Professional Documents
Culture Documents
1 s2.0 S0016236115013058 Main
1 s2.0 S0016236115013058 Main
1 s2.0 S0016236115013058 Main
Fuel
journal homepage: www.elsevier.com/locate/fuel
a r t i c l e i n f o a b s t r a c t
Article history: Cost efficiency and the ecological footprint are becoming more important in the locomotive and maritime
Received 25 November 2015 business. To overcome these needs, manufacturers are forced to develop highly efficient internal combus-
Received in revised form 16 December 2015 tion engines suitable for alternative fuel, such as biogas. In the development process of a new engine, sim-
Accepted 17 December 2015
ulation tools offer economic benefits compared to engine bench tests. To manage the trade-off between
Available online 28 December 2015
enhanced efficiency by increasing the compression ratio and knocking combustion, a reaction mechanism
is necessary to capture the ignition behavior of various fuel blends and conditions during the simulation.
Keywords:
Since previous studies only investigated methane/propane oxidation at pressures up to 50 bar, experi-
Biogas
Knocking combustion
mental investigations have been conducted to include pressure regimes, which are similar to those of
RCM a knocking combustion. Therefore, rapid compression machine (RCM) experiments were performed using
High pressure methane/propane mixtures at lean conditions with air-to-fuel ratios of 1.5, 1.7 and 1.9. Furthermore a
Experiments range of compression temperatures from 800 to 1000 K at pressures of 80, 100 and 120 bar were con-
Modeling ducted. The experiments for mixtures containing 30 mol% propane showed a negative temperature coef-
ficient (NTC) behavior, which is consistent with experiments for other alkane fuels presented in the
literature. In addition, the experimental results were compared with the recently published detailed
chemical mechanism Aramco Mech 1.3. This mechanism was used to simulate the ignition delay times
with consideration of the facility effects of the rapid compression machine. The simulation results, based
on a zero-dimensional homogeneous batch reactor showed good agreement with the measurements over
a wide range of the investigated conditions. In a further step the detailed mechanism has been reduced by
necessity analysis in order to minimize computational efforts for future combined CFD (Computational
Fluid Dynamics) and kinetic simulations. The results of this novel work show the first set of high pressure
screening experiments in a RCM for CH4/C3H8 mixtures combined with a systematic method for mecha-
nism reduction.
Ó 2015 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.fuel.2015.12.044
0016-2361/Ó 2015 Elsevier Ltd. All rights reserved.
140 R.F. Pachler et al. / Fuel 172 (2016) 139–145
knock models which are based on fuel properties such as the Table 1
octane- or methane number in order to predict the knock limit List of specific literature.
[4,7]. To get deeper insight of knocking combustion, there is still No Condition P (bar) ɸ Facility Ref.
a need for more detailed approaches to predict the knock behavior 1 C1–C5 natural gas based mixtures 20 0.5 ST [19]
of different gas mixtures at lean conditions [8]. 2 C3H8 mixtures 21, 27, 37 0.5, 1.0, 2.0 RCM [20]
In this paper, the oxidation of methane–propane blends at very 3 CH4/C2H6/C3H8 mixtures 13–21 0.625, 1.0 RCM [21]
high pressures of up to 120 bar in a rapid compression machine 4 CH4/C2H6 and CH4/C3H8 16, 40 1 ST [22]
5 CH4/C2H6/C3H8 mixtures 1–50 0.5, 1.0, 2.0 ST/RCM [23]
was studied. While previous research [Table 1] investigated CH4/ 6 CH4/C3H8 mixtures 10, 20, 30 0.3–3.0 ST/RCM [24]
C3H8 ignition at pressures only up to 50 bar, the range in this work 7 CH4/C3H8 mixtures 5.3–31.4 0.5–3.0 ST [25]
included pressure regimes that are similar to those of a knocking 8 C1–C5 alkane mixtures 8–30 0.5, 1.0, 2.0 ST/RCM [26]
combustion. Further, the generated data was used to compare with 9 C1–C4 mixtures 3–15 0.45–1.25 ST [27]
10 CH4 air mixtures 16–40 0.7–1.3 ST [28]
a detailed kinetic model.
