|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
DEVELOPMENT OF A NOVEL COMPACT REFORMER FOR PEMFC By JING SU A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009
2 2009 Jing Su
3 To my parents and my brother.
4 ACKNOWLEDGMENTS I would like to express my gratitude to my supervisor Prof. ChangWon Park, for his enthusiastic support and guidance throughout this study. He was always passionate and caring when I met difficulties. He was willing to share his experience and spent plenty of time in giving me instructions. I wish to thank my cochair Dr. Helena E. Hagelin Weaver for her abundant ly helpful and invaluable assistance and suggestion. I also wish to thank Dr. Jason E. Butler and Prof. Renwei Mei for their valuable recommendation as dissertation committee members. Also, I would like to express my sincere thank to Mr. Samuel D. Jones fo r his assistance in gas chromatographic measurements. Special thank to Dr. Jong Kwan Lee for his assistance in preparing the experimental setup This study was partially supported by NASA Lewis/Glenn Research Center, Grant NAG 32930, through the College of Engineering at the University of Florida
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ...............................................................................................................4 LIST OF TABLES ...........................................................................................................................7 LIST OF FIGURES .........................................................................................................................8 ABSTRACT ...................................................................................................................................10 CHAPTER 1 FUEL CELLS .........................................................................................................................12 1.1 Types of Fuel Cell ............................................................................................................12 1.1.1 Alkaline F uel C ell (AFC) .......................................................................................13 1.1.2 Phosphoric A cid F uel C ell (PAFC) ........................................................................13 1.1.3 Molten C arbonate F uel C ell (MCFC) .....................................................................14 1.1.4 Solid O xide F uel C ell (SOFC) ...............................................................................14 1.1.5 Proton Exchange Membrane Fuel Cell (PEMFC) ..................................................15 1.2 Fundamentals of PEMFC and the Use of a Hydrocarbon Fuel ........................................17 1.3 Reformers for H ydrogen S upply ......................................................................................22 1.3.1 Compact R eformer for PEMFC A pplication ..........................................................23 1.3.2 A Common Design Approach ................................................................................24 2 A NEW COMPACT REFORMER ........................................................................................27 2.1 Characteristics of t he P roposed R eformer ........................................................................27 2.2 Determination of the R eformer D imension ......................................................................29 2.2.1 Flow through a Packed Channel and the Pressure Field [86, 87] ...........................29 2.2.2 Dimension of a prototype reformer ........................................................................30 2.2.3 S imulation with more R ealistic R eaction K inetics .................................................37 3 EXPERIMENTAL STUDY ...................................................................................................46 3.1 Experimental S etup ...........................................................................................................46 3.1.1 Calibration of the Oxygen Feed Rate .....................................................................47 3.1.2 Conditions for Steam Reforming with Oxygen ......................................................50 3.1.3 Calibration of Product Gas Flow Rate ....................................................................51 3.1.4 Gas Chromatography Configuration and Operation Processes ..............................52 3.2 Experiments with G lass B ead as the P acking M aterial ....................................................54 3.3 Experiments with Cu/ZnO C atalyst ..................................................................................56 3.3.1 Steam Reforming Experiments ...............................................................................57 220.127.116.11 Low temperature experiment .......................................................................57 18.104.22.168 Temperature effect .......................................................................................61 22.214.171.124 Effect of methanol to water molar ratio of the feed .....................................63
6 3.3.2 Steam Reforming with Oxygen ..............................................................................66 126.96.36.199 Experimental results and theoretical predictions (19 channels of catalyst) ........66 188.8.131.52 Catalyst characterization ..............................................................................73 184.108.40.206 Experimental results and theoretical predictions (5 channels of catalyst) ...74 3.4 Heat T ra nsfer A spects of the R eformer ............................................................................78 4 C ONCLUSIONS AND FUTURE WORK .............................................................................86 APPENDIX A HEAT OF REACTION AT THE OPERATING TEMPERATURE OF THE REFORMER ...........................................................................................................................88 CPOX R eaction ...............................................................................................................88 Steam R eforming R eaction ..............................................................................................90 LIST OF REFERENCES ...............................................................................................................92 BIOGRAPHICAL SKETCH .........................................................................................................99
7 LIST OF TABLES Table page 11 Fuel cel l types and selected features .................................................................................16 12 Companies developing fuel processor technology ...........................................................22 21 Initial flow rates of all species ...........................................................................................32 31 Product gas composition as a function of feed rate at steam reforming conditions (reaction temperature: 220C water to methanol molar ratio=1:1 ) ...................................58 32 Product gas composition as a function of feed rate at steam reforming conditions (reaction temperature: 220C Water to methanol molar ratio=1:1, Catalyst3 was used for measurements, which is a mixture of fr esh and used catalyst from the second batch [Catalyst 2] ) .................................................................................................59 33 Product gas composition as a function of feed rate at steam reforming conditions (reaction temperature: 250C Water to methanol molar ratio=1:1 ) ..................................62 34 Product gas composition at various water to methanol molar ratio s (reformer temperature: 250C) with online GC measurements .........................................................65 35 Theoretical calculation of gas flow rates and compositions as well as maximum conversions ........................................................................................................................67 36 Product gas flow rate as a function of air flow rate at feed rates of 0.102 mol/h of methanol and 0.046 mol/h of water( with product gas composition from gas chromotograph) at T = 250 C. ...........................................................................................68 37 Theoretical calculation of gas compositions and conversions (hig her feed rate) ..............71 38 Product gas flow rate as a function of air flow rate at feed rates of 0.145 mol/h of methanol and 0.066 mol/h of water (with product gas composition from gas chromotograph) at T = 250 C. ...........................................................................................72 39 Theoretical calculation of gas compositions and conversions (various feed rates) ...........75 310 Theoretical cal culation of gas compositions and conversions (various feed rates for CPOX rxn only) .................................................................................................................76 311 Product gas flow rate as a function of air flow rate at two feed rates of methanol and water ( 0.102 and 0.046 mol/h, as well as 0.145 and 0.066 mol/h) and product gas composition from gas chromotography. ............................................................................77
8 LIST OF FIGURES Figure page 11 Schematic drawing of a hydrogen/oxygen fuel cell and its reactions based on the proton exchange membrane fuel cell (PEMFC)  ...........................................................17 12 Methanol conversion in methanol steam reaction as a function of reaction temperature (water content in feed=24, 43, 64 mol%, catalyst loading =1.0 g, GHSV = 1100 h1). Conversion data are from Choi et al. [ 56] ......................................................21 13 Schematic of a micro reformer with serpentine micro channels .......................................25 21 Schematic of the proposed reformer channel layout ..........................................................28 22 Molar flow rates of each chemical species alon g the flow path of the reformer (total length 95 cm) .....................................................................................................................33 23 Pressure profile along the flow path at various particle size (reformer channel Dia.= 2.5 mm) ..............................................................................................................................35 24 Pressure profile along the flow path ..................................................................................36 25 Details of the reformer body and end plate ........................................................................37 26 M olar flow rates of each chemical species along the flow path of the reformer calculated for kCPOX = 13388 L2/mol s gcat and kSR = 1.37 105 mol/s gcat kPa0.09 ......40 27 Molar flow rates of each chemical species along the flow path of the reformer calculated for kCPOX = 13.388 L2/mol s gcat a nd kSR = 1.37 105 mol/s gcatkPa0. 0 9 ...........41 28 Molar flow rates of each chemical species along th e flow path of the reformer calculated for kCPOX = 1.3388 L2/mol s gcat a nd kSR = 1.37 105 mol/s gcatkPa0. 0 9 ...........42 29 Molar flow rates of each chemical species along the flow path of the refor mer calculated for the same values of kCPOX a nd kSR for Figure 2 8, but for a small reformer length near the inlet to show the fast decline of oxygen .....................................42 210 Molar flow rates of each chemical species along the flow path of the reformer calculated for kCPOX = 1.3388 L2/mol s gcat a nd kSR =6.85 106 mol/s gcatkPa0. 0 9 to show the influence of slower reaction kinetics for the SR rxn. .........................................43 211 Molar flow rates of each chemical species along the flow path of the reformer calculated for kCPOX = 1.3388 L2/mol s gcat a nd kSR =2.74 105 mol/s gcatkPa0. 0 9 to show faster reaction kinetics for the SR rxn. .....................................................................43 212 Pressure profile along the flow path based on the reaction conditions for Figure 27 ......44
9 213 Pressure profile along the flow path based on the reaction conditi ons for Figure 211 ....45 31 S chematic of the e xperimental s etup .................................................................................46 32 Air flow rate at various values of the pressure at locatio n 3. .............................................49 33 Calibration curve for 500 cc flow meter. ...........................................................................52 34 Comparison between predicted and measured pressure drop ............................................56 35 Product gas flow rate vs. feed flow rate ( : catalyst batch 1, : catalyst batch 2) at 220C. The solid line indicates the product molar flow rate expected ..............................60 36 Conversion of methanol at various feed rate s and reformer temperatures. .......................62 37 Influence of water to methanol molar ration on the product molar flow rate at vario us methanol feed rates (at 250C). The solid line represents the product gas flow rate at 100% conversion of MeOH, assuming only steam reforming and no reverse water gas shift or methanol decomposition reactions. ......................................................63 38 Influence of water to methanol molar ratio on methanol conversion at various methanol feed rate s (at 250C). ..........................................................................................64 39 Catalysts in reformer channels after experiments ..............................................................74 310 Temporal variation of the reformer temperature when the electrical power supply is cut off .................................................................................................................................80 311 Temporal variation o f [(T T)/(TiT)] when the electrical power supply is cut off ........84
10 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the De gree of Doctor of Philosophy DEVELOPMENT OF A NOVEL COMPACT REFORMER FOR PEMFC By Jing Su May 2009 Chair: Chang Won Park Cochair: Helena Hagelin Weaver Major: Chemical Engineering A compact reformer to produce hydrogen for portable fuel cell applica tions is presented in this thesis Th is reformer is a conventional single path tubular reactor type that is packed with granular catalyst particles. The catalyst is used to induce catalytic partial oxidation reaction (CPOX) and steam reforming reaction (SR ) in series using a mixture of methanol, water and oxygen as the feed. One of the important features of the reformer is the interlaced flow path for efficient heat transfer between the reactor sections where the endothermic SR reaction and the exothermic C POX reaction may be taking place respectively. Flow simulation has been conducted to determine various design features including the dimension of the reformer that can be used for a PEMFC with the energy production capacity of 10 Watt at 50% efficiency. S ubsequently, a prototype reformer has been built to confirm the simulation results and to assess the efficacies of new design features. Experiment with a commercial catalyst (CuO/ZnO/Al2O3) showed a methanol conversion of about 85% at the feed rate enough to generate energy production capacity of 10 Watt at 50% efficiency with an operating temperature of 250 a hydrogen concentration of approximately 7273% in the product gas when SR is the only hydrogen generation reaction.
11 When oxygen is f ed to the reformer a s an additional reactant, CPOX reaction also takes place and the hydrogen concentration decreases slightly as a result. The current study showed a hydrogen concentration of approximately 67 % 72% at 250C at various oxygen feed rates Th ese hydrogen concentrations in the product gas are in reasonable agreement with theoretical predictions and indicate that the CPOX reaction does indeed take place before the SR reaction The experiments indicated that only 5 of the maximum 19 reformer chan nels filled with catalyst are not sufficient to complete both the CPOX and SR reaction s under the conditions used in this study. The experimental results support that the prototype reformer is capable of meeting the preset requirements to produce enough hydrogen for 10W PEMFC application. The results also suggest that proper insulation of the reformer is essential for self sustainability.
12 CHAPTER 1 FUEL CELLS A fuel cell is a device that can convert chemical energy directly in to electrical energy The fir st fuel cell may be dated back to 1839 when Sir William Grove a British jurist and physicist invented a system which converted chemical energy to electrical energy by reacting hydrogen and oxygen on platinum electrodes in a sulfuric acid solution [1 3] At that time the research o n fuel cell was not as popular because the primary energy sources were abundant and readily available at a low cost In the late 20th century, fuel cell emerged as an important technological area because of its capability to con vert chemical energy directly in to electrical energy with a higher efficiency than the conventional thermomechanical electric devices  T he overall efficiency of a fuel cell for energy production is about two times higher than those of conventional combustion engines. Furthermore fuel cell s are environment friendly as they use pure hydrogen as fuel and generate only water as a byproduct without producing any greenhouse gases. R enewable energy sources including wind, water and solar energy will continue to be developed quite extensively in the next few decades. However, these renewable energy resources may not fulfill the entire demand, and f uel cells may work as a complimentary energy source to those [3 7] 1.1 Types of F uel C ell Fuel cells may be classi fied into six different types. Th ey are (1) Alkaline Fuel Cell (AFC), (2) Phosphoric A cid F uel C ell (PAFC) (3) Molten Carbonate Fuel Cell (MCFC), (4) Solid Oxide Fuel Cell (SOFC), (5) Proton Exchange Membrane Fuel Cell (PEMFC) and (6) D irect M ethanol F uel C ell (DMFC). DMFC is in fact very similar to PEMFC thus it may belong to the same category as PEMFC [3 7] Th is classification is based on the electrolyte s used for each type of fuel cell s. Some important features of each type of fuel cells are discussed below :
