Content uploaded by Felix Leach
Author content
All content in this area was uploaded by Felix Leach on Nov 23, 2020
Content may be subject to copyright.
Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Transportation Engineering 1 (2020) 100005
Contents lists available at ScienceDirect
Transportation Engineering
journal homepage: www.elsevier.com/locate/treng
The scope for improving the eciency and environmental impact of
internal combustion engines
Felix Leach
a
,
∗
, Gautam Kalghatgi
a
, Richard Stone
a
, Paul Miles
b
a
Department of Engineering Science, University of Oxford, Parks Rd, Oxford OX1 3PJ, UK
b
Combustion Research Facility, Sandia National Laboratories, 7011 East Avenue, Livermore, CA 94550, USA
Keywords:
Internal combustion engine
Emissions
Eciency
Hybridisation
Currently 99.8% of global transport is powered by internal combustion engines (ICEs) and 95% of transport
energy comes from liquid fuels made from petroleum. Many alternatives including battery electric vehicles (BEVs)
and other fuels like biofuels and hydrogen are being considered. However, all these alternatives start from a
very low base and face very signicant barriers to unlimited expansion so that 85–90% of transport energy
is expected to come from conventional liquid fuels powering combustion engines even by 2040. Hence it is
imperative that ICEs are improved in order to reduce the local and global environmental impact of transport.
This paper considers the scope for such improvement after discussing the basic principles that govern engine
eciency and the technologies to control exhaust pollution. The great scope for such improvement is illustrated
by considering various practical approaches already in the market. For instance, the best in class SI engines
in the U.S. have 14% lower fuel consumption compared to the average. Engine and conventional powertrain
developments alone could reduce the fuel consumption by over 30% for light duty vehicles (LDVs). Implementing
other technologies such as hybridisation and light-weighting could reduce fuel consumption by 50% compared to
the current average for LDVs. Current after-treatment technology can ensure that the exhaust pollutant levels meet
the most stringent current emissions requirements. Indeed, with the most modern diesel vehicles, the exhaust can
be cleaner than the intake air in urban centres. The implications for transport policy, particularly where there
are plans to ban ICEs, are considered in the nal discussion. All available technologies need to be deployed to
mitigate the environmental impact of transport and it would be extremely short-sighted to discourage further
development of ICEs by limiting their sales.
1. Introduction
Transport of goods and people contributes around 25% of global CO
2
emissions from fossil fuel combustion [1] . However, its share of global
greenhouse gas (GHG) emissions including other contributors such as
methane is around 14% [ 2 , 3 ], comparable to the share from livestock
Abbreviations: ASTM, American Society of Testing and Materials; BEV, battery electric vehicle; CAD, crank angle degrees; CFD, computational uid dynamics;
CFR, cooperative fuels research; CI, compression ignition; CO, carbon monoxide; DCN, derived cetane number; DPF, diesel particulate lter; EGR, exhaust gas re-
circulation; EOI, end of injection (CAD); EU, European Union; FSN, lter smoke number; GCI, gasoline compression ignition; GHG, greenhouse gas; GPF, gasoline
particulate lter; HC, hydrocarbons; HCCI, homogeneous charge compression ignition; HEV, hybrid electric vehicles; ICE, internal combustion engine; ID, igni-
tion delay (SOC-SOI); IDW, ignition dwell (SOC-EOI); LPG, liquid petroleum gas; MBT, maximum brake torque; MFB, mass fraction burned; MON, motor octane
number; NO
x
, nitrogen oxides; OOD, octane on demand; PPC, partially premixed compression; RCCI, reactivity controlled CI; RDE, real driving emissions; RON,
research octane number; S, sensitivity (RON-MON); SCR, selective catalytic reduction; SI, spark ignition; SOC, start of combustion (CAD); SOI, start of injection
(CAD); SRG, straight run gasoline; TWC, three way catalyst; AFR, air fuel ratio; AKI, anti knock index; ASC, ammonia slip catalyst; BOE, barrel of oil equivalent;
CN, cetane number; CNG, compressed natural gas; DEF, diesel exhaust uid; DME, dimethyl ether; DOC, diesel oxidation catalyst; GDI, gasoline direct injection;
LDV, light duty vehicle; LNG, liquid natural gas; LNT, lean NOx trap; MTBE, methyl tert-butyl ether; NEDC, new european drive cycle; RFO, residual fuel oil;
PEMS, portable emissions measurement system; PFI, port fuel injection; PHEV, plugin hybrid electric vehicle; PM, particulate matter; PMEP, pumping mean ef-
fective pressure; PRF, primary reference fuel; TN, toluene number; TRF, toluene reference fuel; UHC, unburned hydrocarbons; WLTP, worldwide light duty test
procedures.
