Experimental and chemical kinetic study of CO and NO formation in oxy-methane premixed laminar flames doped with NH3

Combustion and Flame xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Combustion and Flame j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c o m b u s t fl a m e

Experimental and chemical kinetic study of CO and NO formation in oxy-methane premixed laminar flames doped with NH3 Manuel Barbas a, Mário Costa a,⇑, Stijn Vranckx b,c, Ravi X. Fernandes b,d a

IDMEC, Mechanical Engineering Department, Instituto Superior Técnico, University of Lisboa, Lisboa, Portugal Physico Chemical Fundamentals of Combustion, RWTH Aachen University, 52056 Aachen, Germany c Flemish Institute of Technological Research (VITO), Boeretang 200, 2400 Mol, Belgium d Physikalisch Technische Bundesanstalt, Bundesallee 100, 38116 Braunschweig, Germany b

a r t i c l e

i n f o

Article history: Received 16 July 2013 Received in revised form 28 October 2014 Accepted 29 October 2014 Available online xxxx Keywords: Oxy-fuel combustion Experimental Kinetic study CO NO

a b s t r a c t The present work focuses on the oxy-fuel combustion of methane doped with ammonia in a premixed laminar burner operating at atmospheric pressure, and includes both experiments and a chemical kinetic study. CO and NO formation/emission were examined as a function of the stoichiometry and oxidizer composition. The experimental results showed that, for all oxidizer compositions studied, an increase in the excess oxygen coefficient generally decreases both the CO and NO emissions. Moreover, for the O2/CO2 environments, decreasing the oxygen concentration in the oxidizer, for a given excess oxygen coefficient, leads to higher CO emissions, but lower NO emissions. In air firing, the CO emissions were found to be significantly lower than those measured under oxy-fuel conditions, while the NO emissions were higher than those from the oxy-fuel cases. The chemical kinetic study allowed to identify the main reactions that directly (with the aid of a rate-of-production analysis) and indirectly (through a sensitivity analysis) influence both the CO and NO emissions. Under oxy-fuel conditions, CO2 + H M CO + OH and 1 CH2 + CO2 M CH2O + CO significantly contribute to CO formation additionally to those reactions found in air-fired combustion, while CO oxidation takes place through CO + OH M CO2 + H for all studied conditions. Formation of NO occurs for all conditions mainly with HNO as intermediate, particularly through HNO + H M H2 + NO. Once NO is formed, interconversion to NO2 occurs. Ó 2014 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

1. Introduction It seems to be consensual [1] that the current energy demand will be largely supplied by the combustion of fossil fuels in the near future, despite the limited reserves and the environmental issues. Renewable energy sources are being introduced to the energy scenario and, hopefully, will be the major contributor in the future, solving both issues, particularly the increase in CO2 emissions. For the time being, the so-called oxy-fuel combustion technology is one of the most promising solutions to enable CO2 capture and sequestration. In this technology combustion takes place with pure oxygen instead of atmospheric air, which, combined with flue gas recirculation, generates combustion products rich in CO2 that greatly facilitates its sequestration. This study concentrates on CO and NO formation during oxymethane combustion in a laboratory burner. Related previous studies include those reported in Refs. [2–12]. Amato et al. [2] ⇑ Corresponding author at: Mechanical Engineering Department, Instituto Superior Técnico, Av. Rovisco Pais, 1049-001 Lisboa, Portugal. E-mail address: [email protected] (M. Costa).

investigated the CO (and O2) emissions from a methane fired laboratory combustor operating under oxy-fuel conditions. The authors performed measurements and thermodynamic equilibrium and chemical kinetics calculations. They concluded that CO emissions are higher in combustion with O2 diluted with CO2 than in combustion with air. Moreover, for a given residence time, the CO emissions from oxy-fuel combustion are higher than the equilibrium values because of the slow oxidation of the intermediate CO formed in the flame. Also, they found that CO emissions increase exponentially with the flame temperature, and predicted that an increase in pressure would lower the emissions. Glarborg and Bentzen [3] evaluated the chemical effects of the presence of high CO2 concentrations in the oxy-fuel combustion of methane in a laboratorial flow reactor. Their experimental results were interpreted with the aid of a detailed chemical kinetic mechanism for hydrocarbon oxidation. They concluded that the presence of high CO2 concentrations lead to a significant increase in the CO concentrations, in the near burner region. The high levels of CO2 prevent complete oxidation of the fuel at higher temperatures despite the presence of excess oxygen. Heil et al. [4] investigated experimentally the effect of the N2 and CO2 (as bulk gases) on the burning

http://dx.doi.org/10.1016/j.combustflame.2014.10.020 0010-2180/Ó 2014 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Please cite this article in press as: M. Barbas et al., Combust. Flame (2014), http://dx.doi.org/10.1016/j.combustflame.2014.10.020

