Oxygen-Enriched Combustion Studies with the Low NO x CGRI Burner

Transcription

1 IFRF Combustion Journal Article Number 00404, October 004 ISSN X Oxygen-Enriched Combustion Studies with the Low NO x CGRI Burner D. Poirier, E.W. Grandmaison*, A.D. Lawrence 1, M.D. Matovic and E. Boyd Centre for Advanced Gas Combustion Technology Queen s University Kingston, ON, K7L 3N6 Canada 1 IFM Kemiteknik Linkopings Universitet, Linkoping Sweden *Corresponding Author(s): Ted Grandmaison, Department of Chemical Engineering, Queen's University, Kingston, ON K7L 3N6, Canada. Tel.: Fax : grandmai@chee.queensu.ca

2 IFRF Combustion Journal - - Poirier, Grandmaison, Lawrence et. al. ABSTRACT An oxygen-enriched/natural gas combustion study with a modified low NOx CGRI burner has been completed. Effects of oxygen enrichment, at various stack oxygen levels and a single furnace operating temperature, on NOx and CO emissions, fuel efficiency and furnace temperature distribution, were determined. Combined effects of oxygen enrichment and air infiltration were also studied. A single sidewall mounted burner was employed in the pilot scale CAGCT research furnace. The firing rate required to maintain the furnace temperature at 1100 C decreased linearly with increasing oxygen enrichment. At full oxygen enrichment, a reduction of 40-45% in the firing rate was needed to maintain constant furnace temperature. NOx emissions (< 1 mg/mj) were relatively constant with changes in oxygen enrichment levels below ~ 60% and decreased at higher oxygen enrichment. NOx emission increased with increasing stack oxygen concentration at all oxygen enrichment levels. Air infiltration resulted in NOx emissions similar to those observed with no air infiltration but with similar stack oxygen concentrations. The standard deviation of the temperature distribution for the furnace roof and blind sidewall was in the range, 19 7 C with no oxygen enrichment and C with 90% oxygen enrichment. Keywords: oxygen-enriched combustion, low NOx combustion, energy efficiency

3 IFRF Combustion Journal Poirier, Grandmaison, Lawrence et. al. INTRODUCTION Improvements in energy efficiency coupled with reduced emissions are an ongoing objective in many industrial sectors employing combustion technologies. Dilute combustion technology (Milani and Saponaro, 001) has been found to reduce NOx emissions by mixing the fuel and oxidant streams with inert combustion product gases. This technique leads to lower oxygen and fuel concentrations along with lower temperatures in the combustion or reaction zones of industrial furnaces. A burner conceived by the Canadian Gas Research Institute (CGRI) and tested jointly with the Centre for Advanced Combustion Technology (CAGCT) falls into this category of technology (Besik et al., 1996; Sobiesiak et al., 1998; Grandmaison et al., 1998). Oxygen-enhanced combustion is a relatively well developed technology (Baukal, 1998) employed in the combustion industry (e.g. De Lucia, 1991; Delabroy et al., 001; Marin et al., 001). This study combines dilute combustion and oxygen-enriched combustion, with the goal of optimizing the beneficial characteristics of both technologies: energy efficiency (low CO ), low NOx emissions and good heat transfer. CAGCT FURNACE SYSTEM Testing and development of the O -enriched furnace system were conducted at the Centre for Advanced Gas Combustion Technology (CAGCT), Research Furnace Laboratory, Queen s University. The interior of the furnace, Figure 1, is divided into two unequal size chambers by a checker-work, brick end-wall. The first chamber is the main furnace cavity with internal dimensions of 4.5 m long, 3 m wide and 1 m high (177 in. x 118 in. x 39 in.). The second chamber serves as an exhaust plenum with interior dimensions of 0.6 m long, 3 m wide and 1 m high (4 in. x 118 in. x 39 in.). The checker wall, 15 mm thick (8.5 in.), with an 8 x 3 array of openings, 75 mm x 115 mm (3 in. x 4.5 in.), separates these two chambers. The refractory lining for the furnace walls and roof are ceramic fibre blocks, 305 mm (1 in.) thick. The furnace wall structure and refractory is a combined 36 mm (14 in.) thick, as shown in Figure 1. Instrumentation for the furnace includes fixed thermocouples, static pressure taps, orifice meters for gas and air flow measurement and control. Refractory wall surface-thermocouples are located at positions T1 T41 as shown in Figure. These thermocouples, 0.54 mm

