- Email: [email protected]
A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate
ARTICLE IN PRESS
JID: PROCI
[m;October 12, 2016;21:9]
Available online at www.sciencedirect.com
Proceedings of the Combustion Institute 000 (2016) 1–8 www.elsevier.com/locate/proci
A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate Seunghwan Keum∗, Cherian A. Idicheria, Paul M. Najt, Tang-Wei Kuo GM Global Research and Development Center, 30500 Mound Rd, Warren, MI 48090, USA Received 2 December 2015; accepted 12 August 2016 Available online xxx
Abstract In this work, we present a skeletal reaction mechanism for modeling ignition of iso-octane with low temperature plasma. The low temperature plasma was modeled by its main product ozone as a surrogate. A key chain branching reaction of oxygen radical attack on iso-octane was identified from reaction pathway analysis. A new skeletal reaction mechanism was developed by incorporating the key reaction along with ozone chemistry to the skeletal iso-octane reaction mechanism from Ra and Reitz [19]. Reaction rate constants of the key reactions and five other related reactions were calibrated to match predictions from the comprehensive reaction mechanism from Curran et al. [Combust. Flame 129 (2002) 253–280]. The optimized reaction mechanism shows good agreement with the comprehensive mechanism. It provides sufficient accuracy and computational simplicity to enable practical CFD simulation of low temperature plasma ignition in internal combustion engines. © 2016 by The Combustion Institute. Published by Elsevier Inc. Keywords: Plasma assisted ignition; Ozone assisted ignition; Reaction mechanism
1. Introduction With more stringent emission regulations and fuel economy requirements, the paradigm of the gasoline internal combustion engine is shifting toward lean combustion, either with air or exhaust gas recirculation (EGR) dilution. Lean combustion provides better efficiency from higher specific heat capacity ratio and reduced pumping losses. The combustion temperature is lower than stoichiometric combustion, which reduces NOx emissions sig-
∗
Corresponding author. E-mail addresses: [email protected], [email protected] (S. Keum).
nificantly. However, the level of improvement decreases at higher dilution level because of reduced combustion stability. As the efficiency gain is limited by reduced engine stability at the dilution tolerance, significant efforts have been made to extend the dilution limit using novel ignition systems, such as dual coil igniter [1], microwave ignition [2] and low temperature plasma [3]. The low temperature plasma uses a non-thermal plasma that does not transition to high temperature arc and glow phase. Kadono et al. [3] investigated the ignition and combustion using a low temperature corona discharge with different polarities and reported volumetric ignition and faster combustion than conventional spark, which are beneficial for combustion stability. Shiraishi et al. [4,5] investigated low tempera-
http://dx.doi.org/10.1016/j.proci.2016.08.035 1540-7489 © 2016 by The Combustion Institute. Published by Elsevier Inc.
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
JID: PROCI
2
ARTICLE IN PRESS
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
ture plasma ignition in lean homogeneous charge compression ignition (HCCI) engine and reported superior ignition characteristics than conventional spark discharge. Suess et al. [6] performed experiments with corona ignition device in an HCCI engine and reported superior combustion stability with corona ignition than conventional spark, and extension of dilution limit with the corona ignition. The ignition from a corona ignition device showed volumetric ignition, which generated multiple flame fronts that benefited the combustion stability. Mechanism of low temperature plasma ignition has been investigated by a number of researchers. Wu et al. [7] measured radical species concentration in the afterglow of a pulsed discharge and reported that low temperature plasma generates radical pool at temperatures as low as 400–800 K. Kosarev et al. [8] reported that the increase in reaction rate was due to chemical effect rather than thermal effect. The electron impact dissociation of oxygen molecules generates highly reactive oxygen radical, which promotes chain branching reactions to increase the reaction rate. The thermal heating from the plasma was reported negligible [8,9]. Breden et al. [10] performed numerical modeling of plasma igniter to investigate the radical yield and their effect on ignition and combustion. However, the computing overhead from the plasma chemistry and related governing equations are yet prohibitive for computational fluid dynamic (CFD) modeling of the plasma igniter in an IC engine. In the present study, a skeletal reaction mechanism is developed to model the plasma ignition using conventional CFD in IC engine applications. Instead of resolving all plasma chemistry, the nonthermal plasma is modeled by surrogates with oxygen radical and ozone, following Kosarev et al. [8,9]. Conceptually, the surrogate approach is similar to the conventional thermal spark ignition model in IC engine CFD modeling, where the thermal energy is used as a surrogate for the thermal plasma. The oxygen radical yield can be obtained from a separate plasma simulation [10], which can be used as an input to the combustion CFD modeling along with the developed reaction mechanism. 2. Reaction mechanism development: base mechanisms Curran et al. [11] developed a comprehensive mechanism for iso-octane, which is considered a surrogate for gasoline fuel. The reaction mechanism contains detailed oxygen radical pathways and has been used to understand the effect of oxygen radical in combustion [4]. The detailed reaction mechanism is fairly large with 874 species 3796 reactions, and it is not practical for combustion CFD even with the modern computing resources. Such full mechanisms have been mainly served as a refer-
ence for the development and validation of reduced mechanisms that are small enough for CFD simulations. Recently, efforts have been made to utilize the full mechanisms in CFD by taking advantage of advanced chemistry solvers, for example advanced ordinary differential (ODE) equation solver [12] or tabulated chemistry [13,14]. Contino et al. [15] performed CFD simulation with a large reaction mechanism (1062 species) using advanced tabulated chemistry solver [14] to investigate the effect of ozone seeding on HCCI combustion. However, adaptation of such advanced solver requires significant modification on the CFD code, which is beyond the scope of the current work. Instead, the current study focused on developing a reaction mechanism that is capable of capturing the effect of oxygen radical, at the same time small enough for CFD applications. There are several methods to reduce a large mechanism to a smaller mechanism [16,17]. However, such mathematical reduction methods have been generally limited to approximately 200 species, which is still too large for engineering combustion CFD applications. Further reduction of a reaction mechanism often involves grouping similar species into a single representative species (isomer lumping) and calibration of reaction rate coefficients [18]. Ra and Reitz [19] proposed a heavily reduced, skeletal primary reference fuel (PRF, a mixture of iso-octane and n-heptane) with 41 species and 130 reactions. The mechanism reduction was first carried out by conventional model reduction methodology, starting from sensitivity analysis, reaction elimination and chemical lumping. Then, the reduced mechanism is further reduced by combining several reactions into one and removing redundant species. Finally, the reaction rate coefficients were calibrated to meet target values. The skeletal mechanism has shown good prediction in gasoline engine modeling with very efficient simulation performance [20]. As such, the skeletal mechanism from Ra and Reitz [19] has been selected as the base mechanism for plasma ignition model development, instead of generating another reduced mechanism from the full mechanism. The skeletal reaction mechanism has sets of reactions for iso-octane, n-heptane, and small hydrocarbon reactions which are common between iso-octane and n-heptane. The present study used only iso-octane and small hydrocarbon reactions with 36 species and 54 reactions. 3. Reaction mechanism development: reaction pathway analysis A reaction pathway analysis was conducted using the Curran’s detailed mechanism [11] to identify key reaction pathway of oxygen radicals. A commercial chemistry solver Chemkin Pro [21] was used for the reaction pathway analysis. Ignition
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
JID: PROCI
ARTICLE IN PRESS
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
3
mechanism of Ra and Reitz [19]. The new mechanism development began with incorporating the chain branching reaction (1) into the skeletal mechanism. IC8 H18 + O = IC8 H17 + OH
Fig. 1. Oxygen radical pathway analysis with detailed mechanism, A/F = 20:1, 6000 PPM oxygen radical.
