Development of a modular supplemental burner for optimization of distributed generation/combined heat and power system efficiency

Increased interest in distributed power generation has lead to increased interest in combined heating cooling and power (CCHP). One significant issue in implementing a CCHP is that electrical loads and heating or cooling loads are rarely synchronized. Duct burners are frequently used to provide additional heat when the waste heat available from the prime mover does not meet the needs of the heat recovery device. Duct burners are required to operate on vitiated oxidizer streams which can result in poor stability, poor emissions, or both. Rich-burn, quick-mix, lean-burn (RQL) style combustors have been shown to provide low emission and high stability in lean gas turbine applications. An experiment was performed to assess the merits of using a RQL style combustor in a duct burner application. Several burner configurations were tested and it was found that an RQL combustor showed greatly improved stability over a single stage combustor when operating under vitiated conditions without adversely effecting emissions.

2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 Development of a Modular Supplemental Burner for Optimization of Distributed Generation/Combined Heat and Power System Efficiency Elliot-Sullivan Lewis, Richard Hack, and Vincent McDonell 1 UCI Combustion Laboratory University of California Irvine, CA 92697-3550 1 Abstract Increased interest in distributed power generation has lead to increased interest in combined heating cooling and power (CCHP). One significant issue in implementing a CCHP is that electrical loads and heating or cooling loads are rarely synchronized. Duct burners are frequently used to provide additional heat when the waste heat available from the prime mover does not meet the needs of the heat recovery device. Duct burners are required to operate on vitiated oxidizer streams which can result in poor stability, poor emissions, or both. Rich-burn, quick-mix, lean-burn (RQL) style combustors have been shown to provide low emission and high stability in lean gas turbine applications. An experiment was performed to assess the merits of using a RQL style combustor in a duct burner application. Several burner configurations were tested and it was found that an RQL combustor showed greatly improved stability over a single stage combustor when operating under vitiated conditions without adversely effecting emissions. 2 Introduction Combined cooling, heating, and power (CCHP) on the local level (i.e., "distributed generation or DG") has the potential to increase the overall energy efficiency of industrial and commercial operations. While this attribute of CCHP has been long recognized, one of the significant issues in implementing such a system is that electrical, heating, and cooling loads are rarely synchronized in time. The ability to add flexibility to the ratio of heat to power for a given system can increase the utilization and overall efficiency of such a system. To this end, duct burners can be used to provide additional heat when the waste heat available from the prime mover does not meet the needs of the heat recovery device. Duct burners are widely employed in industry to add heat to the exhaust stream of a gas turbine engine either to provide additional heat for steam generation water heating or for some other on site process. 1 Corresponding Author, 949 824-5950, x121, [e-mail] 1|Page 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 Typically these duct burners employ the non-premixed combustion of either natural gas or fuel oil, stabilized behind a simple bluff body. While simple to implement, this burner design lends itself to relatively high emissions of nitrogen oxides. With increasingly stringent emissions regulations, the contribution of emissions from the duct burner can be a significant limiting factor. However, the development of low NOx duct burners presents a somewhat unique challenge to the combustion engineer. In particular, the exhaust gas from a typical gas turbine engine has a reduced concentration of oxygen when compared to standard air. Reduced oxygen concentration will result in inherently poor burner stability. Poor stability, at best, can result in an insufficient burner turndown ratio and, and worst, an inoperable burner due to flame blow off or high carbon monoxide emissions. These are some of the reasons why simple non-premixed systems make good sense. They offer high operability, but with the potential associated penalty of high emissions. To attack the emissions challenge, some industries such as industrial burner OEMs and gas turbine OEMs have widely adopted lean premixed combustion strategies (e.g., Lefebvre, 1999; Ragland and Bryden, 2011). But in light of the stability challenges these systems already face (e.g., Richards, et al., 2001; Lieuwen, et al., 2008), evolving them to deal with a vitiated environment can pose further challenges. Alternatively, "rich-burn, quick-mix, lean-burn" (RQL) style combustors have been shown to provide low emissions and high stability in overall lean gas turbine applications (e.g., Lefebvre, 1999). Current Pratt & Whitney aero gas turbine engines achieve NOx levels that are thirty percent of the current emissions requirements by using RQL type combustors while simultaneously providing wide stability limits and an absence of combustion oscillations (McKinney 2007). The fuel-rich portion of the flame located upstream provides necessary stability and limiting the formation of NOx due to a deficit of excess oxygen. The downstream lean zone facilitates the burnout of CO while maintaining low flame temperatures, avoiding significant formation of NOx. As a result, two obvious advantages to developing an RQL style duct burner are (1) improved stability, and (2) low emissions production rates. A possible third advantage is the potential for the fuel-rich portion of the burner to promote the re-burning of NOx and actually reduce the emissions concentration at the exit of the system. 