4 IFRF Combustion Journal Poirier, Grandmaison, Lawrence et. al. Burner Plenum Wall Top View Furnace Exhaust Refractory Side View 0 Water-cooled floor panels Figure 1: CAGCT research furnace shown with the single sidewall mounted burner used in the present study. All dimensions in mm. diameter Pt/Pt-10%/Rh, are embedded about 5 mm into the refractory walls. The size and positioning of these thermocouples help minimize measurement error. The furnace is also equipped with water-cooled floor panels for heat flux measurements, sampling ports for internal furnace measurements and recuperators for air preheat. In the present work, a single burner was fired from the furnace sidewall, Figure 1. The furnace was operated at positive pressure for the primary set of tests with a selected set of trials performed at negative pressure to study the effect of air infiltration. The burner design, Figure 3, was a modified form of the ultra-low NO X burner initially developed at the Canadian Gas Research Institute (CGRI) and CAGCT (Besik et al., 1996, Sobiesiak et al.,

5 IFRF Combustion Journal Poirier, Grandmaison, Lawrence et. al. Burner T1 T8 T1 T16 T T5 T T9 T14 T17 T3 T6 T7 T10 T15 T18 T1 T T T4 T11 T13 T T6 T5 T4 T8 T7 T31 T30 T9 T33 T38 T35 T37 T3 T34 T36 T41 T40 T Figure : Location of refractory-wall thermocouples in the CAGCT Research Furnace top figure shows refractory roof, bottom figure shows blind-sidewall opposite the burner. UV scanner port Pilot burner port Air/oxidant nozzle Fuel nozzle Figure 3: Schematic diagram of the CGRI showing the locations for the air/oxidant and fuel nozzles.

6 IFRF Combustion Journal Poirier, Grandmaison, Lawrence et. al. 1998, Grandmaison et al., 1998) was used in this work; hereafter this burner is referred to as the CGRI burner. The burner consists of a ring-array of alternating fuel and oxidant nozzles directed at different angles to the burner axis. The burner was modified to include oxygen supply tubes and jets running coaxially to the air supply tube and jets. The oxygen nozzle diameter and the air nozzle annulus were sized so that the momentum of the combined oxidant stream would remain relatively constant with changing O -enrichment level for a constant firing rate. The air and O nozzle angle (10 ), air-nozzle annulus size, fuel jet angle (0 ) and fuel nozzle diameter were maintained at constant values for the results reported in this work The firing rate was adjusted to maintain a constant furnace temperature of 1100 C as O enrichment and excess oxidant was varied. This clearly demonstrated fuel savings gained by O enrichment and provided a better basis for comparison of other data including NO X levels. Oxygen enrichment level, ψ O, is defined as ψ O m& O = m O + m & & OA 100 where m& O and m& OA are the mass flow rates for the pure oxygen and oxygen associated with the air feed streams, respectively. Concentrations of O, CO, CO, NO X and CH 4 in the exhaust gases were continuously measured. Refractory surface temperatures of the furnace walls and ceiling and heat flux to water-cooled floor panels were also continuously monitored. Quasi-steady-state furnace conditions for gas composition measurements were assumed once the furnace control temperature reached the operator set point (1100 C in these trials) and gas analysis readings stabilized. A large number of the data reported in this work were obtained during steel scaling tests reported by Poirier et al. (004) in which the furnace was operated at fixed conditions over 4 8 hour periods. RESULTS AND DISCUSSION To demonstrate the reduction in fuel usage and CO emissions that can be expected with O - enriched combustion, the burner firing rate was monitored at the furnace set point temperature (1100 ± 0 C) under constant furnace load and stack O level for various levels of O enrichment. Results of these tests, shown in Figure 4, indicate that fuel usage (firing

7 IFRF Combustion Journal Poirier, Grandmaison, Lawrence et. al. Figure 4: Furnace firing rate as a function of oxygen enrichment for various stack oxygen levels. Furnace temperature between 1080 and 110 C. rate) decreases linearly with increasing O enrichment level. At ψ O = 0% the average firing rate required to maintain the furnace set point temperature was 353 kw and at ψ O = 100% the firing rate decreased to an average value of 1 kw. This represents a potential fuel savings of ~40% with full oxygen enrichment. A summary of the data at ψ O = 0 and 100% are also given in Table 1 showing the firing rate data as a function of excess oxidant levels with stack oxygen concentrations in the range of 0% < O <.0% and % < O < 4%. These results and the data in Figure 4 show a modest effect of excess oxidant level on the required firing rate. As expected, the firing rate tends to increase with increased stack O, but this trend was only evident at lower values of oxygen enrichment, ψ O < ~30%. At higher oxygen enrichment levels this trend was not evident within the experimental error associated Table 1: Summary of the furnace firing rate conditions (furnace target temperature of 1100 C) and potential fuel savings as a function oxygen enrichment level and stack oxygen concentration. ψ O = 0% ψ O = 100% Stack O, % w.b. Firing rate range, kw, Average firing rate, kw, (number of tests) Firing rate range, kw, Average firing rate, kw, (number of tests) Potential fuel savings 0 < O <.0% kw, 344 kw, (17) 14 4 kw, 19 kw, () 36%.0 < O < 4.0% kw, 366 kw, (10) 0 4 kw, 11 kw, (10) 4%

