Applied Thermal Engineering - Seri 1 PDF [PDF]

Citation preview

PAKET DISEMINASI INFORMASI TERSELEKSI



APPLIED THERMAL ENGINEERING Seri: 1



S



alah satu alasan kenapa masih rendahnya jumlah dan mutu karya ilmiah Indonesia adalah karena kesulitan mendapatkan literatur ilmiah sebagai sumber informasi.Kesulitan mendapatkan literatur terjadi karena masih banyak pengguna informasi yang tidak tahu kemana harus mencari dan bagaimana cara mendapatkan literatur yang mereka butuhkan. Sebagai salah satu solusi dari permasalahan tersebut adalah diadakan layanan informasi berupa Paket Diseminasi Informasi Terseleksi (PDIT). Paket Diseminasi Informasi Terseleksi (PDIT) adalah salah satu layanan informasi ilmiah yang disediakan bagi peminat sesuai dengan kebutuhan informasi untuk semua bidang ilmu pengetahuan dan teknologi (IPTEK) dalam berbagai topik yang dikemas dalam bentuk kumpulan artikel dan menggunakan sumber informasi dari berbagai jurnal ilmiah luar negeri. Paket Diseminasi Informasi Terseleksi (PDIT) ini bertujuan untuk memudahkan dan mempercepat akses informasi sesuai dengan kebutuhan informasi para pengguna yang dapat digunakan untuk keperluan pendidikan, penelitian, pelaksanaan pemerintahan, bisnis, dan kepentingan masyarakat umum lainnya. Sumber-sumber informasi yang tercakup dalam Paket Diseminasi Informasi Terseleksi (PDIT) adalah sumber-sumber informasi ilmiah yang dapat dipertanggungjawabkan karena berasal dari artikel (full text) jurnal ilmiah luar negeri bidang terkait.



DAFTAR ISI



i



Pilih/klik judul untuk melihat full text



1.



6th International Conference on Clean Coal Technologies CCT2013



2.



The Sotacarbo gasification pilot platform: Plant overview, recent experimental results and potential future integrations



3.



Techno-economic analysis of a CO2 capture plant integrated with a commercial scale combined cycle gas turbine (CCGT) power plant



4.



Multi-fuel multi-product operation of IGCC power plants with carbon capture and storage (CCS



5.



Experimental studies of CO2 capture by a hybrid catalyst/adsorbent system applicable to IGCC processes



6.



A complete transport validated model on a zeolite membrane for carbon dioxide permeance and capture



7.



Modelling and simulation of intensified absorber for post-combustion CO2 capture using different mass transfer correlations



8.



Characterization of the oxy-fired regenerator at a 10 kWth dual fluidized bed calcium looping facility



9.



Reprint of “Experimental studies of single particle combustion in air and different oxyfuel atmospheres”



10. Reprint of “Experiences in sulphur capture in a 30 MWth Circulating Fluidized Bed boiler under oxy-combustion conditions”



DAFTAR ISI



11. Oxycombustion for coal power plants: Advantages, solutions and projects 12. Torrefaction: Unit operation modelling and process simulation 13. Understanding pulverised coal, biomass and waste combustion – A brief overview 14. Detailed investigation of flameless oxidation of pulverized coal at pilot-scale (230 kWth) 15. Numerical prediction of processes for clean and efficient combustion of pulverized coal in power plants 16. Low rank coal – CO2 gasification: Experimental study, analysis of the kinetic parameters by Weibull distribution and compensation effect 17. A proposal for SO2 abatement in existing power plants using rich in calcium lignite 18. Modelling and assessment of acid gas removal processes in coal-derived SNG production 19. Performance evaluation of high-sulphur coalfired USC plant integrated with SNOX and CO2 capture sections 20. Hot gas clean-up technology of dust particulates with a moving granular bed filter 21. Study of slagging and fouling mechanisms in a lignite-fired power plant 22. Thermodynamic analysis and comparison of retrofitting pre-drying



DAFTAR ISI



23. Mechanical activation of micronized coal: Prospects for new combustion applications 24. Effect of solution chemistry on cyanide adsorption in activated carbon 25. Hydrogen separation studies in a membrane reactor system: Influence of feed gas flow rate, temperature and concentration of the feed gases on hydrogen permeation 26. Using the second law first: Improving the thermodynamic efficiency of carbon dioxide separation from gas streams in an Endex calcium looping system 27. Dynamic modeling and validation of aminebased CO2 capture plant



Applied Thermal Engineering 74 (2015) 1



Contents lists available at ScienceDirect



Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng



Editorial



6th International Conference on Clean Coal Technologies CCT2013 The 6th International Conference on Clean Coal Technologies CCT2013 was held on 12e16 May 2013 in Thessaloniki, Greece. This was co-organized by the IEA Clean Coal Centre of London, the University of Western Macedonia, the Aristotle University of Thessaloniki, the National Technical University of Athens and the Centre for Research and Technology Hellas, Chemical Process & Energy Resource Institute, Thessaloniki. It was held under the auspices of the Greek Ministry of Macedonia and Thrace and with the support of the Greek Public Power Corporation. It was attended by 140 delegates from 28 different countries including many Member States of the European Union, Australia, Brazil, Canada, China, Egypt, Japan, Korea, Malaysia, Pakistan, Russian Federation, Serbia, Taiwan, Ukraine and the USA. In this series, the first three Conferences were held in Sardinia (2002, 2005 and 2007), the 4th in Dresden (2009) and the 5th in Zaragoza (2011) while the 7th Conference will be held in May 2015 in Krakow, Poland. The aims of the Conference were to examine and promote the use of clean coal technologies. The scope covered: carbon capture including oxy-firing, pre-combustion and post-combustion capture; combustion; gasification; co-firing; mercury; emissions and their control; low rank and low grade coals; coal characterization; coal preparation and upgrading; underground coal gasification; and international perspectives.



http://dx.doi.org/10.1016/j.applthermaleng.2014.11.037 1359-4311/© 2014 Elsevier Ltd. All rights reserved.



There were 84 oral presentations and 22 posters presented at the Conference while 77 papers were included in the Proceedings. This Special Issue contains 26 papers addressing topics of interest to thermal engineering applications, based on the versions presented at the 6th Conference. Authors were invited to submit updated papers to be considered for inclusion in this Special Issue, subject to normal peer reviewing procedure of the journal. It is hoped that the reader will find the papers of this Special Issue both stimulating and enjoyable. Petros A. Pilavachi, Managing Guest Editor* Department of Mechanical Engineering, University of Western Macedonia, Bakola and Sialvera Street, 50100 Kozani, Greece Robert Davidson, Guest Editor IEA Clean Coal Centre, Park House, 14, Northfields, London SW18 1DD, UK E-mail address: [email protected]. George Skodras, Guest Editor Department of Mechanical Engineering, University of Western Macedonia, Bakola and Sialvera Street, 50100 Kozani, Greece E-mail address: [email protected]. * Corresponding author. E-mail address: [email protected] (P.A. Pilavachi).



Applied Thermal Engineering 74 (2015) 2e9



Contents lists available at ScienceDirect



Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng



The Sotacarbo gasification pilot platform: Plant overview, recent experimental results and potential future integrations Alberto Pettinau*, Gabriele Calì, Eusebio Loria, Paolo Miraglia, Francesca Ferrara Sotacarbo S.p.A., c/o Grande Miniera di Serbariu, 09013 Carbonia, Italy



h i g h l i g h t s � The Sotacarbo pilot platform for electrical energy and H2 production is described. � Main results from gasification of different coals and biomass are summarized. � Lignite presents the best performance, with a syngas production of 73 Nm3/h. � Clean syngas with less than 1 ppm of H2S and COS can be directly fed to MCFC stack. � Syngas-fed MCFC can allow an electrical efficiency up to 32e33%.



a r t i c l e i n f o



a b s t r a c t



Article history: Received 25 July 2013 Accepted 30 December 2013 Available online 8 January 2014



Sotacarbo is currently developing several research projects to optimize a coal-to-hydrogen process configuration within the field of hydrogen production through coal gasification for distributed power generation. To achieve this goal, between 2007 and 2008, Sotacarbo built a flexible pilot platform within its Research Centre in the Serbariu former coal mine (Sardinia, Italy), which is still very much into operation. The platform includes a demonstration plant and a pilot air-blown fixed-bed gasifier, the latter equipped with a flexible syngas treatment line for combined power generation and CO2-free hydrogen production. This paper aims to give a brief description of the whole experimental equipment and to summarize the main results obtained during more than 1700 h of experimental tests in the pilot unit. It is also reported the gasification performance under different operating conditions. A number of different fuels and fuel blends were tested, including South African sub bituminous coal, Sardinian high sulfur coal, lignite from Alaska and wood chips from local forests. Alaska’s lignite reached the best gasification performance due to the high reactivity. There is also a quick examination of the syngas cleaning process’s main performance. Finally, the very high efficiency of sulfur compounds removal through a zinc oxide-based hot gas desulphurization process suggested to evaluate the possibility to integrate the plant with a fuel cell system for a high efficiency combined heat and power (CHP) generation. The main results of this theoretical assessment, reached through a properly developed simulation model, are also reported in this work.  2014 Elsevier Ltd. All rights reserved.



Keywords: Coal and biomass gasification Pilot plant Hydrogen production Molten carbonate fuel cell



1. Introduction Coal and the other fossil fuels will remain a significant source of energy all along the transition phase (it will last several decades) towards a sustainable worldwide energy system, mainly based on renewables and nuclear sources [1e3].



* Corresponding author. Tel.: þ39 0781 670 444; fax: þ39 0781 670 552. E-mail address: [email protected] (A. Pettinau). 1359-4311/$ e see front matter  2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.applthermaleng.2013.12.079



Among clean coal technologies, coal gasification could represent a competitive option for power generation and also for the production of chemicals or clean fuels, with particular reference to hydrogen. Hydrogen is universally considered one of the most promising energy carriers [4,5] and characterized by a worldwide production (18% from coal) greater than one billion of cubic meters per day [6,7]. Gasification processes have the distinctive advantage to be easily integrated with pre-combustion CO2 capture systems, typically more efficient and less expensive than the postcombustion processes [8e11].



3



A. Pettinau et al. / Applied Thermal Engineering 74 (2015) 2e9



In this context, Sotacarbo is engaged in a series of research projects in order to develop and to optimize a gasification process and an integrated syngas treatment line for CO2-free power generation and hydrogen production from coal and biomass. To achieve this goal, in a flexible pilot platform located in the Sotacarbo Research Centre, in Carbonia (South-West Sardinia, Italy) more than 1700 h of experimental tests had been performed since June 2008. The platform includes a 1.3 m diameter demonstration unit and a 0.3 m diameter pilot unit, the latter equipped with a complete syngas treatment line for both power generation and hydrogen production. The choice of a pilot platform configuration is the compromise between the need to develop a gasification process for medium and small-scale industrial applications and the interest in the development of coal-to-hydrogen integrated processes to be applied in large-scale power plants. The target of the gasification process (medium and small scale, up to 10e15 MWth) brought with itself the choice of a fixed-bed, air-blown gasifier. These processes handling is easier [12,13] and it is well known that the counter current fluid dynamics ensures a higher efficiency than other gasification technologies [14]. The Sotacarbo gasification technology allows to gasify different kinds of coal (including high sulfur low rank coals) and also biomass. The flexible configuration of the pilot syngas treatment line allows to test and characterize some gas treatment processes and materials (solvents, sorbents and catalysts) for syngas desulphurization, water-gas shift, CO2 removal, hydrogen purification and so on. These specific properties of the pilot unit allow Sotacarbo to provide technical support to third Companies for testing activities on specific fuels, materials and processes. This paper reports a short description of the pilot unit, together with an overview of the main experimental results obtained with the most representative tested fuels, which represents the core of the overall work. Moreover, a preliminary theoretical assessment on the potential integration of the pilot plant with an advanced syngas-feed molten carbonate fuel cell (MCFC) system is also presented. 2. An overview of the Sotacarbo pilot plant



Fig. 1. The overall Sotacarbo platform.



bottom, below the fuel grate, so that they are pre-heated by cooling the bottom ash, which are removed through the grate itself (an electrical pre-heating up to 250  C can be operated in order to avoid steam condensation before the injection into the gasifier). Temperature profile into the reactor can be determined through a probe, located near the reactor vertical axis and equipped with a series of 11 K-type thermocouples (with a measure range between 0 and 1200  C), and through a series of other 34 thermocouples located near the reactor’s wall and in the grate. The start-up of the gasifier is carried out by using a series of three ceramic lamps, located near the bottom of the fuel bed, which heat the fuel (initially wood pellets, usually mixed with a small amount of paraffinic material to promote the ignition) in an inert atmosphere. 2.2. Dust and tar removal system As shown by Fig. 3 [17], raw syngas from the gasification process is sent to an integrated skid which includes a wet scrubber, a first



The Sotacarbo pilot platform (Fig. 1) was built up to test different plant solutions at different operating conditions; therefore, it is characterized by a very flexible and simple layout. Both demonstration and pilot plants are based on a fixed-bed, up-draft and airblown gasification process, suitable to be fed with both coal and biomass. 2.1. Pilot gasification process The Sotacarbo pilot unit is based on a 0.3 m internal diameter gasifier (Fig. 2), equipped with a manual coal charging system, and a flexible syngas treatment system for both power generation and hydrogen production. For the feeding of the gasifier, fuel is provided in big bags; every bag is drown out from the storage area by a heaver and, through a tackle, it is charged in a proper hopper in order to empty the bag itself. Then, fuel is drown out from this hopper and charged into the gasifier. Fuel bed (which operates at about 0.11e0.14 MPa) is characterized by different operating zones, where the coal drying, devolatilization, pyrolisis, gasification and combustion processes take place. As fuel flows downwards, it is heated by the hot raw gas that moves upwards, coming from the gasification and combustion zones [15,16]. The gasification agents (air, if necessary enriched with oxygen, and steam) are introduced into the reactor near the



Fig. 2. Pilot gasifier.



4



A. Pettinau et al. / Applied Thermal Engineering 74 (2015) 2e9



cold gas desulphurization stage and an electrostatic precipitator (ESP). In particular, wet scrubber reduces syngas temperature from 150 to 300  C (depending on the particular operating conditions) to about 50  C and operates a primary dust and tar separation. The electrostatic precipitator allows, if necessary, to complete particulate and tar removal. Due to the need to use coals with very high sulfur contents and to protect the electrostatic precipitator by the effects of acid atmosphere, a first cold gas desulphurization stage (which typically uses an aqueous solution of sodium hydroxide as solvent) is installed immediately upwards of the ESP. Downstream of the ESP, syngas can be sent to the power generation line; moreover, depending on the goals of each experimental test, a portion of syngas (20e25 Nm3/h) can be sent to the hydrogen production line. 2.3. Power generation line Fig. 4. Hot gas desulphurization system.



Power generation line includes the second cold gas desulphurization stage, directly followed by an internal combustion engine (ICE), characterized by a nominal power output of about 24 kW, fed with clean syngas, which can be enriched in hydrogen. In particular, the second cold gas desulphurization stage is a packed column in which hydrogen sulphide is chemically absorbed through an aqueous solution of sodium hydroxide and hypochlorite [18]. During several specific experimental tests, the column has been also used with other solvents such as methyldiethanolamine (MDEA) for H2S absorption or with monoethanolamine (MEA) for carbon dioxide capture. A two-chamber gasometer (with an overall internal volume of 11.3 m3) has been recently installed to overcome problems of pressure variability and low syngas mass flow at the inlet of ICE (in particular when low reactive coal are gasified) and to ensure a constant electric energy production. 2.4. Hydrogen production line Hydrogen production line includes a compressor (just to win the pressure drops through the downwards equipment) followed by an electric heater, a dry hot gas desulphurization process, an



integrated water-gas shift (WGS) and CO2 absorption system and a hydrogen purification section. In particular, hot gas desulphurization process operates between 300 and 500  C (measured by K-type thermocouples located into the reactors’ beds) and includes three main components: a catalytic filter for carbonyl sulphide (COS) conversion and two hydrogen sulphide (H2S) adsorbers. In the catalytic filter, the small amount of COS contained in syngas is converted in H2S through the hydrogenation process, promoted by Ni-MoO3/Al2O3 catalyst [19]. On the other hand, the two hydrogen sulphide absorbers (Fig. 4) are disposed in lead-leg configuration and filled with a zinc oxidebased sorbent [20e22]. In particular, zinc oxide (ZnO) reacts with H2S producing zinc sulphide and steam [23e25]. Clean syngas from hot gas desulphurization system, with a H2S concentration typically lower than 10 ppm by volume (and frequently lower than 1 ppm), is sent to a double stage water-gas shift (WGS) process, with an intermediate and a final CO2 absorption system. In particular, WGS process takes place into two reactors operating at high temperature (HT, between 300 and 450  C) and low temperature (LT, about 250  C), respectively. Both conventional and advanced catalysts have been tested during the clean syngas



primary fuel



COLD GAS WET DESULPH. SCRUBBER (1st stage)



ESP



COLD GAS DESULPH. (2st stage)



electrical energy ICE



offgas



hydrogen-rich gas



syngas UP-DRAFT GASIFIER



PSA



raw syngas



COMPRESSOR ELECTRIC HEATER clean syngas



gasification agents



HT WGS



bottom ash COS HYDROGENATOR



HOT GAS DESULPH.



Fig. 3. Pilot plant simplified scheme [17].



CO2 SEPAR. (1st stage)



CO2 SEPAR. (2st stage) LT WGS



5



A. Pettinau et al. / Applied Thermal Engineering 74 (2015) 2e9



experimental campaigns [17]. Carbon dioxide absorption takes place into two identical bubbling reactors (Fig. 5), in which syngas is injected through 40 diffusers based on ceramic membranes and reacts, at about 30  C and atmospheric pressure, with amine-based solvents (typically MEA 5 M). An amine regeneration unit has been also installed to study the performance of thermal CO2 desorption of exhaust chemical amine-based solvents. The column, which can operate both in batch or continuous modes, is equipped with an electrical reboiler (operating at 120e150  C), a condenser, a mist separator to split CO2 gas from residual water and solvent vapors and a series of heat exchangers to preheat rich solvent and to cool lean solvent. Finally, hydrogen purification section is based on the pressure swing adsorption (PSA) technology, which is widely common in the industrial applications due to its low costs [26]. In particular, PSA is composed by a simple double-stage process based on carbon molecular sieves. The size of the secondary syngas treatment line, even if much smaller than the size of commercial scale plants, has been chosen in order to give reliable experimental data for the scale-up of future plants. Moreover, with the goal to ensure a full plant flexibility, as well as to simplify the management of the experimental pilot plant, the different cooling and heating devices are not fully integrated. 2.5. Control system and data collection In order to support the experimental tests, pilot plant is equipped with a control and data acquisition system which allows the monitoring of the main operating parameters (such as pressures, temperatures, volume flows and so on) and the evaluation of the process performance. Raw syngas flow is measured by a mass flow measurer with three different calibration curves, selected on the basis of hydrogen concentration. Syngas composition is measured by three different systems: (i) a double real-time oxygen analyser, (ii) a micro gaschromatograph and (iii) a real-time monitoring system. In particular, the two real-time oxygen measurers in raw syngas play a double role of safety control, to avoid the formation of explosive atmosphere, and performance indicator of the gasification process [27,28]. Upstream and downstream of each main plant component, a sampling outlet has been located in order to operate syngas analysis through both a micro gas chromatograph Agilent 3000 (which is calibrated with a standard gas mixture before each test and evaluates, every 3 min, the concentration of CO, CO2, H2, O2, CH4, N2, H2S, COS, C2H6 and C3H8 in the selected stream) and an ABB



Fig. 5. CO2 absorbers.



Table 1 Primary fuels characterization. Bitum. coal South Africa Proximate analysis (% by weight) Fixed carbon 72.58 Moisture 3.64 Volatiles 8.81 Ash 14.97 Ultimate analysis (% by weight) Total carbon 75.56 Hydrogen 3.86 Nitrogen 1.40 Sulfur 0.57 Oxygen n.a. Moisture 3.64 Ash 14.97 Thermal analysis (MJ/kg) LHV 27.18



Sulcis Italy



Usibelli Alaska



Wood chips Italy



40.65 7.45 40.45 11.45



31.33 17.64 41.00 10.02



18.30 7.70 73.63 0.37



66.49 6.18 1.41 7.02 n.a. 7.45 11.4



48.56 5.96 0.50 0.18 17.14 17.64 10.02



49.95 6.14 0.11 0.00 35.74 7.70 0.37



21.07



17.75



17.25



integrated analyzer (which gives the real-time composition of every stream by using different methods: a URAS26 IR module is used to analyze CO, CO2 and CH4; a thermal conductivity-based CALDOS25 module is used to measure H2 concentrations; a paramagnetic Magnos 206 module indicates O2 concentration; a UV Limas 11 module is used to evaluate H2S content). Finally, a tar analysis system is currently under set up and it will allow a careful measure of tar content in syngas upwards and downwards of wet scrubber and ESP. 3. Pilot plant performance As mentioned, pilot plant was tested for more than 1700 h of experimental tests since July 2008. The performance analysis here reported comes from the processing of the experimental data automatically collected by the system. 3.1. Gasification performance The experimental tests in the pilot plant have been operated using several kinds of feedstock. In particular, the results here reported are referred to the gasification of (i) a low sulfur South African coal, (ii) a mixture of the latter (80%) with a high sulfur local coal (from the Sulcis coal mine, located near the Sotacarbo pilot platform), (iii) a lignite from Alaska (Usibelli coal mine) and (iv) a local biomass. The latter is chipped stone pine (Pinus pinea) wood from local forests, supplied by the Sardinian Regional Authority for Forests within a specific cooperation. Each feedstock is fed to the reactor with a typical particle size between 5 and 15 mm. The proximate, ultimate and thermal analyses of these fuels, determined in the Sotacarbo laboratories, are shown in Table 1 (where LHV is the lower heating value of each fuel). Proximate analysis was performed by a thermogravimeter LECO TGA-701 (based on the ASTM D5142 “Moisture Volatile Ash” standard); ultimate analysis was obtained by a LECO Truspec CHN/S analyzer (based on ASTM standards); finally, heating values were measured by a LECO AC-500 calorimeter. The reported data are the averaged values of several analyses. The performance of the gasification process with the previously described feedstock is summarized in Table 2. The reported results are typically averaged during at least 6 h of steady-state operation of the reactor (typically, each run is characterized by an overall duration of about 16 h, but several longer runs have been also performed). As expected, due to its high reactivity and high volatile content, Usibelli coal allows to maximize the syngas production: among the



6



A. Pettinau et al. / Applied Thermal Engineering 74 (2015) 2e9



number of fuels tested in the plant, it presents, at now, the best performance. As a matter of facts, the structure of coal particles and the high moisture and volatiles contents (about 18 and 41% by weight, respectively) allow a significant increasing of the particle surface as coal drying and devolatilization processes take place; it involves a high fuel reactivity. In other words, during the first phase of the heating process, moisture and volatile matters are released by the coal particle, and char (the solid residue, basically composed by carbon and ash) assumes a porous structure, characterized by a wide surface. This promotes the gasification process, in which char’s carbon reacts with oxygen and steam. On the contrary, the low reactive South African coal appears not completely suitable for the gasification in the Sotacarbo air-blown process (since it operates at about atmospheric pressure); this is confirmed by the relatively high carbon content frequently found (through a thermogravimetric analysis) in the discharged ash during the experimental tests with such a coal. In this case, the structure of coal particles is subjected to a slight variation during drying and devolatilization phase. Particle surface increasing is not enough to allow a complete reaction of char. Moreover, since the gasification reactions proceeds, the formation of a thin layer of ash can prevent the contact between the gasification agents and the core of the particle, which remains unreacted. Pure Sulcis coal is quite similar to Usibelli lignite in terms of volatiles and ash content, but its very high sulfur percentage does not suggest the use as single fuel in a commercial application: this justifies the choice to mix Sulcis coal with low sulfur South African coal. Finally, the performance reported for wood chips gasification comes from an extrapolation of the experimental results of a preliminary run: the process should be optimized in order to improve the syngas production and stabilize the operating conditions [29]. Raw syngas composition is strongly influenced by gasification parameters and fuel properties [30]. As a matter of facts, an increasing air injection typically involves a higher impact of the combustion process with respect to the gasification one: a higher amount of carbon reacts with oxygen, and CO2 concentration increases, whereas carbon monoxide content in raw syngas typically decreases. In parallel, an increasing steam injection typically shifts



Table 2 Typical gasification conditions and performance. S. African coal



80% S. Africa 20% sulcis



Operating parameters Fuel consumption (kg/h) 8.0 10.5 Air mass flow (kg/h) 36.8 41.2 Steam mass flow (kg/h) 3.7 7.8 Raw syngas composition (molar fractions, dry basis) CO 0.1807 0.1772 0.0947 0.0969 CO2 0.1889 0.2149 H2 0.5128 0.4780 N2 0.0151 0.0151 CH4 H2S 0.0003 0.0006 COS 0.0001 0.0001 0.0074 0.0172 O2 Raw syngas properties (dry basis) Mass flow (kg/h) 46.83 54.84 42.90 51.49 Volume flow (Nm3/h) LHV (MJ/kg) 4.50 4.83 Specific heat (kJ/kg K) 1.23 1.27 Outlet pressure (MPa) 0.14 0.14 Main gasifier performance 1034 1050 Maximum temp. ( C) Cold gas efficiency 96.93% 97.20% 3 5.36 4.90 Gasifier yield (Nm /kg)



Usibelli coal



Wood chips



24.0 57.6 3.7



12.0 11.3 0.0



0.2368 0.0771 0.1779 0.4729 0.0173 0.0002 0.0001 0.0176



0.2207 0.0797 0.3342 0.3418 0.0119 0.0000 0.0000 0.0117



79.67 72.90 5.14 1.23 0.14



23.31 25.48 7.49 1.47 0.14



993 96.13% 3.04



730 84.33% 2.12



Table 3 Coals classification. S. African coala



Sulcis coalb



Usibelli coalc



Macerals content (% by weight) Vitrinite



71.2



74.0



5.6



7.0



11.6



14.0



11.6



5.0



0.5e0.7%



0.3e0.6%



25.8 Inertite 60.0 Liptinite 4.2 Minerals 10.0 Vitrinite reflectance index 0.7% Notes. a Source: Ghetti et al., 2000 [32]. b Source: Ciccu et al., 2010 [33]. c Sources: Walsh et al., 1995 [34]; Hankinson, 1965 [35].



the equilibrium of the endothermic water-gas shift reaction toward the products; as a consequence, hydrogen concentration in raw syngas increases, with a subsequent reduction of CO content, whereas operating temperature typically decreases. Syngas composition is also conditioned by primary fuel properties. For example, in the reducing atmosphere typical of gasification processes, the sulfur contained in primary fuel leads the formation of hydrogen sulphide (H2S) and carbonyl sulphide (COS). In particular, raw syngas from South African coal presents a very low content of sulfur compounds: 274.3 ppm of H2S and 100.9 ppm of COS, both by volume. When a mixture of South African coal with 20% of Sulcis coal is used, H2S content significantly increases (634.4 ppm), whereas the raising in COS concentration is quite low (up to 126.3 ppm). When Usibelli coal is gasified, H2S and COS concentrations are very low (241.0 and 61.5 ppm, respectively) in raw syngas. Finally, sulfur compounds concentrations are negligible for wood chips gasification, due to the absence of sulfur in the considered biomass. 3.2. Fuel petrology In addition to fuel characterization in terms of syngas production and properties, a different approach can be used to predict the suitability of a fuel to be gasified in an atmospheric fixed-bed updraft reactor, such as the Sotacarbo pilot unit. This approach, based on the evaluation of the fuel reactivity, has been successfully assumed by Thimsen et al. [31] during an experimental campaign carried out in early 80s in a fixed-bed gasification plant characterized by similar operating conditions. Fuel reactivity can be determined through the specific gasification rate (SGR, expressed in kg/m2 h), defined as the amount of gasified fuel (kg/h) divided by the area (m2) of the horizontal section of the fuel bed. For fixed-bed gasifiers, SGR corresponds to the so-called grate loading (GL), defined as the amount of fuel gasified per square meter of grate area. According to these definitions, SGR evaluated for the four different feedstock reported in Table 2 amounts to 113, 148, 339 and 169 kg/m2 h, respectively. This confirms the higher reactivity of Usibelli lignite with respect to the other tested feedstock. In case of coal gasification, the reactivity can be theoretically correlated with the petrographic structure of coal itself, with specific reference to maceral content. In particular, Table 3 reports the maceral content of the considered coals, in terms of vitrinite, inertite, liptinite and minerals, together with the vitrinite reflectance index. Vitrinite is the most homogeneous maceral, which mainly contributes to the formation of the so-called cleats, sort of cracks of



7



A. Pettinau et al. / Applied Thermal Engineering 74 (2015) 2e9



the particle surface which increase the structure permeability and porosity, promoting char reactivity. On the contrary, inertite presents a dense and amorphous structure, and its high concentration in coal reduces fuel reactivity. Finally, a high liptinite content promote the production of gas hydrocarbons and tar and conditions coal LHV. Vitrinite reflectance index in oil can be used as an indicator of maturity in hydrocarbon source rocks and it conditions the coal rank. It can also be used as reference to predict the behavior of coal during gasification (low values of this index typically correspond to high reactivity of coal). 3.3. Syngas cleaning performance As reported above, fly ash and tar removal takes place in both wet scrubber and electrostatic precipitator, which assure very low outlet concentrations of both particulate and tar. When high sulfur coal are used to feed the gasification process, first cold gas desulphurization stage is usually fed with an aqueous solution of sodium hydroxide and the process automatically regulates solvent makeup in order to maintain a constant pH of the washing solutions. Typically, the system operates with a relatively low solvent pH, in order to allow a gross H2S removal without a sensible separation of carbon dioxide (to avoid an excessive alteration of CO2 concentration which could conditions the reliability of the experimental tests on water-gas shift and carbon capture sections). In this condition, an H2S removal efficiency up to 70% can be achieved, depending on both solvent pH and initial concentration in raw syngas. Cold-gas desulphurizaton in the packed column (second stage) is significantly more efficient. The experimental tests with both non-regenerative (aqueous solution of sodium hydroxide and hypochlorite) and regenerative solvents (methyldiethanolamine, MDEA) show a removal efficiency typically higher than 98e99%, with a final H2S concentration usually lower than 50e70 ppm (by volume). Differently from the cold-has desulphurization systems, which removes H2S and small amounts of CO2 but not COS, hot has desulphurization system is equipped, as mentioned, with a catalytic COS hydrogenator. It operates an almost complete (99.0e 99.5%) COS conversion [19], allowing to obtain, in clean syngas, a negligible concentration of carbonyl sulphide. Hydrogen sulphide is adsorbed by zinc oxides-based sorbents and the removal efficiency mainly depends on the adsorbers operating temperature. Overall, the hot gas desulphurization system allows to achieve a sulfur compounds (H2S and COS) concentration typically lower than 10 ppm (by volume) and frequently lower than 1 ppm [20]. Table 4 shows, for the four considered feedstock, the typical clean syngas properties and composition at the outlet of hot gas desulphurization (HGD) system, being the first cold gas desulphurization stage not operative due to the relatively low H2S concentrations in raw syngas. Obviously, hot gas desulphurization system is unessential when only wood chips are used as fuel in the gasification process. 3.4. Carbon dioxide removal and hydrogen production performance With reference to Fig. 3, downwards of hot gas desulphurization stage syngas is sent to the double stage water-gas shift system which, as mentioned, has been tested with both conventional (pyrophoric) and advanced (non pyrophoric) catalysts. Both these catalysts allow a CO conversion very close to the equilibrium values [36]. In particular, in order to minimize steam consumption, it is possible to reduce steam/CO molar ratio in the first WGS stage, in which inlet CO content is relatively high, and increase the same ratio in the second stage, when inlet CO content is low. According to



Table 4 Typical syngas composition downwards of HGD. S. African coal Clean syngas properties Mass flow (kg/h) 50.13 47.02 Volume flow (Nm3/h) LHV (MJ/kg) 4.20 Pressure (MPa) 0.140  400 Temperature ( C) Syngas composition (molar fractions) CO 0.1649 0.0864 CO2 0.1722 H2 0.4678 N2 0.0138 CH4 0.0000 H2S COS 0.0000 0.0068 O2 0.0881 H2O



80% S. Africa 20% sulcis



Usibelli coal



Wood chips



58.78 56.43 4.50 0.140 400



85.23 79.88 4.78 0.140 400



25.28 27.94 6.90 0.140 400



0.1618 0.0884 0.1960 0.4362 0.0138 0.0000 0.0000 0.0157 0.0882



0.2161 0.0703 0.1622 0.4314 0.0158 0.0000 0.0000 0.0161 0.0881



0.2013 0.0727 0.3048 0.3117 0.0109 0.0000 0.0000 0.0107 0.0881



this strategy, the fist WGS stage typically operates with a steam/CO molar ratio of 1.5e2.0, allowing a CO conversion efficiency between 83.8 and 89.3%; an overall CO conversion efficiency of 99.5% can be achieved operating the second WGS stage with a steam/CO molar ratio of about 10e12. A CO2 removal efficiency up to 90e92% can be achieved by feeding the bubbling reactors, disposed in series with respect to gas flow, with a liquid/gas ratio of about 3e4 kg of solvent per kilogram of gas. Finally, PSA allows to separate most of hydrogen contained in treated syngas, with a final purity up to 97% (by volume). 4. Potential future developments: integration with an MCFC system In addition to the main experimental results obtained on the Sotacarbo pilot plant, which represent the core of the work, this ancillary section presents one of the most interesting potential developments of the experimental equipment. With the aim to analyze innovative configurations for power generation, heat recovery and hydrogen production, a preliminary theoretical analysis has been performed to evaluate the possibility to apply the Sotacarbo coal and biomass gasification technology for high efficient power generation in a medium- and small-scale combined heat and power (CHP) industrial system [37] by a integration of the gasification process itself with a molten carbonate fuel cell (MCFC). 4.1. Methodology For the ancillary analysis here discussed, mass and thermal balances of the gasification and syngas treatment processes have been evaluated on the basis of the experimental data collected during gasification and hot gas desulphurization tests, with particular reference to the same four gasification conditions summarized in Table 2. On the other hand, the performance of the MCFC system was assessed by a simulation model, based on thermodynamic-electrochemical analysis at steady-state operating conditions [38e42]. The model was developed by Sotacarbo in close cooperation with the University Cagliari (Department of Mechanical, Chemical and Materials Engineering). A detailed description of the model is presented in a previous work [43]. 4.2. Selected plant configuration Most of the studies regarding the integration between gasification processes and fuel cell systems consider complex



8



A. Pettinau et al. / Applied Thermal Engineering 74 (2015) 2e9



Table 5 MCFC energy balance.



