Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610

Contents lists available at SciVerse ScienceDirect

Renewable and Sustainable Energy Reviews

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

Modeling HCCI combustion of biofuels: A review

N.P. Komninos ∗ , C.D. Rakopoulos
Internal Combustion Engines Laboratory, Thermal Engineering Department, School of Mechanical Engineering, National Technical University of Athens,
9 Heroon Polytechniou St., Zografou Campus, 15780 Athens, Greece

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

Article history: The ever increasing energy demands coupled with the limited availability of fossil fuels and the detri-
Received 8 November 2011 mental environmental effects resulting from their use, has guided research toward seeking alternative
Accepted 25 November 2011 fuels to gradually substitute conventional ones. Among these, biofuels have received increasing attention
Available online 18 January 2012
due to their attractive features of being renewable in nature and reducing the net CO2 emissions. Biofuels
have been used in conventional diesel and gasoline engines either as neat fuels or as supplements.
Keywords:
Fortunately, a relatively new combustion concept for internal combustion engines, namely homo-
Biofuels
geneous charge compression ignition (HCCI) combustion, has been evolved in parallel to the biofuel
HCCI
Simulation
research. HCCI combustion seems to be able to take advantage of the diverse properties of biofuels, since
Ethanol in this combustion mode ignition is not externally instigated, but relies on the compression and sub-
Hydrogen sequent autoignition of a fuel–air mixture. This fact allows the utilization of different fuels or blends
DME thereof, in order to regulate the ignition point and provide adequate operation under diverse operating
DEE conditions.
Biodiesel This study provides an overview of existing simulation models for the simulation of biofueled HCCI
Methyl butanoate combustion. Simulation models aid and supplement the experimental research conducted on HCCI com-
Methyl decanoate
bustion, providing a fundamental insight into the physicochemical parameters affecting performance and
emissions formation. The simulation models include single-zone models, multi-zone models, probability
based models, and multi-dimensional models in order of complexity. The vast majority of these models
implement chemical kinetics to simulate the combustion process, not only due to the inherent depen-
dence of HCCI combustion on the physicochemical properties of the fuel, but also due to the sometimes
complex chemical structure of the biofuels, which include esters, ethers and alcohols. The reaction paths
for these homologous series are quite different from the conventional hydrocarbons used to simulate
conventional fuels, and provide the ground for current and future research work.
© 2011 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1589
2. Chemical kinetics modeling – reaction pathways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1592
3. Single-zone modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1592
3.1. Hydrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1596
3.2. Alcohols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1596
3.3. Ethers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1597
3.4. Biodiesel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1598
3.5. Fuel blends and additives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1598
4. Multi-zone modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1600
5. Multi-dimensional modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1602
5.1. Direct injection HCCI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1603
5.2. Fully premixed HCCI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1604
6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1607
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1607

∗ Corresponding author. Tel.: +30 210 7721710.

E-mail address: [email protected] (N.P. Komninos).

1364-0321/$ – see front matter © 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.rser.2011.11.026
 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610 1589

Nomenclature
LTR low temperature reactions
B cylinder bore (m) LTHR low temperature heat release
h molar specific enthalpy (J/kmol) MB methyl butanoate
i any zone MD methyl decanoate
k species index MeOH methanol
K number of species MON motor octane number
n moles MRG methanol reformed gas
P pressure (N/m2 ) MSW municipal solid waste
S cylinder height (m) MZ multi-zone
T temperature (K) NOx nitrogen oxides
t thickness (m) NVO negative valve overlap
U internal energy (J) OHC oxygenated (unburned) hydrocarbons
V volume (m3 ) PDF probability density function
z number of zones PI port injection
PRF primary reference fuel
Greek symbols RES renewable energy sources
 air–fuel equivalence ratio (lambda) RON research octane number
 fuel–air equivalence ratio (phi) SRM stochastic reactor model
SI spark ignition
Subscripts SZ single-zone
chem chemical TC turbocharged
cyl cylinder TDC top dead center
HT heat transfer toe tonnes oil equivalent
i zone index UHC unburned hydrocarbons
in inlet or initial
w wall
1. Introduction
Superscripts
in inflow During the last decades two main concerns have arisen in the
out outflow energy generation and transportation sectors, namely the increas-
tr transferred ing energy demand coupled with the limited availability of oil
reserves and the detrimental environmental effect produced by
Abbreviations
their exploitation to deliver energy and motion. Nuclear energy
a after
could be used to gradually displace fossil fuel based energy pro-
b before
duction, but the late accident in the Fukushima nuclear power
BDC bottom dead center
plant has raised many concerns related to the hazards of nuclear
CA crank angle
power plants; it seems that the efforts toward the establishment
CAD crank angle degrees
and the public acceptance of nuclear energy have yielded some-
CO carbon monoxide
what to skepticism. In an effort to gradually displace fossil fuels,
CFD computational fluid dynamics
gross inland energy consumption1 of renewable energy sources
CI compression ignition
(RES) has increased in the European Union (27 countries) in the
CR compression ratio
decade 1999–2009 from over 5% to approximately 9% [1]. How-
DI direct injection
ever, this increase in the share of renewable energy consumption
DTBP di-tertiary butyl peroxide
seems to be mainly due to the reduction in the solid fuel energy
EDC eddy dissipation concept
consumption share, since the nuclear energy consumption share
EGR exhaust gas recirculation
(∼13–14%) and the combined share of petroleum and natural gas
EtOH ethanol
(∼61%) have remained fairly constant (Fig. 1).
EU European union
The use of fossil fuels has increased the energy dependence of
EVC exhaust valve closing
many countries around the world, since oil production is extremely
EVO exhaust valve opening
localized with the Middle East possessing over 55% of the proved
FID flame ionization detector
oil reserves world wide as of 2009 [2,3]. This impacts all countries
HC (unburned) hydrocarbon(s)
and citizens around the world, since the oil prices are reflected
HCCI homogeneous charge compression ignition
on virtually every commodity. Moreover, it creates and sustains
HRR heat release rate (J/deg)
energy insecurity among countries with high energy dependence.
HTR high temperature reaction
Another detrimental effect of the use of fossil fuels is related to
HTHR high temperature heat release
the environmental impact from their use for energy production and
IA injection angle
transportation. The fuel combustion in internal combustion engines
imep indicated mean effective pressure (bar)
produces – among others – CO2 , a major greenhouse gas that is
ISFC indicated specific fuel consumption (g/kWh)
IT injection timing
IVC inlet valve closing
IVO inlet valve opening 1
Gross inland energy consumption refers to the total energy demand of a country
LES large eddy simulation or region. It covers energy consumption by the energy sector itself, distribution
and transformation losses, final energy consumption by end-users and statistical
differences. It does not include energy provided to maritime bunkers [1].
1590 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610

100% production, and by gasification to produce liquid hydrocarbons,

Renewable Energy Sources
biodiesel, bioDME and bio Synthetic Gas. Additionally, algae seem a
90% promising feedstock for the production of second generation biofu-
Solid Fuels
els and especially diesel fuel. Their advantages are their high growth
80%
rates, tolerance to diverse environmental conditions, high CO2 trap-
70% Nuclear Heat ping and fixation ability, high oil content – sometimes exceeding
80% of their dry weight – and the high microalgae yield per hectare,
60% which is almost an order of magnitude higher than the palm oil
yield [7]. Owing to the fact that other non-fuel products can also be
50% Gas obtained from biomass, an integrated approach is being researched
40% to fully exploit the biomass available and produce all possible prod-
ucts from biomass, including biofuels and a platform of chemicals,
30% in biorefinery systems [4,7]. The production of second generation
biofuels is not yet cost effective since there are a lot of technical
20% Total Petroleum Products barriers to be overcome [7].
In view of the attractive attributes of biofuels, efforts have been
10%
made to introduce them into the energy consumption systems. In
0% Fig. 3 the final energy consumption2 of RES in EU is shown for the
year 2009. It can be seen that the main sector in which biofuels –
1999 2001 2003 2005 2007 2009
especially liquid ones – are currently used, is the transport sector
Year
and particularly road transports. Fig. 3 also indicates that liquid bio-
Fig. 1. Gross inland energy consumption share by fuel in the EU (27 countries). Gas: fuels – including bioethanol, biomethanol, biodiesel, bioDME and
almost exclusively natural gas. Renewable energy sources: biomass and renewable other – are currently the gateway for the penetration of RES into
wastes, hydro, geothermal, wind, solar, wave, tide and ocean energy. the transport sector. The share of liquid biofuels in the transport
Data source: Eurostat (online data codes: nrg 100a, nrg101a, nrg 102a, nrg 103a, sector and especially road transports has increased during the last
nrg 104a, nrg 1071a) [1].
decade as shown in Fig. 4.
These data highlight the importance of biofuels and justify the
known to contribute to the global warming effect. The main issue research effort toward their utilization in conventional (CI and SI)
is not the production of CO2 itself, since the combustion of most engines as well as in Homogeneous Charge Compression Ignition
fuels leads to its production, but the fact that in the case of fossil (HCCI) engines. In Diesel engines fuels such as n-butanol, ethanol,
fuels the CO2 emitted cannot be reconverted to the original fuel by biodiesel, DME, either neat or in blends, have been tested and/or
any short-term, natural – and therefore energy neutral – process. simulated [8–16]. Some of these biofuels, such as esters and DME,
In view of the first issues analyzed, i.e. the energy dependence have an inherent tendency to reduce soot formation due to their
of most of the countries, the instability of the oil prices and the chemical structure; the tendency to form soot precursors decreases
increasing demand of fossil fuel energy, alternative fuels have been as the oxygen content in the fuel molecule increases and/or the
sought after. Among those, biofuels have received increasing atten- number of C–C bonds decreases [16]. This is due to the fact that
tion due to their additional attribute of being renewable. The term the oxygen bound carbon atoms tend to form CO or CO2 rather
“biofuel” refers to any fuel that can be produced from biomass, than participate in soot formation reactions [17]. In spark ignition
such as starch, sugars, lignocellulosic material, animal fat, oleagi- engines hydrogen, biogas, ethanol, DME have been tested and/or
nous plants, etc. The main advantage of a biofuel is that it can be simulated [18–21].
reconstructed by natural processes using the CO2 emitted from its Apart from the conventional combustion modes a relatively new
combustion. This is part of the photosynthetic process of the plants, concept has been developing during the last two decades, i.e. homo-
which use sunlight and CO2 to build their cells. Thus, biofuels have geneous charge compression ignition combustion (HCCI). In HCCI
the potential to be CO2 neutral. This is depicted in Fig. 2, in which combustion a mixture of fuel and air is introduced into the combus-
a comparison is made between the biofuel energy system and the tion chamber early during compression or during the intake stroke.
fossil fuel energy system. The mixture is compressed and autoignites at the point where the
Initially, feedstock such as barley, wheat, corn, sugar beat, sugar- local thermo- and fluid-dynamic conditions are favorable [22]. The
cane, etc. was – and still is – used for ethanol production; oleaginous main advantage of the HCCI concept relative to conventional com-
plants such as soybeans, castor beans, rapeseed, Jatropha and palm bustion modes is the significant reduction of NOx emissions and
were used for biodiesel production. However, the feedstock utilized the absence of soot in the exhaust. Moreover, the fast combustion
for the production of these “first generation” biofuels originates usually encountered and the unthrottled operation are prone to
from edible crops. This led to an unfavorable effect related to food yield high indicated thermal efficiency. However, HC and CO emis-
security in terms of both land use and food prices. With over 1/8th sions are usually high due to bulk quenching at low loads, and wall
of the population worldwide undernourished as estimated in 2010 quenching and crevice flows at higher loads [23].
[5,6], using part of the edible crops to produce biofuels did not seem Added to these unfavorable attributes of HCCI is the inability
a socially just choice. Added to the land use was the issue of rising to directly control the combustion phasing in HCCI engines. This is
food prices due to the increased demand, which was instigated by a major obstacle preventing the implementation of the HCCI con-
the alternative use of food commodities for fuel production [2]. cept. HCCI ignition relies heavily on chemical kinetics and largely
The food security issue raised concerns related to the choice of depends on the fuel used. In an effort to control ignition and the
biofuel production routes and shifted research toward biofuel pro- combustion rate, thereby expanding the HCCI operation over a wide
duction from non-edible feedstock, mainly lignocellulosic material range of load–engine speed, various means have been used, such
or “plant biomass”. The term “second generation” biofuels is used as variable compression ratio, variable valve timing, variable inlet
to depict this shift in the feedstock used, but not the final fuels
produced. The lignocellulosic material is abundant in plants and
mainly consists of polysaccharides [7]; biofuels can be obtained 2
Final energy consumption is the total energy consumed by the end-users, such
from this feedstock by hydrolysis and fermentation, for ethanol as industry, transports, households, services, etc. [1].
 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610 1591

Fig. 2. Comparison of bioenergy and fossil fuel energy systems [4].

temperature, internal or external EGR, direct fuel injection, etc. [23]. conditions [23–25]. The main concept is that fuels with different
A promising method for the combustion control seems to be the autoignition tendencies can be blended at varying proportions to
use of alternative fuels and fuel blends, depending on the charac- regulate the ignition point at various load–speed regions [26]. This
teristics of the engine (compression ratio, etc.) and the operating is a fortunate evolution since the fuels used in the blends may

RES Wood & Wood Waste Biogas Biofuels MSW (ren.)

45000 Transport Sector Road Transport Sector
Final Energy Consumpon in 2009

40000 4,5%
Final Energy Consumpon

35000 4,0%
3,5%
of liquid biofuels

30000
3,0%
(ktoe)