13 1.1.1 Alkaline F uel C ell (AFC) The alkaline fuel cell (AFC) is one of the most well developed fuel cell types which was used for the Apollo mission i n 1960s [6,8] In an AFC syst em, hydrogen and oxygen supply not only electric power but also heat and water. Alkaline fuel cell for which Potassium hydroxide (usually 3045 wt. % in concentration) is used as the electrolyte, operates at a low temperature below 100C. AFC has the high est electrical efficiency among all fuel cell types However, it is very sensitive to impurities such as carbon dioxide that can react with KOH and produce carbonates. The carbonates are destructive to the electrolyte and the performance of the fuel cell decreases sharply as a result Thus it is necessary to use pure hydrogen and oxygen as reactants or to use an extra unit, such as an iron sponge system, to remove carbon dioxide from the system Many different catalysts may meet the requirement of the AFC electrode Thus, inexpensive catalysts can be used lower ing the manufactur ing cost of AFC compared to other types that have limited options such as PEMFC [9 11] 1.1.2 P hosphoric A cid F uel C ell (PAFC) PAFC may be the most broadly commercialized fuel cell technology. PAFC based power plants have been constructed in many places around the world with outputs ranging from 50200 kW for general use to 520 MW for large plants that can supply electric ity heat, and hot water to various buildings in town includin g hospitals  P hosphoric acid, that is the electrolyte for P AFCs begins to solidify at about 40C This makes the startup of PAFC to be difficult thereby restricting the PAFCs to operate continuous ly  The electrochemical reactions of PAFC take place on highly dispersed electro catalyst particles which is supported on carbon black. Platinum or its alloys are used as the catalyst for both electrodes; the anode and the cathode. The advantages of PAFC s include thermal stability, chemical stability and l ow volatility of phosphoric acid at the operating temperature of
14 150~ 200C. PAFC is also considered as a low temperature fuel cell as AFC, PEMFC and DMFC due to its low operating temperature [10 12] 1.1.3 Molten C arbonate F uel C ell (MCFC) As its name ind icates, MCFC use s a molten carbonate as the electrolyte which is stabilized by an alumina based matrix  The typical operating temperature of MCFC is 600~ 700C, which is much higher than those of AFC or PEMFC [8, 10] T his high temperature is required i n order to achieve sufficient conductivity of the carbonate electrolyte. The high operation temperature also makes it possible to utilize the waste heat properly Due to the high operating temperature, nonprecious metals and oxides such as Ni and NiO can be used as the catalysts for both electrodes thereby making it cost efficient [13 15] Unlike most other fuel cells, hydrocarbon fuels can be used as the fuel for MCFCs as the hydrocarbon fuel can be converted to hydrogen within the fuel cell itself by an internal reforming process. Another advantage of MCFCs is that they are quite resistant to impurities such as carbon monoxide or carbon dioxide achieving the electrical efficiency of 50% to 70% with a combined cycle system [16, 17] The primary disadvantage of MCFC is its poorer durability The materials of construction for MCFCs should meet the requirements imposed by the high operating temperature. In addition, they should resistant to the corrosion by the liquid electrolyte Otherwise, leakage of the electrolyte can be a major problem [18, 19] 1.1.4 Solid O xide F uel C ell ( SOFC) The solid oxide electrolyte s for SOFCs that is usually Y2O3stablilized ZrO2, are more stable than those of MCFC s [3 6 ] Furthermore, SOFCs are more durable than MCFCs because the corrosion issue caused by solid oxide electrolytes is less problematic than that by the liquid electrolytes for MCFCs The electrodes of SOFCs are typically made of Ni ZrO2 ceramal and Sr doped LaMnO3 [3 6]. As the operating temperature of SOFC is high, expensive catalyst s are not
15 necessary thereby making i t economical than other types such as PEMFC that require expensive precious metal catalysts [20, 21] Due to the high operation temperature of SOFC that is in the range of 700~ 1000C similar requirements are imposed on the materials of construction as th ose for MCFC. Furthermore, internal reforming process is also possible with SOFC thereby making it possible to use hydrocarbon fuels without a separate reformer unit to convert hydrocarbon to hydrogen [22, 23] 1.1.5 P roton Exchange Membrane Fuel C ell (PE MFC) PEMFC i s named after the electrolyte which is a proton exchange membrane. The same acronym also refer s to polymer electrolyte membrane fuel cell  PEMFCs are known to have a high power density at a very low operating temperature These advantages make PEMFC more suitable for portable applications than any other fuel cells. The most important milestone in the development of PEMFC may be the invention of the membrane called Nafion by DuPont. M ain component of the Nafion membrane is polytetrafluoroethylene (PTFE), and sulfonation of the material imparts high acidity to the membrane and also provide s high conductivity [10, 11] The electrodes for PEMFCs are generally porous for efficient diffusion of gaseous components and for better contact of hydroge n and oxygen with the noble metal catalyst that is usually platinum [24 26] As the working temperature is as low as 60 ~ 120C, PEMFCs can achieve a much faster startup than MCFCs or SOFCs. While pure hydrogen is the suitable fuel for PEMFCs, a special PEMFC can use methanol directly as the fuel without a separate fuel processor. This special type of PEMFC is known as Direct M ethanol F uel C ell (DMFC) [27 29] DMFC has a similar operating temperature as PEMFC although it operates at a slightly higher temperature to improve power density. DMFCs use methanol solution in water (generally 1~2M) as the fuel  The methanol reacts with water at anode to release protons, electrons and carbon dioxide whereas
16 hydrogen is oxidized at anode generating protons and elect rons in PEMFC. DMFC has a lower efficiency because of permeation of the methanol through the membrane, which is called methanol crossover  This methanol crossover is the most serious deficiency of DMFC. Methanol can lead to a mixed potential through the interference of methanol oxidation reaction with the oxygen reduction reaction at cathode. While many factors are related with the methanol crossover, membrane itself is the main focus of research to resolve the issue. For DMFCs, a thicker membrane is usually better to decrease the methanol crossover. As the methanol is consumed at anode, a high performance anode which can oxidize as much methanol in feed will also decrease crossover. Another approach is to develop a methanol tolerant catalyst for cath ode [3, 30]. AFC, PAFC and MCFC have been well developed already whereas further development efforts for SOFC and PEMFC /DMFC are still ongoing. A brief comparison between various types of fuel cells is given in Table 1 1. Table 1 1 Fuel cell types and selected features [ 8 ] Type Operating t emperature ( C) Fuel Electrolyte Mobile ion PEM FC (P olymer E lectrolyte M embrane F uel C ell ) 70110 H2, CH3OH Sulfonated polymers (NafionTM) (H2O)nH+ AFC (A lkali F uel C ell ) 100 250 H 2 Aqueous KOH OH PAFC (P hosphoric A c id F uel C ell ) 150250 H2 H3PO4 H+ MCFC (M olten C arbonate F uel C ell ) 500700 Hydrocarbons, CO (Na, K)2CO3 CO3 2 SOFC (S olid O xide F uel C ell ) 7001000 Hydrocarbons, CO (Zr, Y)O2 O2
17 1.2 Fundamentals of PEMFC and the Use of a Hydrocarbon Fuel As indicated schematically in Figure 1 1, PEMFC which consists of three major parts; anode, cathode and electrolyte. At the anode, hydrogen is oxidized into protons releasing electrons : e H H 2 22 ( 11) T he electrons flow from the anode to the cathode through an external circuit (or a conductor); hence generat ing electrical current to drive an external load The protons migrate toward the cathode through the proton exchange membrane Oxygen is reduced at the cathode by combining with the electrons from the anode, and then form s water by reacting with the protons from the anode through the proton exchange membrane : O H e H O2 22 2 2 / 1 ( 12) Figure 11 Schematic draw ing of a hydrogen/oxygen fuel cell and its reactions based on the proton exchange membrane fuel cell (PEMFC) 
18 PEMFC offer s much higher power density than other types of fuel cells. In addition the intrinsic properties of the materials used for PEMFC m ake them operate at a low temperature between 60 ~ 120C This low operating temperature allows quick startup and rapid loadresponse which are main advantages of PEMFC Because of its low temperature operability, PEMFC is favored as a portable power source for the applications as in aviation, automobile, and consumer electronics (e.g. laptop, cell phones, etc). In fact, only PEMFCs can meet the requirements for such portable applications which are compact size and lightweight besides low temperature operability The best fuel for PEMFC is pure hydrogen as for all other type of fuel cells However, the storage of hydrogen as well as the portability of the hydrogen storage system s are rather problematic for smallsize mobile applications. Thus, hydrocarbons, especially methanol or methane, are recognized as more practical choices as a fuel for PEMFCs [ 31, 32] Use of methanol or other hydrocarbon fuels, however, requires a reformer unit by which hydrogen is produced. It is an additional unit that not only incr eases the system size but also creates other technical difficulties such as CO poisoning of catalyst. Recent advances in various areas including catalyst technology have resolved many of the difficulties although there still exist some obstacles that requi re research efforts at a fundamental level. Between methanol and methane, methanol may be more favorable because it is in liquid phase at standard condition [ 33] It is also much more difficult to reform methane Steam reforming of methane requires more en ergy and significantly higher temperature s compared to methanol reforming [ 3437] Therefore, the current research has focus ed on reforming methanol. The well known reaction to produce hydrogen from methanol is the st eam reforming reaction: [ 38 41]
19 CH3OH + H2O CO2 + 3H2 ( H0 = 130.9 kJ/mol) ( 13) The steam reforming of methanol is commonly carried out over commercially available low temperature shift copper based catalysts [ 42, 43] The s team reforming reaction can be followed by a reverse shift reacti on which establishes the thermodynamic equilibrium: CO2 + H2 CO + H2O ( 14) Another side reaction that is known to occur is the decomposition of methanol [6 44] : CH3OH CO + 2H2 ( 15) This reaction is known to occur when there exist excessive amount s of methanol in the system i.e. at low H2O and O2 concentrations [ 45] CO generated by th e reaction (1 4) or (1 5) should be removed as it acts as a poison for the electrode catalyst in the fuel cell. The steam reforming reaction occurs at about 200~300C and it is an endothermic reaction requiring external heat supply. Although the heat can be supplied by an external source through the partial use of the electric power generated by the fuel cell, t he catalytic partial oxidation (CPOX) of methanol that is highly exothermic (autothermal reforming) can be combined with the steam reforming reaction : [ 4652] CH3OH + 1/2 O2 CO2 + 2H2 ( H0 = 154.9 kJ/mol) ( 16) Among the four reactions described above, only three reactions are linearly independent. Thus, i n analyzing the flow rates and output gas compositions using material balances, only three reaction stoichiomet ries may be used (i.e., reactions 13 and 1 6 plus reaction 14 or 1 5). The combustion reaction or total oxidation of methanol is also a side re action that may take place in the presence of excess oxygen : 2CH3OH + 3O2 2CO2 + 4H2O (1 7)
20 This reaction consumes methanol and oxygen without producing hydrogen. Thus, it should be avoided, if at all possible The formaldehyde, HCHO may also be formed by the following side reaction s : [ 53, 54, 55] CH3OH +1/2O2 H C H O + H2O (1 8) CH3OH H C H O + H2 (1 9) This first formaldehyde forming reaction consumes methanol and oxygen without producing any hydrogen and should also be avoided. While the dehydrogenation of methanol does produce hydrogen, the formaldehyde must be reformed in a second step to obtain the maximum hydrogen production rate. Because the steam reforming reaction is reversible, a higher ratio of water to methanol may be preferable in order to obtain a high conversion of methanol at a low temperature. The s tudy of Cho i et al [ 56] in fact, showed that the methanol conversion increased with the water content at a fixed reaction temperature as indicated in Figure 12. A t a certain G as H ourly S pace V elocity ( GHSV = volume of gas over volume of catalyst in one hour) a near complete conversion of methanol is achieved at a lower temperature when the water content is higher [ 56] Furthermore, the methanol decomposition and the reverse water gas shift reactions are suppressed in the presence of excess water [ 57, 58]. Thus, a higher water content in the feed is desired to decrease the operation temperature of a reformer while maintaining a high conversion of methanol and suppress the CO forming reactions However, by adding m ore water than the stoichiometric need, t he reactor will require more energy to evaporate the excessive water. In actual reactor design, the improvement of conversion and selectivity must be sufficient to offset the additional energy load for evaporating the additional water for this to be econom ical.
21 T he reverse shift reaction (Reaction 14) and the methanol decomposition reaction (Reaction 1 5) produce carbon monoxide that acts as a poison to the catalyst in the anode for the hydrogen oxidation reaction. The gas cleanup may be achieved by selective oxidation using a membrane where CO is removed from H2 by selective oxidation using a catalytically active membrane [ 59] or methanation [ 60] Selective oxidation of carbon monoxide can be achieved by alumina supported Ru/Rh catalyst which provides ne ar complete conversion of CO at a temperature as low as 100C [ 61 ] Therefore, a reformer may have to include a gas clean up unit at the end of the reaction path where the steam reforming and/or the partial oxidation reactions are taking place. 0 20 40 60 80 100 100 150 200 250 300 350 Temperature (Deg. C) Methanol conversion (%) 64% 43% 24% Fig ure 12 Methanol conversion in methanol steam reaction as a function of reaction temperature (water content in feed=24, 43, 64 mol%, catalyst loading =1.0 g, GHSV = 1100 h1). Conversion data are from Choi et al. [ 56]
22 Since the feed is in a liquid form, it has to be vaporized prior to the reactions. Therefore, the reformer should include three functional units in series for feed evaporation, reforming and CPOX reactions, and gas clean up. 1.3 Reformers for H ydrogen S upply In recent years, many companies have co nducted extensive research on fuel processors to produce hydrogen from hydrocarbon fuels (Table 12). These are mainly for the PEMFC for automobile application. Table 12 Companies developing fuel processor technology [ 6264] Corporation Fuel type Primary fuel processor CO conversion processes Tech. status (max. capacity ) Methanol Gasoline SR POX 1 WGS 2 PROX 3 Epyx Second First 50 kW Daimler Benz First Second 50 kW GM First Second 30 kW Honda Sole HBT Sole 7 42kW IFC First Second 100 kW JM First Second 6 kW Mitsubishi Sole 10 kW Nissan Sole Toyota Sole 25 kW 1POX: Partial Oxidation Reaction 2WGS: Water Gas Shift Reaction 3PROX: Preferential Oxidation Reaction Epyx: an Arthur D. Little company one of the leading developers of compact fuel processing system solutions for fuel cell applicatio ns in the micro power and transportation markets. HBT: Hydrogen Burner Technology, Inc. IFC: International Fuel Cells JM: Johnson Matthey Although the fuel processors described in Table 12 are of large scale, smaller ones are also of interest for nu me rous practical applications. Main focus of the present study is on the small scale compact reformers.
23 1.3.1 Compact R eformer for PEMFC A pplication A commercially viable portable reformer requires high efficiency in terms of conversion of hydrocarbon fuel and thermal management, compactness, and easy integration with the fuel cell. Numerous researchers have been working on meeting these requirements, and the current research has been also on investigating and assessing a new design idea that satisfies all these requirements. Patil et al. [ 65] presented a 40 watt methanol fuel processor which consist ed of a vaporizer, steam reformer, and recuperative heat exchanger w hile the weight of entire system was less than 80g. Yamamoto et al. [ 66] reported a multilay ered micro channel reactor which was capable of combined functions as a vaporizer a methanol reformer, a carbon monoxide remov al device, and a catalytic combustor to supply heat for the endothermic steam reforming reaction. T heir micro channel reactor was made of glass with a power generation rate of about 10W for a laptop PC Park et al. [ 67] disclosed an integrated fuel cell system with a methanol fuel processor that consist ed of a fuel vaporizer, heat exchanger, reformer, and catalytic combustor. Kundu et al. [ 68] report ed a Micro Electro Mechanical System ( MEMS ) based micro reformer which was intended for cellular phone application. Kwon et al. [ 69] developed a micro reformer using a silicon wafer as the reactor substrate based on microelectronics fabri cation technology, whereas Re nse et al. [ 70] demonstrat ed a micro reformer using a thin metal plate as the reactor substrate. In the work of Reuse and coworkers, total oxidation of methanol was used to provide the heat needed for the hydrogen generating st eam reforming reaction by adopting a twopassage reactor system for the two separate reactions
24 1.3.2 A Common Design Approach Although most industrial research [ 71 ] on microreformers is rather secretive without much published information, it seems that one of the most popular approaches is to adopt a tubular reactor type that consists of serpentine micro channels etched on a thin substrate such as silicon wafer (Fig ure 13). The cross section of the reactor path is a rectangular shape with its dimension in the order of hundred micrometers and the inner surface of the rectangular channels is coated with a catalyst for the steam reforming reaction. The catalysts (Cu/ZnO type for the steam reforming and Ru/Rh type for the gas clean up both supported on alumi na) are deposited on the surface along the groove by solution coating or sputtering [ 72, 73 ] Temperature control is achieved by heating the substrate plate. It was shown that when the grooved cross section was about 200 by 200 micrometers with the total length of about 30 cm, hydrogen could be produced at a rate of about 1L/hr at standard condition. Because a production rate larger than 15L/hr is required for a 10W usage assuming 3040% efficiency, many of the thin single meander reactors need to be stack ed. When t he hydrocarbon fuel (e.g. methanol solution of water) is fed into the reactor inlet, the reaction occurs and generates hydrogen as the hydrocarbon fuel flows through the serpentine reactor path.
