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

net/publication/273695527

Developments in internal combustion engines and implications for

combustion science and future transport fuels

Article in Proceedings of the Combustion Institute · January 2015

CITATIONS READS

104 5,142

1 author:

Gautam T. Kalghatgi
Retired
151 PUBLICATIONS 9,958 CITATIONS

SEE PROFILE

All content following this page was uploaded by Gautam T. Kalghatgi on 09 May 2018.

The user has requested enhancement of the downloaded file.

 Available online at www.sciencedirect.com
Proceedings
ScienceDirect of the
Combustion
Institute
Proceedings of the Combustion Institute xxx (2014) xxx–xxx
www.elsevier.com/locate/proci

Developments in internal combustion engines

and implications for combustion science
and future transport fuels
G.T. Kalghatgi ⇑
Saudi Aramco, PO Box 62, Dhahran 31311, Saudi Arabia

Abstract

Changes in engine technology, driven by the need to increase the efficiency of the SI engine and reduce
NOx and soot from diesel engines, and in transport energy demand will have a profound effect on the prop-
erties, specifications and production of future fuels. The expected increase in global demand for transport
energy is significantly skewed towards heavier fuels like diesel and jet fuel while the demand for gasoline
might decrease. Abnormal combustion such as knock and preignition will become more likely as spark-
ignition (SI) engines develop to become more efficient and fuel antiknock quality will become more impor-
tant. In current and future SI engines, for a given RON (Research Octane Number), a fuel of lower MON
(Motor Octane Number) has better antiknock quality. Current fuel specifications in several parts of the
world assume that MON contributes to antiknock quality and will need to be revised as the mismatch with
engine requirements widens. The primary challenge for diesel engines is to reduce emissions of soot and
NOx while maintaining high efficiency. This becomes much easier if such engines are run on fuels of extre-
mely low cetane. In the long term compression ignition engines are likely to use fuels with RON in the
range of 70–85 (Cetane number <30) but with no strict requirements on volatility. Such fuels would
require less processing in the refinery than today’s fuels and also help mitigate the expected demand imbal-
ance in favour of heavier fuels. Such an engine/fuels system will be at least as efficient as today’s diesel
engine but could be significantly cheaper. The review concludes with a list of issues for combustion science
that are relevant to this fuel and engine development.
Ó 2014 Published by Elsevier Inc. on behalf of The Combustion Institute.

Keywords: IC engines; Gasoline; Diesel; Autoignition; Preignition

1. Introduction importance to modern industrial society. The glo-

Transport accounts for around 20% of the total
energy consumed globally [1] and is of central
⇑ Tel.: +966 13876 8171.
E-mail address: Gautam.Kalghatgi@aramco.com
bal demand for transport energy is expected to
increase rapidly, by around 40% by 2040 com-
pared to now, mostly in the non-OECD countries
[2–5]. Transport is expected to be powered
overwhelmingly (around 90% share) by internal
combustion (IC) engines using liquid fuels made
from petroleum (crude oil) in the foreseeable

http://dx.doi.org/10.1016/j.proci.2014.08.031
1540-7489/Ó 2014 Published by Elsevier Inc. on behalf of The Combustion Institute.

Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
2 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx
future, just as it is today, in spite of the expected
advances in alternatives [2–5].
As in other energy sectors, the supply and
demand and the sources for transport energy as
well as changes in engine design, and hence future
fuels, are shaped by political and economic factors
such as energy security and by local environmen-
tal concerns and global issues such as climate
change. In addition to legislative demands, future
IC engines also have to respond to consumer
demands for performance, drivability and afford-
ability in an increasingly competitive world.
Engine efficiency, performance and emissions will
be squeezed to the utmost in the future and even
small improvements, perhaps previously not con-
sidered worth pursuing, will become important.
In such a context, traditional combustion prob-
lems such as ignition and flame development will
acquire renewed importance. As engine designs
change problems such as preignition leading to
extremely heavy knock, “superknock”, in SI
engines are threatening to limit efficiency improve-
ments. Other traditionally neglected fuel and com-
bustion related issues such as deposit formation
will also be important. However, the autoignition
behavior of practical fuels provides the major uni-
fying thread to help understand how future fuels
might be affected. This review discusses these
issues – a more detailed discussion can be found
in [2]. Engine technologies such as injector design
and control strategies can affect mixing and hence
combustion in diesel engines and are discussed in
that context. Other engine and vehicle technolo-
gies such as friction reduction and transmission
also affect fuel efficiency, performance and emis-
sions but detailed discussion of such topics is
beyond the scope of this review.
1.1. Internal combustion (IC) engines
IC engines are either spark ignition (SI)
engines or compression ignition (CI) engines
[6,7]. Practical CI engines currently are diesel
engines running on diesel fuels. SI engines use gas-
oline which is premixed with air and the mixture is
compressed – heat release occurs in an expanding
turbulent flame initiated by a spark near the top
of the compression stroke, top dead center
(TDC). In a diesel engine, fuel and air are not fully
premixed and combustion is initiated by autoigni-
tion as the fuel vaporizes, mixes and reacts with
oxygen in the engine cylinder. Currently, the pas-
senger car sector is dominated by SI engines
whereas the commercial sector (heavy duty road,
air, rail and marine) is dominated by diesel
engines using diesel fuel.
Diesel engines are more efficient than SI
engines [6,7] but suffer from high engine-out emis-
sions of particulates and NOx (nitrogen oxides)
which are very difficult to control by after-treat-
ment of the oxygen-rich exhaust through the use
Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
of catalysts. In contrast, SI engines are run at a
stoichiometric mixture strength which enables a
three-way catalyst to be used to treat the exhaust
and reduce tail-pipe emissions of unburned hydro-
carbons (HC), carbon monoxide (CO) and NOx
to extremely low levels. Diesel engines are much
more expensive than SI engines primarily because
of the need to meet increasingly stringent regula-
tions on soot and NOx emissions while using die-
sel fuel [2] because this requires very high pressure
injection systems and complex after treatment.
Recently there has been significant interest in
Homogeneous Charge Compression Ignition
(HCCI) engines because of the potential of high
efficiency and low NOx and low soot operation.
In HCCI engines, as in SI engines, fuel and air
are fully premixed and the mixture is compressed
but heat release takes place by autoignition of
the charge. Unlike in SI engines (spark timing)
or in diesel engines (fuel injection timing), there
is no in-cycle control on combustion phasing. This
makes the control of HCCI engines extremely dif-
ficult. Equally importantly, as the load is
increased for a given speed, the equivalence ratio
increases and the maximum heat release rate, the
pressure rise rate and NOx levels increase. This
puts a severe restriction on the maximum load
achievable in HCCI engines and it is very unlikely
that there will be practical engines working in
HCCI mode over the full operating range. How-
ever there is great potential for developing practi-
cal fuel/engine systems with “premixed enough”
compression ignition to get most of the benefits
of HCCI combustion – see Section 6.
The primary focus for the development of SI
engines is to improve their efficiency while that
for diesel engines is to meet increasingly stringent
soot and NOx emissions requirements without
compromising efficiency. All this has to be
achieved at an acceptable cost and meet other cus-
tomer demands. Future fuel manufacture and
properties will need to evolve to enable these goals
to be met. This evolution will also be influenced
and constrained by factors affecting the supply
and demand of transport energy.
2. Current transport fuels
Liquid fuels have become the fuels of choice
for transport over the past century or so because
of their high energy density and the ease of trans-
port, storage and handling. As a result an exten-
sive global distribution network has been built
up which is very difficult to replace or duplicate.
Transport fuels are primarily made in refineries
starting with petroleum and then blending the
products of several refinery processes [2,8]. Petro-
leum itself contains hundreds of different hydro-
carbons, each with its own boiling point.
Initially it is separated into different boiling range
 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx
fractions by distillation. Up to 2% of the crude
can be dissolved gases which are released when
the temperature is raised above ambient tempera-
ture – these constitute Liquid Petroleum Gas,
LPG. The main component of LPG, around
75%, is propane. The light fraction resulting from
distillation and boiling roughly in the gasoline
range from 20 °C to 200 °C is termed ‘straight
run gasoline’ (SRG). The generic term ‘naphtha’
is used for products, which can come from differ-
ent processes in the refinery, in this boiling range.
Naphtha is usually further processed to improve
its autoignition resistance before it is used as a
gasoline component. The middle distillates span
the kerosene/diesel boiling range of 160 °C to
380 °C. Between 40% and 60% of the weight of
petroleum can be “Residue” with boiling points
above 380 °C, depending on the crude.
These heavier fractions have to be converted
through further processing into lighter fractions
that can be used as fuel components. Refinery
processes involve separation into smaller frac-
tions, a change of composition, removal of con-
taminants and a redistribution of carbon/
hydrogen ratios of the molecules through carbon
rejection or hydrogen addition. A description of
refinery and other processes used for making fuel
components can be found in [2,8,9].
In addition, other processes outside a conven-
tional refinery produce fuel components which
are often used in liquid fuels. These include Biofu-
els made from sugar, starch, vegetable oils, cellu-
losic residue and biological waste; other
oxygenates such as MTBE and Gas to Liquid
(GTL) fuels made using the Fischer-Tropsch pro-
cess for synthesizing syngas, a mixture of carbon
monoxide and hydrogen, into liquid fuels.
Gasoline and diesel are complex mixtures of
hydrocarbons. They differ from each other pri-
marily in terms of their autoignition quality which
is measured by the Research and Motor Octane
Numbers, RON and MON, for gasolines and by
the Cetane Number, CN, for diesels [10]. The
higher the RON, the more resistant is the gasoline
to autoignition while higher CN indicates that the
fuel is more prone to autoignition – there is an
inverse correlation between RON and CN [10].
Gasolines need to be resistant to autoignition to
avoid knock, an abnormal combustion phenome-
non which limits SI engine efficiency whereas con-
ventional diesel fuels auto-ignite easily. Gasolines
are made up of comparatively lighter hydrocar-
bons with carbon numbers usually between 4
and 11 while diesels contain heavier, higher boil-
ing-point compounds typically with carbon num-
bers between 10 and 21.
RON and MON are measured in tests run
according to the procedures set in [11] and are
of fundamental importance to the petroleum
industry since fuels manufacture is primarily dri-
ven by the need to meet the required standards
Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
3
for gasoline octane quality. The RON test is run
in a single-cylinder engine at an engine speed of
600 rpm (revolutions per minute) and an intake
temperature of 52 °C while the MON test is run
at 900 rpm and with a higher intake temperature
of 149 °C. The octane scale is based on two paraf-
fins (alkanes), n-heptane and iso-octane. Blends of
these two primary components are referred to as
primary reference fuels (PRF) and define the
intermediate points in the RON or MON scale.
The RON or MON is the volume percent of iso-
octane in the PRF. A fuel is assigned the RON
(or MON) value of the PRF that matches its
knock behavior in the RON (or MON) test.
Cetane number (CN) is measured for a given
fuel by comparing its ignition characteristics with
reference fuels in a single-cylinder diesel engine
[11]. The reference fuels are made by blending
normal cetane, n-hexadecane, which is defined to
have a CN of 100 and heptamethylnonane, a
highly branched paraffin which is assigned a CN
of 15. More recently, a laboratory test method
based on measuring ignition delay has been devel-
oped to measure the autoignition quality of diesel
fuels [11]. In this test, a small specimen of test fuel
is injected into a heated, temperature-controlled
constant volume chamber which has been previ-
ously filled with compressed air. A Derived
Cetane Number, DCN, is calculated from the
measured ignition delay.
Fuels can be said to be in the diesel autoigni-
tion range if CN > 30 and fuels with
RON > 60 (CN < 30) can be considered to
be in the gasoline autoignition range [10]. Practi-
cal diesel fuels have CN ranging from 40 to 60
while practical gasolines in most markets have
RON between 90 and 100.
Jet fuel or kerosene is the third most important
transport fuel currently accounting for around
12% of total transport energy demand. It is in
the diesel boiling range but with lower final boil-
ing point, lower cetane and lower density com-
pared to typical European diesel fuel.
3. Outlook for supply and demand for transport
fuels
The world demand for transport fuels is very
large – around 1.6 trillion liters per year each for
gasoline and diesel – and accounts for around
60% of the total world oil consumption and is
expected to grow rapidly [3–8]. Currently, around
95% of all transport energy comes from petroleum
[4].
Global oil supply capacity is growing at an
unprecedented level and there is an abundance
of oil to meet the demand in the foreseeable future
[13,14]. New discoveries and improved recovery
rates coupled with increasing availability of non-
conventional petroleum (e.g., oil sands) have
4 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx
ensured that reserves have been increasing and
more than kept pace with increasing demand over
the past few decades [2,3]. The reserves to produc-
tion ratio has increased from around 29 years in
1980 to 54 years in 2012 [15].
It will be difficult for alternative transport
energy sources, even if they grow rapidly, to dis-
place petroleum-based liquid fuels to any substan-
tial extent in the next few decades. Currently, of
the total global transport energy, biofuels supply
about 4% and the rest, which include CNG (com-
pressed natural gas), LPG (liquid petroleum gas)
and LNG (liquefied natural gas) about 1%. The
contribution of electricity is negligible and of
hydrogen, non-existent.
The recent shale gas revolution has more than
doubled the estimate of global natural gas reserves
[16] and it is likely that the availability of cheap
natural gas will bring more of it into the transport
sector but its use will be constrained by infrastruc-
ture limitations. Natural gas can be used in trans-
port as CNG or LNG but this will require
investments in new infrastructure. It could also
be used to make GTL (gas-to-liquids) and this is
a very expensive route. A recent projection sug-
gests that by 2040, natural gas could supply
around 5% of global transport energy [17] but
there are likely to be large regional variations.
Of the various options for electrification of the
transport sector, hybrid electric vehicle (HEV)
technology is expected to be very widely adopted
where possible – e.g., in light duty passenger vehi-
cles working in urban cycles with many stop/starts
– and substantially moderate the increase in fuel
demand from this sector [2,12]. Even by 2040,
around 90% of transport energy is still expected
to come from petroleum [4,5].
The increase in demand for transport energy is
expected to be very substantially skewed towards
the commercial sector [2,4,5,12,17] under the cur-
rent technology trajectory. Even though the num-
ber of passenger vehicles is expected to double to
around 1.6 billion by 2040 [4], the average future
car will be much lighter, smaller and travel less
distance compared to today’s average car and
the scope for efficiency improvements e.g.,
through hybridisation, is much greater in this sec-
tor. As a result there will be a substantial shift in
demand towards heavier fuels like diesel and jet
fuel compared to gasoline as illustrated in Fig. 1
which is a projection from the Freeway scenario,
which assumes that market forces dominate, of
the World Energy Council. The line in the middle
of the figure shows the ratio of (diesel + jet fuel)
to gasoline. This ratio is around 1.5 now and is
expected to increase to around 2.4 by 2040 [2].
Whatever the actual numbers, the trend is clear
and this will require global investments running
to hundreds of billions of dollars in refineries to
make more diesel and jet fuel compared to gaso-
line in the future. It will also lead to a surplus of
Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
160 7
(Diesel+Jet)/Gasoline
140 6
Energy demand EJ
120 5
100
4
80 3.8
3
60 2.4
40 2
1.5
20 1
0 0
2005 2010 2015 2020 2025 2030 2035 2040 2045 2050
Year
Gasoline Jeuels Diesel Rao (D+J)/G
Fig. 1. Projections for gasoline, jet fuel and diesel
growth in the Freeway scenario of the World Energy
Council [5].
low-octane gasoline components such as naphtha.
Alternatively, a significant proportion of CI
engines, especially heavy duty, will have to be
developed to use alternatives like natural gas
and light fuels rather than conventional diesel fuel
in order to keep supply and demand in balance.
Further discussion of these topics can be found
in Ch.1 of [2].
4. Fuel effects on autoignition in engines
As combustion proceeds in SI engines, the
pressure and temperature in the unburned “end-
gas”, ahead of the advancing flame front, increase
and under some circumstances could cause the
endgas to auto-ignite before the flame consumes
it. In SI engines, though the fuel and air are fully
premixed, the mixture is never homogeneous
because of temperature gradients caused by turbu-
lence. Autoignition is initiated around one or
more ignition centers or “hot spots” and could
cause a very rapid release of much of the chemical
energy contained in the endgas and this could lead
to a sharp pressure rise and pressure waves inside
the cylinder. Knock can cause engine damage and
the strategies used to manage knock reduce the
efficiency and the maximum power of SI engines.
In modern cars equipped with knock-sensors, cor-
rective action to protect the engine from knock
damage leads to a reduction in engine power
and acceleration performance [e.g., [10,18,19]].
Autoignition is governed by chemical kinetics
[20] and depends on the chemical composition of
the fuel and on the evolution with time of pressure,
temperature and equivalence ratio. In SI and HCCI
engines, pressure and temperature increase with
time while the global equivalence ratio of the mix-
ture is constant. In contrast, in diesel (conventional
CI) engines, equivalence ratio changes with
time and space as the fuel mixes with air while
pressure and global temperature remain rela-
tively unchanged in the lead-up to autoignition.
 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx 5

Currently chemical kinetics of a practical fuel needs 1000

to be simulated by a “surrogate fuel,” a fuel com- 950

posed of a small number of pure compounds whose K=1

Compression Temperature, T, K
900
behavior matches that of the target fuel [21,22].
Further, such surrogate kinetic models need to be 850

small enough to be coupled to models which predict 800 K=0

turbulent flow and mixture preparation in CI 750
engines and turbulent flame development for SI K = - 0.33
700
engines. Such models are very useful to understand MON Test, K= +1.0
autoignition of a well-characterized fuel in engines. 650 RON test, K= 0

However, for the foreseeable future, autoigntion 600

Knock, K=-0.33 SAE2001-01-3584

quality of fuels has to be understood in terms of 1 1.5 2 2.5

empirical measures such as RON and MON, given Pressure, P, MPa

their importance to the refining industry, regardless

of any advances in the fundamental understanding
of autoignition and the accuracy and the applica-
bility of the models.
Fig. 2. Variation of the temperature of the unburned
mixture with pressure. It depends on engine design and
operating conditions.