25000
2,5%
20000 2,0%
15000 1,5%
10000 1,0%
5000 0,5%
0,0%
0
Indusrty Transport Other 1999 2001 2003 2005 2007 2009
Sector Year
Fig. 3. Final energy consumption of RES, biomass and renewable wastes by sector Fig. 4. Final energy consumption share of liquid biofuels in the transport sector and
in the EU (27 countries) in 2009. MSW (ren.): municipal solid waste (renewable). the road transport sector in the EU (27 countries).
Data source: Eurostat (online data codes: nrg 1071a, nrg 1073a) [1]. Data source: Eurostat (online data codes: nrg 100a, nrg 1073a) [1].
1592 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610

as well be biofuels. Thus, hydrogen [27], ethanol [28–32], ethers 2. Chemical kinetics modeling – reaction pathways
[33–37] and biodiesel [38] have been used in HCCI engines, either
as neat fuels or in blends. High cetane number fuels (such as DME) Almost all of the simulation models used in HCCI combus-
can be used as ignition promoters in the blend and high octane tion modeling and all simulation models presented herein include
number fuels (e.g. ethanol, methanol, etc.) can be used as ignition chemical kinetics. This is the result of the decisive role chemical
suppressors. With the use of closed or open loop control systems, kinetics play in HCCI combustion. Since no external means are uti-
the resulting fuel blend can provide the means to control HCCI igni- lized to instigate ignition, the latter relies on the reaction kinetics of
tion over a wider range than the one provided by the individual the fuel. Ignition commences where the thermodynamic conditions
neat fuels [25]. This has already been demonstrated in HCCI engines of pressure, temperature and species concentration are favorable.
using blends of n-heptane and isooctane [26]. Moreover, the specific characteristics of the reaction paths and
As is the case with conventional combustion types (SI and CI), intermediate species also affect the formation of pollutants such as
the HCCI combustion mode can be studied using simulation mod- NOx , CO and unburned HC. These issues provide the momentum for
els of various degrees of complexity. Simulation models describe the ongoing research oriented toward building chemical reaction
the fundamental processes occurring within the combustion cham- paths, which are able to describe in various levels of detail the oxi-
ber and – if accurate enough – can help toward understanding dation characteristics of fuel–air mixtures under various operating
the effects of individual operating parameters on HCCI combus- conditions. Compared to the diversity of the biofuels used, which
tion. Such an analysis is not always feasible in experiments, since include hydrogen, alcohols, ethers and esters, the chemical mech-
it is usually difficult to isolate the effect of a single parameter anism research has a short history. The difficulties encountered
– be it fuel properties, inlet conditions, etc. – without directly involve the identification of the primary and secondary reaction
or indirectly affecting another parameter. Simulation models can pathways as well as the reduction of the resulting chemical mech-
provide insight into the fundamental processes occurring during anisms. This reduction is paramount if the chemical mechanisms
combustion and/or during the stage of gas exchange processes, are to be used in simulation models, since the simulation time
depending on the sophistication level of the model. Features such increases non-linearly with the increase of species and reactions.
as combustion chemistry, heat and mass transfer can be studied Such considerations are especially important when using the chem-
so that the crucial factors affecting them can be assessed. Espe- ical mechanisms in CFD models, in which the computational time
cially in the case of biofuels, which have not been numerically cost is already aggravated by the solution of the Navier–Stokes
studied extensively until recently, there is a lot to be explored equations.
as regards combustion chemistry, emissions formation, trans- As regards the chemical mechanisms used in modeling HCCI
port properties, etc. Combustion chemistry affects significantly combustion, care should be exercised during the choice of the spe-
the mixture autoignition, combustion rate and emissions forma- cific mechanism, since the various existing mechanisms do not
tion. Transport properties affect heat and mass transfer within the necessarily provide similar results even for well-known fuels such
combustion chamber, which determine the temperature field or as isooctane [39]. Therefore, comparisons of ignition timings and
temperature–mass distribution within the combustion chamber. global combustion rates must be interpreted with caution, espe-
Moreover, in the case of direct injection HCCI, in which the fuel is cially when differences between fuels are assessed [40]. This also
directly injected into the combustion chamber, the fuel transport holds true for the comparison of NOx emissions, which are known to
properties must be adequately modeled to simulate the forma- be sensitive to peak temperature variations and consequently to the
tion, break-up, vaporization and mixing of the fuel jet with the ignition timing predictions. Thus, it is essential for the researcher
air. Such phenomena determine the mixture stratification and the to investigate the conditions under which the mechanisms were
possibility of fuel jet impingement on the combustion chamber developed and the experiments used for their validation (shock
walls. tube, flow reactor, jet stirred reactor, rapid compression machine,
HCCI simulation models are divided into single-zone, multi- laminar flame, HCCI engine, etc.). Preferably, the mechanisms have
zone and computational fluid dynamics (CFD) models, in order of to be validated at low equivalence ratio and high pressure since
increasing complexity. In single-zone models the main assumption these conditions are relevant to HCCI combustion in engines [39].
is the uniformity of the charge properties within the combus- Furthermore, it has to be kept in mind that the fuel oxidation chem-
tion chamber. Multi-zone models consider multiple regions within ical mechanisms usually refer to surrogates of actual fuels, which
the combustion chamber, each of which is considered an open cannot describe in exhaustive detail all the actual fuel properties. It
thermodynamic system of uniform properties. CFD models are is obvious that even the detailed mechanisms have their limitations
the most sophisticated ones, providing detailed description of and their validation cannot be exhaustive, considering the diverse
the in-cylinder gas motion and thermodynamic state on a local operating conditions encountered in HCCI engines (high EGR dilu-
level. However, this increased level of sophistication is not with- tion, fuel injection in hot residuals during negative valve overlap
out compensation, in terms of increased computational time and HCCI, etc.).
effort. Table 1 includes chemical reaction mechanisms for various bio-
In the following sections single-zone, multi-zone and CFD mod- fuels found in the literature. Some of these mechanisms have not
els, which have been constructed for biofueled HCCI combustion been used in the simulation studies presented herein, but are avail-
are presented. Since the vast majority of HCCI simulation models able for future research. However, the vast majority of the oxidation
use chemical mechanisms of various levels of detail to describe mechanisms included in Table 1 has been implemented in sim-
the chemical kinetics of the ignition and combustion processes, a ulation models, which are described in detail in the subsequent
separate section is devoted to the use of chemical kinetics. The sections. The citations describing each mechanism have been num-
HCCI simulation models presented in the following sections are bered consecutively, while the citations of the papers in which
included in this study regardless of whether they produced specific these mechanisms were used have been numbered in order of
biofueled-oriented results. The selection criteria were considered appearance in the text.
satisfied by the mere fact that biofueled HCCI combustion was
simulated. Upon consideration of the relatively short simulation 3. Single-zone modeling
history of the HCCI engine, simulation data for biofueled HCCI
engines have their own obvious merits and provide a reference for Single-zone models are often used in HCCI modeling due to
future studies nonetheless. the short computational time required for the simulation. In the
Table 1
Available reaction mechanisms for biofuel combustion simulation.

# Number of Source Ref. Used in Ref.a NOx b Fuel Simul. model Engine typec CR Pin d (bar) Tin d (K) Load ()e Engine Validationf Comments
species/number speed
of reactions (rpm)

1-Butanol (BuOH)
1. 181/1703 and Dagaut et al. – w n-heptane
573/2701 [41]
2. 234/1399 Black et al. [42] –
3. 251/1990 Dagaut and – w gasoline sur.
Togbé [43]

N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610
4. 1046/4398 Saisirirat et al. Saisirirat et al. – BuOH/n-heptane SZ 4S 16 – 353 0.3 1500 QLT 63% n-heptane
[44] [100] 37% BuOH
(vol.)
Diethyl ether (DEE)
5. 112/484 (+DME, Mack et al. [45] Mack et al. [45] DEE/EtOH/DTBP SZ 4S 16.25 1.7 380–455 0.3 1800 QLT
EtOH)
6. 215/1051 Yasunaga et al. –
[46]
Dimethyl ether (DME)
7. 23/23 Yamada [47] –
8. 26/28 Yao et al. [48], Huang et al. – DME CFD 4S 17.0 1.1 375 0.01–0.14 1400 QNT
Liang et al. [49] [138]
9. 27/35 (+MeOH) Liang [50] Chen et al. [82] DME–MeOH CFD 4S 17.5 1.02 375–380 0.224, 0.226, 1400 QNT DME PI, MeOH
30/44 [133] 0.605 DI.
(+MeOH + NOx )
Yao et al. [137] – DME/MeOH CFD 4S 17.0 1.16 375.6 0.238–0.5 1400 QNT
10. 28/45 Kim et al. [51] –
11. 78/336 Curran et al. Mack et al. [45] See DEE
[52]
Yao et al. [48] – DME SZ 4S 17 1 348 0.215, 0.095 1400 QNT
Mack et al. – DEE/EtOH SZ 4S 16.25 1.7 – 0.3, 0.4 1800 QLT
[101]
Yao et al. [103] – DME/MeOH SZ 4S 17 1 350 DME – 0.17 1400 QNT
MeOH –
0.063
Yao et al. [104] – DME/CH4 SZ 4S 17 1 348 0.297 1400 No
Ogawa et al. – DME/MeOH–H2 O– SZ 4S 18.8 1 293 0.16, 0.25 1200 No
[105] EtOH–H2 –CH4
Yamada et al. – DME/MeOH–O3 SZ 4S 8 1 378–388 0.12–0.42 600 QNT
[108] (DME )
12. 79/351 Fischer et al. Hoffman and [83,84] DME SZ 4S 16 1.07–1.49 300 0.4 2300 No
[53], Curran Abraham [97]
et al. [54],
Kaiser et al.
[55]
Shudo and – DME/H2 SZ 4S 9.7 – Ambient 0.27–0.35 1000 QLT
Yamada [106],
Shudo et al.
[107]
Kim et al. [120] [84] DME CFD 4S 17.8 320 8 mg/stroke 1500 QNT IT 80◦ –0◦ bTDC
Park [139] [84] DME CFD 4S 17.8 1–3 300–460 0.2–2.0 1500 No
Kong [142] [84] DME/MeOH CFD 4S 17.7 1.1 340 0.175–0.435 960 QNT
13. 97/457 Chun-lan et al. Chun-lan et al. [85] DME CFD 4S – 1 350 0.214 1500 QNT Emissions
(Detailed) 36/73 [56] [56] validation
(Reduced)

1593
 1594
Table 1 (Continued)

# Number of Source Ref. Used in Ref.a NOx b Fuel Simul. model Engine typec CR Pin d (bar) Tin d (K) Load ()e Engine Validationf Comments
species/number speed
of reactions (rpm)

Ethanol (EtOH)
14. 38/228 Röhl and Peters Komninos and [86] EtOH MZ 4S 21 1 ∼393 0.1–0.28 994 QNT
[57] Rakopoulos
[40]
Komninos and [86] EtOH MZ 4S 21 1 ∼393 0.15–0.26 994 QNT
Rakopoulos
[113]
15. 39/238 Li et al. [58] Sjoberg and – – – – – – – – – – Examined, not

N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610
Dec [118] used
16. 43/235 Saxena and Viggiano and – EtOH CFD 4S 21 1 363–393 0.16–0.26 1000 QNT
Williams [59] Magi [145]
17. 57/383 Marinov [60] Szybist [95] – EtOH SZ 4S 7–29 1 325,500–550 1, ∼0.25 900, 2000 No NVO
Mack et al. [45] See DEE
Ng and [84] EtOH SZ 4S 15 1 380–660 0.2–0.5 1000 No EGR, neat EtOH
Thomson [92] or Reformed
Martinez-Frias [84] EtOH (wet) SZ 4S 16 3 420 (ivc) 1 (␾EtOH ) 1800 No TC
et al. [93]
Yang et al. [96] – EtOH SZ 4S 13 – – – 1400 No NVO
Mack et al. – EtOH/DEE SZ 4S 16.25 1.7 – 0.3, 0.4 1800 QLT
[101]
Ogawa et al. See DME
[105]
Komninos and [86] EtOH MZ 4S 21 1 ∼393 0.15–0.26 994 QNT
Rakopoulos
[113]
Sjoberg and – – – – – – – – – – Examined, not
Dec [118] used
Yu et al. [143] – EtOH CFD 4S 17.2 1 354, 400 0.303 1200 QNT
Joelsson et al. – EtOH CFD 4S 17.2,17 0.9–1 354, 388 0.303 1200 QNT
[144]
18. 58/310 Curran et al. Sjoberg and – EtOH MZ 4S 12.8 0.84–1.81 418–443 0.25–0.6 500–2400 QNT
[61] Dec [117]
Sjoberg and – EtOH MZ 4S 12.8 1 427 0.4 1200 QNT
Dec [118]
19. 112/484 Curran in Mosbach et al. – EtOH/DEE SRM -PDF 4S 16.25 1.18 400–420 0.3, 0.4 1800 QNT
Mosbach et al. [62]
[62]
20. 167/1591, Dagaut and – w n-heptane
564/2589 Togbé [63]
21. 235/1866 Dagaut and – w gasoline surr.
Togbé [64]
22. 1046/4398 Saisirirat et al. Saisirirat et al. – EtOH/n-heptane SZ 4S 16 – 353 0.3 1500 QLT 63% n-heptane
[44] [100] 37% EtOH (vol.)
Ethyl acetate (Ethyl Acet.)
23. 23/142 Gasnot et al. Cotino and – Ethyl Acet. CFD 4S 13.3 1.35 boost 462 1.5–4 bar 1200 QNT Validation with
[65] Jeanmart [150] imepg iso-octane
Hydrogen
24. 10/19 Mueller et al. –
[66]
25. 10/19 Marinov et al. Goldsborough – H2 SZ 2S 3–38 1.5 300 0.15–0.6 – No Free piston
[67] and Van engine
Blarigan [89]
Komninos et al. [86] H2 MZ 4S 18–20 1–2 390–410 0.2–0.4 1000–3000 No
[112]
Table 1 (Continued)

# Number of Source Ref. Used in Ref.a NOx b Fuel Simul. model Engine typec CR Pin d (bar) Tin d (K) Load ()e Engine Validationf Comments
species/number speed
of reactions (rpm)

26. 11/19 O’Connaire –

N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610
et al. [68]
27. 11/23 Westbrook and Noda and [83] H2 MZ 4S 20 345–410 0.1–1.0 1600 Nо
Dryer [69] Foster [111]
28. 11/25 Kee et al. [70] Fiveland and – H2 SZ 4S 15 1.2.1.5 400–800 0.15–0.4 1500 No
Assanis [88]
29. 12/30 Li and Karim Liu and Karim [71,72] H2 CFD 4S 16–20 0.85 280–320, 515 0.5 800, 900 QLT
[71], Liu [72] [149]
Methanol (MeOH)
30. 20/84 Li et al. [73] –
31. 22/89 Held and Dryer Yao et al. [103] See DME
[74]
32. 26/84 Westbrook Ogawa et al. See DME
et al. [75,76] [105]
33. 52/326 Lindstedt and Yang et al. [96] – MeOH SZ 4S 13 – – – 1400 No NVO, 32
Meyer [77] Species/167
Reactions
Methyl butanoate (MB)
34. 41/150 (neat) Brakora et al. Brakora and [87] MB + n-heptane SZ 4S 18 1.1 300–400 0.298–1.0 1200 No 56 Species/169
53/156 (w [78] Reitz [98] Reactions
n-heptane)
Um and Park [84] MB + n-heptane CFD 4S 17.8 1–3 boost 300–460 0.2–2.0 1500 No
[152]
35. 275/1545 Dooley et al. –
[79]
36. 295/1498 Gaïl et al. [80] Szybist et al. [84] MB + n-heptane SZ 4S 10.5 1 433–498 0.38–0.48 1800 QNT Tin variable
[38]
Methyl decanoate (MD)
37. 3036/8555 Herbinet et al. Hofmann and [83,84] MD SZ 4S 16 1.16–1.65 310, 315 0.4 1670 No
[81] Abraham [97]

QLT, qualitative validation-mostly validation of predicted trends; QNT, quantitative validation-includes direct comparison of simulated quantities to experimental measurements; NUM, numerical validation-comparison of the
HCCI simulation results using a reduced mechanism against predictions from a detailed mechanism.
a
The source mechanisms may have been modified for their use in each study. Moreover, some of the mechanisms may have been updated since the publication of each study. Thus, the number of species and reactions are
subject to change. Symbol “–” used in this column indicates mechanisms, which are not used in the studies presented herein.
b
Symbol “–” in NOx : information not available in the study.
c
4S: four stroke engine, 2S: two stroke engine
d
Subscript “in” refers to inlet or initial conditions
e
Load is expressed in terms of the fuel–air equivalence ratio, unless explicitly stated otherwise.
f
Validation refers to the corresponding HCCI simulation model and not to the reaction mechanism(s).

1595
1596 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610

atoms in their molecules. Besides the study of the combustion

chemistry of neat biofuels, single-zone models can be used to study
the characteristics of various fuel blends or the effect of additives
on biofuel HCCI combustion. This is an important feature, since the
use of fuel blends and additives may provide the means for ignition
timing and combustion rate control. Issues like the aforementioned
provide the impetus for the ongoing HCCI research [88].