∗ Corresponding author.
E-mail address: felix.leach@eng.ox.ac.uk (F. Leach).
farming [4] . The world had an estimated 1.1 billion light duty vehi-
cles (LDV), dened as those weighing less than 8500 lb (3860 kg) and
around 380 million heavy goods vehicles (HGV) in 2015 [5] ; in 2018 the
global production of LDVs was around 70 million and that of commercial
vehicles, around 25 million [6] . The number of vehicles is increasing,
primarily in developing countries, and by 2040 the number of LDVs in
https://doi.org/10.1016/j.treng.2020.100005
Received 18 April 2020; Received in revised form 15 May 2020; Accepted 24 May 2020
2666-691X/© 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license.
( http://creativecommons.org/licenses/by-nc-nd/4.0/ )
F. Leach, G. Kalghatgi and R. Stone et al. Transportation Engineering 1 (2020) 100005
Table 1
Percentage share of global transport en-
ergy demand in 2018 across dierent
transport sectors [10] .
Passenger Sector
Light duty vehicles (LDV) 44%
Rail, buses and 2-3 wheelers 7%
Aircraft 10%
Freight transport
Heavy duty road 26%
Marine 8%
Rail and pipeline 5%
the world is expected to increase to 1.7 to 2 billion [ 5 , 7-10 ]. Currently
transport is almost entirely ( > 99.8%) powered by combustion engines –
land and marine transport runs on reciprocating internal combustion en-
gines (ICEs) while air transport is dominated by jet engines. Combustion
engines also full important roles in industry and power generation.
Liquid fuels have become the energy source of choice for transport
because of their high energy density and because they are easy to trans-
port and store. For instance, at normal temperatures and pressure, the
volumetric energy density of gasoline is around 3100 times and 800
times higher compared to hydrogen and natural gas, respectively. A vast
global infrastructure for the manufacture and distribution of liquid fu-
els which will be expensive and dicult to replace or replicate has been
established over the past century or so. Transport and petroleum (crude
oil) are very closely linked – currently around 95% transport energy is
provided by liquid fuels made from petroleum and around 60% of crude
oil is used to make transport fuels [7-10] .
Table 1 shows the share in demand of transport energy across dif-
ferent transport sectors –it is to be noted that aircraft and land and
marine freight transport account for 50% of global transport energy de-
mand [10] . Around 80% of global LDVs are powered by spark-ignition
(SI) engines using gasoline [8] while the freight sector is dominated by
compression ignition (CI) engines running on diesel fuel. Marine trans-
port mostly uses residual fuel oil (RFO) though a large fraction of RFO
is also used for domestic and industrial heating.
The global demand for transport energy is large - Table 2 shows the
average daily demand for transport fuels for the third quarter of 2018
[11] . The rst column shows the demand in barrels of oil equivalent
(BOE). An exajoule (10
18
Joule) is equivalent to 163.4 million BOE and
assuming a volumetric energy density of 32.5 MJ/l for gasoline, 36 MJ/l
for diesel and jet fuel and 40 MJ/l for residual fuel oil, the fuel volumes
used are shown in the last column. Thus, the world uses, on average,
over 11 billion litres of gasoline, diesel and jet fuel daily . This demand
is growing with the growth expected to occur almost entirely in devel-
oping (non-OECD) economies and is expected to be around 40% higher
in 2040 compared to 2015 [7-10] . The demand for energy for freight
transport and aviation is expected to grow faster than for lighter vehi-
cles because of the greater opportunity for improving fuel eciency and
for electrication in the light duty sector [ 7 , 9 ]. Hence the demand for
diesel and jet fuel (middle distillates) is expected to grow faster than for
gasoline [ 7 , 9 ].
Currently there are many initiatives across the world to develop al-
ternatives to ICEs and petroleum-based conventional fuels driven by
concerns about climate change and local air quality associated with
transport because of emissions of CO
2
, particulates, nitrogen oxides
(NO
x
), carbon monoxide (CO) and hydrocarbons (HC). Indeed, criti-
cism of ICEs is common in the news media and amongst some politi-
cians in many countries and has led to a belief that the elimination of
ICEs is desirable and imminent. Of course, transport policy is also in-
uenced greatly in many countries by the desire for economic growth,
energy independence and energy security. The ICE could be replaced
by a battery or a fuel cell while the alternatives to conventional fuels
include biofuels, natural gas, hydrogen, synthetic fuels, electro-fuels,
liquid petroleum gas (LPG) and methanol [ 12 , 13 ].