4.1. The Octane Index – autoignition quality of

practical gasolines
However, neither RON nor MON describes
the autoignition quality of practical gasolines ade-
quately. The true resistance to autoignition of
practical fuels is given by the Octane Index, OI,
defined as
OI ¼ RON  KS ð1Þ
Here S is the sensitivity of the fuel, (RON –
MON) and is typically around 10 for practical
gasolines whereas it is zero, by definition, for
PRF. OI is the octane number of the PRF that
matches it for autoignition behavior in the partic-
ular test under consideration. K is a constant
depending on the pressure and temperature evolu-
tion in the unburned gas – it is not a primary
property of the fuel. Practical gasolines contain
other classes of compounds such as aromatics,
olefins and oxygenates apart from paraffins which
make up PRF on which the octane scale is based.
A gasoline will match different PRFs, i.e. have dif-
ferent OI, at different operating conditions (differ-
ent K values). The temperature of the unburned
gas in the MON test conditions is higher than in
the RON test condition for the same pressure as
illustrated in Fig. 2. In [10], Tcomp15 was defined
as the compression temperature of the unburned
gas when the pressure is 15 bar (1.5 MPa) and
helps to place the operating condition of interest
in relation to the RON test condition. If Tcomp15
is less than that for the RON test, the test condi-
tion is said to be “beyond RON”, K will be nega-
tive and a non-PRF fuel (S > 0) will have
OI > RON. An example of such a condition
“beyond RON” is also shown in Fig. 2. Studies
in HCCI engines where autoignition can be
obtained at conditions not usually accessible in
knocking SI engines have helped clarify the
autoignition of practical fuels [23–25].
The K value used in Eq. (1) is determined
empirically by measuring a parameter, z, which
depends on the autoignition quality of the fuel
in an engine or a vehicle test, using fuels of differ-
ent RON and sensitivity i.e. different chemical
composition. For instance, z can be CA50, the
crank angle at which 50% of the total heat release
from autoignition occurs in HCCI tests. The fuels
are compared at the same engine operating condi-
tions so that the fuel/air mixture is subjected to
the same pressure/temperature history. It is possi-
ble to find a linear regression equation [10] such
that
z ¼ aRON þ bMON þ c ð2Þ
The Octane Index, OI is then defined with RON
and MON and the coefficients from the linear
regression.
a b
OI ¼ RON þ MON ð3Þ
aþb aþb
Eq. (2) can then be rewritten with the Octane
Index.
z ¼ c þ ða þ bÞOI ð4Þ
K is defined as:
b
K¼ ð5Þ
aþb
K is a measure of how different the actual operat-
ing condition is compared to the RON test condi-
tion. Similarly, the sensitivity S can be taken as a
measure of how different the fuel autoignition
chemistry is from that of PRF. Two fuels of the
same RON and MON will have the same OI
regardless of the composition or the value of K
and will have the same autoignition phasing,
within experimental variation, in SI and HCCI
engines at different conditions. A toluene, iso-
octane and n-heptane mixture can be found to
match any RON and MON combination [26]
and will be a good surrogate fuel for gasoline.
The empirical approach based on the OI and
K value provides a suitable framework for

Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
6 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx
understanding the experimentally observed
autoignition behavior of fuels in the gasoline
autoignition range (RON > 60, CN < 30) over a
wide range of operating conditions both in HCCI
engines and knocking SI engines.
4.2. Ignition delay and the initiation of autoignition
The time taken for autoignition as marked by
heat release at a given temperature and pressure
is known as the induction time or ignition delay,
si, and is commonly measured in shock tubes
and rapid compression machines. The lower the
value of si the more reactive is the mixture. For
a given fuel, the ignition delay changes with tem-
perature and pressure and also depends on equiv-
alence ratio. Chemical kinetic models of
autoignition are usually calibrated using experi-
mental values of ignition delay. In IC engines at
each point during the cycle the mixture will have
a different value of si. It is then hypothesized that
autoignition occurs when the integral of the reci-
procal of si with time, t, (Eq. (6)) reaches unity.
Z te
dt
I¼ ¼1 ð6Þ
0 siðP ;TÞ
This hypothesis suggested by Livengood and Wu
[27] is now commonly used in knock studies as
well as HCCI studies [e.g., [28–30]]. It is very
important to account for the effects of changing
pressure, P. An increase in pressure causes a
decrease in si. In general si can be expressed over
a given temperature range as
 n
si P
¼ f ðT Þ ð7Þ
si0 P0
Here, si0 is the ignition delay measured at a pres-
sure and temperature of P0 and T respectively.
This is illustrated in Figs. 3 and 4. Figure 3 shows
si measured for iso-octane for different pressures
and temperatures; the data are re-plotted from
Isooctane
5
τ, ms.
0.5
1.3 MPa
1.7 MPa
3.4 MPa
4.0 MPa
4.5 MPa
0.05
0.7 0.9 1.1 1.3 1.5
1000/T, T in K
Fig. 3. Ignition delay for iso-octane at stoichiometric
conditions at different pressures and temperatures from
the shock tube studies of Fieweger et al. [31].
Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
Fig. 11 of [31]. The data can be collapsed to a sin-
gle function of temperature if they are scaled to a
single pressure using Eq. (7) as illustrated in Fig. 4
which shows si scaled to 4 MPa by multiplying
each value by a scaling factor (4/P)1.5 – in this
case, n is chosen, by trial and error, to be 1.5.
The functional variation of si with tempera-
ture, T, at a fixed pressure, P, can be represented
by an Arrhenius type expression such as Eq. (8),
in particular temperature ranges.
 