25 stack inlet (side view) (top view) outlet Fig ure 13 Schematic of a micro reformer with se rpentine micro channels The reactor is typically kept at about 200~250C for the reaction, and heat is provided continuously from an external power source like a separate battery unit because the steam reforming reaction is endothermic. This type of approa ch has been demonstrated by Motorola laboratory, Casio and a few other institutions for micro PEMFC applications. However, such reformers are only prototypes that still require breakthroughs for commercial use. Major problems or difficulties with this appr oach include [ 7477] 1. Difficulty in catalyst incorporation and low catalyst efficiency due to small contact area 2. Difficulty in section wise temperature control. 3. Difficulty in leak proof stacking. 4. Difficulty in feed stream branching with uniform flow rate fo r each stack. 5. Difficulty in catalyst replacement after catalyst deactivation Furthermore, because the steam reforming reaction is a gas phase reaction, the reactants should be vaporized prior to be injected into the reactor and that also requires an ext ernal energy source [ 78, 79]. Although it is claimed that the battery which supplies heat for the reaction can
26 be recharged using the energy generated by the fuel cell itself, the energy consumption for the recharging becomes a substantial portion of the e nergy generated by fuel cell [ 8082] In the present study a new design idea is explored that may resolve most of the difficulties described above As it will be described in more detail in the following sections, t he new design concept appears to offer va rious advantages as described above, and the current study is to demonstrate its viability through process modeling and experiment s
27 C HAPTER 2 A NEW COMPACT REFORM ER 2.1 Characteristics of t he P roposed R eformer It has bee n described in the previous section that one of the difficulties in making a compact reform er is the difficulty in catalyst incorporation and low catalyst efficiency associated with small contact area when surface coating method is applied. One idea to inc rease the catalyst efficiency is the conventional tubular reactor in which granular catalyst particles are packed in cylindrical reaction channel s In order to make a reformer to be self sustaining without an external energy source, the exothermic CPOX (re action 1 6) may be induced along with the endothermic steam reforming reaction. A proposed reformer design that may be capable of achieving these two requirements (i.e., improved catalyst efficiency and self sustainability) is shown schematically in Figure 21. It consists of two types of basic plates; the reactor plate and the end plates (front and rear plate in Fig ure 21). The straight cylindrical channels in the reactor plate are filled with catalyst particles and the steam reforming and CPOX reactions take place in the channels. The front and the rear plates have many grooves that connect two neighboring cylindrical channels thereby changing the flow direction back and forth and forming a long meandering reaction path. The back and forth layout of channels make full utilization of space, which is necessary for the reformer to be compact enough for portable application of PEMFC. The most important design feature of this reformer is the layout of the grooves that enable the interlacing of the flow channels The numbers, 1 through 29, written on each cylindrical channel indicate the sequential order of the flow path through which the reactant will flow. Reactants are fed into channel #1 of the reactor through the hole (#1) in the front plate. Then they flow through the channels in the top row #2 to #10 where they are vaporized. ( Although not
28 shown in the figure, the reformer also uses a thin film heater attached on the top surface of the reactor plate. In a complete portable fuel cell, a start up unit (e.g. a battery) is necessary to heat up the reformer to its working temperature to allow the reaction s to take place [8 3, 84 ] The heater is to supply heat for the evaporation of methanol/water mixture and heat up the reformer to an appropriate operation tempera ture for the initial startup.) The vaporized reactants then flow into channel #11 and flow sequentially from #11 to #18. The slant grooves in the end plates make these flow channels #11 through #18 to be alternating in rows and skipping immediately neighboring channels in the same row. The reactants then flow into channel #19 and sequentially from #19 through #28. Channels #19 through #28 are again alternating in rows and skipping immediately neighboring channels in the same row due to the end plate groove layout. By then, the reactions are complete and the reaction products flow through channel #29 and exit the reactor through the hole (#29) in the rear plate. We may notice that channels #11 through #18 are interlaced with the channels #19 through #28. Figure 21 Schematic of the proposed reformer channel layout
29 The feed contains oxygen which is supplied from the air, and the partial oxidation reaction is dominant over the steam reforming reac tion in the presence of oxygen [ 24, 85]. Thus, until the feed oxygen is depleted, the exothermic CPOX is the main reaction that occurs in the channels #11 through #18 where heat is generated. (We assume the CPOX reaction will be completed in the channels #11 through #18, however, it may need more or less channels for the reaction to be completed depending on the catalyst activity.) Once oxygen is depleted, only the endothermic steam reforming reaction occurs in the channels #19 through #28. Because channels #11 through #18 are interlaced with channels #19 through #28, the heat generated in #11 through #18 are transferred effectively to channels #19 through #28. T his interlac ing of the channels that enable efficie nt heat transfer may be one of the most novel features of the reformer design. Furthermore, this reformer is of a modular structure in that multiple units can be combined either vertically or horizontally depending on the required hydrogen production rate. 2.2 Determination of the R eformer D imension 2.2.1 Flow through a P acked C hannel and the P ressure F ield [ 86, 87] For a low Reynolds number flow the pressure drop through a packed bed can be described by the KozenyCarman equation which describes the linear relationship between the p ressure drop and the superficial velocity of the fluid. 3 2 2 2 0) 1 ( 150 p sD V L p ( 21) H ere L p is the pressure drop per unit length, is the viscosity of fluid, is the porosity, s is the sphericity, Dp is the characteristic length scale for the packing material V0 is the superficial velocity which is the volumetric flow rate divided by the cross sectional area of the channel. This
30 equation is equivalent to the Darcys law in that the average velocity (or flow rate) is proportional to the pressure gradient. When the Reynolds number is very large, inertial force is dominant over viscous force and t he pressure drop becomes proportional to the square of average velocity as described by the Burke Plummer equation: 3 2 0) 1 ( 75 1 p sD V L p ( 22) For a flow at a moderate Reynolds number, the Ergun equation is known to describe the relationship between the average velocity and the pressure d rop. The Ergun equation is nothing but the simple addition of the KozenyCarman equation and the Burke Plummer equation: 3 2 0 3 2 2 2 0) 1 ( 75 1 ) 1 ( 150 p s p sD V D V L p (2 3) The Ergun equation has been used in describing the flow of current interest in which the flow is mostly gaseous and the chemical reactions described in section 1.2 are taking place. 2.2.2 Dimension of a prototype reformer One of the design criteria that de fines the size of the reformer is the hydrogen generation rate. In the present study, main focus will be on a micro reformer that can produce a suitable amount of hydrogen for a 10W PEMFC application. Because the reactants are in gas phase, gaseous flow through a packed tube has been investigated using the Ergun equation to determine the suitable channel di mension and the size of the catalyst particles. The Ergun equation (Equation 2 3) is for a flow of an incompressible fluid in which the fluid properties such as density and viscosity are constant. However, the reactants are in gaseous phase, and the flow of current interes t is a compressible flow in which the density may change
31 considerably. However, if the average velocity is low so that the pressure drop is not very significant, use of Ergun equation is justified for a compressible flow. One guideline ma y be the Mach number ( Ma ), in that if Ma is much smaller than 1, incompressible flow equations can be used for compressible flow. Typically, the average velocity in the reformer is in the range of about 0.1~1 m/s. Thus, the Mach number for the flow of curr ent interest is much smaller than 1. In using the Ergun equation for a compressible flow, however, the density variation due to the pressure change has to be taken care of properly. In the present calculation, the Ergun equation was applied in a segment wi se manner assuming a quasi steady state in each segment. That is 2 3 2 3 2 2 2) 1 ( 75 1 ) 1 ( 150i i i i p s i i p s i i i iV b V a D V D V p (2 4) where the subscript i indicates the small i th segment. The physical properties i and i are calculated at the average pressure in each segment, and the average velocity is calculated using the densityi in that segment. The density i can be calculated once the molar flow rate of each chemical component in the gaseous stream is kno wn. Assuming that the conversions of the chemical reactions (i.e., steam reforming and CPOX reactions) are uniform, the molar composition at each segment can be calculated. In F ig ure 2 2, the molar composition inside the reformer is shown for the ideal sit uation. Initially it was assumed that the reformer consisted of 19 segments and that Only CPOX is taking place for the first half of the reaction path where oxygen is consumed, and the steam reforming reaction takes place in the second half. Oxygen is supplied from the air so that nitrogen is also present in the stream. The change in viscosity is negligibly small.
32 Table 2 1 Initial flow rates of all species Species Methanol Water O 2 N 2 H 2 CO 2 Initial flow rate (mol/hr) 0.102 0.046 0.028 0.112 0.00 0.00 Initial flow rate (10 5 mol/s) 2.83 1.28 0.778 3.11 0.00 0.00 Also a ssuming 50% efficiency of a fuel cell, the required hydrogen production rate for 10W is calculated to be 0.252 mol/h Hydrogen is produced by both the steam reforming (SR) and the CPOX reactions. In determining the relative production rate of hydrogen by each reaction, it was assumed that the amount of heat generated by the CPOX reaction is balanced with the heat required by the SR reaction so that the reformer is self sufficient in term of energy requirement. Assuming 100% conversion of both the partial oxidation reaction and the steam reforming reaction, the required feed flow rates of methanol, water and oxygen, to give the desired hydrogen production rate, are given in Table 2 1. For a model case, where the reaction rate is constant irrespective of the concentration of species in the reactor, the molar flow rates of all chemical species are shown in Figure 22. Figure 22 indicates that the molar flow rate of the inert nitrogen is cons tant, and that the oxygen is gradually consumed and depleted in the first half. The molar flow rate of water is constant for the first half and decreases gradually to depletion in the second half where the steam reforming reaction is taking place. The hydr ogen production rate in the second half is greater than that in the first half because 3 moles of hydrogen are produced per mole of methanol by the steam reforming reaction whereas 2 moles of hydrogen are produced per mole of methanol by the CPOX reaction.
33 Fig ure 22 Molar flow rates of each chemical species along the flow path of the reformer (total length 95 cm) With the component molar flow rates known, the average density and the average volumetric flow rate of the fluid in each segment can be estimated using the ideal gas law. In addition, the effective viscosity can be estimated using the following semiempirical formula of Wilke [ 88, 89]: N mixx x1 ( 25) where N is the number of chemical species in the mix ture x and are the mole fraction and the viscosity of species at the system temperature and pressure The dimensionless parameter is defined as: 0.0E+00 2.0E-05 4.0E-05 6.0E-05 8.0E-05 Molar flow rate (mol/s) Methanol Water O2 N2 CO2 H2 Inlet Outlet Water CO 2 O 2 Methanol N 2 H 2
34 2 4 / 1 2 / 1 2 / 11 1 8 1 M M M M ( 26) where M is the molecular weight of species With the density, viscosity and the average velocity of the gaseous mixture in the segment known, the pressure drop can be calculated using equation ( 24) once the inlet pressure is known. However, the inlet pressure is not known a priori and should be determined iteratively. However because the outlet pressure is usually known, the pressure profile can be calculated backward starting from the outlet toward the inlet. If the mass of catalyst is the same, when the diameter of the catalyst particle becomes smaller, the accessible surface area of the catalyst will increase and be beneficial for chemical reactions taken place on the surface. In large particles the diffusion of reactants to and products from the catalyst surface can be severely limiting for the reactio n. To determine the reactor dimensions and reasonable catalyst particle sizes, model calculations were carried out at varying reactor channel diameters and as a function of catalyst particle size to determine th e pressure drops in each case. Assuming the reformer channel diameter is 2.5 mm, by varying the size of the spherical catalyst particle of 150 m 200 m 250 m and 300 m the pressure profile along the reformer flow path is given in Figure 23. These pressure profiles were determined by the backw ard calculation assuming that the outlet pressure is 1.5 atm.
35 150 200 250 300 350 400 450 Pressure(kPa) 150 um 200 um 250 um 300 um Figure 23 Pressure profile along the flow path at various particle size (reformer channel Dia.= 2.5 mm) Considering that a reasonable pressure drop in the reformer is about 2~3 atm, a catal yst particle size of 150 m would on the small side as it gives a pressure drop of close to 3 atm. It appears that a catalyst particle of about 200 m is an appropriate size that can be used at this reformer diameter (2.5 mm ). This particle size will result in an acceptable pressu re drop, and it should keep the diffusion limitations at a reasonable level To determine an appropriate reactor diameter, for a catalyst particle size of 200 m the pressure profile s were determined at reactor channel diameter s of 2.0 mm 2.5 mm and 3.0 mm (see Figure 24) Inlet Out let
36 150 200 250 300 350 400 450 Pressure (kPa) 2.0 mm 2.5 mm 3.0 mm Figure 24 Pressure profile along the flow path According to these calculations, a reactor channel diameter of 2.5 mm is the smallest diameter that will give a reasonable pressure drop (1 2 atm.) when using catalyst particle sizes of 200 m. Decreasing the reactor diameter to 2.0 mm, will result in a pressure drop of close to 3 atm. Considering that these are only model calculations, it is wise to avoid the reactor and catalyst dimensions that give pressure drops close to the up pe r bound. Calculation of the pressure profile using Equation ( 24) will be less accurate if the number of segments is too small. Thus, additional calculations were made by varying the number of segments from 10 to 38. The result with 19 segments was virtual ly the same as those with higher number of segments indicating that 19segment calculation was appropriate without serious calculation error. Based on the result given in Figure 24, the channel diameter has been determined to be 2.5mm for a prototype ref ormer Its overall dimension has been chosen to be 76 x 76 x 18 mm Inlet Out let
37 making the total volume of the prototype reformer to be about 104 cm3 that seems compact enough for portable PEMFC applications In Figure 25, the reformer manufactured according to this d imensional specification is shown. Fig ure 25 Details of the reformer body and end plate 2.2.3 S imulation with more R ealistic R eaction K inetics In the previous section, the underlying assumption that the conversions of the chemical reactions (i.e., ste am reforming and CPOX reactions) are uniform is equivalent to assuming that both reactions are zeroth order. In this section, more realistic reaction kinetics proposed by other researchers [ 45, 85, 90, 91, 92] are used for the prediction of molar compositi on along the reformer. In the refined calculation, the following assumptions are still made CPOX reaction is dominant over SR reaction until oxygen is depleted. Oxygen is supplied from the air so that nitrogen is also present in the stream. The change in viscosity is negligibly small. In 2007, Lin et al. [ 45] suggested the following kinetic model for the catalytic partial oxidation reaction (CPOX ) over a 40 wt% Cu / ZnO catalyst prepared via the co precipitation method.
38 ] ][ [2 3O OH CH k rCPOX CPOX (2 7) Using a feed composition of O2, CH3OH and N2 with O2/CH3OH ratio kept at 0.3, they determined a reaction rate constant, kCPOX, of 13388 L2/mol s gcat at 200C. Here gcat is weight of the catalyst in grams. For the steam reforming (SR) reaction, the follow ing reaction kinetics over a commercial catalyst ( BASF [ S3 85]; consisting of 31.7% CuO, 49.5% ZnO and 18.8% Al2O3) has been suggested by Jiang et al. in 1993: [ 90, 91] ) 7 (2 2 303 0 26 0kPa P P P k rH O H OH CH SR SR (2 8) ) 7 (2 2 2 32 0 03 0 26 0kPa P P P P k rH H O H OH CH SR SR (2 9) Since CPOX react ion is assumed to occur prior to the SR reaction, there will be hydrogen in the reaction by the time SR reaction takes place. As it will be shown, the hydrogen partial pressure is always larger than 7 kPa when SR reaction occurs. Thus, only the 2nd kinetic expression [ i.e., eqn (2 9 ) ] is used in the calculation s According to Agrell et al. [ 85] in 2002, the reaction rate constant for this reaction is 1.37 105 mol/s gcatkPa0. 0 9 at 200C. Apparently, the reaction rate constant for the CPOX reaction is many orders of magnitude larger than that for the SR reaction. Thus, the CPOX reaction is expected to be an extremely fast reaction that may take place within a very short distance from the inlet to the reformer. However, as it will be shown in later chapter i t is likely that the CPOX reaction does not occur as fast over the catalyst with the proposed experimental conditions in the actual reformer Also, this is not necessarily detrimental to the operation of the reformer, as the catalyst can be diluted with in ert material to reduce the reaction rate in the CPOX portion of the reforming.
