Combustion and Flame 160 (2013) 1–8

Contents lists available at SciVerse ScienceDirect

Combustion and Flame

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c o m b u s t fl a m e

Opinion Article

Directions in internal combustion engine research

Rolf D. Reitz
Engine Research Center, University of Wisconsin-Madison, USA

1. Introduction world’s annual production of 37 billion tons of CO2 contributes

to Green House Gases (GHG), which some fear could lead to cli-
Why do we still need research on Internal Combustion (IC) en- mate change with unpredictable consequences. Drastic reductions
gines? Society continues to rely heavily on IC engines for transpor- in fuel usage would be required to make appreciable changes in
tation, commerce, and power generation. That is why engine GHG trends. For example, the International Energy Agency road-
research has enjoyed a more than century-long history, and engine map is to reduce fuel use per kilometer by 30–50% in new road
research is still vibrant today. Reciprocating Internal Combustion vehicles worldwide by 2030, and from all vehicles by 2050. The
(IC) engines range in size from model airplane engines that fit in goal is to limit the global average temperature rise which some cli-
the hand, to gigantic marine engines as tall as a four story building. matologists project for 2050 (the 2DS scenario) [1]. Government
IC engines power all manner of utility devices (e.g., pumps, mow- mandates requiring such reductions impose a tall order on the al-
ers, chain-saws, portable generators, etc.), earth-moving equip- ready high-tech and high-cost automotive industry. And, although
ment, tractors, propeller aircraft, ocean liners and ships, personal 2050 seems a long time in the future, the time required to bring
watercraft and motorcycles, plus the 750 million passenger cars new engines to production, together with the years needed for
on the world’s roads. And, if you just look at the U.S., there are new technology to permeate the overall vehicle fleet, means that
250 million vehicles – that’s roughly one per person. Sixty million major effort (and investment) is needed in IC engine research over
cars will be made in the world in 2012, which is a 50% increase the next decades.
over a decade ago. Part of this increase is due to the fact that China So, why do we still use gasoline and diesel fuels? They are a
became the world’s largest car market in 2011 and is now produc- compact and portable form of energy, providing over 40 MJ per kg
ing a quarter of the world’s cars. A third of all cars produced are of fuel. Considering that the kinetic energy of a thousand kg auto-
made in the European Union, and half are powered by diesel en- mobile traveling at 60 mph is about the amount of energy in only a
gines, partly due to their higher fuel efficiency. So, engine research 1/3 of an ounce, or a teaspoon of gasoline or diesel fuel, the appeal
encompasses both gasoline and diesel engines, and even small is obvious. The high energy density of liquid petroleum fuel comes
improvements in their efficiency have a major impact on economy from breaking and rearranging its molecular bonds, producing CO2
and pollution. and H2O from large molecular weight hydrocarbons. However, we
The huge number of vehicles and engines on the planet burns a do not actually harness all of this energy. Getting the fuel from the
vast quantity of fuel. Indeed, 70% of the 86 million barrels of crude oil well to the gas station requires about a 20% equivalent amount
oil that we consume every day is used in IC engines. And, in the of energy to that contained in the fuel. Once the fuel enters the
U.S., 10 million barrels are used per day in cars and light-duty vehicle’s gas tank the typical Spark-Ignition (SI) automotive engine
trucks, plus 4 million in heavy-duty vehicles, which works out to delivers about 18% of this energy to its drive train. Engine and ex-
about 2.5 gallons per person every day. Since there are insufficient haust losses account for about 62% of the loss, and about 20% is
fuel reserves to meet this consumption, 62% is imported and, at consumed in running accessories and stand-by-idle. More losses
current prices (e.g., $80 a barrel), the U.S. spends about 1 billion occur in the driveline, so that only about 12% of the fuel tank en-
dollars a day on imported oil. This cost is certain to increase as ergy actually makes it to the wheels.
more-and-more economic development drives increased demand In view of the well-documented inefficiencies of the IC engine,
for automotive fuels world-wide. there is perennial discussion about displacing it with some more
Associated with this massive oil use is the accompanying emis- energy-efficient power-plant. However, no such power-plant has
sion of pollutants, including nitric oxides (NOx) and particulates emerged in spite of billions of dollars and decades of research since
(soot), as well as the production of CO2. Pollutant emissions have the invention of the automobile more than a century ago. New
serious environmental and health implications, and thus most gov- ideas are regularly proposed, but cost, complexity and overlooked
ernments have imposed stringent vehicle emissions regulations real-world design shortcomings usually restrict their implementa-
that are continually being tightened further. In addition, the tion. Examples that have not seen much success include Stirling,
rotary, solar and hydrogen power-plants, and recently fuel cell
powered vehicles. Electric hybrid or battery powered vehicles have
E-mail address: [email protected] benefits in some light-duty applications, but their advantages