E
si ¼ f ðT Þ ¼ A exp ð8Þ
RT
Here, E is the activation energy, R the universal
gas constant and A is a pre-exponential constant.
A and E are different for different fuels over a
given pressure and temperature range; their values
also depend on the equivalence ratio for a given
fuel. The ignition delay time at high and low tem-
peratures can be expressed as different Arrhenius
expressions such as Eq. (8) but in the temperature
range between 700 K and 900 K, the so called
NTC (Negative Temperature Coefficient) region,
si might actually increase with increasing temper-
ature for some fuels e.g. [32].
The pressure exponent, n, in Eq. (7), could also
be different in different pressure/temperature
ranges [33] for the same mixture. However, in typ-
ical engine tests, only ignition delays smaller than
about 5 ms contribute significantly to the Liven-
good-Wu integral, I. For instance at 1500 rpm,
5 ms equals 45 crank angle degrees (CAD) and a
one CAD step with such an ignition delay in Eq.
(6) will contribute 0.022 to the integral. In this
range, for PRF, n can be expected to be between
1.5 and 1.64, the values measured for isooctane
and n-heptane while for non-PRF fuels it is much
lower [34,35] (see Table 4.1 in [2]). Thus, if
pressure is increased at a given temperature,
starting from the same si, the ignition delay of a
non-PRF fuel is reduced less i.e. it becomes more
100
Isooctane
τ, ms. Normalised to 4 MPa
10
1
1.3 MPa
1.7 MPa
0.1
3.4 MPa
4.0 MPa
4.5 MPa
0.01
0.7 0.9 1.1 1.3 1.5
1000/T, T in K
Fig. 4. Ignition delay in Fig. 3 scaled to 40 bar by
multiplying each value of si by (4/P)1.5, where P is the
actual pressure in MPa at which si was measured.
 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx
resistant to autoignition compared to the initially
matching PRF fuel.
Both pressure and temperature change in
engines and the difference in the value of n helps
explain why a gasoline might behave like a PRF
with an octane number higher than its RON
under some circumstances (i.e. when K is nega-
tive) – see [36] and Ch.4 in [2]. When the condi-
tions are “beyond RON”, the pressure at a given
temperature of the mixture will be higher than
at the RON test condition and the gasoline will
be more resistant to autoignition (si will be higher
because n is lower), the Livengood –Wu integral
will reach unity later compared to the PRF fuel
it matched in the RON condition i.e. OI > RON.
The challenge now is to explain, in chemical
kinetics terms, why n is lower for non-PRF fuels
in the pressure/temperature range relevant to
autoignition in engines. Leppard [32] invoked the
differences in NTC chemistry to explain why
MON is lower than RON for non-PRF fuels. His
contention was that comparing a PRF and a non-
PRF fuel of the same RON, at the MON condition,
as the temperature is higher, the PRF fuel is more
affected by NTC chemistry and its autoignition
chemistry is slowed down relative to the non-PRF
fuel which will actually match a lower octane PRF.
4.3. Pressure wave development following
autoignition
The initiation of autoignition in SI and HCCI
engines can be predicted by assuming the mixture
to be homogeneous and using the Livengood-Wu
integral. In reality there are always temperature
gradients because of turbulence even in nominally
homogeneous mixtures and autoignition starts at
an ignition center or hot spot. The development
of the pressure wave initiated by the autoignition
as it traverses through the gradient of ignition
delay time within the ignition center has been
studied by Bradley and co-workers in [37–41].
These studies were based on the early work of Zel-
dovich [42] and Oppenheim [43]. When fuel and
air are fully premixed, as in a SI or HCCI engine,
the reactivity gradient can be assumed to be
caused by the temperature gradient. The reaction
front will then propagate at a velocity, ua, relative
to the unburned gas given by:
ua ¼ ð@r=@si Þ ¼ ð@si =@T Þ1 ð@T =@rÞ1 ð9Þ
where r is the distance from the center of the hot
spot. If it is assumed that si can be expressed
locally at constant pressure in Arrhenius form –
Eq. (8):Then from Eqs.(8) and (9)
ua ¼ ðT 2 =ðsi E=RÞÞð@T =@rÞ1 ð10Þ
The amplitude of the pressure wave generated by
autoignition depends upon ua [37,41]. The over-
pressure, above the ambient, generated by the
Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
7
propagation of the reaction front through the
hot spot can be related, in the absence of shock
waves, to a non-dimensional “resonance parame-
ter” pnffiffiffiffiffiffiffiffiffiffi
= a/ua where a, is the acoustic speed,
a ¼ cp=q [44]. The analysis in [41,46] identified
the modes of propagation within the hot spot as
follows:
(i) n = 0, thermal explosion.
(ii) 0 < n < nl, supersonic autoignitive deflagra-
tion. The autoignitive wave is ahead of
acoustic wave.
(iii) nl 6 n < nu, developing and developed
detonation
(iv) nl 6 n < aSl1, subsonic autoignitive defla-
gration where, Sl, is the laminar burning
velocity.
(v) n P aSl1 laminar burning deflagration at
the laminar burning velocity.
However, when there are large changes in den-
sity at the pressure front and when chemical
energy can be fed into the developing pressure
wave at a hot spot, the main energy release occurs
during an excitation time, se, that is much smaller
than si. The strength of the pressure pulse is influ-
enced by the rapidity of this energy release [37,45].
The development of detonation depends not just
on n, but also on a dimensionless “reactivity
parameter”, e. If, r0, is the radius of the hot spot,
e, is defined as
r0
e¼ ð11Þ
ase
e is a measure of the hot spot reactivity – it is the
number of excitation times fitting into the pres-
sure pulse during its time in the hot spot. The
pressure wave generated by the hot spot can cou-
ple with the autoignition reaction front when it
moves into the unburned mixture at approxi-
mately the acoustic speed. The fronts are mutually
reinforced to create a damaging pressure spike
propagating at high velocity within the hot spot
in a developing detonation (DD) [42]. This cou-
pling is complex and a full understanding of a
DD requires direct numerical simulation of the
developing front. Such simulations were made in
[41,46] with CO/H2 – air mixtures under different
conditions. These fuels were selected because the
associated reaction kinetics were relatively well
understood. It was found that the results of the
computations could be usefully generalized by
plotting regime boundary values of n and e as in
Fig. 5 [41]. Chemical resonance corresponds to
n = 1. There is a peninsula in Fig. 5, with an upper
limit, nu, and a lower limit, nl, within which deto-
nations can begin to develop within the hot spot.
Whether autoignition, in a knocking SI engine
or an HCCI engine, leads to DD and hence extre-
mely high pressure pulses, can be assessed by
using Fig. 5 if n and e can be estimated [30,59,60].
8 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx
45
40 ξu
35
30
25
N2
20
ξ
Developing
Detonation
15
10 .
5
S2 ξl
0
0 5 10 15 20
ε
Fig. 5. Conditions for the occurrence of developing
detonations in terms of n and e [41,59]. The broken line
refers to pressure traces shown in Fig. 7.
5. Implications for SI engines
5.1. Fuel antiknock quality
The intensity of knock set up by autoignition is
usually characterized by the maximum peak-to-
peak amplitude of the fluctuating part of the pres-
sure signal averaged over many engine cycles. In
SI engine experiments, at a set condition, as the
spark timing is advanced, knock intensity (KI)
increases and the engine is said to knock when
KI exceeds a threshold value which is often set
at 20 kPa. It is possible to have autoignition in
the endgas e.g., observed in optical experiments
[38], without “knock” (benign autoignition). The
spark advance at which KI reaches the threshold
value is known as the Knock Limited Spark
Advance (KLSA) and is used as a measure of
the antiknock quality of the fuel – the larger the
KLSA, the greater is the knock resistance of the
fuel. Also, as the ignition timing is advanced effi-
ciency and torque reach a maximum value at
MBT (minimum advance for best torque) timing
[6,7]. An engine is said to be “knock limited”
when it cannot be run at MBT timing because
of knock. At a given operating condition, the
Octane Requirement (OR) of the engine is the
OI of the fuel with KLSA the same as MBT. Such
a fuel, or a fuel with OI higher than the OR,
allows the engine to run at the best efficiency
and power without knock.
Most modern high-performance engines are
knock-limited in at least some part of their oper-
ating regime on normally available fuels. Such
engines are usually equipped with knock-sensors.
With the engine operating on a given fuel, when
knock is detected, the engine management system
takes corrective action, usually by retarding the
ignition timing, and this causes a reduction in
power. In such vehicles, if the antiknock quality
Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
of the fuel is increased, the chances of knock are
reduced and the engine is more likely to operate
near the optimum condition so that performance
can be improved up to a limit and either the max-
imum power at a constant speed or an accelera-
tion time can be used as measures of the
antiknock quality of the fuel [10,18,19,28].
Downsizing allied with turbocharging is being
very actively pursued by all engine manufacturers
to improve the efficiency of SI engines by minimiz-
ing throttling at low loads while ensuring high
peak power. As with other approaches to increase
the efficiency and performance of the SI engine
25
(e.g., increasing the compression ratio), this
increases the maximum pressure and temperature
of the unburned gas and makes autoignition and
knock more likely. The antiknock quality of the
fuel becomes more important – the octane
requirement of the engine increases. Moreover,
all such approaches also reduce the temperature
of the unburned gas at a given pressure i.e.,
Tcomp15 is reduced. Indeed, throughout the history
of the SI engine, the average value of K and
hence, the importance of MON, has decreased
[47] as SI engines have become more efficient
and Tcomp15 has decreased on average. In fact cur-
rent engines [10,18,19,28,47–52] and down-sized,
turbocharged engines [53,54] which are being
developed for the future, all have negative K val-
ues in the operating ranges where they are likely to
be knock limited.
This is illustrated in Fig. 6 using P1500, the
power at full-throttle at an engine speed of
1500 rpm, for a knock-sensor equipped car mea-
sured at the wheels on a chassis dynamometer –
the results are from Table 5 in [28]. The car was
tested on ten different fuels of different chemical
compositions with varying RON and MON.
Fig. 6a and b illustrate that P1500 showed some
correlation with RON but little correlation
MON. The K value of 0.51 was established by
using P1500 rather than z in Eq. (2). Fig. 6c shows
that the correlation between P1500 and OI using
this negative value of K is much better.
In many parts of the world fuel specifications
have been set on the assumption that MON con-
tributes to antiknock properties of the fuel. These
specifications were set at a time when MON did
matter but in modern engines with higher efficien-
cies, this is not the case anymore. Such gasoline
antiknock quality specifications are inconsistent
with the requirements of current and future
engines at conditions where knock is more likely
[2] and more effort and energy is being spent in
making fuels that are arguably less suitable for
modern engines. As engines seek to improve their
efficiency, this mismatch between specifications
and engine requirements will get wider and will
need to be addressed. There are many parts of
the world, e.g. Japan and China where fuel anti-
knock quality is only specified by RON (i.e., K
 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx 9

(a) 23 down-sizing do not increase indefinitely with

22.5
increasing OI [54]. There are diminishing returns
with increasing OI. Increasing fuel antiknock
Full-throle Power at the wheels, kW

22
quality could also require greater use of energy
21.5 R² = 0.8668 in making the fuel. In general, raising the RON
21 of the gasoline pool will lead to a further surplus
20.5
of low-octane gasoline (see Section 3 also). A
proper well-to-wheel analysis would be required
20
to assess the true benefits for energy consumption
19.5 attainable through increasing fuel antiknock qual-
19 ity in combination with down-sizing.
18.5
A question that crops up often is whether the
ASTM RON and MON tests need to be revised.
18
85 90 95 100
The problem with the RON and MON tests is
RON not that they use an engine of antiquated design
because autoignition is determined by the pres-
(b) 23
sure/temperature history of the endgas rather than
22.5 engine design. The main problem is that the pri-
Full-throle Power at the wheels, kW

22 mary reference fuels (PRF) used to define the

21.5
octane scale have zero sensitivity and do not prop-
R² = 0.1035
erly reflect the auto-ignition behavior of practical
21
fuels. A better scale might be one based on tolu-
20.5 ene/heptane mixtures which have sensitivities com-
20 parable to practical fuels and will also enable fuels
19.5
which fall beyond the octane scale (RON > 100) to
be rated quantitatively – see Section 4.4 in [2].
19