Fig. 2. Oxygen radical pathway analysis with detailed mechanism, A/F = 20:1 with 60% EGR, 6000 PPM oxygen radical.
delay in a constant volume perfectly stirred reactor (PSR) was examined. The initial temperature and pressure was given as 800 K 10 bar, which is typical temperature and pressure range near the spark timing in a spark ignition engine at part load to medium load operation. Overall air–fuel ratio was 20:1 with 6000 PPM of oxygen radical. The authors examined different amount of oxygen radical to find that the reaction pathway is not affected by the amount of oxygen radical. As combustion products such as CO2 and H2 O are introduced to control combustion in IC engines, 60% exhaust gas recirculation (EGR) condition was tested. The results are shown in Fig. 1 (zero EGR) and Fig. 2 (60% EGR). The reaction pathway analysis showed that most of the oxygen radical reacts with iso-octane to produce octyl radical and its isomers regardless of the presence of the combustion product, particularly water. Small amount of oxygen radical recombined to oxygen, while very small amount of oxygen radical formed carbon monoxide in the presence of CO2 (Fig. 2). From the reaction pathway analysis, it was found that the dominant pathway of the oxygen radical from the low-temperature plasma is the chain branching reaction of oxygen radical with iso-octane, as proposed by other researchers [4,8,9]. It is interesting to note that the oxygen radical did not show any pathway with water in the high EGR case. 4. Reaction mechanism development: skeletal mechanism The chain branching reaction of oxygen radical attack on iso-octane is not included in the skeletal
(1)
Reduced mechanism typically retains highly sensitive reactions and species from the detailed reaction mechanism. As the reactions are tightly coupled via lumped pseudo-species [18], the model performance may deteriorate if any reaction is added or modified in a way that affects the species balance. As the current base mechanism is a heavily reduced skeletal mechanism, calibrating the reaction rate coefficients of reaction (1) alone failed to capture the effect of oxygen radical. Additional reactions were selected by the impact on species balance from reaction (1). The chain branching reaction (1) in an oxygen radical rich condition affects hydroxyl concentration and balance among related species via the following reactions. Hydroxyl can attack the fuel species in another chain branching (2), recombine to hydroperoxyl (3), which participates in another chain branching to form hydroperoxide (4). Hydroperoxyl and hydroperoxide are balanced by hydroperoxyl recombination reaction (5). IC8 H18 + OH = IC8 H17 + H2 O
(2)
OH + OH = HO2 + O
(3)
IC8 H18 + HO2 = IC8 H17 + H2 O2
(4)
HO2 + HO2 = H2 O2 + O2
(5)
Finally, the oxygen radical may recombine with hydroxyl to form oxygen. O + OH=O2 + H
(6)
From these reactions, it is evident that increased oxygen radical concentration would affect all of the five reactions (1)–(5), and inclusion of reaction (1) would require calibration of reaction rate coefficients of all related reactions (2)–(6). It is interesting to note that all reactions are related to hydroperoxyl and hydroperoxide, which have been reported as key species in low temperature chemistry [19]. If the plasma is applied to fresh air without fuel species, plasma-generated oxygen radical may recombined with oxygen to form ozone, which is much more stable than the oxygen radical. This can occur if the plasma igniter is activated during the intake stroke or early compression, prior to any fuel injection. The ozone will dissociate during the compression stroke as the charge temperature becomes higher than 600 K. Therefore, the ozone serves as a carrier of oxygen radical. As such,
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
ARTICLE IN PRESS
JID: PROCI
4
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
Table 1 Ozone–oxygen reaction from Mohammadi et al. [25]. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
O3 + O = 2O2 O3 + H = O2 + OH O3 + H = HO2 + O O3 = O2 + O O3 + M = O2 + O + M O3 + H2 O = O2 + H2 O2 H2 O2 +O3 = H2 O + 2O2 O3 + OH = O2 + HO2 O3 + HO2 = 2O2 + OH HCO + O3 = H + CO2 + O2 CH3 + O3 = CH3 O + O2 O3 + CO = O2 + CO2
Table 2 Variable range for reaction rate coefficient calibration. Equivalence ratio Temperature (K) Pressure (bar) O3 concentration (PPM)
0.5, 1.0 600, 800, 1000, 1200 10 0–5000
the ozone breakdown and recombination are also important in the plasma ignition modeling. There are several ozone chemistry reaction mechanisms. The effect of ozone decomposition has been investigated [22,23]. Halter et al. [24] and Mohammadi et al. [25] proposed ozone mechanism for modeling the effect of ozone on natural gas combustion, while Pinchak et al. [26] developed one to model the ozone effect on ethylene flame. The oxygen radical– ozone chemistry from Ref. [25] was included in the reaction mechanism to model the recombination and breakdown of ozone species, and its interaction to the small hydrocarbon species. The ozone chemistry is reproduced from Ref. [25] in Table 1.
5. Optimization of reaction rate constants Once the oxygen radical related reactions are identified, the reaction rate coefficients for those reactions are optimized. Due to the lack of experimental data on plasma assisted combustion on larger hydrocarbon fuels, the reaction rate coefficients were calibrated to match ignition delay prediction from detailed iso-octane reaction mechanism [11]. The ignition delay was calculated using Chemkin-Pro package with perfectly stirred reactor (PSR) model. The operating conditions for reaction rate coefficient optimization are summarized in Table 2. Even though the oxygen radical is the key species in plasma chemistry, it is more likely that the produced oxygen radical will stay as more stable ozone species unless the ambient pressure and temperature are high enough. Therefore, the plasma was modeled by ozone as its surrogate. Ignition delay was determined at the time when the mixture temperature is increased by 400 K from the initial
Fig. 3. Difference in ignition delay prediction at 10 bar with equivalence ratio 1.0 between detailed and skeletal mechanisms.
temperature [19]. It is noted that Curran’s detailed mechanism and Ra’s skeletal mechanism do not produce identical results even without the presence of oxygen radical/ozone. The predicted ignition delays without ozone are compared in Fig. 3. The detailed mechanism shows very strong NTC behavior at low temperature than the skeletal mechanism. The difference in NTC behavior was discussed and explained in Ra and Reitz [19]. Matching the baseline predictions between the two mechanisms would be an interesting study, but it is beyond the scope of the current study. As the baselines differ between two mechanisms, the optimization process was focused on matching relative reduction of ignition delay, instead of the absolute number. The reaction rate coefficients in reactions (1)–(6) were optimized to minimize the difference in predicted ignition delay from Curran and the current mechanism. The optimization was carried out by archive based micro genetic algorithm (AMGA) method [27], which finds an optimal solution based on Darwin’s theory of survivalof-the-fittest. The optimized rate coefficients are listed in Table 3.