However, all these alternatives start from a very low base and face
signicant barriers to rapid and unrestrained growth. Such alternatives
need to be assessed on the basis of life-cycle analysis and there will be
severe environmental, economic and social consequences if alternatives
are imposed prematurely [12] . For instance, there is great current in-
terest in electrication of transport but there are dierent degrees of
electrication and the need for an ICE is only eliminated by a battery
electric vehicle (BEV). However, the GHG impact of BEVs could be worse
compared to conventional vehicles if the electricity generation and the
energy used in battery manufacture, which depends on battery capac-
ity and can be large, is not suciently decarbonised. In many grow-
ing economies such as China and India coal will play a substantial role
in power generation for decades to come and the carbon intensity of
power generation will be too high for BEVs to be benecial from a GHG
consideration. Indeed, in such countries, even emissions of particulates,
NO
x
and sulfur dioxide (SO
2
) will be worse for BEVs though they will
be displaced away from the vehicle. In several other countries, despite
signicant increases in renewable power generation, the dispatchable
power needed to meet demand during EV charging periods will today
largely be met with generating capacity powered by natural gas. When
this electricity is generated from gas turbines, GHG emissions from BEVs
are comparable to emissions from current ICE-powered vehicles.
In addition, there are very signicant human toxicity implications
associated with the mining of metals needed in batteries [12] which
cannot be ignored as the number of BEVs grows. Large engines in com-
mercial transport and aircraft cannot be run on batteries because of
the size of the batteries required [12] . At the end of 2018, the world
had around 3.3 million BEVs [14] ; in 2019 around 2.3 million BEVs
and plugin hybrid electric vehicles (PHEVs) were sold, of which around
70%,were BEVs, (giving a total of approximately 4.9 million global BEVs
at the end of 2019) [ 15 , 16 ]. More than half of them were sold in China
[ 15 , 16 ] which would actually have worsened the global GHG impact of
transport. Indeed, policy on BEVs in China is being driven by the need to
gain a technological lead in “new energy vehicles ”and concerns about
urban air quality rather than GHG considerations. So by the end of 2019
global sales of BEVs amounted to around 2.5% of global LDV sales and
the global stock of BEVs was just under 0.4% of total LDV numbers.
A hundred-fold increase by 2040 in BEV numbers, compared to 2019,
would bring their number to around 490 million, roughly 29% of global
LDV numbers which would account for around 13% of transport energy
since LDVs use around 44% of transport energy ( Table 1 ). Even under
the unrealistic assumption that all these 490 million BEVs were made
with carbon-neutral energy and run on carbon-free electricity, at best
around 13% of transport-related GHG would be saved by such an enor-
mous increase in BEV numbers. In any case, such a massive increase
in BEV numbers will require very large prior investments in charging
infrastructure, additional electricity generation, nancial incentives to
encourage the sale of BEVs and recycling of batteries [12] which may not
be aordable in most countries. The supply and cost of metals needed,
particularly of cobalt, to support such a large spread of BEVs could also
be very challenging. Similarly, any increase in the share of biofuels, hy-
drogen or other energy sources for transport will face signicant chal-
lenges [12] . Hence most credible projections suggest that even by 2040,
85-90% of transport energy will still come petroleum based liquid fuels
[ 7 , 10 , 17 ].
Thus, for decades to come, global transport will be predominantly
powered by ICEs using petroleum-based fuels. It is imperative that the
eciency and environmental impact of such engines are improved in or-
der to bring about any signicant and realistic improvement in the sus-
tainability of transport. In fact, there is great scope to bring about such
improvements in the short term with better combustion, after-treatment
and control systems and in the medium term with the development of
new fuel/engine systems. The vast existing transport infrastructure can
be used to support such initiatives without requiring too much change
or investment. This paper discusses the scope for such improvements,
particularly for the short and medium term. In the long term, as the
F. Leach, G. Kalghatgi and R. Stone et al. Transportation Engineering 1 (2020) 100005
Table 2
Average global daily demand for transport fuels, third quarter 2018 [11] .
Assuming 32.5 MJ/l for gasoline, 36 MJ/l for diesel and jet fuel, 40 MJ/l for RFO. 1 exa joule = 163.4 million barrels of oil equivalent (BOE)
Total Million barrels of oil equivalent (BOE) Energy, exajoules Fuel Volume, billion litres
Gasoline 26.4 0.162 4.985
Diesel/Gasoil 28.1 0.172 4.778
Jet/Kerosene 8.0 0.049 1.361
Residual Fuel Oil (RFO) 7.0 0.043 1.075
Other
∗ 30.0
Tota l 99.6
∗ Includes naphtha, LPG, ethane, other petrochemicals and energy use in reneries.
overall energy system transforms and the share of renewable energy in-
creases and battery and fuel-cell technology improve, other alternatives
become increasingly practical. However rst we briey consider the ex-
isting engines and fuels and the mechanisms that limit eciency and
aect emissions.
