Abstract
The direct-fired supercritical carbon dioxide cycles are one of the most promising power generation methods in terms of their efficiency and environmental friendliness. Two important challenges in implementing these cycles are the high pressure (300 bar) and high CO2 dilution (>80%) in the combustor. The design and development of supercritical oxy-combustors for natural gas require accurate reaction kinetic models to predict the combustion outcomes. The presence of a small amount of impurities in natural gas and other feed streams to oxy-combustors makes these predictions even more complex. During oxy-combustion, trace amounts of nitrogen present in the oxidizer is converted to NOx and gets into the combustion chamber along with the recirculated CO2. Similarly, natural gas can contain a trace amount of ammonia and sulfurous impurities that get converted to NOx and SOx and get back into the combustion chamber with recirculated CO2. In this work, a reaction model is developed for predicting the effect of impurities such as NOx and SOx on supercritical methane combustion. The base mechanism used in this work is GRI Mech 3.0. H2S combustion chemistry is obtained from Bongartz et al. while NOx chemistry is from Konnov. The reaction model is then optimized for a pressure range of 30–300 bar using high-pressure shock tube data from the literature. It is then validated with data obtained from the literature for methane combustion, H2S oxidation, and NOx effects on ignition delay. The effect of impurities on CH4 combustion up to 16 atm is validated using NOx-doped methane studies obtained from the literature. In order to validate the model for high-pressure conditions, experiments are conducted at the UCF shock tube facility using natural gas identical mixtures with N2O as an impurity at ∼100 bar. Current results show that there is a significant change in ignition delay with the presence of impurities. A comparison is made with experimental data using the developed model and predictions are found to be in good agreement. The model developed was used to study the effect of impurities on CO formation from sCO2 combustors. It was found that NOx helps in reducing CO formation while the presence of H2S results in the formation of more CO. The reaction mechanism developed herein can also be used as a base mechanism to develop reduced mechanisms for use in CFD simulations.
1 Introduction
Supercritical CO2 cycles for power generation have received increased attention recently due to the possibilities for zero-emission, high efficiency, and compactness of the turbomachinery components [1–15]. The direct-fired supercritical carbon dioxide (sCO2) cycles are one of the most promising power generation methods in terms of efficiency [1,16] and can be coupled with both syngas and natural gas fuels. High-pressure carbon dioxide produced during their operation makes carbon capture and sequestration easier making it environmentally friendly. Due to this, direct-fired sCO2 cycles are an attractive choice for power generation. However, the extreme operating conditions such as high temperature (∼1500 K), high pressure (∼150–300 bar), and high CO2 dilution makes the implementation challenges. Additionally, the feed to sCO2 combustor, natural gas, may contain a small amount of impurities in the form of hydrogen sulfide and ammonia depending on the gas/coal fields and processing facilities they are obtained from. During combustion, these get converted to sulfur oxides (SOx) and nitrogen oxides (NOx). Since, a part of combustion exhaust is recirculated in sCO2 combustors, incoming natural gas feed undergoes combustion in the presence of these impurities. To understand the effect of these extreme conditions and impurities on sCO2 combustion, systematic experimental and simulation studies are necessary.
Due to the complexity, safety, and cost of experiments, experimental studies on methane combustion at high pressure (>100 bar) and high CO2 dilution are scarce although many studies exist at lower pressures with CO2 addition in the literature [3,9,11,13,14,17–24]. Shao et al. [25] conducted ignition delay measurements in a shock tube at 30–300 bar for lean and stoichiometric CH4/O2/CO2 and H2/O2/CO2 mixtures for a temperature range of 1045–1578 K. They found measured ignition delay times to be shorter with an increase in pressure and pointed out that accurate reaction rate measurements are required for some reactions with large pressure dependency. Barak et al. [15,26] conducted shock tube ignition delay measurements near 100 atm in mixtures of oxy-syngas and oxy-methane added to CO2 bath gas environments. In total, five mixtures were investigated in their study within a pressure range of 70–100 atm and a temperature range of 1050–1350 K. Another interesting result from this work was that Aramco 2.0 [27] mechanism was able to predict ignition delay for their experiments relatively well. Also, there has been recent work by the current authors on assessing the impact of supercritical CO2 on combustion using theoretical methods [13,28–37]. However, literature still lacks experimental studies on the effect of impurities on natural gas combustion at sCO2 combustor relevant conditions.