11 CH4/n-C4H10 mixtures 10–30 0.32–2.31 ST/RCM [29]
To investigate and understand knocking combustion inside the 12 H2/CH4/CO mixtures 20–80 0.5 RCM [30]
engine cylinder, a transient CFD1-analysis has to be made, where for 13 CH4/H2 mixtures 15–70 1 RCM [31]
each time step the chemistry, the heat transfer and the turbulent
ST – Shock Tube; RCM – Rapid Compression Machine.
mixing process has to be solved [9]. To reduce the computational
complexity while at the same time retain good accuracy in predict-
ing the ignition behavior, established methods such as sensitivity
and flow analysis [10–13] have been used in this paper to reduce a Table 2
Overview of the test matrix [Pc – end of compression pressure; Tc – end of
detailed chemical mechanism.
compression temperature].
150
100
1
50 10
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
t [sec]
Fig. 1. RCM pressure trace for targeted condition of 100 bar Pc and 850 K Tc with 5% 0
10
propane at lean condition of 1.9 lambda. 1 1.05 1.1 1.15
1000/T [1/K]
matrix. The experiments also include mixtures with 5 and 30 per- Fig. 2a. Ignition delay times for 100% CH4 and Lambda 1.9: Experimental data
centage by molar composition of propane at the same targeted (symbols) versus model predictions of Aramco Mech 1.3 (lines) at Pc = 80, 100 and
120 bar.
temperatures. For modeling an RCM experiment a non-reactive
pressure trace is typically required. This can be experimentally
obtained by replacing the content of oxygen with nitrogen in the 1000 K 952 K 909 K 870 K 833 K 800 K
mixture. The non-reactive pressure trace captures the rapid rise 10
2
80bar Experiment
in pressure during the compression stroke and the gradual 100bar Experiment
120bar Experiment
decrease in pressure due to heat loss. A RCM experiment can be 80bar Aramco Mech
represented by an ideal constant volume auto ignition perturbed 100bar Aramco Mech
120bar Aramco Mech
by the associated facility effect due to the compression stroke
Ignition Delay time [ms]
and the heat loss from the core region to the walls of the reaction
chamber [34]. For shorter ignition delay times, the radical pool
effect during the compression stroke overshadows the effect of
1
heat loss after end of compression, thereby a shorter ignition delay 10
Figs. 2–4 show the plot of the logarithm of the ignition delay 1000/T [1/K]
time against the inverse of compressed gas temperature for various Fig. 2b. Experimental data (symbols) versus model predictions using Aramco Mech
conditions on the test matrix. Figs. 2(a)–(c) show the comparison 1.3 (lines) at Pc = 80, 100 and 120 bar using 95% CH4 and 5% C3H8 at lambda 1.9.
of experimental results versus model predictions for 100% CH4,
95% CH4/5% C3H8 and 70% CH4/30% C3H8 mixtures. It is evident
from the figures that as pressure and temperature increase ignition temperature are captured well by the simulation. However, while
delay times decrease as expected. for the mixture with 30% propane simulation results lie again
For simulating the ignition delay times the Aramco Mech 1.3 within the uncertainty of the experimental data, for 5% propane
[36] was used. Former studies [37,38] compared the ignition delay in the mixture the mechanism slightly over-predicts ignition delay
time, obtained from a zero-dimensional homogeneous ignition times. Detailed investigation an optimization of the detailed mech-
process in a batch reactor of four published mechanisms under anism will be the topic of future studies.
