Novel Process Concept for Cryogenic CO2 Capture

Novel process concept for cryogenic CO2 capture

Citation for published version (APA): Tuinier, M. J. (2011). Novel process concept for cryogenic CO2 capture. Technische Universiteit Eindhoven. https://doi.org/10.6100/IR719418

DOI: 10.6100/IR719418

Document status and date: Published: 01/01/2011

Document Version: Publisher’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 version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author 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

General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and 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.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement: www.tue.nl/taverne

Take down policy If you believe that this document breaches copyright please contact us at: [email protected] providing details and we will investigate your claim.

Download date: 28. Sep. 2021 Novel Process Concept for Cryogenic CO2 Capture Samenstelling promotiecommissie:

prof.dr. J. Meuldijk, voorzitter Technische Universiteit Eindhoven prof.dr.ir. M. van Sint Annaland, promotor Technische Universiteit Eindhoven Prof.Dr.-Ing. A. Seidel-Morgenstern Otto-von-Guericke-Universitat¨ Magdeburg prof.dr. G.J. Kramer Shell / Universiteit Leiden prof.dr.ir. J.A.M. Kuipers Technische Universiteit Eindhoven prof.dr.ir. T.H. van der Meer Universiteit Twente dr.ir. D.W.F. Brilman Universiteit Twente

The research reported in this thesis was sponsored by Shell Global Solu- tions International.

c M.J. Tuinier, Eindhoven, The Netherlands, 2011 No part of this work may be reproduced in any form by print, photocopy or any other means without written permission from the author.

Publisher: Ipskamp Drukkers B.V., P.O. Box 333, 7500 AH, Enschede, The Netherlands.

A catalogue record is available from the Eindhoven University of Techno- logy Library.

ISBN: 978-90-386-2900-1 Novel Process Concept for Cryogenic CO2 Capture

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 donderdag 24 november 2011 om 16.00 uur

door

Martin Jan Tuinier

geboren te Wijhe Dit proefschrift is goedgekeurd door de promotor:

prof.dr.ir. M. van Sint Annaland vi Contents

Summary 1 1 Introduction 5 1.1 Climatechange...... 5 1.2 Carboncaptureandstorage ...... 8 1.3 Thisthesis ...... 15 2 Cryogenic packed bed process concept 19 2.1 Introduction ...... 20 2.2 Theprocessconcept...... 20 2.3 Detailednumericalmodel...... 24 2.4 Simpliﬁedmodel: Sharpfrontapproach ...... 31 2.5 Processanalysis ...... 37 2.6 Discussionandconclusions ...... 42 3 Experimental demonstration of the concept 45 3.1 Introduction ...... 46 3.2 Experimentalsetupandprocedure ...... 46 3.3 Results ...... 48 3.4 Simulations...... 52 3.5 Discussionandconclusions ...... 56 4 Mass deposition rates of carbon dioxide 59 4.1 Introduction ...... 60 4.2 Experimental ...... 61 4.3 Results ...... 67 4.4 Developmentofafrostgrowthmodel ...... 73 4.5 Discussionandconclusions ...... 87

vii viii Contents

5 Experimental demonstration in a pilot scale setup 93 5.1 Introduction ...... 94 5.2 Experimental ...... 94 5.3 Experimentalresults ...... 98 5.4 Simulations...... 101 5.5 Discussionandconclusions ...... 107 6 Techno-economic evaluation 111 6.1 Introduction ...... 112 6.2 Processevaluation...... 113 6.3 Comparison with absorption and membrane technology . . . 124 6.4 Conclusions...... 131 7 Biogas puriﬁcation 133 7.1 Introduction ...... 134 7.2 Adsorption ...... 135 7.3 Cryogenicpackedbedconcept ...... 136 7.4 Adsorption versus cryogenic packed bed concept ...... 142 7.5 Hydrogensulﬁderemoval...... 146 7.6 Discussionandconclusions ...... 146 8 Epilogue and outlook 149 8.1 Importantaspectsforfuturedevelopment ...... 150 8.2 Futureoftheproposedconcept ...... 152 Bibliography 154 List of publications 163 Curriculum Vitae 165 Dankwoord 167 Summary

Carbon capture and storage (CCS) is generally considered as one of the necessary methods to mitigate anthropogenic CO2 emissions to combat climate change. The costs of CCS can for a large extent be attributed to the capture process. Several post-combustion CO2 capture processes have been developed, such as scrubbing, membrane processes and pressure swing adsorption. Amine scrubbing is currently the state of the art technology, in which CO2 is being removed by contacting the flue gas with a solvent in an absorber. Regeneration is carried out by heating the loaded solvent in a stripping column. The main disadvantages of this process are the energy costs related to the regeneration step and solvent losses due to degradation. A promising novel option is to freeze out (desublimate) CO2 from flue gases using cryogenically cooled surfaces. High cooling costs could be minimized by exploiting the cold duty available at Liquefied Natural Gas (LNG) regasification sites. No standard process equipment is available to deal with separations based on desublimation. Therefore, a novel process concept has been developed and investigated in this dissertation, based on the periodic operation of cryogenically cooled packed beds. When feeding a flue gas to a previously cryogenically refrigerated packed bed, CO2 will freeze onto the packing surface, while permanent gases such as N2 pass through the bed unaltered. The amount of CO2 depositing onto the packing reaches an equilibrium value, because the amount of cold energy stored in the packing is limited. Therefore, plugging of the bed is intrinsically circumvented. A front of desublimating CO2 will move through the bed, until breakthrough is observed. At that point, the bed is switched to a recovery step, in which all previously deposited CO2 will be removed by recycling gaseous CO2 through the bed. The energy required for the sublimation of CO2 can be provided to the

1 2 Summary

bed by feeding the flue gas at elevated temperatures during the capture step. A process cycle is finally finished with a cooling step, in which the bed is again refrigerated to its initial temperature. The proposed process concept has several advantages: simple and low cost equipment can be used, large pressure drops can be avoided, deep CO2 removal is possible and CO2, H2O and other impurities can be separated simultaneously. The development in time of axial temperature, gas concentration and mass deposition profiles in the packed beds during the different process steps can be well described using a one-dimensional, pseudo- homogeneous, axially dispersed plug flow model, in which the mass and energy balances are solved simultaneously using an advanced numerical scheme. When assuming that no axial dispersion and mass deposition rate limitations occur during the process, the fronts which are developed are very well defined (sharp). Based on this assumption, the process can be well described with a simplified model (the ‘sharp front’ approach). With basic mass and energy conversation laws, the axial temperature, gas composition, and mass deposition profiles and front velocities can be calculated very fast using this model, making it a perfect tool for design and evaluation of the process. In the limit of negligible dispersion in the detailed numerical model, the solution converges to the profiles predicted by the sharp front approach. Outcomes of the sharp front approach show that the specific cooling duty required to capture a certain amount of CO2 increases for lower CO2 concentrations in the feed gas and for higher initial bed temperatures. A small scale experimental setup has been designed and constructed to measure axial temperature profiles and CO2 concentrations at the outlet during the capture step. Experiments have been carried out for N2/CO2 and N2/CO2/H2O mixtures. The results showed that a good separation between the single components is possible. The experimental findings have been compared to results calculated using the numerical model. The front velocities and therefore CO2 breakthrough times and the temperature profiles for different CO2 and H2O concentrations in the flue gas and initial bed temperatures are very well predicted by the developed model. Expressions for mass deposition rates of CO2 are required to accurately describe the process. No information is available in literature, therefore a dedicated experimental setup has been designed and constructed to measure mass deposition rates for different (gas and surface) temperatures, pressures and compositions. It is shown that the Summary 3

rate of desublimation is influenced by the thickness of the CO2 layer deposited onto the cooled surface, indicating the importance of heat transfer through the frost layer. Furthermore, it is found that the presence of N2 in the gas phase has a large effect on the desublimation rates, which indicates the presence of mass transfer limitations. A model has been developed to describe the observed behavior. The frost growth process has been described as a moving boundary problem, in which both mass and heat transfer are taken into account. Based on the experimental results, expressions have been derived to describe the density and heat conductivity of the frost layer. Using these expressions, the model is well able to predict the experimental results. It is shown that under the conditions as prevailing in the packed beds, mass deposition rates are mainly determined by mass transfer from the gas bulk phase to the packing surface. The small scale experimental setup has been used to study the capture step. In order to demonstrate the entire process cycle including the cooling and recovery step, a larger pilot setup has been constructed, containing three beds operated continuously. Test runs of more than 10 hours showed that it is indeed possible to continuously capture CO2 with the proposed concept. Radial temperature differences were observed in the beds, which could be attributed to the influence of the steel wall via simulations with the numerical model after including an additional energy balance for the wall. In a techno-economic evaluation the influence of several process parameters has been investigated; lower initial bed temperatures and higher CO2 concentrations in the feed result in more efficient use of the bed volume. The pressure drop over the system plays an important role in the process economics, due to the high flow rates required in the process. The cryogenic concept has been compared to two competing technologies: amine scrubbing and membrane separation. The results show that the preferred technology highly depends on assumptions related to the availability of utilities. The novel cryogenic capture process can compete with other technologies, provided that cold duty is available at low cost. An alternative promising application for the proposed technology is biogas purification. By operating the capture step at elevated pressures, it is possible to remove CO2 during the regeneration step very efficiently by reducing the pressure. The process performance has been compared with vacuum pressure swing adsorption technology. The required beds for treating a gas mixture containing 45 vol.% CO2 and 55 vol.% CH4 are 4 Summary eight times smaller for the cryogenic packed bed concept and the energy consumption is 22% lower. 1

Introduction

1.1 Climate change

There is a growing worldwide awareness of the fact that the earth’s surface temperatures are changing globally. Although the climate of our planet has been altering continuously during its history, the current changes are taking place with unprecedented pace and are expected to have dramatic consequences for human kind (IPCC, 2007). In the ﬁrst place due to rising sea levels, which is threatening land at lower levels, but also of the expected negative impact on agriculture and fresh water supply. Since the industrial revolution in the late 19th century, fossil fuels started to play an important role in our energy supply for transportation, heating and electricity. The combustion of fossil fuels results in large amounts of CO2, which are emitted into the atmosphere. The increase in CO2 concentrations coincides with the increase in global temperatures. There is a general consensus among most scientists that the rise in CO2 concentrations is responsible for the observed increase in temperatures. In 1988 the Intergovernmental Panel on Climate Change (IPCC) was established in order to evaluate the risks of climate change caused by human activity. Based on their observations, it is expected 6 Introduction

that global temperatures will keep on rising in the next century, as illustrated in Fig. 1.1. In order to prevent or at least minimize further temperature increases, it is necessary to reduce anthropogenic CO2 emissions. Not only for health and safety reasons, but also for economic reasons. In a study on the economical effects of climate change by Stern (2007) it is stated that the costs of mitigating climate change can be limited to around 1% of global GDP each year. Doing nothing (‘business as usual’) and facing the consequences of climate change will be equivalent to los- ing 5% to possibly 20% of global GDP each year. Therefore, immediate action to reduce CO2 emissions is essential. This can be achieved in several ways, in the first place by improving efficiencies. Developments in the automobile industry for example are leading to more and more economic engines, consuming less fuel. The efficiency of power plants is also increasing, and chemical industry is able to save energy by for example heat integration. These developments will contribute to CO2 emission reductions. However, it is not expected that these efficiency improvements will be sufficient to bring down our emissions to acceptable levels, mainly because of increased energy demands by developing countries such as China and India. Therefore more measures are required. A key measure is to switch our energy supply to renewable energy sources such as biomass, solar and wind energy. Not only to reduce our CO2 emissions, but also to bring down our dependency on scarce fossil fuels. However, at this point power supply by renewable energy sources is still under development and is not yet competitive with conventional power generation based on fossil fuels. A third possible route to emission reductions is nuclear power, but safety issues and nuclear waste disposal are causing moral and political con- cerns. Due to the above mentioned reasons, it is expected that fossil fuels will continue to play a significant role in our energy supply for the next decades (U.S. Department of Energy, 2010), as shown in Fig. 1.2. It is therefore considered to be necessary to introduce Carbon Capture and Storage (CCS) to mitigate anthropogenic CO2 emissions as a mid- term solution until a full transition to energy supply by renewables can be realized. 1.1 Climate change 7

Figure 1.1: Projections of surface temperatures for the period 2020-2029 and 2090-2099 relative to temperatures in 1980-1999, based on three scenarios (IPCC, 2007). 8 Introduction

300

Liquids Renewables

Coal Nuclear

Natural Gas

200

100 Energy use [quadrillion Btu]

1990 2000 2010 2020 2030

Year

Figure 1.2: World marketed energy use by fuel type from 1990 to 2035 (U.S. Department of Energy, 2010).

1.2 Carbon capture and storage

The goal of CCS is to remove CO2 from ﬂue gases and to store it for the long term. This process is schematically represented in Fig. 1.3. Before discussing the possible capture processes, ﬁrst attention is paid to storage options.

1.2.1 Storage

It is proposed in literature (Leitner, 1995) to reuse captured CO2 as a raw material for chemical syntheses. However, the CO2 molecule is thermo- dynamically very stable and relatively unreactive. Therefore severe and costly process conditions are required. Another option is to reuse CO2 in greenhouses, which is currently being applied by Shell at their reﬁnery in Pernis, The Netherlands (Shell, 2011). Furthermore it is proposed to reuse CO2 as a feedstock for growing algae (Brennan and Owende, 2010). Although above mentioned options could play a role in some cases, the total amount of CO2 available is in general exceeding the demand by far. 1.2 Carbon capture and storage 9

Therefore, it is also necessary to store CO2 instead of reusing it. Several options are available: mineralization of CO2 (Seifritz, 1990) and storage in geological formations or in oceans. The problem of mineralization is that reaction rates are low, limiting large scale application. The effects of ocean storage on marine life is still uncertain and therefore most attention is paid to geological storage. CO2 could be stored in depleted oil or gas reservoirs or saline formations (IPCC, 2005). CO2 injection could also be used for Enhanced Oil Recovery (EOR) or Enhanced Coal Bed Methane Recovery (ECBM). Experience with CO2 injection in for example the Sleip- ner project in Norway, indicate that storing CO2 in geological formations is a feasible option to mitigate CO2 emissions (IPCC, 2005).

1.2.2 Capture Fossil fuels are normally combusted using air. The flue gas is therefore composed of a large amount of N2 and 5-20 vol.% CO2. Furthermore it contains H2O and impurities such as sulphur and nitrous oxides, depending on the feedstock and process. Compressing and storing the entire flue gas, including N2 will be too costly. Therefore it is necessary to obtain CO2 in purified form first, before it can be stored in geological formations. About 75% of the costs involved in CCS are associated with the capture step (Ebner and Ritter (2009)) and therefore many research projects focus on development or optimization of capture technologies. CO2 capture technologies are often classified into oxyfuel, pre- and post-combustion processes, which are schematically represented in Fig. 1.4. In oxyfuel processes, fossil fuels are combusted using pure oxy- gen, circumventing dilution of CO2 with N2. Disadvantage is that an energy intensive air separation unit is required to obtain pure O2, although this could be avoided using chemical looping combustion (see e.g. Ishida and Jin (1994), Noorman et al. (2007)). In pre-combustion processes fossil fuels are first converted into H2 and CO2 (via (autothermal) reforming or partial oxidation and water-gas- shift), CO2 is subsequently captured and H2 is fed to the combustion chamber or fuel cell. Advantage is that the separation of CO2 and H2 can be carried out at a high pressure, resulting in a high driving force for separation. Disadvantage of pre-combustion is that these processes are applicable to mainly new plants and can therefore not be applied to already operating facilities. 10 Introduction

Figure 1.3: Schematic overview of CCS process (IPCC, 2005). 1.2 Carbon capture and storage 11

Figure 1.4: Pathways to CO2 capture, a. conventional combustion process without capture, b. post-combustion, c. pre-combustion, d. oxy-fuel.

Post-combustion processes are based on capturing CO2 from flue gases from conventional air fired combustion processes. Disadvantage is that CO2 is dilute and at low pressures, reducing the driving force for separation. However, this technology can be retrofitted to already operating power plants and industries. For this reason post-combustion is considered the most realistic technology on the short term, even though the efficiency of the alternatives could be higher (Kvamsdal et al., 2007). Sev- eral post-combustion technologies are under development, an overview is given below.

Scrubbing

The state-of-the-art post-combustion CO2 capture technology is scrubbing. This technology is based on feeding the ﬂue gas to an absorber to selectively absorb CO2. In a desorber column the solvent is stripped from CO2 by changing temperature and/or pressure. Amines, such as Mono Ethanol Amine (MEA), are the most commonly used chemical solvents. Since 1930 amine scrubbing has already been used to remove CO2 from natural gas and hydrogen (Rochelle, 2009). Drawbacks of amine 12 Introduction scrubbing are high energy demands, mainly due to the stripping step at elevated temperature, intolerance of solvents to impurities (SOx, NOx, O2) and equipment corrosion. Many novel amine-based solvents are under development with higher CO2 solubility, faster absorption kinetics and at the same time a better tolerance for impurities, (see e.g. Notz et al. (2007)). Also chemical solvents based on amino acids (Aronu et al., 2010), chilled ammonia (Darde et al., 2010) or carbonates (Figueroa et al., 2008) are under consideration. Instead of chemical solvents, physical solvents, e.g. Selexol (dimethyl ether of polyethylene glycol) or Rectisol (chilled methanol) could be used for CO2 removal. However, physical solvents are considered inferior to chemical solvents because a high partial pressure of CO2 and low temperature is required, both of which are not the case in ﬂue gas treatment (Notz et al., 2011).

Membranes Three types of membrane systems are under consideration for post- combustion CO2 capture: gas separation membranes, membrane absorption and facilitated transport membranes. Polymeric (Powell and Qiao, 2006) and ceramic (Bredesen et al., 2004) membranes could both be used as gas separation membranes. Ceramic membranes require high temperatures in general and will therefore have more potential for pre- combustion capture. The intolerance of polymeric membranes to impurities but also H2O is generally considered as a limitation to application, although in recent research novel polymeric membranes have been developed which are able to separate CO2 and H2O simultaneously (Reijerkerk et al., 2011). In a membrane absorption process, membranes are installed in a reg- ular scrubbing process, to act as a contact area between the flue gas and solvent. Therefore impurities in the gas phase will not directly contact the solvent, reducing solvent losses. Furthermore a higher contact area per unit volume can be created, although the additional mass transfer resistance associated with the introduction of the membrane might reduce or possibly even cancel the advantage of the increased contact area (Notz et al., 2011). In Facilitated Transport Membranes (FTM) a carrier medium selectively interacts with a specific molecule. A CO2 selective FTM would have much lower processing costs and improve the equilibrium driven processes. However, currently FTMs have stability problems, which are 1.2 Carbon capture and storage 13 mainly caused by evaporation of the carrier medium (Ebner and Ritter, 2009). The challenge in developing membrane systems is that two important criteria are inversely related: a high permeability and a high selectivity. CO2 separation from for example natural gas has already been successfully applied commercially, due to a high partial pressure of CO2 in the feed and therefore high driving force for permeation. However, CO2 separation from flue gases is more difficult due to the lower partial pressures of CO2 (Ebner and Ritter, 2009). The low driving force therefore requires compression of the flue gas or requires large membrane areas, resulting in increased operational and capital costs respectively. Despite this reason, some authors claim that membrane processes could compete well with absorption processes (Favre, 2007).

Adsorption

Separation of CO2 from gas mixtures by adsorption is based on differences in interaction with the adsorbent surface. Molecular sieves or activated carbons are common used as adsorbents (IPCC, 2005). The process is normally carried out in several packed beds operated in parallel. Re- generation is done by either Pressure Swing Adsorption (PSA) (Ko et al., 2005), Temperature Swing Adsorption (TSA) (Merel´ et al., 2006) or Elec- trical Swing Adsorption (ESA) (Grande and Rodrigues, 2008). Adsorption technology is already being used for CO2/H2 separation. The energy costs could be lower than scrubbing for post-combustion CO2 capture. How- ever, the disadvantages are the low CO2 selectivity, low loading capacity for CO2 (resulting in large beds), low adsorption rates, relatively large pressure drops over the fixed beds and the high energy demand for regeneration (Notz et al., 2011). Research focuses on improved adsorbents, such as amine-functionalized zeolites (Su et al., 2010), or Metallic Or- ganic Frameworks (MOFs) (Britt et al., 2009). Another adsorption technology under investigation is calcium looping, in which the flue gas is contacted with CaO in a fluidized bed forming CaCO3. In a connected second fluidized bed regeneration is carried out at elevated temperatures (Manovic et al., 2009). 14 Introduction

Cryogenics

Cryogenics are another option for separating CO2 from gas mixtures. The advantages are that no chemical ab- or adsorbents or large pressure differences are needed and that high purity products can be obtained. How- ever, cryogenic CO2 capture is not included in most (economic) comparison studies, as it has been considered as an unrealistic candidate for post-combustion CO2 capture. In the first place due to expected high cooling costs, but also because it has been considered as a gas-liquid separation (Aaron and Tsouris, 2005; Ebner and Ritter, 2009). At atmospheric pressures CO2 will go directly from its gas phase to its solid phase (desublimation). In order to be able to carry out the CO2 removal from flue gases as a gas-liquid separation, it is necessary to compress the gas to ◦ pressures above the triple point of CO2, which is at 5.2 bar and -56.6 C for pure CO2, as shown in the phase diagram of pure CO2 in Fig. 1.5. Compressing flue gases to high pressures is too energy intensive. Expensive refrigeration can possibly be avoided when exploiting the cold duty available at Liquefied Natural Gas (LNG) regasification sites. Currently, LNG is being regasified using seawater or by using water baths which are heated by burning a fuel gas (Ertl et al., 2006). The global LNG market is strongly growing (John and Robertson, 2008), therefore integration of LNG regasification and a cryogenic CO2 process could be beneficial. Clodic and Younes (2002, 2005) have developed a cryogenic CO2 capture process, in which CO2 is desublimated as a solid onto surfaces of heat exchangers which are cooled by evaporating a refrigerants blend. With calculations and experimental tests they showed that their process could compete with other post-combustion CO2 capture processes. The main disadvantage of their system is that the water content in the feed stream to the cooling units should be minimal in order to prevent plugging by ice or an unacceptably high rise in pressure drop during operation. Therefore, several costly steps are required to remove all water traces from the flue gas. In addition the increasing layer of solid CO2 onto heat exchanger surfaces during the capture cycle will adversely affect the heat transfer, reducing the process efficiency. Moreover, the costly heat exchangers have to be switched to regeneration cycles operated at a different temperature, which should be carried out with great care to avoid excessive mechanical stresses. 1.3 This thesis 15

10,000

1,000 supercritical fluid

100

P [bar] critical point

1 200 250 300 350 400 T [K]

Figure 1.5: Phase diagram of pure CO2.

1.3 This thesis

To avoid above mentioned problems, an alternative cryogenic CO2 process concept based on dynamically operated packed has been developed at the University of Twente and Eindhoven University of Technology in co- operation with Shell Global Solutions International. This process concept will be described and analyzed in detail in this dissertation. When feeding a flue gas to a previously refrigerated packed bed, CO2 will freeze onto the packing surface, while permanent gasses such as N2 pass through the bed unaltered. A front of desublimating CO2 will move through the beds, until breakthrough is observed. At that point, a bed is switched to a recovery step, to remove all previously deposited CO2. A process cycle is finally finished with a cooling step, in which the bed is again refrigerated to its initial temperature. The proposed process concept has several advantages: simple and low cost equipment, no large pressure drops (intrinsic circumvention of plugging) and the possibility to separate CO2, H2O and other impurities simultaneously. 16 Introduction

Chapter 2 gives a detailed explanation of this novel process concept. The dynamic behavior of the process will be described by a simplified model, based on thermodynamic equilibrium and an advanced numerical model. The evolution of temperature and mass deposition profiles will be shown. In order to demonstrate the capture step of the proposed process concept, a small scale experimental setup was designed and constructed to measure axial temperature profiles and CO2 concentrations at the outlet during the capture step. Chapter 3 will give the results of the experiments for a wide range of conditions using different compositions and initial bed temperatures. These experimental results are compared to simulation results. Efficient operation of the proposed concept requires fast CO2 desublimation rates. Limited information is available in literature on these rates. Chapter 4 shows the design of a dedicated setup to measure desublimation rates. Obtained experimental results for different temperatures, pressures and compositions will be shown and a frost growth model is developed to describe the observed behavior. In Chapter 5 it is presented how the complete process cycle including all three steps can be demonstrated. The design of a continuously operated setup will be discussed and results will be shown. Chapter 6 will give a techno-economic evaluation of the process concept. The influence of initial bed temperatures, CO2 concentrations and pressure drops will be investigated. Furthermore the cryogenic concept will be compared to two competing technologies: amine absorption and membrane separation. The proposed process could be applied for alternative gas separations. Chapter 7 shows how biogas purification can be carried out using the cryogenic packed bed concept. It is presented how the process can be operated efficiently and compares the outcomes to a competing technology: Vacuum Pressure Swing Adsorption (VPSA). Finally an outlook is given in Chapter 8, in which the status of the technology and the required future research and development are discussed. 1.3 This thesis 17

Acknowledgment

The author would like to thank Shell Global Solutions International for their ﬁnancial support and involvement in the project. 18 Introduction 2

Cryogenic packed bed process concept

Abstract

This chapter elucidates the novel process concept based on cryogenically cooled dynamically operated packed beds. In order to capture CO2 from a flue gas, three process steps are required: a capture, recovery and cooling step. The dynamic behavior of the axial temperature, gas phase concentrations and mass deposition profiles during these steps have been investigated with two different models: An advanced numerical model including effects such as axial dispersion in the beds, and a simplified model in which developed fronts are assumed to be perfectly well defined, referred to as the ’sharp front approach’. Using the advanced model, it has been demonstrated that the simplified model can well capture the most salient process characteristics even quantitatively, in particular for cases with low axial dispersion. The sharp front approach is an excellent tool to evaluate the influence of process parameters such as the inlet composition and initial bed temperatures on the process performance. The outcomes of this evaluation are presented in this chapter. Lower initial bed temperatures result in more CO2 mass deposition per unit of bed volume, while the specific cooling duty (required cooling duty per 20 Cryogenic packed bed process concept

unit of mass of CO2 captured) reaches a constant value at low initial bed temperatures. Lower CO2 concentrations in the ﬂue gas result in less mass deposition per unit of bed volume and increased required cooling duty.1

2.1 Introduction

This chapter starts with a qualitative description of the different process steps involved: the capture, recovery and cooling step. During these different steps axial temperature, gas phase concentration and mass deposition profiles develop in time. An advanced numerical model based on a pseudo-homogeneous one-dimensional plug flow model with superimposed axial dispersion is developed, which is able to describe the dynamic behavior of the process. Subsequently, a simplified, but computationally much easier and faster model will be presented, in which fronts are assumed to be perfectly well defined (sharp front approach). These two models will be compared, and the chapter concludes with giving an evaluation of the influence of several process parameters, such as the initial bed temperature and inlet composition.

