Comparative examination of aerosols of pulverized coal combustion in air and in oxygen with carbon dioxide combustion environments

The purpose of this work was to examine the effects of a change in combustion environment on aerosol formation. Due to the need to provide carbon based energy with carbon capture and sequestration to eliminate carbon emissions, oxy fuel combustion is a technology which is currently under investigation. The main advantage of oxy fuel combustion is that it utilizes a combustion process of fuel, pure oxygen, and recycled CO2 in order moderate flame temperatures. The result is a flue gas which is highly concentrated in CO2 with water vapor which is easily condensed and removed. However, there has been very little research done on the effects of the altered combustion environment from N2/O2 (air) to oxy fired conditions of O2/CO2 on aerosol formation. This work indicates that there are differences in black carbon particle emissions, as well as changes in mechanisms which drive aerosol formation. Of particular interest in the findings was that iron was found in higher relative concentrations in the submicron range of particles in oxy fired conditions compared to air fired conditions. Also, magnesium and calcium appear to be in higher relative concentrations in oxy fired conditions while sodium and potassium seem to be lower in the submicron range. These are all important as these compounds can affect how ash and slag build up on surfaces inside of the furnace.

A COMPARATIVE EXAMINATION OF AEROSOLS OF PULVERIZED COAL COMBUSTION IN AIR AND IN OXYGEN WITH CARBON DIOXIDE COMBUSTION ENVIRONMENTS by William James Morris A thesis submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Master of Science Department of Chemical Engineering The University of Utah December 2009 Copyright © William James Morris 2009 All Rights Reserved THE U N I V E R S I T Y OF UTAH G R A D U A T E SCHOOL SUPERVISORY COMMITTEE APPROVAL of a thesis submitted by William James Morris This thesis has been read by each member of the following supervisory committee and by majority vote has been found to be satisfactory. UNIVERSITY GRADUATE SCHOOL ofthe Chair Jost O. L. Wendt ~· 4~ Eric Eddings ~r ~~.~ eof lcox THE U N I V E R S I T Y OF UTAH G R A D U A T E SCHOOL FINAL READING APPROVAL To the Graduate Council of the University of Utah: 1 of William James Morris in its final form and have found that (1) its format, citations, and bibliographic style are consistent and acceptable; (2) its illustrative materials including figures, tables, and charts are in place; and (3) the final manuscript is satisfactory to the supervisory committee and is ready for submission to The Graduate School. Date Jo/t O. L. Wendt lair: Supervisory Committee Approved for the Major Department JoAnn Lighty Chair/Dean Approved for the Graduate Council Charles A. Wight Dean of The Graduate School UNIVERSITY GRADUATE SCHOOL APPROVAL ofthe I have read the thesis of William James Morris fmal JrP7{2 2--'3) z rJV '7 ABSTRACT The purpose of this work was to examine the effects of a change in combustion environment on aerosol formation. Due to the need to provide carbon based energy with carbon capture and sequestration to eliminate carbon emissions, oxy fuel combustion is a technology which is currently under investigation. The main advantage of oxy fuel combustion is that it utilizes a combustion process of fuel, pure oxygen, and recycled CO2 in order moderate flame temperatures. The result is a flue gas which is highly concentrated in CO2 with water vapor which is easily condensed and removed. However, there has been very little research done on the effects of the altered combustion environment from N2/O2 (air) to oxy fired conditions of O2/CO2 on aerosol formation. This work indicates that there are differences in black carbon particle emissions, as well as changes in mechanisms which drive aerosol formation. Of particular interest in the findings was that iron was found in higher relative concentrations in the submicron range of particles in oxy fired conditions compared to air fired conditions. Also, magnesium and calcium appear to be in higher relative concentrations in oxy fired conditions while sodium and potassium seem to be lower in the submicron range. These are all important as these compounds can affect how ash and slag build up on surfaces inside of the furnace. ABSTRAcT C02 02 02/C02 TABLE OF CONTENTS 1.1.1 Air Fired Combustion 3 1.1.2 Oxy Fired Combustion 5 1.2 Slagging 6 2.1 Overview of Combustion Laboratory 8 2.2 Oxy Fuel Combustor 8 2.3 Coal Feeding System 11 2.5 0 2 / C 0 2 Delivery System 12 2.10 Particle Probes and Sampling System 15 2.11 Equipment Used for Analysis 21 3.1 Air Fired Baseline Experiments 25 3.1.1 Loss of Ignition Tests 25 3.1.2 Scanning Mobility Particle Sizer Experiments 26 3.1.3 Photoacoustic Black Carbon Experiments 28 3.1.4 Size Segregated Fly Ash Partitioning Experiments 29 C-ONTENTS ABSTRACT ....................................................................................................................... iv LIST OF TABLES ............................................................................................................ vii LIST OF FIGURES ............................................................................................................ ix ACKNOWLEDGEMENTS ................................................................................................ xi 1. INTRODUCTION AND BACKGROUND ................................................................... 1 1.1 Coal Combustion ..................................................................................................... 3 1.1.1 Air Fired Combustion ...................................................................................... 3 1.1.2 Oxy Fired Combustion .................................................................................... 5 1.2 Slagging ................................................................................................................... 6 2. LABORATORY FACILITIES ....................................................................................... 8 ...................................................................... ................................................................................................ ............................................................................................. 2.4 Air Delivery System .............................................................................................. 11 02/C02 ....................................................................................... 2.6 Flue Gas Treatment ................................................................................................ 13 2.7 Burner Design ........................................................................................................ 13 2.8 Coal Type ............................................................................................................... 14 2.9 Flue Gas Analyzers ................................................................................................ 14 ................................................................. .............................................................................. 3. EXPERIMENTS ........................................................................................................... 25 ............................................................................. .................................................................................... .............................................. 26 ..................................................... 28 Segregated ....................................... 29 3.2 0 2 / C 0 2 Fired Experiments 3.2.1 Loss of Ignition Tests • 30 3.2.2 Scanning Mobility Particle Sizer Experiments 31 3.2.3 Photoacoustic Black Carbon Experiments , 32 3.2.4 Size Segregated Fly Ash Partitioning Experiments 32 ANALYSIS 4.2 Scanning Mobility Particle Sizer Results 39 4.3 Photoacoustic Black Carbon Results 44 4.4 Chemical Species Results of Size Segregated Fly Ash 45 5.4 Effects of C 0 2 on Particle Speciation 63 vi 02/C02 Fired Experiments ............................. ...... .... .... ....................... ............. ..... 30 3.2.1 Loss of Ignition .................. .................................................................. Experiments ... ........... ..... ........................... Black ........................................ .. .. ......... S!;!gregated .... ............................. ...... . . ... -"'. . . 4. RESULTS AND ANALySIS ....................................................................................... 34 4.1 Loss of Ignition Results ........................................................................... ..... ......... 34 ................. .... ..... .. ....... .. .............. ........... ............. ....... ............... ................................. .44 ........................... ............. .45 5. CONCLUSIONS ...................................................................................... ... .. .... .. ......... 61 5.1 Loss on Ignition .. ...................................................... ........... ........ ..... ... .................. 61 5.2 Particle Size Distribution .............. ... ......... .............................................. ....... ........ 62 5.3 Black Carbon Particle Density ............................................................................... 63 CO2 ......................................................... ........... Appendices A. PROCEDURES ............. .............................................................................................. 65 B. EXPERIMENTAL DATA ......................................................................... ................ .. 79 REFERENCES ... ... ...... .... ......... .. ........................................ ................ ..... ................. ... ... 126 VI LIST OF TABLES 1. Coal and ash composition 14 2. Raw data for air fired LOI results 80 3. Raw data for LOI for O2/CO2 detached case 3 flame 81 4. Data for LOI for 0 2 / C 0 2 attached case 4 flame 82 5. Raw data for 1st set of air fired SMPS scans 84 6. Raw data for 2nd set of air fired SMPS scans 87 7. Raw data for 1st set of case 3 SMPS scans 90 8. Raw data for 2nd set of case 3 SMPS scans 93 9. Raw data for 1st set of case 4 SMPS scans 96 10. Raw data for 2nd set of case 4 SMPS scans 99 11. Photoacoustic raw data without dilution calculation 102 12. Calibration curve for motive air mass flow controller in SMPS/PA 13. Calibration curve for nitrogen mass flow controller using a least squares fit 107 14. Raw chemical species results for 1s t set of air samples 108 15. Calcium correction for 1s t set of air samples 109 16. Mass of element per standard cubic meter of flue gas with dilution calculation for 1s t set of air samples 109 17. Raw chemical species results for 2 n d set of air samples 110 ............................................................................................. ................................................................................. 02/C02 ................................................... 02/C02 ........................................................... 1 st ................................................................ .............................. , ............................... 1 st .................................................................. ................................................................. 1 st .................................................................. ............................................................... ................................................ 1 02 SMPSIP A experiments ...................................................................................................................... 105 ........... 1 07 1 st ............................................ 1 st ........ , ................................................... 1 09 1 st ............................................................................................................. 2nd ........................................... 18. Calcium correction for 2 n d set of air samples Ill nd 19. Mass of element per standard cubic meter of flue gas with dilution calculation for 2 set of air samples Ill 20. Raw chemical species results for 1s t set of case 3 samples 112 21. Calcium correction for 1s t set of case 3 samples 113 22. Mass of element per standard cubic meter of flue gas with dilution calculation for 1s t set of case 3 samples 113 23. Raw chemical species results for 2 n d set of case 3 samples 114 24. Calcium correction for 2 n d set of case 3 samples 115 25. Mass of element per standard cubic meter of flue gas with dilution calculation for 2 n d set of case 3 samples 115 26. Raw chemical species results for 1s t set of case 4 samples 116 27. Calcium correction for 1s t set of case 4 samples 117 28. Mass of element per standard cubic meter of flue gas with dilution calculation for 1s t set of case 4 samples 117 29. Raw chemical species results for 2 n d set of case 4 samples 118 30. Calcium correction for 2 n d set of case 4 samples 119 31. Mass of element per standard cubic meter of flue gas with dilution calculation for 2 set of case 4 samples 119 32. Mass of element per standard cubic meter of flue gas for each run 120 33. Relative weight percentages of each run for each element 122 34. Differential mass log plot data using average Dp 124 viii 2nd ........................................................... 111 nd ............................................................................................................. 111 1 sl sample_s ................... ~ ................... 1 sl ....................................................... 1 st ........................................................................................................ 2nd ...................................... 2nd ...................................................... nd ........................................................................................................ 1 sl ....................................... 1 sl ....................................................... 1 sl ........................................................................................................ 2nd ...................................... 2nd ...................................................... nd ........................................................................................................ ......................... element.. ..................................... ..................................................... Vlll LIST OF FIGURES 1. Schematic of overall OFC research facility (Zhang et al., 2009) 9 2. Top section of the OFC with wall heaters, refractory, and windows (Zhang et a l, 2009) 10 3. Schematic of sampling for LOI experiments 16 4. Schematic of dilution sampling probe illustrating concentric tubes for the water jacket with a nitrogen feed for dilution at the tip to dilute the outgoing sample 18 5. Diagram showing the sampling setup used for SMPS and PA analysis 19 6. Sampling diagram for BLPI sample collection 21 7. Berner Low Pressure Impactor cross section illustrating the number of stages as well as the particle diameter range for each given stage 24 8. Averages of % LOI for all combustion scenarios 34 9. Average of LOI results for stoichiometric ratios 1.05 through 1.15 35 10. Attached oxy flame case 4 (courtesy of Zhang 2009) 37 11. Detached flame from oxy case 3 (courtesy of Zhang 2009) 37 12. Ingition loss as a function of oxygen percentage in the flue gas for all three flame scenarios 38 13. Sample SMPS scan for air fired conditions 39 14. Average air fired mass distribution for 15-660 nm particle diameter 40 15. Sample of case 3 0 2 / C 0 2 fired SMPS scan 41 16. Averaged SMPS mass distribution of case 3 oxy fired conditions 42 17. Sample SMPS scan of case 4 conditions 43 aI., ................................... aI., ................................................................................................................................. ................................................................ ......................... ....................... ............................................................ ........................................................... of% ......................................................... ofLOI ............................ ................................................ ....................................... flame ............................................................................................................................ ................................................................ ....................... .40 02/C02 ............................................................... .41 ............................ .42 ................................................................... .43 18. Averaged SMPS mass distribution for higher temperature case 4 oxy fired conditions 44 19. Sample of results from photoacoustic experiments with air dilution ratio 88:1, case 3 dilution ratio of 88:1, and a case 4 dilution ratio of 124:1 45 20. Averaged relative weight percentage of silicon ...........................................47 21. Differential mass of elemental silicon in terms of mg per standard cubic meter of flue gas 48 22. Average relative weight percentage of aluminum 49 23. Differential mass of elemental aluminum per standard cubic meter of flue gas as a function of particle diameter 50 24. Averaged relative weight percentage of iron in fly ash 51 25. Differential mass of elemental iron per cubic meter of flue gas as a function of particle diameter. The first two points of air fired combustion are omitted due to being below detectable limits and the issues with plotting zeros on a log scale 52 26. Relative percentage of magnesium in fly ash as a function of particle diameter 53 27. Differential mass of elemental magnesium as a function of particle diameter. Data points for low particle diameters are omitted due to being below detectable limits and the issues with plotting zeros on a log scale 54 28. Average relative weight percent of calcium in fly ash 54 29. Differential mass of elemental calcium per cubic meter of standard flue gas as a function of particle diameter 55 30. Relative weight percent sodium in fly ash as a function of particle diameter 56 31. Differential mass of elemental sodium in mg per cubic meter of flue gas as a function of particle diameter 57 32. Averaged relative weight percentage of potassium in fly ash 58 33. Differential mass of elemental potassium in mg per cubic meter of standard flue gas as a function of particle diameter. Data points that were below detection limits have been omitted rather than posted as zeros 59 34. Motive air mass flow controller curve 106 ........................................................................................................................... 88: case 3 dilution ratio of 88:1, and a case 4 dilution ratio of 124:1.. ................................... .45 20. Averaged relative weight percentage otsilicon ....... ~ ..... :.:~ ... ~ .. : ............ : .................... .47 21. Differential mass of elemental silicon in terms of mg per standard cubic meter of flue gas ..................................................................................................................................... 48 ..................................................... .49 .............................................................................................. .............................................. ......................... diameter. ...... ........................ '" ................................................. ............................................... diameter. ............................................................................................. ............ function diameter. ........................................................................................................... .................................... .................................................................................... ...................................................................... x ACKNOWLEDGEMENTS I wish to acknowledge the constant support and encouragement of family, particularly my mother Janice Tidwell and father Jim Morris and stepfather Jerry Tidwell, and friends throughout the pursuit of graduate education. In addition, special thanks to Dr. Jost Wendt, for inclusion in his research group. Also thanks to advisory committee members Dr. Eric Eddings and Dr. Geoff Silcox for their input and classroom instruction. I also greatly appreciate the help of the faculty and staff of the University of Utah department of Chemical Engineering including fellow researchers Jingwei "Simon" Zhang and Dr. Dunxi Yu as well as research engineers Dave Wagner and Ryan Okerlund and the undergraduate contributors Dallin Call and Raphael Erickson. I also wish to thank the Department of Energy for funding my research under awards DE-FC26-06NT42808 and DE-FC08-NT0005015, and Praxair for generously providing liquid storage tanks as well as the consumable C02 and O2 necessary to conduct this research. NTOOOSOlS, C02 02 CHAPTER 1 INTRODUCTION AND BACKGROUND Coal combustion is responsible for CO2 emissions which contribute to anthropogenic climate change. Carbon capture and sequestration (CCS) is an emerging technology designed to allow vast coal reserves in the US, as well as abroad, to be utilized while greatly reducing the net CO2 emissions to the atmosphere. There are many different technologies that are currently under development to aid in CCS such as integrated gasification combined cycle (IGCC), postcombustion capture, chemical looping, and oxy fuel combustion. The focus of the research contained in this thesis is on oxy fuel combustion. Oxy fuel combustion separates O2 and N2 from air in order to provide a pure oxygen stream for fuel combustion. This process has several advantages including the elimination of thermal NOx as well as producing a highly concentrated CO2 exhaust stream which does not require separation from N2 in the flue gas. This allows for an efficient sequestration process. However, there are still many unknowns involving oxy-fuel combustion. Since nitrogen is removed from air and only oxygen is introduced into the combustor, CO2 from the flue gas must also be reintroduced in order to keep flame temperatures low enough for safe operation in a given combustion facility. While CO2 and N2 are both generally inert species, the focus of this research was to determine whether replacing N2 emISSIOns NOx C02 C02 2 in the combustion environment with CO2 would significantly alter the mechanisms by which particles in the flue gas would be formed. In this research, a baseline test was conducted using once through CO2. Once through C 0 2 is CO2 that is not recycled from the flue gas is sourced from an external storage tank providing high purity gas. While this is not completely representative of the combustion conditions in a utility boiler, it would provide baseline data that could be used to determine whether the replacement of N2 with CO2 in the combustion environment would significantly alter the mechanisms that control aerosol formation. This would then allow for comparison to future work involving recycled flue gas which would contain impurities in order to be able to identify the relative contributions to changes in aerosol formation of each change in the combustion environment. The focus of this research involving oxy-fuel combustion is on a retrofit application. Therefore, the study of aerosol formation was of great importance because minerals in coal form fly ash and slag that affect utility boiler operation and efficiency. As a result, the focus of this research was to examine the size distribution and chemical composition of particles formed in the same furnace while utilizing the same Utah bituminous coal and only changing the combustion conditions from O2/N2 (air) to 02/C02 (oxy). A total of three flames were examined. One flame was a baseline air fired scenario. Two others were oxy fired cases. The first oxy case was selected based upon a matching adiabatic flame temperature which would provide a direct comparison between oxy fired and air fired flames of the same temperatures, but in different combustion environments. The final scenario involved a high concentration oxygen attached flame C02 CO2 but efficiency. 021N2 C02 3 that was being studied by another student examining flame behavior in the near burner zone. This case was examined in order to observe the behavior of flames with a greater concentration of oxygen and different flame and burner dynamics. 1.1 Coal Combustion Coal combustion is a highly complicated and long studied process. Although combustion occurs very rapidly, there are a great many steps and factors which can affect the formation of aerosols. Obviously the mineral components of the fly ash are different from one coal to the next and have the most significant impact, but the combustion environment could play a potential role as well, and is the focus of this study. 1.1.1 Air Fired Combustion In air fired combustion, there has been a great deal of research in terms of particle formation and mechanisms. This research has allowed for great insight into the mechanisms that govern aerosol formation in air fired combustors. While many of the governing mechanisms can be quite complex, there are a few fundamental aspects of aerosol formation that are understood. In coal combustion there is a submicron range class of particles which is high in terms of number of particles, while being only a few percentage points of total mass, and having a different chemical composition than the larger particles (Senior and Flagan 1982). In addition to the submicron particles, larger supermicron particles are also formed during coal combustion. The submicron particles are primarily the result of minerals in the coal particles that are volatile and vaporize and condense as they are oxidized (Linak and Peterson 1984). The supermicron particles are formed when the ash affect different playa 4 deposits in the coal are either left behind as the char surrounding burns away, or coagulate with other deposits which are softened and melting as the char burns away forming larger supermicron particles (Linak and Peterson 1984). While the details of certain particle formation, and more specifically speciation, are a subject of ongoing research, in air fired conditions there is a great deal of understanding in the mechanisms of aerosol formation in an air fired boiler. It is generally well understood that more volatile compounds will make up the fume and condense into submicron particles, while the larger particles are formed by larger deposits of less volatile minerals within the coal. There are many mechanisms that affect the volatility of compounds within coal ash. In some cases, such as with silicon oxides, SiC>2 is not volatile, but can be reduced to SiO by CO, and become much more volatile (Buhre et al., 2005b). This more volatile compound can then undergo condensation and reoxidation, contributing to the formation of new primary particles as well as the growth of other particles to which it binds (Buhre et al., 2005b). This phenomenon was used as a part of a predictive model for air combustion which assumed that the only source of C 0 2 in the combustion environment would be found at relatively large distances from the char particle undergoing combustion in a CO rich environment capable of reducing the S i 0 2 to SiO and forming C02 (Senior and Flagan, 1982). While the assumptions put forth in this mechanism are certainly valid for air fired combustion, they would not be valid for oxy fired conditions with very high concentrations of C 0 2 throughout the furnace. As a result, it is obvious that models and mechanisms that may hold for air fired conditions will not necessarily be valid for oxy fired conditions. bums bums Si02 aI., aI., CO2 Si02 CO2 CO2 5 1.1.2 Oxy Fired Combustion In oxy fired combustion, there is a lack of detailed study regarding the mechanisms and properties of ash generated by coal under oxy fired conditions. This is one of the critical areas of future oxy fired research, and was highlighted by Burhe et al. in 2005a as one of the major research steps that has yet to be undertaken in the oxy fired research community. Of particular interest is the interaction between CO and C 0 2 and how the enriched CO2 environment will affect the reduction of compounds by CO to form more volatile suboxides which could have great impact on fouling as well as particle size distribution and composition (Burhe et al., 2005a). Since volatile compounds are more likely to create deposits due to their molten state in the furnace that would allow them to stick to heat exchanger tubes or furnace walls, this is an important consideration for retrofitting air fired furnaces with oxy fired technology. Also, there is now greater interest in the formation of black carbon aerosols due to global warming concerns. It has been shown that higher amounts of O2 are needed in an O2/CO2 environment than an O2/N2 environment to achieve the same burnout levels (Borrego and Alvarez, 2007). This suggests that the combustion environment does affect the mechanism by which coal particles are ignited and burn. It also indicates that a greater amount of black carbon particles would be emitted from an oxy fired boiler than an air fired boiler with the same percentage of oxygen injected into the furnace. This is a significant concern because it means that the thermal efficiency of plant is reduced, while radiation impact of black carbon particles is increased. aI. C02 C02 aI., 02/C02 021N2 affect bum. 6 1.2 Slagging There are many factors that affect whether a particle will remain suspended in the flue gas stream and be emitted through the exhaust stack, captured by emissions control devices, or form a deposit on heat exchangers or furnace walls. It has been shown that heavier particles with high inertia, such as those with Fe compounds are much more likely to be able to break through boundary layers on the surfaces of objects and form deposits than less dense Si compounds (Huang et al., 1995). Ash deposits can be very costly for the operators of combustion furnaces and boilers. In the University of Utah combustion laboratory, the oxy fired combustor (OFC), which was used for the research contained in this thesis, must be shut down after long periods of testing and thoroughly cleaned in order to remove slag and ash deposits from the heat exchangers and sections of the furnace walls. Since it takes time to shut down the facility and perform the required maintenance, a very valid concern of oxy fired retrofits is the amount of ash build up that can be expected under the altered combustion conditions. Of course, as would be expected, the aerodynamics found in a furnace also has significant contribution to how and where deposits are formed (Huang et al., 1995). There is also research to prove, as expected, that deposition rapidly increases with increased flue gas temperatures (Huang et al., 1995). While a specific reason was not given, this finding is not surprising. There could be many contrinbuting factors such as greater temperature gradients and thermophoretic effects as well as partial melting of particles which are then much more likely to bind to a surface upon collision. Models that attempt to predict slagging are often complex and not necessarily accurate due to the high number of variables that contribute to slag formation and ash aI., OFe), aI., aI., 7 deposits (Wang and Harb, 1997). This indicates the need for experimental studies under actual combustion conditions with certain coals to begin to understand the mechanisms of aerosol formation, as well as collecting experimental data for a wide variety of coals under oxy fired conditions. CHAPTER 2 LABORATORY FACILITIES 2.1 Overview of Combustion Laboratory The University of Utah combustion laboratory located at 870 South and 500 West in Salt Lake City, Utah is the facility where all experiments were carried out during this research. The facility contains numerous combustion furnaces of different scales for all types of combustion and small scale explosion research. All experiments were conducted on the Oxy Fuel Combustor referred to as the OFC. 2.2 Oxy Fuel Combustor The OFC was designed and built by Ryan Okerlund and Jingwei "Simon" Zhang for a previous project and utilized as the test furnace for the particle research contained in this body of work as shown in Figure 1. The furnace is a down fired design and utilizes a very simple axial burner with no swirl. This was done as research was being conducted concurrently with a project which required the use of this burner type to facilitate modeling from data that were collected regarding flame performance. The OFC is a 100 kW furnace that burned coal at a feed rate of 10 lbs (4.54 kg) per hour. The furnace itself is refractory lined with electric wall heaters in the flame zone that keep the walls a constant temperature as shown in Figure 2. There are numerous lingwei Figure 1. Schematic of overall OFC research facility (Zhang et al., 2009). 