Energy Conversion and Management 119 (2016) 227–238

Contents lists available at ScienceDirect

Energy Conversion and Management

journal homepage: www.elsevier.com/locate/enconman

Experimental and numerical investigation of hetero-/homogeneous

combustion-based HCCI of methane–air mixtures in free-piston micro-
engines
Junjie Chen ⇑, Baofang Liu, Xuhui Gao, Deguang Xu
School of Mechanical and Power Engineering, Henan Polytechnic University, Jiaozuo, Henan, China

a r t i c l e i n f o a b s t r a c t

Article history: The hetero-/homogenous combustion-based HCCI (homogeneous charge compression ignition) of fuel–
Received 14 January 2016 lean methane–air mixtures over alumina-supported platinum catalysts was investigated experimentally
Received in revised form 9 April 2016 and numerically in free-piston micro-engines without ignition sources. Single-shot experiments were
Accepted 15 April 2016
carried out in the purely homogeneous and coupled hetero-/homogeneous combustion modes, involved
Available online 20 April 2016
temperature measurements, capturing the visible combustion image sequences, exhaust gas analysis,
and the physicochemical characterization of catalysts. Simulations were performed with a two-
Keywords:
dimensional transient model that includes detailed hetero-/homogeneous chemistry and transport, leak-
Free-piston micro-engine
Micro-combustion
age, and free-piston motion to gain physical insight and to explore the hetero-/homogeneous combustion
Hetero-/homogeneous combustion characteristics. The micro-engine performance concerning combustion efficiency, mass loss, energy den-
Mass loss sity, and free-piston dynamics was investigated. The results reveal that both purely homogeneous and
Free-piston dynamics coupled hetero-/homogeneous combustion of methane–air mixtures in a narrow cylinder with a diame-
Detailed kinetic modeling ter of 3 mm and a height of approximately 0.3 mm are possible. The coupled hetero-/homogeneous mode
can not only significantly improve the combustion efficiency, in-cylinder temperature and pressure, out-
put power and energy density, but also reduce the mass loss because of its lower compression ratio and
less time spent around TDC (top dead center) and during the expansion stroke, indicating that this cou-
pled mode is a promising combustion scheme for micro-engine. Heat losses result in higher mass losses.
Heterogeneous reactions cause earlier ignition, which is very favorable for the micro-device.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction micro-engine designer. Quenching is a matter of great concern

Recently, micro-scale combustion has attracted increased atten-
tion because of the growing interest in developing portable power
generation systems, with increasing efforts toward miniaturizing
thermal engines for electricity production of a few tens of Watts
[1]. Specifically, the high energy density of hydrocarbon fuels
allows for the realization of micro-scale power generation systems
[2]. Among the micro-scale system using the chemical energy of
hydrocarbon fuels, micro-scale heat engine is being explored [3].
Micro-engine-based power supplies can deliver large energy den-
sity, and may eventually replace traditional lithium-ion batteries
in various applications [4].
How to prevent thermal and radical quenching and achieve
stable combustion is one of the most challenging issues to the
⇑ Corresponding author at: School of Mechanical and Power Engineering, Henan
Polytechnic University, 2000 Century Avenue, Jiaozuo, Henan 454000, PR China.
E-mail address: concjj@163.com (J. Chen).
because the specific heat transfer rate varies inversely with the
characteristic dimension of the combustion chamber [5]. As engine
size decreases, the surface-to-volume ratio increases, resulting in
increased heat losses and increased potential destructions of active
radicals on the combustion chamber walls. These mechanisms will
increase the chemical reaction time and possibly inhibit the onset
of homogeneous ignition, resulting in thermal or radical quenching
[6]. Another concern is residence time, decreasing the dimension of
the combustion chamber results in significant reduction of resi-
dence time because the flow velocity cannot be reduced accord-
ingly. Insufficient residence time results in partial or incomplete
combustion, in turn resulting in insufficient heat generation and
further quenching [7]. In micro-scales, traditional engine combus-
tion schemes are generally infeasible because of quenching effects
and insufficient residence times. When considering micro-scale
application, HCCI combustion mode is a promising alternative
because it has the following experimentally verified characteris-
tics: ignition is not initiated by an external event; the charge is