high pressure conditions. Based on the investigations carried out From Fig. 3, it can be seen that as the propane percentage
in Pachler et al. [39] and Robert et al. [40] Aramco Mech 1.3 was increases ignition delay time becomes shorter. The key reason
chosen as the suitable mechanism for the simulation of ignition behind the increase in reactivity is the influence of propane chem-
delay time. It can be seen from the Fig. 2(a) that the simulation istry in this temperature regime. Especially for higher propane con-
results are well within the experimental uncertainty. The effect centrations low temperature chain branching pathways present for
of varying pressure and temperature are captured by the simula- propane start dominating the overall reactivity of the mixture
tion verifying the kinetic mechanism for describing the auto- leading to short ignition delay times. The mechanism also repro-
ignition of methane at these high pressures. duces the effect of propane addition fairly well. Fig. 4 shows the
Figs. 2(b) and (c) show the experiment and model prediction for effect of varying lambda for a 95% CH4/5% C3H8 mixture at same
the 5% and 30% propane addition again at pressures of 80, 100 and end of compression pressure of 100 bar. In general as the lambda
120 bar. Also for these cases the comparison of the kinetic simula- value decreases the reactivity increases. This is primarily because
tion results and the experimental data shows a reasonable agree- of the dependence of concentration of fuel on reactivity. In the
ment. In both cases the effect of varying pressure and low and intermediate temperature regimes the fuel is oxidized
142 R.F. Pachler et al. / Fuel 172 (2016) 139–145
1000 K 952 K 909 K 870 K 833 K 800 K 769 K 741 K 714 K 1000 K 952 K 909 K 870 K 833 K 800 K
2 2
10 80bar Experiment 10 λ=1.5 Experiment
100bar Experiment λ=1.9 Experiment
120bar Experiment λ=1.5 Aramco Mech
80bar Aramco Mech λ=1.9 Aramco Mech
100bar Aramco Mech
120bar Aramco Mech
1 1
10 10
0
0 10
10 1 1.05 1.1 1.15 1.2 1.25
1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4
1000/T [1/K] 1000/T [1/K]
Fig. 2c. Experimental data (symbols) versus model predictions using Aramco Mech Fig. 4. Effect of varying lambda; experimental data from RCM versus model
1.3 (lines) at Pc = 80, 100 and 120 bar using 70% CH4 and 30% C3H8 at lambda 1.9. predictions using Aramco Mech 1.3. for 95% CH4 and 5% C3H8 at varying lambda and
100 bar pc.
1111 K 1053 K 1000 K 952 K 909 K 870 K 833 K 800 K 769 K 741 K
2
10 100% CH4 Experiment
5% C H Experiment
3 8
into the chemical kinetic simulation software, LOGESOFT [41]. The
30% C3H8 Experiment reduction approach proposed in the current work is based on spe-
100% CH4 Aramco Mech
cies removal with the necessity analysis method. This approach
5% C H Aramco Mech
3 8
30% C3H8 Aramco Mech uses sensitivity analysis proposed by [10,11] in combination with
Ignition Delay time [ms]
Fig. 3. Effect of propane addition; experimental data from RCM versus model
predictions with Aramco Mech 1.3 for Ignition delay time at 100 bar Pc. Ns and Ns are the total number of species and atoms respectively, SSj;i
contains the information how sensitive species j is to species i, k
and then pyrolysis on the products happens and this process is pri-
denotes a species and Bi is 1 or 0 depending on if i is set as a neces-
marily dependent on the concentration of the fuel [24]. In this a
sary species or not. f i;j is the weighted formation of species i from
study, lambda values of 1.5 and 1.9 have been investigated. As
expected the ignition delay times decrease with decreasing lambda species j by atoms from species j to species i;
in simulation and experiment. As discussed before, for this partic- R t1 PNr
ular fuel mixture containing 5% propane the simulation slightly a t¼0 R¼1 r R ðtÞni
a
v 0jR v 00iR D1n a dt
f i;j ¼ R t1 PNr R
Table 3
Comparison of the detailed and reduced mechanisms.