3.1. Hydrogen

Goldsborough and Van Blarigan [89] studied numerically a free-

piston hydrogen-fueled HCCI engine. The researchers modeled
piston dynamics and the gas exchange process in their free-piston,
two-stroke engine. They also used the HCT [90] chemical kinetics
software to simulate a zero-dimensional thermodynamic process
throughout the engine closed cycle. The hydrogen oxidation was
accounted for using the chemical mechanism of Marinov et al.
[67]. Heat transfer was accounted for by using Woschni’s corre-
lation [91], in which the heat transfer constants were modified
to match experimental data from hydrogen combustion in a rapid
compression machine. It was found that the engine operation was
significantly dependent on the scavenging process, which altered
the amount of fuel–air mixture trapped within the combustion
chamber, the initial temperature of the charge, and the amount
Fig. 5. Schematic of the First Law of Thermodynamics for single-zone models.
of fresh charge that was expelled directly through the exhaust port
during this scavenging process. Overall, the engine was found to
produce low NOx emissions and to provide almost constant volume
single-zone (or zero-dimensional) model the engine cycle is mod- combustion. The rapid combustion, which resulted in the simula-
eled using the First Law of Thermodynamics, taking into account tions, could be partly attributed to the nature of the single-zone
the work absorbed/produced by the piston and – in some cases – model used.
the heat exchanged between the charge and the combustion cham- Fiveland and Assanis [88] constructed a zero-dimensional
ber walls, as shown in Fig. 5. Mass blow-by is usually neglected as numerical model for the simulation of the complete engine cycle
shown in the figure. using methane or hydrogen as a fuel. The hydrogen combustion
The thermodynamic (pressure, temperature, composition, etc.) chemistry scheme included 11 species and 25 chemical reactions,
and transport properties (viscosity, conductivity, etc.) of the mix- and was modeled using the CHEMKIN libraries [70] adapted for a
ture are considered uniform. Since any temperature stratification variable volume plenum. Heat transfer was calculated using a sub-
within the combustion chamber is neglected, the single-zone set of the k–ε model under the isotropic turbulence assumption.
models tend to overestimate the global heat release rate and, con- This method yielded the characteristic velocity that was used to
sequently, the pressure rise rate and the maximum combustion calculate the heat transfer coefficient.
pressure. Moreover, since the mixture composition is also consid- Simulation results for a hydrogen-fueled four-stroke engine
ered uniform, the single-zone model is incapable of describing the with a compression ratio of 15 and 1.5 bar intake pressure required
HC and CO formation processes, unless these are due to global par- 425 K intake charge temperature to produce ignition near TDC. This
tial combustion of the mixture; under normal HCCI combustion high intake temperature requirement, at a moderate compression
conditions, the formation of these pollutants is believed to occur in ratio and at a low fuel–air equivalence ratio of 0.3, was necessary
the relatively low temperature regions of the combustion chamber since hydrogen ignition occurred at about 1050–1100 K in the sim-
as a result of mass transfer between cold and hot regions, in which ulation. Under these conditions the researchers identified peaks in
any HC transferred are partially burned producing CO [23]. H, OH and HO2 fractions near the ignition point. The simulation
Apart from these inherent limitations, single-zone models can model was also used in a parametric investigation, in which it was
be used to provide an estimation of the ignition timing and the found that decreasing the initial temperature or the fuel–air equiva-
formation of NOx emissions. This is due to the fact that both of lence ratio, beyond a certain point, retarded ignition and eventually
these phenomena are primarily controlled by the temperature of led to misfire.
the hottest regions of the combustion chamber, which is close to
the mean gas temperature of the mixture. Moreover, single-zone 3.2. Alcohols
models can be used to assess the relative importance of chemical
reactions and dominant species that are part of a chemical reaction Ng and Thomson [92] constructed a single-zone model to sim-
mechanism describing the fuel–air oxidation. Thus, detailed reac- ulate ethanol HCCI combustion and especially the effect of ethanol
tion mechanisms can be compared to reduced ones under the same reformed gas on the ignition requirement and NOx emissions. They
initial conditions and assumptions provided by the single-zone incorporated Marinov’s set of chemical reactions [60] to simulate
simulations. The latter features are especially important in the case ethanol oxidation and integrated a NOx reaction mechanism from
of biofuels, the combustion chemistry of which is under continuous GRI-Mech [84] to simulate NOx formation. For the estimation of
investigation. The relevant studies are perplexed by the variety of heat losses they used a convective heat transfer correlation (similar
the biofuels, which include organic compounds of diverse homolo- to Newton’s Law of cooling).
gous series, such as alcohols, ethers, and esters. This is a significant The researchers conducted a parametric study, in which the
departure from the combustion chemistry of the common pure parameters varied were the fuel–air equivalence ratio, the EGR
hydrocarbons, which are usually used as surrogates for the simu- amount and the degree of reforming, i.e. the degree of ethanol con-
lation of diesel or gasoline combustion and do not include oxygen version into CO and H2 . Thus, the initial charge contained ethanol,
 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610 1597

CO, H2 and EGR at various proportions. It was found that fuel [60] for ethanol, and Lindstedt and Meyer [77] for methanol. The
reforming resulted in an expansion of the HCCI operating region, researchers studied both the thermal and the chemical effect of
the boundaries of which were determined by misfire and unac- the fuel injection during the NVO period. They found that during
ceptable NOx emissions. However, reforming tended to increase this period both heat and active species such as H2 and CO can
NOx emissions due to the resulting high peak combustion temper- be produced, depending on the fuel type, oxygen availability and
ature. In relation to reformed gas, EGR was more effective toward the in-cylinder temperature of the residuals during injection. At
expanding the simulated operating range at moderate equivalence relatively high oxygen availability in the residuals the tendency
ratios (0.2–0.5). was to produce more heat than active species, since the fuel was
Martinez-Frias et al. [93] investigated the direct use of wet oxidized. In oxygen-poor residuals only part of the fuel was oxi-
ethanol using a simulation model. The use of wet ethanol as a fuel dized producing active species. Methanol and ethanol were less
reduces the energy demands for its production, since the distilla- sensitive than gasoline to O2 availability owing to their oxygenated
tion and/or dehydration processes, which are used to remove the nature. Methanol showed the greatest tendency in forming active
water and provide neat ethanol, require a significant amount of species.
energy (37% of the total energy in ethanol products for corn-based
ethanol production). The objective of the study was to determine 3.3. Ethers
whether benefits could be obtained by the direct utilization of wet
(as opposed to neat) ethanol into a turbocharged HCCI engine. The In [48] Yao et al. used the SENKIN code of the CHEMKIN pack-
researchers constructed a simulation model for a TC HCCI engine age to simulate DME combustion. The single-zone model used
fueled with wet ethanol (35 vol.% EtOH in H2 O). Apart from the was considered adiabatic and the chemical mechanism, which was
TC, the full engine system model also incorporated an exhaust gas used to describe the DME oxidation, was that of Lawrence Liver-
regenerator, a fuel vaporizer and a catalytic converter. more National Laboratory including 79 species and 336 reactions
The HCCI engine was modeled via a single-zone model, using [52–55]. The scope of the study was to analyze the reaction path-
CHEMKIN [94] to solve for chemical kinetics. The ethanol reac- ways of DME oxidation. The researchers validated the simulation
tion mechanism was that of Marinov [60] and the NOx -related results against experimental data, which indicated that the com-
reactions were taken from GRI-Mech [84]. Heat transfer was also bustion rate and the peak combustion pressure were overestimated
incorporated into the reciprocating engine model assuming a wall due to the nature of the single-zone model. However, the ignition
temperature of 430 K. Using this model, the researchers found that timing was adequately predicted, implying a sufficient descrip-
a minimum ethanol-in-water fraction of 35 vol.% (liq.) was neces- tion of the reaction pathways. It was found that DME presented a
sary for a sustainable simulated HCCI operation. This provided an two stage auto-ignition with low temperature reactions (LTR) fol-
overall brake efficiency of 38.7% for the entire simulation system. lowed by high temperature reactions (HTR). HCHO and OH were
The researchers estimated that this direct use of 35% wet ethanol the primary radicals during LTR. DME ignition was controlled by
significantly reduced the energy required for fuel production – rel- the H2 O2 decomposition to two OH radicals, which provided a
ative to neat ethanol production – by eliminating the need for chain branching step. These hydroxyls were the primary attack-
dehydration and limiting the distillation needs to 3% of the total ers for the H atom abstraction from DME and the oxidation of
energy in ethanol products, in the case of corn-based ethanol pro- CO. At relatively leaner DME concentration the LTR was not sig-
duction. nificantly affected but the CO oxidation deteriorated, indicating a
In Ref. [95] Szybist used CHEMKIN-PRO [94] to simulate the HCCI temperature sensitivity of the CO oxidation process. This led to the
combustion of n-heptane, isooctane, toluene and ethanol. The sim- generally accepted conclusion that CO emissions in HCCI engines
ulation model was switched to the adiabatic mode. The mechanism originate from the relatively cold and/or lean regions of the com-
used for the simulation of ethanol combustion was that of Mari- bustion chamber, such as the crevices and the vicinity of the wall,
nov [60] including 56 species and 351 reactions. The author used in which the local temperature is adequately high to induce partial
this model to simulate HCCI combustion with the aforementioned oxidation of the existing HC but not high enough to consume the
fuels under normal naturally aspirated conditions with air dilution resulting CO.
and under negative valve overlap (NVO) conditions. In the latter Hofmann and Abraham [97] also used the internal combustion
case, the mixture had a higher temperature at BDC (>500 K) and a engine model included in the CHEMKIN package [94] to simulate
significant amount of diluents such as O2 , CO2 , H2 O and N2 . These n-heptane, DME and methyl decanoate HCCI combustion. The lat-
conditions resemble HCCI under NVO operation, in which a portion ter was studied as a biodiesel surrogate. The simulation model
of the exhaust gases is trapped within the combustion chamber and incorporated Woschni’s correlation [91] for the estimation of heat
raises the temperature of the incoming fuel–air charge. The results losses from the combustion chamber. The primary objective of the
of the study showed that with low intake temperature (325 K) and study was the investigation of ignition delay and NO formation
no EGR, the most resistant fuel to ignite was ethanol requiring the for the three fuels. For this purpose, the researchers used Her-
highest compression ratio. However, under NVO conditions toluene binet et al. [81] detailed chemical kinetic mechanism for methyl
was the most difficult to ignite. Moreover, with NVO the igni- decanoate – a biodiesel surrogate – consisting of 3036 species
tion timing between fuels converged to within a few CA degrees, and 8555 reactions, and Livermore National Laboratory scheme
with the exception of n-heptane at high O2 dilution. On the con- for DME oxidation including 79 species and 351 reactions [53–55].
trary, ethanol was essentially insensitive to the O2 dilution, under NO formation was simulated using two alternative schemes; the
the conditions of the simulation, indicating a zero-order reaction extended reaction mechanism of Zeldovich [83] and the NO kinetics
rate with respect to O2 . The researcher concluded that ignition from GRI-Mech v.3 mechanism [84]. The Zeldovich NO formation
in HCCI combustion under NVO conditions presented distinct fea- scheme provided reduced NO concentration relative to the more
tures, which were not entirely consistent with the RON and MON extensive NO scheme of GRI-Mech for all fuels. Moreover, it was
parameters. demonstrated that DME showed the shortest ignition delay fol-
Yang et al. [96] also investigated NVO HCCI combustion. They lowed by methyl decanoate and n-heptane. Thus, DME required
used a single-zone adiabatic model to simulate the reactions tak- higher engine speed and the lowest initial temperature. Although
ing place within the combustion chamber with the injection of the primary consideration for NO formation was the peak combus-
gasoline, ethanol or methanol during the NVO period. The reac- tion temperature, fuel chemistry was also found to play a secondary
tion kinetics was incorporated via the mechanisms of Marinov role.
1598 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610

3.4. Biodiesel simulation as regards the dependence of the results on the biodiesel
concentration, and the duration and phasing of the LTHR and the
Apart from Hofmann and Abraham mentioned in the preced- main combustion event. These were attributed to the low cetane
ing sub-section, biodiesel HCCI combustion was also simulated by number of methyl butanoate, the approximation of the IVC tem-
Brakora and Reitz [98] using the SENKIN code of the CHEMKIN perature with the intake manifold temperature, and the inherent
package [94]. The single-zone model was considered adiabatic for inability of the single-zone model to capture any fuel or tempera-
the purpose of the study, which was to investigate the formation ture stratification. The latter resulted to an overestimation of the
of NOx from biodiesel- and diesel-fueled HCCI engines. N-heptane peak combustion temperature and the combustion rate, which led
was used as the diesel surrogate, and a mixture of 2/3 n-heptane to almost 100% combustion efficiency, very low CO and high NOx
and 1/3 methyl butanoate was used to simulate biodiesel combus- emissions.
tion. This proportion of n-heptane and methyl butanoate in the Saisirirat et al. [100] examined the effect of n-heptane blends
biodiesel surrogate was found to yield a better approximation for with ethanol and 1-butanol, both experimentally and numerically.
the actual C:H:O ratio found in the real biodiesel, relative to neat They used the SENKIN single-zone model of the CHEMKIN II code
methyl butanoate or methyl decanoate. The latter two, being rela- [94] to simulate n-heptane/ethanol and n-heptane/1-butanol fuel
tively short chain esters, provided a much higher C:O ratio than blends. The chemical mechanism used was that of Saisirirat et al.
the actual biodiesel fuel. The chemical mechanism used for the [44] consisting of 1046 species and 4398 reactions, which includes
biodiesel surrogate consisted of 56 species and 169 reactions [78]. ethanol, 1-butanol and PRF oxidation sub-schemes. The simula-
NOx formation was incorporated via a 7 species and 19 reactions tion compared the HCCI combustion of neat n-heptane to the one
scheme, which was developed by Yoshikawa and Reitz [87] relying obtained with the two fuel blends, i.e. n-heptane/ethanol and n-
on reactions from the GRI-Mech mechanism [84]. heptane/1-butanol both in 63:37 molar proportions. It was found
Brakora and Reitz found that the ignition timing played a cru- that the addition of ethanol or 1-butanol to n-heptane affected the
cial role in the resulting NOx emissions. With an equivalent energy OH radical production and slowed down the chemical reactions
release and for the same ignition timing, the two surrogates pro- during LTHR, decreasing the concentration of reactive intermediate
vided roughly the same peak combustion temperatures (within species (HCHO, OH, HO2 , etc.). Thus, alcohols induced a chemical
approximately 20 K) and yielded similar NOx emissions at EVO. effect on the LTHR. Owing to this reduction in chemically active
There was a difficulty in connecting the oxygen content of the fuel species and to the low LTHR, which prevented a further tempera-
to the produced NOx emissions, especially at lean mixtures. Under ture rise of the mixture, the main combustion event was delayed
stoichiometric conditions, the biodiesel surrogate yielded a 26% with the addition of alcohols.
increase in NOx relative to the diesel surrogate. However, the pre- Mack et al. [101] constructed a single-zone model for the sim-
dicted difference in peak combustion temperature was about 13 K ulation of ethanol/diethyl ether (DEE) blends. The single-zone
lower for biodiesel, suggesting that the fuel–bound oxygen could model incorporated Woschni’s correlation for the estimation of
play a role in the formation of NOx under limited O2 concentrations. heat losses; it used a DME-derived chemical mechanism [52] for
It was also found that under the same energy content, ignition tim- the chemistry of DEE oxidation, and Marinov’s ethanol oxidation
ing and total N:O ratio in the mixture, for the two fuels, the diesel scheme [60] for the combustion chemistry of ethanol. The com-
surrogate yielded higher NOx emissions. This suggested that hav- bined mechanism consisted of 112 species and 484 reactions. The
ing the same amount of O2 available in the air, as opposed to being purpose of the study was to investigate the relative reactivity of DEE
fuel–bound, increased the NOx emissions due to the alteration of and ethanol in a fuel blend of 25% DEE in ethanol. The experiments
the heat capacity of the mixture, which resulted in higher flame conducted with carbon-14 isotope tracing showed that DEE was
temperatures for the diesel surrogate. The study concluded that more reactive than ethanol, yielding more CO2 relative to ethanol.
under HCCI conditions, in which near stoichiometric local condi- The simulation results supported this conclusion, since it was found
tions are not encountered, any NOx increase due to biodiesel would that DEE began to react earlier than ethanol. However, a signifi-
be small. cant elongation of the heat release, which would be desirable to
control the combustion process, was not observed using the fuel
3.5. Fuel blends and additives blend.
The same research group also studied the effect of small
Apart from neat fuels, single-zone models have been used with amounts of di-tertiary butyl peroxide (DTBP) (up to 3%) on the HCCI
fuels blends, to assess the effects of additives/blends on the HCCI combustion of ethanol and diethyl ether blends [45]. The intention
combustion process. was to elongate the combustion event using the exothermic decom-
Szybist et al. [38] used a single-zone model to simulate the com- position of DTBP to promote early DEE ignition. The single-zone
bustion of biodiesel blends in HCCI engines. In their study they used model used was the one mentioned earlier and the combustion
CHEMKIN [94] for the chemical kinetics simulation coupled with reaction scheme was enhanced by the adiabatic decomposition
chemical mechanisms for n-heptane [99], which represented diesel of DTBP based on the work of Iizuka et al. [102]. It was found
fuel, and methyl butanoate [80], which accounted for biodiesel. The via simulation, and verified by experiments, that the addition of
combined mechanism consisted of 670 species and over 3000 reac- DTBP advanced the ignition timing in the DEE-ethanol blend. The
tions, including NOx formation reactions, which were taken from simulations showed that the decomposition of DTBP raised the
GRI-Mech [84]. With this mechanism it was possible to study dif- temperature of the mixture and advanced ignition, although poten-
ferent blends of n-heptane and methyl butanoate, which were used tial chemical effects of DTBP were not excluded.
as surrogates for the simulation of diesel/biodiesel blends. Thus, the Yao et al. [103] used a zero-dimensional thermodynamic model
researchers simulated diesel/biodiesel blends of 0, 10 and 50% (vol.) coupled with CHEMKIN III to simulate a DME/methanol blend in an
in biodiesel with n-heptane/methyl butanoate blends, the relative HCCI engine. The chemical mechanisms used were the ones of Cur-
proportion of which was chosen to match the oxygen content (wt%) ran et al. for DME [52], consisting of 78 species and 336 reactions,
of the experimental fuel blends. and Held and Dryer [74] for the reactions of methanol, consisting of
The simulation model predicted the effect of inlet temperature 22 species and 89 reactions. The authors compared the simulation
increase on the phasing and magnitude of the low temperature predictions to experimental results of pressure and operating range
heat release (LTHR), and the phasing of the main combustion event. in DME − MeOH map. The analysis of the simulation results showed
However, discrepancies were observed between experiments and that in DME/methanol dual fuel oxidation the low temperature heat
 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610 1599

