Skeletal MethaneâAir Reaction Mechanism for ... - ACS Publications

Subscriber access provided by University of Newcastle, Australia

Article

Skeletal Methane-Air Reaction Mechanism for Large Eddy Simulation of Turbulent Microwave-Assisted Combustion Anders Larsson, Niklas Zettervall, Tomas Hurtig, Elna Johanna Kristina Nilsson, Andreas Ehn, Per Petersson, Marcus Aldén, Jenny Larfeldt, and Christer Fureby Energy Fuels, Just Accepted Manuscript • DOI: 10.1021/acs.energyfuels.6b02224 • Publication Date (Web): 18 Jan 2017 Downloaded from http://pubs.acs.org on January 24, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Energy & Fuels is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Skeletal Methane-Air Reaction Mechanism for Large Eddy Simulation of Turbulent MicrowaveAssisted Combustion A. Larsson†, N. Zettervall†, T. Hurtig†, E.J.K. Nilsson¶, A. Ehn¶, P. Petersson¶, M. Alden¶, J. Larfeldt‡ and C. Fureby† †

Defence Security Systems Technology, The Swedish Defence Research Agency – FOI, SE 147 25 Tumba, Stockholm, Sweden ¶

Combustion Physics, Lund University, PO Box 118, SE-221 00 Lund, Sweden ‡

Siemens Industrial Turbomachinery AB, SE-612 83 Finspång, Sweden

Keywords: Large Eddy Simulation, skeletal mechanism, plasma, turbulence, microwave irradiation

ABSTRACT Irradiating a flame by microwave radiation is a Plasma Assisted Combustion (PAC) technology that can be used to modify the combustion chemical kinetics in order to improve flame stability and to delay lean blow-out. One practical implication is that combustion engines may be able to operate with leaner fuel mixtures and have an improved fuel flexibility capability including bio-fuels. Furthermore, this technology may assist in reducing thermoacoustic instabilities, a phenomenon that may severely damage the engine and increase NOX production.

ACS Paragon Plus Environment

1

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

To further understand microwave-assisted combustion, a skeletal kinetic reaction mechanism for methane-air combustion is developed and presented. The mechanism is detailed enough to take into account relevant features, but sufficiently small to be implemented in Large Eddy Simulations (LES) of turbulent combustion. The mechanism consists of a proposed skeletal methaneair reaction mechanism accompanied with subsets for ozone, singlet oxygen, chemionization and electron impact reactions. The baseline skeletal methane-air mechanism contains 17 species and 42 reactions, and predicts the ignition delay time, flame temperature, flame speed, major and most minor species well, as well as the extinction strain compared to the detailed GRI3.0 reaction mechanism. The amended skeletal reaction mechanism consists of 27 species and 80 reactions and is developed for a reduced electric field E/N below the critical field strength (of around 125 Td) for the formation of a microwave breakdown plasma. Both laminar and turbulent flame simulation studies are carried out with the proposed skeletal reaction mechanism. The turbulent flame studies consist of propagating planar flames in homogeneous isotropic turbulence in the reaction sheets and the flamelets in eddies regimes, and a turbulent low-swirl flame. Comparison with experimental data is performed for a turbulent low-swirl flame. The results suggest that we can influence both laminar and turbulent flames by non-thermal plasmas based on microwave irradiation. The laminar flame speed increases more than the turbulent flame speed, but the radical pool created by the microwave irradiation significantly increases the lean blow-out limits of the turbulent flame, thus making it less vulnerable to thermoacoustic combustion oscillations. Apart from the experimental results from low-swirl flame presented here, experimental data for validation of the simulated trends are scarce, and conclusions build largely on simulation results. Analysis of chemical kinetics from simulations of laminar flames and LES on turbulent flames reveal that singlet oxygen molecule is of key importance for the increased reactivity, accompanied with


2

Page 2 of 63

Page 3 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

production of radicals like O and OH.

1. Introduction

Gas turbine power plants are one of the cornerstones in the generation of electric energy, and its role is increasing. From 1990 to 2009 the use of gaseous fuels has increased from 9% to 23% of the total electricity generation within EU 1. The fuel normally used in industrial gas turbines is natural gas (mainly methane), and the recent boom in shale gas development makes natural gas more abundant. Gas turbine power plants also play a crucial role for the production of electricity in the context of emission reduction targets and increased penetration of intermittent renewable energy sources. Industrial gas turbines are usually optimized for an efficient combustion of natural gas. However, there is a strong demand to be able to use other fuels, or fuel compositions, like biogas, process gas, hydrogen diluted natural gas and oxy-fuel. Hence, a key issue is to have significant fuel flexibility in gas turbine power plants. Since fuels have different combustion characteristics, a way of compensating for fuel variety would be required. Adding electric energy to the combustion provides an opportunity to compensate for the different combustion characteristics of different fuels, thus allowing for lean combustion with minimized emission regardless of fuel composition. Adding electric energy to a combustion process is a promising technology, often referred to as Plasma Assisted Combustion (PAC) 2-5. The main idea of PAC is to add a comparatively small amount of electric energy to a flame to achieve a relatively large flame enhancing effect that improves flame stability. The electric energy is transferred to free electrons that excite, dissociate, ionize and heat gas molecules. Apart from employing PAC in reforming of combustible gases or flue-gas treatment, the electric ener-


3

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

gy can be coupled directly to the flame to increase the flame speed and to stabilize the flame. Adding electric energy directly to the flame is the most attractive use of PAC with the potential to improve flame stability and to delay lean blow-out, thus allowing for leaner fuel mixtures, or for use of biofuels or other fuels for which the combustion system was not originally designed or optimized for. Deposition of electric energy in the flame region has greater potential than reforming due to short-lived radicals and excited states that may not have sufficient lifetimes to be transported from the reforming zone to the combustion zone6. Adding electric energy via microwave radiation is particularly appealing as the microwave radiation will be absorbed where needed as the reduced electric field will be highest where the electron density is the largest since the chemionization take place in the flame. As an illustration, the ion density in the flame front of an atmospheric premixed laminar propane-air flame with an equivalence ratio of φ=1.2 is 4·1017 m–3.7 Microwave radiation has been studied for PAC in various configurations. In several of these studies, the microwave radiation has been sufficiently strong to create a dielectric breakdown of the gas and thereby creating a microwave plasma, either for reforming8 or ignition9-13, or for direct plasma-flame interaction14-19. A continous microwave plasma is typically close to Local Thermodynamic Equilibrium (LTE), where the electrons and molecules have approximately the same temperature. However, if the electric field of the microwave radiation is below the critical field for dielectric breakdown, a state of non-LTE can be achieved with a substantial higher temperature of the electrons compared to the gas temperature. The approach of augmenting a combustion process by applying microwave energy to generate a non-LTE flame in an internal combustion engine was first discussed by Ward20-21. Several experimental studies have been performed of microwave-stimulated flames without formation of a microwave plasma.


