See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/285588860

COMPARISON OF DIFFERENT CH4-AIR COMBUSTION MECHANISMS IN A

PERFECTLY STIRRED REACTOR WITH OSCILLATING RESIDENCE TIMES CLOSE
TO EXTINCTION

Conference Paper · June 2015

CITATIONS READS
0 570

3 authors:

Luigi Acampora Francesco Saverio Marra

ENEA Italian National Research Council
21 PUBLICATIONS   70 CITATIONS    88 PUBLICATIONS   529 CITATIONS   

SEE PROFILE SEE PROFILE

Emanuele Martelli
Università della Campania "Luigi Vanvitelli"
63 PUBLICATIONS   807 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Numerical strategies for the bifurcation analysis of perfectly stirred reactors with detailed combustion mechanisms View project

Fundamentals of solid fuel combustion and gasification View project

All content following this page was uploaded by Francesco Saverio Marra on 03 December 2015.

The user has requested enhancement of the downloaded file.

MCS 9 Rhodes, Greece, June 7-11, 2015

COMPARISON OF DIFFERENT CH4-AIR COMBUSTION

MECHANISMS IN A PERFECTLY STIRRED REACTOR
WITH OSCILLATING RESIDENCE TIMES CLOSE TO
EXTINCTION
L. Acampora*, F.S. Marra** and E. Martelli***
marra@irc.cnr.it
*Dipartimento di Ingegneria - Università degli Studi del Sannio, Benevento, Italy
**Istituto di Ricerche sulla Combustione - CNR, Napoli, Italy
***Department of Industrial and Information Engineering - Seconda Università degli
Studi di Napoli, Aversa, Italy

Abstract
Transient simulations of the methane-air mixture combustion in a Perfectly Stirred Reac-
tor (PSR) with global one-step and two-step, reduced (by directed relation graph method)
and detailed chemical reaction mechanisms, have been conducted. The simulation condi-
tions were carefully selected to observe the dynamic of a PSR with an oscillating residence
time close to the extinction. The effect of the oscillation amplitude and frequency have
been investigated over a wide interval of values. The temperature response of the global
and reduced mechanism has been compared to that of the detailed reaction mechanisms,
the GRI Mech 3.0. The discussion of results points out limits and problems associated
with the adoption of global reaction mechanisms. The results of the comparison indicate
that the adoption of a global mechanism is not adequate in order to obtain an accurate
description of methane-air combustion in a PSR with an oscillating residence time. This
comparison underlines that only a little benefit is added by the adoption of the selected
two-step mechanism. Instead the reduced mechanism shows an excellent agreement with
the detailed reaction mechanism.