2. Current IC engine combustion systems and fuels
Several books [18-21] discuss engine combustion systems and fuels
and this section contains a brief summary.
2.1. Engine combustion systems
Land and marine transport are powered by two major engine com-
bustion systems.
Most light duty engines run on spark Ignition (SI) engines and use
gasoline fuel. Combustion is initiated in a SI engine by an electric spark
and fuel energy is released as a ame propagates through a mixture of
fuel and air that is compressed after premixing. In modern SI engines
operated at lambda = 1 (stoichiometric), CO, HC and NO
x
emissions
in the exhaust are reduced to acceptably low levels through the use of
a three-way catalyst. The numbers of nanoparticles (less than 100 nm
in diameter) are of increasing concern though particulate mass is very
low in the exhaust of a SI engine. Gasoline particulate lters (GPFs) are
likely to be required in the future to tackle these particulate emissions.
SI engined vehicles on average convert only 20-25% of fuel energy to
motive power mainly because of the requirement of throttling at low
loads and knock [ 18 , 19 ] at high loads –the principles are discussed
in greater detail in Section 3 . However, their impact on pollution and
air quality is low because of the use of eective after-treatment. Larger
SI engines, such as those required for commercial vehicles, also have
to run at low speeds, would be more prone to knock on a given fuel
because there is more time available for chemical reactions leading to
knock and their eciency would be reduced. Hence, SI engines are not
usually used in commercial transport requiring large engines.
In compression ignition (CI) engines, fuel is injected into the high-
pressure and high-temperature environment near the top of the com-
pression stroke and heat release is initiated by autoignition as the fuel
mixes with oxygen. Currently all practical CI engines use diesel fuel
–they are diesel engines. Soot (particulates) and NO
x
emissions are
a signicant problem for diesel engines. Technology such as complex
after-treatment and high-pressure injection systems are needed to con-
trol them. Hence modern diesel engines are much more expensive com-
pared to SI engines of similar size but are more ecient compared to SI
engines.
There has been much recent interest in homogeneous charge com-
pression ignition (HCCI) combustion. In a HCCI engine, fuel and air are
fully premixed as in a SI engine but heat release occurs by autoignition
as in knock in a SI engine. The thermal eciency of HCCI engines is very
high but they are constrained to operate at lean mixture strengths and
hence, low loads, because of excessive pressure rise rates at richer equiv-
alence ratios. Friction losses are proportionately higher at lower loads
and the brake eciency of HCCI engines, at the loads they can oper-
ate, will be lower. HCCI engines are not practical for this reason and
also because they are dicult to control. HCCI-like combustion with
low NO
x
and soot and high eciency but which is easier to control can
be ensured by not fully premixing fuel and air. Such ‘premixed-enough’
combustion systems are discussed in Section 5 .
2.2. IC engine fuels
Gasolines need to have high anti-knock quality as specied by RON
and MON, the Research and Motor Octane Numbers or the Anti-Knock
Index (AKI = (RON + MON)/2)) [ 20 , 22 ]. Most market gasolines have
RON > 90 and a MON value about 10 lower than the RON value. Fuel
requirements of SI engines are discussed in greater detail in [23] . For
diesel fuels, autoignition quality is measured by the Cetane Number,
CN. Diesel fuels generally need to have a high CN because they need
to autoignite easily; practical diesel fuels have CN > 40. The higher the
RON of a fuel, the lower is its CN and vice versa [ 20 , 22 ]. The cetane
number of jet fuel is lower than that of conventional diesel fuel and it
is blended using more volatile components than are found in the diesel
boiling range. Marine transport fuels are blended from the heaviest com-
ponents in the fuel pool and have a high sulfur content. Marine engines
could be forced to run on conventional diesel fuel because of current
moves to reduce the sulfur content of marine fuels, further contributing
to the increase in demand for conventional diesel fuel in the future.
Gasoline-like fuels are dened in this paper as fuels with CN < 30 or
RON > 60, i.e., in the gasoline autoignition range as in [22] .
