Second report of PhD Thesis committee:

Transcription

1 UNIVERSITY OF MONS FACULTY OF ENGINEERING Second report of PhD Thesis committee: Capture and purification processes applied to CO 2 derived from cement industry for conversion into methanol. PhD student: Ir Sinda LARIBI Thesis Promoter: Prof. Diane THOMAS Thesis Co-promoter: Prof. Guy DE WEIRELD Research coordinator: Dr Lionel DUBOIS Chemical and Biochemical Engineering Department March 2016

2 ACKNOWLEDGEMENTS The author would like to express the deepest appreciation to all those who provided the possibility to construct this report. First, a special gratitude the author gives to Prof. Diane THOMAS and Prof. Guy DE WEIRELD for their precious help, guidance, constructive feedback and the work they put up with to bring this report into life. Furthermore, the author appreciates the continuous support of Dr. Lionel DUBOIS, who played an undeniably big part in the execution of this project, for his patience, motivation and efficient supervision. Last but cert not least, the author would like to express gratefulness to University of Mons (UMONS) and to the European Cement Research Academy (ECRA) for the technical and financial supports accorded to the ECRA Academic Chair.

3 Table of contents 1. Introduction General context Cement manufacturing process Gaseous emissions and directive CO 2 capture technologies Context of this work Study of the post-combustion CO 2 capture applied to O 2-enriched air combustion: hybrid process Partial oxyfuel combustion capture: definition and innovative aspects Hybrid process: advantages and purpose Energetic effect Chemical effect CO 2 capture using the absorption-generation process Absorption-regeneration process technology Gas-liquid absorption solvents Experimental screening of solvents under partial oxyfuel conditions Experimental device: The cables-bundle contactor Selected solvents Types of tests conducted Experimental results Calculation of the required global energy for the partial oxyfuel combustion process Calculation of the required energy for the Air Separation Unit under partial oxyfuel combustion conditions Simulation of the CO 2 capture using the absorption-regeneration process: Estimation of the regeneration energy under partial oxyfuel conditions Conclusions and perspectives of part Simulation of a CO 2 purification process related to a gas coming from cement industries applying full oxyfuel combustion Oxyfuel combustion technology applied to the cement industry Oxyfuel carbon capture and purification Purification of oxyfuel power plants derived CO Purification of oxyfuel cement plants derived CO Sour-Compression Unit (SCU): De-NOx and De-SOx processes for CO 2 purification Bibliographic review of the SCU complete chemical mechanism I

4 New SCU chemical mechanism deduced from the bibliographic review Complexity of the implementation into Aspen Plus Optimization of the SCU simulated process by (Iloeje et al., 2015) Simulation results of the SCU absorption performances Conclusions and perspectives of the simulation works General conclusion Doctoral Training activities of Sinda Laribi from January 2015 to January 2016.a Bibliographic references..i Annexes..A II

5 List of figures Figure 1: Scheme of the dry manufacturing process of cement (Cement Association of Canada (CAC), 2015)... 2 Figure 2: CO 2 capture technologies... 4 Figure 3: ECRA CHAIR framework at the University of Mons... 5 Figure 4: Scheme of the two pathways detailed in this report (in red)... 6 Figure 5: Scheme of a CO 2 post-combustion capture applied to an O 2-enriched air combustion technology... 8 Figure 6: Energetic effect of hybrid process (adapted from Favre et al., 2009) Figure 7: Chemical effect of the hybrid process (adapted from L. Li et al., 2013)) Figure 8: Illustration of the Absorption-Regeneration process (CO 2CRC, 2015) Figure 9: Examples of different types of amines Figure 10: Semi-developed formulas of: a) PZ and b) TETRA Figure 11: Chemical solvents and physical solvents absorption capacities (Adapted from Bailey et al., 2005) Figure 12: Cables-bundle contactor scheme Figure 13: Effect of Y CO2,in on the absorption rate A (%) solution for a MEA 30% solution Figure 14: Effect of Y CO2,in on G CO2,abs for a MEA 30% solution Figure 15: Results of the continuous tests with simple solvents Figure 16: Results of the continuous tests with activated solvents for [MMEA+activator] Figure 17: Results of the continuous tests with activated solvents for [AMP+activator] Figure 18: Results of the continuous tests with activated solvents for [DEA+activator] Figure 19: Summary of all the results for activated solvents Figure 20: Continuous tests with hybrid solvents Figure 21: Visualization of the demixing phenomenon for [MEA 30% + TOU 35%] (after 45 minutes) Figure 22: Temporal evolution of G CO2,abs Figure 23: Temporal evolution of α CO Figure 24: Semi-continuous tests with activated solvents Figure 25: Best solvents screened in high CO 2 concentrations compared to the reference MEA 30% 27 Figure 26 : Basic layout of partial oxyfuel configurations Figure 27: Scheme of an O 2-enriched air combustion Figure 28: Inputs of the O 2-enriched air combustion Figure 29: Evolution of E ASU with Y CO2 in the flue gas out Figure 30: Global flowsheet of the simulated pilot Figure 31: Conditioning step of the stream entering the absorption column Figure 32: Absorption column Figure 33: Pumping and preheating of the rich solution Figure 34: Regeneration step Figure 35: Cooling of the lean solution and "makes up" Figure 36: Evolution of E regen with Y CO2,in to capture 1,5 t CO 2/h Figure 37: Global energy required for partial oxycombustion process Figure 38: Micro-pilot for absorption-regeneration tests Figure 39: Oxyfuel combustion technology (ECRA technical report phase 3, 2012) Figure 40: Basic layout of full oxyfuel configuration (Carrasco-Maldonado et al., 2016) Figure 41: Global flowsheet of the CO 2 Purification Unit (CPU) III

6 Figure 42: Conversion of nitrogen compounds into N 2O and HADS plus HAMS (Petrissans et al., 2001) & (Normann et al., 2013) Figure 43: Diagram of the two pathways for the Interactions between nitrous acid and hydrogen sulfite Figure 44: Competition between production of N 2O and production of N-S complexes Figure 45: Schematic representation of the interaction between sulfite and nitrite ions in aqueous solutions Figure 46: Reaction pathways for ph levels between 1 and 4 (adapted from Ajdari et al., 2015) Figure 47: Complete SCU Chemical Mechanism for SOx and NOx absorption without (A) or with (C) SOx/NOx interactions (reactions selected for 1<pH<4) Figure 48: SCU Chemical Mechanism for SOx and NOx absorption with simplified SOx/NOx interactions (B) Figure 49: Flowsheet of the two-column process Figure 50: Aspen Plus flowsheet of the simulated SCU Figure 51: Abatement rate of the SO 2 and NOx components in the first column of the SCU IV

7 List of tables Table 1:Industrial Emissions Directive Limits (Meer & Ravail, 2013)... 3 Table 2 : Primary, secondary and tertiary amines configurations Table 3: Kinetic constants relative to different types of amines Table 4: Amine solvents Table 5: Activator used in the solvents [amine+activator] Table 6: Physical solvent used Table 7 : CO 2 loadings of the solvents at the end of the tests (after 90 min) Table 8: Input parameters Table 9: Recapitulative table of the studied cases Table 10: Dimensions and operating conditions of the columns Table 11: Compositions and conditions of the gas to treat-different cases Table 12: Effect of the composition of the flue gas on the energy regeneration costs Table 13: Raw & purified flue gas compositions from basic oxyfuel purification process applied to power plants Table 14: Air Products reactions of the sour compression process for the purification of oxyfuelderived CO Table 15:Updated SOx-NOx reaction network for Air Products Sour-Compression Unit (Santos, 2015) Table 16: New complexes denominations Table 17: Operating parameters taken for the simulations tests Table 18: Kinetic parameters taken for the SCU chemical mechanism Table 19: Kinetic and equilibrium law expressions in Aspen Plus Table 20: Molecular structures of the N-S complexes implemented in Aspen Table 21: Calculation of H 2SO 4 properties by means of a group contribution method Table 22: Required N-S properties introduced in Aspen Plus to do the simulations Table 23: AQU-DATA method used for ions properties estimation in Aspen Plus Table 24: Purity targets for pipeline specifications Table 25: Reactions chosen for the implemented model (Iloeje et al., 2015) Table 26: Design specifications taken in Iloeje and co-workers model Table 27: Flue gas compositions and stream results for the simulated model by ((Iloeje et al., 2015) 61 Table 28: Compositions of an outcoming gas from an oxyfuel cement industry Table 29: Design specifications of the SCU contactors (based on ranges given by Air Products) Table 30: SCU simulation results considering the chemical mechanism without interactions (mechanism A) Table 31: SCU simulation results considering the chemical mechanism with the main interactions (mechanism B) Table 32: SCU simulation results considering the mechanism with all the interactions (mechanism C) V

8 Table of notations General abbreviations Abbreviation ASU CPU ECRA FGD IC IR SCU SHA TC TOC TSA Signification Air Separation Unit CO 2 Purification Unit European Cement Research Academy Flue-Gas Desulfurization Inorganic Carbon InfraRed radiation Sour-Compression Unit Sterically Hindered Amine Total Carbon Total Organic Carbon Temperature Swing Adsorption Solvents Abbreviation Name Formula CAS number AHPD 2-amino-2 hydroxymethyl-1,3-propanediol C 4H 11NO AMP 2-amino-2-methyl-1-propanol C 4H 11NO DEA Diethanolamine C 4H 11NO MDEA N-methyldiethanolamine C 5H 13NO MEA Monoethanolamine C 2H 7NO MMEA Monomethylethanolamine C 3H 9NO PZ Piperazine C 4H 10N TETRA Triethylenetetramine C 6H 18N TOU 2,5,7,10-tetraoxaundecane C 7H 16O SOx/NOx complexes Complete denomination Raw formula Alias Nitrososulfonate ONSO 3 NSS Nitrososulfonic acid ONSO 3H NSSH Hydroxylamine N,N-disulfonate HNO(SO 3) 2-2 HADS Hydroxylamine N,N-disulfonic acid HNO(SO 3H) 2 HADSH Hydroxylamine N-sulfonate HONHSO 3 HAMS Hydroxylamine N-sulfonic acid HONHSO 3H HAMSH VI

9 1. Introduction 1.1. General context Nowadays, concentrations of greenhouse gases within the atmosphere are likely to accelerate the rate of climate change by raising the earth's average temperature. For this reason, lowering the emissions of greenhouse gases in the atmosphere has quickly become one of the most urgent environmental issues. Carbon dioxide emitted from cement industries represent 30% of the total annual CO 2 emitted from industrial sectors (5% of the CO 2 global emissions) (IEA, 2013). Therefore, the cement industry plays a major role in these CO 2 emissions. In addition to CO 2, formation and emissions of SO 2, NOx, HCl, heavy metals, TOC and dust is inevitable and have to be controlled. Carbon Capture and Storage (CCS) is one option for reducing these harmful anthropogenic CO 2 emission. Another option on which we focus in this work is the Carbon Capture and Re-Use (CCU) Cement manufacturing process The cement manufacturing process described in Figure 1 is composed by several steps: extraction of the raw materials: the usual cements are produced from 80% of limestone (CaCO 3) and 20% of clay (SiO 2 and Al 2O 3) homogenization of the raw materials in order to obtain a slurry the produced slurry is sent in a rotary kiln (e.g 90 meters long and 6 meters of diameter) and flows toward a combustion flame of 2000 C, produced by fuel combustion, with CO 2 as a product. The clinkering process occurs at 1450 C in the drying step chains are used to increase heat exchange between slurry and the kiln walls calcining step: in this zone occurs the calcination of limestone by the reaction of production of clinker which requires an important amount of energy and generates CO 2 : CaCO 3 + heat CaO + CO 2 Energy consumption of the global cement production process between 3200 and 4200 kj/ton of clinker. quenching step: the formed clinker is rapidly cooled and gives a physico-chemical combination of calcium silicates and calcium aluminates. clinker is then crushed to powder and some minor amounts of gypsum are often added in different proportions according to the type of cement aimed at the end of the process. 1

10 Figure 1: Scheme of the dry manufacturing process of cement (Cement Association of Canada (CAC), 2015) Since 2007 the major cement manufacturing process used in the kilns is the dry process with about 90 % of Europe cement production. For the dry process, approximately 2/3 of the CO 2 emission is a result of calcination of limestone and 1/3 from the combustion of fuels Gaseous emissions and directive Directive 2010/75/EU of the European Parliament and of the council of 24 November 2010 on industrial emissions (integrated pollution prevention and control) established threshold values referring to production capacities or outputs of different industrial activities. Among them, the production of cement lime and magnesium oxide is concerned: production of cement clinker in rotary kilns with a production capacity exceeding 500 tons per day or in other kilns with a production capacity exceeding 50 tons per day. Therefore, the Industrial Emissions Directive (IED, 2010/75/EU) gives emission limits to be implemented in environmental permits. These emission limits are not to be exceeded. Usually, NOx components have not to exceed 500mg/Nm³ for all types of kilns; in the case of SO 2, it has not to exceed 500 mg /Nm³ and commonly for dust emissions the maximum is 30 mg/nm³. However, we have to take into account that part of the IED structure are the BREF (BREF = Best available techniques Reference document). In these documents are included the Best Available Techniques for each sector (recommended technologies for Greenfield and sometimes existing plants), the BAT associated emission levels (BAT-AEL) for emissions which are not regulated in the IED that are recommendation for possible emissions limits in the permit. Consequently, the actual values in the permit can be different to the BREF values. However, that deviation has to be suggested by the competent authorities (government), explained (and sometimes proven) in the permit. This is the reason why possible differences of values of emissions limits between the IED, 2010/75/EU and the BAT-AEL could exist. Table 1 summarizes the emission limits dictated by the IED, 2010/75/EU and then, marked in green, the BAT-AEL. All emission limit values are calculated at a temperature of 273,15 K, a pressure of 101,3 kpa and after correction for the water vapor content of the waste gases and at a standardized O 2 content of 3 % (for combustion plants). 2

11 Table 1: Industrial Emissions Directive Limits (Meer & Ravail, 2013) Component Emission limit (mg/nm³) Period of sampling Total dust 30 default value < for kiln, cooling, grinding 24 hours HCl <10 24 hours HF <1 24 hours NOx <500 for preheater kilns < exemption: < until 1/1/2016 for Lepol* and long kilns NH 3 (from SNCR) < 30 as default but up to 50 allowed for Lepol and long rotary kilns even higher SO2 <50 Higher values are allowed when not from co-incineration < TOC 10 Higher values are allowed when not from co-incineration 24 hours 24 hours 24 hours Dioxins + furans 0.1 ng/nm 3 8 hours Sb+As+Pb+Cr+Co+Cu+Mn+Ni+V min 8hours Cd + Tl min 8 hours Hg min 8 hours *Lepol kilns design energy-efficient systems known as semi-dry Lepol process: The preheaters consist of a moving flexible endless-chain grate made up of slotted steel plates which carries spherical nodules in a bed m deep. The nodules are made from dry raw meal, by addition of 11-14% water (hence semi-dry), in a nodulizing pan CO2 capture technologies To reduce CO 2 emissions various techniques can be used and were explained in the previous report, a quick reminder of these techniques is given in Figure 2. 3

12 Figure 2: CO 2 capture technologies The pre-combustion capture technique could not be applied to the cement industry since the major part of the CO 2 emitted is originated from the calcination reaction. The post-combustion capture system consists in capturing the carbon dioxide (Y CO2=25-30%) directly in the exhaust flue gas of the plant; consequently, CO 2 is mixed with other components that have to be removed in order to reach a final pure CO 2 stream. As detailed in the previous report, pilot plants exist at Brevik s cement plant (Norcem Brevik, a and b, 2013) where different tests of CO 2 capture are currently conducted. Post-combustion carbon capture technologies have not been applied to the cement industry yet but studies proved the possibility to extend some technologies to an industrial scale. It is important to mention also that as showed in the report of the industrial visit of Boundary Dam CCS plant (Saskatchewan, Canada) carried out in the framework of the PCCC-3 Conference in September 2015 (Boundary Dam CCS plant visit report, 2015), Carbon Capture and Storage is industrially applied for power plants (Y CO2= 5-15%) by amine scrubbing process allowing CO 2 capture and at the same time SOx capture valorized in H 2SO 4 commercialized further. The partial oxycombustion capture (oxyfuel combustion capture) is a combined technology working with a O 2-enriched air combustion of the fuel, hence partial oxycombustion, allowing a CO 2 more concentrated flue gas (20 %<Y CO2<70 %) compared to a conventional combustion. This technology will be detailed in part 2. The full oxyfuel combustion capture system consists in realizing the combustion of the fuel using only oxygen, hence an Air Separation Unit (ASU) is necessary. The objective is to work with an outcoming flue gas from the cement plant concentrated in CO 2 (Y CO2= 70-90%). This technology will be detailed in part 3. 4

13 1.5. Context of this work As mentioned in the previous report, this work is part of an academic chair associating the University of Mons (UMONS) and the European Cement Research Academy (ECRA) since April The ECRA CHAIR framework (Figure 3) schematizes, within the University of Mons, the studied pathways of CO 2 capture, purification and conversion processes providing the reduction of CO 2 emissions in the cement industry. For this purpose, a CO 2 treatment chain from the capture step to the conversion step into valuable products, including gas treatment processes, is required. The final purpose is converting the CO 2 into valuable products. This could be possible either by direct use: fire extinguisher, solvent, refrigerant (R744), food additive (acidity regulator), beverages production (soda, sparkling water) or by means of other techniques depending on the final product aimed: use in chemical transformations: CO 2 is consumed for production of valuable compounds ( as methanol or urea); mineral carbonation: binding with minerals to form inert and stable carbonate minerals for building materials. The comparison of the most interesting CO 2 Re-use pathways is under progress in another PhD thesis of the ECRA chair at UMONS (Rémi Chauvy). Besides, main interesting products like methane and methanol stood out. However, according to the ECRA Technical Report of November 2013, as more hydrogen is required to convert CO 2 into methane, in comparison to methanol, the conversion of CO 2 into methanol is a more economic issue to valorize the CO 2. For this reason, in the context of the ECRA Chair in the University of Mons, the CO 2 conversion into methanol is the issue investigated in another PhD thesis (Nicolas Meunier). The catalytic conversion into methanol is realized under high pressure and at intermediate temperature in a catalytic reactor. The used catalysts are mainly CuO/ZnO/Al 2O 3. Sulfur and nitrogen oxides are known to be potentially harmful for this kind of catalyst, consequently, a gas purification step is required before conversion. Figure 3: ECRA CHAIR framework at the University of Mons 5

14 In order to reach the purpose of CO 2 conversion, several capture and purification techniques are studied in this work. Mainly, three fields are investigated: post-combustion CO 2 capture applied to a conventional combustion; post-combustion CO 2 capture applied to an O 2-enriched air combustion (partial oxyfuel combustion); CO 2 Purification Unit (CPU) applied to a full oxyfuel combustion. In the previous report, general aspects were explained about the cement industry manufacturing process, the CO2 capture technologies schematized in Figure 2, the post-combustion capture techniques (absorption, adsorption ) and the purification techniques dedicated to remove the impurities (SOx, NOx, dust) from the flue gas emitted by the cement industry. First simulation works of a CO 2 Purification Unit (CPU) using Aspen Plus simulation software have been elaborated and were based on previous works applied to power plants. A comparison with the cement plant has been done taking as an input of the flowsheet compositions from cement plant flue gases. The results showed that the flowsheet simulated could be applied not only for power plants but also for cement plants in order to reduce NOx and SOx components from cement plant flue gases. The process is expected to remove about 100% of sulfur and nitrogen oxides which is a good cause to investigate it more deeply and to study different configurations in order to optimize it. Figure 4 summarizes the investigation points realized for the two carbon capture technologies selected to be investigated deeply in this PhD. Figure 4: Scheme of the two pathways detailed in this report (in red) 6

15 This second report of the PhD committee synthesizes the main points investigated during the last year in mainly two parts: Chapter 2 of this report describes works realized in the context of a post combustion CO 2 capture applied to a partial oxyfuel combustion process called hybrid process : different solvents have been tested experimentally at a lab scale in order to evaluate their absorption performances in high CO 2 concentrations conditions; energy regeneration requirements have been identified using Aspen Hysys TM in order to evaluate energetic costs in high CO 2 concentrations conditions. Chapter 3 describes a detailed study about the SOx/NOx reactive absorption into water and in pressurized systems to simulate a purification step included in the global CO 2 Purification Unit (CPU) named the Sour-Compression Unit (SCU). This part involves several points: A comprehensive bibliographic study where the ph influence on the selection of the reactions is identified; a new and more accurate chemical mechanism is elucidated taken into account SOx/NOx interactions; a new complex species arising from these interactions are simulated; a description of the further simulations including an optimization of the configuration of the SCU process. The final chapter provides conclusions after this year of research and a list of the different scientific activities elaborated during the last year. 7