2

M. Barbas et al. / Combustion and Flame xxx (2014) xxx–xxx

rates during the oxy-fuel combustion of methane in a furnace operating under flameless oxidation conditions. In case of combustion in N2/O2 atmospheres, the CO profiles for different O2 concentrations overlap indicating that changing the O2 concentration does not affect the combustion rates. However, in case of combustion in CO2/O2 atmospheres, the CO2 concentration had a significant impact on the CO formation and consumption rates, which was attributed to the CO2 participation in the chemical reactions. Abián et al. [5] used the detailed kinetic mechanism developed by Glarborg et al. [6], with minor changes and updates, to simulate and interpret the experimental results of the CO oxidation in a quartz flow reactor operating at atmospheric pressure. They concluded that adding CO2, in general, inhibits the CO oxidation, and that this suppression is more pronounced for fuel-rich conditions and lower CO2 concentrations. Their results also revealed that water vapor enhances the CO conversion, contrary to some literature results with higher O2 and H2O levels. Very recently, Alves et al. [12] concluded that the CO emissions are significantly higher for the combustion in an oxidizer with 20.9% O2/79.1% CO2 than for the combustion in air, and that, under oxy-fuel conditions, the CO emissions decrease as the oxidizer O2 concentration increases, with the oxidizer composed by 35% O2/65% CO2 presenting levels of CO emissions similar to those obtained for the combustion in air. The addition of high levels of H2O to the oxidizer slightly decreases the CO emissions only for oxidizers with high CO2 concentrations. In oxy-fuel combustion the only source of nitrogen (if any) is present in the fuel itself. Therefore it is critical to understand the fuel-NO formation mechanism under oxy-fuel combustion conditions, particularly, in coal applications. During coal combustion CH4, HCN and NH3 are released to the gas phase during devolatilization [7,8]. Wang et al. [9] studied the conversion of pyridine into NO and N2 in O2/CO2 atmospheres in a flow reactor at temperatures between 1073 K and 1473 K for different stoichiometries. Their experimental results indicate that higher temperatures promoted the formation of NO from pyridine and that increasing the equivalence ratio led to a monotonic increase in NO formation. Their simulation results showed that high CO2 concentrations reduced the availability of oxygen, thus altered the evolution of NO through the promotion of reaction HNO (+M) M H + NO (+M) and limitation of reaction HNO + O2 M HO2 + NO. The authors observed also that the major pathways for NO consumption occurred through reactions with NCO and NH as intermediates. Giménez-López et al. [10] examined experimentally and numerically the oxidation of HCN in O2/CO2 and air atmospheres in a quartz flow reactor at temperatures between 900 K and 1450 K for different stoichiometries. The authors observed an inhibition of the HCN oxidation under high levels of CO2, as compared to N2. This effect resulted from the competition of CO2 and O2 for atomic hydrogen, reducing the formation of chain carriers and thus inhibiting HCN oxidation. The lower HCN burning rate observed was accompanied by a higher CO and HNCO formation and lower NO and N2O concentrations, as compared to air combustion. Mendiara and Glarborg [8] investigated experimentally the oxidation of NH3 during the oxy-fuel combustion of methane in a plug flow reactor and interpreted the results in terms of a detailed chemical kinetic model. The results showed that a high CO2 level enhanced NO formation under reducing conditions, while it inhibited NO production under stoichiometric and lean conditions. According to the authors, the enhanced CO concentrations and alteration in the amount and partitioning of O/H radicals are responsible for the effect of a high CO2 concentration on the NH3 conversion. Watanabe et al. [11] studied experimentally and numerically the NO formation and reduction in staged O2/CO2 and air combustion using a flat CH4 flame doped with NH3. The authors observed that the NOx conversion ratios in O2/CO2 environments were lower than

those in air combustion and that abundant OH radicals were formed in O2/CO2 environments through CO2 + H M CO + OH. It is instructive to examine closer the modeling approaches followed in previous related studies, namely in regard to the temperature profile and reaction model configuration. Amato et al. [2] compared experimental data with results obtained from chemical kinetics calculations under adiabatic conditions. They found actual flame temperatures to be significantly lower than the theoretical adiabatic flame temperature so they decided to use the actual measured temperature at the probe location when comparing the two types of results. However, they did this by diluting the system with CO2 to match actual and adiabatic temperatures at the sampling location, which certainly brings unexpected chemical effects. Coppens et al. [13] adopted a strategy to account for downstream heat losses, which avoids exhaustive temperature profile measurements, by applying a constant temperature gradient to the adiabatic temperature profile calculated by the energy equation at the post flame zone. This can be questionable since heat transfer is a complex problem, which may not result in a linear temperature profile at the post flame zone. However, in their work, experiments and calculations were found in good agreement. In regard to the reaction model configuration, Li et al. [14] performed chemical kinetics analysis of oxy-fuel combustion by combining two reactor models to simulate a fully premixed flame; specifically, a perfectly stirred reactor (PSR) simulates the recirculation zone and a plug flow reactor (PFR) is connected downstream for the post flame zone, which has computational advantages, but the PSR residence time must be chosen somewhat arbitrarily, in the absence of computational fluid dynamics (CFD) calculations that can give more reliable values for a specific combustor. Previous chemical kinetic studies [3,5,8–11] showed that stoichiometry and temperature are very important parameters in establishing CO and NO formation/consumption. Moreover, the presence of hydrocarbon species such as CH4 in the feed stream [3,8,9,11], in contrast to their absence [5,10], may also affect the formation/consumption of CO and NO. Watanabe et al. [11] found that, due to the presence of a CH4 flat flame in their study, no competition between CO2 + H M CO + OH and O2 + H M O + OH was observed in contrast to Giménez-López et al. [10]. However, the study of Watanabe et al. [11] was made only under reducing conditions, where the presence of hydrocarbon species is globally higher due to incomplete oxidation. Most fundamental flow reactor studies [3,5,8,9] pointed out that the competition between those reactions as having a crucial role in establishing the amount and portioning of O/H radicals and thus the formation/consumption of CO and NO. Despite the existence of several studies involving both experimental and chemical kinetics modeling under oxy-fuel combustion conditions, it should be pointed out that most of them are limited to plug flow reactors [3,5,8–10]. Studies on premixed flames under oxy-fuel conditions are relatively scarce and only very few include detailed chemical kinetics analysis [2,11]. The present work is intended to help redress this problem by studying simultaneously CO and NO formation/consumption during oxy-methane combustion under fuel-lean reacting conditions in a laboratory premixed laminar burner operating at atmospheric pressure. This is the first detailed study on the chemical kinetics for these particular conditions. Furthermore, the kinetic model was used with actual temperature profiles and an optimized mechanism for oxy-fuel conditions, yielding further insight in the relevant combustion chemistry compared with previous related studies [2,11].

2. Experimental setup, techniques and conditions Figure 1 shows the experimental set-up used in this study. It consists of a vertically oriented stainless steel tube with an inner

Please cite this article in press as: M. Barbas et al., Combust. Flame (2014), http://dx.doi.org/10.1016/j.combustflame.2014.10.020

3

M. Barbas et al. / Combustion and Flame xxx (2014) xxx–xxx

Exhaust

Water

Filter

Pump

CH4 NH3

O2

Aalborg GFM

Aalborg GFM

Aalborg GFM

Thermocouple 10 cm

FI

Thermocouple 20 cm

Abb

Water

Omega FMA

Thermocouple 40 cm

CO2 Air compressor

Fig. 1. Experimental set-up.