In this work, an optimized reaction mechanism is developed for sCO2 combustor conditions using C1–C3, H2S, and NOx mechanisms available in the literature. Also, presented herein are the shock tube ignition delay times for natural gas feed containing 1% N2O at 100 bar and temperature range of 1440–1574 K at an equivalence ratio of 1. To the best of the authors knowledge, this is the first study to report ignition delay capturing an impurity effect in natural gas combustion near 100 bar.
2 Methodology
2.1 Kinetic Modeling and Optimization.
To develop a base mechanism for sCO2 combustion in the presence of impurities, a mechanism that accurately describes: (1) C1–C3 combustion chemistry, (2) H2S oxidation and pyrolysis chemistry, and (3) NOx formation chemistry was required.
2.1.1 C1–C3 Chemistry.
For CI–C3 chemistry, the GRI 3.0 [38] mechanism was chosen. GRI 3.0 has 53 species and 325 reactions that are well-validated for combustion of natural gas mixtures containing C1–C3 species. GRI 3.0 is validated for methane ignition delay at lean and rich conditions, shock tube species profiles, flow reactor studies, flame speed measurements, and prompt NO under a wide range of conditions. Since the fuels of interest in this work contain mostly methane, GRI 3.0 will be able to capture the combustion process with reasonable accuracy at atmospheric pressure.
2.1.2 H2S Chemistry.
For H2S oxidation and pyrolysis chemistry, the mechanism proposed by Bongartz et al. [39] was used. This mechanism has 157 species and 1011 reactions and was validated for a wide range of conditions including H2S flame speed measurements, ignition delay times, H2S pyrolysis, and flow reactor studies. Some validations were also conducted in the presence of methane, which takes into account the sulfur–hydrocarbon species interactions. Since the feed is expected to have only trace amounts of SO2 or H2S, this mechanism can yield good results.
2.1.3 NOx Chemistry.
The presence of nitrogen in a combustion chamber results in NOx formation. Additionally, the presence of nitrogen oxides in hydrocarbon mixtures is known to reduce the time for auto-ignition [40]. To adequately capture NOx formation and the variations in auto-ignition, we chose the model from Konnov [41]. This mechanism consists of 129 species and 1231 reactions. This mechanism is validated for the ignition delay of hydrocarbon mixtures with NOx impurities up to 16 atm.
By extracting important reactions for NOx chemistry and H2S chemistry from the above mechanisms and merging them with GRI 3.0, a combined reaction mechanism was obtained. This mechanism has 106 species and 916 reactions. This mechanism was used as a base mechanism in this work for generating a sCO2 combustion mechanism in the presence of NOx and SOx impurities.
The base mechanism was optimized using an in-house optimization code written in matlab coupled with ansys Chemkin Pro [42]. The code edits the rate of constant parameters in the mechanism file and executes Chemkin to run the simulation and provides the mean square error with experimental results. Simulations were conducted using a 0D constant volume closed homogeneous reactor with an energy solver.
Reactions R1 and R1r were chosen since our previous studies on sCO2 combustion revealed that the recombination reaction of methyl radical plays an important role in supercritical conditions [13]. By including these two reactions in the mechanism, the original methyl radical recombination reaction derived from GRI 3.0 sub-mechanism was removed. The ethane thus formed undergoes a series of reactions that are represented by the lumped reaction R2 to form hydroxyl radicals. The high sensitivity of R3 to ignition delay also made it a target reaction for optimization.
3 Shock Tube Experiments
Experiments were conducted in the stainless steel shock tube facility at University of Central Florida (UCF). The shock tube has an inside diameter of 14.17 cm with the driver side separated from the driven section using a metal diaphragm. The test section is located 2.00 cm away from the end section on the driven side. It is equipped with eight optical ports to aid in laser absorption studies. For measuring pressure, one port was installed with a pressure transducer (Kistler 603B1-piezoelectric). Four timer counters (Agilent53220A; 0.1 ns time resolution) were used to measure the time interval, triggered using five piezoelectric transducers (PCB 113B26; 500 kHz frequency response) placed at the last 1.4 m of the driven section. These timer counter values are used to measure the shock velocity, which is then extrapolated linearly to obtain the shock velocity at the end wall. Additional details about shock tube and current experimental strategy are provided in earlier publications [9,11,15,22,23,43–54].