2.2 The process concept

In this section the process concept is described in detail, where the ﬂue gas is represented as a mixture of N2, CO2 and H2O to simplify the description. Continuous separation of these components can be obtained when three packed beds are operated in parallel in three different steps: a capture, recovery and cooling step. These three steps are discussed consecutively below focusing on the evolution of axial temperature and mass deposition proﬁles (see Fig. 2.1).

2.2.1 Capture step

When a gas mixture consisting of N2, CO2 and H2O is being fed at a relatively high temperature Tc,in to an initially cryogenically refrigerated packed bed (at T0), an effective separation between these components can

1This chapter is based on the papers: Tuinier et al., Chem. Eng. Sci. , 65(1), 114-119, 2010 and Tuinier et al., Int. J. Greenhouse Gas Control, 5(4), 694-701, 2011. 2.2 The process concept 21

be accomplished, due to differences in dew and sublimation points. The gas mixture will cool and the packing material will heat, until H2O starts to condense at the packing surface. A certain amount of H2O per volume of packing material (indicated as mH2O in Fig. 2.1a) will condense, until a local equilibrium is reached (at a temperature TH2O). Actually a very small part of the H2O at the front will be frozen to ice, but simulations have revealed that this is a very small part of the H2O and has negligible inﬂuence on the resulting axial temperature and mass deposition pro- ﬁles. The cold energy stored in the packing will be consumed and a front of condensing H2O will move through the bed towards the outlet of the bed. At the same time, previously condensed H2O will evaporate due to the incoming relatively hot gas mixture. Therefore, two fronts of evaporating and condensing water will move through the bed, with a faster moving condensing front. After all water being condensed, the gas mixture will be cooled further until CO2 starts to change phase. At atmospheric pressure CO2 will desublimate directly from gas to solid, and therefore solid CO2 is deposited onto the packing surface. Similar as for H2O, two CO2 fronts will move through the bed: an evaporation and a desublimation front. Again an equilibrium is reached, a certain amount of CO2 (mCO2 ) is deposited at the packing surface at a temperature of TCO2 . Note that in this way an effective separation between CO2 and H2O is accomplished. N2 will not undergo any phase change (as long as T0 is not chosen too low) and will therefore move through the bed unaffected. When the CO2 desublimation front reaches the end of the bed, CO2 may break through and the bed should be switched to a recovery step just before that. 22 Cryogenic packed bed process concept 2 H O H T 2 H O H m c,in T r,in T O 2 Axial position Axial position H 0 0 1 2 t t t t (c) Cooling step 0

mpeaure eratu p em T n sitio o ep d ass M 0 1 t t 2 2 CO * CO ass deposition proﬁles for the capture (a), CO * CO T m 2 2 H O H 2 H O H 2 m T CO CO m T c,in T 2 2 2 O 2 Axial position Axial position H CO CO CO 0 0 1 1 t t t t (b) Recovery step r,in

mpeaure eratu p em T n sitio o ep d ass M 0 2 1 2 t t T CO m 2 O 2 H CO 2 2 t t

2 O O c,in 2 2 2 T H O H CO H CO Axial position Axial position T 1 1 m 2 t t H O H T

(a) Capture step

mpeaure eratu p em T n sitio o ep d ass M Figure 2.1: Schematic axial temperature and corresponding m recovery (b) and cooling (c) step respectively. 2.2 The process concept 23

2.2.2 CO2 recovery step

The first zone of the bed has been heated to Tc,in during the capture step. This heat is used in the recovery step to evaporate the condensed H2O and desublimated CO2. A gas flow consisting of pure CO2 is fed to the bed. When feeding a pure CO2 gas flow at a temperature Tr,in to the packed bed, the gas will be heated up to Tc,in and all fronts will move through the bed, as illustrated in Fig. 2.1b. However, during the initial period of the recovery step the ingoing CO2 will deposit onto the packing. Due to the increase in CO2 partial pressure compared to the capture

step, more CO2 is able to desublimate at the packing surface (from mCO2 to m∗ ), and the bed temperature will slightly increase to T ∗ . Pure CO CO2 CO2 2 is obtained at the outlet of the bed after this new equilibrium is reached. Part of the outgoing CO2 should be compressed for transportation and sequestration, while the other part can be used to recycle to the inlet of the bed at a temperature T , which is slightly higher than T ∗ due to r,in CO2 the heat production associated with the compression in the recycle blower and some unavoidable heat leaks. When all CO2 has been recovered, the bed is switched to a step in which H2O is removed and the bed is cooled simultaneously. Alternatively, the deposited CO2 could be recovered as a liquid, avoiding expensive compression costs required for transportation and storage. This could be accomplished by closing the valves connected to the bed and by introducing heat into the bed. CO2 evaporation occurs and pressure builds up until the system reaches the triple point of pure CO2 and liquid CO2 will be formed. The drawbacks of this process alternative are that pressure vessels are required and that heat should be introduced into the bed for example by means of internal tubes. Both measures will result in a signiﬁcant increase in capital costs. Furthermore not all liquid CO2 might be recovered from the packing, due to the static liquid hold up in the bed. This process option is not further explored in this work.

2.2.3 H2O recovery and cooling step In the last step, the bed is cooled down using a gas flow refrigerated before to temperature T0. The cleaned flue gas can be used for this purpose. Cooling can be performed using a cryogenic refrigerator or by evaporating LNG. H2O is evaporated and removed from the bed during the first period of the cooling step. The N2/H2O mixture can be released to the atmo- 24 Cryogenic packed bed process concept

sphere and when all H2O is recovered, the outgoing flow can be recycled to the inlet of the bed, via a cooler. Temperature and mass deposition profiles are shown in Fig. 2.1c. It should be noted that it is not required to cool the entire bed to T0. The last zone can be kept at Tr,in, as during the capture step this last part will be cooled down by the cleaned flue gas.

2.3 Detailed numerical model 2.3.1 Model description The prevailing heat and mass transfer processes in the periodically operated packed beds have been investigated with a pseudo-homogeneous one-dimensional plug flow model with superimposed axial dispersion. The modeling was based on the following main assumptions: • It is assumed that heat losses to the environment are small (i.e. adiabatic operation) and additionally that a uniform velocity profile exists in the absence of radial temperature and concentration gradients allowing the consideration of the axial temperature and concentration profiles only; • Possible heat transfer limitations between the solid packing and the bulk of the gas phase are accounted for via effective axial heat dispersion (pseudo-homogeneous model);

• The rate of mass deposition and sublimation of CO2 was assumed to be proportional to the local deviation from the phase equilibrium, taking a reasonably short equilibration time constant (g) of 1·10-6 s/m, which was assumed independent of temperature. The rate of sublimation of previously deposited CO2 was assumed to approach a ﬁrst order dependency on the mass deposition when this mass deposition approached zero. The mass and energy conservation equations have been listed in Table 2.1. The constitutive equations for the transport parameters and the mass deposition rate have been summarized in Table 2.2 and 2.3, respectively. The gas phase (mixture) properties have been computed according to Reid et al. (1987), using the pure component data supplied by Daubert and Danner (1985). Uniform initial temperature proﬁles were taken without any mass deposited onto the solid packing, where the gas 2.3 Detailed numerical model 25

phase in the bed was initially N2. Furthermore, the usual Danckwerts- type boundary conditions were applied at the inlet and outlet of the beds. The system of strongly non-linear, coupled partial differential equations was solved using a very efficient finite volume discretization technique, using a second order SDIRK (Singly Diagonally Implicit Runge- Kutta) scheme for the accumulation terms, an explicit 5th order WENO (Weighted Essentially Non-Oscillatory) scheme for the convection terms (with implicit first order upwind treatment using the deferred correction method), second order standard implicit central discretization for the dispersion terms and the standard Newton-Raphson technique for the linearly-implicit treatment of the source terms. Moreover, automatic time step adaptation and local grid refinement procedures have been implemented, making effective use of the WENO smoothness indicators and interpolation polynomials (Smit et al., 2005). The steep temperature and mass deposition gradients in combination with the strongly non-linear sublimation kinetics require a very efficient and stable numerical imple- mentation using higher order implicit schemes.

2.3.2 Simulation results Simulations have been carried out for all three process steps. The used bed properties and process conditions are listed in Table 2.4 and 2.5 respectively. A stainless steel monolithic structure was chosen as packing material, because axial dispersion and pressure drop are minimal for this type of packing material while the volumetric heat capacity is relatively high. The axial temperature and mass deposition profiles during the capture step are shown in Fig. 2.2a and 2.2d respectively. At the chosen ◦ initial bed temperature of -140 C, more than 99% of CO2 is recovered. After 600 seconds the CO2 desublimation front reaches the end of the bed and the capture cycle should be stopped. The conditions at 600 seconds are used as initial conditions for the simulation of the recovery step. Fig. 2.2b and 2.2e show that during the recovery step extra CO2 will be deposited onto the packing surface, as explained in section 2.2.2, and that all deposited CO2 is removed after again 600 seconds. Now the data at the end of the recovery step are used as initial conditions for the cooling step. A refrigerated N2 flow is being fed to the bed and profiles will develop as illustrated in Fig. 2.2c and 2.2f. Note that not the entire bed is cooled down, the last zone is cooled down during the capture step. The 26 Cryogenic packed bed process concept

Table 2.1: Model equations for the 1-D pseudohomogeneous model.

Component mass balances for the gas phase:

∂ωi,g ∂ωi,g ∂ ∂ωi,g εgρg = −ρgvg ∂z + ∂z ρgDeﬀ ∂z ∂t nc ′′ ′′ −m˙ i as + ωi,g m˙ i as Xi=1

Component mass balance for the solid phase:

∂m ′′ i =m ˙ a ∂t i s

Total continuity equation for the gas phase:

nc ∂ (ε ρ ) ∂ (ρ v ) ′′ g g = − g g − m˙ a ∂t ∂z i s Xi=1

Energy balance (gas and solid phase):

∂T (ε ρ C + ρ (1 − ε )C ) = −ρ v C ∂T + ∂ λ ∂T g g p,g s g p,s ∂t g g p,g ∂z ∂z eﬀ ∂z nc ′′ − m˙ i as∆Hi Xi=1

Pressure drop over packing:

∂P f 1 2 14.9 dh = −4 ρgvg with: f = 1 + 0.0445 Re ∂z dh 2 Re r L 2.3 Detailed numerical model 27

Table 2.2: Heat and mass transfer coefﬁcients for a monolith packing.

Effective axial heat dispersion (Vortmeyer and Schaefer, 1974):

2 ρgvgCp,g 1 λeﬀ = (1 − εg) λs + εg αg,sas

Gas to solid heat transfer coefﬁcient (Hawthorn, 1974):

0.45 λg dh ρgvgdh αg,s = 2.978 1 + 0.095RePr with: Re = dh Lc ηgεg

Axial mass dispersion (Cybulski and Moulijn, 1994):

2 2 vg dh,c 1 with: Dax = Deff ,i + Deff ,i = nc 192Deff ,i yj,g D Xj=1 i,j

Table 2.3: Mass deposition rate.

Mass deposition rate:

σ σ ′′ g (y P − P ) if y P ≥ P m˙ = i,s i i,s i i g (y P − P σ) mi if y P

Gas-solid equilibrium:

3082.7 − P σ (T ) = exp 10.257 − + 4.08ln T − 2.2658 · 10 2T CO2 T

sub 5 −1 ∆HCO2 = 5.682 · 10 J · kg 28 Cryogenic packed bed process concept

Table 2.4: Bed properties used in the numerical study.

Length bed [m] 8 Diameter bed [m] 3 Packingtype Steelmonolith Solids density [kg · m−3] 7750 Channel diameter [m] 6.76 · 10−4 Wall thickness [m] 1.34 · 10−4 Porosity [-] 0.7 Surface area [m2 · m−3] 4124 5 −1 −1 i Heat capacity [J · kg · K ] ciT Xi=0 with: c0 −203.75 c1 6.4335 −2 c2 −2.4320 · 10 −5 c3 4.6266 · 10 −8 c4 −4.2721 · 10 −11 c5 1.5296 · 10

results show that CO2 and H2O capture can be integrated in one single bed. However, the temperature profile during the recovery step shows that the heat stored in the first zone during the capture step is only just sufficient to remove H2O from the bed. The hot zone is moved through the bed, but due to axial heat dispersion this hot zone will be spread out over the bed, which is well visible in Fig. 2.2b. When feeding the gas mixture at realistic flue gas temperatures (which are generally lower than 250◦C) during the capture step, not sufficient heat is stored in the packing to evaporate previously condensed water again. A possibility would be to introduce extra heat into the bed in the initial period of the recovery step. However, more practical is to carry out the H2O capture step in a separate smaller bed, which can be cooled down to temperatures much higher than the initial bed temperature of the CO2 capture bed. 2.3 Detailed numerical model 29

Table 2.5: Conditions used in the numerical study.

Capture Recovery Cooling ◦ Tin [ C] 250 −70 −140 -1 Φin [kg · s ] 20 100 70

yCO2,in [-] 0.1 1.0 0.0

yH2O,in [-] 0.01 0.0 0.0

30 Cryogenic packed bed process concept

8 8 0 s 0 s 300 s 600 O 0 s 0 O 2 H 6 6 4 4 (f) (c) Axial positionAxial [m] positionAxial [m] 2 2 0 0 0 0 80 60 40 20 50 -50 100 300 250 200 150 100

-100 -150

] ·m g [k n sitio o ep d ass M

mpeaur ° ] [°C re eratu p em T

3-3

8 8 0 s 0 s 300 s 600 ition (d - f) proﬁles for the capture, recovery 6 6 an be found in Table 2.4 and 2.5 respectively. 4 4 0 s 0 s 300 2 2 (e) (b) CO CO Axial positionAxial [m] positionAxial [m] 2 2 O O O s 600 O 2 2 2 H H H 0 0 0 0 80 60 40 20 50 -50 100 300 250 200 150 100

-100 -150

] ·m g [k n sitio o ep d ass M

mpeaur ° ] [°C re eratu p em T

3-3

8 8 0 s 0 s 300 s 600 300 s 300 s 600 2 2 6 6 CO CO O O O 2 2 4 4 H H (a) (d) Axial positionAxial [m] positionAxial [m] 2 2 0 0 0 0 80 60 40 20 50 -50 100 300 250 200 150 100

-100 -150

] ·m g [k n sitio o ep d ass M

mpeaur ° ] [°C re eratu p em T 3-3 Figure 2.2: Simulated axialand temperature cooling (a - step. c) Bed and properties mass and depos operating conditions c 2.4 Simpliﬁed model: Sharp front approach 31

2.4 Simpliﬁed model: Sharp front approach

The process concept can be described by the advanced numerical model as detailed in the previous section. However, when assuming that the fronts which are formed during the different process steps are perfectly well defined (sharp), a simplified and relatively easy to solve and fast model can be developed, referred to as the ‘sharp front approach’. In the first place this approach provides a very useful tool to quickly investigate the influence of process parameters on process behavior. Further- more it can be used in conceptual design studies. This section describes this sharp front approach and compares the outcomes with the more advanced numerical model. Finally, the influences of several process parameters are studied.

2.4.1 Model description Capture step The model is derived for capturing a component i from a binary gas mixture consisting of components i and j (but could be easily extended to multicomponent mixtures). When feeding this mixture to a refrigerated bed, two fronts are formed: a ’frost’ and a ’defrost’ front as depicted in Fig. 2.3a. The defrost front moves from zd,1 to zd,2 during a time period ∆t. The mass of i evaporated is equal to the distance the front moved multiplied with the bed cross sectional area A and the amount of mass deposited per unit of bed volume mi. Due to this evaporation of previously

deposited i, the mass ﬂow of i after the defrost front (Φ1ωi1 ) is equal to the

inlet mass ﬂow (Φ2ωi2 ) plus the amount of i evaporated. This results in the following component mass balance:

Φ1ωi1 = Φ2ωi2 + Amivd (2.1)

In which the front velocity vd is deﬁned as:

(z 2 − z 1) v = d, d, (2.2) d ∆t Due to the evaporation of previously frosted i, the mass fraction after the ﬁrst front (ωi1 ) will be higher than the inlet mass fraction (ωi2 ). Also an overall mass balance can be formulated. The total mass ﬂow after 32 Cryogenic packed bed process concept

(a) (b)

Figure 2.3: Axial temperature, mass deposition and gas concentration proﬁles used in the derivation of the sharp front approach for the capture step (a) and recovery step (b). 2.4 Simpliﬁed model: Sharp front approach 33

the defrost front (Φ1) is equal to the inlet ﬂow (Φ2) plus the amount of component i evaporated:

Φ1 = Φ2 + Amivd (2.3)

At the frost front, i is deposited onto the packing surface, therefore the outlet mass ﬂow of component i is equal to the inlet ﬂow minus the amount of i deposited in time ∆t, which can be written as:

Φ0ωi0 = Φ1ωi1 − Amivf (2.4)

The velocity of this frost front vf is described as:

(z 2 − z 1) v = f, f, (2.5) f ∆t Again, also an overall mass balance can be formulated. The total mass ﬂow after the frost front (Φ0) is equal to Φ1 minus the amount of component i deposited onto the packing surface:

Φ0 = Φ1 − Amivf (2.6)

For both fronts also energy balances can be formulated. At the defrost front heat is required to heat up the packing material from the saturation temperature (T1) to the inlet temperature (T2). Furthermore heat is consumed due to the (endothermic) sublimation of component i. The required energy is provided by the feed gas, which is cooled down from T2 to T1. The energy balance results in:

Avd [ρsCp,s (T2 − T1)+ mi∆Hi] = Φ2 (T2 − T1)(ωi2 Cp,i + ωj2 Cp,j ) (2.7)

At the frost front the exothermic desublimation is included in the energy balance, resulting in:

Avf [ρsCp,s (T1 − T0) − mi∆Hi] = Φ0 (T1 − T0)(ωi0 Cp,i + ωj0 Cp,j) (2.8)

So, three balances have been derived for each front (component mass, overall mass and energy balances) giving a total of six balances. There are eight unknowns (T1, Φ0, Φ1, ωi0 , ωi1 , vd, vf and mi). The two mass fractions of component i after the defrost front (ωi1 ) and in the outlet (ωi0 ) are related to the temperature by the phase equilibrium: 34 Cryogenic packed bed process concept

σ Pi (T ) Mi ωi = (2.9) Ptot M in which M is the average molar weight. Finally eight equations and eight unknowns are obtained. The gas and solid heat capacities are dependent on the temperature and are normally described by polynomial correlations and the vapor pressures as function of the temperature are normally described by exponential relations. Therefore, a system of non-linear equations is obtained, which can be solved by standard root seeking methods such as the Newton-Raphson technique.

Recovery step During the recovery step, the bed is fed with pure component i. Three fronts will develop, as illustrated in Fig. 2.3b. Initially, additional i will deposit onto the packing surface, due to the higher pressure of i compared to the capture step. Therefore, a front will be formed and moves towards the outlet with a velocity vf,r. The new amount of mass deposited ∗ per unit of bed volume is now indicated as mi . The component mass, the overall mass balance and the energy balance for this frost front are listed below:

∗ ∗ ∗ Φ1,rωi1,r = Φ1,rωi1,r − A (mi − mi) vf,r (2.10)

∗ ∗ Φ1,r = Φ1,r − A (mi − mi) vf,r (2.11)

∗ ∗ Avf,r [ρsCp,s (T1 − T1) − (mi − mi) ∆Hi] ∗ = Φ1,r (T1 − T1)(ωi1,rCp,i + ωj1,rCp,j) (2.12)

Also a defrost front will move through the bed during the recovery step, for which the next mass and energy balances can be formulated ∗ (Note that ωi1,r and ωi2,r are both unity when the bed is recovered with pure i):

∗ ∗ Φ1,rωi1,r = Φ2,rωi2,r + Amivd,r (2.13) 2.4 Simpliﬁed model: Sharp front approach 35

∗ Φ1,r = Φ2,r + Amivd,r (2.14)

∗ ∗ Avd,r [ρsCp,s (T2 − T1 )+ mi ∆Hi] ∗ = Φ2,r (T2 − T1 )(ωi2,rCp,i + ωj2,rCp,j) (2.15)

Similar to the capture step, a system of non-linear equations is obtained which can be solved using a root seeking technique. During the recovery step a third front is being formed: The front closest to the inlet (moving with velocity vr). This front is formed due to the difference in inlet temperature (T3) and the temperature of the bed at the initial zone after the capture step (T2). No phase change of component i takes place at this front, therefore the front velocity can be described with:

Φ3,r (ωi3,rCp,i + ωj3,rCp,j ) vr = (2.16) AρsCp,s When the heat capacities of the gas and solid phase are independent of the temperature, the front velocity is not a function of the temperature.

Cooling step During the cooling step no phase change is involved. The bed is cooled down with an inert gas, fed at temperature T0. A temperature front will move through the bed, with a velocity:

Φ0,c (ωi0,cCp,i + ωj0,cCp,j ) vc = (2.17) AρsCp,s All equations for the three steps for the sharp front approach have been summarized in Table 2.6.

2.4.2 Simulation results The outcomes of the sharp front approach are presented in this section, and compared to the simulation results of the advanced numerical model. Fig. 2.4 shows axial temperature and mass proﬁles which are formed after 400 seconds, when feeding a binary N2/CO2 mixture at an inlet temperature of 150◦C. Other conditions are equal to those listed in Table 2.4 36 Cryogenic packed bed process concept

Table 2.6: Equations for sharp front approach.*

Capture step:

Defrost front: Φ1ωi1 =Φ2ωi2 + Amivd Φ1 =Φ2 + Amivd − − Avd [ρsCp,s (T2 T1)+ mi∆Hi]=Φ2 (T2 T1)(ωi2 Cp,i + ωj2 Cp,j )

Frost front: − Φ0ωi0 =Φ1ωi1 Amivf Φ0 =Φ1 − Amivf − − − Avf [ρsCp,s (T1 T0) mi∆Hi]=Φ0 (T1 T0)(ωi0 Cp,i + ωj0 Cp,j )

Recovery step:

Φ3,r (ωi3,r Cp,i+ωj3,r Cp,j ) Temperature front: vr = AρsCp,s

Defrost front: ∗ ∗ Φ1,rωi1,r =Φ2,rωi2,r + Amivd,r ∗ Φ1,r =Φ2,r + Amivd,r − ∗ ∗ − ∗ Avd,r ρsCp,s T2 T1 + mi ∆Hi =Φ2,r T2 T1 (ωi2,rCp,i + ωj2,rCp,j ) Frost front: ∗ ∗ − ∗ − Φ1,rωi1,r =Φ1,rωi1,r A mi mi vf,r ∗ − ∗ − Φ1,r =Φ1,r A mi mi vf,r ∗ − − ∗ − ∗ − Avf,r ρsCp,s T1 T1 mi mi ∆Hi =Φ1,r T1 T1 (ωi1,rCp,i + ωj1,rCp,j ) Cooling step:

Φ0,c(ωi0,cCp,i+ωj0,cCp,j ) Temperature front: vc = AρsCp,s

* Note that for a temperature dependent heat capacity, the term Cp (Ty − Tx) Ty should be replaced by: C dT Z p Tx 2.5 Process analysis 37

and 2.5. It can be observed that the front positions, equilibrium temperature, as well as the amount of mass deposited per unit of bed volume match very well between the two approaches. Although heat and mass dispersion is included in the advanced model, fronts are reasonably sharp during the capture step. Especially the frost front is well defined, which can be attributed to a ‘self sharpening’ effect of the exothermic desublimation. In order to demonstrate that the outcomes of the advanced model approach the simplified model even closer, an additional simulation has been carried out using the detailed model in which the actual heat and mass dispersion coefficients have been decreased by a factor 100. The excellent agreement between the two models can be discerned from Fig. 2.4. Also for other inlet compositions and initial bed temperatures the two models agree very well (results are not included here). The recovery step has also been simulated using the two models. As already explained before, additional CO2 will deposit onto the packing in the initial phase of the recovery step. This is described by both models and again matches well, as observed in the temperature and mass deposition profiles after 10 seconds in Fig. 2.5a and 2.5b respectively. It is observed again that when assuming low axial heat and mass dispersion, the solution of the advanced model is approaching the sharp front approach. Finally the temperature and mass deposition profiles have been computed for the recovery step after 400 seconds, as illustrated in Fig. 2.5c and 2.5d. It can be observed that dispersion is playing a more prominent role during the recovery step, which is related to higher flow rates in comparison to the capture step. For that reason, the sharp front approach is especially suited to describe the capture step.