39 In a differential segment (or element) of an ideal tubular reactor (or a plug flow reactor), the material balance for a reactant A (e.g., methanol) is given as Adx r dV r dN A A A (2 10) Here dNA (mol/s) is the rate of increase of species A within the differential element of the reactor, rA is the production rate of A per unit volume of the reactor (i.e., mol/s L), and dV is the differential volume of the reactor (L) which is the product of the cross sectional area ( A ) and the differential length (dx ) of the reactor. Thus, for the CPOX reaction, the molar rate of change of methanol is given as A O OH CH k dx dNCPOX MeOH ] ][ [2 3 (2 11) Once the reformer volume and the amount of the catalyst in the reformer are known, the reaction rate constant is known. Assuming idea gas law, the molar volume of the reactant mixture can be calculated if the absolute pressure at the given reactor position is known. Hence the volumetric conce ntration of each reactant can be calcu l ated. The numerical calculation has been performed following the procedure described below: 1) Assume the inlet pressure and calculate the initial volumetric concentration of the reactants for a given feed rate of the re actants, see Table 2 1. 2) Calculate the molar rate of change of methanol for a small segmental length of the reactor (e.g., dx = 105 cm) assuming that only CPOX reaction takes place in the beginning. 3) Determine molar rates of change for all other species us ing the reaction stoichiometry. 4) Calculate the pressure drop through the differential segment of the reactor and update the pressure at the down stream position and the molar concentration of all chemical species. 5) March forward following the same scheme unt il oxygen is depleted (i.e., more than 99.9% oxygen is consumed)
40 6) Apply the same calculation scheme but using the kinetic expression for the SR reaction through the remaining length of the reformer 7) If the exit pressure is different from the target value ( e.g., 1.5 atm ) assume a new value for the inlet pressure and repeat the entire calculation procedure. 0.0E+00 2.0E-05 4.0E-05 6.0E-05 8.0E-05 Molar flow rate (mol/s) Methanol Water O2 N2 CO2 H2 Figure 26 Molar flow rates of each chemical species along the flow path of the reformer calculated for kCPOX = 13388 L2/mol s gcat and kSR = 1.37 105 mol/s gcat kPa0. 0 9 Following the procedure given above, calculations were conducted for several different values of the reaction rate constants and the results are shown in Figures 26 through 211 in the same format as Figure 2 2. All results were obtained for a reformer whose diameter is 2.5 mm with the total length of 95 cm. If it is not specified specially, the x axis represents total flow path of reformer (95cm). The total weight of catalyst in the reformer was assumed to be 2 grams, which in fact is the actual amount for the experiment that will be described in the later sections. H 2 N 2 CO 2 Methanol Water O 2 Inlet Outlet
41 The reactant feed rates were 2.83 105, 1.28 105 and 3.89 105 mol/s for methanol, water and air, respectively (as given in Table 2 1) 0.0E+00 2.0E-05 4.0E-05 6.0E-05 8.0E-05 Molar flow rate (mol/s) Methanol Water O2 N2 CO2 H2 Figure 27 Molar flow rates of each chemical species along the flow path of the reformer calculated for kCPOX = 13.388 L2/mol s gcat a nd kSR = 1.37 105 mol/s gcatkPa0. 0 9 H 2 N 2 CO 2 Methanol Water O 2 Inlet Outlet
42 0.0E+00 2.0E-05 4.0E-05 6.0E-05 8.0E-05 Molar flow rate (mol/s) Methanol Water O2 N2 CO2 H2 Figur e 2 8 Molar flow rates of each chemical species along the flow path of the reformer calculated for kCPOX = 1.3388 L2/mol s gcat a nd kSR = 1.37 105 mol/s gcatkPa0. 0 9 Figure 2 9 Molar flow rates of each chemical species alon g the flow path of the reformer calculated for the same values of kCPOX a nd kSR for Figure 28, but for a small reformer length near the inlet to show the fast decline of oxygen 0.0E+00 2.0E-05 4.0E-05 6.0E-05 8.0E-05 0 5 10 15 20 25 Position from inlet along the reformer (cm) Molar flow rate (mol/s) Methanol Water O2 N2 CO2 H2 N 2 CO 2 Water Methanol O 2 Inlet H 2 Out let H 2 N 2 CO 2 Methanol Water O 2 Inlet Outlet
43 0.0E+00 2.0E-05 4.0E-05 6.0E-05 8.0E-05 Molar flow rate (mol/s) Methanol Water O2 N2 CO2 H2 Figure 210 Molar flow rates of each chemical species along the flo w path of the reformer calculated for kCPOX = 1.3388 L2/mol s gcat a nd kSR = 6.85 106 mol/s gcatkPa0. 0 9 to show the influence of slower reaction kinetics for the SR rxn. 0.0E+00 2.0E-05 4.0E-05 6.0E-05 8.0E-05 Molar flow rate (mol/s) Methanol Water O2 N2 CO2 H2 Figure 211 Molar flow rates of each chemical species along the flow path of the reformer calculated for kCPOX = 1.3388 L2/mol s gcat a nd kSR = 2.74 105 mol/s gcatkPa0. 0 9 to show faster reaction kinetics for the SR rxn. N 2 H 2 Methanol Water CO 2 O 2 Inlet Inlet Out let Out let O 2 Water CO 2 Methanol H 2 N 2
44 As Figure 2 6 indicates, if the reaction rate constant for the CPOX reaction is as large as suggested by Lin et al. ( 2007), the reaction is extremely fast and occurs within an unreasonably short length. For this large rate constant, the molar rate of change of both methanol and oxygen is very steep and consequently, the numerical calculation is likely to not be very accurate unless an extremely short segment length is used. When the reaction rate constant for the CPOX reaction is much smaller, its reaction length increases as Figures 2 7 through 2 9 indicate. Similarly, the reformer length for the SR reaction incr eases if the reaction rate constant gets smaller (Figure 2 10), whereas it decreases if the reaction rate constant is larger (Figure 2 11). As the pressure profiles for the reformer design were based on the model calculations given in Figure 22, these pr essure profiles were recalculated for the reaction conditions in Figure s 27 and 211 (see Figure s 2 12 and 213) 150 200 250 300 350 400 450 Pressure(kPa) 2 mm 2.5 mm 3.0 mm Figure 212 Pressure profile along the flow path based on the reaction c onditions for Figure 27 Inlet Out let
45 150 200 250 300 350 400 450 Pressure(kPa) 2 mm 2.5 mm 3.0 mm Figure 213 Pressure profile alo ng the flow path based on the rea ction conditions for Figure 211 As is evident in Figures 212 and 213, these pressure profiles are not much different from those in Figure 24. These calculations suggest that the pressure profile s obtained under very si mplified conditions can give reasonable results, even though the composition along the reactor may be rather unrealistic. Inlet Out let
46 C HAPTER 3 EXPERIMENTAL STUDY 3.1 Experimental S etup Figure 31 shows the experimental setup schematically. The main components of the setup are (1) an air compressor to supply oxygen to the reformer along with nitrogen, (2) syringe pump to supply methanol/water liquid mixture, (3) the com pact reformer shown in Figure 25, and (4) the temperature contro ller to control the operating temperature of the reformer and (5) a cold trap is located at the outlet of the reformer to condense any unreacted methanol and water from the feed There are three pressure gauges to monitor the pressure at three different lo cations. The flow meter s are both small volume rotameter s that are used to monitor the flow rate of the air or the total product flow rate The temperature of the reformer is measured and controlled by a thermocouple attached to the surface of the reformer Because the capacity of the heater ( 100 W ) was rather large and the thermocouple had to be positioned just below the heater (i.e., between the heater and the top surface of the reformer), a temperature overshoot of 4~5 C was realized whenever the heater was turned on. Fig ure 3 1 S chematic of the e xperimental s etup
47 The syringe pump is driven by a stepper motor that is controlled by a PC to provide a tight control of the feed rate of the liquid phase reactants (i.e., methanol/water mixture). The flow rat e of the gas phase reactant (i.e., oxygen) is more difficult to control especially when the flow rate is low as in the present case. In the current study, oxygen was supplied in the form of air stored in the tank of an air compressor. As the molar flow rate of air varies widely depending on the pressure and the temperature, calibration curves have obtained prior to the actual experiment. Details of the air flow calibration curves will be discussed in a later section. The valve at the outlet side of the refo rmer is to control the pressure at position 3 (i.e., the location where the pressure gauge 3 is present) in calibrating the air flow rate. This valve was wide open when the actual experiment was performed to assess the performance of the reformer. The pressure gauges 1 and 2 are analog type whereas pressure gauge 3 was a digital one. 3.1.1 C al i bration of the O xygen F eed R ate Because the syringe pump is a positive displacement device, the feed rate of the liquid phase reactants is not determined by the syst em pressure but by the linear velocity of the piston. In contrast the flow rate of the gas phase reactant (i.e., oxygen) fed from a compressed air tank depends significantly on the system pressure. Thus, the flow meter shown schematically in Figure 31 wa s cali brated prior to the experiment. As mentioned, t here were 3 pressure gauges in the experimental setup (Figure 31). Gauge 1 was the one attached to the regulator that controlled the pressure at the outlet of the compressed air tank of the compressor and gauge 2 was the one measuring the pressure at the inlet to the flow meter. The experiments described later in section 3.3.2 indicated that the typical pressure at the inlet to the reformer (i.e., pressure at gauge 3 in Figure 3 1) was about 20 ~ 25 psi for the typical range of feed rate s and the weight of catalyst used Thus, in calibrating the flow
48 meter, the pressure at location 3 was varied between 15 to 30 psi wh ereas the pressure at the outlet from the tank was between 30 and 35 psi depending on the air flow rate The choice of the pressure at location 1 (i.e., gauge 1 pressure P1) may be rather arbitrary as long as it is set at a value higher than the pressure at location 3. However, the higher the pressure at location 1 is, the more difficult it is to control the air flow rate. Thus, P1 was set at a value as low as possible as long as a desired stable air flow rate was achieved. There were two needle valves to control the air flow rate; one attached to the flow meter itself and the other located at the out let from the reformer (Figure 3 1). From the stoichiometric equation for the CPOX reaction (Equation 16) and the equations ( 31 ) and ( 32 ), the standard oxygen flow rate is Y/4. That is 0.028 mol/h. Thus, the air flow rate is 0.133 mol/h or 49.8 ml/min at the standard condition. Because the air flow rate is so low, the pressure drop through the tubing from the air tank to the flow meter is as low as 1~2 psi. Thus, the pressure reading at location 2 (i.e., P2) was between 28 and 34 psi. T he pressu re at the location 3 (i.e., P3) can be controlled by the two valves. However, the valve attached to the flow meter can provide only minor adjustment whereas the valve at the o utlet of the reformer (Figure 3 1) is the main control val ve to adjust P3. Thus, the calibration procedure for the air flow rate is as follows: 1) The syringe pump is shut down so that no liquid enters the system throughout the calibration procedure. 2) The pressure gauge 1 is set at 30 ~35 psi by adjusting the regulator knob of the com pressed air tank while the valve at the reformer outlet is close. 3) Open the val ve of the flow meter (i.e., rotameter) to a certain extent and gradually open the valve at the reformer outlet so that the pressure reading at location 3 (i.e., P3) is at a de sired val ue between 12 and 29 psi as show n in Figure 32. 4) Adjust the val ve of the flow meter so that the float (a stainless steel ball) of the rotameter is positioned at a desired location from 2 th rough 5 of the rotameter indicator 5) Measure the volumetric flow rate of the air at the exit for each float position as a function of the pressure at location 3 (between 15 30 psi)
49 A graduated cylinder filled with water was connected to the outlet from the reformer so that the air exiting the system displaces the water in the graduated cylinder. Thus, the displaced volume of the water per unit time is equivalent to the air flow rate. In Figure 32, the measured air flow rates are given for various values of the pressure at location 3 (i.e., pressure gauge 3 in Figure 3 1). Also given in Figure 32 are the linear regression lines for each value of the float position. 10 30 50 70 90 110 130 150 10 15 20 25 30 gauge-3 pressure (psi) Air flow rate (ml/min) float position=5 float position=4 float position=3 float position=2 Fig ure 32 Air flow rate at various values of the pressure at location 3. The correlation coefficients for all these lines except for float position 5 are greater than 0.9, and the following equations for the linear regression lines were used to calculate the air flow rates for the experiments described in the following section: Float position 2: Y = 0.747 X + 25.515 (R2 = 0.94) Float position 3: Y = 1.207 X + 42.569 (R2 = 0.91) Float position 4: Y = 1.724 X + 57.342 (R2 = 0.96) Float position 5: Y = 0.590 X + 116.58 (R2 = 0.89)
50 3.1.2 Conditions for Steam Reforming with O xygen The oxygen in the feed is to induce the exothermic CPOX (catalytic partial oxidation) reac tion in addition to the endothermic SR (steam reforming) reaction so that the heat generated by the CPOX reaction can be used for the SR reaction for self sustainability of the reformer. SR reaction: CH3OH + H2O CO2 + 3H2 ( H0 = 130.9 kJ/mol) ( 13) CPOX reaction: CH3OH + O2 CO2 + 2H2 ( H0 = 154.9 kJ/mol) ( 16) Thus, more methanol is needed than water because methanol is consumed by both the CPOX and the SR reactions. For an ideal case in which no side reaction such as the reverse shift reaction occurs, the methanol to water ratio of the feed mixture can be determined for the condition that the amount of heat generated by the CPOX reaction bala nces with the heat needed for the steam ref orming reaction. For this calcuation, the heat of reaction has to be determined first for a situation where all reactants are at room temperature (i.e., 25 C) and the product gas is at the reaction temperature (e.g., 250C). Using appropriate thermodynamic property data for all species involved in the reations, the heats of reactions are calculated to be HSR = 1 59.8 kJ/mol and HCPOX = 132.5 kJ/mol respectively. Methanol and water in the feed are in liquid phase at 25C and the oxygen is gaseous at 25C. Details of the calculation for the heats of reactions are given in the Appendix. For a PEMFC with a power generation rate of 10 Watt (or 72 kJ/h) operating at the efficiency of 50%, the hydrogen feed rate of 0.252 mol/h is needed. Thus, the reformer should be able to produce hydrogen at that rate and should be self sustaining. These constraints can be expressed as the following set of algebraic equations: X + Y = 0.252 ( 31)
51 (159.8/3)X + ( 132.5/2)Y = 0 ( 32) Here X and Y are the numbers of moles produc ed by the steam reforming (SR) and the catalytic partial oxidation (CPOX) reactions, respectively. Equation 31 is for the hydrogen production rate of 0.252 mol/h whereas equation 32 describes the energy balance for self sustainability. The solution to these equations are X = 0.139 and Y = 0.112. Thus, the feed rates of water and methanol should be X/3 and (X/3+Y/2), respectively. That is, 0.046 and 0.102 mol/h for water and methanol, respectively. Therefore, the methanol to water ratio of the liquid phase feed should be 2. 2 for the ideal case where the heat realeased in the partial oxidation reaction equals the heat required for the steam reforming reaction For all experiments that followed, the liquid phase feed were the methanol/water mixture with the m ixing ratio of 2.2. 3.1.3 Calibration of P roduct G as F low R ate The product gas flow rate was measured either by the volume displacement method (as for the air calibration measurements) or via a calibrated flow meter connected to the system after the condenser. The flow meter was necessary to determine the product gas flow rate when the reformer outlet (after the condenser) was connected to the gas chromatograph (GC) for online product gas composition analysis. In this case, the volume displacement method could not be used a t the same time the as composition was monitor ed using the GC Thus, a second flow meter (100 500 ml/min range) was attached to the outlet of reformer after the condenser ( Figure 31) T he outlet of flow meter was then connected to the gas chromatograph. The calibration curve of the flow meter is given in Figure 33. The gas composition used for the calibration wa s approximately 20% CO2, 50% H2 and 30% N2, which is similar to the composition of product gas The calibration was done by volume displacement method
52 50 100 150 200 250 300 350 400 450 0 0.5 1 1.5 2 2.5 3 3.5 4 Scale reading Flow rate (cc/min) Fig ure 33 Calibration curve for 500 cc flow meter 3.1.4 Gas C hromatography C onfiguration and O peration P rocesses The product gas compositions we re measured by a customized Agilent 6890 gas chromatograph (GC). The GC is programmed to take sample with a given volume (e.g 1.0 ml ) through an automatic sampling valve. The samples can be taken either offline by manually injecting samples using a gas sampling bag or online by connecting the outlet of the steam reformer, after the c ondenser, to the inlet of the GC To avoid introducing additional uncertainties in the measurements, online gas monitoring is preferred. This is particularly the case when hydrogen is a significant portion of the product stream, since hydrogen is challengi ng to contain. The GC has two columns and two detectors in series. The first column is a polar Porapak Q capillary column, which separates CO2, methanol and water. The second one is a molecular sieve column that separates the permanent gases H2, O2, N2, CO. A column separation valve is