16 2. Study of the post-combustion CO 2 capture applied to O 2 - enriched air combustion: hybrid process 2.1. Partial oxyfuel combustion capture: definition and innovative aspects To reduce CO 2 emissions from fossil fuels, post-combustion capture by absorption-regeneration and oxyfuel combustion capture are promising technologies. However, solvent regeneration costs reach from 3 to 4 GJ/t CO2 for MEA (30%) solvent and high-purity O 2 production from the air separation unit make the post-combustion capture and the oxyfuel combustion capture intensive energy consumption processes respectively. Consequently, the solution is to use an alternative technology called hybrid technology. A hybrid technology is a combination of two (or more) different technologies allowing increase individual performances or reduce the overall energy consumption in comparison with the separate use of the technologies. According to the literature, there are several examples in the context of CO 2 capture, between other we can cite: combination of membranes gas separation and absorption technologies ; combination of membranes gas separation and cryogenic CO 2 capture technologies ; combination of chemical and physical solvents in a post-combustion CO 2 capture plant (= hybrid solvents). In our case, the hybrid technology represented in Figure 5 is a combination between a partial oxyfuel combustion process with an amine post-combustion CO 2 capture technology. The CO 2 postcombustion capture by amine scrubbing considered in this work will be detailed in point This postcombustion capture technology applied to O 2-enriched air combustion is called partial oxyfuel combustion capture. Partial oxyfuel combustion flue gas treatment chain CO2 post-combustion capture (absorption-regeneration) Figure 5: Scheme of a CO 2 post-combustion capture applied to an O 2-enriched air combustion technology (adapted from Smart & Riley, 2012) 8

17 The hybrid process includes different steps: The first step is an Air Separation Unit (ASU) aiming to separate O 2 from N 2 of air. This will lead to the enhancement of O 2 to the incoming air. Types of ASU and their respective costs were given in the first report. Indeed the combustion can take place with enriched- O 2. The next step is the production of cement in the kiln with a treatment of denitrification, commonly Selective Non Catalytic Reduction (SNCR). The next two steps are also gas treatment steps designed to remove fly ashes and dust from the gas using commonly Electrostatic Precipitators (ESP) or baghouse filters and to remove SOx components by Flue Gas Desulphurization (FGD). Details of SNCR, ESP and FGD techniques were given in the previous report. Studies concerning the recirculation are under progress by ECRA to optimize the position and the rate at which the flue gas has to be recycled. This O 2-Enriched air combustion process allows increasing the CO 2 content of the flue gas (20-35 %<Y CO2<70 %) as it will be clearly demonstrated in CO 2 from the flue gas will be captured using different techniques cited in report 1: absorption processes, adsorption processes, membrane processes or carbonate looping systems. In our case, the capture technique used is the absorption-regeneration in amine solvent based systems. Without any supplementary CO 2 Purification Unit (CPU), the purity of the CO 2 generated by the postcombustion unit should be sufficient for CO 2 storage or re-use (no CPU envisaged in the case of postcombustion); e.g. for the CCS project at SaskPower s Boundary Dam power plant (Boundary Dam, Saskatchewan, Canada) no supplementary purification step is required before storage (Dubois & Laribi, 2015). An important research contribution in the applicability of oxyfuel technologies for cement plants is the Dania Pilot Plant Project (Carrasco-Maldonado et al., 2016), which is, to date, the first pilot test regarding the application of the oxyfuel technology for cement processing. In 2009 FLSmidth, Lafarge and Air Liquide started a joint study on the oxyfuel technology for cement plants. These study confirmed that the retrofitting of a conventional large scale calciner to partial oxyfuel operations is possible Hybrid process: advantages and purpose No reference on this hybrid technology applied to the cement industry, the only reference is linked to power plants through the ECO-SCRUB project (Adeosun et al., 2013) (Smart & Riley, 2012). This project aims to develop a low-cost option for carbon capture in existing modern coal-fired power plants. The concept uses a novel combination of techniques employed in CO 2 capture, such as oxygen enrichment and post-combustion solvent scrubbing, together with measures to increase efficiency, reduce steam consumption and generate power requirements. The project work involved development of the process through laboratory- and pilot-scale tests, simulation and modelling studies, literature reviews and collection of power plants performance data. Optimising the oxygen enrichment level and the rate at which the flue gas is recycled were key parameters for achieving satisfactory combustion characteristics, low NOx emissions and improved heat transfer characteristics, and avoiding ash deposition problems. Preferentially enriching the staged air system with oxygen was an actual option for NOx control. A pilot-scale combustion test facility with oxyfuel capability, simulated flue gas recycling and equipped with a pilot-scale amine 9

18 solvent scrubbing plant was used to evaluate the flexibility of the ECO-Scrub process and solvent performance. Membrane separation systems were shown to be promising alternatives to conventional CO 2 capture methods for enriched CO 2 flue gas Energetic effect The ECO-Scrub project demonstrated that oxygen enriched air combustion is technically and economically a viable option for carbon capture and storage for existing power plants producing CO 2 enriched flue gas that will have an energetic benefit for the overall process as represented by Figure 6. Reduce the size of the Air Separation Unit contrary to a total oxyfuel technology: the partial oxyfuel technology needs less oxygen enrichment so there will be less oxygen per unit of time circulating in the Air Separation Unit and consequently this reduces the size of an air separation unit (Smart & Riley, 2012). As the solvent is more charged with CO 2, it requires less energy into the reboiler for the regeneration leading to approx % reduction in energy demand (according to ECRA data), consequently, the amine degradation could be minimized. Studies from (Doukelis et al., 2009) demonstrated that less energy is required for the reboiler in the case of a partial oxyfuel combustion compared to a full oxyfuel combustion. Despite the complexity, the ECO-Scrub concept seems to have competitive advantages in retrofitting existing power plants since the combination of the partial oxyfuel and the post-combustion CO 2 capture technologies has beneficial results in the energy penalties and the electricity generation costs compared to the incorporation of exclusively one of the above technologies. Moreover, this technology requires less amine mass in the scrubbing plant for CO 2 capture compared to the pure post-combustion CO 2 capture technology. In addition of the project ECO-SCRUB, the paper from (Favre et al., 2009), related to a hybrid process combining oxygen enriched air combustion and membrane separation for post-combustion carbon dioxide capture shows the energetic advantages of working with an intermediate process between conventional and total oxyfuel combustion (Figure 6). E TOT (GJ/t CO2 ) E TOT (GJ/t CO2 ) Post-combustion capture Oxy-fuel combustion capture E TOT = E1 + E2 E1: O 2 production E2: CO 2 capture y O2 =0.21 Optimum between y O2 = 1.00 Post- and total Oxy-fuel Y CO2 Figure 6: Energetic effect of hybrid process (adapted from Favre et al., 2009) 10

19 As represented in Figure 6, the optimum energy corresponding to the sum of the energy for O 2 production from the ASU and the energy required for CO 2 capture by membrane separation is an intermediate case between a full oxyfuel combustion and a post-combustion capture process (conventional combustion). Based on this study, in this work, the applicability of this technology will be tested for the cement industry outcoming gas compositions: -An experimental screening of solvents will be used to determine the solvents absorption performances in CO 2 concentrated flue gases. -The total energy required for this hybrid process in the cement plant case will be estimated by absorption-regeneration simulations in Aspen Hysys and ECRA calculations sheet for E ASU. Results for both cases will be detailed further in this report Chemical effect Under CO 2 concentrated flue gases (high P CO2 ), the screening of different solvents have been tested in this work in order to evaluate the increase of the loading capacities of each solvent Δα CO2 when increasing P CO2 (Figure 7). Figure 7: Chemical effect of the hybrid process (adapted from L. Li et al., 2013)) Figure 7 taken from (L. Li et al., 2013) studies, shows that for the represented solvents ΔαCO 2 loading capacities increase with increasing P CO2. However, for the same increase of CO 2 partial pressure, we can have a different increase of the CO 2 loading capacity specific to each solvent (Δα 1 Δα 2 Δα 3). For this reason, during the screening of solvents we tested the absorption performances of different solvents for CO 2 contents Y CO2,in between 10 % and 60 % under partial oxyfuel conditions (see Figure 14). 11

20 2.3. CO2 capture using the absorption-generation process Absorption-regeneration process technology For CO 2 capture, several technologies can be used. In report 1, several techniques were listed and described like gas-solid adsorption or absorption, membrane processes but in this work our focus will be about the use of the reactive gas-liquid absorption into solvents for the CO 2 capture under partial oxyfuel conditions. The absorption-generation process is a double step process represented in Figure 8: The first step is the reactive gas-liquid absorption at typically 40 C: the flue gas coming from the cement industry is fed at the bottom of the absorber and flows to the top counter-currently with the solvent. It flows through the packing of the absorber, making contact with the solvent as it falls down. The CO 2 is absorbed by the solvent as the flue gas rises so that the gas that comes out at the top of the tower contains very little CO 2. Figure 8: Illustration of the Absorption-Regeneration process (CO2CRC, 2015) The second step is the CO 2 desorption at 120 C for aqueous MEA 30 wt%: in the second column, the temperature is increased by means of the energy produced in the boiler in order to facilitate desorption of the CO 2 (the reaction between CO 2 and amine being reversible). The solvent drops in the desorber and the CO 2 is released from solvent. Concentrated CO 2 is recovered at the top of the column and the lean solvent flows to the bottom of the desorber. The energy demand for the CO 2 regeneration have to be minimized when sizing installations and determination of operating conditions. The lean solvent is then recycled to the absorber. The solvent flows from one column to another, with steady supplies of fresh solvent to compensate the degradation of the absorption solution like thermal degradation, oxidative degradation and degradation due to irreversible reactions with CO 2. 12

21 Moreover an internal heat exchanger allows the lean solution giving calories to the rich solution in order to minimize the energy requirement Gas-liquid absorption solvents We can distinguish different categories of solvents: The first generation of solvents: including chemical solvents like monoethanolamine (MEA), diethanolamine (DEA), triethanolamine (TEA), potassium carbonate (K 2CO 3) and physical solvents like ethers, acetals or alcohols. The second generation of solvents: including blends of the first solvents like a-mdea representing MDEA + PZ, hybrid solvents representing mixtures between physical and chemical solvents. Third generation of solvents: including ionic liquids like imidazolium, pyrolidinium Fourth generation of solvents: including the enzymatic solvents; these solvents use the same principle as the biological fixation of carbon dioxide. The enzyme acts as a catalyst for CO 2 capture (CO 2 solution, 2015). It has to be noted that the 3 rd and 4 th generations are at a lower level of development than the first and second generations on which we will focus in the following pages First generation solvents properties Different categories of solvents are selected for the screening of solvents under high CO 2 concentrations, most of them are alkanolamines called in this report amines including primary alkanolamines, secondary alkanolamines and tertiary alkanolamines and presented in Table 2: Table 2 : Primary, secondary and tertiary alkanolamines configurations They are called "sterically hindered amines" (SHA) when additional branches are grafted at the hydrogen atoms which are substituted. A scheme summarizing the different types of amines is described in Figure 9. Figure 9: Examples of different types of amines 13

22 Depending on the structure of the amine, they react differently with CO 2, hence the different values of kinetic constants specified in Table 3. The CO 2 reaction with primary and secondary amines: Note: in the case of a primary amine, R 2 = H. The overall reaction CO 2-Amine is: 2 R 1R 2NH + CO 2 R 1R 2NCOO - + R 1R 2NH 2 + However, the reaction mechanism between CO 2 and alkanolamines is not as simple as described in the last equations, even for MEA. (Danckwerts, 1979) introduced a reaction mechanism proposed initially by (Caplow, 1968) which describes the reaction between CO 2 and alkanolamines via the formation of a zwitterion followed by the removal of a proton by a base, B. R 1R 2N + HCOO - + B R 1R 2NCOO - + BH + Amines Zwitterion Carbamate The stoichiometry of the overall reaction showed that 0.5 mol CO 2 /mol of amine is the maximum loading that the primary and secondary amines could achieve. Nevertheless, it has been proved that the absorption rate depends on CO 2 partial pressure (consequently the temperature). The CO 2 reaction with tertiary amines: Global reaction: R 1R 2NH + CO 2 R 1R 2 NCOOH R 1R 2 NCOOH+ R 1R 2NH R 1R 2 NCOO - + R 1R 2NH 2 + R 1R 2NH + CO 2 R 1R 2 N + HCOO - R 1R 2R 3N + CO 2 + H 2O HCO R 1R 2R 3NH + The stoichiometry of this reaction indicates that the theoretical absorption capacity of these amines is 1 mol CO 2/mol amine. CO 2 reactions with sterically hindered amines (SHA) (e.g. AMP) occur with the global mechanism: RNH 2 + CO 2 + H 2O RNH HCO 3 - SHA have the same theoretical absorption capacity as tertiary amines (1 mol CO 2/ mol amine). As mentioned, these different reactions lead to different kinetic parameters (Table 3). The simplified kinetic expression is used to represent the kinetic constant (k) relative to each type of amine studied in this work: the CO 2 reaction with an amine can be considered globally as an apparent 2 nd order reaction with first order relatively to each reactants CO 2 and amine: r=k.c CO2 C Amine (Versteeg et al., 1996). 14

23 Table 3: Kinetic constants relative to different types of amines Amine Category Kinetic constant at 25 C (m³/kmol.s) Reference PZ Cyclical diamine (Derks et al., 2006) MMEA Secondary amine 7940 (Patil et al., 2012) MEA Primary amine 5938 (Versteeg & Swaaij, 1988) DEA Secondary amine 3240 (Versteeg et al., 1990) TERA Non-cyclical tetramine 1252 (Amann & Bouallou, 2009) AMP Sterically hindered amine 810 (Xu et al., 1998) AHPD Sterically hindered amine 285 (Bougie & Iliuta, 2009) MDEA Tertiary amine 12 (Versteeg et al., 1996) CO 2 reactive absorption in amines depends on the structure and properties of each amine. The theoretical kinetic parameters of the reactions CO 2-Amine taken from the literature and listed in Table 3 will be illustrated experimentally in point of this report by the comparison of the solvents absorption performances Second generation solvents properties In order to enhance the absorption capacity of the amines, different amine blends can be realized: In order to combine interesting properties, i.e. good reaction kinetics of primary and secondary amines with the high absorption capacity of tertiary or sterically hindered amines, blends of amines could enhance the absorption. An amine blended with an activator and mixed with water in order to obtain activated amines is interesting. The most commonly used activators are Piperazine (PZ) (Figure 10, a), which is a cyclical diamine, and Triethylenetetramine (TETRA) which is a blend of 4 Tetramine isomeres (Figure 10,b). a) b) b) Figure 10: Semi-developed formulas of: a) PZ and b) TETRA 15

24 Hybrid solvents are a mix between physical solvents and chemical solvents: chemical solvents like monoethanolamine (MEA), diethanolamine (DEA) or triethanolamine (TEA) and physical solvents like ethers, acetals or alcohols. The schematic graphic represented in Figure 11 shows that these different types of solvents have good absorption capacities at different conditions. Indeed, performances of the different solvents categories depend on the operating conditions. Figure 11: Chemical solvents and physical solvents absorption capacities (Adapted from Bailey et al., 2005) From Figure 11 it stands out that physical solvents have high absorption capacities at high P CO2 but a better absorption capacity of the chemical solvent is attained at low P CO2. Combining the properties of a physical solvent at high P CO2 and the high absorption kinetics due to the chemical reaction with CO 2 makes the hybrid solvents interesting for further investigations Methodology of choice of the different solvents Many factors could have an important influence on the CO 2 absorption performances into amines (Dubois, 2013). Indeed, since the CO 2 reaction with an amine can be considered globally as an apparent 2 nd order reaction (partial order one for CO 2 and amine), the expression of the CO 2 transferred flow depends on the kinetic factor of the reaction CO 2-amine (Table 3), on the diffusivity and the solubility of the CO2 in the amine. Moreover, a less viscous solvent is more likely to reduce solvent pumping requirements. This explains the importance of the physico-chemical properties of the CO 2 solvent system. Besides, the cost and industrial availability of the solvent, the absorption (mechanism and kinetic constants) and regeneration (energy) performances, the chemical stability (reactions with other components (SOx, NOx, ) of the flue gas, irreversible reactions with CO 2, effects on equipment ) and the volatility are important factors to take into account in the choice of the solvent to use. Consequently, the choice of the solvents used for the screening tests in this work is done considering these different criteria. 16

25 2.4. Experimental screening of solvents under partial oxyfuel conditions Experimental device: The cables-bundle contactor Screening tests of solvents were conducted at the laboratory scale in a gas-liquid contactor, namely the cables-bundle contactor represented on Figure 12 that was not designed specifically for these tests but the purpose was to compare the solvents under the same conditions, even if these conditions are not optimized for maximizing the performances. The developed surface per volume unit (ensuring contact between the gas and the liquid) is relatively low compared to the surface that provides a packed contactor for example, thus, in practice this type of contactor is not used at industrial scale. Figure 12: Cables-bundle contactor scheme The contactor is a column containing 6 twisted cables (made of polypropylene, with a diameter = 1,6 mm) distributed around a central rod (of 20 mm of diameter) to ensure the strength of the column and allow the cable tension. The inside diameter of the column is 45 mm, for an effective height of 540 mm. Two circuits circulate along the device, the gas circuit and the liquid circuit: The gas flows from bottom to the top in the free space of the contactor entering in contact through the cables with the liquid flowing counter-currently (forming a cylindrical sheath with superficial wrinkles). The incoming gas is fed through two cylinders, one containing N 2 and the other CO 2. Each flow rate is measured and controlled by mass flow regulators. The two gases are then mixed together using a mixing valve before being sent to the humidification column. For the gas analysis, a part of the gas is taken to be analyzed at the inlet of the contactor (Y CO2,in% vol.) and at the outlet of the contactor (Y CO2,out %vol.). This gas is thus dried in the membrane dryers and 17

26 quantified by Infrared analysis. The analysis gas and the gas leaving the column are discharged into the air via a hood. Liquid analyses are also conducted by discontinuous manual samplings and quantifications using the TOC analyzer and ph measurements. CO 2 absorption into amines is an exothermic reaction, therefore temperature measurements were conducted at the inlet and the outlet of the contactor to evaluate an eventual thermal effect. Liquid and gas flow rates are kept constant in order to always work in the same hydrodynamic conditions within the contactor (L = 185 ml/min equitably distributed along the 6 cables; G = 12.9 Nl/min in dry conditions). Moreover a thermostat keeps the temperature of the liquid equal to 25 C at the inlet (T in). The hydrodynamic conditions are maintained constant all along the tests. For each test, the IR analyzer (brand Emerson, X-STREAM X2GP model) performs continuous measurements of the CO 2 content at the inlet and outlet of the contactor (respectively Y CO2,in and Y CO2,out). The absorption efficiency can then be calculated. This is defined as the difference between the amount of CO 2 in the input and output, divided by the amount fed: A (%)= Gin YCO2,in - Gout YCO2,out G in Y CO2,in 100 A (%)= G CO2,abs G in Y CO2,in 100 G in = molar gas flow rate at the inlet of the contactor (mol/h) G out = molar gas flow rate at the outlet of the contactor (mol/h) A more accurate calculation taking into account the humidity difference between the inlet and the outlet of the contactor: A (%)= Y CO2,in - Xout X in 1-Y CO2,in X in 1-Y CO2,out X out Y CO2,out Y CO2,in 100 with X in = 1 hum in, hum in is the relative moisture calculated with the temperature of the circulating (T hum) water and X out = 1 hum out, hum out is calculated using the average temperature between the top and the bottom of the contactor Tav = Tin+Tout. 2 Two types of analysis were conducted during experiments tests with this device: Liquid phase analysis: This analyzer measures the carbon content over time. Samplings every 15 minutes are conducted during the semi-continuous tests (which will be detailed further), the CO 2 loading of amine solutions is determined using a TOC-VCSH Shimadzu analyzer. The calibration of the TOC analyser is carried out using a hydrogen phthalate of potassium solution standardized to 1000 mg C/l. Specific dilutions of this solution are made to obtain solutions with carbon content between 100 and 1000 mg C/l. After analysis of these solutions, a calibration straight line is obtained between the peak areas measured by the device to the imposed concentrations. Total Carbon (TC) includes all forms of carbon present in solution, both organic (TOC) and inorganic (IC). The CO 2 concentration can be calculated using the difference between TC of fresh and loaded solution: C CO2 ( mol CO2 l )= TC loaded solution-tc unloaded solution 1000 MM carbon 18

