packed bed chemical-looping combustion : … packed bed chemical-looping combustion, experimental...
TRANSCRIPT
Packed bed chemical-looping combustion : experimentaldemonstration and energy analysisHamers, H.P.
DOI:10.6100/IR783193
Published: 01/01/2015
Document VersionPublisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the author's version of the article upon submission and before peer-review. There can be important differencesbetween the submitted version and the official published version of record. People interested in the research are advised to contact theauthor for the final version of the publication, or visit the DOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and page numbers.
Link to publication
Citation for published version (APA):Hamers, H. P. (2015). Packed bed chemical-looping combustion : experimental demonstration and energyanalysis Eindhoven: Technische Universiteit Eindhoven DOI: 10.6100/IR783193
General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ?
Take down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.
Download date: 04. Jun. 2018
PACKED BED CHEMICAL-LOOPING COMBUSTION
experimental demonstration and energy analysis
First press
© 2014 Paul Hamers
All rights reserved. No part of this publication may be reproduced, stored in a retrieval database or published in any form or by any means, electronic, mechanical or photocopying, recording or otherwise, without the prior written permission of the author.
The research reported in this thesis was sponsored by cato2 with ECN and TNO as project partners under the project number WP1.3F2.
Front cover drawn by Marie-Josée Hamers, www.art-emjee.nl
Publisher: Proefschriftmaken.nl || Uitgeverij BOXPress
A catalogue record is available from the Eindhoven University of Technology Library
ISBN: 978-90-386-3757-0
PACKED BED CHEMICAL-LOOPING COMBUSTION
experimental demonstration and energy analysis
PROEFSCHRIFT
ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de rector magnificus prof.dr.ir. C.J. van Duijn, voor een commissie
aangewezen door het College voor Promoties, in het openbaar te verdedigen op woensdag 7 januari 2015 om 16:00 uur
door
Hubertus Paulus Hamers
geboren te Winschoten
Dit proefschrift is goedgekeurd door de promotoren en de samenstelling van de promotiecommissie is als volgt:
voorzitter: prof.dr.ir. J.C. Schouten
1e promotor: prof.dr.ir. M. van Sint Annaland
copromotor(en): dr. F. Gallucci
leden: prof.dr. P. Canu (University of Padova)
dr. J.C. Abanades (Spanish Research Council)
prof.dr.ir. E.J.M. Hensen
dr.ir. D.W.F. Brilman (University of Twente)
adviseur(s): P.D. Cobden BSc (ECN)
Summary Packed bed chemical-looping combustion, experimental demonstration and energy analysis
Concerns about the role of anthropogenic greenhouse gas emissions on the observed climate change have resulted in great research efforts to develop new technologies that reduce the anthropogenic emission of especially carbon dioxide into the atmosphere. Among the possible solutions proposed, the capture, transport and storage of CO2 (CCS) can contribute to up to 20% of the total emission reduction. The large scale exploitation of CCS is hampered by the high energy penalty (and thus the costs) mainly related to the CO2 capture step. The focus of most of the research on CCS has been related to large scale, concentrated point sources of CO2, and thus in particular the power sector, where scale can play a role in reducing costs and increasing efficiency.
In recent years, many new technology approaches for CCS in power production with reduced primary energy requirements have been suggested. One possibility, which inherently combines power production with CO2 capture in a single process, is Chemical Looping Combustion (CLC). With CLC, the fuel is indirectly combusted by reacting the fuel with an oxygen carrier (metal oxide) that is used to separate oxygen from air in a separate, preceding reaction step. In this way, an almost pure CO2 product stream can be obtained after steam condensation directly from the combustion without additional energy intensive gas separation steps. During the highly exothermic oxidation of the metal, a hot gas stream is produced and this can be converted into power. Higher efficiencies than other capture technologies can be achieved, if the CLC system delivers a hot stream (>1200 °C) to the gas turbine at elevated pressures (ca. 20 bar).
In this work, packed bed reactors have been selected for CLC to better accommodate the combination of high temperatures and high pressures since in packed beds gas/solid separation steps (very challenging at the desired operating conditions) are not required. More specifically, the application of CLC in pressurized packed bed reactors is developed using syngas (from gasified coal) as fuel.
For the exploitation of chemical-looping combustion in pressurized packed bed reactors, the understanding of the effect of the pressure on the oxygen carrier reactivity is crucial. Chapter 2 describes a detailed study to the pressure effect in a magnetic suspension balance operated at high temperatures and up to 20 bar for CuO/Al2O3 and NiO/CaAl2O4 particles (which are typical oxygen carriers suggested in the literature). The experiments have demonstrated that the pressure has a negative influence on the reactivity of both oxygen carriers and this might be caused by the decrease in the number of oxygen vacancies at higher pressures, thereby decreasing the oxygen transport in the grains.
Moreover, the reactant gas fraction has been shown to be an important variable. These effects have been included in the kinetic model leading to a well matching description of the experimental results.
Another important aspect for packed bed CLC is the fact that larger oxygen carrier particles are required compared to fluidized bed applications because of pressure drop considerations. To study the influence of the use of bigger particles on the efficiency of the process, a detailed particle model that considers diffusion limitations and kinetics has been developed. It has been shown in chapter 2 that the impact of diffusion limitations decrease with increasing pressure, due to the decrease in reaction rate and increase in the fluxes caused by increased Knudsen diffusion. The results have been validated by experiments with 1.7 mm NiO/CaAl2O4 particles. Considering these effects, the larger required particle size for packed bed CLC only leads to a limited decrease in effective reaction rates when operated at higher pressures, which is beneficial for the application in packed bed reactors.
Chapter 3 and 4 report on experimental demonstrations of the CLC concept in a lab scale pressurized packed bed reactor using CuO/Al2O3, NiO/CaAl2O4 and ilmenite as oxygen carriers and H2, CO or syngas as fuel. The pressure has been varied between 2 and 7.5 bar (and constant reactant fraction) and it has been demonstrated that the pressure has quite a small effect on the performance of the reactor. The experimental results are described well by a developed reactor model, which shows that the desired high temperatures can be reached in case the reactor is scaled up. Steam addition has been shown to be an effective method to avoid carbon deposition, which would otherwise reduce the carbon capture efficiency. However, the heat management strategy has to be adapted because of the heat produced by the water gas shift reaction.
For CLC, a temperature rise from 450 °C to 1200 °C is required during the oxidation step. This makes the process very demanding for the reactor material itself and in particular for the oxygen carrier particles. The oxygen carrier should thus have high chemical and thermal stability, in combination with a high reactivity over the entire temperature range (450-1200 °C). A novel approach has been introduced in chapter 5 in which the demands on the oxygen carriers are alleviated by carrying out the process in two separate packed beds placed in series, obtaining a so-called Two-Stage CLC (TS-CLC). The first temperature rise (up to around 400 °C) is achieved in the first bed and the final temperature of 1200 °C is reached in the second bed. The oxygen carrier in the first bed should be reactive (in particular for the reduction) at low temperatures (for example a Cu-based oxygen carrier) and the second oxygen carrier should be stable and highly selective at high temperatures. The advantages of this configuration is that the temperature change in each reactor is smaller compared with the classical packed bed CLC system, which makes the oxygen carrier selection easier.
A proof of principle of this configuration is demonstrated with a combination of copper and nickel based oxygen carriers. The results are described well by the developed packed bed reactor model, which is also used for the description of the reactor behavior at power plant scale.
An important aspect to be studied when developing new reactor concepts is the overall efficiency of the plant employing the new reactor. To this end, power plant integration studies have been carried out in chapter 6 and 7 to investigate the effect of the operating conditions during CLC on the overall process efficiency. The combination of the IGCC power plant and CLC (IG-CLC) results in a LHV process efficiency of about 41%, which is much higher than with CO2 capture by amine scrubbing (35%).
Also with TS-CLC a high LHV efficiency can be obtained of 40.8%. Thus, with the TS-CLC still a high efficiency can be achieved, while it gives much more flexibility in the selection of the oxygen carrier and the reactor design. Therefore, TS-CLC can be considered as a good alternative to the one stage CLC in packed bed reactors.
Moreover, the packed bed CLC process has been compared with the more conventional circulating fluidized bed configuration in chapter 7. It has been shown that the process efficiency for both reactor configurations is in the same order of magnitude. Therefore, it can be concluded that the reactor selection will not only be based on process efficiency, but also on the availability, operability and cost of high temperature and high pressure reactor systems, that are still under development and both concepts present pros and cons. Furthermore, it has been shown that an additional efficiency gain of 0.5-1% point can be obtained by adapting the desulfurization method in the power plant.
Concluding, it has been demonstrated that the CLC process can be performed at high pressures without significant problems. The mathematical particle and reactor models have been validated at different operating conditions and can be used for scale up of the system. The overall LHV process efficiency of an IGCC power plant with embedded CLC is calculated at 41%, significantly higher than for power production with CO2 capture with other technologies. A new reactor configuration has been developed (TS-CLC) that can make use of the currently available oxygen carriers and still reach almost the maximum energy efficiency. Therefore (TS-)CLC can be regarded as energy efficient alternative to integrate CO2 capture in power plants.
Samenvatting Packed bed chemical-looping combustion, experimental demonstration and energy analysis
Door de invloed van CO2 uitstoot op klimaatverandering, wordt veel aandacht besteed aan onderzoek naar manieren om de CO2 uitstoot door menselijk toedoen te verminderen. Een van de mogelijkheden is om CO2 af te vangen van puntbronnen met een hoge CO2 emissie, zoals elektriciteitscentrales, en op te slaan. Dit zou kunnen bijdragen aan een CO2 uitstootreductie met 20%. Een groot nadeel van deze methode is dat het rendement voor het opwekken van elektriciteit flink afneemt. Voor een groot deel wordt deze afname veroorzaakt door de CO2 afvangstap. Om deze reden wordt veel onderzoek gedaan om de energieconsumptie voor de CO2 afvang te reduceren.
Veel verschillende technieken zijn recentelijk geïntroduceerd om de CO2 afvangkosten te reduceren, waaronder chemical-looping combustion (CLC). Dit is een proces, waarbij CO2 afvang geïntegreerd is in de elektriciteitsproductie. In CLC wordt de brandstof indirect verbrand via een reactie met een zuurstofdrager (metaaloxide). Doordat de verbranding niet direct met lucht plaatsvindt, kan er een zeer geconcentreerde CO2 stroom worden geproduceerd (na stoom condensatie) zonder dat er extra (energie-intensieve) gasscheidingsstappen nodig zijn. De metaaloxide ontstaat tijdens de voorafgaande oxidatiestap, waarbij het metaal geoxideerd wordt met lucht. Bij de exotherme oxidatie wordt een gasstroom geproduceerd op hoge temperatuur, dat wordt omgezet in elektriciteit. De elektriciteit kan met een hoger rendement opgewekt worden dan bij andere CO2 afvangtechnieken, als de gasturbine gevoed wordt met gas op hoge temperatuur (> 1200 °C) en hoge druk (ca. 20 bar).
In dit proefschrift zijn gepakt bed reactoren geselecteerd voor CLC om de combinatie van hoge druk en hoge temperatuur beter aan te kunnen, omdat er bij deze reactor geen scheiding hoeft plaats te vinden tussen het gas en de zuurstofdragerdeeltjes. Deze scheiding is erg uitdagend bij de vereiste druk en temperatuur. Als brandstof voor het CLC proces wordt syngas gebruikt, dat wordt geproduceerd door kolen te vergassen.
Van groot belang voor het uitvoeren van CLC in gepakt bed reactoren bij hoge druk is het in kaart brengen van de effect van de operatiedruk op de reactiviteit van de zuurstofdragers. Dit wordt behandeld in hoofdstuk 2. Het drukeffect op de reactiviteit van CuO/Al2O3 en NiO/CaAl2O4 (frequent onderzochte zuurstofdagers in de literatuur) is experimenteel onderzocht met een hogedrukbalans, waarbij de druk gevarieerd is van atmosferisch tot 20 bar. De experimenten demonstreerden dat de druk een negatieve invloed heeft op de reactiviteit van beide zuurstofdragers. Het blijkt dat de reactant gas fractie van belang is voor de kinetiek. Bij constante gasfractie neemt de kinetiek in beperkte mate af met de druk.
Dit wordt mogelijkerwijs veroorzaakt door een afname in het aantal zuurstofvacatures bij hogere druk, waardoor het zuurstoftransport vertraagd wordt. De experimentele data kunnen goed worden beschreven als deze effecten zijn geïntegreerd in de kinetiekbeschrijving.
Gepakte bed reactoren bevatten grotere zuurstofdragerdeeltjes dan de (voor CLC meer conventionele) wervelbedreactoren om de drukval in de reactor te beperken. Een deeltjesmodel is ontwikkeld om de reacties en het massatransport in deze deeltjes te beschrijven. De modelsimulaties zijn gevalideerd met experimenten met 1,7 mm NiO/CaAl2O4 deeltjes. De invloed van massatransportlimiteringen blijkt beperkt te zijn bij hoge druk, doordat ook de reactiesnelheid afneemt met de druk, terwijl het transport door Knudsen diffusie toeneemt.
Hoofdstuk 3 en 4 beschrijven experimentele demonstraties van het CLC concept in een gepakte bed reactor op laboratoriumschaal bij verschillende drukken met CuO/Al2O3, NiO/CaAl2O4 en ilmeniet als zuurstofdragers en H2, CO of syngas als brandstof. Bij experimenten met constante gas fractie en variërende druk (van 2 tot 7,5 bar) is aangetoond dat de druk een beperkte invloed heeft op de resultaten. De experimentele resultaten kunnen goed worden beschreven met het ontwikkelde 1D reactor model. Daarnaast toont dit model aan dat de gewenste hoge temperatuur bereikt kan worden in opgeschaalde reactoren. Het toevoegen van stoom is aangetoond als effectief middel om koolstofvorming in de reactor te voorkomen (dat het CO2 afvangrendement kan reduceren). In dat geval moet wel de strategie voor de warmtehuishouding worden aangepast aan de geproduceerde warmte door de water gas shift reactie.
Gedurende de oxidatie in het CLC proces is een temperatuurstijging van 450 °C naar 1200 °C gewenst. Dit bemoeilijkt de selectie van het type zuurstofdrager. De zuurstofdragerdeeltjes moeten thermisch en chemisch stabiel zijn bij 1200 °C en ook een hoge reactiviteit hebben over het gehele temperatuurbereik. Om de eisen voor de zuurstofdrager te verzachten, is er een nieuwe reactorconfiguratie ontwikkeld in hoofdstuk 5, de Two-Stage CLC (TS-CLC). In deze configuratie wordt de gewenste temperatuurstijging verdeeld over twee reactorbedden. In het eerste reactor bed wordt een temperatuurstijging van ongeveer 400 °C bereikt met een materiaal dat reactief is bij relatief lage temperaturen (bijvoorbeeld bij een zuurstofdrager gebaseerd op koper). Hete lucht geproduceerd in deze reactor wordt geblazen naar een tweede reactor, waarin de gewenste temperatuur van 1200 °C gehaald wordt. Deze tweede reactor bevat een materiaal dat stabiel en erg selectief is bij hoge temperatuur. Doordat de temperatuursverandering per reactor kleiner is, komen er meer verschillende zuurstofdragers in aanmerking voor dit proces. Deze configuratie is experimenteel aangetoond met een combinatie van een koper en een nikkel zuurstofdrager.
De resultaten zijn goed beschreven met het reactor model, dat na deze validatie is gebruikt om reactoren op elektriciteitscentraleschaal te beschrijven.
Om te bepalen hoe de nieuwe configuratie zich verhoudt met andere CLC configuraties en om de optimale procescondities te bepalen, zijn energie-evaluatiestudies uitgevoerd over een elektriciteitscentrale, wat beschreven wordt in hoofdstuk 6 en 7. Hieruit is geconcludeerd dat bij het rendement, waarmee elektriciteit kan worden opgewekt, voor CLC rond de 41% ligt en hiermee is het 6%punt hoger dan bij conventionele technieken, zoals absorptie met Selexol. Het rendement van TS-CLC is iets lager (40.8%) dan voor CLC in een stap (41.1%), maar hier komt een grotere flexibiliteit in de zuurstofdragerselectie en het reactorontwerp voor terug. Hierdoor kan TS-CLC als een goed alternatief beschouwd worden voor CLC in gepakte bed reactoren.
Daarnaast is het proces in gepakte bedden vergeleken met de meer conventionele wervelbedreactoren (op hoge druk). Het procesrendement blijkt voor beide reactoren rond hetzelfde niveau te liggen. Om deze reden zal de reactorkeuze niet gemaakt worden op basis van procesrendement, maar op beschikbaarheid, operatie en de kosten van hoge temperatuur en druk systemen. Beide reactorconcepten hebben hierin hun voor- en nadelen. Daarnaast kan nog een rendementstoename van 0.5-1% behaald worden door de ontzwavelingsmethode aan te passen.
Samengevat, dit proefschrift toont aan dat het CLC proces in gepakt bed reactoren op hoge druk kan worden uitgevoerd zonder significante problemen. Mathematische modellen voor de reacties in de zuurstofdrager en de reactor zijn gevalideerd voor verschillende procescondities, waardoor het gerechtvaardigd is deze te gebruiken om het proces op elektriciteitscentraleschaal te beschrijven. Het rendement voor een elektriciteitscentrale geïntegreerd met CLC ligt rond de 41%, wat een stuk hoger is dan voor een centrale met conventionele CO2 scheidingstechnieken. Daarnaast is er een nieuwe reactorconfiguratie ontwikkeld (TS-CLC) dat kan worden gebruikt met de huidig beschikbare zuurstofdrager-materialen. Hierdoor kan CLC in gepakte bedden worden gezien als een energie-efficiënte methode om CO2 afvangst te integreren in elektriciteitscentrales.
Contents
1. Introduction .......................................................................................................................... 2
2. Pressure influence on oxygen carrier reactivity ............................................................. 10
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 .................... 28
4. Pressure effect on performance of CLC in packed bed reactors ............................... 58
5. Two-stage-CLC, a novel configuration for packed bed CLC ..................................... 84
6. Energy evaluation of TS-CLC configurations for the IGCC power plant .............126
7. Comparison of CLC in packed bed and fluidized bed reactors ...............................152
8. Epilogue.............................................................................................................................194
Bibliography ..............................................................................................................................199
Nomenclature ...........................................................................................................................209
Curriculum vitae .......................................................................................................................213
List of publications...................................................................................................................214
Acknowledgement ....................................................................................................................216
|1
1 Introduction
2 |
1. Introduction Since the industrial revolution the world energy demand has continuously been increasing. The required power and heat has mainly been (and is still mainly) generated by combustion of fossil fuels that resulted in large anthropogenic CO2 emissions to the atmosphere. From 1900, the CO2 concentration in the atmosphere has risen from 300 to 390 ppm, and this increase has indeed been associated to fossil fuel combustion, cement production and deforestation (IPCC, 2013). Only 43% of the CO2 released during these activities remains in the atmosphere. The other 57% of the CO2 has been reabsorbed in the terrestrial ecosystems and the oceans (IPCC, 2013). Hence, the acidity of the oceans has increased (pH decrease by 0.1 which corresponds to a 26% increase in hydrogen ion concentration) with obvious negative influence on the environment in the oceans (IPCC, 2013).
The temperature on Earth is dependent on the radiative forces in the atmosphere that influence the solar energy uptake by the earth. Without these forces, the average temperature would be -20 °C instead of 15 °C (Meijer, 2012). Greenhouse gases, like CO2, CH4, O3 and halocarbons, when in the right concentration, have a positive influence on the radiative force and thus on the global temperature. However, with the increase of greenhouse gas concentration, and especially CO2, the balance on the radiative forces is altered and the temperature is increased. During the last century, the temperature has increased by 0.8 °C (mainly in the period 1970-2000) (IPCC, 2013), which also resulted in a sea level increase of 15 cm. Due to this, the climate has changed, which is expressed by an increase of the heat wave frequency, the increase of the frequency and intensity of heavy precipitation events, increased incidence and/or magnitude of extreme sea levels and possibly an increase of the frequency and intensity of drought events and intense cyclone activity (IPCC, 2013).
The expectations for the 21st century are that the anthropometric CO2 emissions will continue and rise to 421-936 ppm (IPCC, 2013). This results in a further global temperature increase of 1-3.7 °C, which means that the impact of human activities on the climate change will increase as well. The ice levels on the artic sea and the Northern Hemisphere will reduce and the sea level is expected to increase by 0.4-0.63 meter (IPCC, 2013). It is also very likely that the ocean circulation will decrease by 11-34%, which affects the regional climates (IPCC, 2013).
The long term solution to this problem (and also for energy security) requires that the fossil fuels will be replaced by sustainable (renewable) resources. On the midterm, however, fossil fuels will still play an important role in the energy conversion processes, thus the effect on the climate can only be mitigated by decreasing the CO2 emissions
1. Introduction |3
1
related to fossil fuel combustion. This can be achieved by applying carbon capture and storage (CCS) to power production and energy intensive processes (IPCC, 2014). In that case, the CO2 is captured from large emission sources (like power or steel plants), compressed and transported to the storage location (or used as building block for other products). The storage location could be ocean or geological storage e.g. a former oil or gas field (IPCC, 2005), where it should be retained for centuries. Currently, 13.5 gigaton CO2/year are emitted from point sources, while 200-2000 gigaton CO2 storage capacity should be available in oil and gas fields or deeps saline formations (IPCC, 2005), which would be sufficient for at least 20 years.
Worldwide, 12 gigaton CO2 is annually emitted for power and heat production, which is 25% of the total emissions (IPCC, 2014). To achieve CO2 capture in power plants, the CO2 separation could be carried out after the combustion step (which is mainly a CO2/N2 separation), before the combustion step (H2/CO2 separation) or the combustion could be carried out with pure oxygen (N2/O2 separation). All these possibilities include an energy intensive gas separation step that will decrease the efficiency of the plant. For instance, with the current available capture technologies, the power production efficiency of a coal-based power plant would drop from 42% to 31-33% (Finkenrath, 2011). For the exploitation of the CCS technology, much research is thus dedicated to reduce this efficiency penalty by improving (or circumventing) the separation steps.
In this respect, the Chemical-looping combustion (CLC) process is a very interesting new technology, in which the separation of CO2 is integrated with the power production process by avoiding direct contact between the fuel and air (and therefore dilution of the CO2 with N2). The fuel is combusted with an oxygen carrier (metal oxide) producing a CO2/H2O mixture while the oxygen carrier (MeO) is reduced as shown in Figure 1.1. The oxygen carrier is subsequently oxidized to its original state with air. The hot stream produced during this very exothermic oxidation reaction is fed to a gas turbine and converted into electricity. The reactions involved with the CLC process are shown in equation 1.1 in case of syngas as fuel. As this equation demonstrates, the same net combustion reaction is performed, while a highly concentrated CO2 stream is obtained (after H2O removal). Because the CO2 capture is integrated in the combustion step, a higher electrical efficiency of the complete plant is achieved compared to the state-of-the-art capture technologies (such as Selexol). Additionally, another advantage of the CLC process is that the NOx-emissions are reduced due to the lower maximum temperature.
4 |
Figure 1.1: The principle of chemical-looping combustion (Gallucci and Van Sint Annaland, 2011).
CLC has been introduced in the 1950s as technology to produce pure CO2 with a lower heat consumption than conventional technologies (Lewis and Gilliland, 1954). In the 1980s, this technology was reintroduced as a possibility to increase the efficiency of the combustion process (Ishida et al., 1987; Richter and Knoche, 1983). From 1994 CLC gained more interest as CO2 capture technology (Ishida and Jin, 1994).
2 2
2
2
2 2 2 2
H MeO H O Me CO MeO CO Me
O Me MeO +2
H CO O H O CO 2
n n n nm m m mn m n m n m
n mn m n m
1.1
A reactor design based on circulating fluidized bed reactors has been proposed in 2001 (Lyngfelt et al., 2001) which recalls the original idea of the 1950s application. During the subsequent decade, the research had mainly been focused on the development of oxygen carriers (Adánez et al., 2004; Cho et al., 2004; Hossain and de Lasa, 2008) and the demonstration of operation with this reactor configuration (Lyngfelt, 2011a). Until now, it has been successfully demonstrated on a maximum scale of 1 MWth (Ströhle et al., 2014) and there are plans for a demonstration at 10 MWth (Marx et al., 2013). The technology can also be used for the production of H2 as chemical, also referred to as chemical-looping reforming (Ryden and Lyngfelt, 2006; Ryden et al., 2006), or as feedstock for power production (Adanez et al., 2012).
Solid, liquid and gaseous fuels are suitable for CLC (Adanez et al., 2012; Lyngfelt, 2013; Moldenhauer et al., 2014). In this work, coal is selected as primary feedstock, because it has the highest carbon content and therefore the highest anthropogenic CO2 emissions reduction can be achieved by applying CCS to coal fired power plants. Although coal
1. Introduction |5
1
could be directly fed to the fuel reactor (Adanez et al., 2012; Lyngfelt, 2013; Mattisson et al., 2009) and react with the oxygen carrier, this is a quite slow reaction which results in very large reactors even in case the oxygen carrier (such as copper, manganese and cobalt based oxygen carriers) can release O2 in the gas phase in the so-called chemical-looping with oxygen uncoupling (Mattisson et al., 2009). In case coal is fed to the fuel reactor, an additional carbon stripper has to be introduced to the setup to separate the ashes from the reduced oxygen carriers.
A more interesting and realistic alternative foresees that the coal is first converted in a gasifier to syngas, which is afterwards fed to the CLC system. This thesis will focus on this second option. So, an integrated gasification chemical looping (IGCLC) process is considered, where the syngas production and post-treatment are performed similarly to conventional integrated gasification combined cycle (IGCC) plants.
One of the product streams from CLC is a hot gas stream that is converted into electricity in a gas turbine. A pre-requisite for the CLC to be exploited industrially is that the energy efficiency is higher when compared to other capture technologies. The highest energy efficiency is achieved when the CLC system is operated at high pressures and temperatures.
Figure 1.2: The pressure and the temperature effect on the process efficiency for natural gas fueled power plants (Naqvi et al., 2007).
6 |
Figure 1.3: The pressure effect on the process efficiency for coal fueled power plants (Spallina et al., 2014).
Figure 1.4: Schematic overview of the circulating fluidized bed system (a) and the dynamically operated packed bed reactors (b).
39.40 40.15 40.54 40.80 40.61
30313233343536373839404142
-50
0
50
100
150
200
250
300
350
11 14 17 20 23
net e
lect
ric e
ffici
ency
, %
Pow
er, M
We
CLC pressure, bar
Gas Turbine Steam Cycle Others CO2 compression
N2
compr
Airblower
Heat
rem
oval
Oxi
datio
n
Redu
ctio
nCO2/H2O
Syngas
Purg
e
Purg
e
Air
Airreactor
Fuelreactor
CO2/H2O
Syngas
Hot O2
depleted air
MeO
Me
4 51 2 3
b)a)
Air
Hot O2
depleted air
N2
1. Introduction |7
1
The effect of the operating pressure and temperature of the hot air (turbine inlet temperature, TIT) is demonstrated in Figure 1.2 for natural gas fired power plants (Naqvi et al., 2007). For coal-fired power plants, a similar trend is observed, but here lower efficiencies are reached as demonstrated in Figure 1.3, which is usual for plants fueled by coal (Spallina et al., 2014). It is demonstrated that the highest process efficiencies are obtained in case the CLC reactors are producing a hot gas stream at elevated pressures and temperatures (in the order of 20 bar and 1200 °C). At these conditions, CLC in combined cycle configuration has conceptually proven to outperform the conventional CO2 capture technologies, with the calculated efficiency of 3-5% points higher than with conventional CO2 separation technologies (Erlach et al., 2011; Spallina et al., 2014).
A circulating fluidized bed system could be used for CLC as proposed by Lyngfelt et al. (Lyngfelt et al., 2001), which consists of a fuel and an air reactor. The syngas is fed to the fuel reactor, where it reacts with the MeO to from CO2 and H2O. In the air reactor, the Me is re-oxidized to MeO and the solids are recirculated between both reactors, as illustrated in Figure 1.4a. At the reactor exits, the gas and the solids are separated by cyclones. However, at the desired operating conditions (high temperatures and pressures), the gas solid separation (essential for the gas to be fed to the turbine downstream of the CLC) is very challenging also because of the fines that are continuously produced due to the solids circulation between the reactors. As alternative, a dynamically operated packed bed configuration has been developed by Noorman et al. (2007, 2010b). The oxygen carrier particles remain stationary in the packed bed reactor, while the reactant gas feed is switched, as shown in Figure 1.4b. In principle, a continuous operation is achieved with multiple reactors working in parallel. The most important advantage is that the difficult gas-solid separation is circumvented, and therefore these reactors can be easily operated at the required conditions. Additionally the scale-up of packed bed reactors is much easier compared to circulated fluidized bed systems. Two other reactor configurations have recently been introduced, based on non-interconnected dynamically operated fluidized beds (Zaabout et al., 2013) and on a rotating bed reactors (Håkonsen and Blom, 2011; Håkonsen et al., 2014), but these will not be discussed in this work.
A packed bed configuration for CLC consists of several dynamically operated reactors working in cycles. A process cycle consists of the following operating steps: oxidation, heat removal, reduction and purge. Starting with a reduced oxygen carrier (OC) and reactor mainly at 450-600 °C, the oxygen carrier is oxidized with air and heat is produced inside the reactor (step 1). When the complete bed has been oxidized, a hot air steam is produced at 1200 °C (step 2) by feeding air at 450 °C (gas temperature after
8 |
adiabatic compression from ambient conditions to 20 bar). After the heat is blown out of the reactor (and used for power production), the reactor is purged with N2 to circumvent contact between O2 and the fuel. Subsequently, the fuel (syngas produced in coal/biomass gasification) is fed to the reactor, the OC is reduced again and a CO2/H2O stream is produced, which is cooled down condensed and the CO2 is compressed and sequestrated. After the reduction, the bed needs to be purged again. To have a continuous process, at least one reactor has to operate in each operating step.
Several oxygen carriers (OCs) could be selected for this process. Usually nickel, copper, iron or manganese based materials are used as OCs (Adánez et al., 2004). The oxygen carrier should be highly reactive, withstand a temperature of 1200 °C, facilitate a temperature rise of 750 °C (from 450 °C to 1200 °C) during oxidation and full fuel conversion during reduction, have low costs, be environmentally friendly and be available.
This study is aimed to get an overall understanding of the chemical looping combustion process in packed bed reactors fueled by syngas obtained from coal gasification. Operation at elevated pressures is crucial to reach a high process efficiency. In chapter 2, the pressure effect on the oxygen carrier reactivity is investigated. After this, the focus is put on the experimental demonstration of CLC in packed bed reactors and the pressure effect on the reactor performance (chapter 3 and 4). Subsequently, a novel reactor configuration is proposed as a solution to find suitable oxygen carriers (chapter 5). Chapter 6 and 7 are focused on the effect of the reactor configuration on the process efficiency. The electrical efficiency of the proposed novel reactor configuration is calculated and compared with alternative configurations, and in particular the circulating fluidized bed system. Afterwards, the effect of the desulfurization method on the process efficiency is discussed. Eventually in chapter 8, conclusions are drawn and it is discussed how the CCS research should proceed.
|9
2 Pressure influence on
oxygen carrier reactivity
This chapter is based on the following paper: Hamers, H.P., Gallucci, F., Williams, G., Cobden, P., Van Sint Annaland, M., 2014. Reactivity of oxygen carriers for CLC in packed bed reactors under pressurized conditions. Submitted to Energy & Fuels.
10 |
2. Pressure influence on oxygen carrier reactivity
Abstract For the design, scale-up and optimization of pressurized packed bed reactors for chemical-looping combustion, understanding of the effect of the pressure on the reactivity of the oxygen carriers is very important. In this chapter, the redox reactivity of CuO/Al2O3 and NiO/CaAl2O4 particles at elevated pressures have been measured in a pressurized high-temperature magnetic suspension balance. The experiments have demonstrated that the pressure has a negative influence on the reactivity and that this effect is kinetically controlled. The negative effect of the pressure might be caused by the decrease in the number of oxygen vacancies at higher pressures. Moreover, the reactant gas fraction has been demonstrated as an important parameter, probably related to competition between different species for adsorption on the oxygen carrier surface. These effects have been included in the kinetic model leading to a good description of the experimental results. The impact of these findings on packed bed CLC applications with larger oxygen carrier particles has been investigated with a particle model that considers diffusion limitations and kinetics. It has been shown that the impact of diffusion limitations decrease with increasing pressure, due to the decrease in reaction rates and the increase in diffusion fluxes caused by Knudsen diffusion. The results have been validated by experiments with 1.7 mm NiO/CaAl2O4 particles. These results corroborate that the selection of larger particles because of pressure drop considerations, does not lead to a large decrease in effective reaction rates, which is beneficial for packed bed CLC applications.
2. Pressure influence on oxygen carrier reactivity |11
2
2.1. Introduction Most power plant integration studies with CLC assume that the process is operated at elevated pressure, but at this moment little information is known about the reactivity of the oxygen carriers at elevated pressure. García-Labiano et al. have published the kinetics at elevated pressure for copper, nickel and iron based oxygen carriers (García-Labiano et al., 2005). The kinetics have been determined at 450-950 °C and a pressure between 1 and 30 bar. It appears that a high total pressure has a negative effect on the kinetics. They observed the same behavior with different particle sizes (0.09-0.25 mm), which indicates that they were measuring kinetics (García-Labiano et al., 2005). The authors have indicated that the decline in reactivity cannot be caused by a decrease in surface area and pore volume. For other gas-solid reactions similar results have been obtained, for example for the reactions between CaO and H2S or calcination of CaCO3. Chauk et al. have found a decrease in the reactivity of CaO with H2S at higher pressures, but have attributed this to a lower pore volume and surface area (Chauk et al., 2000). Similar behavior was found for the calcination reaction where the reactivity was described by increasing the negative pressure effect in Fuller’s equation for molecular diffusion (García-Labiano et al., 2002). From these studies, it can be concluded that the pressure could have a negative effect on the reactivity, but there is no general consensus on the cause for the observed negative effect.
Some experimental work has been carried out to the application of CLC in pressurized circulating fluidized bed reactors. Such a system has been constructed by Xiao et al. (Xiao et al., 2012). Experiments have been carried out with Shenhua bituminous coal as fuel and iron ore as oxygen carrier at three different operating pressures (1, 3 and 5 bar). Stable operation has been reached. A higher combustion efficiency was observed at elevated pressure. However, the oxygen carrier fine production was also higher which is very detrimental for power production in downstream turbines.
To better accommodate the CLC process at elevated pressure, packed bed reactors have been selected in this study. In packed bed reactors, larger oxygen carrier particles are required to maintain a low pressure drop. But the selection of larger particles may imply also that the role of diffusion limitations inside the particle may become more dominant. The effect of diffusion limitations has been described by Noorman et al. for atmospheric applications (Noorman et al., 2011a). This model considers molecular diffusion and Knudsen diffusion. From these studies, it was concluded that Knudsen diffusion was the rate determining step in the oxygen carrier particles considered.
The objective for this chapter is to measure the pressure effect on the kinetics and to evaluate this impact for packed bed CLC applications. For the oxygen carrier reaction
12 |
rate measurements, CuO/Al2O3 and NiO/CaAl2O4 particles have been tested in a high-pressure magnetic suspension balance. An kinetic expression is determined from these measurements and this correlation is used to describe the behavior in packed bed reactors using an advanced particle model. The model is validated by experiments with larger particles.
2.2. Materials and methods
2.2.1. Oxygen carriers The CuO/Al2O3 particles were obtained from Sigma-Aldrich with an active weight content of 13wt% and a particle size of 1.1 mm. For kinetic experiments, the particles were crushed and sieved to a size of 110-150 μm.
The NiO/CaAl2O4–particles used in this work is a Johnson Matthey product, HiFUEL® R110 (Ni based catalyst supported on CaAl2O4 for steam reforming of natural gas), available in pelleted form from Alfa Aesar. The particles were received in the form of shaped pellets, comprising of 4 hole, 4 flute domed cylinders. For this study, the pellets were crushed and sieved to particle sizes of 1.7 mm and 0.15 mm. Separate TGA experiments proved that the mass change (and thus the active weight content) is not the same for each particle. The experiments with CO and O2 were carried out with a particle with an active weight content of 18.5wt%, while an active weight content was measured of 17wt% for the experiments with H2. Before the tests, the oxygen carrier was activated by exposing it to two redox cycles containing reductions with H2 and oxidations with air at 900 °C.
2.2.2. High pressure magnetic suspension balance The experiments have been carried out in a magnetic suspension balance (Rubotherm) that can operate between 200-1200 °C and 1-30 bar. An oxygen carrier sample of 100 mg is placed in a porous quartz glass sample holder. The basket is placed on an Ir wire that is hanging on a permanent magnet. The mass is determined by the strength of the magnet.
The reactant gases are supplied at the top of the reactor. The reactor is surrounded by a vessel that is maintained at lower temperature. Argon is supplied to this vessel to prevent that reactant gases can enter and mix in the insulation layer. A schematic overview of the setup is provided in Figure 2.1.
2. Pressure influence on oxygen carrier reactivity |13
2
Figure 2.1: Schematic overview of the magnetic suspension balance setup.
Before a series of experiments is started, the system is pressurized and the reactor is set at the desired operating temperature. When this temperature is reached and the system has been stabilized, redox cycles are carried out that consist of a 20 min reduction, a 10 min purge, a 10 min oxidation and a 10 min purge. During experiments, a total flow rate of 480 mLn/min. is fed. Some experiments have been carried out with a lower flow rate (320 mLn/min.) to demonstrate that the measured reactivity is not influenced by external mass transfer limitations. Every experiment has been repeated at least two times to assure the reproducibility of the results.
Blank experiments have been carried out with only a sample holder (and no oxygen carrier) so that the influence of the flow change on the measured mass change can be determined and corrected for. The blank experiment data are subtracted from the data obtained with the oxygen carrier sample.
2.2.3. Particle model The conversion of the solid is described by a numerical particle model that assumes a spherical oxygen carrier particle with a uniform porosity, fixed particle diameter and a uniform pore size (Noorman et al., 2011a). The model describes the gas transport
Pre
ssur
eve
ssel
N2AirH2COCO2H2O
Ar
PC
vent
sample
magnet
14 |
inside the particle from the moment that the particle is exposed to a reactant at certain operating conditions. For the gas transport, the reaction kinetics and molecular and Knudsen diffusion and external mass transfer limitations are taken into account. Based on this description, the solid conversion is simulated as a function of time and these data is compared with TGA results. The equations applied in the model are listed in Table 2.4.
The kinetics for CuO are taken from García-Labiano et al. (2004, 2005), while the pre-exponential factor was adapted from the shrinking core to the homogenous model (parameters are listed in Table 2.1. The kinetics for NiO have been determined in separate TGA experiments and are shown in Table 2.2 (Medrano et al., 2014). The particle properties are based on pycnometer (Quantachrome Micro-ultrapyc1200) and BET (Thermscientific Surfer) measurements and are listed in Table 2.3.
Table 2.1: Kinetic parameters for the reactions with CuO/Al2O3. Parameters are based on Garcia-Labiano (2004).
H2 CO O2 k0, s-1 barq 12,000 708 1200 EA, kJ mol-1 33 14 15 n 0.6 0.8 1
Table 2.2: Kinetic parameters for the NiO/CaAl2O4 particles based on TGA experiments.
H2 CO O2 Cs, mol m-3 89960 89960 151200 r0, m 3.13·10-8 3.13·10-8 5.8·10-7 k0, mol1-n m3n-2 s-1 barq 9.00·10-4 3.5·10-3 1.2·10-3 Ea, kJ mol-1 30 45 7 n 0.6 0.65 0.9 Ds,0, mol1-n m3n-1 s-1 1.70·10-3 7.4·106 1 EDs, kJ mol-1 150 300 0 kx 5 15 0 b 1 1 2
Table 2.3: Particle properties of NiO/CaAl2O4 for particle model applications.
Oxygen carrier 17-18.5wt% NiO on CaAl2O4 TGA experiments Particle diameter, mm 1.7 sieved Particle porosity, m3gas/m3particle 0.55 derived from combination
dry and liquid pycnometer Average pore size, Å 130 BET porosimetry Tortuosity 2
2. Pressure influence on oxygen carrier reactivity |15
2
Table 2.4: The equations of the particle model (Noorman et al., 2011a).
Continuity equation 2
21
1 gNtotg g
i ii
r nr M
t rr
Gas phase components 2
,2
1 ig g g ii i
r nr M
t rr
where 1
,, , ,
1
gNg k
i i g i tot g eff ik g i totk
n j n D nr
Solid phase components ,
1
gNs s s j
i ji
r Mt
where , , , , ,1 1ox ox s MeOs MeO s act s Me s Me s act
Me
v MM
Energy balance g,reactants
2, , ,2
1
1 N
g g p g s s p s eff i R ii
T TC C r r Ht r rr
Kinetics
- copper: , 0 5exp10
qn totA
i g p gpEr k c
R T
- nickel: 0
,s p s acti
j
dXrb M dt
02 1
0 03 3
3
1 (1 ) (1 )
ng
s
s s
cb rdX
dtc
r rX Xk D D
0 5exp10
q
totA pT
kR
k E ,0 exp expsD
s s x
ED D k X
R T
Diffusion - Binary molecular diffusion: Knudsen diffusion:
1.75 1 1
, 23 3
0.01013 i kBin ik
i k
T M MD
p v v ,
83pore
Kn ii
d R TDM
- Maxwell-Stefan diffusion matrix 1D B that consists of the elements dik and n gaseous components
with: 1, , ,
( )
1ni k
iikBin in Bin ik Kn ii k
y yB
D D D and
, ,
1 1ik i
Bin ik Bin in
B yD D
- Effective diffusivity ,g peffD D
16 |
2.3. Results and discussion
2.3.1. Pressure effect on kinetics Experiments were carried out varying the total pressure (1-20 bar), while the partial pressure of the reactant was kept constant at 1 bar. In this way, the reactant gas concentration and the temperature were fixed, so that solely the influence of the total pressure is measured. Redox cycles have been measured with CuO/Al2O3 and NiO/CaAl2O4 as oxygen carriers at 600 and 800 °C. In case of CO as reactant, a CO2/CO-ratio of 3 and 1 was used at 600 and 800 °C respectively.
The results plotted as solid conversion (defined in equation 2.1) as a function of time for the experiments at constant temperature and different total pressures are shown in Figure 2.2 and Figure 2.3. During the reduction cycles, full solid conversion was not reached. The maximum solid conversion depends on the reduction temperature and the type of the support material (Adánez et al., 2004; Baek et al., 2011). The support material could be present as inert layer in the solid structure, which might influence the accessibility of the oxygen and thus the degree of reduction. The maximum degree of reduction depends on the operating temperature. The lowest conversion is reached for the H2 experiments with CuO at 600 °C. For demonstration of the pressure effect, the curve has been zoomed in on the first 60 seconds. But after that moment, the particle keeps on reacting and a conversion of 80% is reached after longer times (which is discussed in more detail in chapter 3). From the experiments it can be concluded that the maximum solid conversion does not depend on the operating pressure, but only on the temperature.
observed mass changesolid conversionmaximum mass change for the assumed active weight content
2.1
As can be observed from Figure 2.2 and Figure 2.3, where the partial pressure of the reactant was fixed, but the total pressure was varied, the reaction rate decreases with increasing total pressure. It has to be noted that in the experiments with a higher total pressure, the reactant was thus more diluted, because the partial pressure of the reactant was kept constant. The decreased reactivity with the pressure is observed for all reactants with both oxygen carriers. At higher pressures, more fluctuations in the experimental results can be seen which is related to limitations of the experimental set-up. These fluctuations could in principle be decreased by reducing the total flow rate, but in that case external mass transfer limitations could occur. Despite these fluctuations, a clear trend can still be discerned from the results at 20 bar.