53 located between the columns to avoid contaminating the molecular sieve column. The first detector is a thermal conductivity detector (TCD) and the output from this detector is sent to the flame ionization detector (FID). As t he FID can only detect any species containing C H bonds (e.g. methanol and potentially methane in our case ) a methanizer is placed in be tween the TCD and the FID. The methanizer will reduce CO and CO2 to methane, which will allow detection with the FID an d this also increases the sensitivity to these species. L ines of hydrogen, air and nitrogen are supplied to the GC and serve as the fuel, oxidant and make up gas for the FID, respectively. Only helium is supplied to the TCD and it serves as both reference and carrier gas for the TCD. The TCD can detect all chemical species in the product stream, including those detected by FID. However, the sensitivity to hydrogen is relatively low due to the He reference and carrier gas. As the concentration of hydrogen in the product stream is high, this is not a serious limitation in the experiments. Consequently, t he TCD can perform all the measurements alone, but the additional FID detector provide s better sensitivity for the hydrocarbon species and the CO and CO2 due t o the methanize r To start the GC measurements, the FID temperature is set at 400C. It takes about 10 minutes to reach this temperature. After that, the hydrogen flame is turned on via the GC digital panel manually for the FID. The hydrogen, air and nit rogen (make up gas) flow rates are preset to be 40, 450 and 6 sccm. The TCD temperature is maintained at 250 C and helium is used as the reference gas. The reference Helium flow is set at 33 sccm and the make up Helium flow is set at 4 sccm. Although it is possible to control most things via the Cerity 3.0 software, it is easier to turn on the FID flame manually before starting the software. A method created in the Cerity software is then used to control the sample injection valve, the oven temperature as w ell as the temperatures of and flow rates in the other units (valves, columns, detectors) in the GC. With the
54 temperature program used, e ach sample takes about thirty minutes to complete. Multiple runs are taken for each sample to ensure reliable and reproducible measurements After completion, the collected GC data is analyzed directly by Cerity 3.0. The software will calculate the areas under the peaks in the GC spectra and this area is proportional to the concentration of the spe cies giving rise to that peak. To be able to quantify the concentrations of each species, calibration curves with known compositions of each compound are created. The calibrations in this case had been prepared for another project using argon at a constant flow rate as an internal standard. The response (area under the peak) of each compound was plotted against the concentration of the compound. T he resulting linear regression equation s for CO CO2 and H2 are given by the following equations : CO% =0003*FID(CO)+0.0642 (3 3) CO2%=0.0051*TCD(CO2) + 0.3543 (3 4) H2% =19.744* TCD(H2) 48.452 (3 5) FID(CO) represents integral area under the peak at the retention time for CO resulting from the FID detector, and the TCD(CO2) and TCD(H2) represent integral area under the peak at the retention time for CO2 and H2 on the TCD channel 3.2 Experiments with G lass B ead as the P acking M aterial I n order to confirm the flow simulation results described in the previous chapter, experiments were performed with the reformer filled with glass bead to suppress any chemical reaction even at an elevated temperature. Because the channels 1 through 10 (Figu re 21) of the reformer were intended to be the evaporation zone, these channels were kept empty whereas
55 channels 11 through 29 were packed with spherical glass bead of a diameter in the range of 177~250 m. For this experiment, only the liquid phase reactants were fed into the reformer with the air compressor shut off and the flow regulati ng knob of the rotameter closed. The reformer was maintained at 220C by a thin electrical heater mounted on top of t he reformer. The liquid phase reactants entering the reformer would evaporate in the evaporation zone and the gaseous feed would then flow through the packed zone before exiting to the atmosphere through the exit at the end of channel 29. The overall press ure drop for the flow was monitored by the pressure gauge 3 in Figure 31. Three different types of feed were used for this experiment ; pure methanol, pure water, and equi molar mixture of the two The feed rate was varied from 0.1 to 1.1 mol/h. The experim ent was run for at least 10 to 15 minutes prior to taking any data (i.e., the pressure reading) to ensure steady state operation. The experimental results are shown in Figure 3 4 along with the theoretical predictions for three different sizes of the packing material. The theoretical estimates are for a monodispersed packing material whereas the experimental results are for a polydispersed packing. Considering that the size of the glass bead is between 177 and 250 m, the agreement between the calculated and the measured values seems reasonably good. It is noted that the experimental results for three different types of feed form a single curve indicating that the pressure profile depends only on the total flow rate but not on the materials themselves. This result should be true only if the viscosities of the materials in gas phase are similar to each other. In estimating the pressure drop using the Ergun equation (Equation 24 ), the effective viscosity of the mixtur e was used as estimated by the sem i empirical relation (Equation 25). It should noted, however, that the effective viscosity is rather insensitive to the
56 composition of the gas phase mixture, and using the viscosity of the major component in the mixture a s the representative value is, in fact, sufficient in using Equation (24). 0 50 100 150 200 250 300 350 400 450 500 0 0.2 0.4 0.6 0.8 1 1.2 Molar Flow Rate (mol/hr) Pressure Drop (kPa) 250 um theory 200 um theory 175 um theory pure water 50/50 pure methanol Fig ure 34 Comparison between predicted and measured pressure drop 3.3 E xperiments with Cu/ZnO C atalyst All catalyst e xperiment was conducted using a commercial catalyst, MDC 3, from Sd Chemie Co. Ltd. This catalyst consists of 42% CuO, 47% ZnO, and 10% Al2O3, and is in a cylindrical pellet form (3.23.2 mm) The reaction temperature recommended by the manufacturer is between 220 and 270C. The catalyst pellets were granulate d using a ball mill and the particles in the range of 180 and 212 m were collected using a stack of sieves. As it was described in section 3.2, the reformer channels 11 through 29 were filled with the catalyst particles, and experiments were conducted for va rious feed rates and compositions at a fixed reaction temperature of 220C which is the lowest temperatur e recommended for the catalyst. Some experiments were also carried out at 250 C to determine the influence of the reformer
57 temperature and in one set o f experiments, only the last five channels (channels 25 through 29) were filled with catalyst. 3.3.1 Steam R eforming E xperiments 220.127.116.11 Low temperature experiment In the steam reforming experiments, t he feed was initially an equimolar mixture of methan ol and water, and later the methanol to water molar ratio was varied to 1:2 and 1:3. In the absence of oxygen in the feed, only the steam reforming reaction would take place and the reaction temperature had to be maintained solely by the electrical heater throughout the experiment. Although reduction of the catalyst prior to the experiment i s recommended for activation of the catalyst it was bypassed as prereduction would be difficult in a truly portable reformer. Furthermore, under autothermal conditions, the oxygen added for the partial oxidation reaction would reoxidize at least part of the catalyst, which would decrease the effects of prereduction. In addition, t he hydrogen produced by the steam reforming reaction itself is expected to reduce the cataly st which may cause an induction period in which more H2O and less H2 is formed For this reason, once the reformer was filled with newly prepared catalyst, experiment was conducted initially at a l ow feed rate of about 0.1 mol/h for approximately 12 hr s before initiating the experiments and increasing th e feed rate to higher value s For each experiment, the volumetric flow rate of the product gas exiting the reformer was measured, and samples were taken using gas sampling bags for analysis. The product gas composition measured using gas chromatography (Agilent 6800) is shown in Table 31 for 5 different samples. The samples 1 through 3 were taken from three different experiments conducted using the same catalyst (catalyst 1) at the same temperature (220C ) but at different
58 feed rates. Experiments for samples 4 and 5 were also conducted at the same temperature of 220C but using a different batch of catalyst (catalyst 2). Table 31 Product gas c omposition as a function of feed rate at steam reforming condi tions (reaction temperature: 220 C w ater to methanol molar ratio=1:1 ) E xperiment # 1 2 3 4 5 Catalyst type Cat. 1 Cat. 1 Cat. 1 Cat. 2 Cat. 2 Feed rate (mol/hr) 0.05 0.07 0.1 0.4 0.9 Product gas flow rate (mol/hr) 0.095 0.127 0.18 0.45 0.54 Methanol conversion 95% 91% 90% 56% 30% H2 76% 73% 80% 78% 77% CO2 20% 23% 17% 18% 21% CO 4% 4% 3% 4% 2% The feed rates for these 5 experiments were varied in a wide range from 0.05 to 0.9 mol/h. Despite the low reaction temperature of 220C, the methanol conversion was higher than 90% at the low flow rate s whereas it dropped to 30% for the highest flow rate. The methanol conversion is calculated by assuming the SR reaction is the only reaction to take place. From the stoichi ometry of SR reaction, 1 mol of meth anol and 1 mol of water will generate 1 mol CO2 and 3 mol es H2, or 4 mol es product gas in total. The moles of methanol reacted to products per unit time were thus determined by dividing the measured gas product flow rate (converted from volumetric flow to molar flow) by 4. As the feed was injected in the ref ormer by syringe pump with controlled liquid flow rate, the conversion of methanol was calculated by dividing the methanol converted per unit time by the methanol molar feed rate. The product gas composi tion remained about the same regardless of the methanol conversion. It may be noted that the molar ratio of H2 to CO2 is higher than the stoichiometric ratio of the steam reforming reaction
59 (equation 13). It is speculated that this discrepancy may be asso ciated with the fact that the product gas analyses were done off line in a batch wise manner using a limited amount of sample. If the efficiency of a PEMFC is assumed to be 50%, hydrogen supply at a rate of 0.252 mol/h is required for 10W power generation. This is equivalent to the feed rate of 0.168 mol/h for the reactants that is an equimolar mixture of methanol and water. Thus, for when repeating the steam reforming experiments, the reactant feed rate was varied from 0.074 to about 0.61 mol/h, as this c overed the range of about 0.5 to 4 times the flow rate that is needed to produce hydrogen for 10W power generation. The new results which were done online in a continuous manner with more duplicated compositional analysis measurements showed good agreement with the reaction stoichi o metry of the steam reforming reaction. For e ach feed rate, 3 or more samples were measured by GC to get an average composition. T he standard deviations were also calculated for each data point and added to Table 3 2. Table 3 2 Pr oduct gas composition as a function of feed rate at steam reforming conditions (reaction temperature: 220C W ater to methanol molar ratio=1:1 C atalyst 3 was used for measurements, which is a mixture of fresh and used catalyst from the second batch [Catal yst 2] ) Feed rate (mol/hr) 0.074 0. 146 0. 244 0. 366 0. 488 0. 610 Product gas flow rate (mol/hr) 0.13 0.20 0.28 0.30 0.30 0.30 Methanol conversion 86% 70% 56% 42% 31% 25% H2% 72.36 72.49 72.50 72.61 72.58 72.76 CO2% 27.52 27.42 27.42 27.31 27.34 27.16 CO % 0.11 0.08 0.08 0.08 0.08 0.08
60 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Reactant molar flow rate (mol/hr) Product mol flow rate (mol/hr)) Fig ure 35 Product gas flow rate vs. feed flow rate ( : catalyst batch 1, : catalyst batch 2) at 220C. The solid line indicates the product molar flow rate expected In Figure 35 the molar flow rate of the product gas is shown as a function of the feed flow rate for the two different catalyst batches. The two batches of catalyst are pretreated in the same way, i.e. activation by slow methanol/water feed rates in the absence of oxygen for 12 hours. Also shown in the f ig ure is the line representing the ideal (or maximum) product gas flow rate that is equivalent to 100% conversion of methanol. When the feed flow rate is at 0.15 mol/h, the conversion of methanol is close to 100% and the behavior of Catalyst 2 is very close that of C atalyst 1. However, w hen the feed flow rate is doubled, to 0.3 mol/h, the conversions decrease to about 80% for C atalyst 1 and 62% for C atalyst 2. Further increase in the feed flow rate decreases the conversion even more making it lower than 50% e ven for catalyst 1 when the feed rate is about 0.9 mol/h. Although the catalysts were prepared following the same procedure, the
61 second batch of catalyst appeared to be of lower efficiency than the first batch and the pre reduction of the catalysts might ha ve reduced the differences between them. 18.104.22.168 T emperature effect Low temperature operability is most desirable for a compact reformer for portable application, and the methanol conversion has been shown to be close to 100% for a targeted feed flow ra te i.e. the 0.168 mol/h feed rate of an equimolar mixture of methanol and water, which is needed for a 0.252 mol/h H2 production rate required for power generation in a 10 W PEMFC, even at the lowest temperature recommended for the commercial catalyst. As the demand for hydrogen may increase, it is important to test both feed flow rate s higher than the targeted feed flow rate and higher temperatures. As increasing the feed rate decreases the contact time this results in a decrease in conversion (see Figur e 3 5), to utilize more of the methanol, the temperature must be increased. Therefore, experiment were also conducted at 250C and compared to those conducted at 220 C (see Table 3 3) The C atalyst 3 was used for this experiment as it was the latest batch and showed lower conversion than C atalyst 1 at 220C. The temperature thus expected to have a stronger influence on Catalyst 3 compared to Catalyst 1. The feed was also the equimolar mixture, and the feed flow rate was varied from 0. 074 to 0.61 mol/h.
62 Table 3 3 Product gas composition as a function of feed rate at steam reforming conditions (reaction temperature: 250C Water to methanol molar ratio=1:1 ) Feed rate (mol/hr) 0.074 0. 146 0. 244 0. 366 0. 488 0. 610 Product gas flow rate (mol/hr) 0.14 0.25 0.3 6 0.41 0.41 0.41 Methanol conversion 93% 87% 74% 56% 42% 34% H2% 72.94 72.21 72.54 71.85 72.56 72.48 CO2% 26.72 27.63 27.27 28.04 27.35 27.44 CO% 0.34 0.16 0.19 0.11 0.09 0.08 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 0.00 0.20 0.40 0.60 Total flow rate (mol/hr) Methanol conversion 250C 220C Fig ure 36 Conversion of methanol at various feed rates and reformer te mperatures. As Figure 36 indicates, the methanol conversion is slightly higher at a higher reaction temperature. At the feed flow rate of 0. 146 mol/h, the conversion increases from 70% to 87% when the reaction temperature is increased from 220C to 250C whereas it increases from 42 %
63 to 56% at a feed rate of 0. 366 mol/h. At 220C, the feed rate above which the conversion drops below 50% is about 0.3 mol/h, whereas it is 0.4 mol/h when the reaction temperature is 250C. Consequently, if a higher H2 flow r ate is needed or if the catalyst deactivates slightly so that the desired H2 production rate is not reached at 220 C, it is possible to increase the temperature to increase methanol conversion and thus the H2 yield. 22.214.171.124 Effect of methanol to water mol ar ratio of the feed As i ncreas ing the water to methanol molar ratio of the feed has been shown to result in higher conversions, higher yields of hydrogen and higher selectivities to CO2, the effect of varying water to methanol molar ratio was also investi gated T he results are shown in Figures 37 and 38 as well as in Table 34. In these cases the product gas only consists of the noncondensable H2, CO and CO2. 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.0 0.1 0.2 0.3 0.4 MeOH feed rate (mol/hr) Product molar flow rate (mol/hr 1:1 molar ratio 2:1 molar ratio 3:1 molar ratio Fig ure 37 Influence of water to methanol molar ration on the product molar flow rate at various methanol feed rate s (at 250C) The solid line represents the product gas flow rate at 100% conversion of MeOH assuming only steam reforming and no reverse water gas shift or methanol decomposition reaction s
64 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.0 0.1 0.2 0.3 0.4 MeOH feed rate (mol/hr) MeOH conversion 1:1 molar ratio 2:1 molar ratio 3:1 molar ratio Fig ure 38 Influence of water to methan ol molar ratio on methanol conversion at various methanol feed rate s (at 250C) Only a t the lowest methanol feed rate s, below 0.1 mol/h, does a higher water to methanol ratio appear to result in slightly higher product gas flow rate (Figure 37) and conversion of methanol (Figure 3 8) although the differences are not significant In contrast, at a higher feed rate, the adverse effect s of a larger water to methanol ratio was apparent in that both the product gas flow rate and the methanol conversion were lower than for the case of the equimolar mixture feed. At water to methanol molar ratio of 3 the decrease in conversion was rather significant in that it dropped below 50% at the methanol feed rate of about 0.2 mol/h. This is to be compared with a conversi on of 6569% at the same methanol feed rate, but water to m ethanol ratios of 1:1 and 2:1. This result may be associated with the fact that a higher water content in the feed at a high feed rate would require much more energy for vaporiz ation and this cou ld reduce the temperature in the reaction region A reduced temperature, in turn would reduc e the methanol
65 conversion, potentially significantly which is the observed outcome Having the ability to measure and control the temperature inside the catalyst b ed would be very important to further probe the effects of varying water to methanol ratios and feed rates. As the H2:CO ratio is expected to vary with the H2O to MeOH ratio online GC measurement were performed to determine t he product gas composition in these experiments (see Table 3 4). As is evident in the table, the hydrogen and CO2 selectivities do not appear to change significant ly with the water to methanol ratio. Only a slight decrease in CO concentration was observed with increased water content in the feed. This indicated that the selectivity of the SR reaction is slightly improved over the methanol decomposition and the reverse water gas shift. However, the differences in CO concentrations are not significant in this case. Table 34 Product ga s composition at various water to methanol molar ratio s ( reformer temperature: 250C) with online GC measurements MeOH feed rate (mol/hr) 0. 12 0. 12 0. 12 Water f eed rate (mol/hr) 0. 12 0. 24 0. 36 Product gas flow rate (mol/hr) 0.40 0. 41 0.39 Methanol conve rsion 82% 85% 81% H2% 72.54 72.21 72.04 CO2% 27.27 27.66 27.85 CO% 0.19 0.13 0.12
