Fuel 172 (2016) 139–145

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Reduction and validation of a chemical kinetic mechanism including

necessity analysis and investigation of CH4/C3H8 oxidation at pressures
up to 120 bar using a rapid compression machine
Robert F. Pachler a,⇑, Ajoy K. Ramalingam b, K. Alexander Heufer b, Franz Winter a
a
Reaction Engineering and Combustion, Vienna University of Technology, Institute of Chemical Engineering, Getreidemarkt 9/166, 1060 Vienna, Austria
b
Physico-Chemical Fundamentals of Combustion (PCFC), RWTH University, Aachen, Schinkelstraße 8, 52062 Aachen, Germany

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.

1. Introduction advantage of a large gas engine with an extremely lean combustion

The rising demand for emission reduction, especially CO2 and
NOX will lead to a further decrease of emission limits e.g. EURO 6
or IMO Tier III. Therefore, the complexity of conventional internal
combustion engines will increase in order to meet these limits
[1]. In so-called ECA regions (Emission Control Areas), such as
coastal zones, limits for the sulfur content in the fuel and speed
related nitric oxide emissions are required. Especially in the mar-
itime or locomotive sector, fuel gases such as biogas are popular
to be used due to the content of shorter hydrocarbons compared
to heavy fuel oil (HFO) and the high knocking resistance. The
⇑ Corresponding author.
is the lower raw emission with at the same time similar power out-
puts and thermal efficiencies comparable to a diesel engine [2,3].
Further, increase in combustion efficiency by e.g. higher compres-
sion ratios, increased charge pressure stands in the foreground of
the developments, but is rather limited by knocking combustion
[4,5]. The self-ignition behavior of fuel gas is strongly dominated
by the mixture of its components like methane, propane as well
as substantially amounts of inert gases [6]. The conventional pro-
cedures for optimizing large gas engines are engine bench tests,
which are very resource consuming and expensive. Therefore, the
demand for detailed numerical methods is increasing. The most
common methods are 1D-modeling approaches to predict the load
exchange and the burn rate. They are implemented with empirical

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].

No C3H8 (mol%) lambda (–) CO2 (mol%) Tc range (K) Pc (bar)

2. Experimental
2.1. Rapid compression machine
A detailed description of the construction of rapid compression
machine (RCM) used in this study can be found in Lee et al. [14].
The facility utilizes a creviced piston head to reduce the effect of
roll up vortex by containing the boundary layer inside the crevice
volume [15,16]. The RCM used for this study facilitates compres-
sion times of the order of less than 20 ms. The end of compression
temperature can be varied using a moveable end wall which
changes the compression ratio. This increases the flexibility and
range of conditions achievable with one diluent gas. However, in
the current study the diluent gas was varied to achieve the condi-
tions in the test matrix. The reported experimental work on
methane conducted in other RCM facilities have been in the oper-
ating ranges of 16 [17], 10 and 25 bar [18] and some studies have
used the facility up to 80 bar. Table 1 shows a list of experimental
work carried out in shock tubes and RCM’s with relevance to the
current study. To our knowledge, this is the first experimental
work carried out for CH4/C3H8 mixtures at pressures of 80, 100
and 120 bar in a RCM.
In this study, the auto ignition delay measurements were car-
ried out for a defined test matrix which can be categorized into
two sections. The first section involves the study of CH4/C3H8 mix-
tures, at k = 1.9 for varying pressures of 80, 100 and 120 bar for a
temperature range of 825 –940 K. The second section comprises
of the study of the mixtures for varying lambda at an end of com-
pression pressure of 100 bar for a temperature range of 880–920 K.
Table 2 shows a comprehensive overview of the test matrix.
The test gases used in the present study were supplied by Prax-
air and Westfalen with a high level of purity [CH4, C3H8 – 99.99%
and for diluent gases Ar – 99.996%, N2 and O2 – 99.999%]. The mix-
ture is prepared in a stainless steel mixing vessel which is con-
trolled by two different pressure gauges ranging from 0 to
500 mbar (Co. STS, ATM. 1st) and 0 to 10 bar (Co. STS, ATM. 1st).
Both gauges show temperature compensated pressures with an
uncertainty of 0.05% of full scale for temperatures below 125 °C
ensuring good accuracy in the mixture composition. The initial
temperature is measured using a several T type thermocouples
mounted over the reaction chamber. The pressure in the reaction
chamber during compression and any post compression event
1
Computational Fluid Dynamics.
1 0/5/30 1.9 0 825–940 80/100/120
2 5 1.9/1.5 0 880–920 100
including ignition is measured using a dynamic pressure trans-
ducer (PCB 113B22). However, the non-reactive experiments were
measured using two pressure transducers Kistler 6125C-U20 and
PCB 113B22 in order to quantify the effect of heat shock [32].
The uncertainty of these pressure transducers are 0.07% of full
scale and ±0.04% of full scale respectively. As the end of compres-
sion temperature influences the chemistry and ignition delay time,
the uncertainties leading to its calculation is estimated. By taking
into account uncertainties in initial temperature, initial pressure,
change in air–fuel ratio and possible uncertainty in thermal data
of species; the combined uncertainty of the measurements leads
to a deviation of ±5 K in the end of compression temperature [33].
The Fig. 1 shows the pressure trace for the condition of 100 bar
end of compression, 850 K end of compression temperature with
5% propane at lean condition of k = 1.9. Target end of compression
temperature was achieved by varying the percentage of Argon in
the diluent gas. It is important to mention that pre-ignition was
observed in some of the conditions tested and were regarded as
failed experiments. Pre-ignition is characterized by a gradual
increase of pressure before main ignition resulting from a local
ignition and flame propagation in contrast to a homogeneous igni-
tion in the entire reactor.
The initial temperature Ti, initial pressure Pi along with the reac-
tant composition and the experimentally measured compressed
gas pressure Pc were used to calculate the end of compression tem-
perature Tc in Gaseq. It employs the adiabatic compression/expan-
sion routine and uses the temperature dependence of the ratio of
specific heats, p according to the equation