0010-2180/$ - see front matter Ó 2012 Published by Elsevier Inc. on behalf of The Combustion Institute.
http://dx.doi.org/10.1016/j.combustflame.2012.11.002
2 R.D. Reitz / Combustion and Flame 160 (2013) 1–8

depend heavily on the duty cycle, favoring city more than long dis- engine studies. Much recent research at the ERC and elsewhere
tance driving. Also, it is important to note that the thermal effi- has focused on improving the understanding of ignition, which is
ciency of electricity generation at power stations (to charge plug- highly dependent on the chemical composition of the fuel, and in
in hybrid and battery vehicles) is generally less than 50%. This lim- increasing the quality of the air–fuel mixture for improved com-
its the overall energy efficiency severely from the very outset. In bustion efficiency in both SI and CI engines. Research progress
addition, further losses need to be considered, such as those from has been aided by the advent of new developments in laser-based
mining and transporting the fuel to the power station (e.g., coal/ diagnostics that permit characterization of in-cylinder flows, com-
natural gas), distributing the power to the end user, and energy bustion temperatures and species concentrations in unprece-
storage losses – plus those accessory and driveline losses on the dented detail, together with the development of accurate
vehicle itself. Computational Fluid Dynamics (CFD) models. The latter has been
With this in mind, the relative low cost, high durability and con- facilitated by the more than 10,000-fold increase in computer
venience of the IC engine, coupled with its ability to use high-en- speeds over the past 15 years.
ergy–density liquid fuels, makes it difficult to make the case to Indeed, current multi-dimensional CFD tools have become suf-
displace it. Indeed, a recent NRC report [2] concluded: ficiently mature to guide the design and development of more effi-
cient and cleaner IC engines [3]. Modeling is aiding the
There seems to be little doubt that, regardless of the success of
development of new Low Temperature Combustion (LTC) concepts
any pathways discussed, the internal combustion engine (ICE)
that provide previously unimaginable efficiencies. LTC is also de-
will be the dominant prime mover for light-duty vehicles for
sired because low combustion temperatures result in NOx reduc-
many years, probably decades. Thus, it is clearly important to
tion due to the high activation energy of NO formation reactions.
perform R&D to provide a better understanding of the funda-
LTC strategies include Homogeneous Charge Compression Ignition
mental processes affecting engine efficiency and the production
(HCCI), Premixed Charge Compression Ignition (PCCI) and Reactiv-
of undesirable emissions.
ity Controlled Compression Ignition (RCCI). These and other ad-
vanced combustion technologies are being pursued by all major
In addition, there is no obvious alternative to the IC engine for engine companies, by the U.S. Department of Energy [4] and by
the medium- and heavy-duty commercial vehicle markets, and Government laboratories.
these applications account for a quarter of all fuel used (mostly
diesel). However, it is also clear that the sustainability of the 2.1. Engine modeling
world’s appetite for fossil oil is not justifiable in the longer term,
and this drives the need for more research. Absent the arrival of Instead of describing engine processes empirically, as in classi-
some magic bullet power-plant, the world will continue to rely al- cal thermodynamically-based analyses, comprehensive computer
most entirely on the IC engine which, by the way, can be trans- codes are now heavily used in engine research. The governing Na-
formed to run on non-fossil fuel sources, such as renewable bio- vier Stokes partial differential conservation equations are solved on
fuels. a numerical mesh that follows the time-varying engine geometry
during each engine cycle. Submodels are still required to describe
2. Opportunities and challenges in engine research processes that occur on time and length scales too small to be re-
solved in the simulations. But, these models have universality,
As discussed above, there is great incentive to reduce petroleum since they refer to ‘‘micro-scale’’ processes that are common to
dependence and to make the best use of the world’s limited fossil general flows. For example, most wall boundary layers are too thin
fuel resources by improving IC engine efficiency through research. to be resolved on a practical engine CFD mesh, but adequate wall
It is also important to seek and explore new engine concepts that heat transfer predictions can be obtained with ‘‘law-of-the-wall-
use alternative fuels, such as renewable bio-, or natural gas- and based’’ submodels that use local information about the mean flow
coal-derived fuels, so as to reduce the monopoly of fossil oil. Of in the computational cells at the combustion chamber walls. Sub-
course, these engine performance enhancements must be made model development often relies on experiments conducted in
while meeting stringent pollutant emissions mandates. The engine more controlled environments than engines, such as in spray
must perform reliably over wide ranges of engine speed and load, chambers, steady pipe flows or shock tubes. Direct Numerical Sim-
and must start and operate under all climates and terrestrial ulation (DNS) is too computer time intensive for practical engine
elevations. simulation, but can provide insights about micro-scale processes,
After over a century of intensive development, it is remarkable such as turbulence-chemistry interactions in flames.
that significant improvements to IC engine efficiency are still being
made. This reflects the richness of the combustion process, which 2.2. Engine experiments
still presents opportunities for new discoveries. Engine combustion
can be characterized as a low Mach number, compressible, multi- CFD model development benefits greatly from closely coordi-
phase, high-Reynolds number turbulent flow with chemical reac- nated companion experiments. Engine experiments are difficult
tions and heat transfer, confined in a time-varying geometry. The to perform in the hostile engine environment and they are compli-
combustion process spans multiple regimes that include turbulent cated by phenomena such as cycle-by-cycle variability. However,
flame propagation, mixing-controlled burning, and chemical- they are critical to provide insights into processes that can lead
kinetics-controlled processes, plus their combinations. Describing to engine performance improvements. Diagnostics such as Planar
these complex interacting processes certainly classifies under- Laser Induced Fluorescence or Incandescence (PLIF and LII), chemi-
standing engine physics as a ‘‘grand challenge’’ problem. Indeed, luminescence (imaging and emission spectra) and soot thermal
Prof. Phillip S. Myers, who founded UW-Madison’s Engine Research emission provide detailed information about in-cylinder spatial
Center in 1946, often observed that the IC engine is also the perfect and temporal distributions of important species, including com-
teaching tool, since it relies on all areas of Mechanical Engineering, bustion intermediates, radicals and particulates. Details of the flow
with the possible exception of nuclear reaction! field can be obtained using Particle Image Velocimetry (PIV) or
Engine research activities at the ERC include premixed and par- Phase Doppler Anemometry (PDA) techniques for spray drop size
tially premixed Spark-Ignition (SI) combustion, Conventional Die- and velocity analysis. Of course, use of these techniques requires
sel Combustion (CDC) and advanced Compression Ignition (CI) optical access to the combustion chamber, and this can necessitate
 R.D. Reitz / Combustion and Flame 160 (2013) 1–8 3