3 Review of Previous Work As pointed out, one of the key challenges associated with duct burner development is the need to operate in a vitiated environment. Some of the earliest work dealing with the effects of combustion in vitiated environments like those found in gas turbine exhaust gas was performed by Mullins (1949). In this work, calculations were performed to determine the adiabatic flame temperature of kerosene-air flames for differing inlet oxygen compositions. It was determined that the adiabatic flame temperature is reduced significantly for fuel-rich flames, and is relatively unaffected in fuel-lean flames at a given air/fuel ratio, though vitiation does increase the air/fuel ratio required for stoichiometric combustion. It was also determined that increasing vitiation is detrimental to the fuel-air mixing process due to the increased volume through which the fuel must be distributed. Ignition delay was found to increase inversely with oxygen composition. Visual inspection of the flame found that increasing the length of the flame and renders it less luminous. Mullins found that increasing vitiation narrows the stability limits. Below 15% oxygen, combustion could not be sustained at any air/fuel ratio. It was found that 2|Page 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 decreasing the oxygen percentage to 16% produced the same effect on stability as a decrease in temperature by 100 degrees C. Later work that dealt with the effects of vitiation on combustion was most often for gas turbine afterburner applications. One such experiment was carried out by Miles (1971) and studied the effects of vitiation on the lean and rich blowout of premixed propane-air flames in a well stirred reactor. Vitiated air was provided by a propane fueled slave burner upstream of the well stirred reactor, so inlet reactant temperature increased with increasing vitiation from 294 K to 650 K. The approach to compensate for the different inlet temperatures was to first determine the effect of the inlet temperature on blow off characteristics and then to incorporate the relationship into the vitiated correlation. Miles (1971) found that, for a given equivalence ratio, an increase in vitiation of 10% had the same effect on blow off as increasing the air loading factor (a normalized reactant flow term) by a factor of 2.6. It was also observed that all blow off data for propane and vitiated air can be correlated when the term is plotted vs. equivalence ratio. Where G is the air flow rate, v is the fraction vitiation, T is the temperature, V is the reactor volume, P is pressure and b is temperature roughly proportional to and is constant at a given inlet temperature. This study suggested that a similar behavior pattern could be expected for the blowoff of different fuels at varying levels of vitiation. More recently studies pertaining to combustion in vitiated environments have focused on the effect of NOx formation. While exhaust gas recirculation (EGR) has long been used in industrial combustion systems as a means to help reduce NOx, it is now being considered in systems like gas turbines as an option to further reduce NOx in those systems. Two such studies were carried out by Fackler, et al. (2011) and Ditaranto (2009). Fackler et al. (2011) performed a study to understand how the addition of CO2 and N2 to methane-air flames affects the NOx emissions from lean premixed flames. It was observed that for a fixed firing temperature, that increasing the diluent concentration resulted in an increase in NOx emission index. That, is when increasing the inlet concentration of nitrogen or carbon dioxide there is an increase in the mass of NOx produced per mass of fuel consumed. It was found that diluting the fuel with nitrogen was more effective and generating NOx than diluting with carbon dioxide. A chemical reactor network indicates prompt NOx as the main component of the NOx formed in these experiments. This is because, in all conditions the firing temperature is held at a constant 1800K, below the 1900K NOx "threshold" but the fuel flow rate, and thus the concentration of transient reactive organic gasses, is increased leading to increased prompt NOx. 3|Page 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 Ditaranto (2009) performed a study to determine the factors influencing the reburning of NOx in gas turbine exhaust gas. Reburning of NOx is achieved by subjecting the NOx to prolonged contact with a pool of hydrocarbon radicals that are present in the fuel-rich reaction zone of a flame. The most significant factor dictating NOx reburning efficiency is the pool of hydrocarbon radicals that the NOx is in contact with. Increasing the number of radicals available or the contact time increases the reburning efficiency. It was found that by increasing the inlet concentration