<%BANNER%>

Investigation of Heterogeneous Surface Chemistry for Catalytic Combustion of Methane and Syngas

Permanent Link: http://ufdc.ufl.edu/UFE0022402/00001

Material Information

Title: Investigation of Heterogeneous Surface Chemistry for Catalytic Combustion of Methane and Syngas
Physical Description: 1 online resource (208 p.)
Language: english
Creator: Henry, Cary
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: catalyst, catalytic, combustion, laser, methane, palladium, platinum, raman, synthesis
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre: Mechanical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: In recent years, catalytic combustion has been intensely researched for its ability to reduce NOx emissions and the release of unburned hydrocarbons. One major point of interest is the use of catalytic reactors along with homogeneous combustors to create a hybrid combustor for power generation cycles. By using catlaysis along with traditional homogeneous combustion, the same amount of energy can be released from hydrocarbon fuels at a temperature lower than the adiabatic flame temperature, which results in a reduction in the produced NOx emissions. While studies on the activity of fuel lean catalytic reactors are well published, there have been few investigations for the fuel rich catalytic reactors. Another aspect of catalytic combustion which has little published research is the study of heat transfer effects initiated by a catalytically active surface. The focus of this research is to study and model the chemical kinetics for the fuel rich catalytic combustion of methane, hydrogen, and syngas as well as to investigate the effect of a catalytically active surface on the convective heat transfer coefficient of the catalyst surface.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Cary Henry.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Hahn, David W.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022402:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022402/00001

Material Information

Title: Investigation of Heterogeneous Surface Chemistry for Catalytic Combustion of Methane and Syngas
Physical Description: 1 online resource (208 p.)
Language: english
Creator: Henry, Cary
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: catalyst, catalytic, combustion, laser, methane, palladium, platinum, raman, synthesis
Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
Genre: Mechanical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: In recent years, catalytic combustion has been intensely researched for its ability to reduce NOx emissions and the release of unburned hydrocarbons. One major point of interest is the use of catalytic reactors along with homogeneous combustors to create a hybrid combustor for power generation cycles. By using catlaysis along with traditional homogeneous combustion, the same amount of energy can be released from hydrocarbon fuels at a temperature lower than the adiabatic flame temperature, which results in a reduction in the produced NOx emissions. While studies on the activity of fuel lean catalytic reactors are well published, there have been few investigations for the fuel rich catalytic reactors. Another aspect of catalytic combustion which has little published research is the study of heat transfer effects initiated by a catalytically active surface. The focus of this research is to study and model the chemical kinetics for the fuel rich catalytic combustion of methane, hydrogen, and syngas as well as to investigate the effect of a catalytically active surface on the convective heat transfer coefficient of the catalyst surface.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Cary Henry.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Hahn, David W.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022402:00001


This item has the following downloads:


Full Text

PAGE 1

INVESTIGATION OF HETEROGENEOUS SURFACE CHEMISTRY FOR THE CATALYTIC COMBUSTION OF METHANE AND SYNGAS By CARY ASHER HENRY 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 2008 1

PAGE 2

2008 Cary Asher Henry 2

PAGE 3

For my loving wife Crystal 3

PAGE 4

ACKNOWLEDGEMENTS I would like to express my sincerest gratitude to Dr. David Worthington Hahn for his help and guidance. His thorough knowledge of the subject matter and his enduring patience are greatly appreciated. I would like to express my sincerest gratitude to my fellow labmates (Patricia Dalyander, Prasoon Diwakar, Philip Jackson, Leia Shanyfelt and Bret Windom) for their help with various tasks, such as data analysis and acquisition. I would also like to thank Siemens Power Generation for their financial and intellectual support. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGEMENTS.............................................................................................................4 LIST OF TABLES...........................................................................................................................7 LIST OF FIGURES.........................................................................................................................8 ABSTRACT...................................................................................................................................15 CHAPTERS 1 LITERATURE REVIEW...........................................................................................................16 1.1 Introduction to Catalytic Combustion..............................................................................16 1.2 Hybrid Combustion..........................................................................................................17 1.3 Catalyst Preparation..........................................................................................................19 1.4 Reactor Geometry.............................................................................................................19 1.5 Platinum Based Catalysts.................................................................................................20 1.6 Palladium Based Catalysts................................................................................................21 1.7 Bimetallic Catalysts..........................................................................................................24 1.8 Trimetallic Catalysts.........................................................................................................28 1.9 Heat Transfer Effects........................................................................................................29 1.10 Pressure Effects..............................................................................................................30 1.11 Water Inhibition..............................................................................................................34 1.12 Sulfur Poisoning.............................................................................................................36 1.13 Reactor Modeling...........................................................................................................40 1.14 Kinetics...........................................................................................................................42 1.15 Gaseous Promoters.........................................................................................................43 1.16 Future Work....................................................................................................................45 2 FUNDAMENTAL SCIENCE AND BACKGROUND..............................................................60 3 EXPERIMENTAL METHODS..................................................................................................65 3.1 Reactor Specifications......................................................................................................65 3.1.1 Tubular Reactor......................................................................................................65 3.1.2 Stagnation Flow Reactor........................................................................................67 3.2 Experimental Apparatus...................................................................................................70 3.2.1 Tubular Reactor......................................................................................................70 3.2.2 Stagnation Flow Reactor........................................................................................72 3.3 Experimental Conditions..................................................................................................74 3.3.1 Tubular Reactor......................................................................................................74 3.3.2 Optical Experimentation.........................................................................................75 3.3.3 Heat Transfer Measurement...................................................................................76 3.3.4 Stagnation Reactor..................................................................................................77 5

PAGE 6

3.3.5 Optical Experimentation.........................................................................................78 4 RESULTS AND DISCUSSION.................................................................................................99 4.1 Tubular Reactor................................................................................................................99 4.1.1 Raman Spectroscopy..............................................................................................99 4.1.1.1 Calibration....................................................................................................99 4.1.1.2 Results and discussion................................................................................103 4.1.2 Heat Transfer Enhancement Studies....................................................................108 4.1.2.1 Linear Heat Transfer Model.......................................................................115 4.1.2.2 Exponential Heat Transfer Model..............................................................118 4.1.3 Catalytic Combustion Stoichiometry Studies.......................................................127 4.2 Stagnation Reactor..........................................................................................................131 4.2.1 Raman Spectroscopy............................................................................................132 4.2.1.1 Calibration..................................................................................................132 4.2.1.2 Results and discussion................................................................................135 4.2.2 Catalyst surface studies........................................................................................141 4.3 Summary of Work..........................................................................................................144 4.4 Future Work....................................................................................................................145 APPENDICES A REACTOR PART CAD DRAWINGS....................................................................................195 B CATALYST SURFACE STUDIES........................................................................................199 LIST OF REFERENCES.............................................................................................................203 BIOGRAPHICAL SKETCH.......................................................................................................208 6

PAGE 7

LIST OF TABLES Table page 1-1. Catalysts preparation methods and their loading, particle size, and dispersion [4]...............46 1-2. Activation energy and temperatures at which 10, 30, 50% CH4 conversions were reached for the complete oxidation of methane over Pd and Pt catalysts under standard dry reaction mixture a ...........................................................................................46 1-3. Temperatures at which 10, 30, 50% CH4 conversions were reached or the complete oxidation of methane over Pd and Pt catalysts under wet feed conditions........................46 1-4. Deutschmann reaction mechanism for the catalytic combustion of methane and hydrogen over platinum.....................................................................................................47 1-5. Catalytic activity results for Ni based catalyst......................................................................47 1-6. Common catalysts and their substrates, active states, and poisoning characteristics............47 3-1. Tubular reactor: all necessary components for the acquisition of Raman spectroscopic data.....................................................................................................................................79 3-2. Stagnation point flow reactor: all necessary components for the acquisition of Raman spectroscopic data..............................................................................................................80 3-3. Synthesis gas composition.....................................................................................................80 3-4. Flow rates for tubular reactor................................................................................................81 3-5. Investigated gaseous species and their respective Raman shifts and wavelengths due to laser excitation at 355 nm..................................................................................................81 3-6. Flow rates for stagnation reactor...........................................................................................81 4-1. Typical operating temperatures of the tubular reactor for various fuel feeds.....................146 4-2. Flow Rates for the stoichiometric studies performed on the tubular catalytic reactor........146 4-3. Reduced flow rates for the stagnation reactor burn data.....................................................146 4-4. Surface characteristics of the stagnation point flow reactor catalyst samples.....................146 B-1: Values of effective surface area ratio for the reacting and non-reaction zones on Pt and Pd catalysts.......................................................................................................................201 7

PAGE 8

LIST OF FIGURES Figure page 1-1. Conventional flame combustor and catalytic combustor......................................................48 1-2. Methane conversion on Pd/Al 2 O 3 catalyst. (a) reduced with H2; (b) pretreated with reactant mixture after reduction.........................................................................................48 1-3. Methane conversion as a function of feed temperature on alumina based catalysts, prepared by impregnation (Pd/Al(I)) and sol-gel (Pd/Al(SG) and Pd/Al(SG)Pd).............49 1-4. Methane conversion as a function of feed temperature on titania based catalysts, prepared by impregnation (Pd/Ti(I)) and sol-gel (Pd/Ti(SG) and Pd/Ti(SG)Pd)..............49 1-5. Methane conversion versus Pd Oxidation..............................................................................50 1-6. Activity tests when temperature was varied stepwise for Pd-ref (), PdNi (), and PdPt (). The dotted line represents the feed temperature to the catalysts..............................50 1-7. Combustion rate of methane with increasing temperature at 10 K min -1 ...............................51 1-8. Combustion rate of methane at 800C with time on stream...................................................51 1-9. Temperature profile of bulk fluid flow near heated wall........................................................51 1-10. Influence of pressure on methane conversion at feed temperatures of 500C on Pd/Al2O3 () and PdPt/Al2O3 ()..................................................................................52 1-11. Influence of pressure on PdO/Pd transformation temperature.............................................52 1-12. Rate of methane conversion as a function of pressure.........................................................53 1-13. Rate of methane conversion over Pd/SnO 2 as a function of water vapor feed concentration......................................................................................................................53 1-14. Rate of methane conversion over Pd/Al 2 O 3 as a function of water vapor feed concentration......................................................................................................................54 1-15. Rate of methane conversion over Pd/Al 2 O 3 -36NiO as a function of water vapor feed concentration......................................................................................................................54 1-16. Rate of methane conversion over noble metal catalysts: (a) Pd/Al 2 O 3 ; (b) Pt/ Al 2 O 3 ; (c) Rh/ Al 2 O 3 under varying conditions: (rhombus) freshly reduced catalyst; () poisoned; () regenerated.................................................................................................55 1-17. Rate of methane conversion over Pt/ Al 2 O 3 and the effects of sulfur compounds introduced into the feed stream..........................................................................................56 8

PAGE 9

1-18. Methane conversion over fresh (open symbols) and aged (filled symbols) Pd/Al2O3 (,) and Pt/Al2O3 (,) catalysts versus temperature in wet feed..................................56 1-19. Influence of 10 vol % water vapor addition on methane conversion over fresh (open symbols) and aged (filled symbols) Pd/Al2O3 (,) and Pt/Al2O3 (,) catalysts versus time on stream.........................................................................................................57 1-20. Influence of 100 vol ppm H 2 S addition on methane conversion over fresh (open symbols) and aged (filled symbols) Pd/Al2O3 (,) and Pt/Al2O3 (,) catalysts versus time on stream.........................................................................................................58 1-21. Methane conversion on Pd/Al 2 O 3 catalyst. (a) sulfur free feed; (b) sulfur containing feed.....................................................................................................................................59 1-22. Effect of water vapor inlet concentration. Dots: experimental points; lines: model predictions..........................................................................................................................59 2-1. Physical process of scattering produced from an induced dipole moment.............................63 2-2. Vibrational displacement of the atoms about the equilibrium position..................................63 2-3. Vibrational energy well..........................................................................................................64 2-4. Vibrational energy well showing Raman shifted emission of light........................................64 3-1. Tubular reactor flow schematic..............................................................................................82 3-2. Tubular reactor solder joint flow schematic...........................................................................83 3-3. Downstream catalyst tube holder...........................................................................................84 3-4. Alli-Cat Scientific flow meters used for flow control on both the tubular reactor and the stagnation point flow reactor.............................................................................................85 3-5.Cross-section schematic of the catalytic reactor......................................................................86 3-6. Tubular reactor.......................................................................................................................87 3-7. Cross-sectional view of the tubular reactor optical access window......................................88 3-8. Catalyst puck shown with co-flow and reacting air nozzle above and with the cooling nitrogen nozzle below........................................................................................................89 3-9. Stagnation Reactor.................................................................................................................90 3-10. Top view of experimental apparatus used for gas analysis through the tubular reactor via Raman scattering..........................................................................................................91 9

PAGE 10

3-11. Top view of experimental apparatus used for gas analysis through the stagnation point fow reactor via Raman scattering......................................................................................92 3-12. Example of 406 nm window spectral analysis for tubular reactor......................................93 3-13. Example of 380 nm window spectral analysis for tubular reactor......................................94 3-14. Schematic of tubular reactor with temperature sampling for heat transfer measurements.....................................................................................................................95 3-15. Temperature profile for CH 4 steady state burn.....................................................................96 3-16. Sample spectra taken at window centered at 406 nm..........................................................97 3-17. Sample spectra taken at window centered at 380 nm...........................................................98 4-3. Calibration plot showing calibration points with linear curve fit. Notice the bundle of four points in the center of the plot. These points are the 10 vol % cold calibration points and the 20 vol % hot calibration points.................................................................149 4-4. Data points showing the calculated mole fraction of CH 4 at the entrance to the reactor and at the two optical access ports. Results are shown for both ignition test conditions and steady state test conditions for data taken in Spring 2007. Error bars represent one standard deviation......................................................................................150 4-5. Data points showing the calculated mole fraction of CH 4 at the entrance to the reactor and at the two optical access ports non-dimensionalized by the mole fraction of CH 4 at the entrance to the reactor. The catalyst distance is non-dimensionalized by the total length of the catalyst coating. Results are shown for both ignition test conditions and steady state test conditions for data taken in Spring 2007. Error bars represent one standard deviation......................................................................................151 4-6. Data points showing the calculated mole fraction of H 2 at the entrance to the reactor and at the two optical access ports. Results are shown for both ignition test conditions and steady state test conditions for the data taken in Spring 2007. Error bars represent one standard deviation..............................................................................152 4-7. Data points showing the calculated mole fraction of H 2 at the entrance to the reactor and at the two optical access ports non-dimensionalized by the mole fraction of H 2 at the second optical port for the steady state case. The catalyst distance is non-dimensionalized by the total length of the catalyst coating. Results are shown for both ignition test conditions and steady state test conditions for the data taken in Spring 2007. Error bars represent one standard deviation..............................................153 4-8. Sample spectra for 406 nm centered window for a synthesis gas burn. As can be seen, the H 2 peak is just below the minimum limit of detection. Also shown in the plot is the window interference peak which doesnt allow the collection of H 2 O data..............154 10

PAGE 11

4-9. Cooling flow temperature profile for CH 4 combustion showing the measured points on 1 cm intervals and the interpolated points on 1 mm intervals.........................................155 4-10. Cooling flow temperature profile for CO combustion showing the measured points on 1 cm intervals and the interpolated points on 1 mm intervals.........................................156 4-11. Calculated heat flux to the inner cooling flow for the catalytic combustion of CH 4 ........157 4-12. Total heat flux, heat flux to cooling flow, and heat flux to reacting flow for the case of a single exponential decay total heat release along the length of the reactor for the catalytic combustion of CH 4 ............................................................................................158 4-13. Total heat flux to the reacting flow, classical calculation of heat flux to the reacting flow, and catalytic heat flux to the reacting flow for the single exponential decay fit for the catalytic combustion of CH 4 .................................................................................159 4-14. Total heat flux and catalytically enhanced heat flux for the case of an initial exponential decay total heat release followed by a secondary heat release along the length of the reactor for the catalytic combustion of CH 4 ...............................................160 4-15. Total heat flux, heat flux to cooling flow, and heat flux to reacting flow for the case of a single exponential decay total heat release along the length of the reactor for the catalytic combustion of CO.............................................................................................161 4-16. Total heat flux to the reacting flow, classical calculation of heat flux to the reacting flow, and catalytic heat flux to the reacting flow for the single exponential decay fit for the catalytic combustion of CO..................................................................................162 4-17. Enhancement to the convective heat transfer to the reacting flow due to catalytic surface reactions for both CH 4 and CO............................................................................163 4-18. Modeled and calculated enhancement to the convective heat transfer from a catalytically active surface during the combustion of CH 4 ..............................................164 4-19. Tubular operating temperatures during the catalytic combustion of CH 4 as a function of equivalence ratio. T 1 is the reacting flow inlet temperature, T 2 is the downstream mixing temperature, T P1 is the surface temperature at the first optical port, and T P2 is the surface temperature at the second optical port. The line indicates the standard operating conditions used during the Raman spectroscopic studies. The upper flammability limit of CH 4 is at an equivalence ratio of 1.64...........................................165 4-20. Total heat release as calculated via complete rich oxidation of CH 4 and measured temperature rise as a function of equivalence ratio.........................................................166 11

PAGE 12

4-21. Tubular operating temperatures during the catalytic combustion of syngas as a function of equivalence ratio. T 1 is the reacting flow inlet temperature, T 2 is the downstream mixing temperature, T P1 is the surface temperature at the first optical port, and T P2 is the surface temperature at the second optical port. The line indicates the standard operating conditions used during the Raman spectroscopic studies..............................................................................................................................167 4-22. Sample H 2 O calibration spectra for the stagnation reactor showing the H2O peak and no interference from the windows...................................................................................168 4-23. Raman spectra for pure nitrogen flow showing interference line from 532 nm residual ghost lines and no signal from O 2 in the ambient atmosphere.........................................169 4-24. Raman spectra for pure nitrogen flow showing interference line from 532 nm residual ghost lines and O 2 signal resulting from the flow of air through the reacting nozzle. This Figure along with Figure 4-10 show that it is possible to monitor the O 2 signal in the stagnation reactor...................................................................................................170 4-25. Raman spectra for CO2 calibration flow with and without the 532 nm low pass filter installed in front of the collection optics. Notice how the filter removes the interference lines resulting from the 532 nm residual laser light.....................................171 4-26. Sample spectra centered at 406 nm for the steady stat burn of H2. The spectra shows the Raman scattering peaks for H2O and H2...................................................................172 4-27. Concentration of H2 and O2 as a function of vertical distance from the catalyst surface during the steady state combustion of H2...........................................................173 4-28. Concentration of CO, CO2, H2, and O2 as a function of vertical distance from the catalyst surface during the steady state combustion of synthesis gas..............................174 4-29. Co-flow nozzle test showing Raman signal intensities of the 20 vol % CH4 signa as a function of height from the catalyst disk.........................................................................175 4-30. Concentration of H 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of H 2 gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.................................................................176 4-31. Concentration of H 2 O as a function of vertical distance from the reacting gas nozzle during the steady state combustion of H 2 gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.................................................................177 4-32. Concentration of O 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of H 2 gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.................................................................178 12

PAGE 13

4-33. Concentration of H 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.........................................................179 4-34. Concentration of CO as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.........................................................180 4-35. Concentration of CO 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.........................................................181 4-36. Concentration of H 2 O as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.........................................................182 4-37. Concentration of O 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.........................................................183 4-38. Concentration of H 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of H 2 gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.................................................................184 4-39. Concentration of H 2 O as a function of vertical distance from the reacting gas nozzle during the steady state combustion of H 2 gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.................................................................185 4-40. Concentration of O 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of H 2 gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.................................................................186 4-41. Concentration of CO as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.........................................................187 4-42. Concentration of CO 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.........................................................188 4-43. Concentration of H 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.........................................................189 4-44. Concentration of H 2 O as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.........................................................190 13

PAGE 14

4-45. Concentration of O 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm.........................................................191 4-46. Surface structure image for the non-reacting platinum surface.........................................192 4-47. Surface structure image for the reacting platinum surface................................................192 4-48. Surface structure image for the non-reacting palladium surface of the overheated catalyst.............................................................................................................................193 4-49. Surface structure image for the reacting palladium surface of the overheated catalyst....193 4-50. Surface structure image for the non-reacting palladium surface.......................................194 4-51. Surface structure image for the reacting palladium surface..............................................194 A-1.Window Holder Schematic...................................................................................................195 A-2. Reacting gas nozzle holder.................................................................................................196 A-3. Catalyst Lifting Plate..........................................................................................................197 A-4. Catalyst cooling nozzle plate..............................................................................................198 A-5. Catalyst holder....................................................................................................................199 B-1: Nodal points for calculation of the minimum effective surface area..................................201 B-2: Herons formula for calculation of triangle side length......................................................202 B-3: Method of analysis to calculate the maximum effective surface area................................202 14

PAGE 15

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy INVESTIGATION OF HETEROGENEOUS SURFACE CHEMISTRY FOR THE CATALYTIC COMBUSTION OF METHANE AND SYNGAS By Cary Henry August 2008 Chair: David W. Hahn Major: Mechanical Engineering In recent years, catalytic combustion has been intensely researched for its ability to reduce NO x emissions and the release of unburned hydrocarbons. One major point of interest is the use of catalytic reactors along with homogeneous combustors to create a hybrid combustor for power generation cycles. By using catlaysis along with traditional homogeneous combustion, the same amount of energy can be released from hydrocarbon fuels at a temperature lower than the adiabatic flame temperature, which results in a reduction in the produced NO x emissions. While studies on the activity of fuel lean catalytic reactors are well published, there have been few investigations for the fuel rich catalytic reactors. Another aspect of catalytic combustion which has little published research is the study of heat transfer effects initiated by a catalytically active surface. The focus of this research is to study and model the chemical kinetics for the fuel rich catalytic combustion of methane, hydrogen, and syngas as well as to investigate the effect of a catalytically active surface on the convective heat transfer coefficient of the catalyst surface. 15

PAGE 16

CHAPTER 1 LITERATURE REVIEW This chapter presents an in depth introduction to several aspects important to the catalytic oxidation of methane and syngas. Topics such as catalyst poisoning and sintering, varied catalyst coatings, varied catalyst supports, catalyst preparation, etc will be discussed in this chapter. This material is presented in the form of a literature review of many significant papers recently published for varying aspects of catalytic combustion. 1.1 Introduction to Catalytic Combustion Catalytic combustion has been researched very extensively in recent years due to its wide range of possible applications such as power generation and due to its inherent low emission qualities. Catalytic combustion for the application of power generation is of increased interest due to the positive environmental impacts of producing heat and energy at lower temperatures than standard homogeneous combustion, which in turn reduces the emission of harmful pollutants such as CO, NO x and excess hydrocarbons. Since the lower reacting temperature of catalytic combustion mitigates the formation of NO x catalytic combustion reduces the need for indirect exhaust-gas aftertreatment techniques. Another advantage of lower combustion temperatures is the elimination of cooling techniques after the combustor and prior to the turbine entrance. Since current power generation cycles have a combustion temperature of up to 1500C, the exhaust must be cooled to temperatures as low as 1100C so as not to damage the turbine components. With catalytic combustion, the combustion temperature is lower and can be controlled so as not to damage the turbines. Most recent designs for catalytic combustion used in power generation call for the implementation of a hybrid combustion concept. A schematic of a typical hybrid combustor can be seen in Figure 1-1. 16

PAGE 17

Hybrid combustion consists of a rich burn catalytic reactor where all of the fuel and a portion (5 to 25%) of the airflow is flown through a catalytic reactor. After the catalytic combustion section, the catalytic exhaust stream is mixed with the other remaining airflow and is combusted homogeneously. This hybrid combustion allows all of the energy to be released from the methane at a lower charge temperature, which reduces NO x emissions, and eliminates or reduces the need for exhaust cooling. Another important characteristic of catalytic combustion is that it can be performed over a wide range of air to fuel ratios. This wide operating range allows one to control the energy release and combustion temperature by varying the operating air to fuel ratio. Another practical use of catalytic combustion is for the reduction of unburned hydrocarbon emissions from natural gas fueled vehicles. In general, natural gas fueled vehicles have been shown to reduce harmful NO x and particulate emissions. Natural gas vehicles can operate under lean conditions, which allow an increase in the fuel efficiency of the vehicle. However, one point of concern with natural gas fueled vehicles is the emission of unburned methane, which is a potent greenhouse gas [1]. In order to reduce these emissions, catalytic converters are installed in the exhaust stream of these vehicles to oxidize the excess methane. This use of catalytic combustion presents the unique problem of catalyst poisoning due to water vapor and sulfur containing compounds found in the exhaust stream of natural gas fueled vehicles. Catalysts that are resistant to water and sulfur poisoning must be designed and tested in order to meet these strenuous operating requirements. 1.2 Hybrid Combustion Dalla Betta et al. [2] have thoroughly studied the use of catalytic combustion of methane for achieving ultra low NO x emissions in power generation cycles. The design used in their work consists of a catalytic combustor at the exit of the compressor, followed by homogeneous 17

PAGE 18

combustion prior to the turbine. This is the most commonly studied design for the application of power generation. Their design flows all of the CH 4 and air over the catalyst at 2 vol % CH 4 which is designed to only partially combust prior to complete homogeneous combustion. This catalytic pre-combustor allows all of the energy contained in the fuel to be extracted at lower temperatures, hence resulting in a significant decrease in NO x emissions. These tests were carried out on a honeycomb monolith palladium catalyst with an inlet pressure of 11 atm up to a catalytic reactor outlet temperature of 700 C. This is much lower than the typical combustor outlet temperature of approximately 1500 C. Seo et al. [3] have studied a catlytically stabilized combustor consisting of a packed bed catalytic reactor with a downstream homogeneous combustor. The system was very similar to that of Betta [2], hence the entire mixture of air and fuel flows through the catalytic combustor, and is then combusted in a standard gas phase combustor. Seo et al. [3] have compared packed bed reactors with platinum and palladium as the active catalyst on a La-Mn-hexaaluminate support, as well as a combination of the two. It was found that the palladium based catalyst resulted in the lowest lightoff temperature. It was also found that the platinum catalyst lowers the flame ignition temperature in the gas phase combustor. These two results are used to design a dual stage reactor consisting of a first stage comprised entirely of palladium and a second stage comprised entirely of platinum. The reactor is conFigure d in this manner because it was shown that only the catalyst at the beginning of the reactor affects the catalyst lightoff temperature, and the catalyst at the exit of the reactor more heavily affects the gas phase ignition temperature. It is also shown that placement of a catalytic reactor upstream of the gas phase combustor results in increased flame speeds over standard gas phase combustors. They have shown that a hybrid combustor with 57% catalytic conversion results in a flame speed 19 times higher than in a 18

PAGE 19

standard gas phase combustor. It was also shown that the hybrid combustor resulted in significant decreases in CO emission of approximately 90% as compared to the gas phase combustor. 1.3 Catalyst Preparation Catalysts can be prepared with precursors of chloride, nitrates, and acetylacetonate. These precursors are dried and calcined under heated airflow, which results in the deposition of metal particles onto the support. Another method of catalyst preparation is the polyol method, in which the precursor and the support are dispersed and heated in a liquid polyol, which consists of the active catalyst suspended in liquid ethylene glycol. This process involves a redox reaction between the precursor and the polyol, which leads to the deposition of metal particles on the support [4]. Simplicio et al. [4] have studied the effects of various palladium precursors for the combustion of methane. Their research showed that the catalysts formed by the polyol (PAAP) method and the palladium acetylacetonated (PAA) resulted in a satisfactory dispersion of PdO particles on alumina support, which can be seen in table 1-1. The higher dispersion of these precursors resulted in catalysts with the highest activites and thermal stabilities. The authors claim that there is a relation between the PdO particle size of the catalyst and the methane conversion achieved by said catalyst. Specifically, of the catalysts studied, the smallest PdO particle size and the highest PdO dispersion resulted in the highest activity. 1.4 Reactor Geometry Geus et al. [5] have studied monolith reactors for catalytic oxidation. One of the major advantages of monolith reactors is the minimal pressure drop. This low pressure drop allows extremely high flow rates corresponding to Reynolds numbers of 300,000 or more. This is very beneficial for the application of catalytic combustion for power generation applications due to 19