2. Pressure influence on oxygen carrier reactivity |17
2
0 50 100 150 2000.0
0.2
0.4
0.6
0.8
1.0a) H2
ptot=
pH2=1 barT= 800 C
5bar 10bar 20bar
Sol
id c
onve
rsio
n (-
)
Time (s)
model exp
0 50 100 150 2000.0
0.2
0.4
0.6
0.8
1.0b) CO
2bar 5bar 10bar 20bar
Sol
id c
onve
rsio
n (-
)
Time (s)
pCO=1 barT=800 C
ptot=model exp
0 50 100 150 2000.0
0.2
0.4
0.6
0.8
1.0c) O2
pO2=1 barT= 800 C
5bar 10bar 20bar
Sol
id c
onve
rsio
n (-
)
Time (s)
ptot=model exp
Figure 2.2: The effect of pressure on the redox kinetics of NiO/CaAl2O4 with H2, CO and air at 800 °C, while the partial pressure during all the experiment was kept constant at 1 bar. The markers show the experimental data and the lines the model predictions.
External mass transfer limitations cannot be the cause for the decrease in the reaction rate, because the same conversion curves were obtained from experiments with a lower gas flow rate (320 mLn/min instead of 480 mLn/min). This experiment demonstrates that the reactant flow rate was sufficiently high to refresh the gas around the sample and to supply a sufficient amount of reactants for the gas/solid reactions. This has been validated for all the operating conditions investigated.
Moreover, experiments have been carried out with different particle sizes and again the same conversion curves have been obtained. This is illustrated in Figure 2.4, where the oxidation of Cu/Al2O3 is displayed at 20 bar (with pO2=4 bar and 1 bar) and the particle size was varied between 0.15 mm (lines) and 1.1 mm (markers). The same trends can be observed and for that reason, internal mass transfer limitations can also be ruled out as cause for the observed decrease in reactivity at elevated pressures.
18 |
0 20 40 60 80 1000.0
0.2
0.4
0.6
0.8
1.0a) H2
pH2=1 barT=600 C
5bar 10bar 20bar
Sol
id c
onve
rsio
n (-
)
Time (s)
ptot=model exp
0 20 40 60 80 1000.0
0.2
0.4
0.6
0.8
1.0b) CO
2bar 5bar 10bar 20bar
Sol
id c
onve
rsio
n (-
)
Time (s)
pCO=1 barT=800 C
ptot=model exp
0 20 40 60 80 1000.0
0.2
0.4
0.6
0.8
1.0c) O2
pO2=1 barT=600 C
5bar 10bar 20bar
Sol
id c
onve
rsio
n (-
)
Time (s)
ptot=model exp
Figure 2.3: The effect of pressure on the redox kinetics with CuO/Al2O3 with H2, CO and air, while the partial pressure during all the experiment was kept constant at 1 bar. The markers show the experimental data and the lines the model predictions.
Thus, the decrease in reactivity with increasing pressure has to be kinetically controlled. An expression for the reaction rates including a pressure correction factor has been introduced by García-Labiano et al. (2005) and the kinetic term is demonstrated in equation 2.2.
0
5
exp
10
nAq
tot
k Er cR Tp
2.2
The same method has been followed here. By fitting the experimental data, a number for the parameter q was determined. The determined numbers for q are displayed in Table 2.5 together with the data from literature for comparison. In general, the same trend is observed as by Garcia-Labiano, but somewhat different values for q have been found. There may be different reasons for the observed discrepancies; firstly, the experiments reported in literature may have been carried out with different materials or supports. Furthermore, the quality of the fitting is different especially at high pressures.
2. Pressure influence on oxygen carrier reactivity |19
2
0 50 100 150 2000.0
0.2
0.4
0.6
0.8
1.0
25% air/N2
Sol
id c
onve
rsio
n (-
)
Time (s)
Air
Figure 2.4: The particle size effect on the solid reaction rate for the oxidation of Cu/Al2O3 at 800 °C and 20 bar. The reactivity of 1.1 mm particles is represented by markers and the 0.15 mm particles by lines.
Table 2.5: Determined values for q (in Eqn. 2) from the experimental data and comparison with values found in literature.
q CuO/Al2O3 NiO/CaAl2O4 Gaseous reactant
García-Labiano et al. (2005)
This work García-Labiano et al. (2005)
This work
H2 0.53 1.0 0.47 0.75 CO 0.83 1.2 0.93 0.85 O2 0.68 1.3 0.46 1.05
2.3.2. Discussion In the previously described experiments the reactant partial pressure was kept constant, while the total pressure was varied. The partial pressure was fixed by increasing the dilution at higher pressures. During both the oxidation and the reduction reactions, the reaction rates decrease with increasing pressure, which might to a large extent have been caused by the dilution of the reactant gas.
The following reaction mechanism has been proposed in the literature for redox reactions (Atkins and de Paula, 2002; Fierro, 2006). First, the reactant adsorbs on the oxygen carrier and subsequently, an oxygen atom is transferred from the adsorbed gas to the oxygen carrier or vice versa (Atkins and de Paula, 2002). CO2 or H2O is formed
20 |
during reduction and this molecule is desorbed from the oxygen carrier. In principle, not only the reactant could adsorb to the oxygen carrier surface, but also the other gases that are considered to be inert in the reaction. In case of competitive adsorption, the reactant gas fraction is a relevant parameter for the kinetics.
Furthermore, the surface is not expected to be flat, also because the metal (for example Ni has an atomic radius of 125 pm (Shriver and Atkins, 1999)) has a different atomic radius than the oxide (atomic radius of 66 pm (Shriver and Atkins, 1999)). Due to this difference, cavities could be present on the solid surface. The gas molecule that is present in the cavities could be inert or reactive with the solid. If the space of a cavity is occupied by an inert gas, it blocks the pathway of the reactive gas. Therefore a reactive spot on the solid remains unoccupied and this results in a lower reaction rate. These diffusion limitations are not dependent on the particle size, because the gases are distributed in the particle by pores that are much bigger, so that the gas close to the solid surface still has the feed composition.
Other experiments have been carried out to exclude some effects and to prove what could be the reason for the observed behavior in the experiments. Reductions have been carried out with varying CO2/CO-ratios at different pressures and no effect on the kinetics was observed. Therefore it seems that the desorption of gaseous products is not the limiting step. Furthermore, oxidations have been carried out with air (25% air) that is diluted either by N2 or CO2 (so a 25% air/75% N2 vs. a 25% air/75% CO2 mixture). It is expected that CO2 adsorbs on the solid surface and therefore competitive adsorption with O2 is expected. However, the experiments showed that the reaction rates did not change significantly, when air was diluted with CO2. This means that either N2 adsorbs on the surface by physisorption with the same impact as CO2 or that the adsorption is not a rate limiting step. In the latter case, the blocking of the reactant gases in the cavities could explain the decrease in reactivity when the mixture is more diluted.
2. Pressure influence on oxygen carrier reactivity |21
2
0 20 40 60 80 1000.0
0.2
0.4
0.6
0.8
1.0 ptot=5bar
ptot=20barS
olid
con
vers
ion
(-)
Time (s)
ptot
xCO=0.2xCO2=0.2T=800 C
ptot=1bar
Figure 2.5: The influence of the pressure on the reduction reactivity of NiO/CaAl2O4 with CO, while the CO-fraction is fixed.
When the reactant gas fraction is fixed, while the pressure is increased, a small decrease in reactivity is observed. The temporal evolution of the solid conversion of NiO during reduction with 20% CO in 20% CO2 and 60% N2 at 1, 5 and 20 bar is illustrated in Figure 2.5. The same trend was observed for reductions with H2 and oxidations with air.
Another point to be taken into account is that, during oxidation, not all the metal is available at the surface. In fact, the metal does not form a monolayer between the support material and the pore. According to the Wagner oxidation theory, metal ions and electrons migrate to the surface of the metal grain, while oxygen ions move to the bulk. This transport is carried out by vacancies in the metal oxide structure (Kofstad, 1972). If the solid structure contains more metal than oxygen according to the stoichiometry (a so-called oxygen deficient situation), the formation of oxygen vacancies is possibly the rate limiting step. These vacancies are formed according to equation 2.3. In this equation, V0 is an oxygen vacancy, O(s) is an oxygen atom in the solid matrix and O2(g) is gaseous oxygen molecule.
O 21O s V O g2 2.3
22 |
Equation 2.3 describes that if an oxygen gas molecule reacts with the metal, two oxygen vacancies in the solid matrix are replaced by oxygen atoms. If the oxygen pressure is increased, the equilibrium shifts to the left and the number of vacancies can decrease. Because of the decrease in oxygen vacancies, the diffusion through the solid matrix decreases in case the transport via oxygen vacancies is rate determining. In the some cases with oxygen deficient materials, the dependency of the diffusion flux inside the solid matrix is with pO2-1/6 (Kofstad, 1972). In such a situation, a decrease in kinetics can be observed when the pressure is increased, while the reactant gas fraction is fixed. During reduction the oxygen has to be transported in the opposite direction.
If the reactant gas fraction and the oxygen vacancies in the oxygen carrier have influence on the kinetics, the Arrhenius approach might not be the right approach to describe these gas/solid reactions at different pressures. García-Labiano et al. proposed to include an additional term in the equation that is dependent on the total pressure with which the experiments could be described, equation 2.2. This can be rewritten using Dalton’s law to an expression that has a dependency on the reactant gas fraction, y, and the partial pressure of the reactant as reported in equation 2.4.
5 5
1~ ~
10 10
n nn q n n q q n q
tot totq q n ntot tot
c pr p p y p y pR Tp p
2.4
During the experiments with a fixed reactant partial pressure, the gas fraction and the total pressure was varied, which can be fitted by q. The obtained values for q are in general a factor 0.2-0.4 larger than the reaction order, n. This results in a negative number for n-q with an order of magnitude of p-0.2 to p-0.4. It should be noted that the similar trend is observed in the above mentioned diffusion flux inside a solid matrix. This indicates that the decrease in reactivity with increasing pressure might be related with the reduced rate in the formation of oxygen vacancies at higher pressures.
Concluding, the lumped expression given by García-Labiano et al. can capture the effect of observed phenomena, but a more detailed study preferably with in situ analysis should be carried out to elucidate the phenomena prevailing at elevated pressures in more detail. The experimentally determined pressure effect will be used in the next section to investigate its implications for pressurized packed bed CLC applications with relatively large particles.
2. Pressure influence on oxygen carrier reactivity |23
2
2.3.3. Pressure effect in large particles for packed bed applications An increase in pressure results in a decrease in the reaction rates by about a factor 3 at 20 bar relative to atmospheric pressure. In this section, the effect of reduced reaction rates and the extent of internal mass transfer limitations for the relatively large particles used in packed bed reactors will be studied in more detail.
For packed bed applications, larger particles have to be used to avoid an excessive pressure drop over the reactor, which would reduce the overall process efficiency. In case the particle size is increased, the influence of internal diffusion limitations could increase, which could result in a decrease of the effective reaction rates. Experiments have been carried out with different particle sizes to evaluate the impact on the operation of packed bed reactors. The NiO/CaAl2O4-particles were available with a larger particle size and have been used in this part of the investigation.
A particle model has been developed to describe the effective reaction rates inside oxygen carrier particles considering reaction kinetics, molecular diffusion and Knudsen diffusion through the pores. In the previous section, it has been shown that the redox kinetics have a negative pressure dependency. The molecular diffusion coefficient is inversely dependent on the pressure, whereas the Knudsen diffusion coefficient is independent on pressure. The overall diffusion coefficient is multiplied by the gas density that linearly increases with the pressure. Hence, no pressure dependency is expected in the molecular diffusion limited regime and a positive pressure dependency if Knudsen diffusion is the limiting step. At atmospheric pressure the Knudsen diffusion is by far the most important limitation for gas/solid reactions with oxygen carriers (Noorman et al., 2011a). So, the pressure might reduce the diffusion limitations in the particles and increase the effective reaction rates.
The measured solid conversion as a function of time during the reduction with CO and the oxidation of a NiO/CaAl2O4 particle with a relatively large particle size of 1.7 mm is displayed in Figure 2.6 by the markers. The operating conditions were at 800 °C and the reactant partial pressure was fixed, while the total pressure was varied. The same experiments were simulated with the particle model and the model results are shown by the lines in Figure 2.6. The experiments are described quite reasonably by the model. The effectiveness factor that is obtained from the model, increases with increasing pressure, meaning that the reaction becomes more kinetically controlled. This is caused by two reasons. First, the decrease in the redox kinetics with increasing pressure and second, the increase in the Knudsen diffusion flux with increasing pressure due to the density increase.
24 |
0 50 100 150 2000.0
0.2
0.4
0.6
0.8
1.0a)
2bar 5bar 10bar 20bar
Sol
id c
onve
rsio
n (-
)
Time (s)
pCO=1 barT=800 C
ptot=model exp
0 50 100 150 2000.0
0.2
0.4
0.6
0.8
1.0
5bar 10bar 20bar
Sol
id c
onve
rsio
n (-
)
Time (s)
pO2= 1barT=800 C
b)
ptot=model exp
Figure 2.6: Effective reaction rates of NiO/CaAl2O4 for reductions with CO (a) and oxidations (b) at 800 °C varying the operating pressure and constant reactant partial pressure (particle size=1.7mm).
0 50 100 150 2000.0
0.2
0.4
0.6
0.8
1.0
ptot=20bar, T=800 CxCO=0.5
Sol
id c
onve
rsio
n (-
)
Time (s)
0.2mm 1.7mm 5mm 10mm
dpart=
Figure 2.7: The influence of the particle size on the effective reaction rates. The reduction is carried out with a 50% CO/CO2 mixture at 800 °C and 20 bar.
2. Pressure influence on oxygen carrier reactivity |25
2
The good agreement between the model and the experiments with 1.7 mm particles indicates that the model can be used for the prediction of the behavior of particles in packed bed reactors. For the chemical-looping combustion process in packed bed reactors, an operating pressure of about 20 bar is expected to be optimal (which is discussed in more detail in chapter 7). As an example the modelled solid conversion as a function of time is displayed in Figure 2.7 for a reduction cycle with 50% CO at 800 °C and 20 bar for different particle diameters. It is shown that the particle size does not have a large effect on the effective reaction rates when smaller than about 5 mm, because the extent of internal diffusion limitations decreases with increasing pressure. Hence, the fact that larger particles have to be used to reduce the pressure drop in a packed bed reactor does not have a negative overall impact on the process performance.
2.4. Conclusions The effect of the pressure on the reaction rates of CuO/Al2O3 and NiO/CaAl2O4 particles as oxygen carriers for chemical-looping combustion has been investigated using a pressurized magnetic suspension balance and the experimental results have been compared with a numerical particle model. From the experimental results it was concluded that the pressure decreases the solid reaction rate and it is proven that this decrease is kinetically controlled and can be well described in case a pressure factor with a negative exponent is included in the kinetic description. A possible explanation for the observed pressure effect is the competition between gaseous species for adsorption on the oxygen carrier and the number of oxygen vacancies that decreases with the pressure. Furthermore, it was shown that the solid conversion was independent on the total pressure and thus only related to the temperature.
For packed bed applications, increasing the pressure has also a positive effect on the overall reaction rates, because the internal diffusion limitations (controlled by Knudsen diffusion flux) decrease with increasing pressure. For that reason, still reasonably high reaction rates can be obtained when using relatively large oxygen carrier particles.
|27
3 Demonstration of
CLC in packed beds using syngas and
CuO/Al2O3 This chapter is based on the following paper: Hamers, H.P., Gallucci, F., Cobden, P.D., Kimball, E., Van Sint Annaland, M., 2014. CLC in packed beds using syngas and CuO/Al2O3: Model description and experimental validation. Applied Energy 119, 163-172.
28 |
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3
Abstract The performance of CuO/Al2O3 as oxygen carrier in a packed bed reactor with syngas as fuel is investigated in this chapter, while also studying the (possible) carbon deposition and the effect of sulfur impurities on the stability of the carrier. Both experiments and simulations are used for this purpose. Cyclic experiments (oxidation with air and reduction with syngas) have been carried out in a lab scale packed bed reactor with 13wt% CuO/Al2O3. The experimental results were well described by a 1D reactor model, provided that critical attention was given to the reaction rate for the complete reduction reaction, accounting for the dramatic decrease in the reaction rate at high solid conversions. Feeding syngas (pH2=pCO=0.1bar) resulted in 1.1% carbon deposition of the feed. Steam was proven to be more effective in reducing carbon deposition than CO2. Moreover, it has been found that CuO/Al2O3 catalyzed the water gas shift reaction and the reaction rate was not permanently affected by exposure to H2S, two key factors for CLC operation. The results described in this chapter imply that CuO/Al2O3 is an effective oxygen carrier in packed bed reactors as long as the maximum temperature does not exceed the copper melting point, and that the developed model is able to describe the performance at larger scales accurately.
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |29
3
3.1. Introduction It is still questionable what oxygen carrier should be used that fulfills all the requirements for packed bed CLC. In normal oxidation/heat removal/reduction cycles, a large temperature rise should be achieved in a single bed (450-1200 °C). While thermodynamically possible when considering calculations based on existing materials, it will be a challenge to find an oxygen carrier that realizes such a large temperature rise, while also showing high chemical and mechanical stability and reactivity at the same time. This aspect is discussed in more detail in the next section, focusing on the selectivity of oxygen carriers and the required amount of active material in the oxygen carrier to achieve the desired temperature rise.
Copper oxide is an oxygen carrier that can possibly be used for the CLC process in packed bed reactors. Many researchers have studied copper oxide as an oxygen carrier in CLC processes, but mainly in fluidized bed systems, the most often researched configuration for CLC. Copper oxide was identified by Mattison et al. (2003) to be a fast reacting oxygen carrier, while its kinetics were described with a shrinking core model by García-Labiano et al. (2004). It was shown that it reacts fast, even at low temperatures. It was also observed that during reduction, the particle conversion rate drops dramatically if a solid conversion of about 80% is reached. However, this was not taken into account in the kinetics described in (García-Labiano et al., 2004), because the fluidized bed process was carried out at higher temperatures and lower solid conversions. On the contrary, in packed beds, the oxygen carriers could undergo (almost) complete reduction and oxidation. This could be an advantage in terms of solid inventory, but also requires an extended kinetic model to fully describe the process.
In packed bed reactors, CuO has been studied with methane as fuel by Noorman et al. (with Al2O3 as support) (Noorman et al., 2010b) and Corbella et al. (with TiO2 and SiO2 as support) (Corbella et al., 2006, 2005). Noorman et al. also studied the influence of operating conditions (Noorman et al., 2011b). However, the reduction of CuO with syngas in packed bed CLC reactors has not yet been reported.
When CLC is carried out with syngas as fuel, side reactions could occur, like carbon deposition and the water gas shift reaction. These reactions are discussed in the next section. Furthermore, syngas usually contains some H2S (in low quantities), which could react with the oxygen carrier, according to thermodynamics resulting in Cu2S (Wang et al., 2008). In such cases, the carrier might be poisoned and/or deactivated. According to thermodynamic studies and experimental evidences, H2S is converted to SO2 if oxygen is available (either on the carrier, or present in CO2 or H2O) (Forero et al., 2010;
30 |
Jerndal et al., 2006; Wang et al., 2008). This means that for a fluidized bed system, if the oxygen/fuel ratio is high enough, sulfur does not form solid species with the copper oxygen carrier. However, in packed beds, syngas could come into contact with a completely reduced carrier and in this case, sulfides (or sulfates) might be formed. Tian et al. have carried out thermogravimetric analysis (TGA) experiments with CuO/bentonite and syngas/H2S and observed deactivation within 10 cycles (Tian et al., 2009). In this chapter, the effect of H2S in the syngas on CuO/Al2O3 will be investigated by dedicated TGA experiments.
The goal of this chapter is to demonstrate the performance of CLC with syngas and CuO/Al2O3 in packed bed reactors. Lab-scale packed bed reactor experiments are described by the model and the impact of side reactions is studied. The next section opens with the pros and cons of the most promising oxygen carriers for packed bed CLC. Afterwards, the experimental setup and the model are described. Subsequently, the packed bed experiments with CuO/Al2O3 are discussed (oxidation with air and reduction with syngas), including a comparison with the modeling results. The description of the reduction reaction rate is also discussed with additional TGA experiments. Afterwards, the effect of CO2 and H2O on carbon deposition is evaluated and it is shown that CuO/Al2O3 catalyzes the water gas shift reaction. Finally, the effect of the presence of sulfur in the syngas feed on the process performance is studied by exposing the carrier to sulfur and measuring the reaction rate afterwards.
3.2. Oxygen carriers for CLC with syngas This chapter describes CLC experiments with syngas in a packed bed reactor containing a copper based oxygen carrier. In this section, the influence of the oxygen carrier selection on the reactor performance is studied for CLC with syngas as fuel. A critical aspect in the oxygen carrier selection is its selectivity (i.e. the extent at which the syngas is converted into H2O and CO2), which is studied by thermodynamics. Another important aspect is the maximum temperature rise achieved during the process. Therefore, the temperature change in the reactor is examined for different oxygen carriers, including the temperature change during reduction with syngas.
3.2.1. Thermodynamics of solid phases and reduction reactions In this work, four main oxygen carriers have been studied, based on nickel, copper, iron and manganese. The oxygen carrier materials are supported on materials like Al2O3, MgAl2O4, SiO2, TiO2 or ZrO2. The support material is added to increase the chemical and mechanical stability, to keep the temperature rise limited and to provide a porous structure. The porous structure is particularly desired for packed bed particles to
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |31
3
combine good mass transport inside the particle with a limited pressure drop over the bed. Which support is selected depends mainly on the stability, porosity and costs.
Selectivity of carriers The most studied oxygen carriers are nickel, copper, iron and manganese based. For an oxygen carrier for CLC application it is important that it completely converts syngas to CO2 and H2O in order to avoid fuel slip. In case of fuel slip, not all the reactants have reacted, leading to a decreased process efficiency and some downstream cleaning processes might be necessary as well. Therefore, a selectivity very close to unity is required, which means that all CO and H2 is being converted into CO2 and H2O.
Nickel is present as metallic Ni in reduced form and NiO in oxidized form. The selectivity to CO2 and H2O decreases with increasing temperature and, at 1000 °C, it is about 99% (Jerndal et al., 2006). The reduction will probably be carried out at a lower temperature and, in that case, this might not pose a serious problem. For copper, the selectivity is unity at all conditions. It may consist of three phases: CuO, Cu2O and Cu. However, because of the melting point of Cu at 1085 °C, the maximum temperature of this carrier is limited (and thus so is its applicability to one stage CLC).
Depending on the operating conditions, manganese and iron can be present in several phases. Manganese can be present as Mn2O3, Mn3O4, MnO and Mn. Mn2O3 decomposes at around 950 °C and the reaction to Mn hardly prevails with syngas. So, for this study, only the phases Mn3O4 and MnO are considered. An overview of the reaction selectivities is shown in Table 3.1.
Table 3.1: Properties of different metal/metal oxide pairs.
Metal oxide Lowest melting point,
°C
Selectivity γCO at 800 °C (Jerndal et al.,
2006)
Selectivity γH2 at 800 °C (Jerndal et al.,
2006)
Active weight material required for 750 °C T rise*
Costs, €/ton
(Lyngfelt, 2011b)
NiO/Ni 1455 0.995 0.995 19% >20,000 CuO/Cu 1085 1.000 1.000 21% >3000 Fe2O3/Fe3O4 1565 1.000 1.000 ** >1000 Fe3O4/FeO 1370 0.541 0.526 *** >1000 Mn3O4/MnO 1562 1.000 1.000 51% >1000 *In this calculation Al2O3 was considered as support material. ** It is not possible to reach a 750 °C temperature rise with only the Fe2O3/Fe3O4 oxygen carrier pair. *** Because of the low selectivity, the active weight content for a 750 °C temperature rise is not relevant.
32 |
For Mn3O4 and MnO, the selectivity is not an issue, but could pose a problem for iron based oxygen carriers, which can be present as Fe2O3, Fe3O4, FeO and Fe. Fe2O3 should not be further reduced than Fe3O4, because otherwise the selectivity drops to undesired values (Jerndal et al., 2006). This can be achieved by recycling CO2/H2O or adding steam to the syngas stream, so that the oxygen concentration is kept at a certain level.
Carbon deposition If syngas is used as the fuel, carbon deposition on the oxygen carrier can occur, depending on the type of oxygen carrier and the operating conditions. Carbon formation is especially a concern when the degree of reduction is high, which is the case during reduction in a packed bed reactor since (at the inlet) syngas is in contact with a reduced oxygen carrier (Cho et al., 2005). Carbon deposition is caused by the Boudouard reaction, equation 3.1.
22 CO CO C 3.1
Due to carbon formation, the carbon capture efficiency is reduced. During oxidation, the carbon will be burnt to form CO2. This situation should be avoided as much as possible, because in the end this CO2 is emitted to the atmosphere instead of being captured. The Boudouard reaction occurs at low temperatures and high pressures. This equilibrium is independent of the oxygen carrier considered, but to what extent the Boudouard reaction occurs depends on the kinetics. It is known that nickel and iron are catalysts for the Boudouard reaction (Cho et al., 2005). When these carriers are used, the temperature should be kept high or CO2 and/or H2O has to be added to the syngas feed. For instance, when using Cu/Al2O3 as the oxygen carrier, carbon deposition can be completely avoided at 20 bar and 450-1200 °C by using a H2O/CO ratio of 1.5 (HSC, 2013).
Water gas shift reaction Another side reaction that could influence the process is the water gas shift reaction, equation 3.2.
2 2 2CO H O H CO 3.2
The reaction is exothermic, if it produces H2. The equilibrium is independent of the pressure. If pure syngas is fed to the process, the effect of the water gas shift reaction is negligible at the entrance of the reactor, because it can only prevail when a mixture of
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |33
3
reactants and products is present. If H2 reacts faster than CO with the oxygen carrier, CO can be converted with H2O into CO2 and H2, which reacts faster with the oxygen carrier. In this case, the water gas shift reaction has an advantageous effect (Spallina et al., 2013). To prevent carbon deposition, CO2 and/or H2O can be added to the fuel and, in that case, the water gas shift reaction can play an important role.
3.2.2. Temperature change with syngas as fuel The temperature increase in the packed bed should be large enough to generate power at a high efficiency. The temperature increase in a packed bed reactor can be calculated with equation 3.3 (Noorman et al., 2007). In this equation, the initial bed temperature is assumed to be equal to the gas inlet temperature and the thermodynamic properties are assumed to be independent of the temperature.
,
, , ,0
,
R i
p s act p g g iin
act g i
HT C M C M 3.3
From equation 3.3 it is shown that the temperature increase is not influenced by the flow rate and the reaction kinetics. For this case, a change in flow rate or a slight change in the activity (kinetics) will not affect the temperature of the gas stream fed to the gas turbine. This approach is only valid for separated reaction and heat fronts. If the difference in front velocities is small (due to low solid reactivity or high dilution of gas reactants), no temperature plateau is produced, but rather a moving temperature peak. The height of this peak is influenced by the kinetics.
In Figure 3.1, the temperature rise during oxidation is shown as a function of the active metal content in the oxygen carrier material. It is assumed that the initial and the inlet temperature is 450 °C, but these temperatures have little influence on the temperature rise (because the temperature dependency of the reaction enthalpy and the solid heat capacity is very moderate). Although manganese and iron could be present in several phases, in this calculation, only the reactions between Mn3O4/MnO and the Fe2O3/Fe3O4 are considered. The thermodynamic properties were obtained from (Barin, 1993; Daubert and Danner, 1985). The desired temperature increase during oxidation should be around 750 °C, because air is fed at 450 °C (typical temperature of the air in IGCC plants after compression to 20 bar) and should be heated to 1200 °C to be fed to the downstream gas turbine. According to the results presented in Figure 3.1, the oxygen carrier content can be tuned, such that the desired temperature increase is achieved. For instance, using Ni as the oxygen carrier would require an active content
34 |
of less than 20wt%, which would also result in higher mechanical stability (Hossain and de Lasa, 2008; Hossain et al., 2009; Sedor et al., 2008). The disadvantages of nickel is that the costs are higher compared to the other oxygen carriers and that safety measures have to be taken due to its toxicity (Adanez et al., 2012). Copper could in principle also be used because of its high oxygen capacity, resulting in a steep temperature rise as a function of active content in the oxygen carrier material; however, the low temperature melting point makes this material unsuitable for high temperatures. Using Mn as the oxygen carrier would require a greater amount of active material in the particles to reach the desired temperature rise, with possible consequent detrimental effect on the mechanical stability during the different oxidation/reduction cycles. Finally, using Fe as the oxygen carrier and limiting the reduction to Fe3O4 would not be sufficient to reach the desired temperature rise in a single bed; and further reducing iron is not an option as the selectivity drops to undesired values (Jerndal et al., 2006). The results of this calculation also suggest that a combination of oxygen carriers can be used to achieve the total temperature rise required. For instance, using a first bed with Cu as the oxygen carrier to reach a first temperature rise of about 400-500 °C (well below the melting point of Cu) will then require a second bed with Mn or Fe as the oxygen carrier with lower active material content, and thus also the stability of the second oxygen carrier would probably be enhanced. This option will be discussed as a two-stage configuration in chapter 5.
In Figure 3.2 and Figure 3.3, the temperature change caused by the reduction is also reported. In case syngas is used, the temperature change depends on the CO/H2-ratio. Unlike the reduction with methane, which is often endothermic, the reduction with syngas is an exothermic reaction (except for the reduction of Fe2O3 with H2). The advantage of an exothermic reaction is that the temperature of the reactor will not drop during the reduction, which would have a negative effect on the reaction rates and, therefore, the process efficiency. However, attention has to be paid on how the heat gained during reduction can be used in the most efficient way, which will be discussed in chapter 5.3.
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |35
3
0.0 0.2 0.4 0.60
200
400
600
800
Fe3O4/Fe2O3
MnO/Mn3O4
Cu/CuOT
durin
g ox
idat
ion
with
air
(°C
)
Active weight content (kg/kg)
Ni/NiO
Figure 3.1: The temperature rise for the most general oxygen carriers as a function of the active weight content in the oxygen carrier (air inlet temperature of 450 °C).
0.0 0.2 0.4 0.6
0
100
200
300
Mn3O4/MnO
Fe2O3/Fe3O4
CuO/Cu
NiO/Ni
T du
ring
redu
ctio
n w
ith H
2 (C
)
Active weight content (kg/kg)
Figure 3.2: The temperature change due to the reduction with pure H2 as a function of the active weight content in the oxygen carrier (inlet temperature of 450 °C).
36 |
0.0 0.2 0.4 0.60
100
200
300
400
Fe2O3/Fe3O4
Mn3O4/MnO
NiO/Ni
CuO/Cu
T du
ring
redu
ctio
n w
ith C
O (°
C)
Active weight content (kg/kg)
Figure 3.3: The temperature change due to the reduction with pure CO as a function of the active weight content in the oxygen carrier (inlet temperature of 450 °C).
3.3. Materials and methods
3.3.1. Packed bed reactor for CLC An experimental setup has been constructed in which oxygen carriers can be tested in packed bed CLC. It consists of a high temperature stainless steel tube (OD x ID x L = 35 x 30 x 1500 mm) that contains oxygen carrier particles. Inside the reactor, 48 thermocouples have been installed (Rössel, type K). The reactor is heated by 3 ovens, each with a capacity of 200 W (insulated with KVS125), that surrounds the reactor tube. The reactor and oven are further insulated with vermiculite. A simplified scheme of the setup is depicted in Figure 3.4.
The feed flow rate and composition is controlled by Bronkhorst mass flow controllers. The flow downstream of the reactor is cooled with a water cooler and the produced water is collected. The pressure is controlled by a back pressure controller (Bronkhorst). The dry composition is measured with a mass spectrometer (MKS Cirrus 2). For these experiments there was no additional analyzer for CO and therefore the components CO and N2 were not distinguished by the mass spectrometer.
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |37
3
Figure 3.4: Simplified schematic overview of the packed bed reactor setup.
The experiments have been carried out with a 13 wt% CuO/Al2O3 oxygen carrier (diameter 1 mm, Sigma-Aldrich). With separate TGA experiments it was found that the mass change of the CuO/Al2O3 oxygen carrier corresponds to an active weight content of 12.5wt% (instead of the nominal 13wt%). The reactor also contains inert materials, alumina (diameter 3 mm, Sigma-Aldrich) positioned upstream and downstream of the oxygen carrier.
During the experiments the flow rate of the reactants was kept relatively low (2-4 Ln/min.) to keep a reasonable breakthrough time. The total flow rate was kept constant at 20 Ln/min to ensure that the temperature profile in the reactor is not influenced by a flow rate change. For these reasons, the reductions were carried out with a diluted stream (10% reactant). The inlet pressure is 2 bar and the initial bed temperature is 600 °C. An overview of the operating conditions is given in Table 3.2.
Table 3.2: The selected operating conditions for the experiments.
Fuel Reduction gas Oxidation gas Air,
Ln/min
Initial solid temp.
°C
Inlet pressure,
bar H2,
Ln/min CO,
Ln/min H2O,
Ln/min CO2,
Ln/min N2,
Ln/min
H2 2 18 20 600 2 CO 2 18 20 600 2
Syngas 1 1 0-1 0-7 11-18* 20 600 2 *= the N2 flow is adjusted so that the total gas flow is always 20 Ln/min.
Insulation material
Vent
Mass spectrometerH2O
AirN2H2COCO2
Steam
oven oven oven
Packed bed reactor
PCWater condenser
T48
T1
p p
38 |
3.3.2. Thermo gravimetric analyzer The reduction reactions with copper oxide using CO and H2 have been studied in detail by thermo gravimetric analyzer (TGA) experiments. The mass change of an oxygen carrier sample (100 mg) was measured during reactions with gaseous products (H2, CO, air) under isothermal conditions and at atmospheric pressure. In the TGA experiments, the same reactant partial pressure was used as in the packed bed (pO2=pH2=0.2bar). By monitoring the mass change, the reaction rate was measured. More information about the setup is given in Noorman et al. (Noorman et al., 2010a).
3.3.3. Packed bed reactor model The oxygen carrier reacts with the fuel (reduction) or the air (oxidation). During the process, reaction and heat fronts propagate through the bed, resulting in temperature changes in the bed and a breakthrough curve. The temperature and concentration profiles obtained from the experiments are compared with the results from a numerical 1D pseudo-homogeneous packed bed reactor model. An overview of the mass and energy balances is given in Table 3.4. For the incorporation of the heat losses, it is assumed that the surroundings (oven) has a constant (initial) temperature and the amount of heat that is transferred from the reactor to the oven wall can be evaluated with a constant heat transfer coefficient, α. This coefficient has been estimated from the experiments and is listed in Table 3.3. The axial mass dispersion and the effective heat conductivity are described by the equations in Table 3.5. More details on the model and the numerical solution strategy are provided in Smit et al. (2005) and Noorman et al. (2007).
Table 3.3: The selected parameters for the model.
Oxygen carrier 12.5wt% CuO on Al2O3 Particle diameter, mm 1 Solids bulk density in oxidized state, εsρs, kg/m3 762 Void fraction, m3gas/m3reactor 0.4 Reactor length, m 0.54 Heat transfer coefficient, α , W/(K*m2) 45 (evaluated with independent
measurement) Temperature of environment, °C 600 Effectiveness factors, η
- for oxidation kinetics - for reduction with H2 - for reduction with CO
-2.4196Xs3 + 2.2182Xs2 - 0.4998Xs + 0.7141 0.84 0.70
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |39
3
Table 3.4: The mass and energy balances used in the model.
Component mass balances for the gas phase
, , ,,
i g i g i gg g g g g ax eff s i iv D r M
t x x x
Component mass balance for the solid phase ,0 s j
s s act g j jr Mt
Energy balance (gas and solid phase)
, , , ,4
g g p g s s p s s g p g eff g i R i envr
T T TC C v C r H T Tt x x x d
Reaction rate 0
, 0 , ,exp nAi g p g i s j
Er k c cRT
Momentum balance (Ergun equation, (Ergun, 1952)): 2 2
2 3 3
1 1150 1.75g gg g g g
pp g g
v vdpdx dd
Table 3.5: The heat and mass dispersion descriptions.
Effective axial heat dispersion (Vortmeyer and Berninger, 1982) 2 2
,0
Re Pr Re PrPe 6 1 Nu
g geff bed
ax g
λbed,0 is calculated by the Bauer and Schlünder equation (Bauer and Schluender, 1978). Gunn and Misbah equation (Gunn and Misbah, 1993)
242 0.17 0.33 expRe
Pe241 0.17 0.33 exp
Re
ax
Gunn equation (Gunn, 1978) 2 0.2 1/3 2 0.7 1/3Nu 7 10 5 1 0.7Re Pr 1.33 2.4 1.2 Re Prg g g g
Axial mass dispersion (Edwards and Richardson, 1968)
, 2
0.73 0.5Re Sc 9.7
Re Sc
ax eff g pg
g
D v d
40 |
The reactions that occur in the CLC process over a Cu-based oxygen carrier using syngas as fuel are listed in Table 3.6. It is assumed that syngas is fully converted into CO2 and H2O, which means that carbon deposition and selectivity are not taken into account. The water gas shift reaction is only considered if CO2 or H2O are fed into the reactor. The kinetic parameters are listed in section 2.2.3 (Table 2.1). The reaction rates of the different components have been described by the advanced particle model described in detail in (Noorman et al., 2011a), which includes internal mass diffusion limitations. In the particle model it is assumed that the pore size is 200 Å. The particle model calculates an effectiveness factor, which determines to what extent the reaction is kinetically controlled. The results of the computations (effectiveness factors) are shown in Table 3.3. In the reactor model, the reaction rate is multiplied by the effectiveness factor to end up with a realistic description of the reaction rate.
Table 3.6: Overview of the reactions that occur in the CLC process using a Cu based oxygen carrier and syngas as fuel.
2O 2 Cu 2 CuO oxidation of Cu ΔHr = -304 kJ/mol O2
2 2H CuO H O Cu reduction of CuO with H2 ΔHr = -95 kJ/mol H2
2CO CuO CO Cu reduction of CuO with CO ΔHr = -131 kJ/mol CO
22 CO C CO Boudouard reaction ΔHr = -86 kJ/mol CO
2 2 2CO H O CO H2CO2CO2 reversed water gas shift reaction ΔHr = -36 kJ/mol
2 2C O CO carbon combustion reaction ΔHr = -394 kJ/mol O2
3.4. Results and discussion In the next section, the oxidation and the reduction of the copper based oxygen carrier are demonstrated in the lab-scale packed bed reactor and these experiments are described with the model to prove that the model is able to predict the CLC process in a packed bed reactor. Afterwards, the effect of side reactions is discussed in section 3.4.2, like carbon deposition, the water gas shift reaction and possible reactions with sulfur impurities.
3.4.1. CLC experiments in packed beds compared with model description This section opens with the experimental results and the model description of the oxidation reaction of Cu/Al2O3 with air. Subsequently, the reduction reaction with syngas (CO/H2-ratio of 1) is discussed. In first instance, the model description for the axial temperature profiles is not that well for the reduction, but at the end of this section, a better description is shown by improved fittings for the kinetics (especially at high solid conversions).
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |41
3
Oxidation with air Cu/Al2O3 is oxidized with an air flow rate of 20 Ln/min. The experimentally measured axial temperature profiles in the reactor are depicted in Figure 3.5 (markers). The oxidation is an exothermic reaction and therefore the temperature in the bed rises and a heat plateau can be observed in the bed. Because of recuperative heat exchange between the inlet gas and the reacted oxygen carrier, a heat front is observed, moving with a lower velocity than the reaction front. Between the heat front and the reaction front, a higher temperature is observed, referred to as heat plateau. The maximum temperature rise that is observed is 200 °C. In an ideal situation with infinitely high reaction rates, no mass and heat dispersion and no heat losses, a maximum temperature rise of 244 °C could have been reached (as has been discussed in section 3.2.2). However, due to heat losses the actual temperature rise is lower, which is also predicted by the reactor model. The packed bed has a diameter of only 3 cm, resulting in a high amount of heat losses through the walls of the reactor. With the conditions described here, the oxygen carrier is oxidized completely in 1.5 minutes; at that moment, the reaction front has reached the end of the bed.
The experiments are compared with results from the packed bed reactor model as shown in Figure 3.5 (lines). Switching times of valves and the time required for the flow to reach the reactor inlet were taken into account. The model describes the reaction fronts well. The pressure drop was, as expected from the experiments, between 0.1 and 0.2 bar. In general, two differences can be observed. First, the position of the reaction front is predicted well, but the transition of the reaction front to the heat plateau is not as steep as the model predicts. Furthermore, the velocity with which the heat front propagates along the bed is slightly slower than the model prediction. This might be related with the heat capacity of the reactor wall, which reduces the heat front rate and is not incorporated in the model. The same behavior is shown in Figure 3.6, in which the temperatures at the end of the oxygen carrier bed from the experiments and the model are shown. The temperatures are described relatively well by the model with literature kinetics for oxidation.
In Figure 3.7 the O2 fraction at the outlet is depicted. This profile is less steep than the axial temperature profiles, due to additional dispersion in the downstream water cooler (with a volume of 3 liter). Similar extent of dispersion was also detected in case inert gases were fed. The shape of the breakthrough curve is different, but the moment of oxygen breakthrough is in line with the prediction, based on mass flow rates and oxygen carrier active content used.
42 |
0.0 0.1 0.2 0.3 0.4 0.5
600
700
800
t=0min.
heatfront
t=5min.
t=2.5min.
t=1min.
Tem
pera
ture
(°C
)
Axial position (m)
t=0.5min. reactionfront
Figure 3.5: Axial temperature profile during oxidation with an air flow of 20 Ln/min. The markers are experimental values and the lines are model predictions.
0 2 4 6 8550
600
650
700
750
800
850
exp.
modelOut
let t
empe
ratu
re (°
C)
Time (min.)
Figure 3.6: Temperature at the end of the reactive part of the packed bed versus the outlet temperature from the model as a function of time during the oxidation cycle.
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |43
3
0 1 2 3 4 50
5
10
15
20
25
O2
Dry
gas
frac
tion
at o
utle
t (%
)
Time (min.)
Figure 3.7: Experimentally determined O2 mole fraction at the outlet of the reactor (after the cooler) during the oxidation cycle.
Also at a flow rate of 40 Ln/min, the experiments are predicted well by the model (not reported here). At higher flow rates, the concentration and temperature fronts become less steep.
Reduction with syngas The reduction was carried out with syngas (5% H2, 5% CO and 90% N2) at a flow rate of 20 Ln/min. The experimentally observed axial temperature profiles during the reduction are depicted in Figure 3.8. The reaction front and the heat fronts develop through the bed, like it was the case for the oxidation cycle. However, during the reduction cycle, the heat front is closer to the reaction front, because the reactants were diluted with N2. The reason to dilute the feed was to keep the initial temperature profile similar by keeping the total flow rate at the same level of 20 Ln/min. The close heat and reaction fronts do not result in a heat plateau as occurred during the oxidation cycle, but in a heat peak that moves through the bed. The height of the heat peak is dependent on the steepness of the reaction and heat fronts. Therefore the temperature that is achieved depends on the kinetics. A maximum temperature rise in the bed of 185 °C is observed. Around an axial position of 0.4 m, the temperature rise is somewhat lower. This is probably caused by more heat losses that occur at that position, which can also be observed from the slightly lower initial bed temperature.
44 |
0.0 0.1 0.2 0.3 0.4 0.5
600
700
800
t=0min.
t=5min.
t=2.5min.t=1min.
Tem
pera
ture
(°C
)
Axial position (m)
Figure 3.8: Experimental axial temperature profiles for a reduction cycle of CuO/Al2O3 with 5% H2, 5% CO and 90% N2 at 600 °C and 2 bar.
0 2 4 6 8 100
2
4
6
H2
CO2
Dry
gas
frac
tion
at o
utle
t (%
)
Time (min.)
Figure 3.9: Experimental outlet gas fractions during reduction of CuO/Al2O3 with 5% H2, 5% CO and 90% N2 at 600 °C and 2 bar.
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |45
3
After about 4 min, the reaction front has reached the end of the bed and breakthrough of the reactants occurs. The breakthrough curves can be more clearly observed in the outlet concentration profiles in Figure 3.9. It is shown that first CO2 is produced and no reactants are present at the outlet, which means that no fuel slip occurs. The CO2 concentration (5.2%) is the same as the theoretical value (water is not included in the analyzer sample). After breakthrough of reactants, the concentration of reactants increases with time before the inlet concentration is reached. During that time, some CO2 is still produced, which is caused by the particles being not completely reduced (thus some remaining CuO reacts with CO). In section 3.4.2 it will be discussed that this is not caused by the Boudouard reaction, because only 1.1% of the carbon is deposited. At the same time (5min<t<10min), the H2 concentration is slightly increasing with time, indicating that a reaction is still continued. To what extent the reduction reaction can still occur, will be discussed in the next paragraph.