Pin (kW) Pel (kW) Hloss (kW) Hinc (kW) Hth (kW)



S. African coal



80% S. Africa 20% sulcis



Usibelli coal



Wood chips



58.5 19.9 2.9 22.0 13.6



73.6 23.5 3.7 27.7 18.8



113.2 36.7 5.7 40.6 30.2



48.5 15.5 2.4 19.9 10.7



CO2. These integrations, together with the effects of the very promising co-gasification of coal and biomass [45], are currently under evaluation. Even if preliminary, these results indicate the opportunity to experimentally investigate the potential integration of the Sotacarbo gasification technology with a fuel cell system, with the aim to develop an integrated CHP system for medium- and small-scale commercial applications. 5. Conclusions



configurations in which the power generation unit is fed with hydrogen produced by coal syngas. Due to their complexity, these configurations can be only justified, from the economic point of view, for large-scale commercial applications. But the high desulphurization efficiency observed in the Sotacarbo pilot plant suggests a different integration between gasification process and fuel cell system. In particular, MCFC can be directly fed with clean syngas, avoiding the complexities related to hydrogen production. This relatively easy configuration could be suitable for mediumand small-scale commercial applications. Raw syngas is assumed to be washed in the wet scrubber and fully sent to the catalytic COS hydrogenation system followed by the zinc oxides-based hot gas desulphurization process; the final syngas properties are those reported in Table 4. Finally, clean syngas is heated up to 650 � C and sent to the anode of the MCFC system, which operates at 0.137 MPa. The fuel cell cathode is fed with a gaseous stream composed by N2, CO2 and O2 (55, 30 and 15% by volume, respectively). 4.3. Fuel cell performance Table 5 reports the energy balance of the fuel cell system for the four considered gasification conditions. Sensible heat (Hth) released by the MCFC system has been evaluated by the following energy balance [44]:



Pin ¼ Pel þ Hloss þ Hinc þ Hth



(1)



where Pin is the clean syngas combustion potential (mass flow multiplied by the lower heating value), Pel is the released electric power, Hloss is the heat loss due to irradiation, assumed equal to 5% of inlet energy flow [44] and Hinc is the potential energy of combustion available in the outlet MCFC exhaust (all the terms are expressed in kW). In the pilot-scale hypothetical plant CHP configuration, in which the best performing Usibelli lignite feedstock is considered, the MCFC system can be fed with a syngas flow of about 80 Nm3/h, which gets a combustion potential (Pin) of 113.2 kW. In this conditions, an MCFC electrical efficiency (referred to the chemical power related to feeding syngas) of about 32.4% has been calculated, together with a potential cogeneration (heat and power) efficiency of 59.1%. Considering the integration between gasification, syngas treatment and power generation sections (but excluding from the balance the non-representative power consumption of the different sections of the pilot plant), the overall CHP configuration could present an electrical efficiency (referred to the chemical power related to gasified fuel) of about 31.5% and a cogeneration efficiency of 57.3% [29]. A further improvement of these performance can be achieved by the integration of the MCFC system with a burner to recover the residual energy content (Hinc) in the exhaust gas. In particular, MCFC anode spent gas can be recycled to the burner for the combustion of residual hydrogen, carbon monoxide and methane. At the same time, a partial flow of exhausts at the exit of burner can be recycled to supply MCFC inlet cathode with the necessary flow of



The about 1700 h of experimental tests carried out in the flexible Sotacarbo pilot platform allowed to optimize gasification process and syngas treatment sections in different operating conditions and with different kinds of fuels. The gained experience and the high plant flexibility allow Sotacarbo to use the platform to develop some research projects and to provide technical support to several Companies for the characterization of fuels, processes and materials. Among the different tested fuels, the high reactive lignite from Alaska presented, at now, the best gasification performance, with the production of about 73 Nm3/h of raw syngas, characterized by a lower heating value of 5.14 MJ/kg. This performance is mainly allowed by the high volatile and moisture content of the Usibelli lignite, which promotes a quick and efficient conversion of the primary fuel. As expected, the gasification performance decreases when less reactive feedstock is used, in particular when only South African coal is fed to the reactor. Moreover, due to its low energy density, the gasification of wood chips does not allow a high syngas production. The high efficient hot gas cleaning system, which operates COS hydrogenation and H2S adsorption by using zinc oxides-based sorbents, allows a very clean syngas production, with a final concentration of sulfur compounds frequently lower than 1 ppm by volume. This performance suggested to analyze, by using a properly developed simulation model, the possibility to integrate the Sotacarbo pilot plant with a MCFC system, directly feed with clean syngas, to obtain an innovative CHP plant configuration. The analysis shows that the air-blown gasification of about 24 kg/h of Usibelli could allow the production of 36.7 kW of electrical energy and 30.2 kW of sensible heat. Therefore, whereas the single MCFC system presents an electrical efficiency of 32.4% and a cogeneration efficiency of 59.1% (both referred to the chemical power of the feeding clean syngas), the overall CHP configuration (including gasification, syngas cleaning and MCFC-based power generation) could present an electrical efficiency of about 31.5% and a cogeneration efficiency of 57.3%. Acknowledgements Most of the experimental data reported in this paper come from specific tests carried out in close cooperation with ENEA (the Italian National Agency for New Technologies, Energy and Sustainable Economic Development) and funded by the Italian National Regulatory Authority for Electricity and Gas (AEEG) within the “Electrical System Research” project. Nomenclature Acronyms CHP combined heat and power ESP electrostatic precipitator GR grate loading (kg/m2 h) HGD hot gas desulphurization HT high temperature



A. Pettinau et al. / Applied Thermal Engineering 74 (2015) 2e9



ICE LHV LT MCFC MDEA MEA PSA SGR WGS



internal combustion engine lower heating value (MJ/kg) low temperature molten carbonate fuel cell methyldiethanolamine (C5H13NO2) monoethanolamine (C2H7NO) pressure swing adsorption specific gasification rate (kg/m2 h) water-gas shift conversion



Symbols theoretically achievable maximum reversible potential E� (V) F Faraday’s constant (96,485 C/mol) potential energy of combustion (kW) Hinc heat loss due to irradiation (kW) Hloss sensible heat released by the MCFC (kW) Hth i operating electric current (A) j current density (A/cm2) released electric power (kW) Pel clean syngas combustion potential (kW) Pin hydrogen consumption rate (mol/s) qH2 total irreversible losses (U$cm2) Rtot operating cell potential (V) Vcell z number of electrons (dimensionless) hNernst Nernst loss (V) References [1] M. Lucquiaud, J. Gibbins, On the integration of CO2 capture with coal-fired power plants: a methodology to assess and optimise solvent-based postcombustion capture systems, Chem. Eng. Res. Des. 89 (2011) 1553e1571. [2] T. Koljonen, M. Flyktman, A. Lehtilä, K. Pahkala, E. Peltola, I. Savolainen, The role of CCS and renewables in tackling climate change, Energy Procedia 1 (2009) 4323e4330. [3] R. Kothari, V.V. Tyagi, A. Pathak, Waste-to-energy: a way from renewable energy sources to sustainable development, Renew. Sust. Energy Rev. 14 (2010) 3164e3170. [4] M. Balat, Potential importance of hydrogen as future solution to environmental and transportation problems, Int. J. Hydrogen Energy 33 (2008) 4013e 4029. [5] C.C. Cormos, F. Starr, E. Tzimas, S. Peteves, Innovative concepts for hydrogen production processes based on coal gasification with CO2 capture, Int. J. Hydrogen Energy 33 (2008) 1286e1294. [6] A. Perna, Combined power and hydrogen production from coal. Part A e analysis of IGHP plants, Int. J. Hydrogen Energy 33 (2008) 2957e2964. [7] S.P. Cicconardi, A. Perna, G. Spazzafumo, Combined power and hydrogen production from coal. Part B: comparison between the IGHP and CPH systems, Int. J. Hydrogen Energy 33 (2008) 4397e4404. [8] E. Martelli, T. Kreutz, M. Carbo, S. Consonni, D. Jansen, Shell coal IGCC with carbon capture: conventional gas quench vs. innovative configurations, Appl. Energy 88 (2011) 3978e3989. [9] C. Kunze, H. Spliethoff, Assessment of oxy-fuel, pre- and post-combustionbased carbon capture for future IGCC plants, Appl. Energy 94 (2012) 109e116. [10] A. Giuffrida, M.C. Romano, G. Lozza, Thermodynamic analysis of air-blown gasification for IGCC applications, Appl. Energy 88 (2011) 3949e3958. [11] B.S. Hoffmann, A. Szklo, Integrated gasification combined cycle and carbon capture: a risky option to mitigate CO2 emissions of coal-fired power plants, Appl. Energy 88 (2011) 3917e3929. [12] L. Wang, C.L. Weller, D.D. Jones, M.A. Hanna, Contemporary issues in thermal gasification of biomass and its application to electricity and fuel production, Biomass Bioenergy 32 (2008) 573e581. [13] P. Basu, Combustion and Gasification in Fluidized, CRC Press, Boca Raton, Florida (USA), 2006. [14] E. Supp, How to Produce Methanol from Coal, Springer-Verlag, Berlin, Germany, 1990. [15] M.L. Hobbs, P.T. Radulovic, L.D. Smoot, Modeling fixed-bed coal gasifier, AIChE J. 38 (1992) 681e702. [16] O.H. Hahn, P.D. Wesley, B.A. Swisshelm, S. Maples, J. Withrow, A mass and energy balance of a Wellman-Galusha gasifier, Fuel Process. Technol. 2 (1979) 332e334. [17] A. Pettinau, F. Ferrara, C. Amorino, An overview of current and future experimental activities in a flexible gasification pilot plant, in: M.D. Baker (Ed.), Gasification: Chemistry, Processes and Applications, NOVA Science Publishers, New York (USA), 2011, pp. 55e100.



9



[18] L. Chen, J. Huang, C. Yang, Absorption of H2S in NaOCl caustic aqueous solution, Environ. Prog. 20 (2001) 175e181. [19] X. Yao, Y. Li, Density functional theory study on the hydrodesulfurization reactions of COS and CS2 with Mo3S9 model catalyst, J. Mol. Struct. (THEOCHEM 2009) 899 (special. issue) (2009) 32e41. [20] K. Thambimuthu, Hot Gas Clean-up of Sulphur, Nitrogen, Minor and Trace Elements, IEA Coal Research, London, United Kingdom, 1998. Report No. IEACR/12. [21] J.M. Sánchez-Hervás, J. Otelo, E. Ruiz, A study on sulphidation and regeneration of Z-sorb III sorbent for H2S removal from simulated ELCOGAS IGCC syngas, Chem. Eng. Sci. 60 (2005) 2977e2989. [22] L. Li, D.L. King, H2S removal with ZnO during fuel processing for PEM fuel cell applications, Catal. Today 116 (2006) 537e541. [23] J. Sun, S. Modi, K. Liu, R. Lesieu, J. Buglass, Kinetics of zinc oxide sulfidation for packed-bed desulfurizer modelling, Energy. Fuels 21 (2007) 1863e1871. [24] I. Rosso, C. Galletti, M. Bizzi, G. Saracco, V. Specchia, Zinc oxide sorbents for the removal of hydrogen sulphide from syngas, Ind. Eng. Chem. Res. 42 (2003) 1688e1697. [25] S.S. Tamhankar, M. Bagajewicz, G.R. Gavalas, P.K. Sharma, M. Flytzani-Stephanopoulos, Mixed oxide sorbents for high-temperature removal of hydrogen sulphide, Ind. Eng. Chem. Process Design Dev. 25 (1986) 429e437. [26] P. Chiesa, S. Consonni, T. Kreutz, R. Williams, Co-production of hydrogen, electricity and CO2 from coal with commercially ready technology. Part A: performance and emissions, Int. J. Hydrogen Energy 30 (2005) 747e767. [27] A. Pettinau, C. Frau, F. Ferrara, Performance assessment of a fixed-bed gasification pilot plant for combined power generation and hydrogen production, Fuel Process. Technol. 92 (2011) 1946e1953. [28] A. Pettinau, A. Orsini, G. Calì, F. Ferrara, The Sotacarbo coal gasification experimental plant for a CO2-free hydrogen production, Int. J. Hydrogen Energy 35 (2010) 9836e9844. [29] F. Ferrara, A. Pettinau, S. Palmas, C. Amorino, M. Porcu, Integration of coal and biomass gasification process with fuel cell system for small scale industrial CHP applications, in: Proceedings of the 29th Annual International Pittsburgh Coal Conference, Pittsburgh, Pennsylvania (USA), October 15-18, 2012. [30] M.H. Kim, H.K. Park, G.Y. Chung, H.C. Lim, S.W. Nam, T.H. Lim, S.A. Hong, Effects of water gas shift reaction on simulated performance of a molten carbonate fuel cell, J. Power Sources 103 (2002) 245e252. [31] D. Thimsen, R.E. Maurer, B.Y.H. Liu, D. Pui, D. Kittelson, Executive summary, Fixed-bed Gasification Research using U.S. Coals, Vol. 19, United States Department of Interior, Bureau of Mines, Houston, Texas (USA), 1985. Available at: http://www.osti.gov. [32] P. Ghetti, C. La Marca, J. Riccardi, Scelta e caratterizzazione in laboratorio di miscele ottimali di fanghi e carbone, Report ENELP/RIC/RT-2000/046/ 00þRT.RIC.PI, ENEL Ricerca, Pisa, Italy, 2000. Available at: http://doc.rse-web. it (Italian). [33] R. Ciccu, A. Mazzella, C. Tilocca, Caratterizzazione e prova di miscele di acqua e carbone e sviluppo di un sistema informativo geografico a supporto della scelta di siti idonei allo stoccaggio di CO2, University of Cagliari, Cagliari, Italy, 2010. Available at: http://www.enea.it (Italian). [34] D.E. Walsh, P.D. Rao, O. Ogunsola, H.K. Lin, Hydrothermally Treated Coals for Pulverized Coal Injection, Final technical report, Mineral Industry Research Laboratory, University of Alaska, Fairbanks, Alaska (USA), 1995. Available at: http://www.osti.gov. [35] F.C. Hankinson, Petrographic Evaluation of Coking Potential of Selected Alaskan Coals and Blends, University of Alaska, Fairbanks, Alaska (USA), 1965. [36] G. Cau, G. Mura, S. Murgia, Experimental investigation and numerical simulation of CO to CO2 conversion and hydrogen production from the water gas shift reaction, in: Proceedings of the International Conference HYPOTHESIS VIII, Lisbon, Portugal, April 1e3, 2009. [37] J. Ahrenfeldt, T.P. Thomson, U. Henriksen, L.R. Clausen, Biomass gasification cogeneration e a review of state of the art technology and near future perspective, Appl. Therm. Eng. 50 (2013) 1407e1417. [38] H.K. Park, Y.R. Lee, M.H. Kim, G.Y. Chung, S.W. Nam, S.A. Hong, T.H. Lim, H.C. Lim, Studies of the effects of the reformer in an internal-reforming molten carbonate fuel cell by mathematical modelling, J. Power Sources 104 (2002) 140e147. [39] M. Baranak, H. Atakül, A basic model for analysis of molten carbonate fuel cell behavior, J. Power Sources 172 (2007) 831e839. [40] EG&G Technical Services, Inc, Fuel Cell Handbook, Report prepared for U. S, seventh ed., Department of Energy, National Energy Technology Laboratory, Morgantown (West Virginia, USA), 2004. [41] H. Morita, M. Komoda, Y. Mugikura, Y. Izaki, T. Watanabe, Y. Masuda, T. Matsuyama, Performance analysis of molten carbonate fuel cell using a Li/ Na electrolyte, J. Power Sources 112 (2002) 509e518. [42] C.Y. Yuh, J.R. Selman, The polarization of molten carbonate fuel cell electrodes I. Analysis of steady-state polarization data, J. Electrochem. Soc. 138 (1991) 3642e3648. [43] F. Ferrara, S. Palmas, A. Pettinau, Power generation through integrated coal gasification and fuel cell for medium and small scale industrial applications, in: Proceedings of the 5th International Conference on Clean Coal Technologies, Zaragoza (Spain), May 8e12, 2011. [44] F. Urbani, S. Freni, A. Galvagno, V. Chiodo, MCFC integrated system in a biodiesel production process, J. Power Sources 196 (2011) 2691e2698. [45] L.Q. Wang, Z.S. Chen, Gas generation by co-gasification of biomass and coal in an autothermal fluidized bed gasifier, Appl. Therm. Eng. 59 (2013) 278e282.



Applied Thermal Engineering 74 (2015) 10e19



Contents lists available at ScienceDirect



Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng



Techno-economic analysis of a CO2 capture plant integrated with a commercial scale combined cycle gas turbine (CCGT) power plant Roberto Canepa a, *, Meihong Wang b a b



DIME sez.MASET, University of Genova, Via Montallegro 1, 16145 Genova, Italy School of Engineering, University of Hull, Hull HU6 7RX, UK



h i g h l i g h t s  A post-combustion CO2 capture plant model has been developed and has been extensively validated at pilot scale.  Capture plant model has been scaled up to meet the requirement of a commercial 427 MW CCGT power plant.  Sensitivity analysis has been conducted to reduce thermal energy requirement to 4.1 GJ/tonne CO2.  Levelized cost of the electricity from CCGT power plant is increased by 47% when CO2 capure process is added.



a r t i c l e i n f o



a b s t r a c t



Article history: Received 31 August 2013 Accepted 11 January 2014 Available online 22 January 2014



In this study, a combined cycle gas turbine (CCGT) power plant and a CO2 capture plant have been modelled in GateCycle and in Aspen Plus environments respectively. The capture plant model is validated with experimental data from the pilot plant at the University of Texas at Austin and then has been scaled up to meet the requirement of the 427 MWe CCGT power plant. A techno-economical evaluation study has been performed with the capture plant model integrated with flue gas preprocessing and CO2 compression sections. Sensitivity analysis was carried out to assess capture plant response to changes in key operating parameters and equipment design. The study indicates which parameters are the most relevant (namely absorber packing height and regenerator operating pressure) and how, with a proper choice of the operating conditions, both the energy requirement for solvent regeneration and the cost of electricity may be reduced.  2014 Elsevier Ltd. All rights reserved.



Keywords: Process modelling Process simulation Gas turbine Combined cycle CCGT Post-combustion Carbon capture Techno-economic analysis



1. Introduction 1.1. Background and motivations Carbon Capture and Storage (CCS) is regarded as an essential technology to meet greenhouse gases reduction goals [1]. CO2 capture with chemical absorption using amine solvent is a proven and well established technology. Despite this, CO2 capture from exhaust gas coming from a power plant poses many technical and economical challenges. Current CO2 capture projects involve pilot plants on a scale much smaller than required to capture CO2 from a commercially available power plant. In September 2012, Global CCS Institute has identified 75 large-scale integrated CCS projects (LSIP)



* Corresponding author. E-mail addresses: [email protected] (R. Canepa), [email protected] (M. Wang). 1359-4311/$ e see front matter  2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.applthermaleng.2014.01.014



running globally [2]. An LSIP is defined by Global CCS institute as a project involving the capture, transport and storage of CO2 at a scale of at least 800,000 tonnes of CO2 annually for coal-based power plants or at least 400,000 tonnes of CO2 annually for other emission-intensive industrial facilities (including natural gas-based power generation). More than half of all projects started during 2012 are located in China, and all of these are investigating enhanced oil recovery (EOR) options as an additional source of revenue. Among these LSIPs only 16 are, however, currently operating or in construction, for a global capture capacity of around 36 million tonnes per annum. These projects require investments of the order of dozens millions of Euros. It is expected that a full scale demonstration project for CO2 capture would require over a billion dollars [3]. Accurate modelling of CO2 capture plant, for the insight it can provide, is therefore a necessary intermediate step towards demonstrating full scale CO2 capture. Both technical performance and costs are determinant factors to select optimal operating conditions.



R. Canepa, M. Wang / Applied Thermal Engineering 74 (2015) 10e19



1.2. Novel contributions of this paper CO2 capture process is, with current available technology, a very expensive and energy-intensive process. Despite this, it is gaining attention among researchers and policymakers as a short-mid term solution to contain carbon emissions from existing or yet to be built fossil fuelled power plant. However, as usual for a substantially new technology (at least at the scale required for capturing CO2 from power plants), much resistance remains, mainly due to the uncertainty connected to actual performance and costs. Therefore, accurate modelling constitutes a stepping-stone to increase confidence about CO2 capture process. In this perspective, rate-based modelling procedure adopted by the Authors constitute, when compared to equilibrium based calculations, a superior solution in terms of accuracy and sensitivity to changes in the operating parameters. In addition to this pilot plants currently existing or, even more, large scale demonstration projects currently being built, are limited in the range of parameters that can be changed. Capture plant modelling, if based on a rigorous and trusted modelling procedure, can overcome this intrinsic limitation and following this idea a wide sensitivity analysis has been conducted on the main operating and design parameters in order to identify optimal working conditions, thus reducing uncertainty in thermal and economics characteristics of the process. Furthermore, combining capture plant commercial scale modelling with an extensive validation campaign (over a wide range of L/G ratio and process conditions) constitute an emblematic element of novelty. Summarizing, the main novelties of this article are:



11



therefore regarded as the best capture option for existing power plants. Enough space should be provided for flue gas pipeline and capture related sections (notably flue gas pre-processing, CO2 capture and compression) which should be located in the vicinity of the power plant. The main connections between power plant and the capture plant are as follows: a. Flue gas pre-processing; b. Steam draw-off from the steam turbine in CCGT power plant to feed the reboiler of the regenerator in the CO2 capture plant; c. Condensate return from capture plant to the power plant. The first two processes result in a reduction of electricity output from the CCGT power plant. 3.1. Flue gas pre-processing



a. Extensive validation campaign of capture plant at pilot plant scale combined with commercial scale modelling and simulation b. Capture plant operating conditions and design parameters sensitivity analysis



Exhaust gases coming from the HRSG, before being sent to the capture plant, need to be cooled down to 40e50  C in order to improve absorption and reduce solvent losses due to evaporation [4]. The cooling system consists of a direct contact cooler (DCC) in which a spray of water cools down flue gases to the desired temperature level. This process has been modelled in Aspen Plus environment by using RadFrac block for the DCC, regarded as a two theoretical stages tower with Rashig rings packing. Flue gases are cooled down to 40  C by direct contact with a spray of water at 25  C. During the cooling process water is recovered from the flue gas because of condensation. Finally, a blower increases the pressure of the cooled flue gases to a pressure above the atmospheric level, to balance the pressure losses in the capture plant. In Fig. 1, the entire Aspen Plus flowsheet for flue gas pre-processing is presented. Assuming a blower isentropic efficiency equal to 88.5%, compression power requirement has been found to be equal to 8896 kW.



2. Modelling of CCGT power plant



3.2. Steam draw-off



A commercial Combined Cycle Gas Turbine (CCGT) power plant, targeted to 427 MWe production (before capture) is modelled in GE’s GateCycle software. GateCycle software allows an accurate modelling of design but also off-design power plant components operation. The performance of the steam cycle sections, sized for the reference non-capture case, is automatically scaled to take into account the modified pressure and temperature they will face after retrofitting to capture CO2. The reference commercial CCGT power plant employs a heavy duty single shaft Ansaldo Energia AE94.3A gas turbine from which exhaust gas is led to an unfired heat recovery steam generator (HRSG). The steam cycle consists of three pressure levels (124, 28 and 4.5 bar respectively) with a reheat loop. The steam is condensed in a condenser with outer water at 15  C. Deaeration is attained in the deaerator, which operates at 4.5 bar, by using low pressure steam. The condensate from the condenser is heated by means of a closed cycle loop in order to increase heat utilization from flue gas as much as possible. All the parameters required for the calculation comes from various sources: Ansaldo company private communications, GateCycle software library and common practice for large combined cycle power plants.



The steam required by solvent regeneration in the reboiler is provided by means of a steam bled from the IP/LP crossover. As a result, the LP steam turbine will see a major reduction of steam flow rate, which will result in the reduction of both its efficiency and power output. A throttled pressure configuration is used in this study. Given the reduced mass flow rate going through the LP steam turbine, its inlet pressure would drop. To guarantee a sufficient temperature (and thus pressure) for extraction, a valve has been added at IP/LP crossover. This adds pressure throttling losses to the efficiency penalty connected to reduced LP steam turbine mass flow rate and efficiency. To avoid solvent degradation due to high temperature, the steam has to be cooled down to a temperature just above saturation with a water spray. The waste heat resulting from this process has been partially recovered by combining the steam with some of the condensate coming from the reboiler. In this way steam draw-off is also reduced. The remaining condensate is then returned to the condenser.



3. Integration between CCGT power plant and capture plant An exhaust gas with mass flow rate of 702 kg/s is delivered to flue gas pre-processing and consequently capture sections. Applying post-combustion CO2 capture to a CCGT power plant requires minimal structural changes to the original cycle and is



4. Modelling of CO2 transportation and compression At ambient condition, CO2 is a gas. At a temperature between 56.5 and 31.1  C, it may be turned into a liquid by compressing it up to the corresponding liquefaction pressure. The critical point occurs at 73.825 bar and 31.4  C. Above this critical pressure (and at temperatures higher than 60  C), only supercritical or dense-phase liquid conditions exist. If the temperature and pressure are both above the critical point, supercritical conditions



12



R. Canepa, M. Wang / Applied Thermal Engineering 74 (2015) 10e19



Fig. 1. Aspen Plus flowsheet for flue gas pre-processing section.



are attained: CO2 no longer exists in distinct gaseous and liquid phases, but as a dense-phase with the density of a liquid but the viscosity of a gas. At pressures above, but temperatures below critical conditions, CO2 is a dense liquid, whose density increases with decreasing temperature. Captured CO2 has to be transported to a suitable storage site: pipeline is the most economical method of transport in the context of CCS (Refs. [5,6]). To allow an efficient transportation, CO2 flow coming from capture plants has to be compressed above critical pressure. Indeed, managing of a twophase system might be technically very hard to achieve and, moreover, a greater density allows a smaller, and thus cheaper, pipeline. A very important requirement is that pressure, all along the pipeline, should never drop below critical value. Thus CO2 pipeline inlet pressure needs to be chosen according to a proper pressure drop estimation along the pipeline. To cover great distances, as it likely might be required for many power plants, an additional intermediate boosting station should be provided. In literature [7] a lower limit pressure of 80 bar is usually suggested to guarantee a certain margin. Pressure loss depends on many factors, such as the pipeline diameter and length, the roughness of the pipe and CO2 flow velocity. It might be calculated using the DarcyeWeisbach equation: 2



Dp ¼ 10�5 f



vavg L r Di avg 2



(1)



Assuming a capacity factor, that is the ratio of nameplate power functioning over the year, equal to 75% and 8766 h per year (averaging over leap years), a sizing requirement of 0.918 Mtonne/ yr for the pipeline has been found. CO2 density is very dependent on pressure and temperature. Considering an average pipeline pressure equal to 95 bar and a ground temperature of 10 � C, CO2 density is equal to 810.8 kg/m3. Carbon steel pipes can be adopted thanks to the high degree of control over the water content of the CO2 being transported. The pipe has been chosen from existing tabled pipes [8]: a 10 inches (0.254 m) pipe with a 0.307 inches (7.80 mm) thickness has been adopted. This gives an internal diameter equal to 0.238 m and a velocity of 1.07 m/s. With this assumptions, pressure loss over the



total assumed pipeline length (100 km), is calculated equal to 17.0 bar. An inlet pressure equal to 110 bar gives a final discharge pressure of 93 bar and is thus sufficient to guarantee desired transport condition all along pipeline. If a greater distance had to be covered, pumping stations should be provided to raise CO2 pressure as needed. The compression to 110 bar is achieved with a four stage intercooled centrifugal compressor modelled with Aspen Plus. Water is removed during cooling process. PengeRobinson equation of state is used as a thermodynamic base model. In Table 1 compressor assumptions and performance are given. 5. Modelling and simulation of capture plant 5.1. Methodology Capture plant section model has been developed in Aspen Plus starting from Ref. [9], to which it can be referred for more details. Absorber and regenerator columns have been modelled using a rate-based approach, which ensures higher reliability over equilibrium based one [10]. Electrolyte NRTL activity coefficient model is used to account for liquid phase non-ideality, while the Redliche Kwong equation of state is employed for vapour. Both vapoure liquid equilibrium (VLE) and kinetic reactions are accounted for in the columns. The following set of rate-controlled reactions has Table 1 CO2 compression performance. Parameter



Value



CO2 mass flow rate [kg/s] Outlet pressure [bar] Intercooling temperature [� C] Polytropic efficiency (stage 1) [%] Polytropic efficiency (stage 2) [%] Polytropic efficiency (stage 3) [%] Polytropic efficiency (stage 4) [%] Water recovery [kg/s] Compressor power [kW] Compressor power [kJ/kgCO2]



38.77 110 35 85.5 85 84 78 0.203 11,225 289.51



R. Canepa, M. Wang / Applied Thermal Engineering 74 (2015) 10e19 Table 2 Kinetic parameter for rate-controlled reactions. Reaction



K [e]



E [cal/mol]



Source



2 3 4 5



4.32Eþ13 2.38Eþ17 9.77Eþ10 3.23Eþ19



13,249 29,451 9855.8 15,655



Pinsent et al. [12] Aspentech [13] Hikita et al. [14] Hikita et al. [14]



13



An extensive validation campaign has been conducted on capture section. For this purpose, performance data from Separation Research Program (SRP) at the University of Texas at Austin pilot plant have been employed [11]. Then, capture plant has been scaled-up to meet the requirement of the commercial scale CCGT power plant. Aspen Plus flowsheet for pilot plant is given in Fig. 2. 5.2. Model validation



been defined to represent monoethanolamine (MEA) reaction with CO2:



CO2 þ OH� /HCO3 �



(2)



HCO3 � /CO2 þ OH�



(3)



MEA þ CO2 þ H2 O/MEACOO� þ H3 Oþ



(4)



MEACOO� þ H3 Oþ /MEA þ CO2 þ H2 O



(5)



They are governed by power law expressions which kinetic coefficients are given in Table 2. On the other hand, equilibrium constants for equilibrium reactions are calculated from the standard Gibbs free energy change. To reduce the solvent regeneration heat requirement, a cross heat exchanger is used to pre-heat the rich solvent stream entering the regenerator column by using cascade heat from the hot lean solvent stream coming from the regenerator itself. This would constitute a closed loop within the Aspen Plus flowsheet and therefore would require, rigorously, to provide a tear stream for the cross heat exchanger. This has been avoided by using two different heat exchanger for the hot (lean) and the cold (rich) side of the actual heat exchanger respectively. These heat exchangers are then linked together by a heat stream, to ensure the same heat duty on the two sides. Being absorption process favoured by low temperature, a cooler is needed to further decrease lean solvent temperature. Solvent and water makeup are required to close the loop due to losses in vapour streams leaving both the absorber and regenerator columns.



The pilot plant is a closed loop absorption and stripping (regeneration) facility for CO2 removal from flue gas with 32.5 wt% aqueous MEA solution [11]. Two different kinds of packing have been adopted for the columns: Flexipac 1Y, a structured packing with a specific area of 420 m2/m3 and IMTP no. 40, a random metal packing with a specific area of 145 m2/m3. Out of the 48 experimental cases carried out in the test campaign, 12 cases were chosen for validation to account for different liquid solvent to gas (L/G) molar flows ratios. Table 3 shows the process conditions for the considered cases. With reference to it, solvent loading is defined as the (molar) ratio of CO2 to MEA. Therefore the lean loading is the loading of the (stripped) solvent stream entering the top of the absorber column and the rich loading is the one of the solvent (in which CO2 has been captured) coming from the bottom of the absorber column. In Table 4, the overall performance of the CO2 capture plant model is reported. With reference to it, simulation results are compared with the experimental results and with those obtained by Zhang et al. [10] from their Aspen Plus model. Lean loading has been controlled by a design specification set in the regenerator column, and is not therefore a validation parameter. All CO2 loadings available from pilot plant test campaign have been obtained by means of an empirical equation resulting in a 10% uncertainty level. Taking this in consideration, all the rich loading predictions of the model can be considered satisfactory. Cases 47 and 48 have the largest deviations among the selected cases, with an underestimation of the rich loading when compared with the experimental results which is however within the uncertainty range. Interestingly, this is similarly found by Zhang et al. CO2 capture level is always lower than the experimental results, which



Fig. 2. Aspen Plus flowsheet for capture pilot plant.



14



R. Canepa, M. Wang / Applied Thermal Engineering 74 (2015) 10e19



Table 3 Process conditions for cases selected for validation. Case Lean solvent loading [mol/mol] Lean solvent flow rate [l/min] Lean solvent temperature [K] Flue gas flow rate [m3/min] Flue gas temperature [K] Flue gas pressure [kPa] Flue gas CO2 content [mol%] Regenerator pressure [kPa] Regenerator pressure drop [kPa] Regenerator feed temperature [K] Condenser temperature [K] Reboiler temperature [K] L/G [mol/mol]



28



29



30



31



32



39



41



42



43



44



47



48



0.287 81.92 313.1 11 321.1 105.2 16.54 162.1 1.48 345.2 287.8 388.1 7.4



0.285 54.93 312.9 11.02 324.1 105.5 16.18 162 0.55 348.8 286.2 387.3 5.0



0.284 54.89 313.2 10.98 324.7 105.6 17.08 162 0.6 348.6 286 387.4 5.0



0.281 40.58 313.57 5.52 322.4 103.63 17.41 162.03 0.02 358.16 286.02 386.86 7.5



0.279 40.73 313.71 5.48 319.71 103.49 17.66 162.03 0.02 358.67 286.17 386.68 7.4



0.228 83.24 313.2 11.02 328.4 105 16.89 162 1.48 346.7 288.2 389.8 7.9



0.235 56.71 313.3 10.98 325.5 104.8 17.11 162 0.65 350.6 287.6 389.3 5.3



0.232 56.86 313.5 10.96 325.8 104.5 16.98 162 0.67 351.2 287.3 389.3 5.4



0.231 39.33 313.26 10.99 326.66 104.66 16.96 162.03 0.1 353.07 285.3 389.3 3.7



0.231 39.44 313.3 11.01 324.7 104.7 16.8 162 0.1 353 285.3 389.3 3.7



0.281 30.13 313.3 8.22 332.4 103.4 18.41 68.95 0.37 354.3 297.1 366.3 4.6



0.285 30.13 313.4 8.22 329.1 102.9 17.63 68.95 0.41 353.5 297.7 366.4 4.6



is generally observed also in the model by Zhang et al. Only for cases 43 and 44, Zhang et al. reported an overestimation of the CO2 removal rate. Reboiler duty is slightly underestimated by the Aspen Plus rate based model, but this is in accordance with CO2 capture rate estimation results. Zhang et al. (2009) studied stand-alone absorber performance only, so no reboiler duty estimation is given by them. Temperature profile estimation is probably the most important validation parameter. From temperature depends the kinetics of absorption, equilibrium phase, and fluid transport properties. The balance between the heat released from CO2 and MEA reaction and the heat consumed in the process from various sources (like CO2 stripping, water evaporation, heating of the streams and heat losses) produces a bulge in absorber temperature profile [15]. The position of this bulge is mainly connected to L/G ratio. When the liquid to gas ratio is low the bulge will be located near the top of the absorber and, conversely, when it is high it will be located near the bottom. In the case of CO2 capture for CCGT power plant, given the low CO2 concentration in flue gases, the needed L/G ratio will likely be low. In Ref. [10] three types of temperature profiles are identified. Type A is a result of low L/G ratios and is represented, in the current analysis, by Cases 43, 44, 47 and 48. Type B profile, resulting from medium L/G ratios, is represented by Cases 29, 30, 41 and 42. Finally, in the case of high L/G ratios (Cases 28, 31, 32 and 39), type C temperature profile is obtained. In Figs. 3e5 absorbers and regenerators temperature profiles provided by the rate-based model are compared to experimental results for, respectively, Case 48 (low L/G ratio), Case 42 (medium L/G) and Case 39 (high L/G). An



Table 4 Model validation results. Case



Rich loading [mol/mol] Exp.



28 29 30 31 32 39 41 42 43 44 47 48



0.412 0.448 0.453 0.426 0.428 0.367 0.433 0.43 0.491 0.492 0.539 0.537



CO2 capture level [%]



Model



Zhang et al.



Exp.



Model



Zhang et al.



0.412 0.449 0.455 0.437 0.438 0.362 0.423 0.419 0.467 0.466 0.482 0.480



0.405 0.44 0.452 0.431 0.432 0.355 0.409 0.413 0.453 0.45 0.48 0.481



86 70 70 95 95 94 87 87 72 72 69 69



76 69 68 91 91 83 78 79 69 69 68 69



74 70 64 90 90 86 82 80 76 76 68 65



Reboiler duty [MJ/hr] Exp. 1317 918 918 564 548 1497 1087 1087 756 754 738 775



Model 1099 766 778 517 522 1366 908 921 655 656 701 662



excellent match between calculation results and experimental data has been obtained and, consequently, the Aspen Plus model reliability is thoroughly proven. 5.3. Scale-up and sensitivity analysis Aspen Plus is able to size packed column diameters based on the desired approach to flooding on a specified stage, starting from a (user specified) fist-guess diameter. First-guess needed solution has been obtained using the procedure described by Lawal et al. [16]. This has provided the required number of columns and their (first-guess) diameters. The obtained results are presented in Fig. 6, in which absorber and regenerator diameters are represented as a function of the columns number. Due to structural limitations, columns diameter should not exceed 12.2 m (i.e. 40 feet) (Refs. [16,17]). According to this limitation, a three-column absorber and one-column regenerator configuration was selected. A greater number of absorber would require larger capital costs and footprint without any major benefit. On the other hand one regenerator column is sufficient to strip all the rich solvent flow coming from the absorbers. In order to lower the computational time, only one absorber column has been modelled. Assuming the same performance for all the absorber columns the output streams from the column (the vented stream and the rich solvent flow) have been opportunely multiplied to take into account the actual number of columns. In this way regenerator performance can be taken in proper account and the same happens for makeup calculation. As a result of scale-up procedure columns number and (firstguess) diameter, as well as a reasonable solvent flow rate (and thus the corresponding L/G ratio) have been obtained. However, to model commercial scale capture plant many other parameters are needed. Various design specifications or Calculator block have been assigned in order to ensure good capture plant performance: a. Lean solvent loading and temperature are user input; b. Lean solvent flow rate (and thus L/G ratio) is evaluated in order to obtain, for the actual operating conditions the desired (90%) capture rate. It will mainly depends on user input lean loading. Scale-up procedure result is used to provide the needed firstguess solution; c. Reboiler duty is defined in order to obtain the user input lean loading at regenerator outlet. This ensures that the solvent coming from the regenerator has, taking into account solvent and water losses, the same loading than the one entering the absorber;



R. Canepa, M. Wang / Applied Thermal Engineering 74 (2015) 10e19



15



Fig. 3. Temperature profile for Case 48 (low L/G ratio): (a) absorber and (b) regenerator.



d. Cross heat exchanger outlet temperature is calculated to ensure user input heat exchanger approach temperature. An higher approach temperature will correspond to a lower outlet temperature of the cold stream from the heat exchanger. As a consequence, for a given lean loading target, an higher reboiler duty will be needed; e. Lean solvent cooler heat duty is evaluated in order to ensure, at absorber inlet, the user input lean solvent temperature; f. Columns diameters are sized to obtain the desired flooding percentage. However, to allow model convergence, a first-guess reasonable diameter has to be given. The other sizing-relevant parameter, the packing heights, are user input; g. Columns pressure is user input. While absorber usually operates at atmospheric pressure regeneration process is favoured by greater pressures and its operating pressure will have to be chosen accordingly. Pressure drops have been assumed for both the columns (5 kPa for the absorber and 20 kPa in the regenerator). Considering pilot plant solvent concentration (32.5% aqueous solutions of MEA) and columns packing, a baseline commercialscale capture plant model has been developed, scaling it up from the pilot plant model. Absorber and regenerator column heights



have been set equal to, respectively, 15 and 10 m. Cross heat exchanger approach temperature has been defined equal to 10 K and regenerator pressure is 160 kPa. Most relevant user input parameters have been undergone a sensitivity analysis (varying them from baseline case) in order to highlight their influence on capture plant performance and, notably, reboiler duty requirement. Reboiler duty, despite not being the only relevant parameter, is securely the greatest contributor on techno-economic performance of CCGT power plants with CO2 capture. In Table 5 the main results of this analysis are given. When sizing and designing the operational parameters of a capture plant, absorber and regenerator columns, on which mostly depend process economy and efficiency, need to be considered contextually to avoid to operate at a region requiring higher capital and operational expenditure. For this reason, in Table 5 the influence of the lean loading has been also investigated to determine reasonable columns design in terms of capital costs and operational performance. When a low lean loading is specified (that is, a low CO2 concentration in the regenerated solvent has been targeted) reboiler duty (and, consequently, steam draw off) required to strip the rich solvent of CO2 to the desired lean loading will be larger but, on the other hand, given the increased absorption capacity of the lean stream, the quantity of solvent required to attain the desired capture level is decreased.



Fig. 4. Temperature profile for Case 42 (medium L/G ratio): (a) absorber and (b) regenerator.



16



R. Canepa, M. Wang / Applied Thermal Engineering 74 (2015) 10e19



Fig. 5. Temperature profile for Case 39 (high L/G ratio): (a) absorber and (b) regenerator.



The contrary happens when an high lean loading is targeted. For lean loading greater then approximately 0.250, reboiler duty requirement will increase slightly due to the sensible heat demand of the increased rich solvent flow rate. Therefore, while L/G ratio always increases with increasing lean loading, reboiler duty will typically present a saddle tendency with respect to this parameter. In the already mentioned Table 5, considering base-case, the impact of lean loading specification on sizing requirement, can be assessed. When L/G ratio is increased the absorber diameter, in order to obtain the desired flooding percentage, is increased as well. On the other hand regenerator flooding (and thus diameter) is mainly connected to reboiler duty requirement. However, from a process optimization point of view, absorber columns will be among the two, given their larger number and size, the most determinant factor. Absorber packing height was increased from base-case capture plant (15 m). This would require greater capital and also O&M costs connected to capture. By increasing the packing height and, as a consequence, packing volume, absorption capacity is expected to increase. From Table 5 it is clear the beneficial effect of an increase of absorber packing height on thermal energy requirement. Interesting is the fact that the optimal lean loading



(0.25 molCO2/molMEA) is almost unchanged from base-case. With 25 m of absorber packing height reboiler duty requirement is decreased, considering the optimal lean loading, by 31%. A further increase of absorber packing height to 30 m would only marginally increase this benefit (33% in reboiler duty, if compared to basecase capture plant) and, on the other hand, would increase further capital and O&M costs. From column sizing point of view an increase of absorber packing height is very relevant on, as easily expected, absorber requirement but, as well, it is a determinant factor on regenerator sizing. This is mainly due to the lower reboiler duty requirement granted by an increase in the absorber height, on which, as previously shown, regenerator diameter is mainly related. A packing height equal to 25 m ensures a reduction, compared to base-case, of 5% in absorber diameter and of 20% in regenerator one. Regenerator operating pressure was changed from base-case capture plant. Notably three different values of this quantity where considered: 160 kPa (base-case), 185 kPa and 210 kPa, assuming for everyone of them a pressure drop through the regenerator column equal to 20 kPa. Increasing operating pressure would require greater pumping equipment along major pumping operating costs. From thermal point of view, an increase in



Fig. 6. First-guess solution result: absorber and regenerator diameters as function of the number of columns.