Introduction
Reducing combustion emission and improving fuel consumption efficiency are and will be
an important topic for research and industry. Lean combustion in internal combustion
engines and gas turbine appears an obliged route to reduce the pollutant and greenhouse
gas emissions [1]. However, lean combustion, in particular lean premixed or partially pre-
mixed combustion, is prone to combustion instability [2]. Therefore, the understanding
of combustion processes and of instability generation is fundamental. This consideration
promoted many studies. Both theoretical and experimental studies highlighted the role of
different mechanisms, like unmixedness of reactants [3], flame dynamics response to per-
turbations [4, 5, 6, 7, 8, 9, 10] and thermo-acoustic instabilities [2]. The chemical kinetics
response to imposed oscillations has been clearly recognized for the proper estimation of
pollutant emissions [11].
Nevertheless the need of a reduction of the chemical scheme is doubtless required when
CFD simulations of reactive flows is pursued. Except for the light hydrocarbons such
as hydrogen and methane, or for simple combustion models such as 1-D laminar flame,
detailed mechanisms can not be adopted [12]. Indeed, computational cost is not affordable
in applications involving complex geometries and/or flows. A substantial reduction of the
kinetic mechanism is required in these cases. Usually, the reduction of the chemical
kinetics is performed in such a way that the reduced scheme is able at reproducing global
parameters like the adiabatic temperature, the global burning rate or the laminar flame
speed, rarely with a proper assessment of the ability to reproduce dynamical properties
too [13, 14]. The latter requirement is generally taken into consideration in the numerical
studies of flame ignition and quenching. In these cases the adoption of special tuning of
the global reaction kinetics is unavoidable [15].
In a previous work, the role of chemistry in determining the ability of a mixture to
withstand perturbations in conditions close to extinction in a PSR was investigated [16].
The analysis was limited to the comparison of two different one-step mechanisms with
respect to the behaviour observed by adopting the GRI 3.0 Mechanism. Several issues were
highlighted about the ability of the global reaction to reproduce the behaviour obtained
with the GRI 3.0 .
The PSR, also known as Continuous Stirred Tank Reactor (CSTR) or Well Stirred
Reactor (WSR), appears as the minimum archetype that allows to focus on the influence
of the reaction mechanism on the description of the combustion phenomena. The PSR
has been used in several kinetic mechanisms studies in the last years [17, 18, 19, 1, 20, 21].
Namely, it allows the observation of the evolution of a combustion reaction independently
from interactions with complex fluid dynamics and the transport mechanisms.
In this work the PSR model is used to investigate the ability of two global and one
reduced reaction mechanisms to reproduce the behaviour of a reference kinetic scheme
(the well-established GriMech 3.0 scheme [22] for methane combustion), under dynamic
conditions.
Two main issue are addressed. The first is to enlarge the previous study to include
a characterization of other well-known mechanisms. The second is to verify if by using
a little more complex mechanisms, as the selected two-step mechanism [23], significantly
helps to reduce the gap with the predictions given by adopting the GRI 3.0 . This is
obtained by firstly comparing the adiabatic flame temperature obtained with the different
mechanisms. Then, the extinction line in Perfectly Stirred Reactor (PSR) is determined
and proper conditions are selected to perform the analysis of the dynamic behaviour under
both forced and unforced conditions.

Model Description and Methodologies

Perfectly Stirred Reactor
The equations of time evolving non-adiabatic PSR can be written as [1]:
dYj Yj,f − Yj Wj rj
= + , j = 1, 2, . . . , Ns (1)
dt τ ρV
PNs PNs
dT j=1 Yj,f (hj,f − hj ) j=1 hj Wj rj
= − (2)
dt cp τ ρV cp
Here t, Yj , τ , Wj , rj , ρ, V , Ns , T , h, cp , A refer to time, mass fraction of specie j, residence
time, molecular weight of specie j, molar reaction rate of specie j, density, reactor volume,
number of species, temperature, specific enthalpy and constant pressure specific heat.
The subscripts f and o refers to reactor feeding (inlet) conditions and to conditions at
the reactor outlet. M = N
P s
i=1 Mi is the mass inside the reactor.
The residence time τ is defined as:
ρV
τ= (3)
ṁf
where ṁ is the mass flow rate.

Chemical models
In this work we focus on methane-air combustion (air composition approximated as O2
21 %, N2 79 % by volume). We will make use of a detailed kinetic schemes, a reduced
mechanism and two different global reaction mechanism:
• Gas Research Institute 3.0 mechanism (GRI) [22];
• Lu and Law methane-air reduced mechanism (LL) [17];
• 2S-CM2 global two-step mechanism [24, 23, 25];
• Westbrook and Dryer [26] global one step mechanism (WD).
The Gas Research Institute mechanism, also known as GRIMech, is the most widely
known chemical kinetics mechanism used for modelling methane and natural gas combus-
tion in air. The current version (version 3.0) consists of 325 chemical reactions and 53
species. It is optimized for premixed system in the conditions reported in table 1.