A test mixture was prepared manometrically in a 33-l Teflon-coated stainless steel tank. A fuel mixture (>99.9% purity) identical to natural gas was used, as it is more appropriate for the problem of interest. Argon, oxygen, and CO2 used were of research grade purity (>99.99%). A 100 Torr (MKS Instruments/Baratron E27D) and 10,000 Torr (MKS Instruments/Baratron 628D) full-scale range capacitance manometers were used to measure the pressure during the filing process. The errors of these manometers were 0.12% and 0.25%, respectively. Helium was used as the driver gas for all experiments. After preparing the mixture, it was stirred for more than 8 h using a magnetically driven stirrer to ensure mixture homogeneity.
4 Results and Discussions
4.1 Mechanism Optimization.
The base mechanism generated above has to be optimized for conditions relevant for sCO2 combustion. For this purpose, the high-pressure experimental data from Shao et al. [25] were used. They conducted shock tube studies at 30–300 bar with methane and hydrogen in high CO2 dilution. The experimental data considered for lean and stoichiometric oxidation of methane are shown in Table 1. Table 2 shows the ignition delay data for high-pressure combustion of hydrogen at high CO2 dilution. Data from both Tables 1 and 2 were considered for optimization.
Mixture composition | |||||
---|---|---|---|---|---|
ϕ = 0.5, CO2 dilution—86.17% | ϕ = 1, CO2 dilution—77.5% | ||||
T (K) | P (atm) | IDT (μs) | T (K) | P (atm) | IDT (μs) |
1082 | 266.8 | 1098 | 1045 | 259.7 | 636 |
1086 | 269.4 | 1098 | 1055 | 245.2 | 518 |
1144 | 262.7 | 673 | 1087 | 249.4 | 359 |
1352 | 75.1 | 541 | 1100 | 285.5 | 275 |
1359 | 78.6 | 748 | 1265 | 109.4 | 427 |
1380 | 74.5 | 408 | 1277 | 111.8 | 367 |
1428 | 73.1 | 242 | 1346 | 34.4 | 478 |
1447 | 35.7 | 311 | 1362 | 103.6 | 156 |
1453 | 72.2 | 202 | 1374 | 27.1 | 450 |
1480 | 32.2 | 289 | 1411 | 100.5 | 85 |
1507 | 27.5 | 189 | 1421 | 32.8 | 206 |
1578 | 30.3 | 57 | 1434 | 35.4 | 202 |
Mixture composition | |||||
---|---|---|---|---|---|
ϕ = 0.5, CO2 dilution—86.17% | ϕ = 1, CO2 dilution—77.5% | ||||
T (K) | P (atm) | IDT (μs) | T (K) | P (atm) | IDT (μs) |
1082 | 266.8 | 1098 | 1045 | 259.7 | 636 |
1086 | 269.4 | 1098 | 1055 | 245.2 | 518 |
1144 | 262.7 | 673 | 1087 | 249.4 | 359 |
1352 | 75.1 | 541 | 1100 | 285.5 | 275 |
1359 | 78.6 | 748 | 1265 | 109.4 | 427 |
1380 | 74.5 | 408 | 1277 | 111.8 | 367 |
1428 | 73.1 | 242 | 1346 | 34.4 | 478 |
1447 | 35.7 | 311 | 1362 | 103.6 | 156 |
1453 | 72.2 | 202 | 1374 | 27.1 | 450 |
1480 | 32.2 | 289 | 1411 | 100.5 | 85 |
1507 | 27.5 | 189 | 1421 | 32.8 | 206 |
1578 | 30.3 | 57 | 1434 | 35.4 | 202 |
Mixture composition | |||||
---|---|---|---|---|---|
10%H2/5%O2/85%CO2 | 5%H2/10%O2/85%CO2 | ||||
T (K) | P (atm) | IDT (µs) | T (K) | P (atm) | IDT (µs) |
1083 | 251.4 | 366 | 1170 | 40.067 | 354 |
1087 | 275.4 | 446 | 1201 | 39.287 | 294 |
1141 | 244.5 | 147 | 1221 | 37.055 | 216 |
1143 | 311 | 135 | 1224 | 38.204 | 180 |
1154 | 113 | 258 | 1270 | 37.587 | 90 |
1161 | 111 | 227 | – | – | – |
1182 | 110.92 | 176 | – | – | – |
1291 | 103.4 | 36 | – | – | – |
Mixture composition | |||||
---|---|---|---|---|---|
10%H2/5%O2/85%CO2 | 5%H2/10%O2/85%CO2 | ||||
T (K) | P (atm) | IDT (µs) | T (K) | P (atm) | IDT (µs) |
1083 | 251.4 | 366 | 1170 | 40.067 | 354 |
1087 | 275.4 | 446 | 1201 | 39.287 | 294 |
1141 | 244.5 | 147 | 1221 | 37.055 | 216 |
1143 | 311 | 135 | 1224 | 38.204 | 180 |
1154 | 113 | 258 | 1270 | 37.587 | 90 |
1161 | 111 | 227 | – | – | – |