17



R. Canepa, M. Wang / Applied Thermal Engineering 74 (2015) 10e19 Table 5 Sensitivity analysis results.



Table 6 Commercial scale capture plant design and operational specification. Lean load [mol/mol]



a



L/G [mol/mol] Abs. diameter [m] Reg. diameter [m] Reboiler duty [GJ/tonne] Absorber 30 L/G [mol/mol] height [m] Abs. diameter [m] Reg. diameter [m] Reboiler duty [GJ/tonne] 25 L/G [mol/mol] Abs. diameter [m] Reg. diameter [m] Reboiler duty [GJ/tonne] 20 L/G [mol/mol] Abs. diameter [m] Reg. diameter [m] Reboiler duty [GJ/tonne] Regenerator 210 L/G [mol/mol] pressure Abs. diameter [m] [kPa] Reg. diameter [m] Reboiler duty [GJ/tonne] 185 L/G [mol/mol] Abs. diameter [m] Reg. diameter [m] Reboiler duty [GJ/tonne] 5 L/G [mol/mol] Heater approach [K] Abs. diameter [m] Reg. diameter [m] Reboiler duty [GJ/tonne] Regenerator 20 L/G [mol/mol] height [m] Abs. diameter [m] Reg. diameter [m] Reboiler duty [GJ/tonne] 15 L/G [mol/mol] Abs. diameter [m] Reg. diameter [m] Reboiler duty [GJ/tonne] Base case



0.15



0.20



0.25



0.30



1.14 10.5 11.1 9.60 0.815 10.2 9.40 6.99 0.834 10.2 9.51 7.14 0.890 10.3 9.83 7.60 1.13 10.5 9.61 7.41 1.14 10.5 10.3 8.35 1.13 10.5 11.2 9.54 1.13 10.5 10.7 8.77 1.13 10.5 10.9 9.04



1.42 10.7 10.1 6.99 0.945 10.3 8.22 4.84 0.973 10.3 8.34 4.98 1.07 10.4 8.73 5.40 1.41 10.7 9.02 5.82 1.41 10.7 9.46 6.29 1.42 10.7 10.2 6.93 1.41 10.7 9.8 6.52 1.42 10.7 9.89 6.68



1.92 11.1 10.1 6.27 1.13 10.5 7.94 4.18 1.19 10.5 8.11 4.34 1.35 10.6 8.59 4.78 1.92 11.1 9.41 5.72 1.93 11.1 9.71 5.95 1.92 11.1 10.3 6.17 1.91 11.1 10.0 6.15 1.91 11.1 10.0 6.18



3.18 11.7 11.4 6.97 1.46 10.7 8.36 4.30 1.60 10.8 8.68 4.56 1.99 11.1 9.47 5.23 3.18 11.7 10.7 6.47 3.18 11.7 11.0 6.68 3.18 11.7 11.7 6.66 3.19 11.7 11.4 7.01 3.19 11.7 11.4 7.00



a Absorber packing height: 15 m; regenerator pressure: 160 kPa; heater approach: 10 K; regenerator packing height: 10 m.



regenerator operating pressure corresponds to an increase in driving force and thus a beneficial effect on CO2 mass transfer rate through the regenerator column is expected. Thermal energy requirement has been proven to decrease linearly with regenerator operating pressure. Notably with a pressure equal to 210 kPa reboiler duty is lower by 9% if compared to base-case capture plant. While no effect on absorber diameter has been found, a regenerator pressure equal to 210 kPa led to a 7% reduction in regenerator diameter requirement. Again, the optimal lean loading is not significantly changed from base-case capture plant. Cross heat exchanger provides pre-heating to the cold rich solvent by means of cascade heat from the hot lean solvent. So, a decrease in reboiler duty is expected when cross heat exchanger approach temperature is decreased. On the other hand this would require larger heat exchanger surfaces and thus equipment costs. The beneficial effect on thermal energy requirement is however in percent terms very limited. In correspondence to the optimal lean loading (0.250 molCO2/molMEA) reboiler duty decreases by approximately 2% as the approach temperature decreases from 10 K (base case) to 5 K. Interesting is the fact that, even slightly, a decrease of this temperature difference led to an increase (2%) in regenerator diameter requirement. This is believed to be related to the fact that, by decreasing approach temperature, the temperature of the solvent entering the regenerator column will be affected (notably increased) and thus will be affected flooding capacity on base (entering) stage on which column sizing depends.



Parameter



Value



CO2 capture level [%] CO2 captured [kg/s] Columns flooding [%] Lean loading [mol/mol] Rich loading [mol/mol] L/G [mol/mol] Reboiler duty [kW] Reboiler duty [GJ/tonne CO2] Lean solvent MEA concentration [wt%] Lean solvent temperature [K] Cross heat exchanger approach temperature [K] Absorber columns number [e] Absorber columns pressure [kPa] Absorber columns pressure loss [kPa] Absorber columns packing Absorber columns packing height [m] Absorber columns packing diameter [m] Regenerator columns number [e] Regenerator column pressure [kPa] Regenerator column pressure loss [kPa] Regenerator column packing Regenerator column packing height [m] Regenerator column packing diameter [m]



90 38.77 65 0.200 0.477 0.97 158,766 4.1 32.5 313 10 3 105 5 IMTP no. 40 25 10.3 1 210 20 Flexipack 1Y 15 7.4



As for absorber one, by increasing regenerator packing height column performance will be improved and equipment costs increased. Reboiler duty is decreased by 1.5% from base-case by the adoption of 15 m of packing. A further increase in column packing height doesn’t bring any significant benefits. Regenerator diameter, in correspondence with the optimal lean loading (from thermal energy requirement point of view), is only slightly decreased (0.6%) from base-case. No effect on absorber diameter has been proven. 6. Thermo-economic performance Considering the sensitivity analysis previously shown it has been possible to identify an improved set of capture plant operating parameters. In Table 6 the main equipment design parameters are given. Columns packing heights and regenerator pressure have been increased, given the great influence they have on reboiler duty requirement, while heat exchanger approach temperature has been left unchanged from base case capture plant, given its minor influence on thermal energy requirement. Such a configuration securely allows a reduction on reboiler duty requirement if compared to base case capture plant. As for the previously analyzed cases reboiler duty requirement shows a saddle tendency with respect to lean loading (see Fig. 7). The minimal energy requirement condition (which was 0.250 molCO2/ molMEA for base case) is here between 0.200 and 0.250 molCO2/ molMEA of solvent lean loading. However, despite reboiler duty is securely the most important thermal energy requirement of capture plant (to which correspond the greatest efficiency penalty), net power production is not the only parameter which affects the cost of electricity. As already shown when the lean loading is increased solvent flow rate is increased too. Thus, equipment capital cost and O&M costs are expected to be greater. For this reason, in the already mentioned Fig. 7, the cost of electricity is also depicted. Cost of electricity depends on both thermodynamic performance and total annual revenue requirement (TRR) of the integrated power plant and CO2 related sections (flue gas preprocessing, capture section and CO2 compression). Major details on the integrated modelling and cost related assumptions are given in Ref. [18]. Considering the cost of electricity too as a



18



R. Canepa, M. Wang / Applied Thermal Engineering 74 (2015) 10e19



Fig. 7. Impact of lean loading on reboiler duty and cost of electricity at 90% capture level.



Fig. 8. Impact of lean loading on net power production and capture plant total capital requirement at 90% capture level.



decision parameter the optimal lean loading (see Table 6) has been identified equal to 0.200 molCO2/molMEA. To better understand this, in Fig. 8 net power production and capture plant related total annual revenue requirement as a function of lean loading are shown. It can be noted as, while net power plant power production is mainly dependent on reboiler duty (blower and CO2 compression power requirement are unchanged), capture plant capital annualized and O&M costs, represented by the TRR parameter, increase almost linearly with lean loading. As a result, reboiler duty has been identified to be equal to 4.1 GJ/tonne CO2. To satisfy this requirement a large amount of steam need to be extracted from power plant, considerably reducing its thermal efficiency. Power plant net power production is reduced, from 427 MW of no-capture case, to 368 MW, mainly due to steam draw off. This, in combination with the capital and O&M costs connected to capture related sections, gives for the integrated CCGT and capture plants a levelized cost of the electricity equal to 68 V/MWh, increased by 47% when compared to reference (no capture) power plant.



equilibrium based one. Experimental results from pilot plant facility at the University of Texas at Austin represent an invaluable source of insight on post-combustion capture by means of MEA solvent. An extensive validation campaign based on these data has allowed to thoroughly assess model validity. A sensitivity analysis has been conducted to prove how, with a proper choice of capture plant key operating parameters and equipment design, capture plant thermal energy requirement for solvent regeneration might be reduced to approximately 4 GJ/tonne CO2. It has been demonstrated as the most economical solution is not the one with the lowest thermal energy requirement, being capture plant related costs also very relevant on the cost of electricity, and thus proving the need to optimize capture plant from both the thermal and economical point of view.



Acknowledgements Financial support for this study to the first author was provided by ANSALDO ENERGIA and is gratefully acknowledged.



7. Conclusions While the technology to capture CO2 from exhaust gas exists and is viable, it still needs to be deployed on commercial scale power plant. This makes modelling and simulation an invaluable tool for investigating CO2 capture process integration at commercial scale. Very often capture process is simulated in a simplified way or not properly validated. Rate-based approach, as adopted for CO2 capture section, ensures a higher reliability over traditional



References [1] International Energy Agency (IEA), Energy Technology Perspectives, 2010. http://www.iea.org/publications/freepublications/publication/etp2010.pdf. [2] Global CCS Institute, The Global Status of CCS, 2012. http://cdn. globalccsinstitute.com/sites/default/files/publications/47936/global-statusccs-2012.pdf. [3] H. Herzog, J. Meldon, A. Hatton, Advanced Post-combustion CO2 Capture, Clean Air Task Force, 2009, pp. 1e39.



R. Canepa, M. Wang / Applied Thermal Engineering 74 (2015) 10e19 [4] M. Wang, A. Lawal, P. Stephenson, J. Sidders, C. Ramshaw, Post-combustion CO2 capture with chemical absorption: a state-of-the-art review, Chem. Eng. Res. Des. 89 (9) (2011) 1609e1624. [5] R. Svensson, M. Odenberger, F. Johnsson, L. Strömberg, Transportation systems for CO2 e application to carbon capture and storage, Energy Convers. Manage. 45 (15) (2004) 2343e2353. [6] S.M. Forbes, P. Verma, T.E. Curry, S.J. Friedmann, S.M. Wade, et al., Guidelines for Carbon Dioxide Capture, Transport and Storage, World Resources Institute, 2008. [7] C. Wildbolz, Life Cycle Assessment of Selected Technologies for CO2 Transport and Sequestration (Diploma thesis), Swiss Federal Institute of Technology, Zurich, 2007. [8] Standards of the Tubular Exchanger Manufacturers Association, The Tubular Exchange Manufacturers Association Inc., Tarrytown, New York, 2007. [9] R. Canepa, M. Wang, C. Biliyok, A. Satta, Thermodynamic analysis of combined cycle gas turbine power plant with post-combustion CO2 capture and exhaust gas recirculation, Proc. Inst. Mech. Eng. Part E 227 (2) (2013) 89e105. [10] Y. Zhang, H. Chen, C.C. Chen, J.M. Plaza, R. Dugas, G.T. Rochelle, Rate-based process modeling study of CO2 capture with aqueous monoethanolamine solution, Ind. Eng. Chem. Res. 48 (20) (2009) 9233e9246. [11] R.E. Dugas, Pilot Plant Study of Carbon Dioxide Capture by Aqueous Monoethanolamine (MSE thesis), University of Texas at Austin, 2006. [12] B.R.W. Pinsent, L. Pearson, F.J.W. Roughton, The kinetics of combination of carbon dioxide with hydroxide ions, Trans. Faraday Soc. 52 (1956) 1512e1520. [13] Aspen Plus, Rate Based Model of the CO2 Capture Process by MEA Using Aspen Plus, Aspen Technology Inc., Cambridge, MA, USA, 2012.



19



[14] H. Hikita, S. Asai, H. Ishikawa, M. Honda, The kinetics of reactions of carbon dioxide with monoethanolamine, diethanolamine and triethanolamine by a rapid mixing method, Chem. Eng. J. 13 (1) (1977) 7e12. [15] H.M. Kvamsdal, G.T. Rochelle, Effects of the temperature bulge in CO2 absorption from flue gas by aqueous monoethanolamine, Ind. Eng. Chem. Res. 47 (3) (2008) 867e875. [16] A. Lawal, M. Wang, P. Stephenson, O. Obi, Demonstrating full-scale postcombustion CO2 capture for coal-fired power plants through dynamic modelling and simulation, Fuel 101 (2012) 115e128. [17] M. Ramezan, T.J. Skone, N. Nsakala, G.N. Liljedahl, Carbon Dioxide Capture from Existing Coal-fired Power Plants, US Department of Energy e National Energy Technology Laboratory, 2007. DOE/NETL-401/110907. [18] R. Canepa, Analysis of Combined Cycle Power Plants with Post-combustion CO2 Capture, Including Turbomachinery Performance Estimation (Ph.D. thesis), Genova University, Italy, 2013.



Symbols Dp: pressure loss [bar] f: Darcy friction factor [e] L: pipeline length [m] Di: pipeline internal diameter [m] ravg: average CO2 mass density [kg/m3] vavg: average CO2 velocity [m/s]



Applied Thermal Engineering 74 (2015) 20e27



Contents lists available at ScienceDirect



Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng



Multi-fuel multi-product operation of IGCC power plants with carbon capture and storage (CCS) Ana-Maria Cormos a, Cristian Dinca b, Calin-Cristian Cormos a, * a b



Babes e Bolyai University, Faculty of Chemistry and Chemical Engineering, 11 Arany Janos, RO-400028 Cluj-Napoca, Romania Politehnica University, Faculty of Power Engineering, 313 Splaiul Independentei, RO-060042 Bucharest, Romania



h i g h l i g h t s � Assessment of IGCC-based energy vectors poly-generation systems with CCS. � Optimisation of gasification performances and CO2 emissions by fuel blending. � Multi-fuel multi-product operation of gasification plants.



a r t i c l e i n f o



a b s t r a c t



Article history: Received 27 July 2013 Accepted 30 December 2013 Available online 9 January 2014



This paper investigates multi-fuel multi-product operation of IGCC plants with carbon capture and storage (CCS). The investigated plant designs co-process coal with different sorts of biomass (e.g. sawdust) and solid wastes, through gasification, leading to different decarbonised energy vectors (power, hydrogen, heat, substitute natural gas etc.) simultaneous with carbon capture. Co-gasification of coal with different renewable energy sources coupled with carbon capture will pave the way towards zero emissions power plants. The energy conversions investigated in the paper were simulated using commercial process flow modelling package (ChemCAD) in order to produce mass and energy balances necessary for the proposed evaluation. As illustrative cases, hydrogen and power co-generation and FischereTropsch fuel synthesis (both with carbon capture), were presented. The case studies investigated in the paper produce a flexible ratio between power and hydrogen (in the range of 400e600 MW net electricity and 0e200 MWth hydrogen considering the lower heating value) with at least 90% carbon capture rate. Special emphasis were given to fuel selection criteria for optimisation of gasification performances (fuel blending), to the selection criteria for gasification reactor in a multi-fuel multi-product operation scenario, modelling and simulation of whole process, to thermal and power integration of processes, flexibility analysis of the energy conversion processes, in-depth techno-economic and environmental assessment etc.  2014 Elsevier Ltd. All rights reserved.



Keywords: Gasification Coal & biomass co-processing Energy vectors poly-generation Hydrogen & power co-generation Carbon capture and storage (CCS)



1. Introduction Energy issue is very important and actual giving the double significance of the problem: security of primary energy supply and environmental protection and climate change prevention by reducing the greenhouse gas emissions (mainly CO2). Solid fossil fuels ensure a much bigger energy independence compared to liquid and gaseous fossil fuels [1], but coal utilisation is regarded with concern considering the greenhouse gas emissions. Subsequently, solid fossil fuels can be used in the future only in conjunction with Carbon Capture and Storage (CCS) technologies.



* Corresponding author. Tel.: þ40 264 593833; fax: þ40 264 590818. E-mail address: [email protected] (C.-C. Cormos). 1359-4311/$ e see front matter  2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.applthermaleng.2013.12.080



Also, utilization of biomass (e.g. sawdust, agricultural waste, energy crops etc.) and other different solid wastes (e.g. municipal waste, sewage sludge, organic wastes like meat and bone meal etc.) in energy conversion processes, is become more and more significant and their use is predicted to increase sharply. In this context, European Commission has set as a target, for the whole community block, that until 2020, 20% from the energy mix should be covered by renewable energy sources [2]. Since the carbon footprint of these fuels is much lower than in the case of coal, co-processing represent a viable way to reduce fossil CO2 emissions. Gasification is an energy conversion process in which the solid feedstock is partially oxidised with oxygen and steam to produce syngas. Syngas can be further used, via chemical conversion, to produce different valuable compounds which can be used as energy vectors (e.g. hydrogen, methanol, SNG, liquid fuels) or to



A.-M. Cormos et al. / Applied Thermal Engineering 74 (2015) 20e27



21



generate power in a Combined Cycle Gas Turbine (GGCT). Also, unlike other energy conversion processes (e.g. combustion), gasification is more suitable to process low grade fuels (biomass, solid wastes) as a way to strengthen primary energy source supplies [3]. In addition, Integrated Gasification Combined Cycle (IGCC) is one of the power generation technologies having the highest potential to capture CO2 with the lowest penalties in term of energy efficiency and capital & operational costs. In a modified IGCC design for carbon capture, the syngas can be catalytically shifted to maximize the hydrogen level in the syngas and to concentrate the carbon species in the form of CO2 that can be later captured in a pre-combustion arrangement. Hydrogen-rich gas can be furthermore used for power generation or for other applications (e.g. petro-chemistry, PEM fuel cells). The present paper investigates the usage of IGCC design in a multi-fuel multi-product operation scenario. As illustrative examples, co-gasification process of coal with biomass (e.g. sawdust, agricultural wastes) or solid wastes (e.g. meat and bone meal, municipal waste, sewage sludge etc.), for hydrogen and power cogeneration with CCS were evaluated. An illustrative case for coal and meat and bone meal (MBM) co-gasification for liquid fuel production (FT synthesis) with CCS was also presented. The hydrogen and power co-generation case study presented in the paper is producing around 400e500 MW net electricity with a flexible hydrogen production in the range of 0e200 MWth (based on hydrogen lower heating value). The carbon capture rate of all cases was higher than 90% (considering the total carbon from the feedstock). If consider that some of the feedstock’s carbon is coming from renewable energy sources (biomass), the evaluated power plant concepts can be considered as a significant step towards zero or even negative fossil CO2 emissions power plants. Important details, in terms of plant design, are evaluated using process flow modelling (e.g. fuel blending for optimisation of gasification energy efficiency, mass and energy integration issues, overall technoeconomic and environmental parameters etc.).



used to raise steam (in Heat Recovery Steam Generator e HRSG) which by expansion in a steam turbine generates extra electricity in addition to the one generated by the gas turbine. Along technological development, various gasification reactors were used (moving bed, fluidised bed and entrained-flow). Presently, oxygenblown entrained-flow gasifiers are considered the optimum solution for IGCC designs with carbon capture. The most evaluated commercial entrained-flow gasifiers in conjunction with carbon capture are Shell, Siemens, GE-Texaco, E-Gas etc. Compared to conventional IGCC concept without carbon capture, the modification of the design to introduce the precombustion carbon capture step involves changes in the plant configuration as follow: a catalytically water gas shift stage to convert carbon monoxide to carbon dioxide, a bigger Acid Gas Removal (AGR) unit which captures H2S and CO2, a hydrogen purification stage based on Pressure Swing Adsorption (PSA) and a combined cycle running on hydrogen-rich gas. Captured CO2 stream has to comply with certain quality specification imposed by transport and storage. In this paper, the following capture CO2 quality specification was used (expressed in % vol.): >95% CO2; 25% (vol.) Tail gas recycled to H2S absorption stage Purified hydrogen: > 99.95% (vol.) Purification yield: 85% Tail gas pressure: 1.5 bar (recycled to power island) Type: M701G2 (Mitsubishi Heavy Industries Ltd.) Net power output: 334 MW Electrical efficiency: 39.5% Pressure ratio: 21 Turbine outlet temperature (TOT): 588 � C Three pressure levels: 118/34/3 bar MP steam reheat Integration of steam generated in gasification island and syngas treatment line with CCGT Steam turbine isoentropic efficiency: 85% Power consumption for heat rejection: 1% of the heat discharged Steam wetness ex. steam turbine: max. 10% DTmin. ¼ 10 � C Pressure drop: 1e5% of inlet pressure



Stream



Parameters



Air Separation Unit (ASU)



Table 3 Characterisation of main plant streams (Case 3).



Unit



48.00 80.00 129,220 4018.38



Table 2 Main design assumptions.



1.15 587.83 2,812,186 102530.60



A.-M. Cormos et al. / Applied Thermal Engineering 74 (2015) 20e27



24



A.-M. Cormos et al. / Applied Thermal Engineering 74 (2015) 20e27



Table 4 Steam cycle (Case 3). Stream HP steam from process units (WGS reactors) HP steam from process units (gasifier) HP steam to HP steam turbine MP steam to MP reheater Hot reheated MP steam MP steam to process units LP steam from process units LP steam to LP steam turbine LP steam (6.5 bar) to process units LP steam turbine exhaust Cooling water to steam condenser Cooling water from steam condenser Hot condensate returned to HRSG BFW to HP BFW pumps BFW to MP BFW pumps BFW to LP BFW pumps Flue gas at stack



Temperature ( C)



Pressure (bar)



122.00



339.03



120.00



6.70



571.00



118.00



430.70 464.40 464.40 35.30 80.50 600.20 27.70 600.20 32000.00 32000.00 802.68 424.00 69.00 163.50 2812.18



576.98 387.31 449.84 418.33 204.66 184.81 253.10 31.32 15.00 25.00 115.00 115.00 115.00 115.00 110.65



118.00 34.00 33.50 41.00 3.00 3.00 6.50 0.046 2.00 1.80 2.80 2.80 2.80 2.80 1.02



Flowrate (t/h)



and syngas water quench configuration. The first aspect is important in evaluating co-gasification of various fuels (coal in addition with biomass/solid wastes), the second aspect is important in term of increasing energy efficiency and the third aspect is significant because it ensures the water requirement for shift conversion. The following cases were investigated in this paper: Case 1 Coal only as feedstock; Case 2 Coal in addition with sawdust (80/20% wt. blending ratio); Case 3 Coal in addition with meat and bone meal (80/20% wt. blending ratio); Case 4 Coal in addition with municipal solid waste (80/20% wt. blending ratio); Case 5 Coal in addition with sewage sludge (80/20% wt. blending ratio); Case 6 Coal in addition with waste paper (80/20% wt. blending ratio). As main design assumptions, the cases evaluated in the paper generate about 420 MW electricity using one M701G2 gas turbine (Mitsubishi Heavy Industries Ltd.) and a stream of hydrogen with 99.95% (vol.) purity. Hydrogen output is a flexible one, varying in the range of 0e200 MWth, considering hydrogen lower heating value (10.795 MJ/Nm3). The concepts are designed to capture more than 90% of the carbon from the feedstock. Other main sub-systems



Fig. 3. Energy integration analysis for gasification island and syngas conditioning line.



Fig. 4. Energy integration analysis for the power block (CCGT).



of the plant for co-generation of hydrogen and electricity with CCS are presented in Table 2 [4,9e11]. The IGCC-based power generation schemes with CCS were modelled and simulated using ChemCAD and Thermoflex. The simulation results were validated against experimental and industrial data [4,9,10,12e14]. As illustrative example, Tables 3 and 4 present the main streams within the plant (Table 3) and steam cycle characteristics (Table 4) for Case 3 (coal and MBM co-gasification). After modelling and simulation, the schemes were evaluated in term of energy integration for optimisation of energy efficiency [15e19]. As an illustrative example, Figs. 3 and 4 present hot and cold composite curves for Case 3 (coal and MBM co-gasification) for the main plant sub-systems (Fig. 3 for gasification island and syngas conditioning line and Fig. 4 for power block). The mass and energy balances generated by simulation where furthermore used for quantification of overall techno-economic and environmental key performance indicators. In term of technical indicators the following key parameters were calculated: cold gas efficiency (CGE), syngas treatment efficiency (via shift reactors and AGR unit), gross and net power efficiencies, ancillary consumption. As environmental indicators, carbon capture rate and specific CO2 emissions (considering all carbon stream not divided in fossil and renewable) were considered. For quantification of techno-economic and environmental parameters, the assessment methodology presented in Refs. [9,20e23] was used to evaluate capital and operational costs. First, all three case studies were simulated for electricity production only with CCS in a precombustion arrangement using Selexol (physical gaseliquid absorption). An overview of the main techno-economic plant performance indicators are presented in Table 5. As can be noticed from Table 5, in term of power generation, all case studies generate about 420 MW with a net electrical efficiency in the range of 34.6e37.2%. What is important to notice is the fact that co-gasification of coal with sawdust does not reduce significantly the overall energy efficiency, even more for coal cogasification with MBM the net power efficiency is increasing with about 1.2 net electricity percentage points. This is mainly due to the positive influence on gasification performances of the reducing temperature and subsequently the oxygen consumption. In term of environmental performances, all evaluated case have carbon capture rate of about 92e93% and specific CO2 emissions in the range of 68e76 kg/MWh (conventional IGCC technology without carbon capture has specific CO2 emission in the range of 700e800 kg/MWh). The carbon capture rate was calculated considering the whole carbon stream going into the process (both fossil and renewable). For instance for Case 2 (coal co-processing



25



A.-M. Cormos et al. / Applied Thermal Engineering 74 (2015) 20e27 Table 5 Overall plant performance indicators (pre-combustion capture). Main plant data



Units



Case 1



Solid fuel flowrate (as received) Coal/alternative fuels LHV (as received) Feedstock thermal energy e LHV (A) Thermal energy of the syngas (B) Cold gas efficiency (B/A * 100) Thermal energy of syngas exit AGR (C) Syngas treatment efficiency (C/B * 100) Gas turbine output (1 � M701G2) Steam turbine output Expander power output Gross electric power output (D) ASU consumption þ O2 compression Gasification island power consumption AGR þ CO2 drying & compression Power island power consumption Total ancillary power consumption (E) Net electric power output (F ¼ D � E) Gross electrical efficiency (D/A * 100) Net electrical efficiency (F/A * 100) Carbon capture rate CO2 specific emissions Total investment cost Total investment cost per kW gross Total investment cost per kW net Total fixed O&M costs (year) Total fixed O&M costs (kWh net) Total variable O&M costs (year) Total variable O&M costs (kWh net) Total fixed and variable costs (year) Total fixed and variable costs (kWh net)



kg/h MJ/kg MWth MWth % MWth % MWe MWe MWe MWe MWe MWe MWe MWe MWe MWe % % % kg/MWh MM V V/kW V/kW MV/y V/kWh MV/y V/kWh MV/y V/kWh



165704 180,455 168,453 25.353/16.057/19.263/11.962/12.607/18.771 1166.98 1177.68 1129.35 934.75 934.26 931.93 80.10 79.33 82.52 830.70 831.95 832.03 88.86 89.05 89.28 334.00 334.00 334.00 197.50 200.14 196.13 0.78 0.78 0.78 532.28 534.92 530.91 44.72 45.13 41.53 8.08 8.27 8.11 40.07 40.54 42.09 19.00 19.05 18.95 111.87 112.99 110.68 420.41 421.93 420.23 45.61 45.42 47.01 36.02 35.82 37.20 92.35 92.83 92.24 76.12 71.19 72.23 1102.31 1131.33 1092.95 2070.93 2114.95 2058.63 2621.99 2681.31 2600.83 38.22 39.03 38.06 0.01212 0.01233 0.01208 79.18 73.95 135.79 0.02511 0.02337 0.04309 117.40 112.98 173.85 0.03723 0.03570 0.05517



Case 2



with sawdust), about 88% of the feedstock carbon is fossil the rest being renewable. Considering this aspect and the carbon capture rate, one can consider that the co-gasification cases evaluated in this paper have near zero fossil CO2 emissions. The mass and energy balances resulted from simulation were furthermore used for assessment of economic aspects. For evaluation of capital and operational & maintenance (O&M) costs, cost of electricity and hydrogen as well as cumulative cash flow analysis, the methodology presented in Ref. [22] was used. As investment cost indicators, all six cases have similar capital costs in the range of 1093e1178 MM V. The specific capital investments are in the range of 2600e2800 V/kW net. One can notice a positive situation for MBM co-processing (Case 3) compared to other cases due to higher energy efficiency. For operation & maintenance (O&M) costs, considering Case 1 (coal only) as base case, there are positive differences (e.g. for Case 2 and Case 4) and negative differences for Cases 3, 5 and 6. For most of the cases, the increase of O&M costs is due to higher fuel prices (compared with coal) and lower calorific value. The cumulative cash flow analysis for one illustrative case (Case 3: coal and MBM co-gasification) is presented in Fig. 5. The cost of electricity considered in the analysis was 7.36 ¢/kWh and the CO2 removal and avoidance costs were 30.58 V/t and respectively 38.29 V/t. For flexible hydrogen and power co-production in the range of 0e200 MW hydrogen, the gas turbine is gradually turned down to about 80% in order to displace an energy stream of hydrogen which can be purified by Pressure Swing Adsorption (see Fig. 1). The reason to select 80% as turn down limit of the gas turbine is imposed by the fact that under this value the efficiency drops significantly (variable inlet guide vanes are completely closed and no control of the airflow can be made to maintain the gas turbine inlet temperature). The tail gas resulted from hydrogen purification step is then compressed and recycled back to the power block to be burned in the gas turbine. Table 6 presents the variation of plant performance indicators with hydrogen output for Case 3 (coal and MBM co-gasification). As



Case 3



Case 4



Case 5



Case 6



187,400



192,000



182,900



1180.37 935.62 79.26 832.11 88.93 334 200.93 0.78 535.71 45.79 8.35 40.80 19.05 113.99 421.72 45.38 35.72 93.02 70.68 1148.21 2143.34 2722.69 39.48 0.01248 74.24 0.02347 113.72 0.003595



1216.22 938.91 77.20 833.61 88.78 334.00 205.04 0.79 539.83 50.10 8.59 41.33 19.11 119.13 420.70 44.38 34.59 93.57 68.29 1178.61 2183.31 2801.55 40.17 0.01273 85.35 0.02705 125.52 0.03978



1178.48 932.99 79.17 831.87 89.16 334.00 200.16 0.78 534.94 45.15 8.23 40.01 18.62 112.01 422.93 45.39 35.88 92.24 78.11 1133.75 2119.40 2680.71 39.12 0.01233 94.51 0.02980 133.63 0.04213



noticed from Table 6, for hydrogen and power co-production mode, the overall efficiency of the plant is increasing when the ancillary power consumption is remaining virtually constant. This fact is very important and attractive for plant cycling (modification of the power generated by the plant according to the demand of the electricity grid) considering that for low electricity demand the plant can produce mostly hydrogen which, compared with power, can be stored to be used either for covering peak loads or for other applications (transport sector, petro-chemical sector etc.). Specific CO2 emissions are also decreasing with increasing hydrogen production rate. As overall conclusion, the energy vector polygeneration operation for an IGCC design appears to be promising in term of techno-economic parameters. Co-gasification of coal with various alternative fuels (biomass and solid wastes) can be used for other totally or partially decarbonised energy vectors for instance substitute natural gas (SNG) or



Fig. 5. Cumulative cash flow analysis for Case 3.



26



A.-M. Cormos et al. / Applied Thermal Engineering 74 (2015) 20e27



Table 6 Overall plant performance indicators (hydrogen and power co-production mode for Case 3). Main plant data



Units



Power



Fuel flowrate (as received) Coal/MBM LHV (as received) Feedstock thermal energy e LHV (A) Syngas thermal energy (B) Cold gas efficiency (B/A * 100) Thermal energy of syngas ex. AGR (C) Syngas treatment efficiency (C/B * 100) Gas turbine output (1 � M701G2) Steam turbine output Expander power output Gross electric power output (D) Hydrogen output e LHV (E) ASU consumption þ O2 compression Gasification island power consumption AGR þ CO2 drying & compression H2 compression Power island power consumption Total ancillary power consumption (F) Net electric power output (G ¼ D � F) Gross electrical efficiency (D/A * 100) Net electrical efficiency (G/A * 100) Hydrogen efficiency (E/A * 100) Cumulative efficiency (G þ E/A * 100) Carbon capture rate CO2 specific emissions (power) CO2 specific emissions (energy)



kg/h MJ/kg MWth MWth % MWth % MWe MWe MWe MWe MWth MWe MWe MWe MWe MWe MWe MWe % % % % % kg/MWh kg/MWh



168,453 25.353/19.263 1129.35 931.93 82.52 832.03 89.28 334.00 196.13 0.78 530.91 0.00 41.53 8.11 42.09 0.00 18.95 110.68 420.23 47.01 37.20 0.00 37.20 92.24 72.23 72.23



liquid fuels by FischereTropsch synthesis [24e26]. As an illustrative example [27,28], FT fuel and power co-generation case was evaluated in term of main plant performance indicators. As fuel used, co-processing coal and meat and bone meal (MBM) in a Siemens gasifier was considered. The whole plant concept was fully integrated in term of mass and energy aspects (e.g. light hydrocarbon stream was used for heat and power production to cover the ancillary plant consumptions, the heavy hydrocarbon stream was hydro-cracked to produce usable motor hydrocarbons). The investigated case has only a steam (Rankine) cycle which generates power based on available plant heat sources e e.g. shift and FT reactors, syngas cooling train etc. The overall plant performance indicators for liquid fuels and power co-production with CCS are presented in Table 7. As can be noticed, the overall plant energy efficiency is about 67% with a carbon capture rate in the range of 48% and 39 kg CO2/ MWh specific emissions. As in other cases, the carbon capture rate and specific CO2 emissions were calculated considering the total feedstock carbon. The overall conclusion is that FT fuel and power co-generation based on co-gasification with CCS seems to be a



Table 7 Plant performance indicators for liquid fuels production by FischereTropsch synthesis. Main plant data



Units



Value



Feedstock thermal energy e coal Feedstock thermal energy e meat and bone meal (MBM) Total fuel thermal energy Liquid fuel thermal energy Gross power generation Total power consumption Net power output Net electrical efficiency Net thermal efficiency Cumulative energy efficiency Carbon capture rate CO2 specific emissions (total energy output)



MWth MWth



1159.25 73.78



MWth MWth MWe MWe MWe % % % % kg/MWh



1233.03 700.00 193.12 65.24 127.88 10.37 56.77 67.14 47.58 38.92



Power þ hydrogen



313.65 187.82 0.73 502.20 50.00 41.53 8.11 42.09 0.66 18.52 110.91 391.29 44.46 34.64 4.42 39.06 92.83 77.57 68.78



294.76 179.05 0.68 474.75 100.00 41.53 8.11 42.09 1.33 18.08 111.14 364.61 42.12 32.28 8.85 41.13 92.83 83.24 65.33



273.84 170.11 0.62 445.28 150.00 41.53 8.11 42.09 2.01 17.65 111.39 333.89 39.42 29.56 13.28 42.84 92.83 90.91 62.72



259.95 162.35 0.57 418.41 200.00 41.53 8.11 42.09 2.68 17.24 111.65 306.76 37.04 27.16 17.70 44.86 92.83 98.94 59.89



promising concept which combines high energy efficiency and carbon capture rate. For a flexible FT fuel and power co-generation plant, a combined cycle gas turbine has to be used in the power block. This gas turbine will use part of synthesized FT hydrocarbons as fuel to generate a higher power output compared to the case evaluated here which uses only the available plant heat. 5. Conclusions The paper assesses the suitability of coal and alternative fuels (biomass and solid wastes) co-gasification for energy vector polygeneration with carbon capture and storage. Illustrative cases of hydrogen and electricity co-production and FischereTropsch synthesis schemes with carbon capture based on a modified IGCC plant design were presented. One aim was to evaluate the possibility of coal blending with various biomass and solid wastes as a way to process these alternative fuels, but also to optimise the gasification conversion. As presented, co-gasification of coal with alternative fuels having low ash fusion temperature can be used for increasing the cold gas efficiency and reducing the oxygen consumption. This aspect is very important not only from energy efficiency point of view but also in term of security of supply (finding ways in which low grade fuels, biomass and solid wastes can be converted efficiently into various energy vectors). Another important aspect investigated in the present paper was the poly-generation of various energy vectors based on cogasification processes with pre-combustion carbon capture. This aspect improves both the plant flexibility (the capability of the plant to generate timely the energy vectors required from various energy applications) and overall techno-economic parameters. As presented, for hydrogen and power co-generation and liquid fuels production via FischereTropsch synthesis, gasification process is very promising in poly-generation of totally or partially decarbonised energy vectors. In addition to above mentioned aspects, co-processing fossil fuels with biomass and solid wastes in gasification plants equipped with CCS is very promising concept in transition towards zero fossil CO2 emission power plants.



A.-M. Cormos et al. / Applied Thermal Engineering 74 (2015) 20e27



Acknowledgements This work was supported by a grant of the Romanian National Authority for Scientific Research, CNCS e UEFISCDI, project ID PNII-PT-PCCA-2011-3.2-0162: “Technical-economic and environmental optimization of CCS technologies integration in power plants based on solid fossil fuel and renewable energy sources (biomass)” e CARBOTECH. References [1] Statistical Review of World Energy BP, 2012. www.bp.com. [2] European Commission, Communication from the Commission. 20 20 20 by 2020: Europe’s Climate Change Opportunity, COM, Brussels, Belgium, 2008, 30 final. [3] C.C. Cormos, F. Starr, E. Tzimas, Use of lower grade coals in IGCC plants with carbon capture for the co-production of hydrogen and electricity, Int. J. Hydrogen Energy 35 (2010) 556e567. [4] C. Higman, M. Van der Burgt, Gasification, second ed., Elsevier Science, 2008. [5] A. Kondratiev, E. Jak, Predicting coal ash slag flow characteristics (viscosity model for the Al2O3 e CaO e ‘FeO’ e SiO2 system), Fuel 80 (2001) 1989e2000. [6] Babcock and Wilcox, Steam: Its Generation and Use, forty first ed., McDermond Company, 2005. [7] A.G. Collot, Matching gasification technologies to coal properties, Int. J. Coal Geol. 65 (2006) 191e212. [8] C.C. Cormos, A. Padurean, A.M. Cormos, S. Agachi, Power generation based on coal and low-grade fuels co-gasification with carbon capture and storage, in: Clean Coal Technology Conference e CCT 2011, Zaragoza, Spain. [9] International Energy Agency (IEA), Greenhouse Gas R&D Programme (GHG), Potential for Improvement in Gasification Combined Cycle Power Generation with CO2 Capture, 2003. Report PH4/19. [10] International Energy Agency (IEA), Greenhouse Gas R&D Programme (GHG), Co-production of Hydrogen and Electricity by Coal Gasification with CO2 Capture, 2007. Report 13/2007. [11] C.C. Cormos, A.M. Cormos, S. Agachi, Power generation from coal and biomass based on integrated gasification combined cycle concept with pre- and postcombustion carbon capture methods, AsiaePac. J. Chem. Eng. 4 (2009) 870e 877. [12] Gasification Technologies Council, 2013, www.gasification.org (accessed 25.07.13).