PAGE 20

the high throughputs required in power generation cycles. However, the temperature stability of currently available platinum and palladium monoliths limits reactors to surface temperatures of approximately 800 C. For this reason, ceramic monoliths have been created which can withstand surface temperatures greater than their metallic counterparts. The main shortfall of catalytic monoliths is the significant reduction in heat transfer through the monolith, which can result in major temperature gradients throughout the monolith. 1.5 Platinum Based Catalysts Dupont et al. [6] have studied the combustion of methane on platinum surfaces. This study was performed on platinum foil in a stagnation point flow reactor. One of the benefits of stagnation point flow reactors is the validity of the assumption that concentration gradients are only significant in one dimension. This work compares the experimental and predicted fuel conversion efficiencies and CO product selectivities as a function of the catalyst surface temperature up to and including combustion in the homogeneous regime. It was found that the ageing of the foil showed little effect on the absolute CH 4 conversion. It was also found, however, that the temperature at which the reactor returns to the heterogeneous regime increases with ageing of the catalyst. This means that as the catalyst ages, the minimum temperature to sustain homogeneous combustion increases. It was also found that CO selectivity remained at or near zero until homogeneous ignition was achieved. A study on nitrogen dilution found that a change in the amount of nitrogen present in the combustion air had no effect on either the homogeneous ignition temperature or the catalytic conversion of methane. This behavior is indicative of a very fast single global reaction for the combustion of CH 4 with CO 2 This theory was corroborated by the model used for the conversion. A decrease in the nitrogen dilution was shown to dramatically increase the conversion of methane in the homogeneous regime. This behavior is due to the fact that a reduction of nitrogen in homogeneous combustion results in a 20

PAGE 21

reduction in heat capacity of the reacting gas, and thus results in higher reacting temperatures and reaction rates. The CO selectivity also decreased with a decrease in the nitrogen dilution. This study showed that for conversion of methane in both the catalytic and homogeneous regime can sufficiently be simulated by a global chemical reaction approach. 1.6 Palladium Based Catalysts One of the most important benefits of palladium based catalysts over other noble metal catalysts, such as Platinum, is that palladium based catalysts are highly active for methane oxidation. This high activity towards methane oxidation results in ignition at lower temperatures, which is very important for the application of catalytic combustion for the purpose of power generation. Previously, inline heaters have been necessary after the compressor in order to raise the gas temperatures to a suitable ignition temperature. The low ignition temperature exhibited by palladium based catalysts may no longer require the addition of these inline heaters. It is agreed that the active phase is PdO [7-11], and this active phase begins to decompose to the less active Pd at feed temperatures of approximately 700-800 C. Lee et al. [12] were one of the first groups to have studied the catalytic combustion of methane over palladium based catalysts. Their preliminary studies show that the catalytic activity of palladium is the highest among monometallic catalysts. They have also listed six factors for governing the rate of conversion of methane: Feed ratio: it was found that under fuel lean conditions, methane is only oxidized to CO 2 whereas under fuel rich conditions methane is also oxidized to CO. Catalyst loading: the authors have shown that under low loading conditions, increasing the catalyst loading results in an increase of activity until the activity reaches a maximum. Particle size: their research indicates that for both platinum and palladium catalysts, larger particles are more active than smaller particles. Pretreatment gasses: pretreatment with H 2 and reactant mixtures such as O 2 and CH 4 resulted in an increase in activity (see Figure 1-2), whereas pretreatment with O 2 only led to a decrease in catalyst activity. H 2 O feed: the presence of H 2 O in the feed is shown to increase formation of CO 2 21

PAGE 22

Support material: they have found that varying the catalyst support results in a change in the rate of conversion of methane. They have identified SiO 2 -Al 2 O 3 as the most active support for platinum catalysts. Demoulin et al. [13] have studied the catalytic combustion of methane and carbon monoxide over Al 2 O 3 supported palladium catalysts. This study focused on surface reactions and the reaction pathways. When CO was fed to the catalyst, it was first shown to react with the oxidized catalyst to produce CO 2 Once the PdO surface was consumed, the CO reacted with the Pd surface to produce Pd-CO. This Pd-CO surface is only formed at higher temperatures (>400C) when there is insufficient O 2 to convert the CO into CO 2 When CH 4 was fed to the catalyst, there were no adsorbed CH 4 species detected. Once again, the only surface species formed were Pd-CO species. Some formates/carbonates were also found to form on the surface at low to moderate temperatures with a sufficient CO feed. The reaction pathway proposed by the authors is the following: The oxidized surface of the catalyst is covered with highly reactive PdO The CH 4 feed reacts with this oxidized outer layer to produce CO, and thus creates O vacancies on the surface of the catalyst These vacancies must then be reoxidized by gaseous O 2 or by oxygen coming from particles, if there is a lack of feed O 2 The CO can then adsorb on the surface for Pd-CO, react with the support, or form CO 2 This CO reaction is highly dependent on the reaction conditions such as temperature and oxygen availability. Demoulin et al. [14] have studied the origin of transient species present on the surface of palladium catalysts supported on Al 2 O 3 during catalytic combustion of methane. This study showed that the initial oxidation state of the palladium catalyst had no effect on the activity of the catalyst. The authors have also found that formate/carbonate species are formed for feed temperatures less than 400 C, and disappear with the onset of CO 2 production at higher 22

PAGE 23

temperatures. The authors attribute the generation of these species as the product of reactions between CO and the catalyst support. Carrazin et al. [15] have studied the effects of varying supports for the catalytic combustion of methane over palladium. This study focused on the comparison of two differing substrates namely, Al 2 O 3 and TiO 2 and two preparation methods: impregnantion and sol-gel. The sol-gel method consists of two cases: one where the palladium precursor is added to the mixture before gel formation, and one where the palladium precursor is added to the mixture after gel formation. For the Al 2 O 3 substrate, it was shown that the sol-gel method where the palladium precursor is added after gel formation was the most active for mid range temperatures (250-450C). For feed temperatures outside of this range, the conversion of the three preparation methods is almost identical (Figure 1-3). For the TiO 2 substrate, it was shown that sol-gel method where the palladium precursor is added after gel formation is almost completely inactive. The other two methods show very similar activity except for feed temperatures greater than 400C, where the catalyst formed by the sol-gel method with palladium precursor addition before gel formation achieves an increase in conversion over the impregnation method (Figure 1-4). It should also be noted that all of the Al 2 O 3 substrate catalysts outperformed the TiO 2 substrates for the tested feed temperatures (200-500C). These differences in activities can be attributed to variations in the crystallite size, dispersion of palladium particles, and the BET surface areas of the catalysts. Carstens et al. [16] have studied the catalytic combustion of methane over palladium catalysts supported on ZrO 2 It was found that fully reduced metallic palladium was completely inactive. This agrees with many previous works on palladium based catalysts. It was also found that the activity of palladium catalysts increases steadily with increasing surface coverage of 23

PAGE 24

highly active PdO until coverage reaches approximately 40%. At this point, the activity of the catalyst remains insensitive to increased PdO coverage (see Figure 1-5). Surface studies have shown that PdO formed under reaction conditions primarily consists of crystalline PdO, whereas PdO formed under O 2 is a mixture of amorphous and crystalline PdO. It was also shown that the crystalline form of PdO is more easily reduced to non-reactive metallic Pd. 1.7 Bimetallic Catalysts Palladium catalysts have been shown to have superior activity and lower ignition temperatures than platinum catalysts [12]. However, at temperatures between 700-800C the active palladium oxide decomposes into less active metallic palladium [12]. When the catalyst is cooled, the metallic palladium reverts back to the more active palladium oxide. However, this so called reformation temperature is much lower than the decomposition temperature. This difference in temperature causes instabilities in the catalytic combustor. Another limitation of palladium catalysts is the inability to maintain high conversion with time. Even at temperatures lower than the decomposition temperature, palladium based catalysts notice a decrease in catalytic activity with time on stream. Persson et al. [9] have studied the influence of co-metals on bimetallic palladium catalysts in order to form a more stable high temperature catalyst. Catalysts were formed by combing Pd with Co, Rh, Ir, Ni, Pt, Cu, Ag, and Au. All catalysts were formulated to contain equal molar amounts of palladium and the co-metal. The catalyst surface area, particle size, initial activity, and sustained activity were all measured in order to classify the catalysts. The most promising bimetallic catalyst was determined to be PdPt (see Figure 1-6). This is due to the good stability shown by the PdPt catalyst. The PdAg catalyst also showed good stability, however the activity of the PdPt catalyst was slightly higher. It should be noted that the PdPt catalyst was not the most active catalyst for the continually varied feed temperature. However, for the case of constant feed temperature, which would be the case for a 24

PAGE 25

typical power generation cycle, the PdPt catalyst showed the highest maintained catalytic activity. Another advantage important to the power generation cycle is that the bimetallic PdPt catalyst is less sensitive to water in the feed stream compared to the monometallic palladium catalysts. Persson et al. [10] have studied the catalytic combustion of methane over bimetallic PdPt catalysts with varying support materials. It has been shown previously [9] that the activity of monometallic palladium catalysts decreases under steady-state conditions even at temperatures lower than the decomposition of PdO. Bimetallic PdPt catalysts have been shown to [9] maintain high catalyst activity levels under steady state conditions. The next logical step is to study the effects of various commonly used support materials. Persson et al. [10] have studied the use of support materials such as Al 2 O 3 ZrO 2 LaMnAl 11 O 19 CeZrO 2 and YZrO 2 The effects of support material were studied with two experiments: first was a transient activity test where the feed temperature was constantly increased as the conversion was recorded, which measured the overall activity of the catalyst. The second test was a steady state test which held the feed temperature constant for one hour at 50C feed temperature increments. This second test measured the steady-state activity of the catalysts. As previously discussed, it is well known that at feed temperatures above 770C, PdO decomposes in the less active Pd which results in a drop in conversion. Persson et al. [10] have shown that using PdPt catalysts on different supports can act to suppress this decomposition and resulting decrease in activity. They have found that although less active at lower temperatures, PdPt supported on LaMnAl 11 O 19 continues to increase activity steadily up to and past feed temperatures of 900C. This discovery makes it possible to design catalytic reactors in stages with PdPt supported on LaMnAl 11 O 19 being used for the latter stages with higher feed temperatures. The steady-state activity tests have shown 25

PAGE 26

that although most active at the beginning of each temperature step, monometallic Pd supported on Al 2 O 3 shows a steady decrease in activity with time on stream. Even though the initial activity of the PdPt catalyst supported on ZrO 2 is less than the some of the other catalysts, it maintained initial activity at feed temperature of up to 670C, which is 50C higher than the next most stable catalyst. This along with the transient activity tests, which showed that PdPt supported on ZrO 2 maintains an increase in catalytic activity to temperatures exceeding 900C, makes ZrO 2 one of the more promising substrates. Another important aspect of this research is that they showed that the catalysts supported on ZrO 2 reoxidized Pd to the more active PdO at higher temperatures and in less time than PdPt catalysts supported on Al 2 O 3. Ozawa et al. [17] have studied the effect of adding various oxides to the catalyst washcoat on the catalytic combustion of methane and catalyst deactivation. This study focused on determining the cause of catalyst deactivation. It has been shown [7-11] that catalyst deactivation can be the result of the decomposition of active PdO into inactive metallic Pd, growth of the active catalyst particles, and decrease in the surface area of the catalyst. This work studies the deactivation of the bimetallic PtPdO catalyst and the effect of the addition of oxides to the catalyst washcoat on the overall deactivation of the washcoat. In this study, the authors measured the BET surface area of the catalyst, the particle size of the catalyst, and the composition of the catalyst both before and after deactivation. The deactivation of the catalyst was studied with two experiments: the first was a transient experiment with constant increase in feed temperature and measured conversion. This test focused on determining the high temperature deactivation of the catalysts. The second test was a steady state test, with a fixed feed temperature. This test focused on the steady state deactivation of the catalyst with increased time on stream. It was shown that the addition of La 2 O 3 Nd 2 O 3 and ZrO 2 decreased the reaction 26

PAGE 27

rate for the fresh catalyst. The addition of La 2 O 3 and Nd 2 O 3 was also shown to inhibit the deactivation of the catalyst at temperatures above 750C (see Figure 1-7). The other catalysts tested showed a decrease in activity above 750C, indicating deactivation of the catalyst surface. The catalyst with the addition of La 2 O 3, Nd 2 O 3 and ZrO 2 showed the least decrease in catalytic activity for time on stream greater than four hours at a feed temperature of 800C (see Figure 1-8). This is very significant for the catalytic combustion of methane for power generation cycles due to the extended length of time on stream. The analysis of the deactivated catalyst showed no loss of platinum or palladium, and showed an increase in the active PdO phase, and in the Pd-Pt alloy phase. This suggests that the primary cause of deactivation is not the decomposition of active PdO into nonactive metallic Pd. It was shown that the two primary causes of catalyst deactivation were an increase in the catalyst particle size, and a corresponding decrease in the BET surface area of the catalyst. Ozawa et al. [11] have studied the deactivation of PtPdO supported on Al 2 O 3 during catalytic combustion of methane. The purpose of this study was to compare the decomposition of monometallic PdO catalyst to the bimetallic PtPdO catalyst. The deactivation of the catalysts was studied using a fixed-bed flow reactor with a feed temperature of 800C. To determine the proper channel for the deactivation of the catalyst, Ozawa et al. monitored the surface area of the catalyst, the particle size of the catalyst substrate, and the PdO-Pd transformation of surface species. The research showed that the addition of Pt to PdO supported catalyst results in the formation of an alloy of metallic Pd with Pt. This alloy was shown to promote the growth of PdO particles. The addition of the Pt to the PdO catalyst was shown to initially decrease the activity of the catalyst. However, the activity of the catalyst was shown to increase with increasing Pt content. Also, increasing the Pt content of the bimetallic catalyst resulted in a 27

PAGE 28

decrease in the growth of the PdO particles. The addition of Pt was also shown to decrease the steady-state deactivation of the bimetallic catalyst in comparison to the monometallic catalyst. This decrease in deactivation was also shown to improve with an increase in Pt content. This data agrees with the work done by Persson et al. [10]. The overall opinion of the authors is that the catalysts are primarily deactivated by the transformation of PdO to Pd and the growth of the PdO and Pd-Pt particles. Moreover, the lifetime of both the monometallic Pd catalysts and the bimetallic PtPdO catalysts was shown to be affected more by the growth of the PdO particles rather than the transformation of PdO to the less active metallic Pd. Cho et al [18] have studied the structure of PdPt nanoparticles on bimetallic catalysts for the catalytic combustion of methane. X-ray adsorption was used to study the surface. It was found that platinum was anchored on the palladium core. It was also found that the structure remained unchanged and intact with feed temperatures up to 900C. This study agrees with other studies with the argument that adding platinum to palladium based catalysts can inhibit high temperature deactivation experienced by palladium catalysts. 1.8 Trimetallic Catalysts Kraikul et al. [19] have studied the catalytic combustion of methane on palladium, platinum, and lanthanum catalysts in monometallic, bimetallic, and trimetallic configurations. The first item of study in this work is the calcination effects on the catalytic activity. The method found to be most desirable for calcination of the catalyst was to pre-calcine the monolith at 500C for 3 hours before impregnating with 5 wt % of the total loading. The authors also suggest performing a re-calcination step at 900C for 3 hours prior to using the catalyst. The total loading of all catalysts was 5 wt % of total deposition. The main reason behind using lanthanum is a decrease in overall cost of the catalyst. This paper agrees with previous works about the benefits of bimetallic Pd-Pt over monometallic palladium, and therefore studied the 28

PAGE 29

substitution of platinum with less expensive lanthanum. With small substitutions of lanthanum, the catalyst activity was shown to decrease slightly. However, once the weight ratio of Pd:Pt:La reached 1:1:3, the activity was actually shown to increase. The authors attribute this increase in activity to the co-existence of the Pd-Pt alloy with PdO supported on highly dispersed La/Al 2 O 3 This catalyst was suggested due to the increased conversion and also the lower cost due to substitution with lanthanum. 1.9 Heat Transfer Effects In non-reacting fluid flows, heat transfer in the thermal boundary layer takes place primarily through conduction. At the surface of the catalyst, the no slip condition specifies that the fluid velocity is reduced to zero. The shear stress of the fluid, then slows the next layer of fluid to a near zero velocity. This trend continues to build up until the velocity eventually reaches that of the free stream and hence ends the boundary layer. The thermal boundary layer is very similar to that of the velocity boundary layer. Particles that come in contact with the surface of the catalyst reach zero velocity and hence reach thermal equilibrium with the surface. As in the velocity boundary layer, this trend continues until the fluid temperature nears that of the free stream. Since there is no fluid motion at the surface, and the particles are in thermal equilibrium, the heat transfer at the surface is best described by Fouriers Law of Conduction as can be seen in Figure 1-9. The case is quite different for that of reacting flows. In a heterogeneously chemically reacting flow, such is the case with catalytic combustion, products adhere to the surface with a lower total enthalpy than the products that are released from the surface. Also, the conduction at the surface is no longer valid because there is motion at the surface due to the bulk flow of the reactants adhering to the surface and the products being released from the surface. This bulk flow driven by heterogeneous chemistry may work to enhance the heat transfer from the surface. In the case of catalytic combustion, not only is there 29

PAGE 30

bulk flow from the surface, but there also exists the classical convective heat transfer between the hot catalytic surface and the free fluid stream. The research performed in this field of catalytic heat transfer enhancement has been quite limited. Although there have been a few attempts to document this phenomena [20-22], it still remains largely unexplained. Itaya et al. [21] performed experiments using both a non-reacting heated cylinder in a cross flow duct, and a reacting heated cylinder in the same duct. The heat transfer coefficient was calculated by measuring the difference in temperature between the forward stagnation point on the cylinder and the free stream inlet temperature, and by controlling the heat input to the cylinder. This experiment was carried out for both CH 4 and SO 2 reactions. It was found that the experimental heat transfer coefficient increases whenever the surface temperature is high enough to sustain catalytic activity. It was also found that the increase in the heat transfer coefficient is proportional to the reaction rate, and hence is maximized when the reaction rate is maximized at approximately 721K for the case of SO 2 The overall increase in the heat transfer coefficient was reported as 4-16% for the SO 2 reaction. This increase in heat transfer coefficient is contributed to the enthalpy due to diffusion of the reactants in the boundary layer on the surface of the catalyst [21]. 1.10 Pressure Effects The main purpose of research on catalytic combustion is focused on the application of power generation. Since the combustion process in a power generation cycle occurs at increased pressure, it is important to study the pressure effects on the catalytic combustion of methane. Reinke et al. [23-25] have studied extensively the effect of pressure on homogeneous ignition of fuel lean catalytic combustion of methane over platinum based catalysts. They have studied catalytic combustion over the pressure range of 1 bar < P < 16 bar, which encompasses the practical range of all operating power generation cycles. They have found that the location of 30

PAGE 31

homogeneous combustion is marked by an increase in the OH radical concentration. They used 1-D Raman spectroscopy and plan OH LIF to measure the concentration of major species and of the OH radical. They were able to modify several accepted homogeneous reaction mechanisms to account for the increased pressure of combustion. They credit the chain branching reaction CHO + M = CO + H + M as being quite significant for homogeneous ignition, especially at lower pressures [23]. Also studied by Reinke et al. was the addition of H 2 O and CO 2 in the feed stream at the previously mentioned increased pressures. They conclude that the addition of H 2 O into the feed resulted in the promotion of homogeneous ignition, whereas the addition of CO 2 into the feed had little or no effect [24]. It was speculated that H 2 O promotes homogeneous ignition primarily due to its significantly higher heat capacity. Persson et al. [8] have studied the combustion of methane at high pressure on supported palladium-platinum catalysts. It was determined that a bimetallic palladium-platinum catalyst is more stable during the combustion of methane than a palladium only catalyst. This was determined by both lab scale atmospherics tests, as well as in an experimental reactor with pressures up to 15 bar. The catalytic activity was shown to decrease with increasing pressure as can be seen in Figure 1-10. Although the change in activity was much more pronounced at lower pressures (<10 bar) than at the higher pressures. This pressure dependence of catalytic activity was also shown to be dependent upon the catalyst metal. The palladium only catalyst was shown to continue decreasing in activity through 15 bar, whereas the decrease in activity of the palladium-platinum catalyst began to level off around 12 bar. This decrease in catalytic activity with increasing pressure was attributed to growing mass-transport limitations. Even though the 31

PAGE 32

increase in pressure also results in an increase of the mass throughput and surface reaction rate, the growing mass transport limitations causes a decrease in the overall activity of the catalyst. Reinke et al. [25] have studied the effect of pressure on the catalytic combustion of methane over platinum catalysts. They used Raman spectroscopy to monitor the major species, and used planar LIF to monitor the OH radical to confirm that combustion only occurred heterogeneously. They monitored the catalytic activity of the platinum catalyst in a fuel lean feed of methane and air at pressures between 4 and 16 bar and temperatures between 780 and 1250 K. Although they used a different catalyst material, their results differed greatly from the results of Persson et al. [8]. Reinke et al. [25] found that the catalytic activity increased with an increase in pressure over the entire range of temperatures and pressures. They found that the key to implementing an accurate numerical solution was to account for the decrease in surface free-site availability with increasing pressure. They were not able to reproduce the measured catalytic activity over the entire range with a global catalytic step, but rather used a best-fit global step, which yielded results with less than 15% error. They also found that for certain geometric reactor parameters, and catalytic reactivity, the fractional conversion of methane could become completely independent of pressure effects. This phenomenon is highly desirable for the operation of a catalytic reactor in power generation systems due to the changes of pressure during loading and unloading of the system. Carroni et al. [26] have studied the catalytic combustion of methane over honeycomb palladium catalysts for power-generation applications at pressures up to 15 bar. This study measured wall temperatures and gas temperatures of the reacting flow. It was found that the overall temperature rise was effectively independent of pressure. The wall temperatures recorded increased with an increase in pressure. This effect is attributed to the enhanced surface 32

PAGE 33

activity due to the higher pressures. However, the increase in mass throughput due to increased pressured resulted in little change of the gas temperature. One important aspect of this study was that a substantial length of the reactor contributed minimally to the overall conversion of methane. This shows that an improvement in the design of reactors would allow lightoff at a shorter distance from the catalyst entrance, hence resulting in the same conversion rates from less catalytic material. It was shown that the catalytic activity increased with an increase in pressure. This conclusion disagrees with the work of Persson et al. [8], although these differing conclusions could very well be related to the differing geometry of the different studies. It was also shown that for certain reactor geometries and input parameters, the conversion can increase with pressure while maintaining a constant, pressure independent temperature rise. Kuper et al. [27] have studied the catalytic combustion of methane over palladium and PdPt catalytsts at pressures up to 20 bar. Their work focuses on catalytic reactors for the use of power generation. It was shown that palladium catalysts have a pressure dependent trend for the temperature of PdO decomposition into Pd as can be seen in Figure 1-11. This trend shows that an increase in pressure results in an increase of the decomposition temperature. Thus for a typical power generating combustor, the decomposition temperature will increase about 100C from the atmospheric decomposition temperature. It was also shown that at certain pressures and methane concentrations, homogeneous ignition can start at feed temperatures as low as 650C. Also, it is important to note that by 800C, homogeneous ignition was seen for all pressures and methane concentrations. For gas outlet temperatures about 1050C, the authors note that no CO or CH 4 was found in the exhaust stream, therefore suggesting complete oxidation of methane. Due to the aforementioned decomposition of PdO, the authors suggest it may be necessary to use a PdPt bimetallic catalyst for combustors with a higher desired outlet temperature. The authors 33

PAGE 34

also make the statement that for the concept of hybrid combustion, a preburner is probably necessary. Sinev et al. [28] have studied the catalytic combustion of methane over platinum based catalysts at pressures of 0.05-10 atm with the introduction of inert gases into the feed. This study used two catalysts to study pressure effects and inert feed effects. For the first catalyst, no change was seen in the selectivity or the conversion with the introduction of He into the feed, and only a slight increase in both was noticed with the introduction of Ar into the feed. For the second, more active catalyst, the effects were much more significant. Introduction of He into the feed resulted in an increase in the C 2 selectivity and in the overall conversion. These increases were even more pronounced with the introduction of Ar into the feed. It was also found that the lower the feed temperature, the more affect the inert gases had on the conversion and selectivity. The studies done at reduced pressures (0.1-1.0 atm) show that the conversion decrease is directly proportional to the decrease in total pressure as can be seen in Figure 1-12. The explanation given for the effects discussed here is that the conversion is increased due to an increase of surface-induced gas-phase reactions, such as CH 3 recombination into C 2 H 6 which result in a significant effect of diffusion acceleration. 1.11 Water Inhibition Two of the most practical applications for catalytic combustion are power generation applications, and post combustion treatment for pollution abatement of natural gas fueled vehicles. For both of these applications, a significant amount of water vapor is found in the feed (especially for natural gas fueled vehicles). For this reason, it is important to study the effects of water in the catalytic feed. One must understand which catalyst materials, geometries, and feed conditions are better suited to maintaining high activity levels with the introduction of water vapor in the feed stream. 34

PAGE 35

Pieck et al. [7] studied the effect of water vapor on the activity of Pt-Pd catalysts for the combustion of methane. The study focused on water vapor inhibition with several catalysts that had varying ratios of platinum and palladium. The water vapor was shown to affect the catalysts in two very different ways: at low temperatures (<600C), the water vapor was shown to increase the activity of the catalyst. At higher temperatures, the water vapor was shown to decrease the activity of the catalyst. For most of the catalysts tested, Cl was used in the precursor. Previous research has shown that Cl used in a precursor can actually inhibit the catalytic activity [4]. The increased low temperature activity with water vapor addition was due to the removal of Cl from the catalyst surface. Once the Cl containing catalysts had been reformed with water vapor at 600C, the activity was very similar to the Cl free catalyst of the same Pt-Pd ratio. At high temperatures, the addition of water vapor is shown to decrease the catalytic activity. This decrease in catalytic activity is attributed to sintering of the active metal phase. Furthermore, it was determined that of all the catalysts tested, the most reactive combination was 0.4%Pt-0.8%Pd. Kikuchi et al. [29] have studied the effects of water inhibition for the low temperature catalytic combustion of methane over palladium supported on various oxides. The supports surveyed in this study were: Al 2 O 3 SnO 2 and Al 2 O 3 -36NiO. Each of the catalysts was affected differently by water addition. The SnO 2 catalyst shows a shift in the conversion temperatures for each amount of H 2 O addition, with the inhibiting effect being the same at low and high feed temperatures, as can be seen in Figure 1-13. The Al 2 O 3 catalyst shows a more complex inhibiting effect: for H 2 O feeds of 1 vol %, the conversion is decreased for feed temperatures <450C; above this temperature H 2 O feed has no inhibiting effects; for H 2 O feeds greater than 5 vol %, the inhibiting effect takes place for all feed temperatures studied as can be seen in Figure 1-14. 35

