Oxy-Firing Tests in a Simulated Process Heater

Oxy-firing tests were performed in a simulated process heater as part of a range of activities sponsored by the 3rd phase of the CO2 Capture Project (CCP3). The work was performed at the test facility of John Zink Company in Tulsa, Oklahoma. The objective of the tests was to evaluate the feasibility of utilizing commercial process heater burners for oxy-firing, and to compare burner performance with conventional air-firing. Two different vintages of John Zink low-NOx burners were used in the tests, and two fuel types (natural gas and a simulated refinery fuel gas) were fired. The burner performance under oxy-fired conditions was very similar to that for air-firing; however, small changes in the heat flux profile were observed. The most striking observation was that the heater took a long time (a few hours) to reach a steady nitrogen content in the flue gas ( 15% - 20% range on a wet basis). The high leakage into the firebox was an artifact of higher vacuum levels due to the presence of a boiler upstream of the ID fan to re-circulate the flue gas into the oxidant stream, which created significant backpressure which the fan had to overcome. Under oxy-firing conditions, NOx emissions decreased approximately 50% and 70% for the COOLstar and PSFG burners, respectively. Further reductions were not possible due to the availability of nitrogen in the test furnace. The effect of leakage was lower when the test furnace was operated at lower negative pressure; however, the amount of air leakage will be highly dependent on how well the furnace is sealed.

Page 1 of 10 Oxy-Firing Tests in a Simulated Process Heater Jamal Jamaluddin1 Cliff Lowe2, Nick Brancaccio2 Jaime Erazo3, Charles E. Baukal, Jr.3 Rasik Patel4 1Shell Global Solutions (US), Inc. 2Chevron Energy Technology Company 3John Zink Co., LLC 4On-Quest Abstract Oxy-firing tests were performed in a simulated process heater as part of a range of activities sponsored by the 3rd phase of the CO2 Capture Project (CCP3). The work was performed at the test facility of John Zink Company in Tulsa, Oklahoma. The objective of the tests was to evaluate the feasibility of utilizing commercial process heater burners for oxy-firing, and to compare burner performance with conventional air-firing. Two different vintages of John Zink low-NOx burners were used in the tests, and two fuel types (natural gas and a simulated refinery fuel gas) were fired. The burner performance under oxy-fired conditions was very similar to that for air-firing; however, small changes in the heat flux profile were observed. The most striking observation was that the heater took a long time (a few hours) to reach a steady nitrogen content in the flue gas ( 15% - 20% range on a wet basis). The high leakage into the firebox was an artifact of higher vacuum levels due to the presence of a boiler upstream of the ID fan to re-circulate the flue gas into the oxidant stream, which created significant backpressure which the fan had to overcome. Under oxy-firing conditions, NOx emissions decreased approximately 50% and 70% for the COOLstar and PSFG burners, respectively. Further reductions were not possible due to the availability of nitrogen in the test furnace. The effect of leakage was lower when the test furnace was operated at lower negative pressure; however, the amount of air leakage will be highly dependent on how well the furnace is sealed. Introduction CCP is an international effort funded by six of the world's leading energy companies. For the past ten years, this project has been addressing the issue of reducing greenhouse gas emissions in a manner that will contribute to an environmentally acceptable and competitively priced continuous energy supply for the world. The project objective is to develop new technologies to reduce the cost of capturing CO2 from combustion sources and safely store it underground. This concept is commonly referred to as carbon capture and sequestration or CCS. These CCS technologies will be applicable to many of the large point source CO2 emissions around the world - such as power plants and other industrial processes. Implementing these new technologies during this decade will reduce the impact of continued fossil energy use while cleaner energy sources are being developed. In the first phase of the CO2 Capture Project (CCP1, 2000-2003), it was determined analytically that oxy-firing of refinery heaters and boilers showed significant potential for lower CO2 avoided costs when compared to post-combustion capture1. By using high purity oxygen instead of air for combustion, the