http://dx.doi.org/10.1016/j.enconman.2016.04.055
0196-8904/Ó 2016 Elsevier Ltd. All rights reserved.
228 J. Chen et al. / Energy Conversion and Management 119 (2016) 227–238
consumed very rapidly; ignition occurs simultaneously at numer-
ous locations inside the combustion chamber; and an absence of
traditional flame propagation [8]. In the context of micro-
combustion, the fundamental attributes of HCCI combustion are
that an external ignition system is not required, and that the
charge is combusted essentially without flame propagation. In
addition, auto-ignition suggests that the charge is consumed both
uniformly and rapidly, consequently minimizing quenching effects
and without resorting to complicated ignition schemes. This con-
cept may be a way toward elimination of ignition sources from
micro-scale devices, resulting in further reduction of system size
[9]. Additionally, HCCI combustion rate limited by chemical kinet-
ics rather than transport and essentially no flame propagation
result in shorter charge consumption time and in turn higher
engine speed and efficiency [10]. These traits essentially bring
HCCI combustion closer to a constant volume procedure, pushing
the thermodynamic process closer to an ideal Otto cycle [11]. Fur-
thermore, the ability to sustain homogeneous combustion with
very lean mixtures reduces fuel mass losses through the leakage
between the cylinder and piston, adversely enhancing fuel con-
sumption efficiency [10,11].
However, HCCI engines are more difficult to control than tradi-
tional engines. The combustion occurrence (or ignition) depends
on chemical kinetics and the compression process [12,13]. There-
fore, controlling ignition timing is a challenge because it must be
done indirectly. Experiments have demonstrated that employing
a variable compression ratio is a promising approach to control
HCCI ignition [9]. Therefore, to utilize this strategy, an untraditional
engine such as a free-piston engine (FPE) is necessary, and is a very
promising HCCI engine concept due to its valuable feature of vari-
able compression ratio. The distinguishing feature of this device is
a mechanically unconstrained free-piston; the free-piston motion
is not restricted by a crankshaft mechanism, but is completely
determined by gas pressure forces. In addition, the free-piston
engine is a promising power generation device, offering the benefits
of higher thermal efficiency and heat release rate compared to
those of the traditional engine [14], extensive operation optimiza-
tion [15], mechanical simplicity [16], multi-fuel/combustion mode
flexibility [17], and reduced NOx formation [18,19].
Computational models for simulating combustion and heat
transfer of HCCI engines require detailed chemistry models; this
is primarily because the ignition of HCCI engines is more sensitive
to chemical kinetics. In addition, computational models have
demonstrated that the importance of accounting for the fact that
the in-cylinder mixture is actually in-homogeneous, particularly
in terms of temperature field [20]. This in-homogeneity is driven
by heat transfer from the combustion chamber walls and the tur-
bulent mixing of fuel. Moreover, recent simulations have demon-
strated that the charge inhomogeneity has a significant effect on
the pressure rise rates and the consequential engine performance
[21,22]. The charge in-homogeneity would increase with decreas-
ing cylinder size because in-homogeneities in the cylinder is
caused by the thermal boundary layer adjacent to the cylinder
walls (e.g., the in-homogeneity in the temperature field is caused
by the mixing of the colder gases in the boundary layer into the
bulk gas) [20]. The increased charge in-homogeneity, coupled with
the high surface-area-to-volume ratio, may ultimately constrain
the combustion chamber dimension. Furthermore, homogeneous
combustion is mainly controlled by the temperature boundary
layer and reactant species profiles [23,24]. Consequently, in this
work, a spatial dimensionality of at least two is necessary to cor-
rectly describe interphase transport and homogeneous combustion
in particular.
Additionally, HCCI suffers from comparatively large hydrocar-
bon and carbon monoxide emissions, and especially from poor
power density [25,26]. When considering micro-scale application,
however, HCCI is pursued because the charge can be consumed
rapidly, and especially be ignited without additional ignition
devices. Furthermore, although purely homogeneous combustion
can be stabilized in micro-scales [27,28], coupled hetero-/
homogeneous combustion is the preferred route because of large
surface-to-volume ratio issues, compared to purely homogeneous
combustion [29,30]. Heterogeneous reaction broadens the classical
flammability limits [31], and results in stable combustion even in
the presence of high heat losses [32,33]. Coupled hetero-/
homogeneous combustion becomes an attractive technology with
prototype meso-scale devices (miniature free-piston Stirling
engine) already successfully demonstrated [34]. Recent works
[35,36] have further demonstrated the potential of catalytically-
coated walls in inhibiting intrinsic flame instabilities.
The coupled hetero-/homogeneous combustion methodology
crucially depends upon multidimensional CFD (computational fluid
dynamics) modeling needed for combustor design [2,24,29,30], as
well as upon advances in catalyst preparation technology (develop-
ment of active and thermally stable catalysts) [32,35,37]. Nowa-
days, in the context of micro-combustion, two-dimensional
models with detailed hetero-/homogeneous chemistry [37–40]
have become common study methods for steady simulations. How-
ever, to the best of our knowledge, CFD modeling based on such
detailed chemistry have not been applied to time-accurate tran-
sient simulations, coupled with HCCI and dynamic mesh. Transient
catalytic channel simulations with detailed hetero-/homogeneous
chemistry were reported in previous work [2,41–43] using a com-
mercial CFD software; however, these studies did not address
issues such as compression ignition (or HCCI), leakage, free-piston
movement, turbulent transport, and hetero-/homogeneous chem-
istry coupling. In addition, although the heterogeneous catalytic
oxidation and the free-piston HCCI engine have been widely stud-
ied by means of experimental and numerical methodologies [2,5,6
,8,9,17,24,29,30,32,35,37,44,45], fundamental studies focused on
hetero-/homogeneous combustion in free-piston micro-engines
are still lacking and require further investigation.
The present work undertakes a combined experimental and
numerical investigation of the hetero-/homogenous combustion-
based HCCI of fuel–lean methane–air mixtures over alumina-
supported platinum catalysts in a free-piston micro-engine, which
is candidate for small portable power generation applications.
Single-shot experiments in the purely homogeneous and coupled
hetero-/homogeneous combustion modes were performed in a
free-piston micro-engine with a cylinder bore of 3 mm, involved
temperature measurements, capturing the visible hetero-/
homogeneous combustions image sequences, exhaust gas analysis,
and physicochemical characterization of catalysts. A two-
dimensional transient numerical model incorporating detailed
hetero-/homogeneous chemistry and transport, leakage, dynamic
mesh, turbulence, and thermodynamic–dynamic balance is devel-
oped to interpret the single-shot experimental results as well as to
explore hetero-/homogenous combustion characteristics. Following
numerical validation and interpretation of the single-shot experi-
mental results, the transient model was used to investigate the
purely homogeneous and coupled hetero-/homogeneous combus-
tion characteristics. In addition, the performance of free-piston
micro-engines with regard to fuel conversion efficiency and mass
loss was investigated. The primary objective of this work is to
explore the feasibility of coupled hetero-/homogeneous
combustion-based HCCI in free-piston micro-engines, and to char-
acterize the purely homogeneous and coupled hetero-/
homogeneous combustion at the micro-scale as well as to explore
the free-piston dynamics. Of particular interest in the present work
is to resolve the poor power density and low fuel conversion
 J. Chen et al. / Energy Conversion and Management 119 (2016) 227–238
efficiency caused by exceedingly short charge consumption time
and flow residence time as a result of extremely high free-piston
oscillation frequency in free-piston micro-engines.
This article is organized as follows. In Sections 2, the experi-
mental methodologies are introduced, and results for different
combustion modes are presented in Sections 3. The two-
dimensional transient numerical model and detailed kinetic
schemes are provided in Section 4. In Section 5.1, comparisons
between predictions and measurements are used to assess the apt-
ness of the transient numerical model and the employed kinetic
schemes, followed by the interpretation of the single-shot experi-
mental results to gain physical insight into the hetero-/
homogeneous combustion process in Section 5.2. Conclusions are
finally summarized in Section 6.
2. Experimental
Single-shot experiments were performed to characterize
hetero-/homogeneous combustion at the micro-scale, as well as
to explore the free-piston dynamics. The experimental setup used
consists of a single-shot micro-engine designed to reproduce one
compression–combustion–expansion cycle of the continuous pro-
cess. This allows for very good control of the initial fuel–air mix-
tures in the combustion chamber by avoiding the detrimental
effect of products remaining in the combustion chamber from pre-
vious cycles; it also ensures that other process parameters are the
same for each experiment [46,47]. The hetero-/homogeneous com-
bustion experiment at the micro-scale are illustrated schematically
in Fig. 1 and the general principle is equally simple. A free-piston is
driven into a visual cylinder filled with the combustible mixtures,
while digital movies and thermal image of the total hetero-/
homogeneous combustion process are achieved. The digital movies
capture visible combustions and provide temporal measurements
of velocity and position of the free-piston. Gas chromatography
(GC) provided the exhaust composition, whereas infrared thermal
imager was used to measure the two-dimensional temperature
distributions inside the combustion chamber.
Additionally, since the high surface to volume ratio of miniature
internal combustion engine, the friction between the cylinder
walls and the sealing rings become a major factor, and in some
cases are in the order of the power of miniature engine. For this
reason, no rings are practically installed in these miniature engines
[10]. Therefore, to minimize the piston-cylinder walls gap, high
precision manufacturing is required. In these miniature engines,
the charge leakage is therefore unavoidable, and consequently high
engine speed is required to reduce the mass losses incurred [11].
2.1. Micro-combustion test-rig
Movement of the free-piston is initiated by injection of com-
pressed air through the launch mechanism, consequently the
Fig. 1. Schematic diagram of the single-shot experiment.
229
launch mechanism with an extremely high velocity impulses the
free-piston. The visual cylinder made with plexiglass that has good
pressure-resistance and heat-resistance performance is initially
filled with the methane–air homogenous mixtures, then the free-
piston compresses the charge until it self-combusts. The free-
piston and end-plug is FeCr-alloy, and the upper surface of the
free-piston and the lower surface of the end-plug (i.e. the top of
the combustion chamber) were coated with Pt/c-Al2O3 catalysts,
as shown in Fig. 1. Initial distance between free-piston and end-
plug (chamber length) is 20.0 mm and the cylinder is sealed by a
stainless steel end-plug and two O-rings. The free-piston is a
machined stainless steel gauge pin, having a mass of 0.5 g. The
free-piston is 2.997 mm in diameter with an accuracy of
0.003 mm. Nominally, the inside diameter of the plexiglass cylin-
ders is 3 mm. However, the actual dimensions of these plexiglass
cylinders vary considerably. Therefore, fit-tests is used to match
the free-piston to these plexiglass cylinders. The optimal cylinder
and free-piston combination occurs when the outside diameter
of the free-piston and the inside diameter of the cylinder differ
by approximately 0.006 mm. In order to ensure that the free-
piston is not easy to be distorted and damaged and the intensity
is high when operating at high temperatures, quenching treatment
of the free-piston is used to improve the surface capability. Fur-
thermore, the free-piston surface coated with nickel is prepared
by electroplating, resulting in a high surface flatness and a decrease
in the friction energy loss between cylinder walls and free-piston.
As a result, the fit precision and sealing performance of combustion
chamber is improved. In addition, the free-piston is lubricated to
improve sealing performance. Although these methods ensure that
the free-piston engine has excellent air tightness at ambient tem-
perature and pressure, high in-cylinder pressure and temperature
can aggravate the situation. As a result, most of the mass loss
occurs around TDC, as will be discussed in the combustion charac-
teristics section. Therefore, the gas tightness (or leakage) detection
can hardly be achieved today, and it is still a challenging and inter-
esting task for the future.
2.2. Procedure
At first the combustion chamber is filled with reactants. Initially
the free-piston is in the bottom of the combustion chamber. After
this the compression–combustion–expansion cycle is initiated by
injection of compressed air at a predefined pressure through the
launch mechanism. As the free-piston moves up, at some point
the charge self-combusts if enough energy is supplied, then the
free-piston continues to move until the TDC. Finally, the free-
piston moves down again under the force of the compressed gasses
in the combustion chamber. Here and after, bottom and top repre-
sented in Fig. 1 along with moving up and down are the essential
terms for engines. In addition, two different detection devices are
230 J. Chen et al. / Energy Conversion and Management 119 (2016) 227–238
used to monitor important parameters during the experiment. The
high-speed video and motion analysis (Vision Research PhantomÒ
M310) is used to capture visible hetero-/homogeneous combus-
tions, and to determine the velocity and position of the free-
piston. The maximum frame rate and temporal resolution of the
high-speed video and motion analysis are 3.2 Gigapixels/second
throughput and one microsecond, respectively. The high-speed
video and motion analysis is triggered when the compressed air
solenoid valve is actuated. The laser beam is irradiated to the plex-
iglass cylinder through a collimator lens, and used as the light
source of high-speed video and motion analysis. Infrared thermal
imager (FLIR T640, errors of flame temperature field measure-
ments: ±2 K, emissivity: 0.95) monitors the two-dimensional tem-
perature distributions inside the combustion chamber. Initial
pressures for all experiments are 0.1 MPa; initial temperature is
300 K. All of the experimental results shown are from a series of
experiments using fuel–lean methane–air mixtures.
2.3. Catalyst preparation
The supported platinum catalysts of the present work were 0.8%
platinum (wt) on c-Al2O3, prepared by incipient wetness impreg-
nation of the c-Al2O3 support with a platinum(II) nitrate (Pt
(NO3)2) solution. The impregnated catalyst powder was dried over-
night at 393 K and calcined at 1073 K for 8 hours. A slurry prepared
from the powder and a solvent mixture was sprayed on FeCr-alloy
foils. The coated foils were calcined at 873 K for 6 h and the mea-
sured thickness of the catalytic layer was approximately 20 lm.
The surface composition of catalyst-coated FeCr-alloy samples
was determined with XPS (X-ray photoelectron spectroscopy);
XPS provided information on the oxidation state of the detected
elements present in the samples and also showed that Cr or Fe
did not diffuse in the surface. In addition, surface Raman micro-
scopy was used to determine the crystal structure of the support.
Selected samples were analyzed both before and after the activity
tests to determine any structural changes occurring under reaction
conditions. Finally, the active and total surface areas of both used
and fresh platinum-coated FeCr-alloy foils were measured with
the chemisorption of carbon monoxide and the BET (Brunauer–
Emmett–Teller) method of nitrogen physisorption, respectively
[48,49]. The detailed catalyst characterization is outside the scope
of this study; in the following, only information relevant to the cur-
rent work will be presented.
2.4. Gas analysis
The exhaust composition was analyzed by gas chromatography.
Detailed analysis of the exhaust composition was carried out with
an Agilent GC 7890A equipped with a thermal conductivity sensor
and porous polymer and molecular sieve packed columns. The car-
rier gas of the gas chromatography was helium. The single-shot
experiment, sampling, and GC analysis was continuous, whereas
each analysis was discontinuous at 8 min intervals in order to
remove the remaining compounds. The minimum time for each
analysis was 20 min in order to achieve a high peak capacity and
resolution that represents a characteristic of the separation of
two adjacent peaks. Integration was performed with the Agilent
ChemStation software, which is a software package to control Agi-
lent gas chromatography systems, while a spreadsheet was used to
calculate the final compositions of the exhaust gasses. The gas
chromatography allowed measurements of all key gas species,
such as methane, carbon dioxide, carbon monoxide and nitrogen.
Before the gasses entered the molecular sieve column, the porous
polymer column separated carbon dioxide. In addition, the gas
chromatography has been tested against a selection of different
calibration gas mixtures. The relative error in the gas chromatogra-
phy measurements of the gas species was less than 5% for
methane, carbon dioxide, carbon monoxide and nitrogen, increas-
ing up to 60% for the scarce oxygen.
3. Experimental results and discussion
Representative image sequences from the single-shot experi-
ments in different combustion modes are presented in Fig. 2. It
conclusively demonstrates that both purely homogeneous and
coupled hetero-/homogeneous combustion of methane in a narrow
cylinder with a diameter of 3 mm and a height of approximately
0.3 mm are possible. It is noteworthy because the cylinder is nei-
ther heated nor insulated and the charge is initially at ambient
conditions. Moreover, the free-piston near the TDC of the compres-
sion stroke is virtually stationary during each combustion. As a
result, purely homogeneous or coupled hetero-/homogeneous
combustion is essentially a constant-volume process, as well as
one would expect the fuel conversion efficiency to approach the
Otto cycle limit if the whole charge is consumed and the heat
transfer is negligible. Furthermore, Fig. 2(a) demonstrates that
HCCI at the micro-scale is capable of igniting methane–air mix-
tures that traditionally cannot be combusted. In addition, Fig. 2
(a) also reveals that ignition starts at the center of the combustion
chamber, as well as that HCCI proceeds through localized homoge-
neous reactions although only part of the charge is consumed.
Compared to the coupled hetero-/homogeneous combustion
(Fig. 2(b)), the purely homogeneous flame (Fig. 2(a)) under non-
catalytic conditions is very weak and accompanied by only part
of the reactants is consumed, along with the significantly extended
compression–combustion–expansion cycle as a result of its lower
free-piston velocity during the expansion stroke. In the coupled
hetero-/homogeneous combustion mode, heterogeneous reactions
can significantly improve the fuel conversion efficiency and in-
cylinder temperature and enhance combustion process, resulting
in a shorter cycle period (i.e. higher free-piston oscillation fre-
quency). In order to improve the power density and to achieve reli-
able auto-ignition, the micro-engine must operate at kilohertz
frequencies, placing requirements about the allowable ignition
delay times. In addition, the lower apparent activation energy of
heterogeneous reaction causes earlier ignition, resulting in lower
ignition temperature and pressure. Earlier ignition extends the
reaction time inside the combustion chamber, and consequently
improve the combustion efficiency, which is very favorable for
the micro-device.
Experimental results of the free-piston dynamics in the purely
homogeneous and coupled hetero-/homogeneous combustion
modes are compared in Fig. 3. For the compression path, two com-
bustion modes give similar predictions of the free-piston velocity
and position, but the expansion paths vary greatly. When the
free-piston approaches the TDC, the compression rate or the free-
piston velocity slows down while the in-cylinder charge continue
to leak through the gap between cylinder walls and free-piston
from combustion chamber, at a rate that is proportional to the
cylinder pressure. Then, the in-cylinder charge is ignited and, con-
sequently, the in-cylinder temperature and pressure jump, associ-
ated with a sharp increase in the amount of charge loss. Near the
ignition timing, the free-piston velocity varies rapidly. In addition,
energy released by combustion results in higher free-piston veloc-
ity during the expansion stroke. Coupled hetero-/homogeneous
combustion results in higher free-piston velocity during the expan-
sion stroke. However, during the compression stroke, there is only
marginal difference between the free-piston velocities obtained
from different combustion modes. The free-piston motion profile
differs from that of traditional engines, with the primary difference
being a significantly higher piston acceleration around TDC [14].
 J. Chen et al. / Energy Conversion and Management 119 (2016) 227–238 231