release of DME was inhibited, resulting only in a high temperature

heat release. The OH radicals were mainly formed by the decompo-
sition of H2 O2 at blue-flame reactions, and by H + O2 = O + OH and
O + H2 O = OH + OH at hot flame reactions.
Similar results were found by the same Institute in another study
[104]. The simulation model used has already been described in the
first part of their paper [48] presented in Section 3.3. In Ref. [104]
the researchers studied the HCCI combustion of a DME/methane
blend. Under the simulation conditions, the dual fuel blend showed
a two stage autoignition scheme. This was different from the clas-
sical two stage ignition observed in neat DME oxidation, since it
corresponded to two high temperature heat release rates, one of
DME, which was found to oxidate first, and the other of methane.
It was observed that methane - like methanol - inhibited the LTHR
but promoted the high temperature heat release (HTHR) of DME.
On the other hand, DME promoted methane oxidation both due to
thermal and chemical considerations. The thermal effect was due
to the temperature increase, which was the result of DME oxida-
tion, while the chemical effect was due to HCHO and H2 O2 , which
were formed during DME oxidation and increased the OH radical
pool. The researchers also identified the major reaction paths for
the dual-fuel oxidation.
In [105] Ogawa et al. studied the suppression effects of addi-
tives such as water, methanol, ethanol, hydrogen and methane, on
the combustion of a DME fueled HCCI engine. The researchers used
SENKIN under adiabatic conditions for the single-zone simulations
and the corresponding chemical mechanisms were adapted from
Fig. 6. (a) Concept of HCCI combustion engine system fueled with DME and MRG
Curran et al. [52] for DME, Marinov [60] for Ethanol, and Westbrook onboard-reformed from methanol utilizing engine exhaust gas heat [107] and (b)
et al. [75,76] for Methanol. In the experiments the suppressors were experimental setup [106].
injected directly into the combustion chamber, while the DME was
introduced into the air stream at the intake pipe. The direct injec-
tion of suppressors induced a cooling effect on the charge. Water antagonistic effect for the consumption of OH radicals; in the pres-
injection had a predominantly cooling effect, i.e. it lowered the tem- ence of H2 , OH radicals reacted preferentially with H2 instead of
perature of the mixture during its vaporization, thereby retarding DME, to yield the relatively stable – at low temperatures – interme-
ignition. Simulations conducted with the assumption that gaseous diates HO2 and H2 O2 . This led to a suppression of LTHR, a decrease
water was introduced in the mixture at BDC, resulted in a slight in the in-cylinder temperature rise and a consequent retardation of
delay in radical formation and suggested that the water addition the main combustion event. This thermal effect was the main rea-
also had a minor chemical effect. Similar findings characterized son for the combustion retardation, since the ignition temperature
methanol addition. Its main effects were a thermal one observed was not altered appreciably. These findings implied that hydrogen
during the experiments and a chemical one suggested by simula- or hydrogen-containing additives, such as methanol reformed gas
tions. (MRG), could be used to control the ignition timing of DME. The
As in the case of water addition simulations, methanol addi- researchers also studied the conversion of methanol to DME and H2 ,
tion simulations were conducted assuming that methanol existed via dehydration- and thermal decomposition-methanol reformers,
in gaseous state at the beginning of calculations. It was found that respectively, to conclude that the H2 /DME fraction in MRG could be
methanol addition severely suppressed LTHR to near-zero values, continuously controlled by adjusting the methanol supply in each
significantly reduced OH, H and O radicals, and retarded main com- reformer. The addition of CO in neat DME, which was studied in
bustion. Further simulations with ethanol, hydrogen and methane [106], was also found to suppress the DME oxidation at low tem-
addition showed that LTHR was slightly reduced with hydrogen or perature, but its effect was less pronounced than the H2 addition
methane addition, and essentially disappeared with the addition effect.
of ethanol or methanol, in order of significance. Moreover, it was Yamada et al. [108] studied the effect of methanol or O3 addi-
found that despite the different simulation conditions with respect tion to DME-air mixture under HCCI combustion conditions. They
to additive and its quantity, the OH concentration at the onset of used the SENKIN [94] code to simulate the engine operation, and
LTHR remained fairly constant. This suggested that the chemical implemented a chemical mechanism by Curran et al. [52] to account
effect induced by the additives was linked to a reduction in the OH for the DME and methanol addition chemistry. Ozone addition was
radicals prior to LTHR. accounted for by including the thermal decomposition reaction of
This assumption regarding OH radicals was supported by later ozone according to Heirmel and Coffee [109]. The experimental
studies with H2 and MeOH addition to DME–air mixture. Specifi- and simulation results showed that methanol addition retarded the
cally, Shudo et al. [106,107] investigated the effect of H2 addition, main heat release event significantly; 8% addition of methanol in
which originated from on-board methanol reformed gas, on the DME resulted in a 15◦ CA delay. It was suggested that this delay was
DME HCCI combustion. The concept of the combustion engine sys- due to the competitive consumption of OH by methanol instead of
tem and the experimental setup are given in Fig. 6(a) and (b). DME, which led to the formation of relatively stable intermediates
The SENKIN [94] simulation model was used to describe the such as HCHO and HO2 . In contrast, the addition of low O3 amounts
effect of hydrogen addition, and the reaction mechanism used was (0.015% in DME) advanced the ignition timing by as much as 20◦ CA.
that of Fischer et al. [53–55] consisting of 79 species and 351 reac- This was due to the decomposition of O3 at relatively low temper-
tions. The simulation results showed that the addition of H2 in the atures (∼600 K), which provided oxygen atoms and activated the
DME–air mixture suppressed DME consumption. This was due to an mixture. This supply of oxygen atoms at temperatures lower than
1600 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610

In [62] Mosbach et al. constructed a stochastic reactor model

(SRM) based on a probability density function (PDF) approach. The
PDF was approximated by an ensemble of 100 stochastic particles.
These did not contain any spatial or geometric information. The
random variables were the species concentration and the tempera-
ture. The model included heat transfer to the combustion chamber
wall using Woschni’s correlation [91] and the magnitude of the
temperature fluctuations was regulated by a coefficient. Turbulent
mixing was accounted for via a mixing intensity term that included
the characteristic mixing time. The fuel simulated was a blend of
EtOH/DEE, although a validation of the model was also presented
with neat EtOH as fuel. The validation showed good agreement
between experimental and simulated results as regards pressure
traces. The model also captured qualitatively the experimental vari-
Fig. 7. The First Law of Thermodynamics applied to a zone of a multi-zone model. ation of CO and HC emissions with the combustion phasing, i.e.
the angle at which 50% of the total heat has been released. In
these simulations, the equivalence ratio remained constant and
those of normal cool flames suppressed the production of formalde- the combustion phasing was regulated via the initial temperature.
hyde, which was a chain termination step under low temperatures. The stochastic reactor model was also used to study numerically
Thus, more heat was liberated during cool flame reactions, leading the effect of the DEE in EtOH ratio on the combustion phasing,
to an earlier thermal ignition. the combustion duration and the emissions of the HCCI engine. It
was shown that as the fraction of DEE in the DEE/EtOH blend was
4. Multi-zone modeling increased, the combustion phasing advanced before TDC, the com-
bustion duration was shorter, and the CO emissions were reduced.
The limitations encountered with single-zone models can be This fact provided indication that DEE could be used with ethanol
partly overcome with the use of multi-zone models. The latter to regulate the combustion phasing and duration of the HCCI com-
assume that the combustion chamber is divided into a sufficient bustion.
number of regions or zones, each possessing uniform thermody- Noda and Foster [111] numerically studied hydrogen HCCI com-
namic properties. The sufficiency of the number of zones is usually bustion by constructing a multi-zone model; the different zones
determined by a sensitivity analysis, i.e. the minimum number of catered for any temperature or composition inhomogeneities. The
zones used is defined as the number beyond which the simulation zones were not assigned a specific location within the combus-
results tend to converge. tion chamber, and mass exchange between them was neglected.
The temperature, composition, volume and all thermodynamic The heat losses were calculated via the Woschni [91] correlation
properties of each zone are considered uniform within the bound- using the combustion chamber area and the mean gas tempera-
aries of the zone, but can be different from the thermodynamic ture. The combustion chemistry was accounted for by incorporating
properties of other zones. This provides an opportunity to study 11 species and 23 reactions from Westbrook and Dryer [69] for
the effects of temperature or species stratification within the com- the oxidation of hydrogen, and the extended Zeldovich mecha-
bustion chamber, and usually leads to a better estimation of the nism of Lavoie et al. [83] for NOx formation. The researchers used
combustion duration, the pressure rise rate and the emissions for- 10–30 zones to examine the effect of temperature and fuel inho-
mation. Mass and heat transfer within the combustion chamber mogeneities on HCCI combustion. Both temperature and – to a
play a critical role in the estimation of the aforementioned operat- much lesser degree – fuel inhomogeneities increased the com-
ing characteristics and since the combustion chamber is considered bustion duration compared to single-zone simulations. The mean
an ensemble of thermodynamically distinct zones, the processes gas temperature and the type of temperature inhomogeneity also
of mass and heat transfer between zones must also be described affected combustion duration and the formation of NO emissions.
by the multi-zone model. Fig. 7 shows the first law of thermo- Provided the mean gas temperature was high enough to avoid
dynamics applied to a zone, which in the general case includes bulk quenching of the mixture, the most favorable type of tem-
heat and mass transfer. As regards mass transfer, each zone can be perature stratification included a small amount of hot gases to
thermodynamically defined as a closed or an open thermodynamic induce autoignition and avoid excessive NO formation and a greater
system, depending on whether mass flow between zones is pro- amount of colder mixture. This type of stratification expanded the
hibited or allowed, respectively. Likewise, heat transfer between operating range in a mean gas temperature versus temperature
zones and to the combustion chamber wall can also be permitted inhomogeneity map.
or neglected. Komninos et al. [112] and Komninos and Rakopoulos [40,113]
However, the assumptions regarding mass and heat transfer used a multi-zone model to simulate hydrogen [112] and ethanol
affect the results of the simulation significantly; heat transfer [40,113] HCCI combustion. In all studies, the multi-zone model
affects the temperature stratification within the combustion cham- used divided the combustion chamber into several zones, which
ber, the colder regions of which are sources of unburned HC and CO. occupied specific geometric regions (Fig. 8).
The formation of these pollutants is also affected by mass trans- The only uniformity assumed was that of pressure, while heat
fer between zones, since any HC or CO flowing from a colder to a and mass transfer between zones was accounted for, taking into
hotter region/zone is expected to be (partially) oxidized [110]. The account the geometric configuration of the zones; each zone could
predicted combustion rate, pressure rise rate, combustion duration communicate with its immediate inner and outer zone (Fig. 9). The
and peak pressure are also affected by the choice of the sub- model also accounted for a crevice zone, which was in direct com-
models used to describe heat and mass transfer. The multi-zone munication with the outmost zone of the combustion chamber. In
models examined subsequently are divided into stochastic based Ref. [112], 11 zones were used – including the crevices – to simulate
multi-zone models, in which some properties of the mixture are hydrogen HCCI combustion and the Annand correlation [114] was
considered random variables, and phenomenological multi-zone employed to estimate the heat flux from the outmost zone to the
models, in which probability considerations have been excluded. combustion chamber wall. The chemistry of hydrogen oxidation
 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610 1601

