediante adsorción con sorbentes sólidos regenerables. Este proceso es conocido como reformado de metano mejorado por adsorción (o SE-SMR, su acrónimo en inglés). RESUMEN iv SE-SMR representa una novedosa y eficiente energéticamente ruta para la producción de hidrógeno con captura in situ de CO2. Este proceso ha sido estudiado en un lecho fluido burbujeante, usando sorbentes sólidos de óxido de calcio como captores de CO2. Dos sorbentes sólidos han sido estudiados en laboratorio: uno natural (Dolomita) y uno sintético (CaOCa12Al14O33). Además, varios tratamientos han sido desarrollados para mejorar la capacidad de captura de estos sorbentes. Un completo modelo CFD del proceso de SE-SMR ha sido desarrollado. Una aproximación Euleriana-Euleriana ha sido combinada con la Teoría Cinética de Flujos Granulares para simular la fluidodinámica del lecho fluido burbujeante. Los reacciones químicas de reformado y carbonatación han sido implementadas en el modelo CFD. Se ha incluido un modelo detallado de captura de CO2 para simular el comportamiento de los diferentes sorbentes sometidos a diferentes pretratamientos para mejorar su rendimiento. Asimismo, un modelo de arrastre de partículas ha sido desarrollado para reducir el coste computacional de las simulaciones a escala semi-industrial. Se ha llevado a cabo una extensa campaña de simulaciones para validar el modelo a escala de laboratorio y semi-industrial. Las simulaciones CFD han sido combinadas con un Diseño de Experimentos Robusto, con el objetivo predecir y evaluar la sensibilidad del proceso SE-SMR a diversos factores operativos. xi SCOPE, AIMS AND OUTLINE OF THIS THESIS This thesis describes the computational fluid dynamic (CFD) study of the Sorption Enhanced Steam Methane Reforming (SE-SMR) of the ZECOMIX pilot plant. This project is leaded by ENEA (Italian National Agency for New Technologies, Energy, and the Sustainable Economic Development). ZECOMIX cycle is a novel method for pre-combustion CO2 capture, for producing hydrogen and electricity from coal. The first step of the process consists in the gasification with hydrogen of coal. The methane-rich syngas generated is sent to the carbonator reactor. In the carbonator reactor, SE-SMR takes place in a bubbling fluidized bed, composed by reforming catalyst and a CO2 solid sorbent. The addition of a solid sorbent for the selective removal of CO2 has a double aim. On one hand, the equilibrium of the reforming reaction is shifted beyond their conventional thermodynamic limits, increasing hydrogen production and removing the carbon on the syngas. On the other hand reaction reforming and capture are energy-balanced at lower temperature than conventional process, with an important reduction of energy demand. The hydrogen-rich syngas produced is sent to a hydrogen modified microturbine with a nominal power of 100 kWe. This thesis proposes a comprehensive low-computational cost CFD model for the simulation of the semi-industrial SE-SMR reactor of the ZECOMIX plant. Several topics related this main scope (as sorbent performance, fluidized bed simulation or scale-up of BFB) are discussed in this thesis. In Chapter 1 an introduction to carbon capture and storage (CCS) technologies is presented. Taking as starting point the effect of anthropogenic greenhouse gas (GHG) emission in climate change; and recognizing the role of fossil fuels (mainly coal) in the power supply sector; CCS plays a crucial role in the medium-term energy generation. A brief introduction to the role of coal in Italian and Spanish electrical systems is presented, with particular emphasis in the most important projects related with CCS in both countries. A review of different technologies for capture, transport and storage of CO2 is also summarized. Finally, this chapter explains in detail the ZECOMIX project, and the preliminary results to model the coal hydrogasification (more detailed about Italian Sulcis coal can be found in Appendix A). Chapter 2 gives an overview about the use solids sorbents for high temperature CO2 capture. In this work we studied two sorbents, by means of thermogravimetric (TGA) experiments, a natural sorbent (Dolomite) and a synthetic (CaO-Ca12Al14O33, mayenite) sorbent. Dolomite presents a natural decay on the CO2 carrying capacity with cycling of carbonation/calcination. SCOPE, AIMS AND OUTLINE OF THIS THESIS xii In order to improve the performance of dolomite, a novel “triggered calcination” method has been developed, with very promising results. The manufactured sorbent (mayenite) has been synthetized by means of a modification to the classical synthesis route, increasing its capacity. Some thermal treatments have been tested successfully in order to increase the performance of this sorbent. Moreover, a method for estimate from TGA data kinetic parameters and specific surface area has been developed. The characterization of solids sorbents serves to model CO2 capture reaction. In chapter 3 is presented a comprehensive model for CFD simulations of the SE-SMR process in a BFB. An Eulerian-Eulerian approach, combined with the Kinetic Theory of Granular Flow, has been adopted in order to simulate the hydrodynamic behavior of the bubbling fluidized bed. Chemical behavior includes the CO2 capture reaction (for both sorbents, treated and untreated) and steam methane reforming heterogeneous catalytic kinetics. In addition a modified drag submodel has been implemented in order to reduce the computational cost of the simulations. An extensive CFD campaign has been carried out in order to characterize the BFB hydrodynamics. The global model has been validated at laboratory scale with a good agreement compared with literature data. This model has been scale-up to the semi-industrial scale of ZECOMIX plant. In order to reduce the computational cost of this simulation, a coarse-grid simulation study has been carried out in. The studies of grid and time-step independence, as well as, a sensitive study of the model are presented in Chapter 4. CFD simulations have been combined with a Taguchi Robust Design of Experiments, in order to evaluate the effect of several operational factors in the ZECOMIX reactor response. Finally, Chapter 5 presents a summary and a general discussion on the major results. New contributions and recommendations for continuing research are also included. xiii LIST OF PUBLICATIONS A list of international JCR journal papers published during the development of this work is presented below: a) C. Herce, B. de Caprariis, S. Stendardo, N. Verdone, P. De Filippis. Comparison of global models of sub-bituminous coal devolatilization by means of thermogravimetric analysis.Journal of Thermal Analysis and Calorimetry, (2014), in press. b) S. Stendardo, L.K. Andersen, C. Herce. Self–activation and effect of regeneration conditions in CO2 – carbonate looping with CaO - Ca12Al14O33 sorbent. Chemical Engineering Journal, 220, (2013), 383-394. c) B. de Caprariis, P. De Filippis, C. Herce, N. Verdone. Double-Gaussian Distributed Activation Energy Model for Coal Devolatilization. Energy & Fuels, 26 (2012),6153 – 6159. A list of international JCR journal papers submitted or under preparation derived of the development of this work is presented below: a) C. Herce, S. Stendardo. Parametric study of CO2 capture capacity of dolomite stabilised by a novel two step calcination method. Submitted. b) C. Herce, C. Cortés, S. Stendardo. A CFD-Taguchi combined method for the design of an industrial Sorption Enhanced Steam Methane Reforming reactor. Under preparation. c) B. de Caprariis, M. Scarsella, C.Herce, N. Verdone, P. De Filippis. Double Gaussian distributed activation energy model for biomass pyrolysis. Submitted. The contributions to international conferences done during the development of the Ph.D. Thesis are presented below: a) S. Stendardo, A. Calabrò, G. Girardi, P.U. Foscolo, C. Herce A study of a chemical looping carbon capture for high-hydrogen content syngas. In Proceedings of “1st International Conference in Chemical Looping”. Lyon, March, 17-19, 2010 b) C. Herce, A. Calabrò, S. Stendardo Numerical simulation of a high temperature CO2 capture fluidized bed. In Proceedings of “Processes and Technologies for a Sustainable Energy”. Ischia,June, 27-30,2010 LIST OF PUBLICATIONS xiv c) C. Herce, R. Mecozzi, A. Calabrò. Kinetics of Sulcis coal devolatilization. In Proceedings of “5th International Conference on Clean Coal Technologies 2011”. Zaragoza, May, 8-12, 2011 d) C. Herce, S. Stendardo, R. Mecozzi, A. Calabrò, A. Di Annunzio. Experimental study of sintering on Dolomite CO2 capture efficiency. In Proceedings of “5th International Conference on Clean Coal Technologies 2011”. Zaragoza, May, 8-12, 2011 e) A. Calabrò, S. Attanasi, A. Dedola, S. Cassani, L. Pagliari, S. Stendardo C. Herce. Commissioning of ZECOMIX Pilot Plant. In Proceedings of “5th International Conference on Clean Coal Technologies 2011”. Zaragoza, May, 8-12, 2011 f) S. Stendardo, L.K. Andersen, C. Herce, A. Calabrò. Experimental investigation of synthetic solid sorbents for multi-cycling CO2 uptake. In Proceedings of “5th International Conference on Clean Coal Technologies 2011”. Zaragoza, May, 8-12, 2011. g) S. Stendardo, C. Herce, A. Calabrò. Pretreatment of synthetic sorbent for sequentially carbon dioxide capture. In Proceedings of “2nd International Conference on Energy Process Engineering” Frankfurt, June 20-22,2011 h) S. Stendardo, C. Herce, A. Calabrò. Effect of calcination temperature on cyclic CO2 capture using pretreated dispersed CaO as regenerable sorbent. “3rd IEA High Temperature Solid Looping Network Meeting” Vienna (Austria), August 30-31, 2011 i) C.Herce, Y. Li, Q. Wang, I. Guedea, L.I. Díez, C.Cortés CFD simulation of a 90 kWth oxy-fuel combustion bubbling fluidized bed reactor. In Proceedings of “International Conference on Power Engineering-13 (ICOPE 2013)” Wuhan (China), October, 24-27, 2013. Several ENEA reports on CCS have been compiled in: a) Fossil Fuels and Carbon Capture and Storage. Chapter in: Joint Research Program MSE/ENEA – Research on National Electric System. Antonino Dattola (Ed.) A. Calabrò, S. Stendardo, C. Herce, P.U. Foscolo, E. Sciubba, A. Cavaliere, M. de Joannon, S. Giammartini, G. Girardi, T. Faravelli, E. Ranzi, G. Cau, et al.ENEA Unità Communicazione. Frascati, Italy, November 2011. Previously to this work an international JCR journal paper was published, derived from the final degree project: a) J. Pallarés, A. Gil, C. Cortés, C. Herce. Numerical study of co-firing coal and Cynara Cardunculus in a 350 MWe utility boiler. Fuel Processing Technology, 90 (2009),12071213. xv INDEX ABSTRACT .................................................................................................................................. i RESUMEN .................................................................................................................................. iii SINTESI ........................................................................................................................................ v ACKNOWLEDGEMENTS ....................................................................................................... vii SCOPE, AIMS AND OUTLINE OF THIS THESIS .................................................................. xi LIST OF PUBLICATIONS ...................................................................................................... xiii INDEX ........................................................................................................................................ xv LIST OF FIGURES ................................................................................................................... xix LIST OF TABLES .................................................................................................................... xxv 1 INTRODUCTION ................................................................................................................. 1 1.1 Background on climate change mitigation ..................................................................... 1 1.2 Role of coal in power generation .................................................................................... 7 1.3 Comparison of Italian and Spanish electrical systems ................................................... 9 1.4 Carbon capture, transportation and storage .................................................................. 14 1.4.1 CO2 capture ........................................................................................................... 15 1.4.2 CO2 transportation ................................................................................................. 24 1.4.3 CO2 storage ........................................................................................................... 25 1.5 The ZECOMIX project – A novel pre-combustion cycle ............................................ 28 1.6 Hydrogasification of Sulcis coal................................................................................... 30 2 SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE ............................... 35 2.1 State-of-art of solid sorbents for high temperature CO2 capture .................................. 37 2.1.1 Natural sorbents ..................................................................................................... 37 2.1.2 Synthetic sorbents ................................................................................................. 42 2.2 Dolomite stabilized by means of a novel thermal treatment ........................................ 46 2.2.1 Characterization of Dolomite - Experiments in TGA ........................................... 47 2.2.2 Effect of first calcination: Triggered vs. Standard calcination .............................. 50 2.2.3 Effect of combined pre-treatments ........................................................................ 55 2.2.4 Estimation of kinetic parameters and specific surface area .................................. 59 2.2.5 Conclusions ........................................................................................................... 65 INDEX xvi 2.3 Self-reactivation and regeneration of Mayenite (CaO·Ca12Al14O33) ............................ 66 2.3.1 Synthesis of solid sorbent ...................................................................................... 66 2.3.2 Characterization of Mayenite – Experiments in TGA .......................................... 66 2.3.3 Self-reactivation of Mayenite ................................................................................ 67 2.3.4 Conclusions ........................................................................................................... 77 3 CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR .................................................................................................................................. 79 3.1 A brief introduction to fluidization ............................................................................... 79 3.2 CFD modeling of bubbling fluidized beds ................................................................... 87 3.2.1 Kinetic Theory of granular flows .......................................................................... 89 3.2.2 Modified drag model for coarse grid simulation ................................................... 94 3.3 Gas-solid reaction modeling ......................................................................................... 97 3.3.1 Steam Methane Reforming model ......................................................................... 97 3.3.2 Carbon dioxide capture model .............................................................................. 99 3.4 Preliminary hydrodynamic simulations ...................................................................... 105 3.4.1 Numerical method and boundary conditions ...................................................... 105 3.4.2 Cold flow simulations of an oxy-fuel reactor ...................................................... 107 3.4.3 Preliminary simulations of ZECOMIX reactor ................................................... 113 3.5 Laboratory Scale SE-SMR simulations ...................................................................... 117 3.5.1 State-of-the-art of SE-SMR process .................................................................... 117 3.5.2 Numerical simulations of a laboratory scale SE-SMR reactor ........................... 117 3.6 Summary and Conclusions ......................................................................................... 125 4 NUMERICAL SIMULATION OF THE SE-SMR REACTOR IN ZECOMIX PLANT .. 127 4.1 Scale-up of fluidized beds .......................................................................................... 127 4.2 Semi-Industrial SE-SMR simulations ........................................................................ 131 4.2.1 Grid and time step independence studies ............................................................ 131 4.2.2 Model sensitivity study ....................................................................................... 137 4.3 ZECOMIX configurations – Case studies .................................................................. 144 4.4 Taguchi Method for Design of Experiments .............................................................. 146 4.5 Results of combined CFD-Taguchi method ............................................................... 149 4.5.1 Hydrogen concentration ...................................................................................... 150 4.5.2 Methane conversion ............................................................................................ 151 4.5.3 CO and CO2 concentration .................................................................................. 153 4.5.4 CO2 capture rate .................................................................................................. 154 INDEX xvii 4.5.5 Bed porosity, bed expansion and pressure drop .................................................. 155 4.5.6 Summary of influence of parameters .................................................................. 157 4.6 Summary and conclusions .......................................................................................... 159 5 SUMMARY AND CONCLUSIONS ................................................................................ 161 5.1 Solid sorbents for high temperature CO2 capture ....................................................... 161 5.2 CFD modeling of a Sorption Enhanced - Steam Methane Reforming reactor ........... 162 5.3 Numerical simulation of the SE-SMR reactor in ZECOMIX plant ........................... 163 5.4 Sulcis coal hydrogasification ...................................................................................... 164 5.5 Perspectives for future work ....................................................................................... 165 6 CONCLUSIONES ............................................................................................................. 167 6.1 Sorbentes sólidos para la captura de CO2 a alta temperatura ..................................... 167 6.2 Modelado CFD del reactor de reformado de metano mejorado con sorción .............. 169 6.3 Simulación numérica del reactor de SESMR en la planta piloto ZECOMIX .......... 169 6.4 Hidrogasificación de carbón del Sulcis ...................................................................... 171 6.5 Perspectivas para el trabajo futuro .............................................................................. 172 A1 - Kinetics of Sulcis coal devolatilization ............................................................................. 175 A1. 1 Coal Characterization .............................................................................................. 176 A1. 2 Coal Pyrolysis Experimental Tests ......................................................................... 177 A1. 3 Comparison of global models ................................................................................. 178 A1.3.1 Single Kinetic Rate Model – Kissinger Method ................................................. 179 A1.3.2 Distributed Activation Energy Model ................................................................. 181 A1.3.3 CPD Model .......................................................................................................... 191 A1.3.4 Comparison among the different approaches ...................................................... 194 A1. 4 Conclusions ............................................................................................................. 195 NOMENCLATURE .................................................................................................................. 197 BIBLIOGRAPHY ..................................................................................................................... 203 xix LIST OF FIGURES Figure 1. 1 - Global average surface temperature(a), sea level (b) and Northern Hemisphere snow cover(c) variations [1] .................................................................................................. 1 Figure 1. 2 – Carbon dioxide concentration variations [3-7] ....................................................... 2 Figure 1. 3 – Carbon dioxide emission inventories (a) as function of sources (b) and sectors (c) [1] .......................................................................................................................................... 2 Figure 1. 4 – Stabilization levels and probability ranges of impacts that could be experienced as the world comes into equilibrium with more greenhouse gases for temperature increases [8] .......................................................................................................................................... 3 Figure 1. 5 – Current and proposed emissions trading schemes. [10] .......................................... 5 Figure 1. 6 – Emissions reduction potential by technology [20] .................................................. 7 Figure 1. 7 – World primary energy source consumptions [23] ................................................... 7 Figure 1. 8 – World distribution of energy sources in electricity generation [24] ........................ 8 Figure 1. 9 – Cost estimation of energy sources [25] .................................................................... 8 Figure 1. 10 – European coal consumption. This statistics includes all the uses of the coal, (power, industrial, etc…). Eurostat considers Brown Lignite as lignite. Black Lignite is included into hard coal production [32] ................................................................................ 9 Figure 1. 11 – Italian and Spanish power generation in 2011 by energy source [34-36] ........... 10 Figure 1. 12 – Italian and Spanish coal imports historical evolution adapted from [32] ............ 10 Figure 1. 13 – Historical evolution of Spanish coal production and imports adapted from [32] 11 Figure 1. 14 – Coal-fired power plants in Italy [40] ................................................................... 12 Figure 1. 15 – Coal-fired power plants in Spain and CCS demo projects adapted from [36, 47, 48] ........................................................................................................................................ 14 Figure 1. 16 – Overview of CO2 capture processes and systems [49] ....................................... 15 Figure 1. 17 – Routes to CO2 capture in power generation (by fuel) and industrial applications (by sector) [20] .................................................................................................................... 17 Figure 1. 18 – Scheme of post-combustion system with amines (Intercooled Absorber/Interheated stripper with 8mPZ) [52] .................................................................. 18 Figure 1. 19 – Scheme of post-combustion system with CaL adapted from [72] ....................... 19 Figure 1. 20 – Scheme of a coal oxy-fuel power plant [73] ........................................................ 20 Figure 1. 21 – Scheme of a CLC system ..................................................................................... 22 Figure 1. 22 – Scheme of a Quench-IGCC-slurry with capture of CO2 [85] .............................. 23 Figure 1. 23 – Cost of CO2 transportation (each 250km) as function of diameter pipeline and pressure [49]. ....................................................................................................................... 25 Figure 1. 24 – Methods for storing CO2 in deep underground geological formations [49] ....... 26 Figure 1. 25 – Large-scale CO2 capture projects in operation, under construction or at an advanced stage of planning as of end-2012, by sector, storage type, capture potential and actual or estimated start date [20] ....................................................................................... 27 Figure 1. 26 – Scheme of ZECOMIX cycle ................................................................................ 28 Figure 1. 27 – Layout of ZECOMIX pilot plant. In red is the carbonator reactor. ..................... 30 Figure 1. 28 – Coal conversion at P=30bar for different temperatures and hydrogen flow rate (a) and methane evolution at different pressures at T=600ºC (b) during hydrogasification equilibrium simulation ........................................................................................................ 