Table 1: GRIMech Optimization Ranges

Measure Unit Minimum Maximum

Temperature [K] 1000 2500
Pressure [Pa] 1333.224 1013250
Equivalence Ratio (φ) 0.1 5

The LL reduced mechanism is obtained from GRIMech 3.0 by eliminating the unim-
portant species and reactions and by applying the directed relation graph (DRG) method
[27]. The reduction is based on reaction states sampled from auto-ignition and perfectly
stirred reactor (PSR). The current version (available at http://www.engr.uconn.edu/
~tlu/mechs/mechs.htm) consists of 184 chemical reactions and 30 species. It is optimized
for the conditions reported in table 2.
Table 2: Lu and Law Mechanism Optimization Ranges [17]

Measure Unit Minimum Maximum

Temperature (auto-ignition) [K] 1000 1600
Temperature (PSR) [K] 300
Pressure [Pa] 101325 3039750
Equivalence Ratio (φ) 0.5 1.5

The 2S-CM2 mechanism is a two step global reaction mechanism developed in such
a way that it is able to reproduce GRIMech 3.0 flame speeds for various equivalence
ratios. Particularly, the authors of the mechanism show that the results for the detailed
mechanism and 2S-CM2 are in a good agreement for fresh mixture with T = 300 K,
pressure p = 0.1 MPa and equivalence ratios between 0.6 and 1.0. The reactions and
full specification of Arrhenius parameters are reported in table 3 (A, β, and Ea refer
respectively to pre-exponential factor, temperature exponent, and Activation Energy of
the Arrhenius law).
The WD mechanism is, probably, the best-known one-step global reaction mechanism.
It was developed by following a simple procedure. The authors set to zero the temperature
exponent and they fix activation energy by using appropriate average between values found
 Tag Reactions A β Ea Reaction Orders
2S-CM2-1 CH4 + 1.5O2 −−→ CO + 2H2 O 2 × 1015 0 35000 [CH4 ]0·9 [O2 ]1·1
2S-CM2-2 CO + 0.5O2 −−→ CO2 2 × 109 0 12000 [CO][O2 ]0·5
2S-CM2-2r CO2 −−→ CO + 0.5O2 8.1104 × 1010 0 77194 [CO2 ]

Table 3: Reactions and Arrhenius parameters for 2S-CM2 mechanism [24] (units in cm, s, cal, and mol)

Tag Reactions A β Ea Reaction Orders

WD CH4 + 1.5O2 −−→ CO + 2H2 O 2 × 1015 0 35000 [CH4 ]0·9 [O2 ]1·1

Table 4: Reaction and Arrhenius parameters for Westbrook and Dryer mechanism (units in cm, s, cal,
and mol) [26]

in scientific literature. Then held fixed the reaction orders, the pre-exponential factor was
varied until the model correctly predicted the flame speed for an atmospheric pressure,
stoichiometric fuel-air mixture. This procedure was repeated for different set of reaction
orders in order to find the set that correctly predicted the variation of flame speed with
equivalence ratio. The reaction and full specification of Arrhenius parameters are reported
in table 4. In this work was chosen the set 3 reported in the Westbrook and Dryer work
in order to avoid negative reaction order. This set shows good (but not best) agreement
between detailed mechanism and one-step mechanism [26].

Numerical Methods
Adiabatic Flame Temperature. The adibatic flame temperature was computed by using
the Cantera Matlab Toolbox. This Toolbox was used to manage the reaction mechanisms
and to compute equilibrium solution of the mixture. Particularly, we exploited the Can-
tera’s chemical equilibrium solver, which is invoked by the function equilibrate. This solver
is based on element potential method [28], a method that reduces the equilibrium prob-
lem to solving a set of M (number of chemical elements) nonlinear algebraic equations.
Cantera uses a damped Newton method to solve the nonlinear algebraic equations in the
equilibrium problem[28]. The element potential method is part of the class of equivalent
“nonstoichiometric” methods [29].