unavoidable compromises in engine geometry and/or operating scale reaction. In this case, the sub-grid-scale turbulence-chemis-
conditions. However, the impact of such simplifications can also try interactions within the resolved thickened flame or flame brush
be assessed with CFD modeling. are modeled as a blended timescale between the two limits of large
High-resolution endoscopic imaging of combustion is less intru- scale mixing and volumetric (chemical) heat release.
sive and also provides valuable information. For example, in SI en- Flamelet models solve transformed reaction–diffusion equa-
gines the extent and causes of cyclic variability due to spark kernel tions based on a progress variable or non-reacting scalar, where
growth and knock processes can be studied. In Gasoline Direct the diffusion term is assumed to scale with the large-scale turbu-
Injection (GDI) and diesel engines, details of the penetration and lence. PDF methods use assumed or computed weighting strategies
mixing of injected fuel sprays and their possible impingement on to sample kinetic rates within the flame brush that lie between the
chamber walls can be visualized. The trend toward high-speed burned and unburned gas temperatures. In our opinion the most
(>10 kHz frame rate) imaging diagnostics (enabled by commer- practical modeling approach is the level set method (e.g., G-equa-
cially available high-speed cameras and high-repetition-rate la- tion) which tracks the flame surface evolution. However, this mod-
sers) allows key processes to be followed in time over hundreds el requires that the turbulent flame speed be specified, a priori.
of consecutive engine cycles. This is especially valuable for charac- This is done using semi-empirical correlations that are functions
terizing cyclic variation and rare stochastic events that limit engine of the local turbulent kinetic energy (i.e., rpm) and the mixture
performance. and thermodynamic conditions.
Out-of-engine experiments also provide valuable information An important SI engine research area focuses on the initial spark
for understanding engine processes. For example, shock tube and kernel growth, since this has a major effect on the combustion tim-
rapid compression machine experiments provide needed data ing. In addition to the details of the spark ionization and gas break-
about the ignition characteristics of fuels. Constant volume com- down processes, kernel growth is strongly influenced by the
bustion vessel experiments provide information about spray turbulent flow conditions at the spark plug, which in turn, reflect
vaporization, fuel–air mixing, spray ignition, flame lift-off and fuel complex flow behaviors that that can vary from engine cycle to cy-
effects on particulate (soot) formation and oxidation. Flame speed cle. Some of these flow variations can be traced to flow features
measurements (both laminar and turbulent) are helpful for the generated during the intake process, such as vortex shedding and
development of flame propagation models. However, it is impor- flow separation from intake valves, and even to transient wave
tant that the experiments cover the regimes of temperature, equiv- dynamics in the intake runners. The ignition process occurs below
alence ratio (U), turbulence, pressure and composition (including the scale of resolution of the numerical mesh in practical simula-
exhaust gas recirculation (EGR)) of interest in engines. tions, thus Lagrangian particles are often used to track the develop-
ing flame until it reaches the scale of resolution of the mesh, as
depicted in Fig. 1.
3. SI engines An additional long standing SI engine research topic concerns
engine knock and methods to mitigate it. Knocking combustion is
Arguably, the SI engine features the most complex combustion an abnormal combustion phenomenon which produces loud noise
process of all IC engines. The in-cylinder turbulent combustion is and rapid cylinder pressure rise and can result in engine damage.