4

Page 4 of 63

Page 5 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Hemawan et al.22 introduced an experimental arrangement where a microwave cavity was constructed to fully enclose a premixed laminar methane/air flame. The flow rate could be altered, giving a flame power in the range of 10-40 W. The microwave source was a continuous wave (2.45 GHz and 10-200 W) generator. They developed their arrangement into a cavity that only encloses the nozzle of the burner, with electric field enhancement at the nozzle edges23. Rao et al.24-26 used this configuration to study PAC with between 1 and 15 W of microwave radiation being absorbed in the flame. They identified three stages: (i) the electric field enhanced stage, (ii) the transition stage and (iii) the full plasma stage. Only in the first stage no microwave plasma is observed together with increased flame speed and extended flammability limits, however, without any significant change in OH number density or flame behavior. Sasaki et al.27-29 put a premixed laminar methane-air flame in a waveguide, subjected to 150 to 300 W of continuous microwave power at 2.45 GHz. The assembly was such that there were a standing wave in the waveguide and that the flame was located at a local maximum of the microwave radiation. When irradiating the flame with a microwave power of 300 W the length of the flame decreased at the same time as the burning rate or flame speed increased by a factor of 1.19. The emission spectra of OH and CH were not affected, which can be interpreted as virtually no increase in temperature. However, a significant increase in the optical emission intensity for N2 was found, which was interpreted as indications for electron heating in the flame. This experiment was carried out in the electric field enhanced stage and microwave-induced discharge was avoided. Based on the microwave radiation power, they estimated the reduced electric field to be approximately E/N≈10 Td1 in the center of the flame. Zaid et al.30-33 studied premixed laminar flames in a waveguide, subjected to up to 4.5 kW 1

The SI unit of the reduced electric field E/N is V·m2, but within the discharge physics community the unit 1 Td = 10−21V·m2 = 1 zVm2 has been introduced [Huxley, L.G.H.; Crompton, R. W.; Elford, M. T. Br. J. Appl. Phys. 1966, 17, 1237]. The Td (Townsend) unit was introduced before the introduction of the “zepto” prefix.


5

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

of continuous microwave power at 2.45 GHz. The burner nozzle was flush-mounted with the waveguide wall. The electric field strength at the flame was estimated to be 0.2 MV/m at 4.5 kW of microwave power. Three fuel-air mixtures were studied: methane-air at φ=0.70, propane-air at φ=0.60, and ethylene-air at φ=0.50, which resulted in 68%, 57% and 32% increase in laminar flame speed, respectively. In addition, they presented preliminary studies of a premixed turbulent methane-air flame with a Reynolds number of Re=3500, and φ=0.80, where the turbulent flame speed was estimated to increase by 11.4% at 1.5 kW microwave power. Stockman et al.34 used a similar set-up as Zaid et al.30-33 to study a premixed laminar methane-air wall stagnation flat flame with φ=0.60 to 0.80, subjected to a continuous wave microwave field of 0.5 MV/m. The flame front was observed to move towards the burner nozzle and stabilize at a stand-off distance corresponding to a flame speed increase of roughly 20%. No microwave discharge was observed. The post-flame temperature and OH concentration increased with microwave stimulation. For typical laminar flame conditions, the increase of 150 K in the reaction products corresponds to approximately 20 to 40 W of absorbed microwave power, corresponding to about 10% of the 300 W combustion power of these flames. Stockman et al.35 extended his studies to also contain the influence of continuous wave microwave radiation on a turbulent flame and pulsed microwave radiation on a laminar flame. In the study of the turbulent flame, a premixed methane-air flame, with Re≈3500 and φ=0.85, were used, where a 35% flame speed increase was reported when using 1.4 kW of microwave power. In the preliminary study of pulsed microwave radiation on a laminar flame, a premixed methaneair flame with φ=0.70 to 0.90 was used. Experiments with a 25 W average power, 0.1% duty cycle pulsed magnetron showed a laminar flame speed increase of over 30%. Miles et al.36 used similar arrangement as Stockman et al.35 to study the effect of pulsed


6

Page 6 of 63

Page 7 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

microwave radiation on outwardly propagating laminar methane-air flames. 30 kW peak power microwave pulses with a pulse length of 1 to 3 µs with a duty cycle of 0.001 s were used. They report that the effective flame speed is increased by 25% and that they manage to postpone lean limit extinction from φ=0.55 to 0.30. However, the observed temperature increase indicates that a microwave plasma might occur during each microwave pulse. Ehn et al.37 has recently reported on an experiment of a 30 kW turbulent premixed methane-air flame in a low-swirl burner subjected to continuous microwave radiation. The swirling turbulent flame was contained in a microwave cavity, similar to the one used in the experiments of Hemawan et al.22. The microwave stimulation increases the chemiluminescence as well as the post flame temperature. Planar laser-induced fluorescence data from OH and CH2O show that the flame is shifted closer to the burner nozzle when the flame absorbs microwave energy38. The temperature increase and the shift in flame position indicate an increase in flame speed for the turbulent flame as microwaves are applied to the flame. Development of numerical simulation models for studying electrical aspects of flames is a broad and active research area. Several numerical studies of the electric current passing through a flame caused by a weak electric field include chemical kinetic mechanisms, but without taking excitations by electron impact into account39-40. In studies of high-voltage nanosecond-pulses giving high E/N (E/N>>100 Td) non-LTE discharges41, excitation, dissociation and ionization by electron impact are dominating. Only a few attempts to numerically simulate microwavestimulated flames at intermediate E/N (below the critical field for sustained ionization) where molecules are excited by electron impact are published. Ju et al.,

42

, demonstrated an abridged numerical model for microwave-flame integration

which showed that the electric field of the microwave ration could increase the flame speed by


7

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 63

20% at field strengths just below the critical value to sustain a microwave plasma. The predicted flame speed enhancement qualitatively agreed with experimental data,

31

, but quantitatively un-

derpredicts the observed flame speed increase. Ju et al., 42, used the GRI1.2 mechanism completed with CH* and chemionization sub-mechanisms. Parsey et al.43 developed a kinetic global model framework, designed for extrapolation of the parameter space, to elucidate the key reaction pathways within non-equilibrium PAC, and their roles in the combustion process. Of key importance is the ability to determine possible system dependent reaction mechanism augmentation and specific reaction selectivity. In the model, kinetic plasma and gas-phase chemistry are coupled with a compressible gas flow model and solved with iterative feedback to match conditions from experiments. The model is applied to a microwave-assisted jet flame. It is observed that as the reaction mechanism become more complex, the limits of available data will begin to limit the model, requiring empirical treatment of gaps in experimental data to maintain completeness of the reaction mechanism. In summary; for microwave stimulation (below electronic breakdown) of laminar premixed methane-air flames the reported flame speed increase range from 19% to 68%, whereas for turbulent premixed methane-air flames the reported flame speed increase is in the range from 11% to 35%. In addition, the microwave radiation has been showed to lower the lean extinction limit, and therefore the technique of microwave-stimulation show promising results. Most investigations of microwave-stimulated combustion have been performed experimentally, and only a few simulation models have so far been developed. These models are rather simplistic, in particular from a fluid dynamics point of view. A recent study by the authors of the present work38 include both experiments and simulations of a turbulent low-swirl flame close to the lean blow-off limit, subjected to microwave


8

Page 9 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

stimulation. It was shown that microwave stimulation affects flame stabilization, increase the turbulent flame speed, and the kinetics analysis indicate that singlet oxygen molecules are the dominant flame enhancing species. A finite rate chemistry Large Eddy Simulation (LES) model44-46, was used in combination with a skeletal reaction mechanism for microwave-stimulated methaneair combustion. In the present work we extend this work by a thorough validation of the reaction mechanism and an in depth analysis of the effect of microwave stimulation on flames in different turbulence regimes. The skeletal reaction mechanism is developed for an electric field strength of E/N125 Td). This means that ionizing reactions due to electron impact are few and that the number density of ions and electrons is primarily determined by the rate of the chemionization reaction, CH+O⇒HCO++e (R65). Chemionization is of significant importance in regions where the highest concentration of CH is found. Hence, when selecting the composition for the BOLSIG+ calculations, the composition in the region where CH is most abundant is used. The composition in this layer was determined from laminar flame calculations using the GRI3.0 mechanism45. The species included in the calculation of reaction rates involving electron collisions are: CH4, O2, H2O, CO2, CO, H, O, H2, CH3, CH2, CH and N2. Except for N2 these species are all included in the kinetic mechanism used in the LES computations. The presence of N2 in an actual experiment will considerably modify the electron energy distribution, since N2 has an efficient transfer of electron energy to low-energy vibrational states. Hence, N2 needs to be included in the reaction rates calculation involving electrons. Figure 5a il-