Effect of reduction kinetics on packed bed model description Experiments and model predictions were compared for the case where only H2 was fed during the reduction cycle, because initially the largest difference between the model and the experiments was observed in this case. The axial temperature profiles in the case of 10% H2 and 90% N2 in the feed are shown in Figure 3.10. In this figure, the experimental data are compared with the profiles obtained with the reactor model and kinetics as described above. It is shown that the experimentally determined temperature rise is much smaller than the model predicts. If the reaction rate is slower than expected, not all of the reaction heat is released at the same moment, resulting in a lower maximum temperature. As discussed in the introduction, the kinetics expressions present in literature in general do not describe the reaction rate well at a high solid conversion (a condition that is expected to largely prevail in the packed bed reactor).
This behavior has also been observed in TGA experiments. These experiments have been carried out with 20% H2 at different temperatures and atmospheric pressure. The partial pressure of H2 was 0.2 bar, which was the same as in the packed bed reactor experiments.
In Figure 3.11, the solid conversion as a function of time is shown. The experimental data from the TGA experiments are shown with markers and the lines represent the results of the particle model with literature kinetics. Most of the curve is described reasonably well by the model; however, it appears that at a solid conversion above 80%, the reaction rate decreases dramatically, which is not captured well with the correlation used in the reactor model. It could be an explanation for the fact that the reaction front develops faster through the bed than the model predicts.
46 |
0.0 0.1 0.2 0.3 0.4 0.5
600
700
800
t=0min.
t=5min.t=2.5min.
t=1min.
Tem
pera
ture
(°C
)
Axial position (m)
Figure 3.10: Axial temperature profiles during reduction with 10% H2 and 90% N2 at 600 °C and 2 bar. Experimental values are shown by markers and the lines are the model predictions.
0 10 20 30 40 500.0
0.2
0.4
0.6
0.8
1.0
T=600°C
T=700°C
T=800°C
Sol
id c
onve
rsio
n (-
)
Time (s)
Figure 3.11: Comparison between the kinetics description by the particle model and experiments with CuO/Al2O3 with 20% H2 and 80% N2 at atmospheric pressure and different temperatures.
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |47
3
A decrease at high solid conversion has also been observed during reaction with CO, but less pronounced than with H2.
The lower reaction rate above 80% solid conversion has been observed for other oxygen carriers as well, i.e. on NiO/Al2O3 by Dueso et al. (2012). According to Dueso et al., the decrease in reduction rate was attributed to a reaction between Al2O3 and Ni to NiAl2O4 during the previous oxidation. NiAl2O4 is reduced more slowly than NiO and therefore it takes more time before the whole particle is reduced. The same reaction mechanism could occur for CuO/Al2O3, because the formation of CuAl2O4 has been observed by Zhao et al. in XRD-experiments with CuO/Al2O3, that was exposed to oxidation TGA experiments at 800 °C (Zhao et al., 2013). For the NiO/Al2O3 case, Dueso et al. described the last part of the conversion by diffusion in the product layer of uniform spherical grains of varying size, according to equation 3.4 (Dueso et al., 2012). In this equation, the diffusion rate is high at low solid conversion.
,02
6exp expsDs n
s x red gg
Eb DD k X C
RTr
3.4
Based on this approximation, the reaction rate expression used in this work has been modified to equation 3.5, which describes how the reaction rate is controlled by kinetics at low solid conversion and controlled by diffusion at high solid conversion. This equation has been implemented in the particle model, using the parameters listed in Table 3.7. The results have been compared with the TGA experiments with 20% H2 at 600, 700 and 800 °C, given in Figure 3.12, and show that the model with the modified kinetics predicts the reaction rate reasonably well, and clearly much better than the earlier reaction rate expression using only the simple Arrhenius equation.
,
0 ,0
11 1
exp exp exps
ni g p g
A Ds x red
r c
E Ek D k XR T R T
3.5
Table 3.7: Parameters used in the diffusion correlation.
Ds,0, mol1-nm3n-3s-1 EDs, kJ/mol kx, - H2 1.25*1016 160 20 CO 1.563*1017 170 20
48 |
To investigate whether the difference between the experiments and the reactor model was indeed caused by the used kinetics description, the adapted kinetic expression (equation 3.5) was implemented in the reactor model and the results were compared with the same experimental results as already shown in Figure 3.10. The new comparison is shown in Figure 3.13, where the new modeling results are compared with the experimental data (markers) for the reduction with H2. It is evident that the temperature profiles from the model describe the experimental data much better than before. Both the maximum temperature and heat and reaction fronts are described correctly. Only at the end of the bed a difference can be discerned, which is probably caused by heat losses being underestimated by the reactor model in that region of the bed. The same observation was made for the cases with syngas and CO as the reducing gas, allowing for the conclusion that the slow reaction at a solid conversion above 80% has a great impact on the process performance. So, for a good simulation of the reactions in packed beds, an accurate description for the kinetics at a high degree of solids conversion has to be implemented. It is clear that more detailed experiments are required to elucidate whether the decrease in reaction rate (and the influence of temperature on this) is related to the formation of CuAl2O4. After more detailed information has been obtained, the kinetic expression, equation 3.5 might have to be adapted. Other mechanisms may also result in a similar behavior and an equally well description such as the approach by Li et al, which is based on island size distribution (Li et al., 2012).
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |49
3
0 50 100 150 2000.0
0.2
0.4
0.6
0.8
1.0
T=600°C
T=700°C
Sol
id c
onve
rsio
n (-)
Time (s)
T=800°C
Figure 3.12: The kinetic description including the diffusion effect compared with TGA experiments at 800 °C and with 20% H2.
0.0 0.1 0.2 0.3 0.4 0.5
600
700
800
t=0min.
t=5min.t=2.5min.t=1min.
Tem
pera
ture
(°C
)
Axial position (m)
Figure 3.13: Experimental (markers) and modeled (lines) temperature profiles for reduction with 10% H2 and 90% N2. The diffusion effect is included in the reaction rate expression.
50 |
3.4.2. Side reactions
Carbon deposition and water gas shift reaction Experiments have shown that not only the reduction of CuO with H2 and CO and the oxidation of Cu with O2 prevail in the reactor, but also the Boudouard and water gas shift (WGS) reactions can occur. If the Boudouard reaction occurs, carbon is deposited during the reduction and burnt off as CO2 during oxidation. In that case, carbon is not completely captured during the reduction stage, but emitted during the oxidation stage, resulting in a lower carbon capture efficiency of the process. For that reason, this should be avoided as much as possible. A possibility to decrease or suppress the carbon formation is by adding CO2 and steam to the feed (basically recycling part of the exhaust of the reduction stage).
As a base case, the reduction with syngas has been selected and the effect of adding CO2 or H2O on the extent of carbon deposition has been measured. The flow rate was fixed at 20 Ln/min. The amount of deposited carbon was determined from the measured CO2 gas fraction in the outlet stream during the subsequent oxidation, when the deposited carbon is burnt off and CO2 is formed.
The effect of only CO2 addition is depicted in Figure 3.14a. First, it is observed that the amount of carbon deposition in the base case (with CO2/CO ratio of 0) is 1.1%, which is limited in comparison with what is expected according to thermodynamics, which predict that 33% of the carbon could be deposited on a reduced carrier. In a realistic case on a large scale, the process would be operated at a much higher partial pressure of reducing gases, resulting in even more carbon deposition. For that reason, the effect of adding CO2 and steam is relevant. In case only 1.1% carbon deposition occurs, the temperature rise is hardly influenced by the carbon combustion reaction, because much more oxygen is consumed during oxidation (50-100 times more) than the amount of carbon that is deposited, while the reaction enthalpy is in the same order of magnitude as the oxidation reaction.
The results in Figure 3.14a show that even with a high CO2/CO ratio of 7, carbon deposition is not completely avoided, but very small 0.1%. Thermodynamics predict that carbon deposition could have been avoided under these circumstances.
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |51
3 0 1 2 3 4 5 6 7 80.0
0.5
1.0
1.5
Car
bon
depo
sitio
n (%
)
CO2/CO-ratio(-)
a)
0.0 0.2 0.4 0.6 0.8 1.00.0
0.5
1.0
1.5b)
Car
bon
depo
sitio
n (%
)
H2O/CO ratio (-)
Figure 3.14: The percentage of carbon deposition defined as the number of moles C deposited during the reduction cycle relative to the number of moles of CO fed (at Tbed,initial=600 °C, p= 2 bar, yH2=yCO=5%) varying CO2/CO ratios (a) and H2O/CO ratios (b).
The effect of steam addition on the carbon deposition is shown in Figure 3.14b. Here it can be observed that steam is a much more effective way to avoid carbon deposition. At a steam/carbon ratio of 0.5, carbon deposition is already suppressed to a large extent and at a ratio of 1, it is completely avoided. According to the thermodynamics, a steam/carbon ratio of 0.6 is sufficient at the experimental conditions. These results show that to suppress carbon formation, a wet syngas has to be used for CLC rather than recycling the product gases. Whether recycling product gases suffices to suppress carbon deposition, depends on the CO/H2-ratio of the fuel.
If the process would have been carried out at 450 °C and 20 bar (which is expected on large scale), the steam/carbon ratio should be at least 1.5. However, if a different switching scheme between oxidation, heat removal and reduction is used, the reduction can be carried out at around 1200 °C (Spallina et al., 2013). In that case, less steam is required to avoid carbon deposition.
The water gas shift reaction occurs if the products (H2O and CO2) and reactants (H2 and CO) of CLC are available at the same place. If carbon formation should be avoided, steam has to be added to the feed stream, which makes the WGS prone to occur throughout the reactor. It should be noted that the WGS reaction is in any case beneficial for the process because it consumes CO and produces H2, which has much faster kinetics, and thus the possible fuel slip is avoided (Spallina et al., 2013). An example is shown in Figure 3.15, which contains the outlet gas fraction profiles in the case of a steam/carbon ratio of 1. After the breakthrough of H2, some CO2 is still produced and the amount of H2 is higher than the amount fed. This is caused by the
52 |
WGS reaction consuming the CO and steam in the feed and producing H2 and CO2, independent of the fuel combustion. From this, the conclusion can be drawn that Cu/Al2O3 operates as a catalyst for the WGS reaction. The formation of CO2 is in line with the equilibrium of the WGS reaction at the reactor temperature. The H2-concentration is somewhat higher associated to inaccuracies in the H2 calibration curve of the analyzer at this concentration.
0 2 4 6 8 100
2
4
6
8
10
Dry
gas
frac
tion
at o
utle
t (%
)
Time (min.)
CO2
H2
Figure 3.15: The outlet concentrations during a reduction with 5% H2, 5% CO, 5% H2O and 85% N2 (total flow of 20 Ln/min, Tbed,initial=600 °C and p= 2 bar).
Exposure of the oxygen carrier to H2S Syngas from the gasification plant for use in power production will also contain some H2S (generally in the range of 10-20 ppm in order to conform to sulfur emission standards). To assess whether the oxygen carrier is deactivated by H2S, samples of CuO/Al2O3 were tested in a TGA before and after long exposures to higher amounts of H2S.
First, an activated CuO/Al2O3 sample was heated with N2 to 600 °C. Then, it was completely reduced with H2 for 1 hour (at 600 °C) and subsequently exposed to 1000 ppm H2S in N2 for 24 hours at 600 °C. After 4 hours of H2S exposure, the H2S started to breakthrough and some SO2 was formed. After about 10 hours, the initial concentration was reached so that the carrier cannot uptake more H2S. After 24 hours, the sample was cooled down to 100 °C in N2 and exposed to O2/N2, starting with a
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |53
3
low O2 concentration and then gradually increasing to the concentration of air, in order to passivate the sample.
After this procedure, the influence of H2S on the carrier performance was evaluated by TGA experiments. Eight cycles were carried out, consisting of reduction with 20% H2 and oxidation with air at 800 °C and atmospheric pressure. The reactor was heated to 800 °C in an inert atmosphere. In the first few cycles, the mass change (during reduction and oxidation) was larger than expected according to the thermodynamics, but the kinetic curve still has the same shape (see Figure 3.16). After a few cycles (in this case 5 cycles), the reaction rate curve of the exposed sample is the same as for the sample before sulfur exposure. This is illustrated in Figure 3.17, where the conversion versus time of a fresh sample is compared with the sample that was initially exposed to H2S and then cycled normally 8 times.
Apparently, sulfur that has reacted with the carrier, operates temporarily as an oxygen carrier as well. According to thermodynamics, H2S is formed in environment with little oxygen. If the particles are oxidized at 800 °C and 1 bar, about 90% of the sulfur is converted into SO2 and 10% in CuSO4 (Wang et al., 2008). If it is assumed that 35% of the copper is initially present as Cu2S, the measured behavior can be explained with these figures. So, in the first cycles the sulfur acts as oxygen carrier forming Cu2S (in reduced state) and CuSO4 (in oxidized state), but the amount of sulfur is decreasing with increasing cycle number, because of SO2-formation.
X-ray photoelectron spectroscopy (XPS) was carried out with the H2S-exposed sample just after exposure and the sample that had been oxidized and reduced 8 times after the exposure. The XPS experiments proved that the sulfur was reacted during these cycles. In the scan of the exposed sample, a peak can be observed at 169 eV (as illustrated in Figure 3.18), which means that the sample contains some sulfur. After the oxidations and reductions, no peak can be observed, which means that the sample no longer contains any sulfur.
54 |
0 40 80 120 160 200
-3.0
-2.5
-2.0
-1.5
-1.0
-0.5
0.0M
ass
chan
ge (m
g)
Time (s)
increasingcycle number
Figure 3.16: Mass change as a function of time during the first eight reduction cycles of the H2S-exposed CuO/Al2O3 sample.
0 50 100 150 2000.0
0.2
0.4
0.6
0.8
1.0
Sol
id c
onve
rsio
n (-)
Time (s)
base case previously exposed to H2S
Figure 3.17: Comparison of the solid conversion as a function of time during the reduction with 20% H2 at 800 °C for a CuO/Al2O3 sample and a sample that was previously exposed to H2S and then cycled for 8 times with the usual oxidation/reduction cycles.
3. Demonstration of CLC in packed beds using syngas and CuO/Al2O3 |55
3
156 158 160 162 164 166 168 170 172 174 1762500
3000
3500
4000
4500
5000
5500
6000
exposed to H2S and 8 ox./red. cycles
Inte
nsity
Binding energy (eV)
exposed to H2S
Figure 3.18: XPS spectra of the H2S-exposed sample and the sample that has undergone 8 oxidations and reductions afterwards.
From the results in Figure 3.17 and Figure 3.18, it can be concluded that the exposure to sulfur has no permanent impact on the oxygen carrier performance. For application in a power plant, it depends on the syngas quality if desulfurization systems have to be installed. In case the fuel contains H2S, it is possible that SO2 is formed during reduction, but also during oxidation SO2 will be formed, because during reduction the H2S is in contact with the reduced carrier forming Cu2S. In the following chapters, H2S-free syngas is considered, assuming that a desulfurization step is carried out prior the CLC reactor. The effect of this additional desulfurization step is discussed at the end of this thesis (section 7.5).
3.5. Conclusions In this chapter, Chemical Looping Combustion has been carried out in a lab-scale packed bed reactor with CuO/Al2O3 as the oxygen carrier and syngas as the fuel. In particular the effect of the fuel composition has been studied, including the influence of side reactions such as the Boudouard and water-gas-shift reactions. The results from the experiments have been compared with the 1D reactor model. It is shown that the model is able to describe the process very accurately, also at higher flow rates, when a proper description for the reduction kinetics has been implemented.
56 |
During the reduction of the oxygen carrier, the reaction rate decreases dramatically at a solid conversion above 80% (especially during H2 reduction). If this effect is not considered in the reactor model, the temperature in the bed is overestimated and the reaction front velocity is underestimated. However, if the slower reaction rate is included in the reaction rate expression, the temperature profiles during the reduction experiments in the packed bed reactor can be described very well. A correlation has been fitted to separate TGA experiments, but the approach could be further improved when the physical understanding of the decreased reaction rates at high solids conversion has been elucidated in more detail.
The experimental results have shown the influence of the Boudouard reaction and the water-gas-shift reaction. At a partial H2 and CO pressure of 0.1 bar, little carbon deposition occurs. However, on a large scale the process will be operated at 20 bar and then carbon deposition could have a larger impact. It was found that this could be suppressed by adding CO2 and steam to the reactants, where steam appears to be more effective to avoid carbon deposition. At a steam/carbon ratio of 1, carbon deposition was completely avoided. Moreover, Cu/Al2O3 has been proven to catalyze the WGS reaction. The WGS reaction can be quite advantageous since H2 reacts faster with the oxygen carrier than CO. Presence of H2S does not result in permanent deactivation of the carrier.
All the results show that CuO/Al2O3 can be an appropriate oxygen carrier for packed bed CLC with syngas as long as the maximum temperature is limited because of the relatively low melting temperature of copper.
|57
4 Pressure effect on
performance of CLC in packed bed
reactors This chapter is based on the following paper: Hamers, H.P., Gallucci, F., Williams, G., Van Sint Annaland, M., 2014. Experimental demonstration of CLC and the pressure effect in packed bed reactors using NiO/CaAl2O4 as oxygen carrier. Submitted to Fuel.
58 |
4. Pressure effect on performance of CLC in packed bed reactors
Abstract In this chapter an experimental proof-of-concept is given for packed bed CLC at elevated pressures (where the total pressure has been varied between 2 and 7.5 bar) in a pressurized lab-scale reactor using NiO/CaAl2O4 particles as oxygen carrier material and H2 and syngas from coal gasification as fuel in the reduction step. The experiments have demonstrated that the pressure has quite a small effect on the overall performance of the reactor. The experiments are well described by a numerical reactor model, with which it is shown that the required high temperatures for power production can be reached in case the reactor were scaled-up (decreasing the heat losses). A drawback of the selected oxygen carrier is that the reduction kinetics depend on the temperature during the previous oxidation, which has been fixed during all the experiments reported in this chapter. The reduction kinetics decrease with the oxidation temperature, which is probably caused by interaction with the support material and therefore some further modifications to the support material may be required for actual high-temperature applications. In case the process is carried out with syngas and steam, the heat management strategy has to be tuned to account for the heat produced by the water gas shift reaction.
4. Pressure effect on performance of CLC in packed bed reactors |59
4
4.1. Introduction The operating pressure is a key parameter strongly influencing the efficiency of chemical looping based power plants. While elevated pressures have a positive influence on the power plant efficiency, it is also reported in literature that the pressure has a slightly negative influence on the kinetics (García-Labiano et al., 2005) and this effect might affect the performance in packed bed reactors. Until now, no research has been carried out to investigate the pressure effect for CLC in packed bed reactors.
Several oxygen carriers (OCs) could be selected for the packed bed CLC process (Adanez et al., 2012), in general Ni-based OCs, ilmenite and a combination of Cu/Mn-OCs seem to be promising candidates. Ilmenite has the advantage that it is a natural material and thus that the costs of the raw material are low and it is environmentally friendly. With ilmenite the temperature rise in the reactor should be sufficient to reach 1200 °C while also the fuel conversion to H2O and CO2 is complete. The main drawbacks of the material are that the reaction rates are relatively low at low temperatures and that it needs to be activated at high temperatures to obtain acceptable reaction rates (Ortiz et al., 2014). To work with ilmenite, the operation strategy needs to be adapted, so that the reduction can be carried out at high temperature (Spallina et al., 2013). On the other hand, nickel and copper based OCs have a better reactivity at low temperatures. While the low melting point of copper does not allow it to be used in a single stage CLC process, it remains a very suitable candidate to be used as first material in a two-stage CLC process with another oxygen carrier, like a Mn-based OC as second material. So far, only Ni-based OCs have been identified to be possibly applicable for CLC in single stage due to its good reactivity at low temperatures, high oxygen capacity and thus high temperature rise during oxidation and good stability at high temperatures. Additionally, Ni based catalyst pellets are available for high temperature reforming reactions, thus the technology to produce OC pellets is readily available in industry which would decrease the costs of OC particle development for packed bed operations.
In this chapter, a demonstration is given of a nickel-based OC for (single-stage) CLC in packed bed reactors and the influence of the pressure on the reactor performance is examined. The experiments are compared with simulations using a 1D packed bed reactor model. In the next section, the NiO/CaAl2O4 oxygen carrier material, the packed bed reactor setup and the numerical model are discussed. Subsequently, the results are presented and discussed demonstrating the operability of the packed bed CLC system at elevated pressures with Ni-based OCs. In the end, the process is also demonstrated with ilmenite as OC to show that the process is also possible with other materials.
60 |
4.2. Materials and methods
4.2.1. Experimental setup A schematic overview of the packed bed chemical looping setup is illustrated in Figure 4.1. The packed bed reactor (LxODxID=160x400x63 mm), developed by Array Industries BV can be operated with a maximum temperature of 1200 °C and a pressure of 10 bar. The oxygen carrier particles are packed in an Inconel tube (liner) with a thickness of 6 mm. A thick layer of insulation material (Isofrax 160 kg/m3) surrounds the liner to decrease the heat losses. The outer carbon steel reactor wall is used to allow operation at higher pressures. The reactor wall is heated with tracers to 300 °C to minimize the heat losses. A schematic representation of the reactor cross section is given in Figure 4.2.
The reactant gas feed composition and flow rate are controlled by mass flow controllers (Bronkhorst). The feed gas is heated up to the desired feed temperature in two 2.2 kW ovens, which are installed in series upstream of the reactor. For the steam production, demineralized water is pressurized by a HPLC pump (Hitachi) and mixed with the gas stream in the oven. The line between the oven and the reactor is traced with high temperature tracing to minimize the temperature decrease between the oven and the packed bed reactor.
Inside the reactor, there is a tube (LxOD=700x14mm) with 20 K-type thermocouples (Rössel, 1.5 mm thermocouples) to measure the temperature at different axial positions in the bed. The thermocouples are surrounded by an Inconel protection layer (2 mm) and some air can be present between the protection layer and the thermocouples. Hence, a response time of the thermocouple of approximately 45 seconds has to be taken into account with the temperature measurement. The reactor only contains oxygen carrier material in the area where the temperature is measured. The rest of the reactor (below and above the OC) contains inert material (clay granules).
To allow dilatation due to the different temperatures between the liner and the reactor shell (insulation side) the liner is not welded at the top of the reactor. To avoid that the reactant gases enter the reactor shell, a small N2 flow of 2 Ln/min is fed to the shell and mixed with the reactor gases at the reactor exit. By this method, the composition in the reactor is not influenced by the N2, but the outlet composition is (and this is corrected for in all the results presented here).
4. Pressure effect on performance of CLC in packed bed reactors |61
4
Figure 4.1: Picture of the packed bed reactor setup (left) and schematic overview of the packed bed reactor setup (right).
d=389 mm
d=75 mmd=75 mm
insulation material
inconelliner
stainlesssteel reactorwall with tracing
TC
oxygen carrier
d=75 mmd=63 mm
14 mm
Figure 4.2: Schematic overview of the reactor cross section.
Vent
Mass spectrometer
and CO analyzer
H2O
AirN2H2COCO2
Steam Pack
ed b
ed re
acto
r
PCWater condenser
T20T1
p
p
Oven
62 |
The reactor outlet gasses are cooled by an air cooler with fins and a water cooler, where steam is condensed and separated from the other gasses. Afterwards, a digital back pressure regulator is used to control the system pressure, downstream the pressure regulator the gas is sent to a mass spectrometer (MKS spectra products) and a CO analyzer (Siemens Ultramat 23) working in parallel. The residence time in the ovens, coolers and lines to the analyzers (about 45 s) and this time is subtracted from the experimental data.
4.2.2. Packed bed reactor model The 1D packed bed reactor model is a numerical model that solves the mass and energy balances of the system. It is basically the same model as described in section 3.3.3, but with different correlation for the kinetics and a small adaptation of the energy balance. Only the modified equations are listed in Table 4.1, while the other equations can be found in Table 3.4. The kinetics of NiO/CaAl2O4 have been discussed in section 2.2.3. The water gas shift reaction kinetics are taken from literature (Xu and Froment, 1989).
Table 4.1: The kinetic equation and the energy balances used in the model.
Kinetics 0
,s p s acti
j
dXrb M dt
[in mol gas/m3particle/s]
02 1
0 03 3
3
1 (1 ) (1 )
ng
s
s s
cb rdX
dt rc
rX Xk D D
0
5
exp
10
Aq
tot
ER Tp
kk , ,0 exp expsD
s s x
ED D k X
R T
Energy balances*
, , , , ,4
g g p g s s p s TC steel p steel s g p g eff g i R i rl wr
T T TC C C v C r H T Tt x x x d
,4 4w
liner steel p steel steel rl w le w environmentr r liner
T TC T T T Tt x x d d
* εTC=0.03 and εliner=0.417
4. Pressure effect on performance of CLC in packed bed reactors |63
4
Two energy balances were required to describe the temperature inside the reactor. The first balance is solved for the gas and solids inside the reactor and the thermocouple. The second balance is solved for the liner. Heat transfer occurs between the reactor and the liner (αrl=90 W/m2/K) and between the liner and the surroundings (αls=4.5 W/m2/K). The latter heat transfer coefficient has been separately determined from the measured axial temperature profiles during heating and cooling down of the reactor.
4.2.3. Oxygen carrier Experiments have been carried out with A Johnson Matthey product, HiFUEL® R110 (17wt% Ni based catalyst supported on CaAl2O4 for steam reforming of natural gas), available in pelleted form from Alfa Aesar. The particles were received as fluted rings with a height and diameter of 11 mm. For this study, the particles were crushed to an average particle size of 2 mm. Before the oxygen carrier was loaded in the reactor, the particles were activated by exposing it to two redox cycles with reductions with H2 at 900 °C and oxidations with air (this activation procedure was first verified in separate TGA experiments). The oxygen carrier and reactor properties are listed in Table 4.2.
From the bottom of the reactor, it contains inert materials until the first thermocouple. This is followed by a bed of 0.69 m containing the NiO/CaAl2O4 and finally inert material is used on top to reduce the residence time after the reactive zone.
Table 4.2: Oxygen carrier and reactor properties.
Oxygen carrier 17wt% NiO on CaAl2O4 Particle size 2 mm Particle porosity 0.55 Average pore size 200Å Reactor bed length 0.69 m Reactor bed diameter 0.063 m Solid bulk density 953.7 kg/m3 reactor αbl (bed to liner) 90 W/m2/K αls (liner to environment) 4.5 W/m2/K
27 W/m2/K (at first 0.06 m, different insulation material)
Temperature of surroundings 300 °C
64 |
4.3. Results and discussion The NiO/CaAl2O4 has been tested at different pressures and fuel compositions. First, the oxidation is discussed, subsequently the reduction with H2 is presented followed by the reduction with syngas. An overview of the experiments and the operating conditions is given in Table 4.3.
Table 4.3: Operating conditions used during the packed bed reactor experiments.
Reaction Flow rates Pressure, bar
Initial bed temperature,
°C
time, min.
Oxidation Air: 100 Ln/min. 2-7.5 500 15 Reduction with H2 H2: 20 Ln/min.
N2: 80 Ln/min. 2-7.5 600 30
Reduction with syngas
CO: 7.1 Ln/min. H2: 2.9 Ln/min.
CO2: 1.2 Ln/min. H2O: 10.7 Ln/min.
(steam) N2: 28.1 Ln/min.
2-7 600 30
4.3.1. Oxidation and heat removal The oxidation has been carried out with an air flow of 100 Ln/min at an initial bed temperature of around 450-500 °C, while the pressure was varied between 2 and 7.5 bar. Before the oxidations, the bed has been reduced for 30 minutes with H2 at 600 °C, which is sufficient to reach the maximum solid conversion at these temperature.
From the moment that the reactor is fed with air, a temperature rise is observed due to the exothermic reaction. Although a temperature of 1050 °C could be reached with this oxygen carrier (after reduction with H2 at 600 °C), a lower temperature is observed in the reactor: the maximum measured temperature is 800 °C. This difference is mainly related to heat losses unavoidable at these relatively small scales. From the moment that the bed temperature increases, the temperature difference between the bed and the liner increases and this causes a heat flux to equalize the temperatures. The heat capacity of the liner is quite significant in comparison with the reactor bed (note that the reactor diameter is only 6.3 cm) and therefore a considerable amount of heat that is produced by the exothermic reaction is transferred to the liner and this causes a lower temperature rise.
The thermocouple has a significant heat transfer resistance and due to this, there is a delay in the temperature measurement. This delay is estimated at 45 seconds and is
4. Pressure effect on performance of CLC in packed bed reactors |65
4
taken into account via equation 4.1 (Patil et al., 2006). Because of the 45 seconds response time of the thermocouple and the presence of heat losses, the maximum temperature reached in the reactor is never actually measured.
Figure 4.3 shows the outlet temperature as measured and as predicted by the model with and without accounting for the response time of the thermocouple. It can be seen that the response time of the thermocouple can result in a difference of about 100 °C between the real and the observed maximum; according to the model, the maximum temperature is above 900 °C. At a certain moment, the liner temperature is close to the reactor temperature and from that moment, the insulation material that surrounds the liner becomes the limiting factor and this reduces the heat loss rate. The decrease in the heat losses can also be observed from Figure 4.3. This figure also demonstrates that a good prediction of the thermocouple temperature can be given by the model when accounting for all the phenomena occurring in the reactor.
1exp 1 exptTC TC
t tT T T 4.1
0 10 20 30 40 50
500
600
700
800
900
thermocouple
Out
let t
empe
ratu
re (°
C)
Time (min.)
reactorbed
models
exp.
Figure 4.3: The temperature at the reactor outlet as a function of time on stream in the oxidation cycle compared with model predictions showing the reactor temperature and the thermocouple temperature (i.e. accounting for the time delay).
66 |
The axial temperature profiles during the oxidation are given in Figure 4.5. The markers show the experimental data, while the model predictions are displayed by the lines. Also here quite a good correspondence of the temperature profiles is found. The heat and reaction fronts can be clearly discerned in this figure. At the moment that the bed is completely oxidized, O2 breaks through and this is observed in Figure 4.4 with a good correspondence between the model (that does not account for the Ar present in the air) and the experimental results. After the oxidation, the heat removal step starts, in which air of about 500 °C is fed to the reactor, while hot air is produced. Figure 4.3 shows that the obtained temperature during this phase is lower (about 800 °C) and that the temperature is decreasing with time. The temperature inside the reactor decreases, because the heat front moves the heat through the reactor, but also the temperature drops due to the heat losses.
0 2 4 6 8 100.00
0.05
0.10
0.15
0.20
Dry
gas
frac
tion
at o
utle
t (-)
Time (min.)
O2
Ar
Figure 4.4: Oxygen breakthrough profile during oxidation with 100 Ln/min. of air at 2 bar.
4. Pressure effect on performance of CLC in packed bed reactors |67
4
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
500
600
700
800 t=0min. t=1min. t=2min. t=3min. t=4min. t=5min.
Tem
pera
ture
(°C
)
Axial position (m)
model exp
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
500
550
600
650
700
750
800 t=5min. t=10min. t=20min. t=30min. t=40min.
Tem
pera
ture
(°C
)
Axial position (m)
model exp
Figure 4.5: Axial temperature profiles during oxidation of Ni/CaAl2O4 with 100 Ln/min. The markers describe the experimental data and lines illustrate the model predictions.
68 |
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7400
500
600
700
800
900
1000
1100
t=15min.
Tem
pera
ture
(°C
)
Axial position (m)
tt=3min.
Figure 4.6: Axial temperature profiles from simulation of oxidation without considering heat losses.
In an actual CLC reactor the heat losses will be minimized, and the heat capacity of liner and thermocouples would not influence the operation. To show the performance of the reactor for the ideal case, a simulation has been carried out simulating the oxidation experiment without heat losses and without liner or thermocouple. The resulting axial temperature profiles are shown in Figure 4.6, which shows that with the same settings, an air flow with a constant temperature of 1070 °C is produced for a certain period, which corresponds to the active weight content of 17% NiO. Another marked difference between the simulations with and without heat losses and heat capacity of the liner and thermocouple, is the time required to remove the heat. In the simulation and the experiment with heat losses and heat capacity of the liner and thermocouple, it takes more than 30 minutes to remove all the heat, while this is only 15 minutes without heat losses and additional heat capacity. This difference is caused by the heat that is stored in the liner during the actual experiments that is released back during the heat removal step and influences the reactor temperature. This results in a higher overall heat capacity of the reactor bed and thus in a lower heat front velocity, resulting in a longer time required for the heat removal.
4. Pressure effect on performance of CLC in packed bed reactors |69
4
0 5 10 15400
500
600
700
800
900
1000
1100d=2m
no heat losses
d=1m
d=0.063m
Out
let t
empe
ratu
re (°
C)
Time (min.)
d=0.3m
Figure 4.7: The effect of the reactor diameter on the extent of heat losses.
The extent of the heat losses has been evaluated by carrying out simulations with different reactor diameters, while the liner and insulation material thickness was fixed. The calculated outlet temperature profile for different reactor diameters are given in Figure 4.7. The reactor diameter has been varied between 6.3 cm (lab scale reactor) and 2 m and also the case without heat losses is shown. From the figure it is clear that at least a 1 m diameter reactor is required to significantly reduce the impact of heat losses. It has to be noted that the selected cycle time and oxygen carrier may also have some impact on the result. The longer the selected cycle time, the more heat can be lost during the process. Besides, this oxygen carrier material is quite porous, which implies that the solid bulk density is relatively low. In case a material with higher bulk density is selected, the heat capacity of the solids is higher and then also the effect of the heat losses decreases. It is also possible to increase the thickness of the insulation material or use a material with a lower conductivity. In any case it is demonstrated that at industrial scale, the heat losses can be decreased such that these would not have a significant impact on the overall performance of the reactor.
70 |
4.3.2. Pressure effect on oxidations The pressure during the oxidation has been varied from 2 to 7.5 bar. The breakthrough profiles obtained during these tests are shown in Figure 4.8. The differences between the profiles are considered to be negligible. The mass flow rates are fixed during these tests and for that reason, the residence time is increased with the pressure. This results in more mixing after the reactor and for that reason, the profile for the 7.5 bar experiment differs slightly from the other measurements. The effect of the pressure on the temperature rises that are observed during the experiments is also very small (see Table 4.4). During all the experiments, the measured temperature rise is around 300 °C. Because the pressure effect on the kinetics is quite small and the effectiveness factor increases with the pressure, also hardly any effect is observed in the modeling results (thus not shown here).
Table 4.4: Temperature rise during oxidation at different operating pressures.
Operating pressure, bar Temperature rise during oxidation, °C
2 303 4 302 6 300 7.5 297
0 1 2 3 4 50.00
0.05
0.10
0.15
0.20
2bar 4bar 6bar 7.5bar
O2 f
ract
ion
at o
utle
t (-)
Time (min.)
Figure 4.8: Breakthrough profiles after oxidation with 100 Ln/min of air at different pressures. The previous reduction was carried out at the same operating pressure.
4. Pressure effect on performance of CLC in packed bed reactors |71
4
4.3.3. Pressure effect during reduction with H2 The pressure has also been varied during reductions with H2. Also here, the flow rate fixed at 100 Ln/min, but here the fuel contains 20% of H2, balanced with N2. Initially, the reactor temperature was at 600 °C. This reaction is slightly exothermic and therefore a the temperature changes during this experiment are quite small. For that reason, the temperature profile is not shown in this work. The outlet gas fraction profiles are shown in Figure 4.9 and this is compared with the output from the model.
0 2 4 6 8 100.00
0.05
0.10
0.15
0.20
H2 d
ry g
as fr
actio
n (-
)
Time (min.)
expmodel
Figure 4.9: H2 fraction at the outlet for the 2 bar experiment compared with the model.
72 |
For the model description one important aspect has to be mentioned. During TGA experiments with this oxygen carrier material, it was observed that the reduction kinetics depend on the temperature during the previous oxidation. The effect of this temperature can be quite dramatic as demonstrated in Figure 4.10. Here the solid conversion is shown for a reduction with 20% H2 at 450 °C for different oxidation temperatures (varied between 450 and 800 °C). It appears that in case the material has been oxidized at 450 °C (as in a real CLC operation), the material is reduced quite fast and a high degree of reduction is obtained, but when the oxidation temperature is increased, the reduction slows down. This effect is probably related to the solid structure of the NiO and the CaAl2O4 support (and possible sintering effect at too high temperatures). Indeed, one should remember that during oxidation in this lab scale packed bed reactor, a temperature between 700 and 950 °C could be reached, dependent on the degree of reduction. The kinetics have been obtained considering an oxidation temperature of 800 °C and a correction factor has been included for different oxidation temperatures based on the decrease of the slope of the line during TGA experiments.
As example, the effect of the oxidation temperature on the H2 breakthrough during reduction is shown in Figure 4.11. A steeper breakthrough is observed during reduction, when the previous oxidation has been carried out at a lower temperature. This effect should be distinguished from the effect of the pressure. Hence, in each case that operating conditions are compared, the temperature reached during the previous oxidation was the same.
Considering the above mentioned point regarding the oxygen carrier, it is difficult to describe the reduction reaction in a very accurate way and a more detailed study on the kinetics would be needed. But despite this, still a reasonable description of the breakthrough profile is obtained.
4. Pressure effect on performance of CLC in packed bed reactors |73
4
0 200 400 600 800 10000.0
0.2
0.4
0.6
0.8
1.0
Tox=800 C
Tox=600 CS
olid
con
vers
ion
(-)
Time (s)
Tox=450 C
Figure 4.10: Solid conversion curves for reduction with 20% H2 at 450 °C varying the temperature during the previous oxidation.
0 2 4 60.00
0.05
0.10
0.15
0.20
p=4barTmax, prev.ox.=890 C
Dry
frac
tion
at th
e ou
tlet o
f H2 (
-)
Time (min.)
p=7.5barTmax, prev.ox.=860 C
Figure 4.11: H2 breakthrough in case the previous oxidation is carried out at a different temperature.
74 |
For cases with the same previous oxidation, the pressure during the reduction was varied in the medium scale reactor, while the other operating conditions remained constant. The total flow rate was 100 Ln/min and the temperature was about 600 °C. The H2 breakthrough profiles are illustrated in Figure 4.12. Only a small influence of the pressure on the breakthrough profiles can be observed. The main difference is shown after the breakthrough, where the H2 concentration takes longer to reach the inlet fraction at a higher operating pressures. It is expected that the kinetics decrease with the pressure for high degrees of reduction. After the oxygen carrier was reduced for more than 30 minutes, the bed was purged. The measured H2 fraction during this purge is illustrated in Figure 4.13. It is observed that the extent of mixing is quite small and this will have a negligible influence on the measurement.
The pressure effect has been studied by modeling as well, but also here no significant differences were detected (not shown in this work).
Moreover, the temperature rise during oxidation is hardly influenced (as displayed in Table 4.4). Also from this observation, the conclusion can be drawn that the degree of reduction is independent on the pressure. This is in line with the results obtained in the high pressure TGA that have been discussed in chapter 2.
4. Pressure effect on performance of CLC in packed bed reactors |75
4
0 2 4 6 8 100.00
0.05
0.10
0.15
0.20
2bar 4bar 6bar 7.5bar
H2 f
ract
ion
at o
utle
t (-)
Time (min.)
Figure 4.12: Breakthrough profiles of H2 at 600 °C different operating pressures.
0 1 2 3 4 5
0.00
0.05
0.10
0.15
0.20
0.25
H2 g
as fr
actio
n at
out
let (
-)
Time (min.)
2bar 4bar 6bar 7.5bar
Figure 4.13: H2 fraction measured after switching from the reduction to the purge step.
76 |
4.3.4. Reductions with syngas For experiments with syngas, the total flow rate was set on 60 Ln/min. The same composition regarding the CO, H2, H2O and CO2 concentrations was taken from IG-CLC power plant calculations (that are discussed in chapter 6), but the flow was diluted with some extra N2. Because of the presence of N2, the total flow rate could be increased and this reduces the heat losses between the ovens and the reactor. The axial temperature profile and the breakthrough curves are shown in Figure 4.14 and Figure 4.15 respectively. The reactant mixture contains more CO and H2O in comparison to H2 and CO2 and therefore the water gas shift reaction occurs. The water gas shift activity can be observed by the heat production at the reactor entrance and also in the breakthrough curves, where more H2 is produced than fed to the reactor. If the water gas shift reaction occurs, the same amount of heat is produced, but the distribution of the heat production is different. For that reason, quite a high temperature peak could be observed at the reactor entrance and this has to be taken into account in the heat management strategy.
The experiments have been carried out at 7 bar as well and here the breakthrough curves had the same shape. From this it can be concluded that also for the reduction with syngas the pressure effect on the performance of the reactor is quite small.
4. Pressure effect on performance of CLC in packed bed reactors |77
4
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
500
550
600
650
700
750
t=0min. t=1min. t=2min. t=3min. t=4min. t=5min. t=10min.
Tem
pera
ture
(°C
)
Axial position (m)
Figure 4.14: Axial temperature profile of the reactor bed for syngas as fuel at 2 bar.
0 10 20 30 400.00
0.05
0.10
0.15
0.20
0.25
CO
H2
Dry
gas
frac
tion
at o
utle
t (-)
Time (min.)
CO2
Figure 4.15: Breakthrough curve for the reduction with syngas at 2 bar.
78 |
4.4. Reductions and oxidations with ilmenite as oxygen carrier A test series has also been carried out with ilmenite as oxygen carrier in the same reactor setup. Prior to the experiments, the material has been activated in a separate reactor during 20 cycles containing a reduction with a H2 (30% H2, 15% CO2, balanced with N2) at 850 °C and an oxidation with air. In this work, one experiment is discussed to show the implications of the type of oxygen carrier on the reactor performance.
The first 0.05 m of the reactor was filled with clay. On top of it, a bed of 0.82 m containing granules based on Norwegian ilmenite (75wt%) ore was placed. The theoretical temperature rise with this material is in the same order of magnitude as with the nickel based material. Because of the low reactivity of ilmenite, a different operation procedure needed to be selected than with nickel. The heat produced during oxidation is not directly blown out of the reactor, but a reduction is started (after a short purge) so that the reduction (with relatively slow kinetics) can be carried out at the highest possible temperature. The operating parameters are listed in Table 4.5.
Table 4.5: Operating parameters for the experiment with ilmenite as oxygen carrier in the packed bed reactor.
Oxygen carrier 75wt% ilmenite/25wt% Mn2O3 Granule size L x d = 20 x 3 mm Reactor bed length 0.82 m Solid bulk density 1469 kg/m3 reactor Operating pressure 2 bar Reduction flow rate H2: 12 Ln/min.
CO2: 6 Ln/min. N2: 22 Ln/min.
Reduction time 60 min. Oxidation flow rate Air: 40 Ln/min. Oxidation time 15 min.
The reduction is carried out with a 40 Ln/min. mixture of 30% H2 and 15% CO2 (balanced with N2). The results for the axial temperature profile as function of time and the temporal evolution of the dry gas composition at the reactor outlet are shown in Figure 4.16. Initially, a slight temperature increase can be observed close to the reactor inlet due to the exothermic reduction reaction. During this process, the bed is cooled down, because the gas is fed with a lower feed temperature compared to the initial reactor temperature. In the center of the reactor, a heat front can be observed. In the reactor, also the endothermic reverse WGS reaction takes place evident from the CO observed in the breakthrough profiles. The breakthrough is observed at around 15 min.
4. Pressure effect on performance of CLC in packed bed reactors |79
4
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
500
600
700
800
t=0min. t=5min. t=10min. t=15min. t=20min. t=25min.
Tem
pera
ture
(°C
)
Axial position (m)
0 10 20 30 40 50 600.00
0.05
0.10
0.15
0.20
Dry
gas
frac
tions
at o
utle
t (-)
Time (min.)
CO2
H2
CO
Figure 4.16: Axial temperature profile and dry gas fractions at the reactor outlet during reduction of ilmenite (operating conditions listed in Table 4.5).
