UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Screening adsorbents for a layered adsorbent bed for Hydrogen separation using breakthrough experiments Kovacevic, Stevo B. 2000

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2000-0221.pdf [ 4.89MB ]
Metadata
JSON: 831-1.0058626.json
JSON-LD: 831-1.0058626-ld.json
RDF/XML (Pretty): 831-1.0058626-rdf.xml
RDF/JSON: 831-1.0058626-rdf.json
Turtle: 831-1.0058626-turtle.txt
N-Triples: 831-1.0058626-rdf-ntriples.txt
Original Record: 831-1.0058626-source.json
Full Text
831-1.0058626-fulltext.txt
Citation
831-1.0058626.ris

Full Text

SCREENING ADSORBENTS FOR A LAYERED ADSORBENT BED FOR HYDROGEN SEPARATION USING BREAKTHROUGH EXPERIMENTS by STEVO B. KOVACEVIC B.A.Sc. The University of Novi Sad, 1978 A THESIS SUBMITED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF CHEMICAL ENGINEERING We accepted this thesis as conforming to the^required standard THE UNIVERSITY OF BRITISH COLUMBIA April, 2000 © Stevo B. Kovacevic, 2000 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. The University of British Columbia Vancouver, Canada Department DE-6 (2/88) Abstract The aim of the present study was to apply the breakthrough experiment to screen and select adsorbents used for gas purification. Layered adsorbent beds employed in the Pressure Swing Adsorption (PSA) process for hydrogen purification from Steam Methane Reforming and Partial Oxidation product gasses use activated carbon and 5A molecular sieve in the first and second layer, respectively. Few studies on the use of alternative adsorbents have been reported and the present study was aimed at investigating new adsorbents and generating the adsorption data required to assess the adsorbents suitability for hydrogen purification using PSA. The adsorption of C O 2 , CO, C H 4 and N2 on 7 adsorbents, activated alumina AA-300, CBV 780 zeolite, PCB activated carbon, MHSZ-177 zeolite, 13X, 5A, and VSA6 molecular sieves were investigated using the breakthrough experiment. The first four adsorbents (AA, CBV780, PCB carbon, and MHSZ-177) were considered as candidates for the first adsorbent layer, which is typically an activated carbon. The last three (13X, 5A and VSA6) were candidates for the second adsorbent layer which is typically a 5A molecular sieve The first and second moments, height of equivalent theoretical plates, HETP, axial dispersion coefficient, D L , Henry's equilibrium constant, and the adsorbtion isotherm for each adsorbent/sorbate system were extracted from the breakthrough data. These data were used to select the most promising adsorbents for the first and second adsorbent layer. The MHSZ-177 and VSA6, respectively, were found to be the most ii promising candidates for the first and second adsorbent layer of adsorbent beds. The VSA6 had the sharpest mass transfer zone, MTZ and the longest retention time for all light sorbates and therefore was the most promising candidate for the second adsorbent layer. The MHSZ177 had the highest retention time for CO2 and despite having a broader MTZ than activated carbon it had the higher selectivities for CO2 with regard to light sorbates. This put MHSZ-117 in front of activated carbon as the choice for the first layer. The values for D L calculated from the averaged breakthrough data were in very good agreement with the published values. iii Table of Contents Abstract 11 List of Tables v i i i List of Figures X 1 Nomenclature x v Acknowledgements x v m Chapter 1: Introduction 1 1.1 Background 1 1.2 Motivation for the study 5 1.3 Objectives of the Present Study 6 Chapter 2: Literature Review 7 2.1 Adsorption 2.2 Pressure Swing Adsorption • 10 2.2.1 Packed Bed 12 iv 2.2.2 Layered bed 15 2.2.3 New Adsorbents 18 2.3 Experimental Methods in Gas Adsorption Studies 20 2.3.1 Breakthrough Curve 20 2.3.2 Equilibrium and Kinetic Analysis of Adsorbate Breakthrough 26 2.3.2.1 Moment Analysis 26 2.3.2.2 Calculation of the Adsorption Equilibrium Constant K (Henry Constant) 29 2.3.2.3 Determination of the Axial Dispersion Coefficient D L and the Lumped Mass Transfer Coefficient 30 2.3.2.4 Mathematical Model for an Adsorption Column - Determination of Equilibrium Isotherm from the Experimental Breakthrough Curve 31 2.3.2.5 Correlation Between HETP and the Two Moments, p. and a 2 36 Chapter 3: Experimental Methods 40 3.1 Experimental Apparatus and Method 40 3.1.1 Apparatus 40 v 3.1.1.1 Design of the Adsorbent Column 42 3.1.1.2 Adsorbent Bed Activation 43 3.1.2 General Procedure 43 3.2 Adsorbents 45 3.2.1 Adsorbents Used in This Study 45 3.2.1.1 AA-300 Activated Alumina 46 3.2.1.2 PCB Activated Carbon 46 3.2.1.3 MHSZ-177 47 3.2.1.4 VSA6 Molecular Sieve 47 3.2.1.5 13X Molecular Sieve 48 3.2.1.6 5A Molecular Sieve 48 3.2.1.7 CBV 780 Zeolite 49 3.2.2 Adsorbent Selection Factors 49 Chapter 4: Results and Analysis 52 4. 1 Determination of the System Response (Dead Volume) 52 4.2. Modified Reynolds Number 53 vi 4.3 Breakthrough Curve Experiments 54 4.3.1 Breakthrough Curves (BT) of N2, CH4, CO, and C0 2 54 4.3.2 Comparison of BT Curves 56 4.4 Moments and HETP 65 4.5 Adsorption Equilibrium Constant K (Henry Constant) 84 4.6 D L and the Lumped Mass Transfer Coefficient 87 4.7 The Determination of the Langmuir Adsorption Isotherms 92 4.8 Summary of the Experimental Results 95 Chapter 5: Conclusions and Recommendations for Future Work 96 5.1 Conclusions 96 5.2 Recommendations and Future Work 99 Literature 100 Appendix A: Experimental Adsorption Breakthrough Curves 107 Appendix B: Equilibrium Adsorption Isotherms 118 List of Tables Table 1. Absorbent Characteristics 45 Table 2. First and Second Moments (System Response) 53 Table 3: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO, CH4, N2 and CO2 on PCB activated carbon adsorbent 69 Table 4: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO2 on AA300 activated alumina adsorbent 70 Table 5: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO, CH4, N 2 and CO2 on MHSZ 177 high silica zeolite adsorbent 70 Table 6: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO, CH4, N 2 and C0 2 on CBV 780 Y -zeolite adsorbent 71 Table 7: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO, CH4, N2 and CO2 on 13X 16/40 mesh molecular sieve adsorbent 72 Table 8: First, second moments and HETP calculated from the adsorption viii breakthrough curves obtained for CO, CH4, N 2 and C0 2 on 13X 8/12 mesh molecular sieve adsorbent 73 Table 9: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO, CH4, N 2 and C0 2 on VSA6 Molecular Sieve adsorbent 74 Table 10: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO, CH4, N 2 and C0 2 on 5A molecular sieve adsorbent 75 Table 11: First, second moments and HETP calculated from the desorption breakthrough curves obtained for N 2 on 13X 16/40 mesh molecular sieve adsorbent 76 Table 12: First, second moments and HETP calculated from the desorption breakthrough curves obtained for N 2 and CO on 13X 8/12 mesh molecular sieve adsorbent 76 Table 13: First, second moments and HETP calculated from the desorption breakthrough curves obtained for CO and CH 4 on VSA6 16/40 mesh molecular sieve adsorbent 77 Table 14: First, second moments and HETP calculated from the desorption breakthrough curves obtained for C0 2 on MHSZ-177 high silica adsorbent 77 ix Table 15: First, second moments and HETP calculated from the desorption breakthrough curves obtained for CO, CH4 and N2 on 5A 8/12 mesh molecular sieve adsorbent 78 Table 16: First, second moments and HETP calculated from the desorption breakthrough curves obtained for CO2 on Activated Alumina AA-300 adsorbent 79 Table 17: Calculated values of adsorption equilibrium constants K 85 Table 18: Summary of equilibrium adsorption constant K and equilibrium selectivities for CO2 and CH4 on various adsorbents from other studies .87 Table 19: Axial dispersion coefficients, DL, and the lumped mass transfer resistance coefficients, LMTC, calculated from the first and second moments 90 Table 20: Langmouir Isotherm Constants, b and qs 93 x L i s t o f F igures Figure 1. Principle of Operation of A Fuel Cell 2 Figure 2. Adsorption Isotherms 9 Figure 3. The Basic Pressure Swing Adsorption Process 11 Figure 4: Schematics of a Single Adsorbent and a Layered Fixed Bed 14 Figure 5. Idealized Fixed Bed Breakthrough Curve and Wave Front 21 Figure 6. Breakthrough Curve for a (a) Narrow Mass-Transfer Zone and (b) Wide Mass Transfer Zone 22 Figure 7. Relationship Between Adsorption Fronts and Loading Curves as a Function of Residence Times 24 Figure 8. Determination of Henry Constant 30 Figure 9. Determination of Axial Dispersion and the Bulk Mass Transfer Resistance 31 Figure 10. Graphical Evaluation of Isotherm Constants 35 Figure 11. HETP Variation with Flow Velocity According to van Demeter Equation 39 Figure 12. Experimental Set-up for Breakthrough Measurements 40 xi Figure 13. Breakthrough curves for CO on VSA6 for 65, 80, 120, 160, 200, 240, 280 and 320 cc/min flows, respectively from left to right 55 Figure 14. Breakthrough curves for N2, CH4 and CO in helium on 5A 59 Figure 15. Breakthrough curves for N2, CH4 and CO in helium on VSA6 MS 60 Figure 16. Breakthrough curves for N2, CH4 and CO in helium on 13X 8/12 MS ...60 Figure 17. Breakthrough curves for N2, CH4 and CO, in helium on 13X 16/40 61 Figure 18. Breakthrough curves for N2, CO, CH 4 and C0 2 in helium on CVB-780..61 Figure 19. Breakthrough curves for N2, CO, CH4 and C02 in helium on PCB Activated Carbon 62 Figure 20. Breakthrough curves for N2, CO, CH 4 and C0 2 in helium on MHSZ-177 high silica zeolite 62 Figure 21. Breakthrough curves for CH4 in helium on 13X, 5A and VSA6 63 Figure 22. Breakthrough curves for CO in helium on 13X 8/12, 13X 16/40, 5A, and VSA6 63 Figure 23. Breakthrough curves for N2 in helium on 13X, VSA6, and 5A 64 Figure 24. Breakthrough curves for C02 in helium on CBV780, AA 300, PCB Activated Carbon, MHSZ177, VSA6, 13X 8/12, 13X 16/40 and 5A MS 64 Figure 25. Breakthrough curves for C0 2 in helium on CBV780, AA 300, PCB xii Activated Carbon and MHSZ177 65 Figure 26. Variation of HETP with superficial gas velocity (v) for CO, CH4, and N 2 on 5A molecular sieve 79 Figure 27. Variation of HETP with superficial gas velocity (v) for CO, CH4, and N 2 on VSA6 molecular sieve 80 Figure 28. Variation of HETP with superficial gas velocity (v) for CO, CH4, and N 2 on 13X 16/40 and 13X 8/12 mesh molecular sieve 80 Figure 29. Breakthrough curves of CH 4 on the PCB carbon obtained experimentally and by calculation using Equation 39 81 Figure 30. Breakthrough curves of N 2 on the VSA6 adsorbent obtained experimentally and by calculation using Equation 39 81 Figure 31. Breakthrough curves of CO on the VSA6 adsorbent obtained experimentally and by calculation using Equation 39 82 Figure 32. Breakthrough curves of CH4 on the PCB carbon obtained experimentally and by calculation using Equation 40 82 Figure 33. Breakthrough curves of N 2 on the VSA6 adsorbent obtained experimentally and by calculation using Equation 40 83 Figure 34. Breakthrough curves of CO on the VSA6 adsorbent obtained experimentally and by calculation using Equation 40 83 Figure 35. Determination of adsorption equilibrium constants for N2, CO and CH 4 on xm VSA6 molecular sieve 85 Figure 36. Determination of D L and LMTC for N 2 in 5A and 13X 16/40; for CO in PCB carbon; and C H 4 in 5 A molecular sieve (from averaged data) 91 Figure 37. Determination of D L and LMTC for N 2 in 13X 8/12 mesh and 13X 16/40 mesh (from averaged data) 91 Figure 38. Langmuir Isotherms for C0 2 on VSA6, CBV780, PVB Carbon, MHSZ177, AA300, 5A and 13X calculated from b and qs obtained from Table 17 94 xiv Nomenclature A cross section area of bed b Langmuir adsorption equilibrium constant c sorbate concentration in the gas phase co initial sorbate concentration in the gas phase; final (t > 0) steady state value of c Dc intracrystalline diffusivity D L axial dispersion coefficient D m molecular diffusivity Dp pore diffusivity dp particle diameter F gas flow rate HETP Height of Equivalent Theoretical Plate K Adsorption equilibrium constant, dimensionless Henry's law adsorption equilibrium constant defined as a concentration ratio based on particle volume K c K for zeolite crystal k overall effective mass transfer coefficient xv kf external fluid film mass transfer coefficient L adsorbent bed length 1 length of mixing element N number of theoretical plates PH, P L high and low pressure for P S A system Q gas flow rate q sorbate concentration in the solid phase qo final value of q q value of q averaged over crystal and pallet qs saturation limit q* equilibrium value of q r c crystal or microparticle radius Rep Modified Reynolds Number Rp adsorbent particle (pallet) radius t time t mean retention time u, v superficial gas velocity xvi Vf, V s volume of liquid or solid w wave velocity z distance measured from column inlet 8 voidage of adsorbent bed 6 P porosity of the adsorbent particle yi,Y2 constants in Equation 36a u, first moment Pf gas viscosity v interstitial velocity of fluid p gas density CT standard deviation of pulse or step response CT second moment (variance) Acknowledgement I would like to express my sincere appreciation to Dr. Kevin Smith for his critical evaluation, invaluable guidance, and support throughout my research. I would like to acknowledge Highquest Engineering Inc., and especially Bowie Keefer for his encouragement and support, particularly at the beginning of this work. I am grateful for the help and valuable suggestions rendered by my friends, May Chew and Zoran Pavlovic. Above all I am grateful to my wife Snezana and my daughters Senka and Sanja for their help, encouragement, and infinite patience which made the completion of this work possible. xviii Chapter 1: Introduction 1.1 Background The rapid growth of the world's population, the growth in industrial activity, coupled with the phenomenal increase in the number of automobiles, has made air pollution an increasingly serious problem. Exhaust gases from automotive engines emit over 100 million tons of pollutants into the atmosphere every year. Hence, the long-term burden on the environment (through the large-scale release of carbon dioxide) has increased rapidly and significantly. These concerns for the environment have inspired scientists to search for alternative fuel sources. Hydrogen, as a motor fuel, could be the answer to many environmental problems because: • There are no harmful end use emissions • Hydrogen is a potentially totally renewable resource • There is no danger to the world's climate through the use of H2. Current research in hydrogen utilization is focused on technologies that will most directly facilitate the progression to a hydrogen energy economy. In the next century, many vehicles and aircraft may conceivably be powered by hydrogen. Depending on the source of hydrogen and its intended use, hydrogen production usually requires a number of preparation and purification steps. Fuel cells, as shown in Figure 1, are increasingly being accepted as the power source that 1 will most likely replace the internal combustion engine in the next generation of zero-emission or ultra-low emission transportation vehicles (Appleby and Foulkes, 1989; 1998). They are clean, quiet and efficient. To switch from the internal combustion engine to the fuel cell, hydrogen is needed as a fuel. At present, very pure hydrogen is required, although depending on the type of fuel cell that will prevail, the required purity may change. Load H 2 -Puel — $oz Oxidant Porous Electrolyte p D rous Anode Cathode Anode: H 2 2 H + + 2e Cathode: I 0 2 + 2 H + + 2e - • H 2 0 2 Overall Cell Reaction: H 2 + I o 2 _ * H 2 0 2 Figure 1. Principle of Operation o f F u e l Cell 2 Tremendous amounts of highly purified hydrogen will be required in the future to support a hydrogen-based economy if fuel cells displace the internal combustion engine. 8.7 m of hydrogen at 25°C, and 1 atm per 100 km, is comparable to a vehicle which consumes 8.7 L of gasoline per 100 km (CESHR, 1996). The automotive industry is on the verge of a major new era if the majority of new cars were to be fuelled by hydrogen. US projections call for 10,000 hydrogen-powered fuel cell cars by 2004; the goal is to eventually replace all internal combustion engines with fuel cell engines. Hydrogen is found in nature only in compound form. It must therefore, first be released through the use of energy before it can be made available for industrial or energy purposes. Currently, most hydrogen is produced by the steam reforming of methane or methanol (SMR) (Astanovsky et al., 1994), resulting in approximate end products of 74% H 2 , 24% C0 2 , and less then 1% of CO and CH, (Amphlett et al., 1993; Fujimoto et al. 1987). However, two technological problems remain in the SMR production of hydrogen: a) obtaining pure hydrogen from SMR gas, and b) the management of impurities, which are mostly C0 2 (a major green house gas), CO and CH 4 . If all impurities are released into the atmosphere, overall global pollution would not change. At best, the pollution source would be shifted out of populated urban areas. The proton-exchange-membrane (PEM) fuel cell used by Ballard Power Systems, is at present the most promising fuel cell technology being developed. However, the PEM fuel 3 cell can only use hydrogen with less then 10 ppm carbon monoxide (CO), as higher concentrations of impurities eventually poison the platinum catalyst membrane (Appleby and Foulkes, 1998). SMR gas can be purified to the level required in P E M fuel cells by the pressure swing adsorption process (PSA), which may eventually be the double-edged solution for the two main problems associated with SMR-produced hydrogen. By its nature, the PSA process, on the one hand, provides very pure hydrogen, and on the other hand, it concentrates the impurities. If there is a viable solution for containing the impurities by transforming them into useful, preferably solid or liquid products, it appears that PSA may provide an ideal solution for the future. To provide the vast amounts of hydrogen1 necessary for fuelling a great number of P E M fuel cell vehicles at an economical price, the present PSA technology needs to be improved. One of the ways to do this is to optimize the layered adsorbent bed typically used for hydrogen purification. This can be done not only by optimizing the length of the adsorbent layers in a conventional adsorbent bed, but also by trying to introduce new adsorbent(s) as a more efficient adsorbent pair capable of improving the efficacy of the PSA process. The optimal choice of adsorbents in a layered bed would minimize bed size and maximize bed utilization during the cyclic operation of the PSA process. Therefore, a better adsorbent choice improves the separation efficacy and economics of the PSA 1 lm 3 of H 2 at 25°C and 1 atm would be equivalent to 1 L of gasoline in a vehicle which consumes 8.7 L of gasoline per 100 km (CESHR, 1996). 