27 Note: TC expressed in mg C/l. At the equilibrium, the CO 2 loading representing the maximum quantity of CO 2 that has been absorbed by a mole of amine is quantified: α CO2 ( mol CO2 mol amine )= C CO2 C amine But this quantity α can be determined at every time of the absorption test. Gas phase analysis: During all the tests, the concentration of the gas phase at the inlet and outlet of the absorption column is measured (the gases are withdrawn with a three-way valve, 70l/h for each). These samples are dried (a membrane dryer is placed before analysis) and then analyzed continuously through an IR analyzer (brand Emerson, X-STREAM X2GP model). The accuracy of the CO 2 content measurements is 0,01 %vol. The analyzer calibration is performed with pure N 2, pure CO 2 and a standard mixture containing 15% of CO Selected solvents The used solvents during the absorption tests are a combination between the products described in Table 4, Table 5 and Table 6, with three categories of solvents standing out: Conventional solvents like MEA 30 wt%, DEA 30 wt% Table 4: Amine solvents Chemical solvents Product Manufacturer Purity Quality Developed formula MEA Merck 99% For synthesis MMEA Merck 98% For synthesis DEA Merck 99% For synthesis MDEA Merck 98% For synthesis AMP Merck ~95% For synthesis AHPD VWR (BDH PROLABO) 100% AnalaR NORMAPUR 19

28 Activated solvents using a blend of [amine+ activator] like [MEA 30% + TETRA 5%] or [AMP 30% + PZ 5%] using two types of activators PZ and TETRA. Likewise, the effect of PZ 10% and TETRA 30% is studied separately (see Figure 15). Table 5: Activator used in the solvents [amine+activator] Chemical solvents Product Manufacturer Purity Quality Developed formula PZ Merck 99% For synthesis TETRA Merck ~95% For synthesis See Figure 10 b) Other particular solvents tested in our experiments are a combination between an amine (+water) which is a chemical solvent, and a physical solvent which in our case is an ether named TOU (2,5,7,10-Tetraoxaundecane), [amine+tou]. These solvents are called hybrid solvents e.g. [MEA 30% + TOU 35%]. Table 6: Physical solvent used Physical solvent (ether) Product Manufacturer Purity Quality Developed formula TOU Lambiotte 99,5% Industrial All the scrubbing solvents are aqueous Note: see nomenclature for abbreviations Types of tests conducted Different types of tests were conducted in the context of a master thesis (Pierrot, 2015): A. Continuous tests In these tests the liquid and the gas are continuously fed into the contactor. There is only one circulation of the liquid through the column. Some minutes of test are sufficient to reach a steady state for the CO 2 absorption efficiency (A and G CO2, abs). These tests lead to the comparison of kinetic performances of different solvents. Continuous tests without pre-loading (10% <Y CO2,in < 60% ; α=0): The flue gas outcoming from a O 2-enriched air combustion process can contain up to 60%vol of CO 2. For this reason, tests for each solvent were conducted by varying Y CO2,in from 10 to 60%vol at intervals of 10% in order to observe the influence of a high CO 2 content on the absorption performances. The range of CO 2 concentrations chosen is not limited to hybrid conditions. The tests at 10% allows obtaining the behavior of amines for a gas issued from power plant, the 20 and 30% points correspond 20

29 to conventional cement plants conditions and the higher CO 2 contents (40 to 60%) are relative to the hybrid conditions as showed in Figure 14. Continuous tests with pre-loading (10% <Y CO2,in < 60% and α 0): For some solvents, additional continuous tests were conducted by varying the initial CO 2 loading. The solvent selected was previously loaded into CO 2 by bubbling pure CO 2 (10 l/min) in a double stirred cell contactor. Then the pre-loaded solvent (whose initial value of α CO2 is measured before the absorption test) is used in the cables-bundle contactor with Y CO2,in set at 20, 40 and 60%vol. The results corresponding to these tests are available in Annex A. B. Semi-continuous tests (Y CO2,in =40% and α 0) In these tests the gas is continuously fed into the column but the liquid (without any feed) is recycled. The recirculation in the contactor of a constant volume of 1,3 litre of solvent has been achieved. The objective is to characterize the temporal evolution of Y CO2,out, G CO2,absorbed and the CO 2 loading of the solvents. If the test is achieved with a sufficient duration, an equilibrium can be obtained between gas and liquid phases, leading to CO 2 absorption capacity. As intermediate value of the working range of continuous tests, considering the flue gas composition from partial oxyfuel combustion, Y CO2,in was fixed to 40% for all the semi-continuous tests. A liquid sample is withdrawn every 15 minutes to observe the evolution of ph and the CO 2 loadings via the TC and IC measurements. These tests simulate the effect of a partially regenerated solvent (with a residual CO 2 loading of the solution at the inlet of the absorption column) on the absorption performances Experimental results The most relevant results concerning the screening tests of solvents under high CO 2 concentrations are presented in this section for the 3 types of tests conducted Continuous tests without pre-loading Continuous tests with simple solvents The reference taken all along the tests is the aqueous MEA 30wt%: all results and absorption performances will be compared to the conventional solvent ones. As a base case presented in Figure 14, the absorption performances with MEA increase regularly when increasing Y CO2,in representing conditions for conventional power plant combustion (10% <Y CO2,in < 20%), conventional cement plant combustion (20% <Y CO2,in < 40%) and finally O 2-enriched air combustion (40% <Y CO2,in < 70%). Figure 13 and Figure 14 show the results of continuous tests for MEA 30%. Figure 13 shows that the evolution of the absorption rate is relatively constant with Y CO2,in (A = approximately 40%). Figure 14 illustrates the increase of CO 2 absorption molar flow rate linearly with Y CO2,in. 21

30 A (%) G CO2, abs (mol CO 2 /h) Power plants (previous studies) Cement plants (conventional) Cement/Power plants (O2-enriched air conditions) Y CO2,in (%) Y CO2,in (%) Figure 13: Effect of Y CO2,in on the absorption rate A (%) solution for a MEA 30% solution. Figure 14: Effect of Y CO2,in on G CO2,abs for a MEA 30% solution. Other conventional solvents were also tested and compared. The graphics of G CO2,absorbed= f(y CO2,in) for each simple solvent are presented in Figure 15. This chart confirms the interpretations of Figure 14 that the CO 2 absorbed molar flow rate increases almost linearly with Y CO2,in. YCO2, in (%) Figure 15: Results of the continuous tests with simple solvents MMEA 30% and PZ 10% present the highest absorption performances, even more important than the reference MEA 30%, with respectively 50% and 23% of increase compared to MEA 30%. In opposition, lower absorption performances than MEA are shown for TETRA, AMP, DEA, MDEA and AHPD; this is partially linked to the kinetic constants relative to different types of amines given by Table 3. Indeed, at 298K, MMEA and PZ have more important kinetic constants than MEA and the other amines (TETRA, AMP, DEA, MDEA and AHPD) have lower values of kinetic constants than MEA. Besides, this behavior is also linked to the different properties of the amines (diffusivity and solubility) and to the amine chemical structure (primary, tertiary, sterically hindered ) which have an influence on the CO 2-amine absorption performances. Continuous tests with activated solvents Activated solvents are aqueous blends between an amine and an activator. In this case the activators tested are PZ and TETRA. Results are presented in Figure 16, Figure 17 and Figure

31 G CO2, abs (mol CO2/h) G CO2, abs (mol CO 2 /h) G CO2,abs (mol CO 2 /h) MMEA 30% MMEA 30% + PZ 5% 10 MMEA 30% + TETRA 5% Y CO2, in (%) Figure 16: Results of the continuous tests with activated solvents for [MMEA+activator] AMP 30% AMP 30% + PZ 5% AMP 30% + TETRA 5% Y CO2, in (%) Figure 17: Results of the continuous tests with activated solvents for [AMP+activator] DEA 30% DEA 30% + PZ 5% DEA 30% + TETRA 5% Y CO2, in (%) Figure 18: Results of the continuous tests with activated solvents for [DEA+activator] From these figures, the observations that can be highlighted in the case of the continuous tests with activated solvents are mainly the importance of the activation effect shown for 3 solvents compared to MEA: MMEA, DEA and AMP. This activation effect is more pronounced for DEA and AMP. In all cases the best activation effects are provided by PZ activator. These results are confirmed by Figure

32 YCO2, in (%) Figure 19: Summary of all the results for activated solvents Figure 19 shows that [MMEA+PZ], [MMEA+TETRA], [DEA+PZ] and [AMP+PZ] give better absorption performances than MEA, namely 50% increase for [MMEA+activator] and 12% increase for [DEA+PZ] and [AMP+PZ] compared to MEA. MMEA gives the best absorption performances without necessity of using an activator. Consequently, as shown in Figure 16, Figure 17 and Figure 18, the interest of PZ activator is confirmed. Continuous tests with hybrid solvents Hybrid solvents are also aqueous blends of amine solvents (chemical solvent) with a physical solvent (an ether, TOU in our case). YCO2, in (%) Figure 20: Continuous tests with hybrid solvents The tests conducted with hybrid solvents presented in Figure 20 showed mainly that there is a promoting effect of the addition of TOU to MEA. Indeed, when working with [MEA+TOU] the absorption capacities are slightly increased compared to MEA. It is important to mention that during the tests, two phase change phenomena are observed with TOU that could be beneficial in energy consumption: precipitating phenomenon and demixing phenomenon. The demixing phenomenon was observed when manipulating two types of solvents [MEA 30% + TOU 35%] and [DEA 30% + PZ 5% + TOU 35%] after 30 minutes of recirculation, dividing the solution in two phases: the lean phase in CO 2 (and rich phase in TOU) namely the light phase and the rich phase in CO 2 (rich phase in MEA) namely the heavy phase, as shown in Figure 21. The phases compositions were 24

33 G CO2, abs (mol CO 2 /h) α CO2 (mol CO 2 /mol amine) confirmed thanks to TOC and also Fourier Transform Infrared Spectroscopy (FTIR) Analysis (Gervasi, 2013). FTIR gives quantitative and qualitative analysis for organic and inorganic samples. Fourier Transform Infrared Spectroscopy (FTIR) identifies chemical bonds in a molecule by producing an infrared absorption spectrum. The spectra produce a profile of the sample, a distinctive molecular fingerprint that can be used to screen and scan samples for many different components. Thus, FTIR is an effective analytical instrument for detecting functional groups and characterizing covalent bonding information. Lean phase in CO 2 Rich phase in TOU Rich phase in CO 2 Rich phase in MEA Figure 21: Visualization of the demixing phenomenon for [MEA 30% + TOU 35%] (after 45 minutes) This demixing phenomenon showed that it could be a great potential for reducing the energy regeneration by separating the two phases using a decanter and regenerating only the CO 2-rich phase. The positive effect of demixing solvents will be deeply investigated in a new PhD thesis of the ECRA CHAIR at UMONS (Seloua MOUHOUBI) Semi-continuous tests Semi-continuous tests with simple solvents The most relevant conclusions arising from the comparison of simple solvents results of semicontinuous tests are that after 20 minutes of test, TETRA presents the most important CO 2 loading compared to the MMEA which had the best performances for the continuous tests and compared to the reference MEA. Therefore the representations of the temporal evolutions of G CO2,absorbed and α CO2 of this solvent compared to MEA are given respectively by Figure 22 and Figure TETRA 30% MEA 30% TETRA 30% MEA 30% Time (min) Time (min) Figure 22: Temporal evolution of G CO2,abs for the semi-continuous tests with TETRA and MEA. Figure 23: Temporal evolution of α CO2 for the semi-continuous tests with TETRA and MEA. 25

34 The CO 2 molar flowrate measured (presented in Figure 22) decreases with time as the solution becomes increasingly loaded in CO 2 and this can be linked to the increase of the loading of the solvents (presented in Figure 23) as long as recirculated in the column. Simple solvents Activated solvents Table 7: CO 2 loadings of the solvents at the end of the tests (after 90 min) Solvents αco2 (mol CO2/mol amine) at the end of the test MEA 30% MMEA 30% DEA 30% MDEA 30% AMP 30% AHPD 30% PZ 10% TETRA 30% [MMEA 30%+PZ 5%] [MMEA 30%+TETRA 5%] [DEA 30%+PZ 5%] [DEA 30%+TETRA 5%] [AMP 30%+PZ 5%] [AMP 30%+TETRA 5%] The comparison of the CO 2 loading highlighted also that even if PZ has high CO 2 loading TETRA allows absorbing more CO 2 than the other solvents tested and presents also the highest α CO2 at the end of te test. Results of CO 2 loadings of activated solvents are also presented in Table 7 and confirm the conclusions drawn from Figure 24. Semi-continuous tests with activated solvents Results of absorption performances evolutions with time at 3 moments of the recirculation tests are presented in Figure 24: GCO2, abs (mol CO2/h) Figure 24: Semi-continuous tests with activated solvents 26

35 G CO2,abs (mol CO 2 /h) At the beginning, before the recirculation is applied, [MMEA + PZ] presents the best CO 2 loading, this confirms the results elucidated by the continuous tests. Moreover, after 30 min of recirculation, [DEA + PZ] has the most important absorption performance. [AMP + PZ] presents good absorption performances all along the recirculation tests with a significant CO 2 loading at the end of the test Summary of the important results Figure 25 summarizes the experimental screening of solvents with an evaluation of the absorption performances of a wide variety of solvents in high CO 2 contents conditions. The comparison has been done with the conventional solvent MEA (30%). 9 YCO2, in = 40% MEA 30% MMEA 30%+PZ 5% DEA 30% + PZ 5% AMP 30% + PZ 5% TETRA 30% t=0 min t=30 min t=90 min α CO2 =0 α CO2 0 Figure 25: Best solvents screened in high CO 2 concentrations compared to the reference MEA 30% The best solvent at t=0 (α CO2=0) having the highest G CO2,abs was [MMEA 30% + PZ 5%] but after the CO 2 loading (t=90 min), the best solvents become [AMP 30% + PZ 5%] and TETRA Calculation of the required global energy for the partial oxyfuel combustion process Partial oxyfuel technology was considered as the most likely configuration for retrofitting a cement plant, requiring fewer modifications, as the calciner and one preheater string could be operated in oxyfuel conditions and the residual kiln plant in conventional operational mode (IEA, 2008). This configuration is illustrated here below (Figure 26). 27

36 Figure 26: Basic layout of partial oxyfuel configurations In the technical report from ECRA, various configurations (regarding for example heat exchangers locations) are discussed, leading to different thermal energy efficiencies Calculation of the required energy for the Air Separation Unit under partial oxyfuel combustion conditions Thanks to our ECRA-UMONS collaboration through the ECRA Chair, calculation sheets corresponding to a O 2-enriched air combustion represented in Figure 27 were provided by ECRA. These allow to calculate: A. The composition of the flue gas outcoming from an O 2-enriched air combustion (identical composition of the recirculated and the extracted flue gases) B. The O 2 flow rate to be supplied by the ASU and the energy necessary for this production. Figure 27: Scheme of an O 2-enriched air combustion These calculations include as a first and sufficient starting-point global balances of the major components in the clinker burning process. A schematic view taking into account the inputs into the kiln is represented in Figure

37 Figure 28: Inputs of the O 2-enriched air combustion In this configuration the sum of the flow rates of air and oxidizer inputs and recirculated flue gas is assumed to be constant. The calculation sheet is structured as follows: - Input data - Calculations - Mass balances - Results Input data Production capacity, false air ingress (with a reference value of 6% of the raw gas), coal and raw material compositions, fuel energy demand and air input are the input parameters for the calculations. The air input means the air which is provided to the process in addition to oxidizer (oxygen composition of 95 %, coming from ASU) and recirculated flue gas. To determine the energy necessary in the Air Separation Unit (E ASU) several scenarios could be envisaged by variating the input parameters. If considering a fixed oxygen input demand at the inlet of the kiln, as shown in Table 8 (including primary gas and false air whose flow rates are fixed), parameters will variate in order to keep the oxygen input constant. The value of 0.15 m 3 stp /kg clinker was suggested by ECRA. In consequence in the ECRA calculating sheets variating the Air input from 0 to 0.7 m 3 stp /kg clinker leads to different additional O 2 from ASU (oxidizer). The total oxygen demand is defined by the combustion, corresponding to oxygen needed to oxidize completely the fuel. In order to avoid reducing atmospheres into the kiln (with a negative effect on the quality of the product), oxygen levels at different locations in the plant are required and in general a 3% excess oxygen in the flue gas is satisfactory. 29

38 Clinker production capacity Table 8: Input parameters 3000 t/day kg/h Process flow Raw material 1.6 kg/kg clinker kg/h False air m 3 stp /kg clinker 8400 m 3 stp/h Primary gas m 3 stp /kg clinker m 3 stp/h Cooling gas input 0.89 m 3 stp /kg clinker m 3 stp/h Oxygen demand 0.15 m 3 stp /kg clinker m 3 stp/h Air input Varied from 0 to 0.7 m 3 stp /kg clinker Varying from 0 to m 3 stp/h All the input parameters volumes are calculated according to the production capacity of the cement plant (3000 t clinker/day) Calculations For the oxidizer, we have to calculate for each case the oxidizer volume flow rate which depends on the additional O 2 necessary to be provided from the ASU (O 2 composition from ASU fixed to 0.95) in order to maintain constant the oxygen input in the kiln meanwhile the air input variates: Additional O2 from ASU Oxygen input demand oxygen from air Oxidizer volume = = Mass balance calculations They concern the concentration (in vol.%) the absolute volume (m 3 /h), the specific volume (m 3 /kg clinker) and the absolute mass (kg/h) (taking into account densities) flow rates which can be calculated for each component at the inputs. The mass balances between the inputs and the outputs can be calculated for each flow to deduce the flue gas compositions outcoming from the combustion Results For each variation of the air input result a different recirculation rate, a different flue gas composition and also a different Energy demand for ASU. The results are expressed in terms of Energy demand for ASU, this energy demand is calculated considering the additional O 2 necessary to be provided from the oxidizer in order to maintain constant the oxygen input in the kiln and based on 220 kwh/t O2. This last value has been verified in the literature sources: - for conventional ASU (cryogenic distillation), according to Linde group studies, a conventional ASU consumes 245 kwh/t O2 (Linde, 2015). 30

39 - Air liquid studies (Air liquid, 2009) showed that this power consumption for cryogenic air separation is close to 200 kwh/t O2, which confirms the range reported by ECRA (ECRA Report phase II, 2009). This energy demand for the Air Separation Unit can be estimated as (see table of results below): - E ASU (MW) = ASU Consumption (kwh/t O2). O 2 absolute mass flow rate produced (t O2/h) - a function of CO 2 in the flue gas: Energy demand ASU (MW) E ASU (GJ/t CO2) = CO2 absolute mass flow rate ( tco2 ).1000 s The cases obtained from the exploitation of the ECRA calculation sheets are studied variating the air input fraction from 0 to 0.7, this variation leads to a different energy demand for the ASU in order to satisfy the oxygen input demand fixed at the beginning to 0.15 m³ stp/kg clinker and a different flue gas composition for the post-combustion process. Results for each case are summarized in Table 9. The Recycle rate is defined as: Flow rate of recirculated flue gas / Flow rate of flue gas out with: Flow rate of recirculated flue gas = Flow rate of flue gas out Flow rate of extracted flue gas Air input flow rate (m 3 stp/kg clinker) Oxydiser flow rate (m 3 stp/kg clinker) Table 9: Recapitulative table of the studied cases Recycle rate (recirculated Y O2 Y CO2 flow rate/total flue gas out flue gas out flue gas out (% mol) (% mol) flow rate) E ASU (MW) E ASU (GJ/t CO2) In the investigated conditions, as previously discussed, it has to be mentioned that the flue gas out contains from 41 to 77% of CO 2 for limit cases (0 oxidizer input to maximum oxidizer input). Corresponding graphics: Various curves given in Annex C represent the exploitation of the results obtained in the table from the ECRA calculation sheets. The important result concerns the evolution of E ASU=f (Y CO2) represented in Figure

40 E (GJ/t CO2 ) Y CO2 (% vol) in the flue gas out Figure 29: Evolution of E ASU with Y CO2 in the flue gas out Note: calculations could not be done for a total oxyfuel combustion case due to limitations in the ECRA calculation sheet. This graphic gives the same evolution as the one elucidated by literature sources in Figure 6. The Energy necessary for O 2 production increases when working with more CO 2 concentrated flue gases Simulation of the CO2 capture using the absorption-regeneration process: Estimation of the regeneration energy under partial oxyfuel conditions Operating conditions and description of the flowsheet Simulations of the CO 2 capture flowsheet were conducted in Aspen Hysys V8.6 also in the context of a master thesis (Pierrot et al., 2015) and correspond to a pilot unit used in the European projects (CASTOR/CESAR), applied to the case of the Brevik cement plant taken as base case for flue gas compositions. Since all the design and operating parameters are available in literature (which is not the case with most of the other installations), the CASTOR/CESAR pilot unit was selected as our case study with definite operating conditions. The pilot is sized to handle a flow of 5000 Nm³/h at the inlet of the treatment line resulting in a flow of G in=4000 m³/h at the inlet of the absorption column (after removal of a large portion of water, cooling and compression). The absorption ratio is defined to 90%mol (90% of the molar flow rate of CO 2 entering the absorption column is recovered at the outlet of the regeneration column). These two data lead to a fixed amount of CO 2 absorbed (or regenerated G CO2,reg) of 1.5 t CO2/h. The CO 2 recovered purity is fixed at 98%mol (classical value). Solvent selected: MEA 30% Modelling characteristics used for the simulations: Aspen Hysys V8.6 Acid gas package Thermodynamic models: Peng-Robinson (gas) and e-nrtl (liquid) Reactions sets included in the package (validated by literature). 32