Extinction. In the framework of bifurcation theory, the ignition and extinction correspond
to saddle-node bifurcation points in a classical S-shaped steady-state curve [30, 20]. It is
worth to note that exact location of these limit phenomena can not be obtained by using
temporal simulations tools. The bifurcation analysis and the parametric continuation
technique are the tools of choice for studying these phenomena. In this work we limit
the investigation to the extinction and we consider the residence time as continuation
parameter. For identifying this bifurcation point we use an home-made code. This code
uses the pseudo-arclength continuation algorithm to compute the steady state solution
curve and the fold bifurcation test function [31, 32] to identify the bifurcation. The code
was written within the Matlab R
numerical computing environment by using the Cantera
Matlab Toolbox for managing of chemical kinetic mechanisms.

PSR Dynamic Behaviour. The dynamic behaviour of the PSR was studied under forced
inlet conditions. Following the studies [3], [18] and [33], the sinusoidal perturbation of
the residence time:
τ (t) = τ0 [1 + ε sin(2πf t)] (4)
was imposed to the inlet conditions of the PSR. Several analyses were conducted with
different equivalence ratio (0.4-2.0), frequency (1-10946 Hz) and perturbation amplitude
(2-21%). The residence time τ0 was, arbitrarily, assumed 4% greater than the extinction
residence time at fixed equivalence ratio.
The PSR initial value problem was solved by using the standard Matlab R
Ordinary
Differential Equations solver (ode15s).

Results and Discussion

Analysis of Adiabatic Flame Temperature
A preliminary comparison between the introduced mechanisms was performed by studying
the adiabatic flame temperature for temperature T = 300 K, pressure p = 101325 Pa and
equivalence ratios between 0.4 and 2.0 (figure 1). The purpose of this investigation is to
roughly identify the interval of equivalence ratios where the reduced reaction mechanism
are in good agreement with detailed mechanism.
2500

2250
Adiabatic Flame Temperature [K]

2000

1750

1500

GriMech 3.0
1250 Lu
2S-CM2
Westbrook and Dryer

1000
0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Equivalence Ratio

Figure 1: Adiabatic Flame Temperature for T = 300 K, p = 101325 Pa, and equivalence ratios between
0.4 and 2.0

Figure 1 shows that the LL methane-air reduced mechanism adequately predict the
adiabatic flame temperature in all investigated conditions while global mechanisms yield a
satisfactory prediction in the lean region until a critical value of equivalence ratio. Partic-
ularly, the 2S-CM2 mechanism and WD mechanism show a good agreement with GriMech
3.0 respectively for equivalence ratio lower than 1.3 and 0.9. These results appear to be
consistent with the limits of validity of each reaction mechanism. Solely the WD mech-
anism overestimates the adiabatic flame temperature for stoichiometric mixture. This
result appears to be in contrast with the procedure used to develop the mechanism. How-
ever, it can be explained by considering that in the development of the WD mechanism,
flame speeds were the target parameters, computed using a detailed reaction mechanism
different from GRI and experimental data available at that time.

Extinction Residence Time

Further information about the ability of reduced and global reaction mechanisms to re-
produce the behaviour of the GRI mechanism (the reference mechanism) can be obtained
by studying extinction in the PSR. The extinction residence time is computed at differ-
ent equivalence ratios. The envelope of these results in the plane Equivalence Ratio -
Residence Time forms a curve that we define extinction curve (figure 2).
0,1

Gri Mech 3.0

Lu and Law
2S-CM2
Westbrook and Dryer
0,01

Extinction Residence Time [s]

10−3

10−4

10−5
0,4 0,6 0,8 1 1,2 1,4 1,6 1,8 2
Equivalence Ratio

Figure 2: Extinction curves for the different reaction mechanisms for inlet methane-air mixture at
temperature T = 300 K and pressure p = 101325 Pa