4 process. In addition, it may lead to changes in the conditions of the PSA process (i.e. cycle speed), so that the output of the PSA process can be enhanced. 1.2 Motivation for the Study In recent years, many new synthetic adsorbents have been developed and used for particular applications. As adsorption is increasingly the method of choice for gas separation, one option in attempting to achieve improved separations is to apply these new adsorbents to completely new applications. This is more pronounced in the case of an adsorbent process where layered adsorbent beds are employed. The choice of adsorbents for each layer in this case is virtually endless. To be able to design an adsorber, which will perform better than a single adsorbent bed, or a conventional layered bed, equilibrium and/or kinetic adsorption data for the new adsorbents must be available. Usually, these data are not available unless the choice for the replacement is narrowed to the adsorbents which have been used for the same or similar applications. Most often, these alternative adsorbents have similar disadvantages (but to a lesser degree) as the adsorbent to be replaced. If one wants to go beyond this narrow range in an attempt to introduce new adsorbents that would more efficiently deal with a particular problem, the shortage of available data makes such an effort difficult. Consequently, if one were to rely solely on published references, predicting the suitability of unconventional adsorbents for use in completely new applications would be a discouraging endeavor. 5 Fortunately, a feasible alternative exists for predicting the suitability of an adsorbent for a particular application - the data can be obtained from the breakthrough data measured for the adsorbent and sorbates of interest. 1.3 Objectives of the Present Study The main objective of the present study is to utilize the breakthrough experiment to screen and select the most suitable pair of adsorbents to be used in a layered adsorbent bed in an attempt to improve the pressure swing adsorption process for hydrogen purification. Furthermore, the purpose of this research is to emphasize the importance of using a layered adsorbent bed and to encourage the search for a new set(s) of adsorbents that may be more suited for the given conditions of a chosen PSA process. These factors are becoming increasingly more important as more and more new adsorbents are developed. 6 Chapter 2: Literature Review 2.1 Adsorption The process of adsorption involves the separation of a substance from one phase (gas or liquid phase), accompanied by its accumulation or concentration at the surface of another (solid phase). The adsorbing phase is the adsorbent, and the substance adsorbed at the surface of that phase is the adsorbate (Walter et al., 1985). Adsorption at a surface or interface is largely the result of binding forces between the individual atoms, ions, or molecules of an adsorbate and the surface. Depending on the nature or strength of these forces, three types of adsorption are evident: 1. Ion exchange 2. chemical, and 3. physical. Ion exchange adsorption involves electrostatic attachment of ionic adsorbates to sites of opposite charge at the adsorbent surface, which are subsequently displaced by other ionic species of greater electrostatic affinity. Chemical adsorption involves a chemical reaction between an adsorbate and an adsorbent and is therefore irreversible, since it results in a change in the chemical form of the adsorbate (Walter et al., 1985; McKay, 1995). Almost all adsorptive separation process are based on physical adsorption (Ruthven, 7 1984; McKay, 1995) and physical adsorption is the focus of the present study. Physical adsorption (common for adsorption of a substance from the gas phase to the surface of the solid phase) results from the action of relatively weak van der Waals forces and electrostatic interactions comprising polarization, dipole, and quadrupole interactions. In any adsorption process, the contribution of the van der Waals forces is present, while the electrostatic contributions are significant only in the case of adsorbents which have an ionic structure, such as zeolites. Physical adsorption is characterized by low heats of adsorption (less than two or three times the latent heat of evaporation), which are only significant at relatively low temperatures. Physical adsorption from the gas phase is always exothermic, because the adsorbate molecule has fewer degrees of rotational freedom than molecules in the gas phase, meaning that the entropy change on adsorption must be negative (Ruthven, 1984). Furthermore, physical adsorption is rapid, reversible, non-specific, and does not involve the dissociation of adsorbed species nor electron transfer. However, polarization of the sorbate may occur (McKay, 1995). The adsorption of a sorbate from the gas phase to the surface of the solid adsorbent in a given system leads to a thermodynamically defined distribution of sorbate between the phases when the system reaches equilibrium (when no further adsorption occurs). The adsorption isotherm depicts this distribution by expressing the amount of sorbate adsorbed per unit weight (or volume) of adsorbent, q*, as a function of the residual equilibrium concentration, c, of sorbate remaining in the fluid phase at a constant temperature. 8 In general, three different types of isotherm can be distinguished depending on whether the equilibrium distribution is linear, favorable or unfavorable as depicted in Figure 2 (Ruthven, 1984; Cussler, 1997, Weber et al., 1985). Curves A, B and C depict the linear, favorable, and unfavorable adsorption isotherms, respectively. c Figure 2. Adsorption Isotherms A quantitative equilibrium distribution between phase concentrations is defined for each adsorbate/adsorbent system and a set of system conditions. The distribution character is influenced by the properties of the adsorbate, the adsorbent, and the system in which adsorption occurs. Practically, most isotherms are of the favorable type, which is a required condition for an efficient separation process, but which, is not by itself a sufficient condition as the time required to achieve equilibrium must also be considered 9 for the selection and design of a practical adsorption process (Weber et al., 1985). An isotherm that is favorable for adsorption is unfavorable for desorption and vice versa. The shape of the isotherm influences the way in which the concentration wavefront moves through the adsorbent bed. The more favorable the isotherm, the sharper the adsorption wavefront. 2.2 Pressure Swing Adsorption The pressure swing adsorption (PSA) process, like any other adsorption separation process, involves two principal steps: 1. Adsorption, during which the more readily adsorbed species are preferentially adsorbed from the feed; while species which are not strongly adsorbed pass through the bed of adsorbent. 2. Regeneration or Desorption, during which the strongly adsorbed species are removed from the adsorbent, thus making the adsorbent ready for use in the next cycle. The basic PSA system consists of two beds, which are alternatively pressurized and depressurized, according to a programmed sequence in a continuous process as shown in Figure 3 (Cussler, 1997). Large-scale separation or purification systems use more complex cycles involving three or more adsorbent beds to improve system efficiency (Fuderer and Rudelstorfer, 1976; Warmuzinski and Tanczyk, 1997). 10 Countercurrent purge ensures that the more strongly adsorbed components are pushed back towards the bed inlet. Consequently, this ensures that the raffmate product is not contaminated in the next cycle. Product purity increases as the purge is increased, but only to a certain point, after which the gain becomes minimal (Yang et al., 1997). Detrimental to the separation process are fluctuations in the local bed temperature, which rise during adsorption and drop during desorption. The best separation is achieved when the beds are operated under nearly isothermal conditions. This can be realized with short cycles and low throughput per cycle (Yang, 1987; Ruthven et al., 1993). STEP 1 STEP 2 STEP 3 STEP 4 Pressurization Loading Blowdown Purge l i f t T i m e Figure 3. The Basic Pressure Swing Adsorption Process 11 The adsorption step takes place at high cycle pressure, while the regeneration or desorption step occurs at reduced total pressure (Ruthven et al., 1993). Consequently, the pressure swings during adsorption and regeneration between high and low levels and hence the process is called Pressure Swing Adsorption. Although many of the essential features of the PSA process were known in 1930, the introduction of PSA as a viable industrial process is attributed to the Skarstrom process patented in 1958 (Skarstrom, 1960). However, the PSA process only gained large industrial acceptance in the 1980s (Ruthven et al., 1993). As progress continues in the development of the rapid PSA process, PSA may yet become a dominant gas separation process in the near future (Yang et al., 1997; Smith et al., 1991). 2.2.1 Packed Bed A packed bed is a cylindrical column packed with one granular adsorbent, which preferentially adsorbs certain component(s) of a gas mixture. The less strongly adsorbed gas components will pass through the bed. In this way, the less strongly adsorbed gas is purified as shown in Figure 4a. The use of a single-layer bed was widely considered even before the PSA process became the choice for gas separation, and until recently was the only choice for all PSA processes. Commonly, the single-layer beds are used for hydrogen separation from two component mixtures, or from less complex mixtures in one, two or multi-bed PSA processes as presented in the following studies. 12 Yang et al. (1997) studied a two-bed six-step PSA process using zeolite 5A for the bulk separation of hydrogen from H2/CO and H2/CH4 systems (70/30 vol. %) at ambient temperature and feed pressure of 11 atm. They found that the recovery of the product containing 99.99% H 2 was 75.87% in the H2/CO mixture and 80.38% in the H2/CH4 mixture. They also investigated the effects of different cycle step times, purge/feed ratio and adsorption pressure on the performance of the PSA process. They showed that the experimental data of a two-bed PSA cycle could be simulated by a mathematical model. Kumar (1994) studied the effects of changing the two basic properties of adsorbents (equilibrium and kinetics) on the performance of a PSA process by a mathematical model. A feed containing 25% CH4 and 75% H 2 at the pressure of 20 atm was separated by a four-bed PSA process to produce high purity H2. Kumar concluded that the best adsorbent could only be chosen based on a combination of both the adsorbent selectivity and working capacity. However, increasing the saturation capacity while keeping the selectivity unchanged leads to a better adsorbent, which can be achieved by using a binderless adsorbent. Malek and Farook (1997) developed a mathematical model of a six-bed, ten-step PSA process which resembles an industrial hydrogen recovery process from refinery fuel gas. The adsorbent used was activated carbon, while impurities in the hydrogen stream were methane, ethane, and propane. The traces of higher hydrocarbon contaminants and moisture were not taken into account. The model assumed nonisothermal PSA separation, the linear driving force approximation for particle uptake and the extended Langmuir 13 isotherm for adsorption equilibrium. The computer-controlled laboratory version of the actual industrial PSA unit was used to verify the model predictions. The simulation showed that product purity declines quickly with increase in cycle time and decreasing high operating pressure. The studies mentioned above as well as numerous other studies which researched hydrogen separation from different gas mixtures, commonly employed either 5A molecular sieve or activated carbon adsorbents. Packed Bed (a) Product 888888 UQlJ898 B88888 888888 'OQOQOC Feed Layered Bed (b) Figure 4: Schematics of a Single Adsorbent and a Layered Fixed Bed 14 2.2.2 Layered Bed A layered bed is a packed bed filled with two or more adsorbents, each consisting of its own layer, as depicted in Figure 4b. The concept of using more then one adsorbent in a PSA bed (beyond the need for a guard layer, D, to remove water at the inlet of the bed) is relatively new and relatively few papers have been published on layered beds in adsorption processes. Klein and Vermulen (1975) investigated the cyclic performance of layered beds for binary ion exchange to determine the best ratio of layer thickness (one having favorable and the other unfavorable isotherms), for optimal regenerant efficiency and resin utilization. Pigorini and LeVan (1997) applied the equilibrium theory to analyze the cyclic performance of an isothermal PSA bed composed of two adsorbent layers, one with a strongly unfavorable isotherm, and the other with a less favorable (more linear) isotherm. Their simple equilibrium model, that assumes no resistance to mass transfer, shows that in some cases a combination of two adsorbents, used in a layered bed, is to be preferred over the use of one adsorbent alone. Cheldini and Tondeur (1996), and Park et al. (1998) considered the adsorption of a three-component mixture (carried by an inert gas) in a layered PSA bed consisting of activated carbon, followed by zeolite. They discovered that each gas characteristic concentration pattern gets "refracted" when crossing the boundary between the two layers; the slope changes due to the different adsorptive capacities of the layers. This may complicate the 15 analysis of the system. Watson et al. (1995) proposed the use of a layered bed for oxygen separation in a vacuum PSA process to reduce the temperature gradient along the bed by introducing a weaker adsorbent at the inlet of the bed. The reason for doing this was the loss of selectivity of the stronger adsorbent at lower temperatures. They further found that the higher uptake of O2 at lower temperatures made it more difficult to efficiently regenerate the stronger adsorbent with the applied vacuum. Warmuzinski and Tanczyk (1997) presented a general model of a multicomponent PSA process for hydrogen separation over activated carbon and 5A molecular sieve packed in a two-layer adsorbent bed. They studied the operation of a large-scale PSA process for hydrogen purification as a test case. The highest hydrogen recovery obtained from the two PSA installations in series was 74% (which is also typical of commercial plants utilizing two-layered beds). They proved that the model provides a satisfactory qualitative description of the PSA process analyzed. Lee et al. (1999) and Yang and Lee (1998) used a layered bed packed with activated carbon and zeolite 5A, to study the effects of carbon-to-zeolite ratio on the H2 recovery from coke oven gas (COG) which contained 56.4% H2, 26.6% CH4, 8.4% CO, 5.5% N 2 and 3.1% C0 2. They found that the layered bed gave better purity than the single-layer bed, with activated carbon or 5A zeolite, at the same operating conditions. They developed a model under the assumption that the layered bed consisted of two independent beds with a single adsorbent in each. The mathematical model which incorporated mass, energy and momentum balances agreed well with the PSA 16 experimental data. Dong et al. (1999) proposed a more compact PSA process design for the separation of multicomponent mixtures ( C O 2 , CH4 and N2) by combining two or more adsorbents into a single column. 13X zeolite and carbon molecular sieve (CMS) adsorbent beads were packed in a two-layer bed, which formed the annulus around activated carbon beads (the annulus acted as a single-layer column within a two-layer bed). Assuming isothermal operation, the simulation, by a simple mathematical model, found that the efficiency of the new design compared favorably with the conventional process, which also employed these three adsorbents, although packed in a greater number of single-layer beds. The simulation also indicated that a new, more compact, design has advantages in comparison with a conventional PSA separation which is capable of recovering more than one component from multicomponent gas mixtures. A conclusion common amongst the referenced papers was that separation was strongly dependant on the ratio of the two layers in the adsorbent bed; consequently, the optimal design of the length of each layer is an important step for optimization of the PSA process. In general, the papers discussed in the previous paragraphs investigated either the general behavior of the layered bed, or the particular behavior of commonly used PSA layered beds to determine the optimal ratio of the two layers. Attempts to expand the layered bed approach further by exploring the possibility of employing different adsorbent(s), and perhaps unconventional ones for a given PSA process have not been reported. The layered bed is usually filled with a weaker adsorbent at its inlet, followed by a stronger one. For the purification of hydrogen from SMR gas, 17 the first layer of a layered bed would be configured to adsorb the strongly adsorptive component (CO2), and the second layer following would remove the light components (CO, CH4 and N2). This arrangement maximizes the use of each adsorbent, minimizes the adsorbent bed size, and improves the efficacy of separation. The optimal length for each layer of absorbent for a given gas-solid separation system depends on the feed composition and feed velocity. In the case of H2 separation from reformer off-gas, the conventional layered bed consists of activated carbon and 5A molecular sieve (Warmuzinski and Tanczyk, 1997; Yang, 1987). The higher the concentration of the lighter component (CO) to be removed (via the 5A molecular sieve), the higher the required length of this layer (Park et al., 1998). As feed velocity increases, the optimum length ratio increases (as expressed by the ratio of carbon bed length to total bed length). If the layered bed consists of three layers, the first layer is usually a desiccant (activated alumina or silica gel), which acts as a guard bed to remove moisture (Yang, 1997; Werner and Mersmann, 1994). 