41 The Table 10 specifies the dimensions of the absorber and the stripper, and the operating conditions in each column: Table 10: Dimensions and operating conditions of the columns Absorber Stripper D (m) H (m) 17 (17 x 1 m) 10 (10 x 1 m) Packing used Random packing IMTP 50 Random packing IMTP 50 T up,liquid ( C) P bottom (kpa) The internal heat exchanger fixes the temperature of the liquid at the inlet of the stripper to 110 C. In addition, the pressure drop in the absorber and the stripper are fixed to 0,5 kpa/m Flowsheet description The global flowsheet of the pilot simulated in Aspen Hysys TM is represented in Figure 30: Figure 30: Global flowsheet of the simulated pilot The different parts will be zoomed and detailed hereafter. Different blocs could form the overall flowsheet: (1) First step: conditioning of the stream entering the absorption column (Figure 31) The flue gas comes out from the Brevik s cement plant at 165 C and 100 kpa. It is cooled down to 40 C (E-101) and sent in a first flash unit to separate the condensates from the gas. Then the gas is compressed to 120 kpa (K-100). This compression causes an increase of the gas temperature, so the gas has again to be cooled to 40 C (E-102). A second flash unit separates the liquid and vapour phases in order to prevent the injection of condensates into the absorption column. 33

42 Figure 31: Conditioning step of the stream entering the absorption column (2) Second step: CO 2 Absorption (Figure 32) Figure 32: Absorption column The conditioned gas represented by gas to treat is sent to the bottom of the absorber at 4000 m³/h where the CO 2 is captured counter-currently by the amine-based solvent, MEA 30%wt as fixed in the operating conditions, represented by lean solution in abs (fed at the top of the absorber) in the flowsheet. At the outlet of the column, the gas treated contains hardly any CO 2, and the Rich solution before preheat contains the amine solvent which has already absorbed the CO 2 in the column. Due to the rate-controlled nature of the CO 2 absorption/desorption processes, classical equilibrium stage models are inadequate for describing the behaviour of chemical absorption and desorption processes into the columns. A more effective modelling method used in this case is the non-equilibrium rate-based approach, which considers the effects of the various driving forces across the vapor-liquid interface. (3) Third step: Pumping and preheating of the rich solution (Figure 33) Figure 33: Pumping and preheating of the rich solution Once the CO 2 has been absorbed, the rich solution is pumped (P-100) and preheated to 110 C with the internal heat exchanger (LNG-100). 34

43 (4) Fourth step: Stripping of the gas or solvent regeneration step (Figure 34) The CO 2 is stripped out the solvent thanks to the heating power brought by the reboiler, the CO 2 is then recovered at the top of the regeneration column denoted by produced CO 2 in the flowsheet. The regeneration occurs at 200 kpa and 110 C considering that the boiling point of MEA at such pressure level equals to 120 C. The CO 2 purity of the outlet gas flow is fixed at 98%mol (conventional value). The condenser cooling energy (Q condenser) is therefore automatically calculated in order to satisfy this fixed specification. The regeneration energy is determined for each case of reboiler energy Q reboiler with fixed G CO2,regen = 1500 kg CO2/h by: E regen ( GJ Q GJ reboiler ( h ) = ) t CO2 G t CO2,regen ( CO2 h ) Figure 34: Regeneration step The remaining CO 2 loading in the regenerated solvent denoted by lean solution before the cooling in the flowsheet, corresponds to this heating power of the reboiler. (5) Fifth step: preheating of the rich solution and adjustment of the liquid flow (Figure 35) At the outlet of the absorber and the regenerator, possible losses in amine and water could happen, an alternative solution is to add a makeup unit. This adjustment system is composed by the makeup unit (MAKEUP-100) and the purge (PURGE-1) which automatically adjust the total flow rate by adding and purging if necessary water and MEA in order to keep the desired concentration of amine in the system constant. No influence on the results is caused by the use of this makeup unit. Figure 35: Cooling of the lean solution and "makes up" For energy savings purposes, the lean solution crosses the rich solution in the internal heat exchanger in order to preheat it. After the makeup, the lean solution is cooled down to 40 C before entering again in the absorption column and beginning a new absorption-regeneration cycle. 35

44 Compositions and conditions of the flue gas for the base case (Brevik cement plant) and for the case of partial oxyfuel combustion The Table 11 shows the different compositions and operating conditions of the gas entering the absorber which are tested in the simulations. Three study cases are considered: Inlet compositions of Brevik s cement plant Inlet compositions provided by ECRA from results of partial oxyfuel combustion simulations Intermediate inlet compositions defined by interpolation between the two previous cases regarding Y CO2,in Inputs of the simulations Y in the gas to treat Table 11: Compositions and conditions of the gas to treat-different cases Conventional composition ECRA simulations Brevik (base case) Intermediate composition Hybrid 1 Hybrid 2 Hybrid 3 Air input flow rate (m 3 stp/kg clinker)=0.6 Air input flow rate (m 3 stp/kg clinker)=0.4 Air input flow rate (m 3 stp/kg clinker)=0.2 Mole fractions N CO H 2O O Simulated only for the base case Operating conditions of the absorber CO 1.33E-03 / / / / NO E-06 / / / / SO E-04 / / / / NO 4.74E-04 / / / / Ar / T up,liquid ( C) 40 P bottom (kpa) 120 G in (m³/h) ±4000 NB: The air input flowrates indicated for the ECRA simulations together with Y O2 and Y CO2 in the gas to treat correspond to cases studied in Table 9. The ECRA simulations do not calculate NOx, SOx and CO concentrations Simulation results: Study of the influence of Y CO2,in on the regeneration energy The different gas compositions presented in Table 11 were then fixed at the entrance of the absorption column with the design parameters and operating conditions defined in Table 10. The results are summarized in Table

45 Table 12: Effect of the composition of the flue gas on the energy regeneration costs Y CO2, in (%) Base case Intermediate Hybrid 1 Hybrid 2 Hybrid 3 case L (m³/h) Gin (m³/h) G CO2, regen (kgco 2/h) A (%) Q condenser (kj/h) 2.926E E E E E+06 Q reboiler (kj/h) 5.086E E E E E+06 E condenser (GJ/ t CO2) E regen (GJ/ t CO2) α rich α poor Δα E regen saving/base base (%) Table 12 and Figure 36 show important results concerning the evolution of the energy required for the regeneration step with respect to the flue gas content on CO 2. For a fixed and constant G CO2, regen, the regeneration energy is reduced when Y CO2,in increases. This can be explained if the CO 2 loading values are observed. α rich increases when Y CO2,in increases. For each case, L and Δα are identical. Therefore, it seems logical that the α poor increases when the CO 2 content of the feed gas increases. Consequently, the energy required to regenerate the solvent decreases. The advantages of partial oxyfuel combustion conditions are clarified at this point: for the same amount of G CO2, regen = 1,5 t CO 2/h we have a saving of 24% on the CO 2 Regeneration energy if the CO 2 content is increased from 20 to 44 %. However, we have also to take into account the cost of O 2 production in the air separation unit to realize an O 2-enriched air combustion which needs more oxygen-enriched air to produce more concentrated flue gas in CO 2 outlet of the kilns. G CO2, regen = 1,5 t CO 2/h YCO2, in (% vol) Figure 36: Evolution of E regen with Y CO2,in to capture 1,5 t CO 2/h 37

46 E (GJ/tCO2) The interest of the partial oxyfuel combustion technology applied to CO 2 capture is demonstrated in term of E regen as elucidated in (Doukelis et al., 2009) but naturally the impact on the global chain (from O 2 production to CO 2 conversion) will have to be evaluated for the specific case of cement industry with optimum conditions to be investigated Global energy required for partial oxycombustion process The overlap of Figure 36 and Figure 29 is illustrated by Figure 37 and leads to a slight increase of the total energy required for a post-combustion capture technology applied to O 2-enriched air combustion called partial oxyfuel combustion, in the case of a fixed amount of CO 2 recovered and a fixed O 2 input demand into the kiln E regen for conventional post-combustion EASU Easu+Eregen Eregen Y CO2 (% vol) in the flue gas out Figure 37: Global energy required for partial oxycombustion process Using the data of further simulations carried out by ECRA will interesting to complete this graph with the full and partial oxyfuel parameters (O 2 input and Recycle) like presented in Figure Conclusions and perspectives of part 2. The main conclusions drawn from the study of the post-combustion CO 2 capture applied to O 2- enriched air combustion are that this technology could also be applied to the cement industry but further studies are necessary to evaluate the impact of this technology in the global chain. The screening of solvents identified the most interesting solvents for CO 2 capture in these conditions: the continuous tests showed that MMEA presents the best absorption performances, without necessity of using an activator and PZ gives the best activation effect; the tests with the evolution of the solvents CO 2 loading showed that TETRA gives the best absorption capacity and activated AMP also showed interesting results all along the tests. 38

47 Concerning the evaluation of the regeneration energy, simulation results showed that the regeneration energy is reduced when Y CO2,in increases. The calculation of the global partial oxycombustion energy requirement has been elaborated and confirms quite similar results as the ones presented in Figure 6 namely the interest of higher CO 2 content in the gas to treat in terms of energy consumption. As perspectives of this work, the next step will consist in micro-pilot tests (Figure 38) of absorptionregeneration for the best solvents screened in the context of an internship beginning in March These absorption-regeneration tests will be also simulated with Aspen Hysys software. Figure 38: Micro-pilot for absorption-regeneration tests 39

48 3. Simulation of a CO 2 purification process related to a gas coming from cement industries applying full oxyfuel combustion 3.1. Oxyfuel combustion technology applied to the cement industry Oxyfuel technology relies on pure oxygen instead of ambient air for combustion (see Figure 39). For this purpose the nitrogen is removed in a separation plant (ASU) from the air prior to being supplied to the kiln). Consequently the concentration of carbon dioxide in flue gas is increased significantly. The gas properties are different from those in conventional kiln operation with a corresponding impact on the clinker burning process. Also, the theoretical flame temperature in the sintering zone rises compared to ambient-air-based combustion. To maintain an appropriate flame temperature, part of the flue gas has to be recycled while the recirculation rate adjusts the combustion temperature. Oxyfuel technology is one of the most promising technologies for carbon capture and purification processes allowing a more concentrated CO 2 in the flue gas (70 % for power plants and 83 % for cement plants). Consequently, oxy-combustion and the related CO 2 capture techniques are trending and widely studied topics especially in the case of power plants. However, applying oxy-combustion to the cement industry is highly challenging due to the significant differences between cement plant and power plant flue gases. Indeed, cement plant flue gases have a higher CO 2 concentration essentially because of the decarbonation step during the burning process, and have also higher pollutants concentrations as cement plants usually use alternative fuels in their kilns. As a result, oxycombustion in the cement industry will require specific researches and developments as many of the fundamentals of oxyfuel cement plant still need to be better understood. Figure 39: Oxyfuel combustion technology (ECRA technical report phase 3, 2012) From Figure 39 we can see that the ASU provides oxygen injected directly into the pre-calciner and into the kiln where the production of clinker takes place. The raw meal is grinded and preheated before entering the kiln. A gas recirculation in order to concentrate the flue gas in oxygen takes place. 40

49 The flue gas recirculation loop provides an additional degree of freedom to control the temperature profile in the kiln by adjusting the O 2 content in the oxidant. The CO 2 rich flue gas quits the preheater and undergoes series of purification units before final compression. A more simplified scheme is represented in Figure 40: Figure 40: Basic layout of full oxyfuel configuration (Carrasco-Maldonado et al., 2016) The clinker cooler symbolized in Figure 40 is added to the flue gas recirculation circuit for energy recovery purpose. Considering the cooling demand of a clinker cooler in a full oxyfuel configuration a new concept was proposed that required the cooler to be split into two stages. The first stage is to be operated with recirculated flue gas that is subsequently used for combustion in the kiln, while the second stage is operated with supplementary air that completes the clinker cooling and is then used for raw material drying or fuel preparation. A reliable supply of O 2 for combustion is required for a full oxyfuel configuration. For a medium-sized cement plant with a kiln capacity of 3000 t/day of cement clinker the O 2 demand is estimated to be around to 2/t clinker. That amount of O 2 can only be provided by a non-site O 2 supply system. From all oxygen production methods, cryogenic O 2 production is currently the most mature and reliable technology for the production of large quantities of O 2 (Santos, 2014);(Higginbotham et al., 2011).The O 2 purity produced in the ASU has a direct influence in the composition of the flue gases. For example, considering oxyfuel implementation in power plants an O 2 purity from the ASU of 95 vol% was claimed to be an economic optimum. Due to the high power consumption of the ASU aggregate, the design of more energy efficient methods is still a research necessity. An important research contribution in the applicability of oxyfuel technologies for cement plants is the Dania Pilot Plant Project (Carrasco-Maldonado et al., 2016), which is, to date, the first pilot test regarding the application of the oxyfuel technology for cement processing. In 2009 FLSmidth, Lafarge and Air Liquide started a joint study on the oxyfuel technology for cement plants. Another important project that started in May 2015 is the CEMCAP project (Carrasco-Maldonado et al., 2016). It is a 4-years Horizon 2020 joint European research project involving several research institutions, universities, industrial partners and cement manufacturers. The main purpose of this project is a preparation for large industrial scale implementation of CCS in the European cement industry. Within the CEMCAP project, four CO 2 capture technologies, including oxyfuel combustion will be investigated, analyzed and tested. Pilot and demonstration tests in this project are focusing on three main areas: oxyfuel burners, calciners and the clinker cooler regarding a full oxyfuel configuration. 41

50 This CO 2 outcoming from a cement plant oxyfuel combustion process contains impurities such as water vapor, oxygen, nitrogen and argon derived from the excess oxygen for combustion and from impurities in the oxygen used and any air leakage into the system in addition to SOx and NOx components. Thus, a purification step is required to remove eventual impurities that are likely to poison the catalyzer used for the conversion step of the CO 2 into methanol (PhD thesis within the ECRA CHAIR at UMONS, Nicolas Meunier) Oxyfuel carbon capture and purification Purification of oxyfuel power plants derived CO2 A pilot plant at Vattenfall Schwarze Pumpe power plant has been installed by Air Products (White et al., 2013) to test the CO 2 purification and compression. The process produces NOx-free, SO 2-free CO 2 to meet future specifications of CO 2 for geological disposal and enhanced oil recovery applications. In the flue gas treatment chain the impure CO 2 is cooled down to condense water vapor, removing traces of ash and dissolving soluble gases such as SO 3 and HCl. Then, CO 2 is compressed to about 30 bar. During the process of compression (including reactive absorption during the gas-liquid contact), the CO 2 is purified by removal of SO 2 and NO x compounds as H 2SO 4 and HNO 3 respectively. Compositions from Vattenfall Schwarze Pumpe power plant are listed in Table 13 both for the raw gas and the gas treated at the end of the flue gas treatment chain. Table 13: Raw & purified flue gas compositions from basic oxyfuel purification process applied to power plants % mol in raw flue gas At 35 C and 1.02 bar % mol in CO 2 Product At 35 C and 110 bar CO N O Ar SO NO H 2O The CO 2 compression and purification process elucidated by Air Products has shown a CO 2 product of 96% of purity with a recovery of 89% and a total removal of SOx and NOx components. The Air Products Vattenfall Oxyfuel CO 2 Sour-Compression Unit (White et al., 2013) is a compression step removing SOx and NOx components from the flue gas: one of the most interesting parts of the Air Products pilot plant is the process by which the raw CO 2 is compressed and purified using a concept that Air Products calls the sour compression process, which removes SOx and NOx from the feed CO 2 during compression. The theory of this process is that during compression to 30 bar most of the SOx and NOx in the raw CO 2 will react to form acids. Acid formation could have a negative impact if this happens unexpectedly, or in the wrong place, but since Air Products discovered this concept in 2005 these sour compression reactions have been utilized in a configuration that can be used to remove SOx 42

51 and NOx from the raw CO 2 stream by controlling the formation of acids, potentially saving expensive upstream control options and minimizing potential downstream corrosion Purification of oxyfuel cement plants derived CO2 Regarding to this work, the Air Products Vattenfall Oxyfuel CO 2 Compression and Purification Pilot Plant at Schwarze Pumpe process is simulated in Aspen Plus V 8.6 as described in the previous report. Design parameters described in report 1 have been estimated by means of ranges of operational parameters given by Air Products. Calculations of column diameters and gas/liquid velocities as well as the choice of an adapted column packing and an adapted thermodynamic model for the calculations have required preliminary studies. All these parameters were fixed and maintained constant for the first simulations in order to compare the experimental results given by Air Products pilot plant and the simulation results realized in this work. The CO 2 Purification Unit (CPU) has been simulated in 3 successive blocs as described in Figure 41. Sour-Compression Unit: Figure 41: Global flowsheet of the CO 2 Purification Unit (CPU) After the de-dust step, the gas outcoming from an oxyfuel cement plant combustion is then compressed to 15 bar in an isentropic two-stage compressor and enters at the bottom of the first absorber (DESOX) of the Sour-Compression Unit and flows counter currently to the aqueous scrubbing liquid (98% water plus acids). The absorbers were dimensioned based on approximate ranges taken from Air Products Sour-Compression Unit pilot plant. The first absorber has an internal diameter of 0.15 m and consists of consecutive packed-beds with a total packing height of 12 m and filled with IMPT 25 random packing. This absorber was designed for gas and liquid flow rates of respectively 120 Nm³/h and 0.62 m³/h, and leading to a liquid/gas ratio of and a gas velocity of 0.12 m/s. A splitter is used to recycle a part of the liquid flow to the top of the absorber and a water make-up is also provided. The washed gas then leaves the column by the top and is compressed to 30 bar before entering the second absorber (DENOX) at the bottom and also flowing counter currently to the aqueous scrubbing liquid. Another splitter is used to recycle a part of the liquid flow to the top of the second absorber and a water make-up is also provided. This second absorber column has the same geometry (in terms of diameter, height and random packing) than the first one. The washed gas then leaves the column by the top and goes to the dehydration unit. A detailed flowsheet of the unit is present in Figure 50. Simulations of the Sour-Compression Unit were performed in this work on Aspen Plus V8.6 using rate based calculations and Electrolyte NRTL model selected for thermodynamic properties calculations as various electrolytes are considered. 43

52 Simulation results of this unit will be provided further in this report. Dehydration unit: The dehydration unit is composed by a Temperature Swing Absorption (TSA) dual-bed and water is absorbed at high pressure (30 bar) onto a solid adsorbent which can be silica gel, activated alumina or molecular sieve alumina. Typically, regenerative desiccant dryers supply a dew point of -40 C to -70 C if required. At these temperatures, the vapor pressures of ice are respectively Pa and Pa which leads to a water concentration range of ppm in the gas phase. As a result, water present in the flue gas at the exit of the sour compression unit (typically about 0.2 mol%) is supposed to be completely removed before entering the cryogenic unit. Cryogenic unit: The gas coming from the dehydration unit is cooled and flashed in a first flash at 30 bar. The vapor stream is then cooled and flashed anew in a second flash with a lowest temperature of -55 C to avoid the formation of dry ice (solid CO 2) at this pressure. The vapor stream from the second flash goes then through a turbine which lowers its pressure to 1 atm and electrical energy is recovered from it. Liquid streams from both flashes are finally mixed and compressed to 110 bar for transport or storage. Simulation results of the Dehydration Unit and the Cryogenic Unit are provided in another PhD thesis of the ECRA CHAIR at UMONS by Nicolas Meunier Sour-Compression Unit (SCU): De-NOx and De-SOx processes for CO2 purification The first chemical mechanism chosen was described in the first report including the reactions of the SOx and NOx compounds in the gas/liquid absorbers with oxidation reactions, hydrolyses and equilibria between NOx and SOx species. Concerning these reactions, they could occur either in the gas phase or in the liquid phase, and can be considered either as equilibrium or as kinetic reactions. Results of preliminary sensitivity analysis revealed that for this chemical mechanism, pressure of the system has no influence on the abatement rate of SOx and NOx components and no interactions between NOx and SOx were taken into account. That led us to elucidate a more complete and adapted chemical mechanism for SOx and NOx simultaneous absorption in pressurized systems. 44