Fig. 2. Typical image sequences from the single-shot experiments in different combustion modes. The fuel is methane, the equivalence ratio is 0.8, and the charge is initially
at ambient pressure and temperature.

40

In the coupled hetero-/homogeneous combustion mode, micro-

Free-piston velocity (m/s)

20
engine experiments of different equivalence ratios demonstrated
methane–air combustion with exhaust gas temperatures in excess
0 of 700 K and combustion efficiency up to 90% (without considering
the decreased efficiency induced by the mass loss) is capable of
-20 producing 80 W power, although the leakage, heat loss, friction
loss, use of moving parts, and sealing potentially reduce energy
conversion efficiency. However, in the purely homogeneous com-
-40 bustion mode, homogeneous combustion is obtained only with a
Homogeneous combustion
maximum power of 20 W because of the longer ignition delay
Hetero-/homogeneous combustion
-60 time, lower combustion efficiency, and more mass loss; the latter
0.4 0.6 0.8 1.0 1.2 will be discussed in the free-piston dynamics section of numerical
Time (ms) results. Coupled hetero-/homogeneous combustion results in
(a) Free-piston velocity higher output power and energy density. Therefore, one of the
main challenges is to achieve sustainable combustion with high
20 combustion efficiency for hydrocarbon fuels; apparently, coupled
hetero-/homogeneous combustion has a great advantage, and the
use of catalyst is also a promising and effective approach for the
Free-piston position (mm)