2.2.3 New Adsorbents As discussed in the two previous sections, most studies of hydrogen separation have dealt with activated carbon and 5A molecular sieve as adsorbents in a single-layer or two-layer adsorbent bed. Only a few studies on the screening of larger numbers of adsorbents were reported. They 18 were done in an attempt to solve certain separation problems by introducing the PSA separation process as a possible solution, or to determine some equilibrium parameters for a particular adsorbent/sorbate system as explained below. Choudhary and Mayadevi (1993a,b) studied the adsorption of methane, ethane, ethylene and carbon dioxide on NaX, NaY, HY, H-ZSM-5, Na-ZSM-5, H-ZSM-8, Na-ZSM-8 zeolites, Silicalite and ALPO-5 adsorbents at 303-473 K, using the gas chromatography pulse technique. The purpose was to screen a large number of adsorbents in an attempt to explore the possibility of separating C2 hydrocarbons from methane. They found that the heat of adsorption of methane was the lowest, and that the heat of ethylene was the highest on all adsorbents tested. Furthermore, they observed that the relative retention volume at 323 K » 1, indicating that the separation of methane from the other sorbates is possible. Although they found that the chromatography pulse technique was a good and fast tool for screening a large number of adsorbents, it is fair to say that the method, as applied in their study, is only good for an initial preliminary screening of adsorbent suitability for a particular application. Dunne et al. (1996) measured isosteric heats of adsorption and adsorption isotherms simultaneously in a calorimeter for a series of sorbates (Ar, O2, N2, C H 4 , C 2 H 6 , S F 6 , and CO2) . The adsorbates, as listed, have an increasing size and magnitude of quadrupole moment. Various adsorbents (NaX, H-ZSM-5, and Na-ZSM-5) were examined in this study which highlighted the importance of sorbate size and polarity as well as adsorbent heterogenicity in determining adsorption equilibrium parameters (heat of adsorption and adsorption isotherms). They found that for quadrupolar molecules such as C O 2 and N2 the 19 effect of ion type on the heat of adsorption is significant. However, as far as this author is aware, there are no studies reported on screening novel adsorbents to be used for hydrogen separation from SMR or partial oxidation of methane (POX) gas streams. 2.3 Experimental Methods in Gas Adsorption Studies 2.3.1 Breakthrough Curve The breakthrough curve is the adsorption pattern representing the response of an initially clean adsorbent bed to a step-change in inlet concentration. The shape of this pattern, which constitutes a concentration wave front, depends on the shape of the adsorption isotherm for the given adsorbent/adsorbate system, among other factors, as discussed in later sections (Yang, 1987). In an ideal single-adsorbate plug flow system with infinite contact time, there would be no contaminant in the effluent until all of the contaminant broke through the bed at once, and all of the adsorbent was exhausted or saturated. In reality, what is observed is a wave front or a mass-transfer zone (MTZ), which is the length of gradual change in concentration through a portion of the bed based on the continuous adsorption of the adsorbates. As this wave front appears, it can be monitored and plotted to generate the breakthrough curve, which can be used to determine the useful life of the bed. The typical relationship between the passage of adsorbate through the bed and the 20 development of the breakthrough curve is shown in Figure 5 (Fox et al., 1985; McKay, 1995). C, fl o a ;-, fl o fl o 0 MTZ C Y c Y C o mm P M T Z B r e a k t h r o u g h MTZ T i m e Figure 5. Idealized Fixed Bed Breakthrough Curve and Wave Front E x h a u s t i o n The length of the wave front (or mass-transfer zone) is an important design parameter and is usually influenced by contact time (Frederick and Bernardin, 1985). If the mass-transfer zone is narrow relative to the bed length, the breakthrough curve will be steep, as in Figure 6 (a), and most of the capacity of the solid phase: will be utilized at the break point. A narrow mass-transfer, zone is desirable to make efficient use of the: adsorbent and; to reduce the energy cost associated with regeneration. In the ideal case of no mass-transfer resistance and no axial, dispersion, the mass-transfer zone would be of tint r 21 infinitesimal width, and the breakthrough curve would be a vertical line from 0 to 1 (a step function) when al l o f the solid would be saturated; Time , t (b) Figure 6. Breakthrough Curve for a (a) Narrow Mass-Transfer Zone and (b) Wide Mass-Transfer Zone (where tB, ts and p are the breakthrough, elution and mean retention times, respectively); The width of the mass-transfer zone depends on the mass-transfer rate, the flow rate, and the shape o f the equilibrium curve: To predict the: concentration profiles and zone width, a lengthy computation is often required, and the results may be inaccurate due to uncertainties in the mass-transfer correlations: The results from laboratory-breakthrough 22 tests using a small-diameter bed, filled with particles of the same size and tested at the same superficial velocity as that of a full-scale unit, are usually used for scale-up. Factors that affect the actual shape of the breakthrough curve include (Kovach, 1988): • equilibrium conditions (which affect the shape of the isotherm) • rate-limiting mechanism for adsorption • adsorbate concentration • particle size of the adsorbent • depth of column or bed • flow velocity In general, for a given system, the breakthrough time is increased by: • decreased concentration of adsorbate in solution • decreased flow rate • increased bed depth • decreased adsorbent particle size The shape of the adsorption front, and hence the breakthrough curve, is influenced by the nature of the equilibrium adsorption isotherm. The more favorable the isotherm, the sharper the adsorption front. Since most adsorption processes are not run at equilibrium, the dynamics of adsorption 23 (the adsorption rate) is the primary factor affecting adsorption fronts and breakthrough curves (Fox et al., 1985). Three different adsorption; fronts and the. corresponding loading curves for a solute at different residence times are compared: in Figure 7 (Fox et al., 1985). Each show differences that: will' have a notable effect upon;the conceptual design, sizing, operating cycle selection and so forth. Time (a) (b) Figure 7. Relationship Between Adsorption Fronts and Loading Curves as a. Function of Residence Times Curves A, B, and C in Figure 7a depict the adsorption front within the bed, while curves; A, B, and C of Figure 7b depict the corresponding breakthrough curves at the outlet of the bed. The adsorption front A. has the sharpest MTZ (Figure 7a) and also the longest 24 residence time. The adsorption front B has much broader MTZ and shorter residence time, while front C breaks through the column almost instantly. It can be seen that the shape of the breakthrough curves adequately describe the mass transfer front in the bed. Since the breakthrough curve is an easily measured parameter, it can be used to describe the mass transfer front within the bed. The magnitude of the equilibrium adsorption process is expressed as an isotherm relating the concentration of solute on the adsorbent to a given concentration of solute in the solution. In designing a batch adsorption process, the isotherm predicts the adsorbent behavior at equilibrium. In the design of a flow-through system, the equilibrium isotherm relates such important system properties as selectivity, capacity and throughput per given quantity of adsorbent. However, commercial adsorption systems are seldom operated at equilibrium. In most cases, the design and performance of adsorption systems is limited by the rate of adsorption, which is a molecular diffusion process controlled by either film-diffusion, or particle diffusion. Most adsorption processes are particle diffusion controlled, where the rate of adsorption is governed by the transport (diffusion) of the solute molecules within the pores of the adsorbent particles. Thus, the size of a solute molecule, the shape and volume of the adsorbent pores, and the radius of the adsorbent particle are all important system properties since they strongly influence diffusion within the adsorbent particle. In addition, the most economically critical parameters in the conceptual design of the adsorption mode are usually optimized on a kinetic rather than an equilibrium basis. In general, kinetics have a great deal to say about how an adsorbent is best used. As a 25 consequence, the greatest effort in the conceptual design of an adsorbent system has gone toward dealing with the economic consequences of particle-diffusion-controlled adsorption rates (Fox et al., 1985). 2.3.2 Equilibrium and Kinetic Analysis of Adsorbate Breakthrough 2.3.2.1 Moment Analysis The theory of statistical moments provides a precise way of characterizing chromatographic response peaks and breakthrough curves of any shape. The expressions for the moments may be derived from the solution of the model equations to either a step or pulse input. The solutions are derived from model equations written in Laplace form using van der Laan's theorem (van der Laan, 1958). The moments of the experimental response may be calculated directly by integration (Karger and Ruthven, 1992): Response to Pulse input oo First moment: o Equation 1 oo Jc(f)-<# 0 \c(t)-(t-ii)-dt Second moment: Equation 2 \c(t)dt 0 26 Response to Step input First moment: u= t QO = J(l - c I c«)dt o Equation 3 Second moment: CT2= 2 J(l - c I c»)tdt - u 2 Equation 4 where C/CQ is the concentration of the sorbate measured at the exit of the adsorber at time t, after injection of the sorbate in the column. This indicates that moments include system dead volumes which must also be taken into account. Useful general forms for the first and second moments of the step response for a biporous adsorbent were obtained by Haynes and Sarma (1973): u=/=-£V 1 + S J K Equation 5 where K=sp+( 1 -sp)Kc, and a2 DL+ EV {RP R2 2p} vL L(l-e) 2 \ 3kf \5epDp \5KD 1+-(l-e)Kj -2 Equation 6 For a strongly adsorbed species in a gaseous system, K is often large, so s/(l-e)K is small and Equation 6 simplifies to: 27 r2 DL v ' - ^ + R2 A 3fe- i5*A, i s m ; Equation 7 i 1 w axial film macropore micropore As shown above, the contributions of axial dispersion and the external film, macropore, and micropore diffusional resistances are linearly additive. For the simple Linear Driving Force (LDF) rate model (Equation 8): dq_ dt = k(q*-q) Equation 8 the corresponding expression for the second moment is: DL vf e ) \ + • 2ju2 vL L\\-eJ kK , 1 + , V (l-e)KJ -2 Equation 9 For large K, the expression [1+ (e/l-s)KJ2 can be neglected, and Equation 9 becomes: DL vf e 1 1 2//2 vL L\\-eJkK Equation 10 In order to match the second moments we must set: kK ~ 3k/ + l5spDP + \5KDc Equation 11 28 This relationship provides the extension of the Glueckauf approximation to a system in which more then one diffusional resistance is significant (Ruthven, 1986). For gaseous systems the LDF approximation works very well, unless the column is too short or when the mass transfer resistance is very high (Karger and Ruthven, 1992). 2.3.2.2 Calculation of the Adsorption Equilibrium Constant K (Henry's Constant) In matching an experimental response of the breakthrough curve, the adsorption equilibrium constant, K, may be easily determined from the first moment, p., (Equation 5). If sv=F/A, after rearrangement, Equation 5 becomes: where /, L, A, s, K and F represent respectively, mean retention time, bed length, cross section area of bed, bed voidage, Henry's equilibrium constant and gas flow rate. Since the plot of the corrected mean ju (p^Psorbate-P-sys) versus 1/F is approximately linear, X" may be determined from the slope, S, where S=LA[l+(l-s)/e)K]. Equation 12 29 S=LA{l+[(l-e)/e]K} 1 1/F (sec*cm") Figure 8. Determination of Henry's Constant 2.3.2.3 Determination of the Axial Dispersion Coefficient D L and the Lumped Mass Transfer Coefficient Matching the second moment provides one equation containing in effect two unknown parameters - the axial dispersion coefficient D L and the lumped mass transfer coefficient (LMTC). Rearranging Equation 8 yields: cr2 L „ 1 ( s V RP Rl rl ^ = DL — + 2//v ~ V U - * ' • + + -Iks \5sPDP \5KDc. Equation 13 Within a low Reynolds number regime, a plot of (c^/2/J)(L/v) versus 1/v2 should be linear, with slope DL and intercept corresponding to the lumped mass transfer resistance (Ruthven etal., 1984): 30 Equation 14 1/v2 Figure 9. Determination of Axial Dispersion and the Bulk Mass Transfer Resistance The parameters of axial dispersion and bulk mass transfer resistance are determined through measurements of breakthrough curves conducted over a range of experimental conditions. For such applications, the easiest parameter to vary experimentally is usually the fluid velocity. 2.3.2.4 Mathematical Model for an Adsorption Column - Determination of Equilibrium Isotherm from the Experimental Breakthrough Curve Mathematical models are needed to better analyze and interpret experimental data. The parameters characterizing the kinetics and equilibrium of the separation process can then be derived by matching the experimentally obtained response to the model prediction. 31 Generally, gas flow through a packed column can be represented by the axial dispersed plug flow model. Consequently, the differential fluid phase mass balance on an element of the column can be represented by (Karger and Ruthven, 1992): tfc d , . dc (\-e\dq rt . _ - D L ^ + —(vz) + — + Hr = 0 Equation 15 dz dz dt V s J dt For an isothermal system with negligible pressure drop and a low concentration of absorbable species, the gas velocity through the column can be considered to be constant; Equation 15 then simplifies to: d1 c dc dc (\-s\dq — T + V — + — + — dz oz dt V e J dt D i — r v  v  = 0 Equation 16 Equation 15 is generally used for modeling the dynamic response of an adsorption column. The term dq/dt represents the local mass transfer rate averaged over an adsorbent particle. To obtain the dynamic response of the system, c=c(z,t), it is necessary to solve Equation 16 simultaneously with the appropriate mass transfer rate expression: — = f(c, q) Equation 17 dt which is subject to initial and boundary conditions imposed on the system. For step and pulse perturbations, these conditions are defined as follows: step: c(z.0)=q(z,0)=0, c(0,t)=c0 pulse: c(z.0)=q(z,0)=0, c(0,t)=c0 8(t) 32 Assuming the long column condition, c = 0 • \z —>oc The response to a perturbation in the flow composition involves a mass transfer zone or concentration front which propagates through the column with characteristic velocity determined by the equilibrium isotherm. If the isotherm is favorable for adsorption it will be unfavorable for desorption, and for a sufficiently long column (or system with low flow velocity) the composition should approach equilibrium at the column outlet. A desorption curve measured under these conditions can therefore give the complete isotherm. For isothermal conditions and plug flow of the trace component in an inert carrier stream, Equation 16 reduces to (Ruthven, 1984): Equation 18 The general form of the adsorption isotherm at equilibrium can be written as: q = q = f(c) Equation 19 Assuming mass transfer equilibrium, the rate of adsorption becomes: f - \ Equation 20 33 Substitution of Equation 20 into Equation 18 gives: (dc) + Vdt) + l-e) dq dc 8c dt Equation 21 After rearranging and applying a cyclic rule (Ruthven, 1984), Equation 21 becomes: '5zN 1 + \-e\dq - ( f ) . © . -dt = w(c) Equation 22 s J dc where w(c) represents the velocity of the concentration front through the column. Assuming that the Langmuir equation is a good fit for the experimental data: q = ^ H \ + bc Equation 23 whose derivative, with respect to concentration is: dq _ bqs dc (l + bc)2 Equation 24 and substituting the above into an equation for velocity (Equation 22) and then integrating, the following expression is obtained: M \q0 J b a s z vt-z \-e V £ -1 Equation 25 Rearranging the Langmuir equation and evaluating it at c=co gives: 34 qs _ 1 + bco q0 bc0 Equation 26 and substituting it into Equation 25 yields: f 1 ^ KbcJ bqxz (\-s vt-z V £ -1 Equation 27 or: c cn bcn ( \ - E V S y/vt-; 1 Equation 28 Plotting c/c0 versus 1/4vt-z , the parameters of the Langmuir isotherm (b, qs) may be found. The second term in Equation 28 is the intercept /, and the term in square brackets is the slope S as shown in Figure 10: I 1 s 1 yjvt-Z Figure 10. Graphical Evaluation of Isotherm Constants 35 I = -bc. Equation 29 o S = ^-.bqA bc0 V q.s = bclS2 z 1-eJ e \ Equation 30 This approach may be a convenient way of obtaining a single component isotherm. 