2.5 Process analysis

This section aims at giving an overview of the influences of several process parameters on the process performance, using the sharp front approach. The influences of the initial bed temperature, inlet composition, inlet temperature and packing material are analyzed on the basis of two aspects: the amount of CO2 deposited per unit of bed volume and the required specific cooling duty, which is defined as:

V (1 − ε ) ρ C (T − T0) J Q = bed g s p,s s (2.18) (ΦCO2,in − ΦCO2,out)tstep kgCO2 38 Cryogenic packed bed process concept

100

SF approach SF approach

150

Numerical model ] Numerical model -3

Numerical model - low dispersion Numerical model - low dispersion

100

-50

Temperature [°C] 20

-100 Mass deposition [kg·m

-150

0 2 4 6 8 0 2 4 6 8

Axial position [m] Axial position [m] (a) (b)

Figure 2.4: Simulated axial temperature (a) and mass deposition (b) profiles for the capture step. The line indicated with ‘low dispersion’ shows the profile calculated with the advanced model, in which the axial mass and heat dispersion coefficients have been decreased by a factor 100.

The numerator represents the amount of energy required to cool down the bed after a recovery step from Ts to a temperature T0, which is the initial bed temperature before starting a capture step. The temperature ◦ Ts is assumed to be -70 C for all cases. Initially the bed temperature at ◦ the zone where CO2 is deposited will increase to -78 C during the recovery step, as pure CO2 is being fed to the system and is being deposited additionally onto the packing surface. However, the outlet ﬂow during the recovery step is recycled and will increase slightly in temperature and is assumed to be at -70◦C, which will therefore be the initial bed temperature before a cooling step is started. In practice, not the entire bed will be at this temperature at the end of the recovery step. The zone located close to the outlet of the bed will have a slightly higher temperature. On the other hand, as explained earlier, it is actually not necessary to cool down the entire bed to T0 during the cooling step and therefore the numerator gives a realistic value for the required cooling duty. The denominator in

(2.18) gives the amount of CO2 captured during a capture step. ΦCO2,in and ΦCO2,out are the inlet and outlet mass ﬂow rates of CO2 during a capture step. 2.5 Process analysis 39

120

SF approach SF approach 150 ] Numerical model Numerical model

100 -3

Numerical model - low dispersion Numerical model - low dispersion 100

-50 40 Temperature [°C]

-100

20 Mass deposition [kg·m

-150

0 2 4 6 8 0 2 4 6 8

Axial position [m] Axial position [m] (a) (b)

120

SF approach 150 ] Numerical model

100 -3

Numerical model - low dispersion 100

-50 40 Temperature [°C]

-100

SF approach 20

Numerical model Mass deposition [kg·m

-150

Numerical model - low dispersion

0 2 4 6 8 0 2 4 6 8

Axial position [m] Axial position [m] (c) (d)

Figure 2.5: Simulated axial temperature and mass deposition profiles for the recovery step, after 10s (a), (b) and 400s (c), (d). The lines indicated with ‘low dispersion’ show the profiles calculated with the advanced model, in which the axial mass and heat dispersion coefficients have been decreased by a factor 100. 40 Cryogenic packed bed process concept

The amount of CO2 deposited per unit of bed volume is an important indicator of the required capital costs, while the calculated speciﬁc cooling duty gives a good indication of the energy requirements for different cases. Blowers and compressors are responsible for part of the power consumption. The techno-economic evaluation described in Chapter 6 includes these requirements and discussed the economic feasibility of the newly developed process concept.

2.5.1 Initial bed temperature

The required specific cooling duty and CO2 mass deposition as function of the initial bed temperature are illustrated in Fig. 2.6. Below -130◦C the initial bed temperature hardly affects the specific cooling duty, the additional energy required to cool the bed to lower temperatures will at the same time result in a correspondingly higher CO2 storing capacity of the bed and therefore causing a constant specific cooling duty. However, at initial bed temperatures above -120◦C, the specific cooling duty will increase strongly. This is related to the decreasing amount of CO2 being captured, which will decrease exponentially above -120◦C. For example: when feeding a N2/CO2 mixture containing 10 vol.% CO2 to a bed cooled ◦ at -120 C, 90% of the fed CO2 is recovered. However, when feeding the ◦ same mixture to a bed cooled at -110 C, only 12% CO2 is recovered. Therefore, a relatively low amount of extra cooling will result in much higher CO2 recovery rates. This effect is directly related to the exponential temperature dependency of the equilibrium CO2 vapor pressure.

2.5.2 CO2 inlet concentration

When the CO2 fraction in the ﬂue gas feed decreases, the speciﬁc cooling duty will increase, as shown in Fig. 2.6a. This can be explained by the fact that less CO2 is stored in the bed at lower inlet CO2 concentrations (see Fig. 2.6b), while the energy required to cool down the bed remains the same.

2.5.3 Inlet temperature A higher gas inlet temperature will cause the defrost front to move faster, because more heat is available for desublimation. This causes the CO2 2.5 Process analysis 41

7 80

y : ] y :

CO2,in ] 2

CO2,in 70

0.05 -3 bed 0.05 CO

60 0.10

0.10 -1

0.15

0.2

10 1 Mass deposition [kg·m Cooling duty [MJ·kg

-160 -150 -140 -130 -120 -110 -160 -150 -140 -130 -120 -110

Initial bed temperature [°C] Initial bed temperature [°C] (a) (b)

Figure 2.6: Speciﬁc cooling duty (a) and mass deposition (b) as function of the

initial bed temperature for different inlet CO2 fractions.

concentration after the defrost front to increase at the same time. Also more CO2 will deposit per unit of packing volume and the frost front moves faster, and consequently the step time reduces. Therefore, the specific cooling duty will increase when increasing the inlet temperature. However, the specific cooling duty reaches an asymptotic value when increasing the temperature. This is related to the fact that the heat required to heat up the packing (and cool down the gas) relative to the heat involved in the evaporation of previously deposited CO2 is playing a dominant role at higher inlet temperatures. When for example feeding ◦ a mixture containing 10% CO2 to a bed refrigerated to -140 C at an inlet temperature of 150◦C, the amount of heat required for sublimation is only 8.2% of the amount of energy involved in cooling down the gas. It should be noted that the contribution of desublimation at the frost front does play a significant role in the energy balance.

2.5.4 Bed properties The packing properties (i.e. material or porosity) will not inﬂuence the speciﬁc cooling duty. When the heat capacity of a packing material is for example doubled (and the bed dimensions remain the same), the amount of cooling will also be doubled. However, the amount of CO2 deposited per unit of bed volume will change proportionally, and therefore the spe- 42 Cryogenic packed bed process concept

cific cooling costs will not be influenced. However, bed dimensions are influenced when changing the packing material and keeping step times constant and therefore capital costs are influenced by the packing choice. On the other hand, the packing choice will influence the pressure drop and axial dispersion: Effects which are not accounted for in this analysis, but which will be included in the techno-economic analysis in Chapter 6.

2.6 Discussion and conclusions

A novel cryogenic CO2 capture concept, based on dynamically operated packed beds has been presented in this chapter. An advanced numerical model has been derived, which is able to describe the development of axial temperature, mass deposition and gas concentration profiles in the packed beds. In addition to the advanced numerical model, a simplified model based on a sharp front approach has been presented. This sharp front approach offers the possibility to carry out quick calculations and evaluate process designs. The proposed process concept shows several advantages compared to cryogenic separation based on conventional heat exchangers. In the first place, plugging is intrinsically avoided. Due to the limited amount of cold stored in the packing material, also a limited amount of CO2 is desublimated. In Fig. 2.5b it can for example be observed that the amount of 3 CO2 during the recovery step is approximately 80 kg/m , which is corresponding to a volume fraction of approximately 0.06, depending on the density of solid CO2. The gas void fraction of the used packing was much higher (0.7) and plugging or an increase in pressure drop is therefore avoided. Another advantage of the proposed concept is the very high purity of the treated gas. An initial bed temperature of -140◦C will result in a CO2 fraction in the outlet of less than 0.1% during the capture step. The proposed concept can therefore recover more than 99% of the CO2 when feeding a mixture containing 10 vol.% CO2, while for example scrubbing technology is only able to capture 90% of CO2 at acceptable absorber sizes. Another possible advantage of the proposed concept is that other polu- tants such as NOx, SOx and H2O can be captured simultaneously, avoiding expensive pre-treatment. It was demonstrated in this chapter that H2O and CO2 can be captured simultaneously. However, it was also ex- 2.6 Discussion and conclusions 43 plained that the allowed water content in the flue gas is limited, due to the heat required for evaporation of the condensed H2O in the recovery step. Finally, an additional advantage of the concept is that simple low pressure vessels packed with low cost inert packing material can be used, reducing capital costs.

Notation

A cross sectional area, [m2] 2 3 as specific solid surface area per unit bed volume, [m /m ] ci constant used in polynomial for heat capacity Cp heat capacity, [J/kg/K] D diffusion coefficient, [m2/s] 2 Deff effective diffusion coefficient, [m /s] dh hydraulic diameter, [m] f friction factor, [-] g mass deposition rate constant, [s/m] L monolith length, [m] Mi molecular weight of component i, [kg/mol] 3 mi mass deposition of component i per unit bed volume, [kg/m ] ′′ m˙ i mass deposition rate per unit surface area for component i, [kg/m2/s] nc number of components, [-] P pressure, [Pa] Pr Prandtl number (Cp,gηg/λg)

Q Speciﬁc cooling duty, [J/kgCO2 ] Re Reynolds number (ρgvgdh/ηgεg) t time, [s] T temperature, [K], [◦C] 3 Vbed bed volume, [m ] v front velocity, [m/s] v superﬁcial velocity, [m/s] y mole fraction, [-] z axial coordinate, [m]

Greek letters 2 αg,s heat transfer coefﬁcient solids - gas bulk, [W/m /K] ∆Hi enthalpy change related to the phase change of component i, 44 Cryogenic packed bed process concept

[J/kg] εg bed void fraction, [-] η viscosity, [kg/m/s] λ thermal conductivity, [W/m/K] λeff effective conductivity, [W/m/K] ρ density, [kg/m3] Φ mass ﬂow, [kg/s] ω mass fraction, [-]

Subscripts 0 initial c capture d defrost f frost g gas phase i component i in inlet r recovery s solid phase

Superscripts σ equilibrium sub sublimation 3

Experimental demonstration of the concept

Abstract

The fundamental principles of the novel process concept for cryogenic CO2 capture have been elucidated in Chapter 2. This chapter presents an experimental demonstration of the concept. For this purpose, an experimental setup has been constructed in which the evolution of the prevailing axial temperature profiles can be measured accurately during the capture step. Furthermore, the outlet CO2 concentration has been monitored during measurements. Experiments have been carried out for N2/CO2 mixtures with and without the addition of H2O. The influences of the initial bed temperature and inlet gas composition on the evolution of the axial temperature profiles have been studied. Finally, the experimental profiles have been compared with profiles calculated using the detailed numerical model outlined in Chapter 2. When accounting for radiation effects, the model can very well describe the experimental observations.1

1This chapter is based on the papers: Tuinier et al., Chem. Eng. Sci. , 65(1), 114-119, 2010 and Tuinier et al., Int. J. Greenhouse Gas Control, 5(4), 694-701, 2011. 46 Experimental demonstration of the concept

3.1 Introduction

A novel process concept to cryogenically capture CO2 from ﬂue gases was presented in Chapter 2. This chapter gives an experimental validation of the capture step. The used experimental setup and procedure will be described ﬁrst. After that, experimental results for measurements with N2/CO2 and N2/CO2/H2O mixtures are presented. Finally, the obtained results are compared to outcomes of the advanced numerical model.

3.2 Experimental setup and procedure

A flowsheet and a picture of the experimental setup are shown in Fig. 3.1 and 3.2 respectively. A glass tube (OD × ID × L = 40 × 35 × 300 mm) surrounded by a glass vacuum jacket (OD × ID = 60 × 55 mm) was filled with monodisperse blue spherical glass beads (dp = 4.04 mm, ρs = 3 2547 kg/m ). The bed was cooled with a N2 gas flow which was refrigerated in a coil positioned in a liquid nitrogen bath. After cooling, the feed was switched to a N2/CO2 mixture with or without the addition of H2O. These mixtures were prepared by controlling the gas flow rates of N2 and CO2 with mass flow controllers (Bronkhorst El-flow). This mixture could then be fed to a saturator filled with demineralized water. The water content of the feed flow could be varied by changing the temperature in the saturator. The flow from the saturator could be further overheated by using an electrically traced line. The temperatures in the bed were measured along the bed length in the radial center with 11 thermocouples (Thermo-Electric K-type) at every 3 cm in axial direction. The pressure at the inlet of the bed was monitored using an analogue pressure indicator. The CO2 content in the outlet stream was analyzed with an IR-analyzer (Sick-Maihak, s610, 0-3 vol.%). The front of sublimated CO2 was visually inspected with a camera. Axial temperature profiles have been measured for different initial bed temperatures and inlet mole CO2 and H2O fractions (see Table 3.1). 3.2 Experimental setup and procedure 47 Figure 3.1: Flowsheet of the vacuum insulated packed bed. 48 Experimental demonstration of the concept

Table 3.1: Experimental conditions

Experiment #1 #2 #3 #4 #5 #6 Initial bed temperature [◦C] −140 −140 −140 −140 −120 −140 Inlet CO2 mole fraction [-] 0.2 0.3 0.2 0.1 0.2 0.2 Inlet H2O mole fraction [-] 0 0 0.022 0.020 0.021 0.046 Total mass ﬂow [10−4 kg/s] 2.55 2.68 2.35 2.23 2.35 2.39

3.3 Results

The bed was cooled using refrigerated N2 until the bed reached a station- ary temperature proﬁle. Due to heat radiation into the system this initial proﬁle is slightly increasing (almost linearly) from the inlet. The temperature difference between the inlet and outlet is typically about 20◦C, which is relatively small compared to the temperature difference between the refrigerated bed and the gas being fed during the capture cycle. It should be noted that during experiments no pressure excursions were observed, and the pressure drop was very small.

3.3.1 N2/CO2 mixtures

When feeding N2/CO2 mixtures to the refrigerated bed, a moving front of deposited CO2 was observed visually as depicted in Fig. 3.4. The axial temperature proﬁles at several time steps for experiment #1 are shown in Fig. 3.3a. After approximately 200 seconds the frost front reached the end of the bed and CO2 breakthrough was detected in the outlet stream, as shown in Fig. 3.5. The mixture is fed through the same inlet tube as the refrigerated N2 during the cooling step. Therefore the temperature of the packing at the inlet does not attain ambient temperatures immediately, but increased slowly as cam be seen from Fig. 3.3a. The CO2 content in experiment #2 was increased to 30 vol.% CO2. The increased CO2 inlet concentration resulted in a higher saturation temperature (-91.5◦C versus -94.5◦C) and therefore the packing storage capacity slightly increased. However, due to the higher molar CO2 feed ﬂow rate, the front velocity of the frost front increased. Fig. 3.3b shows that CO2 breakthrough already occurred after about 150 seconds. 3.3 Results 49

Figure 3.2: Picture of the experimental setup. 50 Experimental demonstration of the concept

40 40

50 s 50 s

20 20

100 s 100 s

0 0

150 s 150 s

-20 -20

200 s

-40 -40

-60 -60

-80 -80

-100 -100 Temperature [°C] Temperature [°C]

-120 -120

-140 -140

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10 0.15 0.20 0.25 0.30

Axial position [m] Axial position [m] (a) (b)

Figure 3.3: Experimental (markers) and simulated (lines) axial temperature pro- ﬁles for experiments #1 (a) and #2 (b).

Figure 3.4: CO2 ice formed at the packing surface during a capture step. 3.3 Results 51

Experiment

Simulation 2

10 Vol.% CO

0 50 100 150 200 250 300

Time [s]

Figure 3.5: Experimental (markers) and simulated (line) CO2 concentration in the outlet during experiment #1.

3.3.2 N2/CO2/H2O mixtures Fig. 3.7a shows axial temperature profiles measured during experiment #3. Again deposition of CO2 occurs and a CO2 front develops, similar to experiment #1. The gas feed mixture also contains H2O during this experiment, resulting in H2O condensation at the packing surface, which is shown in Fig. 3.6. In between the zones in which H2O is condensed and CO2 is desublimated, a small amount of H2O ice is also observed. This is caused by the decrease in the H2O concentration and the temperature at the condensing front. At some point the mixtures reaches conditions below the triple point of H2O and desublimation of H2O is observed. The amount of desublimated H2O remains constant during a capture step and no influence on the temperature profiles is observed. The effect of H2O condensation on the axial temperatures profiles is small in the first 200 seconds of the measurement. However, after CO2 breakthrough, the gas flow was not stopped in order to show further development of the H2O front. Fig. 3.7b shows that a water front is also moving through the bed and that an equilibrium is formed at a temperature of approximately ◦ 28 C. Lowering the CO2 inlet concentration from 20% to 10% in exper- ◦ iment #4 results in a lower equilibrium temperature for CO2 (-98 C), as illustrated in Fig. 3.7c. The front of freezing CO2 will move slower in this case, breakthrough is observed after approximately 350 seconds. The development of the H2O front is not influenced by the changed CO2 in- 52 Experimental demonstration of the concept

let concetration, as shown in Fig. 3.7d. Fig. 3.8a and 3.8b show axial temperature profiles for experiment #5 with a higher initial bed temperature. This higher bed temperature results in shorter cycle times, because less CO2 is stored per unit of bed volume. Moreover, the purity of the cleaned flue gas will be somewhat lower (98.8 % N2 compared to 99.9 % in the other experiments). Finally, the effect of an increased H2O inlet concentration can be observed in Fig. 3.8c and 3.8d. The evolution of the CO2 fronts is not much influenced, but a clear effect is visible for the H2O front. The equilibrium temperature reaches a higher value of approximately 40◦C, compared to 28◦C in the other experiments.

3.4 Simulations

The experimental results have been compared with profiles calculated by the advanced numerical model, which was described in Chapter 2. The packing material used for the experiments consisted of spherical particles. Therefore the correlations listed in Table 2.2, which are valid for a structured monolith packing are not applicable here, and are replaced by the correlations summarized in Table 3.2. As no information on the sublimation rates is available in literature, the equilibrium time constant (g) was determined by comparing simulation results with the experimental findings. Results are best described when using a constant of about 1·10-6 s/m. Chapter 4 focuses on desublimation rates in more detail. The initial temperature profile used in the simulations is taken from the experiments. As mentioned before, the initial bed temperature is not totally uniform, but is increasing slightly from the inlet towards the outlet. This is caused by the heat leak into the system. To account for this heat leak in more detail, the tube was first cooled down, then cooling was stopped and the temperature rise was measured as function of the time in the radial center and close to the tube wall. It was found that the temperature difference between these two locations was minimal and that the temperature rise could be well described by an additional radiative energy influx. Therefore, the following contribution was added to the energy balance:

4 4 4 σ Th − Tc J φrad = 3 (3.1) dtube 1 Ac 1 m s ǫ + A ǫ − 1 c h h 3.4 Simulations 53

Figure 3.6: Desublimated and condensed components at the packing surface observed during a capture step (feed enters from the bottom). Three zones are observed (from bottom to top): liquid H2O, solid H2O and solid CO2. 54 Experimental demonstration of the concept

100 40

100 s 400 s 80 20

200 s 1000 s 60

-20

-40

-60

-20

-80

-40

-100

-60 Temperature [°C] Temperature [°C]

-120 -80

-140 -100

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10 0.15 0.20 0.25 0.30

Axial position [m] Axial position [m] (a) (b)

100 40

100 s 400 s 80 20

200 s 1000 s 60

-20

-40

-60

-20

-80

-40

-100

-60 Temperature [°C] Temperature [°C]

-120 -80

-140 -100

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10 0.15 0.20 0.25 0.30

Axial position [m] Axial position [m] (c) (d)

Figure 3.7: Experimental (markers) and simulated (lines) axial temperature pro- ﬁles for experiments #3 (a) (b) and #4 (c)(d). 3.4 Simulations 55

100 40

100 s 400 s 80 20

200 s 1000 s 60

-20

-40

-60

-20

-80

-40

-100

-60 Temperature [°C] Temperature [°C]

-120 -80

-140 -100

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10 0.15 0.20 0.25 0.30

Axial position [m] Axial position [m] (a) (b)

100 40

100 s 400 s 80 20

200 s 1000 s 60

-20

-40

-60

-20

-80

-40

-100

-60 Temperature [°C] Temperature [°C]

-120 -80

-140 -100

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10 0.15 0.20 0.25 0.30

Axial position [m] Axial position [m] (c) (d)

Figure 3.8: Experimental (markers) and simulated (lines) axial temperature pro- ﬁles for experiments #5 (a) (b) and #6 (c) (d). 56 Experimental demonstration of the concept

In which σ is the Stefan-Boltzmann constant, ǫ the integral emissivity and the subscripts h and c stand for the hot and cold side. In the used experimental setup, both the hot side (jacket) and cold side (tube) are composed of glass, and therefore an equal integral emissivity of 0.9 is assumed. Furthermore, it was found that the inlet temperature of the gas was only rising slowly in time after switching to the capture step, due to cold stored in the inlet of the tube, insulation material, piping etc. Therefore, the inlet temperature (and corresponding composition) used in simulations was based on the measured inlet temperature. Finally, it should be noted that the heat capacity of the glass wall was included in simulations (in the accumulation term). The resulting calculated temperature profiles have been plotted in the same figures as in which the experimental profiles were shown. It can be concluded that the developed model is very well capable to describe all experimental findings.

3.5 Discussion and conclusions

The novel process concept for cryogenic CO2 capture has been demonstrated experimentally in this chapter. The developed numerical model is very well capable to describe the experiments for N2/CO2 mixtures, with and without the addition of H2O. Both the axial temperature proﬁles, as the outlet concentration can be very well predicted by the numerical model, when assuming a desublimation equilibration time constant g equal to 1·10-6 s/m. A detailed study on the desublimation rates is presented in Chapter 4. The inﬂuences of initial bed temperature and inlet gas composition were shown. The effect of these parameters on the process economics will be discussed in Chapter 6. This chapter only focused on the capture step. An experimental setup in which all three process steps (capture, recovery and cooling) can be operated in parallel has been constructed as well. A description of this setup and the results are discussed in Chapter 5. 3.5 Discussion and conclusions 57

Table 3.2: Heat and mass transfer coefﬁcients.

Effective axial heat dispersion in a transient packed bed (Vortmeyer and Berninger, 1982):

2 2 RePr λg Re Pr λg λeﬀ = λbed,0 + + Peax 6(1 − εg) Nu

In which Peax is calculated according to Gunn and Misbah (1993): 2 −24 p Re Peax = 1−p p = 0.17 + 0.33exp λbed,0 is calculated according to Zehner and Schlunder¨ (1970). Gas-to-particle heat transfer coefﬁcient (Gunn, 1978):

2 0.2 1/3 Nu = 7 − 10εg + 5εg 1 + 0.7Re Pr + 2 0.7 1/3 1.33 − 2.4εg + 1.2εg Re Pr

Axial mass dispersion (Edwards and Richardson, 1968):

D 0.73 0.5 eﬀ = + 9 7 vgdp ReSc . εg εg 1+ Re Sc 58 Experimental demonstration of the concept

Notation

A surface of tube wall, [m2] Cp heat capacity, [J/kg/K] 2 Deff effective diffusion coefficient, [m /s] dp particle diameter, [m] dtube tube diameter, [m] g mass deposition rate constant, [s/m] ID inner diameter, [mm] L bed length, [mm] Nu Nusselt number (αgsdp/λg) OD outer diameter, [mm] p parameter in axial heat dispersion coefficient Peax Peclet number for axial heat dispersion (ρgvgdpCp,g/λax) Pr Prandtl number (Cp,gηg/λg) Re Reynolds number (ρgvgdp/ηg) Sc Schmidt number (ηg/ρg/D) T temperature, [K] v superficial velocity, [m/s]

Greek letters εg bed void fraction, [-] ǫ integral emissivity, [-] η viscosity, [kg/m/s] λ thermal conductivity, [W/m/K] λbed,0 effective bed conductivity at no ﬂow conditions, [W/m/K] λeff effective conductivity, [W/m/K] ρ density, [kg/m3] σ Stefan-Boltzmann constant, [J/s/m2/K4] 3 φrad radiation heat inﬂux, [J/m /s]

Subscripts c cold side g gas phase h hot side s solid phase 4

Mass deposition rates of carbon dioxide onto a cryogenically cooled surface

Abstract

The rates of CO2 mass deposition onto a cryogenically cooled surface are important for CO2 removal processes based on cryogenics. A slow mass deposition rate would lead to early breakthrough of CO2 in the concept as proposed in this thesis. Fast mass deposition is therefore required. However, no detailed experimental data for kinetic rate expressions are available in literature. Therefore, a dedicated experimental setup to measure CO2 mass deposition rates under well deﬁned conditions was designed. Experiments have been carried out for both pure CO2 as well as CO2/N2 mixtures. The experiments showed that heat transfer through the frost layer will slow down the mass deposition process. Furthermore, it was found that the addition of N2 to the gas phase has a large inﬂuence on the mass deposition rates, due to the introduction of a mass transfer resistance towards the frost surface. To describe the experimentally observed behavior, a detailed frost growth model was developed, based on mass and energy balances. Expressions for the frost density as function of the frost temperature, and for the effective frost conductivity as 60 Mass deposition rates of carbon dioxide

function of the frost density were derived and implemented in the model. When accounting for drift ﬂuxes, the model is well able to describe the observed behavior. For packed bed conditions, the formed frost layer is thin and mass deposition rates are mainly determined by external mass transfer. However, for process concepts in which heat exchanging surfaces are continuously cooled, the frost layer will likley inﬂuence frost growth rates and the developed model is a useful tool for design studies.