6. Model validation Due to lack of experimental data, the developed reaction mechanism was validated against ignition delay prediction from the detailed iso-octane mechanism. The validation conditions are chosen from relevant in-cylinder conditions of typical naturally aspirated internal combustion engine (Table 4). First, the reduction in ignition delay with ozone addition for Curran’s mechanism is shown in Fig. 4. Three different ozone amounts at 500, 1000 and 5000 PPM were examined. The equivalence ratio was 1.0 at 10 bar pressure. At lower temperature (T < 1000 K), ozone is very effective in reducing the
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
ARTICLE IN PRESS
JID: PROCI
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
5
Table 3 Optimized reaction rate coefficients. Reaction
A
b
EA
(1) (2) (3) (4) (5) (6)
3.88 × 105
3.81 2.2 – – – 0
13,291 1511 9544 10,815 −4000 −2692
IC8 H18 + O = IC8 H17 + OH IC8 H18 + OH = IC8 H17 + H2 O OH + OH = HO2 + O IC8 H18 + HO2 = IC8 H17 + H2 O2 HO2 + HO2 = H2 O2 + O2 O + OH =O2 + H
2.78 × 105 6.96 × 1013 4.61 × 1013 7.42 × 1012 6.36 × 1012
Table 4 Test conditions for validation. Equivalence ratio Temperature (K) Pressure (bar) EGR (%) O3 concentration (PPM)
0.3–1.2 600–1200 10 0–30 0–5000
Fig. 5. Ignition delay with different amounts of ozone from the new developed mechanism at 10 bar stoichiometry.
Fig. 4. Ignition delay with different amounts of ozone from detailed mechanism at 10 bar stoichiometry.
ignition delay up to two orders of magnitude. On the other hand, the reduction in ignition delay at higher temperature (T > 1000 K) becomes noticeable only at very high concentration (5000 PPM). The ozone addition, or plasma ignition, will be most effective at the low temperature. It is interesting to note that the NTC behavior is reduced as the ozone amount is increased. The ignition delay with the developed reaction mechanism is shown in Fig. 5. The ozone sensitivity to the ignition delay at low temperature (1000/T > 1.2) and high temperature (1000/T < 1.0) is very well captured with the proposed mechanism. The ozone sensitivity at medium temperature (1 < 1000/T < 1.2) is under predicted compared to the detailed mechanism. The developed mechanism showed significant improvement from the original skeletal mechanism in predicting the effect of plasma represented by ozone. To examine the effect of additional reactions, the ignition delay with different amounts of ozone
Fig. 6. Ignition delay with different amounts of ozone from unmodified skeletal mechanism.
from unmodified, original Ra’s skeletal mechanism is compared in Fig. 6. Ra’s mechanism showed little sensitivity to the ozone amount over all temperature range. One of the main reasons for the lack of sensitivity is the missing chain branching step of oxygen radical reaction on fuel species. The ozone and oxygen radical concentration from the skeletal mechanism is examined in Fig. 7. The test condition is at 600 K with 5000 PPM ozone concentration. The initial ozone is
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
JID: PROCI
6
ARTICLE IN PRESS
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
Fig. 7. Ozone and oxygen radical concentration in mole fraction with Ra’s mechanism at 600 K 10 bar, 5000 PPM initial ozone concentration.
Fig. 9. Temporal evolution of oxygen radical and fuel species with the developed mechanism at 600 K 10 bar, 5000 PPM initial ozone concentration. 10
OH Mole fraction
10
10
10
10
10
10
0
No Ozone Ozone 5000 PPM
-2
-4
-6
-8
-10
-12
0
0.5
1
1.5
2
2.5
3
3.5
4
time [ms]
Fig. 8. Temporal evolution of oxygen radical and fuel species in mole fraction at 600 K initial temperature with 5000 PPM initial ozone concentration.
decomposed into oxygen radical rapidly due to the high temperature. However, the oxygen radical stays inert and does not react with any other species, as the mechanism has no such pathways. The oxygen radical does not react until enough radical pool has been generated from thermal decomposition of iso-octane species. The temporal evolutions of oxygen radical and fuel species are compared in Fig. 8. The depletion of oxygen radical occurs only after the iso-octane has decomposed and generated radical pool for oxygen radical to react. The oxygen radical history is shown in Fig. 9. The oxygen radical quickly reacts with fuel species within 0.2 ms from the start of simulation, generating significant amount of alkyl radical (C8H17), which decomposes further as reaction progresses. It shows that the chain propagation by oxygen radical is the key initiation step in plasma-assisted ignition.
Fig. 10. Change in OH radical concentration with ozone addition at 600 K 10 bar.
The chain propagation by the oxygen radical should increase the radical pool. The change in the radical pool concentration against ozone addition is examined with the OH radical. The results are shown in Fig. 10. The OH concentration from the baseline case (no ozone addition) is shown in solid black line, while the OH concentration with 5000 PPM is shown in dotted line. When the ozone is added, significant increase in the OH concentration was observed, confirming that the developed model captures the increase in the radical pool very well.
7. Ignition at high dilution level One of the primary objectives in the low temperature plasma ignition is to extend the dilution limit. The model performance at high dilution level was examined. First, the ignition delay with 30% exhaust gas recirculation (EGR) was tested. The results are shown in Fig. 11.
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
JID: PROCI
ARTICLE IN PRESS
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
Fig. 11. Ignition delay comparison at stoichiometry with 30% EGR. Thin lines: detailed mechanism, thick lines: new mechanism.
7
Fig. 13. Effect of ozone in HCCI ignition timing, equivalence ratio 0.5.
with ozone amount was well captured with the developed mechanism. 8. Application in IC engine simulations
Fig. 12. Ignition delay comparison at equivalence ratio of 0.5, thin lines: detailed mechanism, thick lines: new mechanism.