2.3.2.5 Correlation between HETP and the two moments, ju and cr2 An alternative approach to describing the fundamental behavior of an adsorption column (in terms of differential mass balance on an element of the column, as described above) is one analogous to the "tanks in series" model for a non-ideal flow reactor, which was originally developed by Martin and Synge (1941). Using this approach, the efficiency of the system is expressed by a finite number of theoretical well-mixed stages (plates), N, or the height equivalent to a theoretical plate, HETP. The number of theoretical plates represents a direct measure of axial dispersion and mass transfer resistance in the system. For an ideal mixing cell in which mass transfer occurs between the fluid and adsorbed phase, the transient mass balance for both phases yields (Ruthven, 1984): dc dq svco = eve + Vf — + Vs —— dt dt Equation 31 36 {\-£)l^ = k{q-q) Equation 32 From the Laplace transforms of Equations 31 and 32, the expression for the transfer function is obtained as (Ruthven, 1984): Co 1+-sl 1 + \-e K \ + {s/k) Equation 33 where K= q*/c. For N identical stages in series: Co 1+-sl 1 + l-e K \ + {s/k) Equation 34 Substituting Equation 34 into the expressions for the first and second moments given by Equations 3 and 4 yields the following: |_ + _L(_ £ ^ 1 2// 2N NlU-eJkKV (l-e)K 1 + Equation 35 Providing that N=L/l=vL/2DL, it is evident that Equations 35 and 9 are identical. If N=2ju2/o2, the number of theoretical plates the adsorption column is equivalent to, then HETP is equal to L/N and can be expressed as: 37 CJ2 . DL f £ ^ HETP = — T L = 1 — + 2v| 1 £ 1 + --2 2/i2 v Kl-eJkKV (l-£)KJ Equation 36 Substituting DL=yiDm+y22Rpv, Equation 36A, to Equation 36 provides an expression which resembles the general form of the van Deemeter equation; (van Deemeter, 1956; Ruthven, 1984): HETP = — + B+Cv Equation 37 v where, A=2yiDm, B=2y2Rp and C»2[e/(l-£)]/kK, assuming K»10. Equations 36 and 37 show very clearly that the various contributions to the total plate height are additive. Term A describes molecular diffusion, term B eddy diffusion, while term C includes mass transfer resistance associated with the external fluid film layer, macropore diffusion, and intracrystalline diffusion. Figure 11 shows the dependence of HETP on gas velocity according to the van Deemeter equation (Thaeron, 1996; Yang, 1987). The "R" curve is the resultant of the contributions of terms A, B and C. 38 V Figure 11. HETP Variation with Flow Velocity According to van Demeter Equation. The minimum in curve "R" represents the smallest plate height (and therefore the largest number of theoretical plates). This minimum also represents the smallest standard deviation. 39 Chapter 3: Experimental Methods 3.1 Experimental Apparatus and Method 3.1.1 Apparatus The breakthrough curves were measured for various adsorbents and adsorbates using an experimental set-up consisting of an adsorbent bed, two mass-flow-meter controllers (MFC), multigas controller, two 3-way valves, a thermal-conductivity detector (TCD) and a bubble flowmeter. The TCD detector was part of a Hewlett Packard 5710 A gas chromatograph. Figure 12 shows a schematic diagram of the experimental set-up for the breakthrough measurements. Sorbate He Figure 12. Experimental Set-up for Breakthrough Measurements 40 The helium carrier gas stream controlled by the mass-flow controller, MFC-2, (Brooks 5850E, 0-200 cc/min) introduces the carrier gas (or carrier gas + sorbate) into the system. The pure sorbate stream is first passed through the second mass-flow controller, MFC-1, (Brooks 5850E, 0-20 cc/min), which controls the sorbate stream before it enters the first three-way valve, which directs it towards the second 3-way valve, where it is mixed with the He carrier gas. This mixed solute stream is then introduced into the adsorbent bed where adsorption occurs. When the sorbate three-way valve switches sorbate flow to vent, only pure carrier gas is introduced and the adsorbent bed undergoes desorption. Effluent from the adsorbent bed is introduced into the sample side of the TCD detector where its composition is monitored and measured. Pure He passes through the reference side of the TCD at all times. Breakthrough experiments for a particular adsorbent and adsorbate were carried out for a range of gas flow-rates through the adsorbent bed. The gas flows were measured using a bubble flowmeter downstream of the TCD detector. When the effluent concentration, measured by the TCD detector, showed a steady signal equaling the step value of the initial sorbate concentration, the adsorption equilibrium was reached in the adsorbent bed, and the adsorption breakthrough curve was established. The sorbate flow was then stopped, or directed to vent. By switching the sorbate stream three-way valve to vent, only the pure carrier gas stream continued through the saturated adsorbent bed so that desorption occurred. Effluent concentrations were then monitored and desorption curves established. 41 3.1.1.1 Design of the Adsorbent Column The experimental adsorbent bed was constructed to satisfy the following constraints and requirements. • The diameter of the bed was to be no less then 10-20 times greater than the adsorbent particle size so as to avoid the influence of wall effect on the experimental results. • The bed length was to be long enough to allow the establishment of the constant pattern of concentration wavefront, or the breakthrough curve. • The overall bed volume must satisfy operating requirements for a range of flow rates (rates of superficial velocities), and at the same time be within the measurable limits of the flow rates of the two mass-flow-meter-controllers. • The breakthrough time of the strongest adsorbable species (CO2) was to be considered as this determined how much data could be collected on a data acquisition system with limited storage capacity. Consequently, data collection was a tradeoff between sampling rate and duration of data collection (assuming the number of variables were constant). In the end, the sampling rate was limited to four samples per second, which turned out to be sufficient, as the highest achieved superficial velocity was 4 cm/s. To satisfy the above mentioned requirements, the cylindrical adsorbent bed was made of 5/8" OD stainless steel tube, 20.5 cm in length. At each end of the bed, 1/16" thick 40 um sintered stainless steel filters were used to contain the adsorbent within the length of the adsorbent bed and to allow for uniform flow distribution. Swagelok compression fittings 42 at each end of the adsorbent bed held the filters in place. Upon assembly, all components were tightened and the entire unit leak-checked to ensure proper gas flow through the bed. 3.1.1.2 Adsorbent Bed Activation Newly-packed adsorbent beds were activated in an oven for 16 hours with 60 cc/min helium flow to purge the adsorbent of impurities. Activation temperatures maintained in the oven ranged from 180-300°C, depending on the adsorbent being activated. After activation, the bed was brought to room temperature and leak-checked. The adsorbents activated in this way were free of any impurities and could be used for the breakthrough tests. 3.1.2 General Procedure The experimental study was conducted with eight different adsorbents and four different sorbates (CO2, CO, CH4, and N2). The range of gas flows covered in this study was from 65 cc/min to 320 cc/min (see Table 2, Chapter 4), and the corresponding superficial gas velocities were 0.8 to 4 cm/sec. The concentration of sorbate in He was 5% v/v. This study of breakthrough curves for single components in an inert gas stream was undertaken with the assumption that these same components (adsorbates) would behave in the gas mixture as if they were single entities in the inert gas stream, and would not 43 show any significant interaction with other components. Other studies in the literature described in the following paragraphs have proven this assumption to be valid for mixtures of light sorbates. Ruthven et al. (1979), studied the kinetics and equilibrium sorption of a binary mixture of CH4 and N2 in 4A molecular sieve. They concluded that the dynamic response curve of the binary system could be satisfactorily interpreted by assuming that both components in the mixture diffused independently with the same intrinsic mobilities as single-component systems. Kumar et al. (1982) came to the same conclusion in their studies of mixtures of CF4 and iC\iHio on 4A molecular sieve, and CH4-C2H6 and C3H8-C2C4 on 5A molecular sieve. They found that this conclusion, to a first approximation, still held true for 5 A molecular sieve, where the measurements of sorbate concentrations extended well beyond the region of Henry's Law. Warner and Mersmann (1994) tested 10 models for the predictive calculation, of adsorption equilibria from single component isotherms of CO, CH4, N 2 and CO2 on molecular sieve 5 A and activated carbon. They found that the Ideal Adsorption Solution Theory and the Statistical Thermodynamic Model showed the best results. The other thermodynamic models showed comparably good results in predicting selectivity dependence of temperature or pressure. In view of the above considerations, similar conclusions can be made regarding C0 2 adsorption in that CO2 adsorption is not, and would not be affected by the presence of relatively small amounts of lighter components. Consequently, the light components would not be adsorbed for most of the length of the first layer as it is already saturated 44 with C0 2 In effect, this leaves a very short bed length available for the adsorption of the light components, which has negligible influence on the adsorption of CO2. Therefore, the study of CO2 as a single component in an inert carrier gas is a valid approach to obtaining insight into its adsorption behavior from a gas mixture with CO, CH4 and N2. 3.2 Adsorbents 3.2.1 Adsorbents Used in This Study The adsorbents used in this study were activated alumina AA 300, PCB activated carbon, CBV 780-X16 ultra stabile Y zeolite, MHZS-177 high silica zeolite, 13X molecular sieve, VSA6 molecular sieve, and 5A molecular sieve. Each adsorbent is briefly described in the following section and is summarized in Table 1. Table 1. Absorbent Characteristics Adsorbent Manufacturer Particle s ize Hydrophobicity C o m m en t s Activated Alumina AA-300 Alcan 8/14 mesh B -PCB Activated Carbon Calgon 12/30 mesh CBV 780-X16 Zeolist Internat. 1/16" extrudate + Ultra stable Y zeolite MHZS-177 UOP 1/16" extrudate + 13X Molecular Sieve UOP 16/40 mesh B -13X Molecular Sieve UOP 8/12 mesh B -VSA6 Molecular Sieve UOP 16/40 mesh B -5 A M S Davison 8/12 mesh B -45 3.2.1.1 AA-300 Activated Alumina AA-300 activated alumina is a desiccant grade adsorbent primarily used for drying air or other gas streams as it has very good adsorption capacity for water vapor. It has a high surface area of 350-380 m2/g, and an extremely good crush strength of 18-23 kg. In general, activated alumina is known to have a certain affinity for CO2. 3.2.1.2 PCB Activated Carbon PCB Activated Carbon is a coconut based activated carbon designed for gas phase separation. The surface area of PCB carbon is 1150-1250 m2/g. The activated carbon surface is basically nonpolar and this unique property gives activated carbon the following properties (Yang, 1987): • Separates and purifies without requiring prior stringent moisture removal (however, it can not be regarded as hydrophobic) • Adsorbs more nonpolar and weakly polar organic molecules than other adsorbents due to its large internal surface area • Lower heat of adsorption so that desorption (regeneration) therefore requires less energy • 10-25 Angstrom diameter pore size for the carbon used for gas separation (polimodal pore-size distribution). The structure may be described as having many small pores branching off from larger ones, which are open throughout the entire particle. The 46 smaller pores are regarded as adsorption pores, while the larger pores are called feeder or transport pores. • Used predominantly for gas separation, the removal of nonpolar gases and organic vapors, and H 2 purification. 3.2.1.3 MHSZ-177 MHSZ-177, a high silica molecular sieve adsorbent, is a hydrophobic adsorbent. Its capacity is not reduced by the presence of high relative humidity in the gas feed nor by the length of the mass transfer zone. It possesses a uniform pore structure, which can adsorb smaller molecules with critical diameters up to 6 nm, such as ethanol, acetone and methanol chloride. MHSZ-177 is commonly used for the removal of volatile organic compounds (VOC) from humid gas streams such as air or from aqueous solutions. It is thermally stable up to 800 °C (dry) and up to 500 °C (moist). 3.2.1.4 VSA6 Molecular Sieve VSA6 molecular sieve adsorbent was specifically developed by UOP for separating oxygen from air in a vacuum PSA process. It is a calcium exchanged adsorbent with a nominal pore size of 8 Angstroms. It is a hydrophilic adsorbent, which means that its capacity is drastically reduced by high relative humidity gas streams. 47 3.2.1.5 13X Molecular Sieve Type 13X molecular sieve is a sodium aluminosilicate of Type X crystal structure. With an effective pore opening of about 10 Angstroms, 13X molecular sieve will adsorb molecules with a kinetic diameter less than 10 Angstroms and exclude larger ones. A hydrophilic adsorbent, it is commonly used for the concurrent removal of H2O and CO2 from gas and air streams, separation of heavy hydrocarbons from natural gas and air stream, removal of H2S and mercaptans from liquid and hydrocarbon fractions, etc1. The 13X molecular sieve is a traditional adsorbent that is widely used. Two grain sizes of the adsorbent were tested in this study as listed in Table 1 so that the effect of particle size on mass transfer resistance could be examined. 3.2.1.6 5A Molecular Sieve The 5A molecular sieve is a calcium aluminosilicate having Type A crystal structure with an effective pore opening of about 5 Angstroms. A traditional adsorbent, 5A is used for the production of high purity nitrogen, oxygen, hydrogen and inert gases from air and other gas streams in PSA processes. 5A is also used for the removal of H2O, CO2, H2S and mercaptans from natural gas, the separation of normal paraffins from branched and cyclic hydrocarbons, and hydrogen separation (Yang, 1987). The 5A molecular sieve is also a hydrophilic adsorbent. 1 From the adsorbent manufacturer's brochure 48 3.2.1.7 CBV 780 Zeolite CBV 780 is a Type Y zeolite having a high SiOyAkOa mole ratio of 80, which classifies it as a hydrophobic adsorbent. It has a relatively large effective pore opening of 24 Angstroms, and a high surface area of 780 m2/g. Type Y zeolites are used by catalyst producers for many important catalytic processes. 3.2.2 Adsorbent Selection Factors The objective of the present study was to identify a new and more efficient pair of absorbents to replace the conventional adsorbents of a typical layered bed, used in H2 purification by PSA processes. Activated carbon and 5A molecular sieve adsorbents are commonly used in the layered adsorbent beds for hydrogen separation (Yang, 1987; Park et al., 1998; Kenney, 1984). However, the two are very different adsorbents, each having different roles in the purification of H2. Usually, relatively complex gas mixtures such as reformate gas or partial oxidation (POX) product gas are used as a feedstock. The mixtures typically contain varying proportions of H2, CO2, CH4, CO, N 2 and small amounts of water. Commercial plants utilize a layered bed which include a guard layer of activated alumina or silica gel (a separate bed in some cases), whose purpose is to remove moisture. This is followed by two adsorbent layers, the first being activated carbon, and the second being the 5A molecular sieve (Park et al., 1998; Yang and Lee, 1998; Lee et al., 1999). The role 49 of activated carbon is to retain CO2 within the length of that layer, thus preventing CO2 breakthrough to the 5A molecular sieve (the stronger adsorbent). In addition, the light impurities CO, CH 4 , and N 2 should not break through the entire bed (Chlendi and Tondeur, 1995). If a stronger adsorbent was employed as the first, or the only layer, it would adsorb CO2 so strongly that most of the length of that layer would be saturated with CO2, which would not be desorbed sufficiently by decreasing pressure in the regeneration step. Consequently, the remaining unsaturated portion of the 5A layer would not be sufficient to manage the lighter impurities. The entire PSA process would suffer in recovery and productivity, and have difficulty maintaining product purity. With the above in mind, the search for new adsorbents was bi-directional. Firstly, to find a weak adsorbent (weaker in regards to CO2 than the 5A molecular sieve) with different physical characteristics, and perhaps a slightly higher CO2 uptake and sharper mass transfer zone than the commonly used activated carbon in the first layer, which has to be replaced. Furthermore, the new adsorbent in the first layer has to be strong enough to effectively retain CO2 within the length of its own layer. In addition to CO2 removal, it was desirous that the replacement adsorbent also be able to remove moisture, or at least be insensitive to it. For this purpose, the MHZS-177 hydrophobic zeolite and CBV780 high silica Y zeolite (also hydrophobic) were chosen as potential replacements for activated carbon. In addition to these two hydrophobic adsorbents, AA300 activated alumina adsorbent (although commonly used as a guard layer) was also considered as a possible candidate 50 for the first layer, since it has certain C O 2 adsorption capabilities. PCB activated carbon, a coconut-based activated carbon with very good mechanical characteristics, was chosen for comparison purposes, as well as being a potential replacement for BPL activated carbon, the conventional choice for the first adsorbent layer. The second objective was to find a suitable replacement for the 5A molecular sieve, which is used to prevent CO, CH4 and N 2 from breaking into the H 2 product stream. Since the key to optimizing the layered beds for a PSA process lies in the adsorption step (Park et al., 1998), the search for a replacement for the 5A molecular sieve was based on assessing candidate adsorbents's adsorption capacity for CO and other lighter impurities (CH4 and N 2 ) . Since these impurities are usually present in smaller amounts, the more strongly they are adsorbed, the lower the risks of impurities breaking through the active bed. The potential replacement was VSA6, which was designed for oxygen separation from air since it has a high uptake for N 2 , and it was hoped that it may also have a high uptake for CO, and 13X, a very common and relatively inexpensive adsorbent. The traditional choice of 5A molecular sieve was tested for comparison purposes. 