The figure 2 highlights some important results. First of all, only the LL mechanism
is in excellent agreement with the GRI. This result confirm that the reduced mechanism
obtained with Directed Relation Graph Method exhibit good performance in predicting
the extinction turning point in PSR [21]. Instead, the global mechanisms prove to be
unsuitable for exactly computing the extinction points. The most noticeable difference
is the location of the minimum extinction residence time. In order to quantify these
differences we compute the percentage change in extinction residence time between the
reduced and global mechanisms and the reference mechanism, defined as:
τ − τref
Percentage change = (5)
τref

The plot in figure 3a confirms the excellent agreement between LL and the GRI mech-
anism (differences lower than 2% over the entire range of equivalence ratios). A slight
deterioration in the agreement can be found for high equivalence ratio (greater than 1.3).
The error in predicting the extinction turning point in PSR by using global mechanism
can be extremely high (3b). This error is lower than ±5% at φ values close to 0.4 in
the lean mixtures side and at φ close to 1.3 in the rich side with the WD mechanism.
These values shift to 0.526 and 1.05 with the 2S-CM2 mechanism. Particularly, for this
mechanism the error is lower than ±35% for the equivalence ratios approximately in the
range [0.45 1.15]. The error keeps much higher, up to more than 100% elsewhere.

Analysis of dynamic behavior under periodic forcing

The previous results helps to identify the proper range of τ and φ to investigate the
predictive capability of reduced and global reaction mechanisms in dynamic conditions.
This investigation was performed by examining the effect of the sinusoidal perturbation
of the residence time defined by equation (4) on the reactor temperature.
The initial value problem 1-2 (IVP) was solved by assuming the reactor initially filled
only with nitrogen at temperature T = 3000 K and pressure p = 101325 Pa and fed with
methane-air mixture at temperature T = 300 K and pressure p = 101325 Pa.
 0,25 200

Westbrook and Dryer

2S-CM2
0 150
Residence Time Percentage Difference [%]

Residence Time Percentage Difference [%]

100

−0,5 50

−1 −50

Lu and Law

−1,25 −100
0,4 0,6 0,8 1 1,2 1,4 1,6 1,8 2 0,4 0,6 0,8 1 1,2 1,4 1,6 1,8 2
Equivalence Ratio Equivalence Ratio

(a) Reduced Mechanism (b) Global Mechanisms

Figure 3: Percentage change in extinction residence time

From the solutions of the IVP we compute the dimensionless amplitude of the tem-
perature response Θ = (Tmax − T min)/Tmean (amplitude divided by the mean value of
the last four periods when regime conditions are attained) and the phase shift (with Fast
Fourier Transform). These values computed with reduced and global mechanisms are
then compared with those of GRI.
Figures 4 and 5 describe, respectively, the dependence of Θ and phase shift on the
frequency and amplitude of the perturbation for all the reaction mechanisms considered in
this work. Following the previously illustrated results, we limit the study to equivalence
ratio less than 1.25 because a proper comparison would not seems meaningful for higher
equivalence ratios.
Figure 4 corroborates the excellent agreement between the reduced mechanism LL and
the GRI mechanism even under dynamic conditions. Moreover, this figure shows several
relevant differences between the global mechanisms (2S-CM2 and LL) and GRI. Particu-
larly, the 2S-CM2 mechanism overestimates the dimensionless amplitude of temperature
response for every sinusoidal perturbation and equivalence ratio. The WD mechanism, in-
stead, underestimates Θ at φ = 0.5, gives a good agreement at φ = 0.8 and overestimates
it otherwise.
We found a different sensitivity of the combustion process to a given perturbation
among the global mechanisms (2S-CM2 and WD) and GRI. Ignited solutions obtained
with the global mechanisms are found at frequencies less than or equal to that obtained
with the GRI for amplitudes greater that 4%. Furthermore, global mechanisms are not
able to reproduce the trend of Θ with frequency at the different value of φ. This is very
apparent at φ = 1.25, where this trend is not monotone for the GRI and LL at the
lowest amplitude of perturbation. The inflection point, defined as the intersection of the
lines drawn through the first and last two points of each curve, moves with the selected
mechanism. Finally, the GRI mechanism show a much complex shape of the curves with
several changes of concavity, more evident at φ = 0.5.
Similar observation can be reported about the behaviour of the phase shifts. The
excellent agreement between the reduced mechanism LL and the GRI mechanism is con-
firmed also with respect to figure 5. Instead the comparison between the GRI and the
 ε = 2% ε = 3% ε = 5% ε = 8% ε = 13% ε = 21%