Mechanism # Species # Reactions Max. error (%) CPU timea (s) Speed-up factor (in %)
Aramco Mech 1.3 253 1542 – 95.7 –
Reduced Aramco Mech 1.3 70 352 +4.03/9.72 28.2 3.4/70.5
a
CPU time of methane/propane combustion simulations including a parameter variation with 72 cases in a constant volume reactor, k = 1.9, P = 80–120 bar, T = 825–
1400 K.
20
6
ignition delay time [ms]
15 4
2
10
5
−2
0 −4
0
−6
10
1.2 −8
1.1
20 1
X [%] 0.9
CH 0.8 −10
3 8 30 0.7 1000K/T
Fig. 5. Visualizing the effect of varying propane addition and temperature variation; Simulation data from detailed versus reduced Aramco Mech 1.3 at 80 bar and lambda 1.9.
The color bar represents the error of deviation. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
20
6
ignition delay time [ms]
15 4
2
10
5
−2
0 −4
0
−6
10
1.2 −8
1.1
20 1
0.9
0.8 −10
X [%] 30 0.7
CH 1000K/T
3 8
Fig. 6. Visualizing the effect of varying propane addition and temperature variation; Simulation data from detailed versus reduced Aramco Mech 1.3 at 120 bar and lambda
1.9. The color bar represents the error of deviation. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
144 R.F. Pachler et al. / Fuel 172 (2016) 139–145
volume auto ignition data, sampled from initial conditions cover- been removed without inducing significant errors in ignition delay
ing 825–1400 K and 80–120 bar. The air fuel ratio k was fixed at times using necessity analysis. A reduction from 253 species and
1.9. 1542 reactions to 70 species and 352 reactions could be obtained
using this methodology, with a maximum error deviation of 9.6%.
Figs. 5 and 6 show the simulation data using the reduced Ara-
4.2. Mechanism reduction mcoMech at 80 and 120 bar and a lambda value of 1.9. The figures
demonstrate the variation of temperature from 825 to 1400 K and
As presented in the previous chapter (Section 3.1) the Aramco propane enrichment up to 30 mol% (molar percentage). The color
Mech 1.3 is capable of describing ignition delay times at high pres- scheme used for every single data point represents the error of
sure condition. Based on this, a model reduction has been per- deviation of the reduced to the detailed mechanism, whereby a
formed in order to reduce the computational effort. Simulations positive error signifies an increased ignition delay time and vice
for model reduction procedure have been performed using a versa.
zero-dimensional homogeneous ignition in an adiabatic batch It can be seen that the modeling error at higher amounts of pro-
reactor. The obtained ignition delay times of the detailed and the pane and elevated initial temperatures increases. The reduced
reduced mechanisms were compared to each other, whereby in mechanism shows an enhanced ignition behavior at these condi-
all steps of the reduction process the detailed mechanism served tions, which results in up to 9.7% shorter ignition delay time. This
as a reference. As mentioned before, species and reactions have behavior can be seen both at 80 and 120 bar initial pressure. In the
remaining areas the reduced mechanism performed very satisfying
within ±4% of modeling error. It can be seen also in Figs. 7a and b
1429 K 1333 K 1250 K 1176 K 1111 K 1053 K 1000 K 952 K 909 K 869 K 833 K that no significant deviations occur between the detailed and
10
2
80bar Aramco Mech full
reduced models when varying fuel blend and pressure over a wide
100bar Aramco Mech full range of initial temperatures.
120bar Aramco Mech full
80bar Aramco Mech reduced
100bar Aramco Mech reduced
120bar Aramco Mech reduced 5. Conclusion
Ignition Delay time [ms]
1
10 The oxidation of methane/propane mixtures has been studied in
the temperature range of 825–940 K, at gas pressures of 80–
120 bar, and at air-to-fuel ratios varying from 1.5 to 1.9 in a rapid
compression machine. The presented experimental data extend the
pressure window of observation of previous published studies to
10
0
exceedingly high pressures. Additionally, these data were simu-
lated using previously published chemical kinetic models includ-
ing facility effects. It was found that the model reproduces the
effect of change in pressure and lambda, predicting that propane-
rich, high-pressure mixtures ignite fastest, while pure methane,
10
−1 low-pressure mixtures ignite slowest. Furthermore, a change of
0.7 0.75 0.8 0.85 0.9 0.95 1 1.05 1.1 1.15 1.2 reactivity from high to low temperatures could be predicted with
1000/T [1/K] the model and measured in experiments for mixtures containing
Fig. 7a. Auto ignition validation of reduced Aramco Mech 1.3 mechanism over a
30 mol% propane. Quantitatively the kinetic model was found to
wide range of initial temperatures and pressures and at varying fuel blends for 100% be in good agreement with the experimental data from the RCM
CH4. facility. In a second step a reduction of this chemical kinetic model
has been performed using necessity analysis. The detailed mecha-
nism could be reduced by more than 70 percent within a wide
1429 K 1333 K 1250 K 1176 K 1111 K 1053 K 1000 K 952 K 909 K 869 K 833 K range of fuel, temperature and pressure variation, with excellent
1
10 80bar Aramco Mech full
agreement. The reduced mechanism showed the maximum devia-
100bar Aramco Mech full tion at increased propane enriched mixtures, higher than 20 mol%
120bar Aramco Mech full
80bar Aramco Mech reduced
and elevated initial temperatures.