18 micro-engine. Other challenges include thermal isolation, materi-

als thermal stability, the increase of the compression ratio, and
bearing wear at high free-piston velocities.
16 The larger alkane such as propane is a fuel of particular interest
for micro-scale energy conversion devices because it liquefies at
moderate pressures and room temperature, and is commercially
14
available in compact containers for numerous consumer applica-
tions [50,51]. It has been shown that the catalytic and gas-phase
Homogeneous combustion reactivity of linear alkanes increases with increasing carbon chain
12
Hetero-/homogeneous combustion
length because of weaker C–C bond strengths of higher hydrocar-
0.65 0.70 0.75 0.80 0.85 0.90 0.95 bons [29,52]. In addition, the ignition delay time and ignition tem-
Time (ms) perature decrease with increasing the carbon chain length of linear
(b) Free-piston position alkanes [53]. Furthermore, the larger alkane in the coupled
hetero-/homogeneous combustion mode exhibits significantly
Fig. 3. A comparison of the measured free-piston velocities and positions between stronger homogeneous combustion, especially at higher pressures
the purely homogeneous (black symbols) and coupled hetero-/homogeneous (red [29,54]. Therefore, the homogeneous combustion contribution in
symbols) combustion modes. (For interpretation of the references to colour in this
the coupled hetero-/homogeneous combustion mode increases
figure legend, the reader is referred to the web version of this article.)
with increasing carbon chain length. As a result, the advantage of
Furthermore, the significantly shorter time spent around TDC in coupled hetero-/homogeneous combustion is gradually lost and
the coupled hetero-/homogeneous combustion mode are evident, the difference between the two combustion modes gradually
compared to that in the purely homogeneous combustion mode. decreases with the increasing of carbon chain length of linear alka-
Specific reasons will be discussed in the free-piston dynamics nes. Note that the hydrocarbon fuels in the above discussion is lim-
section. ited to gaseous alkanes because it is difficult to achieve
232 J. Chen et al. / Energy Conversion and Management 119 (2016) 227–238
homogeneous combustion of liquid alkanes in such small-scales
without the aid of other external devices.
Additionally, many fundamental and application studies have
been carried out for hydrogen enrichment combustion in HCCI
engines, using small amounts of hydrogen as additive to conven-
tional hydrocarbon fuels [55,56]. Previous studies have shown that
the hydrogen enrichment can not only improve ignition and com-
bustion capabilities and engine efficiency, but also reduce emis-
sions [57]. In addition, the self-ignition nature of hydrogen–air
mixtures over platinum under very fuel-lean conditions offers an
opportunity to self-ignite hydrocarbons with the assistance of
small amounts of hydrogen [58], which can be reformed from
hydrocarbons and is subsequently used for startup [59]. Therefore,
it might be worthwhile to explore the feasibility of applying the
technique of hydrogen enrichment or hydrogen assisted self-
ignition of hydrocarbons to free-piston micro-engines, and, if feasi-
ble, to thoroughly study their combustion characteristics and
engine performance in the purely homogeneous and coupled
hetero-/homogeneous combustion modes.
4. Numerical
In general, the test-rig of single-shot hetero-/homogeneous
combustion experiments is too small for typical apparatuses and
instruments to be installed. Therefore, a numerical model must
be relied upon to illustrate the experimental results, to explore
the hetero-/homogeneous combustion characteristics, and to gain
physical insight.
4.1. Physical model and dynamic mesh
The characteristic length of the combustion chamber and the
reacting gas flow path in micro-engines is still sufficiently larger
than the molecular mean-free path of the air and other gases flow-
ing through the system. Hence, the fluid can be reasonably consid-
ered as continuous and the Navier–Stokes equations are applicable
to the present work [27]. Simulations were performed using the
initial parameters of the micro-engine used in the single-shot
hetero-/homogeneous combustion experiments. In the physical
system depicted in Fig. 4, the upper surface of the free-piston
and the lower surface of the end-plug were coated with Pt/c-
Al2O3 catalysts. The computational domain is a two-dimensional
channel of initial length L = 20 mm and cylinder bore B = 3 mm,
coupled with two gaps (between cylinder walls and free-piston)
of length Lg = 6 mm (corresponding to the free-piston length) and
Fig. 4. Schematic of the computational domain (not to scale). The upper surface of
the free-piston and the lower surface of the end-plug were coated with Pt/c-Al2O3
catalysts. L is the initial distance between free-piston and end-plug (chamber
length); B is cylinder bore; x is Cartesian coordinate; mp is piston mass; Dp is piston
diameter; T is temperature; T1 is ambient temperature; V is volume; p is pressure;
p1 is ambient pressure; t is specific volume; Yi is the mass fraction of species i; Lg is
gap length; Dg is gap diameter.
various diameters Dg. The thickness of the catalytic layer is several
orders of magnitude smaller than the combustion chamber, so the
effect of the catalyst volume can be neglected. A two-dimensional,
viscous, compressible, and transient numerical model was used
that includes detailed hetero-/homogeneous chemistry and trans-
port, leakage, heat transfer, dynamic mesh, and thermodynamic–
dynamic balance.
The initial flow domain (t = 0 s; 0 6 x 6 L; 0 6 y 6 Dp) is dis-
cretized with an orthogonal staggered mesh of 200
40 points
(in the axial and transverse directions, respectively, with finer
spacing toward the FeCr-alloy walls coated with Pt/c-Al2O3 cata-
lysts, i.e., the upper surface of the free-piston and the lower surface
of the end-plug), which are sufficient to produce a grid-
independent solution.
The distinguishing feature of this free-piston micro-engine is a
mechanically unconstrained piston. Consequently, reciprocating
motion is the result of gas pressure acting on the free-piston, in
contrast to a traditional crankshaft-equipped engine. Free-piston
motion is the result of a ‘‘thermodynamic–dynamic balance”
[60]. Experimentations have demonstrated that the oscillation fre-
quency of free-piston is determined by the average gas pressures
and the free-piston mass [61]. In order to solve the coupling prob-
lem of free-piston movement and hetero-/homogeneous combus-
tion, we developed a dynamic mesh program of free-piston based
on CFD software: FLUENTÒ Release 6.3 [62], coupled with CHEM-
KIN [63]. The dynamic mesh model of six DOF (degree of freedom)
and the mesh method of layering were used; and a UDF (User-
Defined Function) subroutine written in the C programming lan-
guage was used to define the six DOF parameters and to specify
the boundary conditions of the free-piston motion. CFD-based six
DOF simulations are based on evaluating the conservation equa-
tions for the fluid, such as the unsteady Reynolds-averaged
Navier–Stokes equations (RANS), along with the motion equations
for the solid [64,65]. This approach can significantly simplify the
dynamic mesh settings, in spite of the increased computational
cost.
Different gap sizes between cylinder walls and free-piston was
used to examine the leakage effect. Compared with the experimen-
tal results, it is found that the gap size of 6 lm is appropriate. Finer
mesh was used to discretize the gap in the transverse direction.
4.2. Mathematical model
The gas-phase in the combustion chamber and the two gaps
between cylinder walls and free-piston is described by the conser-
vation equations for mass, energy, and momentum species.
Therein, compressibility effects caused by density variations and
heat release are fully considered; while acoustic waves are
neglected, which allows for longer integration times steps. In addi-
tion, the mathematical model is developed by employing a force
balance to the free-piston.
Leakage, the gas escaping through the gap between cylinder
walls and free-piston from combustion chamber, is described by
two-dimensional transient-state compressible flow; this strategy
is a necessity because we found that the pressure drop occurs in
this gap is appreciable. A mass balance is employed to the control
volume:
1 dV V dt
 _ ¼0
þm ð1Þ
t dt t2 dt
_ is the mass flow rate of gas escaping
where t is the time, and m
through the gap between cylinder walls and free-piston from com-
bustion chamber. The gap dimension is defined by
B  Dp
Dg ¼ ð2Þ
2
 J. Chen et al. / Energy Conversion and Management 119 (2016) 227–238
To account for heterogeneous chemistry, the surface species
coverage equations are solved:
@hk s_ k
¼ k ¼ 1;    ; Ns
@t C 9433 and 6743, respectively. Additionally, in most cases, the Rey-
where hk is the coverage of the kth surface species; C is the surface
site density of the catalyst, s_ k is the molar catalytic production rate
of the kth surface species, and Ns is the number of surface species.
At the gas–solid interface, the species boundary conditions are
given by
h i
n  qY i ð~
~ Vi þ ~
uÞ ¼ As s_ i W i i ¼ 1;    ; N g
Wall
where ~n is the unit inward-pointing normal vector to the surface, q
is the gas density, Yi, ~
V i , s_ i , and Wi are the mass fraction, diffusion
velocity vector, molar catalytic production rate, and molecular
weight of the ith gas-phase species, respectively; Ng is the number
of gaseous species; and As is the ratio of the catalytically active area
to the geometrical channel area, and is a crucial parameter in the
modeling of reactors coated with technical catalysts and can be
assessed in practice via hydrogen or carbon monoxide chemisorp-
tion experiments [66,67]. Finally, ~ u is the Stefan flow velocity, and
the Stefan flow occurs when there is a net mass flux between the
surface and the gas. The induced Stefan flow velocity is given by
1 X
N
g
~
n ~
u¼ As s_ i W
q i¼1
The species diffusion velocity ~V i is computed in both reactor
models using mixture-average plus thermal diffusion for the light
species, and are related to species and temperature gradients by:
1 X
N
~
g
DT rT
Vi ¼  W j Dij rX j  i i ¼ 1;    ; Ng
XiW j – i qY i T
 is the
where Xi is the mole fraction of the ith gas-phase species; W
average molecular weight; Dij is the ordinary multicomponent dif-
fusion coefficient matrix; and DTi are the thermal diffusion
coefficients.
4.3. Chemical kinetics
The oxidation of methane on Pt/c-Al2O3 was described by the
elementary heterogeneous scheme of Deutschmann et al. [68] with
a surface site density of C = 2.7
109 mol/cm2 for the Pt/c-Al2O3
catalyst in this work, containing 9 gaseous and 11 surface species
participating in 24 elementary reactions. The homogeneous chem-
