Low Nox Burner Development for High Temperature Reforming and Cracking Furnaces

LOW NOX BURNER DEVELOPMENT FOR HIGH TEMPERATURE REFORMING AND CRACKING FURNACES By RICHARD MARTIN and DALE NICKESON JOHN ZINK COMPANY Tulsa, Oklahoma ABSTRACT A new industrial process burner that is designed specifically for firing in high temperature reforming and cracking furnaces with minimum formation of NOx during the combustion process has been developed by John Zink Company. A 3 mm Btu/hr prototype burner was built by John Zink Company and tested in John Zink's test facilities in Tulsa, Oklahoma. Tests with the prototype burner using 500° F air preheat and burning fuel gas produced emissions having less than 0.05 pounds of NOx per million Btu fired. The NOx level is well below the 0.4 pounds per million Btu fired reported on some units firing in the field. The burner design is based on several years of development work in areas such as sub-stoichiometric combustion and the destruction of NOx in process streams. INTRODUCTION Energy production consists of the burning of fuels that are, with small exception, of fossil origin. The burning of fossil fuels and the subsequent production of NOx has become a matter of concern to the industry. First, NOx has an adverse effect on the environment and second, there is a growing body of regulation dealing with the emission of NOx into the atmosphere. Because of the complexity of regulations, in many cases, it is no longer possible to establish an exact acceptable emissions level for a specific heater. Often, the user must consider the total NOx emissions from the plant as opposed to fixed levels for a specific heater. He is now faced with achieving NOx levels on specific units that were considered impossible in the recent past. An additional concern is the finite supply of fossil fuels. Because of the limited supply, it is of absolute necessity to conserve. One of the most effective ways to conserve fuel is via air preheat, but this has proven difficult because preheated air combustion normally increases the quantity of NOx produced. In the Refining and Petrochemical Industries, a major source of NOx is the process heater. Many heaters, such as reformers and cracking furnaces, require high flue gas temperatures in the radiant section of the heater to produce the required high process temperatures. When a combustion process takes place in a high temperature environment as opposed to a "cooler" environment, the formation of NOx in the combustion process is increased. In addition, heaters of this type have other definite requirements. For example, a uniform heat flux to the process tubes is of utmost importance. To provide the required uniform heat flux, the burners must have well defined and confined flame patterns. Higher NOx levels are normally associated with burners that produce compact, well defined flames as opposed to burners that produce flames of lower intensity. Because of the unique requirements of these high temperature heaters, many NOx control techniques have limited applications. For example, the low theoretical flame temperatures that result from flue gas recirculation may limit the use of this technique. The lower flue gas temperatures reduce the heat flux rate in the high temperature zone, and in some cases may make it extremely difficult to achieve the high process temperature required. Another unique requirement, as previously discussed, is the well defined flame pattern. Certain forms of offstoichiometric combustion that require combustion air to be introduced to the combustion zone at a point remote from the burner are unacceptable because of the flame pattern deterioration. As in the case with flue gas recirculation, certain forms of off-stoichiometric combustion may not be acceptable. Any form of NOx control utilized on the types of heaters discussed must allow the stringent requirements of heat flux, flame patterns, and high temperature combustion to be achieved, and, in addition, must have the capability of utilizing air preheat. The most desireable form of NOx control for the above application is a burner that is designed to meet the requirements of flame pattern and high temperature combustion utilizing preheated air, yet limit the formation of NOx in the combustion process. Unfortunately, known "Lo NOx" burners that have been developed up to this time do not achieve the high NOx reduction required for these high temperature applications. For example, the type of heaters discussed may have NOx levels in the flue products