48" 150" View/Ash Blm,vour P011 !,I @Jt ,I !@ I I 24" I .D. x 30" O.D. Burner Zone 3" x 39" QuartZ \Villdows 04) 840 \1./ E lectrical Heater~ 10.5" I.D. x 14" O.D . Radiant Zone (/) 3" Opposing Sample Ports ,-----Hea Exchanger ",1 / Hear Exchanger #8, aI., To Exhaus t Figure 2. Top section of the OFC with wall heaters, refractory, and windows (Zhang et al., 2009). sampling ports and thermocouples allowing for temperature readings throughout the furnace, and the opportunity to sample from a variety of locations. In these experiments, all sampling took place at the lowest sampling port before the flue gas was directed horizontally through the heat exchangers and then out the exhaust. This sampling location provided an easy work area with ample room to position equipment and was also at the same level as one of the thermocouples allowing for temperature readings of the flue gas as sampled. A thermocouple located opposite the sampling port also allowed for temperature readings at this position. For combustion experiments, the wall heaters were set to a value of 1850 F (982 C) to ensure consistent combustion conditions within the research furnace. 10 ofthe aI., 11 2.3 Coal Feeding System The coal feeder is a K-tron twin auger type feeder which has an operating range of 0-2,000 rpm and a feed rate of 0-20 lbs of coal per hour. In all tests, the coal was fed at a rate of 10 lbs per hour corresponding to an indicated 1,000 rpm on the coal feeder motor. The coal was fed by the augers through a screen mesh to break up any large clumps of coal and facilitate steady feeding. The coal is then fed into a plastic eductor where it is propelled by the primary air or a mixture of CO2 and O2 into the primary jet of the burner. The coal feeder is also on a set of rollers so that it may be adjusted to line the eductor up directly with the inlet of the coal burner. This proved to be critical for achieving a steady coal feed and stable flame without pulsing. It also allows the feeder to be rolled back away from the top of the furnace while running on natural gas in order to prevent heat damage to the feeder and the plastic eductor. 2.4 Air Delivery System The air is delivered by a 25 horsepower Sullair compressor which provides compressed air through the workshop. The primary air and the secondary air are monitored and controlled by the use of calibrated mass flow controllers which are all controlled with OPTO 22 by a computer work station in the control room of the combustion laboratory. The primary air uses an HFC 202 S/N while the larger secondary flow requires an HFC 203 S/N. For air fired combustion, primary air was fed at a rate of 4.59 kg or 10.1 lbs per hour and secondary air was fed at a rate of 43.5 kg or 95.8 lbs per hour. an 02 SIN SIN. 12 2.5 02/COz Delivery System Both the O2 and the CO2 used in these experiments were generously provided by Praxair. Praxair also provided liquid storage tanks for both the CO2 and the O2. These tanks were both plumbed into the primary and secondary air lines for the burner and controlled through manual valves to select O2 and CO2 instead of air for combustion. The O2 lines were cleaned and used materials specified by Praxair's safety recommendations in order to maintain a safe and consistent flow of O2 for use in experiments. The O2 tank was specially constructed on a concrete drip pad for safety, and was surrounded by concrete pylons and a gated fence to prevent accidental damage or contact with the tank body, heat exchangers, valves, or regulators. All tank and valve adjustments are to be handled by Praxair service employees in order to ensure the safety of the students and staff members conducting research at the facility. Students are not to make any adjustments with the tank, and should only operate the valve that allows oxygen to be fed to 870. Any other adjustments and service should be performed by Praxair technicians for safety reasons. Any malfunction should immediately be reported to laboratory manager Ryan Okerlund and Praxair. The CO2 tank provided the necessary dilution CO2 for the combustion atmosphere under oxy fuel conditions in the OFC. For all experiments contained in this body of work, the CO2 in the combustion environment was once through CO2 from the liquid storage tank provided by Praxair. The CO2 was not provided by recycled flue gas from the furnace itself as the intent of this research was to provide baseline data and insight into mechanisms of ash formation based on the difference of an O2/N2 air combustion lli/C02 02 C02 --air 02 III 02 C02 C02 C02 02/13 environment, and an O2/CO2 combustion environment. Research involving recycled flue CO2 CO2 CO2. It O2 CO2 to O2/CO2 delivery, it is very important that flue gas oxygen analyzers are not sampling the flue gas as an oxygen free 100% CO2 environment will destroy the sensor within the 2.6 Flue Gas Treatment Due to the small scale laboratory experiments, currently an air pollution permit is not required for the University of Utah's combustion facility. As a result, there are no emissions control devices found on the experimental furnace. Motive force for the flue gases is provided by an induced draft fan which provides the necessary suction to draw the flue gases out of the furnace. A damper can be opened to the proper using OPTO 22 controls to maintain the desired pressure within the furnace. 2.7 Burner Design In order to maintain compliance with specifications of concurrent research, an axial burner with no swirl was used in order to provide the coal flame for these experiments. The burner was constructed from stainless steel. The burner consists of two concentric cylinders which comprise the primary jet, through which fuel and motive 02/C02 gas is currently being pursued. The C02 is controlled by one master valve outside of the building adjacent to the tank. After it is opened, two valves must be . switched in order for C02 to flow through the air lines. If this is not done, then air will continue to flow rather than C02. is very important that the 02 flow be shut off from the furnace until the C02 flow through the furnace is established. Additionally, while changing from air delivery 02/C02 C02 analyzer. 14 Table 1. Coal and ash composition. O b y BTU/lb % % % % % % % % % 6 6 . 28 4 6 . 44 Analysis A 1 2 0 3 C aO F e 2 0 3 P 2 0 5 K 2 0 S i 0 2 N a 2 0 SO3 T i 0 2 % % % % % % % % % % % air or O2/CO2 and fed, and the secondary jet, through which the remaining air or O2/CO2 is fed to provide enough oxidant for combustion. The primary jet inner diameter measured I.D. 18.57 mm O.D. 21.34 mm with a wall thickness of 2.769 mm. The secondary sleeve measured I.D. = 42.16 mm with a wall thickness of 3.556 mm. For more information see Zhang et al., 2009. 2.8 Coal Type The coal used for these experiments was a Utah bituminous coal with the composition shown in Table 1. The coal and ash analysis was performed by Huffman Laboratories, INC. 4630 Indiana Street Golden, CO. 80403. 2.9 Flue Gas Analyzers In order to verify correct combustion conditions, stoichiometric ratios, and flame consistency, several flue gas analyzers were used in order to provide the necessary information regarding the combustion environment and flame characteristics. 02/C02 02/C02 = = ~easured LD. = mmwith walHhiCkness aI., Huffman co. Coal Analysis LOD Ash Volatile Fixed HHV 750 Oby 105 C C C H N S diff Matter Carbon BTUIlb 3.03 11.71 66.28 4.63 1.23 0.38 13.07 38.81 46.44 11731 Ash Analysis Al Ca Fe Mg Mn P K Si Na S Ti as as as as as as as as Ah0 3 CaO Fe203 MgO MnO P205 K20 as Si02 Na20 as S03 Ti02 13.3 11.75 5.06 2.88 0.05 0.3 1.49 52.37 1.44 2.44 0.65 Two analyzers were used for recording concentration of oxygen in the flue gas. One was a Yokogawa Zirconium Oxide instrument and the other was a Horiba O2 MPA- 510. Three analyzers were used for measuring CO2 concentrations. The first analyzer, a ZRH infrared C02/CO by California Analytical Instruments, was used for air fired combustion and measured concentrations of less than 20% by volume. It was also used to monitor CO in the 0-1,000 ppm range. The second analyzer, another a ZRH infrared C02/CO by California Analytical Instruments, was used for Oxy Fuel conditions and could measure CO2 up to 100% concentration and CO in terms of percent volume. The third CO2/CO analyzer, California Analytical Instruments ZRE CO2 Gas Analyzer, was easily calibrated and set up for both air and oxy conditions and was used as a second back up analyzer in either condition. In addition to the O2 and CO2/CO analyzers, an NO/NO2 and an NO analyzer were also utilized to monitor combustion conditions. The NO/NO2 analyzer, Thermal Environmental 42 CHL, and the NO analyzer, Horiba NO CLA 51055, were both calibrated using an NO in N2 standard for air fired calibrations and an NO in CO2 standard for oxy fired calibrations. Information regarding the calibration and operation of the flue gas analyzers can be found in Appendix A. 2.10 Particle Probes and Sampling Systems Two particle probes were used to draw samples for the experiments performed in this research. The first probe utilized was very simple and did not utilize dilution. It featured a 1" pipe which was measured to be 7/8" inner diameter. This was then sleeved by 1.5" tubing to form a water jacket to keep the probe cool. It was used in order to take 15 02 CO2/AnalYtical C02 C02/C02/N02 N02 a total ash sample for the loss of ignition measurements. The experimental setup is shown in Figure 3. In order to collect the samples, there had to be a filter housing that enclosed the filter, a vacuum pump to provide vacuum to draw the sample out of the furnace, and a rotameter to control the flow. The rotameter was a 50 SCFH unit from Dwyer. The amount of flow through the probe, which was to be as close to isokinetic as possible, was determined using the following equations from Hinds, 1999. 91 Q0 Us=Uo mm housing il housing. not Figure 3. Schematic of sampling for LOI experiments. 16 IS PtlltP -the-san1ple · Probe and furnace not to scale 90 mro filter housing Dwyer 50 SCFH Rotameter to control flow through filter housing. 3 hp vacuum pump provides flow through the impactor Where Qs is the volumetric flow rate of the sample in the probe of diameter Ds and Qo is the volumetric flow through a circular duct with a diameter Do- As a result, the velocity of the bulk gas in the duct is the same as the sample velocity in the probe. The velocity given for air fired conditions ranged between 2.30ft/s (0.70lm/s) to 2.45ft/s (0.747m/s) as the temperature of the flue gas varied from 1100F (593C) to 1200F (649C). This velocity can be calculated knowing the known inner diameter of the furnace, the known temperature of the gas from a thermocouple reading, the known flue gas composition and coal tiring rate, and the ideal gas law. The ideal gas law has been deemed sufficiently accurate for air pollution sampling (de Nevers, 2000). The air fired cases required a flow rate through the probe of 34.5 to 36.5 cfh or about 16.3 to 17.3 1pm for air fired conditions. These calculations can be repeated to show the flow needed for oxy fired conditions as well, which turn out to be about 11.5 1pm for case 3 and 9.2 1pm for case 4. As particles deposited on the filter, the rotameter valve could be opened further to allow continuing sample flow through the probe and filter. For more information regarding sampling, see Appendix A. The second probe utilized in these experiments was brought with Professor Jost Wendt when he moved to the University of Utah from the University of Arizona. This probe utilizes a stainless steel water jacket as well as a having a nitrogen jet at the tip of the probe allowing for dilution of the sample as well as quenching reactions and reducing the amount of coagulation of small particles (Hinds, 1999). This probe was used for all SMPS, PA, and BLPI data collection. The probe had a measured inner diameter of 0.265" or 6.73mm and is shown in Figure 4. 17 Qs Ds Do. conditiotls 2.30ftls 0.70Im/2.45ft1s IIOOF firing lpm tum lpm Ipm P A, 18 Dirty flue gas enters the probe tip where analyzed it is quenched by the dilution nitrogen. The probe is enclosed in a stainless steel water jacket. Dilution nitrogen is injected through the lower inlet on the sample probe. Figure 4. Schematic of dilution sampling probe illustrating concentric tubes for the water jacket with a nitrogen feed for dilution at the tip to dilute the outgoing sample. When setting up the sampling system for photoacoustic and SMPS analysis or for sample collection with the impactor, the dilution probe was used in conjunction with a mass flow controller that metered the rate of dilution nitrogen into the sampling probe as shown in Figure 5. It was found that at the controller manufacturer's recommended ideal operating feed pressure of nitrogen, the flow of nitrogen pulsed wildly through the probe due to the complicated bends of the nitrogen inlet near the probe tip. As a result, the inlet pressure of nitrogen to the mass flow controller was reduced from the recommended 85 psi to 45 psi to provide a steady flow. As a result of the change in inlet pressure and using nitrogen for dilution rather than air, a calibration curve was set up for the mass flow controller at these conditions, which can be found in Appendix B. qu€nched nitrogen . . The diluted sample to be collected or analyzed "..... . .CF~ .