53 Bibliographic review of the SCU complete chemical mechanism The importance of the interaction reactions: Air Products Sour-Compression process Defining a more accurate chemical mechanism means realizing an additional bibliographic review to identify in the literature the key reactions and parameters likely to influence and to model more precisely the SOx/NOx absorption performances. At pilot scale, several technologies were used to test directly the absorption performances and to measure experimentally the kinetics of the reactions likely to occur during the process The CO 2 Processing Unit (CPU) was tested by Air products, Linde, Praxair and Air Liquide (Santos, 2015) The pilot plants tests in Air Products proved that the Sour-Compression bloc removes SOx/NOx during the first step of the CPU: in fact NO is oxidized into NO 2 which oxidizes SO 2 to SO 3, mercury will be removed reacting with the nitric acid that is formed. The Air Products project studied experimentally reactions in order to provide for the calculations the necessary kinetic parameters so far missing from the literature. The chemical mechanism elucidated for the Air Products Vattenfall Sour-Compression process (Murciano et al., 2011); (Torrente-Murciano et al., 2011) takes into account the main reactions listed in Table 14. Table 14 presents the general kinetic layout that will be updated further by Air Products (Santos, 2015) and presented in Table 15. Table 14: Air Products reactions of the sour compression process for the purification of oxyfuel-derived CO 2 Number Reactions Phase 1 2 NO + O 2 2 NO 2 Gas 2 NO 2 + SO 2 + H 2O NO+ H 2SO 4 Gas/Liquid 3 2 NO 2 + H 2O HNO 3 + HNO 2 Liquid 4 3 HNO 2 2 NO + HNO 3 + H 2O Liquid NOx formed during combustion will be mostly NO. Subsequent conversion of NO to NO 2 in the gas phase was expected to follow the first reaction. At low temperature, the kinetics of reaction 1 favors NO 2 production rather than NO. However at low pressure the rate of the reaction 1 is slow. The interaction SOx/NOx through the reaction of NO 2 with SO 2 to form sulphuric acid (reaction 2), commonly referred to as the Lead Chamber process for the manufacture of sulphuric acid, was believed to be fast. In presence of liquid water, SO 2 is oxidized by NO 2 regenerating NO with the rate of conversion of SO 2 being dependent on the acid concentration in the liquid. The NO 2 formed would be converted to nitric acid by the nitric acid process, described by reactions 3 and 4. The NO formed in reactions 2 and 4 was expected to be reconverted to NO 2 by reaction 1. These reactions give a pathway for SO 2 to be removed as H 2SO 4 and NO and NO 2 to be removed as HNO 3. In order to achieve high removal rate, sufficient pressure, residence time and contact with water must be provided. 45

54 An updated chemical mechanism was presented as follows in Table 15. The main differences compared to Table 14 are that reaction 3, is considered to be the key reaction forming HNO 2 + HNO 3 that in turn will react with the other SOx/NOx components. The interaction SOx/NOx through the reaction of NO 2 with SO 2 (reaction 2 of Table 14) was replaced by the reactions of HNO 2 with SO 2 or H 2SO 3 (reactions 7 to 10 of Table 15). Table 15:Updated SOx-NOx reaction network for Air Products Sour-Compression Unit (Santos, 2015) Number Reactions Phase 1 2 NO + O 2 2 NO 2 Gas 2 2 NO 2 N 2O 4 Gas 3 N 2O 4 + H 2O HNO 3 + HNO 2 Liquid 4 2 HNO 2 NO + NO 2 + H 2O Liquid 5 4 HNO 2 2 NO + N 2O H 2O Liquid 6 SO 2 + H 2O H 2SO 3 Liquid 7 2 HNO SO 2 + H 2O 2 H 2SO 4 + N 2O Liquid 8 2 HNO H 2SO 3 2 H 2SO 4 + N 2O + H 2O Liquid 9 2 HNO SO 2 H 2SO NO Liquid 10 2 HNO 2 + H 2SO 3 H 2SO NO + H 2O Liquid 11 2 NO 2 + H 2O HNO 3 + HNO 2 Liquid It should be noted that for NOx removal up to the exit of the 15 bar column around 80% of the NO+NO 2 should be removed from the gas phase at this point, with the rest being eliminated in the 30 bar column (White et al., 2013). It was found, however, that in certain situations, particularly high SO 2/NOx ratios, some of the NOx would convert into N 2O as we can see in the reaction 7 of Table 15. In the worst case, all of the NOx components are converted into N 2O. In addition to the updated chemical mechanism, White and colleagues estimated that further work is required to understand this behaviour. The important aspects arising from the bibliographic reviews proved that interaction reactions have to be taken into account in the mechanism as they are likely to happen during the SOx/NOx simultaneous absorption under pressurized conditions. Nevertheless, the interaction reactions kinetic parameters of the Air Products Sour-Compression process were measured experimentally but not published The ph influence: Normann and co-workers studies Other types of studies include simulation studies with the same purpose of identifying the more adequate chemical mechanism for SOx/NOx absorption. Normann and colleagues (Normann et al., 2013) evaluated the NOx and SOx chemistry, relevant to process conditions typical of pressurized flue gas systems in an oxyfuel power plant by comparing a state-of-the-art reaction mechanism to the results of experimental investigations performed at 46

55 Imperial College London (Murciano et al., 2011); (Torrente-Murciano et al., 2011) and UHP (Petrissans & Zoulalian, 2001); (Petrissan et al., 2005). The reaction mechanism is based on data for simultaneous SOx and NOx absorption under atmospheric conditions and extended to include high-pressure conditions, a wider temperature range, and shorter residence times. The model of (Normann et al., 2013) confirms that the absorption of NOx and SOx and the subsequent formation of acids are important in pressurized flue gases. According to their studies, two processes are key parameters to the absorption: (1) the oxidation of NO into NO 2, in a reaction that benefits from high pressures and low temperature, governs the absorption of NOx, since NO 2 has greater solubility in water compared to NO; (2) The complex interactions between HNO 2 and H 2SO 3 in the liquid enhance the rates of oxidation to HNO 3 and H 2SO 4, respectively. These reactions enable increased absorption of SOx, which is otherwise hindered by the equilibrium between SO 2 and H 2SO 3. In these reactions, the important species of NO and N 2O may also form. Thus, they revealed the same conclusions as (White et al., 2013) concerning the presence and effect of interactions during the simultaneous absorption of SOx and NOx components. Another point that stood out from this investigations is the influence of the ph of the solution on the importance of these reactions to happen. The model showed comparative results for the NO, NO 2 and SO 2 components with the experimental results (Murciano et al., 2011) but among these results other interaction reactions are likely to happen with the arising formation of complex species in the liquid phase. The results of the modelling and experiments show good correlation in the way that: they both show significant N 2O production at low ph values. N 2O is the dominating nitrogen compound at ph<2. Since HSO 3 and HNO 2 are the main hydrolysis products (together with HNO 3) from NOx and SO 2, the important intermediate NSS (NOSO 3 ) is formed. NSS may react to form either N 2O and H 2SO 4 directly or complex nitrogen-sulphur compounds, e.g., HADS (HNO(SO3) 2 2- ) and HAMS (HNOHSO 3 ), which over time may be converted into H 2SO 4. Table 16 presents the complete denominations of these new complexes deriving from interactions reactions. Table 16: New complexes denominations Complete denomination Raw formula Alias Nitrososulfonic acid ONSO 3 NSS Hydroxylamine N,N-disulfonic acid HNO(SO 3) 2-2 HADS Hydroxylamine N-sulfonic acid HONHSO 3 HAMS Littlejohn (Littlejohn & Chang, 1984) studied these reactions at several ph conditions using Raman spectroscopy for nitrite and bisulfite ions in aqueous solutions in order to identify these complexes. These complex species deriving from the SOx/NOx interactions, named HADS and HAMS (see Table 16 for complete denominations) predominate at higher ph values (ph>2). (Normann et al., 2013) also elucidated that the rapid conversion of HADS to HAMS requires a high concentration of H 3O +, which means that for a ph>3, the formation of HAMS will be low and HADS will predominate. The conversion of nitrogen oxides into N 2O, and HADS+HAMS tested experimentally by Pétrissans and colleagues (Petrissans & Zoulalian, 2001) has been confirmed by the modelled process of (Normann et al., 2013). Indeed, the graphic on Figure 42 shows the ph influence on the conversion of nitrogen compounds: 47

56 Figure 42: Conversion of nitrogen compounds into N 2O and HADS plus HAMS (Petrissans et al., 2001) & (Normann et al., 2013). We can see from this figure that effectively the N 2O predominates at ph levels<2 however N-S complexes (HADS+HAMS) predominate at higher ph values Selection of the reaction pathways related to the operating ph: Ajdari and co-workers studies Further works (Ajdari et al., 2015) of the same team as (Normann et al., 2013) included studies of a rate-based model developed to elucidate the chemistry behind the simultaneous absorption of NOx and SOx under pressurized conditions (pressures up to 30 bar) that are applicable to the flue gases obtained from oxyfuel CO 2 capture systems. The important contribution of these studies resides in the kinetic data used in the model and the corresponding chemical reactions modeled and including different complex species derived from SOx/NOx interactions that are detailed in this case. Indeed, the liquid phase interactions between nitrogen and sulfur are known to be complex because liquid phase chemistry parameters under relevant conditions are lacking. Thus, a reaction model was constructed by Adjari and co-workers to identify the important pathways for the simultaneous absorption of NOx and SOx. The simulated model considers a pressure up to 30 bar, a temperature of 25 C and a L/G ratio of 0.02m³/m³, the chosen value for the L/G ratio is of the same order as typical L/G ratios for gas/liquid contactors for gas absorption Table 17 lists the operating parameters taken for the simulation tests. Thanks to buffer solutions the ph of the mixtures is kept constant. Due to the absorption of acidic gases in water, the ph of these solutions will consequently be low. Table 17: Operating parameters taken for the simulations tests Pressure 30 bar SO 2 inlet 1000 ppmv Total NOx inlet 400 ppmv (10% NO 2, 90% NO) O 2 inlet 3% H 2O(g) inlet 2000 ppmv (v) Temperature 25 C L/G ratio 0.02 m³/m³ 48

57 The most relevant conclusion of this study is that the ph level has a strong influence on the reaction pathway that is followed and the types of products that are formed in the liquid phase. Consequently, different chemical mechanisms pathways depending on the ph level of the solution are considered. The possible interactions between the nitrogen and sulfur species are summarized in two categories: Interactions between nitrous acid HNO 2 (and NO 2- ) and hydrogen sulfite HSO 3- : these interactions are influent under acidic conditions for ph<5-2- Interactions between dissolved NO 2 and SOx (HSO 3 and SO 3 ) for ph>5 (which will not be considered in our case). Indeed, according to the operating ph<5 of the system (see the justification further in this report), the pathway more likely to be considered in our case is the one developing the interactions between HNO 2 (and NO 2- ) and HSO 3-. The flue gas is consequently absorbed in the liquid phase and the major fraction of the absorbed NOx in these processes has been found to be in the form of nitrogensulfur (N-S) complexes which are the compounds produced in the reaction between nitrite and bisulfite ions. In turn, the interactions between nitrite and hydrogen sulfite HNO 2 and H 2SO 3 respectively are described by the so-called Raschig mechanism. In the first step of this complex scheme of reactions, the important intermediate NSS (Nitrososulfonic acid ONSO 3 ) is formed (as proved by (Normann et al., 2013)). The reaction of this component may lead to two other different pathways schematized in Figure 43). Pathway 1: NSS may react to form complex nitrogen-sulphur compounds, e.g., HADS (HNO(SO 3) 2 2 ). Solutions of HADS hydrolize at a measurable rate at room temperature to give HAMS and bisulfate ion or react with hydrogen sulfite to form amine trisulfonic acid (ATS) or also amine disulfonic acid (ADS). Pathway 2: NSS hydrolyzes in the presence of H 3O + to form hyponitrous acid (HNO) which dissociates to form N 2O and H 2O (Oblath et al., 1982). Figure 43: Diagram of the two pathways for the Interactions between nitrous acid and hydrogen sulfite 49

58 Consequently, as schematized in Figure 44, N 2O is the dominating nitrogen compound at ph < 2, while HADS and HAMS predominate at higher ph values (Normann et al., 2013). The formation of HADS becomes important as the ph increases to 2. The formation of N 2O is still significant at ph 2. HAMS is also formed as a result of the hydrolysis of HADS. The rate of this reaction increases with decreasing ph, although it is still rather low at ph 2, resulting in around 10% conversion of HADS to HAMS (Ajdari et al., 2015). At ph 4, the formation of N 2O becomes insignificant and, instead, pathway 1 predominates. These conclusions are confirmed by the studies of (Pétrissans et al., 2005) which claim that in the intermediary conditions for 1<pH<4, a competition between three reactions is observed as schematized in Figure 44. (1) Production of N 2O, (2) Production of HADS and acidic hydrolysis of HADS (3) HADS in turn gives an unidentified product Y likely to represent the complex HAMS (Naiditch & Yost, 1941). Figure 44: Competition between production of N 2O and production of N-S complexes Other studies by Chang and co-workers (Chang and al., 2013) proved that the nitrous acid and sulfite react to from nitrososulfonic acid which can react using several pathways indicated by the chemical reaction scheme represented in Figure Figure 45: Schematic representation of the interaction between sulfite and nitrite ions in aqueous solutions (Chang and al., 2013) 50

59 Pathway 1: The nitrous acid reacts with bisulfite ion to form hydroxylamine disulfonate and amine trisulfonate which can provide themselves sulfuric acid and reduced nitrogen species by hydrolysis. Amine trisulfonate can also react with bisulfite and nitrite. This pathway takes place in a neutral or mildly acidic solutions. Pathway 2: NSS can be hydrolyzed to form sulfuric acid and hyponitrous acid which decomposes to produce nitrous oxide. Pathway 3: NSS can also react with nitrous acid to produce sulfuric acid and nitric oxide. Pathways 2 and 3 become more important when the ph of the solution decreases. The reaction scheme as a result of interactions between sulfite and nitrite ions made by Chang and coworkers confirms the studies of Ajdari since the former proved that these pathways of reactions will depend especially on the ph of the system (but also on the temperature and the concentration of nitrite and sulfite species) Main conclusions of the last studies The relevant conclusions derived from the cited bibliographic studies are that the ph level has a strong influence on the reaction pathway that is followed and the types of products that are formed in the liquid phase. The pressure level and the presence of NOx affect the removal of SO 2 from the flue gas. According to our ph conditions represented further in this report (we are in an intermediate situation of 1<pH<4), interactions between HNO 2 (and NO 2- ) and HSO 3 - will be considered and a competition between the two pathways may occur like indicated in Figure 43: the intermediate product of the reaction of bisulfite and nitrite ions NSS may react to form either N 2O and H 2SO 4 at low ph levels or complex nitrogen-sulphur compounds (N-S complexes) at more elevated ph values. According to studies of (Naiditch & Yost, 1941), the hydroxylamine monosulfonate ion HADS is relatively stable and does not hydrolyze in dilute acid solution except at higher temperatures. As the SOx/NOx absorption in water we study here takes place at 30 C, the hydrolysis of HAMS could be neglected. According to (Ajdari et al., 2015) the formation of HADS becomes important as the ph increases to 2. HAMS is also formed as a result of the hydrolysis of HADS. The rate of this reaction increases with decreasing ph. HADS is the dominant specie in the liquid phase at ph 4. The ph-dependent hydrolysis of HADS to HAMS is not active as the ph is increased (ph>4). When the ph is increased further (to 4), only the production of N-S complexes (exclusively HADS) becomes important. Consequently, the formation of HAMS is low at 1<pH<4 and the other N-S complexes ATS and ADS seem to represent a minor proportion compared to HADS and HAMS, so their formation could be neglected in our chemical mechanism. These reactions may have different relative importance according to the ph ranges so we selected a simplified reaction scheme represented in Figure 46 taking into account all these new different configurations based on a solution ph between 1 and 4. 51

60 ph 2 H 2O H 2O N 2O HNO +HSO 4 - Figure 46: Reaction pathways for ph levels between 1 and 4 (adapted from Ajdari et al., 2015). A complete chemical mechanism representing the SOx and NOx simultaneous absorption into water for pressurized systems in both gas and liquid phases, used to simulate reactions occurring in the Sour-Compression Unit is detailed in Figure 47. It takes into account the last bibliographic studies that highlighted the ph influence on the importance of the interaction reactions pathways that may occur and the N-S deriving complexes New SCU chemical mechanism deduced from the bibliographic review Conclusions of part led us to consider different types of mechanisms for the flue gas De-SOx and De-NOx process. The importance of considering properly the interaction reactions is one of the main inputs of this study. Figure 47 represents in black the chemical mechanism without interactions (A) (18 reactions) and in black + green the chemical mechanism with the selected N-S interactions (C) (27 reactions). The ph influence explained in the bibliographic review has been taken into account for the choice of the considered reactions. 52

61 Figure 47: Complete SCU Chemical Mechanism for SOx and NOx absorption without (A) or with (C) SOx/NOx interactions (reactions selected for 1<pH<4) The complexity of the complete chemical mechanism led to a pronounced difficulty for the calculation convergence in Aspen Plus (see also strategy of implementation and order of reactions in Annex B for more details). Thus, in order to elaborate a sensitivity analysis, a simplified mechanism (B) (20 reactions) schematized in Figure 48 has been selected considering only the two main interaction reactions (reactions 19 and 20) and expecting it to have similar N-S removal performances as the complete one (C) (see results further in this report). Figure 48: SCU Chemical Mechanism for SOx and NOx absorption with simplified SOx/NOx interactions (B) 53

62 Table 18 provides all the kinetic parameters of the reactions schematized in Figure 47 and Figure 48. Table 18: Kinetic parameters taken for the SCU chemical mechanism. Reactions Phase Keq, kcin and r expressions (Ea in J/mol) Reference H2O 1) 2 H2O H 3O + + OH - Liquid Keq (298K)= kmol 2 /m 6 Aspen Plus estimations* CO2 NOx SOx SOx/NOx Interactions N-S complexes dissociation 2) CO H 2O HCO H 3O + Liquid Keq (298K)= kmol/m³ Aspen Plus estimations* 3)HCO H 2O H 3O + + CO 3 2- Liquid Keq (298K)= kmol/m³ Aspen Plus estimations* 4) 2 NO + O 2 2 NO 2 Gas r=970 e (5003 RT ) C NO2 C 1 O2 Kcin (298 K)= 7300 m 6 /kmol².s 5)2 NO 2 N 2O 4 Gas Lnkeq= / T Keq (298 k)= Pa -1 6) NO + NO 2 + H 2O 2 HNO 2 Gas 7) NO 2+NO N 2O 3 Gas [HSO 3- ] 22) ONSO - 3 +H 2O HNO + HSO - 4 Liquid r= e ( [H 3O + ] Lnkeq= /T Keq (298 k)= Pa -1 Lnkeq= /T Keq (298 k)= Pa -1 8)N 2O 3+H 2O 2 HNO 2 Liquid r= e ( RT ) C N2O31 C H2O 1 Kcin (298 K)= m³/kmol. s 9)N 2O 4 +H 2O HNO 3 + HNO 2 Liquid r= e ( RT ) C N2O41 C H2O 1 Kcin (298 K)= 812 m³/kmol. s 10)2 NO 2+ H 2O HNO 3 + HNO 2 Liquid r=5288 e (4088 RT ) C NO22 C 1 H2O Kcin (298 K)= m 6 /kmol².s 11)3 HNO 2 H 2O + 2 NO+HNO 3 Liquid r= e ( RT ) C HNO24 C NO -2 Kcin (298 K)= m³/kmol. s (England & Corcoran, 1975) (Holma & Sohlo, 1979) (Holma & Sohlo, 1979) (Hoftyzer & Kwanten,1972) (Hoftyzer & Kwanten,1972) (England & Corcoran, 1975) (England & Corcoran, 1974) (England & Corcoran, 1974) (Rayson et al., 2012) 12) HNO 2 +H 2O NO 2 - +H 3O + Liquid Keq (298K)= kmol/m³ Aspen Plus estimations* 13) HNO 3 + H 2O NO H 3O + Liquid Value not communicated by Aspen Plus Aspen Plus estimations* 14) SO H 2O H 3O + + HSO 3 - Liquid Keq (298K)= kmol/m³ Aspen Plus estimations* 15) HSO H 2O H 3O + + SO 3 2- Liquid Keq (298K)= kmol/m³ Aspen Plus estimations* 16) SO H 2O H 3O + + HSO 4 - Liquid Value not communicated by Aspen Plus Aspen Plus estimations* 17) HSO H 2O H 3O + + SO 4 2- Liquid Keq (298K)= kmol/m³ Aspen Plus estimations* 18) H 2SO 4 + H 2O H 3O + + HSO 4 - Liquid Aspen Plus estimations* 19) NO 2+ SO 2 NO + SO 3 Gas r= e ( RT ) C NO21 C SO2 1 Kcin (298 K)= m³/kmol. s 20) HNO 2 + HSO 3 - ONSO 3 - +H 2O Liquid 21) ONSO HSO 3 - HON(SO 3) 2 2 Liquid 23) HNO + HNO N 2O +H 2O Liquid r= e ( RT ) C NO2 - C H + C HSO3 - = e ( RT ) C HNO2 1 C HSO3-1 Kcin (298 K)=2.431 m³/kmol. s r= e ( RT ) C NSS1 C HSO3-1 Kcin (298 K)= RT ) C NSS1 C HSO3-1 Kcin (298 K)= m³/kmol. s r= C 2 HNO Kcin (298 K)= m³/kmol. s 24) HON(SO 3) H 2O HONHSO 3 +HSO - 4 Liquid r= e ( RT ) C 1 HADS C 1 H3O+ Kcin (298 K)= m³/kmol. s 26) HON(SO 3H) 2+2 H 2O HON(SO 3) H 3O + (Armitage & Cullis, 1971) (Oblath et al., 1982) (Oblath et al., 1982) (Oblath et al., 1982) (Bonner & Hughes, 1988) ( Naiditch & Yost, 1941) Liquid Value not communicated by Aspen Plus Aspen Plus estimations* equilibriums 25) ONSO3H + H2O ONSO3- + H3O+ Liquid Value not communicated by Aspen Plus Aspen Plus estimations* 27) HONHSO 3H + H 2O HONHSO 3 +H 3O + Liquid Value not communicated by Aspen Plus Aspen Plus estimations* *Aspen Physical Property System Estimation by means of the reference state Gibbs free energy of the system. 54