1182 | 110.92 | 176 | – | – | – |
1291 | 103.4 | 36 | – | – | – |
Optimization was performed to minimize the error in predicting experimental results for the temperature range of 1080–1580 K and 1–300 bar. The results obtained using the optimized mechanism is shown in Fig. 1. It can be seen that the optimized mechanism predicts experimental ignition delay times with good accuracy. Optimized reaction mechanism is available upon request from the authors.
4.2 Mechanism Validation.
To ensure that the optimized mechanism performs well in predicting any impurity’s effect in hydrocarbon combustion, further validations were conducted.
4.2.1 Validation With NO2 as an Impurity.
For validating the effect of NO2 as an impurity, experimental data were obtained from Zhang et al. [55]. They studied ignition delay of methane in the presence of NO2 at 5–16 atm in the temperature range of 1100–1900 K. Figure 2 shows the experimental data along with model predictions. It is clear that the mechanism captures the effect of NO2 on the ignition delay of methane combustion.
4.2.2 Validation With N2O as an Impurity.
The effect of small amount of N2O on ignition delay of methane is studied in Deng et al. [40]. They conducted shock tube studies of methane oxidation at equivalence ratio 0.5 to 2. The experiments were conducted in temperature range of 1220–2336 K and pressure ranging from 1.2 to 16 atm. Figure 3 displays the experimental data along with model predictions for equivalence ratio 1. The results show that the model captures N2O impurity effect satisfactorily.
4.2.3 Validation for High-Pressure Oxidation of H2S.
To confirm that H2S oxidation is predicted well by the model, a validation is carried out with the high-pressure oxidation study conducted by Frenklach et al. [56]. The results obtained are shown in Fig. 4. It can be seen that model predicts experimental results very well in the high-temperature region. In the low-temperature regime, a slight deviation is observed.
4.2.4 Validation to Confirm GRI 3.0 Performance.
Since the model was optimized only for sCO2 combustion, it was necessary to confirm whether the mechanism retains GRI 3.0 mechanism performance at low-pressure conditions. Hence, a set of validations were carried out with the works of Seery and Bowman [57]. The results obtained are shown in Fig. 5. It can be seen that the predictions from the present work (sCO2 + SOx + NOx) and GRI 3.0 mechanism overlap each other. This confirms that the performance of GRI 3.0 mechanism is retained at atmospheric pressure for methane ignition delay.
The Allam cycle sCO2 combustor has an operating pressure close to 300 bar; therefore, it is necessary to validate the mechanism at pressures close to 300 bar. However, data beyond 17 bar (validation case 1 and 2) were not available in literature for methane ignition delay in presence of impurities. Hence, further experiments are necessary for validating/improving the model.
4.3 Experimental Results and Model Comparison.
Experiments were conducted using natural gas feed in the presence and absence of nitrous oxide (N2O) at pressures in the range of 100–110 bar and temperature of 1400–1580 K. The mixtures used and their compositions are listed in Table 3. In the case of mixture 1, N2O is replaced with argon. The ignition delay data obtained from experiments are shown in Fig. 6. It can be seen that ignition delay time is faster in the presence of N2O (the model also predicts faster ignition delay with N2O).