23

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 63

lustrates the difference in mean electron energy as a function of E/N in the mixture listed above with and without N2. The significantly lower mean electron energy in the presence of N2 at atmospheric concentration will affect the rates of all processes involving electron collisions. The effect of N2 is shown in Figure 5b where the rate for electron impact ionization of O2 is plotted against E/N for a mixture with and without N2. The electron energy is lowered by the presence of N2 but is still around 1.0 eV (cf. Figure 5a), which is equivalent to an electron temperature of about 12 kK. The flame temperature at the location where the highest concentration of CH is found is calculated to be 1.4 kK. Accordingly, the electron temperature is one order of magnitude higher than the flame temperature, illustrating that the applied electric field creates a state of strong non-LTE in the flame. The rates of R56 and R69-R76, calculated using BOLSIG+, are displayed in Figures 6a and 6b. Note that the reaction rates for R69 to R71 are calculated for reactions creating O–, but that the reaction O–⇒O+e is assumed to occur instantaneously due to the high flame temperature and thus the product is considered to be O. The reaction rate of R56, O2+e→O2*+e, implemented in the mechanism is decreased compared to the value determined by BOLSIG+. In a more extensive mechanism, collisional quenching would deactivate a significant part of the excited oxygen, while in this simplified scheme tuning of O2* production is necessary. The reaction rate coefficients for R56 used in the Z80 mechanism are lowered by a value that approximately corresponds to deactivation of excited oxygen by the reaction O2*+M→O2+M. Among the E/N dependent reactions, a sensitivity analysis suggests that R56 producing O2* molecules and R74 producing ionized O2 molecules are of particular importance.


24

Page 25 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 6. (a) Rates for electron impact ionization as calculated in the mixture. (b) Rates for electron impact dissociation. Reaction numbers in both figures refer to Table 1.

2.6. Electron Attachment

The electron attachment and dissociation reaction subset consists of four reactions, R77 to R80 in Table 1.59 The attachment reactions close the paths of the positive ions created by the ionization and the rates does not experience an E/N-dependence. Since these reactions reduce the electron concentration they in effect also reduce the radical pool and thereby acts in the same way as chain terminating reactions. Sensitivity analysis shows that the most important of these reactions is R77 producing two oxygen atoms.

3. Microwave Assisted Laminar Flame Simulations

The first steps towards understanding of how microwaves affect a turbulent flame is to carry out laminar flame computations with the Z80 reaction mechanism listed in Table 1 for appropriate reduced microwave field strengths in the electric field enhanced stage (E/N1 suggests that the smallest eddies (if resolved) penetrate into the flame. The flame in Case 2 shows a muddled appearance with different flame topologies that are interacting and splitting-off pockets that burn vigorously. This results in a very thick flame brush (≈20δu), with an intrinsic flame structure. The results imply that chemical reaction rates play a less dominant role in determining the burning rates, but rather that the turbulent mixing rates control the rate of combustion. From these results it is also evident that there is only a very small difference in temperature, flame structure and flow between the cases characterized by E/N=0 Td and E/N=70 Td. Cases 1b and 2b both reveal an increased turbulent flame speed over Cases 1a and 2a, respectively, as the hot product mixture has advanced further into the cold reactant mixture. Cases 1c and 2c (E/N=80 Td) reveal a further increase in the turbulent flame speed, which, however, appears significantly smaller than the corresponding increase in laminar flame speed of almost 150% in Figure 8a.

(a)

(c)


39

(e)

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(b)

(d)

Page 40 of 63

(f)

Figure 11. Instantaneous LES predictions in terms of volumetric renderings of the temperature together with iso-surfaces of the second invariant of the velocity gradient tensor, λ2, for (a) Case 1a (ReI=129, Ka=1.92, Da=4.20, E/N=0), (b) Case 2a (ReI=967, Ka=86,5, Da=0.25, E/N=0), (c) Case 1b (ReI=129, Ka=0.90, Da=8.95, E/N=70) (d) Case 2b (ReI=967, Ka=40.5, Da=0.54, E/N= 70), (e) Case 1c (ReI=129, Ka=0.41, Da=19.4, E/N=80) and (f) Case 2c (ReI=967, Ka=18.6, Da= 1.20, E/N=80).

Based on the CH front propagation we can approximate the turbulent flame speed, sT, for the different cases: For Cases 1a and 2a (with E/N=0 and su=0.13 m/s) the turbulent flame speeds are sT≈0.68 and 1.71 m/s, respectively, resulting in that sT/su≈5 and 13, respectively. These flame speed ratios are in general agreement with turbulent flame speed data84. For E/N=70, the CH front propagation results in that the turbulent flame speeds are sT≈0.75 and 1.82 m/s for Cases 1b and 2b, respectively, so that with su≈0.19 m/s as a results of the microwave irradiation, sT/su≈3.9 and 9.6, respectively. For E/N=80, the CH front propagation results in that the turbulent flame speeds are sT≈0.95 and 2.06 m/s for Cases 1b and 2b, respectively, so that with su≈0.28 m/s as a results of the microwave irradiation, sT/su≈3.4 and 7.4, respectively. The increase in sT due to microwave irradiation is therefore 10% and 40% for Case 1 at E/N=70 and 80, respectively, and 6% and 20% for Case 2 at E/N=70 and 80, respectively, whereas the corresponding increase in su for Cases 1 and 2 is 58% and 116%, respectively. These results suggest that the influence of micro-


40

Page 41 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

wave irradiation is larger for laminar flames than for turbulent flames, and that the relative influence of microwave irradiation decreases with increasing turbulence intensity. The reason for the latter is that the higher the turbulence intensity the smaller is the influence of the chemical kinetics in determining the overall turbulent reaction rate. However, the radical pool persists also for high turbulence levels, and hence, precludes quenching. The influence of microwave irradiation can also be illustrated by comparing the volumetrically averaged composition of species, as illustrated in Figure 12. Compared to the laminar case (Re=0) shown in Figure 12a, the two turbulent cases (Re=129 and 967), shown in Figure 12b, show distinct turbulence mixing effects with more CO2, O and OH, and less CO, CH3, H, CH2O and HCO. While in the laminar flame case the composition is fairly similar for E/N= 70 and 80 Td the turbulence result in a larger effect of E/N, and more variation with E/N. It is also interesting to note that the amount of O2* and O3 is less influenced by the turbulence at E/N=80 Td compared to E/N=70 Td, for the lower E/N the more turbulent case result in comparably low concentrations of the microwave induced species. In general it is found that there are less intermediates and radicals in the turbulent flames than in the laminar flames, but that the intermediates and radicals typically increase in concentration with increasing E/N. These observations together suggest that combustion on average is more complete in the two turbulent cases although the increase in st is less than the increase in su. The primary reason for that is the decrease in τign, which in the case of turbulent flames competes with the turbulent mixing time scale.