0,1

0,01
Θ

10−3
GRI GRI GRI GRI
Φ = 0.5 Φ = 0.8 Φ = 1.0 Φ = 1.25
10−4
0,1

0,01
Θ

10−3
LL LL LL LL
Φ = 0.5 Φ = 0.8 Φ = 1.0 Φ = 1.25
10−4
0,1

0,01
Θ

10−3
2S-CM2 2S-CM2 2S-CM2 2S-CM2
Φ = 0.5 Φ = 0.8 Φ = 1.0 Φ = 1.25
10−4
0,1

0,01
Θ

10−3
WD WD WD WD
Φ = 0.5 Φ = 0.8 Φ = 1.0 Φ = 1.25
10−4
1 10 100 1000 104 1 10 100 1000 104 1 10 100 1000 104 1 10 100 1000 104
Frequency [Hz] Frequency [Hz] Frequency [Hz] Frequency [Hz]

Figure 4: Dimensionless amplitude of temperature response versus frequency at different equivalence

ratio for the four reaction mechanism adopted in this work. The not ignited solutions are not shown.

two global mechanisms (2S-CM2 and LL) highlights an important difference. The phase
shift for the global mechanism is a monotone decreasing function of the frequency over
the range investigated. Instead for the GRI the phase shift has a minimum at each equiv-
alence ratio, whose position move with φ with a maximum in frequency at stoichiometric
conditions.
Remarkably, all curves appear not depending on the amplitude of the perturbation,
apart from the points corresponding to the ignited solutions with lowest frequency (with
reference to ε > 4%). Therefore, the error in the prediction of the phase shift performed
with global mechanism increases significantly with frequency.

Conclusions
An extension of previous works [34, 16], aimed to evaluate the behaviour of simplified
combustion mechanisms under dynamic conditions has been presented.
It results that moving from one-step global mechanism to two-steps global mechanism
does not increase significantly the accuracy of the predictions with respect to the adoption
of detailed mechanism. Indeed, even if some steady properties are better predicted by the
two-step mechanism, as the adiabatic flame temperature and the extinction residence time,
 ε = 2% ε = 3% ε = 5% ε = 8% ε = 13% ε = 21%

20
Phase Shift [deg]

−20

−60
GRI GRI GRI GRI
Φ = 0.5 Φ = 0.8 Φ = 1.0 Φ = 1.25
−100
20
Phase Shift [deg]

−20

−60
LL LL LL LL
Φ = 0.5 Φ = 0.8 Φ = 1.0 Φ = 1.25
−100
20
Phase Shift [deg]

−20

−60
2S-CM2 2S-CM2 2S-CM2 2S-CM2
Φ = 0.5 Φ = 0.8 Φ =1.0 Φ = 1.25
−100
20
Phase Shift [deg]

−20

−60
WD WD WD WD
Φ = 0.5 Φ = 0.8 Φ = 1.0 Φ = 1.25
−100
1 10 100 1000 104 1 10 100 1000 104 1 10 100 1000 104 1 10 100 1000 104
Frequency [Hz] Frequency [Hz] Frequency [Hz] Frequency [Hz]

Figure 5: Phase Shift versus frequency at different equivalence ratio for the four reaction mechanism
adopted in this work. The not ignited solutions are not shown.

this is not sufficient to increase the accuracy the prediction of the dynamic behaviour.
Richer simplified chemical mechanisms seems therefore required in order to have re-
liable predictions under dynamic conditions. The mechanism proposed by Lu and Law
[17] is able to furnish predictions comparable to those obtained with the GRI.
Therefore further investigation effort will be devoted to identify the minimum level of
complexity a chemical mechanism has to have to correctly investigate dynamics combus-
tion phenomena.