Description | Mixture 1—Without N2O | Mixture 2—With N2O |
---|---|---|
CH4 | 0.0157 | 0.0157 |
C2H6 | 0.0008 | 0.0008 |
C3H8 | 0.0003 | 0.0003 |
IC4H10 | 0.0001 | 0.0001 |
C4H10 | 0.0001 | 0.0001 |
O2 | 0.0362 | 0.0362 |
CO2 | 0.4470 | 0.4470 |
AR | 0.4999 | 0.4899 |
N2O | 0.0000 | 0.0100 |
Description | Mixture 1—Without N2O | Mixture 2—With N2O |
---|---|---|
CH4 | 0.0157 | 0.0157 |
C2H6 | 0.0008 | 0.0008 |
C3H8 | 0.0003 | 0.0003 |
IC4H10 | 0.0001 | 0.0001 |
C4H10 | 0.0001 | 0.0001 |
O2 | 0.0362 | 0.0362 |
CO2 | 0.4470 | 0.4470 |
AR | 0.4999 | 0.4899 |
N2O | 0.0000 | 0.0100 |
4.4 Comparison With Recent Mechanisms.
The performance of assembled UCF mechanism was compared with two recent mechanisms for natural gas combustion from literature, Aramco mechanism 3.0 [43] and Methling et al. [58] mechanism. Aramco 3.0 has 581 species and 3037 reactions and Methling et al. mechanism have 83 species and 747 reactions. Figure 7 shows the comparison between experimental data from Shao et al. with all the models. At pressure near 30 bar, present model clearly predicts ignition delay better than Aramco 3.0 and Methling et al. models. Near 100 bar, all three models predict ignition delay satisfactorily. At ∼275 bar, prediction by Aramco 3.0 and present model are in good agreement with experiments while Methling et al. model overpredicts ignition delay.
4.5 Effect of Impurities on CO Formation.
Simulations were conducted in Ansys Chemkin Pro using perfectly stirred reactor model and energy solver. The inlet conditions for “pure methane” simulation were 88.75% CO2, 3.75% CH4, and 7.5% O2. For simulations with impurities, 0.05% of CO2 was replaced with the corresponding impurity. The inlet temperature and pressure were 1000 K and 300 bar, respectively, which resulted in a post-combustion temperature close to 1500 K. Figure 8(a) shows the CO concentration versus residence time with 500 ppm of NO2, N2O, and H2S as impurities. The presence of trace amount of N2O and NO2 reduces the amount of CO produced during sCO2 combustion. On the contrary, H2S presents as an impurity resulted in production of more CO. The higher post-combustion temperatures observed (Fig. 8(b)) for cases with N2O and NO2 can be attributed to the efficient conversion of CO to CO2. However, no significant reduction in post-combustion temperature was observed in presence of H2S as impurity even with considerable increase in CO production. This can be attributed to the energy release from H2S oxidation balancing the reduction in energy due to CO formation.
5 Conclusions
In this work, we developed a detailed reaction mechanism for sCO2 combustion in the presence of NOx and SOx impurities. The developed mechanism was validated extensively with data obtained from the literature. Validations were carried out to check whether the mechanism captured the impurity effects of NOx. It was observed that the mechanism was able to capture the impurity effect of NOx on ignition delay time up to 17 bar. Since further validation targets at higher pressure were not available in literature, we conducted experiments using natural gas identical mixture at UCF’s shock tube facility to study the effect of N2O at 100 bar. Results obtained indicated that faster ignition delay time was observed for this mixture in the presence of N2O (when compared with ignition delay time measured without N2O). These results were then used as validation target for the model; good agreement was observed. The generated reaction mechanism was used to predict CO formation from sCO2 combustor in presence of impurities. It was observed that presence of NOx impurities helped in reducing CO while H2S increased CO formation. The reaction mechanism presented here can be used as a base mechanism to generate reduced mechanisms for CFD simulations.
Acknowledgment
This material is mainly based upon work supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award Number DE-SC0019640. Also, this material is partially supported by the U.S. Department of Energy under award numbers DE-FE0025260 and DE-EE0007982.
Disclaimer
This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.
Conflict of Interest
There are no conflicts of interest.