41

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(a)

Page 42 of 63

(b)

Figure 12. Volumetrically averaged species mass fractions at different reduced electrical fields for (a) laminar flames, and (b) turbulent flames after three eddy turn-over times at Re=129 and Re=967.

4.3. Experiments and LES of a Turbulent Low-Swirl Flame

The next level in complexity is to consider a turbulent low-swirl flame85 offering a reasonable compromise between simplicity and flow complexity. The low-swirl flame was exposed to microwave radiation inside an aluminum cavity with optical access. For this case, both experiments37, 38 and LES computations38 have previously been carried out, but here a more comprehensive joint experimental and computation analysis is provided to get an estimate of the turbulent flame speed enhancement.

4.3.1. Experimental and Computational Configuration


42

Page 43 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

An overview of the experimental setup is seen in Figure 13a including burner, microwave generator and surrounding equipment as well as lasers and optics. The low-swirl burner used in this study, Figure 13b, is an extension of the original design proposed by Bedat & Cheng86 for studying premixed flames propagating in turbulence. Premixed fuel and air is fed to a settling chamber, containing perforated grids to assure a uniform mixture, below the burner (not shown). A pipe transports the mixture through the co-flow arrangement surrounding the burner. Ahead of the swirler two perforated plates are located to disintegrate any large turbulent structures formed in the pipe. The swirler consists of an outer part, with eight swirl-vanes, and an inner part, which consists of a perforated plate that allows for a certain portion of the flow to pass through the swirler section without gaining swirl. The resulting reactant flow is divergent, thus creating an inner low-velocity region in which the flame is stabilized. In the experiments the co-flow and combustible air-flow were controlled by mass-flow controllers from Thermal Instruments and Bronckhorst, respectively. The flame was operated with a Re-number of 20,000 and at two different equivalence ratios, one close to lean blow-off conditions with φ=0.58 and a more stable case with φ=0.62, having theoretical powers of ~24 kW and ~27 kW, respectively. The low-swirl burner is covered by a cylindrical aluminum cavity with a diameter of D=300 mm, with one mesh located at each end to keep a confined volume for the microwaves, Figure 13c. Amplitudes of the incident and reflected microwave radiation were measured to estimate the microwave power that is absorbed by the flame. The position of the mesh located at the top of the circular aluminum cavity is adjustable to allow optimization of the microwave mode pattern, with relation to the flame position. Planar laser-induced fluorescence (PLIF) of OH was performed with and without microwave stimulation of the flame. Hydroxyl excitation was performed using a frequency doubled


43

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 44 of 63

Nd:YAG/dye system (Quantel) using a CCD camera with external intensifier (Hamamatsu) to capture the OH PLIF signal. The camera was equipped with a UG 11 filter and a f=100 mm fused silica lens (Carl Zeiss F/4.0). Two-dimensional Particle Imaging Velocimitry (PIV) was used to characterize the flame mapping up averaged velocity components in the center plane of the burner nozzle. The PIV measurements were performed using a double cavity diode pumped kHz Nd:YLF laser (DualPower 1000-30, Dantec) along with a high-speed camera (SpeedSense 311, 1MP, 3250 fps). A narrowband (+/-10 nm) filter was put in front of the camera to discriminate flame chemiluminescence. Velocity fields were extracted from the PIV data using the DynamicStudio software (v. 4.15). The computational set-up consists of a coarse 3.6 million cell hexahedral cell mesh, displayed in Figure 13d. At the walls of the swirler, adiabatic no-slip boundary conditions are used, whereas at the outlet, located 0.5 m downstream of the nozzle, wave-transmissive conditions87 are applied. The swirler inflow is divided into an outer part and an inner part, for which fixed mass flows are specified, whereas as outside of the swirler the co-flowing air is modeled as laminar. To examine the sensitivity to the spatial resolution, a topologically identical grid with about 7.0 million cells have been tested. The microwave modal pattern was computed using COMSOL Multiphysics modeling software88 using the same dimensions of the aluminum cavity as in the experiments. This results in a mode just above the burner nozzle, encapsulating the base and core of the flame. The LES computations are set-up so that the microwave enhancing reactions (reactions R56 and R69-76) are only active within the modal pattern, whereas the reactions controlling the CH4-air combustion are active also outside of the pattern.


44

Page 45 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 13. (a) Schematic overview of the experimental setup. Semitransparent images of (b) the low-swirl burner nozzle and (c) the microwave cavity with the low-swirl burner nozzle and the co-flow. The computational grid is presented in (d).

An analysis approach enabling extraction of a reaction progress variable, c, for the location of the flame front, based on experimental PLIF data was implemented. About 500 OH-PLIF single-shot images were analyzed in Matlab in each sequence to estimate the position of the flame front with and without microwave stimulation. The average flame front position was estimated by forming c from the OH-PLIF data where the c=0.5 contour was determined. This is the position where OH signal is detected in 50% of the evaluated OH PLIF images. The processing scheme for c (for experimental data) was evaluated by performing edge detection of the progress variable data to construct a synthetic CH2O signal. This analysis and evaluation method is described in detail in Appendix 1. From the LES computations the progress variable, c, was computed based on the temperature, following conventional practice62 so that sulting in that c=1 in the hot burnt gases and c=0 in the cold fresh reactants.

4.3.2. Experimental Results and LES Predictions


45

, re-

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 46 of 63

Next, the results from experiments and LES predictions are presented and discussed. Experimental details for similar conditions have been presented in a previous publication by Ehn et al.37, and experimental data and LES were further discussed in another publication by the same group38. The authors presented averaged OH and CH2O PLIF data in this publication showing a distinct increase in OH-PLIF signal with microwave stimulation as well as a broadening of the flame. Further, the averaged PLIF images indicated that the flame was displaced with microwave stimulation in that it moved closer to the burner nozzle. The experimental and LES results presented below are further analyzed to estimate the relative change in axial burning velocity around the central axis of the burner geometry.

Figure 14. (a) Experimental images of the chemiluminescence with instantaneous microwave absorption showing an increase in flame stimulation. LES predictions of the heat release, Q, at (b) E/N=0 Td, (c) E/N=60 Td, (d) E/N=70 Td and (e) E/N=80 Td.


46

Page 47 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 14a show a series of experimental high-speed chemiluminescence video images of a flame subject to instantaneous microwave absorption during 3 ms. The intensity of the flame chemiluminescence apparently increases with microwave absorption. This absorption could be somewhat irregular, as discussed in our previous work37. This is apparent in Figure 14a where the flame becomes more luminous and expands in volume as it starts absorbing microwave energy. The strength of the reduced electric field in the experiments could not be determined, but further understanding of the effect of increasing field strength was obtained from LES. Microwave stimulation is implemented in the LES model by additional E/N-dependent reactions, as described in Section 2. The effect of increasing microwave stimulation can thus be systematically investigated. Figures 14b to 14e show volumetric renderings of the heat release for different reduced electric field strengths. The absolute heat release at the flame front is not significantly increased with microwave stimulation. Thus, the increase in heat release that follows with increasing reduced electric field strength is due to a larger flame volume, which is seen in Figures 14b to 14e. Combustion LES data of the overall effect of microwave stimulation on the flame is seen in Figures 15. The figure display instantaneous volumetric renderings of the temperature, T, and CH4 mass fraction, YCH4, together with contours of the axial velocity, vx, for φ=0.60, and E/N=0, 60 Td, 70 Td, and 80 Td respectively. All cases computed reveal a fully detached flame, lifted about 0.6·D above the burner rim, with the highest temperature occurring at the richest leading edge of the flame. Downstream of the flame front, approximately between x/D≈1 and 5, the LES predictions indicate the presence of a weak, oblong, recirculation zone. In this region, the fluid mixes with the surrounding air resulting in a leaner mixture and thus a lower temperature in the downstream part of the flame. The flame is not stabilized by the recirculation zone,