driven by complicated aero-thermo-chemical interactions that oc- For downsized and boosted SI engines, knock represents a strin-
cur on vastly different time- and length-scales (i.e., scales ranging gent constraint since it prohibits the use of more advanced spark
from the molecular to the device level). In fact, in spite of decades timings, higher compression ratios and/or higher boost pressures.
of research, the basic fact that SI engines can operate over a speed The spark timing may need to be substantially retarded relative
range from hundreds up to tens of thousands of revolutions per to the Maximum Brake Torque (MBT) timing to avoid pre-ignition,
minute (rpm) can still not be explained satisfactorily. A simple thus preventing maximum thermal efficiency from being achieved.
Mallard–LeChatelier analysis shows that, for the flame speed to Downsize/boosting is a major development trend for SI engines,
scale with rpm in a reaction–diffusion system (as required to ex- and therefore the ability to predict knock is becoming more-and-
plain the speed trend) both the diffusion and reaction terms must more critical. It is generally accepted that engine knock is caused
scale proportionally with rpm. However, the burning rate is known by the volumetric auto-ignition of a portion of the end-gas ahead
to be influenced by fuel effects, and mechanisms that explain the of the propagating flame (also depicted in Fig. 1). The extremely ra-
flame’s dual sensitivity to turbulence and chemistry are largely un- pid heat release associated with end-gas auto-ignition causes
known, with the result that SI engine modeling is still largely sharp increases in local pressure and the propagation of large
empirical. amplitude acoustic pressure waves across the combustion cham-
In an SI engine the combustion duration, which is determined ber. The temperature and pressure histories of the end-gas are, in
by the flame speed, is strongly dependent on turbulence. This is turn, governed by the phasing and rate of development of the prop-
exploited in high performance (e.g., racing) engines by designing agating turbulent flame, thus accurate modeling of end-gas auto-
the combustion chamber to generate high turbulence. However, ignition must be coupled with accurate turbulent flame propaga-
the consequence is that in-cylinder flow velocities are also in- tion models. Multidimensional modeling based on detailed chem-
creased, and this leads to a trade-off between fast combustion for ical reaction kinetics is required for accurate knock prediction,
improved work extraction and higher convective wall heat losses. since the assumption of simpler kinetics (e.g., single-step global
The high heat losses also partly explain the relatively poor thermal chemistry) ignores important auto-ignition characteristics of
efficiency of the SI engine. hydrocarbon fuels, such as cylinder pressure effects and two stage
Combustion models must account for the dynamic balance be- ignition.
tween large-scale mixing, small-scale or sub-grid-scale turbu- The unburned mixture in the end-gas is constantly compressed
lence-chemistry interactions and chemical kinetics processes, as by the propagating turbulent flame, and it ignites under a local vol-
depicted schematically in Fig. 1. Popular models are classified as umetric or HCCI-like condition in normal knock. However, in di-
turbulent mixing-controlled, flamelet or PDF approaches. The sim- rect-injected, boosted, downsized engines operating at low speed
plest mixing-type model introduces a characteristic timescale to and high load, a poorly understood phenomenon known as ‘‘super-
describe large scale mixing or diffusion (i.e., the resolved turbulent knock’’ can occur. Here, one out of every few ten to a hundred
transport on the grid), together with a combination of mixing and thousand engine cycles suddenly exhibits a random large ampli-
kinetic (Arrhenius) species conversion rates for modeling sub-grid- tude knocking event that can, in extreme cases, catastrophically
4 R.D. Reitz / Combustion and Flame 160 (2013) 1–8