flue gas from oxy-firing contains higher concentrations of CO2, thus reducing downstream purification costs. Since combustion occurs, notionally, in an air-free environment, NOx formation is expected to be drastically reduced - which is an added benefit of oxy-firing. In CCP3, the John Zink Company was selected to conduct oxy-fired testing on two of their conventional process heater burners, an SFG staged gas low NOx burner and a COOLstar® Ultra-Low NOx burner. Oxy-Firing Tests in a Simulated Process Heater Page 2 of 10 2012 American Flame Research Committee Meeting (Salt Lake City, UT) Oxy-firing for CCS Traditionally viewed as financially unattractive and operationally challenging for years, oxy-firing has come back to life on account of its promise as a means for CO2 enrichment, separation and sequestration. A large body of laboratory and pilot scale studies, primarily focusing on oxy-firing of pulverized coal, has established that oxy-firing is technically feasible, and switching from air-firing to oxy-firing can be achieved smoothly, and typically in less than an hour under scenarios simulating a coal-fired boiler. A 30 MWTh demonstration project by Vattenfall at Schwarze Pumpe, Germany, has been in operation since 2008, and has been an excellent success story for oxy-firing in the power generation industry2. More recently, one of four 30 MWe units at the Callide power station in Queensland, Australia has been converted to oxy-firing (without storage at this point)3. Oxy-firing, of course, has its own set of challenges. The obvious one is that the adiabatic flame temperature is significantly higher than in air-fuel combustion (about 5000°F or 2760°C, compared to about 3500°F or 1930°C for air-fuel combustion). To ensure integrity of the heater tubes and other hardware components internal to the heater, an adequate quantity of flue gas needs to be re-circulated back through the burner. Both desktop studies1,4 and pilot scale tests have shown that heat fluxes and temperature levels similar to those in an operating heater (or boiler) can be achieved by optimizing the recirculation of flue gas and adjusting the concentration of oxygen in the mixed oxidant. As the air is replaced by a mixture of pure oxygen (typically 97%+ purity) and flue gases, the flammability envelope is affected, and stability of the burner flame may become a matter of concern. Adequate monitoring, and proper IPF (instrumented protective function), will need to be in place to alleviate such concerns in an operating heater. There are additional concerns around handling of oxygen, which is a highly potent gas that is known to cause fires, as well as explosions in the extreme, if proper safety measures are not in place. Objective of the Tests The objectives of this program were to:  Assess the feasibility of utilizing conventional process heater burners for oxy-firing.  Confirm feasibility by conducting single burner oxy-fired testing with flue gas recirculation. The test program was divided into five key tasks: 1. Heater performance modeling - to determine changes in heater performance (efficiency), maximum film/tube temperatures, heater draft, and radiant/convection heat absorption ratio, and the flue gas recycle requirements. 2. Modify the selected test furnace for oxy-firing - to plan, procure and install the equipment and instrumentation needed to run oxy-firing tests with flue gas recirculation, and prepare for the added measurements related to burner oxidant-side pressure drop, heat flux, temperature effects, flame dimensions, etc. 3. Computational fluid dynamics evaluation - to evaluate combustion performance, flame shape, and heat flux distributions for the two burner designs in the test furnaces and in typical process heaters. 4. Single burner oxy-fired testing - to conduct a full range of tests with air (base case) and oxygen with two different fuels. 5. Computational fluid dynamics re-evaluation - to modify and rerun the CFD model developed in Task 3 to conform to the results obtained from Task 4 so that it can serve as a workable predictive tool. Oxy-Firing Tests in a Simulated Process Heater Page 3 of 10 2012 American Flame Research Committee Meeting (Salt Lake City, UT) In this paper, only Tasks 1, 2 and 4 are covered. Heater Performance Modeling The use of oxygen in direct fired process heaters will have a twofold impact on heater design: firstly, the operating parameters will change to maintain the same heater duty; and secondly, the operation of the unit will be more complex due to the additional equipment. For the heater modeling simulations, two process