The rst step in the manufacture of practical transport fuels is the dis-
tillation of crude oil. Gases dissolved in the crude oil are released when
oil is heated above ambient temperature and make up Liquid Petroleum
Gas (LPG). Up to 2% of the crude could be LPG which consists mostly
of propane and some butane. The fraction in the gasoline boiling range,
with boiling points between ~20 °C and ~200 °C, from the initial dis-
tillation is known as Straight Run Gasoline (SRG). Diesel fuels are made
up of heavier components with boiling points in the range of ~160 °C
to ~380 °C. Heavy components, with boiling points higher than 380°C,
could constitute 40%–60% of the weight of petroleum depending on
the source of the crude oil. In the renery, these heavy components are
rst “cracked ” into smaller molecules which are further processed to
produce useful products e.g., by reducing sulfur or by changing their
octane/cetane number. The products in the gasoline boiling range from
dierent parts of the renery are collectively known under the generic
term “naphtha ” which is usually processed further to increase its oc-
tane number; it is also used in the petrochemicals industry. Other non-
petroleum components such as biofuels and high-octane components
like methyl tertiary butyl ether (MTBE) are blended with renery com-
ponents along with some fuel additives to meet the required fuel speci-
cations [ 20 , 21 ].
3. Engine and vehicle efficiency – main principles
3.1. Introduction
It has long been known that high compression ratios and weak, or
fuel-lean, mixtures improve the eciency of SI engines, but since 1980
SI engines have invariably been constrained to operate with stoichio-
F. Leach, G. Kalghatgi and R. Stone et al. Transportation Engineering 1 (2020) 100005
Fig. 3.1. Plot of eciency (Eq. 3.1) and the temperature rise ratio ( 𝜃) during
heat addition, with a temperature rise from combustion of 2000 K, the ratio of
heat capacities of 1.4, and an initial temperature of 300 K –the solid lines are
for an expansion ratio of 10, the dashed for an expansion ratio of 14.
metric mixtures so that three-way catalysts can virtually eliminate un-
wanted emissions. With weak mixtures the catalytic oxidation of carbon
monoxide (CO) and unburned hydrocarbons (UHC) is straightforward;
the challenge is the elimination of nitrogen oxides (NO
x
, a mixture of
NO and NO
2
) as a chemically-reducing environment is needed. If the
mixture is suciently weak, then combustion temperatures will be too
low to form NO
x
, but to meet current emissions regulations these mix-
tures would not necessarily be ammable in a conventional engine. In
the late 1990s lean-NO
x
traps (LNT) were introduced and these use ma-
terials such as barium carbonate to react with the NO
x
. However, regen-
eration is needed to convert the barium nitrate back to barium carbonate
during rich mixture excursions. In any case the trapping eciency was
only about 50%, and as emissions legislation became more demanding
these technologies were no longer adequate and lean burn operation
had to be again abandoned. The challenge of reducing NO
x
emissions
in an overall weak mixture is also applicable to diesel engines and this
has led to Selective Catalytic Reduction (SCR) that uses an aqueous urea
solution (32.5% by mass of C(NH
2
)
2
O) to generate ammonia (NH
3
) that
reduces about 90% of the NO
x
. Such systems have become widely used
in diesel engines since 2010 and could be readily applied to SI engines.
More information on these technologies and their development can be
found in texts such as Heywood [18] and Stone [19] , with some of the
more recent trends discussed later in Sections 4.1 and 4.2
It is also well known that combustion needs to be rapid (so as to em-
ulate the instantaneous heat addition of the so-called Otto cycle analy-
sis) and repeatable (if there are cycle-by-cycle variations in combustion,
then ignition can only be optimised for the mean cycles). The key loss
that needs to be eliminated is from throttling, since this means of re-
ducing the air ow increases the pumping work loss. Options for this
include cylinder de-activation, control of the piston stroke (as in the
original Atkinson engine) or closing the inlet valve mid-stroke of the pis-
ton (as originally proposed by Miller for limiting the maximum cylinder
pressure in diesel engines). Whether inlet valve closure is early or late
the theoretical performance is the same and has the result of reducing
the eective compression ratio. However, an air standard cycle model
[19] can be used to show that the expansion ratio ( r
e
) is much more
important than the compression ratio ( r
c
):
𝜂= 1 −
𝜃𝑟
c
+ 𝑟
γ
c
− 𝑟
γ
e
+ γ( 𝑟
e
− 𝑟
c
) ×𝑟
γ−1
e
𝜃𝑟
c
𝑟
γ−1
e
(3.1)
where: 𝛾is the ratio of heat capacities and 𝜃is the ratio of temperature
rise during heat addition to the temperature at the end of compression.
Fig. 3.2. Fuel-Air Cycle eciency as a function of compression ratio and mix-
ture strength - solid lines, Tizard and Pye [24] ; broken lines, Taylor [25] (the
Taylor data does not include the eects of dissociation so overestimates the ef-
ciency of the stoichiometric mixture).