of NOx, the reburning efficiency increased. That is, increasing the NOx concentration resulted in a higher percent reduction in NOx at the exit of the flame. It was also observed that by increasing the percentage of water in the oxidizer mixture from 3% to 9% the reburning efficiency increased from roughly 20% to 35%, while the concentration of unburned hydrocarbons present in the exhaust increased from below 1000 PPM to over 4000 PPM. This indicates that a somewhat quenched reaction results in increased hydrocarbon radicals, resulting in better reburning. It was found that increasing the inlet percentage of oxygen decreased the NOx reburning efficiency. This was attributed to the increased reaction rate, which decreases the lifetime of the hydrocarbon radical pool, as well as increased flame temperatures which increase the NOx production rate independently of the NOx consumption rate. Finally, it was observed that increasing the inlet temperature resulted in similar results as increasing the oxygen percentage. In terms of general design of RQL configurations, extensive work has been done for gas turbines applications. One study (Holdeman 2000) indicates that the most rapid mixing in a cylindrical duct is obtained when the number of discrete jets, n, is determined with the equation: where J is the jet to cross flow momentum flux ratio. The diameter of the jets is then selected to deliver the required air-mass flow ratio. These design criteria tend to produce quick-mix air jets that penetrate to between 0.25 and 0.35 of the duct radius. Another reference (Lefebvre 1999) indicates that optimal mixing can be achieved when the jets penetrate to 1/3 of the duct diameter. The jet effective diameter is determined from the equation: where dj is the diameter corresponding to the effective area of the individual jet. The number of jets, n, is then selected to deliver the required air-mass flow ratio. Generally, these two methods are in good agreement. Both of these methods dictate a method for determining the jet configurations for cylindrical ducts that produce the most rapid mixing. Theoretically, the most rapid mixing will result in the lowest NOx emissions during the rapid transition from the fuel-rich zone to the fuel-lean zone because the least amount of time will be spent at near-stoichiometric conditions. However, one study (Vardakas, 1999) of actual NOx emissions of an RQL combustor for various numbers of jets at a fixed momentum flux ratio and fixed air-mass flow ratio has indicated that the previous assumption may not be correct. It was observed that the number of jets (with a fixed total effective 4|Page 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 area) have little effect on the overall NOx emissions. It was also observed that the quick mix module with the fewest holes, which demonstrated the slowest mixing for non-reacting flow, produced the lowest NOx emissions in the reacting case. It is theorized that, despite better mixing, quick mix modules with a more jet wakes in which to form NOx. It is further theorized that the overall NOx production is controlled by the fixed nitrogen production in the rich zone rather than the mixing in the lean zone. Studies have also been performed where low-NOx duct burners were designed and tested. Two studies were carried out by Nickeson (2010) and by Gas Technology Institute (2010) (GTI). The burner developed by Nickeson was a premixed, swirl stabilized type designed to operate with oxygen levels of 17-18%, and inlet temperature of 500F. The burner consisted of three concentric sections which could be independently fired. Two different turn down strategies were tested: constant flow, where the flow through the burner was held constant and burner sections were not fired, and diverted flow, where all burner sections were fired and a fixed fuel to air ratio, but oxidizer flow was partially diverted around the burner. It was observed that in the constant flow case, the NOx emissions goals could be met, but that CO emissions were greater than 600 PPM. In the diverted flow case, a turn down of 1.5:1 was achieved while meeting emissions limits. This burner was determined to be "stable" at fuel-to-air ratios corresponding to stack O2 percentages between 4 and 9%. The burner developed by GTI is a partially premixed, staged, natural gas fired burner. The burner incorporated staged combustion. Natural gas was mixed with air entering the primary zone of the burner. The flame is stabilized here by the presence of an internal baffle which induced a recirculation zone from downstream in the primary combustion zone. Oxidizer for the secondary zone is introduced axially by a central tube. The NOx performance of this burner is quite good, reducing the inlet NOx concentration from 3.4 PPMVD (corrected to 15% O2) to 2.2 PPMVD. In summary, previous studies have not considered the potential benefit of the inherent stability of the RQL approach. As a result, the objective of the present work is to evaluate the potential for an RQL based low emissions, modular, natural gas fired burner which can operate effectively in vitiated oxidizer streams like those found the exhaust stack of a gas turbine. 