heater geometries were considered: vertical cylindrical and horizontal box. The process side heat absorbed duty was held constant in all of the cases studied. To understand the full effect, several cases were modeled both with and without flue gas re-circulation at varying levels of excess oxygen. From the initial baseline "as is" simulation with ambient air, the model was re-run with different oxy-firing conditions to evaluate the impact on:  Overall heater efficiency,  Maximum film temperature and/or tube metal temperature limitations,  Duty split between the radiant and convection sections, and  Heater draft. The trend results for both heater geometries were similar, and as such, only details of the vertical cylindrical heater are presented here. The refinery fuel gas used in the simulations contained 30% hydrogen, while the oxidant stream composition used was typical of commercial grade 97% pure oxygen. In order to get a comparative set of results, the heater simulations with oxy-firing were performed with the objective of meeting the following constraints of the original heater specifications:  Total heater absorbed duty = 39.9 x 106 Btu/hr (11.7 MW)  Maximum allowable film temperature = 750°F (400°C)  Maximum allowable tube metal temperature = 975°F (524°C) The first set of simulations was performed with varying levels of excess oxygen in the firebox without any flue gas re-circulation (FGR). As can be seen in Figure 1, without FGR a high level of excess oxygen (>70%) is required to meet the limitations of the design allowable film temperature. 735 740 745 750 755 760 0 10 20 30 40 50 60 70 80 90 100 Temperature, °F Excess Oxygen, % (dry) Film Temperature (VC Heater - 100% Oxygen Air - No FGR) Calculated Film Temperature Max. Allowable Film Temperature 920 930 940 950 960 970 980 990 0 10 20 30 40 50 60 70 80 90 Temperature, °F Excess Oxygen, % (dry) Tube Metal Temperature (VC Heater - 100% Oxygen Air - No FGR) Calculated TMT Design TMT Figure 1 Effect of excess oxygen on film and tube-metal temperature, without flue gas recirculation. Oxy-Firing Tests in a Simulated Process Heater Page 4 of 10 2012 American Flame Research Committee Meeting (Salt Lake City, UT) 736 738 740 742 744 746 748 750 752 0 10 20 30 40 50 60 70 80 Temperature, °F Excess Oxygen, % (dry) Film Temperature (VC Heater - 100% Oxygen Air -With FGR) Calculated Film Temperature Max. Allowable Film Temperature 920 930 940 950 960 970 980 0 10 20 30 40 50 60 70 80 Temperature, °F Excess Oxygen, % (dry) Tube Metal Temperature (VC Heater - 100% Oxygen Air -With FGR) Calculated TMT Design TMT Figure 2 Effect of excess oxygen on film and tube-metal temperature, with flue gas recirculation. In order to minimize the oxygen requirements, the second set of simulations was performed using FGR. The firebox excess oxygen was varied from 30% to 0.5% (wet basis). For each excess oxygen case selected, the amount of flue gas recirculation (FGR) was varied until the process constraint on film temperature was satisfied. The simulations showed that the %FGR required as measured by the ratio of flue gas re-cycled to the total flue gas flow increases as the excess oxygen is reduced. For example, at 2.0% excess oxygen (wet basis) the FGR required was 72% while the corresponding oxygen content in the oxidant stream entering the burner was 22%. Figure 2 shows that the film and tube-metal temperatures remained below the design allowable for all excess oxygen levels considered. The results also showed no appreciable change in firebox or stack temperatures over the cases with conventional air. The main conclusions of the heater simulations were:  The results of the two heater case studies indicate that amount of oxy-firing and FGR is strongly dependent on the service. For example, where film temperature is critical, this will determine the minimum level of FGR required for a given firebox excess oxygen.  As expected an improvement in heater efficiency is seen with oxy-firing.  