*
References
[1] Wada, T., Jarmolowitz, F., Abel, D., Peters, N., “An instability of diluted lean
methane/air combustion: Modeling and control”, Combustion Science and Technol-
ogy 183:1–19 (2010).
[2] Lieuwen, T., Yang, V. (editors), Combustion Instabilities in Gas Turbine Engines:
Operational Experience, Fundamental Mechanisms, and Modeling, vol. 210 of Progress
in Astronautics and Aeronautics, AIAA - American Institute of Aeronautics and
Astronautics, 2006.
 [3] Lieuwen, T., Neumeier, Y., Zinn, B.T., “The role of unmixedness and chemical kinet-
ics in driving combustion instabilities in lean premixed combustors”, Combust. Sci.
and Tech. 135:193–211 (1998).
[4] Sardi, K., Whitelaw, J.H., “Extinction timescales of periodically strained, lean coun-
terflow flames”, Experiments in Fluids 27:199–209 (1999).
[5] Balachandran, R., Dowlingy, A.P., Mastorakos, E., “Non-linear response of premixed
flames to superimposed harmonic inlet velocity fluctuations of multiple frequencies”,
41st AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, Tucson, Ari-
zona, USA (2005).
[6] Lieuwen, T., “Nonlinear kinematic response of premixed flames to harmonic velocity
disturbances”, Proceedings of the Combustion Institute 30:1725–1732 (2005).
[7] Ayoola, B., Hartung, G., Armitage, C., Hult, J., Cant, R., Kaminski, C., “Tempera-
ture response of turbulent premixed flames to inlet velocity oscillations”, Experiments
in Fluids 46:27–41 (2009).
[8] Wang, H., Law, C., Lieuwen, T., “Linear response of stretch-affected premixed flames
to flow oscillations”, Combustion and Flame 156:889–895 (2009).
[9] Balachandran, R., Dowling, A., Mastorakos, E., “Dynamics of bluff-body stabilised
flames subjected to equivalence ratio oscillations”, Proceeding of the European Com-
bustion Meeting (2011).
[10] Schrödinger, C., Paschereit, C., Oevermann, M., “Numerical studies on the impact of
equivalence ratio oscillations on lean premixed flame characteristics and emissions”,
Seventh International Conference on Computational Fluid Dynamics (ICCFD7), Big
Island, Hawaii, July 9-13, 2012 ICCFD7-3401 (2012).
[11] Cuoci, A., Frassoldati, A., Faravelli, T., Ranzi, E., “Formation of soot and nitrogen
oxides in unsteady counterflow diffusion flames”, Combustion and Flame 156:2010–
2022 (2009).
[12] Lu, T., Law, C.K., “Toward accommodating realistic fuel chemistry in large-scale
computations”, Progress in Energy and Combustion Science 35:192 – 215 (2009).
[13] Sidhu, H.S., Forbes, L.K., Gray, B.F., “Forced reaction in a cstr. a comparison
between the full system and the reduced model”, Chemical Engineering Science
52:2667–2676 (1997).
[14] Valorani, M., Paolucci, S., “The g-scheme: A framework for multi-scale adaptive
model reduction”, Journal of Computational Physics 228:4665–4701 (2009).
[15] Lacaze, G., Richardson, E., Poinsot, T., “Large eddy simulation of spark ignition in
a turbulent methane jet”, Combustion and Flame 156:1993–2009 (2009).
[16] Acampora, L., Marra, F.S., Martelli, E., “Non-linear response to periodic forcing
of methane-air in continuous stirred tank reactors”, XXXV Meeting of the Italian
Section of the Combustion Institute (2012).