in excess of 300 PPM corrected to 3$ 0^. These units seldom have NOx levels of less than 150 PPM. However, NOx levels in the flue products of as low as 100 PPM have been recorded when turbine exhaust gas is used for combustion air. At this time, it is unclear whether the 100 PPM level was corrected for the dilution effect caused by using turbine exhaust gas for the combustion oxygen source. The actual NOx level is, of course, dependent on factors such as fuel, air preheat, excess air, and, certainly, burner design. It is obvious that a NOx level in the flue products of 300 PPM is unacceptable. In many cases NOx levels of 100 PPM corrected to 3% 07 in the flue products, if this level has actually been achieved, is unacceptable. The solution to the problem is the development of a "Lo NOx" burner that may be required to achieve a six-fold reduction in NOx while operating on preheated air to meet the demands for energy conservation, and still maintain all of the desirable features normally associated with burners that are used to fire high temperature heaters such as cracker furnaces and reformers. BACKGROUND RESEARCH Development of such a burner obviously requires an understanding of the parameters that contribute to the reduction of the formation of NOx in the combustion process. Methods, such as flue gas recirculation and staged combustion, have been used with varying success depending on the actual requirements. Flue gas recirculation, of course, provides lower theoretical flame temperature, but, in addition, the recirculation of flue gas, because of the dissociation of carbon dioxide and water vapor, provides an environment in the initial combustion zone that is rich in carbon monoxide and hydrogen. The same environment is achieved in staged combustion. The process permits the initial stages of the combustion reaction to take place with a deficiency of oxygen. Staged combustion has long been established as an effective means of reducing NOx emissions. Since both flue gas recirculation and staged combustion have established the effectiveness of an environment rich in reducing agents, an obvious choice for a burner design is one that limits the amount of oxygen present in the primary or inital combustion zone, and effectively forms carbon monoxide and hydrogen. For this type of staged combustion to be effective, an equivalence ratio must be established that maintains an appropriate concentration of reducing agents in the oxygen deficient zone, and also maintains sufficient residence time in the oxygen deficient zone. John Zink Company has been engaged in development work for a number of years to determine design parameter such as practical equivalence ratios that produce CO and hydrogen in sufficient quantities for NOx reduction, but also limits the formation of soot, the time of formation for maximum carbon monoxide and hydrogen concentrations, and the time required in the reducing zone to limit the formation of NOx. John Zink manufactures combustion equipment whose primary function is to produce a gas high in reducing components for certain chemical processes. These reducing gases are produced by burning fuel with a deficiency of oxygen. Even though a theoretical equilibrium concentration in the combustion products can be calculated for combustion taking place with a deficiency of oxygen, John Zink conducted a series of tests in 1976 to determine the actual composition of combustion products for various sub-stoichiometric operating conditions. Also of interest was the onset of soot formation. The goal was to operate with as high of an equivalence ratio as possible without significant soot formation. The series of tests was conducted in a refractory lined vessel that had a residence time for stoichiometric conditions of approximately 40 milliseconds. As the amount of combustion air was reduced, the residence time in the reducing zone was increased. The tests were conducted using methane, propane, and Number 2 fuel oil. The fuel-air mixing was of an intense nature. The end results, especially with regard to soot formation, are of course highly dependent on the mixing capability of the burner device. As already implied, the purpose of the test was to determine the concentration of hydrogen and carbon monoxide that could be achieved, and the extent of soot formation. The time of formation and the quantity of carbon monoxide and hydrogen formed is, as previously discussed, an important factor to be considered in "Lo NOx" burner design. High concentrations of hydrogen and