8 IFRF Combustion Journal Poirier, Grandmaison, Lawrence et. al. with these measurements. The corresponding potential fuel savings was 36% at lower excess oxidant levels (0.0% < stack O <.0%) and 4% at higher excess oxidant levels (.0% < stack O < 4.0%). A large portion of the experimental work was dedicated to examining effects of excess oxidant and O enrichment level, on NO X emissions. Results for the base case of NO X emissions as a function of excess oxidant with no oxygen enrichment are shown in Figure 5. These emission levels are consistent with the results reported by Sobiesiak et al. (1998) for the CGRI burner with low air preheat (the air temperature in the present work was relatively constant at ~ 50 C). The results were typically 8 10 ppm (w.b.) at low stack oxygen concentrations, increasing linearly up to about 14 ppm at 4 % stack oxygen concentration. The firing rate was adjusted to maintain a constant furnace temperature as the O enrichment and the excess oxidant was varied. This provided a good basis for comparison of NO X levels across the data set. Figures 6 and 7 display the data in two slightly different, but revealing ways. Figure 6 permits one to examine the effects of oxygen enrichment on NO X production, while effects of excess oxidant level on NO X emissions can be more clearly seen in Figure 7. The data in Figure 7 include all the data shown in Figures 4 and 6 as well as additional observations at stack oxygen concentrations exceeding 4.0% (w.b.). Furnace conditions were near steady state with an average refractory temperature in the range <T r > = C, no cooling panels were exposed and no air infiltration was permitted. Figure 6 shows how NO X emissions varied with O enrichment levels for various ranges of stack oxygen level. The graph demonstrates that there is no dramatic increase in NO X emissions with increasing O enrichment. NO X emissions, in fact, appear to remain relatively constant in the O enrichment range of 0 60%. This is somewhat different from the case with conventional O -enriched burners, where a sharp increase in NO X emissions is encountered. Conventional oxygen-enriched burners produce a much hotter flame than conventional air-only burners. Emissions of NO X are sensitive to temperature and although nitrogen available for conversion to NO X decreases with increased O enrichment, NO X emissions rise due to the increased peak temperature. The CGRI O -enriched burner is a dilute combustion technology which exhibits much lower peak temperatures than typical O - enriched burners. The relatively low NO X emission levels observed for the CGRI O -

9 IFRF Combustion Journal Poirier, Grandmaison, Lawrence et. al. Figure 5: NO X emission as a function of stack oxygen level with ψ O = 0 % and the furnace temperature between 1080 and 110 C. Figure 6: NO X production as a function of oxygen enrichment for various stack oxygen levels. Furnace temperature between 1080 and 110 C. Figure 7: NO X production as a function of stack oxygen concentration for various levels of oxygen enrichment. Furnace temperature between 1080 and 110 C. Figure 8: NO X production as a function of the furnace nitrogen concentration for various levels of stack oxygen concentration. Furnace temperature between 1080 and 110 C.