experimental results and captured adequately the pressure traces

and HRR for both ethanol and isooctane.
According to the simulation results, CO emissions were formed
near the combustion chamber walls, thereby justifying the finer
zonal resolution used in these regions (see Fig. 10). As shown in
Fig. 10, the amount and location of CO production was found to
depend on load; at higher loads, CO production advanced to the
inner zones.
Total CO emissions were adequately captured but HC emissions
were overestimated, especially for ethanol. An investigation for the
composition of the HC emissions for both isooctane- and ethanol-
fueled HCCI combustion showed that in the ethanol cases examined
the vast majority of unburned HC consisted of oxygenated HC
species (OHC). These were primarily ethanol, acetaldehyde and
Fig. 8. Zone configuration in the multi-zone model [112]. formaldehyde. The predicted percentage of OHC in total HC (aprox.
95% by mass) for ethanol was not very sensitive to the load
variation under the simulation conditions. On the contrary, the
was accounted for via the mechanism of Marinov et al. [54] involv- relative amount of OHC in total HC was lower (19–33% by mass)
ing 10 species and 19 reactions, augmented by a NOx formation in the case of isooctane and decreased with the increase of load.
scheme adapted from the isooctane oxidation reaction mechanism It was concluded and demonstrated that this would induce a
of Golovitchev [86]. A parametric investigation was conducted with significant error in the HC measurement if a measuring device
the multi-zone model to determine the effect of various operating such as the FID was used, since the FID principle possesses an
parameters on hydrogen HCCI combustion. Compression ratio and inherent inability to accurately detect oxygenated carbon atoms
initial temperature were the dominant factors affecting the igni- [116].
tion timing and pressure rise rate. Engine speed also affected the Using the same model in [113] the authors compared two
ignition timing, mainly due to the variation of the available time for ethanol oxidation mechanisms, namely Marinov’s [60] and Röhl
the occurrence of chemical reactions. It was found that increasing and Peters’ [57]. Marinov’s mechanism consisted of 57 species and
the load, i.e. the equivalence ratio, NOx emissions were increased 383 reactions, while Röhl and Peters’ mechanism, which was a
significantly. Inlet pressure increased almost proportional to the reduced version of Marinov’s, consisted of 38 species and 228 reac-
attainable imep, without compromising the fuel conversion effi- tions. Both mechanisms were augmented by NOx related reactions
ciency for the conditions simulated. taken from [86]. The results showed that the two mechanisms
In [40] Komninos and Rakopoulos numerically compared predicted essentially the same dominant HC species, with the
ethanol and isooctane HCCI combustion. The multi-zone model exception of formic acid (HCOOH) and to a lesser extent methanol
used was the one described in the preceding paragraph with some (CH3 OH), both of which were not included in the reduced mech-
modifications; the zones number was increased to 16, the zone anism. The mechanism reduction had a significant impact on
resolution was refined near the combustion chamber wall, and the the computational time: although the species used in Röhl and
wall heat flux estimation approach abandoned the Annand corre- Peters’ mechanism were 2/3 of the species used in the Marinov
lation and was based on the temperature gradient near the wall. mechanism, the Röhl and Peters’ mechanism required half the com-
These modifications were necessary to better estimate the forma- putational time. The simulation results showed a slightly higher
tion of HC and CO emissions [115]. The chemical mechanism used predicted combustion rate by the Röhl and Peters’ mechanism rel-
for the ethanol oxidation was that of Röhl and Peters [57] consist- ative to Marinov’s mechanism, despite the fact that the predicted
ing of 38 species and 228 reactions. Since this mechanism did not initial heat release was the same for the two mechanisms. The
include NOx reactions, it was augmented by 13 reactions involving resulting temperature fields were almost identical as shown in
nitrogen oxides formations [86]. The model vas validated against Fig. 11. Therefore, the two chemical mechanisms did not induce
significant differences in the predicted performance. NOx , CO and
HC emissions were essentially the same for the two mechanisms,
with the greatest disagreement occurring between predicted CO
emissions (∼9%) at the low load cases. This was connected with the
slightly lower combustion rate predicted by Marinov.
Sjoberg and Dec [117,118] used a multi-zone model to study the
effects of engine speed, equivalence ratio, initial temperature and
boost pressure on ethanol HCCI combustion. They used the multi-
zone model provided by the Senkin application of the CHEMKIN
package [94], which treated each zone as a single lumped mass
with uniform composition. The only interaction between zones
was through pressure, which was considered uniform. Thus, the
volume of each zone depended on its mass and its temperature
relative to the other zones. The 11 zones used were adiabatic in
principle. This led to the need to introduce initial temperature
stratification for the 10 active zones to match the experimental
pressure trace. The 11th zone was assigned a temperature low
enough to render it chemically inactive throughout the engine
cycle. This zone included 16% of the total mass and compensated
for the combustion inefficiencies, heat transfer and blow-by, which
were not directly modeled. The simulated compression ratio was
Fig. 9. Schematic of heat transfer between zones and to the cylinder wall [113]. also reduced relative to the experimental one to account for the
1602 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610

Fig. 10. CO simulation results for ethanol; left: in-cylinder CO emissions and right: zone CO production rate [40].

neglected heat losses. The thermal stratification of the zones was increase of simulated EGR. Moreover, it was suggested that the
supported by KIVA simulations and reproduced adequately the presence of water in the EGR promoted ethanol auto-ignition.
variation of combustion phasing with temperature. The chemical
mechanism used consisted of 58 species and 310 reactions [61], 5. Multi-dimensional modeling
and was part of a larger gasoline surrogate mechanism produced
by the joint effort of the National University of Ireland–Galway Besides the aforementioned favorable attributes of the multi-
and Lawrence Livermore National Laboratories [119]. zone models, they also have limitations. Even in the cases where
The simulations above showed that the ethanol oxidation mech- spatial considerations have been accounted for by multi-zone mod-
anism reproduced adequately the effect of equivalence ratio, engine els, they usually do not allow high local resolution. Moreover,
speed and boost pressure on ignition. The same model was used by there is an inherent difficulty to fundamentally describe the motion
these researchers to study the effect of EGR and its constituents on of the charge within the combustion chamber and the effects of
the HCCI autoignition of ethanol [118]. In this study, it was found turbulence. The problem is even more pronounced in the case
that the addition of simulated EGR did not affect significantly the of Diesel-like HCCI engines, in which the mixture formation is
ethanol autoignition timing for the conditions examined. The effect achieved via direct fuel injection within the combustion cham-
of EGR on other fuels, such as isooctane and PRF, was more pro- ber. Direct fuel injection imposes new demands on any simulation
nounced. This was mainly owed to the low sensitivity of ethanol model, since the mixture – and the resulting temperature – strati-
oxidation on oxygen concentration, which was reduced with the fication within the combustion chamber is affected significantly by
 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610 1603

Fig. 12. Cylinder locations according to the injection timing and five injection angles
Fig. 11. Zone temperature distributions produced by Röhl and Peters’ (solid line)
[120].
and Marinov’s (dashed line) mechanisms for the high load ethanol case ( = 0.2641)
[113].
The model was validated against experimental measurements
at three injection timings, with an injection angle of 156◦ . The
the injection characteristics. Phenomena such as wall wetting, fuel
experimental and simulation results at an injection timing of 60◦
dispersion, fuel vaporization under the low pressure-low temper-
BTDC had two peaks in the heat release rate, indicating low- and
ature conditions encountered during early direct injection, must
high-temperature DME oxidation. The validated model was subse-
be described. In these cases, the incorporation of accurate models
quently used for a numerical study aimed at analyzing the effect
for the estimation of the physicochemical fuel properties becomes
of the injection angles and timings shown in Fig. 12. The result-
crucial, especially for biofuels, since their development and use has
ing fuel spray patterns created by these configurations are shown
a relatively short history.
in Fig. 13. These patterns revealed fuel spray-wall interactions and
The aforementioned issues can be dealt with by using multi-
accumulation of fuel in the squish area and, possibly, the crevice
dimensional models, i.e. CFD models, which describe in detail the
regions.
charge motion within the combustion chamber and the turbulence
Generally, the fuel injected vaporized rapidly as it entered the
effects, provide a superior spatial resolution of the combustion
combustion chamber due to the high volatility of DME. At an early
chamber, and can account for the description of DI HCCI combustion
injection timing (60◦ bTDC), the variation of the injection angle
by incorporating the dynamics and kinematics of the fuel injection.
affected significantly the pressure and heat release rate diagrams;
In the remaining part of this section, direct injection HCCI CFD
at wider injection angles the peak pressure and HRR were both
models will be discussed first while dedicated HCCI CFD models
reduced, since some of the injected fuel reached the relatively cold
assuming homogeneous conditions at IVC will be presented subse-
crevice and combustion chamber wall regions. At later injection
quently.
timings, the effect of the injection angle was much less pronounced,
since the fuel was injected within the piston bowl. As the injection
5.1. Direct injection HCCI timing approached TDC, the combustion mode shifted progres-
sively from partially homogeneous to diffusive combustion. Soot
Kim et al. [120] investigated the influence of the spray angle emissions were very low for all cases examined due to the chemi-
and advanced injection timing on the combustion performance cal structure of the fuel, which excludes C–C bonds. Another reason
and emission characteristics in a DME fueled compression ignition was the relatively premixed nature of the mixture, especially at
engine. They used the KIVA software [121], which was supple- early injection timings.
mented with liquefied DME fuel properties and their temperature NOx emissions were very low for the early injection timings (IT
dependence [122–126]. The oxidation process of the DME fuel 40◦ bTDC and earlier) due to the premixed nature of the mixture.
was described with a detailed chemical kinetic model [53–55], As the injection timing approached TDC, NOx emissions increased,
which consisted of 351 reversible elementary reactions among 79 probably owing to the predominance of the in-cylinder inhomo-
species. This mechanism was expanded by including NOx forma- geneities. At advanced injection timings and/or wide injection
tion reactions taken from a modified version of GRI Mech [84]. A angles, the HC and CO emissions were higher due to the introduc-
two-step phenomenological soot model [127] was also applied to tion of fuel into the squish and crevice regions. At narrow angles, the
calculate soot emissions. The DME spray and atomization char- high HC and CO emissions were due to the fuel impingement on the
acteristics were simulated via the Kelvin–Helmholtz (KH) and piston bowl. As regards performance, a narrow angle produced low
Rayleigh–Taylor (RT) hybrid breakup model [128]. Turbulence was ISFC at relatively advanced injection timing. Other studies focusing
accounted for via the modified Renormalization Group (RNG) k–ε solely on the spray characteristics of DME by the research group of
model [129], and wall impingement was simulated via the spray the Hanyang University can be found in [131,132]
wall interaction model [130] including droplet splash/spreading. A Chen et al. [133] investigated experimentally and numerically
computational grid with a unit cell size of about 2 mm3 was formed the effect of methanol injection timing in a DME/MeOH dual
for the evaluation of spatial distribution in the cylinder and the fuel engine, in which the DME was introduced at the inlet port.
computational grid was divided into a 60◦ sector mesh, due to the For the numerical simulation of the closed part of the engine
symmetric arrangement of the six-hole injector. The calculation operation, the researchers coupled chemical kinetics with the CFD
time step was set at 0.1 ␮s. code FIRE [134]. The chemical mechanism used to estimate the
1604 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610

Fig. 14. Computational mesh of combustion chamber at TDC used in [138].

at −6◦ aTDC. They found that increasing the injection pressure from
20 to 30 MPa shortened the combustion duration, increased the
thermal efficiency, and induced an increase in peak temperature
and NO concentration. However, it was difficult to experimentally
examine the effect of increased injection pressure due to the low
viscosity of methanol.

5.2. Fully premixed HCCI

In Ref. [137] Yao et al. examined the HCCI combustion of a dual

fuel engine fueled with DME/MeOH blends. They used the STAR-
CD/KINETICS CFD code to simulate the combustion of the charge
that was uniform at IVC. The RNG k–ε model was included to
account for the effects of turbulence, and the PISO algorithm was
used for the transient flow of the engine. The CFD code was coupled
with the CHEMKIN code to obtain the species concentration evolu-
tion, and the DME/methanol oxidation was described by a reduced
mechanism consisting of 27 species and 35 reactions [50]. The time
step corresponded to 0.1◦ CA at an engine speed of 1400 rpm. The
researchers validated the model results against an experimentally
obtained pressure diagram and found adequate agreement with
the measurements. Under the same experimental conditions they
also conducted an investigation into the formation of HC and CO
Fig. 13. Effect of the injection angle and timing on calculated spray and evaporation emissions. They found that UHC emissions mainly consisted of the
distributions in the cylinder according to the crank angle [120].
unburned fuels and CH2 O. The unburned DME and methanol mainly
resided in the piston-ring crevice region, while most of the CH2 O
methanol/DME reaction rates consisted of 30 species and 44 reac- emissions were located next to the cylinder-liner wall. CO emis-
tions including the fuel chemistry [50], the Zeldovich mechanism, sions were mainly found in the regions near the top surface of the
and the N2 O intermediate mechanism [82]. The effect of turbulence piston.
was accounted for by adopting a compressible version of the stan- The authors also studied the effect of load (in terms of equiva-
dard two-equation k–ε turbulence model. The source terms in the lence ratio) and the relative proportion of DME in the fuel blend on
transport equations of FIRE were calculated with the assumption the HC and CO emissions. Increasing the DME ratio, while main-
that the reaction rate is mainly determined by kinetic and turbulent taining the overall equivalence ratio, improved the combustion
time scales. Major submodels used in the computation included efficiency and reduced the HC and CO emissions. Moreover at all
the WALLJET1 wall interaction model, the Dukowicz evaporation loads examined, a critical value for the DME ratio was found; when
model, and the HuhGosman breakup model [135]. The Han-Reitz reducing the ratio below this value induced a significant increase in
wall heat transfer model was used to estimate the wall heat flux both CO and HC due to incomplete combustion. However, the criti-
[136]. The spray angle was 152◦ formed by a 7-hole injector, which cal DME ratio was not the same for all equivalence ratios examined.
allowed for the simulation of a 51◦ sector consisting of 30,000 At high equivalence ratios, the critical DME ratio shifted toward
grid cells at BDC. The simulations examined the effect of MeOH higher values due to the increase in methanol concentration at
injection at two injection timings, namely −26◦ and −6◦ aTDC. high loads, which induced an inhibiting effect on DME ignition.
It was found that methanol injected at −26◦ aTDC had no A critical value was also found for the peak combustion temper-
effect on the low temperature reaction of DME but accelerated the ature (1400 K), above which the HC and CO emissions were greatly
heat release of the high temperature reaction, producing locally reduced. When the load, i.e. the overall equivalence ratio, was
high temperature regions and increasing NO concentration. In this increased while maintaining the DME proportion, HC and CO emis-
case, the injected methanol was completely oxidized. At the later sions were reduced.
injection timing (−6◦ aTDC) the DME found in the crevices was Using roughly the same simulation model, Huang et al. [138]
higher, and the high temperature region was dispersed in the studied numerically the HCCI combustion of a DME fueled engine.
clearance volume and the region near the combustion chamber A reduced chemical mechanism for DME HCCI combustion pro-
wall, leading to lower NO concentration. Some methanol vapor cess was adopted, which was proposed by Yao et al. [48] and Liang
remained in the combustion chamber prolonging the combustion et al. [49]. The reduced chemical mechanism was derived from the
duration. This led to an investigation aimed at decreasing the com- detailed chemical mechanism and consisted of 26 species and 28
bustion duration at this – relatively late – injection timing. The reactions. A single layer grid was applied with an average cell size
researchers studied numerically the effect of injection pressure on of about 2 mm, yielding about 1658 cells at IVC in the combustion
the combustion duration, while maintaining the injection timing chamber (Fig. 14).
 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610 1605

Fig. 15. Initial locations for LTR and HTR in [138].