istry was described by the detailed mechanism of Warnatz et al.
[69], containing 26 gaseous species participating in 108 elemen-
tary reactions. The aptness of this detailed mechanism has been
demonstrated at high pressures [70]. A modified homogeneous
scheme of Warnatz et al. [69] in the kinetic parameters of the
chain-branching elementary reaction CHO + M , CO + H + M has
been proposed [71] to reproduce homogeneous ignition measure-
ments at lower pressures, which was further supported by kinetic
measurements [72]. The scheme of Warnatz with the aforemen-
tioned modification was used in the ensuing computations. In
addition, surface and gaseous thermodynamic data were included
in the provided schemes. Finally, surface and gas-phase reaction
rates were evaluated with Surface-CHEMKIN [73] and CHEMKIN
[63], respectively; while transport properties are calculated from
the CHEMKIN database [74].
233
4.4. Computation scheme and solution algorithm
In the coupled hetero-/homogeneous and purely homogeneous
combustion mode, the computed maximum Reynolds numbers is
ð3Þ
nolds numbers are above laminar values during the whole cycle.
Therefore, a turbulent viscous flow is considered. RANS simula-
tions were performed in this work. Re-normalization group
(RNG) k–e model is employed as the turbulence model [75], to
account for the effects of smaller scales of motion. At the outer sur-
face of the solid walls, heat losses to the surroundings are calcu-
lated, and both natural convection and thermal radiation are
ð4Þ considered. The fluid thermal conductivity, specific heat, and vis-
cosity are calculated from a mass fraction weighted average of spe-
cies properties, along with the species specific heat is calculated
using a piecewise polynomial fit of temperature. The pressure-
velocity coupling is discretized using the ‘‘PISO (pressure-implicit
with splitting of operators)” method. The energy, species, and
momentum equations are discretized using a two-order upwind
approximation. The transient solid energy equation is solved with
a quadratic backward time discretization and a second-order accu-
rate, fully implicit scheme [2,42,67,76]. The integration time step is
fixed, and the step was Dt = 5.0
108 s used in this work. A solu-
tion for the coupled solid phase and flow equations is obtained at
each time step iteratively: convergence is reached when the solid
temperatures at all axial positions do not vary between successive
iterations by more than 105 K, resulting in up to 400 iterations per
time step.
ð5Þ
All computations were performed on a Beowulf cluster consist-
ing of 8
Xeon E7-8830 processors at 2.13 GHz and 128 GB of
RAM. The MPI (message passing interface) was used to transmit
information between nodes when the parallel processing was used.
In addition, computations were very demanding: the calculation
time of each simulation varied between several days and two
weeks on a single CPU when full hetero-/homogeneous chemistry
ð6Þ was included, depending on the initial guess and the difficulty of
the problem. These CPU time requirements in the system running
clearly illustrate why transient multidimensional models with
dynamic mesh and realistic hetero-/homogeneous chemistry have
not so far been presented in the literature.
5. Numerical results and discussion
5.1. Model validation
The single-shot hetero-/homogeneous combustion model is val-
idated by simulating the experiment that yielded the image
sequence presented in Fig. 2. The model parameters and their ini-
tial values are given in the experimental section. In order to ensure
the independence of the solution from the mesh density and the
accuracy of the predictions, a grid sensitivity analysis was per-
formed. Validation of transient model was carried out by compar-
ing the computed free-piston velocities and positions with the
experimental results. Experimental and predicted results are com-
pared in Fig. 5, indicating that the simulated (blue lines) results are
essentially in agreement with experiments (red symbols). The
transient model is extremely sensitive to the cylinder bore and
the initial velocity of free-piston, resulting a small amount of error.
Other thermodynamic parameters (such as average temperature
and fuel conversion) at different initial velocities of free-piston
have been tested, showing the same essential agreements, but
results are not reported here for brevity. Further experimental test-
ing for model validation has been carried out for a broader range of
conditions but is also not reported here. These confirm the reason-
able accuracy of the numerical scheme adopted in the present
work.
234
40
Free-piston velocity (m/s)
20
0 Predicted two-dimensional temperature distributions at the
-20
-40
-60
0.4 0.6 0.8
Time (ms)
(a) Free-piston velocity
20
Free-piston position (mm)
18
16
14
12
0.65 0.70 0.75 0.80 0.85
Time (ms)
(b) Free-piston position
Fig. 5. Predicted (blue lines) and measured (red symbols) velocities and positions of
the free-piston in the coupled hetero-/homogeneous combustion mode. (For
interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)
5.2. Combustion characteristics
To facilitate performance comparisons between the different
combustion modes, the same simulation conditions were chosen.
Comparisons are elaborated for the free-piston initial velocity of
30 m/s. In the coupled hetero-/homogeneous combustion mode,
two-dimensional temperature contours acquired from transient
simulation at various points in time are provided in Fig. 6; pre-
dicted profile of average temperature as a function of the corre-
sponding time is shown in Fig. 7. The corresponding temperature
contours and average temperature profile in the purely homoge-
neous combustion mode are also shown in the same figures. Igni-
tion is indicated by a sudden temperature rise. Clearly, the coupled
hetero-/homogeneous combustion mode shows a completely dif-
ferent ignition behavior from that of the purely homogeneous case.
The heterogeneous reaction initially causes a more rapid tempera-
ture rise and increases the overall reaction rate, resulting in a
higher fuel conversion and a shorter ignition delay compared to
the purely homogeneous case. Peak temperature and ignition tim-
ing are significantly affected by heterogeneous reaction with lower
activation energy. The catalyst opens new reaction pathways,
extending the range of reactivity toward lower temperatures as
well as accelerating reactions at higher temperatures. In addition,
in the coupled hetero-/homogeneous combustion mode, mass
and heat loss causes exhaust temperatures to decrease faster dur-
ing the expansion stroke.
J. Chen et al. / Energy Conversion and Management 119 (2016) 227–238
Since the physical parameter values between different combus-
Simulation tion modes has large difference, different scales are adopted in the
Experiment
following contour plots to conveniently express the slight differ-
ence in the physical parameter distribution caused by heteroge-
neous reaction.
TDC in the coupled hetero-/homogeneous combustion mode are
provided in Fig. 8; the corresponding temperature distributions
in the purely homogeneous case are also shown in the same figure.
In the purely homogeneous combustion mode, the temperatures in
the fluid region are almost identical and within the maximum dif-
ference of 8 K. However, in the coupled hetero-/homogeneous
combustion mode, the highest temperature is found on the cat-
1.0 1.2
alytic walls, i.e., the upper surface of the free-piston and the lower
surface of the end-plug; it is also significantly higher than that in
the purely homogeneous case. In addition, coupled hetero-/
homogeneous combustion results in lower compression ratios than
the purely homogeneous case (TDC: 0.353 vs 0.342 mm) and less
Simulation
Experiment time spent around TDC, as will be discussed in the free-piston
dynamics section.
To further examine the role of catalyst on the micro-engine per-
formance, Fig. 9 shows the predicted two-dimensional distribu-
tions of temperature and methane mass fraction in different
combustion modes at the end of expansion. After the methane
has been consumed, the reaction stops. There are no significant
temperature and species gradients within this region. In the cou-
pled hetero-/homogeneous combustion mode, combustion occurs
very rapidly, consuming most of the methane in an extremely
short time; complete conversion is achieved, and a higher temper-
0.90 0.95 ature is observed mainly because of the heterogeneous reaction.
However, in the purely homogeneous combustion mode, lower
exhaust temperatures caused by incomplete combustion (only
29.8% conversion).
The catalytic effect on the in-cylinder pressure is shown in
Fig. 10. At first, in both combustion modes, the in-cylinder pressure
grows because of compression. When the free-piston approaches
the TDC, the ignited in-cylinder charge results in a sharp increase
in the in-cylinder pressure, irrespective of combustion mode. Dur-
ing the expansion stroke, the in-cylinder pressure is sharply
reduced because of the increase in the volume as well as the heat
and mass loss. Coupled hetero-/homogeneous combustion results
in significantly higher in-cylinder pressure and, consequently,
higher mass loss rate; it also leads to an increase in the in-
cylinder volume due to the higher free-piston velocity during the
expansion stroke. These cause product pressure to decrease faster
in the coupled hetero-/homogeneous combustion mode during the
expansion stroke, even significantly lower than that in the purely
homogeneous case in most of the expansion process. In addition,
heterogeneous reaction results in earlier ignition and higher peak
pressure.
Additionally, the rate of gaseous reactions increases with pres-
sure, which is equivalent to an increase in concentration of the gas.
As the pressure is increased, the ignition for both coupled hetero-/
homogeneous and purely homogeneous combustion modes shift
toward lower temperatures and shorter ignition delay times (i.e.,
earlier ignition). However, this decrease in ignition delay time is
more pronounced in the coupled hetero-/homogeneous than in
the purely homogeneous combustion mode [77], resulting in a
much earlier ignition.
The predicted free-piston dynamics including the free-piston
velocity and position in various combustion modes are compared
in Fig. 11. As observed, the position traces of free-piston in various
combustion modes are identical during the compression stroke,
but differ in the expansion stroke. Coupled hetero-/homogeneous
combustion results in the significantly shorter time spent around
TDC. For the compression path, two combustion modes give similar
 J. Chen et al. / Energy Conversion and Management 119 (2016) 227–238 235

Fig. 6. Predicted two-dimensional temperature contours at various points in time in different combustion modes.