4 Approach To achieve the objective of the present work, the following tasks were carried out: 1. Develop a artificial exhaust generator. In order to study the performance of a burner under a variety of oxidizer compositions an artificial exhaust gas generator was constructed capable of supplying oxidizer mixtures at up to 40 SCFM and with oxygen levels of down to 12% by volume and inlet temperatures of up to 700°F. 2. Develop a modular RQL type burner. A burner was designed with interchangeable quick-mixing sections that each with an effective area that would produce a different oxidizer mass flow ratio to explore the sensitivity of the performance to factors of interest. 3. Characterize the stability of an RQL burner at varying oxygen concentrations. A key design parameter of a duct burner is to be able to operate at reduced oxygen levels. Therefore, rich 5|Page 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 and lean blow off testing was performed for a typical burner configuration at oxygen levels down to 12% by volume. 4. Characterize the emissions of the RQL burner over a range of equivalence ratios. Each quick-mix liner was evaluated for NOx and CO emissions at 17% oxygen over an equivalence ratio range from 0.4 to 0.7. 5. Characterize the NOx reduction capabilities of the burner over a range of equivalence ratios. Each quick-mix liner was evaluated for NOx emissions for inlet conditions 17% and 18% oxygen with 5 PPMV (corrected to 15% O2).The percent reduction in NOx can then be calculated over a range of operating equivalence ratios. 5 Experiment 5.1 Exhaust Simulator To facilitate evaluating the performance of a burner configuration when operating on a vitiated oxidizer stream an artificial exhaust generation rig was constructed. The purpose of this apparatus was to produce an oxidizer stream with the same composition and temperature as the exhaust stream exiting a gas turbine, or other prime mover. Compressed air was fed through a heater and then through a mixing section where CO2, N2, and water were injected in order to lower the percent oxygen in the air to the desired level (Figure 1 and Figure 2). Figure 1: Artificial exhaust generator schematic Figure 2: Artificial exhaust generator in place 6|Page 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 The test procedure consisted of adjusting the inlet air temperature to the desired value, then adjusting the oxygen percentage of the incoming air. The fuel flow was then adjusted to an equivalence ratio of approximately 1.0. An external pilot torch was then triggered to light the burner. Then the fuel flow rate was either lowered or raised to the desired test point. All gasses used in this experiment (air, N2, CO2 and natural gas) supplied via commercially available mass flow controllers. The water vapor was measured as a liquid with a laminar flow element type flow meter and then introduced via an atomizing nozzle to the high temperature region downstream of the heater where it evaporates. 5.2 Burner Configurations 5.2.1 Rich Burn Sections Two burners were evaluated, one with low swirl (Swirl Number of approximately 0.6) and the second with high swirl (Swirl Number of approximately 1.3). For the low swirl burner configuration, a small steel tube surrounded the single parker injector in order to induce a recirculation zone in the combustion chamber. The combustion chamber of the test burner consisted of an 80 mm inside diameter, 18 inch long quartz tube. The setups can be seen in Figure 3. a) Low Swirl b) High Swirl Figure 3. Rich Burn Configurations. 5.2.2 Quick Mix Sections The quick mix section was developed to be added directly to the rich burn section. Figure 4 illustrates the basic approach. Research has shown that optimal performance of an RQL combustor is achieved when the rich zone equivalence ration is between 1.2 -1.6. Even higher equivalence ratios being more effective and reducing NOx but can become unstable. Optimal lean (overall) equivalence ration range is 0.5-0.7 (Micklow 1993). Leaner overall equivalence ratios will help reduce CO, until the lean zone approaches blow off at which point CO production should again begin to rise. With this in mind a test rig 7|Page 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 was developed in which the preheated vitiated air would be split into two streams, one which was directed to the rich zone, and one which was directed to a quick mix section. The rich combustion zone is a premixed, swirl stabilized flame. The inner diameter of the combustion rich combustion chamber is 80 mm, and the swirler produces a swirl number of 1.2. The quick mix section consists of an annulus with holes on the inner wall which would inject air radially inward to quickly mix and then react with the partially combusted exhaust from the rich zone. The inner wall of the annulus (the quick mix "liner") was designed to be removable so that walls with different injection characteristics could be evaluated. Figure 4. Quick Mix Section atop the Rich Burn Section (High Swirl Configuration shown). Three quick-mix liners with a wide range of jet diameters and number of jets were constructed to provide a wide range of burner mass ratios. The burner mass ratio is defined as the ratio of mass which enters the burner through the quick mix jets to that which enters through the rich zone. The jet geometry of the three quick-mix liners is summarized in. Table 1: Summary of quick-mix liner geometry Ring 1.0-A 1.5-A 2.0-B Quick Mix Liner Geometry Jet Diameter Number of Jets 0.375" 5 0.166" 31 0.171" 24 Effective Area 0.42 in2 0.58 in2 0.47 in2 Mass Ratio 1.8 2.5 2.0 While all of these liners have very different jet geometries, the most important parameter to notice is the mass ratio produced. Previous work (Vardakas, 1999) has indicated that the actual geometry of the quick mix liner does not play a significant role in the emissions performance. Thus, the purpose of these 8|Page 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 tests was to determine the mass ratio which provided the best emissions and stability performance and then proceed towards optimizing the exact burner geometry. 