47

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 48 of 63

but by the inner shear layer which originates at the exit of the swirler, cf. Figure 13b, where the flow passing through the swirler mixes with the flow passing through the perforated plate in the center of the burner. Due to Kelvin-Helmholtz structures and the high Re number, this develop into turbulence, which create vorticity and strong radial velocities that further downstream interacts with the unsteady flame front. Further out in the swirling jet, the flame is pushed downstream. Certain amount of fuel mixes with surrounding air prior to the reaction zone, and is diluted beyond the flammability limit. In agreement with experimental results the LES show a clear widening of the flame with increasing microwave stimulation, and also a flame stabilizing closer to the burner nozzle. This indicates that the LES model in combination with the skeletal reaction mechanism is capable of capturing the dominating mechanism by which the electrical field strength influences the combustion chemistry and the turbulent flow.

Figure 15. Instantaneous volumetric renderings of the temperature, T, and CH4 mass-fractions together with contours of the axial velocity, vx, for (a) E/N=0, (b) E/N=60 Td, (c) E/N=70 Td and (d) E/N=80 Td. The color shadings for T range from opaque white, through yellow, orange and red to semi-transparent black using a linear mapping from 5% to 95% of the peak value. Similarly, the color shadings for YCH4 range from opaque dark green to semi-transparent white using a linear mapping from 5% to 95% of the peak value.


48

Page 49 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 16 show contour plots of the mean or time-averaged progress variable, c, without and with microwave stimulation for experiments at φ=0.58, 17a, and LES at φ=0.60, 16b-e. From experiments as well as from LES predictions it is evident that the flame is located closer to the burner rim with microwave stimulation compared to without. The flame front, which is defined as c=0.5, is in the experimental analysis located at 0.94D and 0.88D above the burner rim for the flame at φ=0.58 without and with microwave stimulation, respectively. This is significantly further away from the burner rim compared to previous experiments on the low swirl burner without microwave stimulation, where the flame location was about 0.6D above the burner rim89, as accurately predicted by the LES. This difference in flame position is caused by the grid used to contain the microwaves in the experimental setup. The grid cannot easily be incorporated into the LES predictions and thus the absolute position of the experimental and LES flames cannot be compared, but the comparison is based on relative difference in flame location as a result of microwave stimulation. In the LES predictions the flame front is located at 0.60D for E/N=0, 0.59D for E/N=60 Td, 0.57D for E/N=70 Td and 0.55D for E/N=80 Td, thus showing a gradual increase in flame stimulation with increasing E/N. The relative difference in flame location was estimated from the experimental data to be about 5 to 7% for φ=0.58 and 2% for φ=0.62. The simulations showed relative difference in flame location of 2, 5 and 9%, respectively, at φ=0.60, for the E/N= 60 Td, E/N=70 Td and E/N=80 Td cases. See Table 3 for a summary of experimental and computational results for the low-swirl burner.


49

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 50 of 63

Figure 16. Contour plot of the mean or time-averaged progress variable, from (a) experiments (φ=0.58) without (left) and with (right) microwave stimulation and LES at φ =0.60 for (b) E/N=0 Td, (c) E/N=60 Td, (d) E/N=70 Td and (e) E/N=80 Td. The flame front marker is the contour at c=0.5.

The validity of the mean progress variable processing and analysis was evaluated by comparing the synthetic CH2O signal (for further detail, see Appendix A) with averaged CH2O PLIF data. This analysis was performed on the data from the leaner flame (φ=0.58) since the flame displacement is more significant here. The experimental data is shown in Figure 17a and corresponding cross-sectional profiles along the centerline, both without and with microwave stimulation are shown in Figure 17b. Similarly, averaged synthetic data are shown in Figure 17d with cross-sectional profiles displayed in Figure 17c. The synthetic CH2O data merely display signal at the outer part of the flame and does not reproduce the actual CH2O signal that is produced in the post flame. Hence, the signal peak 70 mm downstream of the burner nozzle is not seen in the synthetic signal, which only should include the actual flame front position. The CH2O and OH measurement data are not synchronized and therefore the cross-sectional profiles does not fully correspond to each other but the averaged position and displacement due to microwave stimulation seem to be accurately described by the synthetic data. This good agreement between exper-


50

Page 51 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

imental and synthetic data indicates that the mean progress variable processing and analysis is reliable for the experimental data.

Figure 17. Validation of the progress variable analysis of experimental OH data. (a) Averaged PLIF-imaging data of CH2O from experiments without (left) and with (right) microwave stimulation with φ=0.58. (b) Cross-sectional profiles of the experimental CH2O PLIF data, shown in (a). (c) and (d) show corresponding data to (a) and (b) but here synthetic CH2O data is displayed.

Table 3. Summary of global flame quantities for the low-swirl flame, obtained from analysis of experimental and LES data. The turbulent velocity, sT, is here represented by the axial velocity component.

φ

Case 1 2 3 4 5 6 7

Exp Exp Exp LES LES LES LES

0.61 0.58 0.62 0.60 0.60 0.60 0.60

E/N location x/D relative location to Cases su sT sT/sT|E/N=0 (Td) 1 and 4 respectively (%) (m/s) (m/s) 0 0.30 1.00 n/a 5.0 to 7.0 0.45 1.50 n/a 2 0.32 1.07 0 0.602 0 0.13 0.45 1.00 60 0.589 2.2 0.14 0.65 1.44 70 0.571 5.2 0.19 0.70 1.56 80 0.545 9.5 0.28 1.02 2.27


51

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 52 of 63

The axial velocity along the centerline has been determined from measurements both without and with microwave irradiation, and is also readily available from the LES computations, Table 3. Experimental and computational velocity profiles show distinct plateaus where the premixed flame is located. This is caused by the balance between (downward) flame propagation and (upward) convective transport of the reacting CH4-air mixture such that v x e x=s T e x , where ex is the unit vector in the x-direction and sT the turbulent flame speed. Comparing the experimental axial velocities with and without microwave irradiation suggests a ~50% increase in sT in the axial direction for φ=0.58 and a ~7% increase in sT for φ=0.62, cf Table 3. The corresponding results from the LES predictions are 44% for E/N=60 Td, 56% for E/N=70 Td and 127% for E/N=80 Td. For comparison with the cases of the planar flames in homogeneous isotropic turbulence discussed in Section 4.2 we note here that v′/su ≈4 and ℓ/δ u ≈40 such that Da≈10 and Ka≈1.2, placing this flame approximately in the upper part of the corrugated flamelet regime. This low-swirl burner flame is thus comparable to Case 1 in Table 2. The microwave irradiation increase su so that the irradiated flames shifts towards lower v′/su and higher lI/δu in the Borghidiagram, making them more flamelet like and gradually more amenable to microwave irradiation. The large difference in sT between the experimental flames with φ=0.58 and 0.62 could be explained by different reduced electrical fields at the flame front. The electric field strength that is emitted by the microwave generator in the very lean flame (φ =0.58) is about 40% higher than that for the lean flame (φ =0.62). Both flames were exposed to E/N just below electric breakdown but it is unfortunately not possible to determine the value of E/N.