Notice the difference in baseline (No O3 ) plotted in solid lines. This is due to the difference in prediction from detailed and skeletal mechanism. The reduction in ignition delay with different amount of ozone is similar between Curran’s mechanism and the developed mechanism, showing that the current mechanism is capable of reproducing ignition delay reduction from ozone. Although the trends are captured correctly, the one-order-of magnitude differences in predicted ignition delay time between detailed and current mechanisms can be a concern for CFD applications. The ignition delay at lean mixture with equivalence ratio 0.5 is shown in Fig. 12. At the lean case, both the detailed and the skeletal mechanisms showed similar prediction with zero ozone concentration (solid lines), compared to other cases. The comparison with different ozone concentration is reasonably accurate, considering the different mechanism sizes. The trend in the ignition delay
Due to the lack of detailed information ozone production from low temperature plasma in IC engine environment, the developed mechanism is evaluated with prescribed amount of ozone seeding. Mohammadi et al. [25] investigated the effect of ozone addition on natural gas engine and reported that adding small amount of ozone (33 PPM) was able to produce stable combustion in a lean mixture with 0.24 equivalence ratio. Combustion phasing was advanced with increase in ozone amount up to 100 PPM level. Similar findings were reported by Masurier et al. [28] from HCCI combustion with iso-octane, where significant combustion phasing advance was observed with 19.6 PPM ozone. Schönborn et al. [29] reported possible ignition phasing control in HCCI by controlling the amount of ozone. The performance of the developed mechanism was tested to examine if it is capable of reproducing the well-reported trend of reducing ignition delay. An HCCI engine with 14:1 compression ratio was modeled using Chemkin HCCI combustion simulation [21]. A parametric study on ozone amount was conducted at equivalence ratio 0.5. The operating condition was set at 1300 RPM with initial temperature and pressure at 350 K and 1 bar, respectively. The results are shown in Fig. 13. The baseline case without any ozone addition is shown with solid line, and it shows that the engine is misfiring without any ozone addition. Stable combustion was observed when a small amount of ozone (20 PPM) was added to the in-cylinder mixture, and the trend observed in Mohammadi et al. [25] and Masurier et al. [28] was successfully reproduced. The simulation also showed that further
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
JID: PROCI
8
ARTICLE IN PRESS
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
addition advances the combustion phasing, as reported by Mohammadi et al. [25] and Schönborn et al. [29]. Mohammedi et al. [25] reported that the effect of additional ozone is saturated, which is well captured in Fig. 13. The developed mechanism is capable in capturing reported key trends reported in IC engine experiments with ozone addition.
9. Conclusions A skeletal iso-octane reaction mechanism for modeling low temperature plasma ignition was developed. A surrogate based approach was taken to model the plasma using ozone, the major product from low temperature plasma. A reaction pathway analysis using Curran’s detailed iso-octane reaction mechanism showed that the oxygen radical reaction with iso-octane is the key step in the low temperature plasma ignition. The identified reaction and ozone chemistry were merged into a skeletal iso-octane reaction mechanism. The reaction rate coefficients in related reactions were optimized to match the reduction in ignition delay from detailed iso-octane mechanism with different ozone amount. The optimized skeletal reaction mechanism was validated against ignition delay prediction from Curran’s mechanism, using the ozone as a surrogate for low temperature plasma. It was found that the inclusion of the oxygen radical reaction on fuel species is key in reproducing fast ignition with ozone. The developed mechanism is simple enough for CFD application. The developed mechanism showed good agreement to the detailed mechanism, especially at low temperature range. However, noticeable difference in the ignition delay at high dilution level was also observed, which might limit the predictive capability. The developed mechanism was tested using Chemkin HCCI combustion model. It was found that the current mechanism is capable of reproducing key characteristics from ozone addition, such as promoting ignition and controlling combustion phasing.
Acknowledgments The authors appreciate constructive discussion and advice from Dr. Ronald O. Grover Jr. of GM Global Research and Development Center. References [1] T. Alger, J. Gingrich, B. Mangold, C. Roberts, SAE Int. J. Engines 4 (1) (2011) 677–692.
[2] B. Wolk, A. DeFilippo, J.-Y. Chen, R. Dibble, A. Nishiyama, Y. Ikeda, Combust. Flame 160 (7) (2013) 1225–1234. [3] T. Kadono, K. Yoshida, H. Shoji, SAE 2002-321823, 2002. [4] T. Shiraishi, T. Urushihara, SAE 2011-01-0660, 2011. [5] T. Shiraishi, Advanced Ignition Systems for Gasoline Engines, IAV, Berlin, Germany, 2012. [6] M. Suess, M. Guenthner, M. Schenk, H-S. Rottengruber, Proc. IMechE, Part D: J. Automob. Eng. 226 (2012) 275–286. [7] L. Wu, J. Lane, N.P. Cernansky, D.L. Miller, A.A. Fridman, A.Yu. Starikovskiy, Proc. Combust. Inst. 33 (2011) 3219–3224. [8] I.N. Kosarev, N.L. Aleksandrov, S.V. Kindysheva, S.M. Starikovskaia, A.Y. Starikovskii, J. Phys. D: Appl. Phys. 41 (3) (2008). [9] N.L. Aleksandrov, S.V. Kindysheva, I.N. Kosarev, S.M. Starikovskaia, A.Y. Starikovskii, Proc. Combust. Inst. 32 (2009) 205–212. [10] D. Breden, L.L. Raja, C.A. Idicheria, P.M. Najt, S. Mahadevan, J. Appl. Phys. 114 (2013) 444. [11] H.J. Curran, P. Gauffuri, W.J. Pitz, C.K. Westbrook, Combust. Flame 129 (2002) 253–280. [12] M.J. McNenly, R. Whitesides, Proc. Combust. Inst. 35 (1) (2015) 581–587. [13] J.A. van Oijen, L.P.H. de Goey, Combust. Sci. Technol. 161 (1) (2000) 113–137. [14] F. Contino, H. Jeanmart, T. Lucchini, G. D’Errico, Proc. Comb. Inst. 33(2) 3057–3064. [15] F. Contino, J.-B. Masurier, F. Foucher, T. Lucchini, G. D’Errico, P. Dagaut, Fuel 137 2104 179–184. [16] T. Lu, C.K. Law, Combust. Flame 144 (2006) 24–36. [17] Y. Shi, H.-W. Ge, J.L. Brakora, R.D. Reitz, Energy Fuels 24 (2010) 1646–1654. [18] Y. Ra, R.D. Reitz, Combust. Flame 162 (2015) 3456–3481. [19] Y. Ra, R.D. Reitz, Combust. Flame 155 (2008) 713–738. [20] S. Keum, C.A. Idicheria, in: ASME 2013 Internal Combustion Engine Division Fall Technical Conference, 2, ICEF2013-19216, Dearborn, MI, USA, 2013. [21] Chemkin Pro: http://www.reactiondesign.com/ products/chemkin/chemkin-pro/. [22] H. Yamada, M. Yoshii, A. Tezake, Proc. Combust. Inst. 30 (2005) 2773–2780. [23] F. Foucher, P. Higelin, C. Mounaim-Rousselle, P. Dagaut, Proc. Combust. Inst. 34 (2013) 3005–3012. [24] F. Halter, P. Higelin, P. Dagaut, Energy Fuels 25 (2011) 2909–2916. [25] A. Mohammadi, H. Kawanabe, T. Ishiyama, M. Shioji, A. Komada, SAE 2006-01-0419, 2006. [26] M. Pinchak, T. Ombrello, C. Carter, E. Gutmark, V. Katta, Philos. Trans. R. Soc. A 373, 2014, 20140339. [27] S. Tiwari, P. Koch, G. Fadel, K. Deb, in: Genetic and Evolutionary Computation Conference (GECCO 2008), July 12–16, Atlanta, Georgia, USA, 2008. [28] J.-B. Masurier, F. Foucher, G. Dayma, P. Dagaut, Proc. Combust. Inst. 35 (2015) 3125–3132. [29] A. Schönborn, P. Hellier, A.E. Aliev, N. Ladommatos, Fuel 89 (2010) 3178–3184.