PAGE 36

The Al 2 O 3 -36NiO has the most complex inhibition characteristics of all the studied catalysts. For an H 2 O feed of 1 vol %, the conversion is inhibited until the feed temperature reaches 350C, at which time the conversion actually increases slightly over the dry feed. For H 2 O feeds greater than 5 vol %, inhibiting effects are seen for the entire temperature range, but are less pronounced than for the other two catalysts studied, as can be seen in Figure 1-15. It was also shown that ending the H 2 O feed over the deactivated catalysts resulted in gradual recovery to the initial dry feed catalyst activity. This agrees with previous works that H 2 O inhibition is only temporary and that catalyst activity is easily restored in dry feed. 1.12 Sulfur Poisoning One of the more important aspects concerning catalytic combustion is sulfur poisoning. Since many catalysts are used in the exhaust of diesel engines to control the release of harmful unburned hydrocarbons, it is important to understand the mechanisms behind sulfur poisoning of catalytic surfaces. Sulfur compounds are also added to many natural gas fuels to aid in leak detection since sulfur has a strong, distinct odor; and therefore study is also important for the application of exhaust treatment for pollution abatement of natural gas fueled vehicles. Jones et al. [30] have studied the sulfur poisoning and regeneration of catalysts for the combustion of methane. The purpose of their study was to investigate the effects of small quantities of sulfur compounds on the combustion of methane over Al 2 O 3 supported noble metal catalysts. This study focused on three main noble metal catalysts: platinum, rhodium, and palladium. All catalysts contained 2 wt. % of the precious metal on an Al 2 O 3 support. The activities were measured for the three catalysts before any sulfur poisoning had occurred. The activity of the catalysts was quantified by T 50 the feed temperature at which the catalyst achieved 50% conversion of methane. The activities of the fresh catalysts followed the order of Pd > Rh > Pt. Following poisoning with sulfur, all of the catalysts observed a decrease in 36

PAGE 37

activity with a T 50 shift of 36C for palladium, 46C for platinum and 50C for rhodium. The catalysts were then regenerated by reduction in pure hydrogen flow. The purpose of the regeneration process is to reduce the oxidized sulfur back to SO 2 which can then be easily desorbed from the catalyst. The platinum catalyst showed very little regeneration (approximately 6% of initial activity) in activity, whereas the rhodium catalyst actually showed an increase in activity above and beyond the activity of the fresh rhodium catalyst. The activity of the palladium catalyst was also restored to near fresh catalyst activity (75% of initial activity). These results can be seen in Figure 1-16. Clearly the rhodium catalyst most benefited from regeneration. This phenomenon can be attributed to the particle size of the catalyst particles. Both the palladium and platinum catalysts show an increase in the metal particle size after regeneration compared to the fresh catalyst. The rhodium catalyst actually shows a decrease in the metal particle size after regeneration compared to the fresh catalyst. It is important to note that along with the deactivation of the catalyst by sulfur poisoning, the catalyst is also undergoing natural deactivation due to surface changes and sintering of the catalyst support. In order to quantify the deactivation due to sulfur poisoning, and other natural sources such as sinterings the authors compared the sulfur poisoning of the palladium catalyst to the natural deactivation of the catalyst. It was shown that the rate of deactivation due to sulfur poisoning was more than twice that of any natural deactivation taking place. This study showed that although all catalysts are affected by sulfur poisoning, some are more easily restored by reduction than others. Clearly the rhodium catalyst exhibits excellent regeneration, whereas regeneration of the platinum catalyst restores almost no activity. Dupont et al. [31] have studied the kinetics of fuel lean methane oxidation on platinum catalysts in the presence of H 2 S and SO 2 This study focused on the kinetics of sulfur containing 37

PAGE 38

feeds on a platinum catalyst in stagnation point flow reactor. As shown in Figure 1-17, it was found that with the introduction of 60 ppm of either H 2 S or SO 2, the methane conversion was found to increase by approximately 15%, showing that the presence of sulfur compounds actually helps to promote the catalytic oxidation of methane. The authors attribute this phenomenon to the coexistence of sulphates on the alumina support near adsorbed oxygen atoms on platinum. These sites then act as new sites for the oxidation of methane. It was also found that at temperatures between 500C and 700C, the catalyst helps to oxidize SO 2 into SO 3 This behavior may be beneficial for the use of platinum catalysts as a means of pollution control in engine exhaust streams. Gelin et al. [1] have studied the influence of water and sulfur containing compounds on the low temperature combustion of methane over platinum and palladium. This work focuses on the study of long-term aging effects of catalysts used in the exhaust of lean-burn natural gas fueled vehicles for the reduction of harmful unburned hydrocarbons. Of the utmost interest in this study is the contribution of water and sulfur to the aging and poisoning of catalysts. Pd catalysts have been shown to be more active than platinum, but monometallic palladium catalysts are very sensitive to water and sulfur compounds in the feed stream. For the experiments on water inhibition, 10 vol % of water was added to the dry feed at 700C. Similarly, for the experiments on sulfur poisoning 5 vol % of H 2 S was added to the dry feed. This study monitored the BET surface area of the catalyst, the metal content of the catalyst, and the dispersion of the catalyst. The fresh palladium catalyst was shown to be more active than the fresh platinum catalyst. It was found that both the BET surface area and the dispersion decreased for both palladium and platinum catalysts upon dry ageing at 700C as can be seen in Figure 1-2. The BET surface area decreased approximately 15% for each catalyst, whereas the dispersion decreased approximately 38

PAGE 39

30% for the palladium catalyst, and 75% for the platinum catalyst. The dry ageing has shown a negligible effect on the catalytic activity of palladium, whereas it has drastic effects on the catalytic activity of platinum. This clearly shows that the palladium catalyst is better suited for dry feed conditions. The addition of 10 vol % water to the feed stream has drastic effects on the palladium catalyst. The measured T 50 for both the fresh and aged palladium catalyst increases (100C for the fresh catalyst and 70C for the aged catalyst) under wet feed conditions as can be seen in Figure 1-3. The fresh platinum catalyst also shows an increase in the T 50 (40C). However, the aged platinum catalyst shows no change in T 50 which shows that water has negligible inhibiting effect on the aged platinum catalyst. Even though wet feed conditions have a more detrimental effect on the palladium catalyst, the overall activity of the palladium catalyst in a wet feed is still greater than the activity of the platinum catalyst in the wet feed as can be seen in Figure 1-18. This shows the superiority of palladium to platinum for wet feed conditions. Upon suppression of water in the feed stream, both catalysts exhibit restored activity as can be seen in Figure 1-19. This behavior is attributed to a competition between water and methane for the available active sites on the catalyst surface. For the case of H 2 S poisoning, H 2 S was added to the dry feed at 5 vol %. Both the fresh and aged palladium catalysts are quickly poisoned by the H 2 S feed. Continued exposure to H 2 S feed for the palladium catalyst results in an almost complete loss of catalytic activity. When the H 2 S feed is suppressed, the catalyst only regains minimal activity. This is very important because it shows that palladium is very sensitive to sulfur poisoning, and that the activity cannot be regenerated, resulting in destruction of the catalyst itself. The platinum catalysts exhibit much different behavior. The fresh platinum catalyst experiences a 42% loss in activity with the addition of H 2 S in the feed stream. However, the aged platinum catalyst actually shows a slight (5%) increase in activity with the addition of 39

PAGE 40

H 2 S in the feed stream as can be seen in Figure 1-20. This is in agreement with the previous work of Dupont et al. [31]. Upon suppression of the H 2 S feed, neither catalyst experiences much difference in the activity. This shows that the platinum catalyst is much more resistant to sulfur poisoning than the palladium catalyst. Lee et al. [12] have also described the deactivation of catalysts due to sintering or poisoning. They have also shown that sulfur oxides deactivate all monometallic noble catalysts with the exception of platinum. These effects can bee seen in Figure 1-21. 1.13 Reactor Modeling Moallemi et al. [32] modeled the catalytic combustion of methane over platinum and palladium catalysts in a cylindrical monolith reactor. The experimental reactor consisted of an inline NO x chemilluminescence analyzer and a gas chromatograph to measure unburned hydrocarbons and other major species. The surface temperature of the catalyst was measured with an infra-red thermometer. The honeycomb monolith reactor was modeled as a group of adjacent perfectly stirred reactors. The surface code used to model the perfectly stirred reactors was SURFACE PSR, with slight modifications for the sites definition and radiative heat losses. The gas phase code used for homogeneous interaction was that of GRI 2.11. It was found that the palladium catalyst yielded an increase in methane slippages. Also, the palladium catalyst resulted in easier ignition at lower flow rates. The palladium catalyst showed an increase in the catalyst surface temperatures over the platinum catalyst. The code created was unable to find steady burning solutions for the palladium catalyst at lower concentrations. Comparison with the experimental results shows that the model actually under predicts surface temperatures for the palladium catalyst by up to 200C. It should be noted that in this numerical modeling, the authors assume that each individual metal atom on the surface of the catalyst is available as a catalytic site. 40

PAGE 41

Goralski et al. [33] have modeled both heterogeneous and homogeneous reactions in a tubular plug flow reactor for the catalytic combustion of methane over platinum based catalysts. The authors use the previously discussed model of Deutschmann et al. [34], which can be seen in table 1-4, to model the heterogeneous reactions, and GRI Mech 2.11 to model the homogeneous reactions. They modeled cases with pure homogeneous reactions, pure heterogeneous reactions, and both. It was shown that catalytic reactors can operate well with homogeneous flammability limits while still producing fewer emissions such as CO and NO x It was also shown that the combined catalytic reactor significantly inhibits the homogeneous reactions even when operating with homogeneous flammability limits. This is due to the adsorption of free radicals such as O, H, and OH onto the catalytic surface. This inhibition of homogeneous chemistry results in lower amounts of CO and NO pollutants than are predicted for pure homogeneous combustion. The results presented by the authors show that it is possible to model the coupling of heterogeneous and homogeneous chemistry. This is very beneficial due to the fact that one of the most popular applications of catalytic combustion for power generation involves a coupling of heterogeneous and homogeneous combustion. Groppi et al. [35] have studied high temperature combustion of palladium supported on Al 2 O 3 for the measurement of reaction kinetics, and for reactor modeling. The authors have found that the catalyst thickness should be constrained to less than 10 m. This thickness limitation is to inhibit significant internal diffusion limitations. The theoretical analysis performed by the authors has defined that internal diffusion limitation as the most critical phenomenon on kinetic measurements. The authors have also recommended a constraint on the height of the annular reacting flow chamber. They believe the height should be on the order of 0.2-0.3mm or as small as is allowed by the pressure drop. This is done to minimize the 41

PAGE 42

phenomenon of gas-solid diffusion and axial gas diffusion, which can be prominent in methane combustion. These constraints have allowed the authors to take kinetic measurement data for the reacting catalyst. It was shown that H 2 O introduced in the feed continues to have significant inhibiting effects up to feed temperatures of 600C as can be seen in Figure 1-22. Sinev [36] has studied the free radicals in the catalytic combustion of light alkenes. It has been shown that the overall reaction mechanism includes both heterogeneous and homogeneous elementary reactions of primary and secondary free radicals. Sinev states that certain characteristics of surface active sites such as the H-atom affinity and the oxygen binding energy control the kinetics of elementary heterogeneous reactions of the free radicals, and hence contributes to the overall kinetics of the entire process. 1.14 Kinetics Hayes et al. [37] have studied the kinetics of the catalytic combustion of methane over platinum in a monolith reactor. The kinetics are modeled using a Mars and van Krevelen type of rate expression which accounts for the water inhibition. This study supports the findings of previous works regarding the significant inhibition attributed to water in the feed, and the trivial inhibition due to CO 2 in the feed. Hayes et al. [37] have found that the reaction order with respect to methane is one, and the reaction order with respect to oxygen is zero. It was also shown that the apparent activation energy of the reaction showed a marked increase at elevated temperatures, although this is not expanded upon in the paper. Also, this increase in activation energy was shown to be more dramatic when the feed was preheated using a propane burner. The authors attribute this effect to the increase in water present in the feed after propane combustion. It is also shown that the washcoat application has a significant effect on the activity of the catalyst. It was observed that a thinner, more dispersed catalyst is more active than a thicker, less dispersed washcoat. It is stated that a diffusion barrier drastically reduces the 42

PAGE 43

measured reaction rate, and even a thin barrier results in significant losses in conversion. The authors believe this effect can be manipulated in order to control the rate of temperatures rise along the length of the catalyst. 1.15 Gaseous Promoters Demoulin et al. [38] have studied the use of active promoters such as N 2 O and CO 2 for the catalytic combustion of methane and propane. For the N 2 O feed tests, the authors use nickel based catalysts. For all of the N 2 O tests, it is shown that the conversion of propane is decreased, while the yield and selectivity towards propylene increase as can be seen in table 1-5. The N 2 O also induces a decrease in the yield and selectivity to CO 2. The authors suggest that the N 2 O feed inhibits the adsorption of O 2 and hence limits the oxidation rate of the catalysts. This is in agreement with the decrease of propane conversion, and the increase in yield and selectivity of propylene. The effect of CO 2 addition in the feed for palladium catalysts on CeZrO 2 substrates was also studied in this work. These results are quite different than that of N 2 O. When 3% CO 2 is added to the feed stream, the conversion of methane of palladium is seen to increase. This shows that the addition of gaseous promoters to the feed stream is one way to control the catalytic combustion either by enhancement or inhibition. Demoulin et al. [39] have done further studies on site interactions with the introduction of N 2 O, CO 2 and H 2 as gaseous promoters in the feed stream. This study covered the introduction of these feed gasses for the catalytic combustion of propane and methane. For the catalytic combustion of propane, the authors used a NiMoO 4 catalyst. The results of 3% feed of CO 2 was an increase in the conversion of propane, and a decrease in the yield and selectivity of propylene, which is coupled with an increase in the conversion of oxygen. These results are similar to previous works [38] where CO 2 was shown to increase the conversion of methane over palladium based catalysts. When N 2 O is added to the feed at 5%, it results in a decrease in the conversion 43

PAGE 44

of propane, with an increase in the yield and selectivity of propylene. As the N 2 O feed is increased, a slight increase in the propane conversion is noticed, but conversion is still lower than without N 2 O feed. The catalytic combustion of methane was studied with catalysts on substrates of Al 2 O 3 and CeZrO 2. For catalysts with Al 2 O 3 substrate, it was shown that addition of 3% CO 2 in the feed causes a major decrease in the conversion of methane. This is quite the opposite effect of when 3% CO 2 is added to the feed for the catalysts with a CeZrO 2 substrate. Clearly, the effect of CO 2 feed addition is dependent upon the catalyst substrate. When H 2 was added to the feed for the palladium catalyst with the Al 2 O 3 substrate, the results were a little more complicated. The addition of 1% H 2 into the feed resulted in decreased conversion of methane for all feed temperatures. However, as the feed rate of H 2 increased the trend was quite interesting. At an H 2 feed rate of 5%, the conversion for the lower feed temperatures (<400C) is reduced, while higher feed temperatures benefit from a significant increase in the conversion of methane. This research has shown that introducing gaseous promoters into the catalyst feed streams is a practical way to control conversion of the fuels, and the selectivity of the products of reaction. Demoulin et al. [40] have studied the influence of H 2 feed on the behavior of palladium and platinum based catalysts supported on Al 2 O 3 It was found that the effects of H 2 addition varies depending on the operating conditions and the amount of H 2 present in the feed. At low concentrations (1 vol %), H 2 was found to drastically reduce to conversion of methane. This is due to the consumption of O species on the surface of the catalyst by H 2 As the amount of H 2 feed was increased (up to 5 vol %), the conversion showed a similar trend for temperatures lower than 400C. For temperatures greater than 400C, the conversion increases substantially. This enhancement in conversion is due to the highly exothermic combustion of H 2 on the PdO 44

PAGE 45

surface, which helps to counteract the consumption of the surface O species. Increasing the H 2 feed to 6.2 vol % once again results in a reduction of methane conversion for all feed temperatures because deactivation is occurring due to the reduction of the catalyst surface. Results for the platinum catalyst differ somewhat from the results for the palladium catalyst. When H 2 is fed to the platinum catalyst, the conversion is shown to increase at all temperatures, and for all feed rates (1.0-6.2 vol %). This phenomenon is attributed to the reduction of the surface, which creates more active sites for the combustion of methane. Deutschmann et al. [41] have studied the hydrogen assisted catalytic combustion of methane on platinum based catalysts. It is believed that the addition of hydrogen to the entrance feed of a platinum based catalytic reactor may lower ignition temperatures because hydrogen has been shown to react on platinum catalysts at room temperature. This study has shown that hydrogen does in fact allow for lower lightoff temperatures. This effect is due to the highly exothermic combustion of hydrogen increasing the surface temperature of the catalyst enough to begin the catalytic combustion of methane. The concentration of hydrogen required to achieve lightoff varies according to the amount of methane in the feed. Hydrogen is more reactive on platinum catalysts due to the oxygen surface coverage inhibiting methane oxidation at low temperatures. In this work, the authors use the reaction mechanism previously created by Deutschmann et al. [34], which can be seen in table 1-4. This mechanism was created to model the heterogeneous combustion of methane, hydrogen, and carbon monoxide. This work shows that the reaction mechanism can be applied to hydrogen assisted combustion of methane, and that the results of the experiments match those generated by the model. 1.16 Future Work There are several areas that the work discussed here does not cover. One of the subjects of catalytic combustion with the least amount of information readily available is the catalytic 45

PAGE 46

combustion of fuel rich gas flows. Fuel rich catalytic combustion can be very beneficial for the design of a hybrid combustor for a power generation cycle. There has also been very little work performed for the understanding of heat transfer enhancement for the catalytic combustion of methane and syngas. It is the purpose of this paper to further review fuel rich catalytic combustion and its respective heat transfer impact. Table 1-1. Catalysts preparation methods and their loading, particle size, and dispersion [4] Table 1-2. Activation energy and temperatures at which 10, 30, 50% CH4 conversions were reached for the complete oxidation of methane over Pd and Pt catalysts under standard dry reaction mixture a [1] Table 1-3. Temperatures at which 10, 30, 50% CH4 conversions were reached or the complete oxidation of methane over Pd and Pt catalysts under wet feed conditions [1] 46

PAGE 47

Table 1-4. Deutschmann reaction mechanism for the catalytic combustion of methane and hydrogen over platinum [34] Table 1-5. Catalytic activity results for Ni based catalyst [38] Table 1-6. Common catalysts and their substrates, active states, and poisoning characteristics Catalyst Common Substrates Active Phase Poisoning Pt Al 2 O 3 ZrO 2 Pt SO 2 (strong) Pd Al 2 O 3 ZrO 2 PdO H 2 O, SO 2 Pd-Pt Al 2 O 3 ZrO 2 LaMnAl 11 O 19 PdO, Pd-Pt, Pt SO 2 H 2 O Rh Al 2 O 3 Rh SO 2 (strong) 47

PAGE 48

Figure 1-1. Conventional flame combustor and catalytic combustor [42] Figure 1-2. Methane conversion on Pd/Al 2 O 3 catalyst. (a) reduced with H2; (b) pretreated with reactant mixture after reduction [12] 48

PAGE 49

Figure 1-3. Methane conversion as a function of feed temperature on alumina based catalysts, prepared by impregnation (Pd/Al(I)) and sol-gel (Pd/Al(SG) and Pd/Al(SG)Pd) [15]. Figure 1-4. Methane conversion as a function of feed temperature on titania based catalysts, prepared by impregnation (Pd/Ti(I)) and sol-gel (Pd/Ti(SG) and Pd/Ti(SG)Pd) [15]. 49

PAGE 50

Figure 1-5. Methane conversion versus Pd Oxidation [16]. Figure 1-6. Activity tests when temperature was varied stepwise for Pd-ref (), PdNi (), and PdPt (). The dotted line represents the feed temperature to the catalysts [9]. 50

PAGE 51

Figure 1-7. Combustion rate of methane with increasing temperature at 10 K min -1 [17]. Figure 1-8. Combustion rate of methane at 800C with time on stream [17] Figure 1-9. Temperature profile of bulk fluid flow near heated wall [43] 51

PAGE 52

Figure 1-10. Influence of pressure on methane conversion at feed temperatures of 500C on Pd/Al2O3 () and PdPt/Al2O3 () [8] Figure 1-11. Influence of pressure on PdO/Pd transformation temperature [27] 52

PAGE 53

Figure 1-12. Rate of methane conversion as a function of pressure [28] Figure 1-13. Rate of methane conversion over Pd/SnO 2 as a function of water vapor feed concentration [29] 53

PAGE 54

Figure 1-14. Rate of methane conversion over Pd/Al 2 O 3 as a function of water vapor feed concentration [29] Figure 1-15. Rate of methane conversion over Pd/Al 2 O 3 -36NiO as a function of water vapor feed concentration [29] 54

PAGE 55

Figure 1-16. Rate of methane conversion over noble metal catalysts: (a) Pd/Al 2 O 3 ; (b) Pt/ Al 2 O 3 ; (c) Rh/ Al 2 O 3 under varying conditions: (rhombus) freshly reduced catalyst; () poisoned; () regenerated. [30] 55

PAGE 56

Figure 1-17. Rate of methane conversion over Pt/ Al 2 O 3 and the effects of sulfur compounds introduced into the feed stream [31] Figure 1-18. Methane conversion over fresh (open symbols) and aged (filled symbols) Pd/Al2O3 (,) and Pt/Al2O3 (,) catalysts versus temperature in wet feed. [1] 56

PAGE 57

Figure 1-19. Influence of 10 vol % water vapor addition on methane conversion over fresh (open symbols) and aged (filled symbols) Pd/Al2O3 (,) and Pt/Al2O3 (,) catalysts versus time on stream. [1] 57

PAGE 58

Figure 1-20. Influence of 100 vol ppm H 2 S addition on methane conversion over fresh (open symbols) and aged (filled symbols) Pd/Al2O3 (,) and Pt/Al2O3 (,) catalysts versus time on stream. [1] 58

PAGE 59

Figure 1-21. Methane conversion on Pd/Al 2 O 3 catalyst. (a) sulfur free feed; (b) sulfur containing feed [12] Figure 1-22. Effect of water vapor inlet concentration. Dots: experimental points; lines: model predictions [35]. 59

PAGE 60

CHAPTER 2 FUNDAMENTAL SCIENCE AND BACKGROUND This chapter discusses the theories and physical process of the Raman effect of diatomic and polyatomic gasses. The Raman effect was discovered by Indian physicist Sir C. V. Raman in 1928, and resulted in his receipt of the Nobel Prize in 1930. Before the invention of the laser, the relatively low signals of Raman spectroscopy required either a high concentration of the substance in question, or high probe volumes. With the invention of the laser, Raman spectroscopy has seen a resurgence of interest due to its abilities for chemical analysis. Scattering of light occurs when an electromagnetic (EM) wave comes into contact with an obstacle or non-homogeneity such as a solid, liquid, or gas and is redirected onto a different path. As this wave collides with the object, the electron orbits within this object oscillate periodically with the same frequency ( o ) as the electric field of the incident wave. When the electrons inside this object oscillate, it results in a periodic separation of charge within the object known as an induced dipole moment. This oscillating induced dipole moment becomes a source of EM radiation and hence releases light. The majority of the scattered light is scattered elastically at the same frequency ( o ) as the incident light. It is possible, however for light to be scattered inelastically which results in the emission of light of a frequency different from that of the incident light ( o ). Raman scattering is an example of inelastic scattering. Figure 2-1 shows a schematic diagram of scattering resulting from an induced dipole moment. These induced dipole moments are created at three, distinct frequencies, namely o ( o + vib ), and ( o vib ). The first scattered frequency is scattered elastically, and is known as Rayleigh or Mie Scattering. The latter two frequencies are shifted to higher and lower frequencies respectively, and are therefore inelastic scattering processes. This scattered light is 60

PAGE 61

referred to as Raman scattering, where the decreased frequency is known as the Stokes shift and the increased frequency is known as the anti-Stokes shift. In order for Raman scattering to take place, it is necessary for the vibrational displacement of atoms corresponding to a particular vibrational mode to result in a change in the polarizability. This effect is shown in Figure 2-2. Figure 2-2 shows that the ability to perturb the electron cloud by an incident EM wave (defined as the polarizability) depends on the relative position of the atoms. When the atoms are located farthest apart from each other, the atoms have the least interaction with each other and therefore the polarizability is greatest. When the atoms are closest together, the atoms have the strongest effect on each other which results in the minimum polarizability. Since this oscillation results in a change in polarizability of the molecule, the fundamental vibrational mode of the molecule is Raman active and will generate inelastically scattered light at both the Stokes and anti-Stokes bands. It is also important to describe Raman scattering in terms of discrete vibrational energy states. This methodology uses a typical vibrational energy well as shown in Figure 2-3. Each individual vibrational state shown in Figure 2-3 corresponds to the vibrational quantum numbers described by the Boltzmann distribuation: /()iEkTtiingenZT (2-1) where, /0()iEkTiiZTge (2-2) Where n i is the number density of each atomic state, n t is the total number density for all states, g i is the statistical weight of each energy state, E i is the excitation energy to achieve each state, k is the Boltzmanns constant, T is absolute temperature (K), and Z(T) is the partitioning 61

PAGE 62

function. According to the Boltzmann distribution, at low to moderate temperatures the ground state is highly favored over the more energetic levels. This means that at any given point in time, the population in the ground state is much higher than the upper energy states. When an incident EM wave interacts with this molecule, it induces an oscillating dipole moment which results in increasing the energy of the electrons to a virtual state higher than the ground state, but not necessarily equal to any particular electronic quantum energy. During this process however, some energy may be transferred to the molecule which results in the transition to a higher vibrational (energy) state than before the interaction as shown in Figure 2-4. This shift in energy then results in the loss of total energy of the emitted photon and therefore a decrease in the frequency of the emitted light. Because the probability of finding a molecule in an excited state is always less likely than in the ground state, the Stokes scattering shown in Figure 2-4 is much more likely to occur than anti-Stokes scattering. For the case of anti-Stokes scattering, the molecule is originally found in an excited state, and during interaction with the EM wave the molecule decays to the ground state thereby transferring energy to the emitted photon after the reaction, and therefore the light scattered is of higher energy and higher frequency. For large systems of molecules, it is typical that both Stokes and anti-Stokes scattering occurs simultaneously with the intensity of the Stokes scattering being much greater than the anti-Stokes scattering. 62

PAGE 63

Figure 2-1. Physical process of scattering produced from an induced dipole moment [44]. Figure 2-2. Vibrational displacement of the atoms about the equilibrium position [44]. 63

PAGE 64

Figure 2-3. Vibrational energy well [44]. Figure 2-4. Vibrational energy well showing Raman shifted emission of light [44]. 64

PAGE 65

CHAPTER 3 EXPERIMENTAL METHODS Two systems have been studied for the examination of catalytic combustion using Raman spectroscopy. This first system is a tubular reactor consisting of a tubular catalyst placed inside of a stainless steel housing. The reacting gases flow on the outside of this catalyst in the annular region between the catalyst surface on the reactor housing. The second system is a stagnation flow reactor with a catalyst disk located in the center of a stainless steel vacuum cross. The reacting gases flow onto the disk surface via an impinging jet at a fixed distance from the surface. Raman spectroscopy was used in both systems to quantify the type and amount of gases at various points in the reactor. Heat transfer analysis was also performed on the tubular reactor in order to quantify enhancement to the convective heat transfer coefficient in reacting flows. This chapter will describe the experimental Raman spectroscopy configuration, and will also describe the specifications of both reactors as well as the experimental procedures used for data acquisition. 3.1 Reactor Specifications 3.1.1 Tubular Reactor The tubular reactor was the first design to be built and used for testing in the laboratory. This work was performed in collaboration with Siemens Power Generation, and hence the tubular reactor configuration was chosen to model the reactor currently being studied by Siemens PowerGen. A schematic of the tubular reactor can be seen in Figure 3.1. The tubular reactor consists of an insulated stainless steel housing with an outer diameter of 12.7 mm and an inner diameter of 6.6 mm. The catalyst tube is located inside the stainless steel housing and has an outer diameter of 4.7 mm, an inner diameter of 4.2 mm, and an overall length of approximately 240 mm. The catalyst tube is attached to a stainless steel holder via a silver solder joint near the 65