2800 predictions of the free-piston dynamics, but the expansion paths

Homogeneous
vary greatly. This occurs because different fuel conversion efficien-
combustion
Average temperature (K)

Hetero-/homogeneous cies and mass losses occur when the free-piston reverses direction.
2300
combustion When the free-piston approaches the TDC, the free-piston velocity
varies rapidly. During the expansion stroke, coupled hetero-/
1800 homogeneous combustion results in higher free-piston velocity,
while the free-piston velocity in the purely homogeneous combus-
1300 tion mode is only slightly increased as a result of lower combustion
efficiency as well as heat and mass loss. However, during the com-
pression stroke, there is only marginal difference between the free-
800
piston dynamics obtained from different combustion modes.
Comparing the predicted free-piston dynamics curves in Fig. 11
300 yields an interesting insight: Coupled hetero-/homogeneous com-
0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3
bustion apparently reduces the mass loss. Intuitively, one would
Time (ms)
expect the opposite to be true because coupled hetero-/
Fig. 7. Predicted profiles of average temperature in different combustion modes. homogeneous combustion results in significantly higher in-
cylinder pressure and temperature; consequently, the blow-by
mass flow rate and the in-cylinder pressure gradient should
increase. Recognizing that the compression ratio varies is the key
to understanding this counterintuitive result. Coupled hetero-/
homogeneous combustion results in higher in-cylinder pressures
than the purely homogeneous case. This additional pressure rise
causes the free-piston to decelerate faster than the purely homoge-
neous case, resulting in a lower compression ratio and less time
spent around TDC. Consequently, coupled hetero-/homogeneous
combustion results in less time spent at large pressures and max-
imum compression ratio, which in turn yields less mass loss than
the purely homogeneous case. On the contrary, in a traditional
crank engine that the compression ratio is fixed, the residence
times are identical and, consequently, coupled hetero-/
homogeneous combustion maximizes the mass loss. In addition,
since the mass loss significantly reduces the output power and
engine performance [10,78], mass loss is a serious concern about
any micro-engine design using gas compression. Therefore, miti-
gating the leakage problem or designing an effective seal is an
essential step toward developing a high power density micro-
engine.
Fig. 8. Predicted two-dimensional temperature distributions at the TDC in different In order to demonstrate the effect of mass loss, the predicted per-
combustion modes. centages of mass loss in various combustion modes are compared in
236 J. Chen et al. / Energy Conversion and Management 119 (2016) 227–238

40
Homogeneous
combustion

Free-piston velocity (m/s)

20 Hetero-/homogeneous
combustion

-20

-40

-60
0.4 0.6 0.8 1.0 1.2
Time (ms)
(a) Free-piston velocity

20

Free-piston position (mm)

18

16

14

12 Homogeneous
combustion
Hetero-/homogeneous
10 combustion

0.65 0.70 0.75 0.80 0.85 0.90 0.95

Time (ms)
(b) Free-piston position
Fig. 11. A comparison of the predicted free-piston velocities and positions between
the purely homogeneous (black lines) and coupled hetero-/homogeneous (red lines)
combustion modes. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)

25 Homogeneous
combustion
Mass loss percentage (%)

Hetero-/homogeneous
20 combustion
Hetero-/homogeneous
combustion
15
+adiabatic walls
Fig. 9. Predicted two-dimensional distributions of temperature and methane mass
fraction in different combustion modes at the end of expansion. 10

5
50
Homogeneous
combustion 0
0.0 0.3 0.6 0.9 1.2 1.5
Average pressure (MPa)

40 Hetero-/homogeneous
combustion Time (ms)

Fig. 12. A comparison of the predicted percentages of mass loss between the purely
30
homogeneous and coupled hetero-/homogeneous combustion modes.