100bar Aramco Mech reduced
120bar Aramco Mech reduced
Acknowledgements
Ignition Delay time [ms]
0
10
The authors would like to gratefully thank the Austrian
Bundesministerium für Verkehr, Innovation und Technologie
(BMVIT) and the Austrian Forschungs-Förderungsgesellschaft
(FFG) in the framework of the A3plus-program ‘‘Bioflame”, FFG
−1
10 835806 for funding this project.
[2] Murakami S et al. Current status and future strategies of gas engine [23] Healy D et al. Methane/ethane/propane mixture oxidation at high pressures
development. In: Congress: CIMAC, Shanghai; 2013. and at high, intermediate and low temperatures. Combust Flame
[3] Nerheim L. The position of the gas engine in competition with diesel engine 2008;155:441–8.
and gas turbine. In: Conference. 6. dessau gas engine conference, Dessau- [24] Healy D et al. Methane/propane mixture oxidation at high pressures and at
Roßlau; 2009. high, intermediate and low temperatures. Combust Flame 2008;155:451–61.
[4] Holly W et al. Modelling the knocking combustion of a large gas engine [25] Petersen EL et al. Methane/propane oxidation at high pressures: experimental
considering cyclic variations and detailed reaction kinetics. SAE Int 2014;13. and detailed chemical kinetic modeling. Proc Combust Inst 2007;31:447–54.
August. [26] Healy D et al. Oxidation of C1–C5 alkane quinternary natural gas mixtures at
[5] Holly W et al. Prediction of knocking combustion and NOx-formation for fuel high pressures. Energy Fuels 2010;24:1521–8.
gases with different methane numbers. In: Engine combustion processes – [27] Spadaccini LJ et al. Ignition delay characteristics of methane fuels. Prog Energy
current problems and modern techniques, Ludwigsburg; 2015. Combust Sci 1994;20(5):431–60.
[6] Herdin G. Motorische Maßnahmen zur weiteren Verbesserung des [28] Huang J et al. Shock-tube study of methane ignition under engine relevant
Konvertierungswirkungsgrades von Biomasse. In: 12. Conference ‘‘The conditions: experiments and modeling. Combust Flame 2004;136:25–42.
Working Process of the Internal Combustion Engine”, Graz; 2009. [29] Healy D et al. Methane/n-butane ignition delay measurements at high
[7] Holly W et al. Numerical optimisation approach for a large gas engine pressure and detailed chemical kinetic simulations. Energy Fuels
considering different fuel gas qualities. In: 15th conference ‘‘The Working 2010:1617–27. 4 February.