32 Figure 1. 29 – Experimental values of mass loss during coal devolatilization at 100 K min-1 and the predicted ones assuming different global pyrolysis models. .................................. 33 LIST OF FIGURES xx Figure 1. 30 – The distribution activation energy curves as a function of activation energy, in red the curve for 1-DAEM in black for 2-DAEM (left). Comparison between experimental reaction rates and reaction rates obtained with the 2-DAEM kinetic parameters for a heating rate of 100K/min (right) ......................................................................................... 34 Figure 1. 31 – Experimental coal conversion as function of pressure (left) and methane evolution under N2 and H2 atmospheres [116, 118] ............................................................ 34 Figure 2. 1 - The equilibrium pressure of CO2 on CaO. ............................................................. 37 Figure 2. 2 - The decay in maximum carbonation conversion (X) with the number of cycles [160]. Equation (7) corresponds to Equation 2.4 in this work ............................................ 38 Figure 2. 3 – Schematic representation of thermal treatments previous to the natural decay of solids sorbents with cycling from [164] .............................................................................. 39 Figure 2. 4 – Limestone behavior by a) Pre-calcination and b) Self-activation methods from [143] .................................................................................................................................... 40 Figure 2. 5 – Steam reactivation effect on sorbent activity during carbonation in the TGA−sorbent calcined/sintered at 1100 °C for 24 h and hydrated by steam from [148] ... 40 Figure 2. 6 – Results of long-term calcination/carbonation cycling with Strassburg limestone without thermal pretreatment, compared with pretreatment at 1000 °C for 6 h and at 1000 °C for 24 h with two extended periods of carbonation. Calcination/ carbonation cycling was at 850 °C in all cases from [139].................................................................................. 41 Figure 2. 7 – a) Example of the increase in CO2 carrying capacity experienced by two particles cycling through the system (black dots, after 15 carbonation calcination cycles and white dots after 100 cycles). b) Evolution of the CO2 carrying capacity of CaO with the number of cycles (black dots with recarbonation and white dots without recarbonation) from [168]. ............................................................................................................................................. 41 Figure 2. 8 – Schematic representation of the main synthetic sorbents and their carrying capacity and operational temperature window, from [170] ................................................ 42 Figure 2. 9 – Structure of natural hydrotalcite, from [173] ......................................................... 43 Figure 2. 10 - Thermal experiment programming ....................................................................... 47 Figure 2. 11 – Comparison between experimental TGA mass loss under standard calcination and trigger calcination. Temperature history is also reported. ............................................ 50 Figure 2. 12 - CO2 uptake at heating rate of 10°C/min for different calcination methods over 15 cycles. .................................................................................................................................. 51 Figure 2. 13 – CO2 uptake at different heating rates over 15 cycles ........................................... 52 Figure 2. 14 – Uptake of pre-treated dolomite over 50 cycles for different CO2 concentrations ............................................................................................................................................. 53 Figure 2. 15 – CO2 uptake of pre-treated dolomite at 100°C/min heating rate over 50 cycles for different carbonation time ................................................................................................... 54 Figure 2. 16 – CO2 uptake of different calcination methods dolomite over 150 cycles for different CO2 concentration calcination and carbonation ................................................... 55 Figure 2. 17 – CO2 uptake under different carbonation times a) Long carbonation times for standard and trigger calcination b) Comparison of effect of carbonation time for trigger calcined samples .................................................................................................................. 56 Figure 2. 18 – CO2 uptake under different first calcination times for triggered calcined dolomite ............................................................................................................................... 57 Figure 2. 19 – CO2 uptake under different thermal treatments (long time and high temperature) for trigger calcined dolomite and standard pure nitrogen calcination. ................................ 58 Figure 2. 20 – CO2 uptake under periodic trigger calcinations ................................................... 59 1 1 INTRODUCTION 1.1 Background on climate change mitigation In 1992, international concern about climate change led to the United Nations Framework Convention on Climate Change (UNFCCC). The ultimate objective of that Convention is the “stabilization of greenhouse gas concentrations in the atmosphere at a level that prevents dangerous anthropogenic interference with the climate system”. According with the Intergovernmental Panel on Climate Chang (IPCC) Fourth Assessment Report: “Warming of the climate system is unequivocal, as is now evident from observations of increases in global average air and ocean temperatures, widespread melting of snow and ice and rising global average sea level. […] Most of the observed increase in global average temperatures since the mid-20th century is very likely due to the observed increase in anthropogenic greenhouse gas concentrations” [1]. Figure 1. 1 - Global average surface temperature(a), sea level (b) and Northern Hemisphere snow cover(c) variations [1] Since Arrhenius demonstrated in the end of XIX century the correlation between CO2 concentration and temperature in the atmosphere [2], the anthropogenic effect on climate has INTRODUCTION 2 been a very controversial topic. However, nowadays, it seems to be globally accepted that there is a direct relation between greenhouse gas (GHG) emissions and climate change. As shown in Figure 1. 2 the CO2 concentration in the last 400,000 years, changed periodically between 180 and 300 ppm. However, since the Industrial Revolution, the rise of concentration has been very fast, arriving in a century to 380 ppm. Figure 1. 2 – Carbon dioxide concentration variations [3-7] This increase of temperatures is mainly associated to the GHG emissions, and in particular due to CO2 emissions from fossil fuels combustion (56.6%). Other anthropogenic emission sources must be considered as CO2 increase due to deforestation and decay of biomass, and emission of GHG in a lower concentration but more active in terms of GHG potential as methane (1 ton of CH4 corresponds to 21 ton of CO2) and nitrogen oxides (1 to 310 equivalent ton of CO2). As shown in Figure 1. 3, the CO2 emissions are due to several sectors as transport, residential and commercial buildings, industry (mainly cement, steel and iron, and refining processes), agriculture, forestry, and, mainly to energy supply. Figure 1. 3 – Carbon dioxide emission inventories (a) as function of sources (b) and sectors (c) [1] 1.1 Background on climate change mitigation 3 Particularly, the estimation of temperature increase for the XXI century varies from 1.1 to 2.9ºC in the more optimistic scenarios, to 2.4 to 6.4 ºC in the most pessimistic scenario (Figure 1. 4). The UNFCC estimated that an increase higher than 2ºC (corresponding to a CO2 concentration of 450 ppm), respect to 1980-1999 period, should have dramatic and irreversible impacts on systems and sectors as water (increased water availability in moist tropics and high latitudes; decreasing water availability and increasing drought in mid-latitudes and semi-arid low latitudes; hundreds of millions of people exposed to increased water stress), ecosystems (Increasing species range shifts and wildfire risk; ecosystem changes due to weakening of the meridional overturning circulation; significant (more than 40%) extinctions around the globe), food ( complex, localized negative impacts on small holders, subsistence farmers and fishers; productivity of all cereals decreases), coasts (Increased damage from floods and storms; about 30% of global coastal wetlands lost; millions more people could experience coastal flooding each year), and health (increasing burden from malnutrition, diarrheal, cardio-respiratory and infectious diseases; increased morbidity and mortality from heat waves, floods and droughts; changed distribution of some disease vectors; substantial burden on health services). Figure 1. 4 – Stabilization levels and probability ranges of impacts that could be experienced as the world comes into equilibrium with more greenhouse gases for temperature increases [8] INTRODUCTION 4 In order to mitigate the effects of GHG in climate changes, the UNFCC promoted the Kyoto Protocol, which commits its Parties by setting internationally binding emission reduction targets [9]. Recognizing that developed countries are principally responsible for the current high levels of GHG emissions in the atmosphere as a result of more than 150 years of industrial activity, the Protocol places a heavier burden on developed nations under the principle of "common but differentiated responsibilities." The Kyoto Protocol was adopted in Kyoto, Japan, on 11 December 1997 and entered into force on 16 February 2005. The detailed rules for the implementation of the Protocol were adopted in Marrakesh, Morocco, in 2001. Its first commitment period started in 2008 and ended in 2012. Its second commitment period was adopted (but not still accepted) on 21 December 2012 in Doha, Qatar. Under the Protocol, countries must meet their targets primarily through national measures. However, the Protocol also offers them an additional means to meet their targets by way of three market-based mechanisms. The Kyoto mechanisms are: • International Emissions Trading: In which Parties have assigned a number of “assigned amount units” (AAUs) corresponding to CO2 emission rights. Emissions trading allows countries that have emission units to spare - emissions permitted to them but not "used" - to sell this excess capacity to countries that are over their targets. • Clean Development Mechanism (CDM): allows a country with an emission-reduction or emission-limitation commitment under the Kyoto Protocol to implement an emission-reduction project in developing countries. Such projects can earn saleable certified emission reduction (CERs) credits, each equivalent to one ton of CO2, which can be counted towards meeting Kyoto targets. • Joint implementation mechanism (JI): allows a country with an emission reduction or limitation commitment to earn emission reduction units (ERUs) from an emissionreduction or emission removal project in another non-under-development country, each equivalent to one ton of CO2, which can be counted towards meeting its Kyoto target. Notable achievements of the UNFCCC and its Kyoto Protocol are the establishment of a global response to the climate change problem, stimulation of an array of national policies, the creation of an international carbon market and the establishment of new institutional mechanisms that may provide the foundation for future mitigation efforts (as shows Figure 1. 5). However, it has been a very controversial project for some reasons. On the one hand, U.S., which made up 16% of global GHG emissions, has no ratified the Kyoto Protocol. On the other hand, other countries as China (17% of global GHG emissions), and India (5%) have no obligations to plan a reduction of emissions. More reasons as volatile prices of emission trade or over-estimations in the assignations plan have been criticized. For all these reasons, some of the most active countries in the development of renewable energies and in the implementation of Kyoto protocol as Canada, New Zealand or Japan, as well as Russia have no accepted second phase Kyoto commitments. In fact, in 31 December 2011, Canada became the first signatory to announce its withdrawal from the Kyoto Protocol. 1.1 Background on climate change mitigation 5 Figure 1. 5 – Current and proposed emissions trading schemes. [10] European Union has been very active since the beginning in the promotion of Kyoto Protocol. EU compromised a reduction in its emission for the period 2008-2012 of 8%, with respect to 1990 levels. In the 2003/87/CE Directive, the European Emission Trading Scheme (EU-ETS) was created in order to regulate the GHG emission market, according with the International Emissions Trading of the Kyoto Protocol. The EU-ETS is a cornerstone of the European Union's policy to combat climate change and its key tool for reducing industrial greenhouse gas emissions cost-effectively. The first - and still by far the biggest - international system for trading greenhouse gas emission allowances, the EU-ETS covers more than 11,000 power stations and industrial plants in 28 EU members plus Iceland, Norway, and Liechtenstein, as well as airlines, covering the 45% of the European GHG emission. The EU-ETS was divided in three steps: 1. Phase 1. From 1 January 2005 to 31 December 2007. This phase was considered a learning step. During this period was established the AAUs price and the emissions market, as well as the necessary infrastructure in order to catalog and control all the companies involved in the system. 2. Phase 2. From 1 January 2008 to 31 December 2012. This phase was the development of EU-ETS and it coincides with the first commitment period of the Kyoto Protocol. 3. Phase 3. From 1 January 2013 to 31 December 2013. This phase pursues the EU 20/20/20 target. 20/20/20 target set three key objectives for 2020: a 20% reduction in EU greenhouse gas emissions from 1990 levels; raising the share of EU energy consumption produced from renewable resources to 20%; a 20% improvement in the EU's energy efficiency. INTRODUCTION 6 EU is still very active in combating climate change. Additionally to the environmental point of view, the economic effects of climate change have been extensively discussed. In 2006 it was commissioned by UK government the Stern Review about climate change [8]. In this work, the main conclusion is that the benefits of strong, early action on climate change far outweigh the costs of not acting. The measurements need to mitigate climate change suppose nowadays the 1% of GDP, but in the future can achieve the 20%. The Review provides prescriptions including environmental taxes to minimize the economic and social disruptions. Moreover, it is evaluated the potential impacts of climate change on water resources, food production, health, and the environment. This work has been widely discussed with several supporters [11-14] and detractors [15, 16], but it is a very good review of scientific and economic effects of GHG emissions as a global externality of enormous proportions. According with this work, and in order to obtain an alternative to the IET proposed in the Kyoto Protocol, several approaches have been proposed, as carbon tax, etc…[17-19] Kyoto protocol is centered in reduce the CO2 emission and to associate a real price to the ton of CO2 emitted. Thus, in order to reduce the carbon intensity of the energy supply sector, and therefore the economic penalties linked with emissions, a wide range of technologies will be necessary to reduce energy related CO2 emission substantially. Figure 1. 6 shows how different technologies contribute to meeting the energy sector target of cutting CO2 emissions by more than half by 2050, in order to maintain increase of temperature in 2ºC. Energy efficiency in end use can suppose 42% of reductions. Much of this can be realized now, with net benefit. Particularly the most important sectors are buildings and transport, but all sectors are susceptible to increase the energy efficiency. The second technology to reduce GHG emission is the increase of renewable sources in power generation, 21%. Great progress already made in photovoltaic (PV) and wind. Offshore wind and concentrated solar power also have large potential. Nuclear could suppose till 8% of emissions. However, great uncertainties remain in the uses of nuclear. Instead of a majority of countries remain committed to nuclear even after the Great East Japan Earthquake and the accident in Fukushima, in March 2011, long-term nuclear uses is under discussion. End-use fuel switching can made up to the 12%. Some can be done fairly easily, some more difficult, particularly in industry. Hydrogen and biomass are the most promising alternative sources. Power generation efficiency and fuel switching could suppose a 3% of the emissions. Carbon capture and storage (CCS) can suppose a reduction of atmospheric CO2 between 14 and 22% [20, 21]. CCS is the only technology available to mitigate GHG emissions from largescale fossil fuel plants. These technologies consist in obtaining a pure stream of CO2 from fossil fuels energy processes, the CO2 obtained is storage in the subsurface, mainly in deep saline aquifers. These technologies are still struggling and several large-scale demonstrations are needed [22]. A novel technique to the capture of CO2 is the aim of this thesis, and in the next sections a deeper explanation will be provided. 1.2 Role of coal in power generation 7 Figure 1. 6 – Emissions reduction potential by technology [20] 1.2 Role of coal in power generation Approximately 69% of all CO2 emissions are energy related, and about 60% of all GHG emissions can be attributed to energy supply and energy use [1].The IEA World Energy Outlook 2009 [23] projects that, without changes in current and already planned policies, global energy-related CO2 emissions will be 57% higher in 2030 than in 2005, with coal demand increasing by 40%. In 2030, fossil fuels would remain the dominant source of energy. The bulk of the additional CO2 emissions and increased demand for energy, 84% of which will come from using fossil fuels, will come from developing countries. Figure 1. 7 – World primary energy source consumptions [23] Thus, the coal production and use is far from being in declining. Coal is the second source of primary energy in the world after oil, and the first source of electricity generation. Coal consumptions grew from 4600 million tons (Mt) in 2000 to 7200 (Mt) in 2010 (see Figure 1. 8). Growth in coal demand is strongly dependent of the country. In the OECD countries the consumption is stable; meanwhile the increasing is driven primarily by developing economies, such as China and India. INTRODUCTION 8 Figure 1. 8 – World distribution of energy sources in electricity generation [24] 90% of coal exports come from only six countries (Indonesia, Australia, Russia, South Africa, Colombia, and the United States). Hence it is logical suppose the need for further diversification in the coal supply and the increase on the autochthonous coal. For these reasons it is estimated that price of coal will increase in the next years (see Figure 1. 9). Instead of the increase on the price of coal, the volatility of the price will be lower than in the cases of oil and gas. Figure 1. 9 – Cost estimation of energy sources [25] It has been estimated that there are over 861 billion tons of proven coal reserves worldwide. This means that there is enough coal to last us around 112 years at current rates of production. In contrast, proven oil and gas reserves are equivalent to around 46 and 54 years at current production levels [26]. Coal reserves are available in almost every country worldwide, with recoverable reserves in around 70 countries, but they are concentrated in a 70% in four countries (U.S., Russia, China, India and Australia) [27]. However there is a controversy about the useful time of the reserves. The most conservative studies estimate that actual reserves are sufficient to cover the current demand for almost 200 years [28, 29]. Other studies, on the contrary, proposed that coal might not be so abundant, widely available and reliable as an energy source in the future [30, 31]. In the case of European Union, the coal reserves are concentrated in Poland and Germany, with a relative high importance of the East countries as well as UK and Spain. The EU reserves are mainly centered in lignite, and a 25% of the global consumption of coal is imported. The consumption of coal in EU is stable in the last years, as well as imports, as shows Figure 1. 10. This consumer policy of is mainly drive by Germany and Poland, which obtain lower costs on 1.3 Comparison of Italian and Spanish electrical systems 9 the coal extraction than other countries. Usually, the price of autochthonous coal is 2 or 3 times higher than imported one. However, the use of local resources allows a higher independence of the third countries; lower variability on prices; and the maintenance of local structures centered in mining, avoiding the restructuring of coalfields and the socioeconomic cost associated. Figure 1. 10 – European coal consumption. This statistics includes all the uses of the coal, (power, industrial, etc…). Eurostat considers Brown Lignite as lignite. Black Lignite is included into hard coal production [32] 1.3 Comparison of Italian and Spanish electrical systems Italian and Spain generation electrical systems present very different mixes. The Spanish electric system presents a well distributed energy sources, in line with OECD countries. Coal, nuclear and renewable (mainly eolic) suppose approximately the 60% of energy sources, in similar proportion (19.3%, 24.1% and 22.1%). The other energies considered are big hydroelectric (7.7%), co-generation and other special regime no renewable (12.7%) and natural gas combined cycle power plants (14.1%). Moreover, in the Spanish system, the imports/exports of electricity are almost balanced, with a surplus on the exports. The Italian case is strongly different. Since 1987, as consequence of the nuclear power referendum following the Chernobyl accident, Italian government decides shutdown the four operative nuclear power plants, and it stopped the construction of a fifth plant, under commissioning phase [33]. The power supply was substituted by oil and natural gas power plants. The increase of prices of the oil reduced the use of this fuel in the last decades, but its role was very important during the 1990’s. Nowadays, the natural gas is the most used energy source in the electric system (43.6%). The role of coal and renewables are significantly lower than in Spanish case. Moreover, Italian system presents a lack in electricity and it imports approximately the 13% of electricity. 0 200.000 400.000 600.000 800.000 1.000.000 1.200.000 1.400.000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Coal (Thousands of tonnes) Coal consumption in EU 25 Lignite Imports Hard coal Imports Lignite production Hard coal production INTRODUCTION 10 Figure 1. 11 – Italian and Spanish power generation in 2011 by energy source [34-36] Both countries are strong importers of coal (Figure 1. 12). In 2011, coal consumption in Italy is ~25 Mt/year, of which just ~1 Mt/year are extracted from the unique Italian coalfield located in the Sulcis basin (in Sardinia). Sulcis coal is a sub-bituminous coal with a high content in sulphur and a relative high heating value. 2/3 of the coal extracted is used in power plants and the other 1/3 have an industrial purpose [37]. The proved reserves of Sulcis coal are ~50 Mt [27]. Spanish coal consumption is ~28 Mt/year, of which just ~6 Mt/year are autochthonous. Two types of coal are extracted in Spain. In one hand, Bituminous coal and Anthracite coalfields are located in the provinces of Asturias, Leon, Palencia (~3.5 Mt/year) and Ciudad Real (~0.5 Mt/year). In the other hand, in the province of Teruel ~2 Mt/year of Black Lignite are extracted. The Spanish autochthonous is used almost exclusively as a fuel for steam-electric power generation (90%) and the reserves proved reach ~530 Mt [27, 38]. Spanish coal market have presented a decrease in the consumptions due to legislation issues, and depletion of brown lignite coalfields (as shows Figure 1. 13). Figure 1. 12 – Italian and Spanish coal imports historical evolution adapted from [32] 0 5.000 10.000 15.000 20.000 25.000 30.000 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Coal (Thousands of tonnes) Italian Imports Spanish Imports 1.4 Carbon capture, transportation and storage 17 Figure 1. 