diameter of 4 cm and a length of 50 cm, and a premixed laminar water-cooled burner operating at atmospheric pressure, mounted at the bottom end of the stainless steel tube. Methane and small amounts of NH3, from gas cylinders, are fed to the burner as fuel. As for the oxidizer, air, from a compressor, or mixtures of O2 and CO2, from gas cylinders, are supplied to the burner. The flow rates of CH4, NH3, air, O2 and CO2 were measured with mass flow meters. Local mean gas temperature measurements along the centerline of the stainless steel tube were obtained using fine wire (27 lm) thermocouples of platinum/platinum: 13% rhodium. The hot junction was installed and supported on 350 lm wires of the same material as that of the junction. The 350 lm diameter wires were located in a twin-bore alumina sheath with an external diameter of 4 mm and placed inside a stainless steel tube. The uncertainty due to radiation heat transfer was estimated to be less than 5% by considering the heat transfer by convection and radiation between the thermocouple bead and the surroundings. Flue gas composition data were obtained using a stainless steel water-cooled probe, placed at a fixed position (40 cm above the burner), near the top end of the stainless steel tube. The probe was composed of a central 1.3 mm inner diameter tube through which quenched samples were evacuated. This central tube was surrounded by two concentric tubes for probe cooling. The analytical instrumentation included a magnetic pressure analyzer for O2 measurements, a non-dispersive infrared gas analyzer for CO2 and CO measurements, a flame ionization detector for hydrocarbons (HC) measurements and a chemiluminescent analyzer for NOx measurements. Note that NOx measurements include both NO and NO2. The chemical kinetic calculations discussed below confirm that in the post flame zone the concentrations of NO2 were below 0.5% of those of NO. At the measuring location, near the top end of the stainless steel tube, probe effects were negligible and errors arose mainly from quenching of chemical reactions, which was found to be adequate. Repeatability of the flue-gas data was, on average, within 10% of the mean value.

A fixed fuel thermal input of 1.5 kW was used throughout the experiments. The NH3 concentration used to simulate the fuel bound nitrogen was 1.8%, which is a value close to that used by Mendiara and Glarborg [8] and that coincides with the typical nitrogen content found in many coals [7,15–17]. Five excess oxygen coefficients (1.05, 1.1, 1.15, 1.2 and 1.25), defined as (moxygen/mfuel)actual/(moxygen/mfuel)stoichiometric, were used throughout the experiments. For each excess oxygen coefficient, the oxidizer composition was varied by changing the CO2 mass flow rate. The oxidizers studied were 21% O2/79% CO2, 23% O2/77% CO2, 25% O2/75% CO2, 27% O2/73% CO2, 29% O2/71% CO2, and air.

3. Chemical kinetic study The chemical kinetic study was performed with CHEMKIN PRO [18] using PREMIX. A number of approaches have been used in similar modeling studies. For example, Watanabe et al. [11] used a PFR to model their experimental premixed flame and Li et al. [14] used both a PSR and a PFR to model experiments performed in a fully premixed flame, as discussed earlier. Contrarily to Amato et al. [2], who used also PREMIX, the flame structure in the present study was modeled with downstream heat losses. To this end, temperature profiles obtained from the experiments were introduced in the PREMIX. The chemical kinetic mechanism employed in this work was developed by Mendiara and Glarborg [8], and involves 97 species and 779 elementary reactions. This reaction mechanism draws on results from the oxidation of CO/H2, C1–C2 hydrocarbons, NH3 and HCN, as well as interactions of these components. The reaction path, for reactive N conversion in CH4 oxidation, is divided in three main pathways. In the first pathway, NH3 is converted into NO through HNO. In the second pathway, hydrogen cyanide is formed from NH3 via CHxNHy species. Finally, in the third pathway, HCN oxidizes to form NO. Once NO is formed it can stabilize to N2 [8].

Please cite this article in press as: M. Barbas et al., Combust. Flame (2014), http://dx.doi.org/10.1016/j.combustflame.2014.10.020

4

M. Barbas et al. / Combustion and Flame xxx (2014) xxx–xxx

oxy_21 oxy_23 oxy_25

120

oxy_27 oxy_29 air

Temperature (K)

CO (dry volume ppm)

150

90 60 30 0 1

1.05

1.1

1.15

1.2

1.25

2200 2000 1800 1600 1400 1200 1000 800 600 400 200 0

oxy_21 oxy_25 oxy_29 air

0

1.3

5

10

15

20

Fig. 2. Measured CO emissions as a function of the excess oxygen coefficient for all oxidizer compositions studied.

Figure 2 shows the measured CO emissions as a function of the excess oxygen coefficient for all oxidizer compositions studied. Measurements reveal that as the mixture becomes leaner the CO emissions decrease mainly because there is more O2 available for the oxidation of the CO. It is also seen that the CO emissions are lower for the oxidizer with the higher O2 concentration. For the case wherein air is used as the oxidizer, CO emissions are essentially independent of the stoichiometry and always lower than those observed in the O2/CO2 atmospheres. Previous experimental studies [3–5,12] support the CO emission trends obtained here. Figure 3 shows the measured NO emissions as a function of the excess oxygen coefficient for all oxidizer compositions studied. The figure includes a condition where the fuel is composed by 100% of CH4 (fuel w/o NH3 in the figure). Recall that under oxy-fuel conditions NO is entirely formed from the conversion of NH3. The figure reveals that the NO emissions slightly decrease as the excess oxygen coefficient increases and that, under oxy-fuel conditions, NO formation is higher for the oxidizer with the higher content of oxygen (oxy_29). It is also seen that under air-fired conditions the NO emissions are higher than those from the oxy-fuel cases. Previous experimental studies [8,15] support the NO emission trends obtained here. Figure 4 shows typical temperature profiles along the centerline of the stainless steel tube for four oxidizer compositions at a fixed excess oxygen coefficient of 1.15. These profiles, used in the chemical kinetic calculations, were generated from the experimental data following the procedure proposed by Coppens et al. [13]. To account for the downstream heat losses, adiabatic flame profiles were calculated and modified ahead of the flame (beyond 2 cm from the burner) considering the temperature measurements. Figure 5 shows the measured temperatures as a function of the excess

NO (dry volume ppm)

oxy_21 oxy_27

oxy_23 oxy_29

oxy_25 air air (fuel w/o NH3)

1200 1000 800 600 400 200 0 1.05

1.1

1.15

1.2

1.25

1.3

Excess oxygen coefficient Fig. 3. Measured NO emissions as a function of the excess oxygen coefficient for all oxidizer compositions studied.

Temperature (K)

4.1. Experimental results

1

35

40

2100 , ,

2000

1400

30

Fig. 4. Typical temperature profiles along the centerline of the stainless steel tube for four oxidizer compositions at a fixed excess oxygen coefficient of 1.15.