Due to a reduction of flue gas flow rate, draft is not impacted. Description of the Test Furnace and Burners Two different burners were tested. The PSFG is a low NOx diffusion flame burner incorporating staged fuel injection to reduce NOx 5. The COOLstar® burner6 is an ultra-low NOx diffusion flame burner incorporating staged fuel injection and internal furnace gas recirculation to reduce NOx emissions. The burners were fired in the test furnace vertically upward. The test furnace is rectangular in shape with internal dimensions 13 ft (4 m) wide, 7 ft (2.1 m) deep and 31 ft (9.4 m) tall. The furnace is cooled by single pass water tubes to simulate the heat absorption by the process and control the flue gas temperature. Sample ports are located at 18 in. (46 cm) intervals along the vertical centerline of the furnace. Velocity thermocouples for flue gas temperature measurement are located on the floor, half way up and at the top of the furnace. The test furnace does not have a convection section that extracts heat from the flue gas. Therefore, an atmospheric boiler was used to cool the flue gas after it had exited the furnace. A fan downstream of the boiler was used to pump the flue gas through the system. After exiting the fan, the flue gas was exhausted through a stack to the atmosphere. Figure 3 is a schematic diagram of the system setup. For natural draft operation, all of the flue gas was exhausted through the stack while the air was drawn through an inlet. The recycle damper was closed to prevent flue gas from entering the air stream. For oxy-fired operation7, the recycle damper was partially open to re-circulate flue gas to the burner. The air inlet was closed and Oxy-Firing Tests in a Simulated Process Heater Page 5 of 10 2012 American Flame Research Committee Meeting (Salt Lake City, UT) * Arrow Indicates Direction of Flow Duct Work Stack Air Inlet Boiler O2 Injection Burner Fan Recycle Damper Furnace Figure 3: Simplified schematic diagram of oxy-fire setup. the source of oxidant was the injection of oxygen into the re-circulated flue gas stream. A liquid oxygen storage tank with a vaporizer and flow control skid was used to deliver the oxygen to the system. During oxy-fired operation, the oxygen injection was controlled with a digital controller. The input variables were oxygen concentration at the stack and in the oxidant stream (post-injection) on a wet basis. Oxy-firing Tests Two of the oxy-fired operating conditions considered for the heater performance modeling studies were tested. The difference between the two operating modes was the oxygen concentration in the oxidant stream to the burner. An ambient air, natural draft case was also tested for comparison. The oxy-fired condition with oxygen concentration of 22.2 %(vol) in the oxidant will be referred to as Oxy-fire A, while the lower oxygen concentration of 20.6 %(vol) will be referred to as Oxy-fire B. Two fuels were tested, a natural gas fuel (Tulsa Natural Gas or TNG) and a representative refinery fuel gas (RFG) mixture (50% TNG, 25% propane, and 25% hydrogen). The two operating conditions and test fuel properties are summarized in Table 1. Table 1: Summary of Operating Cases and Fuel Properties. Parameter / Units Natural Draft Oxy-Fire Heat Release (MMBtu/hr) Maximum 7.68 6.45 Normal 6.4 5.37 Minimum - PSFG 2.56 2.56 Minimum - COOLstar 1.92 1.92 Oxidant Oxygen Concentration (%,v) 20.9 22.2 / 20.6 Recirculated Flue Gas (%) --- ~ 71 Oxygen Concentration in Flue (%,vd) ~ 3 5.2 / 1.3 Oxidant Temperature (°F) Ambient ~ 500 Fuel Tested - Natural Gas & RFG Both Both Fuel TNG Fuel RFG Molecular Weight 17.15 Molecular Weight 20.11 LHV (BTU/scf) 913 LHV (BTU/scf) 1104 Component % vol Component % vol Methane 93.4 TNG 50 Ethane 2.7 Propane 25 Propane 0.6 Hydrogen 25 Butane 0.2 Nitrogen 2.4 Carbon Dioxide 0.7 Total 100 Total 100 A description of the different test points conducted is provided in Table 2. This test procedure is similar to the burner test procedure described in API-5358. At maximum heat release for both natural draft and oxy-fire operations, the flame length was verified with a CO probe and the incident heat flux profile9 was Oxy-Firing Tests in a Simulated Process Heater Page 6 of 10 2012 American Flame Research Committee Meeting (Salt Lake City, UT) measured. The flame boundary was defined as a time-averaged concentration of 2000 ppmvd of CO. Emissions measurements of CO, NOx, wet and dry O2 were made at the exit of the test furnace. The oxygen concentration on a wet basis was measured before and after the oxygen injection. Dry measurements of O2 and CO2 were made just before the burner inlet so the nitrogen content in the oxidant stream could be determined by mass balance. Table 2: Description of test points. Test Point Description Design - Maximum Design heat release CO Breakthrough Increase flow rate until CO > 250 ppmvd or flame instability Normal Normal heat release Minimum Turndown per normal burner pressure drop oxidant flow Minimum Turndown with controlled oxygen concentration in furnace Absolute Minimum Fuel pressure < 0.5 psig if possible, flame instability or excessive CO The burner pressure drop was measured using pitot static probes located just upstream of the burner inlet. The oxidant temperature