27



[13] F. Casella, P. Colonna, Dynamic modeling of IGCC power plants, Appl. Therm. Eng. 35 (2012) 91e111. [14] M. Pérez-Fortes, A.D. Bojarski, E. Velo, J.M. Nougués, L. Puigjaner, Conceptual model and evaluation of generated power and emissions in an IGCC plant, Energy 34 (2009) 1721e1732. [15] R. Smith, Chemical Processes: Design and Integration, Wiley, West Sussex, England, 2005. [16] C.C. Cormos, Evaluation of energy integration aspects for IGCC-based hydrogen and electricity co-production with carbon capture and storage, Int. J. Hydrogen Energy 35 (2010) 7485e7497. [17] J. Klemes, F. Friedler, Advances in process integration, energy saving and emissions reduction, Appl. Therm. Eng. 30 (2010) 1e5. [18] S. Bandyopadhyay, G.C. Sahu, Modified problem table algorithm for energy targeting, Ind. Eng. Chem. Res. 49 (2010) 11557e11563. [19] M. Kawabata, O. Kurata, N. Iki, A. Tsutsumi, H. Furutani, System modeling of exergy recuperated IGCC system with pre- and post-combustion CO2 capture, Appl. Therm. Eng. 54 (2013) 310e318. [20] J. Davison, Performance and costs of power plants with capture and storage of CO2, Energy 32 (2007) 1163e1176. [21] NETL, Department of Energy, Cost and performance baseline for fossil energy plants, in: Bituminous Coal and Natural Gas to Electricity, vol. 1, 2007. Report DOE/NETL-2007/1281, USA. [22] C.C. Cormos, Integrated assessment of IGCC power generation technology with carbon capture and storage (CCS), Energy 42 (2012) 434e445. [23] N. Tsakomakas, P. Pilavachi, A. Polyzakis, An economic comparison assessment of lignite and biomass IGCC power plants, Appl. Therm. Eng. 38 (2012) 26e30. [24] J. Hetland, L. Zheng, X. Shisen, How poly-generation schemes may develop under an advanced clean fossil fuel strategy under a joint sino-European initiative, Appl. Energy 86 (2009) 219e229. [25] K. Yamashita, L. Barreto, Energyplexes for the 21st century: coal gasification for co-producing hydrogen, electricity and liquid fuels, Energy 30 (2005) 2453e2473. [26] S. Li, L. Gao, X. Zhang, H. Lin, H. Jin, Evaluation of cost reduction potential for a coal based polygeneration system with CO2 capture, Energy 45 (2012) 101e 106. [27] L. Ardelean, Clean Gasoline Production from Coal and Waste Co-gasification via FischereTropsch Synthesis, Master thesis, BabeseBolyai University, Faculty of Chemistry and Chemical Engineering, 2010. [28] C.C. Cormos, Assessment of flexible energy vectors poly-generation based on coal and biomass/solid wastes co-gasification with carbon capture, Int. J. Hydrogen Energy 38 (2013) 7855e7866.



Applied Thermal Engineering 74 (2015) 28e35



Contents lists available at ScienceDirect



Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng



Experimental studies of CO2 capture by a hybrid catalyst/adsorbent system applicable to IGCC processes Marta Maroño*, Yarima Torreiro, Domingo Cillero, Jose Maria Sánchez CIEMAT, Combustion and Gasification Division, Av. Complutense, 40, 28040 Madrid, Spain



a r t i c l e i n f o



a b s t r a c t



Article history: Received 1 August 2013 Received in revised form 2 October 2013 Accepted 26 February 2014 Available online 12 March 2014



This paper presents experimental results about CO2 capture in a hybrid adsorbent/catalyst system at both laboratory and bench scale. The proposed novel system consists of a homogeneous mixture of a K-doped hydrotalcite and a high temperature FeeCr WGS catalyst in a single reactor, which was selected in previous works. Tests were performed using simulated syngas compositions (CO, CO2, H2, N2, H2O) and simplified binary mixtures (N2/CO and N2/CO2) at temperatures in the range of 300 � Ce500 � C and pressures up to 15 bar. The effect of contact time, process temperature and feed gas composition in the CO2 capture capacity of the sorbent was investigated and main results are presented. Moreover, the sorbent showed catalytic activity towards the WGS reaction which was highly dependent on process temperature. Details on the influence of temperature in the catalytic activity of the sorbent are also described in this paper. The influence of temperature and volume ratio adsorbent/catalyst (Vads/Vcat) in the performance of the hybrid system proposed is discussed in terms of CO conversion and CO2 capture capacity of the sorbent.  2014 Elsevier Ltd. All rights reserved.



Keywords: Hybrid systems CO2 capture SEWGS reaction Hydrotalcite sorbents



1. Introduction Some of the most promising technologies for the capture of CO2 with H2 production in gasification processes e as alternative to liquid amines and pressure swing adsorption (PSA) e include the use of regenerable sorbents in the so-called sorption enhanced water gas shift (SEWGS) reaction process [1]. The SEWGS approach combines a high temperature water gas shift (WGS) catalyst and a CO2 sorbent in the same reactor. Industrial WGS processes require two reactors, one at high temperature (350e400 � C) and another one at low temperature (250 � C) to reach a final adequate composition of CO (300  C) hydrotalcitebased sorbents are known to consume water, by adsorbing it and/ or in the CO2 capture processes so for their application to SEWGS processes it seems necessary to increase the excess steam used. From the point of view of minimizing the steam requirements in the SEWGS process a sorbent with a high selectivity towards CO2 in the presence of water is desired so only the required steam in the WGS reaction will be needed. This paper presents experimental results about CO2 capture tests under conditions of WGS reaction conducted by CIEMAT at laboratory and bench-scale using a hybrid system catalysteadsorbent that consists of a K2CO3 doped hydrotalcite based sorbent and a high temperature WGS catalyst at temperatures in the range of 300  Ce500  C and pressures up to 15 bar. Independent processes seems to exist in the sorbent for the adsorptive and catalytic reactions involved in the conversion of CO and capture of CO2 within the range of temperatures studied in this work, so the influence of process temperature and volume ratio Vads/Vcat in the performance of the hybrid system proposed is discussed in terms of CO conversion and CO2 capture capacity of the sorbent. Besides that, the advantages of the proposed hybrid system versus the performance of the catalyst and the sorbent separately are discussed in this paper.



29



2. Experimental 2.1. Test rigs Two different experimental test rigs have been used in the work presented in this paper. First, the experiments carried out to study the sorbent and the hybrid system adsorbentecatalyst at laboratory scale were performed in a Microactivity Pro lab-scale Unit. It is an automatic and computerized laboratory rig which consists of a stainless steel tubular reactor of OD 9.2 mm and 300 mm long housed in a one single zone SS304 oven which is able to heat the reactor up to 700  C. Maximum operating gas flow rate is 4.5 NL/ min and maximum operating pressure is 20 bar. Desired gas mixture is produced synthetically using mass flow controllers (HiTech). Deionized water is metered by a piston pump (Gibson 307) and vaporized before entering the reactor. Fig. 1 shows a diagram of the testing rig and a more detailed description of the reaction system can be found elsewhere [15]. The experiments carried out to test the hybrid system adsorbentecatalyst at bench scale were performed in a high temperature, high-pressure (HTHP) bench scale facility available at CIEMAT for sorbent and catalyst testing. The plant can treat up to 20 m3 h1 (at standard conditions, 298 K, 101 kPa) of a gas mixture simulating the composition of gasification gases. It is designed to operate at a maximum temperature of 700  C and a pressure of 30 bar. The reactor has a height of 1 m and an internal diameter of 80 mm and is housed inside a four zone furnace equipped with separate temperature controllers for each zone. The furnace can heat the reactor up to 700  C. Gas temperature was monitored at different heights along the reactor using 11 Type K thermocouples which allowed tracking of temperature gradients during reaction. Complete description of the rig can be found in Ref. [16]. For the WGS experiments presented in this work, the reactor was operated in fixed-bed, down-flow mode. In a typical run the reactor was pressurized to the operating pressure, and heated to the desired temperature under flow of nitrogen. When the desired



Fig. 1. Diagram of the laboratory-scale reaction system employed in this work.



30



M. Maroño et al. / Applied Thermal Engineering 74 (2015) 28e35



temperature was attained, nitrogen flow was stopped, and simulated gasification gas flow was started. For both series of studies, inlet gas and exit gas compositions are measured by gas chromatography using a CP4900 Varian gas chromatograph equipped with two columns, a Porapack PPQ and a molecular sieve column and with two thermal conductivity detectors. 2.2. Catalyst The catalyst used in this work is a commercial Johnson Matthey FeeCr high temperature water gas shift catalyst that is being used in the 14 MWth pre-combustion CO2 capture plant in Puertollano IGCC plant [17]. This catalyst was supplied in pellets of 6 � 3 mm and showed high catalytic activity towards the WGS reaction at temperatures higher than 350 � C [14]. 5 g of catalyst was used in each test. 2.3. Sorbent The sorbent used in this work is a K-doped hydrotalcite material supplied by SASOL Germany Ltd which was selected in our previous work [18] as the most promising material to be used in the proposed hybrid system. It was supplied in pellets of 5 � 5 mm. The sorbent was prepared by calcination at 600 � C for 4 h. Both the catalyst and the sorbent were crushed and sieved between 1.6 and 2.3 mm to favor a homogeneous mixture in the reactor. 2.4. Sorptionedesorption tests procedure All the sorptionedesorption tests performed at laboratory scale used a total amount of 10 g of solid (adsorbent alone or a mixture of adsorbent/catalyst). Testing procedure was as follows: first the samples were heated in N2 at 500 � C for one hour before performing the sorption tests. Then, the reactor was cooled down and pressurized to the desired process temperature and pressure in a N2 flow. The reactor was then bypassed and water, previously vaporized, was fed to the system. Then, N2 flow was changed by the process gas flow and the wet gas mixture was allowed to enter the reactor. Sorption tests were performed at temperatures in the range of 300e400 � C using feed gas mixtures consisting of CO2/N2 and CO/N2 depending on the experiment. After saturation of the sorbent, the reactor was bypassed again and depressurization to atmospheric pressure and heating at 500 � C was performed to desorb the CO2 and regenerate the sorbent. When the regeneration conditions were reached, the reactor was connected again and the desorbed species were measured by gas chromatography. In the experiments performed at bench scale a total amount of solid of 2.2 kg was used and the procedure was the following: first the reactor with the mixture of adsorbent and catalyst was heated and maintained at 600 � C in a flow of N2 for one hour. Then, temperature and pressure were adjusted to the process conditions. Water, previously vaporized, was fed to the system and feed gas flow was allowed to enter the reactor. After saturation of the sorbent, feed gas flow was substituted by N2, pressure was decreased to atmospheric pressure and temperature was increased to up to 500 � C. In these tests no bypass of the reactor was possible so the exit composition profiles were measured continuously by gas chromatography. Desorption was performed in all tests using N2 at 500 � C and 1 bar. A complete cycle of sorption-desorption usually lasted 3e4 h.



2.5. Design of the hybrid system adsorbent/catalyst The hybrid system adsorbentecatalyst consisted of the combination of both solids in the same reactor. In order to favor homogeneity in the reactor both the sorbent and the catalyst were crushed and sieved to a size between 1.6 and 2.3 mm. For the binary system adsorbent/catalyst, different volume ratios (Vads/Vcat) were used in the tests, from only adsorbent to Vads/Vcat ¼ 10. This parameter must be defined very carefully to guarantee that complete conversion of CO takes place during the sorption step. In the experiments performed at bench scale both the adsorbent and the catalyst were used in their original size to minimize pressure drop in the reactor. 3. Results and discussion 3.1. Effect of space velocity on the performance of the sorbent The hourly space velocity (SV, h�1) represents the contact time between the solid and the gas and usually has a strong influence in the performance of surface processes such as catalytic or adsorptive reactions. In this work we studied at laboratory scale the influence of space velocity in the performance of the sorbent using values in the range of 400 and 1000 h�1 and the feed gas stream consisted of a mixture of 4%CO2/N2. All the tests were performed at high pressure and water contents (P ¼ 13 bar and 50% v/v water) which have been reported to increase the CO2 capture capacity of K-doped hydrotalcite materials [19e21]. Fig. 2 shows the influence of space velocity on saturation time and total capture capacity of the sorbent used in this work. In Fig. 2 a theoretical breakthrough time (t0) has been calculated considering that all the MgO present in the material is transformed into MgCO3 and only Mg were to take part in the sorption process. Assuming this hypothesis the results obtained in these tests showed that the sorbent used in this work (MG61-K2CO3) would have been using only about 30% of the total capacity and this capacity decreased with increasing space velocity. The effect is more noticeable at high space velocities suggesting that mass transfer limitations in the fixed bed can be occurring under those conditions. As the formation of MgCO3 is a rather slow process it seems reasonable that higher space velocities, i.e. low contact times, resulted in lower capture capacities. Other flow configurations such as for example a fluidized bed reactor could probably provide an improved operation of the system.



Fig. 2. Influence of space velocity on CO2 capture capacity of the sorbent used in this work. P ¼ 13 bar, Feed: 4%CO2/N2, T ¼ 300 � C, 50 %v/v steam.



M. Maroño et al. / Applied Thermal Engineering 74 (2015) 28e35



31



3.2. Performance of the sorbent in presence of simulated gasification gases Usually, a WGS unit inlet gas composition consists of a mixture of CO, H2 and CO2 so in this work the performance of the sorbent in presence of a gas stream that simulated a real one such as that of oxygen blown gasifier at the WGS inlet Unit (i.e. ELCOGAS CO2 capture pilot plant [17]) was investigated. For comparison a simplified gas composition with the same CO2 content was also used: Mixture A: 4%CO2/N2; Mixture B: 55%CO; 4%CO2; 20%H2; N2 balance. The sorption tests were performed at 300 � C, 15 bar, 50% v/v steam and an hourly space velocity of 432 h�1. Fig. 3 shows the breakthrough curves obtained for CO2 for the two feed gas composition tested. As it can be observed in Fig. 3, when only CO2 is present in the feed gas (mixture A), the breakthrough of CO2 is very smooth. However, when CO and H2 were added to the feed gas mixture, as occurred in mixture B, the breakthrough of CO2 started almost immediately. It is interesting to mention that as soon as the gas mixture entered the reactor, conversion of CO took place as corroborated by the increase in temperature observed. This suggested that the sorbent might be showing catalytic activity towards the water gas shift reaction under the conditions tested. Due to the high CO content in the feed gas (55%) and the limited amount of sorbent (10 g) used in this test, the sorbent became saturated very soon and it was very difficult to demonstrate & follow the SEWGS process concept. This suggested that the use of a simplified feed gas mixture is more suitable for our laboratory scale tests. 3.3. Study of the catalytic activity of the sorbent towards the WGS reaction Based on the results obtained before, the catalytic activity of the sorbent towards the WGS reaction was studied using a simplified gas mixture of 5%CO/N2 in the temperature range of 250 � Ce400 � C. Ten grams of sorbent were tested at 15 bar and using a 50% v/v of steam. Fig. 4 shows the breakthrough curves obtained for CO and CO2 at the different temperatures studied. As can be seen in Fig. 4 only at 350 � C and 400 � C simultaneous breakthrough of CO and CO2 takes place and complete conversion of CO continues after the



Fig. 3. Performance of the adsorbent under different feed gas compositions: feed gas A: 4%CO2/N2; feed gas B: 4%CO2, 60%CO, 22%H2, N2 balance; T ¼ 300 � C, P ¼ 15 bar.



Fig. 4. CO and CO2 breakthrough curves at different adsorption temperatures. P ¼ 15 bar, feed gas composition: 5%CO/N2, 50% steam.



breakthrough of CO2. However, at temperatures lower than 350 � C, CO broke before CO2 and the final CO conversion reached after the breakthrough of CO2 was lower than 40%. Fig. 5 shows the evolution of CO conversion with time for the different temperatures studied. As can be seen in Fig. 5, as process temperature increased, the conversion of CO increased due to the catalytic activity of the sorbent and the kinetics of the WGS reaction. However, when looking at the CO2 capture capacity of the sorbent, as expected, a decrease in the calculated values was observed as temperature increased. According to the behavior observed it seems that depending on process temperature the catalytic and adsorptive processes that take place in this material might be opposite and independent. While high temperatures clearly favored the conversion of CO, they have a negative influence on CO2 capture capacity of the sorbent. This finding suggests that a compromise is needed for selecting an adequate temperature if the sorbent is being used in a SEWGS process. This material was able to provide a conversion of CO higher than 90% after breakthrough of CO2 at temperatures higher than 350 � C.



Fig. 5. Influence of sorption temperature on the catalytic activity of the sorbent towards the WGS reaction. P ¼ 15 bar; feed: 5%CO/N2; 50% v/v steam.



32



M. Maroño et al. / Applied Thermal Engineering 74 (2015) 28e35



So, as a first attempt, temperatures higher than 350 � C can be considered adequate for using this adsorbent in the hybrid adsorbent/catalyst system. 3.4. Cyclic performance of the sorbent As the process under study requires a stable cyclic behavior, the performance of the sorbent was studied at 300 � C and 350 � C doing up to four sorption-desorption cycles. Sorption steps were carried out at 15 bar using a feed gas which consisted of a mixture of 5%CO/ N2 and 50% v/v of steam and desorption steps were performed at 500 � C and 1 bar using N2. Fig. 6 shows the breakthrough curves obtained for both experiments. As it can be observed in Fig. 6, for the experiment performed at 350 � C, the catalytic activity of the sorbent started to decrease after the first sorptionedesorption cycle from more than 90% to less than 50%. This behavior suggested that deactivation of the sorbent might be occurring due to consecutive cycles. However, for the experiment performed at 300 � C, the decrease in catalytic activity of the sorbent with cycles is less pronounced, reaching values of conversion of CO around 25%. When looking at the CO2 capture capacity of the sorbent, a similar behavior can be observed at both temperatures. Although, as expected, the sorbent showed a higher CO2 capture capacity at 300 � C, only an initial loss in capture capacity around 20% was measured after the first sorption cycle at both temperatures and the CO2 capture capacity remained constant for the other three adsorption cycles in both cases. In Fig. 6 it can also be observed that H2 is partially adsorbed in the sorbent and this effect is more noticeable at 300 � C lowering the production of H2. 3.5. Performance of the hybrid system adsorbent/catalyst at laboratory scale As mentioned above in this paper, the hybrid system designed in this work consisted in a homogeneous mixture of adsorbent and catalyst, and for the tests performed at laboratory scale they were crushed and sieved to a common size between 1.6 and 2.3 mm. The advantages of combining a WGS catalyst and a high temperature CO2 sorbent focus on the process intensification concept, concentrating in a single reactor both processes (reaction and adsorption) and on the enhancing effect that the presence of the adsorbent has on the catalytic activity of the catalyst.



Fig. 6. Cyclic performance of the sorbent at 300 � C and 350 � C. P ¼ 15 bar, 50% steam, feed gas composition: 5%CO/N2.



One of the first operating parameters that must be defined for the hybrid system is process temperature. If we look at the individual performance of the catalyst and the sorbent under the proposed operating conditions we found that as reported in our previous work [14] the catalyst started to show catalytic activity towards the WGS reaction at temperatures higher than 300 � C and only at temperatures above 350 � C the conversion of CO can be considered significant. On the other hand, according to the results obtained in this work, the sorbent showed higher CO2 sorption capacities at temperatures lower than 300 � C and catalytic activity towards the WGS reaction at temperatures higher that 350 � C. This behavior suggested that for the use of these two solids in the hybrid adsorbent/catalyst system proposed, process temperatures in the range of 300e350 � C would be suitable to be able to see the effect of the presence of the sorbent on the conversion of CO (sorptionenhanced reaction effect). Another operating parameter that needs to be defined for the hybrid system is Vads/Vcat, that is, the relative volume of adsorbent (ads) and catalyst (cat) in the solid mixture. The adsorbentecatalyst mixture used in the proposed hybrid system should contain enough catalyst to provide sufficient conversion of CO and enough adsorbent to capture as much CO2 as possible at the selected process temperature. In our case we used two different volume ratios Vads/Vcat (5 and 10) and we performed the tests at temperatures in the range of 300e350 � C using the same excess of steam. Fig. 7 shows the effect of the volume ratio adsorbent catalyst (Vads/Vcat) on the performance of the proposed hybrid system. All these tests were performed at 300 � C, 15 bar, water contents of 50% steam v/v and a feed gas mixture of 5%CO/N2. The results obtained when only adsorbent is used (Fig. 7(a)) have been included for comparison. In this test, breakthrough of CO took place before the saturation of the sorbent and a final stable CO conversion of 32.5% was reached. When a Vads/Vcat ¼ 5 was used complete conversion of CO occurred even after the breakthrough of CO2 (see Fig. 7(b)). In order to investigate the minimum amount of catalyst required to obtain a simultaneous breakthrough of CO and CO2 a volume ratio Vads/Vcat ¼ 10 was used. As can be seen in Fig. 7(c) complete conversion of CO was obtained even after the breakthrough of CO2 so this value was considered adequate for the proposed hybrid system. Another advantage of using this high volume ratio Vads/Vcat is that as more adsorbent is placed in the reactor the total amount of CO2 that can be captured is also higher.



Fig. 7. Effect of the volume ratio adsorbent catalyst (Vads/Vcat) on the performance of the hybrid system. (a) Only catalyst, (b) Vads/Vcat ¼ 5; (c) Vads/Vcat ¼ 10. Feed gas composition: 5%CO/N2; temperature ¼ 300 � C; P ¼ 15 bar, 50% v/v steam.



33



M. Maroño et al. / Applied Thermal Engineering 74 (2015) 28e35



Then, based on the results obtained for the first sorptiondesorption cycle showed in Fig. 7, the cyclic behavior of the hybrid system catalyst/adsorbent with a Vads/Vcat ¼ 10 was studied in up to four sorption desorption cycles. Adsorption was performed at 300 � C, 50% steam and 15 bar and desorption was carried out using N2 at 500 � C and atmospheric pressure. Fig. 8 shows the results obtained. While for the first sorption step complete conversion of CO was obtained and breakthrough of CO and CO2 took place simultaneously, for the following three sorption-desorption cycles CO broke always before the CO2 providing a stable CO conversion of approximately 40%. If we compare this final stable catalytic activity with the catalytic activity showed by the catalyst and the sorbent individually under the same operating conditions it is interesting to note that it is approximately the sum of both contributions. Table 1 summarizes the final stable CO conversion reached and CO2 capture capacities obtained for the hybrid system sorbent/ catalyst in comparison with the values obtained for the catalyst and the sorbent individually when tested under the same operating conditions (pressure, temperature and steam contents). The amount of CO2 captured by unit of time was calculated from the breakthrough curves as the difference between the moles converted of CO and the number of moles of CO2 in the exit gas stream at each time. The CO2 capture capacity was calculated by integration of the number of moles of CO2 captured until the exit gas contained 1% v/v CO2. The conversion of CO was calculated according to Equation (1) as follows:



XCO ¼



�



½CO�i � ½CO�f



�. ½CO�i



(1)



[CO]i ¼ Molar concentration of CO in the feed gas in standard conditions (298 K, 101 kPa) [CO]f ¼ Molar concentration of CO in the exit gas stream in standard conditions (298 K, 101 kPa) It is worth mentioning that at the working pressure used in these tests (15 bar) hydrogen was partially adsorbed in the solid during the sorption steps. This was corroborated by the appearance of a desorption peak during the desorption step (data not showed) which corresponded to desorption of H2 and caused the low values of H2 obtained.



Table 1 CO conversion and CO2 capture capacity for different materials and system studied in this work: catalyst, adsorbent and the hybrid catalyst/adsorbent system proposed at different temperatures. P ¼ 15 bar, Feed gas composition: 5%CO/N2 þ 50% steam. Materials and hybrid system



250 � C XCO



Mol/ kg



300 � C XCO



Mol/ Kg



XCO



350 � C Mol/ kg



400 � C XCO



Mol/ kg



Catalyst Adsorbent (1st cycle) Adsorbent (4th cycle) Hybrid System Vads/Vcat ¼ 10 (4th cycle)



2.9 25.7



e 1.19



9.8 32.5



e 0.99



40 98.8



e 0.88



82 100



e 0.77



23



0.81



45



0.64



41.5



0.74



As can be seen in Table 1 the adsorbent used in this work showed a net CO2 capture capacity between 0.8 and 1.2 mol/kg for the first sorption cycle under the conditions tested loosing approximately 20% of the net CO2 capacity after the first cycle but maintaining this value for at least four cycles. A more drastic loss in catalytic activity occurred at both temperatures studied (300 and 350 � C) which after the fourth cycle resulted reduced in more than 50% for the test performed at 350 � C. While comparable conversion of CO was obtained for the adsorbent alone when tested at 350 � C and for the hybrid system adsorbentecatalyst when tested at 300 � C, the CO2 capture capacity obtained for the hybrid system at 300 � C was higher than that obtained for the adsorbent alone at 350 � C. 3.6. Performance of the hybrid system adsorbentecatalyst at bench scale Based on the results obtained at laboratory scale, the performance of the hybrid adsorbentecatalyst system with a Vads/ Vcat ¼ 10 was tested at bench scale under similar operating conditions in order to investigate possible differences in the performance of the system during scaling-up of the SEWGS process. For these tests a feed gas composition of 10%CO/N2 þ 50% v/v steam was used. The experiments were performed at 15 bar and 300 � C. Fig. 9 shows the breakthrough curves and temperature profiles obtained after three sorption-desorption cycles. Fig. 9 shows a completely different adsorption profile to that obtained under isothermal conditions at laboratory scale. As it can be observed, simultaneous breakthrough of CO and CO2 took place during the sorption step. This was due to the process temperatures reached during the sorption step under adiabatic conditions (between 350 � C and 400 � C) due to the exothermicity of the WGS reaction. During this step two exothermic peaks were observed: one due to the adsorption of water and another one due to the WGS reaction. Besides that, desorption of H2 and CO was observed during the regeneration of the sorbent at 500 � C confirming that both components were simultaneously adsorbed together with CO2 during the sorption step. It is interesting to note that desorption of CO2 took place between 400 � C and 500 � C which confirms that a regeneration temperature of 500 � C will guarantee that almost all CO2 has been released during the desorption step. 4. Conclusions



Fig. 8. Cyclic behavior of the binary system adsorbent/catalyst for a volume ratio Vads/ Vcat ¼ 10. P ¼ 15 bar, T ¼ 300 � C, feed gas composition: 5%CO/N2 þ 50% steam.



In this work the performance of a hybrid system adsorbente catalyst has been studied in terms of CO conversion and CO2 capture capacity. The hybrid system proposed consisted of the combination of a high temperature FeeCr WGS catalyst and a K-doped



34



M. Maroño et al. / Applied Thermal Engineering 74 (2015) 28e35



Fig. 9. Cyclic performance of the hybrid adsorbentecatalyst system at bench scale. Feed gas: 10%CO/N2, P ¼ 15 bar, 50% steam.



hydrotalcite based sorbent selected in previous works. From the results obtained during this work the following conclusions can be drawn: The sorbent is catalytically active towards the WGS for the range of temperatures of interest (300e400 � C) and this catalytic activity increased with increasing temperature. However, cyclic performance of the sorbent alone (adsorptioneregeneration cycles), showed a rapid decrease in the catalytic activity and this effect was more pronounced at high temperatures. As expected, the sorbent used in this work provided higher CO2 capture capacities at low temperatures and under cyclic operation (adsorption-regeneration) only a loss of 20% in capacity was observed after the first adsorption cycle independently of the temperature. The hybrid system adsorbentecatalyst provides clear advantages over the catalyst and the sorbent alone. In our experiments we found that besides providing the necessary catalytic activity to the system the presence of the catalyst gives stability to the hybrid system in terms of stable final CO conversion. Although low temperatures favored the sorption capacity of the sorbent considered in this work, the tests performed at bench scale have confirmed that due to the catalytic activity of the sorbent, higher temperatures, in the range of 350 � Ce400 � C seemed to improve the performance of the hybrid adsorbentecatalyst system reducing the amount of catalyst needed but at the expense of reducing the CO2 capture capacity of the sorbent. A compromise is needed between the volume ratio Vads/Vcat and temperature. Acknowledgements This research is financed by the Spanish Ministry of Science and Innovation through the CAPHIGAS project ENE2009-08002 and by the European Community through the FECUNDUS project RFCS-CT2010-00009. Authors sincerely acknowledge the funding received. References [1] E.R. Van Selow, P.D. Cobden, A. Wright, R.W. Van den Brink, D. Jansen, Improved sorbent for the sorption-enhanced water-gas shift process, Energy Procedia 4 (2011) 1090e1095.



[2] Z. Yong, E. Rodriges, Hydrotalcite-like compounds as adsorbents for carbon dioxide, Energy Convers. Manag. 43 (2002) 1865e1876. [3] Y. Ding, E. Alpay, High temperature recovery of CO2 from flue gases using hydrotalcite adsorbent, Trans. IChemE 79B (2001) 45e51. [4] K.B. Lee, M.G. Beaver, H.S. Caram, S. Sircar, Reversible chemisorption of carbon dioxide: simultaneous production of fuel-cell grade H2 and compressed CO2 from synthesis gas, Adsorption 13 (2007) 385e397. [5] R.J. Allam, R. Chiang, J.R. Hufton, P. Middleton, E.L. Weist, V. White, Development of the sorption enhanced water gas shift process, in: D.C. Thomas, S.M. Benson (Eds.), Carbon Dioxide Capture for Storage in Deep Geologic Formations, vol. 1Elsevier Ltd, New York, 2005, pp. 227e256. [6] J.R. Hufton, S. Mayorga, S. Sircar, Sorption-enhanced reaction process for hydrogen production, Separations 45 (2) (1999) 248e256. [7] A. Iwan, A. Lapkin, Development of CO2 adsorbents and reagents for sorption-enhanced methane steam reforming, in: Proceedings International Hydrogen Energy Congress and Exhibition IHEC 2005, Istanbul, Turkey, 13e 15 July 2005. [8] E.L. Oliveira, C.A. Grande, A.E. Rodrigues, CO2 sorption on hydrotalcite and alkali-modified (K and Cs) hydrotalcites at high temperatures, Sep. Purif. Technol. 62 (2008) 137e147. [9] A.D. Ebner, S.P. Reynolds, J. Ritter, Understanding the adsorption and desorption behavior of CO2 on a K-promoted hydrotalcite-like compound (HTLc) through nonequilibrium dynamic isotherms, Ind. Eng. Chem. Res. 45 (2006) 6387e6392. [10] S. Walspurger, L. Boels, P.D. Cobden, G.D. Elzinga, W.G. Haije, R.W. van den Brink, The crucial role of the Kþ-aluminium oxide interaction in Kþ-promoted alumina- and hydrotalcite-based materials for CO2 sorption at high temperatures, ChemSusChem 1 (2008) 643e650. [11] S.P. Reynolds, A.D. Ebner, J.A. Ritter, Carbon dioxide capture from flue gas by pressure swing adsorption at high temperature using a K-promoted HTLc: effects of mass transfer on the process performance, Environ. Prog. 25 (4) (2006) 334e342. [12] K.B. Lee, A. Verdooren, H.S. Caram, S. Sircar, Chemisorption of carbon dioxide on potassium-carbonate promoted hydrotalcite, J. Colloid Interface Sci. 308 (2007) 30e39. [13] E. Xue, M. Okeeffee, J.R.H. Ross, Water gas shift conversion using a feed with low steam to carbon monoxide ratio and containing sulphur, Catal. Today 30 (1996) 107e118. [14] J.M. Sanchez, M. Maroño, D. Cillero, L. Montenegro, E. Ruiz, Laboratory e and bench scale studies of a sweet water-gas-shift catalyst for H2 and CO2 production in pre-combustion CO2 capture, Fuel (2012), http://dx.doi.org/ 10.1016/j.fue.2012.02.060. [15] M. Maroño, J.M. Sanchez, E. Ruiz, Hydrogen-rich gas production from oxygen pressurized gasification of biomass using a FeeCr water gas shift catalyst, Int. J. Hydrogen Energy 35 (2010) 37e45. [16] J.M. Sánchez, E. Ruiz, J. Otero, Selective removal of hydrogen sulfide from gaseous streams using a Zinc-based sorbent, Chem. Eng. Sci. 60 (2005) 2977e 2989. [17] F. Peña, P. Casero, J. Trujillo, in: Proceedings of the 5th International Freiberg Conference on IGCC & XtL, Leipzig, 21e24 May 2012. Technologies, Book of Abstract page 62.



M. Maroño et al. / Applied Thermal Engineering 74 (2015) 28e35 [18] M. Maroño, Y. Torreiro, L. Montenegro, J.M. Sánchez, Lab-scale tests of different materials for the selection of suitable sorbents for CO2 capture with H2 production in IGCC processes, Fuel 116 (2014) 861e870. [19] S.C. Lee, B.Y. Choi, C.K. Ryu, Y.S. Ahn, T.J. Lee, J.C. Kim, The effect of water on the activation and the CO2 capture capacities of alkali metal-based sorbents, Korean J. Chem. Eng. 23 (3) (2006) 374e379.



35



[20] M.K. Ram Reddy, Z.P. Xu, G.Q. Lu, J.C. Diniz de Costa, Influence of water on high-temperature CO2 capture using layered double hydroxide derivatives, Ind. Eng. Chem. Res. 47 (2008) 2630e2635. [21] M. Maroño, Y. Torreiro, L. Gutierrez, Influence of steam partial pressures in the CO2 capture capacity of K-doped hydrotalcite-based sorbents for their application to SEWGS processes, Int. J. Greenhouse Gas Control 14 (2013) 183e192.



Applied Thermal Engineering 74 (2015) 36e46



Contents lists available at ScienceDirect



Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng



A complete transport validated model on a zeolite membrane for carbon dioxide permeance and capture Evangelos I. Gkanas a, c, *, Theodore A. Steriotis b, Athanasios K. Stubos a, Peter Myler d, Sofoklis S. Makridis a, c a



Institute of Nuclear and Radiological Sciences and Technology, Energy and Safety (INRASTES), ‘Demokritos’, Aghia Paraskevi, 15310 Athens, Greece Institute of Advanced Materials, Physicochemical Processes, Nanotechnology & Microsystems, NCSR “Demokritos”, Aghia Paraskevi, Athens 15310, Greece Institute for Renewable Energy and Environmental Technologies, University of Bolton, Deane Road, Bolton BL3 5AB, UK d Centre for Advanced Performance Engineering, University of Bolton, Deane Road, Bolton BL3 5AB, UK b c



g r a p h i c a l a b s t r a c t



a r t i c l e i n f o



a b s t r a c t



Article history: Received 22 August 2013 Accepted 1 February 2014 Available online 1 March 2014



The CO2 emissions from major industries can cause serious global environment problems and their mitigation is urgently needed. The use of zeolite membranes is a very efficient way in order to capture CO2 from some flue gases. Zeolite membranes are porous crystalline materials with pores of a consistent size and these pores are generally of molecular size 0.3 to 1.3 nm, and enable high selectivity and reduced energy requirements in industrial separation applications. Further, zeolites are thermally stable and have known surface properties. Separation in zeolites is mainly based on dissimilarity of diffusivities, favored absorption between the components and/or molecular sieving effects. The present work is aimed at developing a simulation model for the CO2 transport through a zeolite membrane and estimate the diffusion phenomenon through a very thin membrane of 150 nm in a Wicke eKallenbach cell. This apparatus has been modeled with COMSOL Multiphysics software. The gas in the retentate gas chamber is CO2 and the inert gas is argon. The MaxwelleStefan surface equations used in order to calculate the velocity gradients inside the zeolite membrane and in order to solve the velocity profile within the permeate and retentate gas chamber, the incompressible NaviereStokes equations were solved. Finally, the mass balance equation for both gases was solved with the mass balance differential equations. Validation of the model has been obtained at low and high temperatures suggesting that higher the temperature the more beneficial the outcome.  2014 Elsevier Ltd. All rights reserved.



Keywords: Zeolite membrane CO2 permeation WickeeKallenbach cell MaxwelleStefan diffusivity Quasi-chemical approach



* Corresponding author. Materials, Mechanics and Structures Research Division, Faculty of Engineering, University of Nottingham, Nottingham NG7 2RD, UK. E-mail addresses: [email protected], [email protected] (E.I. Gkanas). http://dx.doi.org/10.1016/j.applthermaleng.2014.02.006 1359-4311/ 2014 Elsevier Ltd. All rights reserved.