 Z Tc
Pc p dT
ln ¼
Pi Ti p1 T
while assuming frozen chemistry during compression.
3. Results and discussion
3.1. Rapid compression machine
Experiments on lean mixtures of methane were carried out at
80, 100 and 120 bar end of compression pressures Pc, for various
targeted end of compression temperatures Tc based on the test
 R.F. Pachler et al. / Fuel 172 (2016) 139–145 141

250 1000 K 952 K 909 K 870 K

End of compression Pressure, P =104.75 bar
c 80bar Experiment
End of compression Temperature, Tc=857 K 100bar Experiment
Fuel: 95 % CH4, 5 % C3H8; λ =1.9 120bar Experiment
200
Ar/N2 % = 0.4; Pi=2.5 bar; Ti=288.5 K 80bar Aramco Mech
2
Ignition Delay time, t =25.85 ms
ign
10 100bar Aramco Mech
120bar Aramco Mech
pressure [bar]

150

Ignition Delay time [ms]

End of compression
ignition delay time

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

time is observed compared to the case where the compression

stroke is not considered. Conversely, for longer ignition delay
times, the heat loss effect dominates the radical pool development
which leads to longer ignition delay times compared to the case
where the heat loss effect is not considered. Hence, it is empha-
sized that, while modeling the RCM behavior it is required to
include the facility behavior of radical pool development and heat 10
0