18.5 5.2. Preignition and superknock in turbocharged SI

18
80
(c)
Fig. 6. Power at the wheels at full throttle with the
engine speed at 1500 rpm (P1500), for a knock-sensor
equipped car plotted against 7a) RON, 7b) MON and
7c) OI = RON + 0.51S. Data taken from Table 5 of [28].
is assumed to be zero) and indeed this might be an
acceptable intermediate compromise. Certainly in
such cases setting an additional MON specifica-
tion would be a retrogressive step.
The question also arises as to whether the anti-
knock quality (minimum OI) of gasolines should
be increased and to what level. The gain in knock
limited power, knock limited fuel efficiency and
the potential for improving efficiency through
engines
85 90 95 100
MON
A more detailed discussion of this problem can
be found in Ch.5 of [2]. Turbo-charged DISI
engines have been observed to suffer from an
abnormal combustion phenomenon known as
preignition [30,55–60] where an expanding flame
front is established before the spark plug fires in
the engine. The early start to combustion from
preignition causes the pressure and the tempera-
ture of the unburned gas ahead of the advancing
flame front, the endgas, to rise more rapidly com-
pared to normal spark timing. If autoignition
occurs in the endgas at high pressure and temper-
ature, it can lead to extremely heavy knock, some-
times described as “superknock”, another
abnormal combustion phenomenon which could
potentially damage the engine. Examples of in-
cylinder pressure from cycles, originally discussed
in [30], showing preignition and superknock are
shown in Fig. 7. Cycle S2 shows knock intensity
of over 140 bar (14 Mpa) and could be considered
as a superknock event. Cycle K2 has preignition
starting later than S2 and a KI of about 24 bar
(2.4 MPa) though, at the time of the start of
autoignition, it has a similar amount, about
40%, of unburned mass fraction as Cycle S2.
5.2.1. Preignition
In order to establish a laminar flame, which
later develops into a turbulent flame in the
engine, the fuel/air mixture in a localized, small
volume needs to be heated to a sufficiently high

Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
10 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx
Start of Knock
220
@128 bar/ -
2°CA/ MFB 65%
170
p [bar]
120 Start of Pre-
Ignition
@-19/-6 ° bTDC
70
20
-20 -10
KLSA= -2.3°
Fig. 7. Preignition and superknock in a turbocharged engine. Examples of in-cylinder pressure traces from experiments
discussed in [30]. (MFB – mass fraction burned).
temperature to cause a thermal runaway. In order
to continue to propagate, the flame should have
grown to a critical size, of the order of the laminar
flame thickness, d, by the time the temperature at
the origin has decreased to around the normal
flame temperature. Below this critical flame size,
the rate of heat release within the inner zone of
chemical reaction is insufficient to compensate
for the rate of heat loss to the surroundings and
the temperature would decrease throughout the
reaction volume in the absence of external energy
input of sufficient power and duration [61].
The initiation of preignition during the com-
pression stroke cannot be caused by autoignition
in the bulk fuel/air mixture because the pressure
and temperature are too low and the ignition
delay would be too large [55,59]. In earlier engines
preignition was observed to start at surface hot
spots so that combustion chamber deposits [2]
played a role. In modern DISI engines, preignition
appears to be initiated away from the combustion
chamber walls by the burning of oil/fuel droplets
which are blown into the cylinder from crevices
[55,58] such as at the piston top land though sur-
face ignition cannot be ruled out. Both autoigni-
tion and catalytic processes, not fully
understood at the moment, seem to be responsible
for the creation of this initial hot spot from the
burning of these droplets. This process is stochas-
tic and unpredictable since it will depend on
engine design, the autoignition properties of the
oil and fuel, lubricant additives which might have
catalytic effects, the engine operating conditions
and the flow fields and other parameters which
might affect the transport of the droplets into
the cylinder.
Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
Superknock. S2
Start of Knock @
90 bar/ 20°CA/ Moderate Knock. K2
MFB 62%
Normal cycle. N2
0 10 20 30
Crank Angle degree
Once the initial flame kernel is formed, the
smaller the value of the laminar flame thickness
d (the critical diameter), the larger the probability
of having a kernel bigger than the critical size and
hence of establishing a laminar flame [61,62].
Further,
d  ðl=qS L Þ ð12Þ
The dependence on pressure and temperature of
SL can be expressed in general as
S L ¼ S L0 ðT =T 0 Þp ðP =P 0 Þq ð13Þ
where SL0 is the laminar burning velocity at some
reference pressure,P0, and temperature,T0
Also, from ideal gas law, (P/P0) = (T/T0)(q/q0)
where q0 is the density at pressure,P0, and temper-
ature,T0. Finally dynamic viscosity, l,varies as
T0.5
.
If l0 is the dynamic viscosity at P0 and T0,
from Eqs.(12), (13)
d ¼ ðl0 =q0 S L0 ÞðP =P 0 Þðq1Þ ðT =T 0 Þð1:5pÞ ð14Þ
Many fuels were rated for their preignition resis-
tance (PR) with a procedure developed by
Ricardo and co-workers using a hot wire to pro-
vide the initial source of ignition [63–65]. For
many of the same fuels Farrell et al. [66] measured
the maximum laminar burning velocity, SLmax, at
3.04 bar (P0) and 450 K (T0) for many of these
fuels. Since the ignition criterion is met by the
hot wire, PR should increase as the critical diam-
eter required for establishing a stable flame (lam-
inar flame thickness) increases. In [59], these
results were combined and PR shows a reasonable
negative correlation with SLmax, as suggested by
 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx 11

160 velocity, a, can be calculated from the tempera-

140 ture at the start of autoignition. A kinetic model
120 [67] for a surrogate fuel (for European Premium
100 gasoline) made up of a mixture of isooctane, tolu-
80 ene and n-heptane was used in [30] to calculate si
60
as a function of pressure and temperature from
PR

which a simpler Arrhenius type relationship for

40
si with a pressure correction was established (Eq.
20
(15)). In Eq. (15) si is in ls, pressure, P, is in
0
bar (0.1 MPa) and temperature, T, is in K.
-20 R² = 0.8502  
6626 1:21
-40 si ¼ 83:1 exp P ð15Þ
0.5 0.6 0.7 0.8 0.9 T
SLmax from Farrell et al [66] m/s
The Livengood-Wu integral was calculated for sev-
Fig. 8a. Preignition Resistance (PR) vs Maximum eral knocking cycles in two different engines using
Laminar Burning Velocity, SLmax at 0.3 MPa and Eq. (15) and the measured in-cylinder pressure
450 K. Data taken from [59]. and temperature. The crank angle position for the
onset of knock estimated using this method was
140
in excellent agreement with the observed value indi-
120 cating that Eq. (15) is a very good predictor of si for
100 a European premium (RON98) gasoline. The
excitation time, se, was defined in [30] as the time
80
from 5% of maximum to the point of maximum
60 R² = 0.2502 heat release rate. With the same kinetic mechanism
PR

40 but using the smallest time increment available in

CHEMKIN [68], the chemistry simulation soft-
20
ware used, a single equation, Eq. (16), could be
0 found for se in ls with, P, in bar and T, in K.
 
- 20 799 0:456
se ¼ 3:07 exp P ð16Þ
- 40 70.0 80.0 90.0 100.0 110.0 120.0
T
RON The temperature gradient in Eq. (9) and the radius
Fig. 8b. PR vs RON. Data from [59].
of the hot spot, r0, in Eq. (11) were assumed and n
and e could be estimated [30,59] for the conditions
of interest. It could then be shown that the cycles
140
which showed extremely high knock intensity,
120 such as cycle S2 in Fig. 7, fell well within the
100 DD peninsula in Fig. 5. However, the “normal”
knocking cycles where autoignition of the endgas
80
started at relatively low pressures and tempera-
60 R² = 0.3309 tures, such as cycle N2 in Fig. 7, fell outside this
peninsula. Rudloff et al. [60] have also explained
PR

40
their experimental results in a similar manner.
20
Thus very high knock intensities in superknock
0 can almost certainly be attributed to DD. These
- 20 points are discussed in much greater detail in
[59] and in Ch.5 of [2].
- 4070.0 80.0 90.0 100.0 110.0 Peters et al. [69] have used a refinement of turbu-
MON lence theory to estimate the joint probability of the
temperature gradient and the length scale, l, over
Fig. 8c. PR vs MON. Data from [59]. which the temperature gradient exists; l is taken as
a measure of r0. They can then estimate, using
Eq. (14) (see Fig. 8a) but very little correlation Fig. 5, the probability of DD occurring in an engine
with RON (Fig. 8b) or MON (Fig. 8c) of the fuels, running at a given pressure and temperature regime
again as expected, since autoignition of the bulk on a fuel whose autoignition behavior can be mod-
mixture plays no role in preignition. elled. This approach, unlike previous analyses,
includes turbulence and does not require any
5.2.2. Superknock assumptions to be made about the size or the tem-
The pressure and temperature at the start of perature gradient associated with the hot spot.
autoignition, Pai and Tai are estimated from the In general, fuels which are more resistant to
measured pressure curve as in Fig. 7. The acoustic autoignition will have lower values of ua [37] as

Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
12 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx
can be inferred from Eq. (10) because they will
have higher ignition delays, si. Hence all else being
equal, the likelihood of a DD decreases as the
resistance of the fuel to autoignition increases
[37,69]. Fuels like ethanol and hydrogen are very
susceptible to preignition because of their high
laminar burning velocities but, because of their
higher resistance to autoignition, preignition
might not lead to superknock. Given the increas-
ing importance of ethanol as a gasoline compo-
nent, this area needs further investigation.
6. Implications for CI engines
Soot formation can be minimized by ensuring
that the equivalence ratio, u, of the mixture pack-
ets where combustion occurs is not greater than
u > 2 whereas NOx formation can be minimized
if the combustion temperatures are kept below
about 2200 K [70,71]. Combustion temperatures
and hence NOx can be reduced by increasing the
use of exhaust gas recirculation (EGR). Engine-
out soot reflects the balance between the forma-
tion and the in-cylinder oxidation of the soot that
is formed. Both the temperature and oxygen con-
tent in the cylinder decrease if EGR level is
increased in order to control NOx, oxidation of
soot decreases and engine-out soot or particulate
matter (PM) increases leading to the well-known
NOx/PM trade-off in diesel engines. If soot for-
mation is avoided in the first place by ensuring
that the equivalence ratio everywhere in the cylin-
der at the time of combustion is below the soot
forming limit, EGR can be used to control NOx
without increasing engine-out soot. In general, if
NOx is controlled through low temperature com-
bustion, HC and CO increase and have to be con-
trolled via exhaust after-treatment.
In HCCI engines the global u < 1 and fuel and
air are fully premixed so that soot formation can
be avoided. Use of EGR combined with boosting
allows the maximum load at which HCCI com-
bustion with low NOx and acceptable pressure
rise rate. However, there is no in-cycle control of
combustion phasing in HCCI engines. Engine sys-
tems which work in HCCI mode where feasible
and switch to SI mode where HCCI combustion
is not desirable, particularly at high load, have
been developed. The fuel requirement as well as
engine design (e.g., compression ratio) of such a
system is determined by the need to avoid knock
in the SI mode at high load. Hence such an engine
will need high anti-knock quality gasoline.
In conventional diesel engines combustion
phasing is controlled by the timing of the final fuel
injection. In such systems, soot formation can
be avoided only if the final injection of fuel is
completed sufficiently before combustion starts.
Premixed compression ignition (PCI) combustion
to avoid soot formation was defined to occur in
Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
[72] when combustion starts after the final fuel
injection is over and the engine-out smoke is very
low- below 0.05 FSN (filter smoke number). A rel-
evant time scale is the Ignition Delay, ID, which is
defined as (SOC – SOI) where SOC and SOI are
respectively, the crank angle at the start of com-
bustion and the start of the final fuel injection.
It is to be noted that ID differs from the ignition
delay in pre-mixed systems in that ID will also
depend on the physical mixing process of the fuel
and air in addition to the chemical structure of the
fuel and the pressure and temperature history.
PCI combustion, and hence simultaneous reduc-
tion of NOx and soot, is very difficult to achieve
with conventional diesel fuel which ignites readily
i.e., ID is very low. It is often not possible to com-
plete the injection of fuel, particularly at higher
loads, before combustion starts with conventional
diesel fuels and the diesel fuel will burn in a quasi-
steady jet diffusion flame [71].
6.1. Gasoline Compression Ignition (GCI)
PCI combustion is much easier with fuels in
the gasoline autoignition range (RON > 60)
because the higher ignition delay allows more time
for mixing of the fuel and oxygen before combus-
tion starts [e.g.,[72–79]]. This approach can be
termed Gasoline Compression Ignition, GCI –
see Chapter 6 in [2] for a more detailed discussion.
Fuel is injected earlier compared to the diesel fuel
to get the same combustion phasing as with diesel
fuel. The extra time available ensures much
greater mixing of fuel and air compared to the die-
sel fuel before combustion starts. However the
fuel must not be injected too early or else the fuel
and air will be too well mixed and conditions
approach HCCI conditions.
GCI combustion is easier to achieve than
HCCI combustion – even at engine operating con-
ditions where HCCI combustion is not possible,
there is an injection timing window available for
each fuel and operating condition for GCI where
combustion is stable and combustion phasing can
be controlled by injection timing [73]. In contrast,
at conditions where GCI combustion is not possi-
ble, HCCI combustion will also not be possible. In
HCCI combustion, the mixture is nominally
homogeneous so that if the pressure and tempera-
ture history is not right, the mixture might not
ignite at all. If there is sufficient inhomogeneity
in the mixture autoignition will occur somewhere
and combustion will start in the cylinder and will
promote combustion of the rest of the charge by
increasing the pressure and temperature. It is pos-
sible that in such a combustion system local flame
fronts might exist and not all combustion will be
via autoignition. Fuel and air must be “premixed
enough” everywhere in the cylinder before com-
bustion occurs to avoid soot formation but must
not be fully premixed as in HCCI combustion.
 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx
At high loads proper stratification or inhomoge-
neity will ensure that heat release is distributed
in time so that the peak heat release rate and the
maximum pressure rise rate will not be too high.
This can be achieved through multiple injections
of the fuel [e.g. [74,76,78]].
6.2. Fuel autoignition quality and volatility effects
in GCI engines
If there is sufficient ignition delay, low NOx
and low soot operation can be obtained even with
a relatively low volatility fuel, in the diesel boiling
range, containing very high levels (75%) of heavy
aromatics [Chapter 6 in [2]]. In PCI combustion, if
two fuels have the same ID at a given operating
condition, they have comparable NOx, smoke,
CO, ISFC and MPRR regardless of any differ-
ences in volatility or composition [72].
Moreover, if the RON and MON can be mea-
sured i.e., if the fuels are in the gasoline boiling
range, ID correlates very well with OI (Eq. (1))
with the K value highly negative [72]. Depending
on engine design, the optimum RON for GCI
appears to be between 70 and 85 [2] – much lower
compared to current gasolines. Ignition delay
depends non-linearly on OI. If the RON of the
fuel is too low, ID will be low and comparable
to conventional diesel fuel – PCI combustion
becomes difficult. On the other hand if RON is
too high, ID will be too high and make starting
and low load operation more difficult.
6.3. Dual fuel or Reactivity Controlled CI (RCCI)
One way of dealing with the requirement of
varying ID at different operating conditions is
the dual fuel approach, termed Reactivity Con-
trolled CI (RCCI) by Reitz and co-workers [80].
In this approach, a fuel with high resistance to
autoignition such as commercially available gaso-
line or a mixture of gasoline and ethanol or natu-
ral gas is injected in the port and ignition is
triggered by the direct injection of a fuel of high
reactivity such as commercially available diesel
fuel near top dead center. The ratio of the two
fuels used is changed depending on the engine
operating condition (e.g., relatively more diesel
fuel is used at low loads). However, over a normal
operating range the amount of diesel fuel used is
around 10% of the total fuel used. As summarized
in [80], RCCI combustion is capable of operating
over a wide range of engine loads with near zero
levels of NOx and soot, acceptable pressure rise
rate and noise, and very high indicated efficiency.
6.4. Advantages and challenges of GCI for engine
design
In GCI, mixing is dominated by the ignition
delay rather than the injection event and hence
Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
13
injection pressures need not be as high as with die-
sel fuel [75,78,79]. Indeed at low loads, lower