4. Results and discussion

1600

25

Axial position (cm)

Excess oxygen coefficient

1900

oxy_21 oxy_29

, ,

oxy_25 air

1800 1700 1600 1500 1400 1300 1200 1

1.05

1.1

1.15

1.2

1.25

1.3

Excess oxygen coefficient Fig. 5. Measured temperatures as a function of the excess oxygen coefficient for four oxidizer compositions at the centerline of the stainless steel tube for axial distances from the burner of 10 cm (open symbols) and 40 cm (closed symbols).

oxygen coefficient for four oxidizer compositions at the centerline of the stainless steel tube for axial distances from the burner of 10 cm and 40 cm. The figure reveals that the differences between the measured temperatures are higher in the flame region, decreasing towards the top end of the stainless steel tube. The cases represented in Figs. 4 and 5, i.e., the oxy_21, oxy_25, oxy_29 and air cases present adiabatic flame temperatures of 1648 K, 1856 K, 2030 K and 2089 K, respectively. Combustion in air yields higher adiabatic flame temperatures because the molar heat capacity of N2 is 1.66 times lower than that of CO2. Under oxy-fuel conditions, as the CO2 concentration in the oxidizer increases, the adiabatic temperature decreases. It is interesting to note that the measured temperatures closer to the flame (at an axial distance from the burner of 10 cm) are higher for air, decreasing as the excess air coefficient increases; and, under oxy-fuel conditions, for a given excess air coefficient, the measured temperatures increase as the oxygen content in the oxidizer increases. However, at an axial distance from the burner of 40 cm, the measured gas temperatures are still higher for air, but increase as the excess oxygen coefficient increases; and, under oxy-fuel conditions, decrease as the oxygen content in the oxidizer increases. Significant changes in the heat transfer characteristics occur when N2 is replaced by CO2. Radiative heat transfer depends strongly on temperature due to its fourth order dependence, but under oxy-fuel conditions the presence of CO2 and sometimes H2O greatly enhances the gas absorptivity and emissivity and, thus, the radiative heat transfer. The convective heat transfer is also different since it depends on the gas properties such as viscosity, thermal conductivity, heat capacity and density, as well as on velocity. Despite being a complex system, it is instructive to look at the residence time allowed for heat transfer between the combustion products and the surroundings. Table 1 shows the unburned (inlet) velocity as a function of the excess oxygen coefficient for four oxidizer compositions. The

Please cite this article in press as: M. Barbas et al., Combust. Flame (2014), http://dx.doi.org/10.1016/j.combustflame.2014.10.020

5

1.05

1.1

1.15

1.2

1.25

40.5 34.5 30.3 40.5

42.3 36.0 31.5 42.3

44.0 37.4 32.8 44.0

45.8 38.9 34.1 45.8

47.6 40.4 35.3 47.6

Table 2 Residence time (ms) as a function of the excess oxygen coefficient for four oxidizer compositions.

33

8

pred_air

31

6

29

4

27

2

25

0 1

1.05

1.1

1.15

1.2

1.25

1.3

Fig. 6. Measured (scale on the left) and predicted (scale on the right) CO emissions as a function of the excess oxygen coefficient for air.

Excess oxygen coefficient 1.05

1.1

1.15

1.2

1.25

189 220 250 170

184 213 241 164

178 207 233 159

172 200 225 153

167 194 218 148

unburned velocity and the temperature profile establish the residence time. Table 2 shows the residence time as a function of the excess oxygen coefficient for four oxidizer compositions. The residence time was calculated dividing the distance between the flame front (assumed to be at 2 cm from the burner, which coincides with the point where the temperature profiles were modified) and the sampling location (40 cm) by the mean axial velocity of the burned mixture. Assuming that the inlet temperature is 300 K for all conditions studied, higher inlet velocities as well as higher gas temperatures lead to lower residence times. Higher residence times allow for higher heat transfer between the combustion products and the surroundings, which is consistent with the temperature profiles behavior addressed earlier. Amato et al. [2] used adiabatic flame calculations to predict emissions from experiments with high heat losses, which were measured at an axial position in their combustor corresponding to a residence time of 40 ms for all conditions. This is a very short residence time, which in the present study would correspond to an axial position of 10 cm, where the measured temperatures behave exactly like the adiabatic flame temperatures when the stoichiometry and the oxygen concentration in the oxidizer are varied. Moreover, CO emissions, obtained from predictions under adiabatic conditions, increase as the oxygen content in the oxidizer increases, unlike what happens in the present experiments. This means that CO concentration depends strongly on temperature. Actually, Persis et al. [19] were confronted with completely disagreeing experimental and predicted CO trends as they modeled experiments with the freely propagating configuration of the PREMIX model, which solves the energy equation together with the transport equations to calculate both the temperature and the gas species. All these aspects corroborate the necessity of using actual temperature profiles, as it is done in this study. 4.2. Chemical kinetics Figures 6 and 7 show the measured and predicted CO emissions as a function of the excess oxygen coefficient for air and three oxidizer compositions, respectively, and Figs. 8 and 9 show the measured and predicted NO emissions as a function of the excess oxygen coefficient for air, with and without NH3 in the fuel, and three oxidizer compositions, respectively. Note that experimental and predicted data are plotted with different scales in order to gain qualitative insight to the results since CO is slightly underpredicted and NO is largely overpredicted (except for the condition where

120

160 pred_oxy_21 pred_oxy_25 pred_oxy_29

exp_oxy_21 exp_oxy_25 exp_oxy_29

140

100

120

80

100

60

80

40

60

20

40

0 1

1.05

1.1

1.15

1.2

1.25

Predicted CO (dry volume ppm)

oxy_21 oxy_25 oxy_29 Air

10 exp_air

1.3

Excess oxygen coefficient Fig. 7. Measured (scale on the left) and predicted (scale on the right) CO emissions as a function of the excess oxygen coefficient for three oxidizer compositions.

1400

2000

1120

1600

840

1200 pred_air pred_air (fuel w/o NH3)

exp_air exp_air w/o NH3

560

800 400

280

0

0 1

1.05

1.1

1.15

1.2

1.25

Predicted NO (dry volume ppm)

Oxidizer

35

Excess oxygen coefficient

Experimental CO (dry volume ppm)

oxy_21 oxy_25 oxy_29 Air

Excess oxygen coefficient

Experimental NO (dry volume ppm)

Oxidizer

Experimental CO (dry volume ppm)

Table 1 Unburned velocity (cm/s) as a function of the excess oxygen coefficient for four oxidizer compositions.