PAGE 66

inlet of the reactor. This holder is then fitted inline into the Swagelok cross, a cross-section view of which can be seen in Figure 3-2. The downstream end of the catalyst is held in place via a press fit tube holder with holes axially drilled throughout. This tube holder centers the catalyst tube inside of the reactor, and allows the reacting gases to flow through the drilled holes to mix downstream with the cooling air. A schematic of the downstream catalyst holder can be seen in Figure 3-3. The catalyst used for all experiments is a proprietary catalyst containing platinum and palladium as the active metals. This catalyst is coated on a Haynes 230 alloy substrate. The catalyst is coated on the outside of the tube, and hence the reacting gasses flow through the annular region between the tubes. The reacting air is heated in a 200W T Type process heater to a temperature of 800K, which is regulated by an Omega CN132 temperature controller. The room temperature (300K) CH 4 fuel is then mixed with the heated reacting air before entering the catalytic reactor. The mixed fuel-air flow temperature is monitored via a Type K thermocouple inserted into the flow stream before entering the annular reacting region. The cooling air is heated in a 700W T Type process heater to a temperature of 800K, which is regulated by an Omega CN132 temperature controller. The cooling air then flows through the center of the catalyst tube to keep surface temperatures within a reasonable limit to prevent sintering of the catalyst and catalyst substrate. The temperature of the cooling airflow is then sampled through the end of the reactor using a Type K thermocouple and an Omega HH-26K handheld thermometer. The flow rates of the gasses were controlled with AlliCat Scientific mass flow controllers as shown in Figure 3-4. The exhaust gases at the exit of the reactor must be cooled and diluted in order to inhibit gas phase ignition through the ventilation system. To quench the exhaust gas, a glass bead flow meter was used to control the flow of fresh air to the 66

PAGE 67

exhaust gas in order to create a fuel lean condition outside of the flammability limits of the fuel. For cooling, the exhaust gas is routed through a chilled water heat exchanger. The chilled water is supplied by a Neslab Merlin M25 water chiller. A cross-section view of the catalytic reactor can be seen in Figure 3-5. In order to gain optical access to the reactor for the use of Raman spectroscopy, Swagelok 4-way crosses (part number: SS-810-4) were swaged onto the reactor housing at approximately 5.5 cm and 15.5 cm from the aforementioned solder joint at the catalyst entrance holder. In order to fit the windows to the reactor housing, the Swagelok crosses were drilled out to a diameter of 12.8 mm. This hole was drilled to a depth of 22.2 mm measured from the edge of the Swagelok cross. This allowed the UV grade silica windows (diameter 12.5 mm, thickness 6.3 mm) to insert smoothly into the fitting. The windows were purchased from CVI laser (part number PW1-0525-UV). The windows were held in place by a stainless steel window holder, which is tightened by the nut supplied with the Swagelok cross. A schematic of the window holder can be seen in Figure A-1. In order to seal the windows, ceramic fiber gaskets were placed between the reactor housing and the window and between the window and the window holder. The gaskets had an inner diameter of 9.5 mm, an outer diameter of 12.7 mm, and a thickness of approximately 1.5 mm. A cross-sectional view of the reactor window can be seen in Figure 3-6 and a picture of the complete tubular reactor apparatus can be seen in Figure 3-7. 3.1.2 Stagnation Flow Reactor The stagnation flow reactor was the second system to be designed and built. The reason for using a stagnation point flow reactor is the simplicity of modeling, as well as a greater optical path length and optical scattering collection angle, thereby enhancing Raman detection limits. In a stagnation point reactor, it can be assumed that the reaction zone varies only in one dimension (the distance from the catalyst surface), and hence can be modeled as a one dimensional reactor. 67

PAGE 68

In a stagnation point reactor, the reacting gases (fuel and air) flow through a vertical nozzle, and onto the catalyst surface. The flow reaches a stagnation point on the surface of the disk, and the reaction takes place on the catalytic surface. The stagnation flow reactor chamber used for the design is a standard 6-way vacuum cross purchased from Huntington Vac. The cross (part number VF-6250) is built with a tubing outer diameter of 2.50 inches. The top port of the reactor is used for both the insertion of the reacting gas nozzle, and for the reactor exhaust. The inlet air is preheated to 800 K in a 200 W T-Type inline process heater, and the temperature is controlled by an Omega CN132 temperature controller. The fuel at room temperature (300 K) is then added to the preheated air through a Swagelok tee joint. The nozzle is made of standard 0.25 inch diameter 316 stainless steel tubing with a wall thickness of 0.040 inches. The nozzle is attached to a micrometer controlled vertically adjustable stage. This adjustable stage allows the nozzle to be raised or lowered to change the nozzle-to-catalyst distance. In order to insure the assumption of one dimensionality, the reacting nozzle is placed inside of a nitrogen nozzle which flows nitrogen onto the catalyst at an identical momentum as the reacting nozzle. This nitrogen nozzle is made of standard 0.375 inch diameter tubing with a wall thickness of 0.040 inches. Press fit into the nitrogen nozzle is a nozzle holder which centers the reacting gas nozzle inside of the nitrogen nozzle in order to achieve uniformity of the nitrogen co-flow. This nozzle holder is press fit 1 inch deep into the nitrogen nozzle so as to allow the nitrogen flow to become fully developed before it exits the nozzle. A schematic of the nozzle holder can be seen in Figure A-2. The exhaust of the reactor is collected via four 0.25 inch Swagelok stainless steel flexible lines mounted to the top flange of the reactor. This allows the exhaust gas to be forced to the top of the reactor due to the inherent pressure in the reactor from the reacting gas flow, the nitrogen co68

PAGE 69

flow, and the cooling nitrogen flow. The exhaust is then vented into the laboratory ventilation system. The bottom port of the 6-way cross is used for the insertion of the cooling nitrogen flow nozzle, and for the linear motion manipulator used for adjusting the height of the catalyst. The cooling flow for the stagnation point flow reactor is room temperature (300 K) nitrogen, and is passed through a nozzle, similar to the reacting nozzle, on the bottom of the catalyst disk to prevent sintering of the catalyst and the catalyst substrate. The linear motion manipulator is manufactured by Huntington Vac (part number ZL-2111). The linear manipulator is mounted to the cross via a Huntington Vac flange adapter. The linear manipulator is then connected to the catalyst lifting plate which can be seen in Figure A-3. The triangular cuts seen in the catalyst lifting plate allow the polyethylene cooling flow tubing to be routed through the lifting plate and attached to the catalyst cooling nozzle plate which can be seen in Figure A-4. The cooling nozzle plate is attached directly above the catalyst lifting plate via three 0.125 inch diameter legs 5.50 inches in length. The center hole in the cooling nozzle is used to mount the cooling nozzle jet. The cooling nozzle jet is built from a 0.25 inch stainless steel Swagelok through hull fitting (part number SS-400-61) swaged to a 0.25 inch stainless steel Swagelok port connector. The catalyst holder is attached directly above the cooling nozzle plate via three 0.125 inch diameter legs 2.50 inches in length. The catalyst holder is made of molybdenum to allow for maximum heat transfer and high melting point, and is designed to hold a round catalyst puck 0.25 inches in diameter. Schematic drawings of the catalyst holder and the catalyst puck can be seen in Figure s A-5 and 3-8, respectively. The catalyst puck is made from Haynes 230 alloy, which is the same material used as the tubular reactor catalyst substrate. Haynes 230 is commonly used for 69

PAGE 70

catalyst substrates due to its excellent binding qualities, relatively high thermal conductivity compared to conventional stainless steel, and high melting point. The four horizontal ports on the 6-way cross are used for optical access to the reactor. As in the tubular reactor, Raman spectroscopy will be used to analyze the component gases in the reaction zone for the stagnation point flow reactor. For this reason, the port used for Raman spectroscopy contains a UV grade quartz window; whereas the other three ports contain regular glass windows. As previously mentioned, both the reacting gas nozzle and the catalyst support system are attached to linear motion manipulators. The reactor is designed in this matter so as to allow complete user control of the nozzle height, catalyst height, and nozzle to catalyst distance. As previously discussed, the reaction on a stagnation point flow reactor varies only in the vertical direction. Therefore, it is necessary to obtain data at varying distances above the surface of the catalyst. By having full user control of the vertical adjustment of the catalyst and reacting gas nozzle, Raman spectroscopic data can be taken at varying distances from the catalyst surface without having to adjust the vertical position of the laser or the reactor housing itself. This allows the user to fix the laser beam to the center of the quartz reactor windows to allow an optimum optical angle. A picture of the stagnation reactor apparatus can be seen in Figure 3-9. 3.2 Experimental Apparatus 3.2.1 Tubular Reactor For the collection of Raman spectroscopic data, backscatter collection is the only option for the tubular reactor. This is due to the fact that the tubular reactor only has two windows at each point to be evaluated, and hence orthogonal collection cannot be achieved. Figure 3-10 shows all optical equipment used for the collection of all Raman data for the tubular reactor. With the exception of the 375 nm high pass filter, the optical layout remained the same for all 70

PAGE 71

scattering experiments. For data taken with the spectroscopic window centered at 380 nm, the 375 nm high pass filter had to be removed in order to collect the spectra signal at shorter wavelengths. A list of equipment for all the items shown in Figure 3-10 can be found in table 3-1 including manufacturers name, part number, and specifications. The excitation source for all Raman spectroscopic experiments is a Big Sky Neodymium: Yytrium-Aluminum-Garnet (Nd:YAG) operating at its fundamental wavelength of 1064 nm, frequency tripled resulting in 355 nm emission. For all experiments, the laser operated at a 5 Hz repition rate, and with 16.5 J/pulse pump energy, which produces a beam of 355 nm light with 35 mJ/pulse. Upon exit of the laser, the beam diameter is narrowed via an adjustable iris aperture in order to obtain a uniform beam profile. The beam was then turned via two 45 dichroic mirrors designed for 355 nm. These mirrors are coated with anti reflection coatings for light at 1064 nm and 532 nm light, respectively. These mirrors allow the residual 1064 nm and 532 nm light to pass through the mirror with 99% efficiency and to the beam dump. This allows the laser light to be free of residual light, and comprised entirely of 355 nm light at the sampling zone. The laser light then passes through a focusing lens which focuses the laser beam from a nominal diameter of approximately 1 cm to a nominal diameter of 1 mm in the sampling zone. The laser light then passes through a 45 pierced mirror and into the sampling zone. Any leftover light that does not interact with the molecules to produce the Raman effect is collected in the cylindrical, conical termination beam dump located directly behind the reactor window. Due to the relatively low efficiency of the emitted Raman signal compared to the pulse energies of the stimulating light, reflections from the outer surface of the quartz reactor window must be minimized. This minimization of reflection was achieved by rotating the reactor assembly by a few degrees in order to misalign the reflection so it would not be collected by the collection optics. 71

PAGE 72

Once the laser light enters the sampling zone, it energizes the molecules as described in chapter 2, and the targeted molecules release the Raman shifted light. As shown in Figure 3-10, this light exits the front reactor window and expands its diameter until the light reaches the pierced mirror, at which time the light is turned 90 toward the collection optics. The pierced mirror is designed so that less than 5% of the Raman signal is lost through the center of the mirror. The Raman light then passes through the 4 inch plano-convex collection lens which focuses the light of approximately 3 inches in diameter to a spot approximately 2 mm in diameter on the fiber optic assembly. After being focused to a minimal spot by the collection lens, the light passes through several filters designed to reject 355 nm light, as shown in Figure 3-10 and listed in table 3-1. After passing through the filters, the focused light was collected on a 17-fiber bundled fiber optic with a 1.2 mm diameter and input into the diffraction spectrometer. The groove spacing used on the intensified charged coupled device (ICCD) for all experiments was 2400 grooves/mm. This setup was used for all experiments on the tubular reactor. 3.2.2 Stagnation Flow Reactor For the collection of Raman spectroscopic data, backscatter collection was chosen as the means of collection for the stagnation point flow reactor. The main advantage of backscatter collection is the insensitivity to minor errors in alignment. In sidescatter, if the collection optics are misaligned, the Raman signal will be misaligned on the fiber optic bundle, which can result in no signal transmitted to the spectrometer. In backscatter, a misalignment simply results in the Raman signal either being under focused or over focused at the fiber optic bundle. This results in a weaker signal, but not a total loss of signal. Figure 3-11 shows all optical equipment used for the collection of all Raman data for the stagnation point flow reactor. The optical layout remained the same for all scattering experiments. A list of equipment for all the items shown in 72

PAGE 73

Figure 3-11 can be found in table 3-2 including manufacturers name, part number, and specifications. The excitation source for all Raman spectroscopic experiments is a Big Sky Neodymium: Yytrium-Aluminum-Garnet (Nd:YAG) operating at its fundamental wavelength of 1064 nm, producing frequency tripled light at 355 nm emission. For all experiments, the laser operated at a 5 Hz repition rate, and with 16.5 J/pulse pump energy, producing a 35 mJ/puls beam. Upon exit of the laser, the beam was turned via two 45 dichroic mirrors rated for 355 nm. These mirrors are coated with anti reflection coatings for light at 1064 nm and 532 nm, respectively. These mirrors allow the residual 1064 nm and 532 nm light to pass through the mirror with 99% efficiency and into the beam dump. This allows the laser light to be free of residual light, and comprised entirely of 355 nm light at the sampling chamber. The laser light then passes through a focusing lens which focuses the laser beam from a nominal diameter of approximately 1 cm to a nominal diameter of 1 mm in the sampling zone. The laser light then passes through a 45 pierced mirror and into the sampling zone. Any leftover light that does not interact with the molecules to produce the Raman Effect is collected in the cylindrical, conical termination beam dump located directly behind the sampling chamber rear window. As discussed in the reactor specifications section in this chapter, both the nozzle and the catalyst disk height are adjustable in the stagnation point flow reactor. This is designed so that the laser can be permanently mounted at a fixed height, which reduces the chance of misalignment due to repositioning of the laser. Also, by fixing the height of the laser and the height of the reactor chamber, the laser enters the reaction chamber in the center of the window, which allows maximum expansion of the Raman shifted light, and hence maximizes the Raman spectroscopic signal. 73

PAGE 74

Once the laser light enters the sampling zone, it energizes the molecules as described in chapter 2, and the energized molecules release the Raman shifted light. As shown in Figure 3-11, this light exits the front reactor window and expands its diameter until the light reaches the pierced mirror, at which time the light is turned 90 toward the collection optics. The pierced mirror is designed so that less than 5% of the Raman signal is lost through the center of the mirror. The Raman light then passes through the 4 inch plano-convex collection lens which focuses the light of approximately 3 inches in diameter to a spot approximately 2 mm in diameter on the fiber optic assembly. After being focused to a minimal spot by the collection lens, the light passes through several filters to reject residual 355 nm light as shown in Figure 3-11 and listed in table 3-2. After passing through the filters, the focused light is collected on a 17-fiber bundled fiber optic with a 1.2 mm diameter and input into the diffraction spectrometer. The groove spacing used on the intensified charged coupled device (ICCD) for all experiments was 2400 grooves/mm. This setup was used for all experiments on the stagnation point flow reactor. 3.3 Experimental Conditions 3.3.1 Tubular Reactor For the tubular reactor, most experimental conditions were specified to be consistent with conditions in common power generation cycles under catalytic combustion. It was specified that both the reacting air and the cooling nitrogen flow enter the reactor at approximately 460C. In order to achieve these desired inlet conditions, both the reacting air and the cooling nitrogen flow are preheated in the inline T-Type heaters to a temperature of 525C. In addition to the preheat temperature of the reacting air and cooling nitrogen, the fuel was mixed at room temperature to the reacting air directly before the entrance to the reactor. The fuel flow rates were also specified to result in fuel rich conditions. Consistent with power generation feeds, two fuels were selected for testing of the tubular reactor: synthesis gas and methane. Synthesis gas 74

PAGE 75

is the product of coal gasification, and is gaining popularity as a way to extract cleaner energy production from coal. Another reason for the ever increasing popularity of coal gasification is the increase in cost of natural gas fuels such as methane. Table 3-3 shows the exact composition of synthesis gas used for the optical experiments on the tubular reactor. Table 3-4 shows the volumetric flow rates for the reacting air, cooling nitrogen, and fuel. When methane was used as the reacting fuel, lightoff could not be achieved by the introduction of the fuel due to the high activation energy for the catalytic combustion of methane. It was necessary to seed the methane feed with approximately 0.40 slpm H 2 until light off was achieved. Lightoff was achieved when the thermocouple placed downstream of the reactor mixing point showed a spike in the downstream temperature, and was confirmed visually by the radiative glowing of the catalyst surface in the optical access windows. As soon as lightoff was achieved, the H 2 flow would be terminated and the catalyst could sustain methane conversion with no fuel seeding. When synthesis gas was used as the reacting fuel, seeding with additional fuel was not necessary due to the high concentration of H 2 found in the synthesis gas. After lightoff was achieved, optical data was taken until the reactor reached equilibrium approximately 30 minutes after lightoff. 3.3.2 Optical Experimentation The optical setup for the tubular reactor is shown in Figure 3-11. In order to take data at both the upstream window (port 1) and the downstream window (port 2), rather than adjusting the laser data acquisition assembly, the reactor system is unbolted from the optical table and moved to align the laser system with the port being analyzed. This results in more reliable alignment of the optical equipment, which is necessary for the incredibly short laser path length for the tubular reactor setup. The frequency tripled 355 nm Nd:YAG laser was pumped with 16.5 J/pulse at a frequency of 5Hz. The delay from the firing of the laser and the initiation of 75

PAGE 76

data acquisition (gate delay) was set to 0.050 s, with a gate width of 0.300 s. The spectra were analyzed in two different optical windows: one centered at 406 nm, and one at 380 nm. The 406 nm window included spectral data for H 2 (416.0 nm, 4156 cm -1 shift) and CH 4 (395.6 nm, 2917 cm -1 shift), an example of which can be seen in Figure 3-12. The 380 window included spectral data for CO (383.9 nm, 2143 cm -1 shift), CO 2 (371.6 nm, 1285 cm -1 shift; 373.1 nm, 1388 cm -1 shift), and O 2 (375.4 nm, 1555 cm -1 shift), an example of which can be seen in Figure 3-13. A table showing the investigated species and their respective Raman shifts and wavelengths can be seen in table 3-5. Each spectrum was averaged over 250 shots. 3.3.3 Heat Transfer Measurement As previously discussed in chapter 1, it is important to analyze the heat transfer from the surface of the catalyst to the reacting gas stream. As discussed, earlier work has proposed an enhancement to the heat transfer in the reacting gas stream due to the interaction with the surface of the catalyst. In order to analyze and quantify the enhancement of the heat transfer in the reacting stream, the temperature profiles of the cooling flow for varying fuels were measured. As can be seen in Figure 3-14, the thermocouple was inserted downstream of the tubular reactor exit, and just upstream of the chilled water heat exchanger. The thermocouple has two bends within the last 15 cm of the tip of the thermocouple so as to allow the thermocouple to be self centering, and thus take temperature measurements of the bulk flow temperature of the cooling nitrogen flow. The type K-24 inch exposed bead thermocouple was inserted fully into the cooling nitrogen flow stream up to the aforementioned solder joint where the catalyst is soldered to the catalyst holder at the beginning of the reactor. Temperature measurements are then taken at 1 cm intervals for steady state reaction conditions. These temperature measurements are repeated until approximately 23 cm from the solder joint, which is where the cooling nitrogen flow and the reacting flow mix. A final temperature measurement is recorded at 30 cm from the 76

PAGE 77

solder joint, which is the location of the downstream permanently installed thermocouple (T 2 ). An example of the measured temperature profile can be seen in Figure 3-15. 3.3.4 Stagnation Reactor For the stagnation reactor, the initial experimental conditions were determined qualitatively by viewing the reacting zone on the catalyst disk, and selecting initial flow conditions based on flow rates which resulted in a relatively large (2.5 cm) reaction zone. As with the tubular reactor, the reacting air is preaheated in an inline T-Type heater to a temperature of 798 K. For the stagnation reactor it was not feasible to also heat the cooling nitrogen flow because in order to have the catalyst height adjustable, the cooling nitrogen flow tubing located inside the reactor chamber must be flexible. In order to satisfy the flexibility requirement, the tubing used is polyethylene, which would melt if heated to 798 K. The fuel used for preliminary testing and modeling of the stagnation reactor is H 2 As in the tubular reactor, the fuel flows in at room temperature (300K) and mixes with the heated reacting air prior to entrance into the reactor chamber. The nitrogen co-flow is also preheated in an inline T-Type heater to a temperature of 525C. Initially, the reactor design did not call for a nitrogen co flow. However, initial testing showed that possible eddy formation from the reactants residing on the edges of the catalyst disk could affect the spectra taken during combustion. To prevent the formation of these eddies; a momentum flux matching nitrogen co-flow was added concentric to the reacting gas flow as described in detail previously in this chapter. The flow conditions for the stagnation reactor can be found in table 3-6. Since the only fuel tested thus far in the catalytic reactor is H 2 seeding of the fuel to obtain lightoff is unnecessary. Lightoff was confirmed visually by the radiative glowing of the catalyst surface in the optical access windows. Upon lightoff, the reactor was allowed to burn for approximately 30 minutes in order to reach steady state conditions before data was taken. 77

PAGE 78

3.3.5 Optical Experimentation The optical setup for the tubular reactor is shown in Figure 3-11. In order to take data at varying heights from the surface of the catalyst, rather than adjusting the laser data acquisition assembly, as discussed previously in this chapter both the reacting gas nozzle and the catalyst disk are adjustable via micrometer linear translators. This results in more reliable alignment of the optical equipment, and allows the focal point of the laser beam to always be at the center of the reactor chamber, thus maximizing the optical angle of the collected Raman spectrum. The frequency tripled 355 nm Nd:YAG laser was pumped with 16.5 J/pulse at a frequency of 5Hz. The delay from the firing of the laser and the initiation of data acquisition (gate delay) was set to 0.050 s, with a gate width of 0.300 s, which was centered on the laser pulse. For the initial stagnation reactor experiments, the Raman spectra were collected at five different heights from the surface of the catalyst on 1 mm intervals. The spectra were analyzed in two different optical windows: one centered at 406 nm, and one at 380 nm. The 406 nm window included spectral data for H 2 (416.0 nm, 4156 cm -1 shift), H 2 O (407.5 nm, 3650 cm -1 shift), and CH 4 (395.6 nm, 2917 cm -1 shift), an example of which can be seen in Figure 3-16. The 380 window included spectral data for CO (383.9 nm, 2143 cm -1 shift), CO 2 (371.6 nm, 1285 cm -1 shift; 373.1 nm, 1388 cm -1 shift), and O 2 (375.4 nm, 1555 cm -1 shift), an example of which can be seen in Figure 3-17. Each spectrum was averaged over 250 shots. 78

PAGE 79

Table 3.1. Tubular reactor: all necessary components for the acquisition of Raman spectroscopic data. Component Specifications Vendor Part Number Quantity Laser Nd:YAG 1064 nm, frequency tripled 355 nm, 5Hz repetition rate Big Sky 19008300-3714 1 1 st Turning Mirror 45 dichroic 355 nm, 1064 nm AR coating, 2 in diameter CVI Laser BSR-31-2025 1 2 nd Turning Mirror 45 dichroic 355 nm, 532 nm AR coating, 2 in diameter CVI Laser BSR-35-2025 1 Pierced Mirror 100 mm x 100 mm x 1 mm, 0.5 inch hole Rolyn Optics 60.2475 1 Square Filter 375 nm high pass filter Comar 375-GY-50 1 Round Filter Low pass filter 1 Focusing Lens 300 mm focal length, 50.8 mm diameter CVI Laser PLCX-50.8-154.5-UV-1064 1 Collection Lens 101.6 mm diameter Comar 160-PG-100 1 Razor Filter 364 nm high pass filter Semrock LP02-364RS-25 1 Spectrometer 0.275 m grating Monochromator/Spectrograph Action Research Corp. SpectraPro 300i 1 CCD Intensified CCD Princeton Instruments PI-MAX 1 79

PAGE 80

Table 3-2. Stagnation point flow reactor: all necessary components for the acquisition of Raman spectroscopic data. Component Specifications Vendor Part Number Quantity Laser Nd:YAG 1064 nm, frequency tripled 355 nm, 5Hz repetition rate Big Sky 19008300-3714 1 1 st Turning Mirror 45 dichroic 355 nm, 1064 nm AR coating, 2 in diameter CVI Laser BSR-31-2025 1 2 nd Turning Mirror 45 dichroic 355 nm, 532 nm AR coating, 2 in diameter CVI Laser BSR-35-2025 1 Pierced Mirror 100 mm x 100 mm x 1 mm, 0.5 inch hole Rolyn Optics 60.2475 1 6-Way Vacuum Cross 63.5 mm OD tubing with quartz windows Huntington Vac VF-6250 1 Purple Filter low pass filter Shott glass 1 Focusing Lens 300 mm focal length, 50.8 mm diameter CVI Laser PLCX-50.8-154.5-UV-1064 1 Collection Lens 101.6 mm diameter Comar 160-PG-100 1 Razor Filter 364 nm high pass filter Semrock LP02-364RS-25 1 Spectrometer 0.275 m grating Monochromator/Spectrograph Action Research Corp. SpectraPro 300i 1 CCD Intensified CCD Princeton Instruments PI-MAX 1 Table 3-3. Synthesis gas composition Gas Percent Composition Volumetric Flow Rate H 2 31.3% 1.75 SLPM CO 54.2% 3.04 SLPM CO2 11.5% 0.64 SLPM CH 4 0.4% 22.4 CCPM 80

PAGE 81

Table 3-4. Flow rates for tubular reactor Flow Rate Cooling Nitrogen 30.0 SLPM Reacting Air 5.0 SLPM Synthesis Gas 5.6 SLPM Methane 1.6 SLPM Table 3-5. Investigated gaseous species and their respective Raman shifts and wavelengths due to laser excitation at 355 nm. Species Raman shift (cm -1 ) Emission Wavelength (nm) H 2 4156 416.0 O 2 1555 375.4 CO 2143 383.9 CO 2 1285, 1388 371.6, 373.1 CH 4 2917 395.6 H 2 O 3650 407.5 Table 3-6. Flow rates for stagnation reactor Flow Rate Cooling Nitrogen 30.0 SLPM Reacting Air 10.0 SLPM H 2 3.5 SLPM Co Flow Nitrogen 19.3 SLPM 81

PAGE 82

Figure 3-1. Tubular reactor flow schematic. 82

PAGE 83

83 Fuel/Com ubstion A ir Cooling N itr ogen Fuel/Com ubstion A ir Figure 3-2. Tubular reactor solder joint flow schematic.