4.1 Introduction

In this thesis a novel process concept has been developed for the removal of CO2 from flue gases by feeding the mixture to a cryogenically refrigerated packed bed. CO2 will freeze onto the packing surface, and an effective separation between CO2 and N2 is obtained. Information on mass deposition rates is important for the design of this process. When mass deposition would be relatively slow, the front of freezing CO2 would disperse and early CO2 breakthrough would occur, resulting in an inefficient use of the cold stored in the bed. For other CO2 removal processes in which the heat exchanging surface is being cooled continuously, mass deposition rates are also important. It was found (Chang et al., 2009; Chang and Smith, 1990) that solid CO2 will build up locally in a heat exchanger, eventually leading to a large pressure drop or even plugging when not switched to a regeneration step in time. Knowledge on desublimation rates and the effect of layer thickness on the desublimation is required to predict the location where CO2 will accumulate and the time when regeneration is necessary. To the best of our knowledge the number of studies on the fundamentals of CO2 desublimation is rather limited. Shchelkunov et al. (1986) studied CO2 desublimation on a plate which is positioned in a flow parallel to the plate, resulting in mass transfer perpendicular to the flow direction. Frost was measured for only one CO2 partial pressure (4.35 kPa) and was not uniform along the plate. Ogunbameru et al. (1973) studied CO2 frost formation on a flat plate, which was refrigerated with liquid N2, but again for only one concentration (2 vol.%). This chapter describes an extensive study on CO2 desublimation rates to obtain more insight in the role of kinetics and mass and heat transfer for different CO2 (partial) pressures, surface and gas phase temperatures and flow conditions. The chapter is organized as follows: first the 4.2 Experimental 61

experimental setup and procedure are elucidated, followed by a description of the results of experiments for pure CO2 and mixtures of N2 and CO2. Then, a frost growth model is derived to describe the experimental ﬁndings. Finally the conclusions and its signiﬁcance for cryogenic CO2 removal equipment is given.

4.2 Experimental

This section gives a description of the dedicated experimental setup and the experimental procedure. Also the interpretation and processing of the obtained experimental data is discussed.

4.2.1 Setup The design of the experimental setup was based on a setup used for measurements of phase equilibria of CO2 containing gas mixtures as described by Le and Trebble (2007). A schematic representation of the experimental setup is given in Fig. 4.1. The setup consists of a stirred cell with glass walls (inner diameter = 85 mm, height = 110 mm), containing a round cooled surface (d = 20 mm) at the bottom, at which CO2 is desublimating during a measurement. The axial propeller type stirrer has a diameter of 50 mm, and has a maximum rotation speed of 1730 RPM. The stirred cell is positioned in a large custom made dewar vessel with viewing stripes (KGW-isotherm). This dewar is filled with special cooling liquid (3M Novec 7200) and contains a large coil which can be fed with liquid N2 for cooling. The flow of liquid N2 is controlled by a magnetic valve in order to control the temperature of the (stirred) liquid and hence of the gas phase in the stirred cell. The dewar contains two (K-type) thermocouples to monitor the temperature of the cooling liquid. The stirred cell is furthermore connected to a vacuum pump, a premix vessel and a CO2 feed line. The cell is equipped with an accurate vacuum gauge (Inficon CDG025D) and contains two K-type thermocouples in the gas phase at two different axial positions. The temperature of the cooled surface is controlled by a refrigerated gaseous N2 flow. This N2 flow is passed through a coil submerged in a dewar filled with liquid N2 and the flow rate is controlled using a mass flow controller (Brooks Smart, 60 NL/min) connected to a PID control loop. The refrigerated N2 is then contacted intensively with the cooled copper made plate in the stirred 62 Mass deposition rates of carbon dioxide cell. Inside the cooled plate four 0.5 mm calibrated T-type thermocouple have been installed at different locations, to monitor the temperature in the cooled surface. The cooled surface is surrounded by PVC, which minimizes heat transfer to other parts of the bottom of the cell due to its low conductivity. Monitoring temperatures and pressures and controlling valves and mass flow controllers is performed using NI LabVIEW. A picture of the setup is shown in Fig. 4.2.

4.2.2 Procedure An experiment is started by evacuating the stirred cell and cooling down the cooling liquid in the dewar vessel surrounding the stirred cell to the required set point temperature. Then the cold surface is brought to the desired temperature. The premix vessel (equipped with a GE Sensing PTX 1400 pressure gauge) is filled with pure CO2 or a mixture of N2 and CO2. Valve V-3 is opened to allow the gas to enter the cell until the user given pressure set point is reached. At this point CO2 immediately starts to desublimate at the surface and the pressure in the cell decreases. By controlling MFC-1 (using a PID control loop), fresh (pure) CO2 is fed to the cell to maintain the set point pressure. The flow through the mass flow controller is accurately monitored and is directly proportional to the amount of CO2 desublimating at the cold surface. Frost growth during the measurement is recorded using a Canon EOS digital camera. After a measurement, cooling of the surface is stopped and the system is flushed with N2. 4.2 Experimental 63 Figure 4.1: Simplified process scheme of the experimental setup. 64 Mass deposition rates of carbon dioxide

Figure 4.2: Picture of the experimental setup (with the dewar vessel lowered).

To make sure that no temperature and concentration gradients are present in the gas phase, the mixing behavior of the stirred cell was ex- amined before starting experiments. This was done by feeding a N2/CO2 mixture to the cell and measuring the CO2 content in the outlet with an IR spectrometer. At some point, the CO2 feed was suddenly stopped and the decrease in CO2 concentration in the outlet was measured in time. This was repeated for several rotation speeds, and the outcomes were compared to the theoretical profile of an ideally stirred tank reactor. It was found that mixing approached ideal behavior at rotation speeds above 130 RPM. Experiments have been carried out under a wide range of conditions. To exclude any effects of dilution with N2, first measurements are performed with pure CO2 in the gas phase. The influence of the pressure of CO2 and the cold plate temperature have been studied. Subsequently, the effect of the amount of N2 present in the gas phase was studied, again for different cold plate temperatures, but also for different gas phase tem- 4.2 Experimental 65

Table 4.1: Conditions for all experiments

◦ ◦ Experiment PCO2 [mbar] PN2 [mbar] T0 [ C] Tg [ C] Ns [%] P100 100 0 -130 -30 100 P150 150 0 -130 -30 100 P200 200 0 -130 -30 100 P250 250 0 -130 -30 100 T0-145 100 0 -145 -30 100 T0-160 100 0 -160 -30 100 50T0-130 100 100 -130 -30 100 50T0-140 100 100 -140 -30 100 50T0-150 100 100 -150 -30 100 10T0-130 100 900 -130 -30 100 10T0-140 100 900 -140 -30 100 10T0-150 100 900 -150 -30 100 10Tg-45 100 900 -130 -45 100 10Tg-60 100 900 -130 -60 100 10N75 100 900 -130 -30 75 10N50 100 900 -130 -30 50

peratures and stirrer rotation speeds. The conditions for all experiments are summarized in Table 4.1.

4.2.3 Data processing This section shows how the experimental data have been processed. This is demonstrated in detail for one single experiment: experiment 50T0- 140. The temperature of the cold plate, of the gas phase in the stirred cell and the temperature of the cooling liquid in the dewar vessel during this experiment are shown in Fig. 4.3a. The temperature of the cold plate could be very well maintained at approximately -140◦C. The gas phase temperature is decreasing slowly during the measurement by approximately two degrees. This can be attributed to the heat exchange with the cold surface at the bottom of the cell, which has a much lower temperature. The pressure in the cell, shown in Fig. 4.3b, could be controlled well by compensating the depositing CO2 by feeding fresh CO2 into the 66 Mass deposition rates of carbon dioxide

cell. The output of the mass flow controller is initially high, and is decreasing in time, as illustrated in Fig. 4.3c. Based on the output of the mass flow controller, the mass accumulated at the cold plate can be calculated, which is shown in Fig. 4.3d. The increase in volume of the frost layer was observed by the camera, the pictures of the growing ice layer are depicted in Fig. 4.4. These pictures clearly show that the ice layer is not only growing in the vertical direction, but also in the horizontal direction and that the ice layer attains a curved surface at the edges. This behavior complicates the derivation of the mass deposition rate per unit of surface area, as the surface area is changing in time. Furthermore, heat transfer within the layer will also take place in the radial direction at the edges of the ice layer. In order to facilitate the interpretation of the experimental results, and later on to be able to develop a model for ice layer growth, it is desired to approach the growth as a one-dimensional process. The pictures show that within a certain radius, the ice layer maintains an almost horizontal surface, even after 900 seconds of measurement time. It can be assumed that within this inner part of the ice layer, radial effects can be ignored. Therefore only the growth of the layer within the inner 1 cm radius is considered. However, the mass deposition rate onto this inner part of the ice layer is not directly measured. The accumulated mass as plotted in Fig. 4.3d is formed at the entire surface, including the curved outer parts. The mass of CO2 depositing onto the inner 1 cm radius is being calculated as follows. First both the total volume of the ice layer as well as the volume within the inner radius of 1 cm have been calculated from the images. To calculate the volume from the (two-dimensional) image, the surface area of a pixel and the distance to the central axis are determined using the image magnification factor. Then, this area is rotated around the central axis of the ice layer to obtain the volume of revolution of the pixel. This is done for the pixels at the left side of the central axis as well as the right side and the result is averaged, as the image is not perfectly symmetrical. This is carried out both for the inner radius, as well as the entire frost layer. Thus, the ratio of the inner volume and the total volume is determined for every image. Fig. 4.5 shows the result of this procedure for experiment 50T0-140. It can be observed that especially in the beginning of the experiment the volume ratio is decreasing, indicating that especially at the start of the experiment the ice layer is growing significantly in the radial direction. Note that when there would be no ice layer formation in the radial direction, the ratio would be 0.25 (due to the total radius of 2 cm compared 4.3 Results 67

to the inner 1 cm). When extrapolating to t = 0, it can indeed be observed that this ratio will approach approximately 0.25. To compute the amount of CO2 deposited in the inner part, the total mass accumulated (as plotted in Fig. 4.3d) is being multiplied with the obtained volume ratio. This is allowed, when assuming that the frost density at the inner part of the ice layer is equal to the density of the entire layer. Fig. 4.5 shows this amount of CO2 accumulated at the inner part of the ice layer. Finally, the mass deposition rate is calculated by ﬁtting the accumulated mass with a power law (also shown in Fig. 4.5), and differentiating the obtained equation. After data processing the following information has been obtained for an experiment: the layer thickness, bulk density and mass deposition rate as function of the time. These results are presented for all experiments in the next section.

4.3 Results

The results for the experiments listed in Table 4.1 are presented and discussed in this section. The effect of the pressure of CO2 (without the addition of N2) is shown in Fig. 4.6. It can be observed that for all measurements the initial mass deposition rate is very high, but is quickly decreasing in time. The only property changing during an experiment is the thickness of the ice layer (shown in Fig. 4.6b). It can therefore be concluded that heat conduction through the ice layer towards the cold plate is playing an important role. Remarkable is the fact that there is no large effect of the CO2 pressure on the deposition process. Apparently, surface kinetics do not play a role at these conditions. To further investigate the effect of heat conduction through the ice layer, experiments have been carried out with different plate temperatures. At lower temperatures, the heat ﬂux through the ice layer becomes higher, and therefore a higher mass deposition rate is measured, as shown in Fig. 4.7. To study the effect of N2 on the deposition rates, measurements have been carried out by adding N2 to the gas, while keeping the CO2 partial pressure constant at 100 mbar. Fig. 4.8 shows results of measurements with pure CO2, 50 vol.% (total pressure of 200 mbar) and 10 vol.% (total pressure of 1000 mbar) mixtures. It can be very clearly observed that mass deposition rates and the layer growth are slowed down by the presence of N2 in the gas phase, which clearly indicates that mass transfer 68 Mass deposition rates of carbon dioxide

250

-20

Cooling liquid dewar

-110

Gas phase cell Temperature [°C] Temperature

-24

200

Cold plate

-120

-28

150

-32

-130

-36

100

-40

-140 Temperature [°C] Pressure cell [mbar]

-44

-150

-48

0 200 400 600 800 0 200 400 600 800

Time [s] Time [s] (a) (b) ] -1

-3

0.25 3.0x10

-3

2.5x10

0.20

-3

2.0x10

0.15

-3

1.5x10

0.10 accumulated [kg] 2

-3

1.0x10

0.05

-4

5.0x10 flow into cell [NL·min 2 Mass CO

0.00 0.0 CO

0 200 400 600 800 0 200 400 600 800

Time [s] Time [s]

Figure 4.3: The measured temperatures (a), pressure (b), ﬂow into the cell (c) and the accumulated mass on the cold plate (d) during experiment 50T0-140. 4.3 Results 69

Figure 4.4: Photographs taken every minute of the CO2 layer for experiment 50T0-140. 70 Mass deposition rates of carbon dioxide

0.5

8 Mass accumulated [kg·m accumulated Mass

Volume ratio

Mass accumulated inner part

0.4

Power law fit

0.3

0.2

2 Volume ratio [-]

0.1 2-2 ]

0.0

0 200 400 600 800

Time [s]

Figure 4.5: The ratio of the ice layer volume formed at the inner 1 cm radius and the entire layer volume plotted as a function of time for experiment 50T0-140. The accumulated mass per area for the inner 1 cm radius is plotted at the right y-axis. 4.3 Results 71 ]

0.012

0.05 -1

P = 100 mbar ·s

P = 100 mbar CO2

CO2 -2

P = 150 mbar 0.010

P = 150 mbar CO2

0.04 CO2

P = 200 mbar

P = 200 mbar CO2

CO2

0.008

P = 250 mbar

P = 250 mbar CO2

0.03 CO2

0.006

0.02

0.004

0.01 Layer thickness [m]

0.002

0.00

0.000 Mass deposition rate [kg·m

0 200 400 600 800 0 200 400 600 800

Time [s] Time [s]

(a) (b)

1600 ] -3

1200

800

P = 100 mbar

CO2

P = 150 mbar

CO2

P = 200 mbar

400

CO2

P = 250 mbar

CO2 Layer density [kg·m

0 200 400 600 800

Time [s]

(c)

Figure 4.6: The experimental (markers) and simulated (lines) mass deposition rate (a), frost layer thickness (b) and layer density (c) as function of the time for the pure CO2 experiments at different pressures (P100, P150, P200 and P250, see Table 4.1). 72 Mass deposition rates of carbon dioxide ]

0.012

0.05 -1

T = -130°C ·s

T = -130°C -2

T = -145°C 0.010

T = -145°C

0.04

T = -160°C

0 0.008

0.03

0.006

0.02

0.004

0.01 Layer thickness [m]

0.002

0.00

0.000 Mass deposition rate [kg·m

0 200 400 600 800 0 200 400 600 800

Time [s] Time [s]

(a) (b)

1600 ] -3

1200

800

T = -130°C

T = -145°C

400

T = -160°C

0 Layer density [kg·m

0 200 400 600 800

Time [s]

(c)

Figure 4.7: The experimental (markers) and simulated (lines) mass deposition rate (a), frost layer thickness (b) and layer density (c) as function of the time for the pure CO2 experiments at different cold plate temperatures (P100, T0-145 and T0-160, see Table 4.1). 4.4 Development of a frost growth model 73

of CO2 from the gas phase to the forst layer plays a significant role. It can also be discerned that the density of the frost layer is influenced by the presence of N2 in the gas phase. For the pure CO2 measurements, the frost density is quickly increasing to a plateau value, while for the mixtures the density is slowly increasing in time. These effects will be further discussed in section 4.4. Concluding, mass transfer has a large influence on the deposition process. On the other hand, the mass deposition rate is still decreasing in time for the measurements with the CO2/N2 mixtures, which implies that heat transfer through the solid layer still plays a role. To study this in more detail, again the plate temperature has been varied, both for the 50 vol.% (Fig. 4.9) as well as the 10 vol.% CO2/N2 (Fig. 4.10) mixtures. Again, it can be observed that there is a general trend of increasing mass deposition rates, when the cold plate temperature is decreasing. However, it should be noted that the differences are small. The influence of the gas phase temperature on the mass deposition rates is shown in Fig. 4.11. A lower gas phase temperature results in slightly faster deposition rates, although the differences are close to the range of the experimental accuracy. Finally, the rotation speed of the stirrer has been changed. The mass transfer towards the frost surface is decreasing at lower rotation speeds, resulting in lower deposition rates, as shown in Fig. 4.12.

4.4 Development of a frost growth model

In order to obtain a better understanding of the desublimation process of CO2 onto a cold surface, a model will be developed in this section. The outcomes will be compared with the experimental results and finally the significance of the findings for the cryogenic packed bed concept for CO2 capture are discussed.

4.4.1 Model development The experimental results revealed that both heat transfer through the formed solid ice layer and mass transfer of CO2 towards the ice surface play an important role. Therefore, the frost growth is described as a moving boundary problem, with both mass as well as heat transfer included. 74 Mass deposition rates of carbon dioxide ]

0.010

0.05 -1

·s Pure CO

Pure CO

2 -2

50 vol.% CO

0.008 2

0.04 2

10 vol.% CO

0.006 0.03

0.02 0.004

0.01

0.002 Layer thickness [m]

0.00

0.000 Mass deposition rate [kg·m

0 200 400 600 800 0 200 400 600 800

Time [s] Time [s]

(a) (b)

1600 ] -3

1200

800

Pure CO

400

50 vol.% CO

2 Layer density [kg·m

10 vol.% CO

0 200 400 600 800

Time [s]

(c)

Figure 4.8: The experimental (markers) and simulated (lines) mass deposition rate (a), frost layer thickness (b) and layer density (c) as function of the time for experiments with different CO2 concentrations (T0-145, 50T0-140 and 10T0-140, see Table 4.1.) 4.4 Development of a frost growth model 75 ]

0.010 -1

·s 0.020 T = -130°C T = -130°C

0 0 -2

T = -140°C T = -140°C

0.008 0 0

T = -150°C T = -150°C 0.015

0 0

0.006

0.010

0.004

0.005

0.002 Layer thickness [m]

0.000

0.000 Mass deposition rate [kg·m

0 200 400 600 800 0 200 400 600 800

Time [s] Time [s]

(a) (b)

1600

T = -130°C

0 ]

T = -140°C -3

T = -150°C 1200

800

400 Layer density [kg·m

0 200 400 600 800

Time [s]

(c)

Figure 4.9: The experimental (markers) and simulated (lines) mass deposition rate (a), frost layer thickness (b) and layer density (c) as function of the time for the 50 vol.% CO2 experiments at different cold plate temperatures (50T0-130, 50T0-140 and 50T0-150, see Table 4.1). 76 Mass deposition rates of carbon dioxide

0.006 ]

0.010 -1

·s T = -130°C

T = -130°C

0 -2

0.005

T = -140°C

0.008 0

T = -150°C

0 0.004

0.006

0.003

0.004

0.002

0.002 Layer thickness [m]

0.001

0.000

0.000 Mass deposition rate [kg·m

0 200 400 600 800 0 200 400 600 800

Time [s] Time [s]

(a) (b)

1600

T = -130°C

0 ] -3

T = -140°C

1200

T = -150°C

800

400 Layer density [kg·m

0 200 400 600 800

Time [s]

(c)

Figure 4.10: The experimental (markers) and simulated (lines) mass deposition rate (a), frost layer thickness (b) and layer density (c) as function of the time for the 10 vol.% CO2 experiments at different cold plate temperatures (10T0-130, 10T0-140 and 10T0-150, see Table 4.1). 4.4 Development of a frost growth model 77

0.006 ]

0.010 -1

·s T = -30°C

T = -30°C g -2

0.005

T = -45°C

0.008 T = -45°C g

T = -60°C

T = -60°C g

0.004 g

0.006

0.003

0.004

0.002

0.002 Layer thickness [m]

0.001

0.000

0.000 Mass deposition rate [kg·m

0 200 400 600 800 0 200 400 600 800

Time [s] Time [s]

(a) (b)

1600

T = -30°C

g ]

T = -45°C -3

T = -60°C 1200

800

400 Layer density [kg·m

0 200 400 600 800

Time [s]

(c)

Figure 4.11: The experimental (markers) and simulated (lines) mass deposition rate (a), frost layer thickness (b) and layer density (c) as function of the time for the 10 vol.% CO2 experiments at different gas phase temperatures (10T0-130, 10Tg-45 and 10Tg-60, see Table 4.1). 78 Mass deposition rates of carbon dioxide

0.006 ]

0.010 -1 ·s

N = 100%

-2 N = 100% s

0.005 s

N = 75% 0.008

N = 75% s

N = 50%

N = 50% 0.004 s

0.006

0.003

0.004

0.002

0.002 Layer thickness [m]

0.001

0.000

0.000 Mass deposition rate [kg·m

0 200 400 600 800 0 200 400 600 800

Time [s] Time [s]

(a) (b)

1600

N = 100%

s ] -3

N = 75%

1200

N = 50%

800

400 Layer density [kg·m

0 200 400 600 800

Time [s]

(c)

Figure 4.12: The experimental (markers) and simulated (lines) mass deposition rate (a), frost layer thickness (b) and layer density (c) as function of the time for the 10 vol.% CO2 experiments at different stirrer rotation speeds (10T0-130, 10N75 and 10N50, see Table 4.1). 4.4 Development of a frost growth model 79

(a) (b)

Figure 4.13: Schematic representation of temperature (a) and CO2 concentration (b) proﬁles during the frost growth process.

A schematic representation of the involved processes together with the used variables is given in Fig. 4.13. The following assumptions are made:

• The temperature proﬁles within the layer are developed instanta- neously (quasi-steady state approach), i.e. the frost growth is much slower than the temporal change of the temperature proﬁles within the layer.

• The desublimated CO2 at the frost surface is in equilibrium with the gas phase.

• The measurements showed that the density is changing in time. It is likely that the formed ice layer has a porous structure, which becomes denser in time. The porosity, density and therefore also layer heat conductivity might be a function of the location within the frost layer. However, this information could not be obtained from our experiments, therefore, it is assumed that the layer has a uniform density and heat conductivity.

• The mass and heat transfer coefﬁcient for the transfer from the gas bulk towards the frost surface are coupled according to the Chilton- Colburn analogy.

CO2 is transfered from the bulk of the gas phase to the solid phase and deposits at the surface. Due to the removal of CO2 from the gas 80 Mass deposition rates of carbon dioxide

phase, a net ﬂow towards the solid is generated and Fick’s diffusion is no longer valid. Drift ﬂuxes have to be taken into account:

• σ NCO2 = xgNtot + ctot kg (xg − x ) (4.1)

There is no net ﬂux of N2, therefore:

NCO2 = Ntot (4.2) resulting in:

σ • (xg − x ) • NCO2 = ctot kg with: kg = kgξ (4.3) (1 − xg) and:

Φ NCO ξ = Φ= 2 (4.4) exp(Φ) − 1 ctot kg and therefore:

• NCO 1 k = 2 (4.5) g NCO ctot 2 exp c k − 1 tot g Substituting Eq. 4.5 in Eq. 4.3 results in the following relation for the net CO2 mole ﬂux to the ice layer, which is known as Stefan ﬂow (Taylor and Krishna, 1993):

1 − xσ NCO2 = ctotkg ln (4.6) 1 − xg The heat transfer from the gas to solid phase by conduction and by radiation, together with the heat formed by desublimation are conducted through the solid layer towards the cold surface. Again the effects of drift ﬂuxes for the heat conduction through the gas towards the ice surface are taken into account:

σ 4 σ4 αg (ξh + Φh)(Tg − T )+ σǫ Tg − T + NCO2 ∆Hs = λeﬀ σ (T − T0) (4.7) δ 4.4 Development of a frost growth model 81

in which:

NCO2 Cp,CO2 Φh Φh = ξh = (4.8) αg exp(Φh) − 1

The concentration of CO2 at the frost surface follows from combining (4.6) and (4.7). The layer thickness is calculated by:

d (ρsδ) = NCO MCO with:δ(t = 0) = 0 (4.9) dt 2 2 which can be solved with a standard ODE solver. Constitutive equations are summarized in Table 4.2. The density ρs and conductivity λeﬀ of solid CO2 are unknown. As- suming these variables to be constant is not correct, as it was concluded from the experiments that the density is increasing in time, but reaches a plateau value for the measurements with pure CO2 in the gas phase relatively quickly. For pure CO2, the CO2 concentration at the frost surface is equal to the gas phase concentration, which means that it can be expected that the surface temperature will be constant during a measurement. For the measurements with mixtures, the surface temperature is not constant, but will increase in time. Therefore, a correlation for the layer density as a function of the surface temperature is proposed. When inspecting the measurements for different pure CO2 pressures in Fig. 4.6, it can indeed be observed that the measured density of the layer is higher for measurements with a higher CO2 pressure in the gas phase (corresponding to a higher interface temperature). To investigate this relation in more detail, the following procedure is followed. Based on the measured mass deposition rates, it is calculated using equation (4.6) what the concentration at the surface should have been. This concentration can again be used to calculate the surface temperature, because of equilibrium. In this way the (measured) density can be plotted as function of the calculated surface temperature for a large number of measurements, as shown in Fig. 4.14a. Although the results are somewhat scattered, there is a clear trend that the density is increasing when the surface temperature is increasing and leveling off at higher temperatures. Although not measured, it can be expected that for lower temperatures the density will also level off. This behavior can be described using the following ﬁt: a ρ = (4.10) s 1+ b exp(−cT σ) 82 Mass deposition rates of carbon dioxide

Table 4.2: Constitutive equations used in the frost growth model.