The computational time for the closed part of the engine cycle due to mass transfer, indicating that their concentration can be
was approximately 18 h in a Pentium IV 3.0 GHz CPU PC with 1 GB affected by the combustion chamber configuration and the exist-
memory. Under homogeneous charge IVC conditions, the DME HCCI ing flow field. At EVO a relatively high concentration of these gases
combustion process does not take place in the entire combustion remained near the wall.
chamber simultaneously. The low temperature reaction was initi- In Ref. [139] Park numerically analyzed the combustion and
ated primarily at the region near the piston surface and the squish emission characteristics and the optimal operating conditions of
region, and subsequently moved toward the cylinder head. The homogeneous charge compression ignition (HCCI) engines fueled
high temperature reaction was initiated in the combustion cham- with dimethyl ether (DME). For this investigation, a modified
ber core zone and squish region, and next proceeded to the rest of version of KIVA-3V code [121] was used to simulate DME HCCI
the combustion chamber (Fig. 15). combustion with a two-dimensional computational mesh. The
The majority of HC emission was unburned fuel and CH2 O, CHEMKIN chemistry solver [70] was integrated into KIVA-3V, in
and resided with CO in the bottom, middle, and upper part of order to solve the chemistry and the chemical mechanism used for
the piston-ring crevice regions, respectively. At very low loads, the simulation of DME consisted of 351 reactions and 79 species
DME was partially oxidized producing unburned HC and CO. As [53–55]. A two-step phenomenological model was used to simu-
the load increased, the unburned HC decreased and the CO emis- late soot emission, since the study extended the investigation into
sions increased and then decreased. This behavior can be explained rich DME combustion. The soot model considered both soot for-
considering the successive steps through which the hydrocarbon mation, using a modified Hiroyasu model that considers acetylene
oxidation takes place, i.e. the oxidation follows the general scheme (C2 H2 ) as a soot precursor [140], and soot oxidation employing
HC → CO → CO2 . At extremely low loads the fuel and any unburned the Nagle–Strickland–Constable model [141]. For NOx emissions,
HC produced by its decomposition fails to be converted to CO to a a mechanism derived from the Gas Research Institute (GRI) NO
great extent. At higher loads, the HC to CO conversion takes place mechanism [84] was used.
more vividly and, at the highest attainable loads, both HC and CO are During the simulations, the charge was assumed homogeneous
oxidized to produce CO2 through the aforementioned generalized at intake valve closure. The researcher generated an equivalence
oxidation scheme. ratio versus peak temperature operating map based on numeri-
Chun-Ian et al. [56] simulated neat DME HCCI combustion. They cal simulations, and compared the DME operation to n-heptane
used a single-zone model to compare two chemical mechanisms: operation. High CO concentration was found at low peak combus-
a detailed mechanism of 97 species and 457 reactions, which was tion temperatures and rich mixtures. CO emissions for DME were
obtained by combining the mechanism from Curran et al. [54] with greater than those for n-heptane, because of a lower initial mole
the NOx formation reactions of Marinov [85], and a simplified reac- fraction of oxygen. HC emissions for DME and n-heptane were quite
tion mechanism of 36 species and 73 reactions. The latter included similar at lean mixtures. However, HC emissions for DME were
a low temperature, a negative temperature coefficient, and a pyrol- lower than those for n-heptane below a peak cycle temperature
ysis and oxidation submodel along with the reaction path for NOx of 1600 K and an equivalence ratio higher than 1.5. Although such
formation. operating regimes are outside the usual HCCI operating conditions,
The FLUENT CFD software was also used to obtain insight into this finding indicated that at low temperature-high equivalence
the formaldehyde and formic acid formation process during the ratio combustion conditions HC emissions can be reduced by sub-
DME HCCI combustion. For this reason the CFD code was coupled stituting DME for diesel fuel. The optimal operating range for the
with the three submodels, i.e. the low temperature, the negative best ISFC or imep is at low peak cycle temperature and below stoi-
temperature coefficient, and the pyrolysis and oxidation submod- chiometric for both fuels, because ignition is near TDC for that area.
els. The RNG k–ε turbulence model and the EDC combustion Total heat release of DME is close to that of n-heptane in this study.
model were implemented in the CFD code. Using this model, the The optimal operating range, which was based on a merit func-
researchers found that both formaldehyde and formic acid existed tion including emissions and performance, was between 1700 and
in the exhaust. Their concentration was close to the experimen- 1850 K peak cycle temperature and below 0.5 equivalence ratio for
tally determined, in the case of formaldehyde, but the measured DME HCCI engines.
formic acid was much less than the one predicted. The CFD cal- Kong [142] studied numerically and experimentally the dual
culations showed that the formation of these species was located fuel DME-methane oxidation in an HCCI engine. The researcher
at the region near the combustion chamber wall and above the used an improved version of the CFD code KIVA-3V [121]. The
ring crevices. Their formation originated from the partial oxida- CHEMKIN chemistry solver [94] was integrated into the KIVA-3V
tion of DME, which had entered the crevices during combustion code for solving the chemistry during multi-dimensional engine
and flowed back to the combustion chamber during expansion. The simulations. A detailed reaction mechanism for DME [53–55] was
partial oxidation of DME gave rise to formaldehyde and formic acid, used to simulate fuel chemistry, consisting of 79 species and 351
the further oxidation of which was suppressed due to the low local reactions. This mechanism was enriched by NOx -related reactions
temperatures induced by the gas expansion and heat transfer to the obtained from a NO mechanism, which was derived from the Gas
cylinder wall. Some of the formaldehyde and formic acid formed Research Institute (GRI) NO mechanism [84]. The final scheme con-
were transferred to the central part of the combustion chamber, sisted of 83 species and 360 reactions. These reactions included the
1606 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610

oxidation chemistry of methane, which was used to simulate nat- engine. These factors prolonged the combustion duration, thereby
ural gas. Turbulence was modeled via the RNG k–ε model and a decreasing the maximum pressure rise rate and the peak pressure.
new wall function was constructed to calculate the wall heat flux, Viggiano and Magi [145] used a multidimensional CFD model
which accounted for the variation of the gas density and the turbu- coupled with chemical kinetics to study ethanol HCCI combustion.
lent Prandtl number in the boundary layer assuming quasi-steady The model used was the REC-200, which solves the 3D Reynolds
conditions. averaged Navier–Stokes equations for transient, two-phase, tur-
Using this model, the researcher found that as DME concen- bulent, chemically reactive flows with sprays [146]. Turbulence
tration decreased ignition delayed and combustion duration was was modeled with a two-equation k–ε model, and the heat and
prolonged. In contrast to methane, DME experienced two-stage momentum fluxes were calculated using the wall function model of
ignition and was depleted before any significant oxidation of Launder and Spalding [147]. Combustion was modeled by coupling
methane. During the first stage of heat release, a noticeable amount the turbulent flow to a detailed reaction mechanism. The approach
of CO was formed, which was later increased during combustion used for the coupling was that of Kong and Reitz [148]. The influence
and decreased rapidly only after both methane and DME were of turbulence on the reaction rates was accounted for by multiply-
nearly consumed. The combustion of methane was induced by the ing the latter with a correction factor, which accounted for both
high temperatures resulting from DME combustion. NOx emissions the kinetic and turbulence timescales. The detailed reaction mech-
were much lower than those in conventional diesel engines, due anism used was that of Saxena and Williams [59], which included
to the low combustion temperature. It was found that there was a 43 species and 235 reactions excluding NOx formation reactions.
minimum requirement in DME concentrations for HCCI combus- The structured grid of the 1-degree sector considered, consisted
tion regardless of natural gas concentrations, in order to obtain of 26 cells in the radial direction. In the axial direction, the cell
stable combustion. Moreover, the operating limits – defined by number varied from 48 at BDC to 8 at TDC by removing cell rows
misfire and knocking combustion – narrowed as the methane con- during compression. The grid resolution near the wall was kept
centration increased, implying that high methane concentrations at 0.2 mm. After the validation of the CFD model against experi-
induce combustion instability under the conditions examined. This ments was accomplished, the researchers investigated the effect
was probably the result of the rapid combustion of methane, which of turbulent timescale, turbulent diffusivity, initial and wall tem-
was triggered by the DME oxidation and prevailed as the amount perature, and equivalence ratio on HCCI combustion. They found
of methane increased. that increasing the turbulent timescale induced a reduction to the
Yu et al. [143] used large eddy simulation (LES) to simulate combustion rate and decreased the pressure rise rate. Increasing
ethanol HCCI combustion, in which the ethanol was port injected. the initial diffusivity delayed ignition and deteriorated the com-
The main scope of the study was to investigate the role of turbu- bustion rate. Inlet and wall temperature affected combustion in a
lence in the HCCI combustion process. The chemical mechanism similar manner, with the effect of inlet temperature being more
used was that of Marinov [60]. The LES simulations were conducted pronounced; specifically, increasing Tin or Tw advanced ignition,
for two different piston configurations, i.e. a disc shaped combus- enhanced combustion, and reduced CO emissions.
tion chamber and a bowl in piston configuration, while maintaining In Ref. [149] Liu and Karim examined numerically the effect of
the compression ratio. A grid consisting of 128 × 128 × 128 mesh various operating parameters on hydrogen HCCI combustion. For
cells was used, with size about 1 mm in the radial direction and the purposes of this investigation, the KIVA3 multi-dimension CFD
0.32–1.5 mm in the axial direction for the square bowl-in-piston code [121] was adopted with some modifications to make it suit-
case and 0.075–1.3 mm for the disc-shaped combustion chamber able for modeling the turbulent combustion processes. Detailed
case. This mesh size was of the order of Taylor microscale and at chemical kinetics of hydrogen oxidation was implemented into
TDC close to the Kolmogorov microscale. Eight parallel processors the code, including 20 reactions with additional reactions involv-
and 48 h were necessary to yield a full cycle simulation for a four- ing nitrogen [71,72]. With this model, the researchers conducted
stroke engine. The simulation results, which were validated against a parametric study, the results of which showed that by increas-
experiments, showed that the piston geometry affected the HCCI ing the swirl ratio heat transfer was enhanced, thereby reducing
combustion process; the square bowl-in-piston design had a much both the minimum and maximum temperature within the com-
slower combustion. Although the levels of turbulence in terms of bustion chamber. As a result, ignition was retarded, combustion
the velocity components were similar, the temperature inhomo- duration was prolonged, and the peak pressure and NOx emissions
geneities were quite different between the two configurations. This were reduced. An increase in engine speed increased the heat loss
was the reason for the longer combustion duration in the case of rate due to the induced increase of the swirl ratio, but the heat
square bowl-in-piston design. losses were reduced because the time available for heat transfer
A similar study was later conducted by Joelsson et al. [144]. In was decreased. The calculations showed that there was an opti-
this study, the researchers compared numerically and experimen- mum engine speed in terms of indicated thermal efficiency. Initial
tally the ethanol HCCI combustion in a metal and an optical engine. temperature and compression ratio affected significantly the onset
The engines were essentially identical, the main difference being of combustion as well as the combustion duration.
the piston material; a metal piston was used in the metal engine Cotino and Jeanmart [150] studied numerically the running zone
and a quartz piston head in the optical engine. It was expected that of ethyl acetate and compare it with the running zone of isooctane.
the quartz piston engine would operate at a higher piston tempera- Ethyl acetate was chosen, since it is a product of the acidogenic
ture, since the quartz piston was air cooled, in contrast to the metal fermentation process used to produce biodiesel. This molecule was
piston that was oil cooled. The scope of the study was to investi- the basic compound and was used as a starting point to character-
gate the resulting thermal stratification differences and their effect ize a group of esters produced. The model used in the numerical
on ethanol HCCI combustion. The same LES model was used in this study was the commercial CFD FLUENT code and the computa-
study, as in the previous one. The expected difference in the piston tional grid was obtained from the commercial software GAMBIT.
temperatures was accounted for by increasing the piston surface The combustion mechanism used for ethyl acetate was a reduced
temperature in the optical HCCI engine simulation. The simulation one, consisting of 41 species and 203 reactions. This mechanism
results showed that the two engines had minor differences in the was derived from more detailed ones, namely the Gasnot et al.
induced turbulence field. However, the thermal stratification was ethyl acetate submechanism [65], and the low carbon combustion
much higher in the optical engine, due to the hotter quartz piston mechanism of Miller and Melius [151]. Turbulence was modeled
temperature and the lower intake temperature used in the optical with the RNG k–ε model and the molecular mixing at the subgrid
 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610 1607

level via the eddy dissipation model (EDC). Except for swirl motion, a chemical mechanism, depending on the conditions under which
3D effects were not included and an axisymmetric geometry was its validation was conducted. These conditions are preferably
chosen, yielding 33,000 cells at IVC. high pressure and low equivalence ratio, since such conditions
The model was validated against experimental results using are encountered in HCCI combustion. It would also be advisable
isooctane as the fuel, and was subsequently used with ethyl acetate to implement more than one chemical mechanism for the same
to determine the ethyl acetate HCCI operating limits (running zone) biofuel in any HCCI simulation model, and compare the results.
at 1200 rpm. The upper load limit was determined by the maximum This helps to avoid reaching conclusions (e.g. on autoignition, or
pressure rise rate, which was set at 5 bar/CAD. The lower limit cri- emissions formation), which are the result of the specific chemical
terion implied combustion instability and was determined taking mechanism used rather than the HCCI combustion process.
into account the combustion timing, the EGR rate, the sensitivity of Single-zone models for biofuel HCCI combustion are very use-
the combustion on the initial temperature, the compatibility of the ful in identifying reaction paths and dominant chemical species
final and initial temperatures, and the CO emissions. The results of during combustion. They have been used to provide insight into
the study showed that the ethyl acetate had a similar running zone the behavior of various biofuels and blends thereof, and to validate
range as isooctane, but it was shifted toward higher loads. More- the corresponding combustion mechanisms versus experiments or
over, ethyl acetate presented extended low temperature reactions other more detailed mechanisms. Owing to the zero dimensional
compared to isooctane. nature of the single-zone models, they have been used mainly to
Um and Park [152] analyzed the chemical effects of biodiesel provide insight into the autoignition properties of the fuel and the
on combustion and emissions of an HCCI engine fueled with formation of NOx emissions. Single-zone models also provide an
diesel/biodiesel blends. For this purpose, they used a modified ver- indication on the effect of various configurations, such as nega-
sion of the KIVA-3V code [121] in a 2D computational mesh, coupled tive valve overlap, heavy EGR, fuel reforming, etc., on biofuel HCCI
with CHEMKIN chemistry solver [70]. The mechanism for biodiesel combustion.
oxidation used was the one suggested by Brakora et al. [78], and Multi-zone models with their added feature of accounting for
was modified by Um and Park [153] to produce a reasonable igni- mixture thermal and fuel stratification, provide better and more
tion delay. The mechanism consisted of 156 reactions among 54 realistic estimates on combustion duration, peak combustion pres-
species. Since it was found that neat methyl butanoate was not sure and emissions formation. Multi-zone models for biofuel HCCI
suitable as a biodiesel surrogate, due to its high oxygen content, it combustion have been used to study the effect of various biofuels
was assumed that 1 mol of biodiesel consisted of 1 mol of methyl and their blends on the in-cylinder thermal stratification, combus-
butanoate and 2 mol of n-heptane. Neat heptane was also used as a tion duration, emissions formation, and the overall performance.
diesel surrogate. A NOx mechanism, reduced from the Gas Research Not only can they provide an insight into the formation of CO and
Institute (GRI) NO mechanism [84], was used for this study. It con- HC emissions, but also to the specific species formed and consumed
sisted of 12 reactions and 4 species (i.e., N, NO, NO2 , and N2 O). in the low temperature regions of the combustion chamber, thereby
Moreover, a soot formation and oxidation model was incorporated, indicating the species expected to be found in the exhaust of a HCCI
since rich mixtures with equivalence ratio up to 2.0 were used. The engine. This is an important feature, since the composition of HC
high equivalence ratio, which is impractical in actual HCCI engines, emissions can vary substantially in the case of biofuels, due to their
is justified considering the nature of the study, which was to exam- non-conventional molecular structure.
ine the chemical properties of the biodiesel blends. Soot formation Multi-dimensional models provide an even higher spatial res-
was calculated using a modified Hiroyasu model, which considers olution of the combustion chamber than the multi-zone ones. Gas
acetylene (C2 H2 ) as a soot precursor [140], while soot oxidation flow during the gas exchange process or within the combustion
was modeled using the Nagle–Strickland–Constable model [141]. chamber can be modeled in more detail, providing a more realis-
The results of the study revealed that at stoichiometric equiva- tic spatial resolution of the species evolution during combustion.
lence ratio, increasing the biodiesel-to-diesel fuel ratio advanced Moreover, mass flow between hot and cold regions is described in
ignition and decreased the total amount of heat released. This detail, providing an insight into the emissions formation process.
reduced the in-cylinder temperatures and induced a decrease in Especially in the case of direct injection biofueled HCCI combus-
NOx emissions with an increase in CO and HC. At richer mixtures tion, the physical properties of the fuel must be described in detail;
( > 1.0) increasing the biodiesel content increased CO, decreased properties such as viscosity, volatility, etc., are essential to describe
NOx and HC emissions, and decreased Soot emissions compared to the evolution of the fuel spray into the combustion chamber. The
neat diesel fuel operation. Below stoichiometric conditions ( < 1.0) importance of these properties cannot be overemphasized, since
CO and HC were formed in the cases where peak cycle temperatures the spray evolution will determine not only the fuel stratification
were below 1400 K, NOx was produced at peak temperatures higher within the combustion chamber but also the fuel impingement on
than 1800 K, while Soot emissions were minimal. At rich mixtures the cylinder liner or the piston crown. The description of the physic-
significant amounts of CO and HC emissions were observed for most ochemical properties of biofuels is an active area of research, which
cases and soot emissions, which were higher for neat diesel, peak- is bound to be further advanced.
ing in the cases with a peak cycle temperature of 1600 K. The NOx
and CO emission maps, i.e. equivalence ratio versus peak tempera- References
ture maps, were similar for the diesel fuel and biodiesel cases.
[1] Energy Statistics, Eurostat. http://epp.eurostat.ec.europa.eu/portal/page/
portal/energy/data/database [accessed on 31.10.11].
[2] Escobar JC, Lora ES, Venturini OJ, Yáñez EE, Castillo EF, Almazan O. Bio-
6. Conclusions
fuels: environment, technology and food security. Renew Sust Energy Rev
2009;13:1275–87.
Chemical kinetics are an essential part of biofueled HCCI com- [3] BP Statistical Review of World Energy, June 2010, http://www.bp.com/
sectiongenericarticle.do?categoryId=9033088&contentId=7060602
bustion modeling, owing to the nature of HCCI combustion and
[accessed on 1.6.11].
the diversity of the biofuels’ molecular structure. The development [4] Cherubini F, Bird ND, Cowie A, Jungmeier G, Schlamadinger B, Woess-
of chemical reaction mechanisms is an active research area, from Gallasch S. Energy- and greenhouse gas-based LCA of biofuel and bioenergy
which biofuel HCCI research can benefit. Efforts should be focused systems: key issues, ranges and recommendations. Resour Conserv Recycl
2009;53:434–47.
on the consistency of the various mechanisms developed for use in [5] Food and Agriculture Organization of the United Nations. http://www.
HCCI engines. Special care should be exercised on the selection of fao.org/docrep/012/al390e/al390e00.pdf [accessed on 1.6.10, also in:
1608 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610