loss [34,35]. 1 1.05 1.1 1.15 1.2 1.25

Figs. 2–4 show the plot of the logarithm of the ignition delay
time against the inverse of compressed gas temperature for various
conditions on the test matrix. Figs. 2(a)–(c) show the comparison
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
delay times decrease as expected.
For simulating the ignition delay times the Aramco Mech 1.3
[36] was used. Former studies [37,38] compared the ignition delay
time, obtained from a zero-dimensional homogeneous ignition
process in a batch reactor of four published mechanisms under
high pressure conditions. Based on the investigations carried out
in Pachler et al. [39] and Robert et al. [40] Aramco Mech 1.3 was
chosen as the suitable mechanism for the simulation of ignition
delay time. It can be seen from the Fig. 2(a) that the simulation
results are well within the experimental uncertainty. The effect
of varying pressure and temperature are captured by the simula-
tion verifying the kinetic mechanism for describing the auto-
ignition of methane at these high pressures.
Figs. 2(b) and (c) show the experiment and model prediction for
the 5% and 30% propane addition again at pressures of 80, 100 and
120 bar. Also for these cases the comparison of the kinetic simula-
tion results and the experimental data shows a reasonable agree-
ment. In both cases the effect of varying pressure and
1000/T [1/K]
Fig. 2b. Experimental data (symbols) versus model predictions using Aramco Mech
1.3 (lines) at Pc = 80, 100 and 120 bar using 95% CH4 and 5% C3H8 at lambda 1.9.
temperature are captured well by the simulation. However, while
for the mixture with 30% propane simulation results lie again
within the uncertainty of the experimental data, for 5% propane
in the mixture the mechanism slightly over-predicts ignition delay
times. Detailed investigation an optimization of the detailed mech-
anism will be the topic of future studies.
From Fig. 3, it can be seen that as the propane percentage
increases ignition delay time becomes shorter. The key reason
behind the increase in reactivity is the influence of propane chem-
istry in this temperature regime. Especially for higher propane con-
centrations low temperature chain branching pathways present for
propane start dominating the overall reactivity of the mixture
leading to short ignition delay times. The mechanism also repro-
duces the effect of propane addition fairly well. Fig. 4 shows the
effect of varying lambda for a 95% CH4/5% C3H8 mixture at same
end of compression pressure of 100 bar. In general as the lambda
value decreases the reactivity increases. This is primarily because
of the dependence of concentration of fuel on reactivity. In the
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

Ignition Delay time [ms]

Ignition Delay time [ms]

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]

flow analysis, which was introduced by [12] and implemented in

mechanism reduction by [13]. This analysis can be used to find
the reactions and species most necessary for formation and con-
1
10 sumption of user-defined targets.
As described by [42], the necessity I of a species i is calculated
iteratively from

Ii ¼ maxðIj f a ; Ij ca ; Ii ; j ¼ 1; Ns ; a ¼ 1; Na Þ
i;j i;j

where the first value for Ii is calculated as

0 1
SSj;i
0
10
B C
0.9 0.95 1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 Ii;0 ¼ max @ ; Bi A
1000/T [1/K] maxðSSj;k Þ
k¼1;ns

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

over-predicts the ignition delay times. However, the effect of vary-

t¼0 R¼1 r R v 0
jR dt
ing lambda is captured well.
Overall, the comparison between the experimental results and Similarly, cai;j is the weighted consumption of atoms from spe-
simulations shows good prediction performance of the detailed cies i to species j;
mechanism. For this reason Aramco Mech 1.3 is used for model R t1 PNr 
reduction as described in the following section. The ignition delay t¼0 R¼1 r R ðtÞni
a
v 0iR v 00jR D1n a dt
cai;j ¼ R t1 PNr  R
time measurements for all RCM experiments in the test matrix
R¼1 r R v jR dt
0
along with initial pressure, compressed pressure and temperature t¼0

are provided as Supplementary Material.

This reduction scheme identifies unimportant or redundant
species that can be removed from the full mechanism until the glo-
4. Mechanism reduction bal error in ignition delay exceeds a user-specified limit. Also reac-
tions containing removed species have also been eliminated during
4.1. Methodology all steps of the reduction process.
Table 3 shows a comparison of the original mechanism with the
The mechanism reduction presented in this paper has been latest step of the reduced mechanism. The results in CPU time,
obtained using a variety of analysis methods, which are integrated speed-up and maximal error were generated using constant
 R.F. Pachler et al. / Fuel 172 (2016) 139–145 143

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.

AramcoMech 70species, p=80bar, Lambda=1.9 modeling error [%]

10

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.)