carbon monoxide in the primary combustion zone will significantly reduce the level of formation of NOx and, in the case of high nitrogen fuels, act as a reducing agent for any NOx formed in the initial stages of combustion. Figures 1 and 2 show the concentration of carbon monoxide and hydrogen for various oxygen conditions when burning methane. The theoretical equilibrium concentrations for each component is shown as a dashed line. Excessive soot formation in the combustion process will, of course, cause extensive fouling of the heat transfer surfaces of the furnace. It is, therefore, of great importance that, when a combustion process is operating with a deficiency of oxygen, careful consideration be given to the level of soot formation. In the discussed series of tests, it was determined that an essentially soot free reducing gas can be generated by operating at sub-stoichiometric conditions for the above mentioned residence time. The unit was operated with equivalence ratios approaching 1.7 with minimal soot formation. The actual particulate level was less than 0.02 grains per standard cubic foot of flue products. HYDROGEN CONCENTRATION IN WET PRODUCT GAS AT DIFFERENT BURNER LOADS, FIRING ON NATURAL GAS % of Stoichiometry CARBON MONOXIDE CONCENTRATION IN WET PRODUCT GAS AT DIFFERENT BURNER LOADS, FIRING ON NATURAL GAS FIGURE 2 -7- In addition, it was demonstrated that, with proper burner design, stability is not a problem for extremely sub-stoichiometric conditions. The tests demonstrated that a burner can be designed to produce a reducing atmosphere that will limit the production of NOx in the combustion process without the formation of carbon particulates. The design of a burner capable of extremely low NOx production is highly dependent on knowing the optimum reducing zone equivalence ratio. John Zink Company has spent a considerable period of time determining the optimum reducing zone equivalence ratio for fuels ranging from natural gas, to #6 oil, to high nitrogen shale oil. Again, a refractory lined vessel that permitted a reducing zone residence time of from 50 to 100 miliseconds, depending on the firing rate and primary zone equivalence ratio, was designed. Additional combustion air was introduced at the exit of the vessel to complete combustion. The NOx levels were measured after combustion was completed. The testing established the optimum reducing zone equivalence ratio for the configuration tested of between 1.4 and 1.7 depending on the fuel fired. Typical data is shown in Figure 3 for the combustor with four different methods of air injection. It should be noted that the equivalence ratios were established with mixing configurations that maintain flame quality. If the flame quality is allowed to deteriorate, the optimum primary equivalence ratio, with respect to NOx formation, increases. While most of the test work was conducted with liquid fuel, it is felt that sufficient data was gathered to predict a satisfactory reducing zone equivalence ratio for gas firing. A third series of tests was conducted to determine the effect of increasing the residence time in the reducing zone on NOx formation. A series of tests were conducted using two refractory lined vessels. The first vessel had a length to diameter ratio of one. For the equivalence ratio tested, the residence time in -8- EQUIVALENCE RATIOS FOR VARIOUS BURNER CONFIGURATIONS - Configuration 1 Configuration 2 Configuration 3 1.0 0.8 Configuration 4 o 0.6 a 1 0.4 0.2 0.6 0.8 10 1.2 1.4 1.6 1.8 2.0 Primary-Combustion Zone Equivalence Ratio FIGURE 3 -9- the reducing zone was approximately 35 milliseconds. A second vessel was designed that was dimensionally similar to the first except the length of the reducing zone was doubled. The second vessel had a combustor with a length to diameter ratio of two, giving a residence time in the reducing zone of approximately 70 millliseconds. Additional combustion air was introduced at the exit of the chambers to complete combustion. Again, the system was designed to give flames with good definition. The reduction in NOx varied from 17% to 30%. The highest percentage reduction occurred with natural gas firing, although the greatest reduction in terms of parts per million occurred with high nitrogen fuels. The results are shown in Figure 4. Results from recent development work on a "Lo NOx" boiler for high nitrogen fuels indicates that additional residence time will give additional reduction, provided the flue gases are maintained at an elevated temperature. The flue gases