5.3 Diagnostics The diagnostics used consist of (1) direct imaging of reactions using a digital camera, (2) documentation of lean and rich blow off limits by means of recording fuel flows at which these phenomena occurred, and (3) measurement of exhaust gas emissions. The emissions analyzer used for these tests was a Horiba PG-250 portable gas emissions analyzer. It utilizes non-dispersive IR detection for the detection of CO, SO2, and CO2 and uses chemiluminescence for the detection of NO. For the detection of NOX, the sample gas is first passed through a NOx converter where any NO2 is converted to NO for detection. O2 is detected using a galvanic cell. Before each test, the analyzer was calibrated by first zeroing with research grade nitrogen, and then spanning with two calibration gasses, one containing 40 PPM NO (balance nitrogen) and the other containing 155 PPM CO, 8% CO2, and 8% O2 (balance N2). Sample conditioning was consistent with CARB protocols. 6 Experimental Results 6.1 Effect of Vitiation on Burner Stability Before development began on the RQL style burner, it was important to further understand the effects of vitiation on the combustion of natural gas. A set of stability loops was generated for the low swirl rich burn section. The goal of these tests was to observe general trends in stability with respect to the level of vitiation. 6.1.1 Low Swirl Configuration A photograph of the low swirl reaction is shown Figure 5. The reaction is taking place with inlet oxygen concentration of 21%, no preheat (room temperature) and equivalence ratio of 0.6. As shown, the reaction is "pear" shaped and approximately 2 duct diameters in length. The reaction appears attached to the outer ends of the cylindrical extension piece. For the low swirl configuration, testing indicated the stability ranges shown in Figure 6 and Figure 7. The results of this test show a considerable decrease in stability range as vitiation level increases. At 18% O2, nominal gas turbine conditions, the width of the stability range has decreased by approximately 25% when compared to the 21% O2 case. In addition to this, it was observed that at no conditions could combustion be sustained when the inlet oxygen composition was below 12%. 9|Page 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 Figure 5: "Generic" burner setup used for evaluating stability limits. 2.5 Equivalence Ratio 2 1.5 400F 1 525F 650F 0.5 0 10 12 14 16 18 20 22 Oxygen Content (%) Figure 6: Blowoff vs. Oxygen percentage for 400°F, 525°F and 650°F with average inlet velocity of 6 ft/sec 10 | P a g e 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 2.5 Equivalence Ratio 2 1.5 400F 525F 1 650F 0.5 0 10 12 14 16 18 20 22 Oxygen Content (%) Figure 7: Blowoff vs. Oxygen percentage for 400°F, 525°F and 650°F with average inlet velocity of 10 ft/sec 6.1.2 High Swirl Configuration In order to evaluate the sensitivity of the stability loops to the swirl strength, studies were carried out for the high swirl configuration. A photograph is shown in Figure 8. In contrast the low swirl case, the reaction fills the width of the tube and does not exhibit the "teardrop" shape of the low swirl case. More streaks are noted for the high swirl case which may be associated with some mixing issues. Figure 8: High swirl burner tested for later use in RQL burner 11 | P a g e 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 In this configuration stability was improved. The lean blowout of the burner with pure air 525°F inlet air temperature was lowered to, 0.40, which is typical of an "good" natural gas combustor. Based on this result, the high-swirl burner arrangement was selected to be used in the in the next stage of development, the RQL style burner. Lean blow off limits vs. inlet oxygen composition for this configuration can be seen in Figure 9 and Figure 10. In comparison to the results for the low swirl configuration, the lean blow off stability is superior, perhaps not surprising given the more substantial recirculation zone of the high swirl configuration. 1 Equivalence Ratio 0.9 0.8 0.7 400F 0.6 525F 0.5 650F 0.4 0.3 10 12 14 16 18 20 Oxygen Content (%) Figure 9: Blowoff vs. Oxygen percentage for high swirl burner with average inlet velocity of 6 ft/sec 1 Equivalence Ratio 0.9 0.8 0.7 400F 0.6 525F 0.5 650F 0.4 0.3 10 12 14 16 18 20 Oxygen Content (%) Figure 10: Blowoff vs. Oxygen percentage for high swirl burner with average inlet velocity of 10 ft/sec 12 | P a g e 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 6.2 RQL Configuration Evaluation A photo of the operating RQL configuration is shown in Figure 11. Two reaction zones can be seen clearly. The lower reaction is fuel rich, operating at an equivalence ratio of 1.4. The upper reaction (partially obscured by the quick-mix hardware) is fuel lean, operating at an overall equivalence ratio of 0.5. Figure 11: RQL burner in use For the RQL configuration, three types of testing were performed. The first round of testing was, again, to observe general trends in stability with respect to the level of vitiation. The tests which were performed earlier in this study were repeated with the new RQL type burner with the 1.0-A quick-mix liner installed. The next test was to perform emissions testing for each quick-mix liner over a range of equivalence ratios, with inlet conditions similar to those found in the exhaust of a natural gas fueled gas turbine. The final test performed was to assess the NOx reburning capability of the burner by adding a known quantity of NO upstream of the burner in addition to the other vitiation elements. 