Predicted CO (dry volume ppm)

M. Barbas et al. / Combustion and Flame xxx (2014) xxx–xxx

1.3

Excess oxygen coefficient Fig. 8. Measured (scale on the left) and predicted (scale on the right) NO emissions as a function of the excess oxygen coefficient for air with and without ammonia in the fuel.

CH4 is not doped with NH3 when burned in air). The relationship between the experimental and predicted CO values is (predicted CO) = (experimental CO)  (K1), where K1 is 25 and 40 for the air and the oxy-fuel cases, respectively. In the case of the NO, the relationship is (predicted NO) = (experimental NO)  (K2), where K2 is 1.4 and 2.7 for the air and the oxy-fuel cases, respectively. The overall qualitative agreement between the experimental data and the predictions is satisfactory. However, quantitatively there are discrepancies, which are more pronounced for the NO results. The use of a pollutant emission rate in ng/J reveals that the oxidizer composition has little effect on NO formation from NH3. Figure 10 shows the measured and predicted NO emissions, in ng/J, as a function of the excess oxygen coefficient for three oxidizer compositions. These trends demonstrate that due to the fact

Please cite this article in press as: M. Barbas et al., Combust. Flame (2014), http://dx.doi.org/10.1016/j.combustflame.2014.10.020

M. Barbas et al. / Combustion and Flame xxx (2014) xxx–xxx

3250

960

2600

720

1950

480

1300 exp_oxy_21 exp_oxy_25 exp_oxy_29

240

pred_oxy_21 pred_oxy_25 pred_oxy_29

650

0

1.0E+01

CO (dry mole fraction)

1200

Predicted NO (dry volume ppm)

Experiemental NO (dry volume ppm)

6

1.0E-01 1.0E-03 1.0E-05 1.0E-07 1.0E-09

1.0E-13 1.0E-15 1.0E-17 0

0 1

1.05

1.1

1.15

1.2

1.25

oxy_21 oxy_25 oxy_29 air

1.0E-11

5

10

15

20

25

30

35

40

Axial position (cm)

1.3

Excess oxygen coefficient Fig. 9. Measured (scale on the left) and predicted (scale on the right) NO emissions as a function of the excess oxygen coefficient for three oxidizer compositions.

Fig. 11. Typical predicted axial CO concentration profiles for four oxidizer compositions at a fixed excess oxygen coefficient of 1.15.

2.5E-03

175

455

150

390

125

325

100

260

NO (dry mole fraction)

520 pred_oxy_21 pred_oxy_25 pred_oxy_29

exp_oxy_21 exp_oxy_25 exp_oxy_29

Predicted NO (ng/J)

Experimental NO (ng/J)

200

2.0E-03 1.5E-03 oxy_21 oxy_25 oxy_29 air

1.0E-03 5.0E-04 0.0E+00 0

1

1.05

1.1

1.15

1.2

1.25

5

10

Excess oxygen coefficient Fig. 10. Measured (scale on the left) and predicted (scale on the right) NO emissions as a function of the excess oxygen coefficient for three oxidizer compositions.

that for a fixed thermal input, varying the oxygen flow rate for each stoichiometry and, then, for each stoichiometry, varying the CO2 flow rate to achieve the desired oxygen composition brings some dilution effects affecting the concentration. However, although affected by dilution, the CO trends are kept, as well as the NO trends as a function of the excess oxygen coefficient. A closer examination of the rates of production of NO for all conditions studied (presented below) led to the conclusion that, contrarily to what is seen in the trends represented in terms of ppm, an increase in the oxygen concentration in the oxidizer results in lower NO formation. Several studies similar to this one adopted the Gas Research Institute (GRI) mechanism [20] for their chemical kinetic studies. In this study, it was made a comparison between predicted CO and NO emissions using the GRI and the mechanism of Mendiara and Glarborg (M&G) [8] as a function of the excess oxygen coefficient for four oxidizer compositions (not shown). It was concluded that the CO predictions using the GRI mechanism are slightly lower than those obtained using the M&G mechanism. In regard to the NO predictions, the differences between the results from both mechanisms are marginal. It is interesting to note that predicted NO emissions using both mechanisms are similar in spite of the significant difference in regard to the chemistry of the ammonia oxidation. In fact, the GRI mechanism comprises 325 reactions, while the M&G mechanism includes 779 reactions. It might be concluded that for the present conditions both mechanisms yield similar results. In order to gain further insight into the present results, a rateof-production (ROP) analysis together with a sensitivity analysis were performed for both CO and NO. Figures 11 and 12 show typical predicted axial CO and NO concentration profiles, respectively, for four oxidizer compositions at a fixed excess oxygen coefficient of 1.15. As expected, the CO formed in the flame region is

15

20

25

30

35

40

Axial position (cm)

1.3

Fig. 12. Typical predicted axial NO concentration profiles for four oxidizer compositions at a fixed excess oxygen coefficient of 1.15.

subsequently oxidized to some extent. NO is initially formed and reaches a plateau level where it is very slightly reduced along the post flame zone. In air-fired combustion NO is globally formed through the entire reaction zone. The set of reactions that have a major contribution to CO formation does not differ much between oxy-fuel and air-fired combustion (cf. Fig. 13):

CH3 þ O $ CO þ H þ H2

ð1Þ

HCO þ O2 $ CO þ HO2

ð2Þ

HCO ðþMÞ $ CO þ H ðþMÞ

ð3Þ

However, in oxy-fuel combustion, due to the presence of large concentrations of CO2, and sufficiently reactive radicals such as methylene, the reaction where the CO2 molecule is broken forming CO has also a major contribution: 1

CH2 þ CO2 $ CH2 O þ CO

ð4Þ

Glarborg and Bentzen [3] found that steps where methylene reacts with CO2 are responsible for 10–20% of the consumption of CO2. The CO oxidation process is quite well established. CO oxidation takes place almost solely through the following reaction:

CO þ OH $ CO2 þ H

ð5Þ

It is unambiguous that this is a key reaction in oxy-fuel combustion where a large amount of CO2 is present. Actually, the reverse step of reaction (5) contributes to CO formation at medium to high temperatures (flame region), due to equilibrium reasons [5,8]. However, for oxidizers with 29% of oxygen it slightly reduces CO in that same region. In the presence of air, this reaction is the only step to CO oxidation, through the entire reaction zone, according to the rate-ofproduction analysis.