[17] Lu, T., Law, C.K., “A criterion based on computational singular perturbation for
the identification of quasi steady state species: A reduced mechanism for methane
oxidation with {NO} chemistry”, Combustion and Flame 154:761 – 774 (2008).
[18] Lilleberg, B., Ertesvåg, I.S., Rian, K.E., “Modeling Instabilities in Lean Premixed
Turbulent Combustors Using Detailed Chemical Kinetics”, Combustion Science and
Technology 181:1107–1122 (2009).
[19] Mze-Ahmed, A., Hadj-Ali, K., Dievart, P., Dagaut, P., “Kinetics of oxidation of a
synthetic jet fuel in a jet-stirred reactor: Experimental and modeling study”, Energy
& Fuels 24:4904–4911 (2010).
[20] Shan, R., Lu, T., “Ignition and extinction in perfectly stirred reactors with detailed
chemistry”, Combustion and Flame 159:2069–2076 (2012).
[21] Wang, Q.D., Fang, Y.M., Wang, F., Li, X.Y., “Skeletal mechanism generation for
high-temperature oxidation of kerosene surrogates”, Combustion and Flame 159:91
– 102 (2012).
[22] Smith, G.P., Golden, D.M., Frenklach, M., Eiteener, B., Goldenberg, M., Bowman,
C.T., Hanson, R.K., Gardiner, W.C., Lissianski, V.V., Qin, Z.W., “GRI-Mech 3.0”,
(2000).
[23] Bibrzycki, J., Poinsot, T., Zajdel, A., “Investigation of laminar flame speed of
CH4/N2/O2 and CH4/CO2/O2 mixtures using reduced chemical kinetic mecha-
nisms”, Archivum Combustionis 30:287–296 (2010).
[24] Bibrzycki, J., Poinsot, T., “Reduced chemical kinetic mechanisms for methane com-
bustion in o2/n2 and o2/co2 atmosphere”, Working note ECCOMET WN/CFD/10
17 (2010).
[25] Bibrzycki, J., Poinsot, T., “Examination of simplified mechanisms of CH4 combus-
tion in N2/O2 and CO2/O2 atmosphere using mathematical modeling”, Archivum
Combustionis 31:255–262 (2011).
[26] Westbrook, C.K., Dryer, F.L., “Simplified Reaction Mechanisms for the Oxidation of
Hydrocarbon Fuels in Flames”, Combustion Science and Technology 27:31–43 (1981).
[27] Lu, T., Law, C.K., “A directed relation graph method for mechanism reduction”,
Proceedings of the Combustion Institute 30:1333 – 1341 (2005).
[28] Goodwin, D.G., Moffat, H.K., Speth, R.L., “Cantera: An object-oriented software
toolkit for chemical kinetics, thermodynamics, and transport processes”, http://
www.cantera.org (2014), version 2.1.
[29] Smith, W., Missen, R., Chemical reaction equilibrium analysis: theory and algorithms,
Wiley series in chemical engineering, Wiley, 1982.
[30] Law, C.K., Combustion Physics, Cambridge University Press, 2006.
[31] Kuznetsov, Y.A., Elements of Applied Bifurcation Theory, Applied Mathematical
Sciences, Springer.
[32] Seydel, R., Practical Bifurcation and Stability Analysis, vol. 5 of Interdisciplinary
Applied Mathematics, Springer New York, New York, NY, 2010.
[33] Oh, C.B., Lee, E.J., Jung, G.J., “Unsteady auto-ignition of hydrogen in a per-
fectly stirred reactor with oscillating residence times”, Chemical Engineering Science
66:4605 – 4614 (2011).
 [34] Marra, F.S., Martelli, E., “Non-linear response to periodic forcing of methane-air
global and detailed kinetics in Perfectly Stirred Reactors”, 23th ICDERS, Interna-
tional Colloquium on the Dynamics of Explosion and Reacive System, University of
California, Irvine - July 24-29, 2011 (2011).

View publication stats