51 Chapter 4: Results and Analysis 4.1 Determination of the System Response (Dead Volume) Knowledge of the system dead volume is very important for obtaining reliable information regarding the mass transfer characteristics of the particular adsorbent/adsorbate system. Dead volume causes a broadening of the concentration wavefront and longer retention time. In order to evaluate the dead volume of the experimental set-up, or system response, the adsorbent bed was removed and the inlet and outlet of the bed were connected directly. The breakthrough curves were obtained from such a set-up for the range of flow rates used in the study. The first and second moments (retention time and variance) were calculated for each flow rate, and used to correct the data obtained when the beds, filled with the different adsorbents, were in place. This was done by subtracting the values of the system response, p.sys and a sys, (no adsorbent bed) from the values of the u. and a 2 calculated when the adsorbent bed was in place. Since the second moment, or variance, is very sensitive to the length of the tail of the breakthrough curve, curve fitting of the data was performed to average the moments. The averaged moment values are shown in Table 2. The dead volume of the thermal conductivity cell is unavoidable, so reducing the tubing length and the number of fittings was the only way to minimize the external dead volume. 52 Table 2. 1st and 2nd Moments (System Response) - Contribution of the Setup without the Adsorbent Bed Flow rate, Q _2 Sy cc/min s s2 65 8.44 21.230 80 7.12 11.468 120 5.40 2.777 160 4.83 0.555 200 4.27 0.222 240 3.85 0.103 280 3.54 0.056 320 3.38 0.039 4.2. Modified Reynolds Number The nature of the gas flow in the adsorbent column is determined by the magnitude of the modified Reynolds Number, Rep. The modified Reynolds number is defined by (Perry, 1984): Re = — . Equation 38 53 where dp, u, pg, p.f and e, represent the adsorbent particle diameter (or equivalent diameter), superficial velocity, the density and the viscosity of the gas which flows through the adsorbent bed, and the void spaces between particles as a fraction of the total volume of the bed, respectively. This dimensionless quantity, RePi is proportional to the ratio of inertial force to viscous force in a flow system. For the breakthrough curve experiments, the carrier gas used was helium, which contained sorbates (CO, CH4, N 2 and CO2) in very small concentrations. However, the gas is considered to be pure helium for the calculation of the modified Reynolds number. The bed vidage s was approximated to be 0.37, which is consistent with values obtained from manufacturers for some adsorbents of the same particle size. For the gas velocities and particle sizes of the different adsorbents in the adsorbent bed, the value of the modified Reynolds number ranged in value from 0.04 to 4.3, indicating laminar gas flow through the adsorbent column for all the ranges of flow rates and particle sizes applied in this study. These values are in the range of values used in most other studies. 4.3 Breakthrough Curve Experiments 4.3.1 Breakthrough Curves of N2, CH 4 , CO, and CO2 for each Adsorbent Breakthrough measurements on seven different adsorbents using a feed gas of 5% each of 54 CO, CH4, N2 and CO2, in helium, were conducted for the range of flow rates shown in Table 2. The bed pressures applied in this study were close to atmospheric pressure, and were the minimum required for achieving a particular flow. The pressure was in the range 102-106 kPa above atmospheric pressure for all experiments reported herein. The experiments were all done at a room temperature of 22 ±0.5°C. Figures A l to A22, in Appendix 1, show the series of breakthrough curves for each adsorbent and adsorbate, for the applied range of gas flow rates. These graphs show the general overview of the breakthrough curves for a given adsorbent/adsorbate system. The higher the superficial velocity the shorter the retention time. Figure 13 shows one example of this trend for CO on the VSA6 molecular sieve. VSA6 - CO - Breakthrough Curves ! 1 0.9 y 0.8 V 0.7 SI 0.6 § 0.5 ! 8- 0.4 ! 2 0.3 : 8 0.2 ! »- 0.1 | 0 , | 0 250 500 750 1000 1250 1500 1750 2000 j Time s | V6C0-6A V6C0-12A V6C0-16A V6CO-20A V6C0-24A V6C0-28A V6CO-32A Figure 13. Breakthrough curves for CO on VSA6 for 65, 120, 160, 200, 240, 280 and 320 cc/min flows, respectively from right to left. 55 4.3.2 Comparison of BT Curves Figures 14 to 25 show a series of breakthrough curves for four sorbates at a flow rate of 280 cc/min for each adsorbent considered in this study. Figures 14, 15, 16, and 17 show the breakthrough curves of N2, CH 4 and CO on the 5A, VSA6, 13X (8/12 mesh) and 13X (16/40 mesh) respectively. The VSA6 has a very sharp mass transfer zone (MTZ) for all three light sorbates, and especially for CO. The mean retention times of the light sorbates (CO, CH 4 and N2) were almost identical on the 13X (16/40 mesh) and 13X (8/12 mesh), but the breakthrough curves on the 13X (16/40 mesh) had a sharper MTZ, due to smaller particle size. Figures 18, 19 and 20 show breakthrough curves of C0 2, CH4, CO, and N 2 on the CBV-780, PCB Activated Carbon and MHSZ-177, respectively. Based on the slope and mean retention time1 of the breakthrough curves, it can be concluded that all three adsorbents have the highest affinity for C0 2, followed by CH4, CO and N2. The uptake (mean retention time) for CH 4 is higher on the PCB activated carbon than on the MHSZ 177, while the CO uptake and MTZ for both of them is about the same. The CBV 780 has an insignificant uptake for CH4, CO, and N2, and slightly higher for C0 2 (Figure 18). Figures 21,22 and 23 show the CH4, CO and N 2 breakthrough curves, respectively on the 13X, 5A and VSA6 adsorbents. 1 mean retention time corresponds to the time at the middle of the breakthrough curve 56 Figure 24 shows the breakthrough curves of CO2 on all adsorbents tested in this study, while Figure 25 shows only the first four (CBV-780, AA300, PCB and MHZ-177) from Figure 24, which were considered as possible candidates for the first layer of the layered adsorbent bed. From Figure 25, it is clear that the CBV 780 adsorbent has the lowest C0 2 uptake of all adsorbents tested. The activated alumina has a much higher C0 2 uptake, but still far lower than PCB or MHSZ-177, therefore it can not be considered as an efficient first layer. Although, if employed as a guard layer (in front of the first layer), besides protecting the first and second layers from moisture, it will considerably reduce the amount of C0 2 entering the first layer. This would be beneficial to the length of the first layer, in which case it could be shortened. Direct comparison of the breakthrough curves for CH4, CO, and N2 sorbates in helium are shown in Figure 21, 22, and 23 respectively for the 5A, VSA6 and 13X adsorbents (candidates for the second layer). From the CH4 breakthrough curves of Figure 21, it is clear that VSA6 has the sharpest MTZ and a slightly longer mean retention time (and therefore higher CH4 uptake) than 5 A. The 13X has the lowest CH4 uptake. From the CO breakthrough curves of Figure 22, the CO uptake, based on the mean retention time, is highest for the VSA6, which also has the sharpest MTZ. The 5A has a relatively high uptake for CO, but extremely wide MTZ which suggests lower bed efficiency. The uptake of CO on the 13X is much smaller. From Figure 23, it is clear that VSA6 also has the sharpest MTZ for N2, while the 5A has the longest mean retention time and surprisingly broader MTZ. Although N2 adsorption, on the VSA6, has a higher mean retention time than N 2 on the 13X 16/40, as expected, 57 the difference between the two adsorbents is smaller than expected1. Also, one would expect that the mean retention times for N 2 on the VSA6 and the 5A are closer1 than shown in Figure 23. From the perspective of the second layer in the layered bed, and based on the above analysis of the breakthrough curves for 5A, 13X and VSA6 adsorbents, it is evident that VSA6 is a better choice for the second layer than the commonly used 5A molecular sieve. The 13X molecular sieve, having a much lower uptake for CO, CH4 and N2 than the VSA6 and even the 5A molecular sieve, should not be considered as a second layer, especially in cases where high purity of hydrogen is required. On the other hand, the 13X molecular sieve should not be employed as a first layer either, since its uptake is too high for C0 2 (Figure 24). The comparison of CO2 breakthrough curves for CBV-780, activated alumina AA 300, PCB carbon and MHSZ-177 adsorbents (the possible candidates for the first layer), is shown in Figure 25. Clearly the uptakes for C0 2 on all of these four adsorbents is much lower than on the 5A, 13X or VSA6 adsorbents. However, only MHSZ-177 has a higher CO2 uptake than the PCB carbon, while having a slightly broader MTZ. Looking back to Figures 19 and 20, it appears that the MHSZ-177 would have a higher separation factor for CO2 (higher than PCB carbon) in comparison with the lighter impurities, CH4, CO, and N2. Furthermore, since the adsorption uptake of CH4 on PCB carbon is much higher than on the second adsorbent layer (5A molecular sieve or on VSA6), when the CH4 front 1 based on the adsorption isotherms obtained from manufacturers of the adsorbents (Appendix B) 58 enters the second layer it may cause a breakthrough of N 2 to occur earlier (Chlendy and Tondeyr, 1995). Based on the higher retention time for C0 2 on the MHSZ-177 than on the PCB carbon, and the significantly lower uptake for CH4, it is concluded that MHSZ-177 should perform better as the first layer in the layered bed, than PCB activated carbon. This conclusion is drawn despite the broader MTZ of the MHSZ-177 compared to PCB carbon, which is ascribed to the shape of the particle of the MHSZ-177. 5A 8/12- BT Curves for N2, CH4 &CO c 8. Q O 0 50 100 150 200 250 300 350 400 450 500 550 600 Time s .5A-N28A-L 5AC1-28A SA-C028A Figure 14. Breakthrough curves for N2, CH 4 and CO in helium (from left to right, respectively) on 5A MS. 59 VSA6 - Breakthrough Curves for N2, CH4 & CO ~1 i f ^ i i l ! 1 j 1 I 1 I ! 1 / i 1 j || \ \ ! ; j / : I f ; ! : - ; •• ' ' 1 ; • : ; [ . I i i f ' ; ; - ! i ! 1 I | J ! ! i ' ! i [I i 1 • : . ' ' 1 ' Ji ; i ; ; i i 50 100 150 200 250 300 350 400 450 500 550 600 Time s i i V6N-28A1 V6C1-28A V6CO-28A | j Figure 15. Breakthrough curves for N2, CH 4 and CO in helium on VSA6 MS (from left to right, respectively). 8 c <n 0) i— Q P o O O 13X 8/12 Mesh - Breakthrough Curves for N2, CH4 & CO 250 • 13N-28A1 13-C128A • 13-C028A ; Figure 16. Breakthrough curves for N2, CH 4 and CO in helium on 13X 8/12 MS (from left to right, respectively). 60 13X 16/40 Mesh - Breakthrough Curves for N2, CH4 & CO -— j j ^ — " / i /-—'. / / / • J (Lj / ! / ; • 8 0 6 f 0.5 100 150 Time s. 200 250 • 13N28A1 13CO-28A • 13C1-28A i Figure 17. Breakthrough curves for N2, CH 4 and CO, in helium on 13X 16/40 (from left to right, respectively). CBV 780 - Breakhthrough Curves for N2, CO, CH4 & C02 CBC2-28A CBCO-28A CBC1-28A CBN28A ] Figure 18. Breakthrough curves for N2, CO, CH4 and C0 2 in helium on CBV-780 (from left to right, respectively). 61 0> c o Q. (/> 0) i — Q P o o 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 PCB Carbon - Breakthrough Curves for N2, CH4, CO & C02 ! (r r i i -i \ /! ! i / • ! ! i i i / ! • / i l l ! ; / 1 ! | i : f i j / i ! i " " 1 " !— / 1 i i 1 I 1 ! i I / 1 / rfj 1—; J • J 1 • PBC228A1 • PBC1-28A - PBCO-28A • PB-N-28A-S 100 200 300 400 500 600 700 800 900 1000 Time s Figure 19. Breakthrough curves for N2, CO, CH 4 and C0 2 in helium on PCB Activated Carbon (from left to right, respectively). 0) to c o Q. CO V l_ Q P o MHSZ177 - Breakthrough Curves for N2, CO, CH 4 & C 0 2 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 f ! i / : 1 i / ! i / I i i i i i . / ! j i / i ! i / I i ; / i J I i / \ i l l | ! 100 200 300 400 500 Time s 600 700 800 900 1000 -MZC2-28A . MZCO-28A -MZC1-28A -MZ-N-28A Figure 20. Breakthrough curves for N2, CO, CH 4 and C0 2 in helium on MHSZ-177 high silica zeolite (from left to right, respectively). 62 13X 16/40, 5A MS & VSA6 - BT Curves for CH4 0.1 0 ! /T i I i I / ! / X 1 1 ] 1/ 1 13X / j i ! / ! / / i ! / 5A / / VSA6 •i i i i / / i '•• i ! i : / / I ! i / / / : y ^ y \ ! 0 25 50 75 100 125 150 175 j Times • j i 13C1-28A V6C1-28A-SS 5AC128A~j I I 1 ! Figure 21. Breakthrough curves for CH 4 in helium on 13X, 5A and VSA6 (from left to right, respectively). 13X 16/40, 13X 8/12, 5A & VSA6 - BT Curves for CO 1 0.9 0.8 Q 0.4 i i I i i / /13X-8/1 2 I ! / ! / i i i / / I ; / ! i i ; / i i 1 ! ! / i / i / i 6/40 i /5A l ; / VSA6 ! 1 7 ! ! 0 50 100 150 200 250 300 350 400 450 500 Time s 13X-28A-CO-16/40 V6CO-28A — 5A-C028A 13-C028A-8/12 Figure 22. Breakthrough curves for CO in helium on 13X 8/12, 13X 16/40, 5A, and VSA6 (from left to right, respectively). 63 13X 16/40, VSA6, & 5A - Breakthrough Curves for N2 0 10 20 30 40 50 60 70 80 90 100 110 Time s | i V6N-28A 5AN-28A 13X-N28A j Figure 23. Breakthrough curves for N 2 in helium on 13X, VSA6 and 5A, (from left to right, respectively). 3? c tn ! Q CBV780, AA 300, PCB Carbon, MHSZ177, VSA6, 13X 8/12, 13X 16/40 and 5A - BT Curves for C02 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1 i / ; / / ; i ! / / I ! j j I j ; / / | / i i / ! / ; / 1 113X8/1? i / / i MHSZ177 ! ! / / 13X16/40! / I 1 VSA6! J j i y5A 1000 2000 3000 4000 Time s 5000 6000 7000 8000 -13C2-28A -V6C228A • PBC2-28A 5A-C228A -MZC2-28A -13-C228A-8/12 -CBC2-28A -AAC2-28A Figure 24. Breakthrough curves for C0 2 in helium on CBV780, AA 300, PCB Activated Carbon, MHSZ177, VSA6, 13X 8/12, 13X 16/40 and 5A MS, (from left to right, respectively). 64 CBV780, AA 300, PCB Carbon and MHSZ177 - BT Curves for C02 ^ • [— ' : —~T—~ 1 1 i \ I \ / i / / i / / \ '••i'i/ j j i ! / i / I ; / . ! T 1 I I / 1 / ' / | ; / 'i i f / i i i ; .1 ^ j i ; / • i : / i / 1 ^BV780 I I PCB / / i ! AA300 ! j ; / j / /MHSZ177 I | 0 100 200 300 400 500 600 700 800 900 1000 j Time s i i ! i CBC2-28A AAC2-28A PBC2-28A MZC2-28A i .-Figure 25. Breakthrough curves for C0 2 in helium on CBV780, AA 300, PCB Activated Carbon and MHSZ177 (from left to right, respectively). 4.4 Moments and HETP The calculations of the first and second moments were based on Equations 3 and 4 and are calculated from experimental breakthrough data, for each adsorbent/adsorbate system and applied gas velocity, using the simple trapezoidal integration rule. The results are shown in Tables 3 to 16. From the first two moments (after correcting them with the values of the system response, since uncorrected values of p and cr2 are reported in Tables 3-16), the height of the equivalent theoretical plate, HETP, was calculated via equation 36. Values of the HETP, for each adsorbate/adsorbent system are also given in Tables 3 to 16. The p represents the first moment, or the mean retention time, while CT is the second moment, or variance, for a particular sorbate/adsorbent system at the measured 65 gas flow. In general, the higher the value of the first moment, the higher the uptake of the sorbate by the corresponding adsorbent. The lower the HETP value for the particular sorbate/adsorbent system at measured gas flow, the more efficient the adsorbent column. Tables 3 to 10 show values for p, a 2 and HETP, for different adsorbate/adsorbent systems, calculated from the data obtained from the adsorption breakthrough curves. The values in Tables 11 to 16, are calculated from the data obtained from the desorption breakthrough curves. This was done since it was observed that the adsorption and desorption curve, for the same gas velocity on the particular adsorbate/adsorbent system, were not mirror images. This may be due to the level of concentration of the sorbate, especially for the more strongly adsorbed species, and perhaps due to the nonlinearity of the TCD detector itself. The comparison of p and HETP will be based mostly on the adsorption data. The desorption data will be used to obtain the average values for [i, <j2 and HETP in the calculation of the axial diffusion coefficient D L , and the lumped mass transfer coefficient, in Section 4.6. Comparing p and HETP, for the PCB activated carbon and the MHSZ-177, calculated from the adsorption breakthrough curves (Table 3 and 5), at 280 cc/min feed flow (3.5 cm/sec superficial gas velocity), it is evident that the MHSZ-177 has a higher mean retention time for CO2 (598 sec vs 467 sec). Also, the MHSZ-177 has a higher HETP, as a result of a broader MTZ, which can be due to the long cylindrical shape of the MHSZ-177 extrudate. The higher HETP value indicates a lower adsorbent efficiency, although, desorption may be slightly easier for the MHSZ-177. From Tables 4 and 6, at the same gas flow, activated alumina AA300, and the CBV 780 had a significantly higher HETP 66 than PCB and MHSZ-177, indicating much lower adsorption efficiency in regard to CO2 separation. Consequently the only viable choices for the first layer of a layered bed adsorber are the PCB carbon and the MHSZ-177. Comparing values of HETP for the 13X, VSA6 and 5A molecular sieve (Tables 7, 8, 9, and 10), at the same flow conditions, it is evident that the VSA6 adsorbent had the lowest value of HETP in the case of CO, CH 4 and N 2 adsorption, followed by the 13X 16/40, 5 A and 13X 8/12 molecular sieve. This suggests that the VSA6 has the highest adsorbent overall performance for the adsorption of light impurities, and therefore is considered the strongest candidate for the second layer in a layered adsorbent bed for hydrogen separation from SMR gas. Plots of HETP versus v (superficial gas velocity) for the 5 A, VSA6 and 13X adsorbents and the light adsorbates are shown in Figures 26 - 28. Most of the curves have a hyperbolic shape similar to the R curve of Figure 11. The increase of HETP with an increase of superficial gas velocity, noticed in the case of the 5 A and especially 13X 8/12 mesh, indicates the existence of mass transfer resistance. A visual comparison of the HETP versus v (superficial gas velocity) curves suggests that the VSA6 adsorbent during the adsorption of CO, CH4 and N 2 would not be affected by mass transfer resistance at higher gas velocities, as much as other adsorbents. In the case of the 13X adsorbents (Figure 28), it is evident that mass transfer resistance would be higher in the case of the larger particle size, 13X (8/12 mesh). Based on the experimental data and the calculated values of the first and second moments (p. and a ), an empirical equation, developed in the present study for the breakthrough 67 curve, is presented in the form of an error function: — = — 11 + erf —f=— Co 2 V2<T Equation 39 The correlation with the experimentally obtained