PAGE 84

Figure 3-3. Downstream catalyst tube holder. 84

PAGE 85

Figure 3-4. Alli-Cat Scientific flow meters used for flow control on both the tubular reactor and the stagnation point flow reactor. 85

PAGE 86

86 Figure 3-5.Cross-section schematic of the catalytic reactor.

PAGE 87

Figure 3-6. Tubular reactor. 87

PAGE 88

Figure 3-7. Cross-sectional view of the tubular reactor optical access window. 88

PAGE 89

89 Reacting Flow Nozzle Co-flow Nozzle Catalyst Disk Cooling Flow Nozzle Figure 3-8. Catalyst puck shown with co-flow and reacting air nozzle above and with the cooling nitrogen nozzle below.

PAGE 90

Figure 3-9. Stagnation Reactor. 90

PAGE 91

91 1064 nm Nd:YAG Laser Aperture 355 nm Dichroic Mirror Beam Dump 355 nm Dichroic Mirror Beam Dump Focusing Lens Collection Lens Fiber Optic Filters Pierced Mirror Reactor Beam Dump Figure 3-10. Top view of experimental apparatus used for gas analysis through the tubular reactor via Raman scattering.

PAGE 92

92 Beam Dump 1064 nm Nd:YAG Laser 6-Way Vacuum Cross Filters Pierced Mirror Fiber Optic Collection Lens Focusing Lens 355 nm Dichroic Mirror Beam Dump 355 nm Dichroic Mirror Beam Dump Figure 3-11. Top view of experimental apparatus used for gas analysis through the stagnation point fow reactor via Raman scattering.

PAGE 93

010020030040050025003000350040004500Intensity (a.u.)Raman Shift (cm-1)CH4H2 Figure 3-12. Example of 406 nm window spectral analysis for tubular reactor. 93

PAGE 94

-50510152025303550010001500200025003000Intensity (a.u.)Raman Shift (cm-1)H2CO2CON2 Figure 3-13. Example of 380 nm window spectral analysis for tubular reactor. 94

PAGE 95

Figure 3-14. Schematic of tubular reactor with temperature sampling for heat transfer measurements. 95

PAGE 96

460480500520540560-5051015202530Temperature (C)Distance (cm) Figure 3-15. Temperature profile for CH 4 steady state burn. 96

PAGE 97

05010015020025003000350040004500intensity (a.u.)Raman Shift (cm-1)CH4H2OH2 Figure 3-16. Sample spectra taken at window centered at 406 nm. 97

PAGE 98

02040608010050010001500200025003000intensity (a.u.)Raman Shift (cm-1)CO2CO2O2CO Figure 3-17. Sample spectra taken at window centered at 380 nm. 98

PAGE 99

CHAPTER 4 RESULTS AND DISCUSSION As discussed in chapter 3, the research presented here focuses on two, separate reactors. The results of the in situ species measurements will be presented and discussed here for both reactors, as well as heat transfer data for the tubular reactor configuration. 4.1 Tubular Reactor The data presented here for the tubular reactor is in situ Raman spectroscopic data for gaseous species identification and also heat transfer data to study the effects of surface reactions on forced convection heat transfer. 4.1.1 Raman Spectroscopy Raman spectroscopy was used to study the composition of the gasses in the reacting flow of the tubular reactor at two optical access points. As discussed in chapter 3, the data were taken at a point in the center of the annular reacting flow at locations 5 cm and 15.5 cm from the entrance to the catalyst reactor. Discussed in this section will be the calibration of the Raman spectroscopic system and the presentation of the reacting gas composition data. 4.1.1.1 Calibration As previously discussed, the intensity of the Raman scattered light is directly proportional to the species number density in the signal state for Stokes scattering. For this reason, the intensity of the Raman scattered light is directly proportional to the number of molecules of a particular gaseous species being excited. As can be seen in Figure 4-1, a typical Raman signal is defined by a peak, which decays on both sides to the baseline spectra. In order to quantify the intensity of this peak, the total area under the peak must be integrated. This integrated area is directly proportional to the concentration of molecules present in the gas flow via calibration. For each Raman spectra taken, a baseline signal must also be recorded in pure nitrogen so as to 99

PAGE 100

obtain an accurate baseline blank signal. This baseline signal is then scaled to match the baseline signal of the reacting stream Raman spectrum at the beginning and the end of the Raman peak rise. Once the blank signal has been scaled to the unknown signal, the area under the Raman peak and the same width area under the scaled blank are integrated. In order to obtain a value of the area contained under the Raman peak, the integrated baseline blank area is subtracted from the integrated Raman peak. For the preliminary Raman data, the integrated peak area was divided by the integrated peak blank in order to obtain a Peak/Base ratio. For the later studies, it was shown that simply correlating the total integrated Raman peak area to the concentration was more accurate. For all studies performed, the unit of measure of concentration of gaseous species is moles/liter. In order to calibrate the Raman system to accurately determine the concentration of each particular species of gas, data was taken at three varying concentrations and at two temperatures. The two temperatures were at room temperature with no preheat, and at the specified reacting conditions with a preheat temperature of 800K. It should be noted that for calibration, the fuel as well as the air were preheated to 800K in order to simulate reaction conditions. During the actual burn of the catalytic reactor, however, the fuel was not preheated but was mixed at room temperature with the preheated reacting feed air. The concentrations calibrated were at 0 vol %, 10 vol %, and 20 vol %. According to the ideal gas equation (shown below), the molar concentration of molecules of a gas species is inversely proportional to the absolute temperature of the gas, NPVRT (4-1) In the above equation, R is the universal gas constant, P is the pressure, and T is the absolute temperature. For our case, all studies are performed at atmospheric pressure, hence the 100

PAGE 101

concentration of molecules in the reacting gas stream is only a function of the absolute temperature of the gas stream. The absolute temperature used to calibrate for the Raman scattering at the aforementioned preheated calibration points was the average of the reactor entrance temperature and the reactor mixing temperature. For most calibration test conditions, this resulted in a hot calibration gas temperature of approximately 600K. This temperature is almost exactly twice the room temperature cold calibration gas temperature, which means that the recorded signal peaks for the hot calibration should be approximately half of the values calculated for the cold calibration at the same temperature, as can be seen in Figure 4-2. This also means that the expected value of the 10 vol % cold calibration should be equal to the expected value of the 20 vol % hot calibration. This is due to the assumption that the increase in temperature of the gas is not significant enough to alter the number of gas particles in the ground state. It can be shown through the calculation of the Boltzmann distribution that the change in number of molecules in the ground state due to a temperature rise from 300 to 600 K is negligible. Therefore, it should be expected that the calibration curve should be a linear function of the concentration of the gaseous species after correction for temperature. An example calibration curve for CH 4 is shown in Figure 4-3. As expected, this calibration curve nicely forms a linear fit, with an R 2 value of 0.98. For the preliminary Raman scattering data, all calibration data were taken prior to the run data. In order to account for various changes in the Raman scattering system such as misalignment, change in optical transmission due to window contamination, laser power, etc; a 20 vol % cold CH 4 signal was taken before and after each set of run data was taken. The signal taken before the run data was averaged with the signal taken after the run data to obtain a daily 20 vol % CH 4 signal. The concentration calculated from the integrated peak of this signal using 101

PAGE 102

the calibration equation was then ratioed to the actual concentration of the 20 vol % cold CH 4 This ratio was then multiplied by the concentrations obtained from the Raman scattering signal for that particular day in order to calibrate the signal based on the operating conditions of each day. actualcalculatedConcentrationCFConcentration (4-2) This allowed the system to be calibrated daily without having to complete a tedious four point calibration each day. For the later Raman scattering experiments, four point calibration was performed before taking the run data for the day. For the calibration for each of the component gasses, the catalyst tube was inserted into the reactor so that the flow during calibration would be similar to the flow during typical run conditions. The cooling nitrogen flow was also established through the center of the catalyst at the specified preheat temperature of 800 K. The total calibration flow volume was kept at 5 slpm, which is similar to the specified conditions for CH 4 burn at a total flow volume of 6.6 slpm. Since calibration was done with the active catalyst in place, it was important to prevent reaction of the feed gasses. In order to prevent reaction, the base gas used in all calibrations was pure nitrogen. For the calibration of each component gas, data was taken for component gas flow rates of 0 slpm (blank), 0.5 slpm, and 1.0 slpm while varying the nitrogen flow rate to keep the total flow rate at 5 slpm. The reactor was calibrated at the 406 nm centered spectral window for CH 4 and H 2 and at the 380 nm centered spectral window for CO and CO 2 In the tubular reactor it was not possible to calibrate for H 2 O due to an interfering signal given off by the reactor window at the same location (407.5 nm) as the expected H 2 O signal. It was also not possible to calibrate for O 2 in the tubular reactor due to the relatively short path length (~5 mm) for Raman scattering. While flowing O 2 through the reactor, the O 2 signal obtained by Raman 102

PAGE 103

scattering was identical to the signal obtained while flowing N 2 through the reactor. This is due to the fact that air is made up of 20% O 2 and that the optical path length through the reactor was only approximately 5 mm. This resulted in the majority of the O 2 Raman signal coming from the ambient atmosphere along the optical path length, and not from the reactor flow. Therefore it was not possible to monitor the O 2 content inside the tubular reactor. 4.1.1.2 Results and discussion As previously discussed in the calibration section, Raman scattering data was recorded for the tubular reactor in two sets. The first set was calibrated before any reaction data was taken, and used the daily spectra of the 20% CH 4 peak to calibrate the daily results for the reaction data. The second set was calibrated each day before reaction data was taken. The most important difference between the first and second data sets was the addition of a nitrogen purge flow at the optical access ports being analyzed. It was discovered that due to the high flow rates for the cooling flow, the backpressure on the reactor increased enough to force leakage from the optical access windows. The ceramic gaskets were too porous to prevent gas leakage, and better seals were not obtainable due to the high temperatures of the reactor. The leaking windows posed several problems which could affect the data. The windows were leaking reacting gas into the dead zone of the Swagelok cross and window holder which is directly in the laser path. This resulted in the Raman spectra including the gas composition of the gas in the reacting zone. Also, the leaking windows allow reacting gas to exit the reactor at the first optical port, which can have possible effects on the accuracy of optical measurements taken at the second optical access port, though the total rate of leakage was deemed minimal with regard to total flow. In order to prevent the first scenario from affecting the data, a nitrogen purge flow was directed toward the optical access ports being analyzed. This purge flow forced successfully displaced any reacting gas in the Raman path length area, thereby preventing any reacting gas from 103

PAGE 104

affecting the signal. The results of both sets of data are discussed here. There are two reactor operating conditions which have been studied. The first operating condition is the catalyst ignition behavior. In order to study the catalyst ignition behavior, Raman scattering data was taken immediately after lightoff and was continuously sampled for approximately 3 minutes (4 sets of 250 shot spectra at 5 Hz sampling rate). This data is important to study the lightoff characterstics of the catalyst in order to understand the kinetics controlling lightoff conditions such as temperature, conversion, etc. The second operating condition studied is the steady state behavior. After the ignition spectral data was sampled, the catalyst was allowed to burn for 30 minutes before steady state combustion data were recorded. The same 4 sets of 250 shot spectra were averaged for the steady state data. The steady state data is important because this is the operating condition the catalyst will be subjected to during most of its lifetime in a power generation cycle. Data were taken for the aforementioned operating conditions with both CH 4 and synthesis gas (i.e. syngas) as the fuel source. It is important to understand the surface physics of catalytic combustion in order to understand the chemical reactions which are taking place on the surface of the catalyst. For this reason, calculations have been performed in order to quantify the amount of CH 4 which adheres to the surface of the catalyst. In order to obtain this value, the collision rate of CH 4 molecules with the surface of the catalyst was calculated via kinetic theory using the following equation: 4124[](2CHRTCollisionRateCHM ) (4-3) Where [CH 4 ] is the concentration of CH 4 molecules in mol/l, R is the Universal Gas Constant in J/molK, and M CH4 is the molecular mass of the CH 4 molecule. Assuming entrance conditions for the concentration of CH 4 the collision rate of CH 4 molecules with the surface is found to be 5.6E26 collisions/m 2 s. In order to calculate the sticking rate of CH 4 molecules on 104

PAGE 105

the surface of the catalyst, one must multiply the collision rate of CH 4 with the sticking percentage. The sticking percentage of CH 4 on various catalyst surfaces has been extensively studied. Valden et al. [45] have studied the sticking probability of CH 4 as a function of incident energy for both Pt and Pd surfaces. Depending on incident energies, the values Valden et al. obtained for sticking probability on a bimetallic PtPd catalyst were found to be on the order of 0.15-5%. The average sticking rate calculated for a bimetallic PtPd surface such as the one used for the tubular reactor catalyst was found to be on the order of 8.4E23 2.8E26 molecules/m 2 s. The spectral data is calibrated to give the molar concentration of the species in question in the units of mole/l. Since the molar concentration is inversely proportional to the gas temperature and the reactor is oxidizing CH 4 thereby increasing the reacting gas temperature, the molar concentration is not an accurate means of quantifying the amount of methane present at the optical access ports. In order to quantify the conversion of CH 4 the molar concentration must be converted to mole fraction, which is completely independent of the gas temperature. In order to convert the molar concentration to mole fraction, the total molar concentration of the reacting gas must be calculated according to the ideal gas equation. Therefore, the reacting gas temperature is necessary to convert to mole fraction. In order to approximate the reacting gas temperature, it is first assumed that the cooling flow and the reacting flow reach an equilibrium temperature before mixing downstream of the catalyst. This is a good approximation due to the fact that the catalyst substrate thickness is less than 0.5 mm, and as confirmed with the spectral data the reaction is mostly complete by the second optical access port which is located 8 cm upstream of the mixing point. In order to determine the temperature at the two optical access ports, the temperature is linearly interpolated between the catalyst entrance temperature and the catalyst exit temperature. 105

PAGE 106

A plot of the CH 4 mole fraction versus the catalyst distance is presented in Figure 4-4. As expected, the CH 4 mole fraction is shown to decrease throughout the reaction zone. It is also important to note that the majority of the reaction takes place in the first 3 cm of the catalyst coating. The reaction continues to propagate between the two optical access ports, but at a much slower rate. This is in agreement with a visual inspection of the optical access ports. While the reactor is running, the catalyst rod glows bright orange in the first optical access port, with the brightest glow coming just upstream of the window. A Sekidenko optical pyrometer gives a surface temperature reading of 1040 K. A table containing all the operating temperatures for both fuels can be found in table 4-1. When CH 4 is used as the fuel, the catalyst displays no noticeable glow in the second optical access ports, hence confirming that the bulk of the reaction takes place at very short distances from the reactor entrance. This is later corroborated by the heat transfer data presented later in this chapter. It is also important to note that ignition and steady state conversion profiles have a very similar shape. This suggests that the reaction is taking place within the same location at ignition and steady state conditions, but only differs due to surface site coverage and overall catalyst temperature. It is also important to note that the shape of the data for the two data sets are very similar and the presented values of mole fraction are within 15% of each other. Possible sources for this discrepancy will be discussed later in the chapter. A plot of the H 2 mole fraction versus the catalyst distance is presented in Figure 4-6. As expected the H 2 mole fraction is shown to increase through the reaction zone. This is expected because it has been shown that the initial step for catalytic CH 4 combustion is the reduction of CH 4 into CO, CO 2 and H 2 This reflects a breakdown of the hydrocarbon in conditions with a lack of sufficient oxidizer to convert the CO and H 2 However, as discussed further, some 106

PAGE 107

significant CH 4 remains, hence ideal fuel rich products (CO and H 2 ) are not achieved. The profile of the H 2 mole fraction supports the data presented previously by the CH 4 mole fraction plot where the majority of methane conversion occurs with the first 3 cm of the catalyst coating. This is apparent due to the sharp spike in H 2 production between the catalyst entrance and the first optical access port. As with CH 4 reduction, the H 2 production rate slows considerably in the 10 cm reacting zone between the two optical access ports. As with the CH 4 mole fraction plot, the shape of the H 2 mole fraction plot is very similar for each of the 4 cases. Once again, this supports the statement that the ignition and steady state conditions are very similar with the exception of site coverage and catalyst operating temperature. With the H 2 mole fraction plot, however, the discrepancy between the two data sets varies by as much as 50%. The variation in values between the two data sets can be attributed to several factors: Catalysts have been shown to exhibit a change in the overall activity with repeated use and time on stream as noted in Chapter 1. The data shown in Figure s 4-4 and 4-6 shows an increase in the activity of the catalyst from the first to second data set, which represents approximately 20 hours of operation. While most data suggests a decrease in the activity of catalysts with time on stream and repeated use, catalysts have also been shown to exhibit an increase in activity after several combustion cycles. It is possible that the different calibration techniques have affected the obtained results. This is particularly true for H 2 detection due to a much lower Raman scattering signal than CH 4 The most likely reason for the discrepancy is the addition of the nitrogen purge at the windows, which was demonstrated to affect the calibration and overall signal strength. While all these factors can contribute to the difference in conversion rates, it is important to note that the overall shape of the conversion profiles is almost identical. Therefore, it is apparent that most of the reaction takes place in the very short entrance length of the reactor. This is a very important observation because it means that the catalysts coating length could possibly be reduced to much shorter lengths, which results in significantly less catalyst usage, hence cost savings. This is important because platinum and palladium are both very expensive noble 107

PAGE 108

metals, and a typical power generation application may require on the order of 50,000 catalyst rods. Data for the synthesis gas fuel is not presented here due to the majority of the gas concentrations being below the experimentally determined limit of detection, as can be seen in Figure 4-8 where the H 2 peak is shown for a steady state synthesis gas burn. When CH 4 is used as the fuel, data is only presented for CH 4 and H 2 mole fractions. All other gases were determined to be below the experimentally determined limit of detection. As previously discussed, the Raman scattering effect is relatively weak compared to elastic scattering. When this is combined with the physical parameters of the tubular reactor (i.e. short optical path and small optical angle) and the operating conditions of the catalyst, the signal is decreased even further. Some of the physical parameters which affect the limit of detection are the optical path length through the reaction zone, and the optical angle of collection, which is decreased due to the 10 mm diameter optical access port. The reactor was designed with these physical parameters in order to match the physical parameters provided by Siemens Power Generation, and to minimize the creation of eddys and areas of decreased velocity which could affect the reaction at the optical port. Also, as previously discussed, the intensity of the Raman scattering is directly proportional to the concentration of molecules, which is inversely proportional to the gas temperature. Due to the high gas temperatures in the reacting gas stream, the signal is decreased even further. For this reason, the stagnation reactor was built to increase both the path length to approximately 50 mm, and the optical angle by increasing the window diameter to 45 mm. 4.1.2 Heat Transfer Enhancement Studies As discussed in chapter 1, very few studies have been reported regarding the effects of convective heat transfer due to heterogeneous surface chemistry. Of the studies performed, it is 108

PAGE 109

suggested that the heterogeneous surface chemistry results in an enhancement in the convective heat transfer from the reacting surface, which in our case is the annular flow. As part of the study on the tubular reactor system, heat transfer was studied using the same proprietary catalyst. In order to quantify the convective heat transfer from the catalyst surface, it is necessary to measure the bulk flow gas temperatures along the length of the rod. Due to the mixing nozzle that is press fit into the reactor and the small annular space, it is not possible to insert a thermocouple into the annular reacting gas stream. Therefore, the cooing nitrogen flow bulk temperature (i.e. inside the catalyst tube) was measured at 1 cm intervals along the length of the catalyst tube. Several mechanisms may explain the enhanced heat transfer with catalytic combustion. In the typical case of convective heat transfer from a hot surface, the cool gas comes into contact with the surface and reaches a zero velocity at the surface. This results in conduction heat transfer to the gas in this boundary layer, as given in the relation 0"fydTqkdy (4-4) where T f is the temperature of the fluid, k is the thermal conductivity of the fluid, and y is the distance from the wall. As the gas gains energy, heat diffuses away from the surface. Since the surface is highly reactive it is possible that the molecules desorb from the surface with higher energy due to a transfer of chemical energy as well as thermal energy. In order for these products to desorb from the surface at this higher energy, their bond energy with the surface must be high enough to overcome the activation energy of the bond on the surface. These respective activation energies are given by Deutschmann et al. [34] for the catalytic combustion of CH 4 The critical surface temperature required for desorption of each particular species can then be calculated via the following equation: 109

PAGE 110

AcrET R (4-5) Where E A is the activation energy in J/mol required for desorption, and R is the Universal Gas Constant in J/molK. The calculated critical temperature for desorption of CO 2 from the surface of the catalyst is calculated to be 2466 K. This critical temperature is much higher than the values of surface temperature obtained empirically with the optical pyrometer. The Steric factor can then be calculated using 1000 K as the value for the surface temperature, and results in a Steric factor of 11.2. This means that approximately 1 out of every 12 adsorbed molecules of CO 2 will be released from the surface at a surface temperature of 1000K. Since the steric factor is an exponential function of the critical temperature divided by the surface temperature, a higher surface temperature results in a significantly higher number of molecules desorbing from the surface of the catalyst. Alternatively, the enhancement of the convective heat transfer coefficient may be explained by either a mole producing reaction or a mole reducing reaction. In a mole producing reaction, there will be a greater molar flux of product gasses leaving the surface than of reactant gasses that are adsorbed onto the surface, and will hence show an enhancement to heat transfer which is not accounted for in Newtons Law of Cooling. Likewise, for a mole reducing reaction one may see a reduction in the heat transfer with the addition of a reacting surface. In order to test these theories, the heat transfer tests were carried out with CH 4 and CO as fuel sources. These fuels were chosen because the rich combustion of CH 4 is a mole producing reaction while the combustion of CO is a mole reducing reaction, as shown here: 1CH 4 + 2O 2 1CO 2 + 2H 2 O (4-6) 2CO + 1O 2 2CO 2 (4-7) 110

PAGE 111

Therefore, if the molar flux theory holds, there should be an enhancement in the convective heat transfer to the reacting flow for the CH 4 reaction, and suppression in the convective heat transfer to the reacting flow with CO. The equivalence ratio for the specified CH 4 ratio was calculated to be 3.147. In order to keep the data comparable, the CO flow rate and reacting air flow rate was modified to match the CH 4 equivalence ratio used for heat transfer studies. This resulted in a CO flow rate of 3.76 slpm and a reacting air flow rate of 2.84 slpm. At equilibrium, these would result in the following, although full equilibrium is not expected based on the Raman spectroscopic data: 1CH 4 + 0.624(O 2 + 3.76N 2 ) 0.102CO 2 + 0.898CO + 0.146H 2 O + 1.854H 2 + 2.346N 2 (4-8) 6.294CO + O 2 + 3.76N 2 4.294CO + 2CO 2 + 3.76N 2 (4-9) It is noted that under fuel-rich, ideal combustion, 1.624 moles of reactants produces 3 moles of products for methane, and 7.294 moles of reactants produces 6.294 moles of products. The cooling nitrogen flow temperature was measured with a type K thermocouple and an Omega handheld thermometer on 1 cm intervals from the solder joint (0 cm) to the mixing point (23 cm). The temperature data points were averaged over the three sets taken, and the points were interpolated on 1 mm intervals using a 4 th order polynomial. Temperature profiles of the cooling flow for CH 4 and CO can be seen in Figure 4-9 and 4-10 respectively. The heat flux to the inner cooling flow was then calculated based on the the temperature change for each 1 mm interval and conservation of energy, and is shown in Figure 4-11, through the relation: ()"(p ) x PqmcT& (4-10) As can be seen in Figure 4-11 there exist two distinct regions of heat release to the cooling flow. It is apparent that the first region is initially of a very high energy release, which decays significantly within 3.5 cm from the start of the catalyst. The second region of heat release is 111

PAGE 112

much less dramatic and is also very short lived. It is important to note that the majority of the heat release is occurring within the first 5 cm of the catalyst. This is consistent with the data from the Raman spectroscopic studies performed on the tubular reactor, which show that greater than 90% of the reaction takes place within the first 5 cm of the catalyst. It is important to determine the underlying cause for these two regions of heat release. In fuel rich combustion, O 2 is depleted very quickly and the amount of O 2 available for combustion directly controls the amount of energy release through combustion. Catalytic combustion is much more sensitive to the availability of O 2 than homogeneous combustion because the O 2 must be adsorbed onto the surface of the catalyst before the reaction can take place. In fuel lean catalytic combustion (i.e. excess oxygen), the catalyst surface is typically saturated with O 2 and the diffusion and adsorption of fuel onto the surface of the catalyst controls the catalytic reaction. In fuel rich catalytic combustion, the opposite is true. It is the diffusion and adsorption of O 2 onto the surface of the catalyst, as well as the overall kinetics, which controls the catalytic reaction. One such hypothesis is that these two regions of heat release are directly related to the diffusion and adsorption of O 2 onto the surface of the catalyst. In order to assess this particular hypothesis, it is necessary to compare the diffusion of O 2 within the boundary layer to the consumption of O 2 on the surface of the catalyst due to catalytic combustion. In order to simplify the calculations, it is required that several assumptions be made regarding the concentration of O 2 The following equation was used to calculate the O 2 boundary layer diffusion rate: 22[][] s OOdiffD (4-11) where D is the binary diffusion coefficient of O 2 into air, [O 2 ] is the concentration of O 2 in mol/l, and is the thickness of the annular region or the hydraulic diameter. This equation maximizes 112

PAGE 113

the difference in species concentrations; however, the diffusion length scale is the full hydrodynamic diameter of the annular region. While the actual concentration differences will be reduced, the effective length scale will also be less than the full channel, hence this is a reasonable approximation. This equation yielded a value of 0.05 mol/m 2 s for the diffusion rate of O 2 into the boundary layer. Next, the consumption rate of O 2 was calculated by using stoichiometry along with the consumption rate of CH 4 required to maintain the initial heat release shown in Figure 4-11, 4initialCHqconsH (4-12) where H CH4 is the heat of combustion (kJ/mol) of CH 4 This value for the consumption of CH 4 was then multiplied by two in order to maintain stoichiometry for the reaction of CH 4 with O 2 This calculation yielded a value of 8 mol/m 2 s for the consumption rate of O 2 This certainly represents a maximum rate based on stoichiometric combustion and the maximum point of heat release. Comparing the diffusion rate of O 2 with the consumption rate of O 2 shows that in order to maintain the initial high energy heat release that O 2 must diffuse into the boundary layer with a rate higher than 8 mol/m 2 s. Since the actual diffusion rate of O 2 is significantly less than the consumption rate, this suggests a very rapid decline in the rate of energy release due to limited O 2 availability on the surface of the catalyst. Such a rate is observed with the data. This rapid decline in energy release continues until the diffusion rate of O 2 into the boundary layer is greater than the consumption rate of O 2 At this point, O 2 is then allowed to build up on the surface of the catalyst in order to start the second region of energy release. In this region, the reaction rate then ultimately declines as the reactants (CH 4 and O 2 ) are diminished, and results in a decrease in energy release for the second region. 113

PAGE 114

An alternative interpretation is the initial rapid release of energy that increases the temperature of the reacting gas stream rapidly. A sharp increase in the outer, reacting gas stream alters heat transfer to the inner, cooling flow, giving the second regime. Preliminary numerical analysis has been performed to determine if the non-reacting convective heat transfer is enhanced or suppressed due to surface reactions. In order to determine if the heat transfer is affected by surface reactions, the total heat release due to combustion is calculated using the overall change in energy (i.e. global conservation of energy) from the inlet of the reactor to the exit of the reactor. This number is then corroborated by calculating the heat of reaction from the enthalpy of formation of the products and reactants. The process was repeated for both the CH 4 and CO combustion. The CH 4 combustion case will be discussed in detail here. For the purpose of data analysis, several assumptions must be made: Rich combustion is complete and there is no remaining O 2 The two gas streams (cooling and reacting) reach an equilibrium temperature near 13 cm from the leading edge of the catalyst inlet. Heat loss to the surroundings is negligible. The products of the combustion process were calculated assuming complete combustion with no remaining O 2 and assuming no dissociation producing minor species. For this case of rich combustion, the water-gas shift reaction was also accounted for. The following equation was used to determine the stoichiometric balance: 22222(3.76)3.76xyCHaONbCOcCOdHOeHaN 22tot (4-13) 22COHOCOH (4-14) (/4)/axy (4-15) The mole fractions can then be calculated from the stoichiometric coefficients: /ii x NN (4-16) 114

PAGE 115

From this calculation of product mole fractions, the mass flow rates of the products were calculated. iimNMW i]4H (4-17) This information was then used along with the entrance temperature data and equilibrium temperature data to calculate the total heat liberated by combustion using the total heat capacity equation along with conservation of energy: [))pffpiiQmcTcT& (4-18) The heat liberated by combustion was found to be 122.7 W for CH 4 combustion and 90.9 W for CO combustion. The heat liberated through combustion was also calculated by using the previously determined stoichiometric flow rates and the heat of formation for each of the reactant and product species. The entrance temperature of the reacting gas mixture was calculated by applying the total energy equation for the air and fuel both before and after mixing, 22224rxnHOHOCOCOCOCOCHCHNHNHNHNH (4-19) The heat released through combustion as calculated from the heat of formation was found to be 115.5 W for CH 4 combustion, which is within 5% of the value calculated from the total heat capacity of both flows, which was 122 W. The heat lost to the surroundings was then calculated to validate the assumption of negligible losses (i.e. perfect insulation). The heat loss to the surroundings was estimated to be less than 1 W, thus validating the assumptions of negligible heat transfer losses. 4.1.2.1 Linear Heat Transfer Model In order to calculate the heat released to the annular flow with no surface reactions, the tubular reactor was modeled using classical correlations for the inner cooling flow, a non-reacting condition that is readily modeled. The heat flux to the cooling flow is calculated using 115

PAGE 116

the mass flow rate of the cooling flow and the change in temperature of the bulk flow for each 1 mm interval. This results in obtaining the heat flux to the cooling flow on 1 mm intervals, as noted above in Equation 4-10. The total heat transferred to the interior flow can then be calculated by integrating the values of the heat flux over the entire length of the catalyst tube. The Reynolds number for the cooling flow was calculated to be 5420, which represents a flow transition from laminar to turbulent conditions for internal flow. The Gnielinski correlation [46] was used to calculate the Nusselt number for the cooling flow: 1/22/3(/8)(Re1000)Pr112.7(/8)(Pr1)ddfNuf (4-20) where f is the friction factor and is calculated from: 2(0.790lnRe1.64)df (4-21) This formula for the calculation of the friction factor was presented by Petukhov and Rohsenow, assumes a smooth pipe and matches the Reynolds number calculated for the cooling flow. The Nusselt correlation used is also valid for the transitioning cooling flow and results in a Nusselt number of 18.7, this was in excellent agreement with other correlations, such as the widely used Dittus-Boelter correlation. In order to calculate the convective heat transfer coefficient, the thermal conductivity of the cooling nitrogen flow must also be calculated. Since the bulk temperature of the nitrogen flow is constantly increasing, the values of the thermal conductivity will also be changing. The values of thermal conductivity were interpolated to match the measured temperature calculated on 1 mm intervals. Once the convective heat transfer coefficient is calculated, the surface temperature of the catalyst tube is then calculated for each 1 mm interval using Newtons Law of Cooling, "surconvqTh T (4-22) 116

PAGE 117

To calculate the heat transfer to the outer flow, a linear rise in the reacting flow bulk temperature along the length of the catalyst was assumed for the preliminary analysis. The Nusselt correlation for annular flow with one heated surface and one insulated surface proposed by Rohsenow and Hartnett was used [47]. This Nusselt correlation is based on the hydraulic diameter, and yields a Nusselt number of 5.625 for the current annular flow conditions. In addition, the thermal conductivity must be calculated for the reacting gases in order to calculate the classical convective heat transfer coefficient. As previously discussed in this chapter, the final products have been calculated assuming ideal rich combustion. To obtain a value for the total thermal conductivity of the reacting gases, the reactants are assumed to be linearly converted to products over the length of the reactor. This linear approximation was later changed to an exponential fit based on the heat transfer model, as discussed in the next section. The percent composition by mass of each reactant gas is then calculated on 1 mm intervals. The thermal conductivity of each reacting gas was then interpolated over the length of the catalyst on 1mm intervals for each value of the reacting gas temperature. This allows for the calculation of the convective heat transfer coefficient at each 1 mm interval. Once the convective heat transfer coefficient is calculated, the classical case heat flux to the annular flow can be calculated at each interval. The total heat transfer to the annular flow can then be calculated by integrating this heat flux over the entire surface of the catalyst. The total classical heat released can then be calculated by adding the total heat released to the cooling flow and the total heat released to the annular flow. In the case of CH 4 the total heat released in the classical model was calculated to be 122 W. This is identical to the heat release calculated by temperature rise, and is 5% higher than the heat release calculated from the enthalpy formation of the reactants and products. Since the calculated value of total heat release for the classical 117

PAGE 118

model is very similar to the actual heat rise, this shows that enhancement due to surface reactions may not be significant on average, although local effects may be important. For CO, the total heat released in the classical model was calculated to be 82 W. This is slightly lower than the heat release calculated from the temperature rise, which means that more heat is transferred in the actual case than in the classical case. Therefore, according to these preliminary results CO combustion results in enhancement to classical heat transfer. It should be noted that for both cases the total heat transfer calculated in classical model and the experimental results are within 10% of each other. Note that these calculations represent a linear temperature profile based on conservation of energy; hence the agreement is not suprising. A more realistic approach should consider a variable rate of reaction, as observed with the Raman data measurements. 4.1.2.2 Exponential Heat Transfer Model By performing these calculations in order to obtain values for the convective heat transfer coefficient from the surface of the catalyst, one must assume a heat release profile for the overall release of energy due to catalytic activity. It is intuitive to calculate a profile similar in shape to the heat flux profile for heat transfer to the cooling flow. Furthermore, an exponential decay, which is consistent with the initial heat release profile, is also consistent with the expected reaction physics, in which the heat release rate should decay exponentially as the reactants are consumed. However, there is no empirical data to support this profile. Therefore, additional temperature measurements were recorded in the reacting stream in order to validate this assumed heat release profile, and attempt to close the model. In order to obtain reacting gas stream measurements, the optical access ports on the tubular reactor were fitted with two 0.0625 inch type K thermocouples. The thermocouples were inserted toward the upstream edge of the optical access port into the center of the reacting gas flow stream. Measurements were repeated for both fuels on three, separate reactor runs. These temperatures were then averaged to obtain reacting 118

PAGE 119

gas flow temperatures for both port one and port two, which are located at 5.5 cm and 15 cm downstream of the reactor inlet. For the fuel rich combustion of CH 4 the reacting gas flow temperatures at port one and port two were found to be 928 K and 782 K, respectively. For the fuel rich combustion of CO, the reacting gas flow temperatures at port one and port two were found to be 914 K and 722 K, respectively. These values for the reacting gas flow temperature were then used to modify the heat transfer model. The total heat release profile was constructed two ways: the first profile was an exponential decay function with variable exponential parameters. This exponential decay is the expected trend for typical combustion processes where reactants are being consumed much faster at the beginning of the process. The second profile more closely modeled the profile of heat transfer to the nitrogen cooling flow. This profile consisted of a primary reaction zone defined by an initial exponential decay function, followed by a secondary reaction zone similar in proportion to the secondary heat release to the nitrogen cooling flow. For both profiles, the heat released to the outer, reacting flow was calculated on 1 mm intervals by subtracting the heat released to the nitrogen cooling flow (based on direct measurements) from the overall heat release profile. From there, the change in temperature of the reacting gas flow was calculated by using the total energy equation of the reacting flow, ("")()totalcoolingflowpqqxPcm& T (4-23) In order to use the total energy equation to calculate the temperature difference, one must calculate the specific heat of the reacting gas flow at each 1 mm interval. The initial reactant feed into the reactor is known, and the final product feed can be calculated assuming ideal rich combustion. The reacting flow composition is then matched to the exponential decay function in order to obtain the species concentrations along each 1 mm interval. The values of specific heat 119

PAGE 120

as a function of temperature are matched to a fourth order polynomial in order to obtain the specific heat of each reacting species as a function of the gas flow temperature. Once the reacting gas flow temperature profile is calculated, the Nusselt number is calculated via the Rohsenow correlation [47]. This Nusselt number is then used to calculate the classical convective heat transfer coefficient from the surface of the catalyst to the reacting gas flow stream. This classical convective heat transfer coefficient can then be used along with Newtons Law of Cooling and the measured surface and calculated reacting gas flow temperatures to calculate the expected classical heat transfer between the surface of the catalyst to the reacting gas flow. In order to obtain the heat released to the reacting flow due to catalytic enhancement, the values obtained for the classical heat transfer to the reacting flow are subtracted from the calculated total heat released to the outer, reacting flow based on the exponential profiles. These calculations are iterated until the reacting flow bulk temperature converges to a steady value. In addition, overall conservation of energy is imposed to match the values calculated by the heat of combustion and the measured total energy releases, namely 122 W for CH 4 and 91.0 W for CO. These calculations are performed in a similar manner for both the single heat release profile and for the dual heat release profile in order to obtain values for the convective heat transfer enhancement to the reacting flow. Figure 4-12 shows the total heat transfer, heat transfer to the cooling flow, and the heat transfer to the outer flow for the single exponential decay total heat release for the catalytic combustion of CH 4 As can be seen from the plot, the heat flux to the outer, reacting flow is very similar in shape to the total heat release fit until 4 cm. At this point, the heat release to the reacting flow begins to decay and the heat release to the cooling flow increases. This inflection point is very important because it is 1 cm short of the window. What the data suggest is that the 120

PAGE 121

catalytic rate of reaction has rapidly decayed at this point. Once again, this is in agreement with the Raman spectroscopic data which show that greater than 90% of the catalytic reaction has been completed by 5 cm. Since the bulk of the catalytic reaction is completed at this point, the enhancement to the convective heat transfer due to catalytic activity is also rapidly decaying. Since the reacting gas flow is now hotter than the surface of the catalyst, the reacting gas flow begins transferring heat to the surface of the catalyst and then into the cooling gas flow stream, which nicely explains the secondary release observed in Figure 4-11. This results in negative values for the heat transfer between the catalyst surface and the reacting gas flow (i.e. heat transfer from the gas to the surface). Figure 4-13 shows the heat flux to the reacting flow, the classical heat flux to the reacting flow calculated via Newtons Law of Cooling, and the heat flux to the reacting flow due to the catalytically active surface. As can be seen in the figure, the heat transfer due to the catalytically active surface decays exponentially with the decay of the overall combustion process. Moreover, the initial heat flux due to the catalytically active surface is on the order of 600,000 W/m 2 This value decays to around 60,000 W/m 2 at 4 cm. This is, once again, in agreement with the previously discussed Raman spectroscopic data which show that greater than 90% of the catalytic process is complete with 5 cm. It is also important to note that the convective heat transfer to the reacting flow calculated for the classical case becomes negative near 1 cm. This occurs because the reacting gas bulk temperature becomes greater than the surface temperature due to enhancement from the catalytically active surface. This is quite remarkable because it suggests that heat transfer from a catalytically active surface can occur from a colder, less energetic surface to hotter, more energetic gas due to the initial energy transferred via the molecular interactions. As the heat release due to the combustion process subsides, this 121

PAGE 122

overheated reacting flow begins to transfer heat back to the catalyst tube, through the surface of the catalyst rod, and into the cooling flow. This transfer of heat is responsible for the secondary heat flux release to the cooling flow as shown in Figure 4-12, and is consistent with classical convective heat transfer. Figure 4-14 shows a plot of the total heat flux released due to combustion and the heat flux to the reacting flow due to catalytic enhancement for the dual heat release model for the catalytic combustion of CH 4 As can be seen in the Figure the heat release profile begins with an initial exponential decay until 2.4 cm, where the secondary heat release begins. As can be seen in the plot, the heat release due to catalytic enhancement is shown to be negative at the entrance to the reactor, and reaches a maximum value of 100,000 W/m 2 near 1.8 cm. This is not physically possible because the heat release due to catalytic enhancement should directly resemble the total heat release from the catalytic combustion of the fuel. The maximum value of heat release due to catalytic enhancement should be at the beginning of the reaction zone, where the total heat release is at a maximum value. For this reason, the single exponential decay function is determined to be the most reasonable model, and will be used exclusively henceforth to model the convective heat transfer for both the catalytic combustion of CH 4 and CO. It is important to note that the calculated values for the reacting gas temperatures are 1108 K and 988 K for port one and port two, respectively. This results in a deviation from the measured values of 19% and 26% for port one and port two respectively. The calculated change in temperature from port one to port two is 120 K, which is 17.8% less than the measured temperature change. The single exponential decay model for the catalytic combustion of methane is shown to have a positive enhancement to the convective heat transfer to the reacting flow. This is in agreement with both of the aforementioned theories. 122

PAGE 123

As seen in Figure 4-15, the single exponential decay heat release model for the catalytic combustion of CO is quite similar to the single exponential decay model for the catalytic combustion of CH 4 Up until 5 cm from the reactor entrance, the heat flux to the outer flow very closely matches the total heat release profile. Once again, this suggests that at short distances from the reactor entrance, the majority of the heat released through combustion is transferred to the reacting gas flow both catalytically and via classical convection. Similarly to the catalytic combustion of CH 4 this trend suggests that the majority of the combustion process is complete by the 5 cm location of port one. Figure 4-16 shows the heat flux to the reacting flow, the classical heat flux to the reacting flow calculated via Newtons Law of Cooling, and the heat flux to the reacting flow due to the catalytically active surface. As can be seen in the figure, the heat transfer due to the catalytically active surface decays exponentially with the decay of the overall combustion process. Moreover, the initial heat flux due to the catalytically active surface is on the order of 450,000 W/m 2 This value decays to around 45,000 W/m 2 at 4.5 cm. This is very similar to the Raman spectroscopic data taken for the catalytic combustion of CH 4 where greater than 90% of the combustion process was shown to be complete by 5 cm. It is also important to note that the convective heat transfer to the reacting flow calculated for the classical case becomes negative near 1 cm. This occurs because the reacting gas bulk temperature becomes significantly greater than the surface temperature due to enhancement from the catalytically active surface. This is identical to the case for the catalytic combustion of CH 4 Modeling work was also performed for a dual heat release profile for the catalytic combustion of CO. This dual heat release model had the same false trend for the enhancement to the convective heat release as the CH 4 model, and thus was discarded. 123

PAGE 124

As shown in the plots, the enhancement to the convective heat transfer to the reacting flow is clearly positive for the catalytic combustion of CO as well as CH 4 This enhancement data is presented in Figure 4-17 for the first 8 cm of catalyst length for CH 4 and 9.5 cm for CO. This trend suggests that the enhancement to the convective heat transfer is more greatly influenced by the desorption of highly energetic products from the surface of the catalyst to the reacting gas flow, rather than by a change in the number of moles of products versus moles of reactants. Some desorption analysis, with respect to the thermodynamics of the process, has been performed for the catalytic combustion of CH 4 and CO. It is theorized that when the products desorb from the surface they enhance typical convective heat transfer by escaping from the surface with a total energy value that lies between the enthalpy of the molecule, and the activation energy required for the molecule to desorb from the surface of the catalyst. An exact partitioning of the excess energy is not available at this time from the literature. In order to validate this theory, the total enthalpy of each reactant and product species was calculated using the appropriate surface and reacting gas flow temperatures. The energy flux into the surface was calculated by multiplying the molar consumption of each reactant species (based on ideal fuel-rich combustion) by the respective total enthalpy of each species calculated based on the reacting gas flow temperature, "(ENhT& ) (4-24) where E is the energy flux, N is the molar consumption/production, and h is the total enthalpy for the given species. The energy flux out of the surface was also calculated based on ideal rich combustion, and was calculated by multiplying the molar production of each product species by the respective total enthalpy of each species calculated based on the catalyst surface temperature. 124

PAGE 125

The energy flux into the surface was then subtracted from the energy flux out of the surface in order to find the net energy released from the surface to the gas, ""netoutinEEE (4-25) This net energy release value was then compared to the enhancement of heat transfer due to catalysis which was previously calculated. An enhancement factor was calculated to determine the net increase in energy flux out of the surface in order to match the net energy release to the convective heat transfer enhancement due to catalytic combustion """catalyticinoutqEfE (4-26) For the catalytic combustion of CH 4 this enhancement factor is found to vary from 0.65 to 6.0, and was inversely proportional to surface temperature. At high surface temperatures near the reactor inlet, the enhancement factor is 0.65, whereas at lower surface temperatures near the end of the reacting zone, the enhancement factor is 6.0. Although this dependence on temperature is not a perfect inverse function, an inverse model for this enhancement factor is better suited than a simple average. By taking this temperature dependence into account, the values for the enhancement factor range from 1.8 to 3.0. Based on the concept of difference in excess energy flux between reactant and product species, it is possible to model the catalytic enhancement and then compare to the experiments. The catalytic heat transfer is directly calculated by the expression: 2222222244"()(/)()(()(/)()(/)()()catalyticCOCOsurfacesurfaceCOCOsurfacesurfaceHHsurfacesurfaceHOHOsurfacesurfaceOOgasCHCHgasqNhTfTNhTfTNhTfTNhTfTNhTNhT&&&&&& /) (4-27) where N i and h i are the molar flux (mole/cm 2 s) and enthalpy (J/mole) of the i th species, and f represents an empirical parameter set to a constant value of 3.2. The empirical constant 125

PAGE 126

represents a scaling of the excess energy in desorbed molecules somewhere between the enthalpy in equilibrium with the surface temperature and the surface activation energy. Figure 4.18 shows the modeled and experimentally predicted catalytic heat transfer as a function of catalyst length throughout the reacting zone. While the agreement is not perfect, given the very significant assumptions in the analysis, it is considered reasonable. The values of the enhancement factor suggest that the mechanism for enhanced heat transfer to the reacting flow requires more energy to be transferred to the reacting flow than simply the total enthalpy of the desorbed species at the surface temperature. It is reasonable to assume that some portion of the activation energy required for the product species to desorb from the catalyst surface are also contained in vibrational and translational energy modes. The results for the catalytic combustion of CO are very similar to those for the catalytic combustion of CH 4 However, the enhancement factors for the catalytic combustion of CO range from 3.0 to 42. Although the overall range of the enhancement factors is similar for CO, the actual values are higher. This means that there is much more energy unaccounted for in the enhancement model for the catalytic combustion of CO. This could possibly be explained by the very high activation energy (152500 J/mol) required for the desorption of CO from a Pt surface. This would further validate the theory that part of the enhancement energy comes from the activation energy during the desorption of the product species and, in this case, leftover unreacted reactant species. Although not enough information is available to definitively determine the mechanism of catalytic enhancement, it is possible that the enhancement to the convective heat transfer to the reacting flow is due to additional energies being transferred along with the products from the surface of the catalyst to the reacting gas flow. In order to better quantify the convective heat transfer enhancement from the surface, it is imperative to obtain more accurate and precise 126

PAGE 127

measurements of the reacting gas flow so that a total heat release profile can be calculated rather than modeled. The most logical way of obtaining these reacting gas flow temperatures is to construct a tubular reactor housing with no optical access ports, and with thermocouples embedded into the housing and extending into the reacting gas flow on 1 cm intervals. This new reactor configuration is straightforward to construct and to obtain the desired measurements. It is suggested that the thermocouples be embedded out of phase along the length of the reactor so that upstream thermocouples will not affect measurements obtained from the downstream thermocouples. 4.1.3 Catalytic Combustion Stoichiometry Studies In order to better understand the effects of varying the overall stoichiometry for the catalytic combustion of CH 4 and syngas, the operating temperatures of the reactor were recorded while varying the overall stoichiometry of the entrance feed. Temperature values from the permanently installed thermocouples at the entrance to the reactor (T 1 ) and downstream of the mixing point (T 2 ) were recorded, as well as surface temperatures at Port 1 (5.5 cm) and Port 2 (15 cm), which were recorded by a Sekidenko Optical Pyrometer. Measurements were carried out for both CH 4 and methane as fuel sources. For the catalytic combustion of CH 4 the equivalence ratio was varied from 2.6 to 3.8. Table 4-2 shows the flow rates used for all measurements. It is important to note that the standard operating conditions used for the Raman spectroscopic data were at an equivalence ratio of 3.15. Since an exact equivalence ratio can not be calculated for syngas, the air flow rate was varied to match the air flow rates for the combustion of CH 4 As the airflow was increased for the combustion of syngas, the increase in surface temperatures was found to result in a premixed flame inside the reactor. This usually occurred with air flow rates between 5.245-5.620 slpm. 127

PAGE 128

Since the most fuel rich case results in the lowest conversion and lowest overall operating temperatures, the first set of data was taken at an equivalence ratio of 3.8. The reactor was ignited at this equivalence ratio and was allowed 45 minutes to reach steady state conditions. After the first measurement was made, the airflow was increased to lower the equivalence ration to 3.6. Upon altering the feed conditions, the reactor was allowed 15 minutes to achieve steady state conditions for each subsequent measurement. It is important to note that although equivalence ratios cannot be calculated for syngas due to the multiple fuel components, equivalence ratios were assigned to the syngas measurements which are based on the same air flow parameters as the calculated equivalence ratios for the combustion of CH 4 For example, at an equivalence ratio of 3.0 the air flow rate for the CH 4 combustion is the same as the air flow rate for the syngas combustion labeled at an equivalence ratio of 3.0. Figure 4-19 shows the four temperature measurements as a function of equivalence ratio for the catalytic combustion of CH 4 The reacting flow inlet temperature (T 1 ) changes the least, from maximum of 595 K at an equivalence ratio of 2.6 to a minimum of 574 K at an equivalence ratio of 3.8. This is simply due to conservation of total energy since the fuel inlet temperature is 300 K, and the fuel-to-air flow rate ratio increases with increasing equivalence ratio. The downstream mixing temperature (T 2 ) also exhibits a maximum of 790 K at an equivalence ratio of 2.6 and a minimum of 748 K at an equivalence ratio of 3.8. The difference in values between T 1 and T 2 is a good indicator of the total heat release through combustion. The surface temperature of the catalyst at the upstream optical port 5.5 cm from the start of the catalyst coating (T P1 ) has the most significant change in temperature with a maximum of 1110 K at an equivalence ratio of 2.6, and a minimum of 994 K at an equivalence ratio of 3.8. This is very important because previous work has shown that at surface temperatures above 1173 K the 128

PAGE 129

surface of the catalyst begins to overheat and lose activity. The surface temperature of the catalyst at the downstream optical port 15 cm from the start of the catalyst coating (T P2 ) shows a much less drastic change in temperature as a function of equivalence ratio, with a maximum of 834 K at an equivalence ratio of 2.6, and a minimum of 765 K at an equivalence ratio of 3.8. This suggests that the majority of the reaction is complete at or near the downstream optical port. This is, once again, in agreement with both the Raman spectroscopic studies and the heat transfer studies performed on the tubular reactor. Figure 4-19 suggests that as more air is fed into the reactant stream to lower the equivalence ratio, then the total heat released through the catalytic combustion of CH 4 also increases, due to the fact that the surface temperatures and the final downstream mixing temperature are increasing. In order to validate this observation, the total heat released through combustion was calculated assuming ideal rich combustion for each equivalence ratio shown in Figure 4-20. This data can be seen in Figure 4-20. It was also vital to calculated the difference in T 2 and T 1 as a function of equivalence ratio since this overall temperature difference is directly related to the heat released through the actual operation of the tubular reactor. As can be seen in Figure 4-20, the heat released through ideal rich combustion reaches a maximum of 193 W at an equivalence ratio of 2.6, and a minimum of 51.9 W at an equivalence ratio of 3.8. This ideal rich combustion model results in a 75% decrease in the heat released through combustion for an equivalence ratio range of 2.6 3.8. However, the temperature difference only results in a decrease from 195 K to 174 K, which is only a ten percent decrease. It is important to note that the actual decrease in energy release based on the temperature difference along the length of the reactor is 31.3% when the change in heat capacity due to the higher total flow rates is accounted for. This is much less than the change in energy release based on ideal rich combustion. This is 129

PAGE 130

due to the fact that the catalytic combustion of CH 4 does not result in complete oxidation of CH 4 as shown in the Raman spectroscopic data for the tubular reactor. Figure 4-21 shows the four temperature measurements as a function of equivalence ratio for the catalytic combustion of syngas. The reacting flow inlet temperature (T 1 ) changes the least from maximum of 576 K at an equivalence ratio of 3.0 to a minimum of 566 K at an equivalence ratio of 3.8. This is simply due to conservation of total energy since the fuel inlet temperature is 300 K, and the fuel to air flow rate ratio increases with increasing equivalence ratio. The downstream mixing temperature (T 2 ) also exhibits a maximum of 817 K at an equivalence ratio of 3.0 and a minimum of 790 K at an equivalence ratio of 3.8. The difference in values between T 1 and T 2 is a good indicator of the total heat release through combustion. The surface temperature of the catalyst at the upstream optical port 5.5 cm from the start of the catalyst coating (T P1 ) has the most significant change in temperature with a maximum of 1193 K at an equivalence ratio of 3.0, and a minimum of 1069 K at an equivalence ratio of 3.8. This is very important because previous work has shown that at surface temperatures above 1173 K the surface of the catalyst begins to overheat and lose activity. This indicates that the catalyst may actually be sintering at these lower equivalence ratios. Interestingly, as the equivalence ratio decreased past 3.0, pre-mixed homogeneous flames were observed to form inside the tubular reactor. This suggests that the upper flammability limit of syngas is near an equivalence ratio of 3.0 under catalytic combustion conditions. This is similar to the published flammability limits of H 2 and CO, which are 2.54 and 6.76, respectively. The rate of increase of the surface temperature with syngas at the upstream optical port for decreasing equivalence ratios is much greater than the rate of increase of the surface temperature for the catalytic combustion of CH 4 This also suggests that syngas has higher flammability 130

PAGE 131

limits than CH 4 The surface temperature of the catalyst at the downstream optical port 15 cm from the start of the catalyst coating (T P2 ) shows a much less drastic change in temperature as a function of equivalence ratio, with a maximum of 875 K at an equivalence ratio of 3.0, and a minimum of 842 K at an equivalence ratio of 3.8. Once again, this data suggest that the heat released through the combustion of syngas is much greater than the heat released through the catalytic combustion of CH 4 This suggests that the majority of the reaction is complete at or near the downstream optical port. This is, once again, in agreement with both the Raman spectroscopic studies and the heat transfer studies performed on the tubular reactor. Due to the much more complicated model for fuel rich combustion of syngas due to the multiple component fuels, the ideal heat release for rich combustion was not calculated for the catalytic combustion of syngas. However, it is apparent from the data presented in Figure 4-21 that the heat released through the catalytic combustion of syngas also increases with decreasing equivalence ratio as is the case for the catalytic combustion of CH 4 This data presented in Figure s 4-19 and 4-21 suggest that the more energy can be released through the catalytic combustion of CH 4 by running the reactor at leaner conditions without causing sintering of the catalyst and substrate. However, Figure 4-21 suggests that the reactor cannot be run leaner with syngas without subjecting the catalyst to surface temperature which may result in sintering and deactivating the catalyst and substrate. This is important because it may be possible to release more energy from CH 4 through the catalytic process, thereby reducing the energy released through homogeneous combustion, which may results in an overall reduction of NO x emissions. 4.2 Stagnation Reactor The stagnation reactor was built after the testing on the tubular reactor revealed the limitations of the optical design. The stagnation reactor was designed to increase the two optical parameters which greatly influence the sensitivity of Raman spectra: optical collection angle, 131