52

Page 53 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

The flow field of the flame with and without microwave stimulation is expected to be somewhat different. However, the difference is small since the change in axial velocity is smooth along the center axis in the location of the determined flame fronts. The data acquired for the very lean case (φ =0.58) should be treated with some care since it is a rather small flame for which the flame position to some extent is affected by competing effects. However, to the best of the author’s knowledge, the current data is the best experimental estimation of turbulent flame speed enhancement due to microwave stimulation at this point. To further elucidate the chemistry responsible for flame enhancement by microwave stimulation the composition of the LES flames are investigated. Figure 18 present volumetric renderings of several key species in simulated flames in a reduced field of E/N=70 Td. CH is found in a thin wrinkled layer in the flame front, 18a, together with OH and O, 18b and 18c. In the LES the chemical reactions of species produced as a result of microwave enhancement are active throughout the flame and therefore e, O2*, and related species like O3 are present in the whole flame volume, as represented by O2* in Figure 18d. As found in the present and previous studies, CH2O production is enhanced by the microwave irradiation, in Figure 18e it is apparent that CH2O is present in large part of the flame volume where O2* is also present. Concentration of O2* is high close to the burner exit, below what can be defined as the flame front. At this early stage the unburnt fuel and oxygen in the premixed gas flow will undergo electron impact ionization reactions (R74, R75), multiplying the number of electrons and producing ions. The electrons will then react with O2 (R69), CH4 (R73), O2+ (R77) and CH4+ (R79) to create an early radical pool. In addition O2* will be produced via R56, which further increase the radical pool. The radicals will increase the reactivity of the unburnt gas mixture and make it ignite earlier and stabilize the flame closer to the burner.


53

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 54 of 63

(a) (b) (c) (d) (e) Figure 18. Volumetric renderings of (a) CH, (b) OH, (c) O, (d) O2* and (e) CH2O for E/N=70 Td. Figure 19 present reaction path diagrams including the 12 most important species for laminar flames at different stages in combustion corresponding to different unburnt fuel fraction, and thus representative of the chemistry at different positions in the turbulent flames. The sequence of fuel breakdown from CH4 to CO2 via intermediate hydrocarbons are the same for all three cases, and so is the presence of O and OH. Figure 19a representing an early stage in the flame have CH4+ and e among the 12 most important species, which signify importance of the electron impact ionization reaction subset mentioned in the discussion of Figure 18, and in particular R75 (CH4+e=>CH4++e+e) multiplying the electrons. At the later combustion stages represented by 19b and 19c not all CO production goes via the CH2O, but via CH2 and CH, that appear among the important species, these contribute directly to CO production. In the case of 50% unburnt fuel fraction also production of electrons from CH play a significant role, while for the case where most of the fuel has burnt (19c) the electrons are no longer among the 12 most important species. At the latest stage O3 again appear important, but it is less strongly connected to CH3O compared to the first case, which can be expected since at this late stage the hydrocarbon chemistry is less pronounced.


54

Page 55 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Figure 19. Reaction path diagrams with 12 important species at φ=0.6 for E/N=70 Td, (a) corresponding to a point in the flame where about (a) 10%, (b) 50% and (c) 90% of the fuel is consumed.

5. Concluding Remarks

The current work involves unique experimental data concerning flame enhancement as a result of microwave irradiation, in combination with an extensive analysis of the chemistry using a new kinetic mechanism suitable for LES of microwave stimulated combustion. Few experimental studies on microwave enhanced combustion exist, and the ones that have been done suffer from that quantification of many parameters, like the reduced electrical field strength E/N, have not been experimentally determined. This mean that all simulations presented in this work have not been experimentally confirmed, but in development of the kinetic mechanism this was partly compensated for by validation against experimental and previous modeling targets for various conditions of selected chemistry sets. The skeletal kinetic mechanism here was thus thoroughly validated, which is presented in detail, and used for simulations of microwave enhanced flames


55

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 56 of 63

at different levels of complexity; laminar premixed flames, premixed flames in homogeneous isotropic turbulence and a turbulent low-swirl flame. The low-swirl case was experimentally and computationally investigated. The overall goal of the study was to investigate the relative increase in flame-speed enhancement and attempt to understand the underlying chemistry, and how the chemical effects interact with turbulence for microwave stimulated flames. We do not attempt to make absolute quantification of the effect of microwave stimulation on the various types of flames. From laminar flame simulations it is clear that increasing the electric field strength (below the dielectric breakdown) result in increasing reactivity and thus faster flame propagation, reduced ignition times, and increased resistance to extinction. As homogeneous isotropic turbulence of different strength was introduced it became clear that the increased turbulence partly overrides the increased chemical reactivity and the enhancement flame reactivity is less pronounced compared to the laminar flame case. Thorough analysis of experimental data and LES for the microwave enhanced low-swirl flame confirmed previous finding37, 38 that the flame is broadened and move closer to the burner as a result of microwave stimulation. In addition the new analysis resulted in that the axial velocity component was found to increase with about 50% in experiments on a flame near the lean blow-off limit, φ=0.58, but only about 5% for a slightly richer flame. These large differences are likely a result of that the two flames were exposed to different electric field strengths, which is at this stage not possible to quantify or control experimentally. Corresponding analysis of the LES flames subject to different reduced electrical fields showed a non-linear increase of axial velocity with increasing E/N, which is in agreement with simulation of laminar flames. The dependence on E/N investigated computationally is such an important factor in further understanding of mi-


56

Page 57 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

crowave stimulated combustion, that future experimental studies should be designed with the aim at controlling this fundamental property. Analysis of chemical reaction paths reveal that reactions involving oxygen are dominating the microwave enhanced chemistry, with singlet oxygen, O2*, as a key player resulting in increased concentrations of, among other species, the important radical OH. At an early stage where the fuel, CH4, is abundant the corresponding ion, CH4+, is important for multiplication of the number of electrons driving the microwave enhanced chemistry. These results illustrate how microwaves can be utilized to stimulate turbulent combustion which hints to possibilities for applications in combustion enhancement and control. Previous investigations have shown increasing absorption of microwave radiation by the flame when the microwaves are pulsed90 which seem to be the obvious way of perusing this line of research and development. The ability to increase absorption would be of interest from an energy conversion perspective for steady stimulation of running combustion at leaner conditions whereas lower absorption would still be beneficial for counteracting thermo-acoustic oscillations.

Acknowledgement

This work has been financed by the Swedish Energy Agency and the advanced ERC grant (TUCLA). Jon Tegnér at FOI is acknowledged for assistance with evaluating the reaction mechanisms.

References

(1) Market Observatory for Energy - European Commission, “Key Figures”, p. 19, 2011, (http://ec.europa.eu/ energy/observatory/countries/doc/key_figures.pdf) (2)

Starikovskaia, S. M.; J. Phys. D: Appl. Phys. 2006, 39, R265-R299.


57

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(3)

Starikovskiy, A.; Aleksandrov, N. Progress Energy Comb. Sci. 2013, 39, 61-110.

(4)

Ju, Y.; Sun, W. Prog. Energy Combust. Sci. 2015, 48, 21-83.

(5)

Ju, Y.; Sun, W. Combust. Flame 2015, 162, 529–532.

Page 58 of 63

(6) Ehn, A.; Zhu, J.; Petersson, P.; Li, Z.; Aldén, M.; Fureby, C.; Hurtig, T.; Zettervall, N.; Larsson, A.; Larfeldt, J. Proc. Combust. Inst. 2015, 35, 3487–3495. (7)

MacLatchy, C.S. Combust. Flame 1979, 36, 171-178.