data was very good (0.996 < r 2 < 0.999). Figures 29 to 31 show some examples of this agreement. The simplicity of this expression, since it is based on the first and second moments may be a convenient way to calculate the breakthrough curve if DL is not known. The Levenspiel and Bischoff analytical solution (Levenspiel and Bischoff, 1963, Ruthven, 1984) for the linear isothermal trace-component system (assuming disperse plug flow) is shown in Equation 40: VZ Equation 40 This equation was also tested in the present study. The agreement with the experimental data was also very good (0.995 < r 2 < 0.999), and some examples are shown in Figures 32 to 34. However, it is noticeable that both equations do not perfectly fit the upper tailing part of the experimentally obtained data. Even though the Equation 39 was obtained empirically, it can be noticed that in the case when the mass transfer resistance is zero and when t/u,«l (which is true if the breakthrough curve is steep and the retention time is not to short) then Equation 40 would 68 reduce to Equation 39. Table 3: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO, CH4, N 2 and C0 2 on PCB activated carbon adsorbent Adsorbate/Adsorbent u u 1 a2 H E T P L R e p data (cm/s) (s) (s2) (s2) (cm) (cm) CO-PCB-AC-12/30 PBCO-6A 0.8 254 64273 1285 0.44 20.5 0.13 PBCO-8A 1.0 206 42488 735 0.38 20.5 0.16 PBCO-12A 1.5 139 19432 256 0.29 20.5 0.25 PBCO-16A 2.0 106 11221 143 0.29 20.5 0.33 PBCO-20A 2.5 86 7406 92 0.28 20.5 0.41 PBCO-24A 3.0 73 5275 59 0.26 20.5 0.49 PBCO-28A 3.5 63 3986 48 0.28 20.5 0.57 CH4-PCB-AC-12/30 BPC1-6A 0.8 705 485560 8011 0.34 20.5 0.13 PBC1-8A 1.0 584 332250 4589 0.28 20.5 0.16 PBC1-12A 1.5 390 148217 1574 0.22 20.5 0.25 PBC1-16A 2.0 295 84403 701 0.17 20.5 0.33 PBC1-20A 2.5 238 54758 428 0.16 20.5 0.41 PBC1-24A 3.0 202 39242 399 0.21 20.5 0.49 PBC1-28A 3.5 176 29888 371 0.25 20.5 0.57 N 2-PCB-AC-12/30 PB-N-6A1 0.8 178 31716 635 0.44 20.5 0.13 PB-N-8A2 1.0 144 20699 360 0.38 20.5 0.16 PB--N-12A 1.5 98 9647 128 0.30 20.5 0.25 PB-N-16As 2.0 75 5682 67 0.27 20.5 0.33 PB-N-20As 2.5 62 3798 43 0.26 20.5 0.41 PB-N24A2 3.0 53 2799 30 0.26 20.5 0.49 PB-N-28A 3.5 46 2133 21 0.24 20.5 0.57 CO 2 -PCB-AC-12/30 PBC228A1 3.5 467 215019 1476 0.14 20.5 0.57 69 Table 4: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO2 on AA300 activated alumina adsorbent Adsorbate/Adsorbent u (cm/s) H (s) u5 (s2) a 5 (s2) H E T P (cm) L (cm) R e p data CO 2-AA-300-14/40 AAC2-12A 1.5 567 315239 34650 2.25 20.5 0.20 AA-C216A 2.0 518 263655 22350 1.74 20.5 0.26 AAC2-20A 2.5 324 102455 6940 1.39 20.5 0.33 AAC2-24A 3.0 257 63852 3216 1.03 20.5 0.40 AAC2-28A 3.5 225 49006 2218 0.93 20.5 0.46 AAC2-32A 4.0 198 37766 2237 1.21 20.5 0.53 Table 5: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO, CH4, N 2 and C0 2 on MHSZ 177 high silica zeolite adsorbent Adsorbate/Adsorbent u u 2 H E T P L R e p data (cm/s) (s) (s2) (s2) (cm) (cm) C O - M H S Z 177 MZCO-6A 0.8 230 52765 923 0.39 20.5 0.30 MZCO-8A 1.0 186 34689 557 0.36 20.5 0.38 MZCO-12A 1.5 127 16097 229 0.32 20.5 0.57 MZCO-16A 2.0 95 9095 141 0.35 20.5 0.75 MZCO-20A 2.5 78 6014 115 0.44 20.5 0.94 MZCO-24A 3.0 66 4356 101 0.54 20.5 1.13 MZCO-28A 3.5 57 3216 83 0.60 20.5 1.32 C H 4 - M H S Z 177 MZC1-6A 0.8 396 150018 3288 0.45 20.5 0.30 MZC1-8A 1.0 322 99338 1987 0.41 20.5 0.38 MZC1-12A 1.5 216 44240 788 0.37 20.5 0.57 MZC1-16A 2.0 166 25914 510 0.40 20.5 0.75 MZC1-20A 2.5 134 16946 361 0.44 20.5 0.94 MZC1-24A 3.0 112 11756 265 0.46 20.5 1.13 MZC1-28A 3.5 98 8891 217 0.50 20.5 1.32 N 2 - M H S Z 177 MZ-N-6A 0.8 128 14399 322 0.46 20.5 0.30 MZ-N8A 1.0 105 9583 193 0.41 20.5 0.38 MZ-N-12A 1.5 72 4423 82 0.38 20.5 0.57 MZ-N-16A 2.0 54 2459 44 0.37 20.5 0.75 MZ-N-20A 2.5 45 1629 30 0.38 20.5 0.94 Mz-N-24A 3.0 35 991 21 0.44 20.5 1.13 MZ-N28A 3.5 31 766 19 0.50 20.5 1.32 C 0 2 - M H S Z 177 MZC2-28A 3.5 599 353982 3862 0.22 20.5 1.32 70 Table 6: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO, CHt, N 2 and C0 2 on CBV 780 Y-zeolite adsorbent Adsorbate/Adsorbent u (cm/s) (s) u1 (s2) (s2) H E T P (cm) L (cm) R e p data C O - C B V 780 CBCO-6A 0.8 65 3162 74 0.48 20.5 0.30 CBCO-8A 1.0 53 1949 33 0.34 20.5 0.38 CBCO-12A 1.5 36 931 16 0.36 20.5 0.57 CBCO-16A 2.0 28 523 13 0.51 20.5 0.75 CBCO-20A 2.5 23 338 8.9 0.54 20.5 0.94 CBCO-24A 3.0 19 235 6.5 0.57 20.5 1.13 CBCO-28A 3.5 17 178 5.4 0.62 20.5 1.32 C H 4 - C B V 780 CBC1-6A 0.8 81 5257 146 0.57 20.5 0.30 CBC1-8A 1.0 66 3482 92 0.54 20.5 0.38 CBC1-12A 1.5 45 1586 38 0.49 20.5 0.57 CBC1-16A 2.0 35 894 23 0.52 20.5 0.75 CBC1-20A 2.5 29 594 15 0.51 20.5 0.94 CBC1-24A 3.0 24 403 9.9 0.51 20.5 1.13 CBC1-28A 3.5 21 319 9.5 0.61 20.5 1.32 N 2 - C B V 780 CBN6A 0.8 52 1887 44 0.48 20.5 0.30 CBN8A 1.0 42 1233 25 0.41 20.5 0.38 CBN12A 1.5 29 555 8.3 0.31 20.5 0.57 CBN16A 2.0 22 296 3.6 0.25 20.5 0.75 CBN20A 2.5 19 204 1.7 0.17 20.5 0.94 CBN24A 3.0 16 142 0.8 0.11 20.5 1.13 CBN28A 3.5 14 106 0.3 0.05 20.5 1.32 C 0 2 - C B V 780 CBC2-28A 3.5 133 16843 542 0.66 20.5 1.32 71 Table 7: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO, CH4, N 2 and C0 2 on 13X 16/40 mesh molecular sieve adsorbent Adsorbate/Adsorbent u u1 a 1 HETP L Re„ data (cm/s) (s) (s2) (s2) (cm) (cm) CO-13X-16/40B 13CO-6A 0.8 541 283166 4746 0.34 20.5 0.09 13CO-8A 1.0 442 188702 2573 0.28 20.5 0.12 13CO-12A 1.5 295 83712 858 0.21 20.5 0.17 13CO-16A 2.0 224 47820 409 0.18 20.5 0.23 13CO-20A 2.5 182 31494 278 0.18 20.5 0.29 13CO-24A 3.0 152 22074 221 0.21 20.5 0.35 13CO-28A 3.5 133 16807 206 0.25 20.5 0.40 CH4-13X-16-40B 13C1-6A3 0.8 256 61208 1146 0.38 20.5 0.09 13C1-8A 1.0 209 40561 675 0.34 20.5 0.12 13C1-12A 1.5 141 18436 199 0.22 20.5 0.17 13C112A1 1.5 142 18546 215 0.24 20.5 0.17 13C1-16A 2.0 109 10766 96 0.18 20.5 0.23 13C1-20A 2.5 89 7113 57 0.17 20.5 0.29 13C1-24A 3.0 73 4795 30 0.13 20.5 0.35 13C1-28A 3.5 64 3652 20 0.11 20.5 0.40 N2-13X-16/40B 13N-6A1 0.8 163 23813 441 0.38 20.5 0.09 13N-8A2 1.0 134 16096 250 0.32 20.5 0.12 13N-12A 1.5 91 7404 78 0.22 20.5 0.17 13N-16A 2.0 70 4239 35 0.17 20.5 0.23 13N-20A 2.5 57 2788 16 0.12 20.5 0.29 13N-24A1 3.0 47 1899 7.9 0.08 20.5 0.35 13N-28A1 3.5 41 1434 4.6 0.07 20.5 0.40 CO2-13X-16/40B 13C2-28A 3.5 4300 18463465 179098 0.20 20.5 0.40 72 Table 8: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO, CH4, N 2 and C0 2 on 13X 8/12 mesh molecular sieve adsorbent Adsorbate/Adsorbent u l i I? a1 H E T P L R e p data (cm/s) (s) (s2) (s2) (cm) (cm) CO-13X-8/12B 13-C06A 0.8 560 304573 8106 0.55 20.5 0.22 13-C08A 1.0 457 202452 4929 0.50 20.5 0.27 13-C012A 1.5 308 91717 2236 0.50 20.5 0.41 13-C016A 2.0 233 52180 1389 0.55 20.5 0.55 13-CO20A 2.5 188 33747 989 0.60 20.5 0.68 13-C024A 3.0 157 23452 805 0.70 20.5 0.82 13-C028A 3.5 137 17719 694 0.80 20.5 0.96 13-C032A 4.0 120 13576 597 0.90 20.5 1.09 CH4-13X-8/12B 13-C16A 0.8 270 68156 1857 0.56 20.5 0.22 13-C18A 1.0 220 45201 1097 0.50 20.5 0.27 13-C112A 1.5 148 20447 443 0.44 20.5 0.41 13-C116A 2.0 113 11628 256 0.45 20.5 0.55 13-C120A 2.5 91 7572 176 0.48 20.5 0.68 13-C124A 3.0 77 5313 135 0.52 20.5 0.82 13-C128A-L 3.5 67 3987 114 0.59 20.5 0.96 13-C132A 4.0 59 3094 101 0.67 20.5 1.09 N 2-13X-8/12B 13N6A 0.8 165 24644 655 0.54 20.5 0.22 13N8A 1.0 136 16698 422 0.52 20.5 0.27 13N12A 1.5 92 7542 179 0.49 20.5 0.41 13X16A 2.0 71 4313 109 0.52 20.5 0.55 13N20A 2.5 57 2819 78 0.57 20.5 0.68 13X-N24A 3.0 48 1981 59 0.61 20.5 0.82 13X-N28A 3.5 42 1501 50 0.68 20.5 0.96 13X-N32A 4.0 30 696 27 0.80 20.5 1.09 C0 2 -13X-8/12B 13-C228A 3.5 4087 16674105 313323 0.39 20.5 0.96 13-C232A 4.0 3009 9035196 314963 0.71 20.5 1.09 73 Table 9: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO, CH4, N 2 and C0 2 on VSA6 Molecular Sieve adsorbent Adsorbate/Adsorbent u a* HETP L Rep data (cm/s) (s) (s2) (s2) (cm) (cm) CO-VSA6-16/40 V6CO-6A 0.8 1600 2532381 4162 0.03 20.5 0.09 V6CO-12A 1.5 875 756027 792 0.02 20.5 0.12 V6CO-20A 2.5 531 277427 402 0.03 20.5 0.17 V6CO-16A 2.0 659 428589 547 0.03 20.5 0.23 V6CO-16A 2.0 660 428728 569 0.03 20.5 0.29 V6CO-24A 3.0 442 191660 301 0.03 20.5 0.35 V6CO-28A 3.5 382 143244 242 0.03 20.5 0.40 V6CO-32A 4.0 335 109720 234 0.04 20.5 0.46 CH4-VSA6-I6/4O V6C1-6A 0.8 409 160449 2058 0.26 20.5 0.09 V6C1-8A 1.0 336 107983 1130 0.21 20.5 0.12 V6C1-12A 1.5 228 49518 357 0.15 20.5 0.17 V6C1-16A 2.0 173 28384 176 0.13 20.5 0.23 V6CI-20A 2.5 140 18463 112 0.12 20.5 0.29 V6C1-24A 3.0 118 12917 69 0.11 20.5 0.35 V6C1-28A 3.5 102 9684 50 0.11 20.5 0.40 V6C1-32A 4.0 90 7460 39 0.11 20.5 0.46 N2-VSA6-16/40 V6N-6A 0.8 159 22540 380 0.35 20.5 0.09 V6N-8A 1.0 129 14848 174 0.24 20.5 0.12 V6N-12A 1.5 88 6874 60 0.18 20.5 0.17 V6N-16A 2.0 68 3961 22 0.11 20.5 0.23 V6N-20A 2:5 55 2617 8.5 0.07 20.5 0.29 V6N-24A 3.0 47 1850 1.5 0.02 20.5 0.35 V6N-28A 3.5 41 1407 0.25 0.00 20.5 0.40 CO2-VSA6-16/40B V6C228A 3.5 2672 7118277 191291 0.55 20.5 0.40 V6C2-32A 4.0 2244 5021345 220331 0.90 20.5 0.46 74 Table 10: First, second moments and HETP calculated from the adsorption breakthrough curves obtained for CO, CH4, N2 and C0 2 on 5A molecular sieve adsorbent Adsorbate/Adsorbent u a 1 HETP L Rep data (cm/s) (s) (s2) (s2) (cm) (cm) CO-5A-8/12B 5A-C06A 0.8 1451 2081501 44691 0.44 20.5 0.22 5A-C08A 1.0 1202 1428701 22818 0.33 20.5 0.27 5A-C012A 1.5 781 601565 7330 0.25 20.5 0.41 5A-C016A 2.0 568 316025 3726 0.24 20.5 0.55 5A-CO20A 2.5 461 207953 2939 0.29 20.5 0.68 5A-C024A 3.0 386 145721 2398 0.34 20.5 0.82 5ACO-28A 3.5 315 96747 1962 0.42 20.5 0.96 5A-C032A- 4.0 286 79631 1811 0.47 20.5 1.09 CH4-5A-8/12B 5ACl-6a 0.8 393 148066 3745 0.52 20.5 0.22 5AC1-8A 1.0 322 99416 2169 0.45 20.5 0.27 5AC1-12A 1.5 219 45769 842 0.38 20.5 0.41 5AC1-16A 2.0 167 26363 474 0.37 20.5 0.55 5AC1-20A 2.5 136 17267 321 0.38 20.5 0.68 5AC1-24A 3.0 114 12055 243 0.41 20.5 0.82 5AC128-A . 3.5 99 9109 201 0.45 20.5 0.96 5AC1-32A 4.0 88 7072 176 0.51 20.5 1.09 N2-5A-8/12B 5A-N6A 0.8 279 73102 1693 0.47 20.5 0.22 5A-N8A 1.0 230 49484 1065 0.44 20.5 0.27 5A-N12A 1.5 157 23082 396 0.35 20.5 0.41 5A-N16A 2.0 120 13233 222 0.34 20.5 0.55 5A-N20A 2.5 97 8671 152 0.36 20.5 0.68 5A-N24A 3.0 82 6136 116 0.39 20.5 0.82 5AN-28A 3.5 72 4638 93 0.41 20.5 0.96 5A-N32A 4.0 63 3560 75 0.43 20.5 1.09 C02-5A-8/12-Ads 5A-C028A 3.5 311 94474 2311 0.50 20.5 0.96 75 Table 11: First, second moments and HETP calculated from the desorption breakthrough curves obtained for N 2 on 13X 16/40 mesh molecular sieve adsorbent Adsorbate/Adsorbent u (cm/s) (s) u1 (s2) a 1 (s2) HETP (cm) L (cm) Re p data N2-13X-16/40B-Desorption 13N-6D-16/40 0.8 164 24212 875 0.74 20.5 0.09 13N12D-16/40 1.5 91 7370 199 0.55 20.5 0.17 13N-16D-16/40 2.0 70 4285 113 0.54 20.5 0.23 13N-20D-16/40 2.5 57 2783 70 0.52 20.5 0.29 13N-24D-16/40 3.0 48 1942 50 0.53 20.5 0.35 13N-28D-16/40 3.5 41 1398 33 0.48 20.5 0.40 Table 12: First, second moments and HETP calculated from the desorption breakthrough curves obtained for N 2 and CO on 13X 8/12 mesh molecular sieve adsorbent Adsorbate/Adsorbent u u1 a 5 HETP L Re p data (cm/s) (s) (s2) (s2) (cm) (cm) N2-13X-8/12B-Desorption 13N-6D 0.8 168 25555 1203 0.96 20.5 0.22 13N8D 1.0 139 17285 788 0.93 20.5 0.27 13N12D 1.5 94 7840 367 0.96 20.5 0.41 13N16D 2.0 71 4419 218 1.01 20.5 0.55 13N20D 2.5 58 2875 161 1.15 20.5 0.68 13N24D 3.0 49 2010 121 1.23 20.5 0.82 13N28D 3.5 42 1517 105 1.42 20.5 0.96 CO-13X-8/12B-Desorption 13-C06D 0.8 576 322511 19858 1.26 20.5 0.22 13-C08D 1.0 464 208668 11560 1.14 20.5 0.27 13-CO-12D 1.5 315 95836 5809 1.24 20.5 0.41 13-C016D 2.0 238 54301 3477 1.31 20.5 0.55 13-CO20D 2.5 192 35357 2369 1.37 20.5 0.68 13-C024D 3.0 161 24786 1826 1.51 20.5 0.82 13-C028D 3.5 140 18514 1382 1.53 20.5 0.96 13-C032D 4.0 162 24936 1917 1.58 20.5 1.09 76 Table 13: First, second moments and HETP calculated from the desorption breakthrough curves obtained for CO and CH4 on VSA6 16/40 mesh molecular sieve adsorbent Adsorbate/Adsorbent u u1 HETP L Re p data (cm/s) (s) (s2) (s2) (cm) (cm) CO-VSA6-Desorption V6CO-6D 0.8 1505 2241137 651105 5.96 20.5 0.09 V6CO-12D 1.5 837 692003 200322 5.93 20.5 0.17 V6CO-16D 2.0 656 424580 140231 6.77 20.5 0.23 V6CO-20D 2.5 544 291663 107032 7.52 20.5 0.29 V6CO-24D 3.0 454 202039 74265 7.54 20.5 0.35 V6CO-28D 3.5 393 151400 57426 7.78 20.5 0.40 V6CO-32D 4.0 349 119827 47511 8.13 20.5 0.46 CH4-VSA6-Desorption V6C1-6D 0.8 418 168034 8605 1.05 20.5 0.09 V6C1-8D 1.0 340 110706 4971 0.92 20.5 0.12 V6C1-12D 1.5 231 50673 1992 0.81 20.5 0.17 V6C1-16D 2.0 175 28902 1051 0.75 20.5 0.23 V6C1-20D 2.5 141 18824 651 0.71 20.5 0.29 V6C1-24D 3.0 118 13113 442 0.69 20.5 0.35 ' V6C1-28D 3.5 103 9880 329 0.68 20.5 0.40 V6C1-32D 4.0 90 7555 245 0.66 20.5 0.46 Table 14: First, second moments and HETP calculated from the desorption breakthrough curves obtained for CO2 on Activated Alumina AA-300 adsorbent Adsorbate/Adsorbent u (cm/s) (s) u2 (s2) a J (s2) HETP (cm) L (cm) Re p data CO-MHSZ 177-Des MZCO-6D 0.8 233 50466 3174 1.29 20.5 0.22 MZCO-8D 1.0 188 32750 1994 1.25 20.5 0.27 MZCO-12D 1.5 129 15192 934 1.26 20.5 0.41 MZCO-16D 2.0 97 8445 563 1.37 20.5 0.55 MZCO-20D 2.5 79 5584 387 1.42 20.5 0.68 MZCO-24D 3.0 66 3814 292 1.57 20.5 0.82 MZCO-28D 3.5 57 2840 224 1.62 20.5 0.96 77 Table 15: First, second moments and HETP calculated from the desorption breakthrough curves obtained for CO, CH4 and N2 on 5A 8/12 mesh molecular sieve adsorbent Adsorbate/Adsorbent u 1^  u1 a5 HETP L Rep data (cm/s) (s) (s2) (s2) (cm) (cm) CO-5A-Desorption 5A-C06D 0.8 1498 2217661 212947 1.97 20.5 0.22 5A-C08D 1.0 1236 1527975 140340 1.88 20.5 0.27 5A-C012D 1.5 817 657917 59228 1.85 20.5 0.41 5A-C016D 2.0 577 327345 30116 1.89 20.5 0.55 5A-CO20D 2.5 478 224749 20998 1.92 20.5 0.68 5A-C024D 3.0 403 159351 15445 1.99 20.5 0.82 5AC028-D 3.5 358 125623 12823 2.09 20.5 0.96 5A-C032D 4.0 302 89045 9345 2.15 20.5 1.09 CH4-5A-Deso rption 5AC1-6D 0.8 403 155380 6568 0.87 20.5 0.22 5AC1-8D 1.0 328 102974 3981 0.79 20.5 0.27 5AC1-12D 1.5 223 47311 1643 0.71 20.5 0.41 5AC1-16D 2.0 170 27157 936 0.71 20.5 0.55 5AC1-20D 2.5 138 17773 627 0.72 20.5 0.68 5AC1-24D 3.0 116 12549 452 0.74 20.5 0.82 5AC1-28D 3.5 100 9214 . 371 0.83 20.5 0.96 5AC1-32D 4.0 88 7183 305 0.87 20.5 1.09 N2-5A-Desorption 5A-N6D 0.8 286 77063 3627 0.96 20.5 0.22 5A-N8D 1.0 234 51580 2287 0.91 20.5 0.27 5A-N12D 1.5 160 23871 918 0.79 20.5 0.41 5A-N16D 2.0 122 13631 552 0.83 20.5 0.55 5A-N20D 2.5 99 9004 373 0.85 20.5 0.68 5A-N24D 3.0 83 6259 267 0.87 20.5 0.82 5A-N28D 3.5 73 4760 224 0.96 20.5 0.96 5A-N32D 4.0 64 3677 197 1.10 20.5 1.09 78 Table 16: First, second moments and HETP calculated from the desorption breakthrough curves obtained for CO2 on Activated Alumina AA-300 adsorbent Adsorbate/Adsorbent data u (cm/s) (s) u5 (s2) a 1 (s2) HETP (cm) L (cm) Rep CO-PCB-AC-Des PBCO-6D 0.8 256 61262 2389 0.80 20.5 0.11 PBCO-8D 1.0 207 40146 1429 0.73 20.5 0.13 PBCO-12D 1.5 141 18397 578 0.64 20.5 0.26 PBCO-16D 2.0 106 10319 319 0.63 20.5 0.33 PBCO-20D 2.5 86 6701 211 0.65 20.5 0.40 PBCO-24D 3.0 73 4747 149 0.64 20.5 0.46 PBCO-28D 3.5 63 3569 114 0.66 20.5 0.53 H E T P vs v CO, CH 4 & N2 on 5A molecular sieve 0.1 -I 0 -j , , , . . , , , 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 v (cm/s) !-*-CO-5A-8/12B -»-CH4-5A-8/12B - a - N2-5A-8/12B | Figure 26. Variation of HETP with superficial gas velocity (v) for CO, CH4, and N2 on 5A molecular sieve. 79 E LU X H E T P vs v CO, CH4, & N 2 on VSA6 molecular sieve • CO-VSA6-16/40 -CH4-VSA6-16/40 N2-VSA6-16/40 Figure 27. Variation of HETP with superficial gas velocity (v) for CO, CH4, and N 2 on VSA6 molecular sieve. H E T P vs v CO, CH„, & N 2 on 13X 16/40 & 13X 8/12 mesh molecular sieve 4.5 (cm/s) P«-CO-13X-16/40B -A-CH4-13X-16-40B -X-N2-13X-16/40B |-»-CO-13X-8/12B -*-CH4-13X-8/12B N2-13X-8/12B Figure 28. Variation of HETP with superficial gas velocity (v) for CO, CH4, and N 2 on 13X 16/40 and 13X 8/12 mesh molecular sieve. 80 CH4 on PCB @ u=2cm/sec y=SK-erf( u,CT) r2=0.99915018 DF Adj r2=0.99914912 FitStdErr=0.011554286 Fstat=1889373.7 ii=295.4 a=26.474 O O O 0 100 400 200 300 Time (sec) Figure 29. Breakthrough curves of CH4 on the PCB carbon obtained experimentally (thicker line) and by calculation using Equation 39 (thinner line). N2 on VSA6 @ u=2cm/sec y=SK-erf(u,cr) r2=0.99831224 DF Adj r2=0.99830475 FitStdErr=0.018914323 Fstat=267357.91 u=67.8 a=4.71 O O o 50 75 100 125 Time (sec) Figure 30. Breakthrough curves of N 2 on the VSA6 adsorbent obtained experimentally (thicker line) and by calculation using Equation 39 (thinner line). 81 C O on V S A 6 @ u=2cm/sec y=SK-erf(M,<j) r2=0.99567555 DF Adj r2=0.99567296 FitStdErr=0.025565832 Fstat=769011.87 u=659.6 a=23.399 O O O 0 200 600 800 400 Time (sec) Figure 31. Breakthrough curves of CO on the VSA6 adsorbent obtained experimentally (thicker line) and by calculation using Equation 39 (thinner line). C H 4 on P C B @ u=2cm/sec Levenspiel-Bischoff y=erf(n,D|_,zv) r2=0.99804477 DF Adj r2=0.99804112 FitStdErr=0.017531245 Fstat=409890.47 ua=295.4 DL=0.3352 zv=110.7 O O O 0 100 400 200 300 Time (sec) Figure 32. Breakthrough curves of CH 4 on the PCB carbon obtained experimentally (thicker line) and by calculation using Equation 40 (thinner line). 82 N2 on V S A 6 @ u=2cm/sec Levenspiel-Bischoff y=erf(i_t,DL,zv) r2=0.9992911 DF Adj r2=0.99928637 FitStdErr=0.012271801 Fstat=317873.02 u=67.7 DL=0.37923 zv=110.70488 O O o 50 75 100 125 Times (sec) Figure 33. Breakthrough curves of N2 on the VSA6 adsorbent obtained experimentally (thicker line) and by calculation using Equation 40 (thinner line). C O on V S A 6 @ u=2cm/sec Levenspiel-Bischoff y=erf(n,DL,zv) r2=0.99502532 DF Adj r2=0.99502085 FitStdErr=0.027424695 Fstat=333930.12 ti=659.6 DL=0.029 2V=110.7 O O o 200 600 800 400 Time (sec) Figure 34. Breakthrough curves of CO on the VSA6 adsorbent obtained experimentally (thicker line) and by calculation using Equation 40 (thinner line). 