AramcoMech 70species, p=120bar, Lambda=1.9 modeling error [%]

10

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-
ing 825–1400 K and 80–120 bar. The air fuel ratio k was fixed at
1.9.
4.2. Mechanism reduction
As presented in the previous chapter (Section 3.1) the Aramco
Mech 1.3 is capable of describing ignition delay times at high pres-
sure condition. Based on this, a model reduction has been per-
formed in order to reduce the computational effort. Simulations
for model reduction procedure have been performed using a
zero-dimensional homogeneous ignition in an adiabatic batch
reactor. The obtained ignition delay times of the detailed and the
reduced mechanisms were compared to each other, whereby in
all steps of the reduction process the detailed mechanism served
as a reference. As mentioned before, species and reactions have
1429 K 1333 K 1250 K 1176 K 1111 K 1053 K 1000 K 952 K 909 K 869 K 833 K
10
2
80bar Aramco Mech full
100bar Aramco Mech full
120bar Aramco Mech full
80bar Aramco Mech reduced
100bar Aramco Mech reduced
120bar Aramco Mech reduced
Ignition Delay time [ms]
been removed without inducing significant errors in ignition delay
times using necessity analysis. A reduction from 253 species and
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-
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
propane enrichment up to 30 mol% (molar percentage). The color
scheme used for every single data point represents the error of
deviation of the reduced to the detailed mechanism, whereby a
positive error signifies an increased ignition delay time and vice
versa.
It can be seen that the modeling error at higher amounts of pro-
pane and elevated initial temperatures increases. The reduced
mechanism shows an enhanced ignition behavior at these condi-
tions, which results in up to 9.7% shorter ignition delay time. This
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
that no significant deviations occur between the detailed and
reduced models when varying fuel blend and pressure over a wide
range of initial temperatures.
5. Conclusion

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.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, in