The signicance of Eq. 3.1 is shown in Fig. 3.1 where it has been as-
sumed that the temperature rise associated with combustion is 2000 K
and the ratio of heat capacities is held at 1.4 with an initial temper-
ature of 300 K. If the compression ratio is half of the expansion ratio
then the loss of eciency will only be a couple of per cent (noting that
the air standard cycle analysis predicts almost double that which might
be achieved) and this compares with a reduction in pumping mean ef-
fective pressure (PMEP) of order 0.5 bar that might represent almost
10% of the output of the engine. Air standard cycle analyses ignore the
real thermodynamic performance of the air fuel mixture and combus-
tion products and dissociation. The importance of these is discussed in
Section 3.2 .
A simpler way of reducing the pumping loss that can be applied with
three-way catalysts is to employ Exhaust Gas Recirculation. Up to 25%
is viable (and it makes little dierence whether this is dened on a vol-
umetric, gravimetric or molar basis), and this could lead to a reduction
in PMEP of about 0.25 bar. However, EGR will slow combustion and in-
crease the level of cycle-by-cycle variations in combustion, so mitigation
by increasing in-cylinder motion is important.
Another strategy to reduce part load losses is to ‘downsize’ the en-
gine; this means supercharging the engine (either directly or with an
exhaust powered turbine) so that a reference operating torque on the
road-load curve corresponds to the engine operating at a higher engine
load than it would with the intake at ambient pressure. However, this
is almost certainly associated with a reduction in compression ratio so
as to avoid combustion knock at full load and this imposes an eciency
penalty across the whole operating envelope.
It is of course much better if part load operation is eliminated and
this is possible with hybrid engine vehicles (most simply but not e-
ciently with a series hybrid) or to a large extent with shunt transmission
systems, and these are discussed in Section 3.4 .
3.2. Real thermodynamic behaviour of combusting mixtures
The air standard Otto cycle analysis uses the properties of air at am-
bient conditions and an idealised heat input. Tizard and Pye [24] were
the rst to consider the real heat capacity of the reactants and products
and the signicance of dissociation. Fig. 3.2 shows their computation
of Fuel-Air cycle eciency for weak mixtures, including the eects of
dissociation. Also included in Fig. 3.2 are the results of the Fuel-Air cy-
cle calculations reported by Taylor [25] , with perhaps not surprisingly
F. Leach, G. Kalghatgi and R. Stone et al. Transportation Engineering 1 (2020) 100005
Fig. 3.3. Experimental measurements of a natural gas fuelled engine with a fast
burn combustion system operating at 1500 rpm and a 13:1 compression ratio
with MBT ignition timing [27] . The exhaust chemical energy is expressed as a
percentage of the input fuel energy.
signicant dierences, not least since the Taylor data have not included
the eects of dissociation. Pye [26] subsequently noted that in the 1920s
there was very limited data on the high temperature heat capacity of
gases and equilibrium constants, but even in the 1930s [26] the val-
ues of the equilibrium constants do not agree well with current values.
None the less, this early work explained why the observed indicated ef-
ciencies were much lower than predicted by the Otto cycle analysis
and why maximum power occurs rich of stoichiometric, and why weak
mixtures give a higher indicated eciency. Another way of quantifying
the impact of dissociation is to calculate the caloric value of the car-
bon monoxide and hydrogen in the exhaust of an engine operating with
stoichiometric combustion –this energy content amounts to 3-4% of the
fuel caloric value.
So, weakening the mixture increases the ideal cycle eciency, but
as Fig. 3.3 shows there are competing factors that will reduce the e-
ciency, namely an increase in the level of cycle-by-cycle variations in
combustion, an increase in the emissions of partially burned fuel, and
an increase in the duration of the main combustion period. In a natu-
rally aspirated engine operating at full load there would be a reduction
in the mechanical eciency as the mixture is weakened since the lower
output makes the mechanical losses more signicant. However, if the
engine is supercharged (perhaps driven by an exhaust turbine) then the
output can be maintained, and to a rst order the boost pressure ratio
will need to equal the lambda (for example a boost pressure ratio of 1.4
for a lambda of 1.4), as will be seen later in Fig. 3.4 .
As soon as lambda 1.1 is reached there will be negligible carbon
monoxide in the exhaust and the eciency increases until lambda
reaches 1.4, just after which there is a rapid rise in the hydrocarbon
emissions and cycle-by-cycle variations in combustion. For the very
weak mixtures (lambda above 1.5) ame propagation is slower, but
more signicantly the ame is more readily quenched. The increase in
burn duration is not signicant so long as ignition timing is optimised;
a 10–90% MFB duration of 20° crank angle (ca) amounts to a 1%-point
loss of eciency, and even if this is doubled to 40°ca the loss of e-
ciency is only 3%-points. The valves do not operate instantaneously at
Fig. 3.4. The eect of dilution by air (solid line) and EGR (broken line) on
the cycle eciency and boost requirement for a xed load. As with the fuel-air
cycle eciency calculations the assumptions are for adiabatic and reversible
compression and expansion, with instantaneous heat input at top dead centre.