The mass transfer coefﬁcient kg is calculated according to:

b c Sh = a Re Sc in which is assumed: a = 0.1,b = 0.75,c = 0.33

in which b and c are taken from Winkelman et al. (2002), and a ﬁtted to experimental results. The binary diffusivity for N2/CO2 mixtures is calculated according to the Fuller- Schettler-Giddings correlation (Fuller et al., 1966):

0 5 1.75 . 0.0143T 1 + 1 1000MA 1000MB DAB = 2 1/3 1/3 Ptot ( υ)A +( υ)B h P P i

in which υ is the sum of atomic diffusion volumes, which is 26.9 for CO2 and 17.9 for NP2. The CO2 mole fraction at the interface is the equilibrium value at the surface temperature T σ (Daubert and Danner, 1985):

σ σ 3082.7 σ −2 σ P (T ) exp 10.257 − σ + 4.08ln T − 2.2658 · 10 T xσ = = T Ptot Ptot

The gas to solid heat transfer coefﬁcient αg is coupled to the mass transfer coefﬁ- cient kg, according to the Chilton-Colburn analogy:

Nu Pr1/3 = Sh Sc1/3

The physical properties of the gas phase were computed at the average gas phase temperature according to Reid et al. (1987), using the pure component data supplied by Daubert and Danner (1985). 4.4 Development of a frost growth model 83

] 1.2 10T0-150 T0-160 10T0-150 T0-160

1600 -1

10T0-140 T0-145 10T0-140 T0-145 ·K

] 10T0-130 P100 10T0-130 P100

1.0 -1 -3

50T0-150 P150 50T0-150 P150

50T0-140 P200 50T0-140 P200

1200

50T0-130 P250 50T0-130 P250 0.8

Fit Fit

0.6

800

0.4

400

0.2 Frost density [kg·m

0.0 0 Frost conductivity [W·m

150 160 170 180 0 400 800 1200 1600

-3

Surface temperature [K] Frost density [kg·m ] (a) (b)

Figure 4.14: The measured density plotted as function of the calculated surface temperature (a) and the conductivity of the ice layer as function of the density (b).

with: a = 1.4768·103, b = 1.122·1015 and c = 2.133·10-1 The effective conductivity of the frost layer is very likely to be related to the bulk density. For higher densities, one would expect higher conduc- tivities. In order to verify this, the surface temperature found by solving equation (4.6) using the measured mass deposition rates, is substituted into the energy balance (4.7). When also the measured layer thickness is substituted into the equation, the effective conductivity of the frost layer can be computed. The calculated conductivity is plotted as function of the measured density, again for a large number of measurements (see Fig. 4.14b). Although the results are scattered again, there is a clear trend visible. This relation is described using a linear ﬁt:

λeﬀ = a + bρs (4.11) -3 with: a = 6.200·10 (which is the CO2 gas conductivity) and b = 6.523·10-4.

4.4.2 Simulation results The model developed in the previous section is used to simulate the mass deposition rate, layer thickness and frost density as a function of the time, for all the measurements listed in Table 4.1. The outcomes have 84 Mass deposition rates of carbon dioxide

been plotted in the same figures as where the experimental results have been plotted. The general conclusion is that the experimental and simulation results agree very reasonably. When looking at the measurements for pure CO2, it can be observed that the fast decreasing mass deposition rates are very well described. Also the initially fast increasing frost density to a constant value is well predicted by the model. It was observed during experiments that the different pressures did not have a large effect on the deposition rates and layer growth, as shown in Fig. 4.6. The difference in surface temperatures for the different measurements is too small to have a significant effect on the heat fluxes through the solid. The simulation results show comparable results. The cold plate temperature had a larger effect (Fig. 4.7), which is again well predicted by the model. The differences between the measurements with different CO2 concentrations, as shown in Fig. 4.8, are very well described by the developed model. The different measurements for the 50 vol.% and 10 vol.% mixtures are in general reasonably described. The mass deposition rates are well described, although the discrepancies in the measured and the computed frost density and layer thickness are sometimes somewhat larger, which can be well attributed to the inaccuracies in the derived correlations for the layer density and conductivity, and the inaccuracy in the measurements. However, the calculated trends are very well matching the trends found in experiments. To show the importance of taking drift fluxes into account, the mass deposition rates have been calculated for experiment 50T0-140 and 10T0- 140 with and without including drift effects. It can be observed in Fig. 4.15 that the effect is quite small for the 10 vol.% CO2 measurement, but has a substantial effect for the 50 vol.% CO2 measurement. Furthermore, it is noted that the contribution due to radiation in the energy balance (4.7) is negligible.

4.4.3 Signiﬁcance for the cryogenic packed bed concept The conditions of the stirred cell are different from those encountered in cryogenically refrigerated packed beds. The cold plate temperature in the stirred cell is kept at a constant temperature by continuous cooling, while in the packed bed the temperature of the packing material will increase in time when a front of desublimating CO2 passes by. In order to show the signiﬁcance of the developed model for the mass deposition rates, 4.4 Development of a frost growth model 85 ] -1

·s 0.020 50T0-140 10T0-140 -2

50 Incl. drift 10 Incl. drift

50 No drift 10 No drift

0.015

0.010

0.005

0.000 Mass deposition rate [kg·m

0 200 400 600 800

Time [s]

Figure 4.15: The experimental mass deposition rates for experiments 50T0-140 and 10T0-140 (markers) and the the simulated proﬁles (lines), with and without taking drift ﬂuxes into account.

the next situation will be evaluated: A refrigerated spherical particle, initially at Tp,0 is suddenly positioned in a gas phase of temperature Tg and

CO2 mole fraction of xCO2,g, which is the equilibrium concentration at the temperature Tg. This is similar to the situation in the packed bed at the frost front, where saturated gas will be contacted with cold packing. Due to this step change in temperature and composition, CO2 will start to desublimate at the particle surface, and the particle will heat until the system is in equilibrium, i.e. the particle temperature Tp is equal to the gas phase temperature. The temperature increase of the particle can be described by:

dTp ′′ ρ V C = (Φ + NCO MCO ∆H )A (4.12) p p p,p dt h 2 2 s p where it is assumed that the heat capacity of the desublimated CO2 is much smaller than the heat capacity of the particle. The particle is being heated by desublimation of CO2 and by heat transfer from the gas ′′ phase to the ice layer (Φh). As given in (4.7), this amount is transferred 86 Mass deposition rates of carbon dioxide

Table 4.3: Properties and conditions used for a case study in which CO2 desubli- mates onto a spherical particle.

Particle diameter [m] 0.004 Particle material [-] glass Solids density [kg · m−3] 2546 Heat capacity [J · kg−1K−1] −2.8 · 10−3T 2 + 3.48T − 47.9 Initial particle temperature [◦C] −140 Gas phase temperature [◦C] −103.3 Gas phase mole fraction CO2 [-] 0.1 Gas mass ﬂow [kg · m−2 · s−1] 0.249

through the solid CO2 layer towards the particle, and therefore (4.12) can be written as:

dTp λeff σ 6 1 = (T − Tp) (4.13) dt δ dp ρpCp,p This heat balance coupled with the frost growth model are solved nu- merically using a standard ODE solver. The frost growth at the spherical particle is computed using the conditions and properties as given in Ta- ble 4.3, which are similar to those used in the experiments described in Chapter 3. The mass and heat transfer coefficients are calculated for the conditions as occurring in the packed bed, according to Gunn (1978). The computed temperature of the particle as function of the time is shown in Fig. 4.16. It can be observed that equilibrium is already reached after approximately 5 seconds. The layer of solid CO2 formed at the particle surface at that moment is only 3.24·10-5 m thick (corresponding to 3 57.1 kg/m packing). Therefore, the heat transfer through the solid layer plays a negligible role and the frost surface temperature is approximately equal to the particle temperature. To illustrate this in more detail, the temperature increase of the particle has been calculated without taking layer formation into account. The concentration of CO2 at the frost surface is calculated using the particle temperature. Fig. 4.16 indeed shows that the computed temperature rise of the particle is hardly influenced. Furthermore, it should be noted that heat transfer from the gas to solid phase plays a minor role, meaning that the mass deposition rate for the conditions used in this case are mainly determined by mass transfer of CO2 from the gas to the solid phase. 4.5 Discussion and conclusions 87

It was shown in Chapter 3 that the axial temperature proﬁles within the packed beds could be well described when assuming the following mass deposition rate equation:

σ NCO2 MCO2 = g (xgPtot − P ) (4.14) with g = 1·10-6 s/m. The increase in temperature of the particle is also calculated using this simplified mass deposition rate for different values of g. Fig. 4.16 shows that indeed a value slightly above 1·10-6 s/m matches the temperature profile calculated using the frost growth model best. It was shown in Fig. 4.15 that for low concentrations of CO2 drift fluxes have negligible effects on the mass deposition rate. Therefore the mass deposition rate could be written as:

σ NCO2 = kgctot (xg − x ) (4.15) Combination of (4.14) and (4.15) and using the ideal gas law to couple the concentration and the pressure, yields:

kgMCO g = 2 (4.16) RT

The values of kg and the average temperature in the ﬁlm layer around the particle T are slightly changing in time. When taking the average -2 ◦ values (kg = 4.45· 10 m/s and T = -105.5 C) the value of g amounts 1.4 ·10-6 s/m, which is close to the assumed value of g. Finally, it is noted that the heat transfer within the particle itself is relatively fast, but could play a role for glass particles. Fig. 4.17 shows that the temperature in a spherical particle is well developed for a Fourier at number ( R2 ) of 0.4, which corresponds to 2.2 seconds for a glass particle with a diameter of 4 mm. Note that for steel particles or monoliths this effect plays no role.

4.5 Discussion and conclusions

This chapter has presented a study on the the mass deposition rates of CO2 onto cold surfaces. A dedicated experimental setup was designed to measure CO2 deposition rates under well deﬁned conditions. The results showed that heat transfer through the formed frost layer plays an important role in the deposition rates. Furthermore, it was shown that 88 Mass deposition rates of carbon dioxide

-5

g = 1·10 Frost growth model

-80

-6

g = 1·10 No layer

-7

g = 1·10

-100

-120 Particle temperature [°C]

-140

0 4 8 12 16 20

Time [s]

Figure 4.16: Temperature increase calculated for a cold particle positioned in a

N2/CO2 mixture at t = 0, using different expressions for the mass deposition rate.

Figure 4.17: Unsteady temperature proﬁles in a spherical particle (Carslaw and Jaeger, 1952). 4.5 Discussion and conclusions 89

diluting CO2 with N2 has a large effect on the mass deposition rate, due to the introduction of mass transfer limitations from the gas bulk towards the frost surface. To describe the results, a frost growth model was developed. Based on the experimental results, fitted correlations for the bulk density of the frost layer as function of the frost surface temperature and a correlation for the effective layer conductivity as a function of the frost density were derived. Using these correlations in the frost growth model, the developed model could well describe the experimental findings of the CO2 mass deposition rate as a function of time at different CO2 concentrations and gas and cooled surface temperatures. To show the significance of the results for the cryogenic packed bed concept, the temperature increase of a cold particle suddenly positioned in a N2/CO2 mixture was calculated using the frost growth model. It was shown that under the packed bed conditions, heat transfer through the solid frost layer plays no significant role and the process was mainly determined by mass transfer of CO2 towards the particle, which coincides with the earlier findings reported in Chapter 3. It should be noted that for other cryogenic CO2 capture technologies based on continuously cooled heat exchanging surfaces (such as proposed by Clodic and Younes (2002)), the increasing frost layer does play a significant role for the frost formation rates. Therefore the frost growth model developed in this chapter could be a valuable tool in the design and study of these types of equipment.

Acknowledgment

The contributions of Nhi Dang and Niels Hietberg to this chapter are highly appreciated.

Notation

A area, [m2] a thermal diffusivity, [m2/s] c concentration, [mol/m3] Cp heat capacity, [J/kg/K] D diffusion coefﬁcient, [m2/s] d diameter, [m] g mass deposition rate constant, [s/m] 90 Mass deposition rates of carbon dioxide

kg gas solid mass transfer coefficient, [m/s] M molecular weight, [kg/mol] N molar flux, [mol/m2/s] Ns stirrer rotation speed, [1/s] Nu Nusselt number (αgd/λg) P pressure, [Pa, mbar] Pr Prandtl number (Cp,gηg/λg) R gas constant, [J/mol/K] 2 Re Reynolds number ρgNsds/ηg or (ρgvgdp/ηg) Sc Schmidt number (ηg/ρg/D) Sh Sherwood number (kgd/D) t time, [s] T temperature, [K], [◦C] V volume, [m3] v superficial velocity, [m/s] x mole fraction, [-]

Greek letters 2 αg gas solid heat transfer coefficient, [W/m /K] δ layer thickness, [m] ∆H sublimation enthalpy, [J/mol] ǫ emissivity, [-] η viscosity, [kg/m/s] λ thermal conductivity, [W/m/K] λeff effective conductivity frost layer , [W/m/K] ξ correction factor for drift flux, [-] ρ density, [kg/m3] υ atomic diffusion volume, [-] σ Stefan-Boltzmann constant, [J/s/m2/K4] Φ correction factor for drift flux, [-] ′′ 2 Φh gas solid heat flux, [J/m /s]

Subscripts 0 cold plate, initial g gas phase p particle s solid phase tot total 4.5 Discussion and conclusions 91

Superscripts σ equilibrium 92 Mass deposition rates of carbon dioxide 5

Experimental demonstration of the novel process concept in a pilot scale setup

Abstract

The proposed novel process concept to cryogenically capture CO2 from ﬂue gases using dynamically operated packed beds consists of a capture, recovery and cooling step. The capture step was extensively studied both by experiments in a small scale setup as well as simulations in Chapter 3. This chapter describes a fully automated and continuously operated experimental pilot plant, in which the entire process cycle is demonstrated, including the capture, recovery and cooling step. Tests of more than 10 hours showed that the process can be continuously operated for long periods. The temperatures measured in the beds during the three process steps have been compared to simulation results obtained with the one-dimensional pseudo-homogeneous numerical model developed in Chapter 2. Breakthrough times can be well described. However, it was observed that the temperatures measured at a radial position close to the wall differed from the temperatures measured in the radial center. Therefore, the model was extended with an additional energy balance for the wall, where heat exchange with the packing is taken into account. It 94 Experimental demonstration in a pilot scale setup

is shown for the cooling step that the high conductivity of the wall causes the temperature profiles to be very smooth close to the wall relative to the profiles in the center. When including a separate energy balance for the wall the measured profiles could be well described by the model.

5.1 Introduction

Cryogenic CO2 capture using dynamically operated packed beds requires three process steps: the capture, recovery and cooling step. The capture step was demonstrated in Chapter 3 in a small scale experimental setup. This chapter gives a demonstration of the entire process cycle including all required process steps. First, the experimental setup and procedure are described, followed by the results for long term experiments. Subse- quently, the temperature proﬁles at different locations within the beds are studied in more detail for all three individual process steps and are compared to simulation results of the one-dimensional pseudo-homogeneous numerical model described in Chapter 2. Finally, the chapter concludes with the description of an extension of the numerical model, in order to provide better understanding of the observed temperature proﬁles during the experiments related to the energy transport to and in the steel vessel wall.

5.2 Experimental

This section provides a description of the constructed experimental setup and the experimental procedure followed in the experiments.

5.2.1 Experimental setup A schematic flow diagram of the experimental rig is shown in Fig. 5.1. Three identical stainless steel vacuum insulated vessels (OD × ID × L = 106 × 100 × 500 mm) were filled with spherical glass beads (dp = 3 mm). All three beds were equipped with 9 K-type thermocouples at several axial and radial positions (see Fig. 5.2) in order to monitor the development of temperature profiles within the beds. Pressure sensors were installed both at the inlet as outlet of all three beds to measure the pressure drops over the beds. Cooling of the beds was performed by feeding a cold N2 5.2 Experimental 95

gas flow, which was refrigerated before using a cryogenic cooler (Stirling SPC-1). The cooler was connected to the beds using polystyrene insulated pipes. Mixtures of N2 and CO2 were fed to the refrigerated bed, of which the composition was controlled by using two mass flow controllers (Brooks). Before feeding the gas mixture to a bed, the gas was heated up using a tracing line winded around the feed tube. The recovery step was performed by recycling CO2 using a 3kW side channel blower (VA- COM SC 425). At initial stages of recovery steps, some additional CO2 deposited onto the packing as explained in Chapter 2. Therefore, a CO2 buffer was required, which consists of three stainless steel vessels (total volume of 36 liter). Low temperature CO2 was not allowed to enter the blower, therefore a tracing line has been installed around the blower inlet tube to heat up the incoming gas to ambient temperature. Furthermore, all beds could be flushed with heated N2, if so required. All valves are pneumatic ball valves, special cryogenic valves (Meca-inox) were used at those positions where low temperatures were prevailing. Signal processing and process control was carried out by NI Labview. A picture of the experimental setup is shown in Fig. 5.3.

5.2.2 Experimental procedure

Before an actual measurement was started, the N2/CO2 mixtures were prepared and heated up, the cooler was started up and the CO2 buffer vessels were filled. After these preparations, the first bed (bed #1) was cooled down until the average temperature of two thermocouples (r = 0 cm and r = 4 cm) at the outlet of the bed reached a certain set point temperature. At that point, the second bed (bed #2) was fed with the cold N2 flow, and at the same time bed #1 was fed with the N2/CO2 gas mixture. The temperature at the outlet of a bed will increase when CO2 starts to break through. Therefore again a certain set point temperature was used to monitor whether a bed finished a capture step. At the point that CO2 breakthrough occurred in bed #1, and bed #2 was cooled down, bed #1 was switched to the recovery step, bed #2 to the capture step and bed #3 to the cooling step. The recovery step was started by switching on the recycle blower for CO2. Due to the additional CO2 which desubli- mates initially onto the packing surface in a relative short time, the pressure decreases. This was counteracted by opening the valve connected to the CO2 buffer vessels. Fresh CO2 flows into the recovery piping, until 96 Experimental demonstration in a pilot scale setup

the additional desublimation of CO2 stopped and the pressure reached a certain set point. At that point CO2 was recycled, and the previously deposited CO2 in the bed was evaporated and removed from the bed. The outlet flow is then larger than the inlet flow, therefore the pressure starts to increase. This is again counteracted, in this case by opening a mass flow controller, controlled using a PID control loop. Again the temperature at the outlet of the bed is used to determine the point to switch. When all CO2 is removed from the packing, the temperature will immediately start to increase and at that point the bed was switched again to a cooling step. In case one of the beds finished its step before the other beds were finished, the bed was put on a standby mode and switched again to the next process step as soon as the other beds were finished. 5.2 Experimental 97 ich TI, PI and FI stand for temperature, Figure 5.1: Simplified flow scheme of the pilot test setup, in wh pressure and flow indication respectively. 98 Experimental demonstration in a pilot scale setup

5.3 Experimental results

Experiments have been carried out using the conditions as listed in Ta- ble 5.1. Long term runs of more than 10 hours proved that the system is able to run stable for a long period of time. The average temperature of the thermocouples at r = 0 cm and r = 2 cm at the outlets of the three beds as function of the time is plotted for 6 hours in Fig. 5.4a. Similar re- peating temperature patterns for the three beds can be observed. To have a closer look, the temperature profile is zoomed in for one bed, showing a cooling, capture and recovery step in more detail (see Fig. 5.4b). During the capture step, the temperature at the outlet of the bed is almost stable at the initial bed temperature (approximately -130◦C), but increases strongly as soon as CO2 starts to breakthrough. When the bed is at a temperature of -100◦C the capture step is stopped. When the recovery step is started, it can be observed that the temperature increases quickly to a temperature of approximately -76◦C corresponding to a saturation temperature of pure CO2 at the operating pressure (1.2 bar). Again it can be observed that the temperature increases at the end of the recovery step. At the point that the average outlet temperature reaches -70◦C, the bed is switched to a cooling step. During the cooling step, the outlet temperature first rises to a temperature of about 20◦C. This is the temperature at which the CO2 is fed to the bed during the recovery step. Note that for commercial operation it is better to feed the recovery CO2 gas flow at lower temperatures (-70◦C) to save cooling duty, but due to practical limitations of the used blower, this could not be realized in the experimental situation. After this relatively hot zone of 20◦C has moved through the bed, the temperature at the outlet of the bed decreases, until the set point is reached (-130◦C). Fig. 5.5a shows the pressure at the outlet of one bed, again zoomed in for one process cycle with the three steps. During the capture step, the bed is at atmospheric pressure. As soon as the bed is switched to the recovery step, it can be observed that the pressure suddenly drops to a low value. As explained earlier, this is related to the gaseous CO2 present in the piping of the recovery step which is depositing onto the packing surface. Immediately the valve connected to the CO2 buffer vessels is opened to allow the pressure to increase again and additional CO2 is deposited at the packing. At that point the valve is closed, and the pressure increases and is controlled at 1.2 bar, by allowing CO2 to leave the system 5.3 Experimental results 99

Figure 5.2: Thermocouple positioning in the packed bed. 100 Experimental demonstration in a pilot scale setup

Figure 5.3: Picture of the experimental pilot setup. 5.4 Simulations 101

Table 5.1: Conditions as used for experiments in the pilot setup.

Capture Recovery Cooling

N2 flow [NL/min] 22 - 285 CO2 flow [NL/min] 5.5 220 - Total flow [NL/min] 27.5 220 285

CO2 gas mole fraction [-] 0.2 1 0 Inlet gas temperature [◦C] 100 20 -155 Set point temperature [◦C] -100 -70 -130 Step time [min] 15 15 15

via a mass flow controller. The flow through this mass flow controller is plotted in Fig. 5.5b. Note that initially the pressure is well maintained at 1.2 bar, but that at some point the pressure increases further to approximately 1.28 bar. This is related to the fact that the mass flow controller is fully opened (15 NL/min) as observed in Fig. 5.5b. During the cooling step, a pressure of 1.2 bar is measured, which is slightly changing during the step, which is likely related to the changing gas phase temperature in the bed. During all steps, the pressures at the inlets and outlets of the bed have been monitored. It was observed that the pressures at the inlet and outlet show virtually equal pressures during all three steps, meaning that the pressure drop over the packed bed is negligible compared to the pressure drop over piping/valves etc.

5.4 Simulations

In this section the experimental results are studied in more detail and have been compared to the simulation results of the numerical model developed in Chapter 2. Subsequently, the model will be extended with a separate heat balance for the wall, to provinde a better understanding of the observed behavior. 102 Experimental demonstration in a pilot scale setup

Bed #1 Bed #2 Bed #3

-40

-80 Temperature [°C]

-120

0 50 100 150 200 250 300 350

Time [min]

(a)

Capture Recovery Cooling 40

-40

-80 Temperature [°C]

-120

110 120 130 140 150 160

Time [min]

(b)

Figure 5.4: Evolution of the temperature at the outlets of the three beds for a long run (a) and zoomed in for one process cycle for one bed (b). 5.4 Simulations 103

1.4 Capture Recovery Cooling Recovery Recovery Recovery

20 bed #1 bed #3 bed #2

1.2

1.0

0.8 vent [NL/min] 2 Pressure [bar]

5 CO

0.6

110 120 130 140 150 160 110 120 130 140 150 160

Time [min] Time [min]

(a) (b)

Figure 5.5: Pressure at the outlet of bed #1 during the three process steps (a) and

the CO2 gas ﬂow towards the vent during the recovery steps of the three beds (b).

5.4.1 Radial temperature profiles The experimental temperature development during the capture step is shown for two different radial positions (r = 0 cm and r = 4 cm) for two axial locations: z = 16 cm (Fig. 5.6a) and z = 28 cm (Fig. 5.6b). It can be observed that the temperature close to the wall (r = 4 cm) is initially higher than the temperature in the radial center. The effect on the temperature development is that the front of freezing CO2 is moving faster through the bed close to the wall. These experimental results are compared with simulations with the one-dimensional model (see Chapter 2, assuming an initial bed temperature averaged over the two radial positions. The resulting breakthrough times match well with the experimental outcomes and is exactly positioned between the two measured temperatures. The evolution of the temperature has also been plotted for the recovery step for z = 16 cm (Fig. 5.7a) and z = 28 cm (Fig. 5.7b). A large difference in the temperature profiles between the two radial positions can be observed. The simulated temperature profile is again in between the measured profiles. Remarkably, the CO2 is removed faster from the radial center than from the zone closer to the walls. Based on the capture step, one would expect lower mass deposition of CO2 close to the walls (because of the lower initial bed temperature) and therefore faster 104 Experimental demonstration in a pilot scale setup

-40 -40

measured, r = 0 cm measured, r = 0 cm

measured, r = 4 cm measured, r = 4 cm

-60 -60

simulated simulated

-80 -80

-100 -100

-120 -120

-140 -140 Temperature [°C] Temperature [°C]

-160 -160

0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14

Time [min] Time [min]

(a) z = 16 cm (b) z = 28 cm

Figure 5.6: Temperatures measured in the radial center (r = 0 cm) and close to the wall (r = 4 cm) and the simulated temperature for two axial locations: z = 16 cm (a) and z = 28 cm (b) during the capture step.

removal. Apparently, the wall is playing a significant role, which is also visible for the cooling step, discussed below. Finally, the temperature development is shown for two axial locations: z = 16 cm (Fig. 5.8a) and z = 28 cm (Fig. 5.8b) for the cooling step. The temperature measured close to the wall shows a much more dispersed temperature profile. Therefore, the zones close to the wall require much longer cooling times in order to reach a low temperature. The simulation outcomes show that the temperature development is again in between the profiles measured at the two radial positions, but the profile is as steep as the profile measured in the radial center. This is attributed to the effect of the vessel wall, which will be studied in more detail in the following section.