63 Complexity of the implementation into Aspen Plus Kinetic and equilibrium data The evaluation of the SCU performances is largely dependent on the selection of the adequate chemical mechanism and of the characteristic parameters, requiring literature reviews and critical comparison of various sources. The kinetic data regrouped in Table 18 are the results of a deep research and comparison of different sources due to a lack of data in the literature and an uncertainty between different sources concerning the available kinetic parameters. In fact, literature sources, for several reactions, provide different kinetic data so an additional study had to be done to determine the most adequate parameter to use in the simulation. Likewise, several reactions were tested with different kinetic parameters in Aspen Plus to evaluate the impact of this change in the absorption performances. For rate-controlled reactions, Aspen Plus provides a built-in power law expression for calculating the rate of reaction depending on the concentration basis selected in the [Ci] basis list. The general power law expression is given in Table 19. In Aspen Plus, if computing Keq from built-in expression, coefficients A, B, C and D have to be implemented, the general expression used by Aspen Plus is also given in Table 19. Expression Table 19: Kinetic and equilibrium law expressions in Aspen Plus Kinetic Equilibrium [Ci] basis molarity [Ci] basis partial pressure r = ke ( E RT ) (C αi i ) r = ke ( E RT ) (P αi i ) Ln Keq = A + B + CLn(T) + D(T) T Units of the preexponential factor, k Units of the kinetic, r kmol s. m³ ( kmol m³ )^ αi kmol s. m³ kmol s. m³ ( N m² )^ αi kmol s. m³ Where: r is the rate of reaction, k is the pre-exponential factor, T is the absolute temperature, E is the activation energy, R is the gas law constant, C i is the concentration of the i th component, P i is the partial pressure of the i th component, α i is the exponent of the i th component; Keq is the equilibrium constant and [A, B, C and D] are the user supplied coefficients. For equilibrium implementation, data from the literature had to be converted in order to obtain the constants to implement in Aspen Plus. Moreover, kinetic data from literature had to be converted in order to suit unit specifications for the kinetic of the reaction. As indicated in Table 19, kinetics of reactions taking place in the gas phase and expressed as a pressure variation are also expressed in Aspen Plus as a temporal concentration variation. Consequently, conversions of data from the literature had to be done for the adequate implementation in Aspen Plus. 55

64 Complex species deriving from the SOx/NOx interaction reactions Taking into account the formation of various complexes deriving from the SOx/NOx interaction reactions, namely NSS, HADS, HAMS, means the implementation of N-S complexes into the simulated process. These species being not available in Aspen Plus databases, their properties had to be introduced manually. The required properties are the molecular structure, the molecular weight and for ions the implementation of the charge, the aqueous enthalpy of formation at infinite dilution and 25 C, the aqueous free energy of formation at infinite dilution and 25 C, the absolute entropy at 25 C and the coefficients for the Extended Antoine vapor pressure equation is also necessary (see Table 22). Identification of the molecular structures The N-S complexes are not present in Aspen Plus databases, it has to be highlighted that in the literature ((Normann et al., 2013) (Ajdari et al., 2015) (Chang and al., 2013)), only the name and the raw formula were specified so there is a lack of precision in the full name of the complexes and their molecular structures in the literature. Therefore, several chemical structures have been tested and Table 20 presents the results of studies and deductions of a chemical structure evolution (with Professor Hantson from the University of Mons) that had to be done in order to clarify the chemical structures of these ions and their respective acid forms (6 complexes). Ions (e.g. NSS) and corresponding acids forms (e.g. NSSH) had to be introduced manually in Aspen Plus. Besides, dissociation equilibriums listed in Table 18 were also implemented in the chemical mechanism. Table 20: Molecular structures of the N-S complexes implemented in Aspen N-S complexes ionic form Nitrososulfonate ONSO - 3 alias NSS N-S complexes molecular form Nitrososulfonic acid ONSO 3H alias NSSH Hydroxylamine N,N-disulfonate HON(SO 3) 2 2 alias HADS Hydroxylamine N,N-disulfonic acid HON(SO 3H) 2 alias HADSH Hydroxylamine N-sulfonate HONHSO 3 alias HAMS Hydroxylamine N-sulfonic acid HONHSO 3H alias HAMSH 56

65 N-S complexes properties As explained before, in order to implement these components in Aspen Plus, as they are not present in the databases, properties had to be implemented properly. The complex thermodynamic properties are not available in Aspen Plus databanks so several researches have been done in order to find a group contribution method that could calculate the contribution of each element for the ions. Methods like the ones elucidated by (Jankowski, 2008) and (J. Li & Li, 2000) were investigated in this purpose and have been proved to be applicable to other simple components like H 2SO 4 by the addition of the single group contributions provided by the NBS tables (Wagman et al., 1982) as detailed in Table 21. The simple addition of properties of 2H + and SO 4 2- present in NBS tables have been done for this group contribution method as specified in (J. Li & Li, 2000). Table 21: Calculation of H 2SO 4 properties by means of a group contribution method Calculations for H 2SO 4: 2H + + SO 4 2- Group Contribution method DHAQFM (enthalpy of formation at 25 C) kcal/mol DGAQFM (free energy of formation at 25 C) kcal/mol S025C (standard entropy of formation at 25 C) cal/mol.k Properties present in Aspen Plus databanks Nevertheless, NBS tables (Wagman et al., 1982) could not provide the information to husk correctly the complex ions and apply this group contribution method as the groups listed in the NBS tables are not present in the required aqueous phase. Further investigations may be required to define a more accurate method to represent the activity of this ionic components. In Aspen Plus, for the implementation of the acids, structure and molecular weight are the two required parameters for Aspen properties estimation system to calculate the other necessary parameters. Nevertheless, for ions, the minimum properties to provide to Aspen Plus are (in addition to the molecular structure), the molecular weight, the charge, the aqueous enthalpy of formation at infinite dilution and 25 C, the aqueous free energy of formation at infinite dilution and 25 C, the absolute entropy at 25 C and a coefficient for the Extended Antoine vapor pressure equation of - 1E+20 (to indicate to Aspen that ions must remain only in the liquid phase). Consequently, due to the lack of information in the literature, assumptions had to be used in order to provide the necessary and minimum information for Aspen Plus to achieve the simulations. The most influent groups in the ionic activity of the complexes are the ones carrying the ionic charge, thus SO 3 - and SO 32. As a result, properties of other ions had to be taken as a first estimation: HSO 3 - ionic properties were taken for NSS and HAMS and SO 3 2- properties were taken for HADS. The required properties of the new components that have been introduced in Aspen Plus in order to do the SCU simulations are listed in Table

66 Table 22: Required N-S properties introduced in Aspen Plus to do the simulations Components NSS NSSH HADS HADSH HAMS HAMSH Molecular weight (g/mol) Group contribution method used in Aspen Plus - Joback* - Joback* - Joback* Ion properties NSS & HAMS simulated as HSO3 - HADS simulated as SO3 2- Charge -1-2 DHAQFM Aqueous enthalpy of formation at kcal/mol infinite dilution and 25 C DGAQFM Aqueous free energy of formation kcal/mol at infinite dilution and 25 C S025C cal/mol.k Absolute entropy at 25 C PLXANT Coefficients for the Extended A -1E+20 A -1E+20 N/m² Antoine vapor pressure equation B 0 B 0 C 0 C 0 D 0 D 0 E 0 E 0 F 0 F 0 G 0 G 0 Tlower 0 Tupper 2000 Tlower 0 Tupper 2000 *Joback is a group contribution method included in Aspen Plus physical property estimation system able to predict the thermodynamic properties (normal boiling point, critical temperature, critical pressure, critical volume, ideal gas heat of formation at K, ideal Gibbs free energy of formation at K) for pure components from their molecular structure. In Aspen Plus, Property Constant Estimation System (PCES) is used for properties calculation. Property Estimation in the Aspen Physical Property System can estimate many of the property parameters required by physical property models, including: Pure component thermodynamic and transport property model parameters Binary parameters for the Wilson, NRTL, and UNIQUAC activity coefficient models In the case of the estimation of ions properties, the AQU-DATA method (Table 23) can estimate standard enthalpy of formation of aqueous species (DHAQHG), standard Gibbs free energy of formation of aqueous species (DGAQHG), and absolute entropy of aqueous species (S25HG) for the Helgeson electrolyte model. This method uses directly experimental data in the databank to estimate the parameters for the Helgeson electrolyte model. Table 23: AQU-DATA method used for ions properties estimation in Aspen Plus This method uses standard enthalpy of formation at infinite dilution (DHAQFM) standard Gibbs free energy of formation at infinite dilution (DGAQFM) absolute entropy (S025C) To estimate standard enthalpy of formation of aqueous species (DHAQHG) standard Gibbs free energy of formation of aqueous species (DGAQHG) absolute entropy of aqueous species (S25HG) 58

67 Optimization of the SCU simulated process by (Iloeje et al., 2015) Another part of this work is the optimization of the SCU simulated process with the selected reaction mechanism. A research elaborated in the Massachusetts Institute of Technology (MIT) by Iloeje and co-workers (Iloeje et al., 2015), based on works of (White et al., 2013) developed a model and made a parametric analysis using a single column absorber of nitrogen and sulfur oxides removal from oxycombustion flue gas applied to power plant cases. First of all, a simulation of two-column process that achieves the complete removal of SOx and NOx from the CO 2 stream was elaborated and then a modification of this two-column process to a single column one was proposed. (Iloeje et al., 2015) demonstrate by means of pressure sensitivity studies that this new design can meet the same separation targets as the two-column process in fewer column stages and half the feed water requirement by exploiting the pressure dependence of the rate determining NO oxidation reaction. Having identified pressure, vapor holdup and water flow rate as key design parameters for the absorber column, they carried out a sensitivity analysis to determine the impact of these parameters on the extent of removal of the NOx and SOx pollutant species, as well as the state and composition of the exit streams. They also analyzed the impact of bottoms liquid recycle on the overall performance of the system. These studies are used to determine the optimal specifications for the column design and operation, and to identify design modifications that improve overall performance, energy consumption and water use. For transport limitations, impurities like SOx and NOx, need to be removed to ensure that the transported CO 2-rich stream stays within specified purity limits for pipeline specifications indicated in Table 24: Table 24: Purity targets for pipeline specifications Components Gullfaks Target specifications for Iloeje studies NOx <50 ppm <15 ppm SOx <10 ppm < 1 ppm The flowsheet of the two-column system for the MIT simulation is presented in Figure 49 Figure 49: Flowsheet of the two-column process 59

68 Without any further justification, the overall reaction scheme in this study is very limited. The reactions implemented in Aspen Plus by (Iloeje et al., 2015) are presented in Table 25. Table 25: Reactions chosen for the implemented model (Iloeje et al., 2015) Reactions Comments (1) NO + ½ O 2 NO 2 Rate limited 3 rd order gas phase reaction (2) NO 2 + SO 2 NO + SO 3 Fast equilibrium reaction (3) SO 3 + H 2O H 2SO 4 Fast equilibrium reaction (4) 2 NO 2 N 2O 4 Fast equilibrium dimerization reaction (5) 3 N 2O H 2O 4 HNO 3 + NO Mass transfer limited reaction The absorber columns are modeled using Radfrac blocks in Aspen Plus. Reactions 2 and 3 are handled directly in Aspen Plus. The rest is implemented using a Radfrac User-Kinetic model which links to an external subroutine that calculates the required reaction rates as well as the rate of generation for each species per stage using Miller s model. It has to be noted that, contrary to our selected chemical mechanism, reaction 2 was taken as an equilibrium reaction for the MIT model because of the lack of kinetic data in the literature. Our preliminary simulations proved that taking this reaction as a kinetic reaction or as an equilibrium reaction will change considerably the SOx removal rate in the first absorber for the MIT design specifications. The thermodynamic ELECNRTL property method was specified in Aspen Plus to describe liquid phase solution equilibrium in the Radfrac columns. This method has been determined to be accurate for the dilute acid conditions in the absorber columns. Design specifications The most relevant aspect here is that the majority (not all) of the design specifications for the absorber columns used in the model for the CO 2 De-NOx and De-SOx process, so far missing in the literature, are available in this study. The column sizing parameters shown in Table 26 are based on results designed to achieve the <15 ppm purity targets by providing sufficient residence time for the rate limiting reactions. Table 26 : Design specifications taken in Iloeje and co-workers model Design specifications Two-column design (Figure 49) Single column De-SOx column De-NOx column design Calculation type Rate-Based Rate-Based Rate-Based Top stage pressure Number of stages Flue gas feed stage Bottom (5) Bottom (8) Bottom (9) Water feed stage Top (1) Top (1) Top (1) Recirculation No No No Vapor holdup per stage (m³) Water flow rate (kg/h) Flue gas flow rate (kg/h) Column diameter (m) Column packing IMTP 25 mm IMTP 25 mm IMTP 25 mm Packed height per stage (m)

69 It has to be noted that the column packing type was not provided. Information about the packing height or the packed height per stage were also missing so the deduced packed height per stage (1.4 m) has been calculated by considering that in a column volume the packing and liquid holdup take 10% and the other 90% of the volume is occupied by the gas. It is based on the vapor holdup per stage which is reported to be 20 m³. Flue gas compositions and stream results for the simulated two-column absorber and single column absorber model are detailed in Table 27. Table 27: Flue gas compositions and stream results for the simulated model by ((Iloeje et al., 2015) Components Flue gas in Flue gas out Two-column absorber Single column 15 bar column 30 bar column absorber Total flow (kg/h) Temperature (K) Mole fractions NO (ppm) NO 2 (ppm) N 2O 4 (ppm) E E E-03 SO 2 (ppm) H 2O (ppm) N 2 (%) O 2 (%) CO 2 (%) CO (%) Ar (%) The results showed that the first column removes totally the SO 2 whereas the NOx components are rather removed in the second column. The same global performances have been reached for the single column process at 30 bar. Single column absorber advantages Exploiting the pressure dependence of the rate determining reactions in the NOx/SOx removal process by working with a single column process allows decreasing the column stages, decreasing the water demand and results in a faster overall conversion process at 30 bar. This means a decrease of the material costs and energy requirements (CAPEX, OPEX). Parametric study The parametric study in this case is realized for the single column process and includes: Pressure sensitivity analysis For these tests, the operating pressure of the system is varied from 2 bars to 40 bars while other design parameters remained at the single column design base case values. Variation of required residence time with pressure was conducted keeping the diameter of the column constant while the total volume was varied until the target exit gas composition specifications are reached. 61

70 The total residence time required in the SOx/NOx removal column is deduced from the ratio of the vapor holdup volume to the volumetric flow rate and provides a handle to the required column size that meets design purity specifications for transport (NOx <15 ppm). (see Table 24) It has been shown from these tests that pressure has the most significant impact on the absorber performance and the total residence time required for the purification process is highly pressure dependent, requiring 1780 s at 10 bars compared to 72 s at 40 bars. Holdup sensitivity analysis In this case, the operating pressure of the absorber is kept constant at 30 bar while the vapor holdup volume per stage is varied from 2 to 100 m 3. Since all the SO 2 is completely removed at 30 bar for the range of holdup values investigated by, the impact of holdup volume on SO 2 removal is not studied in this case. Increasing holdup per stage from 2 m 3 up to 20 m 3 results in significant increase in the level of NOx removal since increasing the design holdup volume per stage increases the effective residence time of the flue gas in the column. Water flow rate sensitivity analysis For this simulations, the water flow rate into the top of the column was varied from 1 kg/s to about 12 kg/s while other design parameters were kept constant. The increase of water flow rates allows reducing the acid resistance requirement of the absorber s material. Nevertheless, simulations results show that even with 12 kg/s flow rate, the liquid ph still remains below 1 and acid resistance will remain an important consideration in material selection. Increasing water flow has only a minor effect on the liquid phase ph which is the most important factor that determines NOx removal. Recycle rate sensitivity analysis These tests were conducted by keeping the total top stage liquid supply constant and gradually replacing the fresh water supply with recycled liquid from the bottoms stream. The main conclusions standing out from these tests are that recycling bottoms liquid can reduce the feed water requirement by up to 40% without significantly affecting the exit gas purity. However, beyond 40%, increasing liquid recycle significantly deteriorates column performance: increasing recycle reduces ph and increases the ionic strength of the liquid phase and thus inhibits NOx solubility, so in order to keep high recycle ratios without changing column dimensions, a neutralizing agent will need to be added to the system to raise the ph. Use of (Iloeje et al., 2015) study: This study gave us very interesting results both for two columns and one column process and we will use a quite similar methodology for our parametric study. Moreover, we will investigate the possibility to carry out the same simulations (same flue gas compositions and design specifications) but using our chemical mechanism. Nevertheless, a real comparison will be difficult to achieve since not all the design specifications are provided (e.g. the column packing type). 62

71 Simulation results of the SCU absorption performances Sour-Compression Unit global flowsheet simulated in Aspen Plus V8.6. Figure 50 represents the simulated flowsheet of the Sour-Compression Unit simulated in Aspen Plus. Figure 50: Aspen Plus flowsheet of the simulated SCU The compositions of the outcoming flue gas from a cement industry working with an oxyfuel combustion listed in Table 28, are given by ECRA calculations: Table 28: Compositions of an outcoming gas from an oxyfuel cement industry Components Mole fractions O 2 (%) 3.27 N 2 (%) Ar (%) 1.34 H2O (%) 1.00 CO (%) 0.04 CO 2 (%) NO (ppm) 861 NO 2 (ppm) 96 SO 2 (ppm) 156 Total gas mole flow (mol/h) 4764 (120 Nm³/h) Total liquid flow rate 10 l/min L/G The design specifications used in the model for the design of the De-SOx and De-NOx gas/liquid contactors are specified in Table