~....... r ----.:J" r- "" Th e pro be enc ose d'I n controller at these conditions, which can be found in Appendix 19 orifice controls flow and a pressure gauge verifies critical flow. CO CO .O 2 .o I t outlets for using instruments simultaneously Air filters to clean motive air and mass flow controller to regulate flow. cylinder no 3 T3 a> in c/i n CL Figure 5. Diagram showing the sampling setup used for SMPS and PA analysis. The equipment setup for use with the SMPS and PA was rather involved. The dilution probe was used with a mass flow controller to meter the amount of nitrogen from either a cylinder or a liquid nitrogen dewar. The sample was withdrawn from the probe using an eductor which utilized clean dry motive air, which was then used for sample dilution. The motive air is supplied by compressed air which is filtered through progressively smaller particle filters and oil and water separators to protect the mass flow controller which regulates the amount of motive air to the eductor. The calibration curve for the motive air mass flow controller can also be found Appendix B. The motive air is then finally sent through HEPA filters to ensure a clean dilution stream. The dilution at Probe and furnace not to scale Dilution Manifold ~ provides multiple ~ outlets for using instruments simultaneously An eductor provides vacuum to draw out sample while a critical m~ Dilution nitrogen set by mass flow controller from nitrogen cylinder n o 3 ".0., IT) VI VI IT) Q. ~.., P A HEP A the probe tip and then at the eductor resulted in a compound dilution, which was important to remember when calculating data. Flow through the probe is controlled by a critical orifice from O'Keefe controls that was measured to have a flow rate of 8.45 1pm using a Gillibrator bubble flow meter. A pressure gauge also verified the necessary 2: 1 pressure drop across the orifice necessary to maintain constant choked flow. After flowing through the critical orifice and eductor, the diluted sample stream was sent into the dilution manifold. The dilution manifold consists of a large tube with several sampling ports. As a result, it is possible to operate the SMPS and the PA simultaneously; however they will be operating at the same dilution, which proved not to be ideal for this range of experiments. The diagram in Figure 5 illustrates the previously described instrumentation and sampling setup. The setup used for sampling with the Berner Low Pressure Impactor (BLPI) was less complicated than the sampling system necessary for the PA and SMPS. It simply utilized the dilution probe with nitrogen from a cylinder or dewar controlled by the mass flow controller as shown in Figure 6. The diluted sample was then sent through the BLPI while a 3 hp vacuum pump provided the necessary vacuum to create choked flow through the impactor where the last stage acts as a critical orifice. The flow through the impactor was measured with a Gillibrator bubble flow meter at 24.1 1pm. This was used in conjunction with the amount of dilution nitrogen to determine the dilution ratio of sample in flowing through the impactor. A diagram of the BLPI sampling set up is shown in Figure 6, while a diagram of the BLPI can be found in section 2.11. 20 III tpm P A lpm. Berner Low Pressure Impactor Dilution nitrogen set by mass flow controller from nitrogen cylinder Figure 6. Sampling diagram for BLPI sample collection. 2.11 Equipment Used for Analysis A wide variety of equipment was used for analysis. A Scanning Mobility Particle Sizer, referred to as an SMPS, was used in order to determine the particle size and mass distribution for particles in the 15-660 nanometer range. The SMPS was a TSI 3080 classifier with a 3081 DMA using a 3022A CPC model. The SMPS was used in conjunction with the dilution manifold and is controlled with a desktop PC work station using proprietary TSI Aerosol Instrument Manager (AIM) software. The data are saved as raw scans, and then must be exported as a tab delimited text file which can be opened Probe and furnace not to scale 21 low 3 hp vacuum pump provides flow through the impactor 22 in Excel for analysis. More detail of the operation and set up of the SMPS can be found in Appendix A. Along with the SMPS, the particle stream was analyzed by a photoacoustic analyzer referred to as the PA, which utilized a laser tuned to black carbon in order to measure the concentration of soot and pure black carbon particles found in the flue gas. For a detailed description of the instrument and its verification, see Arnott et al., 2004. The PA, like the SMPS, was controlled by a desktop PC. However, the PA is controlled through a Labview software interface. The PA continuously analyzed the particle stream and did not collect data as discrete scans. Instead, it collects a data point every 6 seconds. Using the Labview software, a selection switch exists that will allow for recording of the data as a tab delimited text file that can be opened in Excel. Labview will only record data when this is selected. For more information regarding the operation of the PA, see Appendix A. Along with the SMPS and PA, a Fossil Energy Research HFLOI loss on ignition instrument was used in order to gravimetrically determine the loss of ignition of total ash samples that were collected during the experiments where loss of ignition was measured for both oxy and air fired conditions at varying stoichiometric ratios. The instrument comes with a very easy to follow instruction manual and includes a feature for drying samples before they are analyzed. This proved to be incredibly valuable as with the water cooled sample probe, there was no way to maintain the sample temperature above the boiling point at the filter housing. Samples that were collected to be analyzed for chemical composition were collected using a Berner Low Pressure Impactor (BLPI). The BLPI is a low pressure pc. P A F or P A, 23 cascade impactor that size segregates particles to be collected for analysis. It has a much wider submicron range than many other impactors. Also, its lowest stage acts as a critical orifice to maintain constant flow through the impactor. Figure 7 illustrates the 11 stages of the impactor including their aerodynamic cutoff particle diameters in microns. In order to analyze the samples collected with the low pressure impactor for metal concentrations, the samples were taken to Enviropro Laboratories LLC in West Valley City, Utah. There, they were acid digested and analyzed with a Perkin Elmer ICP-AES to give elemental concentration. The samples were prepared using EPA method SW846 3050A and analyzed using EPA method 601 OB. Samples were also collected using the BLPI for SEM/EDS analysis with cooperation from the Department of Physics at the University of Utah, which owns the instrument, for major species by Dr. Dunxi Yu, particularly Si, which requires a special acid digestion and expensive consumable parts for an analytical instrument as a result. 6010B. 24 Orifice Plate Substrate Atmospheric section ~ Low pressure secfon Aerodynamic 50% Cut-Off Diameter Plate 11 7.33 3.77 1.98 0.973 0.535 0.337 0.168 0.0926 0.0636 0.0324 Plate 10 Plate 9 Plate 8 Plate 7 Plate 6 Plate 5 Plate 4 Plate 3 Plate 2 Plate 1 Figure 7. Berner Low Pressure Impactor cross section illustrating the number of stages as well as the particle diameter range for each given stage. [pm] 0IiIIte 15.7 ,Subsb.ta A1mospbaic 98CbI Lowpi8SSW8 secIon 0.0926 0.0324 CHAPTER 3 EXPERIMENTS 3.1 Air Fired Baseline Experiments All experiments were conducted under both air and oxy fired conditions. Air fired experiments were always conducted in order to provide a baseline for comparison with oxy conditions. This allowed examination of the differences and significance of the differences between O2/N2 and O2/CO2 combustion environments to provide insight to potential challenges in retrofit applications of existing air fired boilers. The conditions for the air fired experiments referred to as case 1 were 10 lbs (4.54 kg) per hour of coal, 10.1 lbs (4.58 kg) per hour of primary air, and 95.8 lbs (43.45 kg) per hour of secondary air. This resulted in an adiabatic flame temperature of 3853 F (2123 C). The combustion cases and adiabatic flame temperatures were generously shared by Jingwei "Simon" Zhang who had performed these flame calculations for his research prior to the start of the author's work. 3.1.1 Loss of Ignition Tests The Loss on Ignition, referred to as LOI, tests were the first tests completed in this research. This allowed for insight into the performance and efficiency of the furnace as well as providing a relatively easily sampling procedure to practice sampling techniques for pending research. 021N2 02/C02 95.81bs of3853 lingwei oflgnition 26 In each test, a vacuum pump was utilized to pull samples through the probe which was inserted in the lowest sampling port on the OFC. The vacuum pump pulled through a filter housing containing a cellulose acetate filter (Advantec grade: C045A090C 0.45 micron 90 mm diameter) which was contained in a 90 mm filter housing. Flow through the probe was controlled by a Dwyer rotameter in order to control the flow to maintain sampling as close to isokinetic as possible. Samples were collected on the filter paper for 30 minutes to ensure significant loading on the filter paper. A minimum of two pieces of filter paper were collected for each stoichiometric ratio to allow for four different samples to be analyzed for LOI for each specific case. The samples were collected on the filter paper and then transferred to glass sample vials. The ash was then analyzed in the Fossil Energy Research designated weigh boats using the Fossil Energy Research HFLOI. In the case of air, the stoichiometric ratios of 1.15, 1.1, 1.05, and 1 corresponded to 3.00%, 2.00%, 1.00% and 0.00% oxygen in the flue gas, respectively. While this was obtainable down to 1.00% oxygen, it was not possible to accurately measure 0% oxygen in the flue gas as this condition would ruin the zirconium oxide element in the oxygen analyzer used to monitor flue gas oxygen concentration. As a result, considerable uncertainty surrounded the sampling for LOI at the stoichiometric ratio of 1.00. 3.1.2 Scanning Mobility Particle Sizer Experiments Performing the SMPS experiments required the use of significant dilution in order to bring the number or particle counts within operating range of the SMPS. This was achieved with nitrogen dilution at the tip of the probe using an up to 50 liter per minute mass flow controller, which was calibrated using a Gillibrator bubble flow meter, to different 27 meter a specified flow from a compressed nitrogen gas cylinder or a liquid nitrogen dewar. Of particular importance was that in order to use the mass flow controller with the dilution probe, the inlet pressure on the mass flow controller had to be reduced to 45 psi in order to provide steady flow. The flow was then validated using a Gillibrator bubble flow meter to ensure accurate recording of the amount of actual dilution nitrogen dilution in the probe as shown in section Appendix B. In order to provide enough data for averaging of scans to provide a smooth curve and mass distribution, up to 20 scans were taken of air fired combustion on testing days. This ensured that there would be sufficient data even after erroneous scans were aborted or discarded due to unexpected or unintentional interruptions of the steady state combustion process. Also, the concentrations of flue gas O2, CO2, NO, and NO2 were monitored to verify that scans were occurring during consistent combustion conditions. Tests were also performed on different days to address repeatability and consistency of data and combustion experimental conditions. Experiments were conducted on two separate occasions at least a week apart. The dilution manifold uses an eductor to create the necessary suction to draw the samples through the probe and then the motive air from the eductor is used to dilute the sample as it is driven into the dilution manifold. The motive air is controlled by a calibrated mass flow controller, and the flow through the sampling probe is determined using a critical orifice available from a complete set from O'Keefe controls. This sampling setup was also utilized in all photoacoustic experiments where the dilution manifold fed samples for both the SMPS and the photoacoustic analyzer. flow. 02, N02 photo acoustic Using the methods and equations described in section 2.10, the sampling was kept as close to isokinetic as possible. However, due to small fluctuations in furnace operation as well as small errors in the calibration curves, the calculated ideal isokinetic sample flow rates of 0.8 to 1.5 1pm through the probe were at times too small to provide accurate and consistent flow. As a result, the dilution was set such that the flow of the sample into the probe entered at 2 to 3 1pm in order to achieve consistent, uninterrupted flow as close to calculated isoketic conditions as possible. See Appendix A for more discussion. 3.1.3 Photoacoustic Black Carbon Experiments PA experiments were conducted to measure the concentration of black carbon or soot particles found in the flue gas. These experiments were conducted to provide insight on soot formation differences between O2/N2 environments and O2/CO2 conditions. The PA experiments were conducted simultaneously with SMPS experiments using a dilution manifold which allowed up to three instruments to analyze the sampling stream simultaneously. The PA is controlled by the same PC work station as the SMPS, but is controlled through a different Lab View program specific to the PA. Instead of individual scans with the SMPS, the PA continuously takes data at a rate of one reading per 6 seconds. It was also run on different days and scans of several minutes were taken while collecting the SMPS data. PA scans are recorded by Lab View as a tab delimited text file which can then be imported into Microsoft Excel for analysis. The PA has been tested and validated during experiments involving diesel combustion and jet engine applications (Amott et al., 2005). 28 lpm Ipm P A 02/02/C02 P A PAis P A. P A P A ai., 3.1.4 Size Segregated Fly Ash Partitioning Experiments The size segregated partitioning experiments were conducted utilizing the dilution probe, nitrogen for the dilution, and a Berner Low Pressure Impactor referred to as a BLPI. The BLPI contains a critical orifice which limits the flow through the impactor. A three horsepower vacuum pump is used to provide the necessary vacuum to create critical flow through the impactor which remains constant. The flow through the impactor was measured using a Gillibrator bubble flow meter in order to determine the actual flow through the BLPI, 24.1 1pm, so that the dilution of the sample would be known. The BLPI consists of 11 different stages. Each stage has a different 50% aerodynamic cutoff diameter. The cutoff diameters are 0.030, 0.060, 0.090, 0.170, 0.340, 0.540, 0.980, 1.98, 3.77, 7.33, and 15.7 microns. The stages were each covered with a substrate in order collect the particle samples. The procedure of substrate preparations are detailed in Appendix A. The samples were then acid digested and analyzed using ICP-AES in order to provide elemental concentration for the following elements, Si, Al, Fe, K, Mg, Ca, Na, Se, and Ar. The samples were collected on two different substrates, both of which were tested for background elements. The first, filter 1, was a Millipore Durapore membrane filter type 0.22 um catalog number: GVHP09050. The filters which were most economical and easiest to acid digest were, filter 2, Millipore Isopore membrane filter type 0.4 um catalog number: HTTP09030. These were both found to have traces of calcium, which provide background that had to be subtracted out as shown in Appendix B. When sampling for the air fired conditions, two BLPIs were needed. One BLPI used a preseparator cyclone for collection of samples on stages 1-5. The samples were 29 lpm, AI, urn urn BLPls 30 collected over a time interval of 20 minutes. For the larger diameter particles, a BLPI overloading of the stages 6-11. 