was also measured just upstream of the burner inlet using a Type K thermocouple. Each burner had a thermocouple welded to the flame-holder and staged gas tip. Direct observation and digital photography was used to record flame appearance. Some air leakage was expected in the system as the pressure drop through the system varied while operating under different modes. Under certain conditions the vacuum at the floor of the test furnace would reach ~ -5 inH2O (120 mmH2O), which is much greater than what is designed for in the field. As an estimate, a hypothetical open area in the floor of the furnace was calculated to determine the air leakage into the system under various tested drafts. Once this area was determined, the air leakage and expected nitrogen concentration in the flue gases was calculated using draft values more typical of refinery furnaces. During the testing, nitrogen concentrations of approximately 15-20% (vol) were measured at design conditions with a floor draft between 2-4 inH2O (51 - 102 mmH2O). Calculated nitrogen concentrations with the same equivalent open area at reduced draft was approximately 6-10% (vol). Results and Discussion This was a complex test with numerous objectives. This paper focuses on some of the test results pertaining to the process burner performance under oxy-fired conditions. The burner flame appearance, flame length, NOx emissions and heat flux profile will be compared and discussed. Flame Appearance Photographs comparing the flames from the PSFG and COOLstar® burners are shown in Figures 4 and 5. The comparison in this example is between natural draft and Oxy-fire A operation with RFG. The observable differences in flame appearance from both burners between natural draft and Oxy-fire A operation were minor. Some differences in the color of the burner tile of the COOLstar® burner were observed. During Oxy-fire A operation, the zones where the fuel gas from the ignition ports of the gas tips combusts were larger and slightly brighter than the natural draft case. The flameholder skin temperature on the COOLstar® burner was higher for Oxy-fire A operation than with natural draft operation. Under Oxy-fire A conditions, the concentration of oxygen in the oxidant stream was higher than that of ambient air operation. These results suggest that the combustion of the ignition fuel under oxy-fire conditions was more intense, thus producing higher local temperatures. Similarly, the flameholder on the PSFG appeared to be brighter in color. Oxy-Firing Tests in a Simulated Process Heater Page 7 of 10 2012 American Flame Research Committee Meeting (Salt Lake City, UT) A - PSFG, RFG, Natural Draft B - PSFG, RFG, Oxy-fire A Figure 4: Photographs of PSFG burner flame at normal heat release under natural draft and Oxy-fire operation. A - COOLstar®, RFG, Natural Draft B - COOLstar®, RFG, Oxy-fire A Figure 5: Photographs of COOLstar® burner flame at normal heat release under natural draft and Oxy-fire operation. Flame Length The change in flame length for each burner on RFG under oxy-fire conditions compared to the natural draft baseline is shown in Figure 6. In the comparison between the natural draft and Oxy-fire A operations, the PSFG burner flame length showed very little change, whereas the COOLstar® burner flame length decreased by approximately 20%. Both burners showed an increase in flame length relative to the natural draft operation when firing at Oxy-fire B conditions. The increase in flame length under Oxy-fire B conditions compared to Oxy-fire A conditions is also shown in Figure 6. The comparison between the two oxy-fired conditions is interesting because the difference in flame length is large, while the difference in the oxygen concentrations of the oxidant stream is small. Oxy-Firing Tests in a Simulated Process Heater Page 8 of 10 2012 American Flame Research Committee Meeting (Salt Lake City, UT) -20% -10% 0% 10% 20% 30% 40% 50% Flame Length Difference (%) PSFG and COOLstar® Burner Flame Length Comparison PSFG COOLstar Oxy-Fire A vs. Natural Draft Oxy-Fire B vs. Natural Draft Oxy-Fire Bvs Oxy-Fire A Figure 6: Comparison of flame length change from PSFG and COOLstar® burners under various operating conditions. Heat Flux Profile The incident heat flux profile was measured along the vertical centerline of the test furnace. Figure 7 shows a comparison of the COOLstar burner firing RFG under natural draft and oxy-fire conditions. The peak heat flux occurs at the same elevation for both conditions; however, the ambient air firing case produced a higher absolute heat flux compared to the oxy-fire firing case. This is attributed to the difference in peak flame temperature between the two flames. The calculated adiabatic flame temperature of the ambient air flame is approximately 340°F (190°C) greater than that of the oxy-fire operation. 