80 |
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
600
700
800
900
t=0min. t=2min. t=4min. t=6min. t=8min. t=10min. t=12min. t=14min.Te
mpe
ratu
re (°
C)
Axial position (m)
0 2 4 6 8 10 12 140.00
0.05
0.10
0.15
O2 f
ract
ion
at o
utle
t (-)
Time (min.)
Figure 4.17: Axial temperature profile and O2 breakthrough profile during oxidation of ilmenite.
4. Pressure effect on performance of CLC in packed bed reactors |81
4
The breakthrough is less steep in comparison with the experiments with nickel-based OC, while the reduction is carried out at a higher temperature and the reaction is carried out with a lower flow rate. This results from the much slower reaction rates with ilmenite compared with NiO.
After a purge with N2, the bed is oxidized again with 40 Ln/min of air. Figure 4.17 shows that similar temperature profiles are obtained as during the oxidation of the nickel-based oxygen carrier. With ilmenite a maximum temperature rise of 329 °C is observed, which is higher than with nickel (with ΔTmax=300 °C). The amount of oxygen that reacts per kg oxygen carrier is similar for both oxygen carriers and therefore the theoretical temperature rise is at the same order of magnitude. But the main difference is that these granules have a lower porosity (15%) (Ortiz et al., 2014) and this results in a higher solid bulk density. In adiabatic packed bed processes, the solid bulk density does not influence the temperature rise in the reactor (Noorman et al., 2007). In this experimental setup, the liner that surrounds the reactor has a high heat capacity and this has a considerable influence on the reactor temperature as discussed in section 4.3.1. But the higher the solid bulk density, the smaller the impact of the liner and thus a higher temperature is observed inside the reactor when using ilmenite as oxygen carrier.
4.5. Conclusions In this chapter, the application of chemical-looping combustion has been demonstrated in a packed bed reactor using NiO/CaAl2O4 and ilmenite as oxygen carriers. For large scale application, a temperature rise of 750 °C should occur during oxidation, so that a hot air stream is formed that can be sent to the gas turbine. However, during the experiments, the maximum temperature rise was limited to 300 °C, which is caused by inevitable heat losses in experiments at lab-scale. However, the heat losses were quantified with separate experiments and were accounted for in the model yielding a good description of the axial temperature profiles and outlet composition as a function of time during the oxidation cycle. It was also demonstrated that without heat losses, the process could reach the high temperatures required for efficient power production. Despite the fact that the oxidation temperature influences the reduction kinetics, also a good description of the reduction cycle was obtained.
The effect of pressure on the performance in packed bed reactors has been investigated and it was shown that the pressure effect was quite small for the oxidation and for the reduction with H2 and with syngas. Therefore no problems are expected for high pressure packed bed CLC applications.
82 |
For practical applications, the interaction between the metal oxide and the support is not desired, so that a high conversion can be obtained at low temperatures; a performance similar to the one after reduction at 450 °C. In that case, no fuel slip is expected during reduction, while high temperatures could be reached during oxidation. With some adaptations to the support material, this Ni-based material may be made suitable for packed bed CLC operating at high temperatures and pressures.
Experiments in the same packed bed setup with ilmenite as oxygen carrier have demonstrated that ilmenite may also be a suitable material for packed bed CLC. For ilmenite, slower kinetics were observed and a slightly higher temperature was reached during oxidation, because of the higher solid bulk density, but the heat management strategy needs to be adapted to carry out the reduction at the highest possible temperatures.
|83
5 Two-stage-CLC,
a novel configuration for packed bed CLC
This chapter is based on the following papers:
- Hamers, H.P., Gallucci, F., Cobden, P.D., Kimball, E., Van Sint Annaland, M., 2013. A novel reactor configuration for packed bed chemical-looping combustion of syngas. Int. J. Greenh. Gas Control 16, 1-12.
- Kooiman, R.F., Hamers, H.P., Gallucci, F., Van Sint Annaland, M., 2014. Experimental demonstration of two-stage packed-bed chemical-looping combustion using syngas with CuO/Al2O3 and NiO/CaAl2O4 as oxygen carriers. Submitted to Industrial & Engineering Chemical Research.
84 |
5. Two-stage-CLC, a novel configuration for packed bed CLC
Abstract This chapter starts by quantifying the challenges for CLC fueled by syngas in packed bed operation and introduces a novel reactor configuration with which those challenges can be addressed. Continuous cyclic operation of a packed bed CLC system is investigated using a 1D numerical pseudo-homogeneous reactor model. Importantly, it is demonstrated that the temperature profiles that can occur in a packed bed reactor as a result of the different process steps do not accumulate, and have a negligible effect on the overall performance of the system. Moreover, it has been shown that an even higher energy efficiency can be achieved by feeding the syngas from the opposite direction during the reduction step (i.e. countercurrent operation). Unfortunately, in this operation mode, more severe temperature fluctuations occur in the reactor exhaust, which is disadvantageous for the operation of a downstream gas turbine. Finally, a novel reactor configuration is introduced and experimentally proven in which the desired temperature rise for hot pressurized air suitable for a gas turbine is obtained by carrying out the process with two packed beds in series (two-stage CLC). This is shown to be a good alternative to the single bed configuration, and has the added advantage of decreasing the demands on both the oxygen carrier and the reactor materials and design specification.
5. Two-stage-CLC, a novel configuration for packed bed CLC |85
5
5.1. Introduction In the previous two chapters the packed bed CLC process was experimentally demonstrated using syngas as fuel in reactors at different scales, higher pressures and with different OCs. In addition, a numerical reactor model was developed, and it was shown that the model can describe the experiments quite reasonably when properly accounting for the inevitable heat losses in small scale reactors. However, the described experiments focused on single oxidation or reduction cycles, starting each cycle with a bed consisting of either a completely oxidized or uniformly reduced oxygen carrier at more or less uniform temperature. When performing multiple cycles, heat may remain in the reactor bed non-uniformly after a previous cycle, which could influence the performance in the next cycle. To what extent the cyclic steady state is affected by non-uniform axial temperature profiles when starting a new cycle, is discussed in detail in this chapter.
In the previous chapter, two possibly suitable oxygen carriers, viz. nickel oxide and ilmenite, have been experimentally tested for packed bed CLC using a single packed bed. In this case, a temperature rise of about 750 °C should be achieved in one single reactor, while the material should also be able withstand 1200 °C and have sufficiently fast reduction kinetics at relatively low temperatures of about 450 °C (the air feed temperature). The combination of these requirements for the oxygen carrier proofs to be quite challenging and no oxygen carrier has yet been found that can fulfill all these requirements, despite the large research efforts. Moreover, some difficulties with the reactor construction can be expected because of the very large temperature gradients at high pressure operation. Therefore, a new configuration is proposed with which large temperature differences for the solid material can be avoided, and that is to carry out the process in two stages by using two packed beds in series instead of one single bed. In this two-stage configuration, the first bed is responsible for the first temperature rise, demanding that the oxygen carrier should be reactive at low temperature (e.g. copper-based materials), while the second bed is responsible for the remaining temperature rise, and thus the oxygen carrier should be able to withstand higher temperatures. In this chapter, this two-stage approach is introduced and experimentally demonstrated.
86 |
5.2. Packed bed reactor model The performance in packed bed reactors has been studied using the 1D packed bed reactor model that has already been discussed in detail in section 3.3.3. Industrial application is considered in this case and therefore heat losses are not taken into account (i.e. α=0).
5.2.1. Base case definition For this study, a base case has been defined with CuO/Al2O3 as the oxygen carrier. CuO/Al2O3 has been selected because of its fast reduction reaction rates at relatively low temperatures, high oxygen capacity and high selectivity. Moreover, it is one of the few commercially available oxygen carriers for packed beds and can thus be used in a future work for model validation. Although the melting point of Cu is lower than the 1200 °C required for high efficiency electricity production, a packed bed with CuO/Al2O3 can be used as the first stage in a two-stage system.
The specific oxygen carrier considered consists of 12.5wt% CuO/Al2O3 with a particle porosity of 60% (as commercially available). Considering a gas porosity in the reactor of 40%, a solid bulk density of 999 kg/m3 is reached in the oxidized state.
The highest energy efficiency in the gas turbine can be reached at a pressure of around 20 bar. The air for the oxidation is supplied by the compressor of the turbine and is assumed to be delivered at a temperature of 450 °C. Flow rates and reactor lengths are selected such that the maximum pressure drop is 1 bar (5%) and the duration of a cycle is around 10 minutes. Therefore, a particle size of 3 mm is selected combined with a flow rate of 5 kg/(m2s) for the oxidation cycle. Within 600 seconds, the bed is almost completely reduced at a flow rate of 0.15 kg/(m2s), which is a factor 30 smaller than the flow rate during oxidation.
For the fuel, syngas is selected with a H2/CO ratio of 1. In this study, a simplified approach (with only a constant and first order reaction) was used to describe the kinetics based on reaction rates published in literature (García-Labiano et al., 2004). The constant has been estimated conservatively to account for possible diffusion limitations. The operating conditions for the base case are listed in Table 5.1. The results shown in the next sections are based on these conditions.
5. Two-stage-CLC, a novel configuration for packed bed CLC |87
5
Table 5.1: Defined base case for the reactor model simulations.
Oxygen carrier 12.5wt% CuO on Al2O3 Particle diameter, mm 3 Solids bulk density in oxidized state, εsρs, kg/m3 999 Void fraction, m3gas/m3reactor 0.4 Reactor length, m 4.0 Oxidation/
Heat removal Reduction Purge
Gas inlet flow rate, kg/(m2s) 5 0.15 2 Gas inlet, mol fraction O2: 0.21
N2: 0.79 H2: 0.5 CO: 0.5
N2: 1
Gas inlet temperature, °C 450 450 450 Gas inlet pressure, bar 20 20 20 Reaction rate, ri, mol/(m3s) 50cO2 25cH2
25cCO -
Cycle time, s 730 600 10
5.3. Simulation results First, the axial temperature and concentration profiles for a single oxidation or reduction step are shown, assuming a uniform initial temperature throughout the reactor. These results are used to discuss the heat and reaction fronts developing in the reactor, as well as to evaluate the time period at which high temperature air is produced that can be sent to the gas turbine, and thus to define a process efficiency. Subsequently, results for the cyclic steady state after several alternating cycles are presented and discussed. These results are interesting because, in practice, there is always a remaining temperature profile from the previous cycle, which influences the performance of the next cycle. These temperature effects were not studied in detail or considered in previous literature works for CLC, where it was, for example, assumed that the temperature profiles were equalized via short periods of intermediate mild fluidization (Noorman et al., 2007). To what extent the remaining temperature profile in each cycle influences the performance is discussed in section 5.3.2. Furthermore, different feeding strategies are evaluated with a particular focus on the difference between co-current and counter-current feeding of the oxidizing and reducing feed streams.
88 |
5.3.1. Single cycle
Reduction step During reduction, the bed initially contains only CuO/Al2O3. When syngas is fed, CO and H2 react with CuO to form CO2 and H2O. Due to the reduction of the carrier, a reaction front develops through the reactor (Figure 5.1). The steepness of this front depends on the reaction kinetics and axial mass and heat dispersion.
The same type of front is observed in the temperature profile. Initially, the reactor temperature is uniform at 450 °C. When feeding syngas at 450 °C, the exothermic reduction reaction takes place and the temperature rises in the reactor. The temperature increases exactly where the reaction takes place, thus the temperature profile also shows the reaction front in the bed. As the feed temperature of the syngas is different from the temperature of the bed, a heat front moves through the bed as well, with a velocity lower than the reaction front. These fronts are described by equations 5.1 and 5.2 (Noorman et al., 2007). Since the heat front develops slower than the reaction front, a heat plateau arises in the bed, as shown in Figure 5.1.
0,
ing g g act
rs s act g i
v Mw
M 5.1
,
,
g g p gh
s s p s
v Cw
C 5.2
5. Two-stage-CLC, a novel configuration for packed bed CLC |89
5 0 1 2 3 4
450
500
550
600
Axial position (m)
Tem
pera
ture
(°C
)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
600s400s200s
600s400s
CuO
wei
ght f
ract
ion
(-)
200s
a)
0 1 2 3 4
0.0
0.1
0.2
0.3
0.4
0.5
H2O, CO2 H2, CO
H2O, CO2
Gas
frac
tion
(-)
Axial position (m)
H2, CO
b)
Figure 5.1: Axial temperature and solid concentration profiles in the reactor during the reduction step (a) and the axial gas fraction profiles at t = 200s (b).
90 |
Oxidation and heat removal step The oxidation step is started with a uniform bed temperature of 450 °C and Cu/Al2O3 as the bed material. The reduced carrier reacts with O2 to form CuO (oxidized carrier). This reaction is more exothermic than the reduction reaction, resulting in a larger temperature change. The axial temperature profiles and the oxygen gas fraction profiles during the oxidation cycle are shown in Figure 5.3. At a certain moment in time, the oxygen carrier has been oxidized completely. Then (at about 100 s for this simulation), the heat front reaches the end of the reactor and, from this moment in time, hot air is produced at the same temperature of the bed, which can be fed to a gas turbine to produce electricity. The temperature and oxygen concentration of the outlet stream of the reactor as a function of time is depicted in Figure 5.2. Hot air is produced until the heat front reaches the end of the reactor. Unfortunately, the heat front is not a steep front due to heat dispersion. Thus, some heat cannot be effectively used and would remain in the reactor (or be wasted).
0 200 400 600 800 1000
450
500
550
600
650
700
Temperature Oxygen gas fractionTime (s)
Tem
pera
ture
(°C
)
0.00
0.05
0.10
0.15
0.20
0.25
Oxy
gen
gas
fract
ion
(-)
Figure 5.2: Temperature and oxygen gas fraction profiles of the reactor outlet stream during oxidation and heat removal.
5. Two-stage-CLC, a novel configuration for packed bed CLC |91
5 0 1 2 3 4
450
500
550
600
650
700 50s
800s
700s
500s
300s
100s
Tem
pera
ture
(°C
)
Axial position (m)
a)
0 1 2 3 4
0.0
0.1
0.2100s
O2 g
as fr
actio
n (-
)
Axial position (m)
50s
b)
Figure 5.3: Axial temperature profiles in the reactor during the oxidation and heat removal step (a) and the axial O2 gas fraction profiles (b).
92 |
In actual operation, the bed will experience alternating cycles and thus two main aspects will change compared with the single cycles described above. First, the reduction cycle will be stopped when the extent of fuel slip reaches the maximum value allowed. In this case, the bed is not fully reduced, as was assumed as starting point in the single oxidation cycle. Second, during heat removal, not all of the heat is blown out of the bed because, at a certain moment, the outlet temperature drops below the limit that a gas turbine can handle in an efficient way. Thus, the initial temperature profile for the subsequent reduction is not uniform, as was again assumed in the single cycle case, but instead an axial temperature profile prevails with a higher temperature in the bed close to the reactor outlet. To what extent these effects influence the subsequent cycles is discussed in the next section.
5.3.2. Multiple cycles In this section, the cyclic steady state is investigated after many oxidation and reduction cycles (with intermediate purge cycles). However, the cyclic steady state is typically almost attained already after only a few full cycles, as will be shown later. The reduction cycle is started when the outlet temperature has dropped to a level at which the gas turbine can no longer work efficiently. At that point, the bed has a higher temperature at the end than in the rest of the bed, which has the same temperature as the inlet air during the oxidation cycle, in this case 450 °C (see Figure 5.4a). When the reduction cycle starts, the bed temperature increases because the reduction with syngas is also exothermic. During this time, the outlet temperature declines because the heat at the end of the bed is partially blown out of the bed. When fuel slip occurs, the reduction cycle has to be stopped. To prevent fuel slip, the cycle is stopped before all of the solids have undergone a reduction, especially at the end of the bed. The heat produced during the reduction cycle is stored in the bed. It should be noted that in this simulation the reduction and oxidation cycle times were specified to simplify the discussion and analysis. In actual operation, the moment to stop the oxidation and reduction cycles is controlled by the outlet temperature (or outlet composition).
After the reduction cycle, the reactor is purged with N2 for 10 seconds to prevent that syngas and oxygen are in contact. In this step, the remaining syngas in the bed also reacts with the carrier, causing a slight temperature change at the end of the bed, shown in Figure 5.4a. The purge is carried out at a higher flow rate than during the reduction cycle and, due to the flow rate increase, some fluctuations are observed in the temperature profile at the end of the bed. It should be noted that in actual operation, the purge time would also be optimized, but this is not in the scope of the present chapter.
5. Two-stage-CLC, a novel configuration for packed bed CLC |93
5 0 1 2 3 4400
500
600
700
800Te
mpe
ratu
re (°
C)
Axial position (m)
red, initial red, t=200s red, t=400s red, end purge, end
a)
0 1 2 3 4400
500
600
700
800
900b)
Tem
pera
ture
(°C
)
Axial position (m)
ox, initial ox, t=50s ox, t=100s ox, t=350s ox, end purge, end
Figure 5.4: Axial temperature profiles during (a) reduction and (b) oxidation (including heat removal) for the base case in the cyclic steady state.
94 |
After the short purge, the oxidation cycle is started. When the reaction front develops through the bed, the bed temperature rises as well. At the outlet, a temperature increase is observed because of the heat remaining from the reduction reaction. Then the outlet temperature decreases, until the moment that the reaction front arrives at the end of the bed (i.e. the moment that almost the entire bed has been oxidized again, in Figure 5.4b this is at toxidation≈100s). From that moment, hot air is blown out of the bed at a constant temperature (this is the heat removal step). In the first instance, the air temperature is slightly higher, which is caused by the remaining heat from the previous oxidation cycle. The hot air generation process (heat removal) continues until the temperature drops below the limit set by the gas turbine. After a short purge, the bed is reduced again.
The outlet temperature profiles of the first four reduction and oxidation cycles are depicted in Figure 5.5. After the third cycle, repeated behavior can be observed, which is illustrated in more detail in Figure 5.6. It is shown that during the reduction, the outlet temperature drops slightly because the remaining heat from the previous cycle is partially blown out of the bed; this behavior can also be observed in the axial profiles in Figure 5.4a. After starting the oxidation cycle, the outlet temperature has some small fluctuations until the moment that the reaction front reaches the end of the bed (end of oxidation phase). After that moment, the heat removal phase starts, when hot air is produced. The heat removal phase is the largest phase of the oxidation cycle.
In Figure 5.6, it is shown that hot air is produced with a constant temperature, which is required for the downstream power generation unit. It should also be noted that the flow rate during oxidation is much higher than during reduction. For that reason, 85% of the generated energy during the combustion reactions is converted into air at a temperature above 800 °C, which gives an indication of the energy efficiency of the process.
Simulations with other oxygen carriers lead to very similar profiles with different temperatures due to a difference in reaction enthalpy and active weight content (resulting in another wh/wr-ratio). As shown in section 3.2.2, the reaction enthalpy of the reduction of nickel, iron or manganese with H2 and CO is less exothermic. For that reason, the only difference is the profile at the beginning of the oxidation cycle (before the bed is oxidized and hot air is produced).
As is shown in the gas outlet temperature profile (Figure 5.6), not all the heat that is released due to the exothermic reactions can be effectively used for hot air production. In the next section, the effect of the cycle transition method is discussed to evaluate if the process efficiency can be increased by effectively using this remaining heat.
5. Two-stage-CLC, a novel configuration for packed bed CLC |95
5 0 2000 4000 6000400
500
600
700
800
900cycle 4cycle 3cycle 2
Out
let t
empe
ratu
re (°
C)
Time (s)
cycle 1
Figure 5.5: Evolution of the outlet temperature in time after repeated reduction and oxidation (and heat removal) cycles for the base case.
0 200 400 600 800 1000 1200 1400400
500
600
700
800
900
Tem
pera
ture
(°C
)
Time (s)
oxidationreduction
cycle 1
cycle 2
cycle 3...n
Figure 5.6: Outlet temperature as a function of the time after starting a full cycle for repeated reduction and oxidation (including heat removal) cycles for the base case. Between t = 750 s and t = 1200 s, air with a constant temperature is produced, that can be used for power generation in a gas turbine.
96 |
5.3.3. Cycle transition method The remaining energy from the oxidation cycle could be used efficiently by reversing the flow direction during reduction; in that case the syngas is fed counter-currently (thus from the hottest part of the reactor). In this situation, heat at the hottest part of the reactor (z=L) is blown back into the bed instead of blown out of the reactor. For that reason, this option is more energy efficient. A comparison in terms of temperature profiles with the base case is shown in Figure 5.7.
During the reduction, CO2 and H2O are produced at the temperature at z=0 (equal to the air inlet temperature, 450 °C). After switching to the oxidation and heat removal step (and changing the flow direction), all the remaining heat from the previous cycle that has accumulated at z=L is blown out of the bed. From the peak at the outlet temperature profile, it can be concluded that the remaining heat was not distributed well over the bed. During the reduction reaction, two heat transport processes take place. First, a heat front moves through the bed and, second, heat dispersion occurs. However, those processes occur quite slowly resulting in a misdistributed temperature profile.
In the case of countercurrent flows, a larger temperature peak is observed in the outlet temperature profile. Although more heat can be used, this larger fluctuation in temperature is probably not easily accommodated in the downstream power generation unit. This will also affect the properties of (and requirements for) the oxygen carrier. For these reasons the countercurrent operation is not preferred. Additionally, the countercurrent operation also implies the use of high temperature valves at both ends of the reactor, which makes this operation also less attractive from an economic point of view. An alternative solution to equalize the temperature profiles in the bed, thus utilizing the remaining heat in an effective way, is to fluidize the oxygen carrier for a short period by steam as proposed by Noorman et al. (2007). However, by fluidizing the oxygen carrier, the mechanical stability becomes more of an issue and fines could be created, which could give rise to increased pressure drop during the oxidation cycle.
5. Two-stage-CLC, a novel configuration for packed bed CLC |97
5 0 200 400 600 800 1000 1200 1400400
500
600
700
800
900
1000
1100O
utle
t tem
pera
ture
(°C
)
Time (s)
base case reversed flow
oxidationreduction
Figure 5.7: Comparison between the reversed flow option and the base case.
5.3.4. Process design parameters In addition to the flow direction switching scheme, several other parameters can be tuned to optimize the power generation efficiency. In this section, two main options are discussed that influence the process efficiency.
The length of the oxidation/heat removal time can be extended. The effect is depicted in Figure 5.8 where the oxidation cycle is extended by 100 seconds. In this case, more heat is blown out of the bed during the heat removal step, resulting in a lower outlet temperature of CO2 and H2O during reduction. The period of time at which the hot air is produced at >850 °C does not change. The difference between the two cases is the temperature of the CO2/H2O stream and the air at lower temperature. System integration studies have shown that the highest overall electrical efficiency is reached, if the heat removal process is directly stopped when the temperature of the hot gas starts to drop.
98 |
0 200 400 600 800 1000 1200 1400
500
600
700
800
oxidationO
utle
t tem
pera
ture
(°C
)
Time (s)
base case tox=830s
reduction
Figure 5.8: The effect of increasing the oxidation time by 100 seconds. (tox = 730s and 830s).
0 200 400 600 800 1000 1200 1400450
500
550
600
650
700
750
800
850oxidation
Out
let t
empe
ratu
re (°
C)
Time (s)
base case Tred=100°C
reduction
Figure 5.9: The effect of the syngas temperature (Tred=450 °C and 100 °C).
5. Two-stage-CLC, a novel configuration for packed bed CLC |99
5
Another aspect is that the syngas obtained from the desulfurization plant is at low temperature. Therefore the influence of the syngas temperature has been studied. Feeding syngas at a lower temperature (for example 100 °C) will not have much influence on the conversion and temperature profiles, because the syngas flow rate is much smaller than the air flow rate and enough heat is already available in the bed to heat up the feed. On the other hand, feeding syngas at a low temperature avoids the use of heat exchangers (i.e. to heat up syngas to 450 °C) and thus increases the efficiency of the entire power plant. As shown in Figure 5.9, hot air is produced at the same temperature as in the base case. However, the time in which the hot air is produced is slightly lower; in the example of Figure 5.9 hot air above 800 °C is produced for 585 seconds, compared to 610 seconds in the base case. The reason for this is that the bed is cooled down more because of the lower syngas temperature. The effect of the fuel feeding temperature is discussed in detail in section 7.2.1.
To confirm and quantify the advantages in terms of plant efficiency of the different options, a detailed energy analysis of the entire IGCC plant with CLC has been carried out, discussed in chapter 6 and 7.
5.4. Novel approach: two-stage CLC Currently, a lot of research is carried out to find/develop an oxygen carrier that fulfills all the criteria for the CLC process (high oxygen capacity, high temperature rise and reactivity, high chemical and mechanical stability and high selectivity). It turns out to be quite difficult to obtain an oxygen carrier that is highly reactive at 450 °C and that is also mechanically and chemically stable at 1200 °C and 20 bar (Hossain and de Lasa, 2008). In the previous section, copper oxide was used, but this is not feasible for a single stage CLC process as its melting point is below 1200 °C. Therefore, a two-stage CLC configuration is proposed.
In this two-stage configuration, the heat that is produced in a first bed is transferred to a second bed and there the desired temperature of 1200 °C is reached. A schematic overview of the axial temperature profiles is depicted in Figure 5.10. After heat removal, the temperature in the first bed is equal to the gas inlet temperature of 450 °C and the second bed has a higher temperature (in this example 857 °C), which was previously produced in the first bed. Afterwards, reduction and oxidation reactions are carried out leading to a temperature increase in both beds. Then, the heat is removed again by blowing the hot air of 1200 °C to the gas turbine and transferring the heat from the first bed to the second bed.
100 |
0 1 2 3 4 5 6
400
600
800
1000
1200heat removalTg,out
2nd oxygen carrier
3. heat removal
3. heat removal
2. oxidation
2. oxidation
1. reduction
Tem
pera
ture
(°C
)
Axial position (m)
1. reduction
1st oxygen carrierTg,in
Figure 5.10: A schematic overview of the two-stage configuration for CLC in packed beds.
The demands on the oxygen carrier will become less stringent in the case that the process is carried out in two stages. Smaller temperature differences exist in each reactor, which also makes it less demanding for construction. For this approach, the first bed should contain an oxygen carrier that reacts fast at low temperatures (such as Cu-based oxygen carriers): CuO (with Al2O3 as support) has already been demonstrated as an appropriate oxygen carrier for packed bed CLC using methane (Noorman et al., 2011b, 2010b) or syngas (chapter 3) because of its high reactivity at low temperatures. Since the melting point of copper is lower than the maximum desired temperature for power production (1200 °C), the material cannot be used for the one stage CLC process. But this material is still applicable in the first bed for TS-CLC. For the second bed, it is important that the carrier (that might be more expensive) is stable at the more challenging process conditions. Because of the high temperature, the kinetics might not be the most important criterion. Moreover, as the temperature rise required in the second bed is not extreme, a low oxygen capacity of the oxygen carrier can also be tolerated. In this study, NiO/CaAl2O4 has been selected for the second bed, because this material has been proven to be suitable for packed bed reactors. In that case, it was used for the one stage configuration, but it can also be used for TS-CLC, because the apparent active weight content (the weight fraction of the oxygen carrier that reacts during reduction) is limited.
5. Two-stage-CLC, a novel configuration for packed bed CLC |101
5
5.5. Experimental demonstration of TS-CLC In this section, the TS-CLC configuration is demonstrated by experiments in a lab scale reactor. After the experimental demonstration with syngas as fuel, the effect of operating conditions is discussed, i.e. flow rate, operating pressure, fuel concentration and type of fuel (H2, CO, syngas or methane). In the end, the experimental results are described with a packed bed reactor model, which is also used for a prediction of the performance on industrial scale.
5.5.1. Materials and methods The Cu-based particles (13wt% CuO on Al2O3) have been obtained from Sigma-Aldrich and have an average particle size of 1.1 mm. The Ni-based particles (17wt% NiO on CaAl2O4 support) were obtained from Johnson-Matthey as fluted rings with a height and diameter of 11 mm, and have been crushed and sieved to particle sizes of 2 mm. Before the oxygen carrier was placed in the reactor, the particles were activated by exposing them to two redox cycles with reductions with H2 at 900 °C and oxidations with air. The main reactions that prevail in the reactor with these materials are listed in Table 5.2.
The reactor and the experimental setup have been described in section 4.2.1. One packed bed reactor is used containing two bed materials stacked on top of each other. The bottom section (38 cm) of the tube is filled with 13% CuO/Al2O3, while the upper section (32.5 cm) is filled with 17% NiO/CaAl2O4, as illustrated in Figure 4.2. The properties of the two beds are listed in Table 5.3.
Table 5.2: Main reactions prevailing in TS-CLC using syngas as fuel (Tref = 298 K) (Barin, 1993).
Oxidation of Cu 22 Cu O 2 CuO ΔHr = -304 kJ/mol O2
Reduction of CuO with H2 2 2CuO H Cu H O ΔHr = -95 kJ/mol H2 Reduction of CuO with CO
2CuO CO Cu CO ΔHr = -131 kJ/mol CO Oxidation of Ni
22 Ni O 2 NiO ΔHr = -479 kJ/mol O2 Reduction of NiO with H2 2 2NiO H Ni H O ΔHr = -2 kJ/mol H2 Reduction of NiO with CO
2NiO CO Ni CO ΔHr = -43 kJ/mol CO Water gas shift reaction
2 2 2CO H O H CO2H C2H2 ΔHr = -41 kJ/mol Boudouard reaction
22 CO C COC CO ΔHr = -86 kJ/mol CO Carbon combustion
2 2C O CO2CO ΔHr = -394 kJ/mol O2
102 |
Figure 5.11: Schematic overview of the placement of the bed material in the reactor.
5. Two-stage-CLC, a novel configuration for packed bed CLC |103
5
Table 5.3: Properties of the oxygen carriers used for the demonstration of TS-CLC.
Bed 1 Bed 2 Oxygen carrier 13wt% CuO on Al2O3 17wt% NiO on CaAl2O4 Bed length, m 0.380 0.325 Bed mass, kg 1.079 0.899 Solid bulk density, kg/m3 911 888 Particle diameter, mm 1.1 2 αbl (bed to liner), W/m2/K 90 αls (liner to environment), W/m2/K 5.6·10-3·Tw Environment temperature, °C 300
5.5.2. Experimental demonstration of TS-CLC with syngas as fuel Several cycles consisting of a reduction, purge, oxidation and another purge step have been performed for the demonstration of TS-CLC. The oxidation cycles consists of a 120 Ln/min air flow for 3 minutes. The reduction cycles consist of a 60 Ln/min flow with 20% syngas for 8 minutes. The ratio between CO, H2 and CO2 (2.3:1:0.4) is comparable to the syngas obtained from a gasifier in an integrated gasification combined cycle (IGCC) power plant (as is demonstrated in chapter 6). Steam was added to prevent carbon deposition and to stimulate the water gas shift reaction. The amount of steam has been adjusted such that no carbon deposition occurs above 450 °C according to thermodynamics calculations using HSC 5.1 (HSC, 2013). Between the oxidation and reduction a 2 min purge with N2 was introduced. Cycles with alternating oxidation and reduction were performed at 2 bar. An overview of the operating parameters during the cycles is given in Table 4.3.
A cycle consists of the following procedure. First, the packed bed reactor is heated by a N2 flow at 450 °C. Subsequently, five cycles are performed after which the temperature profiles during the cyclic process become in cyclic steady-state. In Figure 5.12 the outlet temperature is displayed for the first five cycles. This figure shows that when the first reduction is performed, both beds have an initial temperature of 433-485 °C and the gas is fed to the reactor at 480 °C. The heat produced during the reduction and oxidation in the first bed is transferred to the second bed and increases the temperature of the initial temperature profile of the next cycle. After five cycles the outlet temperature becomes close to its cyclic steady state, leading to a maximum outlet temperature of 839 °C (∆T=359 °C) during the oxidation. Mainly due to heat losses, this temperature is lower than the desired temperature of 1200 °C. At the end of this chapter (section 5.5.4), it is described how this temperature can be reached. The axial
104 |
temperature profiles and outlet gas fraction during the fifth cycle are shown in Figure 5.13 (reduction) and Figure 5.14 (oxidation and heat removal). The performance during these steps is discussed in the next two sections.
Table 5.4: Operating conditions of the base case used during the packed bed reactor experiments.
Reaction Flow rates, Ln/min. Pressure, bar
Inlet temperature,
°C
time, min.
Reduction with syngas CO: 8.4 H2: 3.6
CO2: 1.4 H2O: 12.6 (steam)
N2: 34.2
2 450 8
Purge N2: 120 2 450 2 Oxidation Air: 120 2 450 3 Purge N2: 120 2 450 2
0 2 4 6 8 10 12 14400
500
600
700
800
900
cycle 5
cycle 3
cycle 2
Tem
pera
ture
(°C
)
Time (s)
cycle 1
Figure 5.12: The outlet temperature profile during the first 5 cycles of the TS-CLC experiment with syngas (base case, operating conditions listed in Table 4.3).
5. Two-stage-CLC, a novel configuration for packed bed CLC |105
5
Reduction The axial temperature profiles and the outlet gas fractions during the reduction is shown in Figure 5.13. A reaction front is observed that is moving through the first bed and a temperature increase in the first bed is observed (∆Tmax = 87 °C). This temperature increase is the consequence of the water gas shift (WGS) reaction and the reduction reaction of CuO, which are both exothermic. Since the reduction enthalpy of NiO is much lower compared to CuO and due to relatively high heat losses, no temperature increase and reaction front is observed in the second bed. The temperature profiles show that the reduction takes places at an elevated temperature, which increases the reactivity of the second bed material.
The breakthrough profiles show that initially CO2 is formed, since all the fuel is oxidized to CO2 and steam, while the steam is condensed and separated before the analyzer. Breakthrough of syngas occurs after 4 min. In the breakthrough curves in Figure 5.13, a larger amount of H2 is produced after breakthrough than fed (and less CO), which indicates that the WGS reaction occurs in the reactor as expected.
It is important to remark that in industrial application, fuel slip should be avoided. If the reduction was stopped at this point, a lower degree of reduction would be reached with the oxygen carrier material, which makes it very difficult to compare the breakthrough curves with other experiments and the model simulations. Therefore, some fuel slip was tolerated during these experiments. For industrial application, the reactor scale is much larger and then the section with a lower degree of reduction is much smaller.
After a reduction of 8 min, the setup is purged by N2 for 2 min.
106 |
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
450
500
550
600
650
700
750
800
end ofreduction
initial temperatureprofile
Tem
pera
ture
(°C
)
Axial position (m)
t=0min t=2min t=4min t=8min
0 2 4 60.00
0.05
0.10
0.15
0.20
0.25
Dry
out
let f
ract
ion
(-)
Time (min)
CO2
H2
CO
Figure 5.13: The axial temperature profile and the gas fractions at the outlet during the reduction.
5. Two-stage-CLC, a novel configuration for packed bed CLC |107
5
Oxidation and heat removal After the purge, the oxygen carriers are oxidized with air. The temperature and O2 fraction at the outlet during oxidation are reported in Figure 5.14. The reaction front can be observed in Figure 5.14 leading to a temperature increase in both beds (due to the exothermic reactions). The maximum temperature in the Cu/CuO bed is 651 °C (∆Tmax=72 °C), while the maximum temperature in the Ni/NiO bed is 839 °C (∆Tmax=173 °C). The temperature rise in both beds is lower than theoretically possible with this material at adiabatic conditions, because the reactor experiences heat losses and the oxygen carriers are not fully reduced. An apparent active weight content is expected of 10wt% CuO (ΔTCumax = 199 °C) and 11.5wt% NiO (ΔTNimax=384 °C) based on TGA experiments at these conditions. A large extent of the heat that is released during the oxidation is taken up by the liner that surrounds the reactor bed and has a relatively high heat capacity. As a consequence of the heat transport from the reactor bed to the liner and the response delay of the thermocouple, the theoretical maximum temperature is never measured and a lower temperature rise is reached.
In the second bed a higher temperature rise is measured at the exit than in the rest of the bed, because during the previous reduction, the temperature was higher and thus a higher degree of reduction was reached.
In the O2 breakthrough profile in Figure 5.14, an oxygen fraction of 1% is initially measured. This is not related to the experiment itself, but to a small leakage in the analyzer (that is operating below atmospheric pressure). This has been taken into account when analyzing the results. After 1 min, both beds are fully oxidized and oxygen starts to break through.
The heat removal which is required in order to transfer the heat from the first bed to the second bed and to remove the heat from the second bed, is integrated with the oxidation (after 1 min) and also carried out during the subsequent purge. The temperature profile during the heat removal is illustrated in Figure 5.15. These profiles are obviously also influenced by the heat losses occurring in the reactor. It is important to remark that the heat from the first bed is not fully transferred to the second bed, because the heat transfer continues during the subsequent reduction and purge step. The time needed to transfer the heat from the first bed to the second bed should equal the total cycle time (the duration of a complete cycle, containing reductions, oxidations and purges). If the total cycle time exceeds the time that is required to transfer the heat from the first bed to the second bed, the process becomes rather a one-stage CLC process.
108 |
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7400
500
600
700
800
900Te
mpe
ratu
re (°
C)
Axial position (m)
t=0min t=1min t=2min t=3min
initial temperatureprofile
reactionfront
0.0 0.5 1.0 1.5 2.0 2.50.00
0.05
0.10
0.15
0.20
Dry
out
let f
ract
ion
(-)
Time (min)
O2
Figure 5.14: The axial temperature profile and the O2 fraction at the outlet during oxidation.
5. Two-stage-CLC, a novel configuration for packed bed CLC |109
5 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7400
500
600
700
800
900
t=5 min
Tem
pera
ture
(°C
)
Axial position (m)
t=2.5 min
heat removal
heat removal
Figure 5.15: Axial temperature profile during heat removal (oxidation after 1 min and subsequent purge).
5.5.3. Effect of operating conditions The influence of the operating conditions has been investigated with respect to the operability of the TS-CLC process. The studied operation conditions are pressure, reduction flow rate, fuel concentration and fuel type. In the base case, the oxygen carriers have been reduced with H2 (20% H2 balanced with N2) so that no competitive reactions (like WGS reaction) can be observed during reduction. An overview of the operating conditions for the base case is given in Table 5.5. Two variables are used in order to express the performance of the process: maximum temperature increase (∆Tmax) in the beds and the maximum temperature (Tmax) during the oxidation. For all the experiments, the data of the fifth cycle have been used, since the cyclic procedure became steady-state after five cycles.
110 |
Table 5.5: Operating conditions of the base case used during the experiments to investigate the effect of the pressure, reduction flow rate, fuel concentration and fuel type.
Reaction Flow rates, Ln/min. Pressure, bar Inlet temperature, °C
time, min.
Reduction with H2 H2: 12 N2: 48
2 450 8
Purge N2: 120 2 450 2 Oxidation Air: 120 2 450 3 Purge N2: 120 2 450 2
Operating pressure The pressure effect has been measured by operating at 2 and 7 bar. The results are displayed in Table 5.6. The reduction is performed with a 60 Ln/min flow containing 20% H2 (balanced with N2).
When the operating pressure is increased from 2 to 7 bar, the temperature increase during oxidation becomes slightly higher in the Cu/CuO bed and slightly lower in the Ni/NiO bed. These effects cancel each other out, but the maximum temperature reached at the end of the second bed is about 25 °C lower. This is related to very small differences in the initial axial temperature profile at the start of the oxidation cycle shown in Figure 5.16a. The reason for this small difference is not clear, but it cannot be attributed to a difference in the extent of reduction, because the temperature rise and the oxygen consumption do not change significantly. The independency of the oxygen consumption on the pressure is in line with the kinetic study about the pressure effect that has been discussed in chapter 2. This demonstrates that increasing the operating pressure does not affect the TS-CLC performance. Hence, TS-CLC can also be carried out at 20 bar, at which the highest possible process efficiency can be reached (Spallina et al., 2014).
Table 5.6: The operating pressure effect on the temperature differences in the reactor sections and the maximum temperature.
Operating pressure, bar ΔTCumax, °C ΔTNimax, °C Tmax, °C 2* 80 218 865 7 90 212 840 *= base case from Table 5.5.
5. Two-stage-CLC, a novel configuration for packed bed CLC |111
5
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7400
500
600
700a)Te
mpe
ratu
re (°
C)
Axial position (m)
2 bar
7 bar
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7400
500
600
700
Tem
pera
ture
(°C
)
Axial position (m)
40% H2
20% H2
b)
Figure 5.16: Initial oxidation temperature profile followed after reductions with different operating pressures (a) and different fuel fractions (b).
Fuel concentration The effect of the fuel concentration has been investigated by comparing cycles in which the H2 fraction during the reduction cycle is varied. The reduction time has been adjusted to partly compensate for the amount of H2 that is fed. The reduction is performed with a concentration of 20% H2 (for 8 min) and 40% H2 (for 5 min) with a 60 Ln/min reduction flow at 2 bar.
The results are given in Table 5.7, showing the temperature rise during oxidation and the maximum temperature at the thermocouple. The maximum temperature rise during the oxidation is similar for reductions with different concentrations and also the oxygen consumption is the same. Despite similar temperature rises, the maximum temperature in the case of 40% H2 is substantially higher. A higher temperature is reached, because the temperature is initially higher as shown in Figure 5.16b. This is the consequence of two effects: due to the longer reduction time for the 20% H2 reduction, the heat front moves further into the reactor bed and also more time is available for heat losses. Hence, the difference in maximum temperature is not an effect related to the operating parameter, but to the cycle time and this can be adjusted. This implies that the TS-CLC process can be operated with a wide range of fuel concentrations as long as the heat removal time is properly adjusted, so that the heat produced in the first reactor bed is exactly transferred to the second reactor bed.
112 |
Table 5.7: The fuel concentration effect on the temperature differences in the reactor sections and the maximum temperature.
H2 fraction Reduction time, min. ΔTCumax, °C ΔTNimax, °C Tmax, °C 20%* 8 80 218 865 40% 5 74 220 931 *= base case from Table 5.5.
Flow rate during reduction The effect of the reduction flow rate is studied by performing the reduction with 60 Ln/min (8 min) and 120 Ln/min (5 min). The reduction time is increased in order to make sure that similar amounts of H2 are fed to the packed bed reactor. In both cases, the reduction is performed with 20% H2 at 2 bar.
The temperature increase in both reactor beds and the maximum temperature are shown in Table 5.8, which shows that the influence of the flow rate on the maximum temperature and maximum temperature increase is negligible. So, apparently the axial temperature profiles and the degree of reduction is not influenced by the flow rate (which is also to be expected) and the heat front has moved similarly in both cases, because when the flow rate is reduced also the velocity of the heat front decreases proportionally. In this case, the reduction time was also increased to compensate for this effect. Hence, the temperature profiles obtained from both cases can hardly be distinguished (and are not reported here).
Table 5.8: The flow rate effect on the temperature differences in the reactor sections and the maximum temperature.
Flow rate during reduction, Ln/min.
Reduction time, min.
ΔTCumax, °C
ΔTNimax, °C
Tmax, °C
60* 8 80 218 865 120 5 73 223 872 *= base case from Table 5.5.
5. Two-stage-CLC, a novel configuration for packed bed CLC |113
5
Fuel type The influence of the fuel type is investigated for four fuel types: H2, CO, syngas, and methane. The exact compositions of the feed stream during the reduction cycle are listed in Table 5.9. Steam (in case of syngas and methane) or CO2 (in case of CO) have been mixed with the fuel to prevent carbon deposition. The reductions are performed with a flow rate of 60 Ln/min for 8 min at 2 bar.
From Table 5.9 it can be concluded that the temperature increase depends on the type of fuel used in the reduction. Reduction with H2 leads in the Ni/NiO bed to a higher temperature increase than reduction with CO. Apparently, the material is more reduced by H2 compared to CO. Since syngas is a mixture of H2 and CO, the temperature rise (and thus the degree of reduction) is in between the ones for H2 and CO. For the Cu/CuO bed, the situation is the other way around. It is known from chapter 3 that Cu/CuO can be reduced further with CO than with H2, if the reduction is performed for more than 2 min. This could explain the higher temperature increase in the first bed for CO compared to H2 and syngas.
The same conclusion can be drawn from Figure 5.17, where the oxygen outlet fraction is displayed as a function of time. The area under the curve is equal to the amount of oxygen that is not consumed by the oxygen carrier. From the figure it can be deduced that reduction with H2 has the highest oxygen consumption in the subsequent oxidation cycle (which leads to a higher temperature rise), followed by syngas, CO and CH4.