Please cite this article in press as: M. Barbas et al., Combust. Flame (2014), http://dx.doi.org/10.1016/j.combustflame.2014.10.020

7

M. Barbas et al. / Combustion and Flame xxx (2014) xxx–xxx

Absolute Rate of Production CO

Absolute Rate of Production CO

HO2+CO HCO+O2 H + CO + M HCO + M H + H2 + CO O + CH3 HCO + H2O H + CO + H2O CH2(S) + CO2 CO + CH2O HCCO + O2 OH + 2CO H + CO2 OH + CO H + OH + CO CH2(S) + O2 OH + H + CO CH2 + O2 CH3 + CO H + CH2CO HO2 + CO OH + CO2 CO + H2O CH2(S) + O2 CH3 + HCO CH4 + CO O + C2H2 CO + CH2 H2 + CO H + HCO

OH + CO H + CO2 O + CH3 H + H2 + CO HCO + H2O H + CO + H2O HCO + M H + CO + M HCO + O2 HO2 + CO H + HCO H2 + CO OH + HCO H2O + CO CH2(S) + O2 H + OH + CO CH2 + O2 OH + H + CO O + HCO OH + CO O + HCCO H + 2CO CH2(S) + O2 CO + H2O H + HCCO CH2(S) + CO O + CO C + O2 CH2(S) + CO2 CO + CH2O

(a)

0.0 -4.92E-5

(b)

-3.01E-3

9.79E-4

0.0

1.26E-3

Fig. 13. CO absolute rate of production at the flame zone (x  1 cm) for two oxidizer compositions: (a) 21% O2/79% CO2 and (b) air at a fixed excess oxygen coefficient of 1.15.

Normalized Sensitivity CO

Normalized Sensitivity CO

O + OH H + O2 C2H6(+M) 2CH3(+M) HO2 + CH3 OH + CH3O CH3 + O2 O + CH3O CH3 + H2 H + CH4 H + CH3(+M) CH4(+M) H + C2H5 2CH3 CH3 + H2O OH+ CH4 CH3 + O2 OH + CH2O O2 + CH2O HO2 + HCO CO + CH2O CH2(S) + CO2 H + H2O OH + H2 HCO + CH4 CH3 + CH2O 2OH H + HO2 O2 + H 2 H + HO2

OH + CO H + CO2 H + O2 O + OH CH3 + H2O OH + CH4 H + CH3(+M) CH4(+M) CH2 + O2 2H + CO2 H + O2 + H2O HO2 + H2O CH2(S) + H2O OH + CH3 O + CH3 H + CH2O HO2 + CO HCO + O2 H + CO + H2O HCO + H2O H + H2O OH + H2 H + CO + M HCO + M H + CH4 CH3 + H2 HO2 + CH3 OH + CH3O O + CH3 H + H2 + CO

-1.92E0

(a)

4.08E0

0.0

(b)

-3.15E-1

0.0

5.06E-2

Fig. 14. Normalized first-order sensitivity for CO at the flame zone (x  1 cm) for two oxidizer compositions: (a) 21% O2/79% CO2 and (b) air at a fixed excess oxygen coefficient of 1.15.

Normalized Sensitivity CO

Normalized Sensitivity CO

H + O2 O + OH H + O2 + H2O HO2 + H2O HO2 + M H + O2 + M H + CO2 OH + CO C2H6(+M) 2CH3(+M) OH + CH3O HO2 + CH3 O + CH3O CH3 + O2 H + CH4 CH3 + H2 2OH(+M) H2O2(+M) H + CH3(+M) CH4(+M) CH3 + H2O OH + CH4 2CH3 H + C2H5 CH3 + O2 OH + CH2O 2OH OH + H2O CO + CH2O CH2(S) + CO2

OH + CO H + CO2 H + O2 + H2O HO2 + H2O NO + O + M NO2 + M OH + HO2 O2 + H2O H + O2 + N2 HO2 + N2 OH + H2O2 HO2 + H2O NO + OH NO2 + H H + O2 O + OH NO2 + O NO + O2 H + O2 + M HO2 + M 2OH(+M) H2O2(+M) N2O + H NH + NO H + HNO NH2 + O NH2 + H NH + H2 2OH O + H2O

(a)

-3.84E-1

0.0

(b)

1.25E-1 -4.31E-1

0.0 2.0E-2

Fig. 15. Normalized first-order sensitivity for CO at the beginning of the post flame zone (x = 2 cm) for two oxidizer compositions: (a) 21% O2/79% CO2 and (b) air at a fixed excess oxygen coefficient of 1.15.

Figures 14 and 15 show the top 15 reactions to which CO concentration is sensitive according to their first order sensitivity coefficients at the flame and post flame regions, respectively. CO concentration in the flame region is sensitive, (almost) regardless of the oxidizer composition or stoichiometry (not shown here), to chain branching reactions, such as:

H þ O2 $ O þ OH which is the main chain branching reaction.

ð6Þ

For high oxygen concentrations in the oxidizer (25–29%) in oxyfuel combustion, CO oxidation is also sensitive to reactions (2) and (3), which are among those that contribute more to the CO formation according to the ROP analyses. For low oxygen concentrations in the oxidizer (21–25%) it was found that reaction (4) is also present. CO is equally sensitive to steps that consume CH4 (reaction (7)) and that lead to chain propagation (reaction (8)) or termination (reactions (9) and (10)), (almost) regardless of the oxidizer composition or stoichiometry:

Please cite this article in press as: M. Barbas et al., Combust. Flame (2014), http://dx.doi.org/10.1016/j.combustflame.2014.10.020

8

M. Barbas et al. / Combustion and Flame xxx (2014) xxx–xxx

Absolute Rate of Production NO

Absolute Rate of Production NO

HNO + H H2 + NO NO + OH NO2 + H NO2 + OH HO2 + NO HNO + O2 HO2 + NO HNO + OH NO + H2O HCNO + CO HCCO + NO HNO + M H + NO + M NO + O N + O2 NO + OH NH + O2 NH + NO N2 O + H H + HNCO CH2 + NO NO + OH HNO + O NO + H NH + O HCN + H2O CH3 + NO NO + CO N + CO2

HNO + H H2 + NO NH + O NO + H N + O2 NO + O NO + H H + OH HNO + OH NO + H2O NO + OH NO2 + H HNO + O NO + OH CH2 + NO H + HNCO NO2 + OH HO2 + NO NO + O + M NO2 + M HCN + O CH + NO N2 + O H + NO NO + CO NCO + O N + HCO CH + NO H + NO + M HNO + M

-1.49E-5

(a)

0.0

1.88E-5

(b)

-6.02E-6 0.0

6.82E-5

Fig. 16. NO absolute rate of production at the flame zone (x  1 cm) for two oxidizer compositions: (a) 21% O2/79% CO2 and (b) air at a fixed excess oxygen coefficient of 1.15.