E.I. Gkanas et al. / Applied Thermal Engineering 74 (2015) 36e46



Nomenclature R T ui Dij Di qi Ni pi Bij c d n



gas constant, 8.314 J/mol/K temperature, K velocity of species-i with respect to zeolite, m/s MaxwelleStefan diffusivity describing interchange between i and j, m2/s MaxwelleStefan diffusivity for species i, m2 s1 loading of component i in zeolite, molecules per unit cell or mol kg1 molecular flux of species-i, molecules/m2/s or (mol/m2/s) partial pressure of species-i, Pa elements of matrix [B], defined in Eq (10), s/m2 concentration of species, mol/m3 density of CO2, kg/m3 dynamic viscosity, Pa s



1. Introduction There is a growing consensus among the scientific community that the rising atmospheric levels of CO2 as a result of human activities, such as emissions from major industries (power generation, steel and cement industries) [1] are that the origin of the warming effect of the climate [2]. Membrane processes appear to be an attractive option to carry out gas separations in terms of their lower environmental impact and energy cost, compared to more conventional separation technologies. Furthermore, the modular nature of membranes constitutes a positive input [3]. Recently, great efforts of novel synthetic routes on membranes for CO2 removal have been reported [4] mainly based on fabricated cross-linked poly (ethylene-oxide) (PEO) membranes for H2 purification and CO2 capture [5] or composite polyetheramine (PEA)epolyhedral oligomeric silsesquioxane (POSS) membranes for CO2/H2 and CO2/ N2 separation [6]. Further, nanohybrid membranes have been investigated with a CO2/H2 selectivity of 11 at 35  C at 3.5 atm [7]. A wide variety of micro- and meso-porous materials are of potential use in separation applications such as CO2 capture [8e10]. Examples of microporous materials include zeolites (crystalline aluminosilicates) among other materials such as metaleorganic frameworks, covalent organic frameworks and zeolitic imidazolate frameworks. Zeolites are inorganic crystalline structures with uniform pores of molecular dimensions. Different pore sizes and composition of zeolites have been used to prepare membranes, and zeolite membranes with different shapes have been investigated to separate CO2 [11e13]. These materials have unique properties such as a singular pore diameter, well-defined surface properties and high thermal stability making them invaluable in many technical applications [14,15]. In the separation of a mixture by a zeolite membrane the selectivity is a function of the sorption and diffusion and the relevant parameters cannot be simply predicted on the basis of molecular size and shape alone. Key parameters for transport in zeolites by surface and micropore diffusion include temperature, pressure, molecular weight, kinetic molecular diameter, pore diameter, heat of gas adsorption, thermal activation energies for both surface and microporous diffusion. Diffusion is an activated process depending on DHads, on the molecular size of the adsorbate and the fractional coverage [9]. Adsorption is an exothermic, non-activated process which is driven by fugacity. It is a competitive phenomenon [56]. Zeolites can be applied as powders, pellets and as thin films grown on inert support membranes with a larger pore size [16]. In



F R



37



volume force, force per unit volume reaction rate, 1/s



Greek letters chemical potential, J/mol fractional loading of component i, dimensionless density of membrane, number of unit cells per m3 or kg/m3 Gij elements of the matrix of the thermodynamic correction factor [G], dimensionless V gradient operator 2 vector Laplacian V P permeance, mol/m2/s/Pa



m qi r



Superscripts sat referring to saturation loading s referring to surface diffusion



between, numerous zeolite membrane preparations are reported and substantial progress can be stated, examples are the preparations of zeolite membranes of types LTA [17], FAU [18], CHA [9], DDR [20] and mixed tetrahedraleoctahedral oxides [21,22]. Since the separation on these membranes is based on competitive adsorption, the selectivities were found to be low. Most often the MFI type membrane was studied [23e25]. Recently, Tsapatsis et al. [26] have prepared ab-oriented MFI silicalite-1 membrane, and they further showed the performance of h0h and c-oriented silicalite-1 layers at different temperatures [27,28]. For a silicalite-1 membrane a selectivity of about 10 was obtained [29] and also the CO2 permeation from pressurized feeds on a silicalite-1 membrane on different supports has been reported [30]. DDR (0.36 nm  0.44 nm) and SAPO-34 (0.38 nm) have pores that are similar in size to CH4 (0.34 nm) but larger than CO2 (33 nm) [9]. It can be expected that these membranes show high CO2/CH4 selectivities due to molecular sieving. Very efficient SAPO-34 membranes were synthesized by insitu crystallization on tubular support [31]. It has also been reported that SAPO-34 membranes can separate CO2 from CH4 in higher efficiency at lower temperatures with a selectivity of 270 at 20  C [32]. Recently, the tuning of CO2 permeation through a SAPO-34 by ion exchange was reported [33,34]. Hasegawa et al. [35] studied Y-type zeolite membranes and found that the membranes synthesized by hydrothermal process on an a-alumina support showed separation factor of CO2/N2 149 at 35  C. For the applications described above, migration or diffusion of sorption molecules through the pores and cages within the crystal structure of the zeolite membrane is dominant. Configurationally diffusion is the term coined to describe diffusion in zeolites and it is characterized by very small diffusivities (108 to 1014) m2/s [36] with a strong dependence on the size and shape of the quest molecules [37,38] and high activation energy [39]. Further, it is characterized by very strong concentration dependence [40]. The measurement of the diffusivity in zeolites can be obtained by both macroscopic and microscopic methods. Diffusion of components, especially gases, through porous materials can be experimentally studied with the use of WickeeKallenbach (WeK) cells [41,42]. These experimental devices consist of two flow-through components separated by a membrane of porous material through which components can penetrate. A steady gas stream with certain composition flows through the first compartment, while another stream of usually inert gas flows through the second compartment as a sweep gas. Mass transport parameters of component in a porous material are frequently determined from a



38



E.I. Gkanas et al. / Applied Thermal Engineering 74 (2015) 36e46



transport model developed to explain experimental observations with a WeK cell. Different models have been proposed to describe various transport mechanisms in various inorganic membrane materials [43,44]. Theoretical approaches for modeling the diffusion in zeolites and/or other microporous structures fall into two different categories. A kinetic approach and an approach based on irreversible thermodynamics. The former is based on random walk models and/ or transition state theory appropriately modified to account for several additional phenomena such as multilayer adsorption, surface heterogeneity and energy barriers [45e47]. The irreversible thermodynamic approach considers the chemical potential gradient as the driving force for diffusion [48]. Multicomponent interactions occur through competitive equilibrium and/or diffusional sorbateesorbate interactions. In this case, the driving force exerted on any particular species is balanced by the friction this species experiences with the other species present in the mixture and can be accurately described by the generalized MaxwelleStefan model for diffusion [49]. In some cases, the membranes can be very thin and in some other cases the membranes can contain large pore cracks and defects. In other cases, permeation flux would be relatively high and if the gas flow rates through the cell are insufficient the component concentration in the WickeeKallenbach cell may not be homogeneous. Therefore, a need for a simulation model is essential for the accuracy of the transport parameters obtained from the model. According to the literature, there are several models proposed for the CO2 capture through zeolite membrane [10,16,19,49,51]. The model is based on the MaxwelleStefan formulation using a WickeeKallenbach cell geometry which is an experimental common geometry but has not been extensively studied by simulation. Perdana et al. [52] presented very nice results. In the current work, a simulation study for CO2 permeation and capture through a zeolite membrane in a WickeeKallenbach geometry has been obtained. A comparison about different scenarios about MeS diffusivity terms has been performed followed by a transient analysis of the permeation. The model was validated with already published experimental results and can be used in order to calculate some crucial parameters such as the MeS terms for some other gases such as N2 or H2 which are not clearly defined in the literature. 2. MaxwelleStefan theory for diffusion through zeolite membranes



i ¼ 1; 2; .n



qi ¼



(1)



jsi



where �7mi is the force acting on species i tending to move along the surface with velocity ui. The first term on the right-hand side describes the friction exerted by adsorbate j on the surface motion



qi qsat



(2)



According to that fact, for different molecules different amounts are needed to obtain similar levels of fractional occupancies. The fractional occupancies are converted into fluxes using Eq. (3).



Ni ¼ rqsat qi ui ¼ rqi ui



(3)



In this case, the ideal adsorbed solution (IAS) theory as proposed by Myers and Prausnitz (1965) will be used, which is thermodynamically consistent and can be applied using single component isotherms. Dropping the superscripts for the surface diffusivities in the GMS expression for convenience, multiplication of both sides by qi/RT and application of Eq. (2), Eq. (1) can be rewritten as:



q



� i Vmi ¼ RT



n X



qi qj



j¼1 jsi



n X ui � uj qi ui ui � uj qi ui þ ¼ qi qj sat sat þ sat Dij Di q qj Dij qi Di i j¼1



(4)



jsi



Using the definition of fluxes, Eq. (4) can be written as:



�r



qi RT



Vmi ¼



n X qj Ni � qi Nj



j¼1



qsat qsat Dij i j



þ



Ni ; qsat Di i



i ¼ 1; 2.n



(5)



jsi



The gradient of the thermodynamic potential can be expressed by terms of thermodynamic factors [47]:



�r



The generalized MaxwelleStefan (GMS) equations have successfully been applied to many systems to describe diffusive transport phenomena in multicomponent mixtures and single component species [50]. These models mainly based on the principle that in order to cause relative motion between individual species in a mixture, a driving force has to be exerted on each of the individual species. This driving force is balanced by the friction these species experiences with the other species in the mixture and the friction between the species and the surface of the membrane. Krishna et al. [50] described the diffusion through the membrane of adsorbed molecules starting from the equation for an n-component mixture: n X u �u u qj i s j þ RT is ; �Vmi ¼ RT D D ij i j¼1



of species i, each moving with velocities uj and ui with respect to the surface, respectively. The second term deals with the friction between the species i and the surface.Dsij andDsi represent the corresponding MaxwelleStefan diffusivities and qj is the fractional surface occupancy. The GMS formulation Eq. (1) has been applied successfully to describe transient uptake in zeolites and carbon molecular sieves, and in zeolitic membrane permeation. Generally, a multicomponent Langmuir-type adsorption model is used to describe the fractional occupancies. For thermodynamic consistency, however the saturation loading for all species must be equal in the multicomponent Langmuir model hence the fractional occupancies can be defined as:



qi RT



Vmi ¼



n X



qi vpi pi vqj



Gij Vqi ; Gij h



j¼1



(6)



where Eq. (6) represents a thermodynamic factor which can be determined by the adsorption isotherm, chosen to relate the surface coverage qi to the partial pressure pi. In the current work the adsorption is described by the extended Langmuir model, Eq. (2). Eqs. (5) and (6) can be cast in a matrixevector relation:



� ��1 �r½G�ðVqÞ ¼ ½B� qsat ðNÞ



(7)



where [qsat]�1 is a diagonal matrix of saturation loadings and the elements of [B] are given by the following equations:



Bii ¼



n X qi 1 þ Di Dij j ¼ 1 isj



(8)



and



Bij ¼ �



qj Dij



(9)



E.I. Gkanas et al. / Applied Thermal Engineering 74 (2015) 36e46



39



Fig. 1. Geometry of the WickeeKallenbach cell.



The solution of Eq. (7) for the diffusion fluxes is the:



  ðNÞ ¼ �r qsat ½B��1 ½G�ðVqÞ



(10)



For pure species, which in the current study is CO2 Eq. (10) can be written as follows:



rqsat q i Di V i 1 � qi



Ni ¼ �



(11)



3. Model description 3.1. Geometry of the model A three-dimensional model has been developed to resolve the flow pattern and CO2 concentration in the WeK compartment. The development of a three-dimensional model is very important in this case because in the introduction of the partial differential equations describing the CO2 flux in COMSOL Multiphysics is performed with matrices as already discussed from Eq. (10). With a threedimensional model a 3 � 3 matrix describing the CO2 flux can be used in order to be able to calculate the flux in all the possible directions within the membrane. The zeolite membrane is very thin (approximately 150 nm thick). The cell has cylindrical shape with 19 mm diameter and consist of a retentate gas chamber where the gas CO2 enters the chamber, and a permeate gas chamber where an inert gas (in this case argon) enters the system. The two chambers are separated by a cylindrical, solid zeolite membrane, usually held within a support system. In order to simplify the current problem some assumptions were taken into account. These assumptions are: 1) The transport of the absorbing components through the zeolite membrane occurs due to surface diffusion, described by the generalized MeS model. 2) Additional contributions such as gas translation are negligible. 3) The pressure drop along each compartment is assumed to be negligible. 4) The deformation of the zeolite membrane under high pressure is negligible.



5) No support layer is taken into account [52]. 6) The sweep gas does not experience counter diffusion through the zeolite membrane. Both gas chambers thickness are 0.3 mm. This counter-current system feeds a concentrated gas-flow into the retentate gas chamber and the chemical species reach the zeolite membrane. A portion of the species diffuses through the zeolite and is removed from the permeate chamber by feeding an inert sweep. The geometry of the model is illustrated in Fig. 1. The gas flowing in the compartment was modeled using the incompressible NaviereStokes equations, assuming that the gas flowing in the compartment is in the laminar flow regime. The general equation that defines the incompressible flow is given by:



d



vu � nV2 u þ rðuVÞu þ Vp ¼ F vt



(12)



Furthermore, the mass transport in the compartment is due to both convection and diffusion. The momentum balance equation in the retentate compartment is given by:



vc þ Vð�DVcÞ ¼ R � uVc vt



(13)



3.2. Boundary conditions The appropriate boundary conditions for the solution of the problem are described by the following set of equations. 3.2.1. MaxwelleStefan diffusion through the membrane The conditions referred to the zeolite membrane domain boundaries, found between permeate and retentate gas chambers are: the Neumann boundary condition which refers to the edges where no flux occurs and the Dirichlet boundary condition, which are used at the interface between the zeolite membrane surface and the respective permeate and retentate gas chambers. The Langmuir isotherm is used in the current study in order to calculate the surface coverage of sites at the interface between the gas



40



E.I. Gkanas et al. / Applied Thermal Engineering 74 (2015) 36e46



chambers and the solid zeolite membrane. Thus this relationship is defined in the Dirichlet boundary conditions. 3.2.2. Mass balance The boundary conditions between the gas and the walls of the chamber have been set as:



! nN ¼ 0; N ¼ �DVcþc u ðInsulation=SymmetryconditionÞ



(14)



The boundary condition between the gas and the membrane interface is given by:



! �D$Vc þ c u ¼ N0 ;



(15) 2



where N0 is the inward flux (mol/m s) 4. Results and discussion 4.1. Model verification In order to validate the model which is proposed in the current study, a comparison between the results extracted from the 3.0



273 K



a)



298 K



323 K



2.5



Amount Adsorbed (mol/kg)



348 K



2.0



1.5



1.0



0.5



Himeno et al. [54] Proposed Model



0.0 0



500



1000



1500



2000



2500



3000



3500



4000



simulation runs based on the model and experimental results obtained from already published data were performed. As other groups have published their simulation results [51,52] without taking into account the effect of the support resistance on permeation and ultimately analyze the permeation behavior using adsorption and occupancy dependent diffusion within the membrane, in the current study the membrane is treated without support. Himeno et al. [53] measured adsorption isotherms of carbon dioxide at temperatures of 273, 298, 323, 348 K, at a pressure range between 0 and 3500 kPa. In order to compare the simulation results with the experimental data, simulation runs were performed at the same temperatures (273, 298, 323, 348 K). For each temperature, results of CO2 concentration (mol/kg) obtained for individual pressures with a pressure step of 200 kPa, in order to cover the range of 0e3500 kPa. Finally, the results collected from these simulation runs were compared with the experimental results by Himeno et al. in Fig. 2a. A good agreement between the results is obtained. Further, for lower temperatures, the same process was performed in order to compare the simulation results with the experimental data obtained by van der Bergh et al. [54]. Van der Bergh et al. obtained their results at temperatures of 298, 273, 252 and 195 K in a pressure range between 0 and 100 kPa. Again, for each temperature, simulation runs were performed for pressure steps of 10 kPa to cover the pressure range 0e100 kPa. The comparison of the results is presented in Fig. 2b. According to these data it is obvious that for the temperatures of 298 and 273 K the results of the simulation runs are in good agreement with the experimental data, while for the lower temperatures there is a small deviation between the results, but the shape of the isotherms is almost the same. This could be due to the fact that the support of the membrane might play a major role of thermal insulator at lower temperatures. Further, in order to ensure that the proposed model is able to describe the permeation of CO2 through every zeolite membrane, a comparison of the simulation performance of three different membranes has been performed with already published experimental results. The three different membranes chosen were: DDR3 zeolite membrane, 5A zeolite membrane and 13X zeolite membrane. For each type of membrane, an adsorption isotherm was measured at constant temperature and compared to experimental data from van den Bergh et al. [54] for the DDR3 membrane,



Pressure (kPa)



7



4.0



b)



Amount Adsorbed (mol/kg)



3.0



252 K



2.5 2.0



273 K



1.5



298 K 1.0 0.5



Simulation Results van der Bergh et al. [55]



0.0 0



20



40



60



80



100



Pressure (kPa) Fig. 2. Comparison of experimental results extracted from recently published adsorption data by Himeno et al. and the simulation results extracted at same temperatures with a pressure step of 200 kPa (a), and experimental adsorption data recently published by van der Bergh et al. and simulation results extracted at same temperatures with a pressure step of 10 kPa (b).



Adsorbed amount (mol/kg)



3.5



195 K



6



13X Zeolite Membrane (298K)



5 4 3



Z5 Zeolite Membrane (303K)



2 DDR3 Zeolite Membrane (273K) Experimental Results Simulation Results



1 0



0



20



40



60



80



100



Pressure (kPa) Fig. 3. Comparison of experimental results extracted from recently published adsorption data for three different types of zeolite membranes, DDR3 membrane by van den Bergh et al. [54], Z5 membrane by Liu et al. [2] and 13X membrane by Cavenati et al. [57] and the simulation results extracted at 273 K for the DDR3 membrane, 303 K for the Z5 membrane and 298 K for the 13X membrane.



41



E.I. Gkanas et al. / Applied Thermal Engineering 74 (2015) 36e46



3,0



Amount adorbed (mol/kg)



Liu et al. [2] for the Z5 membrane and Cavenati et al. [57] for the 13X membrane. The temperature for the DDR3 membrane was 273 K, for the Z5 membrane was 303K and for the 13X zeolite membrane 298 K. The results of these measurements are presented in Fig. 3. As extracted from Fig. 3, the data from the simulation runs



T=273 K



N2O



2,5



CO2



2,0 1,5



CH4



1,0 0,5 0,0



N2 0



20



40



60



80



100



120



Pressure (kPa) Fig. 5. Adsorption isotherms for carbon dioxide, nitrous oxide, methane and nitrogen in DDR3 zeolite membrane at 273 K.



are in very good agreement with the experimental, indicated that the proposed model is valid for all types of zeolite membranes. 4.2. CO2 permeation through the zeolite membrane For single-component diffusion, transport flux of the component through the membrane is described by Eq. (11). The term Di is referred as MeS surface diffusivity. In the current study, three different MeS diffusivity term scenarios are considered and compared. In the weak confinement scenario the MeS diffusivity acts independently of the fractional occupancy and is equal to the initial zero-loading diffusivity:



Di ¼ Di ð0Þ



(16)



In the strong confinement scenario, the MeS diffusivity presents linear dependence on the occupancy:



Di ¼ Di ð0Þð1 � qi Þ



(17)



Reed and Ehrlich [55] proposed a general occupancy dependence on the MeS diffusivity term:



Fig. 4. Transient response to an increase in feed pressure in a CO2 permeation system. Feed and permeate fluxes for the three different scenarios about the MeS diffusivities for feed pressures a) 10 kPa, b) 100 kPa and c) 1000 kPa.



Fig. 6. CO2 flux as function of the feed pressure at constant temperature 303 K.



42



E.I. Gkanas et al. / Applied Thermal Engineering 74 (2015) 36e46 Table 1 Langmuir parameters for the temperatures 303, 363 and 423 K. T (K)



qsat (mol/kg)



303 363 423



2.48 1.81 1.46



Fig. 7. Adsorption isotherms of carbon dioxide at 195, 252, 273 and 298 K.



Fig. 9. Comparison of the results extracted from simulation runs and the experimental results extracted from van der Broeke et al. [8] for the effect of pressure and temperature on the permeance.



ð1 þ 3 Þy�1 DðqÞ ¼ Dð0Þ  y 1 þ f3



(18)



where y is the coordination number (the maximum number of neighbors in the lattice cavity of the membrane) and the other parameters are given by:







f ¼ exp



3



Fig. 8. Effect of pressure and temperature on the permeation of carbon dioxide through the zeolite membrane.



¼



dE RT







ðg � 1 þ 2qÞf 2ð1 � qÞ



(19)



(20)



Fig. 10. Transient flux of CO2 at 303 K for three different feed pressures 10, 100 and 1000 kPa.



E.I. Gkanas et al. / Applied Thermal Engineering 74 (2015) 36e46



43



Fig. 11. Transient CO2 flux analysis across the z-axis of the zeolite geometry. Panel a shows the CO2 flux profile for feed pressure 10 kPa. Panel b shows the CO2 flux profile for feed pressure 100 kPa. Panel c shows the CO2 flux profile for feed pressure 1000 kPa and panel d shows the z-axis of the zeolite membrane geometry which is the perpendicular axis to the membrane.



g ¼



sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi � � 1 1 � 4qð1 � qÞ 1 � f



(21)



Fig. 4 shows the transient permeation of pure CO2 through the zeolite membrane for three different feed pressures (10, 100 and 1000 kPa) and the comparison of the three different MeS diffusivity scenarios described above. All the measurements were performed at temperature 298 K. The fluxes at the feed and the permeate sides are decreasing and increasing respectively until they reach steady state. As the feed pressure increases from 10 to 1000 kPa, there are larger differences among the transient fluxes for the three MeS diffusivity scenarios. However, even at the higher pressure of 1000 kPa the difference between the strong confinement scenario and the quasi-chemical approach is small and this is



a consequence of the close diffusivities of CO2 obtained from these two estimation methods. The same approach was performed by Lee [51] who also pointed that as the feed pressure is increasing, larger differences in the fluxes between the different scenarios are taking place, where the strong confinement scenario with the quasichemical approach, have almost the same behavior, even at high pressures. The main difference (which seems to be minor) between the strong confinement scenario and the quasi-chemical approach lies on the fact that the strong confinement scenario depends directly to the fractional occupancy which related with the number of the remaining free spaces for CO2 capture within the membrane with time, while the quasi-chemical approach depends on more parameters except the occupancy, such as the number of adjacent atoms near the available cavity which might affect the behavior of



44



E.I. Gkanas et al. / Applied Thermal Engineering 74 (2015) 36e46



the membrane. The weak confinement scenario on the other hand does not takes into account the fractional occupancy indicating that the MaxwelleStefan diffusivity term remains constant with time. This is probably the main reason that the results extracted from this scenario, doesn’t match to the results extracted from the other two scenarios at high pressures. The conclusion of the above results is that in high feed pressures the weak confinement scenario might bring wrong estimated MeS diffusivities while the strong and quasi-chemical approach seems to have better potential in describing the diffusion at higher pressures. At low pressures it seems that all three scenarios can describe with detail the CO2 diffusion through the zeolite membrane. The removal of CO2 is very important for applications such as to purify natural gas, reduce the amount of green gas emission from flue gas and collect methane from landfill gas. Further, industrial gas is a complex mixture containing gaseous hydrocarbons and non-hydrocarbon components. Fig. 5 presents a comparison of the adsorption data for CO2, CH4, N2O and N2 through a DDR3 zeolite membrane at 273 K for a pressure range from 0 to 100 kPa. The CO2 and the N2O isotherms are almost the same and this is something expected because some of the main parameters describing the permeation such as the saturation loading, the adsorption enthalpy, the pro-exponential constant are very close in these two gases. Further, CO2 and N2O can be considered as “highly absorbing gases” with respect to the other two gases. The obtained order is the following: CO2 z N2O > CH4>N2. 4.3. Effect of temperature and pressure on the CO2 permeation through the membrane The temperature effect on the diffusion through a zeolite membrane is very important and the temperature dependence of the permeance for a large number of gases has been well-studied. Fig. 6, presents the simulation results for the flux of CO2 through the membrane at a temperature of 303 K and a pressure range from 0 to 600 kPa at temperature 303 K. It is obvious from the results that the flux of CO2 through the zeolite membrane presents a non-linear dependence on the pressure. Fig. 7, shows the adsorption isotherm of carbon dioxide in the zeolite membrane at 298, 273, 252 and 195 K for a pressure range from 0 to 100 kPa. It is clearly seen from Fig. 6 that the isotherm changes from a non-linear to an almost linear shape if the temperature is increased. The reason of this behavior will be discussed later. When the mass transport through the zeolite membrane is described the flux and the permeance are typically used. The permeance is calculated by from a mass balance at steady state by using the pressure difference between the retentate and the permeate side. The permeance can be defined as:



P ¼



N



Dp



For the lower temperature 303 K both the flux and the permeance present a non-linear behavior on the feed pressure, but this behavior is changing for higher temperatures 363 and 423 K where the permeance for the temperature 423 K is almost constant with pressure. These results are in very good agreement with the equilibrium isotherms that presented in Fig. 7. For an almost linear isotherm the flux shows linear behavior and the permeance seems to be independent of the pressure. This can be explained as follows. The flux through a zeolite membrane is a function of the diffusivity of the component and the amount of the component adsorbed within the zeolite. Diffusion in zeolites is an activated process and the diffusivity increases with the temperature, while the adsorbed amount decreases with the temperature. When decreasing the temperature in zeolite reaches saturation and the decrease of diffusivity begins to dominate due to the asymptotic approach of adsorption saturation. 4.4. Transient analysis of CO2 permeation through the zeolite membrane For the transient analysis of the CO2 permeation through the membrane it is expected that the CO2 flux through the membrane initially will be increased and after some time the equilibrium situation will achieved. Further, as discussed in Chapter 4.2 the strong confinement scenario was taken into account in order to describe the MaxwelleStefan diffusivities. Fig. 10 presents the transient flux for CO2 at temperature 303 K. The feed pressures are 10, 100 and 1000 kPa respectively. From Fig. 10 is extracted that as the feed pressure increases the CO2 flux also reaches higher levels within the membrane. Further, the equilibrium time is low for all the three pressures ranging from 2 to 7 s and for the lower pressure of 10 kPa it seems that the equilibrium is reached faster than the higher pressures. Fig. 11 shows the transient profile for CO2 permeation flux through the membrane across the z-axis of the zeolite membrane geometry which is the perpendicular axis to the membrane as shown in panel d. The results showed that the flux profile has the same distribution for all the three feed pressures but the flux is higher as the feed pressure increases. The maximum flux value seems to be located in an area higher than the middle of the membrane. This distribution indicates that in the current geometry the flux of the CO2 through the membrane has a standard profile for all the feed pressures and



(22)



where Dp is the pressure difference of CO2 over the membrane. Sometimes permeance is better quantity to describe the mass transport, because it takes into amount the pressure difference due to some pressure variation problems that might occur due to the sweep gas diffusion mechanism within the membrane. The effect of the isotherm shape on the behavior of the permeation is illustrated in Fig. 8. Panel a presents the CO2 flux as a function of pressure and panel b the permeance as a function of pressure. The parameters used for these temperatures are presented in Table 1. The simulation results for both the permeance and flux are in great agreement with the experimental results presented by Van der Broeke et al. [12], and the comparison is presented in Fig. 9.



Fig. 12. Transient flux of CO2 at constant feed pressure 1000 kPa for three different temperatures 303, 363 and 423 K.



E.I. Gkanas et al. / Applied Thermal Engineering 74 (2015) 36e46



according the results of Fig. 10 after some seconds the equilibrium is reached and the flux after the first 7 s is constant. Fig. 12 shows the transient flux of CO2 at constant feed pressure 1000 kPa for three different temperatures 303, 363 and 423 K. According to the results extracted from Fig. 12, the CO2 flux through the membrane is higher for the temperature 303 K and as the temperature increases the flux seems to become lower. Further, it



45



seems that for the lower temperature the equilibrium is reached slower than the case of the higher temperatures. Fig. 13 presents the transient analysis of the CO2 flux across the z-axis of the zeolite membrane at constant feed pressure 1000 kPa and at three different temperatures 303, 363 and 423 K respectively and these results also indicate that for the current geometry there is a preferred distribution of the CO2 flux through the membrane. This distribution is presented might due to the way that a WickeeKallenbach cell operates. 5. Conclusions On the basis of the MaxwelleStefan approach expressions have been derived for the description of the CO2 diffusion through the zeolite membrane. A three dimensional study has been performed in a WickeeKallenbach cell with diameter of 19 mm successfully indicating a novel approach in this geometry. Argon used as the sweep gas in the permeate gas chamber. The proposed model validated with experimental results and the similarity of the results was satisfying especially at higher temperatures maybe due to the fact that the support of the membrane might play a major role of thermal insulator at lower temperatures. In order to ensure that the proposed model is valid for all zeolitic membranes simulation runs have been performed for three different type of membranes and compared with experimental results. Three different scenarios for the MaxwelleStefan diffusion term were examined and compared to each other. The results showed that for high supply pressures only the strong and quasi-chemical approach have the potential in describing the CO2 diffusion. The effect of temperature and pressure in the CO2 permeation through the membrane was also studied and the results proved that for an almost linear isotherm the flux also can present linear behavior and the permeance is independent of the supply pressure. Finally, the transient analysis showed that the higher the supply pressure the higher the CO2 flux through the membrane, but for lower pressures the time for reaching equilibrium state is lower. Further, for lower temperatures also the flux is greater comparing to higher temperatures but for the high temperatures the time for reaching equilibrium state is also lower. Finally, a comparison of the permeance behavior of four different gases was studied showed that CO2 and N2O are strongly adsorbing gas whereas N2 and CH4 are weakly adsorbing gases. Acknowledgements The authors are very grateful to Professor Doros N. Theodorou (Head of COMSE Group, Department of Materials Science and Engineering, School of Chemical Engineering, National Technical University of Athens, Greece) for all fruitful discussions on this research work. References



Fig. 13. Transient analysis of the CO2 flux across the z-axis of the zeolite membrane at constant feed pressure 1000 kPa and at three different temperatures 303, 363 and 423 K. Panel a shows the flux for temperature 303 K. Panel b shows the flux for temperature 363 K and panel c shows the flux for temperature 423 K.



[1] Stern Review: The Economics of Climate Change, UK Office of Climate Change (OCC), London, 2005. Report available on: http://hm-treasury.gov.uk/ sternreview_index.htm. [2] L. Zhen, C.A. Grande, P. Li, J. Yu, A. Rodrigues, Adsorption and desorption of carbon dioxide and nitrogen on zeolite 5A, Sep. Sci. Technol. 46 (2011) 434e 451. [3] Y.S. Kim, K. Kusakabe, S. Morooka, S.M. Yang, Preparation of microporous silica membranes for gas separation, Korean J. Chem. Eng. 18 (2001) 106e112. [4] Song Lin Liu, Lu Shao, Mei Ling Chua, Cher Hon Lau, Huan Wang, Shuai Qua, Recent progress in the design of advanced PEO-containing membranes for CO2 removal, Prog. Polym. Sci. 38 (2013) 1089e1120. [5] Lu Shao, Shuai Quan, Xi-Quan Cheng, Xiao-Jing Chang, Hong-Guang Sun, Rong-Guo Wang, Developing cross-linked poly(ethylene oxide) membrane by the novel reaction system for H2 purification, Int. J. Hydrogen Energy 38 (2013) 5122e5132. [6] Mei Ling Chua, Shao Lu, Bee Ting Low, Youchang Xiao, Tai-Shung Chung, Polyetheramineepolyhedral oligomeric silsesquioxane organiceinorganic



46



E.I. Gkanas et al. / Applied Thermal Engineering 74 (2015) 36e46



hybrid membranes for CO2/H2 and CO2/N2 separation, J. Membr. Sci. 385e386 (2011) 40e48. [7] Cher Hon Lau, Songlin Liu, Donald R. Paul, Jianzhong Xia, Yan-Ching Jean, Hongmin Chen, Shao Lu, Tai-Shung Chung, Silica nanohybrid membranes with high CO2 affinity for green hydrogen purification, Adv. Energy Mater. 1 (2011) 634e642. [8] T.C. Merkel, L. Haiqing, W. Xiaotong, R. Baker, Power plant post-combustion carbon dioxide capture: an opportunity for membranes, J. Membr. Sci. 359 (2010) 126e139. [9] J. Caro, M. Noack, Zeolite membraneserecent developments and progress, Microporous Mesoporous Mater. 115 (2008) 215e233. [10] R. Krishna, J.M. Van Baten, In silico screening of zeolite membranes for CO2 capture, J. Membr. Sci. 360 (2010) 323e333. [11] W. Zhu, P. Hrabanek, L. Gora, F. Kapteijn, J.A. Moulijn, Role of adsorption in the permeation of CH4 and CO2 through a silicalite-1 membrane, Ind. Eng. Chem. Res. 45 (2006) 767e772. [12] Van der Broeke, W.J.W. Baker, F. Kapteijn, J.A. Moulijn, Transport and separation properties of a silicalite-1 membranedI. Operating conditions, Chem. Eng. Sci. 54 (1999) 245e253. [13] R. Roque-Malherbe, R. Wendelbo, A. Mifsud, A. Corma, Diffusion of aromatic hydrocarbons in H-ZSM-5, H-Beta, and H-MCM-22 zeolites, J. Phys. Chem. 99 (1995) 14064e14071. [14] J. Xiao, J. Wei, Diffusion mechanism of hydrocarbons in zeolitesdI. Theory, Chem. Eng. Sci. 47 (1992) 1123e1141. [15] J. Karger, M. Bulow, Theoretical prediction of uptake behaviour in adsorption kinetics of binary gas mixtures using irreversible thermodynamics, Chem. Eng. Sci. 30 (1975) 893e896. [16] S. Farooq, I.A. Karimi, Modeling support resistance in zeolite membranes modeling, J. Membr. Sci. 186 (2001) 109e121. [17] M. Lassinantti, J. Hedlund, J. Sterte, Faujasite-type films synthesized by seeding, Microporous Mesoporous Mater. 38 (2000) 25e34. [18] T. Matsufuji, S. Nakagawa, N. Nishiyam, M. Matsukata, K. Ueyama, Synthesis and permeation studies of ferrierite/alumina composite membranes, Microporous Mesoporous Mater. 38 (2000) 43e50. [19] H. Lee, P.K. Dutta, Synthesis of free-standing chabazite-type films, Microporous Mesoporous Mater. 38 (2000) 151e159. [20] T. Tomita, K. Nakayama, H. Sakai, Gas separation characteristics of DDR type zeolite membrane, Microporous Mesoporous Mater. 68 (2004) 71e75. [21] K.G. Shattuck, B. Yilmaz, J. Warzywoda, A. Sacco, Hydrothermal synthesis of oriented ETS-4 films on porous a-alumina substrates, Microporous Mesoporous Mater. 88 (2006) 56e62. [22] Z. Lin, J. Rocha, A. Navajas, C. Tellez, J.Q. Coronas, J. Santamaria, Synthesis and characterisation of titanosilicate ETS-10 membranes, Microporous Mesoporous Mater. 67 (2004) 79e84. [23] C. Algieri, P. Bernardo, G. Golemme, G. Barbieri, E. Drioli, Permeation properties of a thin silicalite-1 (MFI) membrane, J. Membr. Sci. 222 (2003) 181e 190. [24] Y. Takata, T. Tsuru, T. Yoshioka, M. Asaeda, Gas permeation properties of MFI zeolite membranes prepared by the secondary growth of colloidal silicalite and application to the methylation of toluene, Microporous Mesoporous Mater. 54 (2002) 257e268. [25] T.Q. Gardner, A.I. Flores, R.D. Noble, J.L. Falconer, Measurement of adsorption and diffusion properties of H-ZSM-5 zeolite membranes by a transient Technique, AIChE J. 48 (2002) 1155e1167. [26] Z. Lai, G. Bonilla, I. Diaz, J.G. Nery, K. Sujaoti, M.A. Amat, E. Kokkoli, O. Terasaki, R.W. Thompson, M. Tsapatsis, D.G. Vlachos, Microstructural optimization of a zeolite membrane for organic vapor separation, Science 100 (2003) 456. [27] G. Xeromitakis, Z.P. Lai, M. Tsapatsis, Separation of Xylene Isomer Vapors with oriented MFI membranes Made by Seeded growth, Ind. Eng. Chem. Res. 40 (2001) 544e551. [28] G. Xomeritakis, A. Gouzinis, S. Nair, T. Okubo, M.Y. He, R.M. Overney, M. Tsapatsis, Growth, microstructure, and permeation properties of supported zeolite (MFI) films and membranes prepared by secondary growth, Chem. Eng. Sci. 54 (1999) 3521e3532. [29] M.C. Lovallo, A. Gouzinis, M. Tsapatsis, Synthesis and characterization of oriented MFI membranes prepared by secondary growth, AIChE J. 44 (1998) 1903e1913.



[30] V. Sebastian, I. Kumakiri, R. Bredesen, M. Menendez, in: Proc. 9th Int. Conf. on Inorganic Membranes, Lillehammer, Norway, 2006, p. 290. [31] S. Li, J.L. Falconer, R.D. Noble, Improved SAPO-34 membranes for CO2/CH4 separations, Adv. Mater. 18 (2006) 2601e2603. [32] S. Li, J.Q. Martinek, J.L. Falconer, R.D. Noble, T.Q. Gardner, High-pressure CO2/ CH4 separation using SAPO-34 membranes, Ind. Eng. Chem. Res. 44 (2005) 3220e3228. [33] M. Hong, S. Li, H.F. Funke, J.L. Falconer, R.D. Noble, Ion-exchanged SAPO-34 membranes for light gas separations, Microporous Mesoporous Mater. 106 (2008) 140e146. [34] S. Li, J.L. Falconer, R.D. Noble, SAPO-34 membranes for CO2/CH4 separation, J. Membr. Sci. 241 (2004) 121e135. [35] Y. Hasegawa, K. Watanabe, K. Kusakabe, S. Morooka, Influence of alkali cations on permeation properties of Y-type zeolite membranes, J. Membr. Sci. 208 (2002) 415e418. [36] P.B. Weisz, Zeolitesdnew horizons in catalysis, Chemtech 3 (1973) 498e505. [37] J. Wei, A mathematical theory of enchanced paraxylene selectivity in molecular sieve catalysis, J. Catal. 76 (1982) 433e439. [38] M. Post, Diffusion in zeolite molecular sieves, Surf. Sci. Catal. 58 (1991) 391e443. [39] D.M. Ruthven, Principles of Adsorption and Adsorption Processes, Wiley, New York, 1984, pp. 124e163. [40] I.H. Doetch, D.M. Ruthven, K.F. Loughlin, Sorption and diffusion of n-heptane in 5A zeolite, Can. J. Chem. 52 (1974) 2717e2724. [41] P. Capek, V. Hejmanek, O. Solcova, K. Klusacek, P. Schneider, Gas transport in porous media under dynamic conditions, Catal. Today 38 (1997) 31e38. [42] S.T. Kolaczkowski, Measurement of effective diffusivity in catalyst-coated monoliths, Catal. Today 83 (2003) 85e95. [43] M.P. Dudukovic, An analytical solution for the transient response in a diffusion cell of the WickeeKallenbach type, Chem. Eng. Sci. 37 (1982) 153e158. [44] O. Solcova, H. Snajdaufova, P. Schneider, Multicomponent counter-current gas diffusion in porous solids: the Graham’s law diffusion cell, Chem. Eng. Sci. 56 (2001) 5231e5237. [45] V.I. Sikavitsas, R.T. Yang, Predicting multicomponent diffusivities for diffusion on surfaces and in molecular sieves with energy heterogeneity, Chem. Eng. Sci. 50 (1995) 3057e3065. [46] N.C. Karayiannis, V.G. Mavrantzas, D.N. Theodorou, Diffusion of small molecules in disordered media: study of the effect of kinetic and spatial heterogeneities, Chem. Eng. Sci. 56 (2001) 2789e2801. [47] I.G. Economou, V.E. Raptisa, V.S. Melissasa, D.N. Theodorou, J. Petrou, J.H. Petropoulos, Molecular simulation of structure, thermodynamic and transport properties of polymeric membrane materials for hydrocarbon separation, Fluid Phase Equilib. 228e229 (2005) 15e20. [48] P. Valertzis, E.S. Kikkinides, M.C. Georgiadis, On the optimization of gas separation processes using zeolite membranes, Trans. IchemE 81 (2003) 525e536. [49] F. Kapteijn, J.A. Moulijn, R. Krishna, The generalized MaxwelleStefan model for diffusion in zeolites: sorbate molecules with different saturation loadings, Chem. Eng. Sci. 55 (2000) 2923e2930. [50] R. Krishna, J.A. Wesselingh, The MaxwelleStefan approach to mass transfer, Chem. Eng. Sci. 52 (1997) 861e911. [51] S.C. Lee, Prediction of permeation behavior of CO2 and CH4 through silicalite-1 membranes in single-component or binary mixture systems using occupancydependent MaxwelleStefan diffusivities, J. Membr. Sci. 306 (2007) 267e276. [52] I. Perdana, B.W. Tyoso, I.M. Berdiyasa, S.K. Wirawan, D. Creaser, Effect of external mass transport on permeation in a WickeeKallenbach cell, Chem. Eng. Res. Des. 87 (2009) 1438e1447. [53] S. Himeno, T. Tomita, K. Suzuki, S. Yoshida, Characterization and selectivity for methane and carbon dioxide adsorption on the all-silica DD3R zeolite, Microporous Mesoporous Mater. 98 (2007) 62e69. [54] J. van den Bergh, W. Zhu, J. Gascon, J.A. Moulijn, F. Kapteijn, Separation and permeation characteristics of a DD3R zeolite membrane, J. Membr. Sci. 316 (2008) 35e45. [55] D.A. Reed, G. Ehrlichm, Surface diffusion, atomic jump rates and thermodynamics, Surf. Sci. 102 (1981) 588. [56] J. Caro, M. Noack, P. Kölsch, R. Schäfer, Zeolite membranes e state of their development and perspective, Microporous Mesoporous Mater. 38 (2000) 3e24. [57] S. Cavenati, C.A. Grande, A.E. Rodrigues, Adsorption equilibrium of methane, carbon dioxide and nitrogen on zeolite 13X at high pressures, J. Chem. Eng. Data 49 (2004) 1095e1101.