then must be cooled and combustion completed at a relatively low temperature to prevent the re-formation of the oxides of nitrogen. It should be noted that, if the combustion is taking place in a device in which heat is transferred from the flue gases, the final air for oxidation must be introduced and mixed with the reducing gases at a point where the temperature is still high enough to give complete combustion, and minimize the formation of soot. One other method of NOx reduction that can be considered in the burner design is the injection of steam into the combustion zone. The injection of steam provides a small amount of flame cooling which may give some reduction in NOx levels. However, it is believed the major portion of the reduction achieved with steam injection is caused by the increased partial pressure of reducing agents in the primary -10- EFFECT OF RESIDENCE TIME IN REDUCING ZONE ON NOx LEVELS <tf° /°" / Fuel Oil with • ^ ^ 0.2 and <3% by wt. ^ ^ • \ ^ ^ ^ Nj in Fuel - - ^ • 100 0 Z-o 90 a c •o o __ o JrJ 80- ^ \ ^ ~ ^ - ^ ^ v CL 0) • = Natural Gas 70 ) ^^^**~~ , ~^^^ ^"^^. ^^*^*** --^^"^^ ^v. ^"•>. 22 u. _ c S 605 5 50z "«2 >» % 40^ a I 30o 20- 1010 20 30 40 50 60 70 Residence Time in Reducing Zone in Milliseconds FIGURE 4 -11- combustion zone resulting from the dissociation of steam to hydrogen and oxygen. John Zink has conducted several different test programs to determine the most efficient method of injecting steam and the results of steam injection on the formation of NOx for various burner configurations. For gaseous fuel firing, it appears the most effective means of injecting steam is by mixing with the fuel gas. For a test conducted using this method of injection, it was determined that the NOx levels for a standard burner, as opposed to a burner designed to reduce NOx emissions, can be reduced to approximately 50% by injecting a quantity of steam equal to 20 to 30% of the fuel weight. The actual reduction will vary depending on the burner design and the level of NOx in the flue products prior to steam injection. Figure 5 shows the results of steam injection for a natural draft burner firing natural gas. The above information, as a minimum, provides a starting point for the design of a burner to achieve the NOx reductions required for the high temperature cracking furnaces and reformers. DESCRIPTION OF EQUIPMENT The test burner was a flat flame burner designed for firing adjacent to a refractory wall. Although the burner was designed with the capability of firing both liquid and gaseous fuels, most of the data gathered was for natural gas firing. This is consistent with the requirements for the burner since most of the furnaces of the type discussed are gas fired. The burner design provided for three different methods of staged air injection. The purpose was to determine the most effective point of air injection within the mechanical design constraints of the burner. The first method of staged air injection, or Configuration #1, admitted -12- m THE EFFECT OF STEAM INJECTION ON NOx LEVELS ;• FIGURE 5 J -13- the staged air at the level of combustion product discharge from the burner block on the side opposite the refractory wall. The second configuration permitted the air to be injected at the point of combustion product discharge from the burner block on the side adjacent to the wall. The third configuration provided for injection of the staged air adjacent to the wall approximately five inches above the point of combustion product discharge from the burner block. Figure 6 shows sketches of burner Configurations #1 and #2. The background information previously given was used to design the burner. The burner was designed for an equivalence ratio of approximately 1.4 in the reducing combustion zone. In addition, the staged air injection ports were designed to begin mixing with the staged air after approximately 50 milliseconds of sub-stoichiometric combustion. Configuration #3 provided a slightly longer residence time in the reducing zone prior to staged air injection. 