Table 5.9: The influence of the fuel on the temperature differences in the reactor sections and the maximum temperature.
Fuel Composition* ΔTCumax, °C
ΔTNimax, °C
Tmax, °C
H2** 20% H2 80 218 865 Syngas*** 14% CO, 6% H2, 2% CO2, 21% H2O 91 183 839 CO 20% CO, 80% CO2 99 160 774 CH4 10% CH4, 15% H2O 27 19 507 *=balanced with N2, **= base case from Table 5.5, ***=operating conditions listed in Table 5.4.
114 |
0 1 2 3 4 50.00
0.05
0.10
0.15
0.20
syngas
H2
O2 d
ry g
as fr
actio
n at
out
let (
-)
Time (min.)
CH4 CO
Figure 5.17: The oxygen consumption during the oxidation after reductions with different fuels.
In the cases that H2 and syngas is used as fuel, a higher temperature rise is observed and therefore a higher maximum temperature is observed in the reactor; the difference in temperature increase is similar to the difference in maximum temperature. The temperature increase for CO is slightly lower compared to H2 and syngas, but the maximum temperature is much lower. This can be explained by the oxidation temperature profile for CO, which is illustrated in Figure 5.18. This figure shows that the temperature rise during oxidation is lower at the exit of the reactor than in the middle of the Ni/NiO-bed. From this, it can be deduced that the end of the NiO/Ni-bed is less reduced compared to the rest of the bed. It is expected that this is related to the slower kinetics for NiO reduction with CO. If the reduction time was increased, a higher solid conversion would have been reached at the end of the NiO bed and then a higher temperature rise would have been observed in the subsequent oxidation cycle. In that case, more fuel slip would have occurred and the heat front would have moved further into the bed, which makes it more difficult to compare the cases with the different fuels. For that reason, the reduction time was not increased in the experiment. Thus the concept is able to deal with only CO as fuel, after fine tuning of the flow rate during reduction in order to avoid fuel slip, while the reactor bed is used as much as possible.
5. Two-stage-CLC, a novel configuration for packed bed CLC |115
5 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
450
500
550
600
650
700
750
800Te
mpe
ratu
re (°
C)
Axial position (m)
t=0 min. t=2 min. t=3 min.
Figure 5.18: Axial temperature profile during oxidation after reduction with CO.
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7440
450
460
470
480
490
500
510
Tem
pera
ture
(°C
)
Axial position (m)
t=0 min. t=1 min. t=2 min.
Figure 5.19: Axial temperature profile during oxidation after reduction with CH4.
116 |
The reactivity with respect to CH4 is very low. The temperature increase and subsequent maximum temperature is substantially lower compared to the other fuels. Two different mechanisms for the reduction of the bed materials by CH4 are proposed. CH4 could directly reduce the oxygen carrier forming H2O and CO2. Another possibility is that CH4 is combusted indirectly by first the steam methane reforming reaction (CH4 with steam is converted into syngas). Afterwards, the oxygen carrier is reduced by the produced syngas, resulting in H2O and CO2. A condition for the second mechanism is that a catalyst is available for the steam methane reforming, like metallic nickel.
However, a very low temperature rise is observed during the oxidation after the reduction with CH4 as illustrated in Figure 5.19. This means that quite a low solid conversion is obtained during the previous reduction. In the second bed, the temperature even drops, which indicates that the material was hardly reduced by CH4. If some CH4 was converted into syngas, the syngas would have reacted with the oxygen carriers as happened during the syngas experiments. Due to the low solid conversion in the first bed, little heat is produced there during oxidation. As a consequence not much heat can be transferred to the second bed and therefore this bed is also at low temperature, at around 500 °C. At this temperature the NiO/CaAl2O4 is not reactive for the direct reduction with CH4 nor for the steam methane reforming, because the nickel is not available in metallic form (Medrano et al., 2014). Because of the low reactivity of CuO/Al2O3 with CH4 at 450 °C, TS-CLC cannot be demonstrated with methane in this case. However, if an oxygen carrier is found that is reactive with methane at these operating conditions, or when mixing the oxygen carrier with a catalyst active at this low temperature, this concept would also become very interesting for methane as fuel.
The operating parameter study shows that in all the cases, the TS-CLC is feasible as long as the different process steps are tuned to each other so that the heat produced in the first reactor bed is effectively transferred to the second bed. Also, increasing the pressure (to reach a higher overall process efficiency) does not affect the performance of the TS-CLC process.
5.5.4. Description by packed bed reactor model In order to investigate the effect of operating parameters and heat losses, a packed bed reactor model is used to describe the axial temperature and concentration profiles inside the reactor. A description of the model has been given in section 4.2.2. The experimental results with syngas as fuel are compared with the results from the
5. Two-stage-CLC, a novel configuration for packed bed CLC |117
5
simulation. Afterwards, the model is used for scaling up the reactor to study the performance of the TS-CLC on industrial scale.
The kinetics for the gas/solid reactions with copper (oxidation and reduction with H2 and CO) have been taken from Garcia-Labiano (García-Labiano et al., 2004) with the pressure factor as published in chapter 2. During TGA tests with CuO/Al2O3 it has been observed that only about 75% of the material reacts at 450 °C and therefore the apparent active weight content has been adjusted to 10 wt%.
For the reactions with NiO/CaAl2O4, the reduction kinetics depend on the maximum temperature reached during the oxidation in the previous cycle. For that reason, different parameters have been formulated for the kinetics, as shown in Table 2.2. The kinetics for the water gas shift reaction have been taken from Grenoble et al. (Grenoble et al., 1981). Carbon deposition was not included in the model, because it was not observed during the experiments: in fact, no CO2 was produced during oxidation in any of the experiments.
As discussed chapter 4, the thermocouple has a certain response time. In that chapter it was demonstrated that a fair comparison between the model and the experimental data could be achieved, if the temperatures obtained from the model are corrected for a thermocouple response delay of 45 seconds.
Table 5.10: Applied kinetic parameters for the NiO/CaAl2O4 particles. Most of the kinetics are the same as published by Medrano et al. (Medrano et al., 2014), but an adjustment was made for the high temperature reached during oxidation.
H2 CO O2 cs, mol m-3 89960 89960 151200 r0, m 3.13·10-8 3.13·10-8 5.8·10-7 k0, mol1-n m3n-2 s-1 barq 3.6·10-4 1.4·10-3 1.2·10-3 EA, kJ mol-1 30 45 7 n 0.6 0.65 0.9 Ds,0, mol1-n m3n-1 s-1 5.1·10-5 2.2·105 1 EDs, kJ mol-1 150 300 0 kx 5 15 0 b, mol solid/mol gas 1 1 2 q 0.75 0.85 1.05
118 |
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
450
500
550
600
650
700
750
800 t=0min t=1min t=2min t=4min t=8min ox,ini
Tem
pera
ture
(°C
)
Axial position (m)
0.0 0.5 1.0 1.5 2.0 2.50.00
0.05
0.10
0.15
0.20
Dry
out
let f
ract
ion
(-)
Time (min)
O2
Figure 5.20: Axial temperature profile and outlet fraction during the reduction cycle. The operating conditions are listed in Table 5.4. The experimental data are represented by markers and the modeling data by lines.
5. Two-stage-CLC, a novel configuration for packed bed CLC |119
5
The experiment with syngas discussed in section 5.5.2 has been described by the model. The axial temperature profiles during reduction with syngas are shown in Figure 5.20. The experimental data are represented by the markers, while the simulation output is shown by the lines. Two reactions simultaneously occur in the first bed, namely the Water Gas Shift (WGS) reaction and the reduction by CO and H2. In the copper bed, heat is mainly produced at the reactor entrance, because of the reduction with CO and the WGS reaction. The WGS reaction results in the production of H2 and therefore the rest of the bed is mainly reduced by H2, which is less exothermic. The reduction of NiO with H2 is almost neutral and therefore only the moving of the heat front can be discerned. The model describes the temperature during the experiment quite reasonably, except in the copper bed during the first minutes of the reduction. The time required to switch gasses and the residence time between the mass flow controllers and the reactor (oven) leads to a short delay, which is not included in the model.
The outlet dry gas fraction is compared with the results from the model and shown in Figure 5.20. At the beginning of the reduction cycle, CO and H2 are fully converted into CO2 and steam. The expected dry gas CO2 fraction is 22%. This is in accordance with the model, but the CO2 fraction in the experiment is slightly lower, which can be related to the accuracy of the mass spectrometer calibration. After 4 minutes, H2 and CO starts to break through in both the model and the experiment. A similar break through slope points at a sufficiently accurate description of the kinetics in the model.
The axial temperature profiles during the oxidation are shown in Figure 5.21. The initial temperature profile in the copper bed is described well, but the temperature at the exit of the reactor is somewhat over predicted. As also observed during the reduction cycle, the temperature is over predicted in the first minute, probably due to the residence time of the gas in the oven (that is upstream of the reactor), which is not accounted for in the model. After two minutes, the temperature profile inside the reactor is described well. The O2 breakthrough profile (shown in Figure 5.21) is also described well by the model (the difference at the beginning is again due to the small leakage in the analyzer that was not included in the model).
120 |
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7400
500
600
700
800
900
t=0min t=1min t=2min t=3min red,ini
Tem
pera
ture
(°C
)
Axial position (m)
0.0 0.5 1.0 1.5 2.0 2.5 3.00.00
0.05
0.10
0.15
0.20
Dry
out
let f
ract
ion
(-)
Time (min)
O2
Figure 5.21: Axial temperature profile and the O2 fraction at the outlet during oxidation (operating conditions from Table 5.4). The experimental data are represented by markers and the modeling data by lines.
5. Two-stage-CLC, a novel configuration for packed bed CLC |121
5
TS-CLC for industrial applications During the experiments it was shown that the heat produced in the first reactor bed was transferred to the second bed so that a higher exit temperature was obtained. So, by these experiments it was proven that the TS-CLC concept works. However, the temperature of the exit stream was not constant and not sufficiently high so that the stream could be fed to a gas turbine for power production. Due to the small diameter and length of the lab scale reactor, the effect of heat losses and axial dispersion is quite significant. Hence, it is not possible to directly describe how the process would work at industrial scale. This is studied by scaling up the reactor using the model that has been validated in the previous section. For this scaling up, the following parameters were changed. The heat losses were excluded in the model (which correspond to the situation when the reactor has a diameter of at least 1 m, as discussed in section 4.3.1) and the length of the reactor was increased by a factor 5 (to 3.53 m) to reduce the effect of axial heat dispersion. Furthermore, the mass flow rate was increased by a factor 5, so that the same cycle times can be used.
Figure 5.22 shows the outlet temperature profiles during a cycle for the prediction on lab scale and on industrial scale. The maximum temperature is reached just after the oxidation (t=11.5 min), but afterwards the temperature decreases due to axial heat dispersion inside the reactor. The simulation on a larger scale shows that a higher temperature can be reached, due to the absence of heat losses. In this case, a temperature peak is observed when the bed is fully oxidized, but after the peak a hot gas stream is produced at constant temperature. The initial temperature peak is caused by two reasons. First, during the previous reduction the temperature at the exit of the bed was higher and then a higher degree of reduction was reached; this results in a higher temperature rise during the subsequent oxidation. Second, when the oxidation starts, the temperature is still somewhat higher at the exit of the reactor, resulting in a higher final temperature. The temperature peak can be avoided with some fine-tuning of the operating conditions (e.g. cycle time) and the maximum temperature can be increased by modification of the oxygen carriers (weight content of the active species).
122 |
0 2 4 6 8 10 12 14 16600
700
800
900
1000
1100
industrialscale
oxidation purgepurgeO
utle
t tem
pera
ture
(°C
)
Time (s)
reduction
lab scale
Figure 5.22: The outlet temperature profile from the lab scale reactor compared with a prediction of the temperature on industrial scale.
In this case, the maximum temperature is below 1200 °C, but this can be increased by increasing the apparent active weight contents. The CuO content on the Cu-based OC could be increased or the support material could be replaced, so that a larger amount of oxygen reacts during reduction at low temperature. An apparent active weight content of 13wt% CuO (instead of 10 wt%) would be sufficient to increase the final temperature by 120 °C. In the second reactor bed, a non-ideal oxygen carrier has been used, because of the price and the toxicity of nickel. This oxygen carrier might have to be replaced by another material, like a Mn-based material. To avoid interaction between the manganese and the support material (that would reduce the reactivity and stability), it could be supported on ZrO2 (Adánez et al., 2004) or MgZrO2 (Johansson et al., 2006; Zafar et al., 2007). A Mn3O4/MgZrO2 (with an active weight content of 30wt%) might be a good candidate for this reactor bed. After some optimization of the oxygen carriers, a working process configuration can be achieved, in which hot gas is produced at 1200 °C. What efficiency can be obtained in that case will be discussed in the next chapter.
5. Two-stage-CLC, a novel configuration for packed bed CLC |123
5
5.6. Conclusions This chapter has investigated the use of syngas as a fuel in chemical looping combustion carried out in dynamically operated packed beds. By simulating the CLC operation with a 1D reactor model, the evolution of the axial concentration and temperature profiles in the reactor have been investigated. It was shown that after five cycles with repeated oxidation and reduction reactions, a cyclic steady state situation is reached. In this stage, hot air is produced at a constant temperature. The process efficiency can be improved by feeding the syngas from the counter-current direction. However, heat remaining from the previous cycle is accumulated at one position (hot spot formation). This results in a temperature peak in the outlet temperature profile, which is not beneficial for the gas turbine performance.
It has been proven that the CLC process can also be carried out in a two-stage configuration. This concept has the advantage that the oxygen carrier selection becomes less critical, because smaller temperature differences prevail in one reactor. A high pressure/high temperature packed bed reactor is used to show that it is possible to carry out CLC with this reactor configuration, which means that the heat of oxidation produced in the first bed is effectively transferred to the second bed. A final temperature of 839 °C was observed, while the gases were fed at 450 °C. The effect of the operating conditions such as the pressure, throughput, fuel concentration and fuel type was studied in the setup. Small effects of the pressure, fuel concentration and throughput during reduction on the maximum temperature increase have been observed as long as differences in flow rates (which is related to heat front rates) are compensated for fine-tuning the cycle time. The fuel type has a significant effect on the performance, because the degree to which the oxygen carriers are reduced depends on the fuel type. The degree of reduction determines the temperature rise that can be reached during the subsequent oxidation. It was not possible to reduce the CuO/Al2O3 with CH4 at 450 °C, but a working process was demonstrated for syngas as fuel.
The experiments with syngas have been well described by a 1D pseudo-homogeneous packed bed reactor model. A prediction of the performance on a larger scale (with higher reactor length and diameter) has been made by the same model, showing that a gas stream with a temperature between 1000 and 1070 °C could be produced in case heat losses and heat dispersion effects are diminished. This maximum temperature can further be increased to the desired level of 1200 °C after optimization of the cycle time and the oxygen carrier materials.
|125
6 Energy evaluation of
TS-CLC configurations for the
IGCC power plant
This chapter is based on the following paper: Hamers, H.P., Romano, M., Spallina, V., Chiesa, P., Gallucci, F., Van Sint Annaland, M., 2014. Energy analysis and economic evaluation of two-stage packed-bed CLC configurations for an IGCC power plant. Submitted to Energy.
126 |
6. Energy evaluation of TS-CLC configurations for the IGCC power plant
Abstract In this chapter, the integration of the TS-CLC configuration (that has been developed in the previous chapter) in a complete power plant based on coal gasification has been investigated. An extensive energy analysis and an economic evaluation of this integrated gasification CLC (IGCLC) plant has been carried out and the results have been compared with the one stage CLC configuration. Two alternative configurations for TS-CLC have been considered: one with two reactors always operated in series (TS-CLC series) and a novel configuration with the reduction carried out in parallel and the other steps in series (TS-CLC parallel). In case of TS-CLC parallel, the footprint is somewhat smaller. The LHV efficiency of TS-CLC is slightly lower (40.3% for TS-CLC series and 40.8% for TS-CLC parallel) than the one stage CLC efficiency (41.1%). Thus, with the TS-CLC still a high efficiency can be achieved, while it gives much more flexibility in the selection of the oxygen carrier and the reactor design. Therefore, TS-CLC can be considered as a good alternative for the one stage CLC in packed bed reactors. In addition, it has been demonstrated that the total reactor volume is larger for the TS-CLC configuration, because the amount of available oxygen per reactor volume is smaller, which also leads to a larger footprint of the CLC section. However, the initial investment costs have been estimated to be a factor two lower, which is mainly related to the fact that much cheaper oxygen carriers can be used.
6. Energy evaluation of TS-CLC configurations for the IGCC power plant |127
6
6.1. Introduction A new packed bed reactor configuration has been developed and demonstrated in chapter 5, the two stage CLC (TS-CLC). In this chapter, the newly developed TS-CLC configuration is compared with a one stage packed bed CLC integrated in a complete IGCLC plant. The comparison is made on the basis of the electrical efficiency and the economics to ensure that the flexibility in terms of oxygen carrier selection of the TS-CLC is not paid by a much lower overall efficiency or higher investment costs.
In the next section, the power plant configuration and modeling assumptions are described. Subsequently, the CLC reactor design for both cases is discussed and compared, and a preliminary estimation is made of the initial investment costs for both configurations. Afterwards, the process efficiencies of the different packed bed CLC configurations are evaluated.
6.2. Method and assumptions For the case of the one stage CLC, a 19 wt% NiO/Al2O3 was selected as oxygen carrier, because this is currently the only feasible carrier in a packed bed, if the reduction is carried out at low temperature, as has been discussed in chapter 5. In the TS-CLC cases, the first reactor contains CuO/Al2O3 and the second reactor Mn3O4/Al2O3.
6.2.1. IGCLC power plant description The IGCLC power plant has been simulated with the GS software developed at the Department of Energy of Politecnico di Milano (GECOS, 2013). It is a software in which complex power plants are reproduced by assembling basic modules. The main feature of the code is the use of built-in correlations to estimate the efficiency of the turbomachines. In particular, steam and gas turbines are calculated based on a stage-by-stage approach (Chiesa and Macchi, 2004; Lozza, 1990). Gas turbine calculation includes routines to estimate the cooling flows needed in each stage and their effect on the stage efficiency. The thermodynamic properties of gases are based on NASA polynomials, while the water/steam properties are taken from Schmidt (Schmidt, 1982). The temperature of the produced streams in the CLC process depends on how the heat fronts develop through the CLC reactors during operation. To achieve an accurate evaluation of this temperature, a 1D model was used to simulate the packed bed reactors for CLC (Noorman et al., 2007). The packed bed reactors have been sized and the number of reactors is calculated based on the selected total cycle time (total time for all the operation steps) and pressure drop.
128 |
A simplified scheme of the IGCLC power plant is shown in Figure 6.1. A detailed power plant scheme, extended description, detailed mass balance and an overview of the assumptions can be found in Spallina et al. (Spallina et al., 2014). In this work, a global description is given of the gasification section, while the CLC section is discussed in more detail.
The composition of coal is based on bituminous South African Douglas Premium Coal with 8 wt% moisture content (EBTF et al., 2011), which was published as reference for power plant calculations by the European Benchmarking Taskforce (EBTF et al., 2011). First, the coal is pulverized and dried to a moisture content of 2 wt%. Then the coal is pressurized to 44 bar by CO2 in the lock hoppers and this is fed to a Shell gasifier. The oxidants for the gasification is supplied by steam and oxygen from the air separation unit (ASU). After the gasifier, the outlet stream is cooled down and high and intermediate pressure steam is produced. Subsequently, the syngas is treated and desulfurized by Selexol. Afterwards, the pressure is reduced to 21.6 bar and the syngas is fed to the saturator to increase the humidity and the temperature. The syngas leaves the saturator at 151 °C and it is heated up to 300 °C by saturated water withdrawn from the high pressure drum. Then 33.4 kg/s steam (stream #17) is mixed to avoid carbon deposition inside the CLC reactor. The amount of steam is adjusted so that no carbon deposition can thermodynamically occur at 20 bar above 450 °C. The dilution with steam is useful to avoid both coke deposition in the CLC bed and metal dusting in the high temperature heat exchanger.
Before the syngas is fed to the CLC reactor, it is further heated up to 600 °C in a gas-gas heat exchanger with the CO2/H2O stream from the CLC unit. The syngas is extra preheated, because better temperature profiles are obtained in the packed bed reactors. The syngas (#1) is fed to the CLC reduction reactors with the following composition: 32.1% CO, 5.3% CO2, 13.1% H2, 48.3% H2O, 0.7% N2 and 0.6% Ar. In the CLC reactors, a mixture of CO2 and H2O (#2) is produced at a temperature dependent on the configuration of the CLC system. During the cooling of this stream, heat is recovered by generating superheated steam at 565 °C (#13). The temperature of the CO2/H2O is fluctuating (in time) and therefore measures to avoid too large temperature fluctuations of the hot heat exchanger surface are needed to avoid excessive thermal fatigue of the tubes material. For example, proper arrangement of the heat transfer banks, controlled mixing with recycled cooler CO2 or buffering with inert material in a fixed or fluidized bed vessel could be considered to reduce the stress for the heat exchangers. Afterwards, heat is recovered by high pressure water which is heated up to the saturation temperature. Some low pressure steam is formed as well and some low temperature heat is used to increase the temperature of part of the water
6. Energy evaluation of TS-CLC configurations for the IGCC power plant |129
6
from the condenser. Afterwards, the CO2/H2O stream is further cooled to 35 °C with cooling water, condensate is separated and the CO2 is intercooled compressed and pumped to 110 bar (#4). The electricity consumption for the CO2 compression has been calculated in Aspen Plus (Aspen Technology Inc., 2011).
The oxygen carrier is oxidized by atmospheric air (#5, at 15 °C), pressurized to 20 bar and 438 °C (#6). This air is used for the oxidation reaction, the heat removal and for cooling the blades of the first two gas turbine stages. After the oxidation reaction, N2 is obtained at 575 °C. This N2 is reused by mixing it with the air that is going to the CLC reactor for the heat removal step. In this way, the heat of the N2 is reused in the system. As a result, the temperature of the heat removal stream is slightly higher (#8, about 466 °C) and therefore the subsequent reduction can be carried out at a slightly higher temperature. The reduction temperature is one of the most critical parameters for the oxygen carrier selection, so a higher temperature could be quite beneficial as it increases the reduction kinetics. The outlet from the oxidation should be at 20 bar (as the air with which it is mixed) and therefore the air for oxidation (#7) is compressed by an additional compressor to a pressure slightly higher than 20 bar (to overcome the pressure drop in the reactors).
Between oxidation and reduction steps, the reactor needs to be purged with N2 from the ASU. This stream is pressurized and fed to the packed bed reactors at 478 °C and 20.4 bar (#9). The purge outlet is mixed with the heat removal stream.
The heat removal stream is fed to the gas turbine at 1200 °C and 19 bar (#10). In the gas turbine, the gas is expanded and sent to the heat recovery steam generator (#11), where steam is generated at three different pressure levels, 144 bar, 36 bar and 4 bar. The cooled O2 depleted air is finally vented by a stack (#12).
130 |
Figure 6.1: The simplified IGCLC power plant scheme.
6.2.2. Packed bed reactor model for CLC reactors The packed bed reactors are described in more detail by a 1D packed bed reactor model that is discussed in detail in section 3.3.3, but in this case the heat losses are considered to be negligible. The kinetics from Garcia-Labiano have been selected (García-Labiano et al., 2005), including the expected influence of diffusion limitations inside the particles, which are based on the particle model, discussed in section 2.2.3 (Noorman et al., 2011a). The influence of the reaction rate in large packed bed reactors is limited to the shape of the reaction front, which covers only a small part of the reactor. So, as long as the reaction rates are not very slow, the kinetics have a small influence on the outlet temperature and concentration profiles. In fact, the maximum temperature achievable in the packed bed is determined by the active metal in the oxygen carrier, as discussed in chapter 3 (equation 3.3).
6.3. Results and discussion
6.3.1. TS-CLC in series or in parallel For TS-CLC two different configurations can be used. During heat removal both reactors have to be connected in series to transfer the heat generated in the first reactor
5Air
11
N2pcompr
Redu
ctio
n
12
2
gasifiersyngas treatment
ST ST
H2O
3CO2fuel
heater
HP eva+SH
1
174
6
ASUO2
Coal
13 14
HP eco
LP eco+ eva+SH
15 16 19
Air
CO2
syngas cooling
18Pu
rge2
Heat
rem
oval
Oxi
datio
n
Purg
e1
9Air
blower 7
8
GT
10
6. Energy evaluation of TS-CLC configurations for the IGCC power plant |131
6
to the second reactor. But for the reduction and the oxidation mode, the reactors with different oxygen carriers can be connected in series or in parallel. In case the reactors are in series, the gas flows are fed to the first bed and the outlet is fed to the second bed, as illustrated in Figure 6.2. With this configuration it is also possible to operate with one reactor that contains two sections (with two different oxygen carriers), because the two stages are always operated in series. This has the advantage of lower complexity.
An alternative is to operate the reactors in parallel during one or both the oxidation and the reduction steps. In those cases, higher flow rates can be used while still complying with the maximum allowable pressure drop because of the reduced reactor length. So, in the end the total cross-section can be reduced. Another advantage is that the CO2/H2O is produced at lower temperature in the first reactor increasing the amount of heat stored for the heat removal. The main drawback is that also valves have to be installed downstream of the first reactor and thus, a larger number of valves is required.
In principle, it is possible to operate the reduction, oxidation and the purge in parallel. But in case the oxidation is carried out in parallel, the inlet zone of the second bed is cooled down too much and therefore the gas flow obtained from the heat removal has a lower temperature, which reduces the process efficiency. Therefore, it was chosen to operate only the reduction in parallel. In Figure 6.3 it is shown how the process is operated in this case. Because the reactors operate in parallel, more valves are required. The valves for the second beds need to withstand 1200 °C, but the maximum temperature of the valves on the first reactor is lower (around 900 °C).
132 |
Figure 6.2: TS-CLC configuration with all the operation steps in series (TS-CLC series).
Figure 6.3: TS-CLC configuration with the reduction in parallel and the other operation steps in series (referred as: TS-CLC parallel).
Air, 15°C, 1atm
438°C,20bar Heat Recovery Steam Cycle
Gasturbine
N2pcompr
Airblower
Air448°C20bar
1st b
edO
xida
tion
CO2/H2O
N2 from ASU
Syngas, 600°C
1st b
edPu
rge2
N2 purge, 478°C
Air80°C, 1 atm
2nd b
edO
xida
tion
2nd b
edPu
rge2
1st b
edHe
at re
mov
al2nd
bed
Heat
rem
oval
1st b
edPu
rge1
2nd b
edPu
rge1
Air1200°C,19 bar
1st b
edRe
duct
ion
2nd b
edRe
duct
ion
AirCompressor
Air, 15°C, 1atm
438°C,20bar Heat Recovery Steam Cycle
Gasturbine
N2pcompr
Airblower
1st b
edO
xida
tion
CO2/H2O
N2 from ASU
Syngas, 600°C
1st b
edPu
rge2
Air80°C, 1 atm
2nd b
edO
xida
tion
2nd b
edPu
rge2
1st b
edHe
at re
mov
al2nd
bed
Heat
rem
oval
1st b
edPu
rge1
2nd b
edPu
rge1
Air1200°C,19 bar
1st b
edRe
duct
ion
2nd b
edRe
duct
ion
Air448°C21bar
CO2/H2O
AirCompressor
6. Energy evaluation of TS-CLC configurations for the IGCC power plant |133
6
6.3.2. Reactor sizing and initial investment cost estimation A continuous process has been simulated by the 1D reactor model (discussed in section 3.3.3). To design the continuous process, a number of reactors per operation step have to be used and the flow rates have to be calculated. These parameters have been calculated based on the following criteria:
- Two cycle times were selected: 20 and 60 minutes. - The pressure drop should not exceed 5% (1 bar). A pressure drop of 5% has
been selected in this study, but a techno-economic evaluation is required to draw conclusions about this parameter, which is not the scope of this work.
- The inner reactor diameter is set on 2.5 m in case of a cycle time of 20 min and 4 m for a cycle time of 60 min. In both cases, a similar L/d-ratio is obtained, which makes it justified to compare those cases. With these sizes, no problems are expected with the transport of the reactors on trucks.
- The total purge size is equal to the total empty volume of the reactor; this should be sufficient, because the gas mixing inside the reactor is very small.
During the cycle (20 or 60 min.), sufficient oxygen should be available for the indirect combustion of syngas. So, the total volume of all the reactors together is fixed according to equation 6.1. This is based on the oxygen carrier properties (ρmol,oxygen is the amount of atomic oxygen per reactor volume), which are listed in Table 6.1.
2
,
CO H cycleR
mol oxygen
M MV CO H cycle2M MCO H 2MH 6.1
For continuous operation, several reactors have to be in operation. The minimum number is five, because there are five operation steps (oxidation, reduction, heat removal and 2 purge steps). For a purge step one reactor is sufficient, but for the other operation steps the total gas flow is too high to send to only one reactor. The criterion to determine the number of reactors is the pressure drop inside the reactors. For a certain reactor length, the maximum specific flow rate (kg/(m2·s)) is calculated based on the Ergun equation considering the highest possible gas temperature. With this value, the number of reactors in a certain operation step is calculated by dividing the total capacity by this maximum specific flow rate and the cross section of a reactor (equation 6.2). In this study, the reactor diameter was fixed at 2.5 or 4 m (dependent on the cycle time) and also the total reactor volume is fixed by the amount of oxygen (equation 6.1). So, the only parameters to be determined are the length and the number of reactors, which are connected via equation 6.3. By varying the length and the
134 |
number of reactors, an optimum is reached at which the pressure drop does not exceed the imposed limit.
,operation step 2
4R
mNm d 2
4m d4 6.2
2
4R RV N d L 6.3
The total reactor volumes are based on the properties of the selected oxygen carriers. For the one stage case, 19 wt% of NiO/Al2O3 is sufficient to reach 1200 °C in the reactors. The active weight contents for the TS-CLC cases are also based on this outlet temperature. In the TS-CLC cases, the heat removal gas inlet temperature is different, so a different ΔT has to be reached and therefore both cases have a different active weight content. Based on the active weight contents and the density, the amount of available oxygen per reactor volume is calculated (assuming a reduction to Cu, Ni and MnO respectively). As shown in Table 6.1, the result is that the amount of oxygen carried by a given volume of supported metal is smaller for the TS-CLC configurations. Therefore, a larger reactor volume (about two times larger) is required for those cases for a given cycle time.
The following designs were made for a cycle time of 60 min. For the one stage process, the reactor length is 11.8 m and there are 24 reactors in operation. For TS-CLC in series the total length is 22.5 m (bed 1: 12 m and bed 2: 10.5 m), while 27 reactors (54 beds in total) are in operation. These bed sizes were selected to keep the difference between the minimum and the maximum temperature in each reactor as small as possible. This has the advantage that the oxygen carrier particles experience the lowest possible thermal stress, which can be advantageous for the mechanical stability and possibly increase its lifetime. The construction of the reactor itself might become easier as well. Because of the lower ΔT, longer reactors are needed than for the one stage concept, which can be explained by two reasons. First, if the ΔT is smaller, a larger amount of material is required to store heat in the reactor. Second, the lower ΔT is reached with a lower oxygen content of the oxygen carrier material and therefore more material is needed to have the same total oxygen capacity for the combustion. For the TS-CLC in parallel, the first reactor is 12 m and the second is 11.5 m and there are 50 reactors in total; the lengths of the reactors have been chosen so that at the end of the heat removal, the heat front reaches the end of both beds at the same time. The heat front velocity is based on the heat capacity of the oxygen carriers. An overview of the sizing parameters can be found in Table 6.2.
6. Energy evaluation of TS-CLC configurations for the IGCC power plant |135
6
Table 6.1: Properties of the oxygen carriers.
One stage CLC
TS-CLC series TS-CLC parallel bed 1 bed 2 bed 1 bed 2
Oxygen carrier* 19wt% NiO 10wt% CuO
27wt% Mn3O4
10wt% CuO
30wt% Mn3O4
Redox couple NiO/Ni CuO/Cu
Mn3O4/MnO
CuO/Cu
Mn3O4/MnO
Solids bulk density in oxidized state, εsρs kg/m3
1031 989 1000 989 1006
Oxygen availability, kgo2/m3reactor
42.0 19.9 18.9 19.9 21.1
Particle diameter, mm 10 10 10 10 10 Void fraction, m3gas/m3reactor 0.4 *=in all these cases Al2O3 acts as support material
Table 6.2: The sizing parameters for the different reactor concepts.
One stage CLC
TS-CLC series
TS-CLC parallel
Cycle time, min. 20 60 20 60 20 60 Number of reactors 35 24 44 27 42x2 25x2 Diameter, m 2.5 4 2.5 4 2.5 4 Length, m - bed 1, m - bed 2, m
6.9 11.8 6.6 5.2
12.5 10
6.2 5.8
12
11.5 Footprint, m2 172 302 216 340 412 628 Volume, m3 1185 3559 2533 7634 2460 7383 Thickness refractory, mm 250 476 237 452 252 484 Thickness steel, mm 27 44 27 44 27 44 Reactor costs, k€/set of reactors 44 210 64 334 80 424 High temp. valve cost, k€/valve 319 400 278 373 286 390 Number of valves/set of reactors 1 1 2 Oxygen carrier cost, €/ton 50,000 10,000 10,000
TS-CLC series is considered to be carried out in one reactor of 22.5 m that contains two sections. In that case, the footprint is 340 m2. In case, the process is operated in two separate reactors, the total footprint would be a factor 2 larger. TS-CLC parallel has to be carried out in two separate reactors, because the reduction is carried out in parallel and therefore a larger total footprint is required, 628 m2. A preliminary investment cost estimation has been carried out to investigate this effect on the initial investment costs. Table 6.2 also contains the reactor sizes in the case of a cycle time of 20 min. This
136 |
option has been designed to evaluate the effect of the cycle time on the initial investment costs.
The initial investment costs have been estimated based on the costs of the oxygen carrier, the high temperature valves and the reactors. The reactor containing the oxygen carrier is surrounded by an internal refractory, a carbon steel vessel and an external refractory. The internal refractory is required to keep the steel vessel temperature below 300 °C. The thickness of the internal refractory and the steel vessel (ssteel) is reported in Table 6.2, which is calculated by equation 6.5 and 6.6. For the size of the refractory it is assumed that 0.25% of the total heat produced at 1200 °C is lost. With this assumption the heat losses are in line with a heat transfer from the external surface (at 70 °C) to the environment with a typical heat transfer coefficient of 5 W/(m2·K) (Hagen, 1999). The heat conductivity of internal insulation material (internal refractory) is 0.2 W/(m·K). For the steel thickness, a pressure, p, of 30 bar is assumed for safety reasons and the design stress, f, is estimated at 85 N/mm2 (Sinnott, 2005).
The costs of the vessels are calculated considering the steel costs of 500 €/ton (“World steel prices,” 2014) and fire bricks (refractory) of 450 €/ton (density of 480 kg/m3 (ThermalCeramics, 2013)) and multiply them by a factor 3 to include the reactor construction. The cost of a high temperature valve is estimated at € 150,000 in case of a hot gas flow rate (during the heat removal step) of 2 m3/s. The valve cost is scaled up by equation 6.4 (Seider et al., 2004). An overview of the high temperature valve system costs for each case is given in Table 6.2. In case of TS-CLC parallel, the outlet stream of the first reactor has to be managed and thus two valves per set of reactors are required. In the TS-CLC series, this is not required, because the outlet of the first reactor is always going to the second reactor in series. The oxygen carrier costs are estimated at 50,000 €/ton in case of the nickel-based oxygen carrier for the one stage configuration and at 10,000 €/ton for the cheaper copper- and manganese-based oxygen carriers that are used for the TS-CLC process.
0.6
00
VCO COV
0.6VVVVV0V0VV 6.4
max,
2
ln
refractoryCLC steel
reactor refractory
reactor
LQ T T
dd
6.5
4 1.2reactor refractory
steel
p ds
f p 6.6
6. Energy evaluation of TS-CLC configurations for the IGCC power plant |137
6
In Figure 6.4 it is demonstrated that the resulting initial investment costs are about a factor 2 lower in the TS-CLC cases in comparison with the one stage CLC with a nickel-based oxygen carrier, because cheaper oxygen carriers can be used in the TS-CLC cases. The TS-CLC parallel case is slightly more expensive than the TS-CLC series case, because more high temperature valves are required. The initial investment cost can be further reduced by optimizing the cycle times, increasing the reactor diameter, increasing the particle diameter or increasing the tolerated pressure drop. The purpose of this work is just to demonstrate the effect of the TS-CLC configuration and therefore the optimization of the above mentioned parameters is out of scope of this work.
For the process simulation of the reactors, the case with a cycle time of one hour has been selected to decrease the switching frequency of the high temperature valve at the exit of the reactor, which is probably beneficial for its lifetime. It is expected that the operating costs associated with the substitution of the material are higher for the one stage CLC plant, which makes the TS-CLC more attractive, despite the higher costs that are expected for the reactors network.
20, 60 20, 60 20, 600
100
200
300
400
500
600
0
35
71
106
142
177
213
Cycle time (min.)TS-CLC parallelTS-CLC series
Initi
al in
vest
men
t cos
ts (M
€)
Initi
al in
vest
men
t cos
ts (€
/kW
)
High temperature valve costs Oxygen carrier costs Reactor costs
1 stage CLC
Figure 6.4: Initial investment costs for the TS-CLC cases in comparison with the one stage CLC for two different cycle times (20 min and 60 min).
138 |
6.3.3. One stage CLC Based on the above mentioned criteria, simulations have been carried out with the settings shown in Table 6.3. In total 24 reactors are in operation in this configuration, where 5 reactors are in the reduction step, 4 in oxidation, 13 in heat removal and 1 per purge step.
Table 6.3: Settings for the one stage CLC. The outlet temperatures are results from the packed bed model.
Step reactors time, s mass flux, kg/m2/s Tin, °C Tout, avg, °C Reduction 5 750 1.841 600 832 Purge1 1 150 0.746 478 527 Oxidation 4 600 3.520 448 575 Heat Removal 13 1950 4.311 459 1199 Purge2 1 150 0.746 478 1192 24 3600
The outlet temperature and mass flux for one reactor as a function of time is illustrated in Figure 6.5. This reactor follows the sequence: reduction, purge1, oxidation, heat removal, purge2. The other reactors follow the same sequence, but start at a different position, so that a continuous process is achieved.
The axial temperature profiles corresponding to the outlet profile in Figure 6.5 are provided in Figure 6.7. It is shown that when the reduction is started, the outlet of the reactor is still hot from the previous heat removal step. This heat is blown out of the bed during the reduction and therefore a decreasing CO2/H2O outlet temperature is observed. During the reduction reaction, the temperature inside the reactor is 450-550 °C. In this temperature range, the selectivity of the reduction reactions with nickel (to CO2 and H2O) is not a problem according to the thermodynamics. Hence, complete conversion of the fuel is assumed. During oxidation, the temperature of the bed increases to 1200 °C, because of the exothermic reaction. Subsequently, the heat is blown as hot gas stream to the gas turbine (heat removal step).
Several reactors operate during the reduction and the heat removal, which deliver different streams at different temperatures. These streams are mixed and then an outlet temperature is obtained as shown in Figure 6.6. The temperature of the CO2/H2O stream fluctuates between 750 and 900 °C with an average temperature of 832 °C. These temperature variations have to be controlled as previously mentioned. The mixed stream with the average temperature is sent to a heat exchanger, where high pressure steam is generated at 565 °C.
6. Energy evaluation of TS-CLC configurations for the IGCC power plant |139
6
0 500 1000 1500 2000 2500 3000 3500 4000400
600
800
1000
1200
Out
let t
empe
ratu
re (°
C)
p2heat removaloxidationp1
Outlet temperature (°C) Mass flow inlet (kg/(m2s)) Mass flow outlet (kg/(m2s))
Time (s)
reduction
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
Mas
s flo
w (k
g/(m
2 s)
Figure 6.5: Outlet temperature and mass flux for one reactor as a function of time operated in one stage CLC.
0 500 1000 1500 2000 2500 3000 3500 4000
600
800
1000
1200
Air to gas turbine
Out
let t
empe
ratu
re (°
C)
Time (s)
CO2/H2O
Figure 6.6: Outlet temperature during dynamic operation of reactors in the one stage CLC process.
140 |
0 4 8 12 16400
600
800
1000
1200
red,initial red, =1/2 red, end purge1, end
a. reduction
Tem
pera
ture
(°C
)
axial position (m)
0 4 8 12 16
400
600
800
1000
1200
purge1, end ox, =1/2 ox, end
b. oxidation
Tem
pera
ture
(°C
)
axial position (m)
0 4 8 12 16
400
600
800
1000
1200
ox, end HR, =1/3 HR, =2/3 HR, end purge2, end
c. heat removal
Tem
pera
ture
(°C
)
Axial position (m) Figure 6.7: Axial temperature profiles of reactors in the one stage CLC process during reduction (a), oxidation (b) and heat removal (c).
After integrating the obtained streams from the reactor model within the power plant model, a mass balance is obtained as shown in Table 6.6. With this process, a net efficiency can be achieved of 41.05% of LHV. An overview of the energy balance is given in section 6.3.6, which shows the electricity production and consumption in the power plant.
6.3.4. TS-CLC series The same procedure has been carried out for the option in which TS-CLC is carried out with two reactors in series. This reactor is simulated as one large reactor that consists of two beds. The first bed has a length of 12.5 m and contains 10 wt% CuO/Al2O3. The second bed has a length of 10 m and contains 27 wt% Mn3O4/Al2O3. The simulation settings are listed in Table 6.4. The inlet temperature for the heat removal is higher than in the one stage case (574 °C instead of 459 °C). This stream is obtained after mixing the air with the outlet from the oxidation and purge 1. The latter two streams have a higher temperature in this case, which results in a higher heat removal inlet temperature. As is illustrated in Figure 6.8, a constant air flow at 1200 °C can also be produced with this configuration.
6. Energy evaluation of TS-CLC configurations for the IGCC power plant |141
6
Table 6.4: Simulation settings for TS-CLC in series.
Step reactors time, s mass flux, kg/m2/s Tin, °C Tout, avg, °C Reduction 3 400 3.068 600 1130 Purge1 1 133 1.569 478 1032 Oxidation 5 667 2.815 449 970 Heat Removal 17 2267 2.856 574 1198 Purge2 1 133 1.569 478 1202
27 3600
During reduction, CO2 and H2O are produced at a higher temperature than in the one stage case, caused by the increased heat dispersion in the packed bed due to the two stage operation. The larger extent of heat dispersion can be explained as follows. The extent of heat dispersion in the first bed is similar to the one stage configuration. But in this case, the dispersed gas flow obtained from the first reactor is sent to the second reactor, where it experiences some additional heat dispersion. In Figure 6.10 the axial temperature profiles inside the reactor are shown. During reduction, syngas is fed at 600 °C and the temperature in the CuO-reactor rises due to the exothermic reduction reactions. In the Mn3O4-reactor, the temperature rise is smaller, because the reduction reactions are less exothermic. When the reduction is completed, two different temperature plateaus can be observed: the first reactor is mainly at 700 °C and the second reactor mainly at 900 °C. During oxidation, air is fed at only 448 °C and therefore a temperature drop can be observed at the inlet of the first reactor. Due to the exothermic oxidation reaction, the temperature rises by about 200 °C in the first reactor and 300 °C in the second reactor, so that a 1200 °C plateau is reached in the second reactor. After the bed has been oxidized, the heat is blown out of the bed and a continuous flow of air at 1200 °C is produced. The average outlet temperatures are shown in Figure 6.9. In this case, the CO2/H2O stream is produced at a higher temperature than in the one stage CLC case and also the temperature fluctuations are smaller (30 °C instead of 140 °C).
The axial temperature profiles from in Figure 6.10 show a reduced temperature difference among the different steps compared to the one stage CLC configuration. In the first bed, a maximum ΔT of 500 °C is observed. The minimum temperature in the second reactor is at around 850 °C during the oxidation, resulting in a maximum ΔT of 350 °C.