3.2 O7/CO? Fired Experiments The oxy fired experiments examined two different flames. One was a matched adiabatic flame temperature O2/CO2 flame and another was a high temperature oxygen enriched attached flame. The matched adiabatic flame temperature was referred to as case 3 in the flame scenarios. The conditions were a coal feed rate of 10 lbs (4.54 kg) per hour, 12.2 lbs (5.53 kg) per hour CO2 and 2.3 lbs (1.04 kg) per hour O2 primary, and 62.0 lbs (28.12 kg) per hour C 0 2 and 22.2 lbs (10.07 kg) per hour 0 2 secondary. This resulted in the matched adiabatic flame temperature of 3849 F (2121 C). The higher temperature flame, referred to as case 4, was the flame primarily studied by Jingwei "Simon" Zhang for his Ph.D. research and was examined in this work as well. The conditions were a coal feed of 10 lbs (4.54 kg) per hour, 12.2 lbs (5.53 kg) per hour C 0 2 and 2.3 lbs (1.04 kg) per hour O2 primary, and 38 lbs (17.24 kg) per hour CO2 and 22.2 lbs (10.07 kg) per hour O2 secondary which yielded an adiabatic flame temperature of 4576 F (2524 C). Again, the combustion cases and adiabatic flame temperatures were generously shared by Jingwei "Simon" Zhang who had performed these flame calculations for his research prior to the start of the author's work. 3.2.1 Loss of Ignition Tests The LOI samples were collected with the same apparatus and technique as described in section 3.1.1 under air fired conditions. However, with the oxy fired was used without the cyclone and only sampled for five minutes to prevent particle lli/C02 02/C02 Ibs Ibs Ibs 02 Ibs CO2 Ibs 02 lingwei Ibs Ibs CO2 Ibs Ibs Ibs lingwei ofIgnition 31 conditions, there were two conditions that were tested. One condition, referred to as case 3, was an O2/CO2 flame which matched the adiabatic flame temperature of the air fired flame. The other condition, referred to as case 4, was an oxygen enriched flame which was attached to the burner tip. This is the flame that Jingwei "Simon" Zhang was examining for his research. As in the air fired cases, the samples were collected while monitoring flue gas composition to ensure consistency of the flame and combustion conditions. The stoichiometric ratios of the oxy cases were also given as 1.15, 1.10, 1.05, and 1.00. However, the oxy cases had much higher concentrations of oxygen for a given stoichiometric ratio than the air fired cases. As a result, the results are given based upon both stoichiometry and percent O2 in the flue gas in section 4.1. Again, due to the limitations of the oxygen analyzer, it was not possible to precisely measure the oxygen down to a concentration of 0.00%, and thus a great deal of uncertainty resulted. Two pieces of ash laden filter paper were collected for each case and each stoichiometric ratio. This allowed for enough sample to be used to run four total samples for each case and stoichiometric ratio. All samples were analyzed using the Fossil Energy Research HFLOI analyzer following the protocols in the manual, which was consistent with the analysis of the air fired combustion samples. 3.2.2 Scanning Mobility Particle Sizer Experiments The SMPS experiments for the oxy fired cases were carried out using the same methodology described in section 3.1.2 for the air fired analysis. The two oxy fired cases were also examined on separate days at least a week apart to verify repeatability of the experiments. As with the air fired experiments, up to 20 total scans were taken for each 02/C02 lingwei 32 oxy case in order to average out irregularities and to provide a smooth curve of mass distribution. For each scan, the readings of the flue gas analyzers were recorded to compare with scans and to monitor the consistency of the combustion environment during analysis. When performing experiments for both oxy cases, sufficient time was given when changing from case 3 to case 4 to allow for a return to steady state before resuming analysis. In all cases the furnace was operated at a stoichiometric ratio of 1.15 rather than matching the concentration of oxygen in the flue gas. 3.2.3 Photoacoustic Black Carbon Experiments The PA was used to collect data simultaneously with the SMPS as in the air fired experiments. The data collection was the same as described in the air fired section 3.1.3 and detailed in Appendix A. 3.2.4 Size Segregated Fly Ash Partitioning Experiments In the oxy fired conditions, the ash samples were collected using the same equipment as described in section 3.1.4. However there was a difference in the duration of sampling time for case 4. In case 3, the samples were collected on stages 1-5 with use of the cyclone for a total of 20 minutes. For the stages 6-11, another BLPI was used without the cyclone for a sampling time of 5 minutes. When conducting experiments for case 4, the concentration of particles in the flue gas was much higher. This was a result of less CO2 in the secondary stream creating an oxygen enriched attached flame. However, since there was less diluent CO2 injected into C02 33 the furnace, but the same coal feed rate, the concentration of particles in the volume of flue gas was considerably higher. As a result, this led to overloading of particles for the same sampling times. In order to eliminate particle bounce off and overloading, the sampling time was reduced to 12 minutes with use of the cyclone for stages 1 -5 and 2 minutes without the cyclone for stages 6-11. The sampling times for case 4 were determined by filling the impactor stages with blank aluminum foil grease coated substrates. Sampling took place for 2 minute intervals at which point the substrates were checked for overloading. Sampling continued until substrates were overloaded as indicated by particles bouncing off and adhering to walls and other areas of the impactor. This sampling time algorithm was adapted from the method described by Seames in his 2000 dissertation. waS 1-5 and 2 CHAPTER 4 RESULTS AND ANALYSIS 4.1 Loss of Ignition Results The averages of the loss of ignition results are presented in Figure 8. The LOI results showed that the combustion furnace provided very complete burnout leaving behind a very limited amount of unburned char in the fly ash. Due to potential for ruining the oxygen analyzer by running at a stoichiometric ratio of 1 with 0% oxygen in the flue gas, the results for a stoichiometric ratio of 1.00 are subject to substantial error, as it had to be estimated when the oxygen in the flue gas would be 0% based on varying the flow rate of oxygen or air into the furnace. 18 16 c o S 12 29 10 2 8 3 6 4 2 0 Loss of Ignition vs Stoichiometric Ratio •Air fired conditions •02/C02 Case 3 Detached 02/C02 Attached Case 4 1 1.05 1.1 1.15 1.2 Stoichiometric Ratio Figure 8. Averages of % LOI for all combustion scenarios. ANAL YSIS c: 14 0 '''c;:; 12 -!!P 10 0 8 III III 0 6 ...I '* 4 2 0 0.95 ~ '" Ii II '" .~ " l ~"" - - T -" T 1.2 -+-conditions _ Detached ..... of% 35 0 I _____ 1.16 Stoichiometric Ratio Figure 9. Average of LOI results for stoichiometric ratios 1.05 through 1.15. As a result, a clearer picture of the averages of the LOI for all three conditions is shown in Figure 9 ranging from a stoichiometric ratio 1.05 to 1.15. While the oxy cases have the lowest average ignition loss based upon stoichiometric ratio, they also have the highest concentrations of oxygen. The case 3 scenario is 31.2% oxygen while the case 4 scenario is 40.2% oxygen. This is consistent with other drop tube experiments which found that in order to achieve the same burnout performance of an O2/N2 combustion environment, there must be a greater percentage of oxygen in the O2/CO2 environment (Borrego and Alvarez, 2007). While the oxy cases have lower average LOI, the most interesting result is that the case 4 combustion scenario, which is a much higher percentage of oxygen, seems to have greater ignition loss than did the case 3 combustion scenario, though the results are within the error limits. There are several possible mechanisms that may account for this, and they all have to do with enhanced mixing of the fuel and gas mixtures in case 3. Loss of Ignition 12 r - 10 T c LOl ' 0 21N2 0 2/C02 LOl, c: 0 ; ·c -!P 0 .''0."".. ~ 8 6 4 2 ~ - .or . .. - -'f.. . 1: "-l ,i 1.04 1.06 1.08 1.1 1.12 1.14 1.16 Stoichiometric Ratio ~Air fired conditions ~02/C02 Case 3 Detached ~02/C02 Attached Case 4 ofLOl 36 The first factor to be considered is that the velocity of the gases in the secondary stream of case 3 are significantly higher than in case 4 since there are an additional 24 lbs per hour of CO2 flowing through the same cross sectional area. This would create a greater shear stress between the flows of the primary stream which contains the fuel, and the secondary stream which contain nearly all of the oxidant. This greater shear stress improves mixing between the two streams, and helps to more fully oxidize the char particles. Another issue related to mixing of the streams is that case 4 is an attached flame, while case 3 is a detached flame and has a considerable standoff distance from the burner tip. Due to error introduced by fluctuations in the flame, it is not possible to develop an absolute conclusion. However, on average the better mixed flame probably features greater burnout. The attached flame is shown in Figure 10, while the detached flame is shown in Figure 11. As shown in the photographs in Figures 10 and 11, there is a zone of premixing between the primary and secondary streams as they exit the burner before the flame actually ignites. This would allow greater mixing and thus greater char burnout than the attached flame where there is less mixing between the fuel rich inner section of the flame and the surrounding oxygen from the secondary stream. Another important consideration is that air separation units are energy intensive to operate, and thus very expensive to operate as well as reducing the overall efficiency of an oxy fired plant. The Figure 12 shows that the oxy fired conditions seem to be less susceptible to ignition loss when running lower stoichiometric ratios, and therefore less expensive since less excess oxygen is being wasted in the flue gas. flame, fluctuations flame flame Figure 10. Attached oxy flame case 4 (courtesy of Zhang 2009). Figure 11. Detached flame from oxy case 3 (courtesy of Zhang 2009). 37 38 co 2c 8 M 1/1 3 4 2 LOI as a Function of 02 in Flue Gas ••Case 3 Matched Adiabatic 02/C02 Flame Case 4 Attached 02/C02 Flame Figure 12. Ingition loss as a function of oxygen percentage in the flue gas for all three flame scenarios. The results in Figure 12 clearly show that for the matched adiabatic flame temperature, it is possible to get similar combustion efficiency at a stoichiometric ratio of 1.1 with nearly 3% O2 in the flue gas as with air at a traditional stoichoimetric ratio of 1.15 with 3% O2 in the flue gas. This indicates that there may be potential for process optimization to achieve the balance of completeness of combustion with energy and cost required to operate an air separator. Also, it is very obvious that mixing plays a critical role in the combustion of the coal char particles since case 4, which has a combustion environment of 40% oxygen, has greater ignition loss than the case 3 flame which is only 31% oxygen even when there is the same amount of oxygen in the flue gas postcombustion even considering the potential error. 02 L 15 02 '31 % 12 10 c: .0. 'c ~ c: 6 0 III III ..0.. . '* 2 0 0 lOI % 2 4 Percentage of 02 in the Flue Gas 6 -+-Air Flame _ ....-39 Figure 13. Sample SMPS scan for air fired conditions. 4.2 Scanning Mobility Particle Sizer Results The SMPS initially counts the number of particles in each bin size that corresponds to a range of particle diameters. The number of ultrafine particles is shown in Figure 13 to be significantly greater than the number of particles in the 600 nm range at the end of the scale. However, due to the very small diameter of the ultrafine particles in the 15 nm range, the total mass is very small compared to the larger particles. As a result, when the raw counts are converted to a mass basis, the curve looks very different. The mass is determined by assuming spherical particles with a constant density throughout the size range. SIze '-number 4.0 3.5 _ 3.0 (0 -II) ~ 2.5 .u.... ! 2.0 a. o CI ~ 1.5 ..... z 'C 1.0 0.5 0.0 10 100 Diameter (nm) 1000 40 The curve in Figure 14 gives the average over all of the collected scans and shows how the ultrafine fume of condensed vapor particles occurs in the region of less than 30 nm particles and then tapers off until the mass increases again as the next mode becomes apparent. This proved to be a similar distribution to the matched adiabatic flame temperature oxy case 3. However, there were some issues with the SMPS results. As one can see based upon the scan of the raw data in Figure 13, the number of counts is many orders of magnitude higher for the ultrafine particles than for the comparatively large submicron particles in the half micron range. Consequently, there were a low number of counts for larger diameter particles. But, due to the higher percentage of mass of these particles, they have a very significant effect on the mass distribution. The result is that when the scans are averaged to give the mass distribution, there is irregularity in the larger particle size range because there are very few particles counted and thus the 10000 £ Q. a c to I I I I I 1 1 1 1 1 1 1 I T i in 0 JPBBHHBBWT"^^ H l f l ( N | f 0 0 1 H C H > } O O r M « ) N H N r O H H i n ( N r O ( n H ( J l < * O O N O O Figure 14. Average air fired mass distribution for 15-660 nm particle diameter. ·. low Air SMPS Distribution 12000 M 10000 E "..'i:io. 8000 c- o 6000 .E "tJ "';; 4000 III nI E "tJ 2000 a Particle Diameter (nm) 41 Figure 15. Sample of case 3 O2/CO2 fired SMPS scan. statistics are not as optimal. In future work, there will be many more scans which are averaged to find the mass distribution rather than the 10 seen here. In Figure 15 it is obvious that the particle distribution is similar to the air case based upon a raw scan of the number of particles and their diameters. While the distributions are very similar, it is apparent that there is a small shift in the peak of the Gaussian distribution of the raw data between Figure 15 and the air fired data in Figure 13. As expected, the similar raw scan produces a similar size distribution to the air fired case, with the shift in the Guassian peak corresponding to a shift in the first modal peak of the ultrafine flume to a particle diameter of a few nanometers larger. The oxy case 3 data is shown in Figure 16. · and' ·3.0 2.5 -U) Q) :;::::: 2.0 II E ~ ! 1.5 a. o Ol a ~ 1,0 z '0 0.5 0.0 10 100 Diameter (nm) 02/C02 1000 42 Case 3 02/C02 20000 mc c - nDp (u - 10000 ass/dli - E i T H i n r s j m < T > T H c r > ^ o o r s i o o r ^ ^ r N i r r i T H r H L n f N r o c n T H C T i ^ r o o f N o o t H T H ( N f N j r \ i m m ^ ^ i n ^ r ^ o o < T ) r H T H , H , ^ f N r N f N r o r n ^ r ^ r L / 1 U 3 Particle Diameter (nm) Figure 16. Averaged SMPS mass distribution of case 3 oxy fired conditions. The discontinuous nature of the larger particle diameters is due to the dilution required for the large number of small diameter particles. The dilution necessary to bring the number of small particles in range reduces the number of large particles that are counted. As a result, subtle differences in the number of larger particles that are counted, and which bin they are placed in can result in a significant shift in mass. This is why collecting a large number of data sets to be averaged is important to produce a continuous curve. In case 4, the oxy enriched attached flame, there was a significant difference in the SMPS results which is clearly evident in the raw data scan shown in Figure 17. This was likely due to the increased flame temperature and higher concentration of particles per m of flue gas due to the lower amount of CO2 injected into the combustion furnace which would facilitate greater coagulation among the smallest particles. 