6.2.1 Stability Initial testing was focused on determining the stability limits of the RQL burner. In these tests, the 1.0-A quick-mix liner used was used. The lean and rich blow off limits were evaluated for in inlet oxygen concentrations of 21% to 12%, inlet temperature of 400°F, 525°F, and 650°F, and at two different average inlet velocities. The average inlet velocity was taken to be the velocity of the non-reacting fuel and air downstream of the quick-mix section. The tests indicate that both the rich blow off limit and the lean blow off limit are increased. The total range of flammability has been increased by approximately 300% in both the 6 ft/sec case as well as the 10 ft/sec (Figure 12 and Figure 13). 13 | P a g e 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 4.5 Equivalence Ratio 4 3.5 3 RQL: 400F 2.5 RQL: 525F 2 RQL: 650F 1.5 Pre-Mixed: 400F 1 Pre-Mixed: 525F 0.5 Pre-Mixed: 650F 0 11 16 21 Oxygen Content (%) Figure 12: Blowoff vs. Oxygen percentage for RQL type combustor and single stage combustor at 6 ft/sec Equivalence Ratio 3.5 RQL: 400F 3 RQL: 525F 2.5 RQL: 650F Pre-Mixed: 400F 2 Pre-Mixed: 525F 1.5 Pre-Mixed: 650F 1 0.5 0 11 16 21 Oxygen Content (%) Figure 13: Blowoff vs. Oxygen percentage for RQL type combustor and single stage combustor at 10 ft/sec In these tests, the lean blow off limit was defined as the overall equivalence ratio at which the burner heat release rate goes to zero; that is, when there is no combustion taking place in any part of the burner. The improvement in the lean blow off limit when the RQL style burner is in use is due to the fact that the combustion is staged. In these tests approximately 60% of the air enters the burner through the quick mix section. When the overall burner equivalence ratio becomes too lean to support combustion in the lean zone, the rich combustion zone is at near stoichiometric conditions, and the fuel flow rate can still be lowered substantially. The rich blow off limit was similarly improved. At the point where the rich blow off limit was reached there was no flame in rich zone. A possible reason for the improvement 14 | P a g e 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 to the rich blow off limit of the burner is the presence of local near stoichiometric regions of the flame caused by imperfect mixing between the quick-mix air and the incoming fuel and air from rich zone. 6.2.2 Emissions Performance The second phase of testing consisted of gathering emissions data for each quick-mix liner over a range of equivalence ratios (0.40 to 0.70). What was also investigated was the relative importance of water vapor in the inlet oxidizer stream. The inlet conditions for these tests simulated those in the exhaust of a recuperated natural gas fired gas turbine: 550°F, 17% O2. In the wet cases the inlet oxidizer contained 77.6% N2, 17% O2, 3.6% H2O and 1.8% CO2. In the dry cases the inlet oxidizer contained 80.8% N2, 17% O2, and 2.2% CO2. 16 1.0, 17%, dry 14 1.0,17%,wet 12 1.5, 17%, dry [NO] 10 1.5, 17%, wet 8 2.0 B, 17%, dry 2.0 B, 17%, wet 6 4 2 0 0.38 0.48 0.58 0.68 Equivalence Ratio Figure 14: NO emissions (PPMVD corrected to 15% O2) for three quick-mix liners (17% O2, 550°F inlet) It was observed that the NO emissions for the 1.5-A and 2.0-B quick-mix liners were below 10 PPMVD for all test conditions, and that the NO emissions for the 1.0-A liner was below 14 PPMVD for all test conditions. Interestingly, for the 1.0-A ring the NO emissions were higher for the wet conditions than for dry conditions, which goes against the previous expectations. These results are summarized in Figure 14. CO emissions were also observed for the same test conditions and can be seen in Figure 15. The 1.5-A and 2.0-B liners show similar emissions patterns. Both liners have high CO emissions at an equivalence ratio of 0.4 which falls to a minimum and is then followed by a spike and another drop off in emissions. The high CO emissions at the lean end are a result of the fact that at these equivalence ratios, the lean zone approaches the flammability limit. The flickering nature of the flame near blow off gives rise to the increase in CO emissions, typically coupled with a decrease in NOx emissions. As the equivalence ratio increases, the flame in the rich zone becomes longer. The sharp rise in the CO emissions at approximately 0.58 correspond to the point at which the rich flame impinges on the flame in the quickmix/lean region of the flame. This region of high CO emissions, again, corresponds to a region of 15 | P a g e 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 reduced NO emissions. Interestingly, the 1.0-A ring shows a different trend in CO emissions than the other two rings. 