Process of the Internal Combustion Engine”, Graz; 2015. [30] Gersen S et al. The effects of CO addition on the autoignition of H2, CH4, and
[8] Wimmer A et al. LEC-GPN – a new index for assessing the knock behavior of CH4/H2 fuels at high pressure in an RCM. Combust Flame 2012;159:3472–5.
gaseous fuels for large engines. In: Knocking in Gasoline Engines, Berlin; 2013. [31] Gersen S et al. Ignition properties of methane/hydrogen mixtures in a rapid
[9] Holly WE, Lauer T, Pachler RF, Winter F, Murakami S. Impact of the fuel gas compression machine. Hydrogen Energy 2008;33:1957–64.
quality on the efficiency of a large gas engine. In: Proc of 28th CIMAC congress [32] Donohoe N et al. Influence of steam dilution on the ignition of hydrogen,
(int council on combustion engines), Helsinki, Finland; 2016. syngas and natural gas blends at elevated pressures. Combust Flame
[10] Turanyi T. Reduction of large reaction mechanisms. New J Chem 2015;162:1126–35.
1990:795–803. [33] Vranckx S et al. A rapid compression machine study of the low temperature
[11] Rabitz H et al. Sensitivity analysis in chemical kinetica. Annu Rev Phys Chem combustion of cyclohexane at elevated pressures. Proc Combust Inst
1983:419–61. 2013;34:377–84.
[12] Revel J et al. Derivation of a global chemical kinetic mechanism for methane [34] Sung CJ et al. Using rapid compression machines for chemical kinetics studies.
ignition and combustion. J Chim Phys Phys-Chim Biol 1994:365–82. Progr Energy Combust Sci 2014;44:1–18.
[13] Androulakis IP et al. Time-integrated pointers dor enabling the analysis of [35] Heufer KA et al. A comparisong of longer alkane and alcohol ignition including
detailed reaction mechanisms. AIChE J 2004:2956–70. new experimental results for n-pentanol and n-hexanol. Proc Combust Inst
[14] Lee C et al. Rapid compression machine and detailed modeling study. Z Phys 2012.
Chem 2012:1–27. [36] Metcalfe WK et al. A hierarchical and comparative kinetic modeling study of
[15] Mittal G et al. Aerodynamics inside a rapid compression machine. Combust C1–C2 hydrocarbon and oxygenated fuels. Int J Chem Kinet 2013;45
Flame 2006;145:160–80. (10):638–75.
[16] Wuermel J et al. CFD studies of a twin piston rapid compression machine. [37] Bardolf R et al. Comparison of chemical kinetic mechanisms for combustion
Combust Flame 2005;30:417–30. simulation of treated biomass. Holistic Approach Environ 2014;4:65–9.
[17] Brett L et al. Simulation of methane autoignition in a rapid compression [38] Bardolf R et al. Comparison of chemical kinetic mechanisms for lean methane/
machine with creviced piston. Combust Flame 2001;124:326–9. propane mixture ignition at high pressures. In: Proc IV annual meeting of COST
[18] Burke U et al. An ignition delay and kinetic modeling study of methane, action CM0901, Perugia; 2013.
dimethyl ether, and their mixtures at high pressures. Combust Flame [39] Pachler RF et al. Lean methane/propane ignition at high pressures: comparison
2015;162(2):315–30. of chemical kinetic mechanisms. In: 35th International symposium on
[19] de Vries J et al. Autoignition of methane-based fuel blends under gas turbine combustion, San Francisco, California; 2014.
conditions. Proc Combust Inst 2007;31:3163–71. [40] Robert P et al. Investigations of available experimental and modeling data on
[20] Gallagher SM et al. A rapid compression machine study of the oxidation of the oxidative coupling and partial oxidation of methane. In: Proceedings of the
propane in the negative temperature coefficient regime. Combust Flame European combustion meeting; 2015.
2008;153:316–33. [41] LOGE AB – Lund Combustion Engineering. Available: http://www.loge.se/
[21] Heyne S et al. Development of a natural gas reaction mechanism for engine Products/LOGEsoft.html.
simulations based on rapid compression machine experiments using a multi- [42] LOGE-AB. Mechanism reduction. In: Scientific software for modelling of
objective optimisation strategy. Fuel 2008;87:3046–54. chemical kinetic systems; 2014. p. 8–12.
[22] Huang J et al. Experimental and kinetic study of autoignition in methane/
ethane/air and methane/propane/air mixtures under engine-relevant
conditions. Combust Flame 2006;144:74–88.