Absolute Rate of Production NO NO + O + M NO2 + M NO + OH NO2 + H H + NO + M HNO + M HNO + OH NO + H2O NO2 + O NO + O2 NO2 + O N + NO H2 + NO HNO + H NO + O N + O2 NO + H N + OH

Absolute Rate of Production NO NO2 + H NO + OH NO + O + M NO2 + M NO2 + OH HO2 + NO H + NO + M HNO + M HNO + OH NO + H2O NO2 + O NO + O2 HNO + H H2 + NO

(a) -5.38E-7

0.0

8.52E-7

(b)

-1.12E-7

0.0

1.02E-7

Fig. 17. NO absolute rate of production at the beginning of the post flame zone (x = 2 cm) for two oxidizer compositions: (a) 21% O2/79% CO2 and (b) air at a fixed excess oxygen coefficient of 1.15.

CH4 ðþMÞ $ CH3 þ H ðþMÞ

ð7Þ

CH3 þ HO2 $ CH3 O þ OH

ð8Þ

2CH3 ðþMÞ $ C2 H6 ðþMÞ

ð9Þ

H þ O2 ðþMÞ $ HO2 ðþMÞ

ð10Þ

In the post flame region, CO concentrations are greatly and negatively sensitive to the main chain branching reaction (6). This means that an increase in these reaction rates would decrease CO formation. It is interesting to note that in this zone, contrarily to the flame region, reaction (5) appears among the reactions to which CO is more sensitive (negatively). The importance of reactions (5) and (6) gives evidence to the competition between those two reactions, thus limiting the size of the O/H pool, under oxy-fuel conditions. In regard to NO, Fig. 16 reveals that its formation occurs under all conditions mostly with HNO as intermediate:

HNO þ H $ NO þ H2

ð11Þ

HNO þ OH $ NO þ H2 O

ð12Þ

HNO þ O $ NO þ OH

ð13Þ

HNO þ O2 $ NO þ HO2

ð14Þ

Reaction (11) has a major contribution, while reaction (12) contributes less. Reactions (13) and (14) contribute even less to NO formation. Another minor pathway proceeds via NH:

NH þ O $ NO þ H

ð15Þ

And, under some conditions, through N:

As pointed out by Mendiara and Glarborg [8], once NO is formed some NO to NO2 interconversion occurs. This interconversion takes place throughout the entire reaction zone. Under oxy-fuel conditions, in the flame region, only reactions (18) and (19) participate in this interconversion, while reactions (18)–(21) have high rates of production/consumption in the post flame zone (cf. Fig. 17). In air-fired combustion, NO to NO2 interconversion also takes place through reactions (18), (19) and (21) in the flame region and, in the post flame zone, only through reactions (18), (20) and (21):

NO2 þ H $ NO þ OH

ð18Þ

HO2 þ NO $ NO2 þ OH

ð19Þ

NO2 þ O $ NO þ O2

ð20Þ

NO þ O ðþMÞ $ NO2 ðþMÞ

ð21Þ

In the flame region there is also some NO reduction. Reactions (22) and (23) are particularly important in O2/CO2 atmospheres, while reactions (23)–(25) are important in O2/N2 atmospheres:

HCCO þ NO $ HNCO þ CO

ð22Þ

3

ð23Þ

CH2 þ NO $ HNCO þ H

CH þ NO $ HCN þ O

ð24Þ

N þ NO $ N2 þ CO

ð25Þ

The reaction responsible for the NO reduction in the post flame zone, for all oxidizer compositions, was found to be:

H þ NO ðþMÞ $ HNO ðþMÞ

N þ OH $ NO þ H

ð16Þ

N þ O2 $ NO þ O

ð17Þ

ð26Þ

The set of reactions to which NO is sensitive to does not change much from those of CO. Figures 18 and 19 show the top 15 reactions to which NO concentration is sensitive according to their first

Please cite this article in press as: M. Barbas et al., Combust. Flame (2014), http://dx.doi.org/10.1016/j.combustflame.2014.10.020

9

M. Barbas et al. / Combustion and Flame xxx (2014) xxx–xxx

Normalized Sensitivity NO

Normalized Sensitivity NO

H + O2 O + OH 2CH3(+M) C2H6(+M) HO2 + CH3 OH + CH3O O + CH3O CH3 + O2 H + CH4 CH3 + H2 CH4(+M) H + CH3(+M) CH3 + H2O OH + CH4 2CH3 H + C2H5 CH3 + O2 OH + CH2O CO + CH2O CH2(S) + CO2 HO2 + HCO O2 + CH2O HCO + CH4 CH3 + CH2O OH + H2 H + H2O O2 + H2 H + HO2 H + HO2 2OH

H + O2 O + OH NH2 + O H + HNO N2O + H NH + NO H + CH3(+M) CH4(+M) O + CH3 H + H2 + CO HO2 + H2O H + O2 + H2O N2 + O N + NO HO2 + CO HCO + O2 HCO + H2O H + CO + H2O NH2 + H NH + H2 HCO + M H + CO + M N + O2 NO + O OH + CH3O HO2 + CH3 NO + H NH + O OH + CH3 O + CH4

-3.28E0

(a)

0.0

7.58E0

- 5 . 5 8 E- 2

(b)

0.0

1.89E-1

Fig. 18. Normalized first-order sensitivity for NO at the flame zone (x  1 cm) for two oxidizer compositions: (a) 21% O2/79% CO2 and (b) air at a fixed excess oxygen coefficient of 1.15.