5.4.2 Inﬂuence of the wall The previous section showed that the one-dimensional numerical model can predict the average breakthrough times, but is not able to explain the differences between the temperature measurements in the radial center and close to the wall of the bed. The difference in temperature proﬁles between these locations is particularly pronounced for the cooling step. The temperature development close to the wall is very disperse, which cannot 5.4 Simulations 105

measured, r = 0 cm measured, r = 0 cm

80 80

measured, r = 4 cm measured, r = 4 cm

simulated simulated

40 40

0 0

-40 -40

-80 -80 Temperature [°C] Temperature [°C]

-120 -120

0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14

Time [min] Time [min]

(a) z = 16 cm (b) z = 28 cm

Figure 5.7: Temperatures measured in the radial center (r = 0 cm) and close to the wall (r = 4cm) and the simulated temperature for two axial locations: z = 16 cm (a) and z = 28 cm (b) during the recovery step.

120 120

measured, r = 0 cm measured, r = 0 cm

80 80 measured, r = 4 cm measured, r = 4 cm

simulated simulated

40 40

0 0

-40 -40

-80 -80 Temperature [°C] Temperature [°C]

-120 -120

-160 -160

0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14

Time [min] Time [min]

(a) z = 16 cm (b) z = 28 cm

Figure 5.8: Temperatures measured in the radial center (r = 0 cm) and close to the wall (r = 4 cm) and the simulated temperature for two axial locations: z = 16 cm (a) and z = 28 cm (b) during the cooling step. 106 Experimental demonstration in a pilot scale setup

be predicted by the one-dimensional model. It is very well possible that the axial heat conduction in the wall has an inﬂuence on the temperature in the region close to the wall. A detailed description of the temperature in the bed as function of the radial and axial position can be obtained by extending the model to two dimensions. However, it is also possible to get more insight in the behavior by adjusting the one-dimensional model. In stead of accounting for the heat capacity of the wall in the accumulation term, an extra energy balance for the wall is included in the model, in which heat is exchanged with the packed bed:

∂T ∂ ∂T (ρ C ) w = λ − a α (T − T ) (5.1) w p,w ∂t ∂z w ∂z w gw w g The energy balance for the bed (both the gas and the solid phase) will therefore also be extended with an extra contribution due to the heat exchange with the wall:

∂T ∂T (ε ρ C + ρ (1 − ε )C ) g = −ρ v C g + g g p,g s g p,s ∂t g g p,g ∂z nc ∂ ∂Tg ′′ λeff − m˙ a ∆H + a α (T − T ) (5.2) ∂z ∂z i s i b gw w g Xi=1 in which the expressions used to calculate the bed to wall heat transfer coefficient (αgw) are summarized by Tiemersma (2010). Simulations have been carried out assuming that the initial wall temperature is equal to the temperature measured close to the wall, and that the packing has an initial temperature as measured in the radial center. The temperature development for both the wall as well as the bed have been plotted as function of the time for z = 16 cm in Fig. 5.9. It can be clearly observed that the temperature within the wall is very dispersed, which is explained by a much higher value of the conductivity of the solid steel wall compared to the effective conductivity within the bed. Based on this result, the dispersed measured temperature close to the wall can be better understood. Accounting for the wall separately, also leads to a better prediction of the temperature within the packing in the radial center of the bed. The wall will also influence the temperature development during the capture and recovery step. It is very well possible that CO2 will deposit at the tube wall (which is not accounted for in the extended model). This 5.5 Discussion and conclusions 107

120

measured, r = 0 cm

measured, r = 4 cm

simulated, r = 0 cm

simulated wall

-40

Temperature [°C] -80

-120

-160

0 2 4 6 8 10 12 14

Time [min]

Figure 5.9: Effect of accounting for the wall separately in simulations for the temperature development at z = 16 cm during the cooling step.

could well explain the slower removal of CO2 during the recovery step close to the wall. Locally there will be more CO2 deposited, due to the heat capacity of the wall and therefore it will take longer to remove solid CO2 from those zones. An additional effect may be that the pressure drop will locally increase due to the higher amount of deposited CO2, resulting in lower ﬂow rates and therefore even slower solid CO2 removal rates.

5.5 Discussion and conclusions

Chapter 3 showed the experimental demonstration of the capture step in a small scale setup. In this chapter the entire process cycle including the cooling, recovery and capture step was demonstrated in an advanced fully automated experimental pilot setup. Three beds were operated in parallel, therefore enabling the option to operate the process continuously. The numerical model developed in Chapter 2 was again able to describe the temperature proﬁles within the beds, also during the recovery and cooling step. However, the walls of the packed bed caused temperature differences measured in the radial center and close to the wall and also 108 Experimental demonstration in a pilot scale setup

a much more dispersed temperature front closer to the wall, especially for the cooling step. To get better understanding of the wall effects, the numerical model was extended for the cooling step with an additional energy balance for the wall, accounting for the heat exchange with the packing. The simulation results showed that the temperature development within the bed can indeed be better described. Note that these wall effects are important in the interpretation of the experimental results of the pilot setup, but can be neglected for beds with large diameters, where the volume of the wall is much smaller compared to the bed volume.

Acknowledgment

Jeroen Zijp is kindly acknowledged for his substantial contribution to the design of the experimental setup described in this chapter.

Notation

a speciﬁc surface area, [m2/m3] Cp heat capacity, [J/kg/K] dp particle diameter, [mm] ID inner diameter, [mm] L bed length, [mm] ′′ m˙ i mass deposition rate per unit surface area for component i, [kg/m2/s] nc number of components, [-] OD outer diameter, [mm] r radial coordinate, [m] t time, [s] T temperature, [K], [◦C] v superﬁcial velocity, [m/s] z axial coordinate, [m]

Greek letters 2 αgw heat transfer coefﬁcient gas to wall, [W/m /K] ∆Hi enthalpy change related to the phase change of component i, [J/kg] εg bed void fraction, [-] 5.5 Discussion and conclusions 109

λ thermal conductivity, [W/m/K] λeff effective conductivity, [W/m/K] ρ density, [kg/m3]

Subscripts g gas phase i component i s solid phase w wall 110 Experimental demonstration in a pilot scale setup 6

Techno-economic evaluation

Abstract

A techno-economic evaluation of the proposed process cryogenic concept is presented in this chapter. A basic process design focusing on the CO2/N2 separation for a 600 MW coal fired power plant is given and the CO2 avoidance costs have been calculated. The influence of several process parameters have been investigated: lower initial bed temperatures and higher CO2 concentrations in the feed result in more efficient use of the bed volume. The pressure drop over the system plays an important role in the process economics, due to the high flow rates required in the process. The cryogenic concept is compared to two competing technologies: amine absorption and membrane separation. The results show that the preferred technology highly depends on assumptions related to the availability of utilities. If the required cold duty is available at a relatively low price, e.g. via integration with a LNG regasification station, the 112 Techno-economic evaluation

proposed cryogenic process concept can well compete with alternative technologies.1

6.1 Introduction

Several economic studies to CO2 capture methods have been published in literature. Resulting costs vary strongly, as they are highly influenced by the system boundaries such as the CO2 source and therefore inlet concentration, whether or not transport and storage is included, the level of maturity and cost measures and assumptions. For example, the opti- mized costs in a study by Abu-Zahra et al. (2007a,b) to CO2 capture by 30% MEA absorption from a 600 MW bituminous coal-fired power plant e have been estimated at 33 /ton CO2 avoided. On the other hand, in a study by van Straelen et al. (2010) to CO2 capture from a refinery also us- e ing 30% MEA, costs of 90-120 /ton CO2 avoided were reported, dependent on the scale and the CO2 concentration in the flue gas. Merkel et al. (2010) evaluated a process based on CO2 capture using membranes and calculated CO2 capture costs of $39/ton CO2. In a report by McKinsey (2008) the development of costs for CCS (including storage costs) is analyzed over the next twenty years. They expect that early demonstration projects will operate at 60-90 e/ton, but that costs could come down to 30-45 e/ton in 2030, a price level which is expected to make CCS economically self sustaining. More research is required to bring down the costs. Although many studies focus on reducing operational costs, e.g. by finding novel more efficient solvents for amine scrubbing, it is at least as important to reduce capital costs in order to reduce CO2 avoidance costs (Schach et al., 2010). The aim of the work described in this chapter is to evaluate the Cryo- genic Packed Bed (CPB) concept, both on technical aspects as well as on economic performance. Furthermore, the economics of the CPB concept are compared to other post-combustion technologies, viz. amine scrubbing and membrane technology, investigating the importance of various process assumptions. The chapter is organized as follows: first a base case is defined and the costs per ton of CO2 emissions avoided are calculated. Subsequently, a sensitivity analysis of some key process parame-

1 This chapter is based on: Tuinier et al., Techno-economic evaluation of cryogenic CO2 capture - A comparison with absorption and membrane technology. Accepted for publication in the Int. J. of Greenhouse Gas Control, doi:10.1016/j.ijggc.2011.08.013. 6.2 Process evaluation 113

ters is discussed. Finally, the results will be compared to the results of economic studies on CO2 capture via amine scrubbing (Rubin, 2010) and membrane technology (Merkel et al., 2010).

6.2 Process evaluation

In order to be able to compare the CPB concept with other technologies, a basic design was made for a capture plant treating flue gas typically generated by a 600 MW coal fired power plant, which is often used as a base case in literature studies. The bed dimensions and required process conditions are obtained by carrying out simulations using the detailed model, which is described in Chapter 2. The capital and operation costs are then estimated and the costs per ton of CO2 avoided are calculated. This section ends with a parameter study, in which the influence of several key parameters on the capture costs is evaluated.

6.2.1 Base case

To simplify the comparison, only CO2 capture is taken into account without impurities and H2O removal. The assumed flue gas conditions and composition are shown in Table 6.1. The bed dimensions and properties for the base case are detailed in Table 6.2. The initial bed temperature ◦ was set at -150 C, which results in more than 99.9% CO2 recovery. A breakthrough time (duration of each step) of 600 seconds was chosen. The required flow rates, pressures and inlet temperatures are listed in Table 6.3. The resulting pressure drops over the beds (also shown in Table 6.3) are rather small due to the nature of the selected packing; a structured monolith. However, gas distribution over the beds, piping and valves will cause an additional pressure drop, therefore a total pressure drop of 100 mbar is assumed for the capture step and 200 mbar for the cooling and recovery step (because of the higher volumetric flow rates). During the recovery step, the outgoing CO2 flow has a temperature of -78◦C and will be partly recycled to the inlet. Due to a temperature increase during the compression by the recycle blower the inlet temperature during the recovery step is increased to -66◦C. The flue gas temperature is estimated at 150◦C, but will increase in temperature to 162◦C, also because of compression. The resulting axial temperature and mass deposition profiles are shown in Fig. 6.1. It can be observed that the heat 114 Techno-economic evaluation

stored in the bed during the capture step is being used during the recovery step to evaporate previously deposited CO2. Furthermore, it can be observed that during the cooling step not the entire bed has to be cooled down, as the last part of the bed will be cooled down during the capture step.

6.2 Process evaluation 115

4 4 0 s 300 s 600 s 3 3 2 2 (f) (c) Axial position [m] position Axial [m] position Axial 1 1 0 0 0 0 50 80 60 40 20 -50 150 100 100

-100 -150

] Mass deposition [kg·m deposition Mass

Temperature [°C] Temperature

-3

4 4 ition (d - f) proﬁles for the capture, recovery 0 s 300 s 600 s n be found in Table 6.1, 6.2 and 6.3. 3 3 2 2 (e) (b) Axial position [m] position Axial [m] position Axial 1 1 0 s 300 s 600 s 0 0 0 0 50 80 60 40 20 -50 150 100 100

-100 -150

] [kg·m deposition Mass

Temperature [°C] Temperature

-3

4 4 0 s 300 s 600 s 0 s 300 s 600 s 3 3 2 2 (a) (d) Axial position [m] position Axial [m] position Axial 1 1 0 0 0 0 50 80 60 40 20 -50 150 100 100

-100 -150

] [kg·m deposition Mass

Temperature [°C] Temperature -3 Figure 6.1: Simulated axialand temperature cooling (a - step. c) Operating and conditions mass and depos bed properties ca 116 Techno-economic evaluation

Table 6.1: Flue gas conditions and composition.

Temperature [◦C] 150 Pressure [bar] 1.013 Vol.% Flow [kg/s] N2 86.5 510 CO2 13.5 125 Total 635

Table 6.2: Bed dimensions and properties.

Diameter [m] 8.5 Length [m] 4.25 Number of beds [-] 21 (7 per step) Packing Steelmonolithstructure Density solids [kg/m3] 7750 Porosity [-] 0.697

6.2.2 Costs base case

In order to calculate the CO2 avoidance costs, the capital investment costs are ﬁrst calculated using a conceptual cost estimation method with an accuracy of 40%. In this method, the main equipment costs are estimated. Fig. 6.2 shows a simpliﬁed process scheme with all main equipment. The costs for blowers, the heat exchanger and the columns have been calculated using correlations reported by Seider et al. (2004) and Loh and Lyons (2002) and were updated to costs in 2010 using

Table 6.3: Process parameters base case.

Capture Recovery Cooling ◦ Tin [ C] 162 -66 -150 Pin [mbar] 1100 1200 1200 Flow/bed [kg/s] 91 564 357 ∆P packing [mbar] 16.9 82.5 56.7 Total ∆P[mbar] 100 200 200 6.2 Process evaluation 117

the Chemical Engineering Plant Cost Index (CEPCI). The packing costs are calculated using a steel price of $1200/ton steel (market price of $600/ton multiplied with factor two for packing construction). The module costs, including piping, installation etc. are then calculated by multi- plying the equipment costs with a Hand factor. When all the module costs are summed up, 25% is added for contingencies. The total direct investment is subsequently calculated and an allocated investment (for storage, utilities and environmental provisions), start up investments and working capital are added. Finally, the total fixed capital is calculated, results are shown in Table 6.4. The operational costs consist of the electricity costs required for the blowers. The CO2 emitted due to the additional power required by the blowers could be captured as well, but is assumed to be emitted into atmosphere in this study. For this base case the cooling is provided by the evaporation of LNG and no additional costs are assumed. Depreciation, interest, labor and maintenance are calculated using 20% of the total capital charge per year. The used cost parameters can be found in Table 6.5. The operational and final CO2 avoidance costs are summarized in Table 6.6. The total costs per ton of CO2 emission avoided amounts $52.8. The capital costs ($28.9/ton CO2 avoided) and the operational costs ($23.9/ton CO2 avoided) have a similar share in the avoidance costs. 118 Techno-economic evaluation Figure 6.2: Simplified process scheme. 6.2 Process evaluation 119 233 338 345 46.6 Table 6.4: Capital investment costs for the base case. recycle blower 10.09 2.5 25.2 product compressor 15.09 2.5 37.7 cooling blower 10.55 2.5 26.4 2 2 2 EquipmentColumn for packed bedPacking (21) (21)Flue gas compressorCO 0.39 Equipment 1.12 costs [M$]Total direct Hand investment factor (TDI) Totalallocatedinvestment 0.67 Module costsStartupinvestment [M$] 4Total process investment 40%ofTDI (TPI) Working capitalTotal fixed capital 2.5 5%ofTDI 32.9 4 2% 2.8 of TPI 56.5 93 12 7 N CO LNG heat exchangerContingencies 1.03 25% 4.8 4.9 120 Techno-economic evaluation

Table 6.5: Cost evaluation parameters.

Operation hours per year 7000 Capital charge 0.2 Blower/compressor/pump efﬁciencies 0.72 Electricity price [$/kWh] 0.06 CO2 emission due to additional power [ton/MWh] 0.8042 CO2 product pressure [bar] 140 6.2 Process evaluation 121 avoided 2 52.8 MW $/hr $/ton CO Table 6.6: Operational and total costs for the base case. captured [ton/hr]recycle blower 450 34.4 2063 6.1 emitted due to additional power [ton/hr] 109.2 avoided [ton/hr] 340.8 product compressor 56.9 3412 10.0 cooling blower 36.4 2181 6.4 2 2 2 2 2 2 CO Flue gas blowerCO Total electricity blowersCapital/maintenance/labor chargeTotal costs 8.2 9850 135.8 490 8145 28.9 23.9 1.4 CO N CO CO 122 Techno-economic evaluation

6.2.3 Parameter study Initial bed temperature The process has been evaluated for different initial bed temperatures. Both the CO2 avoidance costs and LNG consumption are shown in Fig. 6.3. At a higher initial bed temperature, less CO2 is deposited per unit of bed volume. Therefore, more bed volume is required to maintain similar breakthrough times, resulting in increasing capital costs. Larger flow rates during the recovery and cooling step are required as well, in order to finish in 600 seconds. A larger flow rate during the cooling cycle results in a higher LNG consumption. An initial bed temperature of - 160◦C results in even more efficient use of the beds and therefore slightly lower costs and LNG consumption. However, the temperature difference ◦ between LNG (-162 C) and the refrigerated N2 becomes too small. It can be concluded that a lower bed temperature results in more efficient CO2 capture. It should be noted that this conclusion cannot be drawn when (part) of the cooling is generated by refrigeration. The efficiency of a refrigerator decreases with decreasing temperatures and results in higher cooling costs.

CO2 concentration in ﬂue gas

The CO2 concentration in flue gases depend on the used feedstock and process. A concentration of about 5 vol.% is for example encountered in natural gas fired combined cycle power plants. The effect of the CO2 concentration on the performance of the cryogenic packed bed concept is summarized in Fig. 6.4. The amount of flue gas (635 kg/s) is kept constant for all cases. The front of desublimating CO2 will move slower through the bed at decreasing inlet CO2 concentrations. There- fore smaller equipment can be used (when maintaining an equal breakthrough time) and consequently lower flow rates are required for the recovery and cooling steps. However, at the same time the amount of CO2 captured will decrease, due to the lower CO2 content in the gas. The reduction in equipment size and required flows is cancelled out by this decrease. An inlet concentration of 5 vol.% results in avoidance costs of $95.7/ton, which are substantially higher than for the base case ($52.8/ton). The increase in costs is especially strong when going to even lower concentrations, which is related to the CO2 emissions caused by 6.2 Process evaluation 123

100 eaoain[o/o O C [ton/ton evaporation G N L

Operational costs ]

Capital costs 2

LNG evaporation 80

20 2 Avoidance costs [$/ton CO ]

0 0

-160 -150 -140 -130

Initial bed temperature [°C]

Figure 6.3: Avoidance costs and LNG consumption as function of the initial bed temperature. the extra power required. The ratio of the additional required power to the amount of CO2 captured becomes high at low concentrations. Fig. 6.4 also shows that a CO2 inlet concentration of 15% results in lower avoidance costs. At even higher CO2 concentrations, recovery of the beds becomes more difficult, as the heat stored in the first zone of the bed during the capture step becomes insufficient. Additional heat has to be supplied to the process to recover CO2 in those cases.

Pressure drop A pressure drop of 100 mbar for the capture step and 200 mbar for the recovery and cooling step were assumed. The actual pressure drop depends on packing type, tubing diameters but possibly also to a large extent on the gas distribution over the shallow packed beds. For a better distribution a larger pressure drop is required. A non-uniform distributed feed might result in different freezing/evaporating front velocities at different radial positions and therefore in a non-optimal use of the bed volume. Earlier or less sharp breakthrough might be observed, resulting in a lower capture rate of CO2 or higher LNG consumption. The amount of 124 Techno-economic evaluation

140 6 eaoain[o/o O C [ton/ton evaporation G N L

Operational costs ]

120 2 Capital costs

LNG evaporation

100

20 2 Avoidance costs [$/ton CO ]

0 0

5 13.5 15 10

CO concentration [vol.%]

Figure 6.4: Avoidance costs and LNG consumption as function of the CO2 inlet concentration. maldistribution which is still acceptable is unknown and requires more study. To indicate the significance of the pressure drop on the process performance and economics, two cases with 50% higher and lower pressure drops were evaluated. Fig. 6.5 shows that the pressure drop over the beds has a significant effect on the operational costs, which is explained by higher compression costs. Also the amount of required LNG changes slightly, which is related to the heat generated by compression. It should be noted that some CO2 bypass might be tolerated, since often 90% capture is deemed sufficient.

6.3 Comparison with absorption and membrane technology

The economics of the cryogenic packed bed concept is compared to absorption and membrane technology in this section. In order to present a comparison as fair as possible, costs are calculated based on the same 6.3 Comparison with absorption and membrane technology 125

100 4 eaoain[o/o O C [ton/ton evaporation G N L

Operational costs ]

Capital costs 2

LNG evaporation

20 2 Avoidance costs [$/ton CO ]

0 0

150/300 100/200 50/100

Pressure drop [mbar]

Figure 6.5: Avoidance costs and LNG consumption as function of the pressure drops during the capture step (left value) and recovery and cooling step (right value).

cost parameters as used in the evaluation of the cryogenic concept (as shown in Table 6.5).

6.3.1 Absorption technology

The required input for the evaluation of CO2 capture costs via absorption technology is obtained from the Integrated Environmental Control Model developed by Rubin (2010). A 600 MW power plant with monoethanolamine (MEA) absorption was simulated, resulting in a flue gas of 666 kg/s containing 14 vol.% CO2 (on a dry basis), which is similar to the flue gas composition as used in the evaluation of the cryogenic concept. The costs for all purification steps upstream the capture process (NOx, SO2 and particulates removal) are not taken into account. The equipment costs and the electricity, steam, MEA and corrosion inhibitor consumption are taken from the model. Steam required for stripping is generated with an auxiliary boiler in the simulation, but capital and 126 Techno-economic evaluation

operational costs are not taken into account. The resulting values are presented in Table 6.7 and 6.8.

6.3.2 Membrane technology

Merkel et al. (2010) carried out a basic study on the economics of CO2 capture with membrane technology, treating flue gas of 602 kg/s containing 12.9 vol.% CO2. In their evaluation two process alternatives were evaluated. In the first option the driving force for permeation is generated by a vacuum on the permeate side. In the second option an air sweep is used, which is then fed to the boiler of the power plant. Although the second alternative is more efficient and looks promising, it will not be taken into account in this study, as it will influence the combustion process and might be more difficult to retrofit to existing facilities. The calculated equipment costs only consist of the membrane costs and compressors/expanders, but are multiplied with an installation factor. The capital and operational costs adjusted with the parameters used in this study are also shown in Table 6.7 and 6.8. 6.3 Comparison with absorption and membrane technology 127 Membranes membrane technology. 22.3 Amine scrubbing Table 6.7: Capital investment costs for amine scrubbing and absorber 81.3 Compressors/expanders 100 2 EquipmentFlue gas blowerCO Module costs [M$]Total direct investmentTotal allocated investmentStart up Equipment investmentTotal process investment 5.7Working capitalTotal fixed capital Membranes 119.7 299.3 Module costs [M$] 434 Total 15.0 allocated Total investment direct investment Total Start process up investment investment 8.7 443 200 150 Working capital Total 500 fixed capital 725 25 740 15 Heat exchangersCirculation pumpsSorbent regeneratorReboiler Sorbent reclaimer/processingDrying/compression unitContingencies 15.4 12.7 6.2 46.7 49.1 Installation factor (60%) 59.9 Contingencies 150 100 128 Techno-economic evaluation avoided 2 d membrane technology. 7.0 1.4 9.92.1 - 0.2 11.354.5 36.0 120.9 $/ton CO Amine scrubbing Membranes captured [ton/hr]productcompressor 439 369 9.1 emitted due to additional power [ton/hr] 58 120 avoided [ton/hr] 381 249 2 2 2 2 Table 6.8: Operational and total costs for amine scrubbing an CO Sorbent Inhibitor Reclaimer waste disposalTotal chemicals Flue gas blower CO Total power costs Capital charge (20% total fixed capital/year)Total costs 1.4 33.2 84.9 CO Solvent pump CO 6.3 Comparison with absorption and membrane technology 129

6.3.3 Comparison

The CO2 avoidance costs for all three technologies are compared in Fig. 6.6. Amine scrubbing and the cryogenic concept have comparable costs, while membranes are signiﬁcantly more expensive in this evaluation. The results are highly dependent on the assumptions, especially on the availability of utilities. In the amine case it was assumed that steam is available at no costs, which is unrealistic. When no steam is available at all and the operational and capital costs and additional CO2 emissions related to steam generation in an auxiliary boiler are taken into account, the avoidance costs for scrubbing become high ($133.4/ton). This is related to the large amount of heat required during the stripping of MEA (4.5 MJ/kg CO2). The model developed by Rubin also offers the possibility to select an advanced amine, which is used in Fluors Econamine R FG+ process. The resulting avoidance costs are substantially lower ($84.2/ton), related to less steam required for regeneration and lower degradation rates. The cryogenic concept is attractive when the cold exposed during the regasiﬁcation of LNG could be used for free. If no LNG is available and the entire required cooling capacity should be generated using cryogenic refrigerators, the electricity consumption of the refrigerators would be in the same order of magnitude as the electricity production of the power plant, and can therefore be considered as unrealistic. Furthermore, the evaporation of LNG could be integrated with other processes, therefore LNG might be only available at certain costs. When comparing to the avoidance costs of the advanced amine, a maximum price of $8.7/ton LNG can be allowed. The required cooling power is 248 MW, which corresponds to an LNG consumption of 2.7 kg LNG/kg CO2 avoided. An average sized LNG terminal evaporates about 5 million ton/year. Based on an operation of 7000 hours per year, a total amount of 8.6 million ton of LNG would be required for cooling, which corresponds to more than one terminal. When only one terminal is available and the remaining cooling duty has to be generated by refrigeration, the avoidance costs will be $314.4/ton avoided, which is still excessively high. In the situation that the pressure drops can be reduced as shown in Fig. 6.5, less cooling is required and the total avoidance costs when taking refrigeration into account results in $126.5/ton, which makes it competitive with the other technologies. 130 Techno-economic evaluation

180

Cooling Electricity

160

] Steam Capital charge 2

Chemicals

140

120

100

Avoidance costs [$/ton CO 20

Amine Amine

Adv. amine Cryogenic Cryogenic + steam Membranes

+ steam LNG + cooling

Figure 6.6: Avoidance costs for different technologies.