66 3.3.2 Steam Reforming with O xygen 126.96.36.199 Experimental results and theoretical predictions ( 19 channels of catalyst ) Experiments were carried out with catalyst in all 19 channels a s a function of air flow rate. In this case the feed rates of methanol and water were fixed at 0.102 and 0.046 mol/h, respectively while the air flow rate was varied between 0. 102 to 0.335 mol/h In the case of the low air flow rate (Ex periment #1) there is not enough O2 for autothermal reforming at this feed rate (i.e. to balance the heat released from the CPOX reaction and the heat needed in the reforming reaction). Assuming that the CPOX reaction is much faster than the reforming reac tion, so that all the oxygen is depleted before the steam reforming reaction takes place, the theoretical values for the product flow rates and c ompositions can be calculated. These numbers are given in Table 3 5. In t he theoretical estimate (predicted val ues) for the output gas flow rate it was assum ed that only the steam reforming (SR) reaction and the catalytic partial oxidation (CPOX) reactions are taking place in the reformer All other reactions are neglected. In each case, it is assumed that all the oxygen reacts, if there is a sufficient amount of methanol in the feed. In Experiment #1 methanol is in excess and the steam reforming is limited by the water, while methanol is the limiting reactant in Experiments #2 #4. At the highest air flow rates, t here is more O2 added than needed for the partial oxidation reaction (Experiments #3 and #4), while in Experiment #2 only water is in excess. For example, for Experiment #2 where the average air flow rate was 0. 178 mol/h, the oxygen flow was 0.037 mol/h and it was assumed that 0.037 2 = 0.074 mol/h of methanol was consumed in the CPOX reaction (reaction 16) This leaves 0.102 0.074 = 0.028 mol/h of methanol that can react in the steam reforming reaction. As there is 0.046 mol/h of water in the fe ed, this is sufficient to react all of the methanol in the SR reaction (this is not true in Experiment #1 as there is not enough water to react all the methanol in this case). The H2 and
67 CO2 product flow rates are then calculated according to the stoichiom etries of the CPOX and SR reactions, i.e. one mol of methanol gives one mol of CO2 and two moles of H2 in the CPOX reaction and one mol of CO2 plus three moles of H2 in the SR reaction. All the details of the predicted values for each air flow rate are lis ted in Table 35. Table 3 5 Theoretical calculation of gas flow rates and compositions as well as maximum conversions Exp eriment series # 1 2 3 4 O 2 feed rate (mol/hr) 0.021 0.037 0.055 0.070 N 2 feed rate (mol/hr) 0.081 0.141 0.206 0.265 CH 3 OH feed rat e (mol/hr) 0.102 0.102 0.102 0.102 Water feed rate (mol/hr) 0.046 0.046 0.046 0.046 CH 3 OH consume d by CPOX (mol/hr) 0.042 0.074 0.102 0.102 Limiting reactant a Water MeOH SR MeOH PO MeOH PO CH 3 OH consumed by SR (mol/hr) 0.046 0.028 0 0 H 2 generate d by CPOX (mol/hr) 0.084 0.148 0.204 0.204 H 2 generated by SR (mol/hr) 0.138 0.084 0 0 CO 2 generated by CPOX (mol/hr) 0.042 0.074 0.102 0.102 CO 2 generated by SR (mol/hr) 0.046 0.028 0 0 Total product gas rate (mol/hr) 0.391 0.475 0.516 0.590 Theoreti cal CH 3 OH conversion 86.3% 100% 100% 100% Theoretical water conversion 100% 61% 0% 0% Theoretical H 2 vol.% (N 2 free) 71.6% 69.5% 66.7% 66.7% Theoretical CO 2 vol.% (N 2 free) 28.4% 30.5% 33.3% 33.3%
68 a MeOH SR: there is not enough methanol for the steam reforming reaction. MeOH PO: there is not enough methnol for the partial oxidation reaction. The measured product gas flow rate s and compositions for the se experiments are g iven in Table 3 6. As can be seen in the table, the trend in the measured prod uct gas flow rate and H2:CO2 ratio follows the predicted values, i.e. the product flow rate is increased, while the H2:CO2 ratio decreases with an increasing air flow rate. Table 3 6 Product gas flow rate as a function of air flow rate at feed rates of 0. 102 mol/h of methanol and 0.046 mol/h of water ( with product gas composition from gas chromotograph) at T = 250 C Exp. Pressure Feed Product Flow Rates Nitrogen free p roduct gas composition (%) Exp. Theo series # at 3 psi Air (mol/ h) ( ml/min ) ( mol/h ) Theo (mol/h ) CO CO 2 H 2 H2/ CO 2 H2/ CO 2 1 20.8 0.102 172 0.426 0.391 0.39 28.8 70.4 2.44 2.52 2 24.1 0.178 196 0.488 0.475 0.36 29.8 69.9 2.35 2.28 3 27.6 0.261 211 0.524 0.516 0.33 31.6 68.3 2.16 2 4 30.8 0.335 226 0.56 0 0.59 0 0.24 33.1 66.7 2.02 2 Theoretical predition (see Table 3 5) It should be noted the pressure at location 3 (i.e., P3) was higher than 30 psi for the E xperiment #4 in Table 3 5. These were for the cases where the position of the rotameter float was at 5. Because the air flow rate was relatively high for these cases, the pressure at location 1 (i.e., P1) had to be raised to about 33 psi, and P3 be greater than 30 psi unlike all other cases. Otherwise, the float poison 5 could not be obtained. The measured gas compositions are also reasonably close to the predicted ones ( Table s 35 and 36). Since only the CPOX and the SR reactions are cons idered in the calculations, the
69 predicted results do not contain any CO product. This appears to be a reasonable assumption as the measu red CO concentration is less than 1% for all cases. This indicat es that the contributions from the reverse water gas shift and methanol decomposition reaction s are small under these conditions As the reverse WGS and the MD reactions would reduce the H2:CO2 ratio it is not surprising that the H2:CO2 ratio is lower than the predicted one assuming only C PO X and SR in Experiment #1. The fact that the measured ratio is higher than the predicted in the other cases (Experiments #2#4) may suggest that the SR reaction does in fact compete with the C PO X reac tion under certain conditions. In other words, more methanol is reacted in the steam reforming compared to the assumed amount based on the O2 flow rate and the CPOX reaction (i.e. the values assumed in Table 3 5 ) Although not very significant, the CO concentration appears to decrease slightly with increasing air flow rate The h igher oxygen concentration might have created an environment where oxidation of CO to CO2 was more favorable thus decreasing the CO con centration Despite the fact that the concentration of carbon monoxide is low it is high enough to act as a poison to the catalyst in a PEMFC. Thus, installation of a gas cleaning unit will be necessary to remove CO prior to feeding the product gas to a P EMFC. The data given in Table 35 and 36 combined together clearly indicated that the catalyst performance was sufficient to achieve near 100% conversion of methanol producing hydrogen at the rate between 0.2 and 0.23 mol/h. This production rate of hydr ogen is slightly lower than the target value of 0.252 mol/h for a 10 W PEMFC operating at 50% efficiency. The lower rate was due to oxygen concentrations higher than needed for autothermal reforming, which resulted in more methanol being consumed by the CP OX reaction and less in the SR reforming reaction These results indicate that the CPOX reaction does take place before the SR reaction under these
70 conditions although perhaps not to the point where all the oxygen is consumed when oxygen is in excess. How ever, since the calculations on the feed rates on methanol and water were based on the autothermal case (where the heat generated by the CPOX reaction equals the heat needed in the SR reaction), the lower contribution from the SR reaction resulted in a low er H2:CO2 ratio and thus a lower H2 fl ow rate than the targeted one. As the liquid water flow rate in to the system is so small (0.83 ml/h), quantifying the water conversion by measuring the amount collected in the cold trap was not attempted in these cases. The same is true for the liquid methanol flow rate. It is therefore difficult to accurately determine the methanol conversion, as it is not possible to determine the fraction of methanol reacted in the steam reforming reaction versus that reacted in the partial o xidation reaction. In order to increase the hydrogen production rate, the reactant feed rate or the temperature has to be increased As the reaction does not appear to be conversion limited, higher feed rate experiments were also conducted for wh ich the methanol and the water feed rates were 0.145 and 0.066 mol/h, respectively. These feed rates represent about 43% higher rates than the previous cases. The results from the theoretical predictions are presented in Table 3 7. The numbers have been de termined in the same way as in Table 3 5. As the methanol and water feed ratios are higher in the s e experiments while the air rates are the same, oxygen is never in excess (i.e. the partial oxidation reaction is never methanol limited). However, Experiment #2 in Table 3 7 is near autothermal reaction condition (heat released from the CPOX reaction balances the heat needed in the reforming reaction) in which >99% of reactants are consumed in this case. Therefore, assuming that all of the oxygen reacts, Expe riments #1 and #2 have excess methanol and are limited by the water concentration, while in Experiments #3 and #4 the steam reforming
71 reaction is limited by methanol. Even though the maximum methanol conversion would be 78 % in Experiment #1, the total combined hydrogen generat ion rate by the CPOX and SR reaction s would be 0.29 mol/h, which is 14% higher than the target value of 0.252 mol/h for 10W PEMFC application As the theoretical hydrogen production rate for full methanol conversion increases in the ot her Experiments, it should be practical to obtain this hydrogen production rate under these conditions. However, it is not advisable to run at the highest air feed rates as more of the hydrogen is produced via CPOX reaction, which has a lower H2:CO2 produc tion ratio than the steam reforming reaction. Table 3 7 Theoretical calculation of gas compositions and conversions (higher feed rate) Exp eriment series # 1 2 3 4 O 2 feed rate (mol/hr) 0.0223 0.0389 0.057 0.071 N 2 feed rate(mol/hr) 0.0837 0.146 0.213 0. 268 CH 3 OH feed rate (mol/hr) 0.145 0.145 0.145 0.145 Water feed rate (mol/hr) 0.066 0.066 0.066 0.066 CH 3 OH consume d by CPOX (mol/hr) 0.0466 0.0778 0.114 0.142 Limiting reactant Water Water MeOH SR MeOH SR CH 3 OH consumed by SR (mol/hr) 0.066 0.066 0. 031 0.003 H 2 generated by CPOX (mol/hr) 0.0932 0.156 0.228 0.284 H 2 generated by SR (mol/hr) 0.198 0.198 0.093 0.009 CO 2 generated by CPOX (mol/hr) 0.0466 0.0778 0.114 0.142 CO 2 generated by SR (mol/hr) 0.066 0.066 0.031 0.003 Total product gas rate (m ol/hr) 0.488 0.644 0.679 0.706 Theoretical CH 3 OH conversion 77.7% 99.2% 100% 100%
72 Theoretical water conversion 100% 100% 47% 4.5% Theoretical H 2 vol.% (N 2 free) 72.1% 71.1% 68.9% 66.9% Theoretical CO 2 vol.% (N 2 free) 27.9% 28.9% 31.1% 33.1% The measu red data for these conditions are presented in Table 3 8. As the liquid feed rate was higher compared to the previous cases described in Table 3 5 and 36, the pressure at location 1 (i.e., P1) was set at 35 psi and P2 was between 33 and 34 psi depending on the air feed rate. Table 3 8 Product gas flow rate as a function of air flow rate at feed rates of 0.145 mol/h of methanol and 0.066 mol/h of water (with product gas composition from gas chromotograph) at T = 250 C. Theoretical predition (see Table 3 7 ) As for the results at the lower feed rates, the measured product gas flow rate is in sev eral cases slightly higher than the calculated values also for the higher feed rates The largest deviation between the measured and the predicted values is at the lowest air flow rate (0.106 mol/h) This suggests that the assumptions made in calculating t he product flow rates (Table 3 7) are not completely accurate. As the reaction in Experiment #1 is limited by water, and methanol Exp. Pressure Feed Product Flow Ra tes Nitrogen free p roduct gas composition (%) Exp. Theo* S eries No. at 3 (psi) Air (mol/h) ( ml/min ) ( mol/h ) Theo. (mol/h) CO CO 2 H 2 H2/ CO 2 H2/ CO 2 1 23.0 0.10 6 218 0. 542 0.482 0. 66 2 6.83 7 2.51 2. 7 2.5 8 2 2 6.6 0.1 85 255 0. 634 0. 644 0. 43 2 8 5 8 70.98 2. 48 2. 46 3 2 9.8 0.2 70 2 76 0. 684 0. 680 0.3 6 3 0.28 6 9 .3 6 2. 29 2 .22 4 3 3.5 0.33 9 2 88 0. 716 0. 704 0. 36 3 1 .1 1 6 8.54 2. 20 2 .02
73 is in excess, it is possible that the methanol decomposition reaction contributes to the overall reaction in this case. This i s supported by the higher CO concentration in the product compared to the other experiments at this feed flow rate. The agreement between the predicted and the measured production flow rates in the other experiments, suggest that the assumptions made in calculating the data in Table 3 7 are reasonable in those cases. The results indicate that the reformer can be used to produce sufficient hydrogen to run a 10 W fuel cell and the amount of catalyst in the reformer is sufficient to obtain 100% conversion of t he methanol The measured H2to CO2 ratios for the experiments are also close to the predicted ones. In all cases the measured H2to CO2 ratio is slightly higher than the predicted one. A contribution from the methanol decomposition reaction can explain a higher H2to CO2 ratio as this reaction produces H2 and CO and not CO2. In E xperiment #4, t he higher H2to CO2 ratio and gas production rate compared to the predicted ones, are likely due also to a larger contribution from the steam reforming reaction and a lower than predicted CPOX reaction In general the H2 to CO2 ratio is between 2 to 3, which is in agreement with the stoichiometries of SR (3:1) and CPOX (2:1 H2:CO2) reactions. The theoretical H2 to CO2 ratio for SR reaction is 3 and the ratio is 2 for CPOX reaction. In both Table s 37 and 3 8, the H2 to CO2 ratio is decreasing gradually as the air feed rate increases, which indicates an increasing contribution from the CPOX reaction as expected from the higher O2 concentrations 188.8.131.52 Catalyst c har acterization After completing the series of experiments on the steam reforming reaction with various amounts of oxygen present, it was observed that the catalyst in the different cha nnels exhibited different colors. In fact, the catalyst s particles in Channels 1119 are light er in color than the catalys t particles in Channels 20 29. This indicates a difference in the surface state of copper
74 between these catalysts. As the catalyst particles in Channels 2029 do have a red tint, this could indicate some presence of Cu metal, i.e. reduced catalyst. This would be expected after exposure to reductive conditions, such as the steam reforming reaction. The presence of CuO was confirmed with XPS on the catalyst particles in Channel 11 This appears to indicate that the CPOX reaction indeed does take place in Channels 11 19, while the SR reaction occurs in the later half of the catalyst filled channels. Figure 39 Catalyst s in reformer channels after experiments 184.108.40.206 Experimental results and theoretical predi ctions ( 5 channels of catalyst ) As the kinetics of the commercial catalyst used in the reformer is not evaluated, it is not known if the CPOX reaction is as fast as the kinetics from Lin et al (2007) suggests or if it significantly slower. Also, the results presented above does not reveal if the amount of catalyst
75 used in the reformer is more than enough to complete the reactions ( i.e. the amount of catalyst is excessive) If the reaction can be completed in less channel length, the reformer volume/ dimensi on can be smaller and more compact, which is in favor of the compact reformer target Table 3 9 Theoretical calculation of gas compositions and conversions (various feed rates) Exp eriment series # 1 2 3 4 5 6 7 8 O 2 feed rate (mol/hr) 0.0155 0.0263 0.03 70 0.0641 0.0162 0.0277 0.0401 0.0651 N2 feed rate (mol/hr) 0.0585 0.0988 0.1390 0.2410 0.0608 0.1043 0.1509 0.2449 CH 4 O feed rate (mol/hr) 0.102 0.102 0.102 0.102 0.145 0.145 0.145 0.145 Water feed rate (mol/hr) 0.046 0.046 0.046 0.046 0.066 0.066 0.06 6 0.066 CH 4 O consume d by CPOX (mol/h ) 0.031 0.053 0.074 0.102 0.032 0.055 0.080 0.130 Limiting reactant Water Water MeOH SR MeOH PO Water SR Water SR Auto thermal MeOH SR CH 4 O consumed by SR (mol/hr) 0.046 0.046 0.028 0 0.066 0.066 0.066 0.0148 H 2 gen erated by CPOX (mol/hr) 0.062 0.105 0.148 0.204 0.065 0.111 0.160 0.260 H 2 generated by SR (mol/hr) 0.138 0.138 0.084 0.000 0.198 0.198 0.198 0.044 CO 2 generated by CPOX (mol/hr) 0.031 0.053 0.074 0.102 0.032 0.055 0.080 0.130 CO 2 generated by SR (mol/h r) 0.046 0.046 0.028 0 0.066 0.066 0.066 0.0148 Total product gas rate (mol/hr) 0.336 0. 441 0.473 0. 560 0.422 0. 534 0. 654 0.695 Theoretical CH 4 O conversion % 75.57 96.57 100 100 67.82 83.75 100 100 Theoretical water conversion% 100 100 61 0 100 100 100 22 Theoretical H 2 vol.% (N 2 free) 72.20 71.16 69.47 66.67 72.76 71.78 71.03 67.76 Theoretical CO 2 vol.% (N 2 free) 27.80 28.84 30.53 33.33 27.24 28.22 28.97 32 .24 Experiments were therefore performed with the reformer only filled with 5 channels of catal yst and the rest of the reformer channels left empty. The experiments are similar to the experiments with all 19 channels filled with catalyst. As in the previous section, it is assumed that the CPOX occurs before the SR reaction and the calculated product flow rates and compositions based on the feed rates of each component is given in Table 39.
76 For comparison and easier evaluation of the contribution from the steam reforming reaction, the calculations were repeated for the case where only the CPOX reacti on takes place. In this case it is assumed that the CPOX reaction goes to completion in the first five channels, but no steam reforming will take place. The result s are presented in Table 3 10. Table 3 10 Theoretical calculation of gas compositions and co nversions (various feed rates for CPOX rxn only) Exp eriment series # 1 2 3 4 5 6 7 8 O 2 feed rate (mol/hr) 0.0155 0.0263 0.0370 0.0641 0.0162 0.0277 0.0401 0.0651 N 2 feed rate (mol/hr) 0.0585 0.0988 0.1390 0.2410 0.0608 0.1043 0.1509 0.244 9 CH 3 OH feed rate (mol/hr) 0.102 0.102 0.102 0.102 0.145 0.145 0.145 0.145 CH 3 OH cons umpt. rate (mol/hr) 0.031 0.053 0.074 0.102 0.032 0.055 0.080 0.130 H 2 generation rate (mol/hr) 0.062 0.106 0.148 0.204 0.065 0.111 0.160 0.260 CO 2 ge neration rate (mol/hr) 0.031 0.053 0.074 0.102 0.032 0.055 0.080 0.130 Total product gas rate (mol/hr) 0.152 0.258 0.361 0.560 0.158 0.270 0.391 0.635 Theoretical CH 3 OH conversion 30.39% 51.96% 72.55% 100% 22.07% 37.93% 55.17% 89.66% The oretical H2 vol.% (N2 free) 66.67% 66.67% 66.67% 66.67% 66.67% 66.67% 66.67% 66.67% Theoretical CO2 vol.% (N2 free) 33.33% 33.33% 33.33% 33.33% 33.33% 33.33% 33.33% 33.33% Total product gas rate (mol/hr) is the summation of H2 CO2 with Nitrogen generation rate. The actual measured product flow rates and product composition s are presented in Table 311. By comparing the data of the measured product gas flow rates to those presented in Table 39, it is evident that the measured results are lower in all cases comparing the theoretical estimations assuming complete CPOX and SR reactions. It is also notable that the measured H2/CO2 ratios are lower than predicted ratios in all cases (except in Experiment #8, where measured and predicted value s are similar, s ee Table 3 11) As only 5 of the 19 channels are filled with catalyst, it is expected that both the CPOX and the SR reactions are no t completed.