20000 18000 M 16000 E Qo 14000 ::J - 12000 c. c £ "a 8000 'Vi III 6000 III E 4000 "a 2000 0 m3 C02 43 1 1000 Diameter (nm) Figure 17. Sample SMPS scan of case 4 conditions. This is clearly evident in the SMPS scans that were collected. This is a result of the greater vaporization of metals due to the very high flame temperatures (2524 C vs 2121 C) and a more concentrated flume which is clearly evident in the Figure 18 displaying the particle mass related to the particle diameter. This shows a clear departure from the other combustion scenarios and is mostly due to the higher flame temperature and lower flue gas volume rather than the combustion environment of O2/CO2 rather than air. The lower flue gas volume also facilitates greater coagulation and coalescence of vaporized metals and mineral deposits due to a slightly longer residence time in the furnace as well as a greater probability of particle collisions due to their higher concentration. 4.0 3.5 _ 3.0 <0 OJ ~ 2.5 ~ ! 2.0 a. o Cl o 1.5 z~ '0 1.0 0.5 0.0 10 100 1 coo Diameter (nm) 0 2/C02 44 02/C02 Case 4 M 20000 E IRIIIIMRIHIiHHMMflHI ^ L n r ^ m c T > r H a ^ ^ o o r s i o o r ^ - r H r s i r o r H T H L n r s i r o c r i r H a ^ ' ^ o o r \ i o o T H T H r s i r s i r \ i m r o ^ ^ i ^ i x i r ^ o o c T i r H T H T H , ^ r j o J o j r o r n ^ r ^ u l U 3 Particle Diameter (nm) Figure 18. Averaged SMPS mass distribution for higher temperature case 4 oxy fired conditions. 4.3 Photoacoustic Black Carbon Results The PA was utilized at the same time and the same dilution as the SMPS. Due to the very low LOI in all combustion cases at stoichiometric ratio 1.15, there was very little black carbon in the flue gas relative to the amount of fly ash. The goal was to determine whether the carbon in the LOI was related to soot or unburned char. Unfortunately, at the very high dilutions needed to operate the SMPS, the PA was reading at or near its detection limits during the duration of the experiments for air combustion and oxy fired combustion cases. This made it difficult to draw the most rigorous conclusions. However, Figure 19 does show the same general trend found in the LOI experiments in section 4.1. The case 3 scenario has the lowest average black carbon content for stoichiometric ratio of 1.15 while case 4 and air are very nearly the same as also shown in section 4.1. 25000 m bo ..=. 15000 Q. c .E '1:1 10000 "'Vi' III III ~ 5000 0 ...... LI'l .L..I..'l. r........:. N m 0'\ ...... 01 c:i m <.Ci ,....; LI'l N N N m m ~ 00 N 00 ,.... ...... N m ...... ...... LI'l N m 0'\ ...... 0'\ "'" 00 N 00 r...: ..n M M ..n cxi ...... m LI'l ,.... 0 m I.D ...... LI'l ...... ,.... ...... ...... ...... ...... ...... N N N m m "'" "'" LI'l m LI'l I.D "'" "'" LI'l I.D ,.... 00 0'\ P A 45 PA Black Carbon Data ro 00 3 O 00 -E 3 C O -Q i_ (0 u u re 7.00E+03 5.00E+03 4.00E+03 3.00E+03 2.00E+03 1.00E+03 0.00E+00 J l 1J |j 50 100 Low Dilution Air Low Dilution Case 3 02/C02 Low Dilution Case 4 02/C02 Figure 19. Sample of results from photoacoustic experiments with air dilution ratio 88:1. case 3 dilution ratio of 88:1, and a case 4 dilution ratio of 124:1. For all combustion scenarios, the result does indicate that neither soot nor unburned char is a significant contribution to the total particle loads produced by any of the tested combustion scenarios. In future Ph.D. work multiple dilutions and different stoichiometric ratios will be considered to help determine the mechanisms and significance of black carbon in the fly ash. However, this test does appear to be promising as it does yield similar trends to the more traditional gravimetric LOI method. 4.4 Chemical Species Results of Size Segregated Fly Ash The main focus of this work was to determine whether the altered combustion environment under oxy conditions would change the chemical speciation of particles in the 30 nm to 16 micron size range. The following elements were analyzed by ICP-AES: III IU b.O 6.00E+03 Q.I -;;:::::J 0 M ..E..... b.O :::J c 0 ..0.. uIU O.OOE+OO ~ U IU as 0 50 150 PA instrument time interval (6 s) - - - 1 , of88:1 , scenanos, 46 Si, Al, Fe, Mg, K, Ca, Na, Se, and Ar. Arsenic and Selenium proved to be below detection limits in all cases due to the low sample mass available for digestion. Initially it was determined that Si would require a special acid digestion. However at the time this work was being conducted, the University of Utah Physics department provided use of an SEM-EDS which would be used for analysis of the major species. The ICP data were to be used along with the SEM-EDS data to compare the analytical techniques. The results presented in this section are averaged relative weight percent in terms of concentration per cubic meter of standard flue gas as and differential mass as a function of particle diameter. The chemical concentrations were determined using ICP-AES as the SEM/EDS analysis was performed by Dr. Dunxi Yu. The reason that the results are shown as an average mass percent, is that on one of the air fired sample collections a smaller mass of sample was collected on the impactor plates. However, the relative percentage of each measured element was similar to the other air fired samples indicating that the composition was not affected, but just the total mass. As a result, the mass concentration results are displayed utilizing the run from each combustion scenario that did not have any obvious faults. As a result, the case 4 set of samples that had an obvious low plate aluminum contamination and the air fired case of very low mass were discarded for the differential mass based graphs. The two case 3 runs were very similar, and the first run was selected to compare to the other combustion scenarios. Silicon did not completely dissolve in the solution prescribed by EPA Method SW846 3050A indicating the need for the SEM-EDS data for accurate silicon concentration. As a result, it is not possible to discern definitive conclusions from the AI, differential scenarIOs. 40 35 30 .J 3 aj 20 •2 15 %m 10 5 0 Relative wt. % Silicon •Air avg •case avg case avg 4 6 8 Impactor Plate 12 Figure 20. Averaged relative weight percentage of silicon. plot of silicon relative to the other elements as shown in the following figures. This is because the acid digestion may have been more effective on the particles of smaller diameter rather than larger diameter, resulting in a somewhat skewed relative weight percentage. This highlights the need for the SEM-EDS techniques conducted by Dr. Dunxi Yu, a visiting postdoctoral researcher, which are nondestructive and more cost effective and much safer for the researcher. As expected the silicon concentration trails off at low particle diameters which would be part of the fume as indicated in Figures 20 and 21. Where the lack of digestion is more apparent is the larger particles on the top stages where the silicon concentration falls off. This is likely due to the poor digestion of silicon rather than the lack of silicon. In Figure 21, it is possible to see that the total silicon mass increases with particle diameter under all combustion scenarios before falling off towards the last stage of the impactor. This matches the trimodal distribution described as consisting of an ultrafine 47 -21 , *.... 25 ~ .>C.I.I 20 QI..I.jI 15 10 5 0 0 2 10 ..... Airavg _ case 3 avg -.-case 4 avg Figure 21. Differential mass of elemental silicon in terms of mg per standard cubic meter of flue gas. fume, fine fragmentation at 2 microns, and then a larger bulk fragmentation mode which is present at larger than recorded particle sizes (Seames, 2003). The aluminum data show a similarly expected decrease in concentration at low particle diameters due to the high volatilization temperature of aluminum with the exception of the case 4 results. This is suspected to be the result of contamination as it occurred in only one of the sample runs in the case 4 results. However, when that is averaged with the rest of the results it produces obvious outlying data points. The contamination could be from any bit of aluminum or small aluminum particle from one of the blank aluminum substrates on the upper stages of the impactor when being used with the cyclone. The overall concentration of aluminum seems to be consistent or without discernable differences between oxy fired conditions and air fired conditions as indicated by Figure 22. This would indicate that using once through CO2 instead of N2 in the E ::,J rt'I E tia -E c.. c till .2 -~ ~ ~ 100 10 · 1 0.1 0.01 0.01 Silicon 0.1 1 10 Average Particle Diameter (microns) ""-Air Fired _ 02/C02 Case 3 ....... 02/C02 Case 4 48 offlue C02 49 % 70 60 3? 50 i « 40 m oc 30 20 10 0 J • / 'Air avg •8 Figure 22. Average relative weight percentage of aluminum. combustor does not significantly affect the distribution of aluminum in the particles of the size range of interest. Also, when examining the amount of mass of aluminum in the flue gas in Figure 23, the same distribution seems apparent in both the oxy and air fired scenarios. This would indicate that the combustion environment is not significantly affecting how the minerals containing aluminum form particles in the fly ash under the studied scenarios. Iron was not as prevalent a species in the fly ash as some of the others that were examined in this research. As a result, there were certain small particle diameter impactor stages in the air fired and oxy fired tests where the level of iron was below the detectable limits of the ICP performing the analysis and those data points could not be 80 70 60 '* 50 ...: ~ :>QpJ 40 !U iii a: 30 20 10 0 o Aluminum relative wt. -. _. - - -- ~\ \ I ... ;;;; .,.... - I ~ Th ~ ~ i""': ~ ~ I ;r- .1 .L~~ I ~ '\:T I -- 2 4 6 10 Impactor Plate 12 ~A iravg _ Case 3 avg ....... Case 4 avg 50 Figure 23. Differential mass of elemental aluminum per standard cubic meter of flue gas as a function of particle diameter. plotted in the figures. While the percentage of iron in the smallest particles is much higher for the high flame temperature oxy fired case 4, the relative weight percentage of iron appears to be higher across nearly all of the sub micron range for the oxy fired cases than the air fired cases in Figure 24. This indicates that the change in combustion environment may have an impact on the mechanisms that form particles containing iron. The finding is significant since, as discussed in section 1.2, particles containing iron are more likely to form deposits within the furnace. As expected, the overall mass of iron in the fly ash increases in as the second mode of particle distribution becomes apparent in the super micron region. In the super micron region, where the bulk of the ash mass is concentrated, it again appears as though the overall particle distribution is similar between the air fired and the oxy fired E 100 ~ I m E ~ 10 -E Q. C tID 1 J2 -"ll :E "ll 0.1 0.01 Aluminum 0.01 0.1 1 10 Average Particle Diameter (microns) -+-Air Fired _ 02/C02 Case 3 "'-02/C02 Case 4 51 Figure 24. Averaged relative weight percentage of iron in fly ash. conditions. Again, the case 4 combustion scenario is shown in Figure 25 to have a greater concentration of elemental iron per cubic meter of flue gas because there is a lower volume of flue gas due to the high oxygen concentration and reduced injection of C02 . Magnesium is another species which was not present in high enough amounts to be above detectable limits for all samples. Based on the Figure 26, it is obvious that the significantly higher flame temperature of the case 4 oxy fired scenario affects volatility of magnesium compounds in the combustion environment, as the relative percentage of magnesium is a factor of two higher than in the other combustion scenarios. The magnesium proportions are much higher in the higher temperature case 4 flames, but are still higher in the case 3 flame than in the air fired scenario. This indicates that there is a difference in the way magnesium affiliates with particles under 45 40 35 c: .0: 30 ';/!. 25 .. ~ :>Q;I 20 Q"j' 15 a: 10 5 0 Iron Relative Wt % 0 2 4 6 8 10 Impactor Plate 12 -+-Airavg _ Case 3 avg -"'-Case 4 avg 52 3 roi "So Q. a OA _o -o if -a 100 10 0.1 0.01 Average Particle Diameter (microns) •Air Fired •Figure 25. Differential mass of elemental iron per cubic meter of flue gas as a function of particle diameter. The first two points of air fired combustion are omitted due to being below detectable limits and the issues with plotting zeros on a log scale. oxy fired conditions and under air fired conditions even at the same adiabatic flame temperature. Therefore, it may be that the O2/CO2 combustion conditions also affect magnesium volatility and the mechanisms that drive the formation of magnesium compounds found in the fly ash. Magnesium, as shown in Figure 27, also shares a very similar mass distribution to iron while being a bit lower in total mass. The distributions are very similar indicating again that in the bulk mode of the fly ash, the particle behavior is very similar between oxy fired and air fired cases. Calcium is a more volatile element and is obviously a key element of the fume based upon the proportions of calcium found on the lower stages of the impactor in all combustion cases as shown in Figure 28. The calcium concentration is also very high on -E ::J 100 I m E tio 10 -E C 1 til) .2 -"0 ~ 0.1 "0 0.01 Iron 0.01 0.1 1 10 ~AirFired _ 02/C02 Case 3 ...... 02/C02 Case 4 0 2/C02 53 Impactor Stage Figure 26. Relative percentage of magnesium in fly ash as a function of particle diameter. the third stage of case 4, which corresponds with the larger amount of mass seen on the SMPS mass distributions in section 4.2. There is not as great of a change in the average relative weight percent of calcium between case 4 and case 3 combustion scenarios indicating that calcium is sufficiently volatile to be vaporized even at the lower flame temperature of the case 3 combustion scenarios. Also, the calcium proportions, like the concentrations of magnesium tend to be higher in the oxy fired cases rather than the air fired case as indicated in Figure 29. This similarity is plausible due to the similar chemical structure and behavior of both elements being very similar. This indicates that there may be a correlation between metals that may form 2+ ions and their volatility and the mechanisms which affect their proportions in sub micron particles in the O2/CO2 combustion environment. The calcium mass distributions also shared the same trends as most of the other major species. Magnesium Relative Wt % 02/C02 E ::J 'iii Q.I C 110 III ~ '.*.' ~ Q.I .>.. III Qj a:: 14 12 10 8 6 4 2 0 -+-Airavg _ Case3avg f-----------,I---II+----ll--~-------------- -.-Case 4 avg 0 2 4 6 8 10 12 Impactor Stage 54 E 3 i ro E *S5 aEa . o T3 "D 100 10 0.1 0.01 » - • - 0 2 / C 0 2 A Figure 27. Differential mass of elemental magnesium as a function of particle diameter. Data points for low particle diameters are omitted due to being below detectable limits and the issues with plotting zeros on a log scale. 70 6 0 Figure 28. Average relative weight percent of calcium in fly ash. E 100 :::J m E ~ 10 .§. Q. 0 1 til) 0 -;:; ~ 0.1 "a 0.01 Magnesium 0.01 0.1 1 10 Average Particle Diameter (microns) -+-Air Fired ~02/C02 Case 3 ..... 02/C02 Case 4 70 60 50 *.~.-.. 