10 IFRF Combustion Journal Poirier, Grandmaison, Lawrence et. al. enriched burner with 0 60% O enrichment are due to these lower peak flame temperatures. While local gas temperatures were not measured in the furnace, the temperature distribution of the refractory surfaces, described later in this section, and the results of Wünning and Wünning (1997) support this observation. As O enrichment levels increase beyond 60% enrichment, Figure 6 shows that NO X emissions decrease for all levels of excess oxidant. This is expected since, as even with conventional burners, when firing with nearly pure oxygen, nitrogen available for conversion to NO X is significantly reduced, resulting in lower NO X production. One expects NO X production to drop to zero when pure O (100% O enrichment) is the only oxidant used. This is not the case for the results displayed in Figure 6. Although no nitrogen from air is available for conversion to NO X, there is nitrogen entering the furnace from the fuel, natural gas. In our case, approximately 1.6% of the fuel is nitrogen. This fuel-nitrogen is sufficient for production of the NO X levels observed at 100% O enrichment. Although there is no pronounced trend in NO X production with O enrichment level, the difference in the NO X levels between different excess oxidant levels is obvious. Higher levels of NO X emissions are observed as the stack oxygen level increases. This trend is clearly demonstrated in Figure7 where the NO X emissions are presented as a function of the stack oxygen level for various ranges of O enrichment. This figure clearly shows the relative effects of excess oxidant and O enrichment on NO X emission levels. It is evident that excess oxidant is influential for all levels of O enrichment, while O enrichment is only influential at levels above 60% enrichment. The NO X production rate as a function of the stack N level is shown in Figure8. Riley et al. (000) reported results of an oxygen enrichment study with dilute oxygen combustion. They suggested that an increase of 10% nitrogen in the furnace gas leads to an increase of about 60% in NO X emissions. In the present work this trend appears to be valid up to nitrogen concentrations of about 50% N (w.b.) after which the NO X levels remain relatively constant or decrease slightly. The results of tests to examine the effect of air infiltration on NO X emissions are shown in Table. For these trials, the furnace was operated at a firing rate of 1 kw, 100% oxygen enrichment and 10% excess oxidant. The first row of data corresponds to the no-air-

11 IFRF Combustion Journal Poirier, Grandmaison, Lawrence et. al. infiltration case with a furnace operating pressure of +1.7 mm H O, the second row of data corresponds to the same furnace settings with a negative pressure of 1.7 mm H O. The resulting air infiltration, quantified by the increase of stack O from 5.5% to 7.5% by volume corresponds to an air infiltration rate equal to 0% of the volumetric burner feed. The third row of data in Table corresponds to an infiltration rate equal to 43% of the burner volumetric feed. The increase in NO X with these levels of air infiltration corresponds well with the results for similar stack O levels and effective O enrichment levels without air infiltration shown in Figures 6 8. Table. NO x emissions at three furnace operating pressures at constant firing rate 10% excess oxidant and 100% oxygen enrichment. Furnace pressure, mm H O Firing rate, kw ψ O, % Stack O, % w.b. <T roof > arith, C NO X, mg/mj Temperature distribution is an important aspect of furnace performance and is of particular interest here, since oxygen-enriched combustion typically results in intensified (hotter) combustion zones. The best of the current data sets available for studying the effects of oxygen enrichment on furnace temperature distribution were trials where the furnace operating conditions where maintained at constant levels for extended periods of 4 8 hours. Refractory surface-temperatures were continually logged throughout each trial for the furnace roof and blind sidewall (opposite the burner sidewall), Figure. Data showing the values for the roof and sidewall arithmetic area-mean temperature T r arith = ( 1/S) S T ds r and the radiative area-mean temperature T r rad = 4 ( 1/S) T ) 1/ 4 r ds S at different stack oxygen levels are shown in Table 3. The standard deviation of the local refractory temperatures from the arithmetic mean, T r σ, are also shown in this table. At ψ O = 0%, estimates of the standard deviation ranged from 19 7 C and slightly higher values were observed at ψ O = 90% with T r in the range C. The stack oxygen σ concentration did not appear to have a significant effect on T. r σ

12 IFRF Combustion Journal Poirier, Grandmaison, Lawrence et. al. Figure 9: Difference in local temperatures from the arithmetic area-mean for the interior surface of the furnace (furnace roof in top diagram; blind sidewall in bottom diagram) for various levels of oxygen enrichment. Stack oxygen level of % w.b.; temperature units are C. The data at each location are ordered (top-to-bottom) with the results for ψ O = 0, 4.4 %, 49.8 % and 90.0 % respectively. Temperature mapping for the furnace operating at % stack oxygen (w.b.) expressed in terms of the difference in local temperature from the arithmetic area-mean value is shown in Figure 9. With the single burner operation employed in this work, the refractory temperatures exhibited positive deviations from the arithmetic area-mean value along furnace roof downstream from the burner. A maximum positive deviation was observed in the corner junction of the furnace roof and the blind sidewall directly opposite from the burner. Minimum values were observed near the exhaust plenum and the lower parts of the blind side-wall. This trend was consistent for all levels of oxygen enrichment and the deviations from the arithmetic area-mean were larger at higher oxygen enrichment levels. This trend is