20
Fig. 12. Similarly to the free-piston dynamics, the mass losses are
identical during the compression stroke, but differ in the expansion
10
stroke. Furthermore, in both combustion modes, most of the mass
loss occurs when the free-piston reverses direction. Therefore, the
0 charge mass essentially undergoes a step change at the maximum
0.65 0.70 0.75 0.80 0.85 0.90 0.95
compression ratio. In addition, coupled hetero-/homogeneous com-
Time (ms)
bustion can significantly reduce the expansion time as a result of its
Fig. 10. Predicted profiles of average pressure in different combustion modes. higher free-piston velocity during the expansion stroke. Therefore,
 J. Chen et al. / Energy Conversion and Management 119 (2016) 227–238
as shown in Figs. 11 and 12, coupled hetero-/homogeneous combus-
tion reduces the mass loss as a result of its lower compression ratio
as well as less time spent around TDC and during the expansion
stroke.
It has been demonstrated that the heat losses are more sig-
nificant for small-scale engines, for which the cylinder surface
to the cylinder volume ratio is higher [10]. Here, the effect of
the heat losses through the cylinder walls is presented and its
result is also shown in Fig. 12. Compared to heat loss, adiabatic
walls can not only improve the combustion efficiency, in-
cylinder temperature and pressure, output power and energy
density, but also reduce the mass loss because of its lower com-
pression ratio and less time spent around TDC and during the
expansion stroke, as the coupled hetero-/homogeneous combus-
tion compared to the purely homogeneous case. Therefore, heat
losses result in lower in-cylinder temperature and pressure, but
higher amount of charge is leaked. Heat losses result in higher
mass losses. It is interesting to note that this finding is opposite
to the previous work of Sher et al. [10], who used a small two-
stroke HCCI engine that its compression ratio was fixed. At
higher initial velocities of free-piston, the time interval that is
allowed for the mass and heat losses is short, and consequently
their effect is limited.
6. Conclusions
The hetero-/homogenous combustion-based HCCI of fuel–lean
methane–air mixtures over alumina-supported platinum catalysts
in free-piston micro-engines was investigated experimentally and
numerically. Single-shot experiments were carried out in free-
piston micro-engines; a two-dimensional transient model with
detailed chemistry and transport simulated the single-shot exper-
iments to gain physical insight and to explore the hetero-/
homogeneous combustion characteristics. Specific results are sum-
marized as follows:
 In micro-scales, traditional engine combustion schemes are
generally infeasible because of quenching effects and insuffi-
cient residence times, whereas the coupled hetero-/
homogeneous combustion mode-based HCCI is a promising
alternative.
 Both purely homogeneous and coupled hetero-/homogeneous
combustion of methane–air mixtures in a narrow cylinder with
a diameter of 3 mm and a height of approximately 0.3 mm is
possible, and methane at equivalence ratio of 0.8 is demon-
strated in both cases.
 A two-dimensional transient numerical model which couples
detailed hetero-/homogeneous chemistry and transport, leak-
age, heat transfer, and free-piston motion is developed, as well
as used to model the single-shot experiments in free-piston
micro-engines and to explore the hetero-/homogenous
combustion-based HCCI. This model approximates the single-
shot process, and the model predictions are generally consistent
with the experimental data.
 Heterogeneous reactions cause earlier ignition, reduce the igni-
tion delay time, and consequently extend the reaction time.
These are very favorable for the micro-device.
 Coupled hetero-/homogeneous combustion results in more
complete fuel conversion and higher temperature and energy
release, as well as significantly improves the output power
and energy density.
 Coupled hetero-/homogeneous combustion reduces the mass
loss because of its lower compression ratio as well as less time
spent around TDC and during the expansion stroke, whereas
heat losses result in higher mass losses.
237
Acknowledgment
This work was supported by the National Natural Science Foun-
dation of China (No. 51506048).
References
[1] Akhtar S, Kurnia JC, Shamim T. A three-dimensional computational model of
H2–air premixed combustion in non-circular micro-channels for a thermo-
photovoltaic (TPV) application. Appl Energy August 2015;152(15):47–57.
[2] Brambilla A, Frouzakis CE, Mantzaras J, Tomboulides A, Kerkemeier S,
Boulouchos K. Detailed transient numerical simulation of H2/air hetero-/
homogeneous combustion in platinum-coated channels with conjugate heat
transfer. Combust Flame 2014;161(10):2692–707.
[3] Khu K, Jiang L, Markvart T. Effect of finite heat input on the power performance
of micro heat engines. Energy 2011;36(5):2686–92.
[4] Cho J-H, Lin CS, Richards CD, Richards RF, Ahn J, Ronney PD. Demonstration of
an external combustion micro-heat engine. Proc Combust Inst 2009;32
(2):3099–105.
[5] Zimont VL. Theoretical study of self-ignition and quenching limits in a catalytic
micro-structured burner and their sensitivity analysis. Chem Eng Sci 2015;134
(29):800–12.
[6] Wierzbicki TA, Lee IC, Gupta AK. Combustion of propane with Pt and Rh
catalysts in a meso-scale heat recirculating combustor. Appl Energy October
2014;130(1):350–6.
[7] Specchia S, Tacchino S, Specchia V. Facing the catalytic combustion of CH4/H2
mixtures into monoliths. Chem Eng J 2011;167(2–3):622–33.
[8] Aichlmayr HT, Kittelson DB, Zachariah MR. Miniature free-piston
homogeneous charge compression ignition engine-compressor concept-Part
I: performance estimation and design considerations unique to small
dimensions. Chem Eng Sci 2002;57(19):4161–71.
[9] Aichlmayr HT, Kittelson DB, Zachariah MR. Miniature free-piston
homogeneous charge compression ignition engine-compressor concept-Part
II: modeling HCCI combustion in small scales with detailed homogeneous gas
phase chemical kinetics. Chem Eng Sci 2002;57(19):4173–86.
[10] Sher I, Levinzon-Sher D, Sher E. Miniaturization limitations of HCCI internal
combustion engines. Appl Therm Eng 2009;29(2–3):400–11.
[11] Sher E, Sher I. Theoretical limits of scaling-down internal combustion engines.
Chem Eng Sci 2011;66(3):260–7.
[12] Barari G, Sarathy SM, Vasu SS. Improved combustion kinetic model and HCCI
engine simulations of di-isopropyl ketone ignition. Fuel 2016;164(15):141–50.
[13] Bahlouli K, Atikol U, Saray RK, Mohammadi V. A reduced mechanism for
predicting the ignition timing of a fuel blend of natural-gas and n-heptane in
HCCI engine. Energy Convers Manage 2014;79:85–96.
[14] Mikalsen R, Roskilly AP. A computational study of free-piston diesel engine
combustion. Appl Energy 2009;86(7–8):1136–43.
[15] Mikalsen R, Roskilly AP. The design and simulation of a two-stroke free-piston
compression ignition engine for electrical power generation. Appl Therm Eng
2008;28(5–6):589–600.
[16] Kim J, Bae C, Kim G. Simulation on the effect of the combustion parameters on
the piston dynamics and engine performance using the Wiebe function in a
free piston engine. Appl Energy 2013;107:446–55.
[17] Hung NB, Lim O, Iida N. The effects of key parameters on the transition from SI
combustion to HCCI combustion in a two-stroke free piston linear engine. Appl
Energy 2015;137(1):385–401.
[18] Jia B, Zuo Z, Tian G, Feng H, Roskilly AP. Development and validation of a free-
piston engine generator numerical model. Energy Convers Manage
2015;91:333–41.
[19] Zhang S, Zhao C, Zhao Z, Ma F. Combustion characteristics analysis of hydraulic
free piston diesel engine. Appl Energy 2015;160(15):761–8.
[20] Maigaard P, Mauss F, Kraft M. Homogeneous charge compression ignition
engine: a simulation study on the effects of inhomogeneities. J Eng Gas
Turbines Power 2003;125(2):466–71.
[21] Neshat E, Saray RK. An optimized chemical kinetic mechanism for HCCI
combustion of PRFs using multi-zone model and genetic algorithm. Energy
Convers Manage 2015;92(1):172–83.
[22] Godiño JAV, García MT, Aguilar FJJ-E, Trujillo EC. Numerical study of HCCI
combustion fueled with diesel oil using a multizone model approach. Energy
Convers Manage 2015;89(1):885–95.
[23] Mantzaras J, Benz P. An asymptotic and numerical investigation of
homogeneous ignition in catalytically stabilized channel flow combustion.
Combust Flame 1999;119(4):455–72.
[24] Mantzaras J, Appel C. Effects of finite rate heterogeneous kinetics on
homogeneous ignition in catalytically stabilized channel flow combustion.
Combust Flame 2002;130(4):336–51.
[25] Sharma TK, Rao GAP, Murthy KM. Effective reduction of NOx emissions of a
HCCI (Homogeneous charge compression ignition) engine by enhanced rate of
heat transfer under varying conditions of operation. Energy 2015;93(Part
2):2102–15.
[26] Bedoya ID, Saxena S, Cadavid FJ, Dibble RW, Wissink M. Experimental study of
biogas combustion in an HCCI engine for power generation with high
indicated efficiency and ultra-low NOx emissions. Energy Convers Manage
2012;53(1):154–62.
238 J. Chen et al. / Energy Conversion and Management 119 (2016) 227–238
[27] Wan J, Fan A, Liu Y, Yao H, Liu W, Gou X, et al. Experimental investigation and
numerical analysis on flame stabilization of CH4/air mixture in a mesoscale
channel with wall cavities. Combust Flame 2015;162(4):1035–45.
[28] Yadav S, Yamasani P, Kumar S. Experimental studies on a micro power
generator using thermo-electric modules mounted on a micro-combustor.