The State of Food Insecurity in the World, 2010, http://www.fao.org/ [36] Zheng ZQ, Yao MF, Chen Z, Zhang B. Experimental study on HCCI combustion
docrep/013/i1683e/i1683e.pdf]. of dimethyl ether (DME)/methanol dual fuel, SAE paper no. 2004-01-2993;
[6] http://esa.un.org/unpd/wpp/Excel-Data/population.htm [accessed on 2004.
1.6.10]. [37] Jothi NKM, Nagarajan G, Renganarayanan S. Experimental studies on homo-
[7] Naik SN, Goud VV, Rout PK, Dalai AK. Production of first and second generation geneous charge CI engine fueled with LPG using DEE as an ignition enhancer.
biofuels: a comprehensive review. Renew Sust Energy Rev 2010;14:578–97. Renew Energy 2007;32:1581–93.
[8] Rakopoulos DC, Rakopoulos CD, Papagiannakis RG, Kyritsis DC. Combustion [38] Szybist JP, McFarlane J, Bunting BG. Comparison of simulated and experimen-
heat release analysis of ethanol or n-butanol diesel fuel blends in heavy-duty tal combustion of biodiesel blends in a single cylinder diesel HCCI engine, SAE
DI diesel engine. Fuel 2011;90:1855–67. paper no. 2007-01-4010; 2007.
[9] Rakopoulos CD, Rakopoulos DC, Giakoumis EG, Dimaratos AM. Investiga- [39] Jia M, Xie M. A chemical kinetics model of iso-octane oxidation for HCCI
tion of the combustion of neat cottonseed oil or its neat bio-diesel in a engines. Fuel 2006;85:2593–604.
HSDI diesel engine by experimental heat release and statistical analyses. Fuel [40] Komninos NP, Rakopoulos CD. Numerical investigation into the forma-
2010;89:3814–26. tion of co and oxygenated and nonoxygenated hydrocarbon emissions
[10] Rakopoulos DC, Rakopoulos CD, Giakoumis EG, Papagiannakis RG, Kyritsis from isooctane- and ethanol-fueled HCCI engines. Energy Fuels 2010;24:
DC. Experimental-stochastic investigation of the combustion cyclic variabil- 1655–67.
ity in HSDI diesel engine using ethanol–diesel fuel blends. Fuel 2008;87: [41] Dagaut P, Togbé C. Experimental and modeling study of the kinetics of oxi-
1478–91. dation of butanol–n-heptane mixtures in a jet-stirred reactor. Energy Fuels
[11] Rakopoulos CD, Antonopoulos KA, Rakopoulos DC, Hountalas DT. Multi-zone 2009;23:3527–35.
modeling of combustion and emissions formation in DI diesel engine operat- [42] Black G, Curran HJ, Pichon S, Simmie JM, Zhukov V. Bio-butanol: combus-
ing on ethanol–diesel fuel blends. Energy Convers Manage 2008;49:625–43. tion properties and detailed chemical kinetic modeling. Combust Flame
[12] Rakopoulos CD, Antonopoulos KA, Rakopoulos DC. Experimental heat release 2010;157:363–73. http://c3.nuigalway.ie/biobutanol.html.
analysis and emissions of a HSDI diesel engine fueled with ethanol–diesel fuel [43] Dagaut P, Togbé C. Oxidation kinetics of butanol–gasoline surrogate mix-
blends. Energy 2007;32:1791–808. tures in a jet-stirred reactor: experimental and modeling study. Fuel
[13] Rakopoulos CD, Antonopoulos KA, Rakopoulos DC. Development and applica- 2008;87:3313–21.
tion of multi-zone model for combustion and pollutants formation in direct [44] Saisirirat P, Togbé C, Chanchaona S, Foucher F, Mounaim-Rousselle C,
injection diesel engine running with vegetable oil or its bio-diesel. Energy Dagaut P. Auto-ignition and combustion characteristics in HCCI and JSR
Convers Manage 2007;48:1881–901. using 1-butanol/n-heptane and ethanol/n-heptane blends. Proc Combust Inst
[14] Rakopoulos CD, Antonopoulos KA, Rakopoulos DC. Multi-zone modeling of 2011;33:3007–14.
diesel engine fuel spray development with vegetable oil, bio-diesel or diesel [45] Mack JH, Dibble RW, Buchholz BA, Flowers DL. The effect of the di-tertiary
fuels. Energy Convers Manage 2006;47:1550–73. butyl peroxide (DTBP) additive on HCCI combustion of fuel blends of ethanol
[15] Rakopoulos CD, Rakopoulos DC, Giakoumis EG, Kyritsis DC. The combustion and diethyl ether, SAE paper no. 2005-01-2135; 2005.
of n-butanol/diesel fuel blends and its cyclic variability in a DI diesel engine. [46] Yasunaga K, Simmie JM, Curran HJ, Koike T, Takahashi O, Kuraguchi Y,
Proc Inst Mech Eng Part A: J Power Energy 2011;225:289–308. et al. Detailed chemical kinetic mechanisms of ethyl methyl, diethyl,
[16] Arcoumanis C, Bae C, Crookes R, Kinoshita E. The potential of di-methyl ether methyl tert-butyl and ethyl tert-butyl ethers: The importance of uni-
(DME) as an alternative fuel for compression-ignition engines: a review. Fuel molecular elimination reactions. Combust Flame 2011;158:1032–6.
2008;87:1014–30. http://c3.nuigalway.ie/ethers.html.
[17] Szybist JP, Song J, Alam M, Boehman AL. Biodiesel combustion, emissions and [47] Yamada H. Simplified oxidation mechanism of DME applicable for compres-
emission control. Fuel Process Technol 2007;88:679–91. sion ignition, SAE paper no. 2003-01-1819; 2003.
[18] Rakopoulos CD, Kosmadakis GM, Demuynck J, De Paepe M, Verhelst S. A com- [48] Yao MF, Qin J, Zheng ZQ. Numerical study of the combustion mechanism of
bined experimental and numerical study of thermal processes, performance a homogeneous charge compression ignition engine fuelled with dimethyl
and nitric oxide emissions in a hydrogen-fueled spark-ignition engine. Int J ether and methane, with a detailed kinetics model. Part 1: the reaction
Hydrogen Energy 2011;36:5163–80. kinetics of dimethyl ether. Proc Inst Mech Eng Part D: J Automob Eng
[19] Rakopoulos CD, Kosmadakis GM, Pariotis EG. Evaluation of a combustion 2005;219:1213–23.
model for the simulation of hydrogen spark-ignition engines using a CFD code. [49] Liang X, Yao MF, Zheng ZQ. Numerical study on the mechanism of low tem-
Int J Hydrogen Energy 2010;35:12545–60. perature reaction of DME/methanol HCCI combustion. J Combust Sci Technol
[20] Rakopoulos CD, Michos CN. Generation of combustion irreversibilities in a 2005;11:149–54 (in Chinese).
spark ignition engine under biogas–hydrogen mixtures fueling. Int J Hydrogen [50] Liang X. Numerical study on HCCI combustion of DME/MEOH dual fuels. Mas-
Energy 2009;34:4422–37. ter’s Dissertation, Tianjin University, Tianjin, China, 2005.
[21] Ji C, Liang C, Wang S. Investigation on combustion and emissions of DME [51] Kim H, Cho S, Min K, Reduced chemical kinetic model of DME for HCCI com-
gasoline mixtures in a spark-ignition engine. Fuel 2011;90:1133–8. bustion, SAE paper no. 2003-01-1822 (JSAE 20030263); 2003.
[22] Alkidas AC. Combustion advancements in gasoline engines. Energy Convers [52] Curran HJ, Pitz WJ, Marinov NM, Westbrook CK, Dagaut P, Boettner J-C, et al.
Manage 2008;48:2751–61. A wide range modeling study of dimethyl ether oxidation. Int J Chem Kinet
[23] Yao M, Zheng Z, Liu H. Progress and recent trends in homogeneous 1998;30:229–41.
charge compression ignition (HCCI) engines. Prog Energy Combust Sci [53] Fischer SL, Dryer FL, Curran HJ. The reaction kinetics of dimethyl ether. I:
2009;35:398–437. High-temperature pyrolysis and oxidation in flow reactors. Int J Chem Kinet
[24] Stanglemaier RH, Roberts CE. Homogeneous Charge Compression Ignition 2000;32:713–40.
(HCCI): benefits, compromises and future engine applications, SAE paper no. [54] Curran HJ, Fischer SL, Dryer FL. The reaction kinetics of dimethyl ether. II: Low-
1999-01-3682; 1999. temperature oxidation in flow reactors. Int J Chem Kinet 2000;32:741–59.
[25] Lu X, Han D, Huang Z. Fuel design and management for the control of [55] Kaiser EW, Wallington TJ, Hurley MD, Platz J, Curran HJ, Pitz WJ,
advanced compression-ignition combustion modes. Prog Energy Combust Sci et al. Experimental and modeling study of premixed atmospheric-pressure
2011;37:741–83. dimethyl ether–air flames. J Phys Chem A 2000;104:8194–206, avail-
[26] Olsson J-O, Tunestål P, Johansson B. Closed-loop control of an HCCI engine, able at https://www-pls.llnl.gov/?url=science and technology-chemistry-
SAE paper no. 2001-01-1031; 2001. combustion-dme [accessed on 8.3.11].
[27] Antunes JMG, Mikalsen R, Roskilly AP. An investigation of hydrogen- [56] Chun-lan M, Yu-sheng Z, Yu S, Jian H, Hai-ying S. Experimental and numerical
fuelled HCCI engine performance and operation. Int J Hydrogen Energy study on emission in an HCCI Engine operated with neat dimethyl ether, SAE
2008;33:5823–8. paper no. 2007-01-1888; 2007.
[28] Christensen M, Einewall P, Johansson B. Homogeneous charge compression [57] Röhl O, Peters N. A reduced mechanism for ethanol oxidation. In: European
ignition (hcci) using isooctane, ethanol and natural gas—a comparison with Combustion Meeting. 2009.
spark-ignition operation, SAE paper no. 972874; 1997. [58] Li J, Kazakov A, Chaos M, Dryer FL. Chemical kinetics of ethanol oxidation,
[29] Christensen M, Johansson B, Amneus PJH, Mauss F. Supercharged homoge- Paper no. C26, Fifth US Combustion Meeting, San Diego; 2007.
neous charge compression ignition, SAE paper no. 980787; 1998. [59] Saxena P, Williams FA. Numerical and experimental studies of ethanol flames.
[30] Yap D, Megaritis A. Applying forced induction to bioethanol HCCI operation Proc Combust Inst 2007;31:1149–56.
with residual gas trapping. Energy Fuels 2005;19:1812–21. [60] Marinov NM. A detailed chemical kinetic model for high temperature ethanol
[31] Megaritis A, Yap D, Wyszynski ML. Effect of inlet valve timing and water oxidation. Int J Chem Kinet 1999;31:183–220.
blending on bioethanol HCCI combustion using forced induction and residual [61] Curran H, Serinyel Z, Metalcafe W. Combustion Chemistry Center at NUI Gal-
gas trapping. Fuel 2008;87:732–9. way, Ireland; 2009. Personal communication with M. Sjöberg, J.E. Dec of the
[32] Tsolakis A, Megaritis A, Yap D. Application of exhaust gas fuel reforming in Sandia National Labs.
diesel and homogeneous charge compression ignition (HCCI) engines fuelled [62] Mosbach S, Kraft M, Bhave A, Mauss F, Mack JH, Dibble RW. Simulating a
with biofuels. Energy 2008;33:462–70. homogeneous charge compression ignition engine fuelled with a dee/etoh
[33] Manente V, Tunestal P, Johansson B. Influence of the wall temperature and blend, SAE paper no 2006-01-1362; 2006.
combustion chamber geometry on the performance and emissions of a mini [63] Dagaut P, Togbé C. Experimental and modeling study of the kinetics
HCCI engine fuelled with DEE, SAE paper no. 2008-01-0008; 2008. of oxidation of ethanol–n-heptane mixtures in a jet-stirred reactor. Fuel
[34] Shudo T, Ono Y. HCCI Combustion of hydrogen, carbon monoxide and 2010;89:280–6.
dimethyl ether, SAE paper no. 2002-01-0112; 2002. [64] Dagaut P, Togbé C. Experimental and modeling study of the kinetics of oxida-
[35] Chen Z, Konno M, Oguma M, Yanai T. Experimental study of CI natural- tion of ethanol–gasoline surrogate mixtures (E85 surrogate) in a jet-stirred
gas/DME homogeneous charge engine, SAE paper no. 2000-01-0329; 2000. reactor. Energy Fuels 2008;22:3499–505.
 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610 1609