1000 1.0, 17%, dry 1.0, 17%,wet 100 [CO] 1.5, 17%, dry 1.5, 17%, wet 10 2.0 B, 17%, dry 2.0 B, 17%, wet 1 0.38 0.43 0.48 0.53 0.58 0.63 0.68 Equivalence Ratio Figure 15: CO emissions (PPMVD corrected to 15% O2) for three quick-mix liners (17% O2, 550°F inlet) 6.2.3 NOx Reduction Evaluation The final tests performed assessed the NOx reduction potential of the burner. Approximately 5 PPM of NO was added to the vitiated oxidizer stream and emissions measurements were made downstream of the burner. All three quick-mix liners were evaluated for inlet conditions of 17% O2 (wet) and 18% O2 (wet) with an inlet temperature of 550°F. The composition of the 17% O2 was the same as described earlier. The composition of the 18% O2 case was: 78% N2, 18% O2, 2.7% H2O, and 1.3% CO2. It was found that the all quick-mix liners had the capability of reducing the corrected NOx concentration at some equivalence ratio for both 17% and 18% O2 inlet conditions. While obvious trends are difficult to distinguish, there are two apparent trends, one that applies to the 1.5-A liner, and another that applies to the 1.0-A and 2.0-B liners. For both 17% and 18% cases, the 1.5-A liner starts with a higher percent change in NOx which drops quickly and then rises slowly. Overall the 1.5-A liner provides the most stable NOx reduction performance. The 1.0-A and 2.0-B rings begin with relatively high (positive) percent change in NOx, and remain relatively constant until an equivalence ratio of approximately 0.53 at which point the percent change in NOx drops and in some cases becomes negative. The NOx reduction trends can be seen graphically in Figure 16. The CO emission data were also collected. The characteristic spike when the rich zone flame impinges on the lean/quick-mix zone is still present. In this round of testing, the flow of air to the rich zone as well as quick-mix zone was carefully measured so that a rich zone equivalence ratio could be obtained as well as total equivalence ratio. When the CO emissions are plotted against rich zone equivalence ratio, it becomes obvious that the CO emissions, in particular the location of the spike in emission, are dependent on the equivalence ratio of the rich zone. This correlation can be seen in Figure 17. 16 | P a g e 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 50 1.5A 18% O2 40 1.5A 17% O2 30 % Change NOx 20 2.0B 18% O2 10 1.0A 18% O2 0 -10 1.0A 17% O2 -20 2.0B 17% O2 -30 -40 0.38 0.48 0.58 0.68 Equivalence Ratio Figure 16: Percent change in NOx (PPMVD corrected to 15%) for 18% O2 inlet and 17% O2 inlet a. Vs Overal Equivalence Ratio b. Vs Rich Zone equivalence ratio 10000 2.0B 18% O2 2.0B 17% O2 1.5A 18% O2 1.5A 17% O2 1.0A 18% O2 1.0A 17% O2 1000 [CO] 100 10 1 0.4 0.5 0.6 Total Equivalence Ratio 0.7 1 1.5 2 2.5 Rich Zone Equivalence Ratio Figure 17: CO emissions (PPMVD corrected to 15%) for 18% and 17% O2 inlet plotted vs. total equivalence ratio and rich zone equivalence ratio 17 | P a g e 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 7 Discussion Based on the stability loops for each burner (Figure 12 and Figure 13), the RQL strategy adopted for this burner is capable of producing superior turndown ratios at inlet oxygen concentrations down to 12% when compared to single stage burners. If an equivalence ratio of 1 is taken to be the upper limit of operation of the burner, and the lean blow off limit to be taken as the lower limit of operation, a turndown ratio of 4:1 was achieved. A more practical range of operation would be the range over which the emissions were tested, which correspond to a turndown ratio of at least 1.75:1 for every burner configuration. These turndown ratios are, of course, limited by emissions ceilings. All burner configurations tested were able to achieve good NOx emissions levels. Quick-mix liners 1.5-A and 2.0-B achieved similarly good NOx emissions; both produced less than 10 PPMVD NO over the entire test range. Quick-mix liner 1.0-A had slightly worse (higher) NO emissions, exceeding 10 PPMVD for a majority of the tested range, but staying below 14 PPMVD. The main difference between the three rings that was being evaluated was the mass ratio. The mass ratio of the 1.0-A ring was the lowest tested (see Table 1). This means that over the range tested, the equivalence ratio in the rich section was the closest to stoichiometric, which means that the NOx production rates were higher for that burner configuration than the other two. For the "wet" inlet conditions (Figure 14: NO emissions (PPMVD corrected to 15% O2) for three quick-mix liners (17% O2, 550°F inlet) are considered, the NO emissions follow the expected trend; the liners with higher mass ratios produce lower NO emissions. This correlation seems to become less ideal when the "dry" inlet conditions are considered. At many points, the 1.5-A liner produces more NO than does the 2.0-B liner, which has a lower effective area. In addition to this, the 1.0-A liner produces more NO in the "wet" conditions than it does in the "dry" tests. These perplexing results may be a result of the fact that NO emissions data were collected instead of NOx data. It is possible that the chemistry associated with NOx formation is altered by the presences of water vapor. Emissions of CO for all burner configurations appear to have an inverse relationship with NOx emissions. The characteristic spike is a result of the flame from the rich zone impinging on the quick-mix/lean zone flame. This occurs at a constant rich zone equivalence ratio of about 1.6 (see Figure 17). The equivalence ratio where this spike occurs can be altered or eliminated by changing the length of