Normalized Sensitivity NO

Normalized Sensitivity NO

NH + NO N2O + H NH2 + O H + HNO NH + H2 NH2 + H NH + CO2 HNO + CO N + NO N2 + O NH + NO N2 + OH NH2 + OH NH + H2O NH3 + OH NH2 + H2O NH + O NO + H N + O2 NO + O HNO + H NH + OH HNO + H H2 + NO OH + CH4 CH3 + H2O H + O2 O + OH NH + H2O HNO + H2

NH + NO N2O + H NH2 + O H + HNO NH2 + H NH + H2 N + NO N2 + O NH3 + OH NH2 + H2O N + O2 NO + O NH + O NO + H N2 + OH NH + NO NH + OH HNO + H OH + CH4 CH3 + H2O NO + H N + OH NH2 + OH NH + H2O NHO + H2 NH + H2O H + CH4 CH3 + H2 H + CH3(+M) CH4(+M)

-4.71E-2

(a)

0.0

3.0E-2

-4.78E-2

(b)

0.0

3.69E-2

Fig. 19. Normalized first-order sensitivity for NO at the beginning of the post flame zone (x = 2 cm) for two oxidizer compositions: (a) 21% O2/79% CO2 and (b) air at a fixed excess oxygen coefficient of 1.15.

order sensitivity coefficients at the flame and post flame regions, respectively. It was seen that reactions (2) and (3) are important in establishing the NO concentrations for the same conditions found for CO. However, the relative sensitivity is different. NO is positively sensitive to reaction (5), unlike CO. NO concentrations in the flame region are particularly sensitive to steps that consume CH4 and that lead to chain propagation and termination, such as reactions (7)–(9). In the post flame zone, steps involving amine species seem to affect the NO concentrations, while the chain branching reactions no longer play a significant role. In this region, NO does not seem to vary much with stoichiometry or oxidizer composition.

 In air firing the CO emissions are significantly lower than those found in oxy-fuel firing, while the NO emissions are higher than those from the oxy-fuel cases.  The chemical kinetic study allowed identifying the main reactions that directly and indirectly influence both CO and NO emissions.  CO formation, under oxy-fuel conditions, has the contribution of CO2 + H M CO + OH and 1CH2 + CO2 M CH2O + CO additionally to those found in air-fired combustion, while CO oxidation takes place through CO + OH M CO2 + H for all studied conditions.  Formation of NO occurs under all conditions mostly with HNO as intermediate, particularly through HNO + H M H2 + NO. Once NO is formed some NO to NO2 interconversion occurs.

5. Conclusions The present work focuses on the oxy-fuel combustion of methane doped with ammonia in a premixed laminar burner operating at atmospheric pressure. CO and NO formation/emission were examined as a function of the stoichiometry and oxidizer composition. The study includes both experiments and a chemical kinetic study. The main conclusions of the present study are as follows:  The experimental results showed that, for all oxidizer compositions studied, an increase in the excess oxygen coefficient generally decreases both the CO and NO emissions.  For the O2/CO2 environments, decreasing the oxygen concentration in the oxidizer, for a given excess oxygen coefficient, leads to higher CO emissions, but lower NO emissions.

Acknowledgment This work was partially supported by the Fundação para a Ciência e a Tecnologia under the research contract PEst-OE/EME/ LA0022/2011. References [1] IEA, World Energy Outlook, 2012. [2] A. Amato, B. Hudak, P. D’Souza, P. D’Carlo, D. Noble, D. Scarborough, J. Seitzman, T. Lieuwen, Proc. Combust. Inst. 33 (2011) 3399–3405. [3] P. Glarborg, L.B. Bentzen, Energy Fuels 22 (2008) 291–296. [4] P. Heil, D. Toporov, M. Förster, R. Kneer, Proc. Combust. Inst. 33 (2011) 3407– 3413.

Please cite this article in press as: M. Barbas et al., Combust. Flame (2014), http://dx.doi.org/10.1016/j.combustflame.2014.10.020

10

M. Barbas et al. / Combustion and Flame xxx (2014) xxx–xxx

[5] M. Abián, J. Giménez-López, R. Bilbao, M.U. Alzueta, Proc. Combust. Inst. 33 (2011) 317–323. [6] P. Glarborg, M.U. Alzueta, K. Dam-Johansen, J.A. Miller, Combust. Flame 115 (1998) 1–27. [7] F. Normann, K. Andersson, B. Leckner, F. Johnsson, Prog. Energy Combust. Sci. 35 (2009) 385–397. [8] T. Mendiara, P. Glarborg, Combust. Flame 156 (2009) 1937–1949. [9] B. Wang, L.S. Sun, S. Su, J. Xiang, S. Hu, H. Fei, Appl. Energy 92 (2011) 361–368. [10] J. Giménez-López, A. Millera, R. Bilbao, M.U. Alzueta, Combust. Flame 157 (2010) 267–276. [11] H. Watanabe, J. Yamamoto, K. Okazaki, Combust. Flame 158 (2011) 1255– 1263. [12] M. Alves, C. Rosa, M. Costa, Energy Fuels 27 (2013) 561–567.

[13] [14] [15] [16] [17] [18] [19]

F.H.V. Coppens, J. De Ruyck, A.A. Konnov, Combust. Flame 149 (2007) 409–417. H. Li, A.M. ElKady, A.T. Evulet, Am. Inst. Aeronaut. Astronaut. (2009). C.R. Shaddix, A. Molina, Proc. Combust. Inst. 33 (2011) 1723–1730. E. Croiset, K.V. Thambimuthu, Fuel 80 (2001) 2117–2121. H. Liu, R. Zailani, B.M. Gibbs, Fuel 84 (2005) 833–840. CHEMKIN 10113, Reaction Design, San Diego, 2012. S. Persis, G. Cabot, L. Pillier, I. Gökalp, A.M. Boukhalfa, Energy Fuels 27 (2013) 1093–1103. [20] G.P. Smith, D.M. Golden, M. Frenklach, N.W. Moriarty, B. Eiteneer, M. Goldenberg, C.T. Bowman, R.K. Hanson, S. Song, W.C. Gardiner Jr., V.V. Lissianski, Z. Qin, GRI-Mech (version 3.0). .

Please cite this article in press as: M. Barbas et al., Combust. Flame (2014), http://dx.doi.org/10.1016/j.combustflame.2014.10.020