40 Qj > '';:; 30 ta '.i.i. 20 10 0 0 Relative wt. % Calcium 2 4 6 8 10 Impactor Plate 12 -+-Airavg ~Case3 avg ..... Case4avg 55 Average Particle Diameter (microns) Figure 29. Differential mass of elemental calcium per cubic meter of standard flue gas as a function of particle diameter. Sodium is likely the other main component of the fume, which is not surprising given that it is known to be a more volatile metal. This is clearly evident as it is present in very high proportions in the smallest size ranges in Figure 30. The sodium likely coagulated onto larger particles in the high temperature case 4 flame, while it remained a high proportion of the small diameter flume particles and likely contributed to the very large flume region found in the SMPS mass distribution in section 4.2. It is also interesting to note that in Figure 30 there was no sodium in the lower stages of the case 4 scenario, but an increase in relative weight percent in the 340 nanometer range. This suggests that the sodium that is vaporized is condensing on particles and possibly helping to facilitate coagulation. Also of interest is that, unlike iron, calcium, and magnesium, there seems to be a higher weight percentage of sodium in the air fired cases rather than the oxy fired cases. E~ 100 I M E tio 10 -E Q. 0a .o 1 .2 -"a ~ 0.1 "a 0.01 Calcium 0.01 0.1 1 10 -+-Air Fired _ 02/C02 Case 3 ....... 02/C02 Case 4 56 Impactor Plate Figure 30. Relative weight percent sodium in fly ash as a function of particle diameter. This suggests that the behavior of sodium is different under oxy fired conditions rather than air fired conditions. It also suggests that the mechanisms which are affecting relative concentration of iron, calcium, and magnesium affect sodium differently, or are not the mechanisms which govern the formation of sodium compounds in the fly ash. Due to sodium's more volatile nature, its mass distribution varies from some of the other elements examined in the fly ash. It is possible to see the submicron fume mode in Figure 31, and that the mass of sodium decreases earlier in the supermicron mode indicating that the sodium in the ash is more likely to be vaporized to form ultrafine particles rather than be present in the larger supermicron particles as was the case with some of the other elements. 60 50 '.*. 40 3: ;>C II 30 10 Qi a: 20 10 0 0 2 Relative wt. % Sodium 4 6 8 10 12 14 -+-Airavg ~Case3avg ...... Case 4 avg Iron, 31 , 57 E 3 i CO E "ao 1 i Q. Q _o T J T3 1 Average Particle Diameter (microns) Figure 31. Differential mass of elemental sodium in mg per cubic meter of flue gas as a function of particle diameter. Potassium is chemically similar to sodium, but was not present in as high concentrations as sodium in any of the ash. As a result, the potassium was below detectable limits in some cases. However, it is also chemically similar to sodium, and may have exhibited some similar behaviors. Due to the low proportions of potassium in all of the samples, it is hard to make definitive conclusions about the behavior of potassium in the flue gas. In the upper stages, where there is more mass to digest yielding more dependable ICP results, it appears that the same trend of relatively higher weight percentages in the air fired cases rather than the oxy fired cases is similar to the behavior of sodium. However, the fractions of potassium in the ash are so small, as shown in Figure 32, that it is important not to be overly cavalier with conclusions. It also appears in Figure 33 as though there is a correlation between the behavior of the mass distribution of potassium in the flue gas and sodium in the flue gas. This is E::,J t'I'I E Co -E 10 1 0.1 0.01 0.01 Sodium 0.1 10 -+-Air Fired _ 02/C02 Case 3 -.-02/C02 Case 4 IS 58 o a. > Potassium Relative Wt Air avg case 3 avg case 4 avg Figure 32. Averaged relative weight percentage of potassium in fly ash. most certainly reasonable, as the two elements are very chemically similar. Thus it is reasonable to conclude that mechanisms that may influence which particles are enriched with sodium may also affect how potassium compounds are formed and distributed within the fly ash. According to Bool and Helble, the volatility of elements is not only a function of their elemental vaporization point, but also depends on the way in which the element is found in the coal. For K, Na, and Fe, if they are found bound to silicates or as oxides, then these species tend to not be volatile, while if they are found bound in the organic matrix of the coal, or with pyrite minerals, they are considerably more volatile under combustion conditions (Bool and Helble, 1995). This is likely not a significant issue in the observed behavior of Na, K, and Fe, as these elements are likely found in the same relative concentrations in the organic matrix or the ash of the coal, though as future coals are examined, this mechanism may become more apparent. 9 E 8 ::J "0:;; 7 III ' . .r.a. 0 6 Q. 5 ~... 3: 4 Qj 3 " .. ra 2 Qj a:: 1 0 -.... -.... o 2 % . . . ... ./ ~ .(' " I T / 1 ...... ..... I ../ ./ " ./ ".",A r- ~ .... 4 6 8 10 Impactor Stage 12 ~Airavg _ case 3 avg ..... case4avg -------.. ------.--- --------- - - - - 59 100 r 3 Average Particle Diameter (microns) Figure 33. Differential mass of elemental potassium in mg per cubic meter of standard flue gas as a function of particle diameter. Data points that were below detection limits have been omitted rather than posted as zeros. There are many factors that can affect the volatility of the minerals in the coal and the coal ash. These factors are responsible for the governing mechanisms that affect the way in which certain minerals and elements group together and form particles. For particle formation in coal combustion, coalescence is thought to be the most important factor for ash size distribution (Wigley and Williamson, 1998). In addition, this hypothesis was supported and augmented by the assertion that in bituminous coals, coalescence of included minerals and fragmentation of excluded minerals were found to be the most important mechanisms of ash formation (Yan et al., 2001). This is true because the bulk of the mass is in the larger supermicron particles. So as far as mass distribution is concerned, the mechanisms governing coalescence and fragmentation are the most important for the overall mass distribution. 10 1 0.1 0.1 Potassium 1 10 100 ..... AirFired ~02/C02 Case 3 -.-02/C02 Case 4 aI., However, the coalescence mechanism does not explain the changes observed in the submicron particle range. In this range, where volatility of species plays a critical role in ash particle formation, differences were observed under air fired and oxy fired experiments for midvolatility metals such as Ca, Mg, and Fe. This would indicate a change in the volatility of these species in a CO2 enriched environment. It has been shown that reduction of certain species can affect volatility (Senior and Flagan, 1982). In Senior and Flagan's studies, they examined SiC>2, which was reduced by CO near the char particle. In their study, in air fired conditions, the particle is assumed by surrounded by N2 and O2 with CO2 being formed near the particle only by reduction of the minerals by CO. Thus the reduced mineral partial pressure is equal to the partial pressure of the CO2 near the char particle to establish equilibrium (Senior and Flagan, 1982). In oxy fired conditions, the CO2 partial pressure would be orders of magnitude higher, and could thus affect this equilibrium process near the char particle and be responsible for the apparent changes in the elemental concentrations at given particle diameters for air fired and oxy fired cases. 60 "experiments " C02 Si02, 02 C02 CHAPTER 5 CONCLUSIONS 5.1 Loss on Ignition The loss on ignition results show that the burner influenced fluid dynamics play a very important role in determining the char burnout in the furnace. This affects the mixing of the O2 and fuel and, in the case of the oxy fired conditions, a higher oxygen concentration actually resulted in a greater average loss. However, looking at the matched adiabatic flame temperature cases between oxy fired and air fired combustion is a more reasonable comparison as both are detached flames and benefit from greater mixing. It was clear in the figures in section 4.1 that the oxy fired case had lower loss as a function of stochiometric ratio, but that this was not the case when examined as a function of percent oxygen. When looking at the loss as a function of percent oxygen in the furnace or the flue The loss on ignition results show that the burner influenced fluid dynamics playa very important role in determining the char burnout in the furnace. This affects the mixing of the 02 and fuel and, in the case of the oxy fired conditions, a higher oxygen concentration actually resulted in a greater average loss. However, looking at the matched adiabatic flame temperature cases between oxy fired and air fired combustion is a more reasonable comparison as both are detached flames and benefit from greater mixing. It was clear in the figures in section 4.1 that the oxy fired case had lower loss as a function of stochiometric ratio, but that this was not the case when examined as a function of percent oxygen. When looking at the loss as a function of percent oxygen in the furnace or the flue gas, the air fired conditions were actually more efficient with lower loss. This is consistent with other drop tube experiments (Borrego and Alvarez, 2007). This is a significant concern because not only does the full scale plant lose efficiency by using energy to operate an air separation unit, recycle system, and then CO2 compressor, but it also appears that it loses combustion efficiency as well. 5.2 Particle Size Distribution In the experiments performed in this body of work, it would appear that the changes that take place in terms of size distribution are more sensitive in the smaller submicron range and are based on a number of factors. The matched adiabatic flame temperature cases for air and oxy fired conditions have very similar distributions with a few subtle differences in modal peaks that could be driven by the change in the combustion environment. However, there is a very significant difference in the ultrafine region of the high temperature case 4 scenario indicating greater metal vaporization under high temperatures and greater coagulation as particles are more likely to stick together in a high concentration environment with many still molten particles. The larger supermicron particles showed very similar distributions throughout all combustion cases. Since there is not a significant shift in any of the mass distributions from the examined elements in any of the combustion conditions studied, it would seem as though the change in combustion gas environment from O2/N2 to O2/CO2 does not have a great effect on these mechanisms. This is perfectly reasonable as the fragmentation and coalescence of ash deposits within the coal to form relatively large particles is not based on volatility of the minerals, but more how the minerals are composed and included in the ash in the coal. However, there are changes in the proportions of submicron particle distribution. This may lead to a significant concern that more submicron ash is created in oxy fired combustion, and this may require special air pollution control attention, which could significantly increase the costs of retrofit. 62 ·. 02iN2 02/C02 63 5.3 Black Carbon Particle Density Due to the dilution that was run for the SMPS which was taking data simultaneously with the PA, the PA was operating at a very low level near its detectable limits. As a result, the PA data showed that the total black carbon aerosol concentration was on the order of 1 to 5 mg/of flue gas. In the future, due to increasing interest in black carbon particles due to global warming potential, work will focus on how the concentration changes with varying percentages of oxygen in the flue gas. This will be important to discover how much oxygen is needed in order to keep black carbon particulate emissions to a minimum while also keeping the cost for oxygen, which will be sourced from an air separation unit in an actual plant, to a minimum. This must also be balanced with achieving the best char burnout possible in order to maximize the efficiency of the combustion process. 5.4 Effects of CO? on Particle Speciation Volatility of elements and compounds is definitely affected by flame temperature, and this was clearly seen in the results in section 4.4 where many elements were found to be in higher relative concentrations in the submicron vaporized and condensed mode in the high temperature case 4 combustion test than in the lower air flame temperature tests. It can safely be assumed that the composition of the coal is constant, and that any changes in volatility based upon binding of elements within the coal would have an equal effect in all experiments and not be a significant contributor to the relative differences in concentrations of elements as a function of particle diameter in the studied experiments. The results also suggest that combustion environment may be responsible for changes in the relative concentration of elements found in the submicron region. These P A P A black-. m3 CO2 64 particles, whose compositions are more greatly influenced by volatility than coalescence of mineral deposits, showed a pattern in relative concentration in section 4.4. The elements Ca, Mg, and Fe were all found in higher relative concentrations in the oxy fired cases rather than the air fired cases, while K and Na were in relatively higher abundance in air fired cases, and Al was similar throughout. Si could not be conclusively placed into one category or another. The trends suggest that elements with the same type of chemical structure behave similarly, which is perfectly reasonable. However, it is not clear at this point what mechanism precisely is driving the relatively higher concentrations of metals forming 2+ ions and the relatively lower concentrations of metals forming 1+ ions in the oxy fired scenarios. It seems as though the CO2 rich combustion environment does impact the equilibrium formation of certain compounds that form aerosols, but it has yet to be determined precisely how that mechanism operates. Further analysis and thermodynamic simulation would be useful in predicting how significant the changes in combustion environment are in creating mechanisms which cause the discussed elements to tend toward different particle diameters in the fly ash. Now that the experiments have indicated that there may be novel mechanisms in the O2/CO2 environment, future work should also investigate the theory behind these observed differences. 'Kand . 1 + 02/C02 differences. APPENDIX A PROCEDURES A.l Experimental Procedures A. 1.1 Preparatory Procedures A. 1.1.1 Preheating of the Furnace Before the furnace can burn coal, the refractory of the furnace has to be heated up to operational temperature with the use of natural gas. The natural gas burner must be installed on the top of the furnace with the coal feeder safely rolled back from the furnace on its platform. The secondary air line is then connected to the natural gas, which should NOT be flowing. There is a master valve near the natural gas meter which should be in the off position whenever the natural gas burner is not in use. After the burner is installed on the furnace and the gas air line is connected to the burner, the natural gas line should be installed. Then the electrical lead for the spark ignition should be connected followed by the peeper. When the installation is complete and there are no leaks or hazards, the valve next to the natural gas utility meter may be opened, but this will not yet allow flow of natural gas to the burner. The burner is then started using the OPTO 22 control software on the OFC computer. The startup sequence is very important to avoid accidental explosion: APPENDIXA " I I.I I.I.I bum flowing. 66 1. Turn on the secondary air valve to approximately 30% open. 2. Turn on the spark ignition. 3. Verify that the peeper sees a flame. If not, then either the peeper or sparker is malfunctioning. 4. Once OPTO 22 indicates presence of a flame, then open the valve for the natural gas in OPTO. 5. Immediately turn off the spark ignition. 