72 Table 29: Design specifications of the SCU contactors (based on ranges given by Air Products) Design specifications Column De-SOx Column De-NOx Calculation type Rate-Based* Rate-Based Top stage pressure (bar) Number of stages Flue gas feed stage Bottom (12) Bottom (12) Water feed stage Top (1) Top (1) Liquid holdup per stage Calculated by Aspen Plus with (Stichlmair et al., 1989) Packing Holdup Correlation Calculated by Aspen Plus with (Stichlmair et al., 1989) Packing Holdup Correlation Recirculated flow rate (kg/h) Water flow rate (kg/h) Total liquid flow rate (kg/h) Flue gas flow rate (kg/h) Column diameter (m) Column packing IMTP 25 mm IMTP 25 mm Packed height per stage (m) 1 1 *Two types of modes can be used in Aspen Plus for calculations within the absorbers: the equilibrium mode and the rate-based mode. The equilibrium mode used to simulate absorption columns in Aspen Plus considers that the streams leaving a section of packed column are in equilibrium. In equilibrium mode, either vaporization or Murphree efficiency can be specified. The Murphree efficiency is based on a semi-theoretical model that assumes that the vapor streams incoming and outcoming a section packed column have a uniform composition and are in equilibrium. The same conventions are taken for the liquid incoming and outcoming the section packed column. It is defined for each packed stage according to the separation achieved on the latter. This can be based on either the liquid phase or the vapor phase. For a given component, it is equal to the change in actual concentration in the phase, divided by the change predicted by equilibrium conditions. Rate-based mode takes into account the mass transfer in vapor and liquid phases. Thermodynamic phase equilibrium between vapor and liquid phases exists only in the phase interface. On both sides of the phase interface there are interfacial films that are the places where the mass transfer limitations exists. Further away from the phase interface there are bulk vapor phase and bulk liquid phase that are assumed to be perfectly mixed in each calculation segment. In rate-based mode, RadFrac model (the main separation block in Aspen Plus, the block can perform simulation, sizing, and rating of tray and packed columns) calculates Murphree efficiencies for each component as the fractional approach to equilibrium of the vapor stream leaving each packed stage and reports the height of packing required to achieve equilibrium as the Height Equivalent to a Theoretical Plate (HETP) for that section. In rate-based calculations, an equilibrium-based initialization step is used only as a first step of the rate-based calculations (and then neglected) including tray efficiencies calculations, holdups for reactions and certain convergence parameters. In our case, rate based calculations for the simultaneous SOx/NOx absorption in the contactors are selected. 64

73 The gas components are going to be washed in the contactors by the liquid scrubbing solution. Due to different affinity coefficients relatively to water, the driving forces are different for each gas absorption and consequently the components can more or less dissolve into water. Henry components could be selected in Aspen Plus as O 2, N 2, Ar, CO, CO 2, SO 2, SO 3, H 2SO 4, NO, NO 2, N 2O 3, N 2O 4, HNO 2, HNO 3, N 2O. The Electrolyte NRTL model calculates the Henry s constant for a dissolved gas component (i) in a solvent (j). The Henry s law in Aspen is expressed as follows: Ln H ij = a ij + b ij T + c ijln T + d ij T + e ij T² Where the Henry's constants a ij, b ij, c ij, d ij and e ij are specific to a solute-solvent pair (see table on Annex D). They can be obtained from regression of gas solubility data. The Aspen Physical Property System has a large number of built-in Henry's constants for many solutes in solvents. These parameters can either be obtained using data from Aspen databanks or introduced manually from the literature. In our case, the constants were all taken from the literature and transformed in their form in order to suit Aspen Plus specifications Simulation results of the SCU chemical mechanisms Table 30,Table 31 and Table 32 present the simulations results of the SCU for the different chemical mechanisms schematized in

74 Table 30 : SCU simulation results considering the chemical mechanism without interactions (mechanism A) GAS LIQUID Components/ Mole fractions Inlet De-SOx column Inlet De-NOx column Outlet De-NOx column Inlet De-SOx column Outlet De-SOx column Inlet De-NOx column Outlet De-NOx column O2 3.29E E E E E E E-05 N2 1.12E E E E E E E-05 Ar 1.35E E E E E E E-06 H2O 3.33E E E E E E E-01 OH E E E E-13 H3O E E E E-05 CO 3.99E E E E E E E-07 CO2 8.37E E E E E E E-02 HCO E E E E-06 CO E E E E-14 SO2 1.57E E E E E E E-06 HSO E E E E-05 SO E E E E-10 NO 8.67E E E E E E E-09 NO2 9.63E E E E E E E-11 N2O E E E E E E-15 N2O E E E E E E-19 HNO E E E E E E-06 HNO E E E E E E-09 NO E E E E-07 NO E E E E-06 Total gas mole flow (kmol/h) Temperature ( C) Pressure (bar) ph Abatement rates (%) De-SOx De-NOx column column Total SCU SO NO NO CO

75 Table 31: SCU simulation results considering the chemical mechanism with the main interactions (mechanism B) GAS LIQUID Components/ Mole fractions Inlet De-SOx column Inlet De-NOx column Outlet De-NOx column Inlet De-SOx column Outlet De-SOx column Inlet De-NOx column Outlet De-NOx column O2 3.29E E E E E E E-05 N2 1.12E E E E E E E-05 Ar 1.35E E E E E E E-06 H2O 3.33E E E E E E E-01 OH E E E E-13 H3O E E E E-06 CO 3.99E E E E E E E-07 CO2 8.37E E E E E E E-02 HCO E E E E-06 CO E E E E-13 SO2 1.57E E E E E E E-14 HSO E E E E-13 SO E E E E-17 SO E E E E E E-30 HSO E E E E-19 SO E E E E-17 H2SO E E E E E E-25 NO 8.67E E E E E E E-08 NO2 9.63E E E E E E E-10 N2O E E E E E E-11 N2O E E E E E E-14 HNO E E E E E E-06 HNO E E E E E E-11 NO E E E E-07 NO E E E E-08 NSS E E E E-16 Total gas mole flow (kmol/h) Temperature ( C) Pressure (bar) ph Abatement rates (%) De-SOx De-NOx column column Total SCU SO NO NO CO

76 Table 32 : SCU simulation results considering the mechanism with all the interactions (mechanism C) GAS LIQUID Components/ Mole fractions Inlet De-SOx column Inlet De-NOx column Outlet De-NOx column Inlet De-SOx column Outlet De-SOx column Inlet De-NOx column Outlet De-NOx column O2 3.29E E E E E E E-05 N2 1.12E E E E E E E-05 Ar 1.35E E E E E E E-06 H2O 3.33E E E E E E E-01 OH E E E E-13 H3O E E E E-05 CO 3.99E E E E E E E-07 CO2 8.37E E E E E E E-02 HCO E E E E-06 CO E E E E-13 SO2 1.57E E E E E E E-09 HSO E E E E-08 SO E E E E-12 SO E E E E E E-26 HSO E E E E-13 SO E E E E-12 H2SO E E E E E E-16 NO 8.67E E E E E E E-09 NO2 9.63E E E E E E E-10 N2O E E E E E E-12 N2O E E E E E E-14 HNO E E E E E E-06 HNO E E E E E E-13 N2O E E E E E E-14 HNO E E E E E E-14 NO E E E E-07 NO E E E E-06 NSS E E E E-33 HADS E E E E-16 HAMS E E E E-16 NSSH E E E E E E-07 HADSH E E E E E E-25 HAMSH E E E E E E-13 Total gas mole flow (kmol/h) Temperature ( C) Pressure (bar) ph Abatement rates (%) De-SOx De-NOx column column Total SCU SO NO NO CO NB: The species highlighted in Table 31 and Table 32 are the supplementary species compared to the mechanism without interactions (A) (Table 30). Main conclusions for the gas phase compositions of mechanisms (A), (B) and (C) In the case of the gas phase compositions, Table 30 and Table 32 show a major difference in terms of SOx removal, indeed in the case of considering the chemical mechanism without interactions (A), the SOx abatement rate in the first column is 33.39% compared to 99.39% in the case of considering the chemical mechanism with all the interactions (C) (see Figure 51). At the end, in both cases, the second column removes all the remaining SOx components resulting on a total removal of SOx for the global process. 68

77 For the NOx components, the differences between the chemical mechanism without interactions (A) and the chemical mechanism with all the interactions (C) are slighter than the case of SOx abatement, the NO and NO 2 abatement rates in the first column are about 96% and 91% respectively, compared to 95% for both species in the case of considering the chemical mechanism with all the interactions (C) (see Figure 51). Like the SOx components, in both cases, the second column removes all the lasting NOx components resulting in a total removal of NOx within the global process. Nevertheless, it has to be noted that in both cases a production of HNO 2 is non negligible. Besides, the absorption performances of the chemical mechanism with the main interactions (B) (Table 31 and Figure 51) are almost equivalent to the absorption performances of the chemical mechanism including all the interactions (C): SOx abatement rate for the first column is 99% in the case of mechanism (B) and (C); NO and NO 2 are removed respectively by 96% and 99% in the first case but by 95% in the second case. Figure 51: Abatement rate of the SO 2 and NOx components in the first column of the SCU Consequently, as represented in Figure 51, the consideration of the interaction reactions has an important influence on the removal performances of SO 2, the NOx components being removed with comparable rates for mechanisms (A), (B) and (C). As explained in , due to the great number of reactions and the complex structure and components of some of them in the complete chemical mechanism, Aspen Plus calculations and therefore convergence are very difficult to achieve when variating the operational parameters like required for the sensitivity analysis. As shown for the gas phase analysis in Figure 51, the two mechanisms (B) and (C) satisfy the same absorption targets for the SOx components (SO 3 remaining quantity being negligible compared to SO 2) and for the total NOx components. Therefore, the use of the chemical mechanism with the main interactions (B) will be considered for the parametric study (optimization step) followed by the simulation of the optimized case for the complete chemical mechanism (C). 69

78 Main conclusions for the liquid phase compositions of mechanisms (A), (B) and (C) The ph influence on the importance of the reaction pathways that may occur and the N-S interactions (and deriving complexes) has been highlighted in and Choosing the adequate reaction pathways to construct the chemical mechanism elucidated in Figure 47 has been elaborated by means of the determination of the ph of the solution given directly by Aspen Plus (see Table 30,Table 31 and Table 32). The analysis and comparison of the liquid phase compositions for the chemical mechanisms (A), (B) and (C) confirmed the relevance of the choice of the ph range (1<pH<4) for the selection of the reactions: the mechanism (A) gives a liquid phase composition with a ph between 1.35 and 2.80 while the liquid phase composition of mechanism (B) has a higher ph namely between 2.25 and 3.31 and the liquid phase composition of mechanism (C) has a ph between 2.01 and The ph of the system is then between 1 and 4, which justifies the reactions choice made to construct the chemical mechanism in Figure 47. The ph of the chemical mechanism (A) is globally lower than the ph of (B) and (C), the majority of the removal of N-S components occurring in that step of reactions. Likewise, for all mechanisms, the ph is obviously lower in the first absorber since the SOx/NOx simultaneous reactive absorption is more important in the first column (the scrubbing solution is more acid) as the second contactor removes only the remaining components from the first absorption. Moreover, N 2O is only present for the mechanism (C) as a result of taking into account all the interaction reactions occurring in this case and precisely the implementation of reaction 23 (Table 18). This observation gives an answer to the question present in the conclusions of (White et al., 2013) where N 2O emission was not elucidated. However, emissions of N 2O have to be considered carefully and a control of the amount of N 2O formed is important for environmental aspects as the latter is a harmful greenhouse gas whose global warming potential is 298 times the one of carbon dioxide (mass basis) in 100 years. Consequently, conclusions showed in Figure 44 are confirmed by the analysis of the liquid phase of mechanism (C) (Table 31): At the outlet of the first absorber, the ph of the liquid phase is equal to 2.01, at this ph HADS and N 2O are formed, HAMS is also formed as a result of the hydrolysis of HADS but remains low at 1<pH<4 and in minor proportion compared to the formation HADS. It has to be noted also that the CO 2 abatement rate increases when increasing the ph of the solution from 0.98% for mechanism (A) to 2.98% for mechanism (C) and to 17.86% for mechanism (B). Finally, it can be highlighted from the liquid phase analysis that the NSSH is the major species in the liquid phase (5.74E-5 and 1.63 E-7 respectively at the outlet of the De-SOx column and the De-NOx column). 70

79 Conclusions and perspectives of the simulation works Once the chemical mechanism pathways have been identified and the corresponding kinetic data have been fixed, the SCU performances can be quantified. The addition of reactions of SOx/NOx interactions showed a great increase on SOx removal compared to the mechanism without considering the interactions, the important effect of this type of reactions is then demonstrated. An optimization of the SCU simulated process is currently under investigation using the process modeled by (Iloeje et al., 2015) as described in point Absorption performances of our simulations using compositions and design specifications of MIT team with our chemical mechanism defined for the SCU will therefore be compared with the ones obtained by Iloeje and co-workers. Nevertheless, the conditions of high pressure levels require an important demand of energy (as shown in report 1), i.e. ~0.15 kwh/kgco 2 as electrical needs for the compression specifically. Thus, an optimization of the SCU for the cement plant case is also under progress considering the energetic, environmental and economic aspects through a parametric study. This sensitivity analysis is based on the variations of operational levels of pressure, water flow rate, recycle flow rate and design parameters. A single column approach for the SCU has been investigated by Iloeje et al. in the case of power plants exploiting the pressure dependence of the rate determining reactions in the NOx/SOx removal process and showing the following advantages: a decrease of the column stages, a 50% decrease of the water demand, a faster overall conversion process by working directly with a pressure fixed at 30 bar and significant decreases of the material costs and energy requirements. The final objective of the present work is to take advantage of the pressure dependence of the rate determining reactions and of higher solubilities in the SO x/no x system, in order to reduce the process to a single column absorber that could be applied to cement industries. 71

80 4. General conclusion Full and partial oxyfuel technologies were presented as solutions able to significantly reduce anthropogenic CO 2 emissions and, at the same time, leading to CO 2 concentrated flue gases for conversion step. For CO 2 Capture purpose, in the first case, a CO 2 purification unit is necessary to obtain pure final CO 2. However, in the second case a CO 2 post-combustion capture process as the absorption-regeneration one is required. Nevertheless, both configurations require certain modifications in the cement production and the adoption of new equipment for the on-site oxygen production and CO 2 post-processing. Among these modifications in the partial oxyfuel configuration a higher operational temperature in the calciner is required to achieve an equivalent decarbonation degree of the raw meal as in conventional cement plants. Likewise, other operational aspects have to be more investigated as air ingress control, energy demand optimization, pollutants removal to allow a successful implementation of the full or partial oxyfuel technology to the cement process. Some of these interrogations are treated by pilot plants investigations tests within the ECRA CCS project, the FLSmidth-Lafarge-Air Liquid and the CEMCAP research projects. The CO 2 post-combustion capture by means of the absorption-regeneration process applied to a O 2- enriched air combustion technology showed interesting experimental results for the evaluation of the absorption performances of a wide range of solvents (conventional solvents, activated solvents, hybrid solvents) in high CO 2 content conditions and for the identification of the most efficient solvents in these conditions (see section 2.6 for more details). Likewise, relevant simulation results in Aspen Hysys TM Software for the determination of the solvent regeneration energy under partial oxyfuel conditions were achieved. These results obtained for CO 2 gas compositions computed thanks to a sheet developed by ECRA showed that the regeneration energy decreases as Y CO2 is increased, for a fixed amount of CO 2 absorbed. Further works including absorption-regeneration tests using the micro-pilot unit for the best solvents screened during the absorption tests will be carried out soon. Evaluation of the impact of partial oxyfuel combustion conditions on the global chain (from O 2 production to CO 2 conversion) will also be investigated in terms of global energy consumption. Considering the full oxyfuel kiln configuration, the relevance of the flue gas recirculation ratio is an additional parameter to control the temperature profile and thermal energy demand in the kiln. Another important aspect is the control of the ingress of air to the process that is necessary to maintain high CO 2 concentrations in the generated flue gas. If an efficient impurity removal is not achieved before flue gas recirculation, impurities are recirculated to the cement kiln, this increase will be more pronounced since oxyfuel flue gases are not diluted by N 2 of air. However, an efficient removal as it is expected for SO 2 would avoid this. SO 2, in an oxyfuel cement process as in an air-fired process can be controlled to large extent by self-capture of SO 2 by solids of the clinker. In the oxyfuel process, an additional SO 2 removal is possible in the flue gas condenser that the flue gases pass during recirculation. 72

81 Despite of all these pre-treatments, a purification step is required to remove eventual remaining impurities in the flue gas that are likely to poison the catalyzer further used for the conversion step of the CO 2 into methanol. For this purpose, the CO 2 Processing Unit (CPU) was tested experimentally on pilot plant scales by Air products, Linde, Praxair and Air Liquide and also by models developed by other research teams but all of them are applied to power plants cases. The topic is thus brand-new leading to a clear lack of data about the CO 2 purification from oxyfuel cement plants. Compositions for the outcoming flue gas from a cement industry applying an oxyfuel combustion, given by ECRA calculations, were taken as inputs of the process simulations in Aspen Plus in order to evaluate the absorption performances of a CO 2 De-SOx and De-NOx process in pressurized systems thanks to two successive columns. Results showed that this configuration can be used to remove efficiently SO x and NO x from the fed stream, acid liquid effluents being recovered at the outlet of the columns. Nevertheless, an identification of the key parameters for an accurate representation of the absorption phenomenon is required by considering the interactions between SOx and NOx species (HNO2 + HSO 3 - ), the ph influence, and the characterization of the deriving N-S complexes in order to build a new optimized chemical mechanism for the Sour-Compression Unit. An optimization of the SCU for the cement plant case is currently under progress considering the energetic, environmental and economic aspects through a parametric study. The final purpose of the present work is to take advantage of the pressure dependence of the rate determining reactions and of higher solubilities in the SO x/no x system, in order to reduce the process to a single column absorber that could be applied to cement industries. Finally, scientific activities achieved during this last year of study are listed hereafter. 73

82 Doctoral Training activities of Sinda LARIBI from January 2015 to January 2016 Co-promotor of student master thesis -Master thesis of Guillaume Pierrot, Experimental and simulation study of CO 2 absorption into amine(s) based solvents: applications to cement flue gases coming from partial oxyfuel kilns, from February to June Scientific publication -Meunier Nicolas, Laribi Sinda, Dubois Lionel, Thomas Diane, De Weireld Guy, "CO 2 capture in cement production and re-use: first step for the optimization of the overall process" in Energy Procedia, 63, , doi: /j.egypro Research reports -Thomas Diane, De Weireld Guy, Dubois Lionel, Laribi Sinda, Meunier Nicolas, "ECRA Chair From CO 2 to Energy: Carbon Capture in Cement Production and its Re-use : Second Annual Report, 16/05/ Dubois Lionel, Laribi Sinda, "SaskPower s Boundary Dam power plant: Integrated Carbon Capture and Storage project - Report on the industrial visit ( )", 05/11/ Laribi Sinda, "First PhD Thesis committee: Purification process applied to CO 2 captured from cement industry for conversion into methane and methanol.", 26/01/2015. Oral presentation at a scientific event -Dubois Lionel, Laribi Sinda, Meunier Nicolas, De Weireld Guy, Thomas Diane, "Global optimization of the CO 2 capture and reuse applied in the cement industry" in "Brussels Sustainable Development Summit 2015", Brussels, Belgium, Laribi Sinda, Dubois Lionel, Thomas Diane, "Etude de la capture du CO 2 en postcombustion: screening de solvants aminés pour le procédé d absorption-régénération appliqué aux fumées de cimenteries à hautes teneurs en CO 2" in "7ème Journée des Jeunes Chercheurs en Génie des Procédés et Energétique", Saint-Quentin, France, Laribi Sinda, Pierrot Guillaume, Dubois Lionel, Thomas Diane, "Post-combustion CO 2 capture: screening tests of solvents for the absorption-regeneration process applied to cement flue gases with high CO 2 contents" in "3rd Post Combustion Capture Conference (PCCC3)", Regina, Canada, Laribi Sinda, "Capture, Purification du CO 2 et conversion en méthanol- Cas de l industrie cimentière" in "Institut de Recherche en Energie: Cinquième Journée Thématique : Traitement des effluents gazeux CCS Combustion Bioénergie", Mons, Belgium, a