The Taylor [25] prediction for eciency (dotted line) does not include the eect
of dissociation.
the end of the stroke, and this means that the eective compression and
expansion ratio will be around 11:1 with about a 2%-point loss of e-
ciency. At a lambda of 1.4 the Fuel-Air cycle would imply an eciency
of 0.53 compared with the experimental value of 0.43, so the biggest loss
that has not been plotted in Fig 3.3 is of course heat transfer that will
account for about a 7%-point loss of eciency. Heat transfer has been
reduced in diesel engines by thermal swing coatings [ 28 , 29 ], but with
SI engines if there is less cooling then there is an increased likelihood
of combustion knock. However, auto-ignition can be exploited in Gaso-
line Compression Ignition (GCI) engines, as discussed in Section 5.2 . It
is also pertinent to compare dilution with air with dilution by EGR and
some results from modelling are presented in Fig. 3.4 , and like the data
in Fig. 3.3 this is for a compression ratio of 13:1 and a xed engine out-
put. The data in Fig. 3.4 is from a phenomenological model [30] that
has been constrained to operate adiabatically with instantaneous com-
bustion so as to replicate fuel-air cycle modelling.
As expected, Fig. 3.4 shows that dilution by air increases the cycle
eciency, and the initial increase is rapid since the impact of dissocia-
tion is reduced, and for weaker mixtures the ratio of the heat capacities
is increased. The increase in eciency with EGR is perhaps unexpected
in this context –it is unlike part load operation when the EGR will re-
duce the throttling loss. The eciency has increased with EGR dilution
because the lower cycle temperatures lead to lower levels of dissociation
and there is small increase in the ratio of the heat capacities. The boost
requirement for the xed power density with dilution rises less rapidly
than the dilution factor because of the rise in eciency. These ideal cy-
cle analyses have been for adiabatic compression and expansion, and
it might be thought that the lower combustion temperature will reduce
heat transfer. This is of course true, but to maintain a xed power output
the in-cylinder pressures will be higher so the heat transfer coecients
will increase; so the lower combustion temperatures will not necessarily
improve the overall eciency.
Evaporative cooling of the fuel during the induction in Gasoline Di-
rect Injection (GDI) engines allows a higher compression ratio to be used
(an increase of about 2 for a given quality fuel). GDI engines are also
more amenable to boosting (supercharging or turbocharging) since fuel
F. Leach, G. Kalghatgi and R. Stone et al. Transportation Engineering 1 (2020) 100005
Fig. 3.5. Eciency contours of Toyota 2.5 L Atkinson cycle engine when tested
with an EPA Tier 2 fuel [32] .
need only be injected after exhaust valve closure, so as to avoid short-
circuiting losses. Boosted engines require a reduced compression ratio,
but do enable the steady-state road load at cruising speeds to be at a
part load condition that is un-throttled. Engine optimisation for vehicles
cannot be separated from the transmission system, and this is discussed
further in Section 3.4 .
3.3. Performance of high efficiency, current market engines
3.3.1. Stoichiometric spark ignition engines
There are two current macrotrends in the development of conven-
tional, high-eciency SI engines –adoption of larger, naturally aspi-
rated Atkinson or Miller cycle engines ( r
e
> r
c
) and smaller, lower com-
pression ratio downsized-boosted engines. As explained previously, the
former seeks to enhance eciency using valve timing to reduce pump-
ing losses and knocking propensity while maintaining a large r
e
, while
the latter seeks to reduce pumping losses via turbocharging and to re-
duce the relative importance of friction and pumping losses by virtue of
having a smaller engine that operates at higher loads.
An example of the Atkinson/Miller pathway is Toyota’s spark igni-
tion engine with a 40% brake thermal eciency [31] ; it is a naturally
aspirated 2.5 L in-line four cylinder engine with a maximum power out-
put of 150 kW that uses 91 RON fuel. It uses a long stroke to improve
the volume to surface area ratio and a high turbulence intensity (allow-
ing increased levels of EGR for a given level of combustion stability). As
a long stroke increases the frictional loss (at a given engine speed) and
reduces the volumetric eciency there is a trade-o, and this engine
has a bore of 1.18 times the stroke. The frictional losses are minimised
by reducing the mass of the reciprocating components, using a variable
displacement oil pump, a low viscosity oil (SAE 0W-16), and careful
control of coolant temperature with an electrically driven pump. To op-
timise the part load fuel economy then an Atkinson cycle is used with
a compression ratio reduced to 6.6:1 (from 13:1) to reduce the trapped
volume without so much need for throttling.