50 milliseconds was chosen because it was felt that, based on the previous work, a satisfactory reducing atmosphere could be achieved within that period of time, and some reduction of NOx formed in the initial stages of combustion sould be reduced. It should also be noted that, for the proposed burner design, heat transfer will occur from the reducing zone gases, as opposed to no heat transfer from the reducing zone gases for the test configuration from which the data was gathered to determine the effect of reducing zone residence time on NOx levels. The heat transfer, of course, results in lower ultimate flame temperatures, and may alter the expected results. The burner was tested in the #5 test furnace at the John Zink Research Facilities in Tulsa, Oklahoma. Figure 7 shows the type furnace and gives pertinent dimensions. The furnace is designed specifically for -14- LO-NOX BURNER CONFIGURATIONS "8 Staged Air Discharge - Wall 1 , F ...... \ 0 \ Wall Staged Air Gas Tip 3 Primary Air S BURNER CONFIGURATION 1 tS \ 'v 7^ O Staged Air Discharge Primary Air BURNER CONFIGURATION 2 FIGURE 6 -15- SCHEMATIC OF NO. 5 TEST FURNACE at John Zink Company-Tulsa -*-3- Probe for Gas Sample Temperature Steel Shield 6" Refractory Window Ji i~2 Platform %//wjmr/jMm/////jmm>m Stairs FIGURE 7 -16- testing radiant wall type burners. The furnace is refractory lined, and the firebox is approximately 8 feet by 6 feet by 16 feet tall. Water tubes are located on the wall opposite the burner to provide a heat sink for temperature control. The furnace is operated under slightly negative pressure, with draft provided by a stack approximately 24 feet tall. The sample connection for flue gas sampling and a thermowell for measurement of stack gas temperature are located below the stack damper in a refractory lined section of the stack that prevents any substantial additional heat loss after the gases exit the radiant section of the furnace. The firebox is of welded construction to provide minimum infiltration of air from the atmosphere. There are several view ports in the firebox to permit observation of the burner and burner flame characteristics. All of the burner tests were conducted using fresh air elevated to the desired preheat temperature. Most of the tests were conducted with the combustion air temperature maintained between 500° F and 550° F. Preheated air was provided by a fired shell and tube heat exchanger. It is important that tests of this type be conducted with fresh air. In the past, in some cases, the elevated air temperature has been obtained by direct firing of a burner into an air stream. Direct firing in the air stream can have the same effect as recirculating flue gas, and can give erroneously low NOx levels. The stack was continuously monitored for excess 0p» NOx, and CO. The flue gas sample was pumped to an instrument room containing the NOx, 0«, and CO analyzers. The sample stream was dried prior to measurement. The water was removed by passing the sample through an ice bath and a drop-out bottle. The sample was then passed through a desiccate drier for final removal of any residual water vapor. The -17- NOx measurements were made with a Thermal Electron Chemiluminescent Analyzer. The CO was measured with an infrared meter, and the 0 2 was measured with a Teledyne oxygen analyzer. Standard calibration gas was used to calibrate all meters, and each meter was periodically checked to verify calibration. The flue gas temperature was continuously monitored at the exit of the radiant section. The temperature of the furnace environment, in which the burner is operating, is a prime factor in determining the final NOx level. It was desired to duplicate the same temperature environment that is encountered in actual field operation. Windbox pressure and furnace draft were measured with two slant tube water monometers. Figure 8 shows the general arrangement of the test furance, air preheater, and measurement equipment. TEST RESULTS As discussed, the burner was designed to fire either liquid or gaseous fuels. However, the initial series of tests concentrated on gaseous fuel firing, since most units now operating are gas fired. Each configuration was tested with an air preheat temperature of approximately 500°F. In each test, the performance of the burner was documented at three heat releases. The burners were first tested at a heat release of 3mm Btu/hr, with 3% 0 2 and 2% 0 2 in the flue products. The configurations were then tested at a heat release of 2 mm Btu/hr, with 3% 0 2 and 2% 0 2 in the flue products. Finally, the three configurations were tested at a heat release of 1 mm Btu/hr. The minimum excess oxygen in the flue products that could be supplied at 1 mm Btu/hr was 8%. The excess air was the minimum air flow that could be supplied without over heating the combustion air preheater. -18- I ..... \0 Thermocouple I Sample line Stainless Steel Air Control Damper • Fuel Air Healer 'TI G') C lJ m Q) -- -- --- - ---- - - - -- The first series of tests was conducted with