Fig. 1. Spark ignition engine processes.

destroy an engine. Superknock has variously been attributed to the Network collaboration [5]. The ECN includes round-robin testing
transport and accumulation of oil droplets from the ring-pack or of standardized injectors and compares results from various exper-
past the valves, which can serve as ignition sources, or to the imental and modeling approaches and groups, internationally.
build-up and break-off of combustion chamber deposits or other New insights about diesel spray behavior are being obtained from
‘‘hot spots’’. Methods to eliminate superknock or Low Speed Pre- these and other out-of-engine experiments. Novel techniques, such
Ignition (LSPI) are under intense investigation. These include im- as phase-contrast X-ray radiography and ballistic imaging are pro-
proved spray targeting and the use of mild EGR. The highly inter- viding insights about breakup mechanisms and nozzle cavitation
mittent and random character of LSPI presents an especially and fuel effects.
severe challenge to both experimental diagnostics and predictive Optical engine experiments allow spray penetration and target-
modeling. ing effects on UHC, CO, NOx and soot emissions to be characterized,
In stoichiometric premixed charge SI engines, Unburned Hydro- and provide data for CFD model validation. For example, diesel
carbons (UHC), CO and NOx emissions are effectively controlled combustion simulations are compared with experimental OH PLIF
with the 3-way catalyst, which was a breakthrough technology measurements in Fig. 2 for three different diesel combustion
developed in the 1960s and 1970s. Over 99% of regulated emis-
sions have been removed from light-duty vehicle exhaust since
1970. Since these engines generally operate at very high burned
gas temperatures and also avoid fuel rich equivalence ratios, soot
emissions are also reduced. However, the 3-way catalyst is not
effective for lean burn GDI engines when operated in a partially-
premixed or stratified mode where NOx reduction is not feasible
due to the presence of oxygen in the exhaust gas. In addition, soot
formation increