MODELING OF CATALYTIC CHANNELS AND MONOLITH REACTORS by PETER M. STRUK Submitted in partial fulfillment of the requirements For the degree of Doctor of Philosophy Dissertation Adviser: Dr. T’ien Department of Mechanical & Aerospace Engineering CASE WESTERN RESERVE UNIVERSITY January 2007 UMI Number: 3237867 UMI Microform 3237867 Copyright 2007 by ProQuest Information and Learning Company. All rights reserved.
This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P. Box 1346 Ann Arbor, MI 48106-1346 ii DEDICATION This dissertation is dedicated first and foremost to my family who patiently waited for me over many long days and nights as I completed this work. Also, this work is dedicated to my parents who provided me the opportunity to pursue higher education.
Finally, this dedication extends to my many friends and colleagues from NASA, CWRU, and the NCSER who continually provided guidance and inspiration for this endeavor. iii TABLE OF CONTENTS DEDICATION. iii TABLE OF CONTENTS. iv LIST OF TABLES.
vi LIST OF FIGURES .2 Lumped versus Distributed Models.3 Gas-Phase Quasi-Steadiness.5 Solid-Phase Heat Transfer .2 Spatial Integration in x .3 Integration in Time .1 Case 1: Steady-State Comparisons – Isothermal Platinum Tube .4 Discussion of Case 1 .2 Case 2: Steady-State Comparisons – Monolith .4 Discussion of Case 2 .3 Case 3: Transient Propagation – Single Horizontal Platinum Tube .4 Discussion of Case 3 .1 Physical Model Improvement .2 Experimental Channel Configuration Improvements .3 Further Recommended Studies.4 Potential Improvements to the Performance of the Computer Program. DERIVATION OF GOVERNING EQUATIONS. TIMESCALES & NON-DIMENSIONAL EQUATIONS. THERMOPHYSICAL AND TRANSPORT PROPERTIES.
MONOLITH UPSTREAM HEAT TRANSFER. IR TEMPERATURE ANALYSIS OF PLATINUM TUBE. 149 v LIST OF TABLES Table 1. Absolute (ATOL) and relative (RTOL) error tolerances used in the computations.
The stable and radical species are defined in Appendix E. Dry CO / O2 sub-mechanism on platinum from the work of Deutschmann et al. Summary of experimental configurations modeled in this work. Values denoted with an asterisk (*) were not explicitly stated in the reference but were assumed.
The ambient pressure surrounding the channel for case 3 (denoted by **) was slightly less-than 1 (≈0. Calculated Sherwood numbers for each species using the analogy of heat and mass transfer with Nu=4. The property values are based on the inlet mixture (3% CO by volume in air with saturated water vapor) at 850K. Calculated Reynolds numbers evaluated at the temperature extremes for the catalytic channel experiments of Khitrin and Solovyeva.
Gas-phase timescales and Peclet numbers for various processes of the catalytic channel using the property values shown at the top. Solid-phase timescales and Peclet numbers for various processes of the catalytic channel using the property values shown at the top. Volumetric ratios of the gas to solid heat capacities for all 3 cases presented in the text. Parameter used to gauge the importance of solid axial heat conduction to lateral heat transfer from the gas to the solid.
Thermophysical property values used in the calculations. Homogeneous CO gas-phase mechanism [84]. The reactions used in the dry CO mechanism are shown in red. All reactions are used in the wet CO mechanism.
CH4 / O2 mechanism on platinum [21]. The reactions shown in red are used in the dry CO calculation. All reactions are used in the wet CO mechanism.143 vi LIST OF FIGURES Figure 1. Flowchart of basic solution algorithm.
Discretization of catalytic channel into finite volumes. Solid phase nodes (shown in black) represent the entire cross-sectional volume. The inlet is at x=0. Comparison of steady-state model results (including a plug-flow model) to the experiment of Khitrin and Solvyeva[54] for 3 channel velocities.
The inlet gas consisted of 3% CO (by volume) with the balance being air. Computed steady-state profiles of CO mass fraction and select surface species along the length of the platinum channel for the 34 m/s case (Nu = 4. The initial condition for the surface is ZO(s) = 1. Computed results showing the effect of varying the mass-transfer effectiveness.
The Sherwood number is related to the Nusselt number via the heat and mass-transfer analogy. The surface initial condition for these calculations is ZO(s) = 1. Wall mass fraction, YkW (for CO and O2), and surface site fractions, Zk (for CO(s) and O(s)), at the channel outlet versus Nu for 560K. Also shown is the product ZCO(s) ⋅ ZO(s), which is proportional to the fuel conversion rate.
Wall mass fraction, YkW (for CO and O2), and surface site fractions, Zk (for CO(s) and O(s)), at the channel outlet versus Nu for 600K. Also shown is the product ZCO(s) ⋅ ZO(s), which is proportional to the fuel conversion rate. Wall equivalence ratio, φW, at the channel outlet versus Nu for four temperatures: 560K, 580K, 600K, and 620K. The dashed line corresponds to the approximate φW just prior to light- off (which would occur with any further temperature increase or Nu decrease).
Comparison of different values of Γ (surface site density) on the steady-state model results. The value of Γ0 =2.71 x 10-9 mol/cm2 was estimated from the density of platinum[19, 63]. The initial surface condition was ZO(s)=1. Comparison of different values of Γ (surface site density) on the steady-state model results.
The value of Γ0 =2.71 x 10-9 mol/cm2 was estimated from the density of platinum[19, 63]. The initial surface condition was ZCO(s)=1. Global sensitivity study of the dry CO mechanism conducted by changing the pre- exponential constant or sticking coefficient by 25% and 175% relative to the published value. The data corresponds to 34 m/s and the black lines are the unmodified kinetics.
In these calculations, the surface is initially covered by CO(s). Global sensitivity study of the dry CO mechanism conducted by changing the pre- exponential constant or sticking coefficient by 25% and 175% relative to the published value. The data corresponds to 34 m/s and the black lines are the unmodified kinetics. In these calculations, the surface is initially covered by O(s).
Computed results showing the effect of saturated water vapor (using a wet CO mechanism) in the inlet feed on the steady-state conversion. Schematic representation of monolith configuration tested in this section. The model represents the center channel of the monolith and is used to characterize the entire monolith performance. The left image is from Kee et al.
Measured CO concentration at the outlet of a commercial monolith from the work of Ullah et al. The inlet velocity for an individual channel was 6.23 m/s at a temperature of 623K. The right axis shows the conversion calculated from the concentration measurements. Computed results showing the effect of varying a* on the conversion of CO as a function of reactor length.
The experimental data is from Ullah et al. The top graph shows the conversion while the bottom graph shows the corresponding solid temperatures. Temperature and select species mass and site fraction as a function of monolith length up to 4cm (the calculation used 12 cm). The model parameters are a*=10 and Nu=4.
Temperature and select species mass and site fraction as a function of monolith length up to 4cm (the calculation used 12 cm). The model parameters are a*=30 and Nu=4. Effect of Nu / Sh on the steady-state conversion of CO as a function of reactor length. The Sherwood number comes from the analogy of heat and mass transfer.
The computations use a*=30 and correspond to the mass-transfer limited or high-conversion solution. Ignition and propagation sequence of a CO / O2 (φ = 1) catalytic reaction along the inside of a platinum tube (0. The inlet gas velocity is 2 m/s. Leading edge of catalytic reaction versus time in a platinum tube for φ = 0.1 to 1 with a dry CO / O2 mixture flowing at 2 m/s.
A hot-wire heats the outlet end of the tube for 3 seconds (denoted by the red vertical line). There are up to 3 tests plotted for each φ. The tube outlet is at 3. Leading edge of catalytic reaction versus time in a platinum tube for φ = 1.1 to 2 with a dry CO / O2 mixture flowing at 2 m/s.
A hot-wire heats the outlet end of the tube for 3 seconds (denoted by the red vertical line). There are up to 3 tests plotted for each φ. The tube outlet is at 3. Leading edge of catalytic reaction versus time in a platinum tube for φ = 0.1 to 1 with a wet CO / O2 mixture flowing at 2 m/s.
A hot-wire heats the outlet end of the tube for 3 seconds (denoted by the red vertical line). There are up to 3 tests plotted for each φ. The tube outlet is at 3. Leading edge of catalytic reaction versus time in a platinum tube for φ = 1.1 to 2 with a wet CO / O2 mixture flowing at 2 m/s.
A hot-wire heats the outlet end of the tube for 3 seconds (denoted by the red vertical line). There are up to 3 tests plotted for each φ. The tube outlet is at 3. Measurements and model predictions of the propagation velocity for the catalytic reaction front versus φ for both dry and wet CO tests.
Comparison of model and experiment looking at the propagation of a catalytic reaction front along a Pt tube for φ ranging from 0.2 to 2 with a dry CO / O2 mixture flowing at 2 m/s (ZCO(s) =1). The model predictions show both the position of the 650K isotherm and location of maximum catalytic heat release versus time.5 cm corresponds to the tube outlet. Effect of varying the input power on the transient evolution of the catalytic flame. The test condition was φ = 1 using dry CO with ZCO(s) = 1 initially.
Model results with varying Nu (and Sh via the heat and mass-transfer analogy) that compare the propagation of a catalytic reaction in a Pt tube for φ ranging from 0.2 to 2 with a dry CO / O2 mixture flowing at 2 m/s (ZCO(s) =1).5 cm corresponds to the tube outlet. Comparison of model and experiment looking at the propagation of a catalytic reaction front inside a Pt tube for φ=0.2 to 2 with a wet CO / O2 mixture flowing at 2 m/s (ZCO(s) =1). The model predictions show the position of the 650K isotherm versus time.5 cm corresponds to the tube outlet. Computations showing the effect on catalytic propagation of varying the external heat transfer coefficients.
Two values are compared: h0=100W/m2·K and 2 times the natural convection correlation of Churchill and Chu[74]. External heat transfer coefficient, h0, along the length of the platinum channel for 4, 6, and 8 seconds after the beginning of ignition. The test conditions correspond to φ = 1 with dry CO. The values for h0 come from the correlation of Churchill and Chu[74].
Computed versus measured temperature profiles along the platinum channel at 4, 6, and 8 seconds after the beginning of ignition. Computed temperature profiles along the platinum channel comparing the effect of external heat transfer coefficient. Control volumes of solid and gas phase. Overall mass conservation control volume for gas-phase.
Gas-phase species conservation control volume. Gas-phase energy conservation control volume. Solid phase control volume (including catalytic surface shown in red). Schematic of stagnation point region used to estimate heat transfer from front face of monolith.
False colored image of the platinum tube during catalytic flame propagation taken with the IR camera. The platinum tube is shown schematically overlaid on the image. There are approximately 20 pixels of resolution across the diameter of the tube. Transmission characteristics of the filter used in IR imaging of platinum tube.
Relationship between Blackbody temperature to the camera counts and the parameter κ from equation 58.146 ix ACKNOWLEDGEMENT I wish to acknowledge my academic advisor, Professor James T’ien, whose dedication to his students has left a long legacy of which I am proud to be a member.