77 Table 3 11 Product gas flow rate as a function of air flow rate at two feed rates of methanol and water ( 0.102 and 0.046 mol/ h, as well as 0.145 and 0.066 mol/h) and product gas composition from gas chromotography. Exp. Feed Product Flow Rates Product gas composition (%) Exp. Theo.* S eries No. Air (mol/h) ( ml/min ) mol/h Theo. (mol/h)* CO CO 2 H 2 H2/ CO 2 H 2 / CO 2 Fe ed flow rates: 0.102 mol/h methanol and 0.046 mol/ h water. 1 0. 074 87.3 0. 217 0.336 0. 10 2 7.94 7 1.95 2. 58 2. 60 2 0. 125 10 0. 272 0. 441 0. 18 2 9.42 70.40 2. 39 2. 47 3 0.176 129 0. 320 0. 473 0. 18 3 2.05 6 7.77 2. 12 2 .28 4 0.3 05 156 0. 388 0. 560 0. 21 33. 53 66. 25 1.98 2 .00 Feed flow rates: 0.145 mol/h methanol and 0.06 6 mol/ h water. 5 0. 077 8 7.6 0. 218 0. 422 0. 15 2 8.40 7 1.44 2. 52 2. 67 6 0.1 32 110 0. 274 0. 534 0. 23 30.0 8 69.69 2. 32 2. 54 7 0. 191 158 0. 393 0. 654 0.3 0 3 1.95 6 7.75 2. 12 2 .45 8 0.3 10 176 0 .437 0. 695 0 21 3 2 .1 2 6 7.68 2. 11 2 .10 Theoretical prediti on (see Table 3 9) Therefore, it is important to also compare the measured results to the case of CPOX only, since this reaction is assumed to be the faster one. It can be seen that the measured product gas flow rates are higher than those predicted from the CPOX reaction at low air flow rates, but lower at high air flow rates. Table 3 11 reveal that the H2 concentrations for the low air flow rates are higher than those expected for CPOX reaction only (i.e. they are above 66.67%). This indicates that at the lower air flow rates the contribution from the steam reforming reaction is significant. In contrast, at high air flow rates the measured product gas flow rates and compositions are very close to the ones predicted from the CPOX reaction. Therefore, it appears
78 that the partial oxidation reaction is indeed fast and that no steam reforming takes place as long as there is oxygen present in the reactor. At least the steam reforming reaction will not be initiate d as long as the oxygen partial pressure is above a certain level. This level was not quantified in the study, but oxygen at concentration levels below 1% (~0.10.4%) could be detected in all reactions involving air. Comparing the results between the 5 and the 19 channels of catalyst (Tables 3 6, 38 and 311) it is evident that 5 channels of catalyst are not sufficient to complete the steam reforming reaction under these conditions. This is the reason that the product gas flow rates and the H2/CO2 ratio s are lower in the 5 channel catalyst experiment compared to the 19 channel catalyst experiment As more methanol is converted in the steam reforming reaction for the case of more catalyst channels, the product gas flow rate and the H2/CO2 ratio increase. Co nsidering that a significant amount of reforming occurs at low flow rates, it is likely that 19 channels are not needed for 100% conversion (or to r each the limiting conversion). 3.4 Heat T ra n sfer A spects of the R eformer Self sustainability is one of the important features of the reformer in that the heat produced by the exothermic CPOX reaction is sufficient enough to maintain the endothermic steam reforming reaction. Due to incomplete insulation, however, some heat is l ost to the environment and energy from an external source has to be provided unless the CPOX reaction can fully compensate for the heat loss to the environment. As it was difficult to identify the exact condition in which the heat produced by the CPOX counter balances the heat required by the steam reform ing reaction and the heat loss to the environment, the following experiments were conducted to assess the heat transfer characteristics of the reformer: With the methanol and the water feed rates fixed at 0.145 and 0.066 mol/h respectively, the
79 air feed rate was varied at three levels; 0.0 (i.e., no oxygen supply), 0.186 and 0.339 mol/h. Temperature of the heat controller was set at 250 C, and experiment was run until steady state was reached. Under the steady state condition, the heater mou nted on the reformer would be turned on and off intermittently due to the heat loss to the environment. Once the steady state is reached under each condition, experiment was run further for 10 to 15 more minutes. Then, the electrical supply to the heater was cut off and the temporal variation of the reformer temperature was monitored. Due to incomplete insulation of the reformer, the heat loss to the outside environment would result in gradual decrease of the temperature. In the absence of oxygen supply, only the steam reforming reaction, that is endothermic, would occur and the temperature decrease would be faster than the cases with oxygen supply. Since the steam reforming reaction is endothermic, the temperature decay is expected to be even faster than t he reference case in which no chemical reaction would occur because the catalyst was replaced with the glass beads. In Figure 3 10, the results for the temperature variation are given.
80 Figure 310 Temporal variation of the reformer temperature when the electrical power supply is cut off In the absence of oxygen supply, it took 1592 seconds (26.5 minutes) for the temperature to drop from 250C to 170 C whereas it took 2495 sec (41.6 min) and 5752 sec (95.9 min) when the air feed rate was 0.186 and 0.339 mol/h, respectively. Air feed rate of 0.186 mol/h is equivalent to the case of experiment #2 in Table 3 8, in which 54% of methanol is expected to be consumed by the CPOX reaction and the balance by the SR reaction. When the air feed rate is 0.339 mol/h ( equivalent to experiment #4 in Table 38), on the other hand, most methanol is expected to be consumed by the CPOX reaction. Thus, the time for the temperature decrease would take longer with the increasing air feed rate in accordance with the experimental observation. The reference case is also shown in the figure in which the catalyst was replaced with the inactive glass bead. The curve for the reference case is expected to be located above the
81 case for the s team reforming reaction only because the SR is endothermic inducing faster temperature decay. The experimental results are in fact in accordance with the expectation. It should be pointed out that temperature decrease was realized even when the methanol was consumed mostly by the CPOX reaction (i.e., w hen the air feed rate was 0.339 mol/h) although it took longer than 1.5 hours for the temperature to drop from 250C to 170C. It is mainly due to the fact that the rate of heat loss to the environment is still larger than the rate of heat generation by t he CPOX reaction alth ough the reformer is insulated us ing fiber glass. Heat B alance : The unsteady heat balance for the reformer is given as g out inE E E dt dE E (3 6) Here Eis th e rate of change of energy within the reformer; in E and outE are the input and output rates of energy to the system (i.e., reformer) accompanied by the materials going in and out of the reformer. gE is the net rate of heat generation that accounts for the heat generated by the chemical reactions as well as the heat loss to the environment. Thus, equation (36) is given as loss gen j j j i i iE E H m H m dt dE E (3 7) Here km and kH are the mass flow rate and the specific enthalpy of species k. The subscript i and j are the chemical species in the input and output streams, respectively. Because the input materials (i.e., methanol, water, oxygen and nitrogen) and their flow rates are specified, im and iH are known. However, the flow rates of the materials in the output stream, jm are not known unless the conversion is known. For the es timation of the transient
82 heat transfer characteristics, we considered the case in which the heat generated by the CPOX reaction is balanced with the heat consumed by the steam reforming reaction. That is 0 E H m H mgen j j j i i i (3 8) Thus, lossE dt dE (3 9) It represents the case in which the transient temperature of the reformer is strictly determined by the heat loss to the environment. Equation (39) is valid only when the conversion is 100% for the given flow rates of ideal mixture described in section 3.1.2. As long as the reformer temperature is above about 200C, both CPOX and SR reactions will persist with complete conversion. However, if the reformer temperature becomes lower than about 200C, the conversion will be less than 100% and equations (38) and (3 9) will become invalid. The temperature of the reformer may depend on both position and time. However, the spatial variation of the temperature may be negl igibl e because the thermal resistance within the reformer is much smaller than that for the outside environment including th e insulation. The ratio of these heat resistances is represented by the following Biot number  k hL Bic (3 10) Here k is the thermal conductivity of the material that constitutes the reformer (i.e., stainless steel), Lc is the characteristic length of the reformer and h is the overall heat transfer coefficient that accounts for the i nsulation as well as the air layer outside the insulation. That is, (3 11) where hair is the convective heat transfer coefficient of the air layer, kinsulation is the thermal conductivity of the insulation, and t is its t hickness insulation airk t h 1 h 1
83 If the Biot number is smaller than about 0.1, it is generally accepted that the spatial variation of the temperature within the reformer is negligible and the so called lumped capacitance method can be used for the thermal analysis in that only temporal variation of the reformer is considered for the transient heat transfer problem. For the present case, k for stainless steel is known to be 16.6 and 19.8 W/m K at 400K and 600K, respectively , Lc is 0.9 cm for the reformer, hair is about 10 W/ m2K for natural convection [ 94] the insulation thickness t is 2 cm, and the thermal conductivity of the insulation material is between 0.06 W/m K at 170C and 0.08 W/m K at 250C . Thus, h is calculated to be 2.9 W/m2K and the Biot number is 1.6103. Because the Biot number is much smaller than 0.1, t he lumped capacitance method can be used. Therefore, equation (39) is given as ) ( T T hA dt dT Vc (3 12) Here the density, volume and the heat capacity of the reformer. A is the su rface area of the reformer and h the overall heat transfer coefficient given in equation (311). T is the reformer temperature at time t and T is th e temperature of the air far away from the reformer surface. For constant values of and h, e quation ( 312) is integrated to be t Vc hA T T T Ti) exp( (3 13) w here Ti is the initial temperature of the reformer at t he time when the electrical heater is turned off. The heat capacity of stainless steel is known to be 0.528 J/mol K at 170C and 0.540 J/mol K at 250C [ 96]
84 0.750.80 0.85 0.90 0.95 1.00 0 1000 2000 3000 4000 5000 6000 7000 Time (s) 50% CPOX + 50% SR CPOX only equation (3-13) Figure 311 Temporal variation of [(T T)/(TiT)] when the electr ical power supply is cut off In Figure 311, the dimensionless temperature, = T T T T i, for two different experimental conditions is given along with the prediction by equation (313). Because equation (313) is close to the ideal case for which equation (3 8) is valid, it is appropriate to compare this prediction with the case of 50% CPOX and 50% SR reaction. As it was pointed out previously, equation (3 13) will not be valid if the temperature is lower than about 200C (i.e., < 0.78) and it is plotted until decreases to about 0.78. It is apparent that the expe rimental observation does not match well with the prediction by equation (313). This discrepancy may be due to the fact that equation (313) includes numerous simplifying assumptions such as dimensional simplification as well as uncertainties associated w ith the values for heat transfer coefficient and
85 material properties. It seems that a full three dimensional analysis has to be applied for better prediction of heat transfer characteristics of the reformer. When the oxygen feed rate is high, methanol is t otally consumed the CPOX reaction and the heat generation rate is maximum at about 5.3 W. According figure 310, the reformer temperature decreases even for this case indicating that the heat loss to the environment is greater than 5.3 W. According to equa tion (3 12), the instantaneous heat loss at t=0 is about 11 W which in fact is much larger than 5.3W as expected. According to figure 3 11, the initial temperature decay (i.e., the slope of at t=0) of the experimental observation appears to be larger than predicted value by equation (3 12) indicating that the actual heat loss at t=0 is even larger than 11 W. This result indicates that proper insulation of the reformer is very important for self sustainability and the current insulation is not sufficient. Nevertheless, all these results support that the prototype reformer is capable of meeting the preset requirements to produce enough hydrogen for 10W PEMFC application. It seems that its size probably can be reduced to make it even smaller. Smaller size would be also helpful for better insulation as the external surface area of the reformer would decrease. One aspect that could not be evaluated by the present study is longevity of the catalyst For such evaluation, the reformer may have to be run continuously for months. In the present study, though, the same catalyst was used for numerous intermittent runs that lasted for tens of hours over several months period.
86 C HAPTER 4 C ONCLUSIONS AND FUT URE WORK A new design idea for a compact reformer for portable applications has been proposed. This reformer, in a sense, is a conventional tubular reactor type because it consists of a cylindrical channel packed with catalyst particles although some design features such as interlaced channel structure are rather unique. Flow simulation has been conducted to determine the dimension of the reformer that is appropriate for a PEMFC with the electrical energy production rate of 10 W. A prototype reformer has be en built based on the simulation result and experiments have been conducted to assess the viability of the reformer u sing a commercial catalyst of CuO/ZnO/Al2O3 type and mixtures of methanol and water as the feed In the absence of oxygen, methanol convers ion higher than 80% c ould be obtained at a reaction temperature lower than 250 and at a moderately high pressure of about 2 atm When oxygen was added to the system in the form of air the catalytic partial oxidation (CPOX) reaction occurred along with t he steam reforming (SR) reaction achieving complete conversion of methanol. I t is likely that 19 channels of catalyst are not needed for 100% conversion (or to reach the limiting conversion) and the amount of catalyst may be reduced in fabricating the ref ormer The results of the present study support the viability of the new design idea in that the exothermic CPOX reaction occurs along with the endothermic SR reaction with efficient heat transfer characteristics. Other advantages of this reformer may include compactness, easy incorporation of the catalyst and efficient thermal management. Although the heat from the exothermic CPOX reaction can be used for the endothermic SR reaction eliminating the need for an external energy source for the reformer to function, an external energy source is still requi red for the startup operation. Furthermore, heat loss to the outside environment occurs due to incomplete insulation, and an external energy source may be necessary even during a steady state
87 operation unless the heat loss is also compensated by the exothermic reaction A s the heat loss to the environment is inevitable no matter how good the insulation may be accurate qua ntification of the heat transfer characteristics will be essential for a proper design of the reformer To better characterize the perfo rmance of the reformer, it will also be helpful if the temperatures inside the channels can be measured although it may be very difficult.
88 A PPENDIX HEAT OF REACTION AT THE OPERATING TEMPERATURE OF THE REFORM ER The heat of reaction at the operating temperature of the reformer (i.e., 250C) is calculated here using the values of the standard heat of reaction and the thermodynamic properties of the materials involved. CPOX R eaction The heat of reaction at 25C i s given by: CH3OH (l, 25C) + O2(g, 25C ) CO2(g, 25C) + 2H2(g, 25C) ( H0 = 154.9 kJ/mol) (1 6) where H0 is the standard heat of reaction. CH3OH(l, 25C) + O2(g, 25C ) CO2(g, 250C) + 2H2(g, 250C) ( H) where H is the heat of reaction at the operating temperature of the reformer, 250C. 0 CH 3 OH (l, 25C) + O 2 (g, 25C ) CO 2 (g, 25C) + 2H 2 (g, 25C ) CH 3 OH (l, 66.4C) + O 2 (g, 66.4C ) CH 3 OH (g, 66.4C) + O 2 (g, 66.4C ) CH 3 OH (g, 250C) + O 2 (g, 250C ) CO 2 (g, 250C) + 2H 2 (g, 250C ) 1 2 3 4 5
89 012345 1234= 05. And 5 can be calculated using the heat capacities of carbon dioxide and hydrogen. Heat capacity of CO2 H eat capacity (J/mol) At 25 C (gas) 0.0372 kJ/mol/ C H eat capacity (J/mol) At 2 50 C (gas) 0.0467 kJ/mol/ C heat from 25 C to 250 C 9.44 kJ/mol Cp(T) data. Here the average heat capacity is used and 25)=9.44 kJ/mol Heat capacity of H2 Heat capacity (J/mol) at 25 C (gas) 0.0288 kJ/mol/ C Heat capacity (J/mol) at 250 C (gas) 0.0288 kJ/mol/ C heat from 25C to 250C 6.49 kJ/mol heat capacity. Therefore, the heat of reaction of the CPOX reaction at 250 C is 05= 154.9( 9.446.49*2) = 132.48 kJ/mol
90 Steam R eforming R eaction CH3OH (l, 25C) + H2O (l, 25C) CO2(g, 25C) + 3H2(g, 25C) ( 0 = 130.9 kJ/mol) (1 3) CH3OH (l, 25C) + H2O (l, 25C) CO2 (g, 250C) + 3H2(g, 250C) ( ) 0 CH 3 OH (l, 25C) + H 2 O (l, 25C ) CO 2 (g, 25C) + 3H 2 (g, 25C ) CH 3 OH (l, 66.4C) + H 2 O (l, 66.4C ) CH 3 OH (g, 66.4C) + H 2 O (l, 66.4C ) CH 3 OH (g, 250C) + H 2 O (g, 250C ) CO 2 (g, 250C) + 3H 2 (g, 250C ) 1 2 3 4 5 CH 3 OH (g, 100C) + H 2 O (l, 100C ) CH 3 OH (g, 100C) + H 2 O (g, 100C ) 6 7
91 01234567 12345607 7 is agai n calculated using the heat capacity data. Thus, 07=130.9 ( 9.44 6.49*3) = 159.8 kJ/mol
92 LIST OF REFERENCES [ 1] M. Francis, Modeling: driving fuel cells, Materials Today 5(5) (2002) 34 39.  C. Song, Fuel processing f or low temperature and hightemperature fuel cells Challenges, and opportunities for sustainable development in the 21st century, Catalysis Today 77 ( 2002) 1749 [ 3] L. Carrette, K.A. Friedrich U. Stimming, Fuel cells Fundamentals and applications Fue l Cell 1 (2001) 539. [ 4] V. Mehta, J.S. Cooper, Review and analysis of PEMFC fuel cell design and manufacturing, J ournal of Power Sources 114 (1) (2003) 32 53.  J. St Pierre, D. P. Wilkinson, Fuel Cells: a New, Efficient and Cleaner Power Source, AIChE Journal 47(7) (2001) 14821486. [ 6] Fuel Cell Handbook, 7th edition, U.S. Dept. of Energy, Nov. 2004. [ 7] K. Sopian, W .R.W. Daud Challenges and future developments in proton exchange membrane fuel cells Renewable Energy 31 (5) (2006) 719727 [ 8] S .M. H aile Fuel cell materials and components, Acta Materialia 51 (19) (2003) 59816000. [ 9] G.F. McLean T. Niet, S. Prince Richard, et al., An assessment of alkaline fuel cell technology, International Journal of Hydr ogen Energy 27 (5) (2002) 507526. [ 10] F Alcaide, P Cabot, E Brillas Fuel cells for chemicals and energy cogeneration J ournal of Power Sources 153 (1) ( 2006) 47 60. [ 11] E. Antolini a J.R.C. Salgadob, E.R. Gonzalez a The stability of Pt M (M = first row transition metal) alloy catalysts and its effect on the activity in low temperature fue l cells A literature review and tests on a Pt Co catalyst J ournal of Power Sources 160 (2) ( 2006) 957968.  N. Sammes, R. Bove, K. Stahl, Phosphoric acid fuel cells: Fundamentals and applications Current Opinion in Solid State and Materials Science 8 ( 5) (2004) 372378. [ 13] L. J. Yu, G. P. Ren, X.M. Jiang, Experimental and analytical investigation of molten carbonate fuel cell stack, Energy Conversion and Management 49 (4) (2008) 873 879. [14 ] T. Ishikawa, H. Yasue, Start up, testing and operation of 1000 kW class MCFC power plant, Journal of Power Sources 86 (12) 2000. 145150  M. Bischoff, Molten carbonate fuel cells: A high temperature fuel cell on the edge to commercialization, Journal of Power Sources 160 (2) (2006) 842845.