17 – Routes to CO2 capture in power generation (by fuel) and industrial applications (by sector) [20] 1.4.1.1 Post-combustion These systems normally use a liquid solvent to capture the small fraction of CO2 (typically 3– 15% by volume) present in a flue gas stream in which the main constituent is nitrogen (from air). For a modern pulverized coal (PC) power plant or a natural gas combined cycle (NGCC) power plant, post-combustion capture systems would typically employ organic physical solvents (such as Rectisol, Selexol), and amine-based chemical solvent (such as monoethanolamine (MEA) or 2n-methyldiethanolamine (MDEA)) [50, 58-61]. These two technologies are mature at industrial scale. In chemical absorption, strong bonds are created between solvent and CO2. The rupture of these bonds requires large amounts of energy. In order to reduce the energy penalties due to the regeneration of solvent, new generation solvents (as piperazine (PZ)) has been tested with better reversibility and 20% lower energy consumptions [62]. INTRODUCTION 18 Figure 1. 18 – Scheme of post-combustion system with amines (Intercooled Absorber/Interheated stripper with 8mPZ) [52] On the other hand, in physical absorption the bonds generated between solvent and CO2 are weaker than in chemical absorption. Bonding takes place a very high pressures and CO2 is released when pressure is reduced. Thus, the energy requirements are centered in driving the compressors for gas pressurization. The amount of energy per CO2 ton captured is inversely proportional to CO2 concentration. Thus, this systems are economical competitive only for concentrations higher than 15% CO2 [50]. There exist a large number of other different technologies for CO2 capture with potentially superior thermodynamic efficiency than amine-based solvents, such as ionic liquids, membranes and carbonate looping. Recently, it has been suggested the use of ionic liquids (ILs) as alternative solvents. In addition to a potentially lower energy demand in the solvent regeneration step, ILs have higher stability and easier regeneration than conventional amine-based CO2 extraction [63-65]. Concerning CO2 capture by membrane separation, both polymeric and inorganic membranes are used in order to produce clean fuel from a mixture gas. Inorganic membranes are very attractive for CO2 removal in IGCC power plants even though their costs are very high. In contrast, gas separation using polymeric membranes is today commercially available. Nevertheless, CO2 capture in large-scale power production by means of polymeric membranes still presents inadequate performances due to their lack of high-temperature stability. Research is mainly addressed for improving the performance of the membranes by increasing their selectivity and permeability and also decreasing their cost [66-68]. Calcium looping (CaL) is a CO2 capture scheme using solid CaO-based sorbents to remove CO2 from flue gases. This technique is based in the reversible gas–solid reaction between CO2 and CaO(s) to form CaCO3(s). Calcium looping is under demonstrative stage, but it presents some important advantages compared with amine-based systems. Firstly, for carbonationcalcination it uses circulating fluidized bed reactors, a mature technology at large scale. Secondly, solid adsorbents (usually called sorbents) derive from cheap, abundant and environmentally benign limestone and dolomite precursors. Finally, the efficiency penalty estimated to a power plant is ~7%.Moreover, this technique, as membranes technologies can be 1.4 Carbon capture, transportation and storage 19 applied also to pre-combustion systems. Many excellent reviews have recently been published on the current status of CaL; see for example [69-72]. Figure 1. 19 – Scheme of post-combustion system with CaL adapted from [72] 1.4.1.2 Oxy-combustion Oxy-fuel combustion systems use oxygen instead of air for combustion of the primary fuel to produce a flue gas that is mainly water vapor and CO2. This results in a flue gas with high CO2 concentrations (greater than 80% by volume). The water vapor is then removed by cooling and compressing the gas stream. Oxy-fuel combustion requires the upstream separation of oxygen from air, with a purity of 95–99% oxygen by means of an Air Separation Unit (ASU). The oxygen stream is combined with recycled flue gas (RFG) to produce an oxygen enriched gas for the oxidant. The recycle is necessary to moderate the otherwise excessively high flame temperature that would result from burning in pure oxygen. As a method of CO2 capture in boilers, oxy-fuel combustion systems are in the demonstration phase. Oxy-fuel combustion can be applied to several fuels, including coal, natural gas or blends of biomass and coal. Most interest has focused on oxy-coal combustion due to the abundance, reliability and high carbon content of the fuel. In comparison to air-fired plants, the implementation of oxy-fuel operation supposes strong plant configuration changes and additional unit operations. The optimum recycle ratio is generally 0.7; this yields oxygen levels in the oxidant environment that typically range from 25 to 30% because at these conditions, the flame and heat transfer characteristics reasonably approximate those of air-fired pulverized fuel (PF) boilers. Oxygen excess is 15–20% for air-firing conditions but for oxy-fuel conditions is INTRODUCTION 20 limited to 10% in order to minimize ASU operational costs. Flue gas oxygen content is typically 3%. In general terms, oxy-fuel combustion induces reduction of plant efficiency by 8– 12%, mainly associate to ASU. By recycling the CO2 (and possibly H2O), several changes in heat transfer can be expected due to the changes in gas properties. These changes are affected by gas radiative properties, and gas thermal capacity. Unlike symmetric diatomic gases such as N2, triatomic gases such as CO2 and H2O are not transparent to radiation. Their partial pressures are significantly higher in oxyfuel combustion flue gas than those in air-fuel combustion, and therefore, the absorptivity and emissivity of the flue gas substantially increases. Hence, the radiative heat transfer from the flame will change. In addition, carbon dioxide and water vapor have high thermal capacities compared to nitrogen. This increase in thermal capacity increases the heat transfer in the convective section of the boiler. However, the amount of gas passing through the boiler in the oxy-fuel case is lower, and increased heat transfer in the radiative section of the boiler results in lower gas temperatures entering the convective pass. Both of these factors will act to lower the heat transfer in the convective section of the boiler. The heat transfer in the radiative and convective sections of the boiler will need to be optimized to ensure efficient operation. The flue gas stream should be cooled, scrubbed and dried before being diverted for the primary recycle. Particulates are removed in order to avoid accumulation of solids in the boiler and prevent the flue gas recirculation fan and gas passages from unnecessary wear due to erosion. Moreover, flue gases must be desulphured before recirculation, in order to reduce corrosion problems. Moreover, oxy-fuel combustion provides a way to reduce emissions of NOx. In addition, NOx generation mechanism changes in oxy-fuel flames. Figure 1. 20 – Scheme of a coal oxy-fuel power plant [73] 1.4 Carbon capture, transportation and storage 21 Until recently, the obvious route for oxy-fuel combustion was via conventional pulverized coalfired (PC), and there is already one large European oxy-fuel PC demonstration plant (Vattenfall’s Schwartze Pumpe 30 MWth plant in Germany). However, recently oxy-fired fluidized bed combustion (FBC) has also become increasingly important as a potential technology offering both fuel flexibility and the possibility of firing or co-firing biomass with CO2 capture. For utility applications, circulating fluidized bed combustion (CFBC) is employed, and this technology is available in the supercritical mode at sizes of up to 460 MWe (erected by Foster-Wheeler in Lagisza, Poland). CFBC is now a widely used technology for the power industry for difficult fuels. The oxy-fuel combustion in CFB has been tested at 30 MWth,at CIUDEN. It is worth to remark the FLEXI BURN CFB project (funded by FP7 EU), with participation of CIUDEN and Foster-Wheeler, and other. This project aims to develop and demonstrate a power plant concept based on the Circulating Fluidized Bed (CFB) technology combined with CCS. The plant will be based on the super critical once through (SC-OTU) technology and oxygen-firing with carbon capture, hence, providing high efficiency, operational flexibility and potential for an almost 100% reduction in CO2. Several groups worldwide are working in oxy-fuel development, and many excellent reviews have recently been published on the current status of coal oxy-fuel combustion (e.g. [73-77]). One of the main advantages of the oxy-fuel technologies is the viability of retrofitting and repowering options. However, the energy penalties present a relative high value. In order to increase the net efficiency, some high efficient novel processes have been conceptually proposed as Graz Cycle [78], Advanced Zero Emission Power Plant (AZEP), Water Cycle (WC) [79], Solid Fuel Cell Integrated with a Gas Turbine cycle (SOFC+GT) [80] and Chemical-looping combustion. Chemical-looping combustion can be considered a variant of oxy-fuel combustion. In this technology, metal oxides are used to carry oxygen and heat between successive reaction loops, producing an inherent separation of CO2. This process is configured with two interconnected fluidized bed reactors: an air reactor and a fuel reactor (Figure 1. 21). The solid oxygen carrier is circulated between the air and fuel reactors. In CLC, the fuel is fed into the fuel reactor where it is oxidized by the lattice oxygen of the metal oxide. Complete combustion in the fuel reactor produces CO2 and water vapor. Therefore, the CO2 formed can be readily recovered by condensing water vapor. Once fuel oxidation completed the reduced metal oxide MyOX-1 is transported to the air reactor where it is reoxidized to MyOx. CLC research has mainly focused on gaseous fuels, but in the last years important work has been dedicated to adapting the process to solid fuels. Chemical-looping combustion (CLC) of solid fuels is a technology with the potential of reducing the costs and energy penalty dramatically for CO2 capture. The potential for low costs is based on the similarity to coal combustion in fluidized beds. The concept is being examined and developed on pilot scale, and shows promise for demonstration by 2020 [81-83]. INTRODUCTION 22 Figure 1. 21 – Scheme of a CLC system 1.4.1.3 Pre-combustion In the pre-combustion systems the primary fuel is processed in a gasification reactor (for solid fuels) or in a reforming reactor (for natural gas natural gas, naphtha or heavy oils) with steam and oxygen to produce a mixture consisting mainly of carbon monoxide and hydrogen (“synthesis gas”). Additional hydrogen, together with CO2, is produced by reacting the carbon monoxide with steam in a second reactor (a “shift reactor”). The resulting mixture of hydrogen and CO2 can then be separated into a CO2 gas stream (send to storage), and a stream of hydrogen. Instead of the complexity of gasification and shift reactors, compared with other capture technologies, the high concentrations of CO2 produced by the shift reactor (typically 15 to 60% by volume on a dry basis) and the high pressures often encountered in these applications are more favorable for CO2 separation. Pre-combustion would be used at power plants that employ integrated gasification combined cycle (IGCC) technology, with a higher global performance and lower costs than oxy-fuel and post-combustion techniques [49, 54, 55, 79, 84]. 1.4 Carbon capture, transportation and storage 23 Figure 1. 22 – Scheme of a Quench-IGCC-slurry with capture of CO2 [85] Hydrogen is a carbon-free energy carrier. Hydrogen is a vector capable of generating energy without producing CO2 (in combustion chambers and fuel cells), could be used to produce electricity and heat on a small scale for urban transportation, and can be stored once generated. Instead of the actual system limitations, hydrogen could play a key role in the future of energy [86]. The hydrogen turbine is the unit that is common for all pre-combustion technologies for all fuels. Pure hydrogen presents several complex challenges for flame stability due to its very high flame speed when premixed, and its high temperatures when non-premixed. The high flame temperatures resulting from hydrogen combustion are attenuated by the addition of nitrogen and/or steam. Modifications to the combustors and fuel mixing system are the principal requirements when converting a natural gas turbine to burn hydrogen-rich fuels. Although hydrogen has almost three times more energy by mass than natural gas, by volume the energy density is much lower [87]. Very important advances have been carried out in the hydrogen turbines in the last years. In 2009 the world's first industrial-scale power plant to be fully fed by hydrogen has been built in the Porto Marghera area (Italy). The experimental plant is a combined cycle in which a 12 MWe hydrogen-fuelled gas turbine is well integrated with the existing coal-fuelled plant. Furthermore, the removal of hydrogen in dehydrogenation and synthesis gas reactions with membranes has been widely studied. Two different approaches have been investigated: Integration of a membrane into the reformer and integration of a membrane into the high temperature shift reactor. Product removal may occur by H2 permeation through a Pd-alloy or composite Pd-ceramic membrane or a ceramic porous membrane [66, 67]. Many novel pre-combustion concepts or improvements have been published. Most of the most promising technologies are focused in the auto-thermal reforming (ATR) and the sorption INTRODUCTION 24 enhanced steam methane reforming process (SE-SMR). SE-SMR combines catalytic shift conversion (of carbon monoxide and steam to hydrogen and carbon dioxide) with a high temperature CO2 adsorption system using a mixture of solid catalyst and adsorbent. With this technology, the shift reactor is not necessary and simplicity and performance increase. The conversion and CO2 removal steps are carried out in a multi-bed pressure swing adsorption unit which is regenerated using low pressure steam which is subsequently condensed to leave a relatively pure CO2 stream [88]. Moreover, this process present the advantage of a high flexibility for production of H2 and electricity [89], can be combined with membranes SESMR has been studied alone or integrate with membranes [90] or fuel cells (ZEG Cycle [91]). Different configurations for the gasification of coal have been proposed in order to improve thermal and CO2 capture efficiencies, such as the HyPr-RING process [92], the LEGS process [93] or the “Calcium Looping Process” (CLP) [94]. Moreover, a negative emission cycle with biomass gasification in interconnected CFB have been demonstrated at pilot scale (AER cycle [95]). This SE-SMR method is the base of this thesis and it is explained in deep in subsequent sections. 1.4.2 CO2 transportation CO2 transportation is considered a mature technology. CO2 is regularly transported safely in pipelines across large parts of the U.S. and Canada. Natural gas contents CO2 naturally. During the transport of liquefied natural gas CO2 crystallizes, producing important operative problems. Thus CO2 is separated from natural gas and re-injected in the gas deposit. CO2 is re-injected in order to maintain the pressure in the deposit, reducing the extraction costs. Hence, CO2 is transported from and to refineries and deposits. The best conditions to transport CO2 is under supercritical conditions (P>73.8 bar and T<31.4 ºC). Under these conditions handling and transport are fast and easy. Compression of CO2 implies a cost, higher in the case of atmospheric capture techniques (e.g. CaL in postcombustion) than at pressurized systems (e.g. IGCC pre-combustion). Moreover, the CO2 must be dried and purified previously in order to avoid operational problems [51]. Some transport ways have been proposed, by train, by ship and by pipelines. Train and ship transport is proposed in order to reuse the available technologies for liquefied gas transport. However, they are more limited and expensive than pipelines [49]. In any case, the cost of pipelines is strongly dependent on diameter, as well as the length, thus a good planning is required in order to project the pipelines[86]. 1.4 Carbon capture, transportation and storage 25 Figure 1. 23 – Cost of CO2 transportation (each 250km) as function of diameter pipeline and pressure [49]. 1.4.3 CO2 storage Safe and secure CO2 storage has been demonstrated, and is still being demonstrated, at a number of sites across the world, with multi-year injections of around 1 Mt per year at a number of sites. Total CO2 storage capacity is also being proven, but will be sufficient for many years of CO2 emissions. Geological storage is cheaper and less uncertain option to CO2 storage, at medium-term, than other options as ocean storage or mineral carbonation [49]. Figure 1. 24 shows the four ways suggested by IPCC to underground geological storage: 1. Depleted oil and gas reservoirs. 2. Use of CO2 in enhanced oil and gas recovery. 3. Storage in deep saline formations (offshore or onshore). 4. Use of CO2 in enhanced coal bed methane recovery. INTRODUCTION 26 Figure 1. 24 – Methods for storing CO2 in deep underground geological formations [49] Depleted oil and gas reservoirs are prime candidates for CO2 storage for several reasons. First, the oil and gas that originally accumulated in traps did not escape demonstrating their integrity and safety. Second, the geological structure and physical properties of most oil and gas fields have been extensively studied and characterized. Third, computer models have been developed in the oil and gas industry to predict the movement, displacement behavior and trapping of hydrocarbons. Finally, some of the infrastructure and wells already in place may be used for handling CO2 storage operations. Depleted fields will not be adversely affected by CO2 (having already contained hydrocarbons) and if hydrocarbon fields are still in production, a CO2 storage scheme can be optimized to enhance oil (or gas) production. CO2 has been injected to enhance oil recovery (EOR) in wells for over three decades and has become the second largest EOR technique after steam flooding [96]. The selection of EOR technologies depends on a number of technical and economic variables including oil density and viscosity, the minimum miscibility pressure, microscopic sweep effects, and the formation of vertical and lateral heterogeneities. Thus, CO2-EOR is limited to oilfields deeper than 600 meters where a minimum of 20% to 30% of the original oil is still in place and where primary production (natural oil flood driven by the reservoir pressure) and secondary production methods (water flooding and pumping) have been applied. Few oil fields have reached this stage [50], but several projects are operating in the world as Figure 1. 25 shows. 1.6 Hydrogasification of Sulcis coal 33 In order to model devolatilization step some global models have been applied to thermogravimetric (TGA) tests in order to characterize kinetics of Sulcis coal devolatilization. The scope is to obtain, under laboratory conditions, kinetic parameters that can be applied to industrial conditions. Some models have been evaluated, and in general presents good results compared with similar coals. However, there is not only a global model capable to reproduce very accurately the complete devolatilization process, due to the important influence of secondary pyrolysis in this coal. Figure 1. 29 – Experimental values of mass loss during coal devolatilization at 100 K min-1 and the predicted ones assuming different global pyrolysis models. Thus, a novel 2-Distributed Activation Energy Model (2-DAEM) has been developed in order to model accurately the coal devolatilization. A typical approach divides in primary and secondary pyrolysis the devolatilization step. Between 400-700ºC the primary devolatilization takes place. In this phase, the weakest bridges may break to produce molecular fragments that can recombine producing tars (condensing at room pressure and temperature), or can be released as light volatile compounds (mainly CH4, C2H4, CO2, and, depending on coal sulphur content, COS and SO2). For temperatures above 700 °C the condensation of the carbon lattice with evolution of CO and H2 takes place to produce char during the secondary pyrolysis. The 2-DAEM distributes the relative weight of primary and secondary pyrolysis in the global devolatilization process, as function of the fuel composition. This method has been demonstrated to be a powerful tool to predict devolatilization kinetics in different rank coals and biomasses. 50 55 60 65 70 75 80 85 90 95 100 100 200 300 400 500 600 700 800 900 MASS / % TEMPERATURE / ºC TGA Kissinger Exact DAEM Braun-Burnham CPD INTRODUCTION 34 Figure 1. 30 – The distribution activation energy curves as a function of activation energy, in red the curve for 1-DAEM in black for 2-DAEM (left). Comparison between experimental reaction rates and reaction rates obtained with the 2-DAEM kinetic parameters for a heating rate of 100K/min (right) A more detailed description of models of Sulcis coal devolatilization can be found in Annex A. Devolatilization step is very fast compared with char hydrogasification process. Based in experimental data, intrinsic kinetics of Sulcis char hydrogasification can be obtained [116, 118]. However, instead of the efforts realized in the last years, the global hydrogasification process of Sulcis is not fully characterized, and some studies still must be carried out. Figure 1. 31 – Experimental coal conversion as function of pressure (left) and methane evolution under N2 and H2 atmospheres [116, 118] 35 2 SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE The high temperature calcium looping for CO2 capture is a promising technology to mitigate greenhouse gases emissions in industrial sectors other than power generation. The choice of a high performance sorbent is fundamental to improve the systems based on carbonation/calcination cycles. Calcium-based sorbents have demonstrated a good compromise between cost and performance. CaO-based chemical looping is being studied as a promising technological option for reducing CO2 emission into the atmosphere. One of the key advantage of this option which makes it a valuable route in gas decarbonization is the low cost and wide availability of the starting material and its high reactivity when reacts with CO2. Moreover the CaO based solid sorbent are more environmentally benign compared to other state-of-the-art solutions proposed (e.g. amine-based liquid solvents). When the CaO is converted to the calcium carbonate, the spent solid sorbent is sent back to the regeneration process where an active sorbent is regenerated for a new carbonate looping. Such a technology could be easily accomplished by means of a fluidized bed where gasification, methane steam reforming or water gas shift reaction occurs improving hydrogen production. As reported in the scientific literature [69, 127-131], when naturally occurring material as calcite or dolomite are used as CO2 acceptor in a Calcium Loop (CaL), there is a decay of reversibility. The ideal CO2 sorbent in a CaL must exhibit a number of properties: high and stable CO2 uptake capacity throughout continuous decarbonizing-regeneration cycling, fast reaction kinetics, uptake capacity and kinetics close to theoretical maximum values, and also mechanical stability and sintering resistance. In an attempt to achieve this goal, researchers have developed novel synthetic sorbents based on e.g. CaO dispersed on calcium aluminate ceramic supports. [132-135]. Considerable importance is being put on naturally occurring carbonates e.g. limestone (CaCO3) and dolomite ([Ca,Mg]CO3) as CO2 acceptor for decarbonizing reformed/shifted fuel gases or flue gases. When the material is exposed to CO2, the fresh CaO grains which compose the sorbent are converted into CaCO3. As the sorbent is completely carbonated or the gas-solid reaction shows negligible extent, calcination under atmospheric pressure and temperature greater than 850 °C allows to obtain a regenerated material for further CO2 uptake cycles. The main drawback of such a process is the marked decrease during the first few cycles with regard to CO2 uptake when used limestone or dolomite in repeated cycles of carbonation/sorbent regeneration [136, 137]. This loss of reversibility is due to the CaO sintering which happens SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 36 during the high temperature regeneration step and the pore-mouth blockage [138, 139]: albeit dolomite shows greater resistance to the sintering and pore-plugging when compared to limestone. The higher reversibility is likely due to the presence of MgO which acts as an inert binder during the carbonation of calcined dolomite inhibiting sintering of the active phase grains (CaO) and the pore closure. Due to this ability to retain pore volume through carbonation/calcinations cycles, dolomite derived sorbent is widely studied showing better results among the naturally occurring carbonates. Unfortunately, the negative features of capture decay with cycling shown by naturally occurring sorbents, counteracts the advantage of their low cost and limits their use as CO2 acceptors under industrially relevant conditions. Thus, there is a need to select sorbent material which shows no decay after repeated cycling. Generally, in order to fulfill high and stable CO2 uptake capacity two main routes are adopted by the scientific community: the synthesis of new materials [131, 134, 140] and thermal/chemical pre-treating of naturally occurring carbonates for improving their stability during cyclic operation. With regards to the synthetic materials efforts are being made to manufacture materials in an economically viable way with long life times and then slight environmental footprints [141, 142]. Besides the attempts to develop new synthetic sorbents with tailored properties there is also the option to modify naturally occurring carbonates to improve their properties. Thermal pretreating at high temperature (1000 – 1100 °C) of natural dolomite or limestone [143-145] and chemical pretreating as hydration with steam [146-148] or stabilized with metals (e.g Cu, Cr, Co, Mn) [149-151] improves the CO2 uptake capacity to some extent during cyclic operation at high temperatures. Among the available pretreatments, thermally stabilization is the most promising route as it does not lead to a great raise of carrying capacity in the pretreated when compared to the aforementioned metal-stabilized naturally occurring minerals. Although promising results, most of the thermally stabilized materials have so far exposed at regeneration in a 100% nitrogen atmosphere without taking into account the effects from CO2 on sintering. In fact, when dolomite and limestone are calcined in a CO2 containing atmosphere sintering processes can occur [152-154]. In order to enhance the contact between the solid sorbent and the CO2 and facilitate regeneration step, carbonation/calcinations cycles are accomplished in fluidized bed (FB) reactors. Instead of the FB presents some shortcomings compared to other reactors (e.g., entrained, fixed bed): large volume, thus investment, scalability problems, etc. this mature technology appears as the most suitable to perform carbon dioxide separation. Because it guarantees temperature homogeneity throughout the vessel, good gas-solid mixing, reduced mass and heat transfer resistances due to the possibility to operate with small particles. Hence this chapter is devoted to the study on the, in one hand, the thermal treatment during the first calcination of a natural sorbent (Dolomite – CaCO3·Mg CO3), and, on the other hand, the study of the behavior of a synthetic sorbent (Mayenite – CAO/Ca12Al14O33) under realistic conditions and the effect of thermal treatments in the activation of the sorbent. 