A very comprehensive benchmarking of the engine [32] , provides
enormous detail on the engine operating regimes. The engine operates
with stoichiometric mixtures for most of its envelope and the peak e-
ciency (40%) occurs with 23% cooled EGR and a brake mean eective
pressure (bmep) of just over 9 bar at a speed of 2800 rpm; the mixture is
only enriched for the highest speed and load conditions. Fig. 3.5 shows
the eciency contour map when the engine was independently tested
with an EPA Tier 2 fuel (AKI of 87).
Fig. 3.6. Eciency contours of Honda 1.5 L turbocharged engine when tested
with an EPA Tier 2 fuel.
As well as reporting the brake eciency there are contour plots of:
inlet valve opening and closure, exhaust lambda, valve overlap, EGR
level and ignition timing [32] .
A good example of the downsized-boosted pathway is provided by
Honda’s recent 1.5L turbocharged direct-injection engine [33] . Like the
Toyota engine, the Honda engine features a long-stroke design and var-
ious eciency enhancing features such as sodium-lled exhaust valves
(for knock mitigation) and dual cam phasing (for optimizing EGR and
pumping losses).The corresponding brake eciency contours, measured
by the U.S. EPA, are shown in Fig. 3.6 . Notice that at lower torque levels
(below, say, 100 N-m) the eciency of the Honda engine exceeds that
of the Toyota –even though the peak eciency of the Toyota engine is
higher (this observation still holds true, though to a lesser extent, when
the engines are scaled to provide the same peak power). At higher torque
levels and low-to-moderate speeds, the Toyota has superior eciency.
Thus, we anticipate that the best engine with respect to fuel economy
will be dependent on the relative frequency of high- and low-load driv-
ing, as well as the specic powertrain system the engine is paired with.
Other factors will also impact the choice of engine technology, including
low-speed torque, mass and size, and noise and vibration.
A logical further step toward improving engine eciency would be
to combine the naturally aspirated Atkinson/Miller pathway with the
turbocharged pathway [34] , as exemplied by the Audi TSFI and the
VW EA211 engine series. These engines have demonstrated signicant
fuel economy benets over the previous turbocharged versions –in the
case of the TSFI the fuel consumption map was signicantly improved
over an earlier downsized boosted engine with a smaller displacement
[35] .
Although the engines described above are technology rich, there are
additional technologies that can be applied to signicantly enhance e-
ciency while maintaining stoichiometric operation – recent reviews can
be found in [36-38] . One example, which has been deployed in numer-
ous production engines since 1981 [39] and achieved a market penetra-
tion of approximately 20% of new light-duty trucks sold in the U.S., is
cylinder deactivation. Deactivation has also been successfully applied to
small three- and four-cylinder vehicles, including the VW engine men-
tioned above. By deactivating the valves of one or more cylinders at
lower loads, it is possible to simultaneously reduce both engine friction
as well as pumping and heat losses. In larger engines, the fuel consump-
tion benet can exceed 15% [40] , while in the highly ecient engines
discussed above the benet is estimated to range between 2 and 3% [ 32 ,
41 ] Cylinder deactivation can also be used to help maintain catalyst tem-
perature during periods of prolonged low load operation, making this
technology particularly synergistic with lean combustion systems.
F. Leach, G. Kalghatgi and R. Stone et al. Transportation Engineering 1 (2020) 100005
3.3.2. Lean and dilute SI engines
Section 3.2 demonstrated the desirability of operating with lean or
dilute intake charge to achieve high eciency. Current engine technolo-
gies all use EGR-dilution, as discussed above, to reduce pumping losses
at lower loads and to mitigate knock at higher loads. Consequently, a
great deal of engineering eort is expended to expand EGR tolerance
though either improved in-cylinder ow motion [ 42 , 43 ] or with ad-
vanced ignition systems [44-47] . The overwhelming advantage of EGR-
diluted combustion systems is that they allow use of inexpensive, highly
eective exhaust gas after-treatment – described in more detail below.
Lean-burn engines, on the other hand, oer potentially higher ecien-
cies ( cf Fig. 3.4 ), but require more expensive NO
x
after-treatment. Under
the current emissions regulations, this has limited their application to
premium vehicles.
Recently, however, Mazda has introduced a lean-burn engine into
the European market that does not require expensive NO
x
after-
treatment. The engine features a high compression ratio of 16.3, which
enables very lean low-load combustion using a weakly stratied, spark-
assisted compression-ignition combustion strategy. Here, a spark initi-
ates a