0 5 10 15 20 25 30 4 5 6 7 8 9 Furnace Elevation (ft) Absolute Incident Heat Flux (BTU/(ft2-sec)) Incident Heat Flux Profile: COOLstar® Air - Oxy-fire Comparison with RFG COOLstar - RFG Air COOLstar - RFG OxyFire-A Figure 7: Comparison of incident heat flux profiles of COOLstar burner firing RFG under natural draft and oxy-fire conditions. Oxy-Firing Tests in a Simulated Process Heater Page 9 of 10 2012 American Flame Research Committee Meeting (Salt Lake City, UT) NOx Emissions The relative difference in NOx emissions, corrected to 3% O2 (vol, dry), for each burner with RFG under oxy-fire conditions compared to the natural draft baseline, is shown in Figure 8. There was a significant reduction in NOx emissions under oxy-fire conditions from both burners compared to the natural draft baseline. There was a much smaller difference in NOx emissions between the two oxy-fire conditions. The reduction of NOx emissions was expected as the oxy-fire system operates at a much lower nitrogen concentration. A complete purge of nitrogen was not possible as there were portions of the system operating under vacuum, causing air to leak into the system. At design conditions, a nitrogen concentration of approximately 15-20% (vol) was present. Normally, the radiant section of a process heater operates at a negative pressure, which causes air to leak into heaters through any openings. 0% 10% 20% 30% 40% 50% 60% 70% 80% NOx Reduction (%) PSFG and COOLstar® Burner NOx Reduction Comparison PSFG COOLstar Oxy-Fire A vs. Natural Draft Oxy-Fire B vs. Natural Draft Oxy-Fire Bvs Oxy-Fire A Figure 8: Comparison of NOx reduction from PSFG and COOLstar® burners under various operating conditions. The difference in NOx emissions between the two oxy-fire conditions was greater for the PSFG burner than for the COOLstar® burner. The COOLstar® burner uses multiple techniques for reducing NOx and generates less NOx than the PSFG burner. The larger improvement in NOx for the PSFG burner is expected because it starts from a higher baseline level compared to the COOLstar® burner. Conclusions The oxy-firing tests at John Zink Company confirmed that conventional process heater burners can be employed for oxy-firing, and that combustion and heat transfer performance similar to air-firing can be achieved. The principal learning from the tests, however, was the difficulty in reducing the nitrogen levels in the heater. Although process heaters in refineries do not operate at nearly as high a draft as was maintained during the tests at John Zink, refinery heaters are more prone to air leakage. Achieving less than 10% (wet basis) nitrogen concentration in a refinery heater (especially natural draft heaters) presents a challenge. Excess nitrogen in the system will adversely affect the competitiveness of oxy-firing as a CCS route since the costs of downstream CO2 purification will increase. Design considerations such as positive pressure operation, where possible, can help make oxy-firing more viable, as it has with the boiler industry. Acknowledgements The authors wish to thank the CO2 Capture Project for their guidance and financial support. Oxy-Firing Tests in a Simulated Process Heater Page 10 of 10 2012 American Flame Research Committee Meeting (Salt Lake City, UT) References 1Allam, Rodney, et al; Volume One: Capture and Separation of Carbon Dioxide from Combustion Sources; Elsevier 2005; page 451-475. 2Stromberg, Lars, Plenary Lecture, 2nd International Oxyfuel Combustion Conference, Yepoon, Queensland, Australia (September, 2011). 3Spero, Chris, Plenary Lecture, 2nd International Oxyfuel Combustion Conference, Yepoon, Queensland, Australia (September, 2011). 4Jamaluddin, Jamal, Farmayan, Walter and Shu, Shu, Paper presented at the 2nd International Oxyfuel Combustion Conference, Yepoon, Queensland, Australia (September, 2011). 5C.E. Baukal (ed.), The John Zink Combustion Handbook, CRC Press, Boca Raton, FL, 2001. 6I-P Chung, D. Meinen, R. Poe, J. Lewallen and C. Baukal, Solving the low NOx dilemma, Hydrocarbon Engineering, Vol. 10, No. 8, pp. 77-80, 2005. 7C.E. Baukal (ed.), Oxygen-Enhanced Combustion, CRC Press, Boca Raton, FL, 1998. 8API Standard 535: Burners for Fired Heaters in General Refinery Services, 2nd Edition, issued January 2006, American Petroleum Institute, Washington, DC. 9C.E. Baukal, Heat Transfer in Industrial Combustion, CRC Press, Boca Raton, FL, 2000.