2.1 State-of-art of solid sorbents for high temperature CO2 capture 37 2.1 State-of-art of solid sorbents for high temperature CO2 capture 2.1.1 Natural sorbents The first proposals for the use of calcium-based sorbents to capture CO2 from flue gases at high temperature were realized by Silaban and Harrison [146] and Shimizu et al. [155], during the 1990’s. However, the basic calcination/carbonation loops for the use of limestone as CO2 acceptor were studied from the 1950’s [156-159], and the first patent concerning to the use of lime as CO2 separator (in gasification of carbon by steam) goes back to 1867, by DuMotay and Marechal. The calcination/carbonation reactions are: CaCO 3 (s)  CaO (s) + CO 2 (g) ∆𝐻298 0 = + 182.1 kJ/mol Calcination ( 2.1) CaO (s) + CO2 (g)  CaCO3 (s) ∆𝐻298 0 = - 182.1 kJ/mol Carbonation ( 2.2) Calcination reaction is favored at high temperatures. The cycling of carbonation and calcination are produced by means of thermal swing through the equilibrium of CaO-CO2 system. In other words, calcination is produced if partial pressure of CO2 in the gas around the particles is less than equilibrium pressure; therefore the temperature is increased in order to obtain a higher equilibrium pressure and promote the calcination. A typical expression for equilibrium decomposition pressure is [137]: 𝑃𝑒𝑞 = 4.173 ×107𝑒−20474/𝑇 𝑎𝑡𝑚 ( 2.3) The equilibrium curve is presented in Figure 2. 1. Figure 2. 1 - The equilibrium pressure of CO2 on CaO. 0,001 0,01 0,1 1 10 100 600 700 800 900 1000 1100 PCO2,eq , atm Temperature, ºC CARBONATION CALCINATION SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 38 Instead of the reversibility of carbonation reactions, of CO2 which are able to retain the particles of CO2 in the first carbonation cycle is lower than the CO2 released during the first cycle, and the CO2 uptake capacity of the sorbents decay with cycling, as shown in Figure 2. 2. Figure 2. 2 - The decay in maximum carbonation conversion (X) with the number of cycles [160]. Equation (7) corresponds to Equation 2.4 in this work This decay is due to the evolution of pores in the particles. In the one hand, the large pores increase their diameter with cycling, reducing their activity because the active pores for CO2 capture are under 220nm [161], and the mesopores formed are less active. On the other hand, due to sintering the occlusion of small pores is produced, losing microporosity. The diameter of active pores is limited by the thickness of product layer that controls the conversion of CaO grains. Abanades and Alvarez [160] developed a very simple and accurate correlation in order to obtain the conversion of particles as function of number of cycles, including the mechanism of evolution of pores for a wide range of conditions: 𝑋𝑁= 𝑓𝑚 𝑁 (1−𝑓𝑤)+𝑓𝑤 ( 2.4) Where 𝑋𝑁 is the maximum carbonation conversion in the N cycle. 𝑓𝑚is the fractional loss of the conversion due to microporosity losses: 𝑋𝑁,𝑚= 𝑓𝑚·𝑋𝑁−1,𝑚=𝑓𝑚 𝑁·𝑋0,𝑚 ( 2.5) And 𝑓𝑤is the net loss of total porosity in the particle (also called shrinking), and it is proportional to fraction of specific surface associated with the void volumes forming the mesoporosity, and to the thickness of the product layer according 2.1 State-of-art of solid sorbents for high temperature CO2 capture 39 𝑋𝑁,𝑤= 𝑓𝑤· (1 −𝑋𝑁,𝑚) ( 2.6) Both 𝑓𝑤 and 𝑓𝑚 remain constant with cycling and can be fitted directly from experimental data. This model has been extensively applied to different sorbents under very different operation ways with very accurate results including very long term cycling [162]. Instead of the attempts to reduce the sintering of the sorbents with cycling, the decay of capture capacity is always observed in the cases of constant short-time carbonation/calcination cycles (less than one hour) [161, 163]. Thus, some cost-effective treatments were developed in order to reduce the limitations of sorbents in, mainly, two different approaches. On the one hand, some thermal treatments of fresh sorbent were proposed with the aim to stabilize the internal structure of solids. On the other hand, regeneration of exhausted sorbents (by means of steam or carbonation) was developed. 2.1.1.1 Thermal treatments of natural solids sorbents Figure 2. 3 presents the decay behavior of sorbents (XNage,Nage+1,…) and the previous history of solid. The conversion increases by carbonation, as far as a maximum limited by the natural deactivation. However, if a pre-calcination or a self-reactivation method is applied to the sorbent, the previous history cannot be estimated and the global performance increases. Figure 2. 3 – Schematic representation of thermal treatments previous to the natural decay of solids sorbents with cycling from [164] Manovic and Anthony [143] observed that, if 500µm calcined limestone is exposed to a nitrogen atmosphere for long time (24h) at different temperatures, the carrying capacity is improved at 900ºC due to internal annealing of the particle, and decay with higher temperatures by sintering. This process is known as pre-calcination (Figure 2. 4a). However, when powdering the samples and repeating the experiments an abnormal continuous increase of carrying capacity with cycling (called self-activation) is observed (Figure 2. 4b). SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 40 Figure 2. 4 – Limestone behavior by a) Pre-calcination and b) Self-activation methods from [143] This work is the basis of the development of solid sorbents pellets [165] and the increasing of performance under oxy-fuel calcination conditions [166]. Lysikov et al. [144] proposed a skeleton sintering model that explains the formation of stable structures that prevents the sintering of the particle with cycling. Ozcan et al. observed the same behavior in other sorbents as dolomite or plaster of Paris [167]. In Section 2.2, we proposed a novel pre-calcination method in order improve the carrying capacity of dolomite. 2.1.1.2 Regeneration of spent solids sorbents Other approach to increase the carrying capacity is based in the regeneration of spent sorbents by means of steam reactivation, or recarbonation. Manovic and Anthony [148] demonstrated that the hydration by steam under pressure is a successful method to reactivate spent sorbent in carbonation/calcination cycling, as well as sulphated particles. During the hydration (from 2 to 24h at 1100ºC) the internal structure of the spent solids changes, due to the formation of Ca(OH)2, with a higher molar volume, increasing noticeably the surface area, reducing the size of sintered grains. Figure 2. 5 – Steam reactivation effect on sorbent activity during carbonation in the TGA−sorbent calcined/sintered at 1100 °C for 24 h and hydrated by steam from [148] Lysikov et al. [144], Sun et al.[162], and Chen et al. [139] demonstrated that treated or no pretreated spent sorbents submitted to an intermediate long carbonation cycle resulted in a substantial recovery in CO2 capture ability. In the subsequent cycles, the calcium utilization efficiency again declined as the cycling continued. Therefore, the increase of surface area due to recarbonation is temporal and decays with cycling. 2.1 State-of-art of solid sorbents for high temperature CO2 capture 41 Figure 2. 6 – Results of long-term calcination/carbonation cycling with Strassburg limestone without thermal pretreatment, compared with pretreatment at 1000 °C for 6 h and at 1000 °C for 24 h with two extended periods of carbonation. Calcination/ carbonation cycling was at 850 °C in all cases from [139] Based in the recarbonation effect observed previously, Arias et al. [168] proposed a modification in the activation of spent solids. It is generally accepted that carbonation reaction presents two different regimes. During the first part kinetics is the limiting step in the reaction and, after some seconds (in Figure 2. 7a 60 sec approx.), diffusional effects dominate the reaction, reducing drastically the rate, until reaching a plateau. In their work they demonstrated that once the plateau is reached, if sorbent is exposed to a higher temperature and CO2 partial pressure, the carrying capacity and the residual capacity increase noticeably. This two-step carbonation presents a behavior different to the direct high concentrated carbonation, in which carrying capacity is lower than in low concentrated atmosphere due to the sintering effect of CO2. Figure 2. 7 – a) Example of the increase in CO2 carrying capacity experienced by two particles cycling through the system (black dots, after 15 carbonation calcination cycles and white dots after 100 cycles). b) Evolution of the CO2 carrying capacity of CaO with the number of cycles (black dots with recarbonation and white dots without recarbonation) from [168]. SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 42 2.1.2 Synthetic sorbents Despite of the efforts to improve the carrying capacity of natural lime by increasing the surface area and stabilizing the pore structure, attrition of natural sorbent particles is another main problem in the Ca-looping process. Thus some different types of candidates (i.e. zeolites, activated carbons, calcium oxides, hydrotalcites, organic-inorganic hybrids, metal organics) were proposed in order to improve the CO2 carrying capacity, adsorption-desorption kinetics, thermal and mechanical stability and regeneratibility [169, 170]. Under the operational conditions for SE-SMR process, the most promising synthetic sorbent are hydrotalcites, lithium-based and calcium-based sorbents (see Figure 2. 8). Figure 2. 8 – Schematic representation of the main synthetic sorbents and their carrying capacity and operational temperature window, from [170] 2.1.2.1 Hydrotalcites Hydrotalcite (HTC) is an anionic clay consisting of positively charged layers of metal oxide with interlayers of anions as carbonates (see Figure 2. 9). This combination implies that the calcined carbonates works as CO2 acceptor, and the layer of metal oxides can act as catalyst. Due to these properties and to a very good stability, they were proposed for SE-SMR process [171, 172]. However the adsorption capacity and carbonation rate of natural hydrotalcites are sensibly lower than that of lime. Therefore, some processes of doped synthesis were proposed in order to improve the CO2 capture performance. Reijers et al. [173] proposed the impregnation of HTC with K2CO3, meanwhile Meis et al. [174] proposed doping HTC with Mg(Al)O-crystals, increasing in both cases the rate of capture and obtaining a good 2.2 Dolomite stabilized by means of a novel thermal treatment 49 in order to study the effect of these pre-treatments in trigger calcined dolomite. The details of operation are summarized in Table 2. 2: - Long carbonation: from 30 minutes to 72 h: Runs from 12 to 15 [139]; - Long calcination: from 5 to 90 minutes: Runs from 16 to 18 [164]; - Trigger cyclic regeneration, every 5 cycles: Runs 19 and 26, adapted from [202]; - High temperature long calcination: from 900 to 1100 ºC; Runs from 20 to 25 [143]. Table 2. 2 – Summary of run details for pre-treatment effect study Run Carbonation 1st Calcination (Heating rate 100ºC/min) Nth Calcination (Heating rate 100ºC/min) N cycles Carbonation Time 50% CO2 [min] Atmosphere composition Isothermal 900ºC time [min] Atmosphere composition 12 1st 2 hNth 30 min Mild 5 Mild 50 13 1st 2 hNth 30 min Trigger 5 Mild 50 14 1st 72 hNth 30 min Mild 5 Mild 50 15 1st 72 hNth 30 min Trigger 5 Mild 50 16 30 Trigger 30 Mild 50 17 30 Trigger 60 Mild 50 18 30 Trigger 90 Mild 50 19 30 Trigger 5 Mild (each 5 cycles trigger) 45 20 30 Mild 360 Mild 30 21 30 Trigger 360 Mild 30 22 30 Mild 360 (1000°C) Mild 30 23 30 Trigger 360 (1000°C) Mild 30 24 30 Mild 360 (1100°C) Mild 30 25 30 Trigger 360 (1100°C) Mild 30 26 30 Trigger 5 Trigger 30 SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 50 The number of cycles studied was limited by the useful life of these types of sorbents. Attrition and decay of carrying capacity limit the cycling period of dolomites to a maximum of 30 cycles, being necessary purges in order to maintain the performance of the cycles[203]. 2.2.2 Effect of first calcination: Triggered vs. Standard calcination The first experiments were carried out in order to study the effects of gas composition during the thermal pretreatment on porosity, and, thus, on CO2 carrying capacity. Two typical calcination atmospheres have been studied: 100% N2 (Run 1), 50/50% N2/CO2 (Run 2) in onestep calcination. Moreover, due to the nature of dolomite which presents two different crystalline phases, triggered calcination is realized in order to control the releasing of CO2 through the particle during the calcination of the two main carbonates which dolomite is composed of. Having completed the release of moisture from the solid sample, the calcination of the dolomite occurs at temperatures greater than 500 °C. Particularly the sample exposed to CO2 free atmosphere (100 % N2) shows a smooth weight loss when the temperature exceeds 650 °C indicating that MgCO3 and then CaCO3 are converted into the respective oxides with a release of CO2 and calcination is yielded. When the investigated material is exposed to CO2 containing atmosphere (50 % CO2, N2 balance) the decomposition of the carbonates is inhibited during sample heating. Particularly, the decomposition of MgCO3 happens at temperatures greater than that of CO2 free calcination and the weight decrease of the sample starts at temperatures higher than 750 °C , Upon completed a first weight loss (half-calcination) the presence of CO2 further inhibits the calcination of the remaining carbonate (CaCO3). As shown in Figure 2. 11, CaCO3 is mostly decomposed into CaO when the flow of CO2 was stopped when the temperature reaches 900 °C and the sample is exposed to pure nitrogen atmosphere (Fig. 2.11: curve 2-step ‘trigger’). The sample experiences a very rapid linear weight loss followed by a sharp transition to a much slower reaction rate where the calcium carbonate is completely converted into oxide. Figure 2. 11 – Comparison between experimental TGA mass loss under standard calcination and trigger calcination. Temperature history is also reported. 0 100 200 300 400 500 600 700 800 900 1000 0 1 2 3 4 5 6 7 8 9 0 100 200 300 400 500 600 700 800 Temperature (ºC) Mass Sample (mg) Time (s) 1-step N2 1-step CO2 2-step "trigger" Temperature 2.2 Dolomite stabilized by means of a novel thermal treatment 51 Calcined samples produced with the three calcination methods were subjected to multi-cycling CO2 capture. The three investigated samples showed the same CO2 uptake capacity for the first carbonation step indicating that the first thermal pretreatment does not influence the reactivity of the three calcined sorbents. However, remarkable differences in CO2 uptake capacity were observed during the cycling, as shown in Figure 2. 12. Calcined particles produced under N2/ CO2 atmosphere (Run 2) show, indeed, a greater capture capacity decay during the first 15 cycles (0.21 to 0.14 gCO2/g) when compared to the calcined produced under pure nitrogen (Run1). Gas composition strongly affects sintering of CaO grains, having CO2 more sintering capacity than N2. Interestingly, sample pre-treated dolomite shows a significant better performance in one step, up to 44% capture capacity, than CO2-calcined dolomite and 30% N2 calcined one. Figure 2. 12 - CO2 uptake at heating rate of 10°C/min for different calcination methods over 15 cycles. The variation on carrying capacity is defined as: 𝑋𝐶𝑂2=� (𝐶𝑂2 𝑢𝑝𝑡𝑎𝑘𝑒 𝑟𝑢𝑛 (𝑔𝐶𝑂2/𝑔 𝑑𝑜𝑙𝑜𝑚𝑖𝑡𝑒)− 𝐶𝑂2 𝑢𝑝𝑡𝑎𝑘𝑒 𝑟𝑒𝑓 (𝑔𝐶𝑂2/𝑔 𝑑𝑜𝑙𝑜𝑚𝑖𝑡𝑒)) ( 𝐶𝑂2 𝑢𝑝𝑡𝑎𝑘𝑒 𝑟𝑒𝑓 (𝑔𝐶𝑂2/𝑔 𝑑𝑜𝑙𝑜𝑚𝑖𝑡𝑒)) × 100 (2.11) Where 𝑋𝐶𝑂2 is the CO2 carrying capacity (%), subscript run is the run study case and ref is the reference case. In fact, the pre-treated dolomite keeps constant the CO2 carrying capacity during the first 10 cycles (0.20 gCO2/g), and decreases slowly with the further cycles. The transition from CO2/N2 to pure N2 during calcination could likely triggered a sintering-resistant structure in the particle, 0,00 0,05 0,10 0,15 0,20 0,25 0 5 10 15 CO2 Uptake (gCO2/g Dolomite) Cycle Number R.1-Calcination 100% N2 R.2-Calcination 50/50 CO2-N2 R.3-Calcination 50/50 CO2-N2+100%N2 SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 52 by gas diffusion in the pore. Moreover, the presence of MgO inhibits blockage of the pores during grain growth [138, 161, 204]. 2.2.2.1 Effect of heating rate Experiments showed in Figure 2.12 were carried out at a constant heating rate of 10ºC/min. High heating rates promotes densification, reducing porosity by enhancing sintering of grains, instead of grain growth [205, 206] and, therefore, reducing the CO2 carrying capacity of sorbents. Hence, in order to know the effect of heating rate on CO2 uptake, four heating rates were studied: all calcination/carbonation cycles at 10ºC/min; all cycles at 100ºC/min; first calcination at 10ºC/min and successive cycles at 100ºC/min; first calcination at 100ºC/min and subsequent ones at 10ºC. Figure 2. 13 shows the effects of heating rate on CO2 capture capacity. The highest CO2 uptake, in the first 15 cycles, corresponds to the case in which all cycles were done at 10ºC/min, and decreases at higher heating rates. In the cases of 100ºC/min the effect of a constant uptake for 6 cycles is observed, with higher carrying capacity than the cases of lower heating rate, and a strong decay in next cycles. This behavior is observed independently if the 100ºC/min calcination is in the first cycle (with subsequent 10ºC/min calcinations), or if the first calcination is at 10ºC/min and subsequent 100ºC/min calcinations. Hence, we can consider that pre-treatment in the first calcination produces this constant CO2 carrying capacity in the first cycles, and that the pre-treatment depends on of the heating rate, being more stable the structure in the cases of lower heating rate. Figure 2. 13 – CO2 uptake at different heating rates over 15 cycles 0,00 0,05 0,10 0,15 0,20 0,25 0 5 10 15 CO2 Uptake (gCO2/g Dolomite) Cycle Number R.1-1st Calc N2 - 10K/min R.3-1st Calc CO2/N2+N2 - 10K/min R.4-1st Calc CO2/N2+N2 - 100K/min R.5-1st Calc CO2/N2+N2 100K/min - 10K/min R.6-1st Calc CO2/N2+N2 10K/min - 100K/min 2.2 Dolomite stabilized by means of a novel thermal treatment 53 2.2.2.2 Effect of CO2 concentration In order to widely investigate the effectiveness of the pretreatment presented here, the calcined sample was also subjected to a 2-step calcination and exposed to 12 % CO2 (N2 balance) to emulate post-combustion CO2 capture. The behavior of CO2 uptake is around a 5% lower in the case at 12% than in that of 50% vol. CO2, as shown in Figure 2. 14. This decay is congruent with literature data [137]. Figure 2. 14 – Uptake of pre-treated dolomite over 50 cycles for different CO2 concentrations 2.2.2.3 Effect of carbonation time Sintering is strongly dependent and increases with time [207, 208]. Therefore, a dolomite particle subjected to carbonation or calcination process for long time presents a reduction of the specific surface and reactivity. Figure 2. 15 shows as triggered calcined dolomite presents a high constant CO2 uptake capacity in the first cycles if carbonation lasts 30 minutes than 2 min. Nevertheless up to 30 cycles the sintering affects in a higher degree the 30 minutes-carbonated dolomite than the 2 minute-ones, equating the carrying capacity. It is important to note this unexpected effect of the novel calcination method respect to results obtained elsewhere [159]. 0,00 0,05 0,10 0,15 0,20 0,25 0 10 20 30 40 50 CO2 Uptake (gCO2/g Dolomite) Cycle Number R.4-CO2 50% - 100K/min R.7-CO2 12% - 100K/min SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 54 Figure 2. 15 – CO2 uptake of pre-treated dolomite at 100°C/min heating rate over 50 cycles for different carbonation time In an attempt to study the effect of regeneration condition on pretreated sorbent, both simple (100% (mild) and 50% (severe) N2 calcination atmosphere) and 2-step calcined samples were subjected to successive CO2 capture cycling where the regeneration occurs under 100% N2 (mild regeneration) or 50:50% CO2/N2 (severe regeneration) atmosphere for 150 cycles, in order to obtain the values of the minimum residual carrying capacity [162]. Single mild calcined samples were calcined under mild condition for all 150 cycles (Run 8). Analogously, single severe calcination cycles were repeated for all experiments (Run 9). Trigged calcined samples were regenerated under mild (Run 10) and severe conditions (Run 11). Figure 2. 16 shows as triggered pretreated dolomite presents a higher CO2 capture capacity than one step calcined ones. Under the investigated conditions the improvement of CO2 uptake is 12.3% in the 100% N2 calcination and 24.5% in the 50%CO2 calcination atmosphere, with respect to the one-step calcination, mild and severe respectively. It worth to note that 2-step pretreated sorbent shows a higher CO2 uptake capacity when regenerated under severe condition. It is also observed a lower CO2 uptake in the cases of severe first calcination than in mild, according with Figure 2. 3. 0,00 0,05 0,10 0,15 0,20 0,25 0 10 20 30 40 50 CO2 Uptake (gCO2/g Dolomite) Cycle Number R.10-2min CO2 R.4-30 min CO2 2.2 Dolomite stabilized by means of a novel thermal treatment 55 Figure 2. 16 – CO2 uptake of different calcination methods dolomite over 150 cycles for different CO2 concentration calcination and carbonation 2.2.3 Effect of combined pre-treatments As reported in Figure 2. 15, an increase in carbonation time reduces CO2 uptake. However, a long-term carbonation (72h) was demonstrated to be a good method in order to regenerate exhausted dolomite [139]. Here, such a process is investigated as a possible route to pre-treat the sorbent material. Thus, single-mild calcined and triggered-calcined samples were carbonated for long times, from 2, (Run 12 and 13, respectively) to 72 h (Runs 14 and 15), with the aim to evaluate the combined effect of long carbonation during the first calcination. As shown in Figure 2. 17a the single calcined samples present a slightly increase in the carrying capacity, a result congruent with literature data. However, this effect is not present in the triggered calcined samples. As shown in Figure 2. 17b, there is a decrease in carrying capacity with carbonation time. The CO2 captured with 2 min carbonation is higher than with 30 min. With 2 and 72 hours carbonation, CO2 carrying capacity is reduced ~ 17% and 33%, respectively compared to 30 min calcination. However, samples pretreated with triggered calcination show similar behavior regardless of the precarbonation operation. Long precarbonated samples (2 and 72h, Runs 13 and 15, Figure 2. 17a) and the samples as presented in Figure 2. 15 (no-precarbonation) experience constant carrying capacity in the first 5 cycles, similar asymptotic decay curves (with convergent residual capacity). Moreover, self-activation effect were observed in the 72h carbonated case [143]. Thus we infer that the structure generated by means of the trigger calcination needs a minimum first carbonation time in order to be stabilized. 