142 |
0 500 1000 1500 2000 2500 3000 3500 4000400
600
800
1000
1200p2heat removaloxidationp1red.
Out
let t
empe
ratu
re (°
C)
Outlet temperature (°C) Mass flow inlet (kg/(m2s)) Mass flow outlet (kg/(m2s))
Time (s)
1
2
3
4
Mas
s flo
w (k
g/(m
2 s)
Figure 6.8: The outlet temperature and mass flux profile from TS-CLC with the two reactors always connected in series.
0 500 1000 1500 2000 2500 3000 3500 4000
600
800
1000
1200
CO2/H2O
Air to gas turbine
Out
let t
empe
ratu
re (°
C)
Time (s)
Figure 6.9: Average outlet temperatures from the TS-CLC series process.
6. Energy evaluation of TS-CLC configurations for the IGCC power plant |143
6
0 4 8 12 16 20
400
600
800
1000
1200
red,initial red, =1/2 red, end purge1, end
a. reductionCuO/Al2O3 Mn3O4/Al2O3
Tem
pera
ture
(°C
)
Axial position (m)
0 4 8 12 16 20
400
600
800
1000
1200
purge1, end ox, =1/2 ox, end
b. oxidation
Tem
pera
ture
(°C
)
Axial position (m)
CuO/Al2O3 Mn3O4/Al2O3
0 4 8 12 16 20
400
600
800
1000
1200
ox, end HR, =1/3 HR, =2/3 HR, end purge2, end
c. heat removal
Tem
pera
ture
(°C
)
Axial position (m)
CuO/Al2O3 Mn3O4/Al2O3
t
t
Figure 6.10: Axial temperature profiles during TS-CLC, for the case where the reactors are always connected in series.
6.3.5. TS-CLC with reduction in parallel (TS-CLC parallel) As explained before, it is also possible to operate CLC in two stages with the reduction in parallel to produce the CO2 at lower average temperatures, which leads to a larger gas flow that can be fed to the gas turbine.
The two reactors used for this configuration are simulated separately. The simulation settings are shown in Table 6.5. The outlet temperature profiles of the streams that are processed downstream are shown in Figure 6.12. Also in this case, a gas stream with a constant temperature of 1200 °C is obtained during heat removal from the second reactor.
144 |
Table 6.5: Simulation settings for TS-CLC with reduction in parallel and oxidation, heat removal and purges in series (TS-CLC parallel).
Step reactors time, s mass flux, kg/m2/s Tin, °C Tout, avg, °C
Reduction reactor 1 2 288 2.233 600 801 Reduction reactor 2 2 288 2.369 600 994 Purge1 1x2 144 1.521 478 962 Oxidation 5x2 720 2.816 448 915 Heat Removal 16x2 2304 3.393 547 1192 Purge2 1x2 144 1.521 478 1075
25x2 3600
In Figure 6.12 it is shown that the temperature decrease during the reduction is smaller than in the one stage configuration. In total, 4 parallel reduction reactors are in operation, whose streams are mixed before being sent to the heat recovery section. This mixed stream has smaller temperature differences and this is an advantage for the design of the downstream heat exchanger. The axial temperature profiles are shown in Figure 6.11. The profiles in the first reactor look similar to the one stage case. Only the temperature levels are different (related to the different oxygen carrier used).
On the other hand, profiles in the second reactor are significantly different from the profiles in the other two cases. Before the reduction starts, a large part of the reactor has a temperature of 850 °C. During the reduction step, syngas is fed into the reactor at 600 °C and CO2/H2O is produced at 1050-950 °C. During oxidation, a heat plateau is formed at 1200 °C. This heat is blown out of the bed during the heat removal step. Feeding the second reactor with relatively cold syngas at 600 °C, leads to a significant temperature drop at the inlet. This leads to an irregular temperature profile for the following stages, with an area with local maximum and local minimum temperature which is progressively blown to the exit of the reactor. However, thanks to the heat diffusion in the axial direction, the temperature profile becomes more uniform during the heat removal phase and the temperature of the hot gases produced is sufficiently stable to feed the power cycle.
6. Energy evaluation of TS-CLC configurations for the IGCC power plant |145
6
0 2 4 6 8 10 12400
600
800
1000
1200 red,initial red, =1/2 red, end purge1, end
a. reduction bed 1, CuO/Al2O3
Tem
pera
ture
(°C
)
Axial position (m)
0 2 4 6 8 10 12400
600
800
1000
1200
red,initial red, =1/2 red, end purge1, end
a. reduction bed 2 - Mn3O4/Al2O3
Tem
pera
ture
(°C
)
Axial position (m)
0 4 8 12 16 20 24
400
600
800
1000
1200
b. oxidationMn3O4/Al2O3CuO/Al2O3
Tem
pera
ture
(°C
)
Axial position (m)
purge1, end ox, =1/3 ox, =2/3 ox, end
0 4 8 12 16 20 24
400
600
800
1000
1200
c. heat removalMn3O4/Al2O3CuO/Al2O3
Tem
pera
ture
(°C
)
Axial position (m)
ox, end HR, =1/3 HR, =2/3 HR, end purge2, end
Figure 6.11: Axial temperature profiles during TS-CLC with the reduction in parallel.
In Figure 6.11 it is shown that the maximum temperature difference that the reactors experience, is limited in the first reactor (450 °C), but relatively large in the second reactor (600 °C). A larger ΔT is observed in the second reactor, because the fuel is fed at 600 °C and air is produced at 1200 °C.
When mixing the gases produced by reactors operating in parallel during the same stage, the averaged outlet temperature profiles illustrated in Figure 6.13 are obtained.
146 |
0 500 1000 1500 2000 2500 3000 3500 4000400
600
800
1000
1200
bed 1
p2heat removalox.p1red.
Out
let t
empe
ratu
re (°
C)
Time (s)
bed 2
Figure 6.12: Outlet temperatures from the first and second reactor that are processed downstream.
0 500 1000 1500 2000 2500 3000 3500 4000
600
800
1000
1200
CO2/H2O from 1st reactor
Air fed to 2nd reactor for HR
CO2/H2O from 2nd reactor
Air to gas turbine
Out
let t
empe
ratu
re (°
C)
Time (s)
Figure 6.13: Outlet temperature profiles from the parallel TS-CLC.
6. Energy evaluation of TS-CLC configurations for the IGCC power plant |147
6
6.3.6. Comparison of the different configurations For each of the three configurations, the integration in the power plant has been designed and the efficiency evaluated. The mass balances for the three configurations are shown in Table 6.6. When CO2 and H2O are produced at a higher temperature (#2), more high pressure steam is produced in the downstream cooler. For that reason, the amount of high pressure steam in the TS-CLC series case differs significantly from the other cases (#13). When more energy is taken by the CO2/H2O stream, a smaller hot air stream can be produced for the gas turbine (#10). The super- and reheat temperature in the HRSC is about 25 °C below the turbine outlet temperature (#11).
Table 6.6: The mass balances of the three different packed bed configurations.
# One stage CLC (base case)
TS-CLC series TS-CLC parallel
T,°C p,bar M,kg/s T,°C p,bar M,kg/s T,°C p,bar M,kg/s 1 Syngas 600 20.0 116.3 600 20.0 116.3 600 20.0 116.3 2 CO2/H2O 832 19.0 156.9 1130 19.0 156.9 901* 19.0 156.9 3 CO2/H2O 136 18.0 156.9 127 18.0 156.9 136 18.0 156.9 4 CO2 28 110.0 81.5 28 110.0 81.5 28 110.0 81.5 5 air 15 1.0 786.2 15 1.0 668.3 15 1.0 750.8 6 air 438 20.0 729.8 438 20.0 617.6 438 20.0 696.8 7 air 448 21.0 176.6 449 21.0 176.6 448 21.0 176.6 8 O2 depl. air 466 20.0 698.4 586 20.0 596.8 554 20.0 675.3 9 N2 478 20.4 18.4 478 20.4 39.5 478 20.0 38.2 10 O2 depl. air 1199 19.0 707.6 1198 19.0 616.5 1192 19.0 694.4 11 O2 depl. air 486 1.0 764.0 482 1.0 667.2 482 1.0 748.4 12 O2 depl. air 92 1.0 764.0 81 1.0 667.2 89 1.0 748.4 13 Steam 565 133.9 88.6 565 133.9 130.6 565 133.9 98.0 14 Steam 527 133.9 129.5 544 133.9 157.4 530 133.9 136.3 15 Steam 333 36.0 129.5 346 36.0 157.4 335 36.0 136.3 16 Steam 458 33.1 142.6 453 33.1 161.3 453 33.1 147.1 17 Steam 395 21.6 33.4 390 21.6 33.4 390 21.6 33.4 18 Steam 300 3.5 37.7 300 3.5 30.9 300 3.5 36.1 19 Steam 32 0.05 146.8 32 0.05 158.7 32 0.05 149.8 SH/RH steam temperature from HRSG, °C
460 455 455
Maximum ΔT in CLC reactors, °C 760 500/350 450/600
Maximum ΔT after mixing the CO2/H2O flows, °C
140 30 50
*both CO2/H2O streams are considered to be mixed.
The energy balances of the three configurations are listed in Table 6.7. The chemical-looping cases are compared with an IGCC power plant without CO2 capture (first
148 |
column), characterized by a net efficiency of 45.2% (Spallina et al., 2014). It is also demonstrated that with all the CLC configurations the conventional capture with Selexol (second column) is outperformed by 6-7% points. These configurations are based on current technology. The CO2 avoided and the SPECCA are calculated by equation 6.7 and 6.8 with the IGCC plant without CO2 capture as reference.
2
2,ref
CO2,avoided
CO
ECO 1
E 6.7
,
2, 2
1 1
SPECCA 3600el el ref
CO ref COE E 6.8
Table 6.7: Energy balances of the different configurations.
Power IGCC-NC
(Spallina et al., 2014)
IGCC Selexol® (Spallina
et al., 2014)
One stage CLC
TS-CLC series
TS-CLC parallel
Heat input LHV, MWLHV 812.5 898.8 853.9 853.9 853.9 Gas turbine, MWe 261.6 263.9 225.1* 194.0* 217.3* Heat Recovery Steam Cycle, MWe 179.5 161.2 183.0 208.1 188.3 Gross power output, MWe 441.1 425.1 408.1 402.1 405.6 Syngas blower, MWe -1.0 -1.1 -0.8 -0.8 -0.8 N2 compressor, MWe -34.1 -29.8 ASU, MWe -29.6 -32.7 -33.9 -33.9 -33.9 Lock hoppers CO2 compressor, MWe
-3.1 -3.1 -3.1
Acid Gas Removal, MWe -0.4 -14.7 -0.4 -0.4 -0.4 CO2 compressor, MWe -19.7 -11.0 -11.0 -11.0 N2 intercooled compressor gasifier, MWe
-1.3 -1.3 -1.3
Heat of rejection, MWe -5.5 -6.3 -3.6 -3.8 -3.7 Other auxiliaries, BOP, MWe -3.2 -3.6 -3.4 -3.4 -3.4 Net power generated, MWe 367.4 317.3 350.6 344.5 348.1 LHV efficiency, % 45.21 35.31 41.05 40.34 40.77 CO2 capture efficiency, % 93.0 97.1 97.1 97.1 CO2 purity, % 98.2 96.7 96.7 96.7 CO2 emission, kg CO2 emitted/MWhe
769.8 101.4 24.7 25.1 24.9
CO2 avoided, % 0 86.8 96.8 96.7 96.8 SPECCA, MJ LHV/kg CO2 3.34 1.08 1.29 1.16 * Gas turbine power includes consumption of air blower and nitrogen compressor for purge.
6. Energy evaluation of TS-CLC configurations for the IGCC power plant |149
6
In the TS-CLC series configuration, a larger amount of high pressure steam is produced and this results in a larger electricity production in the HRSC. However, because the flow to the gas turbine is smaller, the gas turbine produces less electricity.
In the one stage case, the highest LHV efficiency is obtained, 41.05%, but the TS-CLC cases can still compete well with the one stage case, with 40.34-40.77%. A slightly higher LHV efficiency can be obtained in the TS-CLC parallel case (40.77%) compared with TS-CLC series (40.34%), because of the lower CO2 temperature. In all the IG-CLC cases, the efficiency is more than 5% points higher than the reference case with Selexol. The CLC cases also have lower specific CO2 emissions (reduced by a factor four) and a lower electricity consumption for CO2 capture (SPECCA) (by factor three) than the plant with conventional CO2 capture.
6.4. Conclusions The performances of different packed bed CLC configurations in an IGCLC power plant have been compared based on process and reactor design and the initial investment costs. In one stage packed bed CLC, the heat is produced one reactor, but recently a different approach has been demonstrated in which heat is produced in two stages (TS-CLC). In this case, two reactors are in operation in series. During the heat removal operation step, heat is blown from the first bed to the second bed and from the second bed to the gas turbine. In this step, both reactors need to be connected in series. But the oxidation, reduction and purge steps can also be operated with both reactors in parallel. The most beneficial option is with the reduction in parallel and the other steps in series and this option has been considered as best alternative.
It has been demonstrated that with the TS-CLC approach a LHV efficiency close to the one stage CLC can be reached (40.3-40.8% compared to 41.1%). In all the CLC cases, the conventional CO2 capture technology based on Selexol is outperformed by at least 5% point. The advantage of the configuration is that there is much more flexibility on the selection of the oxygen carrier. For TS-CLC a larger reactor volume is needed, a lower temperature change per reactor is desired. The lower temperature change is possible if the oxygen content of the oxygen carrier material is decreased and this leads to a larger total reactor volume. Despite the larger reactor volume, the initial investment costs for TS-CLC are estimated to be a factor two smaller, because of the lower costs for the oxygen carrier.
Despite the slightly lower process efficiency, TS-CLC can be considered as a promising approach, because it gives more freedom to select the oxygen carriers and to minimize the investment costs of the reactor material.
|151
7 Comparison of CLC in packed bed and
fluidized bed reactors
This chapter is based on the following papers:
- Hamers, H.P., Romano, M., Spallina, V., Chiesa, P., Gallucci, F., Van Sint Annaland, M., 2014. Comparison on process efficiency for CLC of syngas operated in packed bed and fluidized bed reactors. Int. J. Greenh. Gas Control 28, 65-78.
- Hamers, H.P., Romano, M., Spallina, V., Chiesa, P., Gallucci, F., Van Sint Annaland, M., 2014. Boosting the IGCLC process efficiency by optimizing the desulfurization step. Submitted to Applied Energy.
152 |
7. Comparison of CLC in packed bed and fluidized bed reactors
Abstract In this thesis CLC is carried out in dynamically operated packed bed reactors, but in many other literature works the process is carried out in circulating fluidized bed reactors. In this chapter, the influence of the reactor selection on the overall process efficiency has been estimated. In this comparison it has been assumed that the circulating fluidized bed reactor system can be operated at 20 bar without producing fines, which has however not yet been demonstrated. For the pressurized circulating fluidized bed reactor system with NiO/Al2O3 as oxygen carrier an LHV efficiency of 41.4% has been calculated, assuming the fuel conversion at chemical equilibrium, leading to some unconverted CO and H2 in the reaction products. If full gas conversion is obtained by using another oxygen carrier operating at the same maximum temperature of 1200 °C, this efficiency can be increased to 41.8%. In the packed bed case, large temperature gradients inside the reactors are obtained and the CO2 and steam is produced at a lower (time-)average temperature. In this case a significant amount of steam has to be added or recycled to avoid carbon deposition, leading to an LHV efficiency of 41.1%. If a carrier is selected with a low kinetic activity for the Boudouard reaction, the efficiency can be increased by 1.1% point up to 42.2% of LHV. A small efficiency boost (0.5-1%) can be obtained by improving the desulfurization method.
Because the process efficiency is hardly influenced by the reactor type, the reactor selection will be made based on the availability, operability and cost of high temperature and high pressure reactor systems, that are still under development. Other differences between the two systems are related to the number of vessels needed and the volume of the solids inventory, which is expected to be lower in the circulating fluidized bed system.
7. Comparison of CLC in packed bed and fluidized bed reactors |153
7
7.1. Introduction As it has been discussed in the thesis introduction, several reactor types can be used for CLC. Although the two fluidized and packed bed concepts have been studied separately in some papers (Erlach et al., 2011; Rezvani et al., 2009; Spallina et al., 2014), the performances of both configurations have not been directly compared yet with the same assumptions and modeling tools. In this chapter, the packed bed and the fluidized bed reactor configurations are directly compared for the first time in terms of overall plant efficiency. The IG-CLC power plant is the same as discussed in section 6.2.1. NiO/Al2O3 has been selected as oxygen carrier for this comparison, because it is suitable for high temperature operation and has fast kinetics even at moderate temperatures and it is hence suitable for both reactor configurations (Adanez et al., 2012).
The chapter opens with a discussion on the packed bed configuration, including a sensitivity analysis regarding the steam addition (to avoid carbon formation) and the pressure drop (to reduce the number of reactors). A preliminary estimation of the initial investment costs is made to conclude which parameters (L/d of the reactor, cycle time, oxygen carrier) are critical in the packed bed CLC reactor design. Finally, the process performance in the fluidized bed reactors is discussed. Because many papers have been published about circulating fluidized bed systems, the discussion is less extensive than the packed bed discussion. Then, both configurations are compared. In the end, it is studied if the process efficiency can be increased by adapting the desulfurization method.
154 |
Figure 7.1: Simplified scheme of the IG-CLC power plant and the circulating fluidized bed (a) and the packed bed configuration (b).
5Air
12GT
N2pcompr
Airblower
7
Heat
rem
oval
Oxi
datio
n
Redu
ctio
n
2CO2/H2O
9N2
1
Purg
e1
Purg
e2
13
6Air
Airreactor
Fuelreactor
2CO2/H2O
1Syngas
11
MeO
Me
1 2 3 4 5
b) CLC in packed bed reactorsa) CLC in circulatingfluidized bed reactors
6
11
CLC
11
2
gasifiersyngas treatment
ST ST
H2O
3CO2
fuelheater
HP eva+SH
1
194
6
ASU
O2
Coal
15 16
HP eco
LP eco+ eva+SH
17 18 21
14Air
9
CO2
8
syngas cooling
20
7. Comparison of CLC in packed bed and fluidized bed reactors |155
7
7.2. Packed bed configuration The behavior in the packed bed reactors has been described by the 1D numerical reactor model. With this model, the temperatures of the gases exiting the packed bed reactors were determined. In this work, a 19wt% Ni-based oxygen carrier was used and performance in packed bed reactors is discussed in section 6.3.3. The results from the packed bed reactor model have been integrated with the IG-CLC power plant model (section 6.2.1) and this results in the mass and energy balances. The stream table of the IGCLC plant with packed bed CLC is shown in Table 7.2 and the corresponding simplified plant scheme is displayed in Figure 7.1. The overall efficiency is 41.05% on LHV basis and CO2 capture is around 97%. More detailed information about the energy balance is given further in Section 7.4.
In the following sections, the results of a sensitivity analysis are reported, carried out on the method used to avoid carbon deposition, the pressure drop and the gas turbine pressure ratio.
7.2.1. Carbon deposition prevention method Carbon deposition could occur by the Boudouard reaction, equation 7.1. The equilibrium of this reaction favors carbon formation at high pressure and low temperatures. Since the reduction is carried out at relatively low temperature in the packed beds and the fuel is in contact with a reduced oxygen carrier (i.e. no additional oxygen can be released to the gas phase by the oxygen carrier), carbon deposition could occur in the reactor. Carbon deposition can be prevented by recycling CO2 and H2O and/or by adding steam which favors the carbon consumption through the gasification reaction, equation 7.2. In the case a H2O/CO2 stream is recycled, the outlet stream of the packed bed reactors is cooled down to 450 °C and then some H2O/CO2 is slightly pressurized by a blower and sent back to the reactor inlet. Increasing the flow rate of the recycle stream means that a larger flow enters the CLC reactors, which absorbs a higher amount of heat and has to be cooled down. Hence, more high pressure steam is produced and a lower amount of heat can be extracted during the heat removal phase.
2 R2 CO CO C H 172.4kJ/mol 7.1
2 2 RC H O CO H H 131.3kJ/mol 7.2
156 |
The effects of the different recycle ratios on the amount of steam to be added to avoid carbon deposition are reported in Figure 7.2. For this figure, a minimum temperature of 450 °C in the Boudouard equilibrium calculations is assumed to prevent carbon deposition in the bed. Such a syngas dilution is also expected to be sufficient to prevent metal dusting in CLC fuel heater, where syngas is heated up to 600 °C (a temperature range where metal dusting corrosion can occur) (Natesan and Zeng, 2003). Here it is shown that if the recycle ratio is increased, less additional steam is required to avoid carbon deposition. Moreover, if only a recycle is used, with no dilution with steam from the steam turbine, a very large stream has to be recycled (about 2.75 times the CO2/H2O mass flow rate), because CO2 is not as effective to prevent carbon deposition as steam. Increasing the recycle leads to an increase of the high pressure steam produced from CO2/H2O cooling. To provide sufficient saturated water, the ΔT at the pinch point has to be increased with detrimental effects on process efficiency. Therefore, cases requiring increased ΔT are not reported in Figure 7.2 and curves are interrupted at a certain value of the recycle ratio.
Figure 7.2 demonstrates that in case a larger recycle ratio is selected, the flow to the gas turbine is decreased, with negative effects on the process efficiency: increasing the exhaust recirculation, results in a higher HP steam production in the first heat exchangers after the reactor operated in reduction, which decreases the overall thermal input of the CLC unit and thus, limits the heat available for the topping gas turbine cycle. The same effect occurs with lower preheating temperatures of the syngas to the CLC unit since a higher portion of the heat of reaction is needed for heating the syngas up to the reaction temperature (as shown in the figure). Another effect of the CO2/H2O recycle is related to the electricity consumed by the recycle blower to compensate the pressure drop inside the reactors and the CO2 cooler. These effects are partly balanced by the higher production rate of high pressure steam, which increases the electricity produced by the steam turbine. A reduction of the amount of steam needed for syngas dilution has a positive effect on the steam cycle efficiency, since a higher amount of steam is expanded in the turbine down to the condensing pressure, instead of being mixed with the syngas and eventually condensed at higher temperature during the CO2-rich stream cooling.
The result of these effects is illustrated in Figure 7.3, where the process efficiency is shown as a function of the recycle ratio. It turns out that the size of the recycle flow has a relatively small effect on the process efficiency. The case without recycle has been selected as base case, because this is the case with the minimum flow sent to the reduction reactor and its efficiency is only about 0.25 percentage points lower than the optimal case when a syngas preheating temperature of 600 °C is adopted.
7. Comparison of CLC in packed bed and fluidized bed reactors |157
7
0.0 0.1 0.2 0.3 0.4 0.5 0.67.0
7.5
8.0
8.5
9.0
Tsyngas,red,in=600°C
Tsyngas,red,in=450°C
no syngas heating
Recycle ratio (kgrecycled/kgin CLC reactor)
Flow
to g
as tu
rbin
e (k
g/kg
syng
as fr
om s
atur
ator)
0.1
0.2
0.3
0.4
Ste
am m
ixed
(kg/
kgsy
ngas
from
sat
urat
or)
Figure 7.2: The gas turbine inlet flow rate and the amount of steam mixed at different recycle ratios.
0.0 0.1 0.2 0.3 0.4 0.5 0.640.0
40.5
41.0
41.5
42.0
Tsyngas,red,in=600°C
Tsyngas,red,in=450°C
Recycle ratio (kgrecycled/kgin CLC reactor)
Ele
ctric
al e
ffici
ency
(% o
f LH
V)
no syngas heating
Figure 7.3: The influence of the recycle ratio and the syngas feeding temperature on the process efficiency.
158 |
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.640.5
41.0
41.5
42.0
42.5P
roce
ss e
ffici
ency
(% o
f LH
V)
H2O/CO-ratio
min. H2O/CO-ratioin undiluted syngasfrom saturator
Figure 7.4: The process efficiency as a function of the amount of steam mixed with the syngas. Initial H2O/CO ratio of undiluted syngas from the saturator is equal to 0.37. At a H2O/CO-ratio of 1.5, carbon deposition cannot occur according to thermodynamic equilibrium at 450 °C and higher.
Syngas preheating before feeding the CLC reactor system is performed by cooling the CO2/H2O stream. In addition to the improved process performance, increasing the syngas inlet temperature to 600 °C has another advantage. The inlet temperature is closer to the temperature in the reactor after the reduction reactor. Therefore, only a small temperature change is observed and this leads to fewer temperature fluctuations in the gas turbine inlet stream. A drawback of this heating procedure is that it is achieved in a gas/gas heat exchanger, which is expected to require a relatively large surface area.
In the above mentioned cases, a conservative estimation was selected for the amount of steam needed to avoid carbon deposition, because the thermodynamic equilibrium was followed. However, the kinetics determine how fast a reaction occurs and based on this, it might be that a smaller amount of steam is sufficient to avoid carbon deposition (or limit it to acceptable levels). The effect of a small amount of carbon deposition on the process thermodynamics is small. If 1% of the CO is deposited (evenly distributed) in the bed, the temperature rise during oxidation increases by 2 °C.
7. Comparison of CLC in packed bed and fluidized bed reactors |159
7
If no additional steam and no recycle are fed with the mixture, still a H2O/CO-ratio of 0.37 is fed to the reactor (#1) thanks to the humidification obtained in the saturator. So, the carbon formation reaction is suppressed to a certain extent and this might be sufficient in case the kinetics of carbon deposition are slow. This leads to an efficiency that is 1.18% point higher (42.23% of LHV). So, the effect of adding steam or using a recycle is quite significant. Therefore, the kinetics of the Boudouard reaction are an important aspect in the selection of the oxygen carrier for the packed bed process and should be investigated experimentally. In Figure 7.4 the effect of the added steam on the process efficiency is displayed.
For final considerations, the results just presented can be compared with the outcomes of a similar analysis performed in (Spallina et al., 2014). In that case, a much more pronounced dependence of plant efficiency on both the CO2/H2O recycle and steam dilution, in order to avoid carbon deposition, was obtained. The first reason of the lower dependence obtained in this work is related to the lower average temperature of the CO2/H2O stream produced in the plant assessed in this work, which limits the effect of higher fuel flows on the hot air flow rate to the gas turbine. As a matter of fact, in (Spallina et al., 2014) a different packed bed operation sequence is adopted to manage the lower kinetics of ilmenite on CO oxidation and the reduction step is performed on the hot bed just after the oxidation phase. Therefore, the CO2/H2O stream is also produced at a significantly higher temperature. The second reason is related to the configuration of the heat recovery section for the CO2/H2O stream, which consists in this case of a low pressure evaporation level, which can recover the increasing low temperature heat originating from steam condensation when high steam dilution is adopted. In Spallina et al. (2014), such a LP evaporation level is not present and low temperature heat from steam condensation is not recovered as efficiently.
7.2.2. Compression ratio and pressure drop in packed bed reactor
The effect of the compression ratio and the pressure drop is illustrated in Figure 7.5. It is demonstrated that the compression ratio has little effect on the process efficiency. A compression ratio of 20 was selected, corresponding to the highest plant efficiency. In addition to efficiency, the pressure ratio influences the design and operation of the reactor system in two ways. First, a higher bed temperature during reduction is obtained in case of a higher compression ratio. In an adiabatic compression, the higher the final pressure, the higher the final temperature. Since a compressed air stream is fed to the CLC reactors for the oxidation and the heat removal, when the reduction starts, the bed temperature is equal to the heat removal gas inlet temperature. For the reduction reaction rates, the temperature is quite a critical parameter and therefore a higher
160 |
compression ratio is beneficial. Second, a higher pressure ratio leads to a higher gas density and therefore a larger gas mass flow rate can be sent through the reactor for a given pressure drop. As it will be demonstrated in the next section, the number of reactors needed is lowered by increasing the compression ratio.
For the plant assessed, the optimal compression ratio of 20 is higher than expected for the turbine inlet temperature (TIT) of 1200 °C, based on the experience on conventional natural gas-fired combined cycles. As a matter of fact, optimal pressure ratios are around 12-14 in combined cycles utilizing a gas turbine with a turbine inlet temperature in that range. The higher than expected optimal pressure ratio is due to two reasons: 1) the maximum SH steam temperature is only slightly influenced by the pressure ratio, since most of the steam is superheated by hot CO2 from the CLC reactors; 2) CO2 compression consumption reduces when operating the reactors at a higher pressure.
14 15 16 17 18 19 20 21 2239.5
40.0
40.5
41.0
41.5
p/p=3% p/p=5% p/p=8% p/p=10% p/p=15%
Ele
ctric
al e
ffici
ency
(% o
f LH
V)
Compression ratio
p/p
3%
15%
Figure 7.5: The effect of the compression ratio and the pressure drop on the process efficiency.
7. Comparison of CLC in packed bed and fluidized bed reactors |161
7
In Figure 7.5 the effect of the pressure drop is shown as well. A larger tolerated pressure drop inside the reactors results in a lower process efficiency, because the stream fed to the gas turbine has a lower pressure in that case. In case the gas turbine inlet pressure is lower, the outlet temperature is higher. Therefore, steam produced in the heat recovery steam cycle (HRSC) can be superheated and reheated to a higher temperature since a fixed temperature approach of 25 °C is considered between the gas entering the HRSC and the final superheater and reheater temperatures.
What pressure drop should be tolerated depends on the process economics, since higher pressure drops lead to higher allowable gas velocities and hence a lower number of reactors. This analysis is out of scope of this work but deserves certainly more attention in a future research.
7.2.3. Reactor design strategy The same approach is used to determine the size of the reactors and the number of reactors as discussed in section 6.3.2, but in this case the reactors are sizes by fixing the ratio between the reactor length and the diameter, the L/d-ratio. The selection of the operating parameters has consequences on the reactor design. The fixed costs are expected to be lower in case the process can be operated with a lower number of reactors, because fewer valves are needed in that case. The following measures can be taken to decrease the number of reactors:
- Decrease the L/d ratio of the reactors. If the reactor diameter is increased, the footprint per reactor increases and then less reactors are needed to facilitate the desired flow. On the other hand, a uniform gas flow distribution along the reactor cross section can become difficult at low L/d. In addition, the reactor diameter should not exceed ca. 6 meter, because transport might become problematic and the walls might become too thick.
- Increase the maximum permitted pressure drop. Increasing the maximum permitted pressure drop means that a larger specific flow rate can be used and the reactor length can be increased. Therefore, a smaller overall footprint is required and thus a lower number of reactors.
- Increase the particle diameter. For a given reactor diameter, the larger the particle diameter, the lower the pressure drop, the longer the reactors and the lower the number of reactors. For this parameter, an optimum has to be found between diffusion limitations inside the particles and costs.
- Increase the cycle time. This can be demonstrated on the basis of simple mathematical considerations, as follows. The total reactor volume is dependent on the amount of syngas that has to be processed during the time of a cycle, as
162 |
shown in equation 6.1. The total reactor volume is by definition equal to the number of reactors multiplied by the volume of one reactor (equation 7.3). Based on a fixed L/d-ratio, equation 7.3 can be rearranged so that a function for the length can be formulated, which is given in equation 7.4. Equation 7.5 demonstrates that the superficial gas velocity is equal to the mass flow rate (kg/s) divided by the gas density, the reactor cross section and the number of reactors in a certain step (φstepNR, where φstep represents the portion of reactors operating in the considered step). When these terms are implemented in the Ergun equation, it is shown how the cycle time and the number of reactors correlate with the pressure drop (equation 7.6). Considering only the design of the reactors, many parameters are constant (c1, c2, c1’ and c2’) and then it turns out that in case the cycle time decreases, the number of reactors has to be increased.
22
,4
CO H cycleR R
mol oxygen
M MV N d L CO H cycle2M MCO H 2MH 7.3
2
23
, 4
CO H cycle
Rmol oxygen
M ML
NdL
2CO H cycM MCO H 7.4
22 2
, ,4
gcycleCO H CO Hcycle
g step R g step g stepmol oxygen mol oxygen
m m m Lvd M M M MN
L
m m m L
2 cycleCO HCO H C2M M M MCO H CO HCO H 2 cycleM MM MH COHH CO2
7.5
2
22 3 3
' '2 31 2
1 2 2 1/3 2/3
1 1150 1.75
g gg gg g
pp g g
cycle cycle Rcycle cycle R
p v L v Ldd
c cL Lc cNN
7.6
The influence of the pressure drop and the cycle time is further discussed in the next sections.
7. Comparison of CLC in packed bed and fluidized bed reactors |163
7
7.2.4. Effect of compression ratio on number of reactors
Based on the strategy described in section 6.3.2, the number of reactors has been calculated for each step of the CLC process and for different compression ratios. In Figure 7.6 it is reported that less reactors are needed, if the compression ratio is increased for a specific pressure drop and L/d-ratio. In some cases, quite a large number of reactors is required and therefore it is important to select the operating conditions carefully. In all the considered cases, the total cycle time is 20 min. So, the amount of oxygen carrier, and thus the reactor volume, is always the same. Hence, when the reactor number is lower, the reactors are larger.
When the L/d-ratio is reduced from 4 to 2, the reactor diameter is larger and therefore fewer reactors are needed. In general, lower L/d ratios seem preferable to limit the total number of reactors and the ancillary components (e.g. piping, valves). On the other hand, the optimum L/d-ratio depends on the reactor production method, on how well the flow is distributed over the cross-section when L/d decreases and on the effects on temperature equalization of the outlet gas as discussed in section 6.3.3.
In general, 50-65% of the reactors are in the heat removal step, 2 reactors are in purge and about 20% are in reduction and about 20% are oxidation.
164 |
13 14 15 16 17 18 19 20
20
40
60
80
100
120
140
p/p=5% p/p=8% p/p=15%
Compression ratio (-)
Num
ber o
f rea
ctor
s
3
4
5
6
7
8
9
10
Rea
ctor
leng
th fo
r L/D
=4 (m
)
L/d=4
L/d=2
Figure 7.6: The number of reactors as a function of the compression ratio with different L/d-ratios and pressure drops (cycle time is 20 min and 10mm particle diameter). Because the cycle time is constant, the total reactor volume is constant at 1183 m3. For clarity, the reactor length is only shown for L/d-ratio of 4.
0 20 40 60 80 100
20
40
60
80
100
120
0 1183 2367 3550 4734 5917
Total reactor volume (m3)
L/d-ratio=2
L/d-ratio=4
Num
ber o
f rea
ctor
s
Cycle time (min)
4
6
8
10
12
14
16
18 R
eact
or le
ngth
(m)
Figure 7.7: The influence of the cycle time on the required reactor number and the reactor length (with L/d-ratios of 2 and 4).
7. Comparison of CLC in packed bed and fluidized bed reactors |165
7
7.2.5. Influence of cycle time In the previous calculation, a total cycle time was adopted of 20 min. The same procedure has been carried out for different cycle times and an overview of the results is given in Figure 7.7. As explained in section 7.2.3, the longer the selected cycle time, the lower the number of reactors, the lower the number of high temperature valves required and the larger the reactors. Also in this case a fixed L/d-ratio of 4 was selected. Some examples of the reactor sizes are given in Figure 7.7.
The solids inventory is dependent on the cycle time. If the cycle time is increased, more oxygen has to be available to process syngas for a longer time. Hence, the longer the cycle time, the higher the solid inventory. A cycle time of 20 min corresponds to a total solids inventory of 1200 ton (1750 kg solid/MWth = 261 kg Ni/MWth), which means that 180 ton nickel is required. In case of a short cycle time, the high temperature valves have to be switched more often. Therefore, the lifetime of the valves is expected to be lower.
7.2.6. Preliminary investment costs estimation The impact of the cycle time, the L/d-ratio and the oxygen carrier on the initial investment costs has been evaluated by a simple preliminary economic analysis. The calculation method for the high temperature valve costs and the reactor costs has been described in section 6.3.2. The different reactor sizes and assumed costs are listed in Table 7.1. Based on these numbers, the investment costs have been estimated for different cycle times and this is illustrated in Figure 7.8. It can be concluded that the Ni-carrier is, by far, the most expensive part of the system. This leads to the design guideline to reduce the cycle time as much as possible.
Table 7.1: The considered costs of the reactors components for different cycle times and L/d.
Cycle time, min 10 20 40 60 90 20 (L/d=2)
L/d-ratio 4 4 4 4 4 2 Number of reactors 123 67 39 29 22 21 Reactor length, m 4.6 7.1 10.7 13.5 17 6.6 Reactor inner diameter, m 1.15 1.78 2.68 3.38 4.25 3.3 Thickness refractory, mm 306 383 495 579 692 182 Thickness steel, mm 16 23 33 40 50 33 Reactor costs, k€/reactor 12 37 110 208 397 60 High temperature valves, k€/valve 150 216 299 357 421 433 Oxygen carrier, €/ton 6,000-50,000
166 |
10 min. 20 min. 40 min. 60 min. 90 min. 20 (L/D=2)0
200
400
600
800
0
71
142
213
284
Initi
al in
vest
men
t cos
ts (M
€)
Initi
al in
vest
men
t cos
ts (€
/kW
)
Cycle time (min)
High temperature valve costs Oxygen carrier costs Reactor costs
Figure 7.8: Influence of the cycle time on the costs in case of NiO/Al2O3 (estimated on € 50,000/ton NiO/Al2O3).
10 min. 20 min. 40 min. 60 min. 90 min. 20 (L/D=2)0
20
40
60
80
100
120
140
0
7
14
21
28
35
43
50In
itial
inve
stm
ent c
osts
(M€)
Initi
al in
vest
men
t cos
ts (€
/kW
)
Cycle time (min)
High temperature valve costs Oxygen carrier costs Reactor costs
Figure 7.9: Influence of the cycle time on the costs in case of a cheap oxygen carrier (€ 6,000/ton oxygen carrier).
7. Comparison of CLC in packed bed and fluidized bed reactors |167
7
Nickel is considered as one of the most expensive oxygen carriers and cheaper alternatives might be selected. For example if a natural material is selected (cost estimated at € 6,000/ton, mainly costs of palletizing (“Demonstration of a cost-effective CO2 capture technology (DemoCLoCK project),” n.d.)), the contribution of the high temperature valves on the total cost becomes important. Reducing the cycle time means increasing the number of reactors and then the high temperature valves are the most expensive parts of the system. In Figure 7.9 it is shown that if the cycle time is increased, the oxygen carrier costs become more dominant, because they increase linearly with the cycle time, while the number of reactors decreases (and thus the number of high temperature valves). So, in case of an expensive oxygen carrier (like nickel), the solid inventory has to be reduced as much as possible and in case of a cheaper material it is also important to keep the number of reactors as low as possible.
The case with a 20 min cycle time is also shown for a L/d-ratio of 2 (instead of 4) in Figure 7.9. In that case, the system can be operated with a smaller number of reactors (and high temperature valves) and then the initial investment costs can be further reduced.
The specific investment costs for an IGCC power plant are around € 1950/kWnet and for a plant with CO2 capture, these costs rise to around € 2650/kWnet (EBTF et al., 2011). The initial investment costs of the packed bed reactor systems for the CLC are lower than the 700 €/kWnet. Thus, the order of magnitude of the reactor system costs is acceptable compared to the total plant cost. However, to draw more accurate conclusions a detailed economic evaluation has to be carried out on the complete plant, which also includes the operational costs and a sensitivity analysis on the most critical components. An important component is the high temperature valve system, which cost also depends on the operational cycle time. From this simple preliminary analysis, it can be concluded that the solid inventory is very important for the costs in case nickel is used as oxygen carrier. The oxygen carrier also needs to be replaced after a while and this will have an effect on the operational costs. If a cheaper oxygen carrier is selected, the number of reactors also becomes an important design factor to keep the number of high temperature valve systems as low as possible.
Because no cost information was available about cyclones, the investment cost has only been estimated for the packed bed configuration.
168 |
7.3. Fluidized bed configuration
7.3.1. Reactors operation and solid inventory Because many papers have been published about circulating fluidized bed systems, the discussion is less extensive than for the packed bed case. An ideal fluidized bed system is considered, assuming that it is working at 20 bar without producing fines. This is however not possible with current available oxygen carriers and cyclone systems.
In the configuration with circulating fluidized beds the temperature rise inside the reactor is independent of the active weight content, since it can be controlled by gas excess (or by immersed heat transfer surfaces in case of steam generation). As a matter of fact, thanks to the good mixing of the bed material, small temperature gradients prevail in the reactors without risk of hot spots formation. Therefore, a higher active weight content could be selected than 19wt% NiO on Al2O3, optimized on chemical and mechanical stability rather than imposed by temperature rise control. In the literature, an active weight content of 40wt% is commonly selected (Adanez et al., 2012).
The conversion of the oxygen carrier is something that distinguishes fluidized beds and packed beds. In the packed bed configuration the carrier is almost fully converted (ΔXs=1), while in the fluidized bed the oxygen carried by the metal oxide is utilized only for a certain extent. On the one hand, due to the large air excess in the air reactor, the metal is expected to leave the air reactor in fully oxidized state. Conversely, since the highest possible oxidation of the fuel needs to be achieved in the fuel reactor, the oxygen carrier cannot be completely reduced in fluidized bed fuel reactor and a certain amount of it needs to be kept in the oxidized state. The composition of the solids inside the fuel reactor determines the reactivity of the bed and the solids inventory required to reach full conversion of the syngas. Based on the selected oxygen carrier conversion and on the gas flow rates, the needed solids circulation rate can be calculated. Typically, the solid conversion ΔXs is maintained below 0.5 to keep a good reactivity of the fuel reactor bed. At elevated pressures, the gas transport capacity of the solids is limited by hydrodynamics, because the reactors have a smaller cross sectional area (Abad et al., 2007). Owing to the low transport of solids, a relatively large ΔXs should be selected, or alternative solid transport systems between the reactors should be employed to keep a sufficiently high solids circulation rate.
Abad et al. and Mattisson et al. calculated the solids inventories at around 70 kg/MWth (Abad et al., 2007; Mattisson et al., 2007). This value is calculated based on a ΔXs of 0.5 and the kinetics at 1200 °C and 20 bar. The calculation only includes the air and the fuel
7. Comparison of CLC in packed bed and fluidized bed reactors |169
7
reactors and thus the solids inventory of the total system (including cyclones and loop seals) is higher. The solids inventory in the fuel and air reactor corresponds to 375 molNiO/MWth. Considering that 1 MWth thermal input corresponds to 3.7 molCO-
H2/s (with CO/H2=2.46 as in our IGCC), it means that the total residence time of the oxygen carrier in the two reactors is 51 s (assuming ΔXs=0.5).
Another solids inventory was found based on experiments carried out by Kolbitsch et al (Kolbitsch et al., 2010). During small scale experiments (65 kW) a solids inventory of 450 kg/MWth (141 kg Ni/MWth) was sufficient to convert H2 with nickel oxide (40wt% NiO on NiAl2O4) (Kolbitsch et al., 2010, 2009). A total residence time in the system of 130-350 s was reached. The ΔXs was about 0.1 (Kolbitsch et al., 2010), so quite low in comparison with the above described situation.
In both cases, a lower active weight content was reported than was assumed for the packed bed case. The required solids inventory is lower in the case of circulating fluidized beds, because the residence time of the solids is lower (a couple of minutes) than the time of a packed bed cycle. The oxygen carriers in the packed bed reactors are most of the time not in a reacting mode (reduction and oxidation), but in the heat removal and purge mode. These steps, not needed for the fluidized bed reactors, make the total cycle time longer.
The difference in solids inventory might be compensated to a certain extent by a longer lifetime. The carriers in the packed bed are exposed to much less mechanical stresses and therefore the lifetime might be longer. At this moment, the difference in lifetime in both reactors is unclear and needs to be further explored experimentally.
7.3.2. IGCLC configuration with circulating fluidized beds For the interconnected fluidized bed case, the IGCLC power plant configuration is slightly different from the case with packed bed reactors on four aspects:
First, the temperature of the fuel reactor is close to the air reactor temperature, because of the circulation of the solids that transfer a large amount of heat between the two reactors. Because of the high fuel reactor temperature and the highly uniform temperature in the reactor, the equilibrium of the Boudouard reaction is more on the CO side (no carbon formation). In addition, the oxygen carrier is not fully reduced and well mixed and some oxidized carrier is hence expected to be available in each zone of the fuel reactor. Therefore, carbon deposition is not considered to be a critical issue in fluidized beds and fuel dilution with steam or recycled CO2 is not needed in this case.