−2
10 the online version, at http://dx.doi.org/10.1016/j.fuel.2015.12.044.
0.7 0.75 0.8 0.85 0.9 0.95 1 1.05 1.1 1.15 1.2
1000/T [1/K]
References
Fig. 7b. Auto ignition validation of reduced Aramco mechanism over a wide range
of initial temperatures and pressures and at varying fuel blends for 70% CH4 and [1] International Maritime Organization. Annex VI of MARPOL 73/74 and NOx
30% C3H8. technical Code, London; 1998.
 R.F. Pachler et al. / Fuel 172 (2016) 139–145
[2] Murakami S et al. Current status and future strategies of gas engine
development. In: Congress: CIMAC, Shanghai; 2013.
[3] Nerheim L. The position of the gas engine in competition with diesel engine
and gas turbine. In: Conference. 6. dessau gas engine conference, Dessau-
Roßlau; 2009.
[4] Holly W et al. Modelling the knocking combustion of a large gas engine
considering cyclic variations and detailed reaction kinetics. SAE Int 2014;13.
August.
[5] Holly W et al. Prediction of knocking combustion and NOx-formation for fuel
gases with different methane numbers. In: Engine combustion processes –
current problems and modern techniques, Ludwigsburg; 2015.
[6] Herdin G. Motorische Maßnahmen zur weiteren Verbesserung des
Konvertierungswirkungsgrades von Biomasse. In: 12. Conference ‘‘The
Working Process of the Internal Combustion Engine”, Graz; 2009.
[7] Holly W et al. Numerical optimisation approach for a large gas engine
considering different fuel gas qualities. In: 15th conference ‘‘The Working
Process of the Internal Combustion Engine”, Graz; 2015.
[8] Wimmer A et al. LEC-GPN – a new index for assessing the knock behavior of
gaseous fuels for large engines. In: Knocking in Gasoline Engines, Berlin; 2013.
[9] Holly WE, Lauer T, Pachler RF, Winter F, Murakami S. Impact of the fuel gas
quality on the efficiency of a large gas engine. In: Proc of 28th CIMAC congress
(int council on combustion engines), Helsinki, Finland; 2016.
[10] Turanyi T. Reduction of large reaction mechanisms. New J Chem
1990:795–803.
[11] Rabitz H et al. Sensitivity analysis in chemical kinetica. Annu Rev Phys Chem
1983:419–61.
[12] Revel J et al. Derivation of a global chemical kinetic mechanism for methane
ignition and combustion. J Chim Phys Phys-Chim Biol 1994:365–82.
[13] Androulakis IP et al. Time-integrated pointers dor enabling the analysis of
detailed reaction mechanisms. AIChE J 2004:2956–70.
[14] Lee C et al. Rapid compression machine and detailed modeling study. Z Phys
Chem 2012:1–27.
[15] Mittal G et al. Aerodynamics inside a rapid compression machine. Combust
Flame 2006;145:160–80.
[16] Wuermel J et al. CFD studies of a twin piston rapid compression machine.
Combust Flame 2005;30:417–30.
[17] Brett L et al. Simulation of methane autoignition in a rapid compression
machine with creviced piston. Combust Flame 2001;124:326–9.
[18] Burke U et al. An ignition delay and kinetic modeling study of methane,
dimethyl ether, and their mixtures at high pressures. Combust Flame
2015;162(2):315–30.
[19] de Vries J et al. Autoignition of methane-based fuel blends under gas turbine
conditions. Proc Combust Inst 2007;31:3163–71.
[20] Gallagher SM et al. A rapid compression machine study of the oxidation of
propane in the negative temperature coefficient regime. Combust Flame
2008;153:316–33.
[21] Heyne S et al. Development of a natural gas reaction mechanism for engine
simulations based on rapid compression machine experiments using a multi-
objective optimisation strategy. Fuel 2008;87:3046–54.
[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.
145
[23] Healy D et al. Methane/ethane/propane mixture oxidation at high pressures
and at high, intermediate and low temperatures. Combust Flame
2008;155:441–8.
[24] Healy D et al. Methane/propane mixture oxidation at high pressures and at
high, intermediate and low temperatures. Combust Flame 2008;155:451–61.
[25] Petersen EL et al. Methane/propane oxidation at high pressures: experimental
and detailed chemical kinetic modeling. Proc Combust Inst 2007;31:447–54.
[26] Healy D et al. Oxidation of C1–C5 alkane quinternary natural gas mixtures at
high pressures. Energy Fuels 2010;24:1521–8.
[27] Spadaccini LJ et al. Ignition delay characteristics of methane fuels. Prog Energy
Combust Sci 1994;20(5):431–60.
[28] Huang J et al. Shock-tube study of methane ignition under engine relevant
conditions: experiments and modeling. Combust Flame 2004;136:25–42.
[29] Healy D et al. Methane/n-butane ignition delay measurements at high
pressure and detailed chemical kinetic simulations. Energy Fuels
2010:1617–27. 4 February.
[30] Gersen S et al. The effects of CO addition on the autoignition of H2, CH4, and
CH4/H2 fuels at high pressure in an RCM. Combust Flame 2012;159:3472–5.
[31] Gersen S et al. Ignition properties of methane/hydrogen mixtures in a rapid
compression machine. Hydrogen Energy 2008;33:1957–64.
[32] Donohoe N et al. Influence of steam dilution on the ignition of hydrogen,
syngas and natural gas blends at elevated pressures. Combust Flame
2015;162:1126–35.
[33] Vranckx S et al. A rapid compression machine study of the low temperature
combustion of cyclohexane at elevated pressures. Proc Combust Inst
2013;34:377–84.
[34] Sung CJ et al. Using rapid compression machines for chemical kinetics studies.
Progr Energy Combust Sci 2014;44:1–18.
[35] Heufer KA et al. A comparisong of longer alkane and alcohol ignition including
new experimental results for n-pentanol and n-hexanol. Proc Combust Inst
2012.
[36] Metcalfe WK et al. A hierarchical and comparative kinetic modeling study of
C1–C2 hydrocarbon and oxygenated fuels. Int J Chem Kinet 2013;45
(10):638–75.
[37] Bardolf R et al. Comparison of chemical kinetic mechanisms for combustion
simulation of treated biomass. Holistic Approach Environ 2014;4:65–9.
[38] Bardolf R et al. Comparison of chemical kinetic mechanisms for lean methane/
propane mixture ignition at high pressures. In: Proc IV annual meeting of COST
action CM0901, Perugia; 2013.
[39] Pachler RF et al. Lean methane/propane ignition at high pressures: comparison
of chemical kinetic mechanisms. In: 35th International symposium on
combustion, San Francisco, California; 2014.
[40] Robert P et al. Investigations of available experimental and modeling data on
the oxidative coupling and partial oxidation of methane. In: Proceedings of the
European combustion meeting; 2015.
[41] LOGE AB – Lund Combustion Engineering. Available: http://www.loge.se/
Products/LOGEsoft.html.
[42] LOGE-AB. Mechanism reduction. In: Scientific software for modelling of
chemical kinetic systems; 2014. p. 8–12.