0,00 0,05 0,10 0,15 0,20 0,25 0 25 50 75 100 125 150 CO2 Uptake (gCO2/g Dolomite) Cycle Number R.8-All Calc 100%N2 R.9-All Calc 50/50%-CO2/N2 R.10-1st Calc CO2/N2+N2 - nth 100%N2 R.11-1st Calc CO2/N2+N2 - nth 50%CO2 SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 56 Figure 2. 17 – CO2 uptake under different carbonation times a) Long carbonation times for standard and trigger calcination b) Comparison of effect of carbonation time for trigger calcined samples As commented previously, sintering increases with time and it stabilizes the pore sized distribution, without producing changes on the particle porosity. Nevertheless, the specific surface area decreases strongly if particles remains during time at high temperature, reaching an asymptotic value before 50% conversion [209]. 0,00 0,05 0,10 0,15 0,20 0,25 0 10 20 30 40 50 CO2 Uptake (gCO2/g Dolomite) Cycle Number R.12-1st carbonat 2h - N2 R.13-1st carbonat 2h-CO2 R.14-1st Carbonat 72h - N2 R.15-1st Carbonat 72h - CO2 0,00 0,05 0,10 0,15 0,20 0,25 0 10 20 30 40 50 CO2 Uptake (gCO2/g Dolomite) Cycle Number R.10-2min CO2 R.4-30 min CO2 R.13-1st carbonat 2h-CO2 R.15-1st Carbonat 72h - CO2 a) b) 2.2 Dolomite stabilized by means of a novel thermal treatment 57 Different authors report the reduction of CO2 carrying capacity of limestone with increasing of first calcination time [127, 144, 210]. However, in the experiments carried out in this work, this effect has not been observed in the pure-nitrogen isothermal calcination (increasing calcination time from 5 to 90 min). The behavior of sorbent is independent of calcination time, as shown in Figure 2. 18. Figure 2. 18 – CO2 uptake under different first calcination times for triggered calcined dolomite This fact suggests that the structure formed during thermally pretreatment is likely stable and rich in relatively large pores (diameter ≥ 220-350 nm), more resistant to sintering and closure than minor ones. Those pores could reduce their diameter with cycling because of cracks and channel formation throughout the particle increasing the pore volume (≤ 220nm) [39]. Then a competitive phenomenon could likely happen during the first cycles (sintering and crack formation) explaining the constant carrying capacity in the first calcination/carbonation cycles. Manovic and Anthony [143] studied the effect of thermal treatment on CaO-based sorbents. They observed that calcination under high temperature (1000–1100 ºC) for long time (6-48 h) the sorbent structure changes, produces a stable porous skeleton, which increases sorbent conversion with cycling. However, this effect has been scarcely investigated in naturally occurring dolomites. Thus, six experiments were carried out in order to evaluate the effects of thermal pre-treatment in N2 atmosphere and trigger calcination, and whether self-activation is presented (Runs 20, 22 and 24 for single mild calcination and Runs 21,23 and 25 for triggered calcined; at 900, 1000 and 1100ºC for 6h). The experiments of single calcined pre-treatments have not shown an increase of CO2 uptake with cycling, and a reduction of carrying capacity has been observed (Figure 2. 19). This effect is consistent with the sintering general theory in which the porosity of particles decrease with 0,00 0,05 0,10 0,15 0,20 0,25 0 10 20 30 40 50 CO2 Uptake (gCO2/g Dolomite) Cycle Number R.4 - 5 min R.16 - 30 min R.17 - 60min R.18 - 90 min SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 58 long-time high-temperature exposure [207]. In the case of trigger calcined dolomite, carrying capacity performance decreases (respect to 900 ºC 5-minutes isothermal pretreated) in 11, 8 and 18% for 6 hours and 900, 1000 and 1100ºC respectively. In the case of 900 and 1000ºC during the first 6 cycles, CO2 carrying capacity remains constant. However for the 1100 ºC calcination case, the decay of performance starts on the 3rd cycle, due to a strong effect of sintering. Thus, for both cases (simple and trigger calcined dolomites) the thermal pretreatment is not effective in order to increase the carrying capacity. This behavior can be explained for the presence of MgO, which creates a more porous structure in dolomite than in limestone, at the same time more susceptible to be affected for sintering. Moreover, it is possible to hypothesize that during trigger calcination there is a formation of micropores, which are more affected by sintering and they collapse during long high-temperature calcination, reducing CO2 uptake capacity with respect to the 5-minute pretreated one. A more detailed explanation about the texture changes in the particle is described in the next section. Figure 2. 19 – CO2 uptake under different thermal treatments (long time and high temperature) for trigger calcined dolomite and standard pure nitrogen calcination. The triggered calcination presented an increase of performance with respect to classical calcination in all cases. However, in order to evaluate the effect of the triggered calcination over the first calcination, two experiments were carried out. In one set of experiments trigger calcination were carried out for all cycles (Run 26). In other experimental run, the trigger calcination has been used as a regeneration process each 5 cycles (4 calcinations in N2 + 1 trigger calcination, Run.19), as suggested for other regeneration studies [147, 148, 211]. Figure 2. 20 shows the results of these approaches, compared with the reference case. It is observed as, in the 5th cycles, when trigger calcination is used as regeneration, carrying capacity increases for that cycle, but decays strongly in the subsequently ones. However in the 0,00 0,05 0,10 0,15 0,20 0,25 0 5 10 15 20 25 30 CO2 Uptake (gCO2/g Dolomite) Cycle Number R.20-Pret 6h - N2 - 900C R.21-Pret 6h - CO2 - 900C R.22-Pret 6h - N2 - 1000C R.23-Pret 6h - CO2 - 1000C R.24-Pret 6h - N2 - 1100C R.25-Pret 6h - CO2 - 1100C 2.2 Dolomite stabilized by means of a novel thermal treatment 65 2.2.5 Conclusions It has been observed that first calcination defines the behavior of multi-cycling carbonation performance. Pre-treated dolomite by means half-calcination in CO2 -containing atmosphere with a subsequent 900 °C isothermal calcination presents better carrying capacity than one-step calcination under N2 or CO2 atmosphere, due probably to the structural changes in the lattice of dolomite particle. Triggered-calcined samples present a stable porosity with cycling. This fact provides better performance than other calcination methods in which porosity decreases with cycling, and, hence, the CO2 capture capacity also decreases. This theory was validated by means of an indirect method in order to obtain the specific surface area, and relate S with CO2 capture performance. The effect of several variables (such as CO2 concentration, heating rate, calcination and carbonation time) and some pre-treatments has been tested under triggered calcination, producing a reduction of CO2 uptake, due to the internal structure effects. The best performance sample was the triggered simplest calcination under industrial condition (low calcination and carbonation time, without other thermal treatment). Thus, this simple method can be applied directly in carbon capture cycles, increasing noticeably the performance of dolomite as CO2 acceptor SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 66 2.3 Self-reactivation and regeneration of Mayenite (CaO·Ca12Al14O33) 2.3.1 Synthesis of solid sorbent The synthesis has been performed according to [132] where powdered CaO (about 99.8%, after calcination) and aluminum nitrate Al(NO3)3·9H2O (> 98%) have been selected as precursors in the synthesis of the binder phase. A wet method has been used to ensure an intimate mixing of starting materials: Distilled water has been mixed with 2-propanol acting as surfactant. The calcium oxide has been calcined at 900°C for 2 h in the presence of air to remove humidity and decompose any traces of CaCO3 into CaO. The amounts of CaO and aluminum nitrate were chosen such that the mass ratio of the active phase CaO to binder phase was 75:25 and 85:15. The compounds were added to the water and the mixture has been stirred at 75 °C and 700 rpm. After 60 minutes stirring, the solution has been dried at 120 °C for 18 h to obtain a dried cake. In order to form the binder Ca12Al14O33 the material has been ground and heated up to 850 °C. Before reaching the temperature for Ca12Al14O33 formation the sorbent precursor has been dried at 500 °C for 180 minutes to evaporate nitric oxides and produce Al2O3 in a controlled mode. The solid-solid reaction between Al2O3 and CaO to form Ca12Al14O33 requires several hours, but the reaction rate can be improved by a further grinding of the manufactured material as specific surface area is maintained higher and fresh surface is brought in contact. As a consequence, after cooling to room temperature the material has been ground again in a mortar and heated up to 850 °C for 90 minutes in order to react Al2O3 with CaO to form Ca12Al14O33. After having completed the final calcination and the subsequent cooling to the room temperature the sorbent material has been ground and sieved. The powder used in this investigation had particle sizes in the range 180 to 500 µm. 2.3.2 Characterization of Mayenite – Experiments in TGA A number of experimental investigations at a laboratory scale have been performed to characterize a synthetic solid sorbent to accomplish the uptake of CO2 from a gaseous mixture. The reactivity and CO2 uptake capacity during cycling of the sorbent were analyzed by using a GC-10 Mettler-Toledo thermo-gravimetric analyzer (TGA). This apparatus can measure minute mass changes of solid samples placed in a furnace with a variable and well-controlled temperature and gas atmosphere. Blank runs were conducted with an empty crucible to record the disturbances in the mass change readings when moving the experiment from calcinations to carbonation process. To avoid the effect of the sample size on carbonation and regeneration processes, such as the external mass transfer resistance of CO2 through the sample, ~2.90 mg samples were used. The experimental protocol used for performing the solid chemical looping consisted of three steps: • Regeneration phase: This phase is conducted by heating the sample up to a programmed temperature at 100 °C/min rate (maximum heat rate allowed by this apparatus). A mixture of nitrogen and carbon dioxide flows over the sample. During this phase the CaCO3 is decomposed into CaO. The sample is calcined for 15 min at 900 or 1000 °C in order to guarantee the complete calcination of sorbent. Three representative regeneration conditions have been used (N2 balance gas): 2.3 Self-reactivation and regeneration of Mayenite (CaO Ca12Al14O33) 67 - Mild condition: temperature 900 °C, with 14 % CO2; - Moderate condition: temperature 1000 °C, with14 % CO2 ; - Severe condition: temperature 1000 °C; with 86 % CO2. • Cooling and carbonation phase: The temperature is then lowered with the rate of 100 °C/min to reach 600 °C selected for CO2 sorption. The atmosphere remains 84% nitrogen and 16% carbon dioxide. When the temperature goes below 700-750 °C the CO2 begins reacting with CaO to form CaCO3. • Isotherm carbonation phase: Having completed the cooling phase, the temperature is maintained constant at 600 °C for 20 minutes. The reacting atmosphere has been programmed to study the effect of CO2 content in the gas to decarbonise: (N2 balance gas): - Higher CO2 content: atmosphere composed of 25 % CO2 - Lower CO2 content: atmosphere composed of 14 % CO2. 2.3.2.1 Treatment of the sorbent precursor During the synthesis process water could be added to the powder obtained with the second sorbent precursor grinding. Few water drops have been added to the particle in order to obtain a sort of paste: CaO reacts with H2O to form Ca(OH)2. Then the hydrated material has been heated up to 850 °C to form the binder phase as reported earlier (Sec 2.1). 2.3.2.2 Treatment of the sorbent material The synthesized material has been thermally treated to see whether the advantage found for the dolomite and calcite [139, 143] persists on the investigated material. Thus solid specimens have been carbonated at 600 °C for 80 minutes, exposing them to a prolonged carbonation with a controlled atmosphere composed of 14% CO2 (N2 balance gas). Long carbonation has been realized with the aim to stabilize the sorbent structure as reported in [162]. After the thermal treatment in TGA has been completed, the solid sorbent has subsequently undergone the multicycling regeneration-carbonation process in order to avoid influences on the investigated specimen (e.g reaction between sample and CO2 in the room atmosphere). Comparison between synthesized sorbent and thermally pretreated sorbent are presented in this work. Particularly the capacity to retain carbon has been investigated by cycling the material up to 150 cycles. In order to study the effect of a cyclic treatment on the material, solid specimens have been exposed to a further prolonged carbonation with a previous calcination step (1000 °C with 14 % CO2, N2 balance). When the sorbent is treated in such a manner, the material is named here as double-treated. 2.3.3 Self-reactivation of Mayenite Figure 2. 24 shows conversion-time curves for different numbers of regeneration-carbonation cycles regenerated in moderate condition. At the beginning of each single cycle, besides a short nucleation period, a linear kinetically-controlled mass growth was found followed by a SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 68 transition to a much slower reaction rate. This transition has been found to be smoother when compared with that occurred during the carbonation of naturally occurring sorbent investigated in the dolomite experiments. No plateau is reached during the last slower phase and the surface reaction continues to play a key role in the whole process. Such a smooth transition was found in [143] where experimental test on CO2 capture for pre-treated limestone were presented. Moreover, Figure 2. 24b shows that with increase in cycle number the initial slope of the conversion-time curves increases from the first to the tenth cycle. The increase of the sample weight with first cycles could likely explained with the phenomenon, named self-reactivation. Analogous effect was observed also in the field of chemical looping combustion where a particular O2-carrier showed a gain in reactivity during reduction/oxidation cycle performed in a thermo gravimetric analyzer [224] . After the 10th cycle the linear growth shows the same slope up to the 80th cycle carbonation-regeneration loop of the experimental test. Moreover, after that cycle, as reported in Figure 2. 24a, TG curves were found to be similar denoting a good reversibility through the multi–cycling CO2 capture. In addition, Figure 2. 24b shows that with the increase of cycling up to 10th carbonation–regeneration step the sample mass achieved after each single loop has been increased when compared with the previous loop. In order to compare the self-reactivation of synthetic sorbent to the performance of naturally occurring sorbent, dolomite sample (Bianchi dolomite: 55.6% CaCO3 and 44.2% MgCO3) has been selected and exposed to the same condition reported in Figure 2. 23. The uptake of the two sorbents for each single cycle was evaluated according to the following ratio: 02 mΔm uptakeCO N = ( 2.15) where ∆ mN represents the mass augment of the solid specimen at each single N-th cycle and m0 represents the weight of the sample inserted in TGA. Figure 2. 23 – Comparison between 75 % CaO synthetic sorbent and “Bianchi” dolomite in term of carbon retain capacity. 2.3 Self-reactivation and regeneration of Mayenite (CaO Ca12Al14O33) 69 Figure 2. 24 – TG curves collected for specimens subjected to moderate regeneration condition up to 80 cycles: (a) mass growth shows no plateau; (b) self reactivation effect: the slope of initial linear mass growth increases with the number cycle. Even if the dolomite shows a larger uptake when compared to that of synthetic sorbent, a decay in the capacity to retain carbon during the first few cycles was found for Bianchi dolomite whereas the synthetic sorbent shows higher uptake while the material is cycled during CO2 capture tests. 2.3.3.1 Effect of sorbent precursor hydration Improvement of sorbent activity has been also observed for the sorbent obtained by chemical pretreatment of the sorbent precursor. By adding water after the second grinding (see Sec 2.3.2.1) the obtained sorbent shows greater uptake when compared with the sorbent whose precursor was dry (see Figure 2. 25). Carbon dioxide uptake of the former sorbent was ~94% higher when compared to the latter at the second cycle. In subsequent cycles the uptake has been at least ~60% higher than that obtained with dry precursor. Such an improvement could be likely be explained with the formation of calcium hydroxide in the sorbent precursor due to the presence of water and the subsequently calcination (for the binder formation see Sec 2.3.2.2) leaving more pore volume. In fact the molar volume of Ca(OH)2 is greater than that of CaO. Thus when the hydrated precursor would undergo calcination process water vapor would be formed and it would leave the particle producing extra pore volume. Moreover the migration of water vapor towards the outer part could likely create cracks throughout the sorbent particle SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 70 exposing more specific surface area to the carbon dioxide. Consequently the specific surface area is higher and more CaO surface is brought into contact with the CO2. Figure 2. 25 – Comparison between material obtained from hydrated precursor and dry precursor when subjected to moderate regeneration condition for the 75% CaO sorbent. 2.3.3.2 Pore size distribution curves The specific surface and the pore volume distribution were analyzed in a Micromeritics Tristar II using N2 physisorption and desorption isotherms at −196 °C (Brunauer–Emmett–Teller BET method [225]). The pore volume distribution was evaluated by using Barrett-Joyner-Halenda (BJH) method [226]. The formation of larger and smaller pores has been indeed found as reported in Figure 2. 26 where the BJH curves for the two kinds of sorbent are presented. As it is possible to see the sorbent synthesized with dry precursor shows a lower single–peaked pore size distribution (average size 30 nm) whereas for the material obtained from hydrated precursor a wider pore size distribution has been observed. Particularly, the formation of higher volume with larger pores (100 nm) have been detected which permits CO2 to get the inner core of the particle with a major CaO utilization whereas smaller pores play a key role to the rapid carbonation of the sorbent material. The hydrated precursor derived sorbent present a peak on ~100 nm below 220 nm, which were found to be the more active sizes for CO2 capture [161]. These pores provide almost all the surface area and determinate the apparent carbonation rates. Then the experimental results presented in the remainder of this work are collected from the sorbent obtained from the hydrated precursor. The pore size distribution (PSD) curves for sorbents after different carbonation time (20 and 80 minutes) in Figure 2. 26 show that if the sorbent is exposed at different carbonation time the internal structure of the material shows an increase in the pore volume particularly in the range of 100–200 nm diameter pores. Having completed the treatment, indeed, material shows an increase in the specific surface area when subjected to a single calcinations step: for an instance the 85% CaO sorbent’s surface is increased from 1.5 m2/g to 2.6 m2/g. This specific surface 2.3 Self-reactivation and regeneration of Mayenite (CaO Ca12Al14O33) 71 increase could likely explained with the larger amount of CaO which reacts with the CO2 during the pretreatment of the material. In fact, when the solid specimen is exposed for longer time to the pretreatment the CO2 diffuses through the particle reaching at its inner core and reacting with more CaO grains. As a consequence when the material is subjected to the calcination step, more CO2 would be drive off the particle with a formation of cracks and channels throughout the particle increasing the pore volume. It worth to note that the pore volume shows a remarkable increase when the material is cyclically used in a carbonation–regeneration loop (see Figure 2. 26). As a consequence more specific calcium oxide surface would be exposed during the first cycles enhancing the reaction between active phase (CaO) and CO2. Moreover the pore size shifts from 100 nm to 60 nm permitting a major CaO utilization with an increase in specific up to 4.0 m2/g for the sorbent 85% CaO sorbent. Such a phenomenon could likely explain the augment of the slope shown by the uptake curves in Figure 2. 24. A similar behavior has been found for the sorbent obtained from dry precursor (Figure 2. 26): with the cycling the average pore size moves from 30 nm to 60 nm. Then it appears that cycling the synthetic sorbent there is an increasing of the pore volume which is favorable to CaO–CO2 reaction. Such a positive effect of cycling on the growth of pore volume decreases with increased the calcination temperature as described in more details in the following section. Figure 2. 26 – Pore size distribution of material obtained with hydrated precursor and dry precursor. Pore volume growth after 30 cycles of carbonation–calcination: PSD curves for the cycled 85 % CaO sorbent show an increase in pore volume peaked at 60 nm SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 72 2.3.3.3 Influence of the regeneration condition Figure 2. 27 and Figure 2. 28 show the influence of regeneration condition on synthesized sorbent uptake. In particular, it can be seen that with the increase of both temperature and the CO2 molar fraction in the atmosphere during the sorbent regeneration the self-reactivation period decreases. • Mild condition When the sorbent is regenerated with mild condition, the self-reactivation period is extended up to ~60th cycle for the untreated and single treated 85% CaO sorbent (Figure 2. 27a). When such a material is double-treated and regenerated in mild condition, no significant self-reactivation period has been detected but the material shows a very good stability up to the 150th cycle. As for the 75% CaO sorbent a self-reactivation period has been found up to the 30th cycle for the 75%CaO sorbent (untreated singleor double treated) (Figure 2. 28a). • Moderate condition As reported in Figure 2. 27b – Figure 2. 28b, if the material is regenerated in moderate condition the self-reactivation period is reduced to the first 15 cycles for the untreated sorbent. When the material is single treated self-reactivation is further reduced. However, when the material is double treated, self-activation is not observed. Moreover a lower uptake capacity and loss of stability with cycling have been detected. Particularly, the double treated 85% CaO sorbent shows self-reactivation at later cycles whereas the 75% CaO decreases its performances at each single cycle. Such a result is consistent with [152-154] where the authors have found that the presence of CO2 in the calcination step favors sintering process. • Severe condition The 85% CaO sorbent regenerated in severe condition does not present considerable selfreactivation period (see Figure 2. 27c). Besides an initial short period (10 cycles) where the untreated 85% CaO sorbent shows stability in CO2 capturing, a slight loss of performances has been found up to the 100th cycle. In addition the ability to retain carbon decreases in later cycles even if the material is singleor double treated. Perhaps the most remarkable result is reported in Figure 2. 28c where the uptake of the sorbent subjected to severe regeneration condition is shown. In contrast with 85 % CaO sorbent, the untreated 75 % CaO sorbent shows self-reactivation period up to the 60th cycle of the carbonation looping. Besides an initial decrease of the performance, the single pretreated sorbent shows a continuously increment of its CO2 capture capacity up to 150th cycle. In order to confirm such an unexpected behavior another experimental run has been accomplished confirming the previous observation (see black dots in Figure 2. 28c) with a good reproducibility. Finally for the double treated specimen no self-reactivation period was found at the beginning where a loss of CO2 uptake has been observed followed by a period of good stability; beyond such a period a self-reactivation was observed reaching a maximum at ~150th cycle. 2.3 Self-reactivation and regeneration of Mayenite (CaO Ca12Al14O33) 73 Figure 2. 27 – Carbon capture capacity of 85% CaO sorbent when subjected to different regeneration condition: (a) mild regeneration: 900 °C, 14/86 %v (CO2/ N2); (b) moderately severe regeneration: 1000 °C, 14/86 %v (CO2/ N2); (c) severe regeneration: 1000 °C, 86/14 %v (CO2/ N2). SOLID SORBENTS FOR HIGH TEMPERATURE CO2 CAPTURE 74 Figure 2. 