CO2 removal using membrane technology is more costly than cryogenics and scrubbing, due to its high capital costs. When the costs of the membrane modules could be reduced in the future, this option may become competitive, especially when cold or steam utilities are not available/expensive. The cryogenic concept shows the advantage that deep CO2 removal can be obtained, generating both a very pure cleaned ﬂue gas as CO2 ◦ product. When cooling to -150 C, the vapor pressure of CO2 is only 8 Pa, resulting in more than 99.9% CO2 removal, compared to 90% for the other technologies. To quantify the exact advantage of this ‘deep’ CO2 removal, the costs for CO2 emissions should be known. The removal of impurities is not incorporated in this study. The cryogenic concept has the potential to remove water and for example sulfur containing impurities simultaneously, as vapor pressures are low at the used temperatures. In that case it could be necessary to install a small separate bed, which allows separate regeneration. Future work will focus on these aspects. 6.4 Conclusions 131

6.4 Conclusions

The costs of cryogenic CO2 capture using dynamically operated packed beds depend strongly on initial bed temperatures and CO2 concentrations in the feed gas. At lower initial temperatures the cold stored in the bed can be used more efficiently, resulting in more CO2 deposited per unit of bed volume. At low CO2 inlet concentrations, the relative costs for the amount of CO2 avoided increase strongly. Due to high flow rates required during the process, the pressure drops over the system substantially influence the CO2 avoidance costs. It is expected that required gas distribution plays an important role in the resulting pressure drop. Future research should focus on the effects of gas (mal)distribution (and hence the required pressure drop over the gas distributor) on the process performance. In the comparison with other technologies it was found that the preferred technology depends heavily on the availability of utilities. The cryogenic concept requires a cold source, such as the evaporation of LNG at a regasification terminal, while amine scrubbing requires low pressure steam in order to strip the solvent. When both LNG and steam are not available at low costs, membrane technology shows advantages. When steam is available at low costs, especially when using an advanced amine, scrubbing is the preferred technology. The cryogenic concept could be the preferred option, when LNG is available at low costs. Especially when pressure drops can be decreased and the simultaneous removal of impurities can be incorporated in one process, the concept could become a serious candidate for capturing CO2 from flue gases.

Acknowledgment

Paul Hamers is kindly acknowledged for his excellent contribution to the work described in this chapter.

Notation

P pressure, [mbar] T temperature, [◦C] 132 Techno-economic evaluation

Abbreviations: CEPCI Chemical Engineering Plant Cost Index CPB Cryogenic Packed Bed LNG Liqueﬁed Natural Gas MEA Mono Ethanol Amine TDI Total Direct Investment TPI Total Process Investment 7

Biogas puriﬁcation

Abstract

This chapter demonstrates with numerical simulations the option to use the proposed process concept for biogas treatment. The performance is compared to Vacuum Pressure Swing Adsorption (VPSA) on the basis of several criteria: purity and recovery of the obtained product, bed dimensions and energy requirements. Simulation results reveal that the purity and recovery of CH4 are higher for the cryogenic packed bed (CPB) concept, while also the bed capacity is much higher: the productivity (deﬁned -1 -3 as kgCH4 h m packing) is a factor eight higher. The recovery is carried out with air and when operated in reversed ﬂow mode, the CPB technology

requires a 22% lower energy duty (2.9 MJ/kgCH4 vs. 3.7 MJ/kgCH4 for the VPSA process). Furthermore, simultaneous deep H2S removal is possible using the proposed concept, although initial bed temperatures as low as -150◦C are required.1

1This chapter is based on a paper submitted to Industrial & Engineering Chemistry Re- search. 134 Biogas puriﬁcation

7.1 Introduction

Biogas is formed at e.g. landfill sites, waste water treatment facilities or is produced by anaerobic fermentation of manure. Biogas mainly consists of CH4 (50-70 vol.%) and CO2 (25-45 vol.%) and furthermore may contain contaminants such as H2O, H2S and siloxanes (Patterson et al., 2011). CO2 is a well known greenhouse gas, however, CH4 has a relative global warming potential 25 times higher than CO2 (Sejian et al., 2011). It is reported that greenhouse gas emissions from the agricultural sec- tor account for about 25.5% of the total global anthropogenic emissions (Sejian et al., 2011). It is therefore critical to bring down CH4 emissions. The cheapest option to avoid emissions is to collect biogas and send it to a flare. However, biogas based methane has the potential to serve as a renewable energy resource, to generate power or to be used as a transportation fuel. To convert biogas to commercial grade CH4, several separation and purification steps are required. H2S is formed by the anaerobic fermentation of sulfur containing pro- teins. H2S removal is necessary, as combustion will lead to the envi- ronmentally hazardous SO2 and could form H2SO4, causing corrosion of process equipment. H2S can be removed by oxidation to elementary sulfur or by scrubbing the biogas with an aqueous alkaline solution. Dis- advantage of the latter is that CO2 has a higher reactivity with the solvent, causing low selectivities for H2S removal (Abatzoglou and Boivin, 2009). Siloxanes are a group of molecules containing Si, which are found in landfill biogases. The combustion of siloxanes leads to silicates and micro crystalline quartz. These solids will damage engines and turbines, and are therefore highly undesirable. Several possible technologies to remove siloxanes are available: (reactive) absorption with liquids, adsorption and cryogenics (Abatzoglou and Boivin, 2009). CO2 is present in high concentrations in biogas. Removal of CO2 is necessary to increase the energy content of the biogas. The available technologies are: scrubbing (using water, a physical or chemical solvent), cryogenic separation, membranes or Pressure Swing Adsorption (PSA) (Basu et al., 2010; Patterson et al., 2011). Similar technologies are applied or studied for CO2 removal from flue gases. However, differences with flue gas treatment are the higher CO2 content in the feed, the lower temperature at which the gas is available and the product requirements. The desired product in flue gas treatment is CO2, while biogas treat- 7.2 Adsorption 135

ment is focused on obtaining CH4. The removed CO2 is normally not considered for sequestration, as the amounts of CO2 produced are lower, making CO2 collection and transportation relatively expensive. All technologies listed above have been commercially applied to remove CO2 from biogas, but energy requirements are normally high due to required pressure differences or elevated temperatures for solvent recovery. This chapter describes the possibilities of using the cryogenic capture technology developed in this thesis and compares it to adsorption technology on the basis of product purity and recovery, bed dimensions and energy requirements. To be able to compare the performance and requirements of the two processes, a speciﬁc detailed study on adsorption from literature has been selected (Grande and Rodrigues, 2007), in which both dimensions and properties of the beds as well as energy requirements have been provided. First, some details about the selected adsorption process are given. Subsequently, the possibility of using the CPB concept is explored by carrying out simulations using the numerical model as described in Chapter 2. Finally, the two processes are compared for the removal of CO2 from a CH4/CO2 mixture. Finally, additional simulations are presented which show the possibility of simultaneously removing H2S using the proposed process concept.

7.2 Adsorption

Adsorption technology is a widely applied separation technology. Sep- aration of CO2 from gas mixtures by adsorption is based on differences in equilibrium capacities at the adsorbent surface (e.g. zeolite 13X) or on differences in uptake rates (e.g. carbon molecular sieve 3K). Regeneration of a bed is normally obtained by reducing the pressure. An adsorption process therefore consists of several packed beds which are operated simultaneously in different process steps, e.g. feed, recovery and pressurization. The removal of CO2 from biogas using adsorption is widely discussed in literature (Grande and Rodrigues, 2007; Kapoor and Yang, 1989; Kim et al., 2006; Esteves et al., 2008; Ribeiro et al., 2008). Grande and Rodrigues (2007) analyzed Vacuum Pressure Swing Adsorption (VPSA), in which the recovery step is carried out at sub-atmospheric pressure. The purity and recovery of CH4 and energy consumption by changing pressures and cycle times have been investigated for two different adsorbents, Carbon Molecular Sieve (CMS) 136 Biogas puriﬁcation

Table 7.1: Bed properties

Property CPB VPSA Length [m] 1.65 4.667 Radius [m] 0.3 0.4667 Porosity [-] 0.7 0.33 Packing material SS CMS Type of packing monolith particles Bulk density [kg m-3] 2347 715.4 Number of beds [-] 2 2

Table 7.2: Operating conditions

Property CPB VPSA Feed ﬂow rate [SLPM] 16000 16000 Feed pressure [bara] 5 8 Recovery pressure [bara] 1.1 0.1 Initial bed temperature [◦C] -110 20 Cycle time [s] 140 140 Purity [CH4%] 99.1 98.1 Recovery [CH4%] 94.3 79.7 -1 -3 Productivity [kg CH4 h m packing] 350.2 43.1

3K and zeolite 13X. Lowest energy requirements have been obtained in a run with CMS as adsorbent. In this run a gas mixture of 16 000 SLPM (0.312 kg/sec), containing 45 vol.% CO2 and 55 vol.% CH4 is treated. It is assumed that the biogas is available at 2 bara and 25◦C. Furthermore it is assumed that water and other contaminants have been previously removed. The bed properties and operating conditions are listed in Ta- ble 7.1 and 7.2 respectively. For further details the reader is referred to their work.

7.3 Cryogenic packed bed concept

Although the process is originally developed to capture CO2 from ﬂue gases, it can also be applied to a wide range of other gas separations, such as biogas treatment. In case of CO2 capture from ﬂue gases, gaseous CO2 7.3 Cryogenic packed bed concept 137

100 ]

70 s -3

0 s

140 s

70 s

140 s

-20

-40

-60

-80

20 Temperature [°C]

-100 Mass deposition [kg·m

-120

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Axial position [m] Axial position [m]

(a) (b)

Figure 7.1: Simulated axial temperature (a) and mass deposition (b) proﬁles during the capture step.

is fed to a bed during the recovery step in order to obtain pure CO2 at the outlet. For biogas treatment, CO2 capture is not required, therefore air can be used in the recovery step. The evolution of axial temperature, mass deposition and concentration profiles within the packed beds can be described using the pseudo-homogeneous one dimensional axially dispersed plug flow model described in Chapter 2. In order to compare the CPB concept with VPSA, simulations have been carried out for the cryogenic concept using an equal flow rate and gas composition as used by Grande and Rodrigues (2007). The bed is cooled down to -110◦C and the feed is pressurized to 5 bar. Under these conditions the outlet CH4 purity is 99.1%. The dimensions of the beds (see Table 7.1) are chosen in such a way, that the capture/feed step takes 140 s, similar to the VPSA case. The recovery and cooling step together also take 140 s, therefore the process can be operated continuously when two beds are operated in paralel. Axial temperature and mass deposition profiles during the capture step are shown in Fig. 7.1a and 7.1b respectively. It should be noted that the process cycle for the VPSA case consists of a (1) feed (60 s), (2) depressurization (10 s), (3) blow down (140 s), (4) purge (50 sec) and (5) pressurization step (70 s). Although their analysis is based on two beds, three beds are required if continuous operation is desired. 138 Biogas purification

100 ]

0 s

0 s -3

3 s

0 6.5 s

-20

-40

-60

-80

20 Temperature [°C]

-100 Mass deposition [kg·m

-120

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Axial position [m] Axial position [m]

(a) (b)

30 s -80

70 s

140 s

-90

-100

-110 Temperature [°C]

-120

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Axial position [m]

(c)

Figure 7.2: Simulated axial temperature (a) and mass deposition (b) profiles during the recovery step in the first seconds and the temperature profiles later in time (c). 7.3 Cryogenic packed bed concept 139

When CO2 breakthrough is observed, the bed should be recovered. By reducing the pressure to atmospheric pressures and by flushing the system with (dry) air, very fast recovery can be obtained. Therefore the recovery and cooling step can be integrated in one step by using an air ◦ flow of 5 kg/s, refrigerated to -110 C. All CO2 is recovered after 6.5 seconds, as can be observed from Fig. 7.2a and 7.2b. In this first 6.5 seconds, the outlet flow cannot be recycled to the inlet of the bed via the cooler, due to the presence of CO2. Therefore an amount of 32.5 kg of dry air is required per step (average flow of 0.23 kg/s). After all CO2 is recovered, the entire bed is further cooled down to -110◦C, in order to start a new capture step again (Fig. 7.2c). A simplified process flow diagram including flow compositions and process conditions is shown in Fig. 7.3. 140 Biogas purification or the base case. Figure 7.3: Simplified process scheme including conditions f 7.3 Cryogenic packed bed concept 141 recovery/cooling step in reversed flow mode. Figure 7.4: Simplified process scheme including conditions, 142 Biogas purification

7.4 Adsorption versus cryogenic packed bed concept

The two processes are compared on:

• Product purity and recovery of CH4 from the biogas • Required bed dimensions

• Energy requirements

Table 7.2 shows that the product purity in VPSA is 98.1% CH4 against 99.1% in the CPB concept. The recovery of CH4 in VPSA is lower (79.4%), due to losses in purge steps. The losses of CH4 in the CPB concept are limited to the amount of CH4 present in the bed before switching to the recovery. When reducing the pressure and flushing the system with air this amount of CH4 is lost: the recovery is 94.3%. The beds required in the cryogenic concept are much smaller (Table 7.1). The productivity is calculated for both processes: this value is more than 8 times higher for -1 -3 the CPB concept (350.2 vs. 43.1 kgCH4 h m packing). Grande and Rodrigues (2007) estimated the energy requirements for the adsorption process, which are summarized in Table 7.3. The total power required amounted 3.7 MJ/kg CH4. The energy requirements for the CPB concept are based on the process flow scheme shown in Fig. 7.3. Three compressors are required, the first one to pressurize the feed to 5 bar (C-1), the second one to pressurize the product to 200 bar (C- 2) and a third one to recycle air through the bed and the cooler (C-3). Simulations showed that the pressure drop in the packed bed is only 0.04 bar, which is explained by the selected nature of the structured packing (monolith). It is assumed that the cooler causes another pressure drop of 0.06 bar, therefore an outlet pressure of 1.1 bar is required for compressor C-3. All compressor duties have been calculated assuming isentropic compression and an efficiency of 72%. Results are shown in Table 7.4. Furthermore, energy is required by the refrigerator in order to cool down the air. As described earlier, an average amount of 0.23 kg/s of fresh dry air is required to recover the bed. It is assumed that this air ◦ is available at 25 C. After CO2 is removed from the bed, the refrigerated air can be recycled. The temperature of the air at the outlet of the bed is changing during the recovery/cooling step, as shown in Fig. 7.5. The average outlet temperature is -101.7◦C, which is mixed with fresh air 7.4 Adsorption versus cryogenic packed bed concept 143

Table 7.3: Energy requirements for VPSA

Step Power (kW) Compression of feed to 8 bar 57.7 Compression of product to 200 bar 73.2 Power for purge step 28.1 Blowdown step 129.8 Total power 288.8

of 25◦C, resulting in an average inlet air temperature of -95.8◦C. Due to compression in C-3, the stream heats up and will be cooled down in the refrigerator to -110◦C. The required cooling duty is 102.4 kW. However, the efficiency of a cooler in this power and temperature range is approximately 40% Timmerhaus and Flynn (1989), and therefore the power input for the refrigerator is higher. The resulting cooling costs are also listed in Table 7.4. The total energy requirement for the process then amounts to 4.3 MJ/kg CH4, which is 16% higher than for the VPSA process. The energy requirements are mainly caused by the cooling step. A relatively large air flow of 5 kg/s is required to cool down the bed to -110◦C. When taking a closer look at the temperature profiles in Fig. 7.2, it can be observed that the hot zone at the inlet of the bed (which is at the biogas inlet temperature, 25◦C), is moved through the entire bed. This is also visible in the outlet temperature as function of the time in Fig. 7.5. Therefore, it is more efficient to reverse the flow direction during the recovery/cooling step, as illustrated in the process flow scheme in Fig. 7.4. Fig. 7.6 shows the resulting axial temperature and mass deposition profiles. In this case a lower total air flow is required (3 kg/s). All CO2 is recovered after 10 seconds, therefore an average amount of 0.21 kg/s of fresh dry air is required. The resulting outlet temperature as function of the time is also plotted in Fig. 7.5. When comparing the two profiles for the base case and the reversed flow mode, it can be concluded that the average outlet temperature is lower in the reversed flow mode. Further- more, the waste stream containing CO2 is emitted at a relatively higher temperature, avoiding cold losses. Finally, the lower flow rate causes a lower power consumption by compressor C-3. The required energy for all steps are also included in Table 7.4. A total amount of 2.9 MJ/kgCH4 is required, which is 22% lower than for the VPSA process. 144 Biogas purification

Base case

Reverse recovery

-20

-40

-60

-80

-100 Outlet temperature [°C]

-120

0 20 40 60 80 100 120 140

Time [s]

Figure 7.5: Outlet temperature as function of time

Table 7.4: Energy requirements for cryogenic packed beds

Power (kW) Step Base case Reversed ﬂow Compression of feed to 5 bar 37.9 37.9 Compression of product to 200 bar 62.6 62.6 Air recycling 34.2 20.4 Cooling power 256 142.5 Total power 390.7 263.4 7.4 Adsorption versus cryogenic packed bed concept 145

100 ] 0 s

0 s -3

20 5 s

5 s

80 10 s

-20

-40

-60

-80

20 Temperature [°C]

-100 Mass deposition [kg·m

-120

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Axial position [m] Axial position [m]

(a) (b)

30 s -80

70 s

140 s

-90

-100

-110 Temperature [°C]

-120

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Axial position [m]

(c)

Figure 7.6: Simulated axial temperature (a) and mass deposition (b) profiles during the recovery step in the first seconds and the temperature profiles later in time (c). The bed is operated in reversed flow mode. 146 Biogas purification

7.5 Hydrogen sulﬁde removal

The H2S content in biogas can range from 0.0001 - 1 vol.% (Abatzoglou and Boivin, 2009). When feeding a mixture containing 1 vol.% H2S, 45 vol.% CO2 and 54 vol. % CH4 to a cryogenically refrigerated bed (equal conditions and properties as those listed in Table 7.1 and 7.2), CO2 and H2S can be separated simultaneously. However, with a bed ◦ temperature of -110 C the H2S content can only be reduced to 0.6 vol.%. If deep H2S removal is required, the initial bed temperature should be lower. The H2S content can be reduced to 40 ppmv when the initial bed temperature is -150◦C. Fig. 7.7 shows the resulting axial temperature and mass deposition proﬁles when feeding the mixture to a bed initially cooled to -150◦C. It can be observed, that both CO2 as well as H2S will deposit onto the packing surface, although the zone where H2S deposits also contains CO2. This can be explained by having a closer look at the saturation vapor pressures of both components in Fig. 7.8. When the mixture is cooled ◦ down, CO2 starts to deposit at -88 C, as indicated by the dotted line. The temperature will decrease further and below -105.1◦C the saturation vapor pressure of H2S becomes lower than its partial pressure in the feed (0.05 bar) and H2S starts to deposit as well. During the cooling of the mixture from -105.1◦C to the initial bed temperature, both components will deposit onto the packing. Therefore, it can be concluded that components with saturation vapor pressures close together will deposit in the same region. Nevertheless, these results show that both components can be efﬁciently removed from the feed gas.

7.6 Discussion and conclusions

In this chapter it is proposed to purify biogas using the process concept studied in this thesis. The possibility has been studied with numerical simulations. The cryogenic packed bed concept shows to be favorable compared to adsorption technology on several aspects. The purity and the recovery of the obtained CH4 are higher than in VPSA. Higher purity is possible (up to 99.99%) at lower initial bed temperatures. Furthermore, required bed sizes are signiﬁcantly smaller, resulting in much lower capital investments. However, good insulation of the cryogenic packed beds is required to avoid heat leaks into the system from the surroundings. 7.6 Discussion and conclusions 147

120

2 20 ] -3

H S

100 2 0

-20

-40

2 -60

-80

40 -100 Temperature [°C] -120

CO + H S

-140

2 2 Mass deposition [kg·m

-160

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Axial position [m] Axial position [m]

(a) (b)

Figure 7.7: Simulated axial temperature and mass deposition proﬁles after 140 seconds when feeding a CH4/H2S/CO2 mixture to an initially refrigerated packed bed.

8 Vapor pressure CO

2 10

7 Vapor pressure H S

2 10

Feed pressure CO

10 2

Feed pressure H S 10

10 Vapor pressure [Pa]

-1 Initial bed T

-2

-160 -140 -120 -100 -80 -60 -40

Temperature [°C]

Figure 7.8: Saturation vapor pressures of CO2 and H2S as function of the temperature. 148 Biogas puriﬁcation

Additionally, energy requirements are quite competitive with VPSA, even if a cooler efficiency of 40% is taken into account. Energy requirements are especially low when the recovery/cooling mode is operated in reversed flow mode. Furthermore, dried air (or nitrogen) is required, which may raise operating costs somewhat. The amount of dried air can be reduced by operating the recovery/cooling step at sub-atmospheric pressures, at the expense of higher operation costs. Furthermore, it should be noted that the upgraded gas is compressed to 200 bar in this study. However, if the cleaned product would be liquefied, the CPB technology shows the advantage that the product gas leaving the packed beds is already at a low temperature of -105◦C, reducing further cooling costs. This study focused on the CH4/CO2 separation and showed the possibility to simultaneously remove H2S. Results in Chapter 3 showed that H2O can be separated simultaneously using the CPB concept. Furthermore, siloxanes could also be separated in the same or a separate bed. Therefore, it can be concluded that the proposed CPB technology is a very promising process for biogas treatment.

Notation

P pressure, [bar] T temperature, [◦C]

Abbreviations CMS Carbon Molecular Sieve CPB Cryogenic Packed Bed PSA Pressure Swing Adsorption VPSA Vacuum Pressure Swing Adsorption 8

Epilogue and outlook

This dissertation discusses a novel process concept for cryogenic CO2 capture based on dynamically operated packed beds. When feeding a flue gas to a refrigerated packed bed, a separation between CO2 and N2 can be obtained. Two models to describe this process have been developed in this work, a detailed one-dimensional pseudo-homogeneous model taking axial dispersion effects into account and a simplified, but fast ‘sharp front’ model. These models have been validated using experiments in a small scale experimental setup. This setup was used to demonstrate the capture step for N2/CO2/H2O gas mixtures, showing simultaneous separation of CO2 and H2O. The entire process, including the recovery and cooling step, was demonstrated in a continuous fully automated experimental pilot setup. An unknown parameter required for the design was the desublimation rate of CO2. Therefore, a dedicated setup was designed to measure these desublimation rates under a wide range of conditions. A model to describe frost growth rates was successfully developed. It was demonstrated that mass transfer towards the cold packing is the rate de- termining step under the conditions in the packed beds. The economics of the novel concept were studied and compared with competing technologies. Especially when a cold source, such as the regasification of LNG is available at low costs, the process could compete well with other 150 Epilogue and outlook

post-combustion CO2 capture technologies. Finally, it was demonstrated that the concept could be very promising for biogas puriﬁcation, even when cooling needs to be provided by refrigeration. By using a simultaneous temperature and pressure swing after the capture step, deposited CO2 was very efﬁciently removed from the bed. It can be concluded that already many important aspects of the process concept have been discussed in this dissertation. It is attempted in this chapter to give a realistic future outlook for the concept and to list some important points of attention for future research activities.