170 |
Second, the oxidation and the heat removal steps are integrated in one step. During oxidation, a large excess of air is fed at 440 °C, only a fraction of the oxygen is reacting (about 25%) and a stream is obtained at 1200 °C. This temperature is reached by tuning the air gas flow rate. Only one compressor is required to obtain the air at 20 bar. So, no additional blower has to be installed for the oxidation.
Third, because the solids are transferred, the reactors do not have to be purged, so no purge flow is required. Actually, some steam or recycled CO2 should be used in the reactors as sealing gas and to assist solids circulation in the loop seals. However, this flow rate is expected to be relatively small and is not taken into account in this work.
Fourth, since the CO2/H2O stream is produced at high temperature, a large amount of high pressure steam can be produced during cooling down of this stream. To produce high pressure steam also low temperature heat is required. Because this low temperature heat is limited, some adaptations were made to the plant design. The saturator outlet temperature was reduced to 123 °C because not much steam is required to avoid carbon deposition.
The temperature of the air reactor was set to 1200 °C, but the temperature of the fuel reactor is dependent on the solids conversion and the syngas inlet temperature. The syngas can be fed at 123 °C directly from the saturator outlet, at 300 °C by heating with high pressure saturated water or at 600 °C by feeding it to the CO2/H2O-cooler. The syngas inlet temperature effect on the temperature of the fuel reactor is shown in Figure 7.10. It is shown that the fuel reactor temperature is always around 1200 °C. In section 7.2.1 it was discussed that the higher the fuel feeding temperature, the higher the process efficiency. When the syngas is fed at 600 °C, the temperature in the fuel reactor is above the limit of 1200 °C. Therefore a syngas feeding temperature of 300 °C is selected, leading to an almost isothermal reactor system.
The pressure drop is assumed to be the same as for the packed bed case (5%). This depends on the solids inventory in the reactors, the reactor cross section and the pressure drop in the cyclones and the distribution plate. As was shown in section 7.2.2, the effect of the pressure drop on the process efficiency is quite small, when kept below about 8%.
7. Comparison of CLC in packed bed and fluidized bed reactors |171
7
0.0 0.2 0.4 0.6 0.8 1.01000
1050
1100
1150
1200
1250
1300
Tsyngas,in=600°C
Tsyngas,in=300°C
T CO
2/H
2O (°
C)
Xsolid
Tsyngas,in=120°C
Figure 7.10: The fuel reactor temperature as function of the solids conversion and the fuel inlet temperature.
In case NiO is used as oxygen carrier, the conversion of the fuel is limited by thermodynamics. In that case, the selectivity of CO oxidation is 0.970 and the selectivity of H2 oxidation is 0.985. It is assumed that the remaining CO and H2 is combusted downstream with a stoichiometric amount of O2 produced in the ASU (with 95% purity). The temperature of the CO2/H2O stream after combustion with O2 is 1295 °C. All these assumptions lead to a mass balance as shown in Table 7.2. In this table, streams related to the gasification island are not included, since they do not differ from the plant with packed bed CLC reactors. The net efficiency of this process is 41.73% of LHV. More details about the energy balance are given in section 7.4.
Contrary to the packed bed configuration, the performance and behavior of the fluidized bed reactors has not been modeled in detail. Ideal assumptions have been considered, not considering some effects that might reduce the process efficiency. First of all, it is assumed that the gases are always converted as much as thermodynamically possible, which is achievable with a proper inventory, gas residence time and ΔXs. Second, in circulating fluidized beds, also some CO2 leakage could occur from the fuel reactor to the air reactor and some N2 might be transported in the other direction (CO2 dilution) (Abad et al., 2006). Also these effects, which could be minimized by utilizing small flow rates of steam as sealing gas, are not accounted for.
172 |
Table 7.2: Comparison streams in packed bed and fluidized bed configuration (stream numbers, #, refer to Figure 7.1).
Packed bed configuration Fluidized bed configuration stream #
M, kg/s
T, °C
p, bar Composition M,
kg/s T, °C
p, bar Composition
1 116.3 600 20.1
Ar: 0.55%, CO: 32.09%, CO2:
5.28%, H2: 13.07%, H2O: 48.29%, N2:
0.72%
72.7 300 20.1
Ar: 0.97%, CO: 56.41%, CO2:
9.27%, H2: 22.97%, H2O: 9.11%, N2:
1.27%
2 156.9 832 19.0 Ar: 0.55%, CO2: 37.37%, H2O:
61.36%, N2: 0.72% 113.3 1295 19.0
Ar: 1.01%, CO2: 65.64%, H2O:
32.07%, N2: 1.29% 3 156.9 136 18.0 Same as #2 113.3 138 18.0 Same as #2
4 81.5 28 110.0 96.70% CO2 81.5 28 110.0 96.63% CO2
5 786.2 15 1.0
Air: Ar: 0.92%, CO2: 0.03%, H2O: 1.03%, N2: 77.28%,
O2: 20.73%
695.3 15 1.0 Air
6 729.8 438 20.0 Air 645.1 438 20.0 Air 7 176.6 448 21.0 Air - - - -
8 698.4 466 20.0
Depleted air: Ar 0.94%, CO2:0.03%, H2O: 1.06%, N2:
81.86%, O2: 16.11%
- - - -
9 18.4 22 1.1 N2 - - - - 10 18.4 478 20.4 N2 - - - -
11 707.6 1199 19.0 same as #8 605.5 1200 19.0
Depleted air: Ar 0.97%, CO2: 0.03%,
H2O 1.10%, N2: 81.81%, O2:
16.09%
12 764.0 486 1.0
Depleted air: Ar 0.94%, CO2: 0.03%,
H2O: 1.06%, N2: 81.52%, O2: 16.45%
655.8 487 1.0
Depleted air: Ar 0.97%, CO2: 0.03%,
H2O 1.09%, N2: 81.46%, O2:
16.45% 13 764.0 92 1.0 Same as #12 655.8 86 1.0 Same as #12 14 120.7 15 1.0 Air 125.2 15 1.0 Air 15 88.6 565 133.9 Steam 133.1 565 133.9 Steam 16 129.5 527 133.9 Steam 156.0 547 133.9 Steam 17 129.5 333 36.0 Steam 156.0 349 36.0 Steam 18 142.6 458 33.1 Steam 160.4 458 33.1 Steam 19 33.4 395 21.6 Steam - - - - 20 37.7 300 3.5 Steam 0.3 300 3.5 Steam 21 146.8 32 0.05 Steam 160.7 32 0.05 Steam
7. Comparison of CLC in packed bed and fluidized bed reactors |173
7
7.4. Comparison of packed bed and fluidized bed CLC In the previous sections, the process efficiency for CLC with packed beds and circulating fluidized beds has been determined. In case of the packed bed system, the reduction could be carried out at lower temperatures. Therefore, the CO2/H2O stream was produced at a lower temperature (832 °C), but a large steam flow had to be mixed to reach the thermodynamic equilibrium to prevent carbon deposition at 450 °C and higher temperatures. This has a large impact on the process efficiency, because in case an oxygen carrier is selected with minimal activity for the Boudouard reaction, a smaller amount of steam is sufficient and then the process efficiency increases from 41.05% to 42.23% of LHV.
In case of the interconnected fluidized bed system, syngas dilution is not necessary to avoid carbon deposition. The fuel reactor operates at higher temperature and a higher fraction of the thermal input is hence available in the fuel reactor off-gas, which is recovered by raising steam. Therefore, a lower portion of heat is converted by the gas turbine based combined cycle. In addition, the selected oxygen carrier influences the process performance, especially if it introduces fuel conversion limitations. As a matter of fact, one of the sources of efficiency penalty in this plant is the incomplete fuel conversion due to thermodynamic limitations entailed by using Ni as oxygen carrier. In case an oxygen carrier allowing complete syngas oxidation is selected instead of NiO, the efficiency could be increased by 0.4% points, from 41.37 to 41.78%.
The energy balances of the assessed cases are reported in Table 7.3. The results of the packed bed and the fluidized bed Ni-based plants are reported in the second and the third column. The different power share of the two cases is evident, with the packed bed case generating more electricity through the gas turbine (55% of the gross power) and less by the heat recovery steam cycle (45%). This is related to the fact that CO2 and H2O are produced at lower temperature. In the fluidized bed case, a higher portion of the gross power (about 53%) is produced by the steam cycle. Mainly due to the loss associated with steam dilution in the packed bed case, the process efficiency is higher in the fluidized bed case by about 0.4 percentage points.
Moreover, two ideal cases have been compared with ideal oxygen carriers (fourth and fifth column in Table 7.3). Ideal means that in the packed bed case, a H2O/CO-ratio of 0.37 is sufficient to avoid a large extent of carbon deposition (i.e. with no further steam dilution after the saturator). In the fluidized beds, the ideal OC allows full conversion of the fuel. As a consequence of these improvements, net efficiency improves by 1.1 and 0.4 percentage points in packed bed and fluidized bed cases respectively. In this
174 |
scenario, the ideal packed bed case is slightly (0.4% point) more efficient than the ideal fluidized bed case.
As far as CO2 emissions are concerned, CLC-based plants allow a very high CO2 capture rate, since the CO2 is lost only from the lock hopper-based coal feeding system and from syngas combustion for coal drying. Additional CO2 losses would occur from the CO2 purification section if the 96.7% purity achieved were not sufficient for the storage site and the transport infrastructure.
Because the process efficiency does not depend significantly on the reactor type in which CLC is carried out, it is not expected that the process efficiency is a decisive factor in the reactor type selection. Before the reactor types could be implemented in practice, a critical unit has to be developed for high temperature/pressure application. For the packed beds, the most challenging part is probably represented by the high temperature valve system, while for the fluidized bed, hot gas filtering, loop seals and control system allowing for stable solids circulation at high pressure need to be developed. The level of development achievable for these units, their availability, operability and cost will most likely determine which technology should be used to carry out CLC integrated with the IGCC power plant.
7. Comparison of CLC in packed bed and fluidized bed reactors |175
7
Table 7.3: Energy balances of the different cases considered for the packed bed and the fluidized bed.
Power PB with NiO
CFB with NiO
Ideal PB, no steam
added
Ideal CFB, full gas
conversion Heat input LHV, MWLHV 853.9 853.9 853.9 853.9 Gas turbine, MWe 225.1* 192.1 232.2* 197.9 Heat Recovery Steam Cycle, MWe 183.0 220.0 185.7 216.3 Gross power output, MWe 408.1 412.1 417.9 414.2 Syngas blower, MWe -0.8 -0.8 -0.8 -0.8 N2 compressor, MWe ASU, MWe -33.9 -35.1 -33.9 -33.9 Lock hoppers CO2 compressor, MWe -3.1 -3.1 -3.1 -3.1 Acid Gas Removal, MWe -0.4 -0.4 -0.4 -0.4 CO2 compressor, MWe -11.0 -11.0 -11.0 -11.0 N2 intercooled compressor gasifier, MWe -1.3 -1.3 -1.3 -1.3 Heat of rejection, MWe -3.6 -3.7 -3.4 -3.6 Other auxiliaries, BOP, MWe -3.4 -3.4 -3.4 -3.4 Net power generated, MWe 350.6 353.3 360.6 356.7 LHV efficiency, % 41.05 41.37 42.23 41.78 CO2 capture efficiency, % 97.1 97.1 97.1 97.1 CO2 purity, % 96.7 96.6 96.7 96.7 CO2 emission, kg CO2 emitted/MWhe 24.7 24.5 24.0 24.3 CO2 avoided, % 96.8 96.8 96.9 96.8 SPECCA, MJ LHV/kg CO2 1.08 0.99 0.75 0.88 * Gas turbine power includes consumption of air blower and nitrogen compressor for purge.
7.5. Boosting up the process efficiency by changing the desulfurization method
7.5.1. Introduction Sulfur compounds can be separated from the fuel by several methods. The conventional method in an IGCC power plant (that has been considered in chapter 6 and the previous sections), is first to produce the syngas, cool it down and then remove the sulfur by absorption with Selexol, cold gas desulfurization (CGD). A disadvantage of this method is that the separation process is carried out close to the ambient temperature, which implies that the syngas has to be cooled down to low temperature, before it is heated up and sent to the CLC reactors. This represents an energy penalty and a capital cost for the different heat exchangers used and for the higher duty of the heat rejection system.
176 |
Two alternative methods are studied in this work, in which the cooling down and heating up are prevented and therefore a higher process efficiency might be achieved. A first alternative is to apply hot gas desulfurization (HGD) with ZnO (Giuffrida et al., 2010, 2013). It has been reported that an efficiency gain of 2.5% points can be obtained if HGD is implemented in an IGCC instead of CGD (Giuffrida et al., 2010). Hence, a higher process efficiency can also be expected for the IGCLC process. HGD has been studied in combination with CLC (in combination with H2 production) by Sorgenfrei (Sorgenfrei and Tsatsaronis, 2013), but the authors did not compare the performances with cold gas desulfurization.
The second alternative considered here is to carry out the desulfurization downstream of the CLC reactors. In this case, the syngas stream that is fed to the CLC reactors, contains sulfur, so the oxygen carrier needs to be sulfur tolerant. In the CLC fuel/reduction reactor, H2S can be converted to SO2 or it leaves the reactor as H2S. Afterwards, the CO2 stream is desulfurized by FGD (flue gas desulfurization) and compressed. Another possibility that could occur in the reduction reactor is the reaction between the sulfur compounds and the oxygen carrier forming metal sulfides (MeSx) or sulfates (MexSO4), like Ni3S2. During oxidation either SO2 is formed and then the O2-depleted air stream needs to be purified as well or the sulfur remains on the oxygen carrier, which might then become deactivated. The oxidation and reduction reaction rates decrease in the presence of H2S for nickel, copper, iron and manganese oxygen carriers, which has been proven by thermo gravimetric analysis at 800-900 °C (Tian et al., 2009). Hence, the MeSx-formation should be circumvented. Several thermodynamics studies have been carried out on the formation of sulfur compounds on oxygen carriers. It has been demonstrated that this reaction can be suppressed at high temperature, low pressure and high oxygen content (on oxygen carrier or in fuel (CO2, H2O)) (Jerndal et al., 2006; Wang et al., 2008).
In this section, the process efficiency of the three desulfurization methods are compared considering both (CFB and PBR) reactor configurations. In these IGCLC configurations, hot air is produced in the CLC reactors at 1200 °C and 19 bar, which is fed to the gas turbine and the heat recovery steam generator. First, sulfur tolerance of the oxygen carriers is discussed and then the different processes are compared.
7.5.2. Model and assumptions The same power plant model is used as discussed in section 6.2.1, while three different desulfurization methods are used. A power plant overview is given in Figure 7.11. The following section contains a description of the different desulfurization methods that have been integrated in the power plant.
7. Comparison of CLC in packed bed and fluidized bed reactors |177
7
Figure 7.11: Simplified IGCLC power plant scheme in which the effect of the desulfurization method on the process efficiency is determined (method A: CGD, method B: HGD, method C: post-CLC desulfurization).
Method A: Cold Gas Desulfurization (CGD) The syngas that has been cooled down to 200 °C (#2) contains some H2S and COS. Because Selexol is only selective towards H2S, COS is converted to H2S in a catalytic packed bed that operates at 180 °C before the syngas stream is desulfurized. Then the stream is further cooled down to 35 °C, at which the desulfurization is carried out with Selexol. The sulfur is recovered from the Selexol in a reboiler, which is heated by low pressure steam. Afterwards, elementary sulfur is obtained via the Claus process.
After the Selexol process, the desulfurized syngas is depressurized to 21.6 bar and then sent to the saturator to increase the steam content and to heat up the stream again. Subsequently, it is heated up to 300 °C with high pressure saturated water and in the
178 |
packed bed case, mixed with steam to avoid carbon deposition and further heated up to 600 °C by the downstream CO2/H2O-heat exchanger. Then the desulfurized syngas is fed to the CLC reactors (#3).
Method B: Hot Gas Desulfurization (HGD) In HGD the H2S and COS are separated from syngas by reacting it with an oxygen carrier, ZnO/TiO2. The process consists of a circulating fluidized bed system containing a desulfurizer and a regenerator, as illustrated in Figure 7.12. In the desulfurizer, the H2S and COS reacts with the ZnO and ZnS is formed, according to equation 7.7 (Giuffrida et al., 2010). The ZnS is reacted back to ZnO in the regenerator and here SO2 is formed, according to equation 7.8 (Giuffrida et al., 2010). The SO2 is separated afterwards.
Desulfurizator: 2 2 RZnO H S ZnS H O H 76.9 kJ/mol 7.7
Regenerator: 2 2 R3ZnS O ZnO SO H 441.3 kJ/mol2
7.8
The first difference with the CGD case is that the syngas is quenched by a stream with a higher temperature (364 °C instead of 200 °C) and therefore a larger syngas recycle stream is required to reach 900 °C after quench. A lower recycle temperature could be preferable to reduce the syngas cooling ducts, but would somewhat reduce the overall efficiency, since it would increase the amount of heat removed at the lower temperature levels. After the quench, the fines are filtered from the syngas and then it is fed to the HGD-unit, where the H2S and COS reacts with ZnO. The desulfurizer operates adiabatically at 400 °C and this temperature is controlled by the syngas stream, which has a temperature of 386 °C. After desulfurization, the syngas is sent to a filter again to avoid the entrainment of Zn-sorbent particles. A small amount of CO2 is therefore also fed to clean the filters. After filtration, a small amount of syngas is combusted to dry the coal to the target moisture content of 2 wt%. Before the syngas is fed to the reduction reactor for CLC, the steam content is increased (in case of the packed bed configuration) to avoid carbon deposition. The amount of steam that is mixed is larger than in the CGD-case, because in this configuration, there is no saturator that could increase the moisture content at low temperature by means of warm (liquid) water.
7. Comparison of CLC in packed bed and fluidized bed reactors |179
7
Figure 7.12: Schematic overview of the hot gas desulfurization process (method B).
The ZnS formed in the desulfurizer is regenerated with air that is diluted with N2 taken from the ASU. During the ZnS oxidation ZnSO4 might be formed, but this reaction is circumvented by keeping the O2 concentration low (2%) in the regeneration gas stream. While the proper O2 dilution level to avoid ZnSO4 is highly uncertain, it was demonstrated that the effect of O2 concentration on plant efficiency is very low (Giuffrida et al., 2010). The temperature of the regenerator is kept at 750 °C. A ZnS/ZnO ratio of 0.1 is assumed for the solids that exit the reactor. After regeneration, the gas is expanded to ambient pressure and then the SO2 is separated by the wet lime-limestone desulfurization process (FGD). An electric consumption of 1483 kWh/ton SO2 has been assumed for the FGD process (EBTF et al., 2011). The assumptions for the HGD process are listed in Table 7.4.
Method C: Post CLC desulfurization If the H2S and COS are oxidized by the oxygen carrier during the CLC process, SO2 is formed in the CLC reactors and this can be separated after the CLC process. In that case, the CO2/H2O/SO2-stream is cooled down and subsequently, a flue gas desulfurization (FGD) is carried out. In this case, FGD operates at unusual conditions, featuring high pressure and highly CO2-concentrated environment. Suitable operating parameters are therefore uncertain. However, the same electric consumption of a conventional FGD has been assumed as in the HGD case. Another option, which
40 b
ar
N2+2%O2
20 bar
Flue Gas Desulfurization N2+O2
Desu
lfuriz
ator
Rege
nera
tor
Syngas, #2
ZnOTiO2
(ZnS)
ZnSZnOTiO2
Filter
Sulfur
CO2
Syngas (excl. H2S), #3
180 |
should be verified according to the CO2 specifications of the storage site, is SO2 co-sequestration. In case this is possible, consumption for SO2 absorption from the CO2-rich stream could be avoided.
Table 7.4: List of assumptions regarding the desulfurization methods.
Method A: CGD (Selexol solvent) Temperature of absorption tower 35 °C Syngas pressure loss 1% Heat of low pressure steam for sour water stripper 20.95 MJth/kgH2S Sulfur removal and recovery auxiliaries 538.2 kWhe/ton H2S Method B: HGD ZnS/ZnO-ratio after regenerator 0.1 TiO2 (inert) content 50mol% Desulfurization temperature 400 °C H2S conversion to SO2 100% Regenerator temperature 750 °C O2 content of regeneration stream 2% Polytropic efficiency of regeneration stream compressor 85% Polytropic efficiency of regenerator gas expander 92% Mechanical efficiency of the turbomachines 98.3% Electricity consumption for SO2 separation 1483 kWh/ton SO2 Method C: Post-CLC desulfurization Electricity consumption for SO2 separation 1483 kWh/ton SO2
7.5.3. Thermodynamics of oxygen carriers and sulfur The post-CLC desulfurization process can only be carried out if no sulfur species are formed on the oxygen carrier, which depends on the oxygen carrier active metal and the process operating conditions. The process conditions depend mainly on the selected CLC reactor type, circulating fluidized bed (CFB) or dynamically operated packed beds (PBR). In the CFB-reactors, the temperature in the fuel reactor is close to the temperature of the air reactor due to the solids circulation. So, the temperature in the fuel reactor is high (around 1200 °C) and in general, the oxygen carrier has a high oxygen availability. In the PBRs the reduction reactor temperature is usually lower; the temperature at the inlet is equal to the temperature of the fuel (300-600 °C) and the temperature of the rest of the reactor could be 450-1200 °C and this depends on the heat management strategy (Spallina et al., 2013). The formation of the sulfates and sulfides is thermodynamically favored at lower temperature. The pressure and the steam content have an influence on the equilibrium, but this effect is negligibly small.
7. Comparison of CLC in packed bed and fluidized bed reactors |181
7
In this section, it is demonstrated to what extent sulfur species can be expected on nickel, copper, iron and manganese based oxygen carriers in packed bed reactors and in circulating fluidized bed reactors based on thermodynamics. The thermodynamics are calculated with HSC Chemistry 5.1 (2013). The oxygen availability is defined in equation 7.9. In case the oxygen availability is 1, the stoichiometric amount of oxygen is available for the combustion of the fuel, including the H2S. The sulfur components that are included in the calculations are listed in Table 7.5.
oxygen available on oxygen carrieroxygen availabilityamount of oxygen required for fuel combustion
7.9
Table 7.5: The sulfur components that are included in the thermodynamic calculations.
Nickel H2S, SO2, Ni3S2, NiS, NiS2, Ni3S4, NiSO4 Copper H2S, SO2, Cu2S, CuS, CuSO4, Cu2SO4 Iron H2S, SO2, FeS, Fe0.877S, FeS2, FeSO4, Fe2(SO4)3 Manganese H2S, SO2, MnS, MnSO4,
Thermodynamics in packed bed reactors The syngas used for the reduction in packed bed reactors is fed at 600 °C, which is taken as reference temperature for the thermodynamic calculations. At the packed bed reactor inlet, the syngas is in contact with a reduced oxygen carrier while inside the packed bed, it is in contact with an oxidized carrier. For that reason, the effect of the different states of oxidation of the oxygen carrier is considered.
In Figure 7.13 the formation of sulfur species on different OCs is shown as a function of the oxygen availability. A total flow of 5 mol/s of H2S (980 ppm) is fed to the reactor and this number is used as reference. It is shown that in case nickel-based oxygen carriers are used, some Ni3S2 is formed for any oxygen availability value. With manganese or iron, the MnS or FeS could be formed. However, the formation of these species can be avoided, if the H2S fraction is limited to 160 ppm. For copper-based oxygen carriers, Cu2S is formed. Hence, based on thermodynamics it is demonstrated that with all the oxygen carriers some sulfur species could be formed on the carrier.
182 |
0.0 0.2 0.4 0.6 0.8 1.0
0
1
2
3
4
5
SO2
Ni3S2H2S
Sul
fur s
peci
esat
equ
ilibriu
m (m
ol/s
)
Oxygen availability (-)
0.0 0.2 0.4 0.6 0.8 1.0
0
1
2
3
4
5 Cu2S
SO2H2SSul
fur s
peci
esat
equ
ilibriu
m (m
ol/s
)
Oxygen availability (-)
0.0 0.2 0.4 0.6 0.8 1.0
0
1
2
3
4
5
FeS
H2S
SO2
Sul
fur s
peci
esat
equ
ilibriu
m (m
ol/s
)
Oxygen availability (-)
0.0 0.2 0.4 0.6 0.8 1.0
0
1
2
3
4
5
MnS SO2
H2S
Sul
fur s
peci
esat
equ
ilibriu
m (m
ol/s
)
Oxygen availability (-)
Figure 7.13: The formation of sulfides and sulfates on the nickel, copper, iron and manganese based oxygen carrier in the packed bed reactor at 600 °C and 20 bar. In cases of iron and manganese, the Fe0.947O/Fe2O3 and the MnO/Mn3O4 redox couples are considered.
However, the kinetics might be slow and therefore the application for post-CLC desulfurization in packed beds can still be possible (but this should be experimentally demonstrated, which is out of the scope of the present study). For the manganese and the iron post-CLC desulfurization might be possible in case of a lower H2S content.
Thermodynamics in circulating fluidized bed reactors In the fluidized bed reactors, the temperature is higher (1200 °C) and oxidized oxygen carriers are available all around the fuel reactor thanks to good mixing of the solids. So, the oxygen availability is high, but might be lower locally, where some MeS might be formed in case of fast kinetics. However, the risk of MeS formation is much lower in circulated fluidized bed reactors (CFB) and might be suppressed by adapting the solid circulation rate and by good mixing of the solids. Similar thermodynamic calculations have been carried out as for the packed bed case. In the CFB-case, a smaller amount of steam is required to be mixed with the fuel to avoid carbon deposition and therefore the H2S content is higher, 1830 ppm.
7. Comparison of CLC in packed bed and fluidized bed reactors |183
7
0.4 0.6 0.8 1.0 1.2 1.4
0
1
2
3
4
5
Ni3S2
SO2
Sul
fur s
peci
esat
equ
ilibriu
m (m
ol/s
)
Oxygen availability (-)
H2S
0.4 0.6 0.8 1.0 1.2 1.4
0
1
2
3
4
5
FeS
SO2
H2S
Sul
fur s
peci
esat
equ
ilibriu
m (m
ol/s
)
Oxygen availability (-)
0.4 0.6 0.8 1.0 1.2 1.4
0
1
2
3
4
5
MnS
SO2H2S
Sul
fur s
peci
es
at e
quilib
rium
(mol
/s)
Oxygen availability (-) Figure 7.14: The thermodynamic equilibriums for nickel, iron and manganese oxygen carriers in circulating fluidized bed reactors at 1200 °C and 20 bar.
Figure 7.14 reports thermodynamics calculations of possible products under these circumstances. Copper is not included in this figure, because the melting point is too low to allow operation at 1200 °C. In case of a nickel-based oxygen carrier, Ni2S3 could be formed, even at an oxygen availability of 1. It depends on the local oxygen availability and the kinetics, if it is formed. However, with a nickel based oxygen carrier no full fuel conversion can be achieved due to thermodynamic limitations (Jerndal et al., 2006). Iron and manganese based oxygen carriers could be applied in fluidized bed reactors operating at 1200 °C and 20 bar.
|18
4 Ta
ble
7.6:
Com
paris
on st
ream
s for
the
diffe
rent
des
ulfu
rizat
ion
met
hods
for C
LC in
pac
ked
bed
reac
tors
.
M
etho
d A:
CG
D
Met
hod
B: H
GD
M
etho
d C:
Pos
t-CLC
des
ulfu
rizat
ion
# M
, kg
/s
T,
°C
p,
bar
Com
posi
tion,
mol
%
M,
kg/s
T
, °C
p,
ba
r Co
mpo
sitio
n,
mol
%
M,
kg/s
T
, °C
p,
ba
r Co
mpo
sitio
n, m
ol%
1 34
.1
15
1 Co
al 34
.1
15
1 Co
al 34
.1
15
1 Co
al
2 78
.9
165
42
Ar:
0.89
%, C
O:
51.5
7%, C
O2:
8.48
%,
H2:
21.0
0%, H
2O:
16.7
5%, H
2S: 0
.16%
, N
2: 1.
16%
69.8
38
1 43
Ar:
1.04
%, C
O:
62.7
4%, C
O2:
7.30
%, H
2: 21
.90%
, H2O
: 5.
49%
, H2S
: 0.
18%
, N2:
1.35
%
69.8
39
5 43
Ar:
1.04
%, C
O:
62.7
4%, C
O2:
7.30
%,
H2:
21.9
0%, H
2O:
5.49
%, H
2S: 0
.18%
, N
2: 1.
35%
3 11
6.3
600
20
Ar:
0.55
%, C
O:
32.0
9%, C
O2:
5.28
%,
H2:
13.0
7%, H
2O:
48.2
9%, N
2: 0.
72%
118.
9 60
0 20
Ar:
0.56
%, C
O:
33.1
2%, C
O2:
4.93
%, H
2: 11
.56%
, H2O
: 49
.09%
, N2:
0.74
%
116.
4 60
0 20
Ar:
0.56
%, C
O:
33.4
5%, C
O2:
3.93
%,
H2:
11.6
8%, H
2O:
49.5
7%, H
2S: 0
.10%
, N
2: 0.
72%
4 15
6.9
832
19
Ar:
0.55
%, C
O2:
37.3
7%, H
2O:
61.3
6%, N
2: 0.
72%
15
9.5
832
19
Ar:
0.56
%, C
O2:
38.0
5%, H
2O:
60.6
5%, N
2: 0.
74%
157.
2 83
2 19
Ar:
0.56
%, C
O2:
37.3
4%, H
2O:
61.2
9%, N
2: 0.
72%
, SO
2: 0.
10%
5 81
.5
28
110
Ar:
1.44
%, C
O2:
96.7
0%, N
2: 1.
76%
81
.5
28
110
Ar:
1.44
%, C
O2:
96.7
0%, N
2: 1.
76%
81
.5
28
110
Ar:
1.44
%, C
O2:
96.7
0%, N
2: 1.
76%
6 78
6.2
15
1
Air:
Ar:
0.92
%, C
O2:
0.03
%, H
2O: 1
.03%
, N
2: 77
.28%
, O2:
20.7
3%
789.
1 15
1
Air
792.
2 15
1
Air
7 72
9.9
438
20
Air
731.
5 43
8 20
A
ir 73
5.5
438
20
Air
8
707.
6 11
99
19
Ar 0
.94%
, CO
2:0.0
3%, H
2O:
1.06
%, N
2: 81
.86%
, O
2: 16
.11%
709.
3 11
99
19
Ar:
0.94
%, C
O2:
0.03
%, H
2O:
1.06
%, N
2: 81
.85%
, O2:
16.1
2%
713.
1 11
99
19
Ar:
0.94
%, C
O2:
0.03
%, H
2O: 1
.06%
, N
2: 81
.85%
, O2:
16.1
2%
9 76
4.0
486
1
Ar 0
.94%
, CO
2: 0.
03%
, H2O
: 1.0
6%,
N2:
81.5
2%, O
2: 16
.45%
765.
8 48
6 1
Ar:
0.94
%, C
O2:
0.03
%, H
2O:
1.06
%, N
2: 81
.51%
, O2:
16.4
6%
769.
8 48
6 1
Ar:
0.94
%, C
O2:
0.03
%, H
2O: 1
.06%
, N
2: 81
.51%
, O2:
16.4
6%
10
764.
0 92
1
Sam
e as
#9
765.
8 95
1
Sam
e as
#9
769.
8 95
1
Sam
e as
#9
11
120.
7 15
1
Air
120.
7 15
1
Air
120.
7 15
1
Air
12
28.9
18
0 47
A
r: 3.
09%
, N2:
1.91
%,
O2:
95.0
0%
28.9
18
0 47
A
r: 3.
09%
, N2:
1.91
%, O
2: 95
.00%
28
.9
180
47
Ar:
3.09
%, N
2: 1.
91%
, O
2: 95
.00%
13
18.4
47
8 20
N
2 28
.9*
478
20
N2
18.4
47
8 20
N
2 14
23
.7
80
56
Sam
e as
#5
23.7
80
56
Sa
me
as #
5 23
.7
80
56
Sam
e as
#5
15
13.9
80
1
Sam
e as
#5
13.9
80
1
Sam
e as
#5
13.9
80
1
Sam
e as
#5
16
129.
5 52
7 13
4 St
eam
13
7.8
532
134
Stea
m
137.
4 53
1 13
4 St
eam
17
12
9.5
333
36
Stea
m
137.
8 33
4 36
St
eam
13
7.4
333
36
Stea
m
18
142.
6 45
8 33
St
eam
15
1.1
458
33
Stea
m
150.
9 45
8 33
St
eam
19
33
.4
395
22
Stea
m
47.1
39
5 22
St
eam
47
.1
395
22
Stea
m
20
37.7
30
0 4
Stea
m
47.8
30
0 4
Stea
m
48.5
30
0 4
Stea
m
21
146.
8 32
0
Stea
m
152.
0 32
0
Stea
m
152.
3 32
0
Stea
m
*=10
.5 k
g/s N
2 is i
nclu
ded
for t
he re
gene
ratio
n of
the Z
nS (H
GD
)
186 |
7.5.4. Results and discussion
Desulfurization for packed bed CLC with syngas For the packed bed CLC system a sufficient amount of steam has to be mixed with the fuel to prevent carbon deposition. In the CGD-configuration, the steam content in syngas is increased by the saturator, but this operation unit is not present in the two alternative cases. Therefore, more steam has to be mixed with the syngas (47 instead of 33 kg steam/s) to avoid carbon deposition. These amounts are listed in #19 of Table 7.6, which contains the mass balances of the different configurations. If the kinetics of the carbon deposition reaction are slow, the amount of steam could be decreased and a higher efficiency can be obtained, as has been discussed in section 7.4.
The overall results of the mass and energy balances for the different desulfurization technologies are shown in Table 7.7. The power for the compressor and expander for the HGD have been accounted for in the electrical net output from the gas turbine. The gas turbine output is hardly influenced by the desulfurization technology, because the hot gas flow sent to the turbine does not change a lot as well, which results from a constant thermal input in the CLC reactors. In cases B and C (HGD and post-CLC desulfurization), more power is produced by the heat recovery steam generator (6.2 MWe) than with CGD, because no steam is consumed for the reboiler and no heat is lost for cooling of the syngas to 35 °C and heating up. On the other hand, more electricity is consumed by the desulfurization process (1.5 MWe) and more intermediate pressure steam is mixed with the fuel, which is not used for power production. The heat rejection is also higher, because the steam cycle condenser produces more heat as well. Because of these effects, the process efficiency is slightly increased in case an alternative desulfurization methods is selected (0.5-0.7%point increase).
Some small other differences are shown in the energy balances. In the HGD case, the consumption for the CO2 compression is somewhat higher, because more CO2 is mixed with the fuel in the filter downstream the desulfurization step. This is due to the need of the additional filtration unit downstream the HGD process to remove the entrained H2S sorbent particles. The syngas blower consumes more electricity in the cases B and C, because the temperature of the quenching gas is higher and therefore the mass flow rate is increased to quench the syngas in the gasifier.
7. Comparison of CLC in packed bed and fluidized bed reactors |187
7
Table 7.7: Energy balances for the CGD, HGD and post-CLC desulfurization in combination with packed bed CLC.
Power Method A: CGD in
PBR
Method B: HGD in
PBR
Method C: Post-CLC in
PBR Heat input LHV, MWLHV 853.9 853.9 853.9 Gas turbine, MWe 225.1* 225.2*,** 226.8* Heat Recovery Steam Cycle, MWe 183.0 189.3 189.3 Gross power output, MWe 408.1 414.5 416.1 Syngas blower, MWe -0.8 -1.5 -1.5 ASU, MWe -33.9 -33.9 -33.9 Lock hoppers CO2 compressor, MWe -3.1 -3.1 -3.1 Acid Gas Removal or Flue Gas Desulfurization, MWe -0.4 -1.9 -1.9
CO2 compressor, MWe -11.0 -11.3 -11.0 N2 intercooled compressor gasifier, MWe -1.3 -1.5 -1.5 Heat of rejection, MWe -3.6 -3.7 -3.7 Other auxiliaries, BOP, MWe -3.4 -3.4 -3.4 Net power generated, MWe 350.6 354.3 356.1 LHV efficiency, % 41.05 41.50 41.71 CO2 capture efficiency, % 97.1 97.0 97.0 CO2 purity, % 96.7 96.7 96.7 CO2 emission, kg CO2 emitted/MWhe 24.7 25.5 25.2 CO2 avoided, %, equation 6.7 96.8 96.7 96.7 SPECCA, MJ LHV/kg CO2, equation 6.8 1.08 0.96 0.90 * Gas turbine power includes consumption of air blower and nitrogen compressor for purge. ** The power of the HGD compressor and expander are accounted for.
The equipment costs for the HGD and post-CLC are expected to be lower than for the CGD technology, because a lower heat exchange surface is needed for the gas-gas heat exchanger for syngas heating. For the post-CLC configuration, the equipment related to syngas cooling is not needed, but it can only be carried out in case no sulfur species are formed on the oxygen carrier.
Desulfurization for fluidized beds with syngas The mass balance for the circulating fluidized bed process is slightly different. The fuel reactor temperature is 1200 °C and at that condition no steam has to be added to avoid carbon deposition. Full gas conversion is assumed, which is possible in case of a Mn or Fe based OC. With these OCs, no sulfur species are expected to be formed on the OC. In case nickel is used, the gas conversion is lower due to thermodynamic constraints.
| 18
8 Ta
ble
7.8:
Com
paris
on st
ream
s for
the
diffe
rent
des
ulfu
rizat
ion
met
hods
for C
LC in
circ
ulat
ing
fluid
ized
bed
reac
tors
.
M
etho
d A:
CG
D
Met
hod
B: H
GD
M
etho
d C:
Pos
t-CLC
des
ulfu
rizat
ion
# M
, kg
/s
T,
°C
p,
bar
Com
posi
tion,
mol
%
M,
kg/s
T
, °C
p,
ba
r Co
mpo
sitio
n, m
ol%
M
, kg
/s
T,
°C
p,
bar
Com
posi
tion,
mol
%
1 34
.1
15
1 Co
al 34
.1
15
1 Co
al 34
.1
15
1 Co
al
2 78
.9
165
42
Ar:
0.89
%, C
O:
51.5
7%, C
O2:
8.48
%,
H2:
21.0
0%, H
2O:
16.7
5%, H
2S: 0
.16%
, N
2: 1.
16%
69.8
38
1 43
Ar 1
.04%
, CO
: 62
.74%
, CO
2: 7.
30%
, H
2:21.
90%
, H2O
: 5.
49%
, H2S
: 0.1
8%,
N2:
1.35
%
69.8
39
5 43
Ar:
1.04
%, C
O:
62.7
4%, C
O2:
7.30
%,
H2:
21.9
0%, H
2O:
5.49
%, H
2S: 0
.18%
, N
2: 1.
35%
3 72
.7
300
20
Ar:
0.97
%, C
O:
56.4
1%, C
O2:
9.27
%,
H2:
22.9
7%, H
2O:
9.11
%, N
2: 1.
27%
71.8
39
0 20
Ar:
1.05
%, C
O:
61.4
1%, C
O2:
9.15
%,
H2:
21.4
4%, H
2O:
5.60
%, N
2: 1.
36%
69.2
39
5 20
Ar:
1.04
%, C
O:
62.6
7%, C
O2:
7.36
%,
H2:
21.8
7%, H
2O:
5.53
%, H
2S: 0
.18%
, N
2: 1.
35%
4 11
3.3
1200
19
A
r: 0.
97%
, CO
2: 65
.68%
, H2O
: 32
.07%
, N2:
1.27
%
112.
4 12
00
19
Ar:
1.05
%, C
O2:
70.5
5%, H
2O:
27.0
3%, N
2: 1.
36%
11
0.1
1200
19
Ar:
1.04
%, C
O2:
69.8
9%, H
2O:
27.5
3%, N
2: 1.
35%
, SO
2: 0.
18%
5 81
.5
28
110
Ar:
1.44
%, C
O2:
96.7
0%, N
2: 1.
76%
81
.5
28
110
Ar:
1.44
%, C
O2:
96.7
0%, N
2: 1.
76%
81
.5
28
110
Ar:
1.44
%, C
O2:
96.7
0%, N
2: 1.
76%
6
716.
2 15
1
Air
736.
3 15
1
Air
742.
2 15
1
Air
7 66
4.8
438
20
Air
682.
5 43
8 20
A
ir 68
9.2
438
20
Air
8 62
4.2
1200
19
Ar 0
.97%
, CO
2: 0.
03%
, H2O
1.0
9%,
N2:
81.7
9%, O
2: 16
.12%
642.
0 12
00
19
Ar:
0.97
%, C
O2:
0.03
%, H
2O: 1
.09%
, N
2: 81
.66%
, O2:
16.2
5%
648.
4 12
00
19
Ar:
0.97
%, C
O2:
0.03
%, H
2O: 1
.09%
, N
2: 81
.64%
, O2:
16.2
6%
9
675.
6 48
7 1
Ar 0
.97%
, CO
2: 0.
03%
, H2O
1.0
9%,
N2:
81.4
5%, O
2: 16
.46%
694.
6 48
4 1
Ar:
0.97
%, C
O2:
0.03
%, H
2O: 1
.09%
, N
2: 81
.33%
, O2:
16.5
8%
701.
4 48
4 1
Ar:
0.97
%, C
O2:
0.03
%, H
2O: 1
.09%
, N
2: 81
.32%
, O2:
16.6
0%
10
675.
6 89
1
Sam
e as
#9
694.
6 10
1 1
Sam
e as
#9
701.
4 10
2 1
Sam
e as
#9
11
120.
7 15
1
Air
120.
7 15
1
Air
120.
7 15
1
Air
12
28.9
18
0 47
A
r: 3.
09%
, N2:
1.91
%,
O2:
95.0
0%
28.9
18
0 47
A
r: 3.
09%
, N2:
1.91
%,
O2:
95.0
0%
28.9
18
0 47
A
r: 3.
09%
, N2:
1.91
%,
O2:
95.0
0%
13
- -
- -
10.5
47
8 20
N
2 -
- -
14
23
.7
80
56
Sam
e as
#5
23.7
80
56
Sa
me
as #
5 23
.7
80
56
Sam
e as
#5
15
13.9
80
1
Sam
e as
#5
13.9
80
1
Sam
e as
#5
13.9
80
1
Sam
e as
#5
16
155.
3 54
1 13
4 St
eam
15
6.0
542
134
Stea
m
154.
7 54
2 13
4 St
eam
17
15
5.3
344
36
Stea
m
156.
0 34
2 36
St
eam
15
4.7
342
36
Stea
m
18
157.
7 45
8 33
St
eam
16
0.5
458
33
Stea
m
159.
7 45
8 33
St
eam
19
-
- -
- -
- -
- -
- -
- 20
0.
1 30
0 4
Stea
m
-20.
6 17
7 4
Stea
m
-20.
0 30
0 4
Stea
m
21
157.
8 32
0
Stea
m
161.
7 32
0
Stea
m
161.
4 32
0
Stea
m
190 |
The mass balances are shown in Table 7.8. The main difference with respect to the packed bed case is the temperature and the steam content of the syngas fed to the CLC system. In the CGD base case, the syngas inlet temperature is 300 °C. This temperature is slightly higher in case of the HGD and the post-CLC desulfurization, because the syngas is heated in the hot desulfurization process (method B) or cooled down to a lesser extent in the syngas coolers (method C). As far as steam content is concerned, it decreases from 48-50% in the PBR cases to 5-17% in the CFB cases. Among the CFB cases, the maximum steam content is obtained for the CGD case, due to the use of the saturator for low temperature syngas humidification and preheating.
Table 7.9 contains the efficiencies of the three different configurations. More power is produced by the gas turbine in the cases B and C, because the syngas inlet temperature is higher. Therefore more energy is fed to the CLC system and a larger hot air flow can be generated. The other main differences are similar to the packed bed cases, except that here no steam is mixed with the fuel. Hence, a higher efficiency gain can be achieved than in the packed bed cases (0.8-1.0% point). Because the CO2 emissions are at the same order of magnitude, the higher efficiency results in a lower SPECCA number.