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
JID: PROCI
[m;October 12, 2016;21:9]
Available online at www.sciencedirect.com
Proceedings of the Combustion Institute 000 (2016) 1–8 www.elsevier.com/locate/proci
A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate Seunghwan Keum∗, Cherian A. Idicheria, Paul M. Najt, Tang-Wei Kuo GM Global Research and Development Center, 30500 Mound Rd, Warren, MI 48090, USA Received 2 December 2015; accepted 12 August 2016 Available online xxx
Abstract In this work, we present a skeletal reaction mechanism for modeling ignition of iso-octane with low temperature plasma. The low temperature plasma was modeled by its main product ozone as a surrogate. A key chain branching reaction of oxygen radical attack on iso-octane was identified from reaction pathway analysis. A new skeletal reaction mechanism was developed by incorporating the key reaction along with ozone chemistry to the skeletal iso-octane reaction mechanism from Ra and Reitz [19]. Reaction rate constants of the key reactions and five other related reactions were calibrated to match predictions from the comprehensive reaction mechanism from Curran et al. [Combust. Flame 129 (2002) 253–280]. The optimized reaction mechanism shows good agreement with the comprehensive mechanism. It provides sufficient accuracy and computational simplicity to enable practical CFD simulation of low temperature plasma ignition in internal combustion engines. © 2016 by The Combustion Institute. Published by Elsevier Inc. Keywords: Plasma assisted ignition; Ozone assisted ignition; Reaction mechanism
1. Introduction With more stringent emission regulations and fuel economy requirements, the paradigm of the gasoline internal combustion engine is shifting toward lean combustion, either with air or exhaust gas recirculation (EGR) dilution. Lean combustion provides better efficiency from higher specific heat capacity ratio and reduced pumping losses. The combustion temperature is lower than stoichiometric combustion, which reduces NOx emissions sig-
∗
Corresponding author. E-mail addresses: [email protected], [email protected] (S. Keum).
nificantly. However, the level of improvement decreases at higher dilution level because of reduced combustion stability. As the efficiency gain is limited by reduced engine stability at the dilution tolerance, significant efforts have been made to extend the dilution limit using novel ignition systems, such as dual coil igniter [1], microwave ignition [2] and low temperature plasma [3]. The low temperature plasma uses a non-thermal plasma that does not transition to high temperature arc and glow phase. Kadono et al. [3] investigated the ignition and combustion using a low temperature corona discharge with different polarities and reported volumetric ignition and faster combustion than conventional spark, which are beneficial for combustion stability. Shiraishi et al. [4,5] investigated low tempera-
http://dx.doi.org/10.1016/j.proci.2016.08.035 1540-7489 © 2016 by The Combustion Institute. Published by Elsevier Inc.
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
JID: PROCI
2
ARTICLE IN PRESS
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
ture plasma ignition in lean homogeneous charge compression ignition (HCCI) engine and reported superior ignition characteristics than conventional spark discharge. Suess et al. [6] performed experiments with corona ignition device in an HCCI engine and reported superior combustion stability with corona ignition than conventional spark, and extension of dilution limit with the corona ignition. The ignition from a corona ignition device showed volumetric ignition, which generated multiple flame fronts that benefited the combustion stability. Mechanism of low temperature plasma ignition has been investigated by a number of researchers. Wu et al. [7] measured radical species concentration in the afterglow of a pulsed discharge and reported that low temperature plasma generates radical pool at temperatures as low as 400–800 K. Kosarev et al. [8] reported that the increase in reaction rate was due to chemical effect rather than thermal effect. The electron impact dissociation of oxygen molecules generates highly reactive oxygen radical, which promotes chain branching reactions to increase the reaction rate. The thermal heating from the plasma was reported negligible [8,9]. Breden et al. [10] performed numerical modeling of plasma igniter to investigate the radical yield and their effect on ignition and combustion. However, the computing overhead from the plasma chemistry and related governing equations are yet prohibitive for computational fluid dynamic (CFD) modeling of the plasma igniter in an IC engine. In the present study, a skeletal reaction mechanism is developed to model the plasma ignition using conventional CFD in IC engine applications. Instead of resolving all plasma chemistry, the nonthermal plasma is modeled by surrogates with oxygen radical and ozone, following Kosarev et al. [8,9]. Conceptually, the surrogate approach is similar to the conventional thermal spark ignition model in IC engine CFD modeling, where the thermal energy is used as a surrogate for the thermal plasma. The oxygen radical yield can be obtained from a separate plasma simulation [10], which can be used as an input to the combustion CFD modeling along with the developed reaction mechanism. 2. Reaction mechanism development: base mechanisms Curran et al. [11] developed a comprehensive mechanism for iso-octane, which is considered a surrogate for gasoline fuel. The reaction mechanism contains detailed oxygen radical pathways and has been used to understand the effect of oxygen radical in combustion [4]. The detailed reaction mechanism is fairly large with 874 species 3796 reactions, and it is not practical for combustion CFD even with the modern computing resources. Such full mechanisms have been mainly served as a refer-
ence for the development and validation of reduced mechanisms that are small enough for CFD simulations. Recently, efforts have been made to utilize the full mechanisms in CFD by taking advantage of advanced chemistry solvers, for example advanced ordinary differential (ODE) equation solver [12] or tabulated chemistry [13,14]. Contino et al. [15] performed CFD simulation with a large reaction mechanism (1062 species) using advanced tabulated chemistry solver [14] to investigate the effect of ozone seeding on HCCI combustion. However, adaptation of such advanced solver requires significant modification on the CFD code, which is beyond the scope of the current work. Instead, the current study focused on developing a reaction mechanism that is capable of capturing the effect of oxygen radical, at the same time small enough for CFD applications. There are several methods to reduce a large mechanism to a smaller mechanism [16,17]. However, such mathematical reduction methods have been generally limited to approximately 200 species, which is still too large for engineering combustion CFD applications. Further reduction of a reaction mechanism often involves grouping similar species into a single representative species (isomer lumping) and calibration of reaction rate coefficients [18]. Ra and Reitz [19] proposed a heavily reduced, skeletal primary reference fuel (PRF, a mixture of iso-octane and n-heptane) with 41 species and 130 reactions. The mechanism reduction was first carried out by conventional model reduction methodology, starting from sensitivity analysis, reaction elimination and chemical lumping. Then, the reduced mechanism is further reduced by combining several reactions into one and removing redundant species. Finally, the reaction rate coefficients were calibrated to meet target values. The skeletal mechanism has shown good prediction in gasoline engine modeling with very efficient simulation performance [20]. As such, the skeletal mechanism from Ra and Reitz [19] has been selected as the base mechanism for plasma ignition model development, instead of generating another reduced mechanism from the full mechanism. The skeletal reaction mechanism has sets of reactions for iso-octane, n-heptane, and small hydrocarbon reactions which are common between iso-octane and n-heptane. The present study used only iso-octane and small hydrocarbon reactions with 36 species and 54 reactions. 3. Reaction mechanism development: reaction pathway analysis A reaction pathway analysis was conducted using the Curran’s detailed mechanism [11] to identify key reaction pathway of oxygen radicals. A commercial chemistry solver Chemkin Pro [21] was used for the reaction pathway analysis. Ignition
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
JID: PROCI
ARTICLE IN PRESS
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
3
mechanism of Ra and Reitz [19]. The new mechanism development began with incorporating the chain branching reaction (1) into the skeletal mechanism. IC8 H18 + O = IC8 H17 + OH
Fig. 1. Oxygen radical pathway analysis with detailed mechanism, A/F = 20:1, 6000 PPM oxygen radical.