injection pressures significantly improve combus-
tion stability and reduce CO and HC emissions,
apparently by increasing the stratification. There
is great scope for reducing the cost of the injection
system compared to modern diesel engines by
reducing the injection pressure.
With GCI, HC and CO emissions will, in gen-
eral, be higher than in conventional diesel com-
bustion and will have to be controlled by
appropriate exhaust after-treatment. The focus
of exhaust after-treatment shifts to oxidation of
HC and CO, which should be relatively easy in
the oxygen-rich exhaust, compared to the control
of NOx as in conventional diesel engines. There is
scope to simplify the after-treatment system com-
pared to advanced diesel engines. However, if
NOx is controlled in-cylinder at sufficiently low
levels, exhaust gas temperatures will be lower than
found in conventional diesel engines and this will
make oxidation of HC and CO more challenging.
Very high efficiencies can be achieved with
GCI. Indicated fuel efficiencies of up to 56% have
been measured with single fuel [76] as well as dual
fuel [80] operation in heavy duty engines running
at medium to full load. In light duty engines, com-
pared to running the engine on gasoline in SI
mode, fuel consumption could be reduced by up
to 26% [78,79].
6.5. Modelling requirements for GCI engines
There are many engine operating and design
parameters such as speed, injection pressures,
injection strategies, EGR levels, valve timings,
intake pressure and temperature, compression
ratio and injector design that can be varied to
meet required targets for efficiency, emissions
and pressure rise rates. Reliable models which
can guide experiments and optimization would
be valuable in the development of this approach.
Some examples of such modeling efforts are
reported in [e.g., [81–83]].
The models need to predict the spatial and
temporal development of the mixture in the cylin-
der and combine this with an appropriate chemi-
cal kinetic model which predicts autoignition in
the cylinder. The kinetic model needs to be small
enough so that it can be merged with a mixing
model. The simplest kinetic surrogate for gasoline
is a mixture of iso-octane, n-heptane and toluene
which has the same RON and MON [72]. There
is a need to develop simpler, well-validated kinetic
schemes for such surrogates which can be used
with CFD mixing models.
6.6. Outlook for GCI engines
GCI offers the possibility of a relatively cheap
but efficient engine and after-treatment system
14 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx
with acceptable emissions. Moreover, the fuel
needed could have much lower octane (70–85
RON) than today’s gasolines and could contain
components in the diesel boiling range. Indeed
refinery components like straight run gasoline
from the initial distillation of crude oil would have
octane numbers near this range and could be used
with little or no upgrading in such engines.
Development work is needed to ensure reliable
cold start, idle and transient operation, acceptable
driveability and compliance with emissions regu-
lations, particularly on CO and HC. The fuel will
need to have sufficient lubricity and detergency (to
keep the injector clean) and this can be achieved
by the use of appropriate additive packages.
In the short term, say up to 2020, we should
expect that diesel fuel (CN > 40) will continue to
be used in CI engines because there has been so
much investment made by all the OEMs in devel-
oping the engine and after-treatment technology
for such systems. However, such engines will try
to use PCI combustion as much as possible and
will have to use higher injection pressures to
achieve this while using conventional diesel fuel.
Such systems will also have to use extensive
after-treatment to meet increasingly stringent
emissions standards on NOx and soot. This will
increase their cost and also might reduce their effi-
ciency (e.g., because of the extra fuel to be used in
the regeneration of diesel particulate filters or
because of lower compression ratios used to
increase ignition delays). Simultaneously global
demand for diesel fuel will increase more than
for gasoline and large investments, running to
hundreds of billions of dollars, will be needed in
the refining industry to supply this increased
demand and meet the changed demand structure.
Even with such investments, light fractions such
as straight run gasoline with low octane numbers
which normally would have been further processed
to make gasoline will be in relative abundance.
Indeed, this imbalance in demand growth between
light and heavy fuels is likely to become a more
compelling driver for the development of GCI
engines than the technical benefits from both the
engine and fuels points of view discussed above.
7. Summary and concluding remarks
(a) Even by 2040, transport energy will mostly
(90% share) come from petroleum-based
liquid fuels.
(b) There will be sufficient oil globally to meet
the increasing global demand for transport
energy.
(c) Under the current technology trajectory,
the growth in global transport fuels demand
is very heavily skewed towards commercial
transport and hence heavier fuels.
(d) The primary development trend in SI
engines is to improve efficiency. This will
Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
make fuel antiknock quality more impor-
tant. Moreover for a given RON a lower
MON fuel will have higher antiknock qual-
ity where the engine is knock limited in such
engines. Higher fuel antiknock quality is
also likely to mitigate “superknock” in
turbo-charged SI engines. Fuel specifica-
tions which assume that MON contributes
to higher antiknock quality might need to
be changed as the gap between engine
requirements and specifications widens.
(e) The primary development trend in diesel
(conventional CI) engines is to reduce soot
and NOx at affordable cost without com-
promising efficiency. This is best achieved
by running CI engines on “low quality” gas-
oline – RON between 70 and 85 and no
strict demands on volatility. This approach
can be termed Gasoline Compression Igni-
tion (GCI).
(f) GCI has the following advantages – 1.
Opens up a path to mitigate expected
demand imbalance between heavy and light
fuels, 2. Offers at least diesel-like efficiency
at lower cost compared to running diesel
engines on conventional diesel fuels, 3. Uses
fuels that are “less processed” compared to
current gasoline and diesel that might be in
surplus otherwise. Such fuels are likely to
require less energy to produce compared
to current gasoline and diesel fuels.
Thus, current trends in engine development
will cause the increase in transport fuel demand
to be skewed towards middle distillates (diesel
and jet fuel) and high octane gasoline. It will
require investments running into hundreds of bil-
lions of dollars globally by the refining industry
and will lead to a surplus of low-octane compo-
nents such as naphtha. This is likely to prove very
challenging for the refining industry and also, in
general, for the efficient use of petroleum Devel-
opment of new efficient fuel/engine systems which
can use such low-octane components will help in
meeting these challenges.
Based on the preceding discussion, a personal,
and hence partial, list of combustion science issues
that are of relevance to fuel and engine studies is
given below.
(a) Explain, in chemical kinetic terms why the
pressure exponent, n, in Eq. (7) is different
for PRF and non-PRF fuels.
(b) Autoignition phasing in knock, HCCI and
PCI combustion is very well predicted by
OI regardless of whether the sensitivity of
gasoline arises due to olefins, aromatics or
oxygenates such as ethanol or MTBE
[18,24,50,52,72]. Volume concentrations of
oxygenates were high in some of these studies
e.g., ethanol at 20% in [24] and 63.5% in
 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx
[50]. A mixture of toluene and PRF (iso-
octane and n-heptane), TPRF, can match
RON and sensitivity and hence OI for a [9]
gasoline at all operating conditions and will
have the same autoignition phasing if both [10]
fuels are subjected to the same pressure
and temperature evolution with time. Of [11]
course the detailed chemistry, e.g., for soot
formation as well as other effects like charge
cooling of such a surrogate and the gasoline [12]
might be different. The need is for simpler,
well-validated kinetic models for such sur- [13]
rogates that can be coupled to other CFD
models (e.g., mixing). [14]
(c) Measure laminar burning velocity for prac-