8.1 Important aspects for future development

This section discusses some important aspects of the proposed concept. First the effects of the recovery step on the process performance are discussed, followed by a discussion on the possible consequences of radial profiles within the beds. The cooling duty required to capture a certain amount of CO2 depends on aspects such as the CO2 concentration in the feed gas and the initial bed temperature before a capture step is started. However, it is also indi- rectly related to the operation mode of the recovery step, as the recovery mode dictates the temperature of the bed before a cooling step is started. During the recovery step, a driving force is required for the sublimation of the solid CO2 deposited during the capture step. This driving force could be in the form of heat, which was proposed for flue gas treatment. Pure CO2 is fed to the system and the heat stored in the initial zone of the bed during the capture step is used to heat up the rest of the bed and evaporate CO2. The result of this way of operation is that the bed temperature will be at least -78◦C before starting the cooling step. In chapter 7 it was shown for biogas treatment that it is possible to provide the driving force for sublimation in another way: by reducing the pressure. This reduction is in the first place attained by letting the total pressure decrease, but at the same time feeding a CO2 lean gas (such as air) to the bed. Due to the large driving force, CO2 will evaporate and is removed very fast by the air flow. A coinciding advantage of this way of recovery is that the sublimation heat is returned to the packing material. Therefore, cooling of the bed to the required initial temperature is for a large extent covered by the evaporation of CO2, causing the cooling costs to be minimal. This way of recovery is difficult for flue gas treatment. In the first place be- 8.1 Important aspects for future development 151

cause the capture step cannot be performed at elevated pressures, due to the large amounts of flue gas and related high compression costs. An- other difference with biogas treatment is that in flue gas treatment the captured CO2 is the product, while for biogas treatment the gas passing through the bed (CH4) is the product. When CO2 capture is not required for biogas treatment air can be used to recover the bed. The recovery step is not only having effect on the required cooling duty, but is also for a large extent responsible for the electricity costs due to the recycle blower. The gas flow required during the recovery step is five to six times as large as the flue gas flow. Especially for power plant scale, these flows will be large and even a small increase in pressure drop will have a substantial effect on the process economics, as explained in Chapter 6. It could be very advantageous to develop an alternative recovery mode in future work. An alternative option was discussed in Chapter 2, in which the deposited CO2 was molten to the liquid phase by blocking a vessel after the capture step and introducing heat by internal heat pipes. It is expected that increased capital costs will not balance the energy savings. However, it can be recommended to carry out an accurate economic evaluation of this concept. Another alternative could be to operate the process using moving beds, in which the packing materials will be circulated. After CO2 has been deposited onto the packing, it could be removed from the bed and trans- ported to a separate vessel where CO2 could be removed (possibly under pressure) to finally transport the particles to a cooling step. In this way a continuously operated process could be obtained, at the expense of operational complications related to the handling of moving solids. Radial temperature profiles were observed during experiments with the pilot setup, as described in Chapter 5. These radial profiles could be explained by the heat transfer in and to the vessel wall, an effect which is expected to play a minor role for large scale beds, where the volume of the vessel wall is small compared to the bed volume. However, it could very well be that on large scale operation still radial profiles may be important in the beds. To minimize pressure drops over the packing, shallow packed beds were chosen in the economic analysis in Chapter 6, having a diameter larger than the length of the beds. Gas distribution becomes an important aspect for such kind of beds. A perfectly distributed flow will result in a possibly unacceptably high pressure drop. It might very well be that a certain amount of maldistribution of the feed has to be tolerated in order to minimize the pressure drop. The effects on the front 152 Epilogue and outlook development is unknown. It could be that due to flow maldistribution, not all parts of the bed will have the same temperature after a cooling step, before starting the capture step. The consequence could be that the frost fronts will move faster through the zone in which the temperature is lower, and early breakthrough of CO2 might be the result. That will cause an inefficient use of the cold stored in the packing, as the bed should be switched to a recovery step as soon as breakthrough is observed. How- ever, it could also be that heat transfer within the bed will level out the temperature differences. Another option is that the zones in which more CO2 is deposited will cause an additional pressure drop, and gas will be automatically redirected to zones in which no CO2 is deposited yet. To get a better insight in the possible consequences, the model should be extended to two dimensions and the Navier-Stokes equations should be incorporated to calculate the velocity profiles within the bed.

8.2 Future of the proposed concept

Scaling up a new technology such as discussed in this thesis directly from laboratory scale to refinery or power plant scale is not realistic. There are still several aspects to be studied in more detail in future work, as discussed in the previous section. Biogas purification is identified as a very interesting application of the developed technology. The smaller scale compared to flue gas treatment and therefore the lower risks, could make biogas treatment an ideal intermediate scale up step before going to flue gas scale. It is concluded in the techno-economic evaluation in Chapter 6 that for flue gas treatment the concept could be economically viable when a LNG regassification terminal is available to provide the required cooling duty at low cost. If this is not the case, refrigeration is required, which will become expensive and other capture technologies might be preferred. Therefore, the technology presented in this dissertation is not providing the single solution to bring down greenhouse gas emissions, but could of- fer a promising option in some situations. At this stage no capture technology is clearly outperforming competing technologies. The preferred technology will likely depend on the local conditions, such as the LNG availability for the cryogenic concept or low steam costs for scrubbing. Therefore it is necessary to progress research on all capture technologies, including cryogenic CO2 capture. Due to the very high degree of CO2 re- 8.2 Future of the proposed concept 153

moval possible with the cryogenic process, an interesting option could be to used the cryogenic capture process as a second stage for deep CO2 removal after a ﬁrst capture stage based on scrubbing to remove about 90%. However, the progress in CCS (Carbon Capture and Storage) development is taking place at a slower pace than required. This is mainly related to the limited public acceptance of CCS, which is again related to the uncertainties in safety risks or at least the inability to judge the signiﬁcance of these risks. The successfulness of CCS may therefore be much more dependent on social and political aspects, in stead of technology developments. 154 Bibliography Bibliography

D. Aaron and C. Tsouris. Separation of CO2 from ﬂue gas: A review. Separation Science and Technology, 40(1-3):321–348, 2005.

N. Abatzoglou and S. Boivin. A review of biogas puriﬁcation processes. Biofuels, Bioproducts and Bioreﬁning, 3(1):42–71, 2009.

M.R.M. Abu-Zahra, J.P.M. Niederer, P.H.M. Feron, and G.F. Versteeg. CO2 capture from power plants. part ii. a parametric study of the economical performance based on monoethanolamine. International Journal of Greenhouse Gas Control, 1(2):135–142, 2007a. M.R.M. Abu-Zahra, L.H.J. Schneiders, J.P.M. Niederer, P.H.M. Feron, and G.F. Versteeg. CO2 capture from power plants. part i. a parametric study of the technical performance based on monoethanolamine. Inter- national Journal of Greenhouse Gas Control, 1(1):37–46, 2007b.

U.E. Aronu, H.F. Svendsen, and K.A. Hoff. Investigation of amine amino acid salts for carbon dioxide absorption. International Journal of Green- house Gas Control, 4(5):771–775, 2010.

S. Basu, A.L. Khan, A. Cano-Odena, C. Liu, and I.F.J. Vankelecom. Membrane-based technologies for biogas separations. Chemical Soci- ety Reviews, 39(2):750–768, 2010.

R. Bredesen, K. Jordal, and O. Bolland. High-temperature membranes in power generation with CO2 capture. Chemical Engineering and Process- ing: Process Intensiﬁcation, 43(9):1129–1158, 2004.

L. Brennan and P. Owende. Biofuels from microalgae - a review of technologies for production, processing, and extractions of biofuels and co-

155 156 Bibliography

products. Renewable and Sustainable Energy Reviews, 14(2):557–577, 2010.

D. Britt, H. Furukawa, B. Wang, T.G. Glover, and O.M. Yaghi. Highly efﬁ- cient separation of carbon dioxide by a metal-organic framework replete with open metal sites. Proceedings of the National Academy of Sciences of the United States of America, 106(49):20637–20640, 2009.

H.S. Carslaw and J.C. Jaeger. Conduction of Heat in Solids. Clarendon Press, Oxford, second edition, 1952.

H.M. Chang and J.L. Smith. Mass transfer of super saturated contaminants in cryogenic helium heat exchangers. International journal of heat and mass transfer, 33(6):1299–1306, 1990.

H.M. Chang, M.J. Chung, and S.B. Park. Cryogenic heat-exchanger design for freeze-out removal of carbon dioxide from landﬁll gas. Journal of Thermal Science and Technology, 4(3):362–371, 2009.

D. Clodic and M. Younes. A new method for CO2 capture: frosting CO2 at atmospheric pressure. In 6th International Conference on Greenhouse Gas control Technologies, GHGT6 (Kyoto, Japan), pages 155–160, 2002.

D. Clodic and M. Younes. CO2 capture by anti-sublimation - thermo- economic process evaluation. In Fourth Annual Conference on Carbon Capture & Sequestration (Alexandria, USA), 2005.

A. Cybulski and J.A. Moulijn. Monoliths in heterogeneous catalysis. Catalysis Reviews - Science and Engineering, 36(2):179–270, 1994.

V. Darde, K. Thomsen, W.J.M. van Well, and E.H. Stenby. Chilled ammonia process for CO2 capture. International Journal of Greenhouse Gas Control, 4(2):131–136, 2010.

T.E. Daubert and R.P. Danner. Data compilation tables of properties of pure compounds. American Institute of Chemical Engineers, New York, 1985.

A.D. Ebner and J.A. Ritter. State-of-the-art adsorption and membrane separation processes for carbon dioxide production from carbon dioxide emitting industries. Separation Science and Technology, 44(6):1273– 1421, 2009. Bibliography 157

M.F. Edwards and J.F. Richardson. Gas dispersion in packed beds. Chemical Engineering Science, 23(2):109–123, 1968.

B. Ertl, C. Durr, D. Coyle, I. Mohammed, and S. Huang. New lng receiving terminal concepts. In World Petroleum Congress Proceedings, volume 2006, 2006.

I.A.A.C. Esteves, M.S.S. Lopes, P.M.C. Nunes, and J.P.B. Mota. Adsorp- tion of natural gas and biogas components on activated carbon. Sepa- ration and Puriﬁcation Technology, 62(2):281–296, 2008.

E. Favre. Carbon dioxide recovery from post-combustion processes: Can gas permeation membranes compete with absorption? Journal of Mem- brane Science, 294(1-2):50–59, 2007.

J.D. Figueroa, T. Fout, S. Plasynski, H. McIlvried, and R.D. Srivastava. Advances in CO2 capture technology - the u.s. department of energy’s carbon sequestration program. International Journal of Greenhouse Gas Control, 2(1):9–20, 2008.

E.N. Fuller, P.D. Schettler, and J.C. Giddings. New method for prediction of binary gas-phase diffusion coefﬁcients. Industrial and Engineering Chemistry Research, 58(5):18–27, 1966.

C.A. Grande and A.E. Rodrigues. Biogas to fuel by vacuum pressure swing adsorption i. behavior of equilibrium and kinetic-based adsorbents. Industrial and Engineering Chemistry Research, 46(13):4595– 4605, 2007.

C.A. Grande and A.E. Rodrigues. Electric swing adsorption for CO2 removal from ﬂue gases. International Journal of Greenhouse Gas Control, 2(2):194–202, 2008.

D.J. Gunn. Transfer of heat or mass to particles in ﬁxed and ﬂuidised beds. International Journal of Heat and Mass Transfer, 21(4):467–476, 1978.

D.J. Gunn and M.M.A. Misbah. Bayesian estimation of heat transport parameters in ﬁxed beds. International Journal of Heat and Mass Transfer, 36(8):2209–2221, 1993. 158 Bibliography

R.D. Hawthorn. Afterburner catalysts effects of heat and mass transfer between gas and catalyst surface [recent advances in air pollution control]. Amer. Inst. Chem. Eng. Symposium Series, 70(137):428–438, 1974.

IPCC. IPCC Special Report of Carbon Dioxide Capture and Storage [Metz, B., Davidson, O., De Coninck, H.C., Loos, M. and Meyer, L.A. (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 2005.

IPCC. Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmen- tal Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M.Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 2007.

M. Ishida and H. Jin. A new advanced power-generation system using chemical-looping combustion. Energy, 19(4):415–422, 1994.

A. John and S. Robertson. Lng: World: Strong growth forecast for global lng expenditure. Petroleum Review, 62(732):34–35, 2008.

A. Kapoor and R.T. Yang. Kinetic separation of methane-carbon dioxide mixture by adsorption on molecular sieve carbon. Chemical Engineering Science, 44(8):1723–1733, 1989. M.-B. Kim, Y.-S. Bae, D.-K. Choi, and C.-H. Lee. Kinetic separation of landﬁll gas by a two-bed pressure swing adsorption process packed with carbon molecular sieve: Nonisothermal operation. Industrial and Engineering Chemistry Research, 45(14):5050–5058, 2006.

D. Ko, R. Siriwardane, and L.T. Biegler. Optimization of pressure swing adsorption and fractionated vacuum pressure swing adsorption processes for CO2 capture. Industrial and Engineering Chemistry Research, 44(21):8084–8094, 2005.

H.M. Kvamsdal, K. Jordal, and O. Bolland. A quantitative comparison of gas turbine cycles with CO2 capture. Energy, 32(1):10–24, 2007. Bibliography 159

T.T. Le and M.A. Trebble. Measurement of carbon dioxide freezing in mixtures of methane, ethane, and nitrogen in the solid-vapor equilibrium region. Journal of Chemical and Engineering Data, 52(3):683–686, 2007.

W. Leitner. Carbon dioxide as a raw material: The synthesis of formic acid and its derivatives from CO2. Angewandte Chemie (International Edition in English), 34(20):2207–2221, 1995.

H.P. Loh and J. Lyons. Process equipment cost estimation. Technical report, DOE, 2002. V. Manovic, J.-P. Charland, J. Blamey, P.S. Fennell, D.Y. Lu, and E.J. An- thony. Inﬂuence of calcination conditions on carrying capacity of cao- based sorbent in CO2 looping cycles. Fuel, 88(10):1893–1900, 2009. McKinsey. Carbon capture & storage: Assessing the economics. Techni- cal report, McKinsey Climate Change Initiative, 2008.

J. Merel,´ M. Clausse, and F. Meunier. Carbon dioxide capture by indirect thermal swing adsorption using 13x zeolite. Environmental Progress, 25(4):327–333, 2006.

T.C. Merkel, H. Lin, X. Wei, and R. Baker. Power plant post-combustion carbon dioxide capture: An opportunity for membranes. Journal of Membrane Science, 359(1-2):126–139, 2010.

S. Noorman, M. van Sint Annaland, and J.A.M. Kuipers. Packed bed reactor technology for chemical-looping combustion. Industrial and En- gineering Chemistry Research, 46(12):4212–4220, 2007.

R. Notz, N. Asprion, I. Clausen, and H. Hasse. Selection and pilot plant tests of new absorbents for post-combustion carbon dioxide capture. Chemical Engineering Research and Design, 85(4 A):510–515, 2007.

R.J. Notz, I. Tonnies,¨ N. McCann, G. Scheffknecht, and H. Hasse. CO2 capture for fossil fuel-ﬁred power plants. Chemical Engineering and Technology, 34(2):163–172, 2011. cited By (since 1996) 0.

A.N. Ogunbameru, P.L.T. Brian, and R.C. Reid. On carbon dioxide frost formation. Industrial and Engineering Chemistry, 12(3):385–387, 1973. 160 Bibliography

T. Patterson, S. Esteves, R. Dinsdale, and A. Guwy. An evaluation of the policy and techno-economic factors affecting the potential for biogas upgrading for transport fuel use in the uk. Energy Policy, 39:18061816, 2011.

C.E. Powell and G.G. Qiao. Polymeric CO2/N2 gas separation membranes for the capture of carbon dioxide from power plant ﬂue gases. Journal of Membrane Science, 279(1-2):1–49, 2006.

R.C. Reid, J.M. Prausnitz, and B.E Poling. The properties of gases and liquids. McGraw-Hill, Inc., New York, fourth edition, 1987. S.R. Reijerkerk, R. Jordana, K. Nijmeijer, and M. Wessling. Highly hy- drophilic, rubbery membranes for CO2 capture and dehydration of ﬂue gas. International Journal of Greenhouse Gas Control, 5(1):26–36, 2011.

R.P. Ribeiro, T.P. Sauer, F.V. Lopes, R.F. Moreira, C.A. Grande, and A.E. Rodrigues. Adsorption of CO2, CH4, and N2 in activated carbon hon- eycomb monolith. Journal of Chemical and Engineering Data, 53(10): 2311–2317, 2008.

G.T. Rochelle. Amine scrubbing for CO2 capture. Science, 325(5948): 1652–1654, 2009.

E.S. Rubin. Integrated environmental control model. Technical report, Center for Energy and Environmental Studies, Carnegie Mellon Uni- versity, Pittsburgh, 2010.

M. Schach, R. Schneider, H. Schramm, and J. Repke. Techno-economic analysis of postcombustion processes for the capture of carbon dioxide from power plant ﬂue gas. Industrial and Engineering Chemistry Research, 49(5):2363–2370, 2010.

W.D. Seider, J.D. Seader, and Lewin D.R. Product & Process Design Prin- ciples. New York, John Wiley & Sons., second edition, 2004.

W. Seifritz. CO2 disposal by means of silicates. Nature, 345(6275):486, 1990.

V. Sejian, R. Lal, J. Lakritz, and T. Ezeji. Measurement and prediction of enteric methane emission. International Journal of Biometeorology, 55 (1):1–16, 2011. Bibliography 161

V.N. Shchelkunov, N.Z. Rudenko, Yu.V. Shostak, and V.I. Dolganin. Sur- face desublimation of carbon dioxide from binary gas mixtures. Journal of Engineering Physics, 51(6):1432–1435, 1986.

Shell. Growing tomatoes with reﬁnery CO2, 2011. http://www. shell.com/home/content/innovation/bright_ideas/growing_ tomatoes/. J. Smit, M. van Sint Annaland, and J.A.M. Kuipers. Grid adaptation with weno schemes for non-uniform grids to solve convection-dominated partial differential equations. Chemical Engineering Science, 60(10): 2609–2619, 2005. N. Stern. The Economics of Climate Change. the Stern Review. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 2007.

F. Su, C. Lu, S.-C. Kuo, and W. Zeng. Adsorption of CO2 on amine- functionalized y-type zeolites. Energy and Fuels, 24(2):1441–1448, 2010. R. Taylor and R. Krishna. Multicomponent Mass Transfer. John Wiley & Sons, New York, NY, USA, 1993. T.P. Tiemersma. Integrated autothermal reactor concepts for combined oxidative coupling and reforming of methane. PhD thesis, University of Twente, The Netherlands, 2010. K.D. Timmerhaus and T.M. Flynn. Cryogenic Process Engineering. Plenum Press, New York, NY, USA, 1989. U.S. Department of Energy. International Energy Outlook 2010. U.S. Energy Information Administration, Ofﬁce of Integrated Analysis and Forecasting, U.S. Department of Energy, 2010. J. van Straelen, F. Geuzebroek, N. Goodchild, G. Protopapas, and L. Ma- hony. CO2 capture for reﬁneries, a practical approach. International Journal of Greenhouse Gas Control, 4(2):316–320, 2010. D. Vortmeyer and R. Berninger. 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 Journal, 28(3): 508–510, 1982. 162 Bibliography

D. Vortmeyer and R.J. Schaefer. Equivalence of one- and two-phase models for heat transfer processes in packed beds: one dimensional theory. Chemical Engineering Science, 29(2):485–491, 1974.

J.G.M. Winkelman, O.K. Voorwinde, M. Ottens, A.A.C.M. Beenackers, and L.P.B.M. Janssen. Kinetics and chemical equilibrium of the hydra- tion of formaldehyde. Chemical Engineering Science, 57(19):4067–4076, 2002.

P. Zehner and E.U. Schlunder.¨ Thermal conductivity of granular materials at moderate temperatures. Chemie-Ingenieur-Technik, 42:933–941, 1970. List of publications

Journal papers

1. M.J. Tuinier, M. van Sint Annaland, G.J. Kramer, and J.A.M. Kuipers. Cryogenic CO2 capture using dynamically operated packed beds. Chemical Engineering Science, 65(1):114-119, 2010.

2. M.J. Tuinier, M. van Sint Annaland, and J.A.M. Kuipers. A novel process for cryogenic CO2 capture using dynamically operated packed beds - An experimental and numerical study. International Journal of Greenhouse Gas Control, 5(4):694-701, 2011.

3. M.J. Tuinier, H.P. Hamers, and M. van Sint Annaland. Techno- economic evaluation of cryogenic CO2 capture - A comparison with absorption and membrane technology. International Journal of Greenhouse Gas Control, In press, doi:10.1016/j.ijggc.2011.08.013, 2011.

4. M.J. Tuinier, and M. van Sint Annaland. Biogas Puriﬁcation using Cryogenic Packed Beds. Submitted to Industrial and Engineering Chemistry Research, 2011.

5. M.J. Tuinier, and M. van Sint Annaland. Desublimation rates of CO2 onto cryogenically cooled surfaces. Prepared for submission to International Journal of Heat and Mass Transfer. 6. M.J. Tuinier, and M. van Sint Annaland. Experimental proof- of-concept for cryogenic CO2 capture using dynamically operated packed beds. To be prepared for submission.

163 164 List of publications

7. M.J. Tuinier, and M. van Sint Annaland. A short-cut model to evaluate the conceptual feasibility for the cryogenic separation of contaminants from a gaseous stream using dynamically operated packed beds. To be prepared for submission.

Patents

1. G.J.B. Assink, G.J. Kramer, M. van Sint Annaland, and M.J. Tuinier. Process for the separation of CO2 from a gaseous feed stream, patent WO/2009/047341.

2. M. van Sint Annaland, and M.J. Tuinier. Process for the separation of contaminant or mixture of contaminants from a CH4 comprising gaseous feed stream. European patent application No. 10188933.5, ﬁled 26th of October 2010 (unpublished). Curriculum Vitae

Martin Tuinier was born on the 10th of August in Wijhe, The Nether- lands. He grew up in Deventer, where he attended elementary school and obtained his pre-university diploma in 2000 from the Alexander Hegius Lyceum. In September 2000 he started studying Chemical Engineering at the University of Twente in Enschede, The Netherlands. He special- ized in Process Engineering and carried out an internship for Akzo Nobel Polymer Chemicals in Ningbo, China. During this internship, he worked on the optimization of a crystallization section of an organic peroxide production plant. He graduated in May 2007 in the group ‘Fundamentals of Chemical Reaction Engineering’ on the ‘Kinetic aspects of Oxidative Cou- pling of Methane on a Mn/Na2WO4/SiO2 Catalyst’. After obtaining his Master degree, he started a PhD project in the same group, under the supervision of prof.dr.ir. Martin van Sint Annaland. In September 2010, the project was continued at the Technical University of Eindhoven, The Netherlands. The results of this project are described in this dissertation.

In September 2011 Martin started working as a process engineer in the Chemical Reaction Technology group of Evonik Industries A.G. in Marl, Germany.

165 166 Curriculum Vitae Dankwoord

Promoveren of de industrie in. Een lastige keuze, die niet alleen ik, maar meerdere collega’s hebben moeten maken aan het einde van hun studie. Eindelijk je kennis toepassen in het bedrijfsleven, of je toch nog vier jaar lang verdiepen in een bepaald onderwerp? De keuze werd in mijn geval makkelijker (of was het nou juist moeilijker?) gemaakt door de interes- sante opdracht die me werd aangeboden door Martin van Sint Annaland: het ontwikkelen van een nieuw proces concept voor het cryogeen afvan- gen van CO2. Hier had ik wel oren naar en dacht: ik grijp deze kans. En zonder enige spijt. Met het schrijven van dit dankwoord sluit ik een leerzame, nuttige en ook leuke periode af. Dat heb ik aan vele mensen te danken, waarvan ik er een aantal in het bijzonder wil bedanken. Martin van Sint Annaland wil ik in de eerste plaats hartelijk bedanken voor zijn vertrouwen in mij en het aanbieden van deze promotie opdracht. Zijn voorverkennende werk bij Shell was erg waardevol, waardoor we al snel met de eerste resultaten naar buiten konden komen. Tijdens het promotie traject heeft Martin me subliem geholpen met deskundig advies en als het nodig was met motiverende woorden. Martin, heel erg bedankt! Daarnaast wil ik Hans Kuipers bedanken, die vooral initieel nauw betrokken was bij het project. Het bewaken van de grote lijn was een belangrijke bijdrage. Vanuit Shell is Gert Jan Kramer betrokken geweest bij het project. Ik ben erg dankbaar voor zijn input tijdens de project besprekingen, welke zonder uitzondering nuttig en inspirerend waren. Ook Jonathan Barsema en Marijke Hogenbirk van Shell wil ik hartelijk danken voor het beoorde- len van ons werk op eventueel patenteerbare resultaten en het vervolgens vrijgeven ervan. Experimenteel werk besloeg een aanzienlijk deel van mijn promotie- werk. De daarom erg belangrijke technische ondersteuning van Johan

167 168 Dankwoord

Agterhorst en Robert Meijer was van hoge kwaliteit. Ik vond dat we een sterk team vormden en waardeer het zeer dat jullie ook tijdens de onzekere periode rond de verhuizing van Twente naar Eindhoven hard zijn blijven doorwerken aan mijn opstellingen. Ook ben ik erg dankbaar voor de hulp van Erik Analbers, Wim Leppink en Gerrit Schorfhaar bij het snel oplossen van problemen met de opstellingen. Ik heb het geluk gehad een aantal sterke studenten te hebben mo- gen begeleiden, die een goede bijdrage hebben geleverd aan de resultaten beschreven in dit proefschrift. Nhi Dang, thank you very much for the help during the design of the kinetic setup and carrying out the ﬁrst exploring experiments. Niels Hietberg ben ik dankbaar voor het verder optimaliseren van deze opstelling en voor het uitvoeren van de vele ex- perimenten. Jeroen Zijp heeft een wezenlijke bijdrage geleverd aan het ontwerp van de pilot plant: Jeroen, bedankt! Paul Hamers heeft erg goed werk verricht in het kader van de techno-economische evaluatie, waar ik hem erg dankbaar voor ben. De secretariele¨ ondersteuning van Nicole Haitjema in Twente en Ju- dith Wagters in Eindhoven was altijd goed geregeld, bedankt daarvoor! Ik wil ook graag Quint Segers en Paul Hamers alvast bedanken voor het bijstaan tijdens de verdediging als mijn paranimfen. Tenslotte, wil ik alle collega’s van FCRE/SMR bedanken voor de goede samenwerking en gezellige pauzes, borrels en uitjes!

-Martin Tuinier