Applied Thermal Engineering 74 (2015) 47e53



Contents lists available at ScienceDirect



Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng



Modelling and simulation of intensified absorber for post-combustion CO2 capture using different mass transfer correlations Atuman S. Joel, Meihong Wang*, Colin Ramshaw Process/Energy Systems Engineering Group, School of Engineering, University of Hull, HU6 7RX, UK



h i g h l i g h t s � Post-combustion carbon capture using Rotating Packed Bed for intensified absorber. � Mass transfer correlations implemented in visual FORTRAN in Aspen Plus. � Two sets of mass transfer correlations compared through model validations. � Intensified absorber using RPB improves mass transfer significantly.



a r t i c l e i n f o



a b s t r a c t



Article history: Received 1 September 2013 Accepted 22 February 2014 Available online 18 March 2014



This paper studied mass transfer in rotating packed bed (RPB) which has the potential to significantly reduce capital and operating costs in post-combustion CO2 capture. To model intensified absorber, mass transfer correlations were implemented in visual FORTRAN and then were dynamically linked with Aspen Plus rate-based model. Therefore, this represents a newly developed model for intensified absorber using RPB. Two sets of mass transfer correlations were studied and compared through model validations. The second set of correlations performed better at the MEA concentrations tested as compared with the first set of correlations. For insights into the design and operation of intensified absorber, process analysis was carried out, which indicates: (a) With fixed RPB equipment size and fixed Lean MEA flow rate, CO2 capture level decreases with increase in flue gas flow rate; (b) Higher lean MEA inlet temperature leads to higher CO2 capture level. (c) At higher flue gas temperature (from 30 � C to 80 � C), the CO2 capture level of the intensified absorber can be maintained. Compared with conventional absorber using packed columns, the insights obtained from this study are (1) Intensified absorber using rotating packed bed (RPB) improves mass transfer significantly. (2) Cooling duty cost can be saved since higher lean MEA temperature and/or higher flue gas temperature shows little or no effect on the performance of the RPB.  2014 Elsevier Ltd. All rights reserved.



Keywords: Post-combustion CO2 capture Chemical absorption MEA solvent Process intensification (PI) Rotating packed bed (RPB) Process simulation



1. Introduction Carbon dioxide (CO2) emission has become crucial environmental concern in recent years because of its contribution to global warming. Combustion of coal and petroleum accounts for the majority of the anthropogenic CO2 emissions. Petroleum is mostly used as a transportation fuel for vehicles while coal is used mostly for electricity generation, for instance about 85.5% of coal is used for electricity generation in 2011 in the UK [1]. Moulijn et al. [2] stated that among the greenhouse gases, CO2 contributes more than 60%



* Corresponding author. Tel.: þ44 (0) 1482 466688; fax: þ44 1482 466664. E-mail addresses: [email protected], [email protected] (M. Wang). http://dx.doi.org/10.1016/j.applthermaleng.2014.02.064 1359-4311/ 2014 Elsevier Ltd. All rights reserved.



to global warming. Atmospheric CO2 concentration is close to 400 ppm which is higher than the pre-industrial level of about 300 ppm [3]; this increased atmospheric concentration of CO2 affects the radiative balance of the Earth and, consequently, its temperature and other aspects of its climate. In order to meet the set target of 50% emission reduction as compared to the level of 1990 as proposed by Intergovernmental panel on climate change (IPCC) [4], carbon capture and storage (CCS) is an important option for that to be achieved. The International Energy Agency (IEA) [5] identifies CCS as a significant and low-cost option in fighting climate change. The most matured CO2 capture technology is post-combustion CO2 capture (PCC) with chemical absorption as reported in MacDowell et al. [6] which is also believed to be a low-risk technology and promising near-term option for large-scale CO2 capture.



48



A.S. Joel et al. / Applied Thermal Engineering 74 (2015) 47e53



Post-combustion CO2 Capture for coal-fired power plants using conventional absorber has been reported by many authors. Dugas [7] carried out pilot plant study of post-combustion CO2 capture in the context of fossil fuel-fired power plants. Lawal et al. [8e10] carried out dynamic modelling of CO2 absorption for postcombustion capture in coal-fired power plants. In these studies, one of the identified challenges to the commercial roll-out of the technology has been the large size of the packed columns needed. This translates to high capital and operating cost and unavoidable impact on electricity cost. Approaches such as heat integration, inter-cooling among others could reduce the operating cost slightly. However, they limit the plant flexibility and will make operation and control more difficult [11]. Process intensification (PI) has the potential to meet this challenge [12e14]. Fig. 1 shows the geometrical similarities and differences between intensified absorber using RPB and convention absorber using packed column. One of the operational benefits of using RPB absorber is its ability to be operated at higher gas and/or liquid flow rates owing to the low tendency of flooding compared to that in the conventional packed bed [15]. Another benefit of using RPB is its better self-cleaning, avoidance of plugging in the system, and being unaffected by a moderate disturbance in its orientation [16]. Because of much less gaseliquid contact time that occurs in RPB than that in a conventional packed bed, the selection of an absorbent with a fast reaction rate with CO2 is crucial [17,18]. This may necessitate the use of higher concentration of solvent, but this comes with another challenge known as corrosion as reported by Barham et al. [19], this problem can be managed by the use of expensive construction material such as stainless steel since the volume of the RPB can be significantly reduced compared to that conventional packed column [12,14,15,20e22]. 1.1. Motivation Over 8000 tonnes of CO2 per day will be released from a typical 500 MWe supercritical coal fired power plant operating at 46%



efficiency (LHV basis) [23]. This big volume of flue gas will result in the use of high amount of solvent and big size of packed columns if conventional technology is to be applied. Lawal et al. [24] reported dynamic modelling study of a 500 MWe sub-critical coal-fired power plant using the conventional packed column. From the analysis, two absorbers of 17 m in packing height and 9 m in diameter will be needed to separate CO2 from the flue gas. These huge conventional packed columns will mean higher capital and operating costs. This could increase electricity costs by over 50% and has been a major impediment to commercializing the technology. 1.2. Novel contributions of the paper There are two novel aspects in this paper: (a) A new first principle model for intensified absorber was developed which was implemented in Aspen rate-based model by replacing different correlations for mass transfer, interfacial area and liquid hold-up. Steady state validation of the intensified absorber is performed, where comparison is made by using two different sets of mass transfer correlations and the results indicated that Set 2 correlations give better predictions at higher (i.e. 75 wt%) and lower (i.e. 56 wt%) MEA concentration than Set 1. (b) With the models developed using Set 2 correlations and validated, process analysis of the intensified absorber with RPB involving different process scenarios was carried out to gain insights for process design and operation. These process scenarios are: (i) when the RPB absorber size is fixed and lean MEA flow rate is fixed, the impact of flue gas flow rate on CO2 capture level was explored; (ii) the effect of higher lean MEA temperature on CO2 capture level for RPB absorber was explored; (iii) the effect of higher flue gas temperature on CO2 capture level was explored. The results indicate that higher leanMEA temperature increases CO2 capture level and CO2 capture level can be maintained at higher flue gas temperature. These results indicate that cooling duty for lean MEA and flue gas can be greatly reduced.



Lean Gas Lean Gas Lean-MEA Lean-MEA



D



Fibre



H



Flue gas



Flue gas Rich-MEA Rich-MEA



Motor



Fig. 1. Geometrical similarities and differences between RPB and conventional absorber.



49



A.S. Joel et al. / Applied Thermal Engineering 74 (2015) 47e53



Comparison between this paper and Joel et al. [20] reveals that there are three main differences: (a) In Joel et al. [20] only Set 1 correlations was used while in this paper Set 1 and Set 2 correlations were compared. (b) Process analysis was done using Set 1 correlations in Joel et al. [20] while in this paper Set 2 correlations were used. (c) Process scenarios considered in Joel et al. [20] were effect of lean MEA concentration, effect of rotating speed, effect of lean MEA temperature on CO2 capture level and temperature bulge analysis in RPB. While in this paper process analysis scenarios considered are: Effect of flue gas flow rate, effect of lean MEA temperature and effect of flue gas temperature on CO2 capture level. 2. Model development To model intensified absorber using RPB, the default mass transfer correlations of the Aspen Plus rate-based absorber model were changed with subroutines written in Intel visual FORTRAN. Since the model for intensified absorber using RPB does not exist in any commercially available model library (including Aspen Plus). To do that, two sets of correlations were written separately in visual FORTRAN and then dynamically linked with Aspen Plus ratebased absorber model; by so doing the model now represent an intensified absorber using RPB. The first set of correlations studied include liquid phase mass transfer coefficient given by Tung and Mah [25], gas phase mass transfer coefficient given by Onda et al. [26], Interfacial area correlation modified by updating the gravity term in the equation with centrifugal acceleration given by Onda et al. [26], and Liquid holdup evaluated using Burns et al. [27] correlation. The second set of correlations include: liquid phase mass transfer coefficient given by Chen et al. [28], gas-phase mass transfer coefficient given by Chen [29], interfacial area correlation estimated by Luo et al. [30] and Liquid hold-up correlation given by Burns et al. [27]. Electrolyte Non-Random-Two-Liquid (ElecNRTL) activity coefficient model in Aspen Plus is used to describe the vapoureliquid equilibrium, the chemical equilibrium and the physical properties of the system. The coefficient of equilibrium constant and equilibrium reactions which are assumed to occur in the liquid film and Kinetics reactions equations and parameters are obtained in AspenTech [31]. The electrolyte solution chemistry which is used in property calculation is modelled with chemistry model and all the ionic reactions are assumed to be in chemical equilibrium as shown in equations (1)e(5) [31].



Equilibrium



2H2 O4H3 Oþ þ OH�



(1)



Equilibrium



CO2 þ 2H2 O4H3 Oþ þ HCO� 3



(2)



Equilibrium



þ 2� HCO� 3 þ H2 O4H3 O þ CO3



(3)



Equilibrium



MEAHþ þ H2 O4MEA þ H3 Oþ



(4)



Equilibrium



MEACOO� þ H2 O/MEA þ HCO� 3



(5)



Kinetic reaction model used for the intensified absorber calculation is specified by equations (6)e(12) in the reaction part of the absorber model in the Aspen Plus [31].



Equilibrium



H2 O þ MEAHþ 4MEA þ H3 Oþ þ



�



(6)



Equilibrium



2H2 O4H3 O þ OH



(7)



Equilibrium



2� þ HCO� 3 þ H2 O4CO3 þ H3 O



(8)



Kinetic



CO2 þ OH� /HCO� 3



(9)



Kinetic



� HCO� 3 /CO2 þ OH



(10)



Kinetic



MEA þ CO2 þ H2 O/MEACOO� þ H3 Oþ



(11)



Kinetic



MEACOO� þ H3 Oþ /MEA þ CO2 þ H2 O



(12)



Power law expressions (To not specified) are used for the ratecontrolled reactions. The kinetic parameters for reaction in equations (9)e(12) are in Table 1



� N � E Y r ¼ kT n exp � C ai RT i ¼ 1 i



(13)



In this typical RPB absorber, flue gas and lean-MEA solvent were contacted counter-currently. Mixed flow model which is default in Aspen Plus was used for this study. Process parameters can be found in Jassim et al. [32]. 2.1. Liquid phase mass transfer coefficient An expression was introduced by Tung and Mah [25] using the penetration model to describe the liquid mass transfer behaviour in the RPB.



�a �1=3 kL dp t 1=2 2=3 1=6 ¼ 0:919 ScL ReL Gr L DL a



(14)



gc in the Grashof number GrL is taken as gc ¼ rw2 to account for the effect of rotation in the RPB absorber. This correlation was developed without considering the Coriolis force or the effect of the packing geometry. Chen et al. [28] developed liquid phase mass transfer correlation that put into consideration the end effect and packing geometry. The correlation was found to be valid for different sizes of the RPBs and for viscous Newtonian and non-Newtonian fluids.



� � kL adp Vo V 1 � 0:93 � 1:13 i DL at Vt Vt ¼



0:17 0:3 Gr0:3 0:35Sc0:5 L ReL L WeL



at a0p



!�0:5 �



sc sw



�0:14



(15)



2.2. Gas phase mass transfer coefficient Onda et al. [26] correlation for calculating gas-side mass transfer coefficient was developed for conventional packed column. Sandilya et al. [33] suggested that the gas rotated like a solid body in the rotor because of the drag that was caused by the packing and that, consequently, the gas-side mass transfer coefficient should be similar to that in a conventional packed column. 1=3 �



kG ¼ 2:0ðat DG ÞRe0:7 G ScG



at dp



��2



(16)



Table 1 Parameters k and E in Equation (13) [31]. Reaction no.



k



E, cal/mol



(9) (10) (11) (12)



4.32eþ13 2.38eþ13 9.77eþ13 2.18eþ13



13,249 29,451 9855.8 14,138.4



50



A.S. Joel et al. / Applied Thermal Engineering 74 (2015) 47e53



Chen [29] presented local gas-side mass transfer coefficient correlation using two-film theory for RPB.



!1:4 � � kG a Vo 0:31 1:13 0:14 0:07 at Re Gr We 1 � 0:9 ¼ 0:023Re G L L G 0 Vt ap DG a2t



(17)



2.3. Total gaseliquid interfacial area Total gaseliquid interfacial area is calculated with the Onda et al. [26] correlation.



� � �s �0:75 a c 0:2 �0:05 ¼ 1 � exp � 1:45 Re0:1 L WeL Fr L s at



(18)



a ¼ 66; 510Re�1:41 Fr �0:12 We1:21 4�0:74 L L L at



(19)



Similarly, gc in the Froude number FrL is taken as gc ¼ rw2 to account for the effect of rotation in the RPB absorber. Luo et al. [30] studied gaseliquid effective interfacial area in an RPB considering different types of packing, also taking into account the effect of fibre diameter and opening of the wire mesh.



Fig. 2. Methodology used in this paper.



2.4. Liquid hold-up Liquid holdup correlation by Burns et al. [27] is given as



�



˛L ¼ 0:039



gc go



��0:5 �



� � � U 0:6 v 0:22 Uo vo



(20)



go ¼ 100 m s�2 ; Uo ¼ 1 cm s�1 ; vo ¼ 1 cS ¼ 10�6 m2 s�1 U ¼



QL 2prZ



(21)



2.5. Modelling and simulation methodology The procedure used in this paper for modelling and simulation of the RPB is shown in Fig. 2. 3. Model validation The experimental data used for model validation was obtained from Jassim et al. [32]. From their experiments, two lean-MEA concentrations (average 55 wt% and 75 wt%) were selected so as to fall within a reasonable range of MEA concentration to minimize the problem of corrosion and maximize CO2 absorption rate. Two sets of correlations were used for the validation. The sets of correlations are presented in Table 2 and the input condition is shown in Table 3. Validation results were presented in terms of CO2 capture level which is defined as



CO2 capture level ð%Þ ¼



yCO2;in � yCO2;out yCO2;in



!



� 100



(22)



The CO2 loading of lean-MEA and rich-MEA in Tables 4 and 5 was evaluated in mole basis as



Loading ¼



Moles of all CO2 carrying species Moles of all MEA carrying species



(23a)



i h i � h � 2� þ MEACOO� ½CO2 � þ HCO� 3 þ CO3 � � � � Loading ¼ ½MEA� þ MEAþ þ MEACOO�



(23b)



In Table 4, the model predictions were compared with experimental data for the two correlation sets in Table 2 and for input conditions in Table 3. For Run 1 (56 wt% MEA concentration and 600 rpm rotor speed), the CO2 capture level of Set 1 correlation is 92.90% while that of Set 2 is 96.36%. For both sets of correlation the model reasonably predicts the experimental data with relative errors less than 3%. In Table 5, the simulation predictions were compared with experimental data at the input conditions shown in Table 3. The error prediction of Run 3 and 4 using Set 2 correlation gives better agreement with the experiment data, the reason for this could be from the liquid and gas phase mass transfer resistance where Chen et al. [28] and Chen [29] account for the effect of viscosity and packing geometry. These results show that the model developed using Aspen Plus� rate-based absorber model modified for RPB with correlations implemented in visual FORTRAN is able to reasonably capture the behaviour of an RPB absorber. As a result, the model can be used to analyse typical RPB behaviour at different input conditions. 4. Process analysis In this section, the model developed and validated is used to analyse the process characteristics of the RPB absorber.



Table 2 Model correlation sets used for the modelling and simulations. Correlations



Set 1



Set 2



Liquid-phase mass transfer coefficient Gas-phase mass transfer coefficient Interfacial area Liquid hold-up



Tung and Mah [25] Onda et al. [26] Onda et al. [26] Burns et al. [27]



Chen et al. [28] Chen [29] Luo et al. [30] Burns et al. [27]



51



A.S. Joel et al. / Applied Thermal Engineering 74 (2015) 47e53



4.2. Effect of Lean-MEA temperature on CO2 capture level



Table 3 Input process conditions for Run 1 to Run 4 [32]. Variable



4.2.1. Justification for case study The study is performed to investigate the effect of lean MEA temperature on the performance of RPB absorber. The key driving forces for absorption are mass transfer and chemical reaction which are known to respectively decrease and increase with temperature [34]. Conventional absorber performance is already known to be hindered by increase in lean MEA temperature due to the possibility of temperature bulge within the absorber column [35]. Based, on this, capture performance with lean MEA temperature should be studied for RPB absorbers.



Runs Run 1



Run 2



Run 3



Run 4



Rotor speed (RPM) Lean temperature ( C) Lean pressure (atm) Flue gas flow rate (kmol/h)



600 39.6 1 2.87



1000 40.1 1 2.87



600 41 1 2.87



1000 40.2 1 2.87



Flue gas composition in (vol%) CO2 H2O N2



4.710 0.159 95.131



4.480 0.159 95.361



4.400 0.159 95.441



4.290 0.159 95.551



Lean-MEA flow rate (kg/s)



0.66



0.66



0.66



0.66



Lean-MEA composition (wt%) H2O CO2 MEA



40.91 3.09 56.00



40.91 3.09 56.00



22.32 2.68 75.00



23.41 2.59 74.00



4.2.2. Setup of the case study Set 1 correlation is use in the implementation of the case study. Process conditions are shown in Table 6. Rotor speed of 1000 rpm, lean MEA flow rate of 0.66 kg/s and lean MEA temperature which is varied from 25  C, 30  C, 35  C, 40  C . to 80  C at 55 wt% and 75 wt % lean MEA concentrations were used.



4.1. Effect of flue gas flow rate on CO2 capture level 4.1.1. Justification for case study In designing RPB absorbers, flue gas flow rate is an important parameter in determining the size of the absorption column, while this process analysis is also necessary in order that the CO2 emission target can be met. 4.1.2. Setup of the case study For this study, Set 2 of the correlations in Table 2 was used and the input conditions in Table 3 for Run 2 and Run 4 having constant rotor speed of 1000 rpm were selected for the analysis. The RPB absorber size is fixed, as well as the lean MEA flow rate. The flue gas flow rate was varied from 0.02 kg/s to 1 kg/s. 4.1.3. Results and discussions Fig. 3 shows that for both Runs CO2 capture level decrease as the flue gas flow rate increases. This is associated with decrease in contact time (i.e. residence time) between the flue gas and liquid MEA solvent resulting in more CO2 escaping the RPB without being captured. Also from Fig. 3, it can be seen that whatever the MEA concentration of the solvent the trend is the same.



4.2.3. Results and discussions Fig. 4 shows the effect of varying lean MEA temperature on CO2 capture level at different lean MEA concentrations (55 wt% MEA and 75 wt% MEA). The results show that CO2 capture level increases significantly from 25  C to 50  C lean MEA temperatures. Lean MEA temperature increase above 50  C has no significant impact on the CO2 capture level. Improvement of RPB performance as temperature increases can be associated to decrease in viscosity of the lean MEA solvent as explained by Lewis and Whitman [36] that the ratio of viscosity to density (kinematic viscosity) of the film fluid is probably the controlling factor in determining film thickness. Haslam et al. [37] said that if film resistance is directly proportional to film thickness, then film conductivity is the inverse of kinematic viscosity. The effect of temperature on density of gas is great, but temperature affects the density of lean MEA only slightly [38]. Again an increase in temperature causes an increase in viscosity of a gas but the same increase in temperature might greatly lower the viscosity of lean MEA. This improves mass transfer due to thinner liquid film since absorption of CO2 into alkanolamines solutions is a liquid film controlled process [32]. Also increasing lean solvent temperature leads to increase in chemical reaction rate.



Table 4 Simulation results with 2 different sets of correlations compared to the experimental data [32] for Run 1 and Run 2. Variable



CO2 loading of Lean MEA, (mol CO2/mol MEA) CO2 loading of Rich MEA, (mol CO2/mol MEA) Average Lean MEA/Rich MEA, (mol CO2/mol MEA) CO2 capture level (%)



Run 1



Run 2



Expt.



Set 1



0.0772 0.0828 0.0800 94.9



0.0772 0.0827 0.0800 92.9



Error 1



Set 2



0.1208 0.0000 2.1075



0.0772 0.0832 0.0801 96.65



Error 2



Expt.



Set 1



Error 1



Set 2



Error 2



0.4831 0.1250 1.8440



0.0772 0.0828 0.0800 95.4



0.0772 0.0825 0.0799 93.26



0.3623 0.1250 2.2432



0.0772 0.0827 0.0801 96.95



0.1208 0.1250 1.6247



Error 1



Set 2



Error 2



2.1569 1.0060 4.0308



0.0483 0.0524 0.0503 98.66



2.7451 1.2072 1.1897



Table 5 Simulation results with 2 different sets of correlations compared to the experimental data [32] for Run 3 and Run 4. Variable



CO2 loading of Lean-MEA, (mol CO2/mol MEA) CO2 loading of Rich-MEA, (mol CO2/mol MEA) Average Lean-MEA/Rich-MEA, (mol CO2/mol MEA) CO2 capture level (%)



Run 3



Run 4



Expt.



Set 1



0.0492 0.0531 0.0512 98.20



0.0492 0.0530 0.0511 93.28



Error 1



Set 2



0.1883 0.1953 5.0102



0.0492 0.0531 0.0512 97.36



Error 2



Expt.



Set 1



0.0000 0.0000 0.8554



0.0483 0.0510 0.0497 97.50



0.0483 0.0521 0.0502 93.57



52



A.S. Joel et al. / Applied Thermal Engineering 74 (2015) 47e53



Fig. 3. Effect of flue gas flow rate of CO2 capture level.



Fig. 4. Effect of lean-MEA temperature on CO2 capture level.



4.3. Effect of flue gas temperature on CO2 capture level 4.3.1. Justification for case study Moisture content of a flue gas is dependent on temperature, pressure and the type of fuel used. Study of flue gas temperature is necessary since additional cost will be incurred in cooling flue gas prior to entering conventional absorber [11,34]. 4.3.2. Setup of the case study Set 2 correlations in Table 2 were used for the formulation of this case study. Run 2 and 4 were selected which are at 56 wt% and 74 wt% MEA concentration respectively. The simulations were run at rotor speed of 1000 rpm, the lean-MEA temperature was kept constant 40.1  C for Run 2 and 40.2  C for Run 2, in both case flue gas temperature was varied from 30  C to 80  C. 4.3.3. Results and discussions Fig. 5 shows the effect of flue gas temperature on CO2 capture level, where the results show that the CO2 capture level is maintained despite increase in the flue gas temperature. Run 2 and Run 4 gives the same trend, this show that even if the solvent is having higher MEA concentration, CO2 capture level behaves the same way. The reason for this behaviour is because of no temperature bulge as reported in Joel et al. [20] since the evaporated vapour condensate does not have enough residence time for energy build-up in the column. Again because of high liquid to gas (L/G) ratio in an RPB, makes the CO2 capture level not sensitive to the flue gas temperature change. The maintained CO2 capture level shown in Fig. 5 indicates that flue gas cooling energy cost can be avoided.



5. Conclusions Modelling, validation and analysis of a post-combustion CO2 capture with MEA using intensified absorber was carried out in this paper with two sets of correlations. The RPB absorber was modelled in Aspen Plus which is dynamically linked with visual FORTRAN. Rate-based modelling approach was used and chemical reactions are assumed to be at equilibrium. Experimental data used for validation were obtained from Jassim et al. [32]. Two sets of correlations were implemented for the validation of the intensified absorber model and the model predictions showed good agreement with the experimental results. The second set of correlations gives better prediction compared to the first set of correlation. Process analysis of the effect of flue gas flow rate, rotational speed, lean-MEA temperature and flue gas temperature on CO2 capture level was studied. It was found that as the lean-MEA temperature increases the CO2 capture level increases and as the flue gas flow rate increases the CO2 capture level can be maintained, which mean that cooling duty cost in RPB can be greatly reduced. The result shows mass transfer is improved with the use of RPB; also since the RPB absorber is operated at higher temperature reaction rate is enhanced.



Table 6 Process conditions for lean MEA temperature studies. Variable



55 wt% MEA con.



75 wt% MEA con.



Rotor speed (RPM) Lean pressure (atm) Flue gas flow rate (kmol/h)



1000 1 2.87



1000 1 2.87



Flue gas composition (vol%) H2O CO2 N2



17.1 4.4 78.5



17.1 4.4 78.5



Lean-MEA flow rate (kg/s)



0.66



0.66



Lean-MEA composition (wt%) H2O CO2 MEA



41.03 3.97 55.00



22.32 2.68 75.00



Fig. 5. Effect of flue gas temperature on CO2 capture level.



A.S. Joel et al. / Applied Thermal Engineering 74 (2015) 47e53



Acknowledgements The authors would like to acknowledge financial support from UK Research Councils’ Energy Programme (Ref: NE/H013865/2) and EU FP7 (Ref: PIRSES-GA-2013-612230). References [1] Department of Energy and Climate Change, Solid Fuels and Derived Gases Statistics: Data Sources and Methodologies, 2012. Available at: http://www. decc.gov.uk/en/content/cms/statistics/energy_stats/source/coal/coal.aspx (accessed May 2012). [2] J.A. Moulijn, A. Stankiewicz, J. Grievink, A. Górak, Process intensification and process systems engineering: a friendly symbiosis, Comput. Chem. Eng. 32 (1e2) (2008) 3e11. [3] T.H. Oh, Carbon capture and storage potential in coal-fired plant in Malaysia e a review, Renew. Sustain. Energy Rev. 14 (2010) 2697e2709. [4] Intergovernmental Panel on Climate Change, Contribution of Working Group III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, United Kingdom/New York, United States, 2007. [5] International Energy Agency, Carbon Capture and Storage Model Regulatory Framework, 2010. Available at: http://www.iea.org/publications/ freepublications/publication/model_framework.pdf, (accessed May 2013). [6] N. MacDowell, N. Florin, A. Buchard, J. Hallett, A. Galindo, G. Jackson, C.S. Adjiman, C.K. Williams, N. Shah, P. Fennell, An overview of CO2 capture technologies, Energy Environ. Sci. 3 (2010) 1645e1669. [7] E.R. Dugas, Pilot plant study of carbon dioxide capture by aqueous monoethanolamine (M.S.E. thesis), University of Texas, Austin, USA, 2006. [8] A. Lawal, M. Wang, P. Stephenson, H. Yeung, Dynamic modelling of CO2 absorption for post combustion capture in coal-fired power plants, Fuel 88 (2009) 2455e2462. [9] A. Lawal, M. Wang, P. Stephenson, H. Yeung, Dynamic modeling and simulation of CO2 chemical absorption process for coal-fired power plants, in: R.M. de Brito Alves, C.A. Oller do Nascimento, E. Chalbaud Biscaia Jr. (Eds.), Computer Aided Chemical Engineering, Elsevier, 2009, pp. 1725e1730. [10] A. Lawal, M. Wang, P. Stephenson, G. Koumpouras, H. Yeung, Dynamic modelling and analysis of post-combustion CO2 chemical absorption process for coal-fired power plants, Fuel 89 (2010) 2791e2801. [11] H.M. Kvamsdal, J.P. Jakobsen, K.A. Hoff, Dynamic modeling and simulation of a CO2 absorber column for post-combustion CO2 capture, Chem. Eng. Process. Process Intensif. 48 (2009) 135e144. [12] D. Reay, The role of process intensification in cutting greenhouse gas emissions, Appl. Therm. Eng. 28 (2008) 2011e2019. [13] C. Ramshaw, R.H. Mallinson, Mass Transfer Process, 1981. U.S. Patent, 4,283,255. [14] M. Wang, A. Lawal, P. Stephenson, J. Sidders, C. Ramshaw, Post-combustion CO2 capture with chemical absorption: a state-of-the-art review, Chem. Eng. Res. Des. 89 (2011) 1609e1624. [15] C. Lin, Y. Lin, C. Tan, Evaluation of alkanolamine solutions for carbon dioxide removal in cross-flow rotating packed beds, J. Hazard. Mater. 175 (2010) 344e351. [16] G.Q. Wang, Z.C. Xu, Y.L. Yu, J.B. Ji, Performance of a rotating zigzag bed e a new HIGEE, Chem. Eng. Process. 47 (2008) 2131e2139. [17] H.H. Cheng, C.S. Tan, Reduction of CO2 concentration in zinc/air battery by absorption in a rotating packed bed, J. Power Sources 162 (2006) 1431e1436. [18] H.H. Cheng, J.F. Shen, C.S. Tan, CO2 capture from hot stove gas in steel making process, Int. J. Greenh. Gas Control 4 (2010) 525e531. [19] H. Barham, S. Brahim, Y. Rozita, K. Mohamed, Carbon steel corrosion behaviour in aqueous carbonated solution of MEA/bmim DCA, Int. J. Electrochem. Sci. 6 (2011) 181e198. [20] A.S. Joel, M. Wang, C. Ramshaw, E. Oko, Process analysis of intensified absorber for post-combustion CO2 capture through modelling and simulation, Int. J. Greenh. Gas Control 21 (2014) 91e100. [21] L. Agarwal, V. Pavani, D. Rao, N. Kaistha, Process intensification in HiGee absorption and distillation: design procedure and applications, Ind. Eng. Chem. Res. 49 (2010) 10046e10058. [22] Y.S. Chen, H.S. Liu, C.C. Lin, W.T. Liu, Micromixing in a rotating packed bed, J. Chem. Eng. Jpn. 37 (2004) 1122e1128. [23] BERR, Advanced power plant using high efficiency boiler/turbine (Report BPB010), BERR, Department for Business Enterprise and Regulatory Reform, 2006. Available at: www.berr.gov.uk/files/file30703.pdf (accessed April 2012). [24] A. Lawal, M. Wang, P. Stephenson, O. Obi, Demonstrating full-scale postcombustion CO2 capture for coal-fired power plants through dynamic modelling and simulation, Fuel 101 (2012) 115e128. [25] H.H. Tung, R.S.H. Mah, Modeling liquid mass transfer in HIGEE separation process, Chem. Eng. Commun. 39 (1985) 147e153. [26] K. Onda, E. Sada, H. Takeuchi, Gas absorption with chemical reaction in packed columns, J. Chem. Eng. Jpn. 1 (1968) 62e66. [27] J.R. Burns, J.N. Jamil, C. Ramshaw, Process intensification: operating characteristics of rotating packed beds d determination of liquid hold-up for a highvoidage structured packing, Chem. Eng. Sci. 55 (2000) 2401e2415. [28] Y.S. Chen, F.Y. Lin, C.C. Lin, C.Y. Tai, H.S. Liu, Packing characteristics for mass transfer in a rotating packed bed, Ind. Eng. Chem. Res. 45 (2006) 6846. [29] Y.S. Chen, Correlations of mass transfer coefficients in a rotating packed bed, Ind. Eng. Chem. Res. 50 (2011) 1778.



53



[30] Y. Luo, G.W. Chu, H.K. Zou, Z.Q. Zhao, M.P. Dudukovic, J.F. Chen, Gaseliquid effective interfacial area in a rotating packed, Ind. Eng. Chem. Res. 51 (2012) 16320e16325. [31] AspenTech, Aspen Physical Properties System e Physical Property Methods, 2010. Available at: http://support.aspentech.com/ (accessed May 2012). [32] M.S. Jassim, G. Rochelle, D. Eimer, C. Ramshaw, Carbon dioxide absorption and desorption in aqueous monoethanolamine solutions in a rotating packed bed, Ind. Eng. Chem. Res. 46 (2007) 2823e2833. [33] P. Sandilya, D.P. Rao, A. Sharma, G. Biswas, Gas-phase mass transfer in a centrifugal contactor, Ind. Eng. Chem. Res. 40 (2001) 384. [34] H.M. Kvamsdal, J. Hetland, G. Haugen, H.F. Svendsen, F. Major, V. Kårstad, G. Tjellander, Maintaining a neutral water balance in a 450 MWe NGCC-CCS power system with post-combustion carbon dioxide capture aimed at offshore operation, Int. J. Greenh. Gas Control 4 (2010) 613e622. [35] S. Freguia, G.T. Rochelle, Modeling of CO2 capture by aqueous monoethanolamine, AIChE J. 49 (2003) 1676e1686. [36] W.K. Lewis, W.G. Whitman, Principles of gas absorption, Ind. Eng. Chem. 16 (1924) 1215e1220. [37] R.T. Haslam, R.L. Hershey, R.H. Keen, Effect of gas velocity and temperature on rate of absorption, Ind. Eng. Chem. 16 (1924) 1224e1230. [38] R. Maceiras, E. Álvarez, M.Á. Cancela, Effect of temperature on carbon dioxide absorption in monoethanolamine solutions, Chem. Eng. J. 138 (2008) 295e300.



Nomenclature a: gaseliquid interfacial area (m2/m3) at: total specific surface area of packing (m2/m3) a0p : surface area of the 2 mm diameter bead per unit volume of the bead (1/m) c: width of the square opening (mm) d: diameter of the stainless steel fibre (mm) D: column diameter (m) DG: diffusivity coefficient of gas (m2/s) DL: diffusivity coefficient of liquid (m2/s) dp: diameter of packing pore (m) G: superficial gas velocity (m/s) gc: gravitational acceleration or acceleration due to centrifugal field (m2/s) go: characteristic acceleration value (100 m2/s) H: height of packing (m) kG: gas phase mass transfer coefficient (m/s) kL: liquid phase mass transfer coefficient (m/s) L: superficial liquid velocity (m/s) MEA: monoethanolamine QL: volumetric flow rate of liquid (m3/s) R: radial position (m) ri: inner radius of the packed bed (m) ro: outer radius of the packed bed (m) rs: radius of the stationary housing (m) T: temperature (K) U: superficial flow velocity (m/s) Uo: characteristic superficial flow velocity (1 cm/s) Vi: volume inside the inner radius of the bed ¼ pri2 Z (m3) Vo: volume between the outer radius of the bed and the stationary housing ¼ pðrs2 � ro2 ÞZ (m3) Vt: total volume of the RPB ¼ prs2 Z (m3) yCO2 ;in : mole fraction of CO2 in inlet stream yCO2 ;out : mole fraction of CO2 in outlet stream Z: axial height of the packing (m) Greek letters 3 L:



liquid holdup



m: viscosity (Pa s) rL: liquid density (kg/m3) rG: gas density (kg/m3) s: liquid surface tension (N/m) sc: critical surface tension (N/m) sw: surface tension of water (kg/s2)



vL: kinematic liquid viscosity (m2/s) vG: kinematic gas viscosity (m2/s) u: angular velocity (rad/s) Dimensionless groups FrL: liquid Froude number (L2at/gc) GrG: gas Grashof number ðd3p gc =n2G Þ GrL: liquid Grashof number ðd3p gc =n2L Þ ReG: gas Reynolds number (G/atnG) ReL: liquid Reynolds number (L/atnL) ScL: liquid Schmidt number (nL/DL) WeL: liquid Webber number (L2rL/ats) 4: c2/(d þ c)2



Applied Thermal Engineering 74 (2015) 54e60



Contents lists available at ScienceDirect



Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng



Characterization of the oxy-fired regenerator at a 10 kWth dual fluidized bed calcium looping facility Glykeria Duelli (Varela)*, Ajay R. Bidwe, Ioannis Papandreou, Heiko Dieter, Günter Scheffknecht Institute of Combustion and Power Plant Technology (IFK), University of Stuttgart, Pfaffenwaldring 23, 70569 Stuttgart, Germany



h i g h l i g h t s � CaL experiments at a 10 kWth DFB system under high CO2 concentration. � Performance of the regenerator in terms of calcination efficiency is studied. � Regenerator efficiency is correlated to reactor geometry and lime properties. � Lime chemical decay over time is recorded. � Material losses are measured in realistic CFB conditions.



a r t i c l e i n f o



a b s t r a c t



Article history: Received 30 August 2013 Accepted 17 March 2014 Available online 26 March 2014



Calcium looping is a CO2 capture technology, developed in the last years with great potential to contribute to the reduction of greenhouse gases emissions. It consists of a carbonator where CO2 contained in power plant flue gas, is absorbed by CaO producing CaCO3 and a regenerator where the CaCO3 is calcined to CaO releasing a highly concentrated CO2 stream. A key aspect of the process is the oxycombustion of the fuel in the regenerator to provide the heat for the endothermic calcination reaction. This implies high CO2 concentration in the off-gas and requires a flue gas recycle. Aim of this work is the characterization of the regenerator operation under this environment and thus experiments were performed at the 10 kWth IFK Dual Fluidized Bed facility. Regenerator efficiency and sorbent performance in terms of chemical activity and mechanical stability are the two basic parameters investigated, while the active space time is used for interpretation of the results. It is found that high calcination conversions can be achieved with temperatures less than 920 � C. Sorbent achieves a residual CO2 carrying capacity similar to the one referred to the literature. The sorbent proved to be mechanically strong thus optimistic percentage of material loss was recorded to be 0.8 wt.%/h or 0.024 mol Ca/mol CO2.  2014 Elsevier Ltd. All rights reserved.