93  P. Tomczy k, MCFC versus other fuel cellsCharacteristics, technologies and prospects, Journal of Power Sources 160 (2) (2006) 858862  S. Campanari, Carbon dioxide separation from high temperature fuel cell power plants, Journal of Power Sources 112 (1) (2002) 273289. [ 18] S. Scaccia, S.Frangini, S. Dellepiane, Enhanced O2 solubility by RE2O3 (RE = La, Gd) additions in molten carbonate electrolytes for MCFC, Journal of Molecular Liquids 138 (1 3) (2008), 107 112. [ 19] M. Z. Hong, S. C. Bae, H. S. Lee, et al. A study of the Cocoated Ni cathode prepared by electroless deposition for MCFCs, Electrochimica Acta 48 (28) (2003) 42134221.  S. P. S. Badwal Stability of solid oxide fuel cell components, Solid State Ionics 143 (1) (2001) 3946.  N. Q. Minh, So lid oxide fuel cell technology features and applications, Solid State Ionics 174 (14) (2004) 271277.  M. Yano, A. Tomita, M. Sano, Recent advances in singlechamber solid oxide fuel cells: A review, Solid State Ionics 177 (39 40) (2007) 33513359. [ 23] J. W. Fergus, Oxide anode materials for solid oxide fuel cells Solid State Ionics 177 (1718) (2006) 15291541.  M. Lyubovsky, S. Roychoudhury, Novel catalytic reactor for oxidative reforming of methanol, Applied Catalysis B: Environmental 54 (4) ( 2004) 203215. [ 25] J .H Wee Applications of proton exchange membrane fuel cell systems R enewable Sustainable Energy Reviews 11 (8) (2007) 17201738. [ 26] H.A. Gasteiger, J.E. Panels, S.G. Yan. Dependence of PEM fuel cell performance on catalyst loading J ournal of Power Sources 127 (1 2) ( 2004 ) 162 171. [ 27] S. Y. Hwa ng, H I Joh, M.A. Scibioh, et al., I mpact of cathode channel depth on performance of direct methanol fuel cells, Journal of Power Sources 183 (1) (2008) 226231. [ 28] E Bostic, N Sifer, C Bolton, et al., The US army foreign compa rative test fuel cell program, J ournal of Power Sources 137 (1) ( 2004) 7679. [ 29] R. Dillon, S. Srinivasan, A.S. Aric, et al. International activities in DMFC R&D: status of technologie s and potential applicat ions, Journal of Power Sources 127 (1 2) (2004) 112126. [ 30] V. Neburchilov, J. Martin, H. Wang, et al. A review of polymer electrolyte membranes for direct methanol fuel cells Journal of Power Sources 169 (2) (2007) 221238.
94 [ 31] J.R. RostrupNielsen, Conversion of hydrocarbons and alcohols for fuel cells. Phys. Chem. Chem. Phys., 3 (2001) 283288. [ 32] J.O. Jensen, Q. Li, C. Pan, et al, High temperature PEMFC and the possible utilization of the excess heat for fuel processing. 32 (2007) 15671571. [ 33] C.L.Yaws, Chemical properties handbook physical, thermodynamic, environmental, transport, safety, and health related properties for organic and inorganic chemicals. McGraw Hill (1999) [ 34] P.M. Biesheuvel, J.J.C. Geerlings, Thermodynamic analysis of direc t internal reforming of methane and butane in proton and oxygen conducting fuel cells. Journal of Power Sources 185 (2008) 11621167. [ 35] L. Zhou, Y. Guo, Q. Zhang, et al, A novel catalyst with plate type anodic alumina supports, Ni/NiAl2O4 Al2O3/alloy, for steam reforming of methane. Applied Catalysis A: General 347 (2008) 200207. [ 36] Y. Chen, Y. Wang, H. Xu, et al, Hydrogen production capacity of membrane reformer for methane steam reforming near practical working conditions. Journal of Membrane Science 322 (2008) 453459. [ 37] F. Gallucci, A. Basile, Pd Ag membrane reactor for steam reforming reactions: A comparison between different fuels. International Journal of Hydrogen Energy 33 (2008) 16711687. [ 38] P. Yaseneva, S. Pavlova, V. Sadykov, et al, Hydrogen production by steam reforming of methanol over CuCeZrYOxbased catalysts. Catalysis Today 138 (2008) 175182. [ 39] A. Mastalir, B. Frank, A. Szizybalski, et al, Steam reforming of methanol over Cu/ZrO2/CeO2 catalysts: a kineti c study. Journal of Catalysis 230 (2005) 464475. [ 40] Y. Liu, T. Hayakawa, K. Suzuki, et al, Highly active copper/ceria catalysts for steam reforming of methanol. Applied Catalysis A: General 223 (2002) 137145. [ 41] P. Kurr, I. Kasatkin, F. Girgsdies, et al, Microstructural characterization of Cu/ZnO/Al2O3 catalysts for methanol steam reforming A comparative study. Applied Catalysis A: General 348 (2008) 153164 [ 42] B.A. Peppley, J.C. Amphlett, L.M. Kearns, et al, Methanol steam reforming on Cu/ZnO/Al2O3. Part 1: the reaction network. Applied Catalysis A: General 179 (1999) 2129. [ 43] H. Purnama, T. Ressler, R.E. Jentoft, et al, CO formation/selectivity for steam reforming of methanol with a commercial CuO/ZnO/Al2O3 catalyst. Applied Catalysis A: Gen eral 259 (2004) 8394.
95 [ 44] G .L. Ohl J .L. Stein G .E. Smith, A Dynamic Model for the Design of Methanol to Hydrogen Steam Reformers for Transportation Applications, Journal of E nergy Resources Technology 126 (2) (2004) 149158. [ 45] Y. Lin, K.L. Hohn, S.M. StaggWilliams, Hydrogen generation from methanol oxidation on supported Cu and Pt catalysts: Effects of active phases and supports. Applied Catalysis A: General 327 (2007) 164172. [ 46] S. Velu, K. Suzuki, M. Okazaki, et al, Oxidative steam reforming of methanol over CuZnAl(Zr)Oxide Catalysts for the selective production of hydrogen for Fuel cells: Catalyst characterization and performance evaluation. Journal of Catalysis 194 (2000) 373384. [ 47] S. Murcia Mascars, R.M. Navarro, L. GmezSainero, et al, Oxidative methanol reforming reactions on CuZnAl catalysts derived from hydrotalcite like precursors. Journal of Catalysis 198 (2001) 338347. [ 48] S. Velu, K. Suzuki, M. P. Kapoor, et al, Selective production of hydrogen for fuel cells via oxidative steam reforming of methanol over CuZnAl(Zr) oxide catalysts. Applied Catalysis A: General, 213 (2001) 4763. [ 49] S. Velu, K. Suzuki, T. Osaki, Oxidative steam reforming of methanol over CuZnAl(Zr) oxide catalysts; a new and efficient method for the produ ction of CO free hydrogen for fuel cells. Chemical Communications, 1999 (23) 23412342. [ 50] M. Turco, G. Bagnasco, U. Costantino, et al, Production of hydrogen from oxidative steam reforming of methanol: I. Preparation and characterization of Cu/ZnO/Al2 O3 catalysts from a hydrotalcite like LDH precursor Journal of Catalysis, 228 (2004) 4355. [51 ] M. Turco, G. Bagnasco, U. Costantino, et al, Production of hydrogen from oxidative steam reforming of methanol: II. Catalytic act ivity and reaction mechanism on Cu/ZnO/Al2O3 hydrotalcite derived catalysts. Journal of Catalysis, 228 (2004) 5665. [ 52] T. L. Reitz, P. L. Lee, K. F. Czaplewski, et al, Time r esolved XANES i nvestigation of CuO/ZnO in the oxidative m ethanol r eforming r eaction Journal of Catalysis, 199 (2001) 193201. [ 53] S. Sakong, A. Gro Density functional theory study of the partial oxidation of methanol on copper surfaces. Journal of Catalysis, 231 (2005) 420429. [ 54] S. Rabe, F. Vogel, A thermogravimetric study of the partial oxidation of m ethanol for hydrogen production over a Cu/ZnO/Al2O3 catalyst. Applied Catalysis B: Environmental 84 (2008) 827834. [ 55] L. Zhou, S. Gnther, R. Imbihl, Low pressure methanol oxidation over a Cu(110) surface under stationary conditions: (II) adsorbate cove rages and reactivity. Journal of Catalysis 232 (2005) 295301.
96 [ 56] Y Choi, H .G. Stenger Kinetics, simulation and optimization of methanol steam reformer for fuel cell applications Journal of Power Sources 142 (1 2) (2005) 8191 [ 57] S.D. Jones, L.M Neal, H.E. HagelinWeaver, Steam reforming of methanol using CuZnO catalysts supported on nanoparticle alumina. Applied Catalysis B: Environmental 84 (2008) 631642. [ 58] D. Chadwick, K. Zheng, The importance of water in the mechanism of methanol decom position on ZnO. Catalysis Letters 20 (1993) 231242. [ 59] H. Yasuhisa, S. KenIchiro, K. Katsuki, et al, Selective oxidation of CO in H2 by permeation through catalytically active zeolite membranes. Journal of chemical engineering of Japan 35 (2002) 1244 1251. [ 60] Z. Li, W. Mi, J. Gong, et al, CO removal by twostage methanation for polymer electrolyte fuel cell. Journal of Natural Gas Chemistry 17 (2008) 359 364. [ 61] S.H. Oh, R.M. Sinkevitch, Carbon Monoxide Removal from HydrogenRich Fuel Cell Feedst reams by Selective Catalytic Oxidation, Journal of Catalysis, 142 (1) 1993 254262 [ 62] A Qi, B. Peppley, K Karan Integrated fuel processors for fuel cell application: A review, Fuel Processing Technology. 88 (1) (2007) 322. [ 63] P. Agnolucci, Economic s and market prospects of portable fuel cells, International Journal of Hydrogen Energy 32 (17) (2007) 43194328. [ 64] D.J. Moon, K. Sreekumar S D Lee et al. Studies on gasoline fuel processor system for fuel cell powered vehicles applicationApplied C atalysis A: General 215 (2001) 19 [ 65] A .S. Patil, T .G. Dubois, N Sifer, et al. Portable fuel cell systems for Americas army: technology transition to the field, Journal of Power Sources 136 (2) (2004) 220225. [ 66] K. Yamamoto, Y. Kawamura, N. Ogur a, et al., in: Proceedings of the 2004 Fuel Cell Seminar, San Antonio, TX, USA, 2004. [ 67] G Park, S Yima Y Yoon, et al., Hydrogen production with integrated microchannel fuel processor using methanol for portable fuel cell systems Journal of Power So urces, 145 (2) (200 5) 702706. [ 68] A Kundu, J .H Jang, H R Lee, et al., MEMS based micro fuel processor for application in a cell phone, J ournal of Power Sources, 162 (1) (2006) 572578. [ 69] O .J Kwon, S .M. Hwang, I .K Song, et al., A silicon based mi niaturized reformer for high power electric devices, Chem ical Eng ineering J ournal 133 (1 3) (2007) 157163 [ 70] P Rense, A Renken, K HassSanto, et al., Hydrogen production for fuel cell application in an autothermal micro channel reactor, Chem ical Eng ineering J ournal 101 (1 3) (2004) 133141.
97 [ 71] N. Edwards, J.C. Frost, A. Jones, et al., Self sustaining hydrogen generator (1998) US5,762,658. [ 72] J.M. Sohn, Y.C. Byun, J.Y. Cho, et al., D evelopment of the integrated methanol fuel processor using micro channel patterned devices and its performance for steam reforming of methanol. International Journal of Hydrogen Energy 32 (18) (2007) 5103 5108. [ 73] J .D. Holladay, E .O. Jones, M Phelps, et al., Microfuel processor for use in a miniature power supply, J ournal of Power Sources, 108 (1 2) (2002) 21 27. [ 74] K. Shah, R. S. Besser Understanding thermal integration issues and heat loss pathways in a planar microscale fuel processor: Demonstration of an integrated silicon microreactorbased methanol steam r eformer, Chemical Engineering Journal 135 (2008) S46S56 [ 75] T Kim, S Kwon Catalyst preparation for fabrication of a MEMS fuel reformer, Chemical Engineering Journal 123 (3) (2006) 93 102. [ 76] A. Kundu, J. H. Jang, H. R. Lee, et al., Micro fuel cells Current development and applications, J ournal of Power Sources 170 (1) (2007) 6778. [ 77] S. Hwang, O. J. Kwon, J. J. Kim. Method of catalyst coating in micro reactors for methanol steam reforming, Applied Catalysis A: General 316 (1) (2007) 8389. [ 78] L Pan, S Wang. A compact integrated fuel processing system for proton exchange membrane fuel cells, International Journal of Hydrogen Energy 31 (4) (2006) 447454. [ 79] D .E. Park, T Kim, S Kwon, et al. Micromachined methanol steam reforming system as a hydrogen supplier for portable proton exchange membrane fuel cells, Sensors and Actuators A: Physical 135 (1) (2007 ) 5866. [ 80] O. J. Kwon, D. H. Yoon, J.J. Kim. Silicon based miniaturized reformer with methanol catalytic burner Chemical Engineering J ournal 140 (1 3) (2008) 466472 [ 81] D. J. Seo, W. Y, Y. Yoon, et al., Development of a micro fuel processor for PEMFCs, Electrochimica Acta, 50 (2 3) (2004) 719723. [ 82] Y. Men, G. Kolb, R. Zapf, et al., A complete miniaturized microstructured methanol fuel processor/fuel cell system for low power applications, International Journal of Hydrogen Energy, 33 (4) (2008) 13741382. [ 83] T. Prescop, Fuel cells for portable computing and communications: Extended power away from the grid. UltraCell Corporation, 2007. [8 4] O. Zerbinati, C. Campisi, F. Guglielmino, Control unit for fuel cells for portable applications. 16th World Hydrogen Energy Conference, 13 16 June 2006, Lyon, France.
98 [ 85] J Agrell, H Birg ersson, M. B outonnet Steam reforming of methanol over a Cu/ZnO/Al2O3 catalyst: a kinetic analysis and strategies for suppression of CO formation J ournal of Power Sources 106 (12) (2002) 249 257 [ 86] J. Bear. Dynamics of Fluids in Porous Media. 1988 [ 87] W. L. McCabe, J. C. Smith, P. Harriott. Unit operations of chemical engineering ( 6th ). Published by McGraw Hill Companies, Inc. [ 88] S .E. Quiones Cisneros, C .K. Zberg Mikkelsen E .H. Stenby, The friction theory (f theory) for viscosity modeling, Fluid Phase Equilibria 169(2), (2000) 249276. [ 89] R.B. Bir d, W.E. Stewart, E.N. Lightfoot, Transport Phenomena (2nd Ed.), Wiley (2002) 2729. [ 90] C.J. Jiang, D.L. Trimm, M.S. Wainwright, et al, Kinetic study of steam reforming of methanol over copper based catalysts. Applied Catalysis A : 93 (1993) 245255. [ 91] C.J. Jiang, D.L. Trimm, M.S. Wainwright, et al, Kinetic mechanism for the reaction between methanol and water over a Cu ZnOAl2O3 catalyst. Applied Catalysis A : 97 (1993) 145158. [ 92] C. Chuang, Y. Chen, J.D. Ward, et al, Optimal design of an experimental methanol fuel reformer. International Journal of Hydrogen Energy 33 (2008) 70627073. [9 3] F.P. Incropera, D.P. DeWitt, Fundamentals of Heat and Mass Transfer (5th edition), J.Willey and Sons, New York, 2002 [ 94] C.J. Lasance, R.E. Simons, Advances In H ighPerformance Cooling For Electronic. 11 (2005) Number 4. [9 5] H. Wang, R.B. Dinwiddie, K.E. Wilkes, et al, G Plus report to Owens CorningThermal Conductivity Measurements of Fiberglass ORNL TM 2003/98. (April, 2003) [ 96] S.V. Stankus, I.V. Savchenko, A.V. Baginskii, et al, Thermal Conductivity and Thermal Diffusivity Coefficients of 12Kh18N10T Stainless Steel in a Wide Temperature Range. High Temperature, 46 (2008) 731733
99 BIOGRAPHICAL SKETCH Jing Su was born in Tianjin, China, on January 9, 1978. He received both Bachelor of Engineering and Master of Science degrees in Chemical Engineering Department at University of Tsinghua University, Beijing, China in 2000 and 2002, respectively. Then he worked in Sunny Chemical Co. Ltd. in Shanghai and Guangdong Province of China. He entered the doctoral research program in Chemical Engineering Department at University of Florida in August 2004.