Fig. 2. Oxygen radical pathway analysis with detailed mechanism, A/F = 20:1 with 60% EGR, 6000 PPM oxygen radical.
delay in a constant volume perfectly stirred reactor (PSR) was examined. The initial temperature and pressure was given as 800 K 10 bar, which is typical temperature and pressure range near the spark timing in a spark ignition engine at part load to medium load operation. Overall air–fuel ratio was 20:1 with 6000 PPM of oxygen radical. The authors examined different amount of oxygen radical to find that the reaction pathway is not affected by the amount of oxygen radical. As combustion products such as CO2 and H2 O are introduced to control combustion in IC engines, 60% exhaust gas recirculation (EGR) condition was tested. The results are shown in Fig. 1 (zero EGR) and Fig. 2 (60% EGR). The reaction pathway analysis showed that most of the oxygen radical reacts with iso-octane to produce octyl radical and its isomers regardless of the presence of the combustion product, particularly water. Small amount of oxygen radical recombined to oxygen, while very small amount of oxygen radical formed carbon monoxide in the presence of CO2 (Fig. 2). From the reaction pathway analysis, it was found that the dominant pathway of the oxygen radical from the low-temperature plasma is the chain branching reaction of oxygen radical with iso-octane, as proposed by other researchers [4,8,9]. It is interesting to note that the oxygen radical did not show any pathway with water in the high EGR case. 4. Reaction mechanism development: skeletal mechanism The chain branching reaction of oxygen radical attack on iso-octane is not included in the skeletal
(1)
Reduced mechanism typically retains highly sensitive reactions and species from the detailed reaction mechanism. As the reactions are tightly coupled via lumped pseudo-species [18], the model performance may deteriorate if any reaction is added or modified in a way that affects the species balance. As the current base mechanism is a heavily reduced skeletal mechanism, calibrating the reaction rate coefficients of reaction (1) alone failed to capture the effect of oxygen radical. Additional reactions were selected by the impact on species balance from reaction (1). The chain branching reaction (1) in an oxygen radical rich condition affects hydroxyl concentration and balance among related species via the following reactions. Hydroxyl can attack the fuel species in another chain branching (2), recombine to hydroperoxyl (3), which participates in another chain branching to form hydroperoxide (4). Hydroperoxyl and hydroperoxide are balanced by hydroperoxyl recombination reaction (5). IC8 H18 + OH = IC8 H17 + H2 O
(2)
OH + OH = HO2 + O
(3)
IC8 H18 + HO2 = IC8 H17 + H2 O2
(4)
HO2 + HO2 = H2 O2 + O2
(5)
Finally, the oxygen radical may recombine with hydroxyl to form oxygen. O + OH=O2 + H
(6)
From these reactions, it is evident that increased oxygen radical concentration would affect all of the five reactions (1)–(5), and inclusion of reaction (1) would require calibration of reaction rate coefficients of all related reactions (2)–(6). It is interesting to note that all reactions are related to hydroperoxyl and hydroperoxide, which have been reported as key species in low temperature chemistry [19]. If the plasma is applied to fresh air without fuel species, plasma-generated oxygen radical may recombined with oxygen to form ozone, which is much more stable than the oxygen radical. This can occur if the plasma igniter is activated during the intake stroke or early compression, prior to any fuel injection. The ozone will dissociate during the compression stroke as the charge temperature becomes higher than 600 K. Therefore, the ozone serves as a carrier of oxygen radical. As such,
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
ARTICLE IN PRESS
JID: PROCI
4
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
Table 1 Ozone–oxygen reaction from Mohammadi et al. [25]. (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
O3 + O = 2O2 O3 + H = O2 + OH O3 + H = HO2 + O O3 = O2 + O O3 + M = O2 + O + M O3 + H2 O = O2 + H2 O2 H2 O2 +O3 = H2 O + 2O2 O3 + OH = O2 + HO2 O3 + HO2 = 2O2 + OH HCO + O3 = H + CO2 + O2 CH3 + O3 = CH3 O + O2 O3 + CO = O2 + CO2
Table 2 Variable range for reaction rate coefficient calibration. Equivalence ratio Temperature (K) Pressure (bar) O3 concentration (PPM)
0.5, 1.0 600, 800, 1000, 1200 10 0–5000
the ozone breakdown and recombination are also important in the plasma ignition modeling. There are several ozone chemistry reaction mechanisms. The effect of ozone decomposition has been investigated [22,23]. Halter et al. [24] and Mohammadi et al. [25] proposed ozone mechanism for modeling the effect of ozone on natural gas combustion, while Pinchak et al. [26] developed one to model the ozone effect on ethylene flame. The oxygen radical– ozone chemistry from Ref. [25] was included in the reaction mechanism to model the recombination and breakdown of ozone species, and its interaction to the small hydrocarbon species. The ozone chemistry is reproduced from Ref. [25] in Table 1.
5. Optimization of reaction rate constants Once the oxygen radical related reactions are identified, the reaction rate coefficients for those reactions are optimized. Due to the lack of experimental data on plasma assisted combustion on larger hydrocarbon fuels, the reaction rate coefficients were calibrated to match ignition delay prediction from detailed iso-octane reaction mechanism [11]. The ignition delay was calculated using Chemkin-Pro package with perfectly stirred reactor (PSR) model. The operating conditions for reaction rate coefficient optimization are summarized in Table 2. Even though the oxygen radical is the key species in plasma chemistry, it is more likely that the produced oxygen radical will stay as more stable ozone species unless the ambient pressure and temperature are high enough. Therefore, the plasma was modeled by ozone as its surrogate. Ignition delay was determined at the time when the mixture temperature is increased by 400 K from the initial
Fig. 3. Difference in ignition delay prediction at 10 bar with equivalence ratio 1.0 between detailed and skeletal mechanisms.
temperature [19]. It is noted that Curran’s detailed mechanism and Ra’s skeletal mechanism do not produce identical results even without the presence of oxygen radical/ozone. The predicted ignition delays without ozone are compared in Fig. 3. The detailed mechanism shows very strong NTC behavior at low temperature than the skeletal mechanism. The difference in NTC behavior was discussed and explained in Ra and Reitz [19]. Matching the baseline predictions between the two mechanisms would be an interesting study, but it is beyond the scope of the current study. As the baselines differ between two mechanisms, the optimization process was focused on matching relative reduction of ignition delay, instead of the absolute number. The reaction rate coefficients in reactions (1)–(6) were optimized to minimize the difference in predicted ignition delay from Curran and the current mechanism. The optimization was carried out by archive based micro genetic algorithm (AMGA) method [27], which finds an optimal solution based on Darwin’s theory of survivalof-the-fittest. The optimized rate coefficients are listed in Table 3.