Keywords: CO2 capture Calcium looping Fluidized bed Oxy-combustor/regenerator



1. Introduction Carbon dioxide is a major greenhouse gas produced at large quantities from fossil fuel combustion, especially at power plants. The global energy demand in 2040 is expected to be increased by 30% in comparison to 2010, while 40% of energy consumption will be due to electricity generation [1]. This demand will be mostly covered by fossil fuel power plants, while coal-fired power plants are responsible for 33% of the world’s CO2 emissions [1]. Therefore the need for efficient ways of making such power plants environmentally friendly is imperative. This is why the focus has been * Corresponding author. Tel.: þ49 (0) 711 685 67803; fax: þ49 (0) 711 685 63491. E-mail address: [email protected] (G. Duelli (Varela)). http://dx.doi.org/10.1016/j.applthermaleng.2014.03.042 1359-4311/ 2014 Elsevier Ltd. All rights reserved.



shifted to extensive research in the development of Carbon Capture and Storage (CCS) technologies, including the calcium looping process. The calcium looping process, which is already studied for syngas CO2 removal from the 1960’s [2], was firstly proposed by Heesink and Temmink in 1994 [3] as one of the zero emission coal technologies. The general process scheme is shown in Fig. 1 and takes advantage of the reversible carbonationecalcination reaction (1).The process is realized in a dual fluidized bed system (DFB). CaO(s) þ CO2(g) 5 CaCO3(s), DH25� C ¼ �178.2 kJ/mol



(1)



The separation of the CO2 is done by means of the exothermic carbonation reaction of CaO so that CaCO3 is formed and CO2 lean



G. Duelli (Varela) et al. / Applied Thermal Engineering 74 (2015) 54e60



Fig. 1. Calcium looping general process scheme.



gas stream is produced. The regeneration of the CaO makes use of the endothermic calcination reaction and is carried out in the regenerator. In the regenerator coal is burnt under oxy-combustion conditions to drive the high temperature calcination reaction [4]. A CO2 rich gas stream is produced and after purification is ready for storage. Fresh lime is added continuously to the system to compensate for loss of the material’s chemical activity [5], but also for mass losses due to attrition phenomena [6]. The O2/CO2 combustion process provides an additional advantage and makes the technology economically competitive [7e10]. Results from bench scale facilities [11e13] as well as from pilot plant tests [14e17] confirm the feasibility of the concept and provide data for the process commercialization. The investigations performed up to now are mainly focused on the carbonator operation as well as the calcination reaction [18e 20] and recently a model for the oxy-fuel combustor/regenerator was proposed [21]. However most of the studies are performed under air-fired conditions thus there is lack of data from a large scale facility where the regenerator is operating under realistic process conditions. For this purpose, IFK performed experiments at a 10 kWth DFB facility investigating the regenerator performance under high CO2 volumetric concentration as imposed by the oxyfuel combustion.



The performance of the regenerator is of significant importance for the whole process as it determines the quality and quantity of the sorbent delivered to the carbonator. As it can be seen in Fig. 1, a molar flow of particles (FCa) with a carbonate content (Xcalc) is entering the carbonator and after capturing part of the CO2 ðFCO2 Þ is directed to the regenerator with a carbonate content of Xcarb. Consequently and by implying simple mass balance equations the amount of CO2 captured in the carbonator is:



FCO2 $ECO2 ¼ FCa $ðXcarb � Xcalc Þ



cause sorbent deactivation as well as ash accumulation resulting in high make-up demands [24,25]. In both cases the heat demands would be increased [26]. Higher residence times would theoretically allow complete calcination but as reported in the literature [24] the average lime CO2 carrying capacity (Xave) would be decreased due to excessive sintering. Considering these facts the temperature should be kept as low as possible allowed by the thermodynamic equilibrium, in order to decrease the heat demands of the regenerator but also to guarantee high degrees of sorbent calcination conversion. Several formulas based on different models are proposed by the literature to calculate the calcination reaction rate [23,27,28]. Recent TGA kinetic studies on calcination reaction of Ca based sorbents for calcium looping applications showed that limes follow an apparent homogeneous conversion pattern at particle level [18]. Thus the rate of calcination proposed by Fang et al. [29] based on the grain model by Szekely and Evans [30] is applied as derived from the following equation:



� � � X � Xcalc 2=3 � rcalc ¼ k$ 1 � carb $ Ceq: � CCO2 Xcarb



(3)



. CCO2 eq: ¼ 4:137$107 $expð�20; 474=TÞ$rg MCO2



(4)



In Equation (3) the kinetic constant k (kmol/m3 s) depends on the lime type and the temperature [20]. CCO2 is the concentration of the CO2 in the reactor and depends on the fuel type as well as the amount of recirculated flue gas. Ceq. is the equilibrium CO2 concentration and is calculated through the Equation (4) below as proposed by Baker [27]:



The regenerator efficiency is defined as per Rodríguez et al. [12] and Charitos et al. [19] as the fraction of the calcined particles, where Xcarb is the carbonate content of particles entering the regenerator and Xcalc carbonate content of particles exiting the reactor (Fig. 1):



Ereg ¼



2. Theoretical background



(2)



This equation indicates that higher the carbonate content of the particles exiting the carbonator (Xcarb) and the lower the carbonate content of the particles entering the carbonator, namely Xcalc, the lower FCa will achieve a certain CO2 capture efficiency. Previous TGA and small lab scale studies report that the high CO2 concentration in the regenerator where the calcination takes place enhances sintering thus the particles CO2 carrying capacity (Xave) is reduced [18,22]. It has been found that both the surface area and the porosity of the produced CaO are affected by the presence of CO2 [22]. Other studies have shown that an increase in the CO2 partial pressure results in a decrease in the calcination rate [23]. Under these limiting conditions, feeding fresh lime (make-up) in the system can compensate for the loss of sorbent activity by improving the sorbent’s carrying capacity. In addition, while high temperatures would ensure full calcination [23], they would also



55



Xcarb � Xcalc Xcarb



(5)



As shown in Fig. 1 above a molar flow of particles (FCa) with a carbonate content of Xcarb is directed to the regenerator. The two main process parameters that affect the regenerator efficiency [19] are the residence time (tres) and the sorbent’s carbonate content (Xcarb). Assuming that there is no make-up flow to the system, the particle residence time is defined as the ratio of the moles of Ca present in the regenerator and the molar flow of carbonated particles available for calcination:



tres ¼



NCa FCa



(6)



Charitos et al. [19] experimentally observed that the regenerator efficiency increases by decreasing the load of the reactor expressed as the amount of particles to be calcined in a certain time. This load is formulated as follows:



Lreg ¼



Xcarb tres



(7)



Moreover, the reaction rate is considered constant until it reaches a critical time tc after which it becomes zero (Equation (8)). This means that only the fraction of particles with a residence time less than the critical time is available for calcination (fa). This concept is proposed by Martinez et al. [20].The critical time is calculated using Equation (9) which results from integration of the Equation (3) for Xcalc ¼ 0. Finally assuming the fluidized bed



56



G. Duelli (Varela) et al. / Applied Thermal Engineering 74 (2015) 54e60



regenerator as a perfectly mixed reactor, this fraction available for calcination is expressed as per Equation (10):



rcalc ¼



Xcarb ; t < tc tc



(8)



rcalc ¼ 0; t � tc tc ¼



3$Xcarb   k$ Ceq � CCO2



fa ¼ 1 � expð � tc =tres Þ



(9) (10)



Under the consideration that full sorbent calcination is taking place in the regenerator and there is no continuous make up flow, the basic mass balance of the calcium looping system is formulated as per Equation (11). It indicates that the CO2 disappeared from the gas phase in the carbonator is released from the solid inventory in the regenerator and is disappeared from the solid phase entering the regenerator:



FCO2 $ECO2 ¼ NCa $fa $rcalc ¼ FCa $ðXcarb � Xcalc Þ



(11)



Based on Equations (11) and (5) the following equation can be obtained:



NCa $fa $rcalc ¼ FCa $Ereg $Xcarb



(12)



and when considering Equation (6) the regenerator efficiency is expressed as follows:



Ereg ¼



tres $fa $r Xcarb calc



(13)



Thereby, the characteristic active space time parameter proposed in Equation (14) is justified as a characteristic parameter and is used in this work for the interpretation of the experimental results.



sactive ¼



tres $fa Xcarb



(14)



3. Experimental The experiments were carried out in IFK’s 10 kWth DFB facility which is described in detail elsewhere [6] and shown in Fig. 2. The circulating fluidized bed reactor was used as the carbonator and the bubbling fluidized bed reactor as regenerator. Even though in process up-scaling a CFB reactor will be implemented, this set up was selected due to limitations imposed by the facility in terms of heat supply. A cone valve was used to control the mass flow from the carbonator to the regenerator. Both reactors are electrically heated and the desired process temperature is set through N type thermocouples. The LabView software program was used control the facility and the data. All gas flows are measured and controlled with mass flow controllers. The operational conditions of the experiments are summarized in Table 2. The used lime is originating from south Germany and the chemical composition as per Table 1 was determined by TITAN in a SRS 3400 from Bruker. The experimental procedure was as follows. Calcined lime was initially fed to the system, which was electrically heated up to 900 � C in the regenerator and 600 � C in the carbonator. During regeneration the bed was fluidized with N2 and a CO2 stream whereas the carbonator and the loop seals with N2. After coupling the reactors, hydrodynamic stability in terms of constant pressure



Fig. 2. The 10 kWth DFB facility: (1) carbonator e CFB riser (2) cyclones (3) candle filters (4) upper loop seal (5) cone valve (6) regenerator e BFB (7) lower loop seal (8) electrical heaters.



profiles and solid flows, was achieved by adjusting the loop seal aeration and the cone valve opening. When the facility was stable a stream of 14 vol.% CO2, (dry) was fed to the carbonator to simulate power plant flue gas. Steam and SO2 were not present during the experiments for the sake of simplicity. The carbonator and the regenerator operated under the fast and bubbling fluidization regime respectively. The data presented here were collected during steady state periods, defined as periods of time lasting at least 10 min during which pressure, temperature, outlet gas concentration and solid circulation rates remain constant. Freshly calcined lime was added to the system from time to time to make up for the losses due to attrition and the possible cyclone inefficiency. The reactor masses are calculated through the pressure drop profiles which are provided by transducers. The carbonator and the regenerator outlet gas were continuously measured by an ABB Advance Optima 2020 and an ABB Easy Line 3020 analyzer respectively. The circulation rates were measured manually through a by-pass pipe specially designed for this purpose. By closing a butterfly valve the solids are accumulating in a pipe and the time was measured up to the point that the solids reach a certain height in this pipe. Samples were collected after each steady state from both the regenerator and the carbonator outlet, at the bottom of the loop seals. The carbonate content of the samples and the average CO2 carrying capacity (Xave) was measured by means of a TGA 701 by LECO and an innovative TG analyzer developed in cooperation with Linseis Thermal Analysis at IFK laboratory respectively. The particle size distribution was determined through sieve analysis. 4. Results and discussion The experimental data validation was carried out by means of the simple carbon mass balance for the carbonator and the regenerator. The mass balance takes into consideration that the CO2



Table 1 Limestone chemical composition in wt.%. CaO



SiO2



MgO



Al2O3



Fe2O3



Na2O



TiO2



CO2



56.01



0.298



0.26



0.13



0.08



0.02



0.01



43.19



G. Duelli (Varela) et al. / Applied Thermal Engineering 74 (2015) 54e60



57



Table 2 Basic experimental conditions. Main parameters



Regenerator



Carbonator



Fluidization regime Bed pressure drop (mbar) Inlet CO2 in vol.% Temperature (� C) Solids tres (min)



Bubbling 25e30 0 up to 65 905/920 5 up to 8



Fast 75e80 14 630 2 up to 4



captured in the carbonator is released in the regenerator (Fig. 3a), while on Fig. 3b the measured vol. concentration of CO2 at the outlet of the regenerator is plotted vs. the calculated one through the CO2 released from the solids. As it can be seen the mass balance closes successfully for the most part, while deviations are due to errors during sampling, analysis and measurement of the circulation rates. Another important observation is that for the specific experimental set up limitations on the maximum achievable regeneration efficiency are implied. As indicated by Fig. 4, the maximum efficiency achieved was not more than 90% regardless of the reactor load. In this figure the efficiency is plotted vs. the active space time. For the calculation of the active space time the kinetic constant was



Fig. 3. Closure of the mass balance a) carbonator b) regenerator.



Fig. 4. Limitations on the regenerator efficiency.



derived by means of TG analysis and implying a fitting constant. The kinetic constant used will be presented in detail in a separate publication where the focus will be the validation of the regenerator model proposed by Martinez et al. It is observed that high residence times in the range of minutes (5e8 min) thus resulting in high active space times in the range of h are required to calcine the solid flow entering the regenerator. This calcination time is much higher when compared to the literature where calcination is completed within some seconds i.e. in TGA experiments [18,22,23]. This phenomenon may be related to the fluidization of the bed mass. It seems that the bubbling bed with low velocities result is low heat and mass transfer coefficient as well as gas-bypassing through the bubbles that allows higher local partial pressures of CO2 in the emulsion phase where actual calcination takes place. This explains the fact that even with low measured mean CO2 concentrations around 25% in vol. and temperatures as high as 920 � C the calcination is not completed independently of the active space time. Another reason can be that the quality of heat provided through the electrical heaters where the heat is transferred from the outside source inside the bed and not generated in the bed as when combustion takes place. In this case the heat transfer rate is low enough allowing for temperature differences in the different areas of the bed as well as in the particle itself. A justification for this assumption can be seen in Fig. 4, where by increasing the overall temperature, the efficiency increases. An additional explanation is related to the particle size, since literature reports resistances on the reaction that can be imposed with increasing the particle size [23] which for the used lime in these experiments was between 150 and 450 mm (dp50 ¼ 325 mm). To conclude; the above mentioned factors are controlling the calcination reaction in our test rig which justify the high solids average residence time (5e 8 min). In Fig. 5 the regenerator efficiency is plotted vs. the active space time for a mean CO2 volumetric concentration of 50%, for two different values of temperature. For the calculation of the active space time the kinetic constant was derived by means of TG analysis and implying a fitting constant. The kinetic constant used will be presented in detail in a separate publication where the focus will be the validation of the regenerator model proposed by Martinez et al. It is obvious that in this environment with high partial pressure of CO2 concentration the regenerator efficiency is an increasing function of the active space time and the temperature. For this bubbling electrically heated regenerator a minimum active



58



G. Duelli (Varela) et al. / Applied Thermal Engineering 74 (2015) 54e60



Fig. 5. The effect of active space time and temperature on regenerator efficiency.



space time of approximately 1.2 h (average solids residence time of around 7 min) is required to achieve a regenerator efficiency of more than 70%, while the efficiency is not drastically increased at 920  C. This means that for a specific type of lime by changing the carbonate content of the solids and the residence time the optimum point can be set for high regenerator efficiencies, which means that the particles delivered to the carbonator have little to none carbonate content so the available CaO to capture CO2 is maximized. In Fig. 6 the effect of the CO2 concentration on the regenerator efficiency is plotted for a temperature of 920  C, for two different active space time values. As expected the higher the partial pressure of CO2 during sorbent regeneration is, the lower the regenerator efficiency. But it can be seen that for this reactor there is a critical active space time of 1.35 h (respective solids residence time of around 8 min) for which the partial pressure influence appears to be minimized and the efficiency achieved is more than 80%. This finding is important as it indicates the optimum operational conditions for this facility in order to achieve almost full calcination,



Fig. 6. The effect of CO2 vol. concentration on regenerator efficiency.



taking into consideration the limitations we have mentioned in the previous paragraph. The average CO2 carrying capacity (Xave) is a metric to evaluate the lime’s ability to capture CO2. It is defined as the conversion that can be achieved by the average solid in the carbonator at the end of the fast reaction period [19] and is measured in a TGA. In Fig. 7 the evolution of the decay of Xave along the experimental hours (in continuous mode) is shown for two different levels of CO2 concentration. After 12 h of continuous operation of the dual system the sorbent residual activity stabilizes at approximately 10%. This residual activity is comparable to the one achieved from studies in the same facility for the air fired case [11]. It is noticed that after approximately 3 h of operation a batch of 1.5 kg was added to the system in order to compensate for mass losses at the cyclones and therefore the Xave increases and afterward it decreases to the previous recorded value. In comparison to the lime performance in TGA tests [31], the Xave decay is more rapid in DFB experiments due to the comparatively higher heating rates. Nevertheless, in both cases the residual activity is comparable. Another important finding is that the regeneration under high CO2 concentration does not accelerate or worsen the decay of the sorbent CO2 sorption capacity. This may be explained by the fact that in the DFB system the uncompleted calcination and subsequently the partially carbonated particles as well as the short residence times may reduce the sintering effects. Finally the mechanical stability of the sorbent was studied through the PSD analysis of the sorbents as well as measurements of the sorbent found in the cyclones. In Fig. 8a the evolution of the initial particle size and after 3 and 10 h of operation is depicted. The sorbents are collected from the carbonator and the regenerator outlet. It is generally found that the particles become smaller in the course of the experiment. In this experimental set up we have noticed no difference between the particle size of the material entering and exiting the regenerator. This may be happening because the regeneration takes place in a bubbling mode and the collision between the particles and/or the walls are less pronounced than in a turbulent fast fluidizing bed. Moreover as shown in Fig. 8b, after 10 h of continuous operation the dp10 of the particles is 125 mm while the losses calculated through the pressure drop of the reactors (as well as confirmed through the mass balance at the end of the experiment) are measured around 0.8 wt.%/h. This material loss would impose a make up requirement of 0.024 mol/h fresh limestone for each mol/



Fig. 7. The decay of the lime average CO2 carrying capacity.



G. Duelli (Varela) et al. / Applied Thermal Engineering 74 (2015) 54e60



59



value of 1.35 h (respective solids residence time of around 8 min) and a temperature of 920  C proved to be necessary for regenerator efficiencies around 90% for the specific experimental set up. Regarding the chemical activity of the sorbent, it is found that the sorbent achieves a residual activity of approximately 10% while the presence of high CO2 concentration does not have a special effect in the deactivation curve. Regarding the mechanical stability of the sorbent, a continuous decay of the mean particle size is recorded while very few fines are generated. The overall losses measured are approximately 0.8 wt.%/h and result in make-up demands of 0.024 mol Ca/mol CO2. The experimental data were validated by a successful mass balance closure. Small deviations are explained through errors during sampling and analysis of the sorbent as well as during measurement of the circulation rate. Limitations were imposed from the specific experimental set up to the maximum achievable regeneration efficiency. Full sorbent regeneration could not be achieved primarily due to the bubbling fluidization regime of the bed and the quality of the heat provided by the electrical heaters through the walls to the bed. These limitations impose solid residence times up to 8 min for the calcination of the solids instead of seconds as recorded in TGA investigations as well as industrial applications such as clinker production. This will not be the case when process up scaling where these limitations will not be present due to the fast fluidization regime and the combustion atmosphere, thus the data presented here is suggested to be treated qualitatively and not quantitatively. Acknowledgements We thank the RFCS Program of the European Commission (RFCR- CT- 2010- 00013) for funding this work under the CAL-MOD Project and Mrs Maria Elena Diego for her active participation in this work. Nomenclature Fig. 8. a) The cumulative particle size distribution b) the evolution of particle size.



h CO2 entering the carbonator. Compared to the numbers reported in the literature [6] these losses are lower. In the literature it is reported that the fines elutriation rate is relatively large after the first calcination and is decreased with the number of carbonatione calcination cycles, since the combined chemicalethermal treatment affects the particle structure making it increasingly hard [32,33]. The presence of a high CO2 concentration during calcination may have led to this low value of fines generation primarily due to the fact that the sintering hardens the particle surface [33] as well as the low calcination reaction rates, leading to lower internal particle pressures due to CO2 lower diffusion rate. Another reason may be the pre-calcination of the material as well as the regeneration under low fluidization velocities. The dp50 of the particles is around 275 mm and is proved to be adequate for a very stable and long-time continuous operation of the system. 5. Conclusions Experimental data from the 10 kWth DFB facility revealed the effect of the high vol. concentration of CO2 during sorbent calcination on the regenerator’s efficiency, and sorbent performance in terms of chemical activity and mechanical stability. The active space time as proposed by Martinez et al. is used as the main regenerator design parameter. A critical value active space time



FCa FCO2 ECO2 Ereg Xcarb Xcalc Xave rcalc k Ceq. CCO2



rg



MCO2 Lreg NCa tres fa tc



sact.



Ca looping rate between the reactors, mol/min CO2 flow to the carbonator, mol/min carbonator CO2 capture efficiency, e regenerator efficiency, % sorbent carbonate content entering the regenerator, mol CaCO3/mol Ca sorbent carbonate content exiting the regenerator, mol CaCO3/mol Ca average lime CO2 carrying capacity, mol CaCO3/mol Ca calcination reaction rate, 1/s kinetic constant of the calcination, m3/kmol s equilibrium CO2 concentration, kmol/m3 regenerator CO2 concentration, kmol/m3 gas density, kg/m3 CO2 molar mass, kg/mol regenerator load, min amount of Ca in the regenerator, mol solids mean residence time in the regenerator, min fraction of particles available for calcination, e necessary time for full sorbent calcination, min regenerator active space time, h



References [1] IEA, World Energy Outlook 2011, Executive Summary, 2011. www.iea.org. [2] G. Curran, C. Fink, E. Gorin, CO2 acceptor Gasification Process Studies of Acceptor Properties, Fuel Gasification. Advances in Chemistry 69, American Chemical Society, 1967, pp. 141e165.



60



G. Duelli (Varela) et al. / Applied Thermal Engineering 74 (2015) 54e60



[3] A. Heesink, H. Temmink, Process for Removing Carbon Dioxide Regeneratively from Gas Streams, patent: PCTWO94/01203, 1994. [4] T. Shimizu, T. Hirama, H. Hosoda, K. Kitano, M. Inagaki, K. Tejima, A twin fluidbed reactor for removal of CO2 from combustion processes, Chem. Eng. Res. Des. 77 (1999) 62e68. [5] D. Alvarez, J.C. Abanades, Pore-size and shape effects on the recarbonation performance of CaO submitted to repeated calcination/recarbonation cycles, Energy Fuels 19 (2005) 270e278. [6] A. Charitos, C. Hawthorne, A.R. Bidwe, S. Sivalingam, G. Scheffknecht, Parametric investigation of the calcium looping process for CO2 capture in a 10 kWth dual fluidized bed, Int. J. Greenhouse Gas Control 4 (5) (2010) 776e 784. [7] N. Poboß, A. Schuster, G. Scheffknecht, Feasibility Study of the Carbonate Looping Process for Removal of CO2 from Power Plant Flue Gases, Cooretec Final Report FKZ 0327771B, 2008. [8] C. Hawthorne, M. Trossmann, P. Galindo Cifre, A. Schuster, G. Scheffknecht, Simulation of the carbonate looping power cycle, Energy Procedia 1 (2009) 1387e1394. [9] T. Kuramochi, A. Ramírez, W. Turkenburg, A. Faaij, Comparative assessment of CO2 capture technologies for carbon-intensive industrial processes, Prog. Energy Combust. Sci. 38 (1) (2012) 87e112. [10] A. Martínez, Y. Lara, P. Lisbona, Luis M. Romeo, Energy penalty reduction in the calcium looping cycle, Int. J. Greenhouse Gas Control 7 (2012) 74e81. [11] A. Charitos, N. Rodríguez, C. Hawthorne, M. Alonso, M. Zieba, B. Arias, Experimental validation of the calcium looping CO2 capture process with two circulating fluidized bed carbonator reactors, Ind. Eng. Chem. Res. 50 (16) (2011) 9685e9695. [12] N. Rodríguez, M. Alonso, J.C. Abanades, A. Charitos, C. Hawthorne, G. Scheffknecht, Comparison of experimental results from three dual fluidized bed test facilities capturing CO2 with CaO, Energy Procedia 4 (2011) 393e401. [13] D.Y. Lu, R.W. Hughes, E.J. Anthony, Ca-based sorbent looping combustion for CO2 capture in pilot-scale dual fluidized beds, Fuel Process. Technol. 89 (12) (2008) 1386e1395. [14] C. Hawthorne, H. Dieter, A. Bidwe, A. Schuster, G. Scheffknecht, S. Unterberger, CO2 capture with CaO in a 200 kWth dual fluidized bed pilot plant, Energy Procedia 4 (2011) 441e448. [15] H. Dieter, C. Hawthorne, A. Bidwe, M. Zieba, G. Scheffknecht, The 200 kWth Dual Fluidized Bed Calcium Looping Pilot Plant for Efficient CO2 Capture, Proc.. of 21st Int. Conf. on Fluidized Bed Tech. Naples, 2012, pp. 397e404. [16] J. Kremer, A. Galloy, J. Ströhle, B. Epple, Continuous CO2 capture in a 1MWth carbonate looping pilot plant, Chem. Eng. Technol. 36 (9) (2013) 1518e1524. [17] B. Arias, M. Diego, J. Abanades, M. Lorenzo, L. Diaz, D. Martínez, et al., Demonstration of steady state CO2 capture in a 1.7 MWth calcium looping pilot, Int. J. Greenhouse Gas Control 18 (2013) 237e245.



[18] I. Martínez, G. Grasa, R. Murillo, B. Arias, J.C. Abanades, Kinetics of calcination of partially carbonated particles in a Ca-looping system for CO2 capture, Energy Fuels 26 (2) (2012) 1432e1440. [19] A. Charitos, N. Rodriquez, C. Hawthorne, et al., Validation of a Carbonator Model and Proposal of a Parameter Interdependence Scheme for the Calooping CO2 Capture Process, Proc. of 21st Int. Conf. on Fluidized Bed Tech. Naples, 2012, pp. 311e318. [20] I. Martínez, G. Grasa, R. Murillo, B. Arias, J.C. Abanades, Modelling the continuous calcination of CaCO3 in a Ca-looping system, Chem. Eng. J. 215 (2013) 174e181. [21] J. Ylätalo, J. Parkkinen, J. Ritvanen, T. Tynjälä, T. Hyppänen, Modeling of the oxy-combustion calciner in the post-combustion calcium looping process, Fuel 113 (2013) 770e779. [22] R. Borgwardt, Calcium oxide sintering in atmospheres containing water and carbon dioxide, Ind. Eng. Chem. Res. 28 (1989) 493e500. [23] F. García-Labiano, A. Abad, L. de Diego, P. Gayán, J. Adánez, Calcination of calcium based sorbents at pressure in a broad range of CO2 concentrations, Chem. Eng. Sci. 57 (2002) 2381e2393. [24] B. González, G. Grasa, M. Alonso, J.C. Abanades, Modeling of the deactivation of CaO in a carbonate loop at high temperatures of calcination, Ind. Eng. Chem. Res. 47 (2008) 9256e9262. [25] Y.Y. Lee, T. Hyppänen, A Coal Combustion Model for Circulating Fluidized Bed Boilers, Int. Conf. on Fluidized Bed Tech., 1989, pp. 753e764 [26] N. Rodríguez, M. Alonso, G. Grasa, J.C. Abanades, Heat requirements in a calciner of CaCO3 integrated in a CO2 capture system using CaO, Chem. Eng. J. 138 (2008) 148e154. [27] E.H. Baker, The calcium oxideecarbon dioxide system in the pressure range of 1e300 atmospheres, J. Chem. Soc. 70 (1962) 464e470. [28] G. Silcox, J. Kramlich, D. Pershing, A mathematical model for the flash calcination of dispersed CaCO3 and Ca(OH)2 particles, Ind. Eng. Chem. Res. 28 (1989) 155e160. [29] F. Fang, Z. Li, N. Cai, Experiment and modeling of CO2 capture from flue gases at high temperature in a fluidized bed reactor with Ca-based sorbents, Energy Fuels 23 (2009) 207e216. [30] J. Szekely, J.W. Evans, A structural model for gasesolid reactions with a moving boundary, Chem. Eng. Sci. 25 (1970) 1091e1107. [31] J. Blamey, E.J. Anthony, J. Wang, P.S. Fennell, The calcium looping cycle for large-scale CO2 capture, Prog. Energy Combust. Sci. 36 (2010) 260e279. [32] Z.X. Chen, C.J. Lim, J.R. Grace, Study of lime particle impact attrition, Chem. Eng. Sci. 62 (2007) 867e877. [33] A. Coppola, F. Montagnaro, P. Salatino, F. Scala, Attrition of lime during fluidized bed calcium looping cycles for CO2 capture, Combust. Sci. Techol. 184 (2012) 929e941.



Applied Thermal Engineering 74 (2015) 61e68



Contents lists available at ScienceDirect



Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng



Reprint of “Experimental studies of single particle combustion in air and different oxy-fuel atmospheres”  Ewa Marek*, Bartosz Swiątkowski Department of Thermal Processes, Institute of Power Engineering, Augustowka 36, 02-981 Warsaw, Poland



h i g h l i g h t s � Particle temperature during combustion was lower in O2/CO2 than in O2/N2 mixture. � Greater temperature differences were observed for coal then for char particles. � CO2 hindered volatiles release and inhibited particle swelling during combustion. � Presence of H2O in oxy-fuel atmosphere increased temperature of combusted particle.



a r t i c l e i n f o



a b s t r a c t



Article history: Received 31 August 2013 Accepted 30 January 2014 Available online 20 May 2014



In this work, direct observation of char and coal single particle combustion in different gases mixtures has been performed. Investigation focused on the influence of atmosphere composition on combustion process and especially on the comparison between combustion in air-like versus oxy-fuel dry and oxyfuel wet conditions. For these tests, particles from Pittsburgh coal and South African Coal were prepared manually to cubical shape (approximately 2 mm and 4 mg). To investigate fuel type influence on oxy-fuel  w mine. Experiments were combustion, some tests were also conducted for Polish lignite coal from Turo carried out in a laboratory setup consisted of an electrically heated horizontal tube operated at 1223 K with observation windows for high speed video recording (1000 frames per second). During the experiments, particle internal temperature was measured to obtain comprehensive temperatureetime history profile. Results revealed that particles burned at higher temperatures in high water vapour content mixtures than in dry O2/CO2 mixture. This behaviour was attributed to lower molar specific heat of water than of CO2 and four times higher reaction rate for chareH2O gasification reaction than chareCO2 reaction. Also visible dynamic of combustion process recorded with the high speed camera differs for experiments carried with water vapour addition. © 2014 Elsevier Ltd. All rights reserved.



Keywords: Single particle Oxy-fuel Coal Water addition Combustion



1. Introduction Oxy-fuel combustion is a technology introduced with aim to help reduce CO2 emission, which is especially urgent in recent times when demand for coal is still growing. In Poland, where more than 90% of electricity is generated from coal, the oxy-fuel technology with possible option of boilers' retrofitting, seems to be an especially attractive variant for CO2 mitigation. However, oxy-fuel technology is only at pilot-scale and the knowledge of



DOI of original article: http://dx.doi.org/10.1016/j.applthermaleng.2014.01.070. * Corresponding author. Tel./fax: þ48 22 642 8378. E-mail addresses: [email protected] (E. Marek), bartosz.swiatkowski@ien.  com.pl (B. Swiątkowski). http://dx.doi.org/10.1016/j.applthermaleng.2014.05.026 1359-4311/© 2014 Elsevier Ltd. All rights reserved.



combustion mechanisms in changed atmosphere can be still perceived as insufficient. Exhaust gas from oxy-fuel combustion contains mostly CO2 an H2O. Part of produced flue gas must be recycled to maintain proper heat exchange and safe operation within the boiler. Whether the recycled stream is dried or contains a significant amount of water is the matter of later optimization of combustion process as well as technical and economic analysis. But lately an agree is emerging, that at least some amount of water in recycled flue gases is inevitable [1,2]. So far a lot of effort was undertaken to investigate the difference between air and dry oxy-fuel combustion [2,3]. But it should be remembered, that H2O as well as CO2 can participate in char gasification reactions and from that point of view, possible interaction of H2O in oxy-fuel combustion process should be better understood.



62



E. Marek, B. Swiątkowski / Applied Thermal Engineering 74 (2015) 61e68



Char gasification reactions can significantly compete with combustion reactions but only under specific conditions. Those are high temperature and/or low oxygen concentration in gas mixture. In comparison to O2echar reaction, gasification either with CO2 or H2O is much slower and requires a lot of external energy to take place. Thereby, if amount of oxygen molecules near char surface is sufficient, quick and exothermic combustion reaction is always promoted. On the other hand, when gas temperature is high enough, O2echar oxidation is too quick for sufficient supply of oxygen molecules and reaction becomes limited by O2 diffusion (both internal and external). In this case, gasification can be promoted, because char surface is still surrounded by plenty CO2 and H2O molecules. Fig. 1 (adapted from Chen et al. [3]) presents the diagram that summarizes the conclusions from char oxidation and gasification experiments found in the literature. Diagram shows three temperatureeO2 concentration dependent regions, among which Regions B and C represent conditions in which gasification reactions are expected to noticeably contribute to char consumption, whereas in Region A only combustion reaction was found significant. Boundaries imposed on the regions are qualitative illustration only and are not conclusive, as emphasized by Chen et al. [3]. The questions that remain interesting are how and which gasification reaction influences char consumption more. While CO2echar reaction was widely investigated in oxy-fuel combustion conditions, only a little effort was taken to study influence of variable steam concentration in oxy-atmosphere on parameters of combustion process [1]. The aim of this work is to experimentally study and compare the behaviour of coal and char particles during high temperature combustion in 21 and 35% oxygen concentrations with different concentrations of CO2, H2O and N2, introduced as the diluent gases.



Fig. 2. Schematic diagram of Single Particle Combustion stand (SPC stand). 1. Coal/char particle, 2. Thermocouple, 3. Oil shield tube, 4. Reactor, 5. Gas inlet, 6. High speed camera, 7. Quartz window, 8. Gas outlet.



Table 1 Experimental matrix for experiments with coal and char particles at reactor temperature 950  C. Particle type



Char/coal Char/coal Char/coal Char/coal Char/coal Char/coal Char/coal



Atmosphere composition [%] O2



N2



CO2



H2Ovapour



21 35 21 35 21 21 21



79 65 0 0 0 0 0



0 0 79 65 64 54 44



0 0 0 0 15 25 35



2. Experimental setup 2.1. Single particle combustion stand description Research stand for ignition and combustion of single particle allows carrying out experiments of quick fuel-particle combustion in controlled temperature and demanded gas mixtures. A schematic idea of Single Particle Combustion stand (SPC stand) is shown in Fig. 2. The main part of the rig is the reactor zone, which basically is a horizontal furnace, electrically heated up to 1000  C (4), with observation windows at both ends. At the quarter of reactor length



Fig. 1. Dominant reactions in char oxidation and gasification experiments (adapted from Chen et al. [3]).



from quartz windows (7) are located two thermocouples used for heating control and setting of experimental temperature. Tip of a 0.5 mm thermocouple (2) was inserted into the hole drilled in the coal particle (1) and with the thermocouple as a support, the particle was then inserted into the movable oil-cooled shield tube (3) inside the reactor zone. This tube, located at the vertical axis of furnace, created cool space inside the reactor and protected particle prior to the experiment's beginning. When demanded temperature conditions were stable, shield lock was released resulting in quick removal of the screen-tube from the reactor zone. Since that moment, investigated particle was exposed to the high temperature and oxidizing gases and this was considered the beginning of experiment. Ignition and combustion of fuel particle was recorded by high speed camera (6), Phantom v310 with applied recording speed of 1000 fps (frames per second). Camera was activated simultaneously with shield lock release. Temperature of particle when placed into the cool shield tube before the experiment was about 110  C. Thus when the shield blockade was released, experiment started with particle heating up. Experiments were carried out with different gas mixtures (O2, N2, CO2, H2O, air), slightly above atmospheric pressure to prevent air leakage to the reactor zone. Gases from cylinders (O2, N2, CO2, air) passed through an electric pre-heater (not shown) where water was vaporized and all gaseous components were mixed and preheated before entering the reactor (5). Water was supplied to the pre-heater by a peristaltic pump while gases flows were setup with flow meters. After passing the reactor zone (8), gas mixture reached a FTIR analyser which provided additional control of mixture composition.



E. Marek, B. Swiątkowski / Applied Thermal Engineering 74 (2015) 61e68



63



Table 2 Properties of CO2 and H2O at 950  C and 1 atm with reference to N2 (from Aspen Properties database). Property Density Specific heat capacity Thermal conductivity Mass diffusivitya (binary diffusion of O2 in X) Absorptivity/emissivity a



kg/m3 kJ/kmol K W/m K m2/s



CO2



H2O



N2



O2



Ratio CO2/N2



Ratio H2O/N2



0.438 56.60 0.080 1.7E-04 >0



0.179 44.04 0.127 2.8E-04 >0



0.279 33.83 0.079 2.2E-04 0



0.319 35.75 0.086 e 0



1.57 1.67 1.00 0.79 e



0.64 1.30 1.61 1.29 e



From Ref. [8].