83 Poster presentations at scientific events -Laribi Sinda, Dubois Lionel, Thomas Diane, "Post-combustion CO 2 capture applied to cement plant flue gases: screening tests of innovative solvents for the absorption-regeneration process" in "10th European Congress of Chemical Engineering (ECCE 10)", Nice, France, Pierrot Guillaume, Gervasi Julien, Laribi Sinda, Dubois Lionel, Thomas Diane, "Post-combustion CO 2 capture: optimization of the absorption-regeneration process for the application to cement flue gases" in "3rd Post Combustion Capture Conference (PCCC3)", Regina, Canada, Pierrot Guillaume, Gervasi Julien, Laribi Sinda, Dubois Lionel, Meunier Nicolas, De Weireld Guy, Thomas Diane, "Innovative solvents for the post-combustion CO 2 capture absorption-regeneration process applied to cement plant flue gases" in "International Conference on Carbon Dioxide Utilization (ICCDU XIII)", Singapore, Singapore, Meunier Nicolas, Laribi Sinda, Dubois Lionel, Thomas Diane, De Weireld Guy, "CO 2 capture and re-use from oxyfuel cement kilns: process simulation of the CO 2 purification and catalytic conversion into methanol" in "International Conference on Carbon Dioxide Utilization (ICCDU XIII)", Singapore, Singapore, Pierrot Guillaume, Laribi Sinda, Dubois Lionel, Thomas Diane, "Solvent screening for the postcombustion CO 2 capture applied to flue gases coming from conventional and partial oxyfuel combustion cement kilns" in "8th Trondheim Conference on CO 2 Capture, Transport and Storage (TCCS- 8)", Trondheim, Norway, Laribi Sinda, Meunier Nicolas, Dubois Lionel, De Weireld Guy, Thomas Diane, "Simulation of a CO 2 purification unit applied to flue gases coming from oxy-combustion cement industries" in "8th Trondheim Conference on CO 2 Capture, Transport and Storage (TCCS-8)", Trondheim, Norway, Laribi Sinda, Meunier Nicolas, Dubois Lionel, De Weireld Guy, Thomas Diane, "Simulation of a CO 2 purification unit applied to flue gases coming from oxy-combustion cement industries" in "8ème édition de la Matinée des Chercheurs 2015 (MdC2015)", Mons, Belgium, ECRA Academic Chair activities -Meetings and oral technical presentations during the ECRA Academic Chair Scientific Committees, 30/03/2015 in Düsseldorf, Germany and 22/10/2015 in Mons, Belgium. Participation to scientific conferences -Participation to the Norcem International CCS Conference (HeidelbergCement Langesund (Norway) and meeting at the Telemark University College of Porsgrunn, from 19/05/2015 to 21/05/2015. Awards - Award to the best oral communication: Laribi Sinda, Dubois Lionel, Thomas Diane, "Etude de la capture du CO 2 en postcombustion: screening de solvants aminés pour le procédé d absorptionrégénération appliqué aux fumées de cimenteries à hautes teneurs en CO 2" in "7ème Journée des Jeunes Chercheurs en Génie des Procédés et Energétique", Saint-Quentin, France, b

84 BIBLIOGRAPHIC REFERENCES Adeosun, A., Muthiah, A., & Abu-Zahra, M. R. M. Evaluation of Oxygen-Enriched Air Combustion Process Integrated with CO 2 Post-Combustion Capture. Thermal & Enviromental Engineering, 5(2), , Ajdari, S., Normann, F., Andersson, K., & Johnsson, F. Modeling the nitrogen and sulfur chemistry in pressurized flue gas systems. Industrial and Engineering Chemistry Research, Amann, J.-M. G., & Bouallou, C. Kinetics of the Absorption of CO 2 in Aqueous Solutions of N - Methyldiethanolamine + Triethylene Tetramine. Industrial & Engineering Chemistry Research, 48(8), , Armitage, J. W., & Cullis, C. F. Studies of the reaction between nitrogen dioxide and sulfur dioxide. Combustion and Flame, 16(2), , Becker, K. H., J. Kleffmann, R. Kurtenbach, and P. Wiesen. Solubility of nitrous acid (HONO) in sulfuric acid solutions. J. Phys. Chem., 100, , Berdnikov, V. M. and N. M. Bazhin. Oxidation-reduction potentials of certain inorganic radicals in aqueous solutions. Russ. J. Phys. Chem., Engl. Transl., 44, , Bonner, F. T., & Hughes, M. N. The Aqueous Solution Chemistry of Nitrogen in Low Positive Oxidation States. Comments on Inorganic Chemistry, 7(4), , Bougie, F., & Iliuta, M. C. Kinetics of absorption of carbon dioxide into aqueous solutions of 2-amino- 2-hydroxymethyl-1,3-propanediol. Chemical Engineering Science, 64(1), , Caplow, M., Kinetics of carbamate formation and breakdown, J. Am. Chem, Soc., 90, , Carrasco-Maldonado, F., Spörl, R., Fleiger, K., Hoenig, V., Maier, J., & Scheffknecht, G. Oxyfuel combustion technology for cement production State of the art research and technology development. International Journal of Greenhouse Gas Control, 45, , Chameides, W. L. The photochemistry of a remote marine stratiform cloud. J. Geophys. Res., 89D, , CO2CRC (2015), CO 2 capture/separation technologies, accessed 07/01/2016, Danckwerts, P. V., The reaction of CO 2 with ethanolamines, Chem. Eng. Sci., 34, , Derks, P. W. J., Kleingeld, T., van Aken, C., Hogendoorn, J. A., & Versteeg, G. F. Kinetics of absorption of carbon dioxide in aqueous piperazine solutions. Chemical Engineering Science, 61(20), , Doukelis, A., Vorrias, I., Grammelis, P., Kakaras, E., Whitehouse, M., & Riley, G. Partial O2-fired coal power plant with post-combustion CO 2 capture: A retrofitting option for CO 2 capture ready plants. Fuel, 88(12), , Dubois, L. Etude de la capture du CO 2 en postcombustion par absorption dans des solvants aminés : application aux fumées issues de cimenteries, Dubois, L. SaskPower s Boundary Dam power plant : Integrated Carbon Capture and Storage project Presentation of the Boundary Dam CCS Project, (November), 1 11, Durham, J. L., J. H. Overton, Jr., and V. P. Aneja. Influence of gaseous nitric acid on sulfate production and acidity in rain. Atmos. Environ., 15, , i

85 England, C., & Corcoran, W. Kinetics and mechanisms of the gas-phase reaction of water vapor and nitrogen dioxide. Industrial & Engineering Chemistry, 13(4), England, C., & Corcoran, W. H. The Rate and Mechanism of the Air Oxidation of Parts-per-Million Concentrations of Nitric Oxide in the Presence of Water Vapor. Industrial & Engineering Chemistry Fundamentals, 14(1), 55 63, Favre, E., Bounaceur, R., & Roizard, D. A hybrid process combining oxygen enriched air combustion and membrane separation for post-combustion carbon dioxide capture. Separation and Purification Technology, 68(1), 30 36, Gervasi, J., Etude exploratoire de la capture du CO 2 en post-combustion par absorption dans des solvants hyrides, Master thesis (UMONS), Gmitro, J. I. and T. Vermeulen. Vapor-liquid equilibria for aqueous sulfuric acid. AIChE J., 10, , Higginbotham, P., White, V., Fogash, K., & Guvelioglu, G. Oxygen supply for oxyfuel CO 2 capture. International Journal of Greenhouse Gas Control, 5(SUPPL. 1), S194 S203, Hoenig, V., Helmut, H., Koring, K., & Lemke, J. ECRA CCS Project Report on Phase III, 107, Hoftyzer P. J. and Kwanten J. G., Absorption of nitrous gases, Nonhebel. Butterwoyhs, London, pp , Holma H. and Sohlo J., A mathematical model of an absorption tower of nitrogen oxides in nitric acid production, Comput. Chem. Eng., vol. 3, no. 1 4, pp , Jan Iloeje, C., Field, R., & Ghoniem, A. F. Modeling and parametric analysis of nitrogen and sulfur oxide removal from oxy-combustion flue gas using a single column absorber. Fuel, 160(X), , Jankowski, M. Group Contribution Method for Thermodynamic Analysis of Complex Metabolic Networks. Biophysical Journal, 95(3), , Li, J., & Li, B. Calculation of thermodynamic properties of hydrated borates by group contribution method. Phys Chem Minerals, , Li, L., Li, H., Namjoshi, O., Du, Y., & Rochelle, G. T. Absorption rates and CO 2 solubility in new piperazine blends. Energy Procedia, 37, , Lide, D. R. and H. P. R. Frederikse, editors. CRC Handbook of Chemistry and Physics, 76th Edition. CRC Press, Inc., BocaRaton, FL, Littlejohn, D., & Chang, S. Identification of Species in a Wet Flue Gas Desulfurization and Denitrification System by Laser Raman Spectroscopy. Environmental Science & Technology, 310(5), , Meer, R. Van Der, & Ravail, H. Industrial Emissions Directive Update, (January), Murciano, L. T., White, V., Petrocelli, F., & Chadwick, D. Sour compression process for the removal of SOx and NOx from oxyfuel-derived CO 2. Energy Procedia, 4, , Naiditch, S., & Yost, D. M. The Rate and Mechanism of the Hydrolysis of Hydroxylamine Disulfonate Ion, 3 7, Normann, F., Jansson, E., Petersson, T., & Andersson, K. Nitrogen and sulphur chemistry in pressurised flue gas systems: A comparison of modelling and experiments. International Journal of Greenhouse Gas Control, 12(2), 26 34, ii

86 Oblath, S. B., Markowitz, S. S., Novakov, T., & Chang, S. G. Kinetics of the Initial Reaction of Nitrite Ion in Bisulfite Solutions. J. Phys. Chem, 86, , Patil, G. N., Vaidya, P. D., & Kenig, E. Y. Reaction kinetics of CO 2 in aqueous methyl- and dimethylmonoethanolamine solutions. Industrial and Engineering Chemistry Research, 51(4), , Perry, R. H. Perry s Chemical Engineers Handbook, fourth edition. McGraw-Hill, Inc., Petrissans, M. S., & Zoulalian, A. Influence of the ph on the Interactions between Nitrite and Sulfite Ions. Kinetic of the Reaction at ph 4 and 5. Ind. Eng. Chem, , Pétrissans, S. M., Pétrissans, A., & Zoulalian, A. Experimental study and modelling of mass transfer during simultaneous absorption of SO 2 and NO 2 with chemical reaction. Chemical Engineering and Processing: Process Intensification, 44(10), , Pierrot, G. Simulation study of CO 2 absorption into amine ( s ) based solvents : application to cement flue gases coming from partial oxyfuel kilns, Rayson, M. S., Mackie, J. C., Kennedy, E. M., & Dlugogorski, B. Z. Accurate Rate Constants for Decomposition of Aqueous Nitrous Acid. Inorganic Chemistry, 51(4), , Sander, R. and P. J. Crutzen. Model study indicating halogen activation and ozone destruction in polluted air masses transported to the sea. J. Geophys. Res., 101D, , Santos, S. Oxy-Coal Combustion Power Plant with CCS Current Status of Development. 7th Dutch CCS Symposium, (June), Santos, S. Update and Status of Development of CO 2 Processing Unit, (October), Smart, J. P. P., & Riley, G. S. S. Use of oxygen enriched air combustion to enhance combined effectiveness of oxyfuel combustion and post-combustion flue gas cleanup Part 1 - combustion. Journal of the Energy Institute, 85(3), , Stichlmair, J., Bravo, J. L., & Fair, J. R. General model for prediction of pressure drop anc capacity of countercurrent gas/liquid packed column. Gas Separation & Purification, 61(12), 19 28, Torrente-Murciano, L., White, V., Petrocelli, F., & Chadwick, D. Study of individual reactions of the sour compression process for the purification of oxyfuel-derived CO 2. International Journal of Greenhouse Gas Control, 5(SUPPL. 1), S224 S230, Versteeg G.F., Van Dijk L.A.J. and Van Swaaij W.P.M., On the kinetics between CO 2 and alkanolamines both in aqueous and non-aqueous solutions. An overview, Chem. Eng. Comm., 144, pp , Versteeg, G. F., Kuipers, J.. M., Van Beckum, F. P. H., & Van Swaaij, W. P. M. Mass transfer with complex reversible chemical reactions-ii. parallel reversible chemical reactions. Chemical Engineering Science, 45(1), , Versteeg, G. F., & Van Swaaij, W. P. M. Solubility and diffusivity of acid gases (carbon dioxide, nitrous oxide) in aqueous alkanolamine solutions. Journal of Chemical and Engineering Data, 33(1), 29 34, Wagman, D. D., Evans, W. H., Parker, V. B., Schumm, R. H., Halow, I., Bailey, S. M., Nuttall, R. L. The NBS tables of chemical thermodynamic properties. Journal of Physical and Chemical Reference Data, 11(2), White, V., Wright, A., Tappe, S., & Yan, J. The Air Products Vattenfall Oxyfuel CO 2 Compression and Purification Pilot Plant at Schwarze Pumpe. Energy Procedia, 37, , iii

87 Wilhelm, E., R. Battino, and R. J. Wilcock. Low-pressure solubility of gases in liquid water. Chem. Rev., 77, , Xu, G., Zhang, C., Qin, S., Gao, W., & Liu, H. Gas Liquid Equilibrium in a CO 2 MDEA H 2O System and the Effect of Piperazine on It. Industrial & Engineering Chemistry Research, 37(95), , iv

88 Annex A Continuous tests with pre-loading (10% <YCO2,in < 60% and initial αco2 0): Based on the best results obtained with the continuous and semi-continuous loading tests, a study of the variation of Y CO2 by varying the α CO2 was conducted with MEA 30% (benchmark), MMEA 30% (best performances during continuous tests), and TETRA 30% (best results in semi-continuous tests after a whole recirculation, surpassing even the MMEA in terms of G CO2, abs). In these tests, experiments were also carried out with a double stirred cell contactor to determine the maximum value of the achievable loading for the solvent (α sat). CO 2 bubbling was done in order to load up to 200 ml of solvent. The CO 2 loading values of the solvents are shown in Table 1. Table 1: α sat of the different solvents Solvent αsat (mol CO 2/mol amine) MEA 30% 0.54 MMEA 30% 0.59 TETRA 30% 1.32 For the MEA, the measured value is α sat = 0.54 mol CO 2/mol MEA (for a total of 122 minutes of bubbling).this value is close to the maximum theoretical capacity (0,5 mol CO 2/mol amine). For MMEA 30%, α sat is equal to 0.59 mol CO 2/mol MMEA (for a total of 95 minutes of bubbling). TETRA 30% allows loading more CO 2 with α sat=1.32 mol CO 2/mol TETRA (for a total of 115 minutes of bubbling) as showed in the results of the semi-continuous tests without pre-loading. MEA 30% As explained below, when the loading increases, the absorption performances of the amine decrease, slippage is observed from the curve A downwards (Figure 1). YCO2,in = 40% Y CO2, in (%) Figure 1: Effect of CO 2 loading on the absorption performances of the MEA 30 Figure 2: Comparison of G CO2,abs obtained in case of semicontinuous test and with the preloaded solutions (Y CO2,in = 40%) A

89 MMEA 30% YCO2,in = 40% Y CO2, in (%) Figure 3: Effect of CO 2 loading on the absorption performances of the MMEA 30% Figure 4: Comparison of G CO2,abs obtained in case of the semi-continuous test and with the preloaded solutions (MMEA 30%; Y CO2,in = 40%) TETRA 30% YCO2,in = 40% Y CO2, in (%) Figure 5: Effect of CO 2 loading on the absorption performances of the TETRA 30% Figure 6: Comparison of G CO2,abs obtained in case of the semi-continuous test and with the preloaded solutions (TETRA 30%; Y CO2,in = 40%) For the three systems tested, two mains conclusions could be highlighted: -The absorption performances decrease when the solution has been pre-loaded with CO 2 (Figures 1, 3, 5) compared to the absorption performances of a non-pre-loaded solution. -The comparison of G CO2 abs for the pre-loading tests at Y CO2= 40% are the same as G CO2 abs for the semi-continuous tests with recirculation at fixed Y CO2= 40% (Figures 2, 4, 6). B

90 Annex B Strategy of implementation in Aspen Plus Several elements can complicate the calculations and thus the convergence in Aspen Plus. The complexity of the system comes from the complex new species to introduce in Aspen Plus (6 new components not present in Aspen Plus databanks), the number of reactions (27 reactions) and also the total number of components that intervene in these reactions. Several simulations variating the following different implementation parameters were conducted in order to study their impact in the simulation results. Order of the reactions After several preliminary implementations, the order in which the reactions were implemented in Aspen Plus has shown to be important since a species cannot react if it is not already present in in the flue gas inputs or generated after a reaction implemented in Aspen Plus. Hence, reactions with species already present in the flue gas were firstly implemented and then the other reactions were implemented progressively following the appearance order of the species. Some reactions like reaction 11, 14, 15, 17, 19 in Figure 1 are key reactions and define highly the absorption performances. It has to be noted that in Aspen Plus, the order in which the reactions take place cannot be defined automatically, therefore, the implementation of the reactions in the required order has to be done manually. Figure 1: Order of the implementation of the reactions in Aspen Plus C

91 ABATEMENT RATE (%) Equilibrium/rate-based mode As described in , two modes could be used in Aspen Plus calculations: equilibrium and ratebased mode. Due to an easier convergence for the calculations in equilibrium mode than in rate based mode, when changing an operating parameter in the flowsheet it is required to run the simulation first with the equilibrium calculation mode. Then, the results of the equilibrium calculations could be taken as an initialization for the calculations in rate-based mode. Temperature and compositions estimates in the columns Besides, after a simulation run, generating estimates of compositions and/or temperatures within the columns could facilitate the convergence for the next calculations. Influence of the tolerance factor The tolerance factor is a relevant parameter to define for the rate-based calculations, to study its effect on the simulation results given by Aspen Plus, we made, fixing the initial composition and operational parameters, several simulations varying only the tolerance factor. As we can see in Figure 2, starting from a tolerance factor of 0.001, results become constant and stable SO2 abatement rate (%) NO abatement rate (%) NO2 abatement TOLERANCE FACTOR Figure 2: Influence of the tolerance factor Nevertheless, the study presented in Figure 2 has been done for a simpler chemical mechanism just to show the importance of this parameter but in our case there is a slight variation between the results presented with a tolerance factor of 0.01 and 0.1, besides, due to the complexity of the simulated system it is not possible to run the simulation with a tolerance factor more precise than 0.01 for both columns. Influence of the number of stages To assure the maximum precision in the results, several configurations were tested for the De-SOx column to identify the optimum number of stages for the calculations considering two different configurations: - Selecting a packing height per stage and then specifying the packing section (starting stage and ending stage) and a total column height that equals to the total packing height): 6 2m; 12 1m; ; m; m Figure 3. D

92 Component mol flow Kmol/h Component mol flow Kmol/h Number of column stages for the calculation SO2 NO NO2 Figure 3: Compositions at the outlet of the De-SOx selecting the packed height per stage Selecting a number of stages of 6 is imprecise to represent the results. From 16 stages to 24, a tolerance factor of is necessary to obtain a constant abatement rate at the outlet of the column so the run time is considerably increased. Working with 12 stages gives constant abatement rate at the outlet of the column using a tolerance factor of 0.01 or Selecting a section packed height and a packing section that equals to this section and a total column height that equals to the total packing height Figure Number of column stages for the calculation SO2 NO NO2 Figure 4: Compositions at the outlet of the De-SOx selecting a section packed height Selecting a number of stages <12 is imprecise to represent the results. From 14 stages to 20, a tolerance factor of is necessary to obtain a constant abatement rate at the outlet of the column so the run time is considerably increased. Working with 12 or 13 stages gives a constant abatement rate at the outlet of the column using a tolerance factor of 0.01 or A number of stages of 12, packing section: 1->12 and a section packed height of 12m are selected to assure a good precision for the rate based calculations and to give precise results without slowing too much the run time. E

93 E (GJ/t CO2 ) Recirculation rate E (GJ/t CO2 ) E (GJ/t CO2 ) Annex C Additional graphs for the ECRA sheet interpretation The graphs below represent complementary exploitation of the results obtained in the table from the ECRA calculation sheets (Table 9). 0.2 E(ASU)=f(Y O2 ) 0.2 E (ASU) =f(r) Y O2 (%vol) in the flue gas out Recirculation rate Figure 1: Evolution of the energy of the ASU with the O 2 composition in the flue gas out Figure 2: Evolution of the energy of the ASU with the recirculation rate The Figures 1 and 2 show respectively that the energy required for O 2 provided from the ASU increase with the O 2 fraction in the flue gas out (obviously) and with the recirculated gas into the kiln E (ASU)=f(air input flow rate) Air input flow rate (m 3 stp/kg clinker) R=f(air input flow rate) Air input flow rate (m 3 stp/kg clinker) Figure 3: Evolution of the energy of the ASU with the air input fraction Figure 4: Evolution of the recirculation rate with the air input fraction Figure 3 and Figure 4 show that the energy required for O 2 provided from the ASU and the recirculation rate decrease when increasing the air input flow rate into the kiln since the O 2 concentration in the inlet of the kiln provided by the air input flow rate is increased to satisfy the oxygen input demand. F

94 Annex D Henry components constants The Electrolyte NRTL model calculates the Henry s constant for a dissolved gas component (i) in a solvent (j). The Henry s law in Aspen is expressed as follows: Ln H ij = a ij + b ij T + c ijln T + d ij T + e ij T² Where the Henry's constants a ij, b ij, c ij, d ij and e ij are specific to a solute-solvent pair. Sources: [1]: (Lide & Frederikse, 1995) [2]: (Wilhelm et al., 1977) [3]: (Berdnikov & Bazhin, 1970) [4]: (Sander & Crutzen, 1996) [5]: (Gmitro & Vermeulen, 1964) [6]: (Durham et al., 1981) [7]: (Becker et al., 1996) [8] : (Chameides, 1984) [9] : (Perry, 1963) G