28 – Carbon capture capacity of 75% CaO sorbent when subjected to different regeneration condition: (a) mild regeneration: 900 °C, 14/86 %v (CO2/ N2); (b) moderately severe regeneration: 1000 °C, 14/86 %v (CO2/ N2); (c) severe regeneration: 1000 °C, 86/14 %v (CO2/ N2). Further runs in (b) e (c) have been presented to show the good reproducibility of the experimental data. 3.1 A brief introduction to fluidization 81 Figure 3. 2 – Geldart classification of particles for air ambient conditions. Adapted from [234] Geldart particle types are, from smallest to largest particles: - Group C: cohesive, or very fine powders (dp < 30 μm) (e.g. flour or cement). Normal fluidization of this type of particles is very difficult because van der Waals and other cohesive forces are stronger than those due to gas. Channeling occurs instead of fluidization. Mechanical powder compaction, prior to fluidization, greatly affects the fluidization behavior of the powder, even after the bed had been fully fluidized previously. A very good characterization of this type of particles can be found in [235]. - Group A: aeratable, or small mean particle size and/or low particle density (<1.4g/cm3) (major example is the FCC catalyst). Easily fluidized particles with smooth fluidization at low gas velocities; and controlled bubbling with small bubbles at higher gas velocities. Large bed expansion before bubbling starts. When fluidized by air at ambient conditions, result in a region of non-bubbling fluidization beginning at Umf , followed by bubbling fluidization as fluidizing velocity increases. Gas bubbles rise faster than the rest of the gas. - Group B: sand-like, or particle size 40 μm to 500 μm and density 1.4 to 4 g/cm3. Majority of gas-solid reactions occur in this regime, based on particle size of raw materials. Gas bubbles appear at the minimum fluidization velocity Umf. Small bubbles form at the distributor and grow and coalesce as the rise though the bed. Bubble size increases linearly with distance and it is independent of particle diameter. - Group D: spoutable, or large and/or dense particles. Very erratic behavior giving large exploding bubbles, or severe channeling, or spouting behavior. Bubbles coalesce rapidly and flow to large size. Bubbles rise more slowly than the rest of the gas percolating through the emulsion. Dense phase has a low voidage. These particles are used in drying and roasting process (e.g. coffee beans wheat, lead shot). CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 82 Geldart classification is very useful to predict the behavior of solids under fluidization conditions. However is not sufficient to predict the regime in which the bed will be. For engineering applications it is necessary know the type of particle and also the fluidization regime to design the reactors and predict the gas-solid contacting mode. That is, the Geldart B particles present a behavior different to Geldart A ones, but, moreover it is necessary know if the particle will be under bubbling or circulating regime (or another regime). Figure 3. 3 is an excellent summary of the possible paths through particulate regimes as the velocity of the fluidizing gas is increased within the bed. Figure 3. 3 – Possible paths of fluidization, from [228] Some investigators have mapped these regimes, but the most widely used is the Grace’s diagram [233, 236]. With a methodology similar to the Geldart work, Grace stabilished the relation between fluidization regime and Geldart particle groups. Moreover, he compared the different theories to calculate minimum fluidization velocity and the boundaries between particle types. Thus, as shown in Figure 3. 4, it is possible predict the fluidization regime and particle type by means of dimensionless superficial gas velocity (𝑈∗) and dimensionless particle diameter (𝑑𝑝 ∗) (𝑈∗and 𝑑𝑝 ∗ are derived from dimensionless particle Reynolds 𝑅𝑒𝑝 and Archimedes 𝐴𝑟numbers). 3.1 A brief introduction to fluidization 83 Figure 3. 4 – Grace diagram of fluidization regimes, from [233] As explained previously, this work is centered in the study of SE-SMR reaction. Due to the particularities of this process, BFB is most convenient reactor. In order to improve gas-solid mixing and to operate in the most stable regime the particles are ground to meet Geldart Bgroup. Thus, further explanation about hydrodynamics will be centered in these systems. Bubbling fluidized reactors are commonly divided in three parts: - Distributor which serves to homogenize the injection of fluidizing gas and to stop solids from flowing back during normal operation and shut down support the bed. This part ha as a considerable effect on proper operation of the fluidized bed. - Bubbling bed. The bed height depends on gas contact time, length to diameter (L/D) ratio needed to provide staging, space needed for internal heat exchangers and solids retention time. - Freeboard. The freeboard or disengaging height is the distance between the top of the fluid bed and the gas-exit nozzle. At least two actions can take place in the freeboard: classification of solids and reaction of solids and gases. When gas velocity is equal to minimum velocity, bubbles are generated at the exit of the distributor. The rising bubbles move gas and particles upward as well as a corresponding flow of particles downward. As a bubble reaches the upper surface of a fluidized bed, the gas breaks through the bubble geysering upward the solids. The downward pull of gravity and the upward pull of the drag force of the upward-flowing gas act on the particles. The larger and denser CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 84 particles return to the top of the bed, and the finer and lighter particles are carried upward. The distance above the bed at which the entrainment becomes constant is the transport disengaging height (TDH). Figure 3. 5 – Bubbling Fluidized Bed with entrained particles in the freeboard region above the dense bubbling bed, from [237] In order to model fluidized bed mechanics is crucial to describe accurately the behavior of bubbles to predict the gas-solid contact. Bubbles form at the distributor exit ports where fluidizing gas enters the bed. Bubbles are generated because the velocity at the interface of the bed just above the distributor holes represents a gas input rate in excess of what can pass through the interstices with a frictional resistance less than the bed weight and hence the layers of solids above the holes are pushed aside until they represent a void through whose porous surface the gas can enter at the incipient fluidization velocity. If the void attempts to grow larger the interface velocity becomes insufficient to hold back the walls of the void and hence they cave in from the sides cutting off the void and presenting a new interface to the incoming gas. This mechanism is illustrated in Figure 3. 6. Figure 3. 6 – Bubble formation mechanism, from [238] 3.1 A brief introduction to fluidization 85 The number, velocity and size of bubbles affect strongly the movement in the bed. A common approach to understand and attempt to model the behavior of the bed is to suppose that the aggregative fluidization consist in two phases: the bubble phase and the emulsion (o particulate phase). The flow rate through the emulsion phase is equal to the flow rate for minimum fluidization, and the voidage is essentially constant at εmf. Any flow in excess of that required for minimum fluidization appears as bubbles in the separate bubble phase. Figure 3. 7 – Bubbling fluidized bed hydrodynamics, , from [237] However bubble phase is more complex to hypothesize. The first complete model for bubble phase was developed in 1961 by Davidson [239]. This theory was capable of explaining many phenomena relating to bubbles in fluidized beds observed experimentally. This model supposes that the bubble is spherical and it is surrounded by a “cloud” of particles which serves simultaneously to change the motion of the bubble and to exchange properties with emulsion phase. Davidson solved these equations in terms of particle motion, the pressure distribution within the fluidizing fluid, the absolute velocities of the fluidizing fluid, and the exchange between the bubble and the particulate phase. He found that the geometry of the stream function is crucially affected depending whether the bubble velocity is larger or smaller than interstitial minimum fluidization velocity. The fluidizing fluid in this case moves downward relative to the bubble motion. The fluid flows past the fictitious cloud sphere. Inside the cloud, the fluid leaves the roof of the bubble and recirculates back to the base of the bubble as shown in Figure 3. 8. Obviously this model was very simplified and it has been modified and improved, for example to approximate better the form of particles and include the effect of bubble wakes (Figure 3. 8). CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 86 Figure 3. 8 – Bubble formation in a 2D BFB (left) [240]. Davidson fast bubble model scheme (right)[230]. Several of the approximations done by Davidson are still very useful to model the bubbling beds in micro and macro-scales. On the one hand, the velocity of bubble affects strongly the behavior of the bed (in terms of bed expansion and voidage). The rising velocity can be calculated directly from the bubble diameter and it is independent of the type of bed materials [241, 242]. On the other hand an empirical rule for estimate bubble grow was proposed to link bubble velocity and diameter. The bubbles in fluidized beds grow in size due primarily to four factors: the hydrostatic pressure on the bubbles decreases as they rise up the bed; the bubbles may coalesce by one bubble catching up with another; the bubbles which are side by side may coalesce; and, the bubbles may grow by depleting the continuous phase locally. Several models have been developed to predict the growth of bubbles as shown in Figure 3. 9 [238, 243, 244]. Figure 3. 9 – Bubble grow mechanism according with Darton et al. (left) and Zenz (right) models [229]. The first complete general two-phase model was proposed by Kunii and Levensipel [228], and it includes the effect of cloud and wake of the bubbles, more detailed analysis of conditions of emulsion and bubble phase, and freeboard effects. The K-L model has been implemented in engineering applications successfully, and, with some modifications, it is widely used nowadays (e.g. [245] or [246]). 3.2 CFD modeling of bubbling fluidized beds 87 3.2 CFD modeling of bubbling fluidized beds In the last decade, a paradigm shift has occurred in the study of fluidization from two-phase flow theory to computational fluid dynamics (CFD). An example of the acceptation of CFD in the fluidization community is the plenary lecture by Prof. J.A.M. Kuipers at FLUIDIZATION IX in 1998 [247]. CFD is a very powerful tool to simulate fluidized beds but present some disadvantages compared with global system models, as shown in Table 3. 1 [248] Table 3. 1 – Classification of the various models used for simulating dense gas-solid flow CFD Models Global System Models Advantage More exact solution available Phenomena follow from calculation a priori Simple models and simple solutions facilitate understanding After adjustment of parameters accurate macroscale prediction Disadvantage Detailed knowledge required about the elementary processes Macroscopic behavior not always accurately predicted Experimental validation and adjustment of parameters necessary Meaning of parameters sometimes unclear due to lumping Over the years, several research groups have developed a large number of numerical models in order to predict the behavior of gas-solid systems. From a generalist point of view there are two different approaches to model both phases: Eulerian and Lagrangian. The Eulerian approach adopt a continuum description of the phase, which is then governed by a Navier-Stokes-type equation. The Lagrangian approach considers the phase as a collection of a discrete number of particles that obey Newton’s law. According with these two options for each phase, the fluidized bed CFD models can be classified as shown in Table 3. 2 and Figure 3. 10 [249]. Table 3. 2 – Classification of the various models used for simulating dense gas-solid flow Name Gas phase Solid phase Gas-solid coupling Scale 1 Discrete bubble model Lagrangian Eulerian Drag closures for bubbles Industrial (10 m) 2 Two-fluid model Eulerian Eulerian Gas-solid drag closures Engineering (1 m) 3 Unresolved discrete particle model Eulerian (unresolved) Lagrangian Gas-particle drag closures Laboratory (0.1 m) 4 Resolved discrete particle model Eulerian (resolved) Lagrangian Boundary condition at particle surface Laboratory (0.01m) 5 Molecular dynamics Lagrangian Lagrangian Elastic collisions at particle surface Mesoscopic (<0.001 m) CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 88 Figure 3. 10Graphical representation of the models summarized in Table 3. 2. The grid indicates the scale on which the continuum phase is solved. The Discrete Bubble Model (model 1) is a very promising technique for very large reactors, but it is still under development, particularly for the problems to refined the rules of coalescence and break-up of the bubbles [250, 251]. In the Resolved Discrete Particle Model (model 4), particles motion is solved by means of a Lagrangian method, and gas phase is fully resolved, by means of Direct Numerical Simulations (DNS) at Kolmogorov scales, or by means of Lattice Boltzmann methods. This method is very expensive in terms of computational resources, and it is not fully developed due to the problems of implementing the stick boundary condition, in which the fluid velocity vanishes at the surface of the solid spheres [252]. In the Molecular Dynamics Model (model 5) both gas phase and solid phase are represented by particles. Thus gas-solid interactions are represented as collisions between gas particles with solid particles. This model can be implemented by Lattice Gas Cellular Automata’s, but it is useful only at very small scales, far from fluidized bed interest [253]. Therefore, the most useful methods, from an engineer point of view are the Unresolved Discrete Particle Model (also termed as Discrete Element Model, DEM, model 3) and the TwoFluid Model (TFM, also known as E-E model, model 2). DEM solves the solid flow by means of a Lagrangian approach, whilst the gas flow is solved by means of the classical Eulerian CFD approach (finite-difference/finite-element techniques). DE models are devoted to investigate particle-particle, particle-gas and particle-wall interactions, and their effects in mixing and formation of heterogeneous flow structures. Very interesting reviews of these methods can be found in [254] and [255]. Meanwhile TFMs abandons the concept of solid phase formed by a discrete number of particles, and it considers that both gas phase and solid are described as fully interpenetrating continua, using a set of generalized Navier-Stokes equations. TFM include indirectly the effects of particle-particle interactions by means of appropriate closure equations. 3.2 CFD modeling of bubbling fluidized beds 89 These two methods (TFM and DEM) have been extensively compared (e.g. [256-260]) with very good results compared with experiments, and comparable computational costs and accuracy at laboratory scale simulations. However, when the size of system increases to semiindustrial or industrial size the TFM model presents a notable computer saving with respect to DEM, that surpass the efforts to develop most efficient numerical methods for solving the Lagrangian phase [261]. Therefore, we used in this work the two-fluid model, embedding the kinetic theory of granular flow for particulate phase stress, for the numerical simulation of gas fluidization. 3.2.1 Kinetic Theory of granular flows TFM model solves governing equations of continuity and momentum equations for each phase. Thus, proper modeling is necessary to obtain accurate and better results. The coupling between the phases depends on the type of mixture being involved in the system. A common approach to close conservation equations is using kinetic theory in case of granular flows (KTGF). KTGF assumes that particle heat and mass transfer characteristics are due to nearly elastic random oscillations of particles suspended in fluids. Thus, solid phase presents a behavior similar to gases, and kinetic theory can be assumed with some modifications. A Maxwellian velocity distribution can be assumed for the particles. And the KTGF can be used for closure of the solid stress terms, with a drag function which serves to calculate the momentum exchange coefficients. Other constitutive equations are needed to close the granular-phase momentum equation. The energy dissipation due to the collision of inelastic particles is calculated with a particle temperature model. In KTGF the basic concept is the granular temperature, which is like thermal temperature in kinetic theory of gases. It measures the random oscillations of particles. The thermal temperature has a conversion factor from energy to degrees, the Boltzmann constant. For granular temperature, this constant is one. In granular flow particles dissipate energy due to inelastic collisions. Hence, equipartition of energy does not hold. Therefore, one usually defines the granular temperature as random kinetic energy per unit mass, which is just the average of the three squares of the velocity components in the three directions. Thus the granular temperature is the average of the three variances of the velocities of the particles. With an accurate modeling of the granular temperature equation, it is possible to model the random oscillations of motion and energy including turbulence, without the necessity of use a devoted closure equation for turbulence (as, for example, kε models). Thus KTGF model computes the viscosity from the granular temperature, with the principal input of effective restitution coefficient. Moreover, the solution of solid shear stress model includes the collisional, the kinetic and the frictional effects. The Kinetic Theory of Granular Flows was mainly developed during the 1980’s and early 1990’s by Dimitri Gidaspow at Illinois Institute of Technology, and a very complete description can be found in his book [262]. Based in this theory, Syamlal, Rogers and O’Brien, CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 90 from U.S. Department of Energy, developed and reformulated some parts of Gidaspow’s theory in an open-source code called MFIX [263, 264], which has been subsequently refined and implemented in CFD commercial codes such as FLUENT, being extensively used with very good results [265]. For a better understanding we summarized the governing and constitute equations of the KTGF. For simplicity, an isothermal non-reactive flow is supposed, thus the explanation is centered in the motion equations of the model. Considering s as solid phase (for simplicity we consider just one solid phase) and g as gas phase, and αs and αg are the volume fraction of each phase, the continuity and momentum equations are: Continuity equation for gas phase: ( ) ( ) 0=⋅∇+ ∂ ∂ ggggg v t  ραρα (3. 1) Continuity equation for solid phase: ( ) ( ) 0=⋅∇+ ∂ ∂ sssss v t  ραρα (3. 2) Momentum equation for gas phase: ( ) ( ) ( ) FvvKgpvvv tgsgsggggggggggg +−++⋅∇+∇−=⋅∇+ ∂ ∂ ραταραρα (3. 3) Momentum equation for solid phase: ( ) ( ) ( ) FvvKgppvvv t sggsssssssssssss +−++⋅∇+∇−∇−=⋅∇+ ∂ ∂ ραταραρα (3. 4) As explained before, the kinetic energy associated with the particle velocity fluctuations is represented by the granular temperature ( s Θ ) which is proportional to the mean square of the random motion of particles: ( ) ( ) ( ) ( ) ss sssssssssss kvIpv tΘΘ −Θ∇⋅∇+∇+−=      Θ⋅ ∇+Θ ∂ ∂ γταραρ  : 2 3 (3. 5) The constitutive equations of KTGF are used for determinate the stresses ( s τ ), viscosity ( s µ ), and pressure ( s p ) of the solid phase, that are dependent of the granular temperature. Several closure models have been developed for different groups and implemented in commercial codes [266]. However the most used are those by MFIX group [263]. If any model is from different author reference is specified. 3.3 Gas-solid reaction modeling 97 3.3 Gas-solid reaction modeling The performance of a reactor depends on the input, the kinetics, and the flow and contacting pattern of gas with solid. The kinetics can usually be evaluated in an experimental unit or can be gotten from literature. However, the flow and contacting has to be evaluated in the piece of equipment to be used. Figure 3. 12 –Information needed to relate output to input of a process reactor, from [285] As commented in previous section the flow and contacting pattern have been modeled by means of CFD two-fluid model with KTGF. This section is devoted to the implementation of kinetics of steam methane reforming and CO2 capture. Two kinetic schemes from literature have been adopted [138, 286], with the necessaries modifications for this work. 3.3.1 Steam Methane Reforming model Steam methane reforming reaction has been extensively studied due to the industrial interest of the process [287, 288]. Nowadays, practically the 95% of hydrogen is produced by means of methane steam reforming. Hydrogen is predominantly used for petroleum refining and the production of industrial commodities as ammonia, and it is proposed as an energy vector alternative to the fossil fuel based. Moreover, SMR is a very mature technology which presents the most competitive costs respect among all hydrogen production pathways [289]. Steam methane reforming is a heterogeneous catalyst reaction. Thus, reactions involved in the kinetic scheme of SMR are surface reactions. There are a widely variety of catalyst, however, the most extended commercially is the Ni-O, supported in alumina. Several heterogeneous kinetic schemes has been proposed [290-294], however, the most extensively used model was proposed in 1989 by Xu and Froment [286, 295]. This model supposes that there are three dominant reactions: 𝐶𝐻4 +𝐻2𝑂 ↔𝐶𝑂+ 3 𝐻2 , ∆𝐻298 0=206.2 𝑘𝐽 𝑚𝑜𝑙– R. 1 (3. 36) 𝐶𝐻4 + 2𝐻2𝑂 ↔𝐶𝑂2+ 4 𝐻2 , ∆𝐻298 0=165.0 𝑘𝐽 𝑚𝑜𝑙 – R. 2 (3. 37) 𝐶𝑂+𝐻2𝑂 ↔𝐶𝑂2+ 𝐻2 , ∆𝐻298 0=−41.2 𝑘𝐽 𝑚𝑜𝑙 – R. 3 (3. 38) CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 98 Two first reactions correspond to SMR to carbon monoxide and dioxide. Reaction 3 occurs contemporary to the other ones, and it corresponds to water-gas shift (WGS) reaction. The thermal global balance of the reactions is endothermic and, usually, a part of natural gas is burned to ensure the necessary heat for the reaction. In this work we proposed that this heat is provided by carbonation. The capture reaction provides the heat necessary and it enhances the equilibrium of reactions to the right, improving the SMR global performance. The kinetic scheme proposed by Xu and Froment includes the catalytic reactions (1-3) as: 𝑟1=𝑘1 𝑃𝐻2 2.5(𝐷𝐸𝑁)2·�𝑃𝐶𝐻4𝑃𝐻2𝑂− 𝑃𝐻2 3𝑃𝐶𝑂 𝐾𝐼� (3. 39) 𝑟2=𝑘2 𝑃𝐻2 3.5(𝐷𝐸𝑁)2·�𝑃𝐶𝐻4𝑃𝐻2𝑂 2− 𝑃𝐻2 4𝑃𝐶𝑂2 𝐾𝐼𝐼 � (3. 40) 𝑟3=𝑘3 𝑃𝐻2(𝐷𝐸𝑁)2·�𝑃𝐶𝑂𝑃𝐻2𝑂− 𝑃𝐻2𝑃𝐶𝑂 𝐾𝐼𝐼𝐼 � (3. 41) 𝐷𝐸𝑁= 1 + 𝐾𝐶𝑂𝑃𝐶𝑂 +𝐾𝐻2𝑃𝐻2+ 𝐾𝐶𝐻4𝑃𝐶𝐻4+𝐾𝐻2𝑂𝑃𝐻2𝑂 𝑃𝐻2 (3. 42) where the kinetic constants, kj for j=1,2,3, the equilibrium constants (KI, KII, KIII) and the adsorption constants, Ki for i= CH4, CO, H2O, H2 were evaluated experimentally as a function of temperature according with Arrhenius law. The reaction rate, equilibrium and the adsorption constants are, respectively: 𝑘1= 1.842 ×10−4exp �−240100 𝑅�1 𝑇−1 648��𝑘𝑚𝑜𝑙 𝑏𝑎𝑟0.5/𝑘𝑔𝑐𝑎𝑡 ℎ (3. 43) 𝑘2= 2.193 ×10−5exp �−243900 𝑅�1 𝑇−1 648��𝑘𝑚𝑜𝑙 𝑏𝑎𝑟0.5/𝑘𝑔𝑐𝑎𝑡 ℎ (3. 44) 𝑘3= 7.558 exp �−67130 𝑅�1 𝑇−1 648�� 𝑘𝑚𝑜𝑙 /𝑘𝑔𝑐𝑎𝑡 ℎ 𝑏𝑎𝑟 (3. 45) 𝐾𝐼= 4.707 ×1012exp �−224000 𝑅𝑇 �𝑏𝑎𝑟2 (3. 46) 𝐾𝐼𝐼 =𝐾𝐼·𝐾𝐼𝐼𝐼 𝑏𝑎𝑟2 (3. 47) 𝐾𝐼𝐼𝐼 = 1.142 ×10−2exp �37300 𝑅𝑇 � (3. 48) 𝐾𝐶𝐻4= 0.179 exp �38280 𝑅�1 𝑇−1 823��𝑏𝑎𝑟−1 (3. 49) 𝐾𝐻2𝑂= 0.4152 exp �−88680 𝑅�1 𝑇−1 823�� (3. 50) 𝐾𝐶𝑂 =40.91 exp �70650 𝑅�1 𝑇−1 648��𝑏𝑎𝑟−1 (3. 51) 𝐾𝐻2= 0.00296 exp �82900 𝑅�1 𝑇−1 648��𝑏𝑎𝑟−1 (3. 52) 3.3 Gas-solid reaction modeling 99 The reaction rate of the species in the solid phase also depends of the pore diffusion resistance and surface phenomena. Di Carlo et al. [296] estimated effectiveness factors ( η j) in order to model in a simple way these effects, being for the reactions η 1 = 0.7, η 2 = 0.4 and η 3 = 0.8. Therefore, in order to include the intrapore-diffusion and the dispersion of the catalyst in the bed, the effective kinetics of the reactions 𝑟𝑗∗ (for j=1,2,3) are: 𝑟𝑗∗=𝜂𝑗𝜌𝑐𝜀𝑐𝑟 (3. 53) Where 𝜌𝑐is the density of catalyst and 𝜀𝑐 is the volume fraction of catalyst. This heterogeneous catalytic kinetic scheme has been implemented in a commercial CFD code (Fluent 6.3), by means of a C programmed User Defined Function (UDF1). 3.3.2 Carbon dioxide capture model The reaction of CaO carbonation takes place simultaneously to SMR. This heterogeneous reaction also implies the change of phase of CO2 during the reaction. The mass transfer phenomena affect the bed fluid dynamics due to density changes in gas and solid phase. Several kinetic models have been proposed in the technical literature in order to describe the gas-solid reaction, such as the well-known uniform and un-reacted shrinking core models. However, Stendardo and Foscolo [138] noted that these models do not predict accurately the conversion process of calcium oxide, and proposed a novel grain model. In such a model the molar rate of CaO conversion per unit volume of particle can be represented as a function of the local conversion, X, of calcium oxide and the average radius of calcium oxide grain δCaO: 𝐶𝑎𝑂+𝐶𝑂2 ↔𝐶𝑎𝐶𝑂3 , ∆𝐻298 0=−178 𝑘𝐽/𝑚𝑜𝑙 (3. 