2.2. Experimental conditions The experimental matrix is presented in Table 1. Basically, experiments were carried out for both coal and char particles in 950  C with different atmosphere compositions (temperature was constrained due to capabilities of the heaters). Composed oxy-fuel atmosphere had different physical properties than air. Table 2 summarizes properties of gases used in this work, at experimental temperature. Gas mixture flow at reactor inlet was always 5 dm3/min thus the laminar flow of gas did not disturb volatiles release and burning. For such experimental conditions the average heating rate of particle was about 200 K/s. Although that slow heating rate is not comparable with industrial PC combustion (104e106 K/s), it allows to investigate sequential combustion of particle, which is essential for our studies. From 3 to 5 coal or char particles were combusted for every experimental setup. For lignite coal tests were limited to one series and coal particles only. Totally, about 110 experiments were performed. For every tested particle, video recording of particle combustion as well as particle internal temperature were obtained. Signal from 0.5 mm thermocouple was collected every 10 ms to receive a comprehensive ‘particle internal temperature-time history profile’. Char preparation was conducted in the SPC stand at 950  C with the nitrogen flow of 5e8 dm3/min. Single particle was put on top of a thermocouple and then introduce to the reactor with the oilshield tube closed. When the shield lock was released, the particle heating started, the increase of particle internal temperature was observed and species like CO2, CO and CH4 in exhaust gas were noticed. After the extinction of volatile products, the newly created char was kept inside reactor for some time to assure completion of the degassing process. Then, the screen tube was once again lowered and locked. Summarizing, particles were exposed in the hot furnace during this procedure for 2 min. Since the SPC stand was working at overpressure, nitrogen flowing through reactor was also present inside the shielding tube after its lowering and in that atmosphere devolatilized coal was cooled to temperature approximately 120  C. Due to this action no oxygen should have been absorbed on particle surface before char combustion in experiment. 3. Methodology 3.1. Particle preparation For the purpose of this study, particles from two bituminous coals: Pittsburgh and South African Coal were prepared manually. At first, fuels were dried for 24 h and the big coal lumps were crushed with the



Fig. 3. Coal particles used for experiments.



hammer. Then, in resulting smaller pieces (~1 cm) a hole was drilled with 0.5 mm drill. Finally, particles were shaped with diamond friction disks into approximately cuboidal solids, with an average size of ~2 mm and weight of 4 mg (Fig. 3). To compare the influence of the fuel rank on oxy-fuel combustion, few additional experiments were  w. Particles from this fuel also conducted for Polish lignite coal: Turo were prepared in the same way as for bituminous coals, but weight of a typical lignite coal particle with a size of ~2 mm, was 1.5e2 mg due to density differences. Results of proximate and ultimate analysis of fuels used in this study are shown in Table 3. Combustion of a relatively big particle cannot be directly compared to the combustion of PC in industrial units since some processes depend on the particle size, i.e. ignition mechanism, fragmentation, etc. Nonetheless, experiment with the use of 2 mm particle can provide more specific insight, especially when coupled with particle temperature profile measurements. Hence it can be of use for a development and validation of mathematical models. 3.2. Particle combustion Coal particle was first inserted into the shield-tube and when demanded atmosphere in the reactor was obtained, the shield lock was released and the particle was exposed. Combustion was considered complete when the temperature measured with the thin thermocouple dropped to the ambient temperature which was the reactor operating temperature and when there were no more visible changes happening within observed particle. Similarly, char particle combustion was conducted. After particle devolatilization and proper cooling inside the oil-shield tube, the atmosphere was switched from nitrogen to oxidizing mixture and then combustion was conducted in the same manner as for coal particles. Degassed chars were never removed from reactor before combustion experiment. 4. Results and discussion 4.1. Visual observations of the combustion When coal-particle heats up, volatile matter are released and at the temperature of volatiles ignition the gaseous yellow flame Table 3 Properties of the investigated coals. Fuel



Pittsburgh No8



South African Coal



Proximate analysis as received (on a dry basis) Moisture (%) 2.2 2.7 Volatile matter (%) 31.5 25.3 Fixed carbon (%) 53.0 57.7 Ash (%) 13.3 (13.6) 14.3 (14.7) LCV (MJ/kg) 28.35 26.01 Ultimate analysis (on a dry basis) Carbon (%) 74.2 71.4 Hydrogen (%) 4.8 4.0 Oxygen (%) (by diff.) 5.3 7.8 Nitrogen (%) 1.3 1.6 Sulphur (%) 0.89 0.63



w Turo 12.9 48.9 33.4 4.8 (5.5) 22.86 66.3 5.8 24.3 0.6 0.55



64



E. Marek, B. Swiątkowski / Applied Thermal Engineering 74 (2015) 61e68



Fig. 4. Pictures from high-speed recording of single particle combustion of Pittsburgh coal in different atmospheres (all experiments without water vapour addition and one experiment with 35% H2O). Numbers under frames indicate time-scale of particle combustion, in ms. Zero represents the beginning of combustion (first visible sign).



Fig. 5. Pictures from high-speed recording of single particle combustion of South African Coal in different atmospheres (all experiments without water vapour addition and one experiment with 35% H2O). Numbers under frames indicate time-scale of particle combustion, in ms. Zero represents the beginning of combustion (first visible sign).



E. Marek, B. Swiątkowski / Applied Thermal Engineering 74 (2015) 61e68



65



 w coal in different atmospheres (all experiments without water vapour addition and one experiment Fig. 6. Pictures from high-speed recording of single particle combustion of Turo with 35% H2O). Numbers under frames indicate time-scale of particle combustion, in ms. Zero represents the beginning of combustion (first visible sign).



appears which indicates the combustion process. This phenomenon corresponds to the homogeneous ignition and takes place away from the coal surface. Once the volatile flame extinguishes, the devolatilized particle starts to combust and that is the second stage of coal combustion. Sometimes both stages can happen at the same time, when the volatile flame is still present but particle surface also starts to glow. This is joint heteroehomogeneous combustion and usually can be observed for pulverized coal particles. Volatile-free char particles ignite and combust only in heterogeneous mechanism, like the described second stage of coal combustion, where reactions take place between gaseous oxygen and the particle surface. The visible sign of the progressing char combustion process is particle glowing. At first vertices and edges start to glow, then combustion progresses over the entire visible surface. In Figs. 4e6 pictures of combustion of single particles are presented (photos are selected from high speed recordings). First picture in row (0 ms) was taken at the moment of ignition which in this study was intended as the first visible sign of combustion. It is worth noticing, that the particle surface between homogeneous and heterogeneous combustion was relatively dark which assures that stages of particle combustion did not overleap and took place one after another. Visual observations from experiments of charparticle combustion are herein not presented because of identical nature as the heterogeneous stage of coal-particle combustion. 4.1.1. Bituminous coals Pittsburgh and SAC coal-particles combusted in similar way (Figs. 4 and 5). When experiments were conducted with 35% H2O addition, the particle ignited and slowly developed volatile flame, at first visible only near the solid surface. After 100 ms, the particle was surrounded by a translucent flame (the particle was still



visible) which front started to move away from the surface. At 700 ms the flame was very high and looked fully developed, but was weakly luminous which indicated that only a little amount of tars and soot appeared within this stage of combustion. At approximately 1000 ms, more rapid burning of the particle began. Volatiles streams were escaping from the particle very quickly, in different places, sometimes getting far away from the solid and instantaneously combust, which was seen as small explosions. Jets of volatiles created some kind of escaping routes within the particle, where almost all remaining volatiles flowed later on. A flame from quickly escaping volatiles was very bright which indicated that soot and tars were main burning species. This effective combustion took place until the flame extinguished and the first stage of combustion was over, which for presented particles happened approximately 2.5 s after the ignition. Then, the dark particle surface started to glow, which indicated that the char combustion was progressing. The temperature of particle was still rising but at the time when the whole particle was already glowing, the temperature reached maximum and maintained around this value for few seconds of char stable combustion (Fig. 7). As heterogeneous combustion was progressing, particle was being “used up”, its size visually reduced until only small remain of ash was present. At this moment the particle glow faded away and the temperature started decreasing till it reached the temperature of surrounding. Combustion in air and 21% O2eCO2 mixture was similar to the process described above. At first a volatile flame appeared and after its extinguishing, char combustion proceeded. In nitrogen diluent mixtures, the volatile matter was often released in a jet form and combusted very brightly. In O2/CO2 atmosphere the flame was more translucent, which means that less tar and soot combusted. Also volatiles flow was more obstructed by thicken surrounding



66



E. Marek, B. Swiątkowski / Applied Thermal Engineering 74 (2015) 61e68



Fig. 7. Sample SAC char-particles temperature profiles during combustion in different atmosphere compositions.



and the long flame tail was created (upwards or downwards). What was characteristic for experiments carried out in air, was that at the end of the volatile's combustion, significant changes within solid particle occurred. A lot of a fly ash was released and the particle was swelling. This means that the pressure inside the particle was very high and altered particle internal structure. Explanation for this phenomenon is probably connected with the higher flame and particle temperature in nitrogen diluent atmosphere than in CO2 atmosphere experiments (Fig. 8). Higher particle temperature accelerates devolatilization which increases pressure of volatiles trapped in the particle even more and causes the particle swelling. On the other hand, CO2 presence was found inhibiting to the particle swelling but beside the influence of lower particle temperature, mechanism of that inhibiting behaviour is not fully explained. Borrego and Alvarez concluded that CO2 may participate in the surface process of cross-linking which they believe reduces the swelling [5]. On the contrary e in increased oxygen atmosphere, the swelling behaviour was visible for both diluent gases. Then also in CO2 atmosphere, the pressure within the particle was very high because combustion in enriched O2 atmosphere in both diluent gases was quicker and took place closer to surface, resulting in faster particle heating.



When comparing the combustion behaviour of particles in nitrogen, carbon dioxide and mixture of water vapour/carbon dioxide, the last one looked similar both to the combustion in nitrogen and combustion in CO2. The conclusion that can be drown is that high amount of water vapour addition cancelled some part of CO2 influence on the combustion. Experiments with lower amount of water addition (herein not shown) visually looked similar to experiments carried in O2/CO2 atmosphere only. But 35% of water vapour content caused significant change to physical properties of oxidizing O2/CO2 mixture. In addition, visual similarity of combustion phenomena in N2 and H2O/CO2, indicates that in pursuing overall similarity between oxy-fuel and conventional combustion, recirculation of wet exhaust gases should be promoted. Slightly higher temperature during combustion in mixture with H2O addition, also means that the higher flue gas recycle ratio will be needed to match the temperature inside the retrofitted boiler in case where wet recirculation is considered. Also the higher recycle ratio is preferential because of better velocity distribution inside boiler and better convective heat transfer. The same observation regarding wet flue gas recirculation arises from Wall et al. [6] theoretical calculations and was pointed out in review by Toftegaard et al. [2]. 4.1.2. Lignite coal In general, for lignite coal, the homogeneous stage of particle combustion was quicker than for bituminous coals (Fig. 6). Flame was fully developed within 20e60 ms after ignition in air and mixtures containing high amount of water vapour. In O2/CO2 without H2O addition, particle surrounding by flame took longer, at least 60 ms. Once again in nitrogen diluent mixtures, a lot of fly ash was released which at the pictures looks like sparks at front of the flame. The same occurrence presented itself when particle burned in 35% H2Oe44% CO2e21% O2 mixture. Bright luminosity of the flame was observed in every experimental conditions, but within mixtures containing 21% of oxygen the particle was still visible, while in the higher oxygen concentration was veiled by the flame. In 35% O2eN2 atmosphere flame was present at the beginning of volatiles combustion, while in CO2 mixture its occurrence took place after 700 ms from ignition and lasted till flame extinguished. Very interesting phenomenon was observed in this atmosphere during the next stage of combustion. When particle started to glow, indicating char combustion, also gaseous particle surrounding started to glow. Luminescent areola lasted to the end of particle combustion. While char contains no more of volatile matter, this glow should be attributed only to reactions engaging gaseous species that are formed during incomplete heterogeneous combustion or gasification reaction. 4.2. Temperatureetime history profiles



Fig. 8. Particles average temperatures during combustion in N2 and CO2 diluent atmospheres.



Based on measurements from 0.5 mm thermocouple, temperatureetime history profiles for single particles combustion were obtained. Fig. 7 presents sample temperature profiles for typical SAC char particles in every experimental conditions. For all combusted particles, the average temperatures were obtained, taking the average temperature from 5 or 2.5 s of particle stable combustion for bituminous coals and lignite coal respectively. In case of  w coal, the shorter time was include to the calculation, because Turo particles of lower rank coal were more reactive thus combustion was quicker and the stable part of it lasted less than the stable part of bituminous combustion. Figs. 8 and 9 show comparison of the average particle temperature during experiments in all investigated oxidizer compositions Results are divided for more clarity into two diagrams: the first one compiles experiments in oxy-fuel and nitrogen diluent atmospheres while the second one presents data for



E. Marek, B. Swiątkowski / Applied Thermal Engineering 74 (2015) 61e68



Fig. 9. Particles average temperatures during combustion in oxy-fuel atmosphere in regards to H2O content.



oxy-fuel conditions only, in relation to the amount of water vapour addition. The highest temperatures during particle combustion, up to 1320  C were noticed for every fuel in experiments conducted in 35% O2/65% N2 mixture. When nitrogen was switched for CO2, particle temperature was lower, but difference was greater in case of coal particle combustion than in char particles combustion. The only distinction between coal and char particle combustion is devolatilization stage and volatile matter combustion. So it may be assumed that the presence of CO2 hinders volatile matter release and influences its combustion, resulting in lower combustion temperature. Similar results were observed for lower oxygen concentration (21% O2). When experiments were carried in air, temperature curves reached 1220  C. When again nitrogen was replaced with 79% CO2, temperatures of burning bituminous particles were usually the lowest temperatures from all profiles obtained in experiments (beside Pittsburgh char particles). In case of experiments with chars, the lower particle temperature when combusted in O2/ CO2 mixture than in O2/N2 mixture can be explained by changes in heterogeneous reactions. It can be caused either by more difficult O2 diffusion through CO2 molecules or by the gasification reaction between solid carbon and CO2. The Boudouard reaction may contribute to the char consumption but is strongly endothermic and demands 172 kJ energy for every reacted mol of solid C [1]. Because of this endothermic character, the gasification should lower the particle temperature which is consistent with presented results. On the other hand, limited O2 diffusion could also not be excluded, because experimental conditions chosen for this study fall on the border between region A and B, where in region B diffusion control of combustion is present as well as gasification (see Fig. 1). When particles burned with the water vapour addition, their temperature profiles were close to each other for every tested H2O w coal concentration. Only in case of Pittsburgh char and Turo particles all temperatures measured within H2O enriched experiments were almost identical. For the rest of tested fuels, temperatures did not vary significantly, but were the highest for the largest water content in oxidizer and the lowest in the dry oxy-fuel conditions. Gasification reaction that involves H2O is less endothermic than the reaction of carbon and CO2. The amount of energy needed for the H2Oechar reaction equals 131 kJ per mol of C solid [7], which is approximately 24% less than amount of energy necessary



67



for the CO2echar reaction. For the H2Oechar reaction also the activation energy is lower (230 kJ/mol for H2Oechar reaction versus 250 kJ/mol for the CO2echar reaction [1]) and this means that the H2O gasification reaction is more promoted. What follows from above facts is that when the H2Oechar gasification occurs, the particle temperature should be expected to be higher than when the CO2echar reaction takes place, which can explain results presented herein. One should remember that when the water vapour addition in the mixture was increased, at the same time CO2 concentration decreased, due to fixed O2 fraction. Observed temperature differences can be attributed to the four times higher reaction rate for the H2Oechar gasification than the CO2echar gasification [4] as well as to H2O lower than CO2 molar specific heat (Table 2). Despite the fact that lignite coal particles weighted less than bituminous coal particles, temperatures obtained during lignite combustion were only slightly lower than in case of higher rank coal combustion. In atmospheres with 35% O2 and in air, the temw particles almost equalled Pittsburgh char partiperature of Turo cles. On the other hand, in 21% O2 oxy-fuel conditions with and without H2O addition, the lignite particle temperature was always the same (around 1090  C) and was the lowest temperature of all particles. The temperature of lignite combustion was not sensitive to water vapour presence in oxidizer, but more tests should be performed for this type of coal to confirm these results. It was pointed out by Chen et al. [3], that the coal type can be an important factor that also should be taken into consideration if the temperature under oxy-fuel conditions should match the temperature in conventional combustion. Results presented herein indicate that N2 replacement with CO2 in the experimental mixture lowered lignite temperature less than bituminous coals tempera w, and around ture (temperature reduction around 40  C for Turo 60  C for bituminous). This would suggest that the higher flue gas recycle ratio is necessary in case of the lower rank lignite coal combustion than in higher rank coals but again further investigations are required to confirm these findings. 5. Conclusions Single particle combustion with 2 mm particles as tested objects can provide good insight into fundamental knowledge of combustion and ignition. Introduced SPC stand was used to investigate coal and char particles combustion in air and under oxy-fuel conditions, with and without additional water vapour content. Conclusions from this study may be summarized as follows:  Particle in O2/CO2 mixture burned with lower temperature than in N2 diluent atmosphere (beside Pittsburgh char particles). The highest temperature difference (70  C) was observed for bituminous coal experiments. Water vapour addition in oxy-fuel atmosphere increased particle temperature during combustion in case of Pittsburgh coal, SAC char and SAC coal particles. This behaviour can be attributed both to H2O lower than CO2 molar specific heat and more promoted, less energy demanding H2O gasification reaction.  Visual similarity was observed between particle combustion in air and in 35% H2Oe44% CO2e21% O2 mixture.  The halo effect observed during lignite coal particle combustion in high oxygen containing oxy-fuel environment is interesting and cannot be definitely explained. Whether there were gasification reactions involved and CO combustion took place away from particle surface, should be carefully considered and more extensively tested (for example with the use of more sophisticated optical and spectroscopic methods). Observed occurrence of intensive glow around lignite char and interpretation of this



E. Marek, B. Swiątkowski / Applied Thermal Engineering 74 (2015) 61e68



68



phenomenon should be considered as a subject of open question. Acknowledgements Presented research regarding bituminous coals was funded by the European Commission 7th FP through RELCOM project, No. 268191. Lignite coal investigation was sponsored by National Research Development Centre through Strategic Program, Grant No. SP/E/2/666420/10. Also contribution from Dr Jarosław Hercog is appreciated. References [1] E.S. Hecht, C.R. Shaddix, M. Geier, A. Molina, B.S. Haynes, Effect of CO2 and steam gasification reactions on the oxy-combustion of pulverized coal char, Combust. Flame 159 (2012) 3437e3447.



[2] M.B. Toftegaard, J. Brix, P.A. Jensen, P. Glarborg, A.D. Jensen, Oxy-fuel combustion of solid fuels, Prog. Energy Combust. Sci. 36 (2010) 581e625. [3] L. Chen, S.Z. Yong, A.F. Ghoniem, Oxy-fuel combustion of pulverized coal: characterization, fundamentals, stabilization and CFD modeling, Prog. Energy Combust. Sci. 38 (2012) 156e214. [4] M.F. Irfan, M.R. Usman, K. Kusakabe, Coal gasification in CO2 atmosphere and its kinetics since 1948: a brief review, Energy 36 (2011) 12e40. [5] A.G. Borrego, D. Alvarez, Comparison of chars obtained under oxy-fuel and conventional pulverized coal combustion atmospheres, Energy Fuels 21 (2007) 3171e3179. [6] T. Wall, Y. Liu, C. Spero, L. Elliott, S. Khare, R. Rathnam, et al., An overview on oxyfuel coal combustiondstate of the art research and technology development, Chem. Eng. Res. Des. 87 (2009) 1003e1016. [7] E.S. Hecht, C.R. Shaddix, A. Molina, B.S. Haynes, Effect of CO2 gasification reaction on oxy-combustion of pulverized coal char, Proc. Combust. Inst. 33 (2011) 1699e1706. [8] B.E. Poling, J.M. Prausnitz, J.P. O'Connell, The Properties of Gases and Liquids, McGraw-Hill, 2001.



Applied Thermal Engineering 74 (2015) 69e74



Contents lists available at ScienceDirect



Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng



Reprint of “Experiences in sulphur capture in a 30 MWth Circulating Fluidized Bed boiler under oxy-combustion conditions”  mez a, *, A. Ferna ndez a, I. Llavona a, R. Kuivalainen b M. Go a b



n Ciudad de la Energía (CIUDEN), II Avenida de Compostilla n 2, 24400 Ponferrada, Spain Fundacio Foster Wheeler Energia Oy, Relanderinkatu 2, 78201 Varkaus, Finland



h i g h l i g h t s  Sulphur capture efficiency (%) was higher in oxy-combustion compared to air-combustion in a 30 MW thermal CFB boiler using anthracite and limestone as sulphur sorbent.  For a Ca/S molar ratio higher than 2.6 there was barely any improvement on sulphur capture efficiency for both air-combustion and oxy-combustion conditions in a 30 MW thermal CFB boiler using anthracite and limestone as sulphur sorbent.  Optimum temperature for sulphur capture at a fixed Ca/S molar ratio is around 880e890  C under oxy-combustion conditions and for anthracite coal with limestone as sorbent in a 30 MW thermal CFB boiler.



a r t i c l e i n f o



a b s t r a c t



Article history: Received 28 August 2013 Accepted 7 January 2014 Available online 21 May 2014



CO2 and SO2 from fossil fuel combustion are contributors to greenhouse effect and acid rain respectively. Oxy-combustion technology produces a highly concentrated CO2 stream almost ready for capture. Circulating Fluidized Bed (CFB) boiler technology allows in-situ injection of calcium-based sorbents for efficient SO2 capture. CIUDEN's 30 MWth CFB boiler, supplied by Foster Wheeler and located at the Technology Development Centre for CO2 Capture and Transport (es.CO2) in Spain, is the first of its kind for executing test runs at large pilot scale under both air-combustion and oxy-combustion conditions. In this work, SO2 emissions under different scenarios have been evaluated. Variables such as limestone composition, Ca/S molar ratio and bed temperature among others have been considered along different test runs in both air-combustion and oxy-combustion conditions to analyse its influence on SO2 abatement. Fly and bottom ash, together with flue gas analysis have been carried-out. Desulphurization performance tests results are presented. © 2014 Published by Elsevier Ltd.



Keywords: CFB technology Oxy-combustion Limestone in-situ desulphurization



1. Introduction CO2 and SO2 from fossil fuel combustion are contributors to greenhouse effect and acid rain respectively. Carbon Capture and Storage (CCS) technologies are one of the options to mitigate the CO2 released into the atmosphere. CCS technologies involve capturing CO2 at large point sources, transporting it to a suitable location and storing it permanently in safe deep geologic formations. Among the three main CCS technologies, pre-combustion, post-combustion and oxy-combustion, the latest is the one that produces a high CO2 concentration flue gas stream at the outlet of the combustor. Oxy-fuel technology is currently undergoing rapid DOI of original article: http://dx.doi.org/10.1016/j.applthermaleng.2014.01.012. * Corresponding author. mez). E-mail address: [email protected] (M. Go http://dx.doi.org/10.1016/j.applthermaleng.2014.05.025 1359-4311/© 2014 Published by Elsevier Ltd.



development towards commercialization with a number of demonstration projects [1]. Circulating Fluidized Bed (CFB) boiler technology is adequate for oxy-combustion due to its operating flexibility and allowing in-situ injection of calcium-based sorbents for efficient SO2 capture. CIUDEN's 30 MWth CFB boiler, supplied by Foster Wheeler and located at the Technology Development Centre for CO2 Capture and Transport (es.CO2) in Spain, is the first of its kind for executing test runs at large pilot scale under both aircombustion and oxy-combustion conditions. In this work, SO2 emissions under different scenarios have been evaluated. Variables such as limestone composition, Ca/S molar ratio and bed temperature among others have been considered along different test runs in both air-combustion and oxycombustion conditions to analyse its influence on SO2 abatement. At typical CFB temperatures (800e950  C) and overall oxidizing conditions, CaSO4 is the favoured final product of the sulphation



70



mez et al. / Applied Thermal Engineering 74 (2015) 69e74 M. Go



reaction and is thermodynamically stable, although its stability decreases with increasing temperatures [2]. The sulphation process is normally viewed as continuing until significant blocking of external pores occurs, leading to the formation of an impenetrable CaSO4 shell which leaves a significant amount of unreacted CaO core [3]. The main objective of this study was to evaluate SO2 emissions at CFB installations working under both air-combustion and oxycombustion conditions when adding limestone and comparing the experiences with those results previously obtained at laboratory and small scale pilot plants. The influence of Ca/S molar ratio and bed temperature on sulphur capture for both air-combustion and oxy-combustion conditions was studied. This work presents an overview of the work done by CIUDEN under the FLEXIBURN project and only presents one part of the experiments carried out, demonstrating the capacity to reduce SO2 in CFB installations using anthracite fuel. 2. Experimental section 2.1. Materials Local Spanish anthracite was selected for this study. Table 1 gives average values for proximate and ultimate analyses of the coal. Crushed coal gives a particle size distribution (PDS) shown on Table 2. Limestone composition is shown on Table 3 and Fig. 1 shows its PDS. Silica sand with composition shown on Table 4 and its PSD below on Fig. 2 was used as inert bed material. 2.2. Experimental installation. CFB boiler The experimental installation consisted in a 30 MW thermal CFB boiler located at the es.CO2. Fig. 3 illustrates schematically the configuration of the CFB boiler. It is a natural circulation, balanced draft boiler designed for using air or a mixture of recycled flue gas and oxygen as oxidant (oxy-fuel). The major components of the boiler are: combustion chamber, solids separator (cyclone), loop seal and fluidized bed heat exchanger chamber with super heaters surfaces (INTREX™). Solid inputs to the boiler are fuel, limestone and bed material; also fly ash can be fed through the fly ash recycling system. No fly ash was recycled in the performed tests. These solid materials enter the furnace above the grid, where the primary oxidant is provided. The oxidant is also fed in different levels: primary oxidant upper nozzles and secondary oxidant (upper and lower nozzles). Part of the oxidant, so called high pressure oxidant, is also used to fluidized the INTREX and loop seal and as a conveying gas for pneumatic solid transport. Table 1 Anthracite proximate and ultimate analyses (average). Proximate analysis (wt. %) Moisture content (as received) Ash (dry basis) Volatiles (dry basis) Fix carbon (dry basis) Ultimate analysis (wt. % dry basis) C H N S HHV (Kcal/kg, wet basis)



4.95 34.14 7.57 53.34 58.33 1.95 0.79 0.98 5040



Table 2 PDS for crushed coal. PSD (mm)



Cumulative %



8e16 4e8 2e4 1.4e2 1e1.4 0.4e1 0.2e0.4 0.125e0.2 800 � C) and the free lime particles due to rapid decomposition break into fine particles with high surface area. At lower temperatures this reaction does not take place at all. Typical Northern Greece lignite carries a large amount of limestone that is well above the stoichiometric ratio for complete desulphurization (free CaO/S). Although desulphurization efficiency is well over 70%, free lime does not capture all SO2 from flue gas, unless the limestone concentration is extremely high.



The sulfur content in lignite is rather low and constant to around 0.5% on natural basis. Although as a percentage this value looks small it compares well with sulfur content in typical hard coal because lignite LHV is very low and in general it takes the same amount of sulfur to be fed to a boiler per kWh produced. Natural desuphurization in lignite mainly depends on soft sedimentary calcium carbonate content (not sulphur content itself) since flue gas rich in free lime has a great potential for desulphurization. Also, as told previously the deSOx reaction takes place inside the furnace and temperatures above 800 � C. Therefore low boiler load, that is lower flue gas flow rate, results in lower flue gas velocity. At lower Table 4 Typical Calcium carbonate Sedimentary formation. Calcium carbonate sedimentary formation (South Field mine)



Table 3 Typical lignite ash chemical analysis. Typical lignite ash chemical analysis SiO2



Al2O3



Fe2O3



TiO2



P2O5



CaO



MgO



Na2O



K2O



SO3



Free lime



31.0



12.9



6.7



0.7



0.3



40.0



4.6



0.3



0.7



2.9



9.4



Moisture, % CaCO3 SiO2 and Al2O3, % LHV kJ/kg CO2 dry basis %



35 59 6 �1.903 40



122



A. Efthimiadou et al. / Applied Thermal Engineering 74 (2015) 119e127 Table 5 Typical Lignite seam composition. Typical lignite seam composition Moisture % Ash dry basis % CO2 dry basis % LHV Kcal/kg



50.7 35.0 7.8 1.342



flue gas velocities, there is time for the desulphurization reaction to proceed to a further extent. Let us consider the boiler operation at full load, CS being the average sulfur content in lignite and Vg (Nm3/kg, at 6% O2, dry) the flue gas produced per kg of lignite feed (the typical value is 2.45). This flow rate is obviously independent of air ratio (l). The SO2 flue gas concentration is given by a simple mass balance by the equation:



. � . � � � CSO2 ¼ 2* 1 � hdeSOx *CS *106 Vg in mg SO2 Nm3 ; dry6%O2



In some rare cases local lignite that comes from rich in clay rather than calcium seams has produced SO2 concentration a little over 4.000 mg/Nm3 but this rare occurrence appears for a handful of hourly SO2 concentration values per year. On the other hand, low SO2 concentrations in flue gas occur very often and concentration below 200 mg/Nm3 are expected when there is low load operation (especially during night shift) or when the low load operation has to do with low quality (low LHV, rich in CO2 content). It turns out that the degree of desulphurization has to reach 96% for SO2 emissions to remain below the 2016 emission limit value (ELV). 5. Statistical analysis of SO2 flue gas concentrations As reported previously SO2 emissions depend not only on lignite sulphur content but lignite quality (LHV and ash quality and content) as well as load. At low loads flue gas velocity are reduced and desulphurization of flue gas is more efficient. On the other hand, low lignite quality is usually related to high calcium carbonate content, hence quality deterioration will lead to low load operation and therefore reduced SO2 emissions. A statistical analysis on hourly SO2 emission values is really enlightening due to the size of the sample. If we collect all hourly SO2 emissions values for five units of Agios Dimitrios Power plant fed with the same quality lignite, we conclude that distribution with respect to groups of 200 mg/Nm3 are similar, Fig. 4. The small



Fig. 3. Linear correlation of CO2 content on LHV.



difference has to do with the schedule that units are loaded on a daily basis and some particularities of boiler/burner design. Unit I is the first to lower load at non peak hours, that is, it operates at lower load, therefore emissions are lower in general. It is observed that over 50% of the total operating hours this unit is below the new emission limit. On the other hand Unit V, being the one with the higher nominal power load and more economical has high emissions since is usually loaded at high loads. Unit IV has high furnace temperatures and high flue gas velocities resulting in high SO2 emissions. Fig. 5 with hourly SO2 emission values shows how emissions are affected by load. At lower loads 90% of the operation time values are below 1.000 mg/Nm3. At peak loads emissions almost never exceed 3.000 mg/Nm3. This proves the claim that wet desulphurization is a rather extravagance solution for these aged units. 6. A proposal for SO2 abatement for lignite fed power plants Aged units with over 30 years of operation are definitely not the best candidates for wet flue gas desulphurization units. Retrofiting of old units is not always possible not only because of the high capital cost involved but also due to space limitations. Semi dry FGD cuts the capital cost by at least 30% and seems to be a rational solution but space limitation still exist. Induced Draft fan modifications may also be required. The photo of Fig. 6 shows the space restrictions that are involved in such modifications. In such complex cases with several limiting factors, the solution has to be complex and several factors have to be considered. Before going further into proposals let us take a another look at the whole population of SO2 emission values over a period of a year. Fig. 7 illustrates all emissions over 200 mg/Nm3 that one needs to take care of. 7. Lignite quality Over a period of a whole year, the lignite feed fluctuates within a large range. Even at low loads there are periods of high SO2 emissions. This calls for a thorough homogenization of the lignite fed to the Unit Silos. Homogenization depends on line lignite quality monitoring as well as large sized bunker for proper mixing. Power Plants bunker has a total storing capacity of 1.800.000 tones which is not large enough given the space needed, since lignite has poor quality and daily consumption exceeds 60.000 tones. However, the SO2 distribution shows that there exist a lignite feed that is appropriate for complete flue gas desulphurization without the need of any other desulphurization technology. It turns out from the investigation of the hourly emissions values that 40% of the operation time SO2 emissions are below 200 mg/Nm3 as well as 13.5% of the time of peak load operation. The ‘ideal’ lignite feed for minimum SO2 emissions at high loads is easily detected by lignite sampling from mill feeders during peak load operation and low SO2 emissions ( 3) the cost of sorbent is dominant compared to the investment cost (e.g. DSI). In natural desulphurization a Ca/S ratio can be derived as follows, with Ca denoting the available Calcium for desulphurization and S the sulphur capture:



Calcium captured by SO2 in flue gas þ calcium measured in ash as free lime Sulphur captured in flue gas



The percentage of the decrease in SO2 emissions with respect to the percentage of load decrease (from nominal load) is given in Fig. 11 as a collection of operation data. It can be safely assumed that a 20% reduction in load form the nominal value reduces the SO2 emissions in half (50%).



Table 6 Lignite characteristics and corresponding SO2 emissions produced. Typical lignite characteristics for daily lignite samples year 2011, Unit III (Agios Dimitrios power plant)



Moisture % Ash % Fuel % LHV Kcal/kg CO2 dry basis % SO2 Emissions mg/Nm3



Lowest LHV



Average LHV



Highest LHV



43.3 28.5 27.3 870 14.5 0



49.5 19.4 31.1 1.253 8.8 565



54.3 13.2 34.7 1.537 4.2 3.511



Calcium captured by sulphur dioxide in flue gas is calculated from total sulphur in fuel and SO2 emission level in flue gas at the stack (simple mass balances). In addition, free lime content in fly ash (as analyzed in the lab) provides the excess of free lime not reacted with SO2 in flue gas. It must be noted that CaO from ash chemical analysis is around 40% in total, while free lime in fly ash is normally around 10%. At power loads close to maximum load, the natural desulphurization efficiency depends on the Ca/S ratio. Since sulphur content in local lignite does not vary dramatically, it is the free lime availability in ash, that is in flue gas that determines the efficiency. Table 7 illustrates collected data from several maximum load trials (2009 and 2010). Similarly, at low loads, Table 8, efficiency is usually very high and on some situations reaches 95%, that is SO2 emissions at the stack are within the new EU limits. These results compare well with results obtained by DSI technologies [1] but obviously do not come even close to wet FGD where for all practical purposes the Ca/S ratio is close to 1 for an efficiency of >96% at any load.



A. Efthimiadou et al. / Applied Thermal Engineering 74 (2015) 119e127



125



11. A proposal for SO2 abatement in existing power plants Existing power plant with a limited service life of less than 20 years may face opt out by the end of this decade. Open cast mines have a potential to support the existing power plants for well over 20 years and the shutdown of the existing power plant will have a tremendous long term effect on the economy of the region. With limited resources and funds for new investment there is an urgent need for the existing plants to operate for base load support. The economics and technology for wet FGD are well understood and Table 9 summarizes the parameters and economics for such an investment: In order to develop a tool for the comparison of a wet FGD option with a potential DSI installation there is a need for several assumptions:



Fig. 8. Location of DSI installation and injection point.



When DSI technologies are involved, adiabatic saturation temperature plays the most important role in desulphurization efficiency. When flue gas temperature is possible to be lowered to around 65e70 � C, the efficiency comes close to wet FGD for quite impressive Ca available/S removed ratios of less than 2. However if flue gas temperature cannot be lowered as much as it should be, Ca/S ratio of 4.5 to 5.0 are experienced when temperatures are kept to a minimum of 130 � C. From the above results it is obvious that in order to avoid wet FGD, low load operation is necessary.



� Duct sorbent injection can fight SO2 emissions of 1.000 mg/Nm3 dry 6%O2 (maximum) with a ratio of Cainjected/Sremoved ¼ 5.0 � If SO2 emissions exceed the above mentioned value of 1.000 mg/ Nm3 dry 6%O2 power load must be lowered accordingly for the SO2 emissions to reach 1.000 � The prediction of the load drop will follow the rule of 20% power load drop per 50% SO2 emissions drop (established from daily operational data). This rule holds for % drop from the nominal power and in case of lower loads it will be safer to assume that it also holds true since flue gas velocities will drop even further. � The calculations for low load operation and load drop will be performed on annual basis of hourly values (emissions and load). For the annual operation of the unit under investigation (Unit III, Agios Dimitrios Power Plant) the SO2 emission profile for all 7.861 h of operation is as illustrated in Table 10. Based on the above assumptions the design characteristics of DSI (duct sorbent injection) are listed in Table 11. Using the assumptions from Tables 9 and 11 and comparing the net present value of both investments (wet FGD and Duct Sorbent Injection) for an interest rate of 6% (actually considered rather low for the current economic situation in this part of Europe) the DSI option offers obvious advantage. Net present values of both investments for 20 years of payout time and interest rate of 6% show that wet FGD NPV is nearly double. 12. Conclusions



Fig. 9. a. Significant SO2 emission drop at low load operation during the night (due to the grit). b. SO2 emission reduction: low load operation due to bad lignite quality.



Northern Greece lignite, rich in sedimentary limestone content carries a high potential for natural desulphurization inside the boiler. The rest of the desulphurization needed for Northern Greece Power plant to comply with the EU directive in 2016 has to be carried out with low cost investment rather than wet FGD. The desulphurization of flue gas to the desired percentage of 96% is not achievable with DSI (duct sorbent injection) unless flue gas temperature is very low (around 70 � C). In existing utilities boilers and space limitations, it is not possible to accommodate this demand. Therefore the only option is to reduce power load in cases of high SO2 emissions in order to lower emission to the vicinity of 1.000 mg/Nm3 so that DSI can be successful. As it turns out the energy consumption for wet FGD is similar to the annual power load drop needed for DSI to successfully lower emissions to 200 mg/Nm3. Although the sorbent cost is tremendous compared to limestone cost needed for wet FGD, the capital needed for wet FGD is extremely high for such a small payout time to compensate for the other costs involved. In essence wet FGD is a capital intensive process compared to DSI which is operation cost intensive in terms of the sorbent supply.



126



A. Efthimiadou et al. / Applied Thermal Engineering 74 (2015) 119e127



Fig. 10. SO2 emissions reduction with load reduction although lignite quality is constant (mill switching).



Table 9 Main design characteristics for new wet FGD in existing plant. Main design characteristics for new wet FGD in existing plant Power (MW) Maximum SO2 emission (mg/Nm3, dry) Annual Energy Consumption (MWh) Hours of operation (hr) Ca/S (stoichiometric) Annual quantity of S to be removed Hourly quantity of S to be removed Estimated limestone Consumption (tn) Limestone Cost (V/t) Annual Maintenance cost (106 V, 2012) Labor (106 V, 2012/year) Investment Cost (106 V, 2012) Major retrofitting (IDFs, Cooling Tower) Service life (years)



310 3.500 62.000 8.000 1.02 3581 tn/year 0,4476 tn/hr 22.800 5 1 0.3 80 20



Fig. 11. Data points collected during load reduction with subsequent SO2 emissions reductions.



Table 7 Natural desulphurization efficiency related to Ca/S ratio at max load. Ca/S



Desulphurization efficiency %



Maximum load (MWe)



1.87 2.21 2.61 2.81 2.93 3.24 3.41



68.6 86.0 90.0 82.5 83.8 89.3 91.7



316 299 316 305 306 314 316



However, DSI has an inherent risk involved since lignite quality (sulphur content and specifically limestone content) has a degree of uncertainty. In case lignite carries less than expected limestone, power load losses may turn out to be too high to compensate for the desulphurization capacity needed. Lignite feed needs to be completely homogenized in order to avoid high SO2 emissions peaks. This calls for smart system for



Table 10 Operation Characteristics for DSI operation. Operation characteristics for DSI operation



Table 8 Natural desulphurization efficiency related to Ca/S ratio at low load operation. Ca/S



Desulphurization efficiency %



Low load (MWe)



1.86 1.94 1.98 2.13 2.14 2.49 2.53 2.78



81.3 94.9 90.1 98.8 95.4 99.4 99.3 97.8



268 276 269 239 276 263 280 283



Hours of operation Maximum SO2 emission (mg/Nm3, dry 6% O2) Average SO2 emission (mg/Nm3, dry 6% O2) Hours of operation with emissions200 & 1000 & 2000 mg/Nm3



Electrical Energy loss due to DSI limitations (MWh)



7.691 3.511 565 3.069 no action 2.934 (SDI) 1.483 (SDI with load drop