tical fuels and fuel components and develop
models to predict the laminar burning [15]
velocity of blends. This will be useful to
understand pre-ignition as well as flame
development in SI engines. [16]
(d) Ignitability of lean mixtures and stability of
combustion in lean and dilute (high EGR) [17]
mixtures in SI engines. [18]
(e) Fuel effects on turbulent flame development [19]
in lean and dilute (high EGR) mixtures. [20]
(f) Improve the understanding of droplet com-
bustion, particularly of lubricant droplets, [21]
to understand the initiation of preignition.
This is likely to include the study of autoig-
nition as well as possible catalytic effects. [22]
(g) Does GCI combustion simultaneously
involve flames and autoignition? How
would practical models handle this? [23]
(h) Understand catalytic oxidation of CO and
HC to enable development of low-tempera- [24]
ture catalysts.
(i) Understand mechanisms of deposit forma- [25]
tion in fuel injectors and combustion cham-
bers better. [26]
[27]
References
[28]
[1] U.S. Energy Information Administration, at
<http://www.eia.gov/tools/faqs/faq.cfm?id=447 [29]
&t=1>.
[2] G.T. Kalghatgi, Fuel/Engine Interactions, SAE [30]
International, Warrendale, PA, 2014.
[3] U.S. Energy Information Administration at <http://
www.eia.gov/forecasts/ieo/pdf/0484(2013).pdf>.
[4] ExxonMobil at <http://www.exxonmobil.co.uk/
corporate/files/news_pub_eo2012.pdf>. [31]
[5] World Energy Council, Global Transport Scenarios
2050, WEC, London, 2011. [32]
[6] J.B. Heywood, Internal Combustion Engine Fun- [33]
damentals, McGraw Hill Book Co., New York,
1988. [34]
[7] R. Stone, Introduction to Internal Combustion
Engines, 4th ed., Palgrave Macmillan, Basingstoke,
2012. [35]
[8] J.G. Speight, Petroleum Refinery Processes. In:
Wiley Critical Content: Petroleum Technology –
Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
15
Vol. 1, John Wiley & Sons Inc., Hoboken,
NJ, 2007.
K. Owen, T. Coley, Automotive Fuels Reference
Book, SAE International, Warrendale, PA,
1995.
G.T. Kalghatgi, SAE Paper 2005-01-0239,
2005.
Annual book of ASTM Standards, Vol. 5.01-5.05,
ASTM International, West Conshohocken PA.,
2012.
G.T. Kalghatgi, Int. J. Engine Res. 15 (2014) 383–
398.
V. Smil, Energy Myths and Realities: Bringing
Science to the Energy Policy debate, American
Enterprise Institute, Washington, 2010.
L. Maugeri, at <http://belfercenter.ksg.har-
vard.edu/files/Oil-%20The%20Next%20Revolution.
pdf>.
BP, at <http://www.bp.com/en/global/corporate/
about-bp/statistical-review-of-world-energy-2013.
html>.
U.S. Energy Information Administration at <http://
www.eia.gov/analysis/studies/worldshalegas/, 2013>.
ExxonMobil at <http://www.exxonmobil.co.uk/
Corporate/files/news_pub_eo.pdf>.
G.T. Kalghatgi, SAE Paper 2001-01-3585, 2001.
A. Bell, SAE Paper 2010-01-1454, 2010.
C.K. Westbrook, Proc. Combust. Inst. 28 (2000)
1563–1577.
W.J. Pitz, N.P. Cernansky, F.L. Dryer, F. Egolfo-
poulos, J.T. Farrell, D.G. Friend, H. Pitsch, SAE
Paper 2007-01-0175, 2007.
J.T. Farrell, N.P. Cernansky, F.L. Dryer, F.G.
Friend, C.A. Hergart, C.K. Law, R. McDavid, C.J.
Mueller, A.K. Patel, H. Pitsch, SAE Paper 2007-01-
0201, 2007.
G.T. Kalghatgi, R.A. Head, Int. J. Engine Res. 7
(2006) 215–236.
G.T. Kalghatgi, P. Risberg, H.-E. Ångström, SAE
Paper 2003-01-1816, 2003.
P. Risberg, G.T. Kalghatgi, H.-E. Ångström, SAE
Paper 2003-01-3215, 2003.
N. Morgan, A. Smallbone, A. Bhave, M. Kraft, R.
Cracknell, G. Kalghatgi, Combust. Flame 157
(2010) 1122–1123.
J.C. Livengood, P.C. Wu, Proc. Combust. Inst. 3
(1955) 347.
G.T. Kalghatgi, K. Nakata, K. Mogi, SAE Paper
2005-01-0244, 2005.
D. Bradley, R.A. Head, Combust. Flame 147 (2006)
171–184.
G.T. Kalghatgi, D. Bradley, J. Andrae, A.J. Har-
rison, in Ind. Mech. Eng. Conference on Internal
combustion engines: Performance, Fuel Economy
and Emissions, in London, December 8–9, 2009,
Chandos Publishing, 2009, ISBN 9781843346074.
K. Fieweger, R. Blumenthal, G. Adomeit, Combust.
Flame 109 (1997) 599–619.
W.R. Leppard, SAE Paper 902137, 1990.
B.M. Gauthier, D.F. Davidson, R.K. Hanson,
Combust. Flame 139 (2004) 300–311.
J. Herzler, M. Fikri, K. Hitzbleck, R. Starke, C.
Schulz, P. Roth, G.T. Kalghatgi, Combust. Flame
149 (2007) 25–31.
M. Fikri, J. Herzler, R. Starke, C. Schulz, P. Roth,
G.T. Kalghatgi, Combust. Flame 152 (2008)
276–281.
16 G.T. Kalghatgi / Proceedings of the Combustion Institute xxx (2014) xxx–xxx
[36] D. Bradley, C. Morley, H. Walmsley, SAE 2004-01-
1970, 2004.
[37] D. Bradley, Philos. Trans. Royal Soc. A 370 (2012)
689–714.
[38] G. König, R. Maly, D. Bradley, A.K.C. Lau,
C.G.W. Sheppard, SAE Paper 902136, 1990.
[39] J. Pan, C.G.W. Sheppard, A. Tindall, M. Berzins,
S.V. Pennington, G.M. Ware, SAE Paper 982616,
1998.
[40] C.G.W. Sheppard, S. Tolegano, R. Wooley, SAE
Paper 2002-01-2831, 2002.
[41] D. Bradley, C. Morley, X.J. Gu, D.R. Emerson,
SAE Paper 2002-01-2868, 2002.
[42] Ya.B. Zeldovich, Combust. Flame 39 (1980) 211.
[43] A.K. Oppenheim, Philos. Trans. R. Soc. Lond. A
315 (1985) 471–508.
[44] D. Bradley, G.T. Kalghatgi, Combust. Flame 156
(2009) 2307–2318.
[45] A.E. Lutz, R.J. Kee, J.A. Miller, H.A. Dwyer,
A.K. Oppenheim, Proc. Combust. Inst. 22 (1988)
1683–1693.
[46] X.J. Gu, D.R. Emerson, D. Bradley, Combust.
Flame 133 (2003) 63–74.
[47] V. Mittal, J.B. Heywood, SAE Paper 2009-01-2622,
2009.
[48] K. Akihama, M. Taki, S. Takasu, T. Ueda, Y.
Iwashita, J.T. Farrell, W. Weissman, SAE Paper
2004-01-1950, 2004.
[49] M.D., Gerty, J.B. Heywood, SAE Paper 2006-01-
0228, 2006.
[50] V. Mittal, J.B. Heywood, SAE Paper 2008-01-2414,
2008.
[51] CRC, Fuel Antiknock Quality – Engine Response
to RON vs MON, Scoping Tests – Final Report-
Report No. 660, Coordinating Research Council
(CRC), 2011.
[52] G.T. Kalghatgi, SAE Paper 2001-01-3584, 2001.
[53] T. Davies, R. Cracknell, G. Lovett, L. Cruff, J.
Fowler, SAE Paper 2011-01-1985, 2011.
[54] A. Amer, H. Babiker, J. Chang, G. Kalghatgi, P.
Adomeit, A. Brassat, M. Guenther, SAE Paper
2012-01-1634, 2012.
[55] C. Dahnz, K.-H. Han, U. Spicher, M. Magar, R.
Schiessel, U. Maas, SAE Paper 2010-01-0355,
2010.
[56] C. Dahnz, U. Spicher, Int. J. Engine Res. 11 (2012)
485–498.
[57] M. Amann, D. Mehta, T. Alger, SAE Paper 2011-
01-0342, 2011.
[58] A. Zadeh, P. Rothenberger, W. Nguyen, M. Anb-
arasu, S. Schmuck-Soldan, J. Schaefer, T. Goebel,
SAE Paper 2011-01-0340, 2011.
Please cite this article in press as: G.T. Kalghatgi, Proc. Combust. Inst. (2014), http://dx.doi.org/
10.1016/j.proci.2014.08.031
View publication stats
[59] G.T. Kalghatgi, D. Bradley, Int. J. Engine Res. 13
(2012) 399–414.
[60] J. Rudloff, J.-M. Zaccardi, S. Richard, J.M. Ander-
lohr, Proc. Combust. Inst. 32 (2013) 2959–2967.
[61] B. Lewis, G. von Elbe, Combustion, Flames
and Explosions of Gases Ch.5, Academic Press,
1961.
[62] G. Joulin, Combust. Sci. Technol. 43 (1985)
99–113.
[63] D. Downs, F.B. Theobald, Proc. Inst. Mech. Eng.
(Auto Division) 178 (1963) 89–102.
[64] H.R. Ricardo, J.G.G. Hempson, The high-speed
internal combustion engine. Ch. 4, Blackie and Son
Ltd., 1972.
[65] R.H. Thring, SAE Paper 831685, 1983.
[66] J.T. Farrell, R.J. Johnson, I.P. Androulakis, SAE
Paper 2004-01-2936, 2004.
[67] J. Andrae, P. Björnbom, R.F. Cracknell, G.T.
Kalghatgi, Combust. Flame 149 (2007) 2–24.
[68] Reaction Design, CHEMKIN, http://www.reac-
tiondesign.com/products/open/chemkin.html.
[69] N. Peters, B. Kerschgens, G. Paczko, SAE Paper
2013-01-1109, 2013.
[70] T. Kamimoto, M. Bae, SAE Paper 880423, 1988.
[71] J.E. Dec, Proc. Combust. Inst. 32 (2009) 2727–2742.
[72] G.T. Kalghatgi, L. Hildingsson, A.J.L. Harrison,
B. Johansson, Int. J. Engine Res. 12 (2011) 452–465.
[73] G.T. Kalghatgi, P. Risberg, H.-E. Ångström, SAE
Paper 2006-01-3385, 2006.
[74] G.T. Kalghatgi, P. Risberg, H.-E. Ångström, SAE
Paper 2007-01-006, 2007.
[75] L. Hildingsson, B. Johansson, G.T. Kalghatgi, A.J.
Harrison, SAE Paper 2010-01-0607, 2010.
[76] V. Manente, B. Johansson, W. Canella, Int. J.
Engine Res. 12 (2011) 194–208.
[77] R. Hanson, D. Splitter, R. Reitz, SAE Paper 2009-
01-1442, 2009.
[78] M.C. Sellnau, M. Foster, K. Hoyer, W. Moore, J.
Sinnamon, H. Husted, SAE Int. J. Engines 7, (SAE
Paper 2014-01-1300), 2014.
[79] J. Chang, Y. Viollet, A. Amer, G. Kalghatgi, SAE
Paper 2013-01-2701, 2013.
[80] S.L. Kokjohn, R.M. Hanson, D.A. Splitter, R.D.
Reitz, Int. J. Engine Res. 12 (2011) 209–226.
[81] A. Dempsey, R. Reitz, SAE Int. J. Engines 4 (2011)
338–359.
[82] Y. Ra, P. Loeper, R. Reitz, et al., SAE Int. J.
Engines 4 (2011) 1412–1430.
[83] A. Bhave, A. Coble, S. Mossbach, M. Kraft, N.
Morgan, G. Kalghatgi, SAE Paper 2011-01-1184,
2011.