6. Model validation Due to lack of experimental data, the developed reaction mechanism was validated against ignition delay prediction from the detailed iso-octane mechanism. The validation conditions are chosen from relevant in-cylinder conditions of typical naturally aspirated internal combustion engine (Table 4). First, the reduction in ignition delay with ozone addition for Curran’s mechanism is shown in Fig. 4. Three different ozone amounts at 500, 1000 and 5000 PPM were examined. The equivalence ratio was 1.0 at 10 bar pressure. At lower temperature (T < 1000 K), ozone is very effective in reducing the
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
ARTICLE IN PRESS
JID: PROCI
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
5
Table 3 Optimized reaction rate coefficients. Reaction
A
b
EA
(1) (2) (3) (4) (5) (6)
3.88 × 105
3.81 2.2 – – – 0
13,291 1511 9544 10,815 −4000 −2692
IC8 H18 + O = IC8 H17 + OH IC8 H18 + OH = IC8 H17 + H2 O OH + OH = HO2 + O IC8 H18 + HO2 = IC8 H17 + H2 O2 HO2 + HO2 = H2 O2 + O2 O + OH =O2 + H
2.78 × 105 6.96 × 1013 4.61 × 1013 7.42 × 1012 6.36 × 1012
Table 4 Test conditions for validation. Equivalence ratio Temperature (K) Pressure (bar) EGR (%) O3 concentration (PPM)
0.3–1.2 600–1200 10 0–30 0–5000
Fig. 5. Ignition delay with different amounts of ozone from the new developed mechanism at 10 bar stoichiometry.
Fig. 4. Ignition delay with different amounts of ozone from detailed mechanism at 10 bar stoichiometry.
ignition delay up to two orders of magnitude. On the other hand, the reduction in ignition delay at higher temperature (T > 1000 K) becomes noticeable only at very high concentration (5000 PPM). The ozone addition, or plasma ignition, will be most effective at the low temperature. It is interesting to note that the NTC behavior is reduced as the ozone amount is increased. The ignition delay with the developed reaction mechanism is shown in Fig. 5. The ozone sensitivity to the ignition delay at low temperature (1000/T > 1.2) and high temperature (1000/T < 1.0) is very well captured with the proposed mechanism. The ozone sensitivity at medium temperature (1 < 1000/T < 1.2) is under predicted compared to the detailed mechanism. The developed mechanism showed significant improvement from the original skeletal mechanism in predicting the effect of plasma represented by ozone. To examine the effect of additional reactions, the ignition delay with different amounts of ozone
Fig. 6. Ignition delay with different amounts of ozone from unmodified skeletal mechanism.
from unmodified, original Ra’s skeletal mechanism is compared in Fig. 6. Ra’s mechanism showed little sensitivity to the ozone amount over all temperature range. One of the main reasons for the lack of sensitivity is the missing chain branching step of oxygen radical reaction on fuel species. The ozone and oxygen radical concentration from the skeletal mechanism is examined in Fig. 7. The test condition is at 600 K with 5000 PPM ozone concentration. The initial ozone is
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
JID: PROCI
6
ARTICLE IN PRESS
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
Fig. 7. Ozone and oxygen radical concentration in mole fraction with Ra’s mechanism at 600 K 10 bar, 5000 PPM initial ozone concentration.
Fig. 9. Temporal evolution of oxygen radical and fuel species with the developed mechanism at 600 K 10 bar, 5000 PPM initial ozone concentration. 10
OH Mole fraction
10
10
10
10
10
10
0
No Ozone Ozone 5000 PPM
-2
-4
-6
-8
-10
-12
0
0.5
1
1.5
2
2.5
3
3.5
4
time [ms]
Fig. 8. Temporal evolution of oxygen radical and fuel species in mole fraction at 600 K initial temperature with 5000 PPM initial ozone concentration.
decomposed into oxygen radical rapidly due to the high temperature. However, the oxygen radical stays inert and does not react with any other species, as the mechanism has no such pathways. The oxygen radical does not react until enough radical pool has been generated from thermal decomposition of iso-octane species. The temporal evolutions of oxygen radical and fuel species are compared in Fig. 8. The depletion of oxygen radical occurs only after the iso-octane has decomposed and generated radical pool for oxygen radical to react. The oxygen radical history is shown in Fig. 9. The oxygen radical quickly reacts with fuel species within 0.2 ms from the start of simulation, generating significant amount of alkyl radical (C8H17), which decomposes further as reaction progresses. It shows that the chain propagation by oxygen radical is the key initiation step in plasma-assisted ignition.
Fig. 10. Change in OH radical concentration with ozone addition at 600 K 10 bar.
The chain propagation by the oxygen radical should increase the radical pool. The change in the radical pool concentration against ozone addition is examined with the OH radical. The results are shown in Fig. 10. The OH concentration from the baseline case (no ozone addition) is shown in solid black line, while the OH concentration with 5000 PPM is shown in dotted line. When the ozone is added, significant increase in the OH concentration was observed, confirming that the developed model captures the increase in the radical pool very well.
7. Ignition at high dilution level One of the primary objectives in the low temperature plasma ignition is to extend the dilution limit. The model performance at high dilution level was examined. First, the ignition delay with 30% exhaust gas recirculation (EGR) was tested. The results are shown in Fig. 11.
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
JID: PROCI
ARTICLE IN PRESS
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
Fig. 11. Ignition delay comparison at stoichiometry with 30% EGR. Thin lines: detailed mechanism, thick lines: new mechanism.
7
Fig. 13. Effect of ozone in HCCI ignition timing, equivalence ratio 0.5.
with ozone amount was well captured with the developed mechanism. 8. Application in IC engine simulations
Fig. 12. Ignition delay comparison at equivalence ratio of 0.5, thin lines: detailed mechanism, thick lines: new mechanism.
Notice the difference in baseline (No O3 ) plotted in solid lines. This is due to the difference in prediction from detailed and skeletal mechanism. The reduction in ignition delay with different amount of ozone is similar between Curran’s mechanism and the developed mechanism, showing that the current mechanism is capable of reproducing ignition delay reduction from ozone. Although the trends are captured correctly, the one-order-of magnitude differences in predicted ignition delay time between detailed and current mechanisms can be a concern for CFD applications. The ignition delay at lean mixture with equivalence ratio 0.5 is shown in Fig. 12. At the lean case, both the detailed and the skeletal mechanisms showed similar prediction with zero ozone concentration (solid lines), compared to other cases. The comparison with different ozone concentration is reasonably accurate, considering the different mechanism sizes. The trend in the ignition delay
Due to the lack of detailed information ozone production from low temperature plasma in IC engine environment, the developed mechanism is evaluated with prescribed amount of ozone seeding. Mohammadi et al. [25] investigated the effect of ozone addition on natural gas engine and reported that adding small amount of ozone (33 PPM) was able to produce stable combustion in a lean mixture with 0.24 equivalence ratio. Combustion phasing was advanced with increase in ozone amount up to 100 PPM level. Similar findings were reported by Masurier et al. [28] from HCCI combustion with iso-octane, where significant combustion phasing advance was observed with 19.6 PPM ozone. Schönborn et al. [29] reported possible ignition phasing control in HCCI by controlling the amount of ozone. The performance of the developed mechanism was tested to examine if it is capable of reproducing the well-reported trend of reducing ignition delay. An HCCI engine with 14:1 compression ratio was modeled using Chemkin HCCI combustion simulation [21]. A parametric study on ozone amount was conducted at equivalence ratio 0.5. The operating condition was set at 1300 RPM with initial temperature and pressure at 350 K and 1 bar, respectively. The results are shown in Fig. 13. The baseline case without any ozone addition is shown with solid line, and it shows that the engine is misfiring without any ozone addition. Stable combustion was observed when a small amount of ozone (20 PPM) was added to the in-cylinder mixture, and the trend observed in Mohammadi et al. [25] and Masurier et al. [28] was successfully reproduced. The simulation also showed that further
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035
JID: PROCI
8
ARTICLE IN PRESS
[m;October 12, 2016;21:9]
S. Keum et al. / Proceedings of the Combustion Institute 000 (2016) 1–8
addition advances the combustion phasing, as reported by Mohammadi et al. [25] and Schönborn et al. [29]. Mohammedi et al. [25] reported that the effect of additional ozone is saturated, which is well captured in Fig. 13. The developed mechanism is capable in capturing reported key trends reported in IC engine experiments with ozone addition.