54) 𝑑𝑋 𝑑𝑡 = 𝜎0,𝐶𝑎𝑂 𝑘𝑠 (1−𝑋)2/3�𝐶𝐶𝑂2−𝐶𝐶𝑂2,𝑒𝑞� 1 + 𝑘𝑠 𝑁𝐶𝑎 2𝐷𝑃𝐿 𝛿𝐶𝑎𝑂√1−𝑋 3 �1−�1−𝑋 1−𝑋+𝑋𝑍 3 � (3. 55) We observed that, equation 3.55 depends on several parameters. 𝑁𝐶𝑎 is the number of moles of calcium carbonate per unit of volume of sorbent particle (in kmol/m3). 𝛿𝐶𝑎𝑂 is the average diameter of CaO grains in the sorbent (in m). 𝑍 is dimensionless molar volume ratio of calcium carbonate to calcium oxide VCaCO3/VCaO. 𝐶𝐶𝑂2 is the molar concentration of CO2 (in kmol/m3) and 𝐶𝐶𝑂2,𝑒𝑞 is the equilibrium concentration of carbon dioxide as a function of temperature according with [137]: 1 A UDF, is a function that you program that can be dynamically loaded with the FLUENT solver to enhance the standard features of the code. CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 100 𝑃𝐶𝑂2,𝑒𝑞 = 4.137 × 107exp �−20474 𝑇� 𝑎𝑡𝑚 ; 𝐶𝐶𝑂2,𝑒𝑞 = 𝑃𝐶𝑂2,𝑒𝑞 𝑅𝑇 (3. 56) The initial specific active surface, 𝜎0,𝐶𝑎𝑂, is proportional to the number of calcium containing grains per unit particle volume, and to the “active” surface of each spherical grain, which corresponds to the volume occupied by un-reacted calcium oxide. This relation establishes the dependency of the specific active surface on the physical and chemical properties of the sorbent, the particle grain size, and the calcium oxide conversion: 𝜎 𝐶𝑎𝑂 = 𝜎 0,𝐶𝑎𝑂 (1−𝑋)2/3= 𝑁𝐶𝑎 𝑉𝐶𝑎𝑂 𝜋 6𝛿𝐶𝑎𝑂 3×𝜋𝛿 𝐶𝑎𝑂 2 (1−𝑋)2/3 (3. 57) The intrinsic rate constant of CaO-CO2 reaction, 𝑘𝑠, can be determined experimentally supposing reaction order n=0 with two complementary derivations from grain model under kinetic control proposed by Szeleky et al. [213]. In one hand, Bhatia and Permutter [201] supposed zero activation energy in the temperature range (823-998 K) and obtained a constant value of k = 5.97·10-7 m4/kmol s. 𝑘𝑠=𝑘 /𝑁𝐶𝑎 (3. 58) On the other hand Sun et al.[214] proposed an Arrhenius law for 𝑘𝑠 for dolomites : 𝑘𝑠=𝐴0exp �−𝐸 𝑅𝑇�;𝐴0= 1.04 × 10−3 𝑚𝑜𝑙 𝑚2𝑠; 𝐸=24 ± 6 𝑘𝐽/𝑚𝑜𝑙 (3. 59) As explained in section 2.2.4, 𝑘𝑠 and 𝑆0 are correlated by means of reaction rate according with: ln 𝑟𝑜=ln(56 𝑘0𝑆0/3)−𝐸 𝑅𝑇 (3. 60) Thus, it is possible suppose that, for a sorbent in which molar ratio of CaO, 𝑘0 and E are constant, reaction rate is directly proportional to surface area, under kinetic control regime: 𝑟0 ∝ 𝑆0 (3. 61) Therefore, it seems reasonable introduce an effectiveness correction factor in the kineticallycontrolled reaction which allows to introduce the effect of variation of surface area in the reaction rate. Supposing as reference case the conditions of first calcination and carbonation 3.3 Gas-solid reaction modeling 101 used by Bhatia and Permutter and considering that 𝑘𝑠 is constant; and applying the model of Sun et al. it is possible to obtain 𝑆0,𝑟𝑒𝑓 (for dolomite in this study corresponds to R.1 in table 2.3, 𝑆0,𝑟𝑒𝑓 =19.3 𝑚2/𝑔). For different cases, as cycled and/or pre-treated sorbents, the effective intrinsic rate constant 𝑘′𝑠 is defined as: 𝑘′𝑠=𝑘𝑠 𝑆0 ′ 𝑆0,𝑟𝑒𝑓 (3. 62) where 𝑆0 ′ is the surface area of sorbent, obtained from TGA data applying the Sun et al. method. Finally, in order to define completely the Stendardo and Foscolo carbonation reaction kinetics (Eq. 3.55) it is necessary consider the diffusion of carbon dioxide though the product layer of calcium carbonate that, at more advanced stages of the process, comes to occupy the whole external region of the grains, thereby preventing direct contact between carbon dioxide in the pore space and the active, unconverted calcium oxide surface. The coefficient of diffusion thorough the product layer defined (𝐷𝑃𝐿) as: 𝐷𝑃𝐿(𝑋)= 𝐷𝑃𝐿,0exp (−𝑎·𝑋𝑏) (3. 63) where 𝐷𝑃𝐿,0 is the initial value of carbon dioxide solid diffusion coefficient (we have assumed a reasonable initial solid diffusivity of 𝐷𝑃𝐿,0=2×10-5 m2/s) and a and b are fitting parameters (a is a decay parameter). By fitting the grain model to the experimental data, we can obtain, for the final stage of the particle carbonation process, the best estimation for the solid state diffusion coefficient of carbon dioxide through the product layer [297] (Table 3. 3). Table 3. 3 DPL at 2500 s (temperature 700 °C). X N D PL [-] [m 2 /s] 1st 0.868 1.82×10 -17 2nd 0.810 1.77×10 -17 3rd 0.739 1.79×10 -17 4th 0.675 1.78×10 -17 In Table 3. 4, these estimatations are summarized for the first cycle of the multi-cycling carbonation at different temperatures. These values represent the diffusivity through the product layer which is located at the outer surface of the particle where the resistances due to the diffusion are greater compared with the inner part of the particle. Thus, for values of DPL lower than shown in this table is possible to suppose that reaction is under kinetic regime. CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 102 Table 3. 4 Model result for dolomite carbonation at different temperatures Temperature D PL [°C] [m2/s] First cycle 500 1.01×10 -16 600 1.99×10 -15 700 3.45×10 -14 Arrhenius type dependence was envisaged for the product layer diffusion coefficient. Figure 3. 13 displays the Arrhenius plot for the values reported in Table 3. 4 to obtain, by a least square method, the pre-exponential factor and the activation energy. The values are reported in Eq. (1). Figure 3. 13 – DPL exponential decay behavior as function of temperature Arrhenius plot for the diffusion through the product layer The activation energy found in this work (181.62 KJ/mol) is consistent with that calculated, by Bhatia and Perlmutter [201] (179.20 KJ/mol). Thus the diffusion coefficient through the product layer can be written as a function of the temperature in the following manner: 𝐷𝑃𝐿(𝑇)= 1.01 ·10−16exp �181620 𝑅�1 773 −1 𝑇�� (3. 64) This higher activation energy is likely due to solid state lattice diffusion process. It is worth noticing that, as reported in Mess et al. [212], the diffusion coefficient through a single crystal of CaCO3 assumes a high value: 352 kJ/mol. As a consequence, the presence of impurities and defects in the calcite lattice could be envisaged. -39 -37 -35 -33 -31 -29 -27 -25 0.001 0.00105 0.0011 0.00115 0.0012 0.00125 0.0013 0.00135 1/T [K-1] ln(DPL) 3.3 Gas-solid reaction modeling 103 3.3.2.1 Implementation of CO2 capture model In an operating plant, it is likely that most interest would be on the kinetically controlled regime, during which most of the achievable uptake of CO2 occurs. This regime is dominant at the beginning of the CO2 capture, typically for 1-2 min for a single particle. However, in a BFB, this time spreads over minutes (or hours) as function of the amount of solids, because all particles are not exposed to CO2. During this phase, the coefficient of diffusion thorough the product layer defined (DPL) is very large. Under these conditions, diffusional effects can be neglected. Thus, in the general equation of capture reaction rate (Eq. 3.55), denominator becomes 1. And, under kinetically controlled regime Eq. 3.55 becomes: 𝑑𝑋 𝑑𝑡 =𝜎0,𝐶𝑎𝑂 𝑘′𝑠 (1−𝑋) 2/3 �𝐶𝐶𝑂2−𝐶𝐶𝑂2,𝑒𝑞� (3. 65) Therefore, this equation was implemented in FLUENT, by means of a UDF. This assumption can be considered valid for two reasons. On one hand, this simplified reaction rate improves the convergence of numerical simulations. In the other hand, CFD simulations are strongly time intensives. The simulation of several minutes of BFB operation needs months of wall-clock time. Thus, CFD fluidized bed simulations are limited at few minutes, in which diffusional effects are negligible and kinetically controlled regime dominates the reaction. In this work two different sorbents were studied, dolomite and CaO-mayenite. Moreover it is interesting to study the effect of treatments in the kinetic-controlled phase (see Chapter 2). Thus the effective kinetic constant rate has been included. Table 3. 5 summarize the values of parameters needed for the modeling of carbonation reaction. CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 104 Table 3. 5 Values of parameters used in simulation for dolomite and mayenite sorbents Dolomite Mayenite Parameter Units CaO.MgO (1:1) CaO·Ca 12 Al 14 O 33 (85:15%) Moles of Ca per unit volume of sorbent particle NCa (kmol/m3) 16.21 a 5.86 c CaO molar volume VCaO (m3/ kmol) 1.69·10-2 a 2·10-2 c Intrinsic rate constant ks (m4/ kmol s) 5.97·10-7 a 5.97·10-7 c Reaction constant k (m/s) 9.68·10-6 a 3.5·10-6 c Grain diameter δ CaO (m) 1·10-6 b 0.7·10-6 c Surface area, Reference S0,ref (m2/g) 19.3 2.6 d Surface area, Non-treated sorbent S’0 (m2/g) 19.3 2.6 d Surface area, Treated sorbent S’0 (m2/g) 23.6 4.0 d a from [138] b from [129] c from [141]d from [215] Under these conditions the carbonation has been compared with TGA data with good agreement as shown in Figure 3. 14. This approach can be considered correct for the first seconds of carbonation reaction, when diffusional effects are negligible, and kinetics dominates the reaction rate. Figure 3. 14 – Comparison of experimental and simulated conversion of sorbents 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 0,4 0 5 10 15 20 1-(1-X)1/3 Time (s) Dolomite-Stand-Exp Dolomite-Stand-Model Dolomite-Trigged-Exp Dolomite-Trigged-Model Mayenite-Activated-Exp Mayenite-Activated-Model Mayenite-NO-Activated-Exp Mayenite-NO-Activated-Model 3.4 Preliminary hydrodynamic simulations 105 3.4 Preliminary hydrodynamic simulations Previously to the simulation of the complete SE-SMR system, two numerical simulation campaigns were carried out in order to establish the best framework to simulate accurately the hydrodynamic behavior of the fluidized beds. On one hand, a 90kWth oxy-fuel test rig (the CIRCE combustor) was simulated, under no reactive conditions, and results compared with experimental data and empirical models [298, 299]. On the other hand, the ZECOMIX reactor was simulated under CO2 capture regime, without including steam methane reforming reactions. The aim of this section is to present the highlights of the CFD simulations, as a practical guide to simulate accurately a BFB, using the commercial CFD code FLUENT 6.3. More than 100 simulations were carried out. Thus, we will show the most useful results. For more detailed information we refer to references [300-304]. 3.4.1 Numerical method and boundary conditions Non-axisymmetric or axisymmetric two dimensional domains were assumed in the simulation for simplicity. Initially, in the case of CIRCE combustor simulations a non-axisymmetric domain was studied. In the ZECOMIX reactor, due to present a reactor diameter five times higher than CIRCE one, an axisymmetric approach has been adopted. The same numerical methods was applied to the simulation of CIRCE and ZECOMIX reactors. An Eulerian-Eulerian description was adopted to resolve the coupling of gas and particulate phase, and it was modeled by solving the Unsteady-state Reynolds Averaged Navier-Stokes equations (URANS) along the computational domain. Non-slip conditions were selected for gas phase and free-slip for solid phase, at the walls. Isothermal simulations were carried out. In this case, isothermal term refers to the fact that energy conservation equation is not considered during the calculation. Isothermal conditions are exact for non-reacting bed. Moreover, in the cases of reacting beds is quite approximate due to a high mixing rate, and because in the case of SE-SMR, the energy of reactions are balanced. The second order QUICK scheme was used to evaluate the convective terms because it presents less numerical diffusion than first order schemes, improving the bubbles resolution [305]. The pressure-velocity coupling is resolved by means of the phase-coupled SIMPLE (PC-SIMPLE) algorithm[306]. The residual tolerances were set of 10-3 for each scaled residual. The bed was assumed to be at minimum fluidization condition at the beginning of the simulation. The initial solid volume fraction (εs) was 0.55 and the maximum value was set at 0.63. The value of restitution coefficient between particles was 0.9 [307], which represent a realistic behavior of Geldart B particles. Table 3. 6 lists the main conditions used in the both simulation campaigns (CIRCE and ZECOMIX). Moreover, CFD simulations of BFB are very sensitive to closure models. Table 3. 7 shows the closure models of KTGF evaluated. All these models are available by default in FLUENT. CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 106 Table 3. 6 Simulation parameter constants Flow field Unsteady RANS Energy Equation Isothermal Multiphase model E-E KTGF Closure Pressure-velocity coupling PC-Simple Discretization 2nd order QUICK BC Constants Initial solids packing 0.55 Maximum solids packing 0.63 Restitution coefficient (e ss ) 0.9 Inlet boundary conditions Velocity Outlet boundary conditions Pressure Time steps (s) 0.001 - 0.0001 Convergence criteria 10-3 Table 3. 7 Closure model of KTGF evaluated Models evaluated Turbulence Laminar / k-ε standard Drag model Gidaspow/ Syamlal-O’Brien/Wen-Yu Solid pressure Lun/ Syamlal-O’Brien Lun/ Syamlal-O’Brien Radial distribution Granular kinetic viscosity Gidaspow/ Syamlal-O’Brien Frictional Viscosity None /Schaeffer Johnson/KTGF/ Syamlal-O’Brien Frictional Pressure Boundary conditions are strongly different for the CIRCE oxy-fuel reactor (OF) and ZECOMIX reactor (ZC). On one hand, both are cylindrical, but diameter and height are different, as well as particle diameter. Thus, both are Geldart B BFB, but scales are different. On the other hand, there are experimental data of OF operation; meanwhile there are not data of ZC reactor, so the results obtained in OF can be compared quantitatively and ZC just qualitatively. Table 3. 8 and summarizes the boundary conditions used in the simulations. 3.4 Preliminary hydrodynamic simulations 113 3.4.3 Preliminary simulations of ZECOMIX reactor The first approach to simulate the carbonator reactor of ZECOMIX plant was based in the specifications provided by [310]. As a first approach, a two dimensional rectangular and axysymmetrical section formed by 3000 cells was studied [311]. This simplification supposes a reduction of computational cost to evaluate the effect of closure equations and the implementation of CO2 capture gas-solid reaction. Firstly the Gidaspow (G) and Siamlal-O’Brien (S-O) drag closures were compared. For the first seconds there are not observed differences in the models. However, before the stabilization of the bed (usually after 5 sec [256]) the resolution of formation of bubbles in the case of G model is lower than in S-O model. Moreover in the case of G model there is observed a strong densification in the central zone. Figure 3. 24 – Solid phase fraction at t =10s with Gidaspow (left) and Syamlal-O’Brien (right) drag functions Initially, turbulence-free flow was assumed throughout the bed, ignoring the possibility of local regions of turbulence. In general, accounting for time-averaged turbulent behavior and turbulent interactions between phases can make simulation predictions more realistic for beds operating at high Reynolds number. However, unless an appropriate turbulence model with the correct empirical constants and closures is chosen, the model predictions may be less consistent with experimental data than the turbulence-free model [312]. Ding and Gidaspow included the turbulence effects in the definition of Granular Temperature (Eq. 3.5) with good results, demonstrating that it is not necessary include a turbulence model to obtain accurate predictions in dense beds [313]. This approximation has been extensively used (e.g. [307, 314-317]) and it has been applied in this work. We compare in Figure 3. 25 the effect of turbulence, obtaining no significant global differences. By using k-ε turbulence model has been observed a lower bubble definition and an increase in computational costs. Thus, we assume that turbulence can be neglected and the turbulence models available in FLUENT are not used. CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 114 Figure 3. 25 – Volume of fraction of solids at t =20s with Syamlal-O’Brien drag functions without turbulence model (left) and with k-ε model (right) This 2D axisymmetric grid of 3000 cells do not presents serious problems with the convergence. However the hydrodynamics of the systems is unrealistic. On the one hand the formation of bubbles is not well predicted when simulation time increases. On the other hand, the circulation pattern observed is like that presented in Figure 3. 27a. However for U>10 Umf it can be expected the “gulf stream” effect, like that presented in Figure 3. 27b. Figure 3. 26 – Solid phase fraction with time in preliminary 2D axysymmetric simulation ZECOMIX reactor Figure 3. 27 – Movement of solids in BFB. a)Aspect ratio(h/d) <1, U low b)Aspect ratio <1,U high, from [228] 3.4 Preliminary hydrodynamic simulations 115 Applying the same grid size that the 2D axisymmetric but not imposing symmetric conditions (increasing grid size from 3000 to 6000 cells), the simulations were repeated. More realism in the bubble formation and in the solids circulation (with “gulf stream” behavior) is observed (see Figure 3. 28). Pressure drop is independent of the mesh type. Thus, 2D-axysymmteric grids are not recommendable BFB (but it can be a good approach to spouted beds [311, 318, 319]). Figure 3. 28 – Solid phase fraction at t =10s with a 6000 cells non-axisymmetric mesh Wang et al. [279] proposed an optimal cell size for BFB of 2-4 particle diameter. With this cell size, mesh size increase enormously (until 1 million cells). Pressure drop is independent of the grid size but the resolution of bubbles disappears. Figure 3. 29 – Solid phase fraction at t =10s with a 1,000,000 cells non-axisymmetric mesh CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 116 Finally, based in the CIRCE combustor case, a simulation including a perforated plate as gas distributor has been carried out. A slightly better bubble formation has been obtained in the opr of the bed, respect to the 6000 cells case, and a sensible improvement respect to the 1 million cells case. The number of the cells in the mesh must increase to include jet effect in the holes of the perforated plate (30.000 cells, x5 with respect to uniform inlet). However, calculation time is reasonable and global pressure drop is not altered. Figure 3. 30 – Solid phase fraction at t =10s with a 30,000 cells non-axisymmetric-detailed-distributor mesh. In conclusion, according to these findings we will assume a 2D non-axisymmetric coarse grid, with no turbulence model and the Syamlal-O’Brien model for drag in the complete SE-SMR simulations. 3.5 Laboratory Scale SE-SMR simulations 117 3.5 Laboratory Scale SE-SMR simulations 3.5.1 State-of-the-art of SE-SMR process Nowadays, Sorption-Enhanced Steam Methane Reforming (SE-SMR) process is still under development. Currently, there is any industrial plant which produces hydrogen with this technology. Thus, all literature about SE-SMR refers to laboratory scale experiments or models. Instead of the first attempts proposed by Brun-Tsekhovoi et al. [320],the group of D.P. Harrison, from Louisiana State University, was the first to propose the SE-SMR process as an alternative to classical SMR process [88]. They demonstrated experimentally the process at medium pressure (15 atm), 500-600ºC, and with a typical steam-carbon ratio (S/C) of 4, in fixed bed using Ni-based catalyst and sorbents as dolomite [321], CaO from Ca(OH)2 precursor [322], and under multicyling operation [323]. Other groups have continued working in SE-SMR at atmospheric pressure, demonstrating the continuous operation [324]; proposing other sorbents as hydrotalcites [325] or CaO pellets[326]; developing hybrid particles or pellets which contains Ni-based catalyst and CO acceptor simultaneously as CaO-Ca12Al14O33 [196], hydrotalcite and CaO-MgO-based sorbents [327]. Johnsen et al. [328] were the first to demonstrate the SE-SMR process in BFB, using separately particles of Ni catalyst and dolomite, at atmospheric pressure and 600ºC. These experiments were reproduced by Di Carlo et al. [296] with slight differences. Moreover, Di Carlo et al. simulated by means of a commercial CFD code their experiments with good agreement. Recently Chen et al. extended this model to hydrotalcites [329]. Modeling of SE-SMR reactors have been also studied at industrial scale with fixed beds [330, 331], with bubbling beds [332-334]. In the modeling works on SE-SMR it is remarkable the work of the group of H.A. Jakobsen, from Norwegian University of Science and Technology (NTNU), which has developed K-L and in-house CFD two-fluid models for SE-SMR process in bubbling and circulating fluidized beds, and hybrid pellets intraparticle diffusion models (e.g. [197, 245, 335-338]). However, these models have not been validated experimentally at full scale. 3.5.2 Numerical simulations of a laboratory scale SE-SMR reactor Previously to the full-scale simulation of SE-SMR ZECOMIX reactor, the model proposed in previous sections has been validated with literature data. In order to validate the model, the most relevant operational conditions of ZECOMIX reactor [310] have been compared with experimental data reported by Johnsen et al.[328] and Di Carlo et al. [296] (as shown ). CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 118 Table 3. 9 Operational conditions of lab scale case studies JOHNSEN DI CARLO CFD Reactor diameter (m) 0.1 0.08 0.1 Reactor height (m) 1 0.6 1 Dolomite particle diameter (μm) 125-300 180-425 225 Ni-Catalyst particle diameter (μm) 150-250 180-425 225 Dolomite particle density (kg/m3) Calcined / No calcined N.R. 1580 / 2300 1550 / 2200 Ni-Catalyst particle density (kg/m3) N.R. 2100 2100 Bulk density of the mixture (kg/m3) 1300 1000-1260 1040 Static bed height (m) 0.3 0.2 0.3 Reforming temperature (ºC) 600 600-700 600 Superficial gas velocity (m/s) 0.032 - 0.096 0.15 - 0.3 0.1-0.2 Dolomite-to-catalyst mass ratio (dimensionless)Dol/Cat 0.4 4-5 2 Steam-to-carbon molar feed ratio (dimensionless) S/C 3 4 4 In order to validate the model, three simulations were carried out. 1 minute of operation time has been simulated in a single 2.2 GHz CPU, using FLUENT 6.3. A 4000 uniform-square-cell mesh (5x5mm) has been used for the simulations (Figure 3. 31). Gas composition was fixed to a S/C=4 (20 %CH4, 80%H2O). Calcined dolomite is selected as CO2 acceptor. A first standard calcination is supposed (kinetic parameters are listed in Table 3. 5). Two different velocities (0.1 and 0.2 m/s) and two different drag models (Syamlal-O’Brien and Modified-Wang) were studied. Simulated time was one minute, at least. The other simulation parameters are listed in Table 3. 10. Figure 3. 31 – Schematic view of the 2D-BFB studied and mesh used 3.5 Laboratory Scale SE-SMR simulations 119 Table 3. 10 Numerical parameters of CFD simulations Flow field Unsteady RANS Energy Equation Isothermal Multiphase model E-E KTGF Closure Pressure-velocity coupling PC-Simple Discretization 2 nd order QUICK BC Constants Superficial gas velocity 0.1 - 0.2 m/s Gas Composition 20 %CH4, 80%H2O Grid number cells 4000 - Uniform 5x5mm Distributor details Uniform Initial solids packing 0,6 Maximum solids packing 0.63 Restitution coefficient (ess) 0.9 Inlet boundary conditions Velocity Outlet boundary conditions Pressure Time steps (s) 0.001 Convergence criteria 10-3 Closure models Turbulence Laminar Drag model Syamlal-O’Brien / Modified Wang Solid pressure Syamlal-O’Brien Radial distribution Lun et al. Granular kinetic viscosity Syamlal-O’Brien Frictional Viscosity Schaeffer Frictional Pressure Syamlal-O’Brien A hybrid particle approach has been used in the simulations. Dol/Cat ratio was fixed in 2, thus the particles are formed by 1/3 of catalyst, and 2/3 of dolomite. As dolomite presents a CaO/MgO ratio of 1:1, the composition of particle is 1/3 catalyst, 1/3 CaO and 1/3 MgO (inert). Moreover, as demonstrated by Di Carlo, there is not observed segregation of the phases using different particle phase for catalyst and sorbent (if velocity is at least 5 times Umf). Hence, only one Eulerian solid phase can be assumed. Furthermore if Dol/Cat ratio is up to 2, the carbonation reaction provides the heat necessary for steam methane reforming, obtaining an autothermal process. This fact validates the isothermal simulation hypothesis. The simulations present good agreement with values reported by Di Carlo (Figure 3. 32). The hydrogen mole fraction obtained under the different experiments is ~ 0.92, thus chemical equilibrium is not achieved as in Johnsen experiments (~0.98). The lack of equilibrium is supported by the incomplete conversion of methane (~0.98) and incomplete CO2 capture (CO2 outlet mole fraction ~ 0.02). The species distribution across the bed is shown in Figure 3. 33. CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 120 Figure 3. 32 – Methane conversion and H2, CH4, CO and CO2 mole fraction in dry basis Figure 3. 33 – CaO conversion (in terms of CaCO3 conversion) and H2, CH4, CO, CO2 and H2O distribution species in gas phase for t=30s U=0.2m/s 3.5 Laboratory Scale SE-SMR simulations 121 Equilibrium is not achieved mainly due to the relatively high gas velocity, which reduces residence time of the gas in the fluidized bed. If residence time is not enough, CO2 capture is not complete. As shown in Figure 3. 