The efficiency gain of the HGD (method B) compared with CGD (method A) is 0.8% point and this is not as high as in the case of Giuffrida et al. (2.5% point) (Giuffrida et al., 2010). The IG-CLC plants have lower turbine inlet temperatures than the IGCC and therefore the gas turbine cycle is less efficient. Because of the lower efficiency, an increase in sensible heat of the inlet stream has a smaller effect on the net electrical efficiency as well. Additionally, in this study the power consumption for the SO2 separation has been included, resulting in an efficiency decay of 0.2% points.
7. Comparison of CLC in packed bed and fluidized bed reactors |191
7
Table 7.9: Energy balances for the CGD, HGD and post-CLC desulfurization in combination with fluidized bed CLC.
Power Method A: CGD in
CFB
Method B: HGD in
CFB
Method C: Post-CLC
in CFB Heat input LHV, MWLHV 853.9 853.9 853.9 Gas turbine, MWe 197.9 206.1 208.6 Heat Recovery Steam Cycle, MWe 216.3 217.5 216.5 Gross power output, MWe 414.2 423.6 425.1 Syngas blower, MWe -0.8 -1.5 -1.5 ASU, MWe -33.9 -33.9 -33.9 Lock hoppers CO2 compressor, MWe -3.1 -3.1 -3.1 Acid Gas Removal or Flue Gas Desulfurization, MWe -0.4 -1.9 -1.9
CO2 compressor, MWe -11.0 -11.3 -11.0 N2 intercooled compressor gasifier, MWe -1.3 -1.5 -1.5 Heat of rejection, MWe -3.6 -3.6 -3.6 Other auxiliaries, BOP, MWe -3.4 -3.4 -3.4 Net power generated, MWe 356.7 363.5 365.3 LHV efficiency, % 41.78 42.57 42.78 CO2 capture efficiency, % 97.1 97.0 97.0 CO2 purity, % 96.7 96.7 96.7 CO2 emission, kg CO2 emitted/MWhe 24.3 24.9 24.6 CO2 avoided, %, equation 6.7 96.8 96.8 96.8 SPECCA, MJ LHV/kg CO2, equation 6.8 0.88 0.66 0.61
7.6. Conclusions The influence of the CLC reactor type, viz. packed beds vs. fluidized beds, on the process efficiency has been studied. Syngas is produced in a Shell gasifier and after low temperature gas cleanup it is fed to CLC reactors filled with NiO/Al2O3 oxygen carrier, which are operated at 20 bar and 1200 °C. It has been shown that for both reactor types a process efficiency (LHV basis) around 42% can be achieved.
The packed bed systems have the advantage that the reduction can be carried out at lower temperatures and therefore the CO2 is produced at lower temperature. Hence, more heat is available to the gas turbine. The drawback is that more steam is required to avoid carbon deposition thereby reducing the efficiency. In case an oxygen carrier featuring slow kinetics for the Boudouard reaction is used, no additional steam has to be fed if the H2O/CO-ratio of 0.37 reached with the saturator is assumed to be sufficient. In that ideal case, the process efficiency is increased by 1.18% points to 42.23% of LHV. So, the kinetics of the Boudouard reaction are an important factor for the oxygen carrier selection/development.
192 |
In the fluidized bed cases, the temperature of the CO2/H2O-stream is higher and this leads to a lower process efficiency. In addition, because NiO was selected as oxygen carrier, the gas conversion is relatively low and some additional oxygen was fed to reach full conversion of the gases. In the end, an LHV efficiency of 41.37% is reached. In case of a different oxygen carrier with which full gas conversion can be achieved, an LHV efficiency of 41.78% can be reached. For the circulating fluidized beds, ideal reactors were assumed (with gas conversion that follows the thermodynamics and no gas leakages) which requires proper CLC reactor system design.
The electrical efficiency can be enhanced by 0.5-1% point by modification of the desulfurization method. In the alternative cases (HGD and post-CLC), there is no saturator and therefore more steam has to be mixed with the fuel for the packed bed configuration. Because of this, the effect of the alternative desulfurization method is rather small (0.5-0.7% point). Post-CLC can only be used if no sulfur species are formed on the oxygen carrier during reduction, but this cannot be excluded by thermodynamics. In the circulating fluidized bed reactors, a higher efficiency gain can be achieved by using a different desulfurization strategy (0.8-1.0% point), if full gas conversion is reached in the reactors. This is possible in case iron and manganese based oxygen carriers are used, that are also resistant for post-CLC. But the differences are quite small.
From these results, it can be concluded that the selection of the reactor type does not have a large influence on the process efficiency. In the packed bed case, a high temperature valve system needs to be designed, while for the fluidized beds, high temperature filtering and a solid circulation control system have to be developed able to be operated at high pressures, which seems to be a much bigger challenge compared to a valve system. The development, operability and costs of these parts will determine which type of reactor is more suitable for large scale CLC application.
|193
8 Epilogue
194 |
8. Epilogue Scope
CO2 capture and storage has been proposed as a mid-term solution to reduce anthropogenic CO2 emissions so that the greenhouse gas effect and climate change can be kept under control. The CO2 capture step is very energy intensive with current available technologies and therefore alternative technologies are under development. One of these alternatives is chemical-looping combustion (CLC), a technology for power production with integrated CO2 capture that is studied in this thesis. In CLC, the fuel is indirectly combusted via an oxygen carrier and therefore no additional expensive gas separation steps are needed to obtain a concentrated CO2 stream for sequestration. This last chapter contains an overview of the results that were obtained during this study and a discussion how this could contribute to reduce climate change.
This thesis
To be able to reach high process efficiencies the CLC process has to be operated at high pressure. The effect of the operating pressure on the redox kinetics of the oxygen carriers has been experimentally studied. A decrease in reduction reaction rates as a function of the total pressure has been observed and is in line with literature findings (García-Labiano et al., 2005). The decrease in reaction rate might be related to the decrease in the number of effective oxygen vacancies in the solid particles at higher pressures. It has also been observed that the gas fraction of the reactant has a large impact on the observed kinetics. A correlation has been found to describe the pressure effect, but no evidence has been found about the mechanism that causes this effect. The gas/solid reactions in packed bed reactors are not only dependent on kinetics, but are also to a large extent controlled by internal mass transfer limitations inside the oxygen carrier particles. These limitations are decreased at elevated pressure, because the mass flux controlled by Knudsen diffusion is increased. Therefore the overall effect of the total pressure is not very pronounced in packed bed reactors, which has also been demonstrated by experiments in a packed bed reactor with NiO/CaAl2O4 at different total pressures.
The performance of packed bed reactors has been investigated with CuO/Al2O3, NiCaAl2O4 and ilmenite as oxygen carriers. A packed bed reactor model could describe the experiments quite reasonably, provided that a good description of the kinetics is incorporated. Dedicated TGA experiments have shown that the reduction reaction rate drops suddenly at a high degree of solid conversion and full conversion cannot be reached at low temperatures. This behavior is most likely dependent on the type of
8. Epilogue |195
8
support material and more research is required to explain the observed marked decrease in reduction reaction rate and to evaluate why the support material is apparently involved in the reactions so that a better prediction and description can be developed. In any case the support material has to be selected carefully for packed bed applications.
Because it is very difficult to find/develop an oxygen carrier that fulfills all the requirements on the redox reactivity, thermal and chemical stability and oxygen transport capacity, a new packed bed process configuration has been developed with more flexibility in the oxygen carrier selection. The novel reactor configuration, referred to as the two-stage CLC (TS-CLC), has been studied in detail and successfully demonstrated experimentally. Also for this case, the model was able to describe the experiments well and it was shown that after upscaling to power plant size, hot gas with a constant high temperature can be produced that can be fed to the gas turbine for power production. Some further development is required on the oxygen carriers, but the combination of CuO/MgAl2O4 and Mn3O4/MgZrO2 looks very promising.
An integrated gasification chemical-looping combustion power plant with TS-CLC has been modeled in this thesis, showing that a LHV efficiency of 40.3-40.8% can be achieved. This is only slightly lower than the efficiency of a one-stage CLC process (41.1%), but a much higher flexibility in the selection of the oxygen carrier(s) is obtained. In all simulated packed bed CLC cases the conventional (pre-combustion) CO2 capture with Selexol is outperformed by 6% points. If CLC is carried out in pressurized circulating fluidized beds, the process efficiency is in the same order of magnitude as with CLC in packed beds, but the technology and operation of circulating fluidized beds at high pressures still needs to be developed and demonstrated. Therefore in case CLC is used in power plants, the reactor type will be selected mostly on the basis of the availability of high temperature valves and cyclones (at high pressure) and on costs. Moreover, it has been demonstrated that the process efficiency can be even further increased with a maximum of 1% point by modification of the desulfurization method.
Some recommendations for further research
For the implementation of the packed bed CLC process, cost-effective high-temperature valves have to be developed and in addition, long term cyclic tests need to be performed to ensure that the oxygen carrier(s) can withstand longtime high-temperature operation, while the reactivity at low temperature is not affected.
196 |
In addition to CLC, the knowledge obtained during this study could be used to develop other packed bed chemical-looping technologies with a low energy intensity at lower temperature, like technologies based on chemical looping (steam) reforming in which H2 is produced at relatively low temperature for power generation. Another interesting possibility could be the use of chemical-looping for oxygen production so that the conventional energy intensive air separation unit (for oxyfuel combustion or syngas generation) can be replaced (Shah et al., 2013).
Value CLC for society
The CCS technology has to compete with other power generation technologies with zero operational CO2 emissions, like solar, wind and hydropower energy. CCS has the relative disadvantage that – by definition – CO2 is produced that has to be stored in a reservoir (or deposited via mineralization), which creates a responsibility to monitor it forever and the dependency on fossil fuels remains and may even increase, because extra energy is required for the CO2 capture and the compression. These disadvantages might be accepted, if the integral costs are considerably lower than the costs for available alternatives. The CO2 might be reused instead of stored, but it is questionable whether this is a realistic solution on the large scale that CO2 is produced.
The International Energy Agency has recently analyzed the electricity costs of the different available power production technologies. This resulted in levelised electricity costs for each energy source, which are depicted in Table 8.1 (International Energy Agency, 2013a, 2013b). The large ranges reflect the large differences in resources, local conditions and the selected sub-technologies. The overview shows that in favorable circumstances onshore wind and hydropower energy can compete with a newly built fossil fuel power plant: in both cases the electricity cost are close to 50 USD/MWh. Hydropower, which generates 16% of the electricity worldwide, can only be carried out on places, where a natural water force is available for power generation. Wind energy is applicable on more places, but the power output has to be balanced by for example pumped storage hydropower, which is expected to have a price below 10 USD/MWh (International Energy Agency, 2005). The current output of a windmill is around 1 MWe and it is expected to increase to 10-20 MWe in the coming 20 years by increasing the size of the rotor and the blades (International Energy Agency, 2013c). Due to technological developments in the past five years, the wind electricity costs have decreased by 33% and further reductions are expected. In Brazil wind turbines are in operation with an electricity cost of 42 USD/MWh (International Energy Agency, 2013c).
8. Epilogue |197
8
Table 8.1: Levelised electricity cost for different power sources (published by IEA in 2013). Levelised costs are the total cost of building and operating a generating plant over its economic life, converted to equal annual payments. The large ranges result from differences in resource, local conditions and choice of sub-technology (International Energy Agency, 2013a, 2013b).
Energy source for electricity production
Levelised cost of electricity, USD/MWh
Coal (new plant) 40-90 Natural gas (new plant) 40-120 Coal with CCS 104* Bioenergy co-firing 80-140 Wind onshore 50-160 Geothermal 40-200 Hydropower 50-230 Bioenergy 80-240 Solar photovoltaic 125-250 Concentrated solar power 140-300 Wind offshore 150-340 *= calculation is based on an IGCC coal fired power plant including CCS with a LHV efficiency of 33.1%. Other state-of-the-art power production and CCS methods resulted in costs with the same order of magnitude (International Energy Agency, 2013b).
The costs of power production with CCS are estimated at 104 USD/MWh for a conventional CCS-integrated power plant with a LHV efficiency of 33.1% (International Energy Agency, 2013b). By technological developments, like CLC, the efficiency can be further increased up to 41% as demonstrated in this thesis. This results in a reduction of the power production costs (including CCS), but in the best case it will reach the minimum of 40 USD/MWh (power generation without CCS). Due to the currently low carbon emission allowance prices and the lack of public acceptance, many CO2 storage projects have been cancelled in recent years and due to this, the application of CCS on large scale will be shifted forward in time. Due to these aspects, the competition with the zero CO2 emission power production technologies is expected to be difficult.
On the other hand, CCS is easier to be used on large scale, while a very limited area is needed for it. Moreover, power plants with CCS can act as buffer to compensate output fluctuations from solar or wind energy supplies. In this respect, the packed bed CLC system offers high flexibility to operate at different capacities, because the temperature rise during oxidation is independent on the flow rate in the reactors. If a lower power plant output is desired, the flow rates can be reduced, while the cycle time is increased.
198 |
This will not have an effect on the performance, as long as the gas and the steam turbines can handle the capacity reduction.
Because the power output from packed bed CLC can simply be adjusted while sustaining a relatively high LHV efficiency, CLC qualifies as a promising technology to contribute to reducing CO2 emissions and thereby to reducing the impact of global warming. More particularly, CLC technology may serve as a flexible complementary power generation technology to compensate temporary shortfalls in wind and solar energy in new multi-generic power grids, and thereby contribute a value to society which may last longer than only facilitating the transition from conventional fossil fuel power to long term sustainable power generation.
Bibliography |199
Bibliography
Abad, A., Adánez, J., García-Labiano, F., de Diego, L.F., Gayán, P., Celaya, J., 2007. Mapping of the range of operational conditions for Cu-, Fe-, and Ni-based oxygen carriers in chemical-looping combustion. Chem. Eng. Sci. 62, 533–549.
Abad, A., Mattisson, T., Lyngfelt, A., Ryden, M., 2006. Chemical-looping combustion in a 300W continuously operating reactor system using a manganese-based oxygen carrier. Fuel 85, 1174–1185.
Adanez, J., Abad, A., Garcia-Labiano, F., Gayan, P., de Diego, L.F., 2012. Progress in Chemical-Looping Combustion and Reforming technologies. Prog. Energy Combust. Sci. 38, 215–282.
Adánez, J., de Diego, L.F., García-Labiano, F., Gayán, P., Abad, A., Palacios, J.M., 2004. Selection of Oxygen Carriers for Chemical-Looping Combustion. Energy & Fuels 18, 371–377.
Aspen Technology Inc., 2011. Aspen Plus 7.3.2.
Atkins, P., de Paula, J., 2002. Atkins’ physical chemistry, 7th ed. Oxford University Press, New York.
Baek, J.-I., Kim, J.-W., Lee, J.B., Eom, T.H., Ryu, J., Ryu, C.K., Yi, J., 2011. Effects of Support on the Performance of NiO-Based Oxygen Carriers. Oil Gas Sci. Technol. – Rev. d’IFP Energies Nouv. 66, 223–234.
Barin, I., 1993. Thermochemical Data of Pure Substances. VCH, Weinheim.
Bauer, R., Schluender, E.U., 1978. Effective radial thermal conductivity of packings in gas flow - 2. Thermal conductivity of the packing fraction without gas flow. Int Chem Eng 18, 189–204.
Chauk, S.S., Agnihotri, R., Jadhav, R.A., Misro, S.K., Fan, L.-S., 2000. Kinetics of high-pressure removal of hydrogen sulfide using calcium oxide powder. AIChE J. 46, 1157–1167.
Chiesa, P., Macchi, E., 2004. A Thermodynamic Analysis of Different Options to Break 60% Electric Efficiency in Combined Cycle Power Plants. J. Eng. Gas Turbines Power 126, 770–785.
200 |
Cho, P., Mattisson, T., Lyngfelt, A., 2004. Comparison of iron-, nickel-, copper- and manganese-based oxygen carriers for chemical-looping combustion. Fuel 83, 1215–1225.
Cho, P., Mattisson, T., Lyngfelt, A., 2005. Carbon Formation on Nickel and Iron Oxide-Containing Oxygen Carriers for Chemical-Looping Combustion. Ind. Eng. Chem. Res. 44, 668–676.
Corbella, B.M., De Diego, L., García, F., Adánez, J., Palacios, J.M., 2005. The Performance in a Fixed Bed Reactor of Copper-Based Oxides on Titania as Oxygen Carriers for Chemical Looping Combustion of Methane. Energy & Fuels 19, 433–441. doi:10.1021/ef049832z
Corbella, B.M., de Diego, L., García-Labiano, F., Adánez, J., Palacios, J.M., 2006. Characterization and Performance in a Multicycle Test in a Fixed-Bed Reactor of Silica-Supported Copper Oxide as Oxygen Carrier for Chemical-Looping Combustion of Methane. Energy & Fuels 20, 148–154.
Daubert, T.E., Danner, R.P., 1985. Data compilation tables of properties of pure compounds. American Institute of Chemical Engineers, New York.
Demonstration of a cost-effective CO2 capture technology (DemoCLoCK project), n.d.
Dueso, C., Ortiz, M., Abad, A., García-Labiano, F., de Diego, L.F., Gayán, P., Adánez, J., 2012. Reduction and oxidation kinetics of nickel-based oxygen-carriers for chemical-looping combustion and chemical-looping reforming. Chem. Eng. J. 188, 142–154.
EBTF, Anantharaman, R., Bolland, O., Booth, N., van Dorst, E., Ekstrom, C., Sanchez Fernandes, E., Franco, F., Macchi, E., Manzolini, G., Nikolic, D., Pfeffer, A., Prins, M., Rezvani, S., Robinson, L., 2011. European best practice guidelines for assessment of CO2 capture technologies. www.gecos.polimi.it/research/EBTF_best_practice_guide.pdf.
Edwards, M.F., Richardson, J.F., 1968. Gas dispersion in packed beds. Chem. Eng. Sci. 23, 109–123.
Ergun, S., 1952. Fluid flow through packed columns. Chem. Eng. Prog. 48, 89–94.
Bibliography |201
Erlach, B., Schmidt, M., Tsatsaronis, G., 2011. Comparison of carbon capture IGCC with pre-combustion decarbonisation and with chemical-looping combustion. Energy 36, 3804–3815.
Fierro, J.L.G., 2006. Metal oxides. Chemistry and applications. CRC Press.
Finkenrath, M., 2011. Costs and Performance of Carbon Dioxide Capture for Power Generation.
Forero, C.R., Gayán, P., García-Labiano, F., de Diego, L.F., Abad, A., Adánez, J., 2010. Effect of gas composition in Chemical-Looping Combustion with copper-based oxygen carriers: Fate of sulphur. Int. J. Greenh. Gas Control 4, 762–770.
Gallucci, F., Van Sint Annaland, M., 2011. A Review on Recent Patents on Chemical and Calcium Looping Processes. Recent Patents Chem. Eng. 4, 280–290.
García-Labiano, F., Abad, A., de Diego, L.F., Gayán, P., Adánez, J., 2002. Calcination of calcium-based sorbents at pressure in a broad range of CO2 concentrations. Chem. Eng. Sci. 57, 2381–2393.
García-Labiano, F., Adánez, J., de Diego, L.F., Gayán, P., Abad, A., Cuadrat, A., 2005. Effect of Pressure on the Behavior of Copper-, Iron-, and Nickel-Based Oxygen Carriers for Chemical-Looping Combustion. Energy & Fuels 24, 1402–1413.
García-Labiano, F., de Diego, L.F., Adánez, J., Abad, A., Gayán, P., 2004. Reduction and Oxidation Kinetics of a Copper-Based Oxygen Carrier Prepared by Impregnation for Chemical-Looping Combustion. Ind. Eng. Chem. Res. 43, 8168–8177.
GECOS, 2013. GS software. doi:www.gecos.polimi.it/software/gs.php
Giuffrida, A., Romano, M.C., Lozza, G., 2013. Efficiency enhancement in IGCC power plants with air-blown gasification and hot gas clean-up. Energy 53, 221–229.
Giuffrida, A., Romano, M.C., Lozza, G.G., 2010. Thermodynamic assessment of IGCC power plants with hot fuel gas desulfurization. Appl. Energy 87, 3374–3383.
Grenoble, D.C., Estadt, M.M., Ollis, D.F., 1981. The chemistry and catalysis of the water gas shift reaction. 1. The kinetics over supported metal catalysts. J. Catal. 67, 90–102.
202 |
Gunn, D.J., 1978. Transfer of heat or mass to particles in fixed and fluidised beds. Int. J. Heat Mass Transf. 21, 467–476.
Gunn, D.J., Misbah, M.M.A., 1993. Bayesian estimation of heat transport parameters in fixed beds. Int. J. Heat Mass Transf. 36, 2209–2221.
Hagen, K.D., 1999. Heat transfer with Applications. Prentice Hall, Upper Addle River, New Yersey, USA.
Håkonsen, S.F., Blom, R., 2011. Chemical looping combustion in a rotating bed reactor--finding optimal process conditions for prototype reactor. Environ. Sci. Technol. 45, 9619–26.
Håkonsen, S.F., Grande, C.A., Blom, R., 2014. Rotating bed reactor for CLC: Bed characteristics dependencies on internal gas mixing. Appl. Energy 113, 1952–1957.
Hossain, M.M., de Lasa, H.I., 2008. Chemical-looping combustion (CLC) for inherent CO2 separations--a review. Chem. Eng. Sci. 63, 4433–4451.
Hossain, M.M., Lopez, D., Herrera, J., de Lasa, H.I., 2009. Nickel on lanthanum-modified γ-Al2O3 oxygen carrier for CLC: Reactivity and stability. Catal. Today 143, 179–186.
HSC, 2013. HSC Chemistry software, version 5.1.
International Energy Agency, 2005. Variability of wind power and other renewables. Management options and strategies. Paris.
International Energy Agency, 2013a. Medium-term renewable energy market report 2013. Paris.
International Energy Agency, 2013b. Technology roadmap - Carbon capture and Storage. Paris.
International Energy Agency, 2013c. Technology roadmap - Wind energy. Paris.
IPCC, 2005. Carbon Dioxide Capture and Storage. Cambridge University Press, UK.
IPCC, 2013. Climate change 2013: The physical science basis. Stockholm.
IPCC, 2014. Climate change 2014: Mitigation of Climate Change.
Bibliography |203
Ishida, M., Jin, H., 1994. A new advanced power-generation system using chemical-looping combustion. Energy 19, 415–422.
Ishida, M., Zheng, D., Akehata, T., 1987. Evaluation of a chemical-looping-combustion power-generation system by graphic exergy analysis. Energy 12, 147–154.
Jerndal, E., Mattisson, T., Lyngfelt, A., 2006. Thermal Analysis of Chemical-Looping Combustion. Chem. Eng. Res. Des. 84, 795–806.
Johansson, M., Mattisson, T., Lyngfelt, A., 2006. Investigation of Mn3O4 With Stabilized ZrO2 for Chemical-Looping Combustion. Chem. Eng. Res. Des. 84, 807–818.
Kofstad, P., 1972. Nonstoichiometry, diffusion, and electrical conductivity in binary metal oxides. Wiley-Interscience, Olso, Norway.
Kolbitsch, P., Bolhàr-Nordenkampf, J., Pröll, T., Hofbauer, H., 2010. Operating experience with chemical looping combustion in a 120kW dual circulating fluidized bed (DCFB) unit. Int. J. Greenh. Gas Control 4, 180–185.
Kolbitsch, P., Pröll, T., Bolhar-Nordenkampf, J., Hofbauer, H., 2009. Characterization of Chemical Looping Pilot Plant Performance via Experimental Determination of Solids Conversion. Energy Fuels 23, 1450–1455.
Lewis, W.K., Gilliland, E.R., 1954. Production of pure carbon dioxide. US2665972A.
Li, Z., Sun, H., Cai, N., 2012. Rate Equation Theory for the Carbonation Reaction of CaO with CO 2. Energy & Fuels 26, 4607–4616.
Lozza, G., 1990. Bottoming steam cycles for combined gas-steam power plants. A theoretical estimation of steam turbine performances and cycle analysis, in: Proceedings of the 1990 ASME Cogen-Turbo. New Orleans (USA), pp. 83–92.
Lyngfelt, A., 2011a. Oxygen Carriers for Chemical Looping Combustion - 4 000 h of Operational Experience. Oil Gas Sci.Technol.– Rev.IFP Energies Nouv. 66, 161–172.
Lyngfelt, A., 2011b. Communication during ICEPE 2011 conference (Frankfurt) presentation.
Lyngfelt, A., 2013. Chemical-looping combustion of solid fuels – Status of development. Appl. Energy 113, 1869–1873.
204 |
Lyngfelt, A., Leckner, B., Mattisson, T., 2001. A fluidized-bed combustion process with inherent CO2 separation; application of chemical-looping combustion. Chem. Eng. Sci. 56, 3101–3113.
Marx, K., Bertsch, O., Pröll, T., Hofbauer, H., 2013. Next Scale Chemical Looping Combustion: process Integration and Part Load Investigations for a 10MW Demonstration Unit. Energy Procedia 37, 635–644.
Mattisson, T., García-Labiano, F., Kronberger, B., Lyngfelt, A., Adánez, J., Hofbauer, H., 2007. Chemical-looping combustion using syngas as fuel. Int. J. Greenh. Gas Control 1, 158–169.
Mattisson, T., Järdnäs, A., Lyngfelt, A., 2003. Reactivity of Some Metal Oxides Supported on Alumina with Alternating Methane and OxygenApplication for Chemical-Looping Combustion. Energy & Fuels 17, 643–651.
Mattisson, T., Lyngfelt, A., Leion, H., 2009. Chemical-looping with oxygen uncoupling for combustion of solid fuels. Int. J. Greenh. Gas Control 3, 11–19.
Medrano, J.A., Hamers, H.P., Ortiz, M., Ramirez, A., Van Sint Annaland, M., Williams, G., Gallucci, F., 2014. NiO/CaAl2O4 as active oxygen carrier for low temperature chemical looping applications, in: 3rd International Conference on Chemical Looping.
Meijer, H.A.J., 2012. Increase of greenhouse gas concentrations, the cause of the climate discussion. Presentation at cato2 New Years Event, Feb 7.
Moldenhauer, P., Rydén, M., Mattisson, T., Younes, M., Lyngfelt, A., 2014. The use of ilmenite as oxygen carrier with kerosene in a 300W CLC laboratory reactor with continuous circulation. Appl. Energy 113, 1846–1854.
Naqvi, R., Wolf, J., Bolland, O., Kjelstrup, S., Hustad, J.E., Gundersen, T., Lior, N., 2007. Part-load analysis of a chemical looping combustion (CLC) combined cycle with CO2 capture. Energy 32, 360–370.
Natesan, K., Zeng, Z., 2003. Study of Metal Dusting Phenomenon and Development of Material Resistant to Metal Dusting. Final Report for DOE. ANL-03/33.
Noorman, S., Gallucci, F., Van Sint Annaland, M., Kuipers, J.A.M., 2010a. Experimental Investigation of a CuO/Al2O3 Oxygen Carrier for Chemical-Looping Combustion. Ind. Eng. Chem. Res. 49, 9720–9728.
Bibliography |205
Noorman, S., Gallucci, F., Van Sint Annaland, M., Kuipers, J.A.M., 2011a. A theoretical investigation of CLC in packed beds. Part 1: Particle model. Chem. Eng. J. 167, 297–307.
Noorman, S., Gallucci, F., Van Sint Annaland, M., Kuipers, J.A.M., 2011b. Experimental Investigation of Chemical-Looping Combustion in Packed Beds: A Parametric Study. Ind. Eng. Chem. Res. 50, 1968–1980.
Noorman, S., Van Sint Annaland, M., Kuipers, J.A.M., 2007. Packed Bed Reactor Technology for Chemical-Looping Combustion. Ind. Eng. Chem. Res. 46, 4212–4220.
Noorman, S., Van Sint Annaland, M., Kuipers, J.A.M., 2010b. Experimental validation of packed bed chemical-looping combustion. Chem. Eng. Sci. 65, 92–97.
Ortiz, M., Gallucci, F., Snijkers, F., Van Noyen, J., Louradour, E., Tournigant, D., Van Sint Annaland, M., 2014. Development and testing of ilmenite granules for packed bed chemical-looping combustion. Chem. Eng. J. 245, 228–240.
Patil, D.J., Smit, J., Van Sint Annaland, M., Kuipers, J.A.M., 2006. Wall-to-bed heat transfer in gas-solid bubbling fluidized beds. AIChE J. 52, 58–74.
Rezvani, S., Huang, Y., McIlveen-Wright, D., Hewitt, N., Mondol, J.D., 2009. Comparative assessment of coal fired IGCC systems with CO2 capture using physical absorption, membrane reactors and chemical looping. Fuel 88, 2463–2472.
Richter, H.J., Knoche, K.F., 1983. Reversibility of combustion processes. ACS Symp. Ser. 71–85.
Ryden, M., Lyngfelt, A., 2006. Using steam reforming to produce hydrogen with carbon dioxide capture by chemical-looping combustion. Int. J. Hydrogen Energy 31, 1271–1283.
Ryden, M., Lyngfelt, A., Mattisson, T., 2006. Synthesis gas generation by chemical-looping reforming in a continuously operating laboratory reactor. Fuel 85, 1631–1641.
Schmidt, E., 1982. Properties of Water and Steam in S.I. Units. Springer-Verlag, Berlin.
206 |
Sedor, K.E., Hossain, M.M., de Lasa, H.I., 2008. Reactivity and stability of Ni/Al2O3 oxygen carrier for chemical-looping combustion (CLC). Chem. Eng. Sci. 63, 2994–3007.
Seider, W.D., Seader, J.D., Lewin, D.R., 2004. Cost Accounting and Capital Cost estimation (chapter 16), in: Product & Process Design Principles. John Wiley & Sons Inc., New York, pp. 473–562.
Shah, K., Moghtaderi, B., Wall, T., 2013. Effect of flue gas impurities on the performance of a chemical looping based air separation process for oxy-fuel combustion. Fuel 103, 932–942.
Shriver, D.F., Atkins, P.W., 1999. Inorganic chemistry, Third edit. ed. Oxford University Press, Oxford.
Sinnott, R.K., 2005. Chemical Engineering Design, section 13.5, in: Chemical Engineering Design. Elsevier, Oxford, p. 815.
Smit, J., Van Sint Annaland, M., Kuipers, J.A.M., 2005. Grid adaptation with WENO schemes for non-uniform grids to solve convection-dominated partial differential equations. Chem. Eng. Sci. 60, 2609–2619.
Sorgenfrei, M., Tsatsaronis, G., 2013. Design and evaluation of an IGCC power plant using iron-based syngas chemical-looping (SCL) combustion. Appl. Energy 113, 1958–1964.
Spallina, V., Gallucci, F., Romano, M.C., Chiesa, P., Lozza, G., Van Sint Annaland, M., 2013. Investigation of heat management for CLC of syngas in packed bed reactors. Chem. Eng. J. 225, 174–191.
Spallina, V., Romano, M.C., Chiesa, P., Gallucci, F., Van Sint Annaland, M., Lozza, G., 2014. Integration of coal gasification and packed bed CLC for high efficiency and near-zero emission power generation. Int. J. Greenh. Gas Control 27, 28–41.
Ströhle, J., Orth, M., Epple, B., 2014. Design and operation of a 1MWth chemical looping plant. Appl. Energy 113, 1490–1495.
ThermalCeramics, 2013. JM23 saving you energy brochure [WWW Document]. http://www.morganthermalceramics.com/products/insulating-fire-bricks-ifb/.
Bibliography |207
Tian, H., Simonyi, T., Poston, J., Siriwardane, R., 2009. Effect of Hydrogen Sulfide on Chemical Looping Combustion of Coal-Derived Synthesis Gas over Bentonite-Supported Metal−Oxide Oxygen Carriers. Ind. Eng. Chem. Res. 48, 8418–8430.
Vortmeyer, D., Berninger, R., 1982. Comments on the paper, theoretical prediction of effective heat transfer parameters in packed beds by Anthony Dixon and D. L. Cresswell[AIChE J., 25, 663(1979)]. AIChE J. 28, 508–510.
Wang, B., Yan, R., Lee, D.H., Liang, D.T., Zheng, Y., Zhao, H., Zheng, C., 2008. Thermodynamic Investigation of Carbon Deposition and Sulfur Evolution in Chemical Looping Combustion with Syngas. Energy & Fuels 22, 1012–1020.
World steel prices [WWW Document], 2014. . http://www.worldsteelprices.com. URL http://www.worldsteelprices.com
Xiao, R., Chen, L., Saha, C., Zhang, S., Bhattacharya, S., 2012. Pressurized chemical-looping combustion of coal using an iron ore as oxygen carrier in a pilot-scale unit. Int. J. Greenh. Gas Control 10, 363–373.
Xu, J., Froment, G.F., 1989. Methane steam reforming, methanation and water-gas shift: I. Intrinsic kinetics. AIChE J. 35, 88–96.
Zaabout, A., Cloete, S., Johansen, S.T., Van Sint Annaland, M., Gallucci, F., Amini, S., 2013. Experimental Demonstration of a Novel Gas Switching Combustion Reactor for Power Production with Integrated CO 2 Capture. Ind. Eng. Chem. Res. 52, 14241–14250.
Zafar, Q., Abad, A., Mattisson, T., Gevert, B., Strand, M., 2007. Reduction and oxidation kinetics of Mn3O4/Mg–ZrO2 oxygen carrier particles for chemical-looping combustion. Chem. Eng. Sci. 62, 6556–6567.
Zhao, H., Cao, Y., Orndorff, W., Pan, W., 2013. Study on modification of Cu-based oxygen carrier for chemical looping combustion. J. Therm. Anal. Calorim. 113, 1123–1128.
Nomenclature |209
Nomenclature
Abbreviations AGR acid gas removal apparent active weight content the weight fraction of the oxygen carrier that reacts
during reduction ASU air separation unit BOP balance of plant CCS carbon capture and storage CFB circulated fluidized bed system CGD cold gas desulfurization CLC chemical looping combustion EBTF European benchmarking task force eco economizer eva evaporator FGD wet lime-limestone desulfurization process GS gas-steam cycles software GT gas turbine HGD hot gas desulfurization HHV high heating value HP high pressure HRSC heat recovery steam cycle HRSG heat recovery steam generator IGCC integrated gasification combined cycle IGCLC integrated gasification chemical-looping combustion IP intermediate pressure L/d length/diameter ratio LH lock hopper LHV low heating value LP low pressure OC oxygen carrier PB packed beds configuration SH super heater SPECCA specific primary energy consumptions for CO2
avoided, MJLHV/kgCO2 ST steam turbine TIT turbine inlet temperature TSA temperature swing adsorption TS-CLC two-stage chemical-looping combustion USD US Dollar WGS water gas shift XPS x-ray photoelectron spectroscopy
210 |
List of symbols b gas solid reaction stoichiometric factor, mol solid/mol gas [B] inversed diffusion matrix c concentration, mol m-3 CO cost of component, € Cp heat capacity, J mol-1K-1 Dax axial dispersion coefficient, m2 s-1 Di,k diffusivity, m2 s-1 Ds,0 pre-exponential factor for diffusion term [D] diffusion matrix d diameter, m EA activation energy, J mol-1 EDs activation energy, J mol-1 ECO2 specific CO2 emissions, kg MWhe-1
f design stress of carbon steel, Pa HR heat removal j diffusive mass flux, kg m-2s-1 kx solid diffusion decay constant, - k0 pre-exponential factor
L reactor length, m M molar mass, kg mol-1 Ṁ molar flow rate, kmol s-1 m mass flow rate, kg s-1 ṁ specific mass flow rate, kg m-2s-1 n reaction order in gas, - ni gas flux of component i, kg m-2s-1 N number of components NR number of reactors, - Nu Nusselt number, - ox oxidation p pressure, Pa p1 purge 1 p2 purge 2 Pin pressure of the vessel, Pa Pr Prandtl number, - Q heat losses, W R gas constant, J mol-1K-1 r particle radius, m ri reaction rate of gaseous component i, mol m-3s-1 rg grain radius, m Re Reynolds number, - red reduction ssteel thickness of steel vessel, m
Nomenclature |211
Sc Schmidt number, - T temperature, K t time, s V gas volume flow rate, m3 s-1 v superficial gas velocity, m s-1 VR volume of reactor, m3 wh heat front velocity, m s-1 wr reaction front velocity, m s-1 x axial position, m X particle conversion, - y mole fraction in gas feed, -
Greek letters α heat transfer coefficient, Wm-2K-1 ΔHR reaction enthalpy, J mol-1 ΔXs conversion of the oxygen carrier, - Δt time difference between two measurements, s ε porosity, m3m-3 ζ stoichiometric factor, mol gas/mol solid η net electric efficiency, - ηg dynamic gas viscosity, kg m-1s-1 λ thermal conductivity, W m-1K-1 λeff effective heat dispersion coefficient, W m-1K-1 ν reaction stoichiometric factor (negative for reactants), - νi diffusion volume of component i, m3mol-1 νs stoichiometric factor for solids, mol MeO/mol Me ρ density, kg m-3 ρmol,oxygen amount of atomic oxygen per m3 of reactor, kmol m-3 τ tortuosity, - τcycle cycle time, s τTC time delay in the thermocouple, s φstep the portion of reactors operating under the considered step ω mass fraction, kg kg-1
212 |
Subscripts act active avg average Bin binary el electrical eff effective g gas i gas component number in inner j solid component number k gas component number Kn Knudsen ls from liner to surroundings max maximum MeO metal oxide Me metal mol molecular n number of gaseous components p particle rl from reactor to liner r reactor red reduction ref reference plant without CO2 capture s solid TC thermocouple th thermal tot total w reactor wall 0 reference case
Superscripts in inlet o in oxidized state
Curriculum vitae |213
Curriculum vitae Paul Hamers was born on the 19th of January 1986 in Winschoten (NL). After finishing gymnasium in 2004 at Bonhoeffer College in Castricum (NL), he studied Chemical Engineering at University of Twente in Enschede (NL). In 2010 he graduated within the Fundamentals of Chemical Reaction Engineering group on a “Techno economic evaluation of cryogenic CO2 capture using dynamically operated packed beds”. From December 2010 he started a PhD project at Eindhoven University of Technology (NL) of which the results are presented in this dissertation.
214 |
List of publications i. Hamers, H.P., Gallucci, F., Cobden, P.D., Kimball, E., Van Sint Annaland, M.,
2013. A novel reactor configuration for packed bed chemical-looping combustion of syngas. Int. J. Greenh. Gas Control 16, 1-12.
ii. Kimball, E., Hamers, H.P., Cobden, P.D., Gallucci, F., Van Sint Annaland, M., 2013. Operation of fixed-bed chemical looping combustion. Energy Procedia 37, 575-579.
iii. Hamers, H.P., Gallucci, F., Cobden, P.D., Kimball, E., Van Sint Annaland, M., 2014. CLC in packed beds using syngas and CuO/Al2O3: Model description and experimental validation. Applied Energy 119, 163-172.
iv. Hamers, H.P., Romano, M., Spallina, V., Chiesa, P., Gallucci, F., Van Sint Annaland, M., 2014. Comparison on process efficiency for CLC of syngas operated in packed bed and fluidized bed reactors. Int. J. Greenh. Gas Control 28, 65-78.
v. Hamers, H.P., Gallucci, F., Williams, G., Van Sint Annaland, M., 2014. Experimental demonstration of CLC and the pressure effect in packed bed reactors using NiO/CaAl2O4 as oxygen carrier. Submitted to Fuel.
vi. Kooiman, R.F., Hamers, H.P., Gallucci, F., Van Sint Annaland, M., 2014. Experimental demonstration of two-stage packed-bed chemical-looping combustion using syngas with CuO/Al2O3 and NiO/CaAl2O4 as oxygen carriers. Submitted to Industrial & Engineering Chemical Research.
vii. Hamers, H.P., Romano, M., Spallina, V., Chiesa, P., Gallucci, F., Van Sint Annaland, M., 2014. Energy analysis and economic evaluation of two-stage packed-bed CLC configurations for an IGCC power plant. Submitted to Energy.
viii. Hamers, H.P., Gallucci, F., Williams, G., Cobden, P., Van Sint Annaland, M., 2015. Reactivity of oxygen carriers for CLC in packed bed reactors under pressurized conditions. Will be submitted to Energy & Fuels.
List of publications |215
ix. Hamers, H.P., Romano, M., Spallina, V., Chiesa, P., Gallucci, F., Van Sint
Annaland, M., 2015. Boosting the IGCLC process efficiency by optimizing the desulfurization step. Will be submitted to Applied Energy.
x. Medrano, J.A., Hamers, H.P., Van Sint Annaland, M., Williams G., Gallucci, F., 2015. NiO/CaAl2O4 as active oxygen carrier for low temperature chemical looping applications. Will be submitted to Applied Energy.
xi. Gallucci, F., Hamers, H.P., Van Zanten, M., Spallina, V., Van Sint Annaland, M., 2015. Experimental demonstration of chemical-looping combustion of syngas in packed bed reactors with ilmenite. Will be submitted to Chemical Engineering Journal.
216 |
Acknowledgement Although only my name is on the cover, some other people have contributed to this work, who deserved to be mentioned. This section is specially dedicated to them.
First of all, I would like to thank my promoter Martin van Sint Annaland with whom I had many meetings and discussions, which has increased the quality of this work. The most important meeting was when we developed the idea of the two stage configuration, while we were preparing a project partners meeting. This idea that was developed on a Wednesday morning in June 2011 has finally become the basis of this dissertation. You gave me the freedom to set the priorities for the research, while you gave feedback and addressed critical points, which has demonstrated to be a good method to finish the thesis in time. I would also like to thank you for the confidence in me and the patience that you had, when you offered me the PhD position. Fausto Gallucci was my daily supervisor, who forms a complementary team with Martin, because both have a different focus and therefore I am very happy with you as supervisors. Fausto, what is very special from you is that you were almost always available for questions and you are a good troubleshooter. Moreover, you have a good view how to transform obtained results to a publication. Furthermore, I would like to thank representatives of the project partners, Paul Cobden (ECN) and Erin Schols (TNO), for all the input during the discussions we had.
During my PhD I spent six months at the GECOS group of Politecnico di Milano in Italy. Here I had to adapt to a different culture, where everything is differently organized and people work longer (note that The Netherlands is the country with the lowest number of working hours per person in Europe). Paolo Chiesa and Matteo Romano allowed me to study at their group and supervised me during this period. I am very thankful for the detailed and critical way how you reviewed the work. Vincenzo Spallina also helped me a lot during this period to get familiar with the software that describes the power plant, which I very appreciate. This period has been very productive for the project, because most of the results from chapter 6 and 7 were obtained during this time.
For experimental work I had very good technical assistance from Joris Garenfeld, Lee McAlpine and Joost Kors. With Joris I worked most of the time, because he built the packed bed reactor setup. Several times you opened and closed the big reactor to substitute the oxygen carrier particles, which saved me much time that I alternatively had to spend in the sports center to be able to seal the big reactor. You always came up with good suggestions how to deal with the problems that are unavoidable with a new
Acknowledgement |217
reactor system. I am also thankful to Herbert Fiedler for the automation of the setups and Judith Wachters for the administrative work.
During this project I supervised quite some students, which diversified the work for me and increased the output from the project. Rajesh Ravindran, Vincent van Duijnhoven, Gerardo Duarte Murillo, Martijn van Zanten, Alan Ramirez Rojas did their MSc graduation project with me and I am thankful for your work and enjoyed the time working with you. I also liked the collaboration during the two months projects with Stijn Smits, Nicky Oppers, Marc van den Bergh, Steven Rademakers, Michael Aubert and Roeland Kooiman. Despite that the results of most of your theses can only indirectly be retrieved from this dissertation, your contributions were quite valuable for the project.
Except with students, I also collaborated with colleagues who worked on similar topics for which I am very thankful. Especially, I would like to thank Vincenzo Spallina, Maria Ortíz Navarro, José Antonio Medrano Jimenez, Marian San Pio Bordeje, Tommaso Melchiori, Remco Lancee and Ivo Filot for the collaborations and I hope that my input was also useful for your projects. Much more people would be mentioned, if the acknowledgement was written about me as person. Because I limit it to the project related acknowledgements, this book contains only one personal element and that is the front cover, for which I am very thankful to the drawer, Marie-Josée Hamers.
In life, the travel experience is more important than the destination. Therefore it is important that personally I have had a nice time and I am very grateful to all my relatives, friends and colleagues who made this possible.