Configuration #1 which injected the staged air from the side of the burner away from the refractory wall. As was the case for all tests, the temperature of the flue gases exiting the radiant section was maintained at approximately 1800°F and the air preheat temperature was 500°F. The NOx emission for Configuration #1 when firing 3 mm Btu/hr with 3% 0 2 in the flue products was 87 PPM corrected to 3% 0^ The NOx level dropped to 78 PPM corrected to 3% 0 2 when firing at 3 mm Btu/hr when the excess oxygen in the flue products was reduced to 2%. The second firing rate for Configuration #1 was 2 mm Btu/hr. The NOx level was again approximately 87 PPM for 3% 0 2 firing. The level was slightly lower with 2% 0 2 in the products than experienced when firing 3 mm Btu/hr. The measured level was 73 PPM corrected to 3% °2When Configuration #1 was fired at 1 mm Btu/hr, the NOx level in the flue products increased. The measured levels ranged form 96 PPM to 103 PPM corrected to 3% 0 2 - As previously stated, the burner was operated at a minimum excess air dictated by the air preheater operation. The 07 in the flue products for this minimum excess air was 8%. The burner was designed to operate with an equivalence ratio of 1.4 in the reducing zone with an excess air of 10%. There was no means provided for adjustment of the equivalence ratio when the excess air was other than 10%. Because of the high excess air condition at minimum firing rate, the equivalence ratio in the primary (reducing) combustion zone could not be maintained. The primary combustion zone actually had a slight excess of oxygen thereby accounting for the increased NOx levels that were measured when the burner was firing at a minimum rate. -20- Next, the configuration of the burner was changed to Configuration #2 to allow injection of the staged air from the side of the burner adjacent to the wall. For the Configuration #2 burner, the air is injected at the level of combustion product discharge from the secondary tile. The NOx levels were considerably lower than levels measured when testing Configuration #1. When the burner was firing at 3 mm Btu/hr with 3% 0 2 in the flue products, the measured NOx level was 49 PPM corrected to 3% 0 2 > Again the preheated air temperature was maintained at 500° F, and the temperature exiting the radiant section was approximately 1800° F. The firing rate of 3 mm Btu/hr was maintained and the 0 2 was reduced to 2% in the flue products. The NOx level for this condition was 43 PPM corrected to 3% 0 2 > Next, the firing rate was reduced to 2 mm Btu/hr. With this firing condition, the NOx level was 54 PPM with 3% 0 in the flue products and 46 PPM with 2% 0 2 in the flue products. The NOx level for Configuration #2 firing at 1 mm Btu/hr was in excess of 120 PPM. Again the 0 2 in the flue products could not be reduced below 8% and the explanation of the high NOx level is the same as discussed for Configuration #1. The third configuration was similar to the second configuration except the air injection point was approximately 5" above the secondary tile. Again the air was directed vertically between the flame and the radiant wall. The combustion air preheat was 500°F and the flue gas temperature in the radiant zone of 1800° F was maintained. When the burner was fired at a rate of 3 mm Btu/hr with 3% 0 2 in the flue products, the NOx measured was 44 PPM. When the 0 2 level in the flue products was decreased to 2%, the NOx level decreased to approximately 41 PPM. The same test conditions were repeated for a heat release of 2 mm Btu/hr. For this case, NOx levels were slightly higher. The level measured for 3% 0 2 in the flue products was 49 PPM, and the level for 2% in the flue products was 43 PPM. -21- NOx VARIATION WITH EXCESS AIR FOR 3MM BTU/HR. FIRING RATE 80 Configuration "1 70" Configuration *2 60CM O 8* CO 50- "O £ 2 40- Configuration ^3 ha o > Q x O 30- Q. 20-0. 