34, on the one hand, the velocities of reforming reactions (R.1-R_CO Eq. 3.36; R.2-R_CO2 Eq. 3.37) are very high and they occur in the bottom of the FB (under 0.1 m); and the water-gas shift reactions (R.3-R_WGS Eq. 3.39) is practically in equilibrium. On the other hand, the CO2 capture reaction rate is higher in the lower bottom of the reactor, but it spreads over the entire reactor, with fluctuations due to the bubbles. The CO2 capture reaction affects majorly to reforming conversion to CO2 (R_CO2) and, in a lower degree to reforming to CO (R_CO). For example the fluctuations in which capture increases, R_CO2 also increases (marked near 0.2m height in Figure 3. 34). Thus CO2 capture is the limiting step of the SE-SMR reaction. Figure 3. 34 – Reaction rate of SMR (a) and CO2 capture (b) bed height. U=0.1 m/s There are slight differences in the species prediction between the cases in which SymalalO’Brien (S-O) and the Modified-Wang (M-W) drag models have been used. As shown in Figure 3. 35 bed behavior at U=0.1 is very similar using S-O and M-W models. Comparable bed expansion and volume fractions have been obtained, with a difference in the resolution of some bubbles. Moreover, applying a Fast Fourier Transform (FFT) to the transient pressure drop signal the spectra are very similar, with a slight increase in the magnitude. Thus in terms 0 2 4 6 8 10 12 14 16 18 20 00,1 0,2 0,3 0,4 0,5 Reaction Rate (mol/ m3 s) Bed Height (m) R_CO R_CO2 R_WGS 0 1 2 3 4 5 6 7 00,1 0,2 0,3 0,4 0,5 CO2 Cpature rate (mol/m3s) Bed height (m) x=0,25m x=0,5m x=0,75m a) b) CFD MODELING OF A SORPTION ENHANCED - STEAM METHANE REFORMING REACTOR 122 of macroscale and microscale bed behavior there are not strong differences, and, therefore there are not differences in the species distribution. However, there are bigger differences in the response of the system under U=0.2 m/s. Figure 3. 35 – Volume fraction of solids at t=30 s Figure 3. 36 – Power spectra of simulated U=0.1 m/s with S-O and M-W drag models The increas of velocity affects strongly the behavior of the SE-SMR reactor. On the one hand, methane conversion and hydrogen volume fraction present a relative reduction of ~ 2%, CO volume fraction increases ~ 35% and CO2 concentration decreases ~ 25%. These variations are due to the lower residence time of methane in the bed, and, to the formation of larger bubbles, BIBLIOGRAPHY 225 [316] X. Gao, C. Wu, Y.-w. Cheng, L.-j. Wang, X. Li, Experimental and numerical investigation of solid behavior in a gas–solid turbulent fluidized bed, Powder Technology, 228 (2012) 1-13. [317] S.D. Sharma, Numerical Simulation of Catalytic Partial Oxidation of Methane to Synthesis Gas in a Bubbling Fluidized Bed Reactor Proceedings "6th international symposium on Catalysis in Multiphase Reactors" - Pune (India), (14-17 January 2007). [318] S.H. Hosseini, G. Ahmadi, M. Olazar, CFD simulation of cylindrical spouted beds by the kinetic theory of granular flow, Powder Technology, 246 (2013) 303-316. [319] G.K. Khoe, Mechanics of Spouted Beds, University Press, 1980. [320] A.R. Brun-Tsekhovoi, A.N. Zadorin, Y.R. Katsobashvili, S.S. Kourdyumov, The process of catalytic steam reforming of hydrocarbons in the presence of carbon dioxide acceptor, in: Hydrogen Energy Progress VII: Proceedings of the 7th World Hydrogen Energy Conference 1988, pp. 885-900. [321] C. Han, D.P. Harrison, Simultaneous shift reaction and carbon dioxide separation for the direct production of hydrogen, Chemical Engineering Science, 49 (1994) 5875-5883. [322] B. Balasubramanian, A. Lopez Ortiz, S. Kaytakoglu, D.P. Harrison, Hydrogen from methane in a single-step process, Chemical Engineering Science, 54 (1999) 3543-3552. [323] A. Lopez Ortiz, D.P. Harrison, Hydrogen Production Using Sorption-Enhanced Reaction, Industrial & Engineering Chemistry Research, 40 (2001) 5102-5109. [324] Z.-s. Li, N.-s. Cai, J.-b. Yang, Continuous Production of Hydrogen from SorptionEnhanced Steam Methane Reforming in Two Parallel Fixed-Bed Reactors Operated in a Cyclic Manner, Industrial & Engineering Chemistry Research, 45 (2006) 8788-8793. [325] Y. Ding, E. Alpay, Adsorption-enhanced steam–methane reforming, Chemical Engineering Science, 55 (2000) 3929-3940. [326] D.K. Lee, I.H. Baek, W.L. Yoon, Modeling and simulation for the methane steam reforming enhanced by in situ CO2 removal utilizing the CaO carbonation for H2 production, Chemical Engineering Science, 59 (2004) 931-942. [327] M. Broda, V. Manovic, Q. Imtiaz, A.M. Kierzkowska, E.J. Anthony, C.R. Müller, HighPurity Hydrogen via the Sorption-Enhanced Steam Methane Reforming Reaction over a Synthetic CaO-Based Sorbent and a Ni Catalyst, Environmental Science & Technology, 47 (2013) 6007-6014. [328] K. Johnsen, H.J. Ryu, J.R. Grace, C.J. Lim, Sorption-enhanced steam reforming of methane in a fluidized bed reactor with dolomite as -acceptor, Chemical Engineering Science, 61 (2006) 1195-1202. BIBLIOGRAPHY 226 [329] Y. Chen, Y. Zhao, C. Zheng, J. Zhang, Numerical study of hydrogen production via sorption-enhanced steam methane reforming in a fluidized bed reactor at relatively low temperature, Chemical Engineering Science, 92 (2013) 67-80. [330] G.-h. Xiu, P. Li, A. E. Rodrigues, Sorption-enhanced reaction process with reactive regeneration, Chemical Engineering Science, 57 (2002) 3893-3908. [331] Y.-N. Wang, A.E. Rodrigues, Hydrogen production from steam methane reforming coupled with in situ CO2 capture: Conceptual parametric study, Fuel, 84 (2005) 1778-1789. [332] K. Johnsen, J.R. Grace, S.S.E.H. Elnashaie, L. Kolbeinsen, D. Eriksen, Modeling of Sorption-Enhanced Steam Reforming in a Dual Fluidized Bubbling Bed Reactor, Industrial & Engineering Chemistry Research, 45 (2006) 4133-4144. [333] G.C. Koumpouras, E. Alpay, F. Stepanek, Mathematical modelling of low-temperature hydrogen production with in situ CO2 capture, Chemical Engineering Science, 62 (2007) 28332841. [334] S. Stendardo, P.U. Foscolo, Modelling of a multi-particle reactor for carbon dioxide capture and hydrogen production, Proceedings "5th Conference on Clean Coal Technologies " - Zaragoza (Spain), (8-12 May 2011). [335] H. Lindborg, H.A. Jakobsen, Sorption Enhanced Steam Methane Reforming Process Performance and Bubbling Fluidized Bed Reactor Design Analysis by Use of a Two-Fluid Model, Industrial & Engineering Chemistry Research, 48 (2008) 1332-1342. [336] Y. Wang, Z. Chao, H.A. Jakobsen, CFD modelling of CO2 capture in the SE-SMR process in the fluidized bed reactors, Chemical Engineering Transactions 21 (2010) 601-606. [337] Y. Wang, Z. Chao, H.A. Jakobsen, 3D Simulation of bubbling fluidized bed reactors for sorption enhanced steam methane reforming processes, Journal of Natural Gas Science and Engineering, 2 (2010) 105-113. [338] Y. Wang, Z. Chao, H.A. Jakobsen, Effects of Gas–Solid Hydrodynamic Behavior on the Reactions of the Sorption Enhanced Steam Methane Reforming Process in Bubbling Fluidized Bed Reactors, Industrial & Engineering Chemistry Research, 50 (2011) 8430-8437. [339] O. Levenspiel, Difficulties in Trying To Model and Scale-Up the Bubbling Fluidized Bed (BFB) Reactor, Industrial & Engineering Chemistry Research, 47 (2007) 273-277. [340] V.V. Kelkar, K.M. Ng, Development of fluidized catalytic reactors: Screening and scaleup, AIChE Journal, 48 (2002) 1498-1518. [341] F.M. White, Fluid Mechanics, McGraw-Hill, 2003. [342] L.R. Glicksman, M.R. Hyre, P.A. Farrell, Dynamic similarity in fluidization, International Journal of Multiphase Flow, 20, Supplement 1 (1994) 331-386. BIBLIOGRAPHY 227 [343] E. Buckingham, On Physically Similar Systems; Illustrations of the Use of Dimensional Equations, Physical Review, 4 (1914) 345-376. [344] M. Rüdisüli, T.J. Schildhauer, S.M.A. Biollaz, J.R. van Ommen, Scale-up of bubbling fluidized bed reactors — A review, Powder Technology, 217 (2012) 21-38. [345] J.B. Romero, L.N. Johanson, Factors affecting fluidized bed quality, Chem. Eng. Prog. Symp. Ser. , 58 (1962) 28-37. [346] T.E. Broadhurst, H.A. Becker, The application of the theory of dimensions to fluidized bed, (1973) 10-27. [347] T.J. Fitzgerald, S.D. Crane, Cold fluidized bed modeling, Proceedings of 6th International Conference on Fluidized Bed Combustion (Atlanta), III (1980) 815-820. [348] L.R. Glicksman, Scaling relationships for fluidized beds, Chemical Engineering Science, 39 (1984) 1373-1379. [349] T.B. Anderson, R. Jackson, Fluid Mechanical Description of Fluidized Beds. Equations of Motion, Industrial & Engineering Chemistry Fundamentals, 6 (1967) 527-539. [350] L.R. Glicksman, M. Hyre, K. Woloshun, Simplified scaling relationships for fluidized beds, Powder Technology, 77 (1993) 177-199. [351] J. Werther, Scale-up modeling for fluidized bed reactors, Chemical Engineering Science, 47 (1992) 2457-2462. [352] P.U. Foscolo, R. Di Felice, L.G. Gibilaro, L. Pistone, V. Piccolo, Scaling relationships for fluidisation: the generalised particle bed model, Chemical Engineering Science, 45 (1990) 1647-1651. [353] M. Horio, A. Nonaka, Y. Sawa, I. Muchi, New Similarity Rule for Fluidized Bed ScaleUp, AIChE Journal, 32 (1986) 1466-1482. [354] C.M. Van Den Bleek, J.C. Schouten, Can deterministic chaos create order in fluidizedbed scale-up?, Chemical Engineering Science, 48 (1993) 2367-2373. [355] B. Leckner, P. Szentannai, F. Winter, Scale-up of fluidized-bed combustion – A review, Fuel, 90 (2011) 2951-2964. [356] B.G.M. van Wachem, J.C. Schouten, R. Krishna, C.M. van den Bleek, Validation of the Eulerian simulated dynamic behaviour of gas–solid fluidised beds, Chemical Engineering Science, 54 (1999) 2141-2149. [357] R. Krishna, J.M. van Baten, Using CFD for scaling up gas–solid bubbling fluidised bed reactors with Geldart A powders, Chemical Engineering Journal, 82 (2001) 247-257. BIBLIOGRAPHY 228 [358] J. Wang, M.A. van der Hoef, J.A.M. Kuipers, Why the two-fluid model fails to predict the bed expansion characteristics of Geldart A particles in gas-fluidized beds: A tentative answer, Chemical Engineering Science, 64 (2009) 622-625. [359] Y.P. Tsuo, D. Gidaspow, Computation of flow patterns in circulating fluidized beds, AIChE Journal, 36 (1990) 885-896. [360] B. Chalermsinsuwan, P. Piumsomboon, D. Gidaspow, A computational fluid dynamics design of a carbon dioxide sorption circulating fluidized bed, AIChE Journal, 56 (2010) 28052824. [361] P. Khongprom, D. Gidaspow, Compact fluidized bed sorber for CO2 capture, Particuology, 8 (2010) 531-535. [362] A. Almuttahar, F. Taghipour, Computational fluid dynamics of high density circulating fluidized bed riser: Study of modeling parameters, Powder Technology, 185 (2008) 11-23. [363] J. Li, M. Kwauk, Particle-fluid two-phase flow: the energy-minimization multi-scale method, Metallurgical Industry Press, 1994. [364] Z. Shi, W. Wang, J. Li, A bubble-based EMMS model for gas–solid bubbling fluidization, Chemical Engineering Science, 66 (2011) 5541-5555. [365] B. Lu, N. Zhang, W. Wang, J. Li, J.H. Chiu, S.G. Kang, 3-D full-loop simulation of an industrial-scale circulating fluidized-bed boiler, AIChE Journal, 59 (2013) 1108-1117. [366] Y. Zhao, H. Li, M. Ye, Z. Liu, 3D Numerical Simulation of a Large Scale MTO Fluidized Bed Reactor, Industrial & Engineering Chemistry Research, 52 (2013) 11354-11364. [367] A. Srivastava, S. Sundaresan, Analysis of a frictional–kinetic model for gas–particle flow, Powder Technology, 129 (2003) 72-85. [368] Y. Igci, A.T. Andrews, S. Sundaresan, S. Pannala, T. O'Brien, Filtered two-fluid models for fluidized gas-particle suspensions, AIChE Journal, 54 (2008) 1431-1448. [369] A. Gobin, H. Neau, O. Simonin, J.-R. Llinas, V. Reiling, J.-L.c. Sélo, Fluid dynamic numerical simulation of a gas phase polymerization reactor, International Journal for Numerical Methods in Fluids, 43 (2003) 1199-1220. [370] J.-F. Parmentier, O. Simonin, O. Delsart, A functional subgrid drift velocity model for filtered drag prediction in dense fluidized bed, AIChE Journal, 58 (2012) 1084-1098. [371] A. Ozel, P. Fede, O. Simonin, Development of filtered Euler–Euler two-phase model for circulating fluidised bed: High resolution simulation, formulation and a priori analyses, International Journal of Multiphase Flow, 55 (2013) 43-63. [372] J.R. van Ommen, M. Teuling, J. Nijenhuis, B.G.M. van Wachem, Computational validation of the scaling rules for fluidized beds, Powder Technology, 163 (2006) 32-40. BIBLIOGRAPHY 229 [373] F.P. Di Maio, A. Di Renzo, Verification of scaling criteria for bubbling fluidized beds by DEM–CFD simulation, Powder Technology, 248 (2013) 161-171. [374] A. Calabrò, Precombustion CO2 capture in coal power plants: research activities at ENEA, "CO2 Capture & Storage: Towards a UK-Italy Common Strategy within a Global Framework." Erice (Italy), (1-7 November 2007). [375] S. Stendardo, P. Deiana, A. Calabrò, Short-cut dynamic model of a catalytic reactor for the CO2 capture by means of dolomite., in: Technical report RSE/2009/21, ENEA, Rome, 2009. [376] L.F. Richardson, The Approximate Arithmetical Solution by Finite Differences of Physical Problems Involving Differential Equations, with an Application to the Stresses in a Masonry Dam, Philosophical Transactions of the Royal Society of London. Series A, Containing Papers of a Mathematical or Physical Character, 210 (1911) 307-357. [377] L.F. Richardson, J.A. Gaunt, The Deferred Approach to the Limit. Part I. Single Lattice. Part II. Interpenetrating Lattices, Philosophical Transactions of the Royal Society of London. Series A, Containing Papers of a Mathematical or Physical Character, 226 (1927) 299-361. [378] I.B. Celik, U. Ghia, P.J. Roache, C.J. Freitas, H. Coleman, P.E. Raad, Procedure for Estimation and Reporting of Uncertainty Due to Discretization in CFD Applications, Journal of Fluids Engineering, 130 (2008) 078001-078001. [379] ASME, Standard for Verification and Validation in Computational Fluid Dynamics and Heat Transfer in, 2009. [380] J.R. Grace, F. Taghipour, Verification and validation of CFD models and dynamic similarity for fluidized beds, Powder Technology, 139 (2004) 99-110. [381] H. Lindborg, M. Lysberg, H.A. Jakobsen, Practical validation of the two-fluid model applied to dense gas–solid flows in fluidized beds, Chemical Engineering Science, 62 (2007) 5854-5869. [382] R. Courant, K. Friedrichs, H. Lewy, Über die partiellen Differenzengleichungen der mathematischen Physik, Math. Ann., 100 (1928) 32-74. [383] A. Gel, R. Garg, C. Tong, M. Shahnam, C. Guenther, Applying uncertainty quantification to multiphase flow computational fluid dynamics, Powder Technology, 242 (2013) 27-39. [384] S. Sasic, B. Leckner, F. Johnsson, Characterization of fluid dynamics of fluidized beds by analysis of pressure fluctuations, Progress in Energy and Combustion Science, 33 (2007) 453496. [385] F. Johnsson, R.C. Zijerveld, J.C. Schouten, C.M. van den Bleek, B. Leckner, Characterization of fluidization regimes by time-series analysis of pressure fluctuations, International Journal of Multiphase Flow, 26 (2000) 663-715. [386] R.A. Fisher, The design of experiments, Oliver and Boyd, 1935. BIBLIOGRAPHY 230 [387] D.R. Cox, N. Reid, The Theory of the Design of Experiments, Taylor & Francis, 2000. [388] D.C. Montgomery, Design and Analysis of Experiments, John Wiley & Sons, 2008. [389] G. Taguchi, System of experimental design: engineering methods to optimize quality and minimize costs, UNIPUB/Kraus International Publications, 1987. [390] R.K. Roy, A Primer on the Taguchi Method, Society of Manufacturing Engineers, 2010. [391] J.C. Miranda, Design of Experiments using the Taguchi Method, in: Product design: techniques for robustness, reliability and optimization. Class Notes., Instituto Tecnologico y de Estudios Superiores de Monterrey, Toluca (Mexico), 2004. [392] R. Spence, J. Amaral-Teixeira, A CFD parametric study of geometrical variations on the pressure pulsations and performance characteristics of a centrifugal pump, Computers & Fluids, 38 (2009) 1243-1257. [393] Q. Chen, M. Zeng, J. Zhang, Q. Wang, Optimal design of bi-layer interconnector for SOFC based on CFD-Taguchi method, International Journal of Hydrogen Energy, 35 (2010) 4292-4300. [394] J.-M. Wang, H.-J. Yan, J.-M. Zhou, S.-X. Li, G.-C. Gui, Optimization of parameters for an aluminum melting furnace using the Taguchi approach, Applied Thermal Engineering, 33– 34 (2012) 33-43. [395] I.E.A., World Energy Outlook: 2011 - executive summay, IEA/OECD, Paris, 2011. [396] R. Cyprès, S. Furfari, Hydropyrolysis of a high-sulphur-high-calcite Italian Sulcis coal. 1. Hydropyrolysis yields and catalytic effect of the calcite, Fuel, 61 (1982) 447-452. [397] P.R. Solomon, T.H. Fletcher, Impact of coal pyrolysis on combustion, Symposium (International) on Combustion, 25 (1994) 463-474. [398] J. Yu, J.A. Lucas, T.F. Wall, Formation of the structure of chars during devolatilization of pulverized coal and its thermoproperties: A review, Progress in Energy and Combustion Science, 33 (2007) 135-170. [399] P.R. Solomon, M.A. Serio, E.M. Suuberg, Coal pyrolysis: Experiments, kinetic rates and mechanisms, Progress in Energy and Combustion Science, 18 (1992) 133-220. [400] S. Niksa, C.-W. Lau, Global rates of devolatilization for various coal types, Combustion and Flame, 94 (1993) 293-307. [401] T.F. Wall, G.-s. Liu, H.-w. Wu, D.G. Roberts, K.E. Benfell, S. Gupta, J.A. Lucas, D.J. Harris, The effects of pressure on coal reactions during pulverised coal combustion and gasification, Progress in Energy and Combustion Science, 28 (2002) 405-433. BIBLIOGRAPHY 231 [402] P. Carbini, L. Curreli, M. Ghiani, F. Satta, Desulphurization of europeans coal using molten caustic mixtures, in: Processing and utilization of high sulphur coals III, Markuvzewski, R. Wheelock, T. D., 1990. [403] R.C. Borah, P. Ghosh, P.G. Rao, A review on devolatilization of coal in fluidized bed, International Journal of Energy Research, 35 (2011) 929-963. [404] P.R. Solomon, D.G. Hamblen, R.M. Carangelo, M.A. Serio, G.V. Deshpande, General model of coal devolatilization, Energy & Fuels, 2 (1988) 405-422. [405] A. Sarwar, M. Nasiruddin Khan, K. Azhar, Kinetic studies of pyrolysis and combustion of Thar coal by thermogravimetry and chemometric data analysis, Journal of Thermal Analysis and Calorimetry, 109 (2012) 97-103. [406] W.R. Ladner, The products of coal pyrolysis: properties, conversion and reactivity, Fuel Processing Technology, 20 (1988) 207-222. [407] S. Scaccia, A. Calabrò, R. Mecozzi, Investigation of the evolved gases from Sulcis coal during pyrolysis under N2 and H2 atmospheres, Journal of Analytical and Applied Pyrolysis, 98 (2012) 45-50. [408] A. Arenillas, F. Rubiera, C. Pevida, J.J. Pis, A comparison of different methods for predicting coal devolatilisation kinetics, Journal of Analytical and Applied Pyrolysis, 58–59 (2001) 685-701. [409] A.K. Burnham, R.L. Braun, Global Kinetic Analysis of Complex Materials, Energy & Fuels, 13 (1998) 1-22. [410] M.V. Kök, Thermal Analysis Applications In Fossil Fuel Science. Literature survey, Journal of Thermal Analysis and Calorimetry, 68 (2002) 1061-1077. [411] M.E. Brown, A.K. Galwey, The significance of “compensation effects” appearing in data published in “computational aspects of kinetic analysis”: ICTAC project, 2000, Thermochimica Acta, 387 (2002) 173-183. [412] M. Maciejewski, Computational aspects of kinetic analysis.: Part B: The ICTAC Kinetics Project — the decomposition kinetics of calcium carbonate revisited, or some tips on survival in the kinetic minefield, Thermochimica Acta, 355 (2000) 145-154. [413] V. Strezov, J.A. Lucas, T.J. Evans, L. Strezov, Effect of heating rate on the thermal properties and devolatilisation of coal, Journal of Thermal Analysis and Calorimetry, 78 (2004) 385-397. [414] A.K. Galwey, M.E. Brown, Arrhenius parameters and compensation behaviour in solidstate decompositions, Thermochimica Acta, 300 (1997) 107-115. [415] A.K. Galwey, Is the science of thermal analysis kinetics based on solid foundations?: A literature appraisal, Thermochimica Acta, 413 (2004) 139-183. BIBLIOGRAPHY 232 [416] S. Vyazovkin, A.K. Burnham, J.M. Criado, L.A. Pérez-Maqueda, C. Popescu, N. Sbirrazzuoli, ICTAC Kinetics Committee recommendations for performing kinetic computations on thermal analysis data, Thermochimica Acta, 520 (2011) 1-19. [417] A. Figen, O. İsmail, S. Pişkin, Devolatilization non-isothermal kinetic analysis of agricultural stalks and application of TG-FT/IR analysis, Journal of Thermal Analysis and Calorimetry, 107 (2012) 1177-1189. [418] S. Badzioch, P.G.W. Hawksley, Kinetics of thermal decomposition of pulverized coal particles, Industrial and Engineering Chemistry: Process Design and Development, 9 (1970) 521-530. [419] G.H. Ko, D.M. Sanchez, W.A. Peters, J.B. Howard, Application of first-order singlereaction model for coal devolatilization over a wide range of heating rates, Preprints of Papers, American Chemical Society, Division of Fuel Chemistry; (USA), (1988) 112-119. [420] M.a.-J. Lázaro, R. Moliner, I. Suelves, Non-isothermal versus isothermal technique to evaluate kinetic parameters of coal pyrolysis, Journal of Analytical and Applied Pyrolysis, 47 (1998) 111-125. [421] H.E. Kissinger, Reaction kinetics in differential thermal analysis, Analytical Chemistry, 29 (1957) 1702-1706. [422] K. Slopiecka, P. Bartocci, F. Fantozzi, Thermogravimetric analysis and kinetic study of poplar wood pyrolysis, Applied Energy, 97 (2012) 491-497. [423] R.L. Braun, A.K. Burnham, Analysis of chemical reaction kinetics using a distribution of activation energies and simpler models, Energy & Fuels, 1 (1987) 153-161. [424] K. Miura, A new and simple method to estimate f(E) and k0(E) in the distributed activation energy model from three sets of experimental data, Energy & Fuels, 9 (1995) 302307. [425] D.M. Grant, R.J. Pugmire, T.H. Fletcher, A.R. Kerstein, Chemical model of coal devolatilization using percolation lattice statistics, Energy & Fuels, 3 (1989) 175-186. [426] A. Savitzky, M.J.E. Golay, Smoothing and Differentiation of Data by Simplified Least Squares Procedures, Analytical Chemistry, 36 (1964) 1627-1639. [427] J.A. Caballero, J.A. Conesa, Mathematical considerations for nonisothermal kinetics in thermal decomposition, Journal of Analytical and Applied Pyrolysis, 73 (2005) 85-100. [428] A. Williams, M. Pourkashanian, J.M. Jones, The combustion of coal and some other solid fuels, Proceedings of the Combustion Institute, 28 (2000) 2141-2162. [429] P.R. Solomon, D.G. Hamblen, Finding order in coal pyrolysis kinetics, Progress in Energy and Combustion Science, 9 (1983) 323-361. BIBLIOGRAPHY 233 [430] J. Pallarés, I. Arauzo, A. Williams, Integration of CFD codes and advanced combustion models for quantitative burnout determination, Fuel, 86 (2007) 2283-2290. [431] H.S. Harold, Lignites of North America, in: Coal Science and Technology, Vol. 23, Elsevier, 1995, pp. 202. [432] M.J.G. Alonso, D. Alvarez, A.G. Borrego, R. Menéndez, G. Marbán, Systematic Effects of Coal Rank and Type on the Kinetics of Coal Pyrolysis, Energy & Fuels, 15 (2001) 413-428. [433] C.C. Lakshmanan, N. White, A New Distributed Activation Energy Model Using Weibull Distribution for the Representation of Complex Kinetics, Energy & Fuels, 8 (1994) 1158-1167. [434] T.C. Ho, R. Aris, On apparent second-order kinetics, AIChE Journal, 33 (1987) 10501051. [435] C.C. Lakshmanan, M.L. Bennett, N. White, Implications of multiplicity in kinetic parameters to petroleum exploration: distributed activation energy models, Energy & Fuels, 5 (1991) 110-117. [436] C.P. Please, M.J. McGuinness, D.L.S. McElwain, Approximations to the distributed activation energy model for the pyrolysis of coal, Combustion and Flame, 133 (2003) 107-117. [437] R. Brun, F. Rademakers, ROOT - An object oriented data analysis framework, Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 389 (1997) 81-86. [438] M.A. Serio, D.G. Hamblen, J.R. Markham, P.R. Solomon, Kinetics of volatile product evolution in coal pyrolysis: experiment and theory, Energy & Fuels, 1 (1987) 138-152. [439] H.A.G. Chermin, D.W. Van Krevelen, Chemical structure and properties of coal XVII - A mathematical model of coal pyrolysis, Fuel, 36 (1957) 85-104. [440] A.K. Burnham, B.J. Schmidt, R.L. Braun, A test of the parallel reaction model using kinetic measurements on hydrous pyrolysis residues, Organic Geochemistry, 23 (1995) 931939. [441] B. de Caprariis, P. De Filippis, C. Herce, N. Verdone, A Double-Gaussian Distributed Activation Energy Model for Coal Devolatilization, Energy & Fuels, 26 (2012) 6153–6159. [442] T.H. Fletcher, A.R. Kerstein, R.J. Pugmire, D.M. Grant, Chemical percolation model for devolatilization. 2. Temperature and heating rate effects on product yields, Energy & Fuels, 4 (1990) 54-60. [443] T.H. Fletcher, A.R. Kerstein, R.J. Pugmire, M.S. Solum, D.M. Grant, Chemical percolation model for devolatilization. 3. Direct use of carbon-13 NMR data to predict effects of coal type, Energy & Fuels, 6 (1992) 414-431. BIBLIOGRAPHY 234 [444] Fletcher TH. 13C NMR parameter calculator. In: Chemical Percolation Devolatilization (CPD) Model. 2006. http://www.et.byu.edu/~tom/cpd/correlation.html. Last accessed 24 Nov 2013. [445] D. Gera, M. Mathur, M. Freeman, Parametric Sensitivity Study of a CFD-Based Coal Devolatilization Model, Energy & Fuels, 17 (2003) 794-795. [446] I. Ion, F. Popescu, G. Rolea, A biomass pyrolysis model for CFD application, Journal of Thermal Analysis and Calorimetry, 111 (2013) 1811-1815. [447] J.O. Pou, Y.E. Alvarez, J.K. Watson, J.P. Mathews, S. Pisupati, Co-primary thermolysis molecular modeling simulation of lignin and subbituminous coal via a reactive coarse-grained simplification, Journal of Analytical and Applied Pyrolysis, 95 (2012) 101-111.