(8) Kim, Y.; Abbate, S.; Ziock, H.; Anderson, G.K.; L.A. IEEE Trans. Plasma Sci. 2007, 35, 1677-1681. (9) Shibkov, V.M.; Shibkova, L.V.; Gromov, V.G.; Karachev, A.A.; Konstantinovskii, R.S. High Temp. 2011, 49, 155-167. (10) Babaritskii, A.I.; Baranov, I.E.; Bibikov, M.B.; Demkin, S.A.; Zhivotov, V.K.; Konovalov, G. M.; Lysov, G,.V.; Moskovskii, A.S.; Rusanov, V.D.; Smirnov, R.V.; Chebankov, F.N. High Energ. Chem. 2004, 38, 407-410. (11) Davydov, A.M.; Gritsinin, S.I.; Kossyi, I.A.; Shikhman, Y.M.; Vinogradov, V.A. IEEE Trans. Plasma Sci. 2008, 36, 2909-2917. (12) Esakov, I.I.; Grachev, L.; Khodataev, K.; Van Wie, D. AIAA-2005-0989. (13) Esakov, I.I.; Grachev, L.P.; Khodataev, K.V.; Vinogradov, V.A.; VanWie, D. IEEE Trans. Plasma Sci. 2006, 34, 2497-2506. (14) Bang, C.U.; Hong, Y.C.; Cho, S.C.; Uhm, H.S.; Yi, W.J. IEEE Trans. Plasma Sci. 2006, 34, 1751-1756. (15) Hong, Y.C.; Uhm, H.S. Phys. Plasmas 2006, 13, 113501-1--8. (16) Hammack, S.; Tonghun, L.; Carter, C. IEEE Trans. Plasma Sci. 2012, 40, 3139-3146. (17) Hammack, S.; Xing, R.; Tonghun, L.; Carter, C. IEEE Trans. Plasma Sci. 2011, 39, 33003306. (18) Wang, C.J.; Wu, W. Combust. Flame 2014, 161, 2073-2084. (19) Wu, W.; Fuh, C.A.; Wang, C. Combust. Sci. Technol. 2015, 187, 999-1020. (20) Ward, M.A.W. Combustion in an Internal Combustion Engine, 1976, USA Patent 3934566. (21) Ward, M.A.W. J. Microwave Power 1977, 12, 187-198. (22) Hemawan, K.W.; Romel, C.L.; Zuo, S.; Wichman, I.S.; Grotjohn, T.A.; Asmussen, J. Appl. Phys. Lett. 2006, 89, 141501-1--3. (23) Hemawan, K. W.; Wichman, I. S.; Lee, T.; Grotjohn, T. A.; Asmussen, J. Rev. Sci. Instrum. 2009, 80, 053507-1--9. (24) Rao, X.; Hemawan, K.; Wichman, I.; Carter, C.; Grotjohn, T.; Asmussen, J.; Lee, T. Proc.


58

Page 59 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

Combust. Inst. 2011, 33, 3233-3240. (25) Rao, X.; Hammack, S.; Carter, C.; Grotjohn, T.; Asmussen, J.; Lee, T. IEEE Trans. Plasma Sci. 2011, 39, 3307-3313. (26) Rao, X.; Hammack, S.; Carter, C.; Lee, T. IEEE Trans. Plasma Sci. 2011, 39, 2354-2355. (27) Shinohara, K.; Takada, N.; Sasaki, K. J. Phys. D: Appl. Phys. 2009, 42, 18200-1--4. (28) Sasaki, K.; Shinohara, K.; Takada, N. AIAA-2011-1023. (29) Sasaki, K.; Shinohara, K. J. Phys. D: Appl. Phys. 2012, 45, 455202-1--7. (30) Zaidi, S.H.; Macheret, S.O.; Ju, Y.; Miles, R.B.; Sullivan, D.J. AIAA-2004-2721. (31) Sullivan, D.J.; Zaidi, S.H.; Macheret, S.O.; Ju, Y.; Miles, R.B. AIAA-2004-3713. (32) Zaidi, S.H.; Macheret, S.O.; Ju, Y.; Miles, R.B.; Sullivan, D.J.; Efthimion, P.C. AIAA2005-0992. (33) Zaidi, S.H.; Stockman, E.; Qin, X.; Zhao, Z.; Macheret, S.; Ju, Y.; Miles, R.B.; Sullivan, D.J.; Kline, J.F. AIAA-2006-1217. (34) Stockman, E.S.; Zaidi, S.H.; Miles, R.B.; Carter, C.D.; Ryan, M.D. Combust. Flame 2009, 156, 1453-1461. (35) Stockman, E.; Zaidi S.; Miles, R.B. AIAA-2007-1348. (36) Michael, J.B.; Cheng, T.L.; Miles, R.B. Combust. Flame 2013, 160, 796-807. (37) Ehn, A.; Hurtig, T.; Petersson, P.; Zhu, J.; Larsson, A.; Fureby, C.; Larfeld, J.; Li, Z.-S.; Aldén, M. J. Phys. D: Appl. Phys. 2015, 49, 185601. (38) Ehn, A.; Petersson, P.; Zhu, J.; Li, Z.-S.; Aldén, M.; Nilsson, E.J.K.; Larfeldt, J.; Larsson, A.; Hurtig, T.; Zettervall, N.; Fureby, C. 2016, Accepted for presentation at the 36th Int. Symp. On Comb. (39) Pedersen, T.; Brown, R.C. Combust. Flame 1993, 94, 433-448. (40) Speelman, N.; de Goey, L.P.H.; van Oijen, J.A. Combust. Theor. Model. 2015, 19, 159187. (41) Starikovskaia, S.M. J. Phys. D: Appl. Phys. 2014, 47(35), 353001. (42) Ju, Y.; Macheret, S.O.; Shneider, M.N.; Miles, R.B.; Sullivan, D.J. AIAA-2004-3707. (43) Parsey, G.; Guclu, Y.; Verboncoeur, J.; Christlieb, A. IEEE International Conference on Plasma Science (ICOPS), San Francisco, USA, 2013. (44) Bulat, G.; Fedina, E.; Fureby, C.; Meier, W.; Stopper, U. Proc. Combust. Inst. 2015, 35, 3175. (45) Frenklach, F.; Wang, H.; Yu, C.L.; Goldenberg, M.; Bowman, C.T.; Hanson, R.K.; Davidson, D.F.; Chang, E.J.; Smith, G.P.; Golden, D.M.; Gardiner, W.C.; Lissianski, V. GRI-Mech 3.0; http://www.me.berkeley.edu/gri_mech.