10- % 0 2 in Dry Flue Products FIGURE 9 -22- NOx VARIATION WITH EXCESS AIR FOR 2 M M BTU/HR. FIRING RATE 80 Configuration^ 70Configuration ^ 2 60CM O 9« CO 50- "O o Configuration **3 2 40o O X 30- O 8: 20- 10- % 0 2 in Dry Flue Products FIGURE 10 -23- Again the level measured for the firing rate of 1 mm Btu/hr with high excess air was near 120 PPM. The results for the tests are shown in Figures 9 and 10. A final series of tests was conducted in an attempt to achieve even lower NOx levels. The three burner Configurations, as previously discussed, were all tested under exactly the same firing conditions as discussed except a small amount of steam was premixed with the fuel gas. The result was an additional decrease in NOx. Each configuration was tested with steam injection rates of 0.05, 0.10, and 0.15 pounds of steam per pound of fuel. For the first configuration, the steam injection rate of 0.15 pounds of steam per pound of fuel reduced the NOx level for the 3 mm Btu/hr firing rate from 87 PPM to 64 PPM when firing with 3% 0 2 in the flue products. When the same configuration was tested with 2% 0 2 in the flue products, the NOx level was reduced from 76 PPM to 60 PPM corrected to 3% 0 2 . When the second configuration was tested with a steam injection rate of 0.15 pounds of steam per pound of fuel, the NOx level was reduced from approximately 50 PPM to 38 PPM. When the 0 2 was reduced to 2% in the flue products, the steam injection resulted in similar reductions. When the final configuration was tested, the measured NOx level, when firing at 3 mm Btu/hr with 3% 0 2 in the flue products, was reduced from 43 PPM to approximately 36 PPM. When the oxygen in the flue products was reduced to 2% 0 2 , the steam injection reduced the NOx level to 35 PPM. In addition to the above, all three configurations were tested for a heat release of 2 mm Btu/hr at 3% and 2% 0 2 in the flue products with similar steam injection rates. The results of the steam injection tests are given in Figures 11 and 12. As expected, the percentage reduction was not as great as achieved when steam injection was -24- 100 90~ 80 I ~ 70 ... ... 60 C'? 0 N <.TI I '0 ------ -- ~- N 0 - - -...r.'L ~-- -- ---- - - --------0 ------- • CD U ......CD 0 50 ... 40 0 >0 ~ __ r- - - - - -= =-= -=- -=-t=---_-- -- -- -- --- -- --Q-_ - - - --...= -=- -=-----=-.:1 )( 0 z 30 :IE C1. C1. 20 () Configuration 1 - 3" 02 .2"02 6 Configuration 2 - 3" 02 10 • 2"02 o Configuration 3 - 3" 02 • 2"02 , .05 .1 0 .15 Steam Injection Into Fuel #/# ~ ~L- _ _______ _____ ______________________________________________________ ~ 100 90 ~ '-. 80 N 0 ae I N M I 0 - en -... '0 Q) U Q) ... 70 ...>0 ................ 60 ----- -- - - -{!r - - - - - - --~ &----- - - - . - - - - - - - - - - --.,.,.-- ~- ----- --~---50 ~ - - - - - - - - -{!}-- - - - - ~ ~ h • 0 0 ~ 40 - - - - -- - - -- - - - -.- - - - - -. - - -- - -- --"0 ~ _ --~----==.::.--& - - - - - -a- - - - - - - - -=---=i )( 0 z 30 :e Q. Q. 20 o Configuration 1 - 3'1. 02 .2%02 [::, Configuration 2 - 3"4. 02 10 .2%02 Configuration 3 - 3% 02 • 2%02 o .05 " .10 .15 Steam Injection Into Fuel #/# C> C JJ ...mNL-______________________________________ -=~----------------------------~ applied to a standard design burner. However, a significant additional reduction can be achieved with steam injection, with the magnitude of the reduction depending on the burner design and the level of NOx present in the flue products prior to the steam injection. CONCLUSIONS A significant reduction in the NOx level for a burner that can be used in reforming and cracking furnaces was demonstrated in the test program. The NOx levels were reduced from values measured in field tests of from 150 PPM to in excess of 300 PPM to a level of approximately 40 to 45 PPM. In addition, it was demonstrated that steam injection can further reduce the NOx levels in this type of burner. Additional reduction for steam injection ranged from 15 to 35% depending on the burner configuration and the NOx levels produced prior to steam injection. Some additional testing was conducted with fuel oils that are typically fired in radiant wall heaters. The preliminary tests gave NOx levels ranging from approximtely 62 PPM to 80 PPM for the configuration with the air injected between the tile and the wall. These levels were measured with an air preheat temperature of 500° F. The temperature of flue gases exiting the radiant section were 1800° F. Since the first objective of the test was to determine the reductions possible for gaseous firing, only spot checks were made for fuel oil firing. Future testing is planned for the fuel oil firing case. In conclusion, it has been demonstrated that a low NOx burner can be designed that will significantly effect the NOx produced in radiant wall type heaters. The burner meets the criteria of maintaining -27- a well defined flame pattern that provides uniform heat flux to the process, and maintains the high combustion temperature that is required for satisfactory operation of the radiant section of the furance. This development is a significant advancement in the efforts to achieve minimum NOx levels in industrial combustion processes. -28-