PAGE 132

optical path length. The path length in the stagnation reactor was increased to approximately 50 mm, and the optical solid angle was increased by increasing the diameter of the windows to 45 mm. Another reason for the construction of the stagnation reactor was for ease of modeling. In the tubular reactor there are concentration gradients in both the axial and radial directions. In a stagnation reactor the concentration gradients are essentially one dimensional and vary only in the vertical distance from the catalyst surface. 4.2.1 Raman Spectroscopy Raman Spectroscopy was used to study the composition of the gasses in the reacting flow of the stagnation reactor as a function of vertical distance from the catalyst surface. As discussed in chapter 3, the data were taken at several points from near surface (0 mm) to 5 mm from the surface on 1 mm intervals. This was obtained by translating both the nozzle and the catalyst disk in the vertical direction. Discussed in this section will be the calibration of the Raman spectroscopic system and the presentation of the reacting gas composition data. It is important to note that these preliminary Raman spectra were obtained prior to the installation of the nitrogen co-flow nozzle. 4.2.1.1 Calibration The Raman scattering calibration for the stagnation point flow reactor is very similar to the calibration method used for the tubular reactor. However, hot calibration sets were not taken due to the inability to effectively measure the gas temperature at different heights. The stagnation reactor was calibrated for the same gas species as the tubular reactor, with the additional calibration for H 2 O. The optical windows used in the stagnation reactor were shown to not have the same interference spectra as the windows on the tubular reactor, as can be seen in Figure 4-22. This allowed the stagnation reactor to be calibrated for H 2 O as well as the other gasses studied in the tubular reactor. Another benefit of the increased path length in the stagnation 132

PAGE 133

reactor is the ability to reliably monitor O 2 Due to the increased path length, the Raman signal contains only spectra of the gasses inside of the reactor, and the ambient environment has no effect on the O 2 signal. As seen in Figure 4-23, the Raman scattering signal for pure nitrogen flow shows absolutely no signal for the O 2 content originating from the ambient air, and when air is flown through the reactor, the O 2 signal can be seen in the Raman spectra as shown in Figure 4-24. The additional interference line shown on these two plots was found to be the result of ghost lines from the 532 nm residual light in the laser path. The insertion of a 470 nm low pass glass filter (Schott UG-5) in front of the collection optics, as discussed in chapter 3, solved this problem and removed the interference line from all further spectra. An example of the 20 vol % CH 4 calibration before and after insertion of the low pass filter can be seen in Figure 4-25. H 2 CH 4 and H 2 O were all studied for a spectral window centered at 406 nm. CO, CO 2 and O 2 were all studied for a spectral window centered at 380 nm. In order to prevent surface reactions from taking place all during collection, gasses were flown at 0, 5, 10, and 20 vol % concentrations with the rest of the gas concentration made up entirely of inert nitrogen. For all gasses except H 2 O, the gasses were flown through the Alli-Cat Scientific mass flow controllers. The calibration for H 2 O proved to be much more difficult. In order to add moisture to the flow stream, nitrogen was flown through into the reactor at a total flow rate of 5 slpm through two lines. One line flowed dry nitrogen and connected directly to a 4-way Huntingtion Vac vacuum cross. The second line flowed nitrogen through a basic water bubbler which added H 2 O to the flow stream and then connected to the 4-way cross. Both streams mix upon entering the 4-way cross. Mounted inside of the 4-way cross was an Omega Thermo-Hygrometer which measured the temperature and the relative humidity of the combined nitrogen flow. In order to obtain a four point calibration, the percentage of nitrogen flown through the bubbler was varied 133

PAGE 134

from 0 100 vol %. The empirical correlation proposed by P.R. Lowe and J.M. Ficke 1974 was used to calculate the vapor pressure of water from the relative humidity measurements: 123456((((()))))oeaTaTaTaTaTaTa (4-28) where T is temperature in units of C, and a i is given from empirical results. From this, the following equations were solved simultaneously to give the molar concentration of water vapor as a function of relative humidity: 2100()HOPRHeT (4-29) 22[] H OairHOxn 22 H OHOPxP (4-30) 100(273.15)airPnRT (4-31) 22[] H OairHOxn (4-32) In the above, where RH is relative humidity, P H2O is the partial water vapor pressure, P is the total pressure, x H2O is the molar fraction of water vapor, and n air is the total molar concentration of the air. The intensity of the H 2 O peak in the Raman spectra can then be solved as a function of the molar concentration in order to calibrate the system for H 2 O. Similar to the first calibration scheme used for the tubular reactor, two sets of four point calibrations were performed prior to the operation of the stagnation reactor. In order to account for day to day changes in the optical system, a daily calibration of the 20 vol % CH 4 signal was taken before and after the reacting data sets. As with the tubular reactor, CH 4 was used for these daily calibrations due to the fact that CH 4 has the strongest Raman signal of all the calibrated gasses, and hence is the most reliable signal. 134

PAGE 135

4.2.1.2 Results and discussion Raman scattering data collection for the stagnation reactor thus far has consisted of several averaged sets of H 2 burn data and several averaged sets of synthesis gas burn data. The air and fuel flow rates were determined based on a steady and uniform hot reacting zone on the surface of the catalyst of about 2 cm in diameter. For the H 2 burn tests, the reacting air flow rate into the reactor was 10 slpm, and H 2 flow rate into the reactor was 3.5 slpm. For the synthesis gas burn tests, the reacting air remained at 10 slpm, and the total flow rate of synthesis gas was 5.6 slpm at the same component percentages discussed previously in this chapter. The catalyst used in these tests was a pure platinum catalyst on a Haynes 230 alloy substrate. For these tests, the nitrogen co-flow nozzle was not used, but the nitrogen cooling flow of the bottom surface of the catalyst substrate was set to 30 slpm at room temperature. The data was collected for both ignition and steady state cases at vertical heights from near surface (0 mm) to 4 mm on 1 mm intervals. Data was collected for H 2 O 2 and H 2 O at both the 406 nm and 380 nm centered spectral windows for H 2 combustion. CO and CO 2 data were not collected due to lack of hydrocarbon fuel and hence CO x formation. For the synthesis gas burn, however, data was collected for H 2 O 2 CO, and CO 2 As shown in Figure 4-26, the H 2 O signal was easily quantified for the stagnation reactor, and was unhindered by interference from the optical window as in the case of the tubular reactor. This will prove beneficial to monitor the decomposition of H 2 in the combustion process. A plot of the concentration of O 2 and H 2 as a function of vertical distance from the catalyst during the combustion of H 2 over a platinum catalyst can be seen in Figure 4-27. As can be seen in the plot, H 2 and O 2 are both shown to increase near the surface and decrease farther away from the surface. A plot of the concentration of key reacting species for the combustion of synthesis gas over the same platinum catalyst can be seen in Figure 4-28. As with the Raman scattering data from the combustion of H 2 the reactants are shown to be at higher concentrations near the 135

PAGE 136

surface. This is in contrast with the preliminary modeling effort [48]. It is believed that due to a dead space on the edges of the catalyst disk, the reacting gasses may be building up and circulating above the surface of the disk, which results in an increase in signal for these reacting gasses. For this reason, the co-flow nozzle previously discussed in chapter 3 was designed and installed to maintain a stagnation point flow and to flush the reacting gasses to the outside of the disk and into the exhaust stream. The co-flow will also help to maintain the stagnation jet profile of the reacting gasses, and hence allow for more accurate Raman scattering data to be obtained from the stagnation reactor. Further testing was done with co-flow nozzle installed to determine if the problem is corrected. There are several other issues with the stagnation flow reactor which must be corrected in order to obtain reliable data. In a stagnation flow reactor, reaction gradients are of the utmost importance inside the thermal boundary layer. The modeled boundary layer thickness for the flow rates used in the study of combustion and syngas is approximately 1mm. This extremely small boundary layer makes it difficult to obtain molecular species data. The diameter of the laser pulse at the center of the reactor is 2.2 mm. Since the diameter of the laser is approximately twice the thickness of the boundary layer it is difficult to obtain reliable molecular species data within the boundary layer. In order to resolve this issue, the flow rates were then reduced by 33.3% in order expand the modeled boundary layer to 1.5 mm. These modified flow rates can be seen in table 4-3. Additionally, a new best form focusing lens has been installed in front of the reactor to reduce the laser beam diameter to 0.9 mm at the center of the reactor. This lens also helps to ensure a more uniform laser beam profile with the stagnation flow reactor. This smaller beam diameter along with the expanded boundary layer allows for the collection of several data points within the boundary layer. 136

PAGE 137

With the installation of this N 2 co-flow nozzle, it is important to understand how it will affect the fluid dynamics of the reacting gas flow. For this reason, some preliminary calculations were performed in order to determine the diffusion rate of the N 2 co-flow into the reacting gas flow stream. Initial calculations for the Reynolds number show that both the N 2 co-flow and the reacting gas flow are in the laminar regime with values of 466.3 and 1786 for the reacting flow and the co-flow, respectively. The diffusion of the co-flow into the reacting flow was calculated using equation 4, and the rate of diffusion was found to be about 0.4 mol/m 2 s. This diffusion rate is very minimal, and should have a negligible effect on the overall reaction. Upon installation of the N 2 co-flow nozzle, the 20 vol % CH 4 peak was monitored both with and without the co-flow. A plot of this data can be seen in Figure 4-29. As can be seen in the figure the data taken without the co-flow nozzle shows a large fluctuation in the Raman signal intensity as a function of height, with a maximum deviation from average of approximately 5%. For the tests performed with the co-flow nozzle the Raman signal intensity is approximately 30% less than without the co-flow nozzle. Since the Raman signal intensity is directly proportional to the optical path length, this decrease in signal is indicative of a shorter path length for the CH 4 concentration. This means that the nitrogen co-flow is forcing the CH 4 flow to maintain a stagnation profile. This stagnation profile results in less entrainment and recirculation of the CH 4 into the stagnation flow. The data taken with the co-flow nozzle show much less fluctuations in the Raman signal intensity as a function of height, with an overall deviation from average of 1%. This data shows that the co-flow nozzle is functioning as designed and is forcing the reacting gas flow to maintain a stagnation profile while minimizing entrainment and recirculation of exhaust gases in to the reaction zone. 137

PAGE 138

Upon installation of the co-flow nozzle and new best form lens, a new set of Raman spectroscopic data was obtained for both H 2 and synthesis gas burns on both Pd and Pt surfaces. In order to maintain surface reactions for the synthesis gas burn on the Pt catalyst, the cooling N 2 flow rate was decreased to 10 slpm. At 20 slpm, the N 2 cooling flow was removing too much heat from the catalyst to maintain the required surface temperature for the catalytic combustion of synthesis gas. Measurements were obtained from 0 2.0 mm from the surface of the catalyst on 0.5 mm intervals. The nozzle to catalyst distance for all tests was 19.0 mm; therefore the data are plotted as a function of nozzle distance with the area of interest being from 17.0 19.0 mm from the reacting gas nozzle. For the Raman spectroscopic data taken with the modified stagnation reactor conditions, the calibration was carried out in the same manner as before. Three sets of calibration data were taken for each species of interest. In order to maintain uniformity between the calibration data and the reacting data, the calibration data were taken at the same total flow rates as the reacting conditions. In order to account for day-to-day changes with the optical system, a daily CH 4 20% peak was analyzed and ratioed to the calibration curves in order to calculate the aforementioned calibration factor for each set of reactor burn data. With the modified reactor specifications, burn data were taken for both H 2 and synthesis gas burns on both the Pt and Pd catalysts. Figures 4-30, 4-31, and 4-32 show the concentration of H 2 H 2 O, and O 2 respectively as a function of distance from the reacting gas nozzle for the catalytic combustion of H 2 gas on a Pd surface. In Figure 4-30, the data suggest that the concentration of H 2 gas is increasing toward the catalyst surface. This is clearly counterintuitive since the concentration of H 2 should decrease with the combustion of H 2 However, an additional point is noted, namely that the data are in absolute units of mol/l, which is directly influenced by temperature. Hence in the presence of 138

PAGE 139

strong temperature gradients which are certainly present, it is difficult to judge the actual species gradient. Figures 4-31 and 4-32 are more consistent with the expected trends within the boundary layer. As shown in Figure 4-31, the concentration of H 2 O is shown to increase from 0.0011 to 0.0030 mol/l within the boundary layer. Also, Figure 4-32 suggests that O 2 decreases within the boundary layer from 0.0036 to 0.0019 mol/L. It is important to note that the data trends were much more accurate within 1.0 mm of the catalyst surface. This infers that the boundary layer at the prescribed flow conditions is approximately 1.0 mm in thickness. This boundary layer thickness of 1.0 mm was confirmed with a detailed kinetic model combined with a stagnation flow model [48]. The inaccuracies of the data outside this predicted boundary layer are accredited to fluid dynamics phenomena such as turbulent mixing and eddy formation at the edge of the catalyst disk. These trends for H 2 O and O 2 are consistent with the kinetic model, and the values obtained for the concentration are on the order of those predicted by the kinetic model. Concentration profiles for the species of interest for the catalytic combustion of synthesis gas on a Pd surface are shown in Figures 4-33 through 4-37. With the exception of the concentration profile of H 2 shown in Figure 4-33, and the concentration profile of CO 2 shown in Figure 4-35, the concentration profiles for the catalytic combustion of synthesis gas on a Pt surface match the predicted trends. Even though the trends for the concentration of H 2 and CO 2 within the boundary layer do not match the trends predicted by the model, the overall values for the respective concentrations are on the order of those predicted by the model. As shown in Figure 4-32, the concentration of O 2 is shown to decrease from 0.0035 to 0.0020 mol/l within the boundary layer. This decreasing concentration trend and these values are consistent with the kinetic model. Figure 4-36 shows the concentration of H 2 O increasing from 0.0004 to 0.0015 mol/l within the boundary layer. Figure 4-37 shows the concentration of O 2 as decreasing from 139

PAGE 140

0.0031 to 0.0015 mol/l within the boundary layer. Both of these trends agree with the kinetic model, and the values are similar to those predicted by the model. Figures 4-38 through 4-40 show the concentration profiles of the species of interest for the catalytic combustion of H 2 gas on a Pt surface. Although the concentration profiles shown in the Figure s display no definite trends, the concentration values obtained agree reasonably well with the values predicted within the boundary layer by the kinetic model. The same is true with Figures 4-41 through 4-45, which show the concentration profiles of the species of interest for the catalytic combustion of synthesis gas on a Pt surface. The addition of the N 2 co-flow nozzle has been shown to improve the Raman spectroscopic molecular species sampling. Some of the concentration profiles have predicted very well the trends also predicted by the kinetic model. However, the majority of the Raman spectroscopic data showed no obvious trend within the boundary layer. This could possible be due to the strong temperature gradients within the boundary layer affecting absolute molar concentration of the species. The values obtained for the concentrations of the species of interest, however, helped to modify the kinetic model in order to match the order of the predicted concentrations within the boundary layer. There are several reasons which can be given as causing the poor resolution with the Raman spectroscopic data: By removing the exhaust products through the top of the reactor, the N 2 cooling flow must be pulled out the top of the reactor as well. This cooling flow then interacts with both the reacting gas flow and the N 2 co-flow in order to form eddies at the edge of the catalyst surface. Since the method of collection for the Raman spectroscopic data is backscatter, a significant portion of this area is being sampled along with the reacting region above the surface of the catalyst. These fluctuations in the flow on the edge of the catalyst surface then results in a fluctuation in the spectroscopic data. A possible method to resolve this fluctuation in the data would be to exhaust the products through the side of the reactor housing. This would result in maintaining stagnation profiles for the reacting gas flow, the N 2 co-flow, and the N 2 cooling flow. 140

PAGE 141

The method of Raman spectroscopic signal collection can be changed from backscatter so as not to obtain a signal from the turbulent eddies formed on the edge of the catalyst surface. For example, collection of the Raman signal at 90 to the incident laser pulse can allow the Raman signal to only be collected with the reacting region on the catalyst surface. The latter method of altering the collection optics configuration was unsuccessfully tried in the laboratory. The collection configuration was modified to a side scatter collection system. The system was designed to collect the Raman spectroscopic signal from a 1 cm section in the center of the reacting flow. A 1 cm square base cuvet was filled with a Rhodamine 610 Chloride laser dye in order to align the collection system. Upon proper alignment, air flow was established through the reacting flow nozzle in order to obtain an O 2 and N 2 Raman signal. The N 2 signal was just slightly above the detection limit. Since the signal strength was so small, this method of optical collection does not have a high enough sensitivity to detect the species of interest within the reacting gas flow stream without using a laser with a much higher power output. Therefore, it is suggested that the exhaust from the stagnation flow reactor be changed to side exhaust in order to reduce the formation of eddies on the edge of the catalyst surface, and/or a higher power laser be used. 4.2.2 Catalyst surface studies In order to better understand the physical parameters of the catalyst surface, atomic force microscopy (AFM) was used to obtain surface roughness measurements for both the platinum and palladium catalyst disks used in the stagnation point flow reactor. An Asylum Research atomic force microscope coupled with an MFP3D molecular force probe controller was used in AC tapping mode to record surface structure of the catalyst disks. All AFM work was performed by Josh Lowitz in the Surface Science and Nanotribology Laboratory in the Materials Science and Engineering Department. 141

PAGE 142

Three catalyst disks were scanned to obtain information on the surface structure. Each disk was scanned at two locations: the center of catalyst was scanned to obtain surface characteristics in the reacting region of the disk, and a point located halfway between the reacting zone and the disk edge was also scanned so as to obtain surface information from a non-reacting region on the surface of the catalyst. One platinum disk in good condition with approximately 20 hours burn time was scanned, as well as one palladium disk in good condition with approximately 10 hours burn time and one palladium disk which had gone through a high temperature cycle with approximately 10 hours burn time. Scans were performed for 625 m 2 area sections. Each scan area was comprised of a 512 x 512 pixel matrix. Igor Pro 5.04B data processing software was used to construct three dimensional surface plots of the catalyst surfaces. Igor Pro software was also used to calculate the RMS surface roughness of each catalyst as well as the effective surface area of the section. Figure 4-46 shows a three dimensional view of the catalyst surface for the non-reacting section of the platinum catalyst. The RMS roughness for the non-reacting section of the platinum catalyst is 362.7 nm, and the effective surface area of the section is 829.4 m 2 which yielded an actual-to-predicted surface area ratio of 1.327. Figure 4-47 shows a three dimensional view of the catalyst surface for the reacting section of the platinum catalyst. The RMS roughness for the reacting section of the platinum catalyst is 425.6 nm, and the effective surface area of the section is 889.9 m 2 which yielded an actual-to-predicted surface area ratio of 1.424. The RMS roughness and effective surface area of the reacting section for the platinum catalyst surface is 17.3% and 7.3% greater, respectively than for the non-reacting section, which suggests that undergoing reaction conditions results in roughening of the platinum catalyst surface. However, it is difficult to make clear statements given the rather small sample space. 142

PAGE 143

Figure 4-48 shows a three dimensional view of the catalyst surface for the non-reacting section of the overheated palladium catalyst. The RMS roughness for the non-reacting section of the overheated palladium catalyst is 529.0 nm, and the effective surface area of the section is 1025 m 2 which yielded an actual-to-predicted surface area ratio of 1.640. Figure 4-49 shows a three dimensional view of the catalyst surface for the reacting section of the overheated palladium catalyst. The RMS roughness for the reacting section of the overheated palladium catalyst is 917.1 nm, and the effective surface area of the section is 1112 m 2 which yielded an actual-to-predicted surface area ratio of 1.779. The RMS roughness and effective surface area of the reacting section for the overheated palladium catalyst surface is 73.4% and 8.5% greater, respectively than for the non-reacting section, which suggests that undergoing reaction conditions results in roughening of the palladium catalyst surface. This is the same trend observed for the platinum catalyst, however the RMS roughness increases much more drastically in the reacting zone for the overheated palladium catalyst surface than for the platinum catalyst surface. In general, the overheated palladium surface is much rougher than the platinum surface. Figure 4-50 shows a three dimensional view of the catalyst surface for the non-reacting section of the palladium catalyst. The RMS roughness for the non-reacting section of the palladium catalyst is 897.8 nm, and the effective surface area of the section is 990.8 m 2 which yielded an actual-to-predicted surface area ratio of 1.585. Figure 4-51 shows a three dimensional view of the catalyst surface for the reacting section of the overheated palladium catalyst. The RMS roughness for the reacting section of the palladium catalyst is 1033 nm, and the effective surface area of the section is 968.6 m 2 which yielded an actual-to-predicted surface area ratio of 1.550. The RMS roughness of the reacting section for the palladium catalyst surface is 15.1% greater than for the non-reacting section. However, the the surface area of the 143

PAGE 144

reacting section is 2.2% less than the non-reacting section. This means that the average surface elevation fluctuations are greater for the reacting section, while the surface area remains relatively constant. It is important to note that the RMS roughness for the reacting section of the overheated palladium catalyst is 11.0% less than the RMS roughness for the reacting section of the regular normal palladium catalyst. However, the surface area for the reacting section of the overheated palladium catalyst is 14.8% greater than the surface area for the reacting section of the regular palladium catalyst. It is important to note that these effective surface area ratios are extremely high for all cases. A more detailed treatment of effective area and alternative values is presente in Appendix B. 4.3 Summary of Work This section will discuss the major conclusions of the work contained within this dissertation: Molecular sampling of the gaseous species via Raman spectroscopy was performed for the catalytic combustion of CH 4 and syngas. Measurements were taken at two points along the length of the reactor in order to validate a combustion model written for the catalytic combustion of CH 4 and syngas. Although some of the species (CH 4 H 2 etc) could be monitored, the majority of the species could not be monitored due to low sensitivity and signal interference from the optical access ports. Heat transfer studies for the catalytic combustion of CH 4 and CO in the tubular reactor have shown that catalytic combustion results in an enhancement of the heat transfer from the catalytically active surface regardless of the fuel being combusted. Although the exact mechanism of this enhanced convective heat transfer has not been definitively determined, it is possible that the enhancement is due to the activation energy required to release the products from the surface of the catalyst. Stoichiometric studies have been performed for the catalytic combustion of CH 4 and syngas for the tubular reactor. These studies have shown that as more air is added to the reacting gas feed, while maintaining constant fuel flow rates, that the surface temperatures increase drastically, and therefore the catalyst can release more energy. This study has shown that it is possible to release more energy during the catalytic combustion of CH 4 while still limiting the surface temperatures to less than 1150 K to reduce sintering effects. By releasing more energy in the catalytic reactor it is possible to reduce the amount of NO x formed during homogeneous combustion. Raman spectroscopic studies have been performed for a stagnation point flow configuration system for the catalytic combustion of H 2 and syngas. The system has been modified to 144

PAGE 145

include a nitrogen matching co-flow in order to sustain the stagnation profile of the reacting flow. The stagnation reactor allowed the collection of all of the species of interest as well as the collection of concentration information for H 2 O. Species concentration measurements were obtained at three distinct points within the boundary layer as a function of height from the surface of the catalyst. These concentration measurements were used to write a predictive model for the catalytic combustion of H 2 and syngas. Although the exact concentration gradients of the data do not match the modeled profiles, the model was able to be adjusted to similar values within the boundary layer for the combustion of H 2 Atomic force microscopy studies were performed on the surface of the catalyst in order to quantify the roughness of the catalyst surfaces both inside and outside of the reaction zone. Studies were also performed on a palladium catalyst which was subjected to very high thermal loading in order to study high temperature effects on the surface area of the catalyst. In all cases, the reacting zone had a higher RMS roughness. This was especially true for the case of the overheated palladium catalyst where the RMS roughness was 73% greater compared to 15% for the standard palladium catalyst. The RMS roughness for the platinum catalyst was found to be 17.3% greater within the reaction zone. 4.4 Future Work As discussed previously in this chapter, much investigation is required to fully understand the kinetics and natural phenomena of the catalytic combustion of methane and syngas. Specifically, there are four areas of interest that will be further investigated: The design of a tubular reactor with thermocouple access points located on 1 cm intervals to measure the reacting gas flow temperature on the same intervals as the cooling flow. Detailed calibration and assessment of species detection limits for the new concentric nozzle design in the stagnation flow reactor with the side exhaust configuration. Measurement of full species profiles for the stagnation flow reactor using the modified exhaust collection for comparison with analytical models. Explore non-dimensional groups for the heat transfer from a catalytically active surface such as the Damkohler number. Aging studies of the catalyst. Explore commonality of the enhancement to the convective heat transfer coefficient for the catalytic combustion of CH 4 and CO. Investigate sticking coefficients of CH 4 onto the catalyst surface. It is possible that the values given in the Deutschmann model are several orders of magnitude too high. Explore the possibility of homogeneous combustion in parallel with the catalytic combustion of CH 4 and syngas in the tubular reactor. 145

PAGE 146

Table 4-1. Typical operating temperatures of the tubular reactor for various fuel feeds Fuel Inlet Temperature (K) Downstream Mixing Temperature (K) Optical Port 1 Temperature (K) Optical Port 2 Temperature (K) CH 4 583 767 1040 798 Syngas 571 810 1145 858 Table 4-2. Flow Rates for the stoichiometric studies performed on the tubular catalytic reactor Equivalence Ratio Air Flow Rate Syngas Flow Rate CH 4 Flow Rate 2.6 6.052 5.6 1.6 2.8 5.62 5.6 1.6 3.0 5.245 5.6 1.6 3.2 4.916 5.6 1.6 3.4 4.627 5.6 1.6 3.6 4.37 5.6 1.6 3.8 4.14 5.6 1.6 Table 4-3. Reduced flow rates for the stagnation reactor burn data Fuel Fuel Flow Rate (slpm) Reacting Air Flow Rate (slpm) N 2 Cooling Flow Rate (slpm) N 2 Co-Flow Rate (slpm) H 2 2.31 6.67 10/20 12.8 Syngas 3.13 5.77 10 12.8 Table 4-4. Surface characteristics of the stagnation point flow reactor catalyst samples. Catalyst Reacting RMS Roughness (nm) Reacting Surface Area (m 2 ) Non-Reacting RMS Roughness (nm) Non-Reacting Area (m 2 ) Platinum 425.6 889.9 362.7 829.4 Overheated Palladium 917.1 1112 529.0 1025 Palladium 1033 968.6 897.8 990.8 146

PAGE 147

010020030040050025003000350040004500Intensity (a.u.)Raman Shift (cm-1) Figure 4-1. Typical Raman spectra showing the Raman peak and baseline. 147

PAGE 148

-100010020030040050028502900295030003050Intensity (a.u.)Raman Shift (cm-1)20% CH410% CH4 Figure 4-2. Raman spectra showing CH 4 peaks at 10 vol % and 20 vol %. 148

PAGE 149

0204060801000.0000.0020.0040.0060.0080.010P/B RatioConcentration (mol/l) Figure 4-3. Calibration plot showing calibration points with linear curve fit. Notice the bundle of four points in the center of the plot. These points are the 10 vol % cold calibration points and the 20 vol % hot calibration points. 149

PAGE 150

0.000.050.100.150.200.25-505101520 CH4 Ignition CH4 Steady StateCH4 Mole FractionCatalyst Distance (cm) Figure 4-4. Data points showing the calculated mole fraction of CH 4 at the entrance to the reactor and at the two optical access ports. Results are shown for both ignition test conditions and steady state test conditions for data taken in Spring 2007. Error bars represent one standard deviation. 150