83 4.5 Adsorption Equilibrium Constant K (Henry Constant) The adsorption equilibrium constants for all the sorbates and all the adsorbents considered in this study were determined from Equation 11, using the graphical method depicted in Figure 8. The obtained results are shown in Table 17. Examples of some of the results are depicted graphically in Figure 35. From Figure 35, it is clear that the p. versus 1/F experimental data fall on a straight line, starting from the origin. From Table 17, it is also evident that the MHZS-177 has a higher value of K for C 0 2 than the P C B activated carbon. This confirms that the MHZS-177 has a higher equilibrium uptake than the P C B carbon. Since the KN2, KCH4, and Kco for the MHSZ-177, are lower than for the P C B activated carbon, the separation factors KcWKcm, Kco2/Kco> and KCO2/KN2 for MHZS-177 are much higher than those for the P C B carbon. This suggests that the MHSZ-177 would be more efficient in separating C 0 2 from the lighter impurities. The VSA6 has a higher value of K for C O and CH4 (the main impurities in the SMR gas), than the 5A molecular sieve. This suggests that the VSA6 is a slightly stronger candidate for the second layer in the layered adsorbent bed. The K value for N2 for the 5A molecular sieve is much higher (~ 80%) than expected in comparison with the K N 2 value for the VSA6 (which was rather low). Since, the N2 isotherms for both adsorbents are very similar (Appendix B ) , it would be expected that the KN2 values are much closer, and perhaps somewhere in between the values for the VSA6 and the 5A. 84 Table 17: Calculated values of adsorption equilibrium constants K (Ts22±0.5°C) Mesh Manufact. Weight g KN2 KcH4 Kco K-C02 Kc02/KcH4 Kco2/Kco Kc02/K-N2 13X-8/12 8/12 UOP 18.9 22.3 36.2 78.2 2377 65.7 30.4 106.8 13 X-16/40 16/40 UOP 20.6 22.9 36.0 73.7 2556 71.1 34.7 111.8 AA-14/20 14/20 Alcan 24.0 - - - 149.6 - - -CBV-780 1/16" Zeolyst 13.8 11.1 79.2 7.1 PCB-12/30 12/30 Calgon 13.4 24.8 98.5 33.9 278 2.8 8.2 11.2 MHSZ177 1/16" UOP 19.2 16.6 54.3 30.4 356 6.6 11.7 21.4 5A 8/12 Davison 18.9 40.1 55.1 199.4 3524 64.0 17.7 87.9 VSA6 16/40 UOP 20.0 22.1 57.2 221.1 1507 26.3 6.8 68.3 l V S A 6 -16/40B - First Moment v s 1/F 0.2 0.4 0.6 0.8 1 1/F (sec-cm3) • N2-VSA6-16/40 • CO-VSA6-16/40 CH4-VSA6-16/40 Figure 35. Determination of adsorption equilibrium constants for N2, CO and CH4 on VSA6 molecular sieve (Ts22+0.5°C). 85 Table 18 shows values of K for CO2 and CH4 obtained on different adsorbents in other studies. It seems that the values of K for the same adsorbent/ adsorbate system can differ significantly among different studies, which is rather surprising. In the case for silicalite, reported values of Kco2 from the studies of Ress, (1994) and Choundary and Mayadevy (1993) , respectively, were 50 and 1700. The K values for most of the adsorbent/sorbate systems calculated in this study were in agreement with the amount of sorbate adsorbed by the adsorbent. The higher sorbate uptake should result in a higher K value (i.e. KN213X < KN2-5A < KN 2 -PCB, Table 17). This consistency did not hold in the cited studies. For instance, the uptake of CH4 on activated carbon (PCB) is much higher than on 1 3 X and on 5 A molecular sieve (Chlendy and Tondeyr, 1995) and therefore the value of KCH4 on activated carbon has to be much higher than on 1 3 X and 5 A . However, the opposite was true, as shown in Table 18. Considering these differences, the data from this study are within the values obtained from the cited works. The consistency and repeatability achieved in this study was remarkable, as demonstrated in the case of 1 3 X 8/12 and 1 3 X 16/40 mesh adsorbents (Table 17). Since both 1 3 X adsorbents are from the same manufacturer and they only differ in particle size, the K values for the particular sorbate are expected to be very close if not identical, for each adsorbent. The differences between values of K for different sorbates were within a range of only 3-7%. 86 Table 18: Summary of equilibrium adsorption constant K and equilibrium selectivities for C O 2 and CH 4 on various adsorbents from other studies Adsorbent T <K) N:o2 N:H4 N:O2^*N:H4 Reference f C B carbon (Ualgon) '2% 97.5 26 3.8 Ritter<fc Vang (1991) BPL carbon 301 51 18.6 2.7 Reich etal. (1990) CMS carbon MS 298 160 28 5.7 Kapoor& Yang (1989) 5A zeolites (pellets) 333 3000 12 250.0 Haq & Ruthven (1986) 5A zeolites (pellets) 298 9300 30.3 306.9 Rolnaik & Kobayashi4 5A zeolites 298 - 78 - Triebe etal.1 (1996) 13X zeolites (crystals) 298 7100 43.6 162.8 Rolnaik & Kobayashi4 13X zeolites 298 - 10 - Triebe et al.2(1996) Silicalite (crystals) 298 1700 460 3.7 Rees(1994) Silicalite 303 50 - - Choundary & Mayadevi3 H-ZSM-5 zeolite 303 95 - - Choundary & Mayadevi3 H-ZSM-8 zeolite 303 200 - - Choundary & Mayadevi3 H-Y zeolite 303 100 - - Choundary & Mayadevi3 'interpolated; 2extrapolated; 3obtained from specific retention volumes; 4unpublished data 4.6 D L and the Lumped Mass Transfer Coefficient The axial dispersion coefficient, D L , and the lumped mass transfer coefficient, LMTC, were estimated by the graphical method depicted in Figure 9, and from the calculated data for the first and second moments (p and c ), found in Tables 3 to 16. The results are shown in Table 19. The results for DL and LMTC, were calculated from the data obtained from the adsorption breakthrough curves for all the light sorbates on most of the adsorbents, as well as from the averaged data (obtained from the adsorption and desorption curves, as mentioned in Section 4.4) to correct their non-symmetry. For the lighter sorbates, which are strongly adsorbed by certain adsorbents (e.g. CO on the 87 VSA6, 5A and even 13X), the "linearization" by averaging the adsorption and desorption data was not good enough, and produced erroneous results for DL. The values for DL were 30-400% higher than the value for CO on PCB carbon (Table 17) or the value calculated from Equation 41. The higher the non-symmetry between the adsorption and desorption breakthrough curves, the more erroneous the results. Certain average results are shown in Figures 36 and 37. Figure 36 shows the determination of D L and the LMTC for N 2 on the 5 A and 13X 16/40, for CO on the PCB carbon, and for CH4 on the 5A molecular sieve. From this graph, it is evident that the axial dispersion coefficient, DL, is almost the same for N2, CO and CH4 on all these adsorbents, since all the lines are almost parallel to each other. This is expected, since the magnitudes of molecular diffusion, Dm, for N2-He, CO-He and CH4-He at 22°C are 0.693, 0.702, and 0.657 cm2/sec (Satterfield, 1970), respectively. From Figure 36, it is also notable that the low value of the intercept shows that the LMTC is insignificant in comparison to D L . Figure 37 shows the determination of D L and the LMTC on the 13X 16/40 and 13X 8/12 mesh for N2. It is obvious that the LMTC is higher on the 13X 8/12 due to larger particle size, while the D L are both about the same, as expected. Calculated values of DL, from the averaged experimental data, were within the expected range, and in satisfactory agreement with values predicted by Equation 36A, or from a similar correlation of Edwards and Richardson (1973): DL= 0.73Dm+vRp/[l+4.85 Dm/vRp] Equation 40 88 which for low gas velocities, reduces to: DL=0.73Dm Equation 41 For the systems used in this study at a temperature of about 22°C, D L values calculated from Equation 41 for CO, CH 4 and N 2 were 0.512, 0.493, and 0.506 cm2/sec, respectively. However, the D L calculated from the averaged data (from the adsorption and desorption runs) were slightly higher then those predicted by Equation 41. Similar findings were obtained in the study by Kumar et al. (1982), where the values for D L were also slightly on the high side, although they were calculated only from the adsorption data. This good agreement, found in these studies is somewhat in contrast with studies by Wakao et al. (1974) and Yi and Mancel (1972), who reported lower results for DL, and atributed that to the nature of moment analysis. The superficial velocities used throughout the experiments were in the range used in most of the cited studies, using the same approach (van Deemeter, 1956; Kumar et al., 1982; Ruthven, 1979). However, most of these studies were conducted using much longer adsorbent beds - 1 meter versus 20.5 cm (as used in this study). It appears that the bed length is unlikely to have any influence on the values of DL, although at the higher gas velocities and shorter retention times, the determination of the second moments were more sensitive to the tailing of the breakthrough curve. 89 Table 19. Axial dispersion coefficients, DL, and the lumped mass transfer resistance coefficients, LMTC, calculated from the first and second moments. Calculated from Adsorption Data From Averaged Data Adsorbent Adsorbate DL LMTC D L LMTC (cm2/sec) (cm /sec) (cm /sec) (cm2/sec) PCB CO 0.41 0.009 0.57 0.019 N2 0.42 0.008 CH4 0.34 0.007 13X 16/40 CO 0.34 0.007 N2 0.42 0.000 0.57 0.010 CH4 0.42 0.003 13X 8/12 CO 0.41 0.037 0.72 0.058 N2 0.45 0.032 0.60 0.050 CH4 0.49 0.025 VSA6 CO 0.03 0.002 2.30 0.182 N2 0.39 -0.004 CH4 0.27 0.003 0.66 0.016 5A CO 0.39 0.013 1.07 0.060 N2 0.45 0.017 0.66 0.030 CH4 0.48 0.018 0.64 0.027 CBV 780 CO 0.64 0.026 N2 0.53 0.001 CH4 0.51 0.028 MHSZ-1777 CO 0.28 0.023 0.65 0.052 N2 0.40 0.021 CH4 0.38 0.023 90 0 0 0.05 0.1 0.15 0.2 \N2 (sec2/cm2) ! x CO-PCB-AC-AVG • N2-13X-16/40-AVG • CH4-5A-AVG A N2-5A-AVG j Figure 36. Determination of D L and LMTC for N 2 in 5A and 13X 16/40; for CO in PCB carbon; and CFL in 5A molecular sieve (from averaged data). 0 _ ; i 0 0.05 0.1 0.15 0.2 0.25 Mv2 (sec2/cm2) I • N2-13X-16/40-AVG A N2-13X-8/12-AVG Figure 37. Determination of D L and LMTC for N 2 in 13X 8/12 mesh and 13X 16/40 mesh (from averaged data). 91 4.7 The Determination of the Langmuir Adsorption Isotherms The calculations of the Langmuir isotherm parameters, b and qs, were done from corresponding desorption breakthrough curves. The results are presented in Table 20 and the calculated Langmuir isotherms, based on these coefficients, are shown in Figure 38. The Langmuir isotherms obtained from the calculated coefficients, (b, qs), although having approximately the same saturation point as the isotherms obtained from the adsorbent manufactures (i.e. based on comparison for CO2 on 13X), appear to approach the saturation point much faster. This is probably due to the fact that the isotherms were calculated from the desorption curves, which were close to the linear range. Also, this may be due to the fact that the Langmuir isotherm may not provide the best fit for the data. In general, it is possible to extract the isotherm from the desorption breakthrough curves, but it turns out that the measurements of the desorption breakthrough curves have to be done under conditions (concentration) more within the range of the nonlinear portion of the isotherm. To be in the nonlinear range of an isotherm, the experiments had to be conducted by applying a much higher concentration of sorbate in the helium stream. However, a much higher concentration of sorbate would result in a significantly longer desorption time which would be beyond the limit of the data acquisition collection capability that was available for this study. Since we were within, or just barely outside of the linear range, these results should not be taken into serious consideration without confirmation by some more exact measurements, such as gravimetric measurements. 92 Table 20. Langmuir Isotherm Constants, b and qs Adsorbent Adsorbate I S Co z s/(l-s) b q s (m ) (m3/m3) (m) (m3/m3) VSA6-28D C0 2 -0.432 11.75 0.05 0.205 0.587 46.3 45.8 CBV-28D C0 2 -1.209 5.57 0.05 0.205 0.587 16.5 3.68 PCB-28D C0 2 -1.562 13.28 0.05 0.205 0.587 12.8 16.18 MHSZ-28D C0 2 -1.082 10.20 0.05 0.205 0.587 18.5 13.78 AA300-8D C0 2 -0.602 4.39 0.05 0.205 0.587 33.2 4.59 AA300-28D C0 2 -6.349 4.33 0.05 0.205 0.587 3.2 0.42 5A-28D C0 2 -0.184 9.68 0.05 0.205 0.587 108.6 72.8 13X-28D C0 2 -0.391 13.39 0.05 0.205 0.587 51.2 65.7 13X-6D-8/12 N 2 -2.566 5.48 0.05 0.205 0.587 7.8 1.68 13X-16/40 N 2 -2.594 5.45 0.05 0.205 0.587 7.7 1.64 5A-N2-8D N 2 -2.491 7.16 0.05 0.205 0.587 8.0 2.95 5A-N2-6D N 2 -2.582 7.36 0.05 0.205 0.587 7.7 3.01 VSA6-6D N 2 -2.714 5.61 0.05 0.205 0.587 7.4 1.66 13X-CO-20-D7 CO -2.783 11.27 0.07 0.205 0.587 5.1 9.15 VSA-CO-6D CO -0.811 6.77 0.05 0.205 0.587 24.7 8.09 VSA-CO-6D CO -0.819 6.83 0.05 0.205 0.587 24.4 8.17 5A-6D CO -1.560 11.37 0.05 0.205 0.587 12.8 11.87 5A-6D CH4 -2.546 8.71 0.05 0.205 0.587 7.9 4.27 VSA6-6D CH 4 -2.267 8.04 0.05 0.205 0.587 8.8 4.09 93 C02 I sotherms on Different A d s o r b e n t s VSA6-C2-28D x r .RV-7»n , PP.R-T.arhnn MHSZ-177 x AA-300-8D « 5A-MS _ | — 1 3 X | Figure 38. Langmuir Isotherms for C0 2 on VSA6, CBV780, PVB Carbon, MHSZ 177, AA300, 5A and 13X calculated from b and qs obtained from Table 17. 94 4.8 Summary of the experimental results Since all tests were done in the low Reynolds number regime, the determination of the axial dispersion coefficient, DL, and lumped mass transfer resistance, LMTC, was possible by applying the method depicted in Figure 9. The calculated values for D L and LMTC for all adsorbent/sorbate systems show that axial dispersion is dominant, while LMTC, and therefore external mass transfer, macropore and micropore resistances are relatively insignificant, although present. The higher LMTC for 13X 8/12 than for the 13X 16/40 mesh is due to a higher macropore or external film mass transfer resistance, since they are directly proportional to the square of the particle radius (Equation 7). The values of DL obtained in this study were in very good agreement with the values reported in the other studies, and when applied to the Levenspiel and Bischoff (Equation 40) solution for the breakthrough curve for the disperse plug flow of isothermal trace component system, it demonstrated very good agreement. The analysis of the first and second moments, HETP, as well as values of Henry's constant, K, led to the conclusion that the MHSZ-177 and VSA6 zeolites, respectively, are better choices for the first and second layer, than the commonly used activated carbon and 5A molecular sieve, for a layered adsorbent bed. Testing the layered beds comprised of MHSZ-77 and VSA6 adsorbents through a PSA cycle (process) and comparing them with the layered beds comprised of activated carbon and 5A molecular sieve, would confirm that MHSZ and VSA6 are a better choice for hydrogen separation. Also, cost and other factors must be considered before making the final decision. 95 Chapter 5: Conclusions and Recommendations for Future Work 5.1 Conclusions Screening adsorbents for a layered adsorbent bed, intended for use in a PSA process for hydrogen purification from SMR or POX product gases, using breakthrough experiments, was investigated in the present study. The main observations and conclusions resulting from this research are: • The breakthrough experiment is relatively simple and a worthy tool for screening adsorbents to be used in new applications, when most of the data regarding the particular adsorbent/sorbate system are not known. • Visual comparison of the breakthrough curves gives a quick general impression of how the particular adsorbent/adsorbate system will perform in a layered adsorbent bed. • The adsorption equilibrium constants, K, calculated from the mean retention times, or first moments, are of expected magnitude. From a ratio of these constants, the selectivity factor can be obtained for any pair of sorbates for a particular adsorbent. The increase in HETP for some of the adsorbent/adsorbate systems, with an increase in the linear gas velocity, suggests that at higher velocities the system enters the mass transfer limited range. This is confirmed by an increase in the value of overall mass 96 transfer resistance. By conducting further tests, it may be determined which mass transfer resistance (film, macropore or micropore), is controlling. By making the measurements over a range of particle sizes it would be possible to separate the film and macropore resistance from micropore resistance, since the micropore resistance is independent of gross particle size. • The axial dispersion coefficients, D L , extracted from the first and second moments, appear to be in agreement with findings in other studies using the same or similar method, or with findings estimated from an empirical equation. • The experimental results obtained in this study, as well as the calculated values of the first and second moments (u. and a2) and the values of the axial dispersion coefficient ( D L ) , were successfully applied to model the experimental breakthrough curve with the Levenspiel and Bischoff analytic solution, and with an empirical solution developed in this study. This very good agreement proves that the breakthrough experiment is a viable tool for obtaining useful information regarding a particular adsorbent/adsorbate system. • The VSA6 has a much sharper MTZ for all three light sorbates (CO, CH4 and N2) than the 5A molecular sieve, and therefore is a more promising candidate for the second adsorbent layer. • Due to the above-mentioned characteristics, the VSA6 adsorbent would be capable of more efficiently managing the light impurities. It seems possible that the required length of the second layer of the VSA6 might be significantly shorter than the length 97 of the 5A layer without affecting hydrogen purity and perhaps enhancing the hydrogen recovery and productivity. The present research strongly suggests that MHSZ177 and VSA6 are better choices, for a layered bed employed for hydrogen separation from SMR gas, than activated carbon and 5A molecular sieve. This research indicates that it is worth trying to explore new possibilities outside of the conventional adsorbents for a particular PSA process. Furthermore, the research sheds new light on how to select the appropriate adsorbents for the purification of a light component from multi-component gas mixtures, particularly when components of very different adsorptivity have to be separated from the gas mixture. Taking into account all the information obtained from the breakthrough data, for all adsorbents of interest, we can better predict how one adsorbent will behave in a layered adsorbent bed. This will allow us to select candidates for a layered bed, which may be more effective than others. 98 5.2 Recommendations for Future Work • Test other adsorbents from the MSHZ family of high silica zeolites, including silicalite. Since the MSHZ-177 zeolite in general showed to be comparable and even slightly better as a choice for the first adsorbent, than the commonly used activated carbon, it is expected that one of the adsorbents from the MHSZ family including silicalite may have a sharper MTZ and perhaps a slightly higher CO2 uptake. • Test VSA6 molecular sieve of different particle sizes to determine the influence of particle size on the MTZ. • Explore the suitability of a new generation of Li-exchanged adsorbents, which are replacing VSA6 in O2 separation from air, since they have a much higher uptake for N2, the component that is most likely to breakthrough first into the hydrogen product stream. • Test layered beds comprised of MHSZ-177 and VSA6 adsorbents, through a PSA cycle, and compare them with the beds comprised of activated carbon and 5A molecular sieve, as well as with ones comprised of activated carbon and VSA6 adsorbents. • To exactly determine the efficiency of a new layered bed, aside from designing one and testing it on a PSA cycle, it may be possible to undertake a relatively complex modeling of the PSA process involving layered beds using the data obtained in this study. 