59

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 60 of 63

(46) CHEMKIN 10112, Reaction Design: San Diego, 2011. (47) Smooke, M.D.; Giovangigli, V. In Lecture Notes in Physics: Reduced Kinetic Mechanisms and Asymptotic Approximations for Methane-Air Flames, Smooke M.D. Ed., Springer-Verlag: New York, 1991, 384. (48) Sher, E.; Refael, S. Combust. Sci. Technol. 1988, 59, 371-389. (49) Goswami, M.; Derks, S.; Coumans, K.; Slikker, W.J.; de Andrade Oliveira, M.H.; Bastiaans, R.J.M.; Luijten, C.C.M.; de Goey, L.P.H.; Konnov, A.A. Combust. Flame 2013, 160, 1627-1635. (50) Wang, Z.H.; Yang, L.; Li, B.; Li, Z.S.; Sun, Z.W.; Aldén, M.; Cen, K.F.; Konnov, A.A. Combust. Flame 2012, 159, 120. (51) Starik, A.M.; Kozlov, V.E.; Titova, N.S. Combust. Flame 2010, 157, 313. (52) Starik, A.M.; Kozlov, V. E.; Titova, N.S. J. Phys. D: Appl. Phys. 2008, 41, 125206. (53) Sharipov, A.; Starik, A. J. Phys. Chem. A 2011, 115, 1796-1803. (54) Pedersen, T.; Brown, R. C. Combust. Flame 1993, 94, 433-448. (55) Hagelaar, G.J.M.; Pitchford, L.C. Plasma Sources Sci. Techn. 2005, 14, 722-733. (56) Morgan, W.L. Plasma Chemistry and Plasma Processing 1992, 12, 477. (57) Phelps, A. V. (private communication) available at http://jilawww.colorado.edu/~avp/ (58) UNAM database, http://www.lxcat..net. (59) Aleksandrov, N.L.; Kindysheva, S.V.; Kukaev, E.N.; Starikovskaya, S.M.; Starikovskii, A. Yu. Plasma Physics Reports 2009, 35, 867–882. (60) Menon, S.; Fureby, C. In Encyclopedia of Aerospace Engineering; Eds. Blockley R.; Shyy W., John Wiley & Sons, 2010. (61) Hawkes, E.R.; Chatakonda, O.; Kolla, H.; Kerstein, A.R.; Chen, J.H. Combust. Flame 2012, 159, 2690-2703. (62) Poinsot, T.; Veynante, D. Theoretical and Numerical Combustion, Edwards: Philadelphia, USA, 2001. (63) Pitsch, H. Annu. Rev Fluid Mech. 2006, 38, 453. (64) Yoshizawa, A.; Horiuti, K. J. Phys. Soc. Japan 1985, 54, 2834-2839. (65) Fureby, C. Ercoftac Bulletin, March issue 2007. (66) Sabelnikov, V.; Fureby, C. Combust. Flame 2013, 160, 83-96. (67) Zettervall N.; Fedina E.; Nordin Bates K.; Heimdal Nilsson E.; Fureby C. AIAA-20154020. (68) Zettervall, N.; Nordin-Bates K.; Nilsson E. J. K.; Fureby C. Combust. Flame. 2017


60

Page 61 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

(69) Nordin-Bates K.; Fureby C.; Karl S.; Hannemann K. Proc. Combust. Inst. 2016 http://dx.doi.org/10.1016/j.proci.2016.07.118. (70) Fedina E.; Gottiparthi C.; Fureby C.; Menon S. Combustion in Afterburning Behind Explosive Blasts”, In Coarse Grained Turbulent Mixing, Ed. Grinstein F., Cambridge University Press 2015. (71) Pope, S.B. Prog. Energy Comb. Sci. 1985, 11, 119. (72) Yeung, P.K.; Pope, S.B.; Sawford, B.L. J. Turb. 2006, 7, 1. (73) Weller, H.G.; Tabor, G.; Jasak, H.; Fureby, C. Comp. in Physics 1997, 12, 629. (74) Drikakis, D.; Fureby, C.; Grinstein, F.F.; Liefendahl, M. In Implicit Large Eddy Simulation: Computing Turbulent Fluid Dynamics, Eds. Grinstein F.F., Margolin L.; Rider B., Cambridge University Press, 2007, 94. (75) Issa, R.I. J. Comput. Phys. 1986, 62, 40. (76) Strang, G. SIAM J. Numerical Analysis 1968, 5, 506. (77) Rosenbrock, H.H. The Computer Journal 1963, 5, 329. (78) Poinsot, T.J.; Lele, S.K. J. Comput. Phys. 1992, 101, 104. (79) Borghi, R. Recent Advances in Aeronautical Science 1985, 23, 117. (80) Eswaran, V.; Pope, S.B. Comp. Fluids. 1988, 16, 257-278. (81) Fureby, C.; Tabor, G.; Weller, H.; Gosman, D. Phys. Fluids 1997, 9, 1416. (82) Tennekes, H.; Lumley, J.L. A First Course in Turbulence MIT Press, 1972. (88) Han, I.; Huh, K.Y. Combust. Flame 2008, 152, 194. (84) Peters N. Turbulent Combustion, Cambridge University Press, 2000. (85) Petersson, P.; Olofsson, J.; Brackman, C.; Seyfried, H.; Zetterberg, J.; Richter, M.; Aldén, M.; Linne, M.A.; Cheng, R.K.; Nauert, A.; Geyer, D.; Dreizler A. Appl. Optics 2007, 46, 3928. (86) Bédat, B.; Cheng, R.K. Combust. Flame 1995, 100, 485. (87) Poinsot, T.J.; Lele, S.K. J. Comput. Phys. 1992, 101, 104. (88) Comsol Multiphysics, Comsol, https://www.comsol.com (89) Nogenmyr, K.J.; Fureby, C.; Bai, X.S.; Petersson R.; Collin, R.; Linne, M. Combust. Flame 2009, 156, 25. (90) Stockman, E.S., PhD thesis, Microwave Enhancement Combustion of Laminar Hydrocarbon Flame Fronts, Princeton University 2009.


61

Energy & Fuels

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 62 of 63

Appendix 1 The processing scheme for c was evaluated by performing edge detection of the progress variable data to construct a synthetic CH2O signal. This synthetic CH2O signal was then compared to the actual CH2O signal for the measurement case to verify that the position of the synthetic and actual CH2O signal agree it the central axis of the burner. The first step of selecting the OH distribution used for c was threshing of the OH signal. The threshold value was selected to 15% of the OH signal peak found right after the flame front. The flame is run in lean conditions close to the point of lean blow-off, causing colder air to entrain the post flame region. This gas transport interferes with the post flame and disturbs the supersaturated OH distribution in this region. A processing routine for filling in these holes in the OH distribution downstream of the flame front were applied on each PLIF image to merely focus on the lower parts of the flame where the actual flame front is located. Each OH-PLIF image resulted in a binary image (BOH) describing the OH distribution in the flame cup. A distribution of OH occurrence was formed by averaging the binary images. A pixel value of 1 represents a location that is within the post flame region for all evaluated images whereas a value of 0 represents an area that is located in the unburned region for all the PLIF images that was evaluated. The contour of the region in the image with a value of 0.5 is estimated as the flame front since half of the evaluated images have OH present in this region. The edge of the OH signal was estimated by using a convolution filter on the binary images (BOH) that was used to estimate c. The idea behind such an operation is to form a distribution that covers an area that is slightly larger than BOH. This will define the location of the OH edge just outside of the OH distribution which is the location of the CH2O distribution (that is seen in the flame cup of the low-swirl flame). The convolution filter was a Gaussian kernel (40x40 pixels) with a full width at half maximum


62

Page 63 of 63

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Energy & Fuels

(FWHM) that corresponds to 1 mm in the image. The Gaussian kernel is convolved with BOH, forming SOH, which has an edge at the flame front that is a step response function to the Gaussian kernel. The location of the actual CH2O signal is found about 0.5 to 0.6 mm outside of the OH distribution. This location is found where SOH is between 0.05 and 0.1. A second binary image, using the condition 0.05

Skeletal MethaneâAir Reaction Mechanism for ... - ACS Publications

Recommend Documents

Skeletal MethaneâAir Reaction Mechanism for ... - ACS Publications