13 IFRF Combustion Journal Poirier, Grandmaison, Lawrence et. al. indicated by the results shown in Figure 9 and the higher standard deviations from the mean value noted in Table 3 at ψ O = 90 %. Table 3: Arithmetic area-mean temperature, radiative area-mean temperature and standard deviation of the arithmetic area-mean temperature from the mean value for the roof and blind sidewall refractory surfaces. Results are shown for different oxygen enrichment and stack oxygen levels. Stack oxygen = 1 %, w.b. ψ O, % <T roof > arith., C <T roof > rad., C <T roof >, C ψ O, % <T sidewall > arith., C <T sidewall > rad., C <T sidewall >, C Stack oxygen = %, w.b. ψ O, % <T roof > arith., C <T roof > rad., C <T roof >, C , % <T sidewall > arith., C <T sidewall > rad., C <T sidewall >, C Stack oxygen = 4 %, w.b. ψ O, % <T roof > arith., C <T roof > rad., C <T roof >, C ψ O, % <T sidewall > arith., C <T sidewall > rad., C <T sidewall >, C

14 IFRF Combustion Journal Poirier, Grandmaison, Lawrence et. al. CONCLUSIONS The objective of this work was to test a modified version of the CGRI low NO X burner with oxygen-enriched combustion in the CAGCT research furnace at 1100 C. This technology has potential to reduce energy costs and emissions of CO and NO X. The reduction in CO emissions arises directly from the expected savings in fuel with O -enrichment and potential savings of ~ 40% in fuel usage at 100% oxygen enrichment were observed. NO X emissions up to ~1 mg NO x /MJ were observed with this modified version of the CGRI burner. Oxygen-enrichment had little effect on NO X emission up to an enrichment level of about ~60%. At higher oxygen-enrichment, emission levels decreased but not to zero because of fuel nitrogen present in the natural gas supply. NO X emission increased with increasing stack oxygen concentration (up to ~ 6% O w.b. in the present work) at all oxygen levels. Air infiltration also had an effect on NO X levels leading to emissions similar to those observed with no air infiltration but with similar stack oxygen concentrations. Oxygen enrichment level had the most significant effect on the temperature distribution of the roof and blind side wall of the furnace. The standard deviation of the temperature variation was in the range, 19 7 C with no oxygen enrichment and C with 90% oxygen enrichment. ACKNOWLEDGEMENT This work was performed under the U.S. Department of Energy (DOE) / American Iron and Steel Institute (AISI) Cooperative Agreement DE-FC07-97ID13554, Technology Roadmap Research Program for the Steel Industry. The support and participation of Air Liquide Corporation, BOC Gases, Dofasco Inc., Fuchs Systems and Stelco Inc. in this program is greatly appreciated. REFERENCES Baukal, C.E., Oxygen-Enhanced Combustion, CRC Press, New York, Besik, F.K., Rahbar, S., Becker, H.A. and Sobiesiak, A., U.S. Patent Application No. 08/56,999, Nov and International Patent Application No. PCT/CA96/00334, May 4, Delabroy, O., Louédin, O., Tsiava, R., Le Gouefflec, G. and Bruchet, P., Oxycombustion for Reheat Furnaces Major Benefits Based On ALROLL TM, A Mature Technology,

15 IFRF Combustion Journal Poirier, Grandmaison, Lawrence et. al. AFRC/JFRC/IEA 001 Joint International Combustion Conference, Kauai, Hawaii, Sept. 9-1 (001). De Lucia, M., Journal of Energy Resources Technology, 113: 1 (1991). Grandmaison, E.W., Yimer, I., Becker, H.A. and Sobiesiak, A., Combustion and Flame, 114: 381 (1998). Marin, O., Bugeat, B., Macadam, S. and Charon, O., Oxygen Enrichment in Boilers, AFRC/JFRC/IEA 001 Joint International Combustion, Kauai, Hawaii, Sept. 9-1 (001). Milani, A. and Saponaro, A., Diluted Combustion Technologies, IFRF Combustion Journal, Article Number 00101, February 001. Poirier, D., Grandmaison, Matovic, M.D., Barnes, K.R. and Nelson, B.D., High Temperature Oxidation of Steel in an Oxygen-enriched Low NO X Furnace Environment, AFRC/JFRC 004 Joint International Combustion Conference, Maui, Hawaii, Oct (004). Riley, M.F., Ryan, H.M. and Kobayashi, H., Application of dilute oxygen combustion (DOC) technology for steel reheating furnaces, American Flame Research Committee (AFRC) International Symposium, Newport Beach, CA, USA (000). Sobiesiak, A., Rahbar, S. and Becker, H.A., Combustion and Flame, 115: 93 (1998). Wünning, J.A. and Wünning, J.G., Prog. Energy Combust. Sci., 3: 81 (1997).