9. Conclusions A skeletal iso-octane reaction mechanism for modeling low temperature plasma ignition was developed. A surrogate based approach was taken to model the plasma using ozone, the major product from low temperature plasma. A reaction pathway analysis using Curran’s detailed iso-octane reaction mechanism showed that the oxygen radical reaction with iso-octane is the key step in the low temperature plasma ignition. The identified reaction and ozone chemistry were merged into a skeletal iso-octane reaction mechanism. The reaction rate coefficients in related reactions were optimized to match the reduction in ignition delay from detailed iso-octane mechanism with different ozone amount. The optimized skeletal reaction mechanism was validated against ignition delay prediction from Curran’s mechanism, using the ozone as a surrogate for low temperature plasma. It was found that the inclusion of the oxygen radical reaction on fuel species is key in reproducing fast ignition with ozone. The developed mechanism is simple enough for CFD application. The developed mechanism showed good agreement to the detailed mechanism, especially at low temperature range. However, noticeable difference in the ignition delay at high dilution level was also observed, which might limit the predictive capability. The developed mechanism was tested using Chemkin HCCI combustion model. It was found that the current mechanism is capable of reproducing key characteristics from ozone addition, such as promoting ignition and controlling combustion phasing.
Acknowledgments The authors appreciate constructive discussion and advice from Dr. Ronald O. Grover Jr. of GM Global Research and Development Center. References [1] T. Alger, J. Gingrich, B. Mangold, C. Roberts, SAE Int. J. Engines 4 (1) (2011) 677–692.
[2] B. Wolk, A. DeFilippo, J.-Y. Chen, R. Dibble, A. Nishiyama, Y. Ikeda, Combust. Flame 160 (7) (2013) 1225–1234. [3] T. Kadono, K. Yoshida, H. Shoji, SAE 2002-321823, 2002. [4] T. Shiraishi, T. Urushihara, SAE 2011-01-0660, 2011. [5] T. Shiraishi, Advanced Ignition Systems for Gasoline Engines, IAV, Berlin, Germany, 2012. [6] M. Suess, M. Guenthner, M. Schenk, H-S. Rottengruber, Proc. IMechE, Part D: J. Automob. Eng. 226 (2012) 275–286. [7] L. Wu, J. Lane, N.P. Cernansky, D.L. Miller, A.A. Fridman, A.Yu. Starikovskiy, Proc. Combust. Inst. 33 (2011) 3219–3224. [8] I.N. Kosarev, N.L. Aleksandrov, S.V. Kindysheva, S.M. Starikovskaia, A.Y. Starikovskii, J. Phys. D: Appl. Phys. 41 (3) (2008). [9] N.L. Aleksandrov, S.V. Kindysheva, I.N. Kosarev, S.M. Starikovskaia, A.Y. Starikovskii, Proc. Combust. Inst. 32 (2009) 205–212. [10] D. Breden, L.L. Raja, C.A. Idicheria, P.M. Najt, S. Mahadevan, J. Appl. Phys. 114 (2013) 444. [11] H.J. Curran, P. Gauffuri, W.J. Pitz, C.K. Westbrook, Combust. Flame 129 (2002) 253–280. [12] M.J. McNenly, R. Whitesides, Proc. Combust. Inst. 35 (1) (2015) 581–587. [13] J.A. van Oijen, L.P.H. de Goey, Combust. Sci. Technol. 161 (1) (2000) 113–137. [14] F. Contino, H. Jeanmart, T. Lucchini, G. D’Errico, Proc. Comb. Inst. 33(2) 3057–3064. [15] F. Contino, J.-B. Masurier, F. Foucher, T. Lucchini, G. D’Errico, P. Dagaut, Fuel 137 2104 179–184. [16] T. Lu, C.K. Law, Combust. Flame 144 (2006) 24–36. [17] Y. Shi, H.-W. Ge, J.L. Brakora, R.D. Reitz, Energy Fuels 24 (2010) 1646–1654. [18] Y. Ra, R.D. Reitz, Combust. Flame 162 (2015) 3456–3481. [19] Y. Ra, R.D. Reitz, Combust. Flame 155 (2008) 713–738. [20] S. Keum, C.A. Idicheria, in: ASME 2013 Internal Combustion Engine Division Fall Technical Conference, 2, ICEF2013-19216, Dearborn, MI, USA, 2013. [21] Chemkin Pro: http://www.reactiondesign.com/ products/chemkin/chemkin-pro/. [22] H. Yamada, M. Yoshii, A. Tezake, Proc. Combust. Inst. 30 (2005) 2773–2780. [23] F. Foucher, P. Higelin, C. Mounaim-Rousselle, P. Dagaut, Proc. Combust. Inst. 34 (2013) 3005–3012. [24] F. Halter, P. Higelin, P. Dagaut, Energy Fuels 25 (2011) 2909–2916. [25] A. Mohammadi, H. Kawanabe, T. Ishiyama, M. Shioji, A. Komada, SAE 2006-01-0419, 2006. [26] M. Pinchak, T. Ombrello, C. Carter, E. Gutmark, V. Katta, Philos. Trans. R. Soc. A 373, 2014, 20140339. [27] S. Tiwari, P. Koch, G. Fadel, K. Deb, in: Genetic and Evolutionary Computation Conference (GECCO 2008), July 12–16, Atlanta, Georgia, USA, 2008. [28] J.-B. Masurier, F. Foucher, G. Dayma, P. Dagaut, Proc. Combust. Inst. 35 (2015) 3125–3132. [29] A. Schönborn, P. Hellier, A.E. Aliev, N. Ladommatos, Fuel 89 (2010) 3178–3184.
Please cite this article as: S. Keum et al., A skeletal iso-octane reaction mechanism for low temperature plasma ignition with ozone surrogate, Proceedings of the Combustion Institute (2016), http://dx.doi.org/10.1016/j.proci.2016.08.035