rich combustion chamber because as the equivalence ratio in the rich zone is increased, the flame becomes longer. It was expected that it would be possible to reduce NOx through a reburning mechanism in the rich zone of the burner. It was not expected that the lean zone would play role in NOx formation outside of producing small quantities of it while the remaining fuel species are consumed. However, Figure 16 indicates that the lean zone does, at least, partially contribute to the reduction of NOx. After the point that the rich zone flame impinges on the lean zone flame, the rich zone flame becomes weaker, eventually lifting off and partially merging with the lean zone flame. This transition occurs at an overall equivalence ratio from 0.5 to 0.56. Figure 16 indicates that there is still NOx reduction taking place in many burner configurations after the point where the rich zone flame has merged with the lean zone flame. The reason for this reduction is also a result of reburning, in this case similar the reburning that occurs in a tangential firing array of a boiler. Radicals generated in the rich zone flame are introduced into the air entraining region of the lean zone flame and reduce NOx to N2. 18 | P a g e 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 8 Summary Based on the current data it was determined that an RQL type supplemental burner can be used to reduce the NOx present in the exhaust gas of a gas turbine engine. However, the CO emissions are often very high at the same conditions where NOx is reduced the most, typical of many combustion systems. It appears that, for the tested burner geometry, the quick-mix liner that provides the best compromise between NOx and CO emissions is achieved with an air mass ratio of 2. The results of this study indicate that prior to the rich flame impingement on the lean zone, CO emissions are in decline, and NOx emissions are fairly constant. If additional length were to be added to the rich combustion chamber, the onset of rich zone impingement could be delayed until a higher equivalence ratio, delaying the spike in CO emissions and providing a lager true-RQL operating range. To meet emissions standards, additional length should be added to the burner, either in the rich section to avoid rich and lean flame interaction, or to the lean zone to facilitate the further burnout of CO. Future development of this burner concept will include adding length to the rich combustion zone as well as a parameter study to determine the effect of rich zone swirl, and quick-mix geometry (for a mass ratio of 2) on burner performance. 9 Acknowledgments The authors would like to acknowledge the California Energy Commission for financial assistance under contract PIR-09-015 (Rizaldo Aldas, program manager). In addition, the assistance of Anthony Jordan and David Beerer is greatly appreciated. 10 References Ditaranto, M. , Hals, J. and Bjorge, T. (2009). Investigation on the in-flame NO reburning in turbine exhaust gas, Proceedings of the Combustion Institute, 2659-2666 Fackler, K.B. , Karalus, M.F., Novosselov, I.V., Kramlich, J.C., Malte, P.C., (2011). Experimental and numerical study of NOx formation from the lean premixed combustion of CH4 mixed with CO2 and N2, Proceedings of ASME Turbo Expo, 1-9 Gas Technology Institute, 2010, Integrated CHP using ultra-low-NOx supplemental firing, Final report for CARB grant No. ICAT 05-1, 1-43 Holdeman, J.D., Samuelsen, G.S. and Leong, M.Y., (2000). Optimization of Jet Mixing into a Rich, Reacting Crossflow, Journal of Propulsion and Power, pp. 729-735 McKinney, R.G., Sepulveda, D., Sowa, W., Cheung, A.K., (2007). The Pratt & Whitney TALON X low emissions combustor: Revolutionary results with evolutionary technology, Proceedings of the 45th AIAA Aerospace Sciences Meeting and Exhibit, 1-8 Lefebvre, A.H., 1999, Gas Turbine Combustion, London, Taylor & Francis 19 | P a g e 2011 AFRC Meeting, Houston TX, 18-21 Sept, 2011 Lieuwen, T., McDonell, V., Petersen, E., and Santavicca, D. (2008). Fuel Flexibility Influences on Premixed Combustor Blowout, Flashback, Autoignition, And Stability (2008). ASME J. of Engineering for Gas Turbines and Power. Vol. 130, January, pp. 011506-1 - 011506-10 Micklow, G.J., Roychoudhury, S., Nguyen, H., Cline, M.C.,. (1993). Emissions reduction by varying the swirler airflow split in advanced turbine combustors, Journal of engineering for gas turbines and power, 563-569. Miles, G.A. (1971). A well-stirred-reactor investigation of the combustion of propane with vitiated air, Proceedings of the International Symposium on Combustion, 483-487 Mullins, B.P. and Barr, J.,( 1949). Combustion in Vitiated Atmospheres, Fuel, 181-188; 205-206;225-231 Nickeson, B. (2010). Development of an ultra low NOx duct burner for small high efficiency CCHP systems, Proceedings of the International colloquium on environmentally preferred advanced power generation (ICEPAG 2010), 1-7. Ragland, K.W. and Bryden, K.M. (2011). Combustion Engineering, 2nd Edition, CRC Press, Boca Raton, FL. Richards, G.A., McMillian, M.M., Gemmen, R.S., Rogers, W.A., and Cully, S.R. (2001). Issues for lowemissions, fuel-flexible power systems, Prog. Energy Combustion Science, Vol. 27, pp. 141-169. Vardakas, M.A., Leong, M.Y., Brouwer, J., Samuelsen, G.S., Holdeman, J.D., 1999, The effect of air preheat on the formation of NOx in the quick-mix section of an axially staged combustor, NASA Technical Memorandum 1999-209431, 1-23 20 | P a g e