Energy Convers Manage 2015;99(15):1–7.
[29] Karagiannidis S, Mantzaras J, Boulouchos K. Stability of hetero-/homogeneous
combustion in propane- and methane-fueled catalytic microreactors: channel
confinement and molecular transport effects. Proc Combust Inst 2011;33
(2):3241–9.
[30] Pizza G, Mantzaras J, Frouzakis CE, Tomboulides AG, Boulouchos K.
Suppression of combustion instabilities of premixed hydrogen/air flames in
microchannels using heterogeneous reactions. Proc Combust Inst 2009;32
(2):3051–8.
[31] Baigmohammadi M, Tabejamaat S, Zarvandi J. Numerical study of the behavior
of methane–hydrogen/air pre-mixed flame in a micro reactor equipped with
catalytic segmented bluff body. Energy 2015;85(1):117–44.
[32] Tolmachoff ED, Allmon W, Waits CM. Analysis of a high throughput n-
dodecane fueled heterogeneous/homogeneous parallel plate microreactor for
portable power conversion. Appl Energy September 2014;128(1):111–8.
[33] Fanaee SA, Esfahani JA. Two-dimensional analytical model of flame
characteristic in catalytic micro-combustors for a hydrogen–air mixture. Int
J Hydrogen Energy 2014;39(9):4600–10.
[34] Gomez A, Berry JJ, Roychoudhury S, Coriton B, Huth J. From jet fuel to electric
power using a mesoscale, efficient Stirling cycle. Proc Combust Inst 2007;31
(2):3251–9.
[35] Tolmachoff ED, Booth AD, Lee IC, Allmon WR, Waits CM. Modeling and
experimental analysis of n-dodecane oxidation in platinum-coated channels.
Combust Flame 2015;162(10):3674–80.
[36] Merotto L, Fanciulli C, Dondè R, De Iuliis S. Study of a thermoelectric generator
based on a catalytic premixed meso-scale combustor. Appl Energy 2016;162
(15):346–53.
[37] Schultze M, Mantzaras J, Grygier F, Bombach R. Hetero-/homogeneous
combustion of syngas mixtures over platinum at fuel-rich stoichiometries
and pressures up to 14 bar. Proc Combust Inst 2015;35(2):2223–31.
[38] Yan Y, Pan W, Zhang L, Tang W, Chen Y, Li L. Numerical study of the
geometrical parameters on CH4/air premixed combustion in heat recirculation
micro-combustor. Fuel 2015;159(1):45–51.
[39] Maestri M, Beretta A, Faravelli T, Groppi G, Tronconi E, Vlachos DG. Two-
dimensional detailed modeling of fuel-rich H2 combustion over Rh/Al2O3
catalyst. Chem Eng Sci 2008;63(10):2657–69.
[40] Cao C, Zhang N, Chen X, Cheng Y. A comparative study of Rh and Ni coated
microchannel reactor for steam methane reforming using CFD with detailed
chemistry. Chem Eng Sci 2015;137(1):276–86.
[41] Grimm M, Mazumder S. Numerical investigation of wall heat conduction
effects on catalytic combustion in split and continuous monolith tubes.
Comput Chem Eng 2008;32(3):552–60.
[42] Michelon N, Mantzaras J, Canu P. Transient simulation of the combustion of
fuel-lean hydrogen/air mixtures in platinum-coated channels. Combust Theor
Model 2015;19(4):514–48.
[43] Robbins FA, Zhu H, Jackson GS. Transient modeling of combined catalytic
combustion/CH4 steam reforming. Catal Today 2003;83(1–4):141–56.
[44] Kirkpatrick MJ, Odic E, Zinola S, Lavy J. Plasma assisted heterogeneous catalytic
oxidation: HCCI Diesel engine investigations. Appl Catal B 2012;117–118
(18):1–9.
[45] Zhang C, Li K, Sun Z. Modeling of piston trajectory-based HCCI combustion
enabled by a free piston engine. Appl Energy 2015;139(1):313–26.
[46] Roestenberg T, Glushenkov MJ, Kronberg AE, Verbeek AA, vd Meer ThH.
Experimental study and simulation of syngas generation from methane in the
pulsed compression reactor. Fuel 2011;90(5):1875–83.
[47] Roestenberg T, Glushenkov MJ, Kronberg AE, vd Meer ThH. On the
controllability and run-away possibility of a totally free piston, pulsed
compression reactor. Chem Eng Sci 2010;65(16):4916–22.
[48] Schneider A, Mantzaras J, Jansohn P. Experimental and numerical investigation
of the catalytic partial oxidation of CH4/O2 mixtures diluted with H2O and CO2
in a short contact time reactor. Chem Eng Sci 2006;61(14):4634–49.
[49] Eriksson S, Wolf M, Schneider A, Mantzaras J, Raimondi F, Boutonnet M, et al.
Fuel-rich catalytic combustion of methane in zero emissions power generation
processes. Catal Today 2006;117(4):447–53.
[50] Karagiannidis S, Mantzaras J, Bombach R, Schenker S, Boulouchos K.
Experimental and numerical investigation of the hetero-/homogeneous
combustion of lean propane/air mixtures over platinum. Proc Combust Inst
2009;32(2):1947–55.
[51] Smyth SA, Kyritsis DC. Experimental determination of the structure of catalytic
micro-combustion flows over small-scale flat plates for methane and propane
fuel. Combust Flame 2012;159(2):802–16.
[52] Aryafar M, Zaera F. Kinetic study of the catalytic oxidation of alkanes over
nickel, palladium, and platinum foils. Catal Lett 1997;48(3):173–83.
[53] Campbell MF, Davidson DF, Hanson RK. Ignition delay times of very-low-
vapor-pressure biodiesel surrogates behind reflected shock waves. Fuel
2014;126(15):271–81.
[54] Stefanidis GD, Vlachos DG. Controlling homogeneous chemistry in
homogeneous-heterogeneous reactors: application to propane combustion.
Ind Eng Chem Res 2009;48(13):5962–8.
[55] Song H, Song S. Predicting performance of a methane-fueled HCCI engine with
hydrogen addition considering knock resistance. Int J Hydrogen Energy
2015;40(45):15749–59.
[56] Kozlov VE, Chechet IV, Matveev SG, Titova NS, Starik AM. Modeling study of
combustion and pollutant formation in HCCI engine operating on hydrogen
rich fuel blends. Int J Hydrogen Energy 2016;41(5):3689–700.
[57] Guo H, Neill WS. The effect of hydrogen addition on combustion and emission
characteristics of an n-heptane fuelled HCCI engine. Int J Hydrogen Energy
2013;38(26):11429–37.
[58] Norton DG, Vlachos DG. Hydrogen assisted self-ignition of propane/air
mixtures in catalytic microburners. Proc Combust Inst 2005;30(2):2473–80.
[59] Jeon SW, Yoon WJ, Jeong MW, Kim Y. Optimization of a counter-flow
microchannel reactor using hydrogen assisted catalytic combustion for
steam reforming of methane. Int J Hydrogen Energy 2014;39(12):6470–8.
[60] Karabulut H. Dynamic analysis of a free piston Stirling engine working with
closed and open thermodynamic cycles. Renew Energy 2011;36(6):1704–9.
[61] Aichlmayr HT, Kittelson DB, Zachariah MR. Micro-HCCI combustion:
experimental characterization and development of a detailed chemical
kinetic model with coupled piston motion. Combust Flame 2003;135
(3):227–48.
[62] FLUENT. 6.3 user’s guide. Lebanon, New Hampshire: Fluent Inc.; 2006.
[63] Kee RJ, Rupley FM, Meeks E, Miller JA. CHEMKIN-III: a Fortran chemical
kinetics package for the analysis of gas-phase chemical and plasma kinetics,
Report No. SAND96-8216, Technical Report, Sandia National Laboratories;
1996.
[64] Bettle MC, Gerber AG, Watt GD. Using reduced hydrodynamic models to
accelerate the predictor-corrector convergence of implicit 6-DOF URANS
submarine manoeuvring simulations. Comput Fluids 2014;102(10):215–36.
[65] Dunbar AJ, Craven BA, Paterson EG. Development and validation of a tightly
coupled CFD/6-DOF solver for simulating floating offshore wind turbine
platforms. Ocean Eng 2015;110(Part A):98–105.
[66] Eriksson S, Schneider A, Mantzaras J, Wolf M, JärÅs S. Experimental and
numerical investigation of supported rhodium catalysts for partial oxidation of
methane in exhaust gas diluted reaction mixtures. Chem Eng Sci 2007;62
(15):3991–4011.
[67] Schneider A, Mantzaras J, Eriksson S. Ignition and extinction in catalytic partial
oxidation of methane-oxygen mixtures with large H2O and CO2 dilution.
Combust Sci Technol 2008;180(1):89–126.
[68] Deutschmann O, Maier LI, Riedel U, Stroemman AH, Dibble RW. Hydrogen
assisted catalytic combustion of methane on platinum. Catal Today 2000;59
(1–2):141–50.
[69] Westbrook CK, Mizobuchi Y, Poinsot TJ, Smith PJ, Warnatz J. Computational
combustion. Proc Combust Inst 2005;30(1):125–57.
[70] Reinke M, Mantzaras J, Schaeren R, Bombach R, Inauen A, Schenker S. High-
pressure catalytic combustion of methane over platinum: In situ experiments
and detailed numerical predictions. Combust Flame 2004;136(1–2):217–40.
[71] Reinke M, Mantzaras J, Bombach R, Schenker S, Inauen A. Gas phase chemistry
in catalytic combustion of methane/air mixtures over platinum at pressures of
1 to 16 bar. Combust Flame 2005;141(4):448–68.
[72] Reinke M, Mantzaras J, Schaeren R, Bombach R, Inauen A, Schenker S.
Homogeneous ignition of CH4/air and H2O and CO2-diluted CH4/O2 mixtures
over Pt; an experimental and numerical investigation at pressures up to 16
bar. Proc Combust Inst 2005;30(2):2519–27.
[73] Coltrin ME, Kee RJ, Rupley FM, Meeks E. SURFACE CHEMKIN-III: a Fortran
package for analyzing heterogeneous chemical kinetics at a solid-surface/gas-
phase interface, Report No. SAND96-8217, Technical Report. Sandia National
Laboratories; 1996.
[74] Kee RJ, Dixon-Lewis G, Warnat J, Coltrin ME, Miller JA, Moffat HK. A Fortran
computer code package for the evaluation of gas-phase, multicomponent
transport properties, Report No. SAND86-8246B, Technical Report, Sandia
National Laboratories; 1998.
[75] Han ZY, Reitz RD. Turbulence modeling of internal combustion engines using
k–e models. Combust Sci Technol 1995;106:267–95.
[76] Karagiannidis S, Mantzaras J. Numerical investigation on the start-up of
methane-fueled catalytic microreactors. Combust Flame 2010;157
(7):1400–13.
[77] Chattopadhyay S, Veser G. Heterogeneous-homogeneous interactions in
catalytic microchannel reactors. AIChE J 2006;52(6):2217–29.
[78] Mabrouk MT, Kheiri A, Feidt M. Effect of leakage losses on the performance of a
b type Stirling engine. Energy 2015;88:111–7.