6. Verify that the flame is still present in OPTO not only with the flame indicator but also by checking thermocouple readings in the furnace to make sure there is a rise in temperature. If the flame is not present or the thermocouples do not indicate a rise in temperature immediately turn off the natural gas valve. 7. If the flame is present, adjust the secondary air valve to achieve the desired stoichiometric ratio, usually about 1.3-1.4. This will not be the total stoichiometric ratio, but just the S.R. given for secondary air. Be sure the primary air is set at 10 lbs per hour to keep the plastic eductor cool on the coal feeder. 8. Finally set the peeper chart to on, and the furnace is ready to warm overnight. A. 1.1.2 Flue Gas Analyzer Calibration Every day the gas analyzers must be calibrated. If experiments will be performed both in air fired and oxy fired conditions, the gas analyzers must be recalibrated using standards with a CO2 background rather than an N2 background. For calibrating the gas analyzers follow the following procedures: 1. Make sure all three valves by the standards are to the right selecting air standards or the left for oxy standards. I. Tum Tum tum tum coolon I.I.2 N2 I. 67 2. Make sure the valves behind the flow meters in the control room in the up position for air or down for oxy. 3. For the first O2 analyzer set the flow to 1.3 scfh and the reference air to 1.3 as well. Select dry air and then select calibrate. For the low calibration select low cal on the rotating valve, and follow the instrument prompts. 4. For the ZRH C 0 2 analyzers, set the flow to 2.2. When selecting high cal, select span on the analyzer for each component. Then select low cal on the rotating valve and zero the analyzers. 5. For the thermo environmental NOx analyzer, make sure that the calibrated NO and NOx values correspond to the air or oxy standard as desired. If not, change the set values. Calibrate the zero value with a flow rate of 1.5 on the flow meter. Then calibrate NO and then NOx. Repeat as necessary. 6. For the Horiba 02 analyzer, set the flow to 1.3 and the rotating valve to dry air. Then open and shut the internal valve. Make sure the nitrogen cylinder for the analyzer is on during the day, and shut off at night. Span the analyzer with dry air, then set the rotating valve to zero and zero the analyzer. 7. For the ZRE CO2 analyzer select the 0-20% range for CO2 for air fired conditions and 0-100% for oxy fired conditions and set its corresponding flow meter to 1. Then select low cal on the rotating valve and zero the analyzer. Then select high cal and span the instrument. 8. For the Horiba NO analyzer set the calibration value to the standard's value for NO. Set its flow meter to 1.5 and select low cal on its selection valve. Then zero the analyzer. Then set the valve to high cal and span the analyzer. scth the. CO2 C02 68 A. 1.1.3 Burner Changeover In the morning, the burner must be changed from the natural gas burner to the coal coaxial burner. When changing the natural gas burner to the coal burner the final steps should be performed: 1. Turn off the peeper chart. 2. Turn off the natural gas using the OPTO 22 program on the OFC workstation in the control room. 3. Turn on the wall heaters of the top section of the furnace. 4. Turn off the natural gas valve next to the meter. 5. Now that the natural gas is double checked to be shut off from the burner, the spark ignition lead should be removed followed by removal of the peeper. 6. Remove the secondary air line from the burner. 7. Unclamp the burner from the top of the furnace. 8. Make sure there is no fuel or other flammable substance collected on top of the furnace or in the area where the burner will be stored. 9. Very carefully remove the burner from the furnace making absolutely sure that there is enough insulation in the form of gloves and towels to prevent burns and that footing is completely secure. 10. Very carefully step down the ladder. A helpful trick is to remove the burner from the furnace and carefully set it on its side on the top of the furnace while stepping down the ladder. Make sure the burner tip or the sparker is damaged by rough handling. l.l.3 Tum Tum Tum ofthe Tum bums 69 11. Set the burner in a safe location. This is usually on the grated floor away from foot traffic with the burner tip and sparker not contacting the floor. A prop underneath the burner housing by the peeper attachment is often necessary. 12. Insert the coal burner on the top of the furnace using two bolts to ensure proper alignment of the burner and gasket. 13. The burner does not need to have the bolts fastened and may be clamped into position. This speeds installation and removal and reduces the risk of burns. 14. Once the burner is securely clamped into position, attach the secondary air line, being careful not to over tighten and maintain proper alignment of the union. 15. Roll the coal burner to the edge of the platform. 16. Adjust position of the coal burner so that the feed of coal is almost completely in line with the burner. 17. Connect the hose from the end of the coal feeder eductor to the burner and gently snug the hose clamps. Over tightening will create a rigid connection between the eductor and burner and vibration and thermal expansion will cause errors in the scale reading and coal feed. 18. Begin feeding coal and the requisite amount of air. A. 1.1.4 Berner Low Pressure Impactor Media Preparation Two separate types of filters were used for sample collection in the BLPIs. The filters were both tested for background contamination which can be found in Appendix B. Filter one was an expensive filter which was more difficult to digest, but much easier to handle. It is ideal for taking samples for SEM/EDS analysis. The filter was a Millipore Durapore membrane filter type 0.22 um catalog number: GVHP09050. The filters which necessary . . cOlil · bums. BLPls. urn 70 were most economical and easiest to acid digest were Millipore Isopore membrane filter type 0.4 urn catalog number: HTTP09030. The filters are used on plates where samples shall be collected for analysis. On plates where samples will not be collected when sampling for short times aluminum blanks should be used. When using the cyclone, stages 6-11 should utilize aluminum blanks. As a punch the correct size could not be found that was the same size as the impactor stages, 90 mm filters were cut out using a razor knife on a clean glass background. Sheets of aluminum were cut out in the same fashion. The filters and blanks were then tacked into place on a piece of cardboard and then air brushed with a mixture of Fischer Scientific brand HPLC grade hexane and Apezion H grease. The grease mixture is approximately 15 grams of grease per liter of hexane. A light even coat should be applied to the substrates. The substrates should then be dried in an oven at no higher than 50 degrees Celsius for 12 hours and then placed in a desiccator overnight before use. This procedure was adapted from Seames' dissertation in 2000. The next day the substrates and blanks may be used to collect samples with the BLPI. A. 1.2 Testing Procedures A. 1.2.1 Beginning Firing with Coal After the coal burner has been installed and the furnace has been preheated with natural gas, it is time to begin firing with coal. 1. Set the wall heaters to 1850 F using OPTO controls. 2. Set the secondary air preheater to 550 F. . · l.2 l.2.1 Fusing 71 3. Begin feeding 5 lbs per hour of coal with 10 lbs per hour of primary air and approximately 45 lbs per hour of secondary air. 4. When the temperature of the walls and in the upper section of the furnace is steady, usually after 10 to 15 minutes, increase coal feed rate to 7.5 lbs per hour and secondary air to approximately 70 lbs per hour for another 10 to 15 minutes. 5. Then increase coal feeding to 10 lbs per hour and the secondary air to 95.8 lbs per hour. 6. Verify that on the manual control module of the coal feeder that the motor rpm is very near 1,000 rpm indicating a feed rate of 10 lbs per hour. If it is significantly different, such as more than 10% off, then adjust the coal feed rate with OPTO accordingly until the indicated rpm reads 1,000. 7. Allow the furnace to achieve steady state with a flue gas oxygen concentration of as near 3.00% as can be achieved before beginning testing. A.l.2.2 Firing Air Fired Conditions Air fired conditions are given by a stoichiometric ratio of 1.15 with a coal feed rate of 10 lbs per hour. By following the procedures in A. 1.2.1, the air case should be A.l.2.3 Firing Q2/CQ2 Fired Conditions In order to fire oxy fired cases, the furnace must be preheated with natural gas, and then changed over to coal and air. After the furnace is running at a consistent temperature under air fired conditions with 10 lbs per hour of coal feed, it is possible to switch to oxy fired conditions. If time permits, it is more advantageous to bring the Sibs increase' feed' . different, I.I.2.1, ready to test. I.02/C02 72 furnace to steady state temperatures under air fired conditions as it is significantly cheaper than using the CO2 and O2 from their liquid storage tanks. 1. Verify the furnace is at steady operating temperature under air fired conditions. 2. Turn all gas analyzer valves to "Dry Air" during the transition. If not, there is a risk of ruining the zirconium oxide element in the Yokogawa oxygen analyzer. 3. Turn off the coal feeding. 4. Close the air valves using OPTO controls. 5. Open the main valves for the outside storage CO2 tank and O2 tank. These valves are located adjacent to the respective tanks. 6. Inside the combustion lab, first open the CO2 valve and then close the air valve. 7. Open the O2 valves for both primary and secondary air lines. 8. Using the OPTO controls, open the CO2 valves for the primary and secondary air lines. Begin feeding 12.2 lbs per hour primary CO2 and 62 lbs per hour of secondary C02 . 9. Using OPTO controls open the O2 valves for the primary and secondary air lines. Begin feeding 2.3 lbs per hour of O2 in the primary stream and 22.2 lbs per hour of O2 in the secondary stream. 10. Resume coal feeding at 10 lbs per hour. 11. Calibrate the high range CO2 analyzers and recalibrate the NOx and NO analyzers for the CO2 environment. 12. Return the gas analyzer valves to "Sample" and begin tuning CO2/O2 and coal feeding to achieve desired combustion conditions. IS ·Tum Y okogawa Tum 02 C02 C02 CO2 . 02 C02 C02 C02/02 73 A. 1.2.4 Sampling Fly Ash The furnace pressure is usually operated very slightly positive relative to the ambient atmosphere on the order of .10" of water equivalent pressure. However, for safety concerns, the damper of the furnace should be adjusted to be very slightly negative while inserting the sampling probe. This prevents ash and other toxic flue gases from being blown out of the sampling port towards the samplers. Usually -.10" of water is sufficient. 1. Verify using OPTO controls that the furnace is reading a negative pressure relative to ambient. 2. Make sure that the sampling probe is clean and contaminant free by forcing compressed air through all orifices. Use acetone or isopropyl alcohol to remove particularly dirty deposits. 3. Remove the sample port flange and the insulation. 4. Insert the probe required for the given type of sample to be collected. 5. Using the same four nuts and bolts for the flange, fasten the sampling probe in place. 6. Adjust flow rate with mass flow controller or rotameter to achieve the necessary dilution and or flow through the probe to maintain as close to isokinetic as possible. A.l.2.5 Operation of the SMPS Research engineer Dave Wagner with the University of Utah should be contacted for instrument service and operator training. The SMPS is operated by the TSI Aerosol Instrument Manager software. order to take data with the SMPS, the PC controlling the instrument must be turned on and the AIM software launched. The software will l.2.4 . · In 74 prompt for several inputs before beginning. The inputs affect the set up of the instrument and are as follows: Classifier Model 3080 DMA Model 3081 DMA Inner Radius(cm) 0.00937 DMA Outer Radius(cm) 0.01961 DMA Characteristic Length( cm) 0.44369 CPC Model 3022 Low Flow Gas Viscosity (kg/(m*s)) 1.82E-05 Mean Free Path (m) 6.65E-08 Channels/Decade 64 Multiple Charge Correction TRUE These should always be used when sampling with the current SMPS instrument. Also save the file with a recognizable file name including the date. Now, with the main screen in view, it is possible to collect scans. Click on the green button with the cursor and a 135-second scan commences. Verify the Gaussian distribution of the curve. If there is an unusual shaped distribution or scattering of data, verify that sample is flowing through the probe and that there are no clogs and the instrument is set up properly. It was found that due to small errors in the N2 calibration curve along with minor perturbations in the combustion process that the theoretical isokinetic flow rates of sample through the probe of 0.8 to 1.5 1pm were often so small that could be affected by the error range of the controllers and the minor disturbances in the furnace. Therefore, was determined that the dilution N2 injected into the probe should be such that the lpm it 75 difference between N2 in and sample out should be 1.5-3 1pm in order to achieve consistent results and enough vacuum so that the probe draws in sample even if there is a very slight fluctuation of mass flow through the combustor. A.l.2.6 Operation of the Photoacoustic Analyzer The PA is operated by the PC as the SMPS. However, it uses a Labview Virtual Interface to control the instrument. After opening the program, it is possible to give the text file a name. The name should obviously indicate the date and reflect the test conducted. Each scan should be given a different name to avoid confusion, as otherwise each successive scan will be appended to the previous text file. This can create massive confusion trying to determine which line is the end of one test and beginning of another. The PA will record data as long as it is turned on and controlled by the Labview VI. The data however will not be saved until clicking on the switch in the top left hand corner with the cursor. At this point, the tab delimited text file will be created and updated with a new data point every 6 seconds which can later be analyzed in Excel. Before collecting sample data, it is important that the PA is zeroed out. order to zero the PA, place a HEPA filter on the inlet of the PA. Wait a few minutes until the PA is reading consistently, and then select manual zero on the interface screen. Leave the HEPA filter in place until the PA consistently reads zero. Remove the filter and reconnect the PA to the dilution manifold. When the PA is connected to the dilution manifold, check the noise level on the Labview VI to verify that there is not interference from the eductor or any other equipment. The noise level should be on the order of a 1-5 ug/m . If not, begin trouble shooting the instrument. Research engineer Dave Wagner lpm P A comer In HEP A P A uglm3 • 76 with the University of Utah should be contacted for instrument service and operator A. 1.3 Posttesting Procedures A.l.3.1 Analytical Equipment Maintenance The PA and SMPS are very expensive instruments and should be treated with care. The SMPS should not be moved unless necessary, and the n-butanol reservoir should be drained before the instrument is moved. Again, research engineer Dave Wagner with the University of Utah should be contacted for instrument service and operator training. If used in a dirty environment, such as the combustion lab, the PA, SMPS, and dilution manifold should ideally be covered and staged in the cleanest area