PAGE 151

0.200.400.600.801.0000.20.40.60.81 CH4 Ignition CH4 Steady StateCH4 Mole FractionCatalyst Distance Figure 4-5. Data points showing the calculated mole fraction of CH 4 at the entrance to the reactor and at the two optical access ports non-dimensionalized by the mole fraction of CH 4 at the entrance to the reactor. The catalyst distance is non-dimensionalized by the total length of the catalyst coating. Results are shown for both ignition test conditions and steady state test conditions for data taken in Spring 2007. Error bars represent one standard deviation. 151

PAGE 152

0.000.050.100.150.200.25051015 H2 Ignition H2 Steady StateH2 Mole FractionCatalyst Distance (cm) Figure 4-6. Data points showing the calculated mole fraction of H 2 at the entrance to the reactor and at the two optical access ports. Results are shown for both ignition test conditions and steady state test conditions for the data taken in Spring 2007. Error bars represent one standard deviation. 152

PAGE 153

0.000.501.001.502.0000.20.40.60.81 H2 Ignition H2 Steady StateH2 Mole FractionCatalyst Distance Figure 4-7. Data points showing the calculated mole fraction of H 2 at the entrance to the reactor and at the two optical access ports non-dimensionalized by the mole fraction of H 2 at the second optical port for the steady state case. The catalyst distance is non-dimensionalized by the total length of the catalyst coating. Results are shown for both ignition test conditions and steady state test conditions for the data taken in Spring 2007. Error bars represent one standard deviation. 153

PAGE 154

02040608025003000350040004500Intensity (a.u.)Raman Shift (cm-1)H2WindowInterferenceInterference Figure 4-8. Sample spectra for 406 nm centered window for a synthesis gas burn. As can be seen, the H 2 peak is just below the minimum limit of detection. Also shown in the plot is the window interference peak which doesnt allow the collection of H 2 O data. 154

PAGE 155

74076078080082002468101214 CH4 Cooling Flow Temperature (K) CH4 Interpolated Cooling Flow TemperatureCH4 Cooling Flow Temperature (K)Catalyst Distance Figure 4-9. Cooling flow temperature profile for CH 4 combustion showing the measured points on 1 cm intervals and the interpolated points on 1 mm intervals. 155

PAGE 156

73074075076077078079080081002468101214 CO Cooling Flow Temperature (K) CO Interpolated Cooling Flow Temperature (K)CO Cooling Flow Temperature (K)Catalyst Distance Figure 4-10. Cooling flow temperature profile for CO combustion showing the measured points on 1 cm intervals and the interpolated points on 1 mm intervals. 156

PAGE 157

05 1041 1051.5 105-202468101214q"inner (W/m2)x (cm) Figure 4-11. Calculated heat flux to the inner cooling flow for the catalytic combustion of CH 4 157

PAGE 158

02 1054 1056 1058 105-202468101214 q"inner q"released q"outerq" (W/m2)Distance (cm) Figure 4-12. Total heat flux, heat flux to cooling flow, and heat flux to reacting flow for the case of a single exponential decay total heat release along the length of the reactor for the catalytic combustion of CH 4 158

PAGE 159

-1 10501 1052 1053 1054 1055 1056 105-202468101214 q"classical q"outer q"catalyticq" (W/m2)Distance (cm) Figure 4-13. Total heat flux to the reacting flow, classical calculation of heat flux to the reacting flow, and catalytic heat flux to the reacting flow for the single exponential decay fit for the catalytic combustion of CH 4 159

PAGE 160

-5 10405 1041 1051.5 1052 1052.5 1053 105-202468101214 q" catalytic q" totalq" (W/m2)Distance (cm) Figure 4-14. Total heat flux and catalytically enhanced heat flux for the case of an initial exponential decay total heat release followed by a secondary heat release along the length of the reactor for the catalytic combustion of CH 4 160

PAGE 161

-1 10501 1052 1053 1054 1055 1056 105-202468101214 q"inner q"released q"outerq" (W/m2)Distance (cm) Figure 4-15. Total heat flux, heat flux to cooling flow, and heat flux to reacting flow for the case of a single exponential decay total heat release along the length of the reactor for the catalytic combustion of CO. 161

PAGE 162

-1 10501 1052 1053 1054 1055 105-202468101214 q"classical q"outer q"catalyticq" (W/m2)Distance (cm) Figure 4-16. Total heat flux to the reacting flow, classical calculation of heat flux to the reacting flow, and catalytic heat flux to the reacting flow for the single exponential decay fit for the catalytic combustion of CO. 162

PAGE 163

01 1052 1053 1054 1055 10502468 CH4 Catalytic Enhancement CO Catalytic EnhancementCatalytic Enhancement (W/m2)Distance (cm) Figure 4-17. Enhancement to the convective heat transfer to the reacting flow due to catalytic surface reactions for both CH 4 and CO. 163

PAGE 164

05 1051 1061.5 1062 1062.5 106-10123456 Modeled Enhancement Calculated EnhancementEnhancement (W/m2)Catalyst Distance (cm) Figure 4-18. Modeled and calculated enhancement to the convective heat transfer from a catalytically active surface during the combustion of CH 4 164

PAGE 165

5006007008009001000110012002.533.54 T1 T2 TP1 TP2Temperature (K)Equivalence Ratio StandardOperatingConditions Figure 4-19. Tubular operating temperatures during the catalytic combustion of CH 4 as a function of equivalence ratio. T 1 is the reacting flow inlet temperature, T 2 is the downstream mixing temperature, T P1 is the surface temperature at the first optical port, and T P2 is the surface temperature at the second optical port. The line indicates the standard operating conditions used during the Raman spectroscopic studies. The upper flammability limit of CH 4 is at an equivalence ratio of 1.64. 165

PAGE 166

0501001502002500501001502002502.533.54 T2-T1 (K) Heat Release (W)T2-T1 (K)Heat Release (W)Equivalence Ratio Figure 4-20. Total heat release as calculated via complete rich oxidation of CH 4 and measured temperature rise as a function of equivalence ratio. 166

PAGE 167

5006007008009001000110012002.833.23.43.63.84 T1 T2 TP1 TP2Temperature (K)Equivalence Ratio StandardOperatingConditions Figure 4-21. Tubular operating temperatures during the catalytic combustion of syngas as a function of equivalence ratio. T 1 is the reacting flow inlet temperature, T 2 is the downstream mixing temperature, T P1 is the surface temperature at the first optical port, and T P2 is the surface temperature at the second optical port. The line indicates the standard operating conditions used during the Raman spectroscopic studies. 167

PAGE 168

-4-202468101225003000350040004500Intensity (a.u.)Raman Shift (cm-1)H2O Figure 4-22. Sample H 2 O calibration spectra for the stagnation reactor showing the H2O peak and no interference from the windows. 168

PAGE 169

020040060080010001200140050010001500200025003000Intensity (a.u.)Raman Shift (cm-1)N2InterferenceO2 Figure 4-23. Raman spectra for pure nitrogen flow showing interference line from 532 nm residual ghost lines and no signal from O 2 in the ambient atmosphere. 169

PAGE 170

02004006008001000120050010001500200025003000Intensity (a.u.)Raman Shift (cm-1)N2O2Interference Figure 4-24. Raman spectra for pure nitrogen flow showing interference line from 532 nm residual ghost lines and O 2 signal resulting from the flow of air through the reacting nozzle. This Figure along with Figure 4-10 show that it is possible to monitor the O 2 signal in the stagnation reactor. 170

PAGE 171

050010001500200050010001500200025003000W/O FilterRaman Shift (cm-1)InterferenceInterferenceN2CO2CO2Interference Figure 4-25. Raman spectra for CO2 calibration flow with and without the 532 nm low pass filter installed in front of the collection optics. Notice how the filter removes the interference lines resulting from the 532 nm residual laser light. 171

PAGE 172

02040608025003000350040004500Intensity (a.u.)Raman Shift (cm-1)H2OH2 Figure 4-26. Sample spectra centered at 406 nm for the steady stat burn of H2. The spectra shows the Raman scattering peaks for H2O and H2. 172

PAGE 173

0.00220.00240.00260.00280.00300.00320.00340.0036012345 H2 Concentration (mol/l) O2 Concentration (mol/l)Concentration (mol/l)Distance (mm) Figure 4-27. Concentration of H2 and O2 as a function of vertical distance from the catalyst surface during the steady state combustion of H2. 173

PAGE 174

0.00100.00150.00200.00250.00300.00350.00400.0045012345 CO CO2 O2 H2Concentration (mol/l)Distance (mm) Figure 4-28. Concentration of CO, CO2, H2, and O2 as a function of vertical distance from the catalyst surface during the steady state combustion of synthesis gas. 174

PAGE 175

6000650070007500800085009000950001234 Coflow No Coflow Intensity (a.u.)Height (mm) Figure 4-29. Co-flow nozzle test showing Raman signal intensities of the 20 vol % CH4 signa as a function of height from the catalyst disk. 175

PAGE 176

0.00100.010016.517.017.518.018.519.019.5 H2 Model H2H2 (mol/l)Height (mm) Figure 4-30. Concentration of H 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of H 2 gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 176

PAGE 177

0.0000000010.0000000100.0000001000.0000010000.0000100000.0001000000.0010000000.01000000016.517.017.518.018.519.019.5 H2O Model H2OH2O (mol/l)Nozzle Distance (mm) Figure 4-31. Concentration of H 2 O as a function of vertical distance from the reacting gas nozzle during the steady state combustion of H 2 gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 177

PAGE 178

0.00010.00100.010016.517.017.518.018.519.019.5 O2 Model O2O2 (mol/l)Nozzle Distance (mm) Figure 4-32. Concentration of O 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of H 2 gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 178

PAGE 179

0.00000.00010.00100.010016.517.017.518.018.519.019.5 H2 Model H2H2 (mol/l)Nozzle Distance (mm) Figure 4-33. Concentration of H 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 179

PAGE 180

0.00010.00100.010016.517.017.518.018.519.019.5 CO Model COCO (mol/l)Nozzle Distance (mm) Figure 4-34. Concentration of CO as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 180

PAGE 181

0.00000.00010.00100.010016.517.017.518.018.519.019.5 CO2 Model CO2CO2 (mol/l)Nozzle Distance (mm) Figure 4-35. Concentration of CO 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 181

PAGE 182

0.00040.00060.00080.00100.003016.517.017.518.018.519.019.5 H2O Model H2OH2O (mol/l)Nozzle Distance (mm) Figure 4-36. Concentration of H 2 O as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 182

PAGE 183

0.00000.00010.00100.010016.517.017.518.018.519.019.5 O2 Model O2O2 (mol/l)Nozzle Distance (mm) Figure 4-37. Concentration of O 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pd surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 183

PAGE 184

0.0010.0116.51717.51818.51919.5 Model H2 H2 (mol/l)H2 (mol/l)Nozzle Distance (mm) Figure 4-38. Concentration of H 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of H 2 gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 184

PAGE 185

10-710-610-50.00010.0010.0116.51717.51818.51919.5 H2O (mol/l) Model H2OH2O (mol/l)Nozzle Distance (mm) Figure 4-39. Concentration of H 2 O as a function of vertical distance from the reacting gas nozzle during the steady state combustion of H 2 gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 185

PAGE 186

0.0010.0116.51717.51818.51919.5 O2 (mol/l) Model O2O2 (mol/l)Nozzle Distance (mm) Figure 4-40. Concentration of O 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of H 2 gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 186

PAGE 187

0.00010.0010.0116.51717.51818.51919.5 CO (mol/l) Model COCO (mol/l)Nozzle Distance (mm) Figure 4-41. Concentration of CO as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 187

PAGE 188

0.00010.0010.0116.51717.51818.51919.5 CO2 (mol/l) Model CO2CO2 (mol/l)Nozzle Distance (mm) Figure 4-42. Concentration of CO 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 188

PAGE 189

10-50.00010.0010.0116.51717.51818.51919.5 H2 (mol/l) Model H2H2 (mol/l)Nozzle Distance (mm) Figure 4-43. Concentration of H 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 189

PAGE 190

0.00010.0010.0116.51717.51818.51919.5 H2O (mol/l) Model H2OH2O (mol/l)Nozzle Distance (mm) Figure 4-44. Concentration of H 2 O as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 190

PAGE 191

10-50.00010.0010.0116.51717.51818.51919.5 O2 (mol/l) Model O2O2 (mol/l)Nozzle Distance (mm) Figure 4-45. Concentration of O 2 as a function of vertical distance from the reacting gas nozzle during the steady state combustion of synthesis gas on a Pt surface. Catalyst surface is at 19.1 mm, nozzle exit is at a distance of 0 mm. 191

PAGE 192

Figure 4-46. Surface structure image for the non-reacting platinum surface. Figure 4-47. Surface structure image for the reacting platinum surface. 192

PAGE 193

Figure 4-48. Surface structure image for the non-reacting palladium surface of the overheated catalyst. Figure 4-49. Surface structure image for the reacting palladium surface of the overheated catalyst. 193

PAGE 194

Figure 4-50. Surface structure image for the non-reacting palladium surface. Figure 4-51. Surface structure image for the reacting palladium surface. 194

PAGE 195

APPENDIX A A REACTOR PART CAD DRAWINGS This appendix includes various cad drawings for parts machined for both the tubular reactor and the stagnation point flow reactor. Figure A-1.Window Holder Schematic. 195

PAGE 196

Figure A-2. Reacting gas nozzle holder. 196

PAGE 197

Figure A-3. Catalyst Lifting Plate. 197

PAGE 198

Figure A-4. Catalyst cooling nozzle plate. 198

PAGE 199

Figure A-5. Catalyst holder. B CATALYST SURFACE STUDIES This section contains information regarding the catalyst surface analysis performed on the measurements obtained through atomic force microscopy. The analysis was performed to calculate the effective surface area due to the rough surface. Two methods were used for this analysis. The first method calculates the minimum surface area and is calculated using Herons formula. Herons formula makes use of each pixel and the heights of each of its eight neighboring pixels, as shown in Figure B-1. The length, of the sides of each triangle, is then calculated using Herons formula as shown in Figure B-2. Where the length of the sides are calculated by: 199

PAGE 200

22212121()()(axxyyzz 2) (B-1) A number s is then defined by: 2abcs (B-2) The surface area can then be calculated as: ..()()()SAssasbsc (B-3) This surface area calculation is repeated for each pixel in the 512x512 matrix. Table B-1 shows the effective surface area ratios obtained from using Herons Formula. The second method use to calculate the effective surface area of the catalyst disk is described in Figure B-3. For this method, each pixel is considered a flat, level surface. The flat surface area of the top surface is added to the surface area of each exposed side of the pixel. The surface area is calculated via this equation: if H(i,j)>H(i+1,j) (B-4) ..((,)(1,))SAdxdyHijHijdy if (H(i,j)>H(i,j-1) ((,)(,1)HijHijdx if H(i,j)>H(i-1,j) ((,)(1,))HijHijdy if H(i,j)>H(i,j+1) ((,)(,1))HijHijdx This method of analysis calculates the maximum effective surface area due to assuming each pixel containing a perfectly flat top. The effective surface area ratios obtained via this calculation are shown in Table B-1. 200

PAGE 201

Table B-1: Values of effective surface area ratio for the reacting and non-reaction zones on Pt and Pd catalysts. Catalyst Projected Area (m 2 ) Minimum Surface Area (m 2 ) Minimum Surface Area Ratio Maximum Surface Area (m 2 ) Maximum Surface Area Ratio Non-reacting Pd 625 1084.6 1.735 1562.7 2.500 Reacting Pd 625 1098.6 1.757 1611.0 2.578 Non-reacting Pt 625 1034.4 1.655 1506.9 2.411 Reacting Pt 625 932.3 1.492 1398.6 2.378 H ( i+1 ,j +1 ) H ( i ,j +1 ) H ( i-1 ,j +1 ) H ( i ,j) H ( i+1 ,j) H ( i-1 ,j) H ( i+1 ,j -1 ) H ( i-1 ,j -1 ) H ( i ,j -1 ) Figure B-1: Nodal points for calculation of the minimum effective surface area. 201

PAGE 202

(x1, y1, z1) b a (x3, y3, z3) c (x2, y2, z2) Figure B-2: Herons formula for calculation of triangle side length. Figure B-3: Method of analysis to calculate the maximum effective surface area. Figure B-3: Method of analysis to calculate the maximum effective surface area. 202

PAGE 203

LIST OF REFERENCES [1] Glin P, Urfels L, Primet M, Tena E (2003). Complete oxidation of methane at low temperature over Pt and Pd catalysts for the abatement of lean-burn natural gas fuelled vehicles emissions: influence of water and sulphur containing compounds. Catalysis Today 83, 45-57. [2] Dalla Betta R, Schlatter J, Yee D, Loffler D, Shoji T (1995). Catalytic combustion technology to achieve ultra low NO x emissions: Catalyst design and performance characteristics. Catalysis Today 26, 329-335. [3] Seo Y, Cho S, Kang S, Shin H (2000). Experimental and numerical studies on combustion characteristics of a catalytically stabilized combustor. Catalysis Today 59, 75-86. [4] Simplcio LMT, Brando ST, Sales EA, Lietti L, Bozon-Verduraz FB (2006). Methane combustion over PdO-alumina catalysts: The effect of palladium precursors. Applied Catalysis B: Environmental 63, 9-14. [5] Geus J, van Giezen J (1999). Monoliths in catalytic oxidation. Catalysis Today 47, 169180. [6] Dupont V, Zhang S, Williams A (2001). Experiments and simulation of methane oxidation on a platinum surface. Chemical Engineering Science 56, 2659-2670. [7] Pieck C, Vera C, Peirotti E, Yori J (2002). Effect of water vapor on the activity of PtPd/Al 2 O 3 catalysts for methane combustion. Applied Catalysis 226, 281-291. [8] Persson K, Ersson A, Carrera A., Jayasuriya J., Fakhrai R., Fransson T., Jrs S (2005). Supported palladium-platinum catalyst for methane combustion at high pressure. Catalysis Today 100, 479-483. [9] Persson K, Ersson A, Jansson K, Iverlund N, Jrs S (2005). Influence of co-metals on bimetallic palladium catalysts for methane combustion. Journal of Catalysis 231, 139150. [10] Persson K, Ersson A, Colussi S, Trovarelli A, Jrs S (2006). Catalytic combustion of methane over bimetallic Pd-Pt catalysts: The influence of support materials. Applied Catalysis B: Environmental 66, 175-185. [11] Ozawa Y, Tochihara Y, Watanabe A, Nagai M, Omi S (2004). Deactivation of PtPdO/Al 2 O 3 in catalytic combustion of methane. Applied Catalysis A: General 259, 1-7. [12] Lee J, Trimm D (1995). Catalytic combustion of methane. Fuel Processing Technology 42, 339-359. 203

PAGE 204

[13] Demoulin O, Navez M, Ruiz P (2005). Investigation of the behaviour of a Pd/-Al 2 O 3 catalyst during methane combustion reaction using in situ DRIFT spectroscopy. Applied Catalysis A: General 295, 59-70. [14] Demoulin O, Navez M, Ruiz P (2006). Origin of transient species present on the surface of PdO/-Al 2 O 3 catalyst during the methane combustion reaction. Catalysis Today 112, 153-156. [15] Carrazn S, Mateos R, Rives V, Ruiz P (2006). Influence of the solid state properties of Pd/MO x (M=Ti, Al) catalysts in catalytic combustion of methane. Catalysis Today 112, 161-164. [16] Carstens J, Su S, Bell A (1998). Factors affecting the catalytic activity of Pd/ZrO 2 for the combustion of methane. Journal of Catalysis 176, 136-142. [17] Ozawa Y, Tochihara Y, Watanabe A, Nagai M, Omi S (2004). Stabilizing effect of Nd 2 O 3 La 2 O 3 and ZrO 2 on Pt-Pd/Al 2 O 3 during catalytic combustion of methane. Applied Catalysis A: General 258, 261-267. [18] Cho S, Kang S (2004). Structural transformation of PdPt nanoparticles probed with Xray absorption near edge structure. Catalysis Today 93-95, 561-566. [19] Kraikul N, Jitkarnka S, Luengnaruemitchai A (2005). Catalytic methane combustion on Pd-Pt-La catalysts and their surface models. Apllied Catalysis B: Environmental 58, 143-152. [20] Hayes R, Kolaczkowski S (1999). A study of Nusselt and Sherwood numbers in a monolith reactor. Catalysis Today 47, 295-303. [21] Itaya Y, Hiroi T, Matsuda K, Hasatani M (2002). Convective heat transfer of exothermic reactive gas flowing across a heating cylinder coated with platinum catalyst in a narrow duct. International Journal of Heat and Mass Transfer 45, 1815-1821. [22] Kolaczkowski S, Thomas W, Titiloye J, Worth D (1996). Catalytic combustion of methane in a monolith reactor: heat and mass transfer under laminar flow and pseudo-steady-state reaction conditions. Combustion Science and Technology 118, 79-100. [23] Reinke M, Mantzaras J, Bombach R, Schenker S, Inauen A (2005). Gas phase chemistry in catalytic combustion of methane/air mixtures over platinum at pressures of 1 to 16 bar. Combustion and Flame 121, 448-468. [24] Reinke M, Matzaras J, Schaeren R, Bombach R, Inauen A, Schenker S (2005). Homogeneous ignition of CH 4 /air and H 2 O and CO 2 -diluted CH 4 /O 2 mixtures over Pt; an experimental and numerical investigation at pressures up to 16 bar. Proceedings of the combustion institute 30, 2519-2527. 204

PAGE 205

[25] Reinke M, Matzaras J, Schaeren R, Bombach R, Inauen A, Schenker S (2004). Highpressure catalytic combustion of methane over platinum: In situ experiments and detailed numerical predictions. Combustion and Flame 136, 217-240. [26] Carroni R, Griffing T, Mantzaras J, Reinke M (2003). High-pressure experiments and modeling of methane/air catalytic combustion for power-generation applications. Catalysis Today 83, 157-170. [27] Kuper W, Blaauw M, van der Berg F, Graaf G (1999). Catalytic combustion concept for gas turbines. Catalysis Today 47, 377-389. [28] Sinev M, Tulenin Y, Kalashnikova O, Bychkov V, Korchak V (1996). Oxidation of methane in a wide range of pressures and effect of inert gases. Catalysis Today 32, 157162. [29] Kikuchi R, Maeda S, Sasaki K, Wennerstrom S, Eguchi K (2002). Low-temperature methane oxidation over oxide-supported Pd catalysts: inhibitory effect of water vapor. Applied Catalysis A: General 232, 23-28. [30] Jones J, Dupont V, Brydson R, Fullerton D, Nasri N, Ross A, Westwood A (2003). Sulphur poisoning and regeneration of precious metal catalysed methane combustion. Catalysis Today 81, 589-601. [31] Dupont V, Jones J, Zhang S, Westwood A, Twigg M (2004). Kinetics of methane oxidation on Pt catalysts in the presence of H 2 S and SO 2 Chemical Engineering Science 59, 17-29. [32] Moallemi F, Batley G, Dupont V, Foster T, Pourkashanian M, Williams A (1999). Chemical modeling and measurements of the catalytic combustion of CH 4 /air mixtures on platinum and palladium catalysts. Catalysis Today 235-244. [33] Goralski C, Schmidt L (1999). Modeling heterogeneous and homogeneous reactions in the high-temperature catalytic combustion of methane. Chemical Engineering Science 54, 5791-5807. [34] Deutschmann O, Schmidt R, Behrendt F, Warnatz J (1996). Proceedings of the 26 th International Symposium on Combustion, The Combustion Institute, Pittsburgh, PA. 1747-1754. [35] Groppi G, Ibashi W, Valentini M, Forzatti P (2001). High-temperature combustion of CH 4 over PdO/Al 2 O 3 : kinetic measurements in a structured annular reactor. Chemical Engineering Science 56, 831-839. [36] Sinev, M (2003). Free radicals in catalytic oxidation of light alkanes: kinetic and thermochemical aspects. Journal of Catalysis 216, 468-476. 205

PAGE 206

[37] Hayes R, Kolaczkowski S, Li P, Awdry S (2001). The palladium catalysed oxidation of methane: reaction kinetics and the effect of diffusion barriers. Chemical Engineering Science 56, 4815-4835. [38] Demoulin O, Dury F, Navez M, Gaigneaux E, Ruiz P (2004). Modification of active catalytic sites with N 2 O and CO 2 as gas promoters during oxidizing reactions. Catalysis Today 91-92, 27-31. [39] Demoulin O, Seunier I, Dury F, Navez M, Rachwalik R, Sulikowski B, GonzalezCarrazan S, Gaigneaux E, Ruiz P (2005). Modulation of selective sites by introduction of N 2 O, CO 2 and H 2 as gaseous promoters into the feed during oxidizing reactions. Catalysis Today 99, 217-226. [40] Demoulin O, Seunier I, Navez M, Ruiz P (2006). Influence of the addition of H2 upon the behavior and properties of a Pd (2wt%)/-Al 2 O 3 catalyst and a comparison with the case of the Pt-based catalyst. Applied Catalysis A: General 300, 41-49. [41] Deutschmann O, Maier L, Riedel U, Stroemman A, Dibble R (2000). Hydrogen assisted catalytic combustion of methane on platinum. Catalysis Today 59, 141-150. [42] Stambler I, Gas Turbine World, March (1993) p. 23. [43] Incropera F, DeWitt D (2002). Fundamentals of Heat and Mass Transfer 5 th Edition. [44] Hahn D (2007). Raman Scattering Theory. [45] Valden M, Hirsimaki M, Suhonen S, Pessa M (1997). Activated adsorption of methane on clean and oxygen-modified Pt{111} and Pd {110}. Surface Science 377-379, 605-609. [46] Gnielinski V (1976). Introduction to Chemical Engineering, 16, 359. [47] Kays W, Perkins H, Rohsenow W, Hartnett J (1972). Handbook of Heat Transfer, Chapter 7, McGraw Hill, New York. [48] Private communication with Professor David Mikolaitis. [49] Forzatti P, Groppi G (1999). Catalytic combustion for the production of energy. Catalysis Today 54, 165-190. [50] Ikeda H, Sato J, Williams F (1995). Surface kinetics for catalytic combustion of hydrogen-air mixtures on platinum at atmospheric pressure in stagnation flows. Surface Science 326, 11-26. 206

PAGE 207

[51] Kolaczkowski S, Thomas W, Titiloye J, Worth D (1996). Catalytic combustion of methane in a monolith reactor: heat and mass transfer under laminar flow and pseudosteady-state reaction conditions. Combustion science and technology. 118, 79-100. [52] Glin P, Primet M (2002). Complete Oxidation of methane at low temperature over noble metal based catalysts: a review. Applied Catalysis B: Environmental 39, 1-37. [53] Thorne A, Litzen U, Johansson S (1999). Spectrophysics: Principles and Applications. [54] Lowe P, Ficke J (1974). The computation of saturation vapor pressure. Technical Paper No. 474, Environmental Prediction Research Facility, Naval Postgraduate School, Monterey, CA p. 27. [55] Petukhov b, Irvine T, Hartnett J (1970). Advances in Heat Transfer, Volume 6, Academic Press, New York. 207

PAGE 208

BIOGRAPHICAL SKETCH Cary Henry was born in New Orleans, Louisiana to Craig and Mary Henry. Cary graduated from Gulf Breeze High School in Gulf Breeze, Florida in May 2001. From there he began his undergraduate studies at the Florida State University in Tallahassee, Florida. Upon completion of his B.S. in mechanical engineering, he entered the University of Florida in August 2004 under the direction of Professor David Hahn. This work is the culmination of his research for a Ph.D. in mechanical engineering. 208