[65] Gasnot L, Decottignies V, Pauwels JF. Kinetics modelling of ethyl acetate oxi- [98] Brakora JL, Reitz RD. Investigation of NOx predictions from biodiesel-fueled
dation in flame conditions. Fuel 2005;84:505–18. HCCI engine simulations using a reduced kinetic mechanism, SAE paper no.
[66] Mueller MA, Kim TJ, Yetter RA, Dryer FL. Flow reactor studies and kinetic 2010-01-0577; 2010.
modeling of the H2 /O2 reaction. Int J Chem Kinet 1999;31:113–25. [99] Curran HJ, Gaffuri P, Pitz PJ, Westbrook CK. A comprehensive modeling study
[67] Marinov NM, Westbrook CK, Pitz WJ. Detailed and global chemical kinetics of n-heptane oxidation. Combust Flame 1998;114:149–77.
model for hydrogen. In: Chan SH, editor. Trasnport Phenomena in Combus- [100] Saisirirat P, Foucher F, Chanchaona S, Mounaïm-Rousselle C. Spectroscopic
tion. Washington, DC: Taylor and Francis; 1996. p. 118–41. measurements of low-temperature heat release for homogeneous combus-
[68] O’Connaire M, Curran HJ, Simmie JM, Pitz WJ, Westbrook CK. A comprehensive tion compression ignition (HCCI) n-heptane/alcohol mixture combustion.
modeling study of hydrogen oxidation. Int J Chem Kinet 2004;36:603–22. Energy Fuels 2010;24:5404–9.
[69] Westbrook CK, Dryer FL. Chemical kinetic modeling of hydrocarbon combus- [101] Mack JH, Flowers DL, Buchholz BA, Dibble RW. Investigation of HCCI com-
tion. Prog Energy Combust Sci 1984;10:1–57. bustion of diethyl ether and ethanol mixtures using carbon 14 tracing and
[70] Kee RJ, Rupley FM, Miller JA. Chemkin-II: A FORTRAN chemical kinetics pack- numerical simulations. Proc Combust Inst 2005;30:2693–700.
age for the analysis of gas-phase chemical kinetics, Sandia National Labs [102] Iizuka Y, Surianarayanan M. Comprehensive kinetic model for adiabatic
Report SAND89-8009B; 1991. decomposition of di-tert-butyl peroxide using BatchCAD. Ind Eng Chem Res
[71] Li H, Karim GA. Knock in spark ignition hydrogen engines. Int J Hydrogen 2003;42:2987–95.
Energy 2004;29:859–65. [103] Yao M, Zheng Z, Liang X. Numerical study on the chemical reaction kinetics of
[72] Liu KC. An experimental and analytical investigation into the combustion DME/methanol for HCCI combustion process, SAE paper no. 2006-01-1521;
characteristics of HCCI and dual fuel engines with pilot injection, PhD thesis 2006.
in Mechanical Engineering, University of Calgary, Canada; 2006. [104] Yao M, Qin J, Zheng Z. Numerical study of the combustion mechanism of
[73] Li J, Zhao Z, Kasakov A, Chaos M, Dryer FL, Scire JJ. A comprehensive a homogeneous charge compression ignition engine fuelled with dimethyl
kinetic mechanism for CO, CH2 O, and CH3 OH combustion. Int J Chem Kinet ether and methane, with a detailed kinetics model. Part 2: the reaction kinet-
2007;39:109–36. ics of dimethyl ether and methane dual-fuel. Proc Inst Mech Eng Part D: J
[74] Held TJ, Dryer FL. A comprehensive mechanism for methanol oxidation. Int J Automob Eng 2005;219:1225–36.
Chem Kinet 1998;30:805–30. [105] Ogawa H, Miyamoto N, Kaneko N, Ando H. Combustion control and operating
[75] Westbrook CK, Dryer FL. A comprehensive mechanism for methanol oxida- range expansion with direct injection of reaction suppressors in a premixed
tion. Combust Sci Technol 1979;20:125–40. DME HCCI engine, SAE paper no. 2003-01-0746; 2003.
[76] Westbrook CK, Dryer FL. Prediction of laminar flame properties of methanol- [106] Shudo T, Yamada H. Hydrogen as an ignition-controlling agent for HCCI com-
air mixtures. Combust Flame 1980;37:171–92. bustion engine by suppressing the low-temperature oxidation. Int J Hydrogen
[77] Lindstedt RP, Meyer MP. A dimensionally reduced reaction mechanism for Energy 2007;32:3066–72.
methanol oxidation. Proc Combust Inst 2002;29:1395–402. [107] Shudo T, Shima Y, Fujii T. Production of dimethyl ether and hydro-
[78] Brakora JL, Ra Y, Reitz RD, McFarlane J, Daw SC. Development and validation of gen by methanol reforming for an HCCI engine system with waste heat
a reduced reaction mechanism for biodiesel-fueled engine simulations. SAE recovery–continuous control of fuel ignitability and utilization of exhaust gas
Int J Fuels Lubr 2008;1:675–702 (SAE paper no. 2008-01-1378). heat. Int J Hydrogen Energy 2009;34:7638–47.
[79] Dooley S, Curran HJ, Simmie JM. Autoignition measurements and a validated [108] Yamada H, Yoshii M, Tezaki A. Chemical mechanistic analysis of additive
kinetic model for the biodiesel surrogate, methyl butanoate. Combust Flame effects in homogeneous charge compression ignition of dimethyl ether. Proc
2008;153:2–32. Combust Inst 2005;30:2773–80.
[80] Gaïl S, Thomson MJ, Sarathy SM, Syed SA, Dagaut P, Diévart P, et al. A [109] Heirmel JM, Coffee TP. The unimolecular ozone decomposition reaction. Com-
wide-ranging kinetic modeling study of methyl butanoate combustion. Proc bust Flame 1979;35:117–23.
Combust Inst 2007;31:305–11. [110] Komninos NP. Assessing the effect of mass transfer on the formation of HC
[81] Herbinet O, Pitz WJ, Westbrook CK. Detailed chemical kinetic oxidation mech- and CO emissions in HCCI engines, using a multi-zone model. Energy Convers
anism for a biodiesel surrogate. Combust Flame 2008;154:507–28. Manage 2009;50:1192–201.
[82] Bowman CT. Control of combustion-generated nitrogen oxide emissions. In: [111] Noda T, Foster DE, A numerical study to control combustion duration of
Proc. of the 24th Symposium (International) on Combustion, The Combustion hydrogen-fueled HCCI by using multi-zone chemical kinetics simulation, SAE
Institute. 1992. p. 859–78. paper no. 2001-01-0250; 2001.
[83] Lavoie GA, Heywood JB, Keck JC. Experimental and theoretical investigation of [112] Komninos NP, Hountalas DT, Rakopoulos CD. A parametric investigation of
nitric oxide formation in internal combustion engines. Combust Sci Technol hydrogen HCCI combustion using a multi-zone model approach. Energy Con-
1970;1:313–26. vers Manage 2007;48:2934–41.
[84] Smith GP, Golden DM, Frenklach M, Moriarty NW, Eiteneer B, Goldenberg M, [113] Komninos NP, Rakopoulos CD. Comparison of a detailed and a reduced ethanol
et al. GRI-Mech v.3.0; 2011. http://www.me.berkeley.edu/gri mech/. oxidation mechanism in HCCI combustion using a multi-zone model. Open
[85] Marinov NM, Pitz WJ, Westbrook CK, Hori M, Matsunaga N. An exper- Renew Energy J 2011;4:47–59.
imental and kinetic calculation of the promotion effect of hydrocarbons [114] Annand WJD. Heat transfer in the cylinders of reciprocating internal combus-
on the NO–NO2 conversion in a flow reactor. Proc Combust Inst 1998;27: tion engines. Proc Inst Mech Eng 1963;177:973–90.
389–96. [115] Komninos NP. Modeling HCCI combustion: modification of a multi-zone
[86] Isooctane oxidation mechanism; 2003, http://www.tfd.chalmers. model and comparison to experimental results at varying boost pressure.
se/∼valeri/MECH.html. Appl Energy 2009;86:2141–51.
[87] Yoshikawa T, Reitz RD. Development of an improved NOx reaction mechanism [116] Schofield K. The enigmatic mechanism of the flame ionization detector: Its
for low temperature diesel combustion modeling, SAE paper no. 2008-01- overlooked implications for fossil fuel combustion modeling. Prog Energy
2413; 2008. Combust Sci 2008;34:330–50.
[88] Fiveland SB, Assanis DN. A four-stroke homogeneous charge compression [117] Sjoberg M, Dec JE. Ethanol autoignition characteristics and HCCI performance
ignition engine simulation for combustion and performance studies, SAE for wide ranges of engine speed, load and boost, SAE paper no. 2010-01-0338;
paper no. 2000-01-0332; 2000. 2010.
[89] Goldsborough SS, Van Blarigan P. A numerical study of a free piston IC engine [118] Sjoberg M, Dec JE. Effects of EGR and its constituents on HCCI autoignition of
operating on homogeneous charge compression ignition combustion, SAE ethanol. Proc Combust Inst 2011;33:3031–8.
paper no. 1999-01-0619; 1999. [119] Mehl M, Pitz WJ, Sjöberg M, Dec JE. Detailed kinetic modeling of low-
[90] Lund CM. HCT: a general computer program for calculating time-dependent temperature heat release for PRF fuels in an HCCI engine, SAE paper no.
phenomena involving one-dimensional hydrodynamics, transport and 2009-01-1806; 2009.
detailed chemical kinetics, Lawrence Livermore National Laboratory Report [120] Kim HJ, Park SH, Lee KS, Lee CS. A study of spray strategies on improvement of
UCRL-52504, rev; 1995. engine performance and emissions reduction characteristics in a DME fueled
[91] Woschni G. A universally applicable equation for the instantaneous heat diesel engine. Energy 2011;36:1802–13.
transfer coefficient in the internal combustion engine, SAE paper no. 670931; [121] Amsden AA. KIVA-3V release 2. Improvement to KIVA-3V. Los Alamos
1967. National Laboratory, LA-UR-99-915; 1999.
[92] Ng CKW, Thomson MJ. A computational study of the effect of fuel reforming, [122] Teng H, McCandless JC, Schneyer JB, Thermochemical characteristics of
EGR and initial temperature on lean ethanol HCCI combustion, SAE paper no. dimethyl ether—an alternative fuel for compression-ignition engines, SAE
2004-01-0556; 2004. paper no. 2001-01-0154; 2001.
[93] Martinez-Frias J, Aceves SM, Flowers DL. Improving ethanol life cycle energy [123] Teng H, McCandless JC, Schneyer JB. Viscosity and lubricity of (liquid)
efficiency by direct utilization of wet ethanol in HCCI engines. Trans ASME J dimethyl ether e an alternative fuel for compression-ignition engines, SAE
Energy Resour Technol 2007;129:332–7. paper no. 2002-01-0862; 2002.
[94] http://www.reactiondesign.com; 2011. [124] Teng H, McCandless JC, Schneyer JB. Compression ignition delay (phys-
[95] Szybist JP. Fuel-specific effect of exhaust gas residuals on HCCI combustion: ical + chemical) of dimethyl ether e an alternative fuel for compression-
a modeling study, SAE paper no. 2008-01-2402; 2008. ignition engines, SAE paper no. 2003-01-0759; 2003.
[96] Yang D, Wang Z, Wang J-X, Shuai S. Experiment and chemical kinetics analysis [125] Teng H, McCandless JC, Schneyer JB. Thermodynamic properties of dimethyl
of active atmosphere with different fuels in HCCI combustion. Energy Fuels ether—an alternative fuel for compression-ignition engines, SAE paper no.
2010;24:4872–8. 2004-01-0093; 2004.
[97] Hoffman SR, Abraham J. A comparative study of n-heptane, methyl decanoate, [126] Teng H, McCandless JC. Comparative study of characteristics of diesel-fuel
and dimethyl ether combustion characteristics under homogeneous-charge and dimethyl-ether sprays in the engine, SAE paper no. 2005-01-1723;
compression–ignition engine conditions. Fuel 2009;88:1099–108. 2005.
1610 N.P. Komninos, C.D. Rakopoulos / Renewable and Sustainable Energy Reviews 16 (2012) 1588–1610

[127] Kong SC, Sun Y, Reitz RD. Modeling diesel spray flame lift-off, sooting ten- [141] Han Z, Uludogan A, Hampson GJ, Reitz RD. Mechanism of soot and NOx emis-
dency and NOx emissions using detailed chemistry with phenomenological sion reduction using multiple-injection in a diesel engine, SAE paper no.
soot model. Trans ASME J Eng Gas Turb Power 2007;129:252–60. 960633; 1996.
[128] Beale JC, Reitz RD. Modeling spray atomization Kelvin–Helmholtz/ [142] Kong S-C. A study of natural gas/DME combustion in HCCI engines using CFD
Rayleigh–Taylor hybrid model. Atomiz Sprays 1999;9:623–50. with detailed chemical kinetics. Fuel 2007;86:1483–9.
[129] Han Z, Reitz RD. Turbulence modeling of internal combustion engines using [143] Yu RX, Bai XS, Vressner A, Hultqvist A, Johansson B, Olofsson J, et al.
RNG ke3 models. Combust Sci Technol 1995;106:267–95. Effect of turbulence on HCCI combustion, SAE paper no. 2007-01-0183;
[130] O’Rourke PJ, Amsden AA. A spray/wall interaction submodel for the KIVA-3 2007.
wall film model, SAE paper no. 2000-01-0271; 2000. [144] Joelsson T, Yu R, Bai XS, Vressner A, Johansson B. Large eddy simulation and
[131] Park SH, Kim HJ, Lee CS. Macroscopic spray characteristics and breakup per- experiments of the auto-ignition process of lean ethanol/air mixture in HCCI
formance of dimethyl ether (DME) fuel at high fuel temperatures and ambient engines, SAE paper no. 2008-01-1668; 2008.
conditions. Fuel 2010;89:3001–11. [145] Viggiano A, Magi V. Multidimensional simulation of ethanol HCCI engines,
[132] Kim HJ, Suh HK, Lee CS. Numerical and experimental study on the compari- SAE paper no. 2009-24-0031; 2009.
son between diesel and dimethyl ether (DME) spray behaviors according to [146] Magi V. REC-2000: A multidimensional code for engines, Engine research
combustion chamber shape. Energy Fuels 2008;22:2851–60. laboratory report, School of Mechanical Engineering, Purdue University;
[133] Chen Z, Yao M, Zheng Z, Zhang Q. Experimental and numerical study of 2000.
methanol/dimethyl ether dual-fuel compound combustion. Energy Fuels [147] Launder BE, Spalding DB. The numerical computations of turbulent flows.
2009;23:2719–30. Comput Method Appl Mech 1974;3:269–89.
[134] AVL FIRE UsersGuide, version 8.4, AVL LIST GmbH; 2005. [148] Kong S-C, Reitz RD. Numerical study of premixed HCCI engine combustion
[135] AVL FIRE Spray, version 8.4, AVL LIST GmbH; 2005. and its sensitivity to computational mesh and model uncertainties. Combust
[136] Han ZW, Reitz RD. A temperature wall function formulation for variable- Theor Model 2002;7:417–33.
density turbulent flows with application to engine convective heat transfer [149] Liu C, Karim GA. A simulation of the combustion of hydrogen in HCCI engines
modeling. Int J Heat Mass Transfer 1997;40:613–25. using a 3D model with detailed chemical kinetics. Int J Hydrogen Energy
[137] Yao M, Huang C, Zheng Z. Multidimensional numerical simulation on 2008;33:3863–75.
dimethyl ether/methanol dual-fuel homogeneous charge compression igni- [150] Contino F, Jeanmart H. Study of the HCCI running zone using ethyl acetate,
tion (HCCI) engine combustion and emission processes. Energy Fuels SAE paper no. 2009-01-0297; 2009.
2007;21:812–21. [151] Miller JA, Melius CF. Kinetic and thermodynamic issues in the forma-
[138] Huang C, Yao M, Lu X, Huang Z. Study of dimethyl ether homogeneous charge tion of aromatic compounds in flames of aliphatic fuels. Combust Flame
compression ignition combustion process using a multi-dimensional compu- 1992;91:21–39.
tational fluid dynamics model. Int J Therm Sci 2009;48:1814–22. [152] Um S, Park SW. Numerical study on combustion and emission characteristics
[139] Park SW. Numerical study on optimal operating conditions of homogeneous of homogeneous charge compression ignition engines fueled with biodiesel.
charge compression ignition engines fueled with dimethyl ether and n- Energy Fuels 2010;24:916–27.
heptane. Energy Fuels 2009;23:3909–18. [153] Um S, Park SW. Modeling effect of the biodiesel mixing ratio on combus-
[140] Kong S-C, Sun Y, Reitz RD. Modeling diesel spray flame liftoff, sooting ten- tion and emission characteristics using a reduced mechanism of methyl
dency, and NOx emissions using detailed chemistry with phenomenological butanoate. Fuel 2010;89:1415–21.
soot model. Trans ASME J Eng Gas Turb Power 2007;129:245–51.