99 Literature Amphlett J. C, Klassen R. D., Mann R. F. and Peppley B. A., Methanol, Diesel Oil and Ethanol as Liquid Sources of Hydrogen for PEM Fuel Cells IECEC Vol. 1, pp. 1.221-1.1226, 1993. Appleby A. J. and Foulkes F. R., Fuel Cell Handbook, Van Nostrand Reinhold, New York, 1989; 1998. Astanovsky D. L., Astanovsky L. Z., Raikov B. S. and Korchaka N. I., Reactor for Steam Catalytic Hydrocarbon Conversion and Catalytic CO Conversion in Hydrogen Production, Int. J. Hydrogen Energy, Vol. 19, No. 8, pp. 677-681, 1994. CESHR, Center for Electrochemical Systems and Hydrogen Research, Texas Engineering Experiment Station, http://engineer.tamn.edu/tees/ceshr/center.html 1996. Chattoraj D. K., and Birdi K. S., Adsorption and the Gibbs Surface Excess, Plenum Press, New York, USA. Choundhary V. R. and Mayadevi S., Adsorption of Methane, Ethane, Ethylene, and Carbon Dioxide on X, Y, L, and M Zeolites Using a Gas Chromatography Pulse Technique, Separation Science and Technology, 28 (8), pp. 1595-1607,1993. Choundhary V. R. and Mayadevi S., Adsorption of Methane, Ethane, Ethylene, and Carbon Dioxide on High Silica Pentasil Zeolites and Zeolite-like Materials Using a Gas Chromatography Pulse Technique, Separation Science and Technology, 28 (13&14), pp. 2197-2209,1993. Cussler E. L., Diffusion Mass Transfer in Fluid Systems, Second Edition, Cambridge University Press, United Kingdom, 1997. Dominguez J. A., Psaras D. and LaClava A., Langmuir Kinetics as an Accurate Simulation of the rate of Adsorption of Oxygen and Nitrogen Mixtures an Non-Fickian 100 Carbon Molecular Sieves, in Adsorption and Ion Exchange, AIChE Symposium Series, pp 73-88, 1986. Dong F., Lou H., Kodama A., Goto M. and Hirose T., A New Concept in the Design of Pressure-Swing Adsorption Processes for Multicomponent Gas Mixtures, Ind. Eng. Chem. Res, 38, pp. 233-239, 1999. Dunne J. A., Rao M., Sircar S., Gorte R. J. and Myers A. L., Calorimetric Heats of Adsorption Isotherms. 2. 02, N2, Ar, C02, CH4, C2H6, and SF6 on NaX, H-ZSM-5, and Na-ZSM-5 Zeolites, Langmuir, Vol 12, pp. 5896-5904, 1996. Fox C. R. and Kennedy D. C, Conceptual Design of Adsorption Systems (in Adsorption Technology), Marcel Dekker, Inc., New York, USA, 1985. Frederick E. and Bernardin Jr., Experimental Design and Testing of Adsorption and Adsorbates, (in Adsorption Technology), Marcel Dekker, Inc., New York, USA, 1985. Fuderer A. and Rudelstorfer E., U.S. Patent 3,986,849 (1976), to Union Carbide Corporation. Fujimoto S., Isihara H. and Tsurino S., A Study on a Steam Reformer for Methanol, JSME International Journal, Vol 30, No 267, 1987. Garg D. R. and Ruthven D. M., The Performance of Molecular Sieve Adsorption Columns: Systems with Micropore Diffusion Control, Chemical Engineering Science, Vol. 29, pp. 571-581, 1974. Haq N. and Ruthven D. M., J. Coloidal Interface Sci. 112 (1), 1986. Haynes H. W. and Sarma P. N., A Model for the Application of Gas Chromatography to Measurements of Diffusion in Bidisperse Structured Catalysts, AIChE Journal, Vol. 19, No. 5, 1973. 101 Kapoor A. and Yang R. T, Kinetic Separation of Methane Carbon-Dioxide Mixture by Adsorption on Molecular -Sieve Carbon, Chem. Eng. Science, Vol. 44, No. 8, pp. 1723-1733, 1989. Karger J. and Ruthven D. M., Diffusion in Zeolites and Other Microporous Solids, John Wiley & Sons, Inc., New York, USA, 1992. Kenney C. N. and Kirby N. F., Pressure Swing Adsorption, in Zeolites: Science and Technology, edited by Ramoa R. et al., Martinus Nijhoff Publishers, The Hague, in cooperation with NATO Scientific Affairs Division, 1984. Klein G. and Vermulen T., Cyclic Performance of Layered Beds for Binary Ion Exchange, A.I.Ch.E. Symp. Ser. 71 (152), pp. 69-76, 1975. Kovach L. J., Gas-Phase Adsorption,'m Separation Techniques, second edition, Editor: Schweitzer P. A., McGraw-Hill, 1988. Kumar R., Duncan R. C. and Ruthven D. M, A Chromatographic Study of Diffusion of Single Components and Binary Mixtures of Gases in 4A and 5A Zeolites, The Canadian Journal of Chemical Engineering, Vol. 60, pp. 493-499, 1982. Langer G., Roether A., Roether K. P. and Gelbin D., International Journal of Head and Mass Transfer, 21, 751, Qaqaswe, 1978. Lee C-H, Yang J. and Ahn H., Effects of Carbon-to-Zeolite Ratio on Layered Bed H2 PSA for Coke Oven Gas, AIChE Journal, Vol. 45, No. 3, 1999. Levenspiel O. and Bischoff K. L. M., Advances in Chemical Engineering, Vol. 4, 95, Academic Press, New York, 1963. Lu Z. P., et al., Intraparticle Convection Effect on Pressurization and Blowdown of Adsorbents, AIChE Journal, Vol. 38, No. 6, 1992. 102 Malek A. and Farook S., Study of a Six-Bed Pressure Swing Adsorption Process, AIChE Journal, Vol. 43, No. 10, 1997. McKay G., Use of Adsorbents for the Removal of Pollutants from Wastewaters, CRC Press, Boca Raton, Florida, USA, 1996. Park J-H, Kim J-N, Cho S-H, Kim J-D, and Yang R. T., Adsorber Dynamics and Optimal Design of Layered Beds for Multicomponent Gas Adsorption, Chemical Engineering Science, 1998. Perry R. H. and Green D., Perry's Chemical Engineers' Handbook, Sixth Edition, Mc GrawHill, 1984. Pigorini G. and Le Van D. M., Equilibrium Theory for Pressure Swing Adsorption. 2. Purification and Enrichment in Layered Beds, Ind. Eng. Chem. Res., 36, pp. 2296-2305, 1997. Rees L. V. C, Exciting New Advances in Diffusion of Sorbates in Zeolites and Microporous Materials, Zeolites and Related Microporous Materials: State of the Art 1994, 84: 1133-1150, PartA-C, 1994. Reich M. H, et al., The Application of SAXS to Determine the Fractal Properties of Porous Carbon-Based Materials, Journal of Colloid and Interface Science, 135: (2), 1990. Ritter J. A. and Yang R. T, Air Purification and Vapor Recovery by Pressure Swing Adsorption - A Comparison of Silicalite and Activated Carbon, Chem. Eng. Comm., Vol. 108, 1991. Rudzinski W., Adsorption of Gases on Heterogeneous Surfaces, Academic Press, San Diego, CA, USA, 1992. Ruthven D. M., Raghavan N. S. and Hassan M. M., Adsorption and Diffusion of Nitrogen and Oxygen in a Carbon Molecular Sieve, Chem. Eng. Sci., Vol. 41, No. 5, pp. 1235-1332, 1986. 103 Ruthven D. M., Principles of Adsorption Processes, John Wiley & Sons, New York, 1984. Ruthven D. M,. Farooq S. and Knaebel K. S., Pressure Swing Adsorption, VHC Publisher Inc., New York, 1993. Ruthven D. M. and Kumur R., A Chromatographic Study of the Diffusion of N2, CH4 and Binary CH4-N2 Mixtures in 4A Molecular Sieve, The Canadian Journal of Chemical Engineering, Vol. 57, pp 342-348, 1979. Satterfield C. N., Mass Transfer in Heterogeneous Catalysis, M.I.T. Press, Cambridge, 1970. Schweitzer P. A., Handbook of Separation Technologies for Chemical Engineers, McGrawHill, 1988. Scott J., Zeolite Technology and Applications, Noyes Data Corporation, Park Ridge, New Jersey, U.S.A., 1980. Skarstrom, C. W., U.S. Patent No. 2,944,627 (July 12, 1960), to Exxon Research and Engineering. Slejko F. L., Adsorption Technology, Marcel Dekker, Inc., New York, USA, 1985. Smith O. J. and Westerberg A. W., The Optimal Design of Pressure Swing Adsorption Systems, Chemical Engineering Science, Vol. 46, No. 12, pp. 2967-2976, 1991. Stock R. and Rice C. B. F., Chromatographic Methods, Chapman and Hall, and Science Paperbacks, New York, 1988. 104 Strazhesko D. N., Adsorption and Adsorbents, John Wiley & Sons, New York, USA, 1973. Thaeron C, Theoretical and Experimental Study of a Parallel Passage Adsorbent Contactor in a Dual Piston Rapid Pressure Swing Adsorption System, Ph.D. Thesis, The University of New Brunswick, 1996. Triebe R. W., Tezel F. H., and Khulbe. K. C, Adsorption of Methane, Ethane and Ethylene on Molecular Sieve Zeolites, Gas Sep. Purif, Vol. 10, No 1, pp. 81-84, 1996. Van Deemter J. J., Zuiderweg F. J. and Kinkenberg A., The Mechanisms of Band Broadening in Linear, Nonideal Chromatography, Chem. Eng. Sci., 5, pp. 271-289, 1956 Van der Laan E. Th., Chem. Eng. Sci., 7, 187, 1958. Vermeulen T., Theory for Irreversible and Constant-Pattern Solid Diffusion, Industrial and Engineering Chemistry, Vol. 45, No.8, 1986. Wakao N. and Iida Y., Determination of Fluid Dispersion Coefficients in Packed Beds, Journal of Chemical Engineering of Japan, Vol. 7, No.6, 1974. Walter J. Weber, Jr, Adsorption Theory, Concepts and Models (in Adsorption Technology), Marcel Dekker, Inc., New York, USA, 1985. Warmuzinski K. and Tanczyk M., Multicomponent Pressure Swing Adsorption -Part I. Modeling of Large-scale PSA Instalations, Chemical Engineering and Processing, Vol. 36, pp. 89-99, 1997. Weber T. W. and Chakravorti R. K., Pore and Solid Diffusion Models for Fixed-Bed Adsorbers, AIChE Journal, Vol. 20, No. 2, 1974. Werner S. and Mersmann A., Single and Multicomponent Adsorption Equilibria of Carbon Dioxide, Nitrogen, Carbon Monoxide and Methane in Hydrogen Purification 105 Processes, Chem. Eng. Technol., 17, pp. 325-337, 1994. Yang J. and Lee C-H., Adsorption Dynamics of a Layered Bed PSA for H2 Recovery Coke Oven Gas, AIChE Journal, Vol. 44, No. 6, pp. 1325-1334, 1998. Yang J., Lee C-H and Chang J-W, Separation of Hydrogen Mixtures by a Two-Bed Pressure Swing Adsorption Process Using Zeolite 5A, Ind. Eng. Chem. Res, Vol. 36, pp. 2789-2798, 1997. Yang R. T., Gas Separation by Adsorption Processes, Butterworth Publishers, Stoneham, 1987. Yi Hua Ma and Mancel C, Diffusion Studies of C02, NO, N02, and S02 On Molecular Sieve Zeolites by Gas Chromatography, AIChE Journal, Vol. 18, No. 60, pp 1148-1153, 1972. Zvibel I., Gariepy R. L. and Schnitzer J. J., Fixed Bed Desorption Behavior of Gases with Non-Linear Equilibrium: Part I, AIChE Journal, Vol. 18, No. 6, 1972. 106 Appendix A Experimental Adsorption Breakthrough Curves 5A 8/12 - BT Curves for N2 Time s 5A-N6A 5A-N8A 5A-N12A-S 5A-N16A-S 5A-N20A 5A-N24A 5A-N32A 5AN-28A-L | Figure A l . Breakthrough curves for N 2 on 5A for 65, 80, 120, 160, 200, 240, 280 and 320 cc/min flows, respectively from right to left. 5A 8/12 - BT Curves for CH4 Time s 5AC1 "1 ° A «ri.1BA SAPI-OnA WM-M* .SAr.1.?HA SAr.i.3?A I Figure A2. Breakthrough curves for CH 4 on 5A for 65, 80, 120, 160, 200, 240, 280 and 320 cc/min flows, respectively from right to left. 108 5A 8/12- CO - Breakthrough Curves 0 250 500 750 1000 1250 1500 1750 2000 2250 2500 Time s i 5A-C06A-L 5A-C012A 5A-C016A 5A-CO20A 5A-C024A 5A-C028A 5A-C032A-S 5A-C08A-S Figure A3. Breakthrough curves for CO on 5A for 65, 80, 120, 160, 200, 240, 280 and 320 cc/min flows, respectively from right to left. 13X 8/12 - N 2 - Breakthrough Curves Time s 13N6A 13N8A 13N12A 13X16A 13N20A 13N24A 13N28A1 I Figure A4. Breakthrough curves for N 2 on 13X 8/12 mesh for 65, 80, 120, 160, 200, 240, and 280 cc/min flows, respectively from right to left. 109 13X - 8/12 Mesh - C H 4 - Breakthrough Curves ! 0 50 100 150 200 250 300 350 400 450 I ! Time s 13-C116A 13-C16A 13-C18A 13-C112A 13-C116A 13-C120A 13-C124A 13-C128A-S j Figure A5. Breakthrough curves for CH 4 on 13X 8/12 mesh for 65, 80, 120, 160, 200, 240, and 280 cc/min flows, respectively from right to left. 13X-8/12 - C O - Breakthrough Curve s 0 100 200 300 400 500 600 700 800 900 Time s 13- C06A 13- C08A 13- C012A 13- C016A 13- CO20A 13- C024A 13- C028A 13- C032AJ Figure A6. Breakthrough curves for CO on 13X 8/12 mesh for 65, 80, 120, 160, 200, 240, 280 and 320 cc/min flows, respectively from right to left. 110 13X 16/40 - N 2 - Breakthrough Curve s Time s 13N-6A1 13N-8A2 13N12A 13N-16A 13N-24A1 13N-28A1-sh 13N-2QA Figure A7. Breakthrough curves for N 2 on 13X 16/40 mesh for 65, 80, 120, 160, 200, 240, and 280 cc/min flows, respectively from right to left. 13X 16/40 - CH4 - Breakthrough C u r v e s 0 50 ' 100 150 200 250 300 350 400 Time s 13C1-6A2 13C1-8A 13C1-12A 13C1-16A 13C1-20A 13C1-24A 13C1-28A I Figure A8. Breakthrough curves for CH 4 on 13X 16/40 mesh for 65, 80, 120, 160, 200, 240, and 280 cc/min flows, respectively from right to left. I l l Figure A9. Breakthrough curves for CO on 13X 16/40 mesh for 65, 80, 120, 160, 200, 240, and 280 cc/min flows, respectively from right to left. VSA6 - N2 - Breakthrough Curves I 1 ! 0.9 j 0.8 j <u 0.7 to ! | 0.6 ! 3 0.5 ! Q 0.4 i H 0.3 ! 0.2 i 0.1 1 0 ! _ — i — I f r r r \ / A / i / i ! | • j j • / - / — / _ j — / 1 — : . \-i 1—-/ I i j 1 / : J y J i y : 1 0 25 50 75 100 125 150 175 200 225 250 . Time s I | V6N-6A V6N-8A V6N-12A V6N-16A V6N-20A V6N-24A-S V6N-28A1 ! j Figure A10. Breakthrough curves for N 2 on VSA6 mesh for 65, 80, 120, 160, 200, 240, and 280 cc/min flows, respectively from right to left. 112 Figure A l l . Breakthrough curves for CH 4 on 13X 8/12 mesh for 65, 80, 120, 160, 200, 240, 280 and 320 cc/min flows, respectively from right to left. «> 0.7 m | 0.6 S 0.5 Q 0.4 £ 0.3 0.2 0.1 0 VSA6 - CO - Breakthrough Curves ! ; ! • / ; • i / i ! : : / 1 " ! '• i / ! I . ;• I / 1 i i : / I j / i : ; ! ; y i 250 500 750 1000 Time s 1250 1500 1750 2000 .V6CO-6A . .V6CO-12A . .V6CO-16A . .V6CO-20A . .V6CO-24A . .V6CO-28A . .V6CO-32A Figure A12. Breakthrough curves for CO on VSA6 for 65, 120, 160, 200, 240, 280 and 320 cc/min flows, respectively from right to left. 113 CBV-780 - N2 Breakthrough Curve s i i 0 10 20 30 40 50 60 70 80 90 100 Time s C B N 6 A C B N 8 A C B N 1 2 A - L C B N 1 6 A - L C B N 1 2 A - L CBN12A-S C B N 2 4 A - L C B N 2 4 A - S I Figure A13. Breakthrough curves for N 2 on CBV-780 for 65, 80, 120, 160, 200, 240, 280 and 320 cc/min flows, respectively from right to left. | CBV-780 - C H 4 - Breakthrough Curve s ! 25 50 75 100 125 150 Time s C B C 1 - 6 A 2 C B C 1 - 8 A C B C 1 - 1 2 A C B C 1 - 1 6 A - L C B C 1 - 2 0 A C B C 1 - 2 4 A C B C 1 - 2 8 A Figure A14. Breakthrough curves for CFf4 on CBV-780 for 65, 80, 120, 160, 200, 240, and 280 cc/min flows, respectively from right to left. 114 CBV780 - C O - Breakthrough Curves 0 25 50 75 100 125 150 Time s CBCO-6A1 CBCO-8A CBCO-16A CBCO-20A CBCO-24A-S CBCO-28A-S CBCO-12A-M I Figure A15. Breakthrough curves for CO on CBV-780 for 65, 80, 120, 160, 200, 240, 280 and 320 cc/min flows, respectively from right to left. A A 14/40 - C 0 2 - Breakthrough Curve s 0 250 500 750 1000 1250 1500 1750 2000 2250 2500 2750 3000 3250 3500 Time s AAC2-6a AAC2-8A AAC2-12A AAC2-16A AAC2-20A AAC2-24A AAC2-28A AAC2-32A Figure A16. Breakthrough curves for CO2 on activated alumina AA-300 for 65, 80, 120, 160, 200, 240, 280 and 320 cc/min flows, respectively from right to left. 115 P C B - A C - N 2 - Breakthrough Curve s 1 0.9 0.8 0.7 0.6 0.5 ! - 0.4 0.3 0.2 0.1 I "> ! to i c ! O Q . i to Q O j t— I i i i I I .... j i ! 50 100 150 200 250 Time s 300 350 400 450 500 . P B - N - 6 A 1 P8-N-8A2 P B - - N - 1 2 A PB-N-16A PB-N-20A PB-N-24A P B - N - 2 8 A | I Figure A17. Breakthrough curves for N 2 on PCB activated carbon for 65, 80, 120, 160, 200, 240, 280 and 320 cc/min flows, respectively from right to left. — — i P C B - A C - C H 4 - Breakthrough Curve s j Time s RPP.1.KA PRfM-RA PRr.-l.17A PHm . l f iA PRfI1.?4A PRC1.38A I Figure A18. Breakthrough curves for CH 4 on PCB activated carbon for 65, 80, 120, 160, 240, and 280 cc/min flows, respectively from right to left. 116 P C B Act ivated Ca rbon - C O - Breakthrough Curve s 0 50 100 150 200 250 300 350 400 Time s PBC0-6A PBCO-8A PBCO-12A PBCO-16A PBCO-20A PBCO-24A PBCO-28A Figure A19. Breakthrough curves for CO on PCB activated carbon for 65, 80, 120, 160, 200, 240, and 280 cc/min flows, respectively from right to left. M H S Z 1 7 7 - N 2 - Breakthrough Curve s 0 50 100 150 200 250 I Time s MZ-N-6A1 MZ-N8A MZ-N-12A MZ-N16A1 MZ-N-20A MZ-N-24A M2-N-28A j Figure A20. Breakthrough curves for N 2 on MHZS-177 for 65, 80, 120, 160, 200, 240, and 280 cc/min flows, respectively from right to left. 117 M H S Z 1 7 7 - C H 4 - Breakthrough Curve s | Time s MZC1-6A M2C1-8A MZC1-12A MZC1-16A MZC1-20A MZC1-24A MZC1-28A Figure A21. Breakthrough curves for CH 4 on MHZS-177 for 65, 80, 120, 160, 200, 240, and 280 cc/min flows, respectively from right to left. M H S Z 177 - C O - Breakthrough Curve s Time s MZCO-6A MZCO-8A MZCO-12A MZCO-16A MZCO-20A MZCO-24A MZCO-28A Figure A22. Breakthrough curves for CO on MHZS-177 for 65, 80, 120, 160, 200, 240, and 280 cc/min flows, respectively from right to left. 118 Appendix B Equilibrium Adsorption Isotherms A d s o r b e n t s , u o p PRODUCT INFORMATION MOtSJV ADSORBENTS a product o< UOP VSA-6 Description/Application MOLS1V* type VSA-6 is a molecular sieve adsorbent specifically developed by UOP for use in systems which employ a vacuum pressure swing cycle for the generation of oxygen or nitrogen. VSA-6 was designed to minimize power consumption, while maintaining its high capacity and selectivity for nitrogen. VSA-6 is the fifth generation in a series of products devel-oped by UOP to meet our customers needs for better unit performance and efficiency via product improvement. Chemical Fotnula CaNa x((A10 2)y(Si0 2) z) • X H 20 Shipping Information 55 Gallon Drum, 300 pounds net weight or 2100 lb lined bags. Sample available. Handling and Loading See UOP Molecular Sieve Safety Brochure entitled. "Precautions and Safe Practices For Handling Molecular Sieve In Process Units" Form M-100B, or contact our Sales Engineers, see sales office listing on back. Typical Properties PARTICLE FORM 8x12 beads Nominal Pore Size 8 Angstroms Panicle Size 0.08 in. Density, settled 38-41 lb/ft3 Water Content, As shipped <1.5 wt% Crush Strength 4.0 lb, Equilibrium H 2 0 Capacity<'> 28.5 wt% Equilibrium C 0 2 Capacity*2) 19.5 wt% (1) Measured at 17.5 mm Hg H 2 0 and 25°C (2) Measured ac 250 mm Hg COj and 25°C Nitrogen and Oxygen Adsorption on VSA-6 PURE COMPONENT ISOTHERMS AT 0°C. 20°C. AND 40°C L oading, wt% Nitrogen 0°Q_ ^ 0* , - / -<*----« t * , w 1 ' , ' . . . . . . . r .T ..~~~r. - " 20°C_ -4 0 ° C i r~" T2o*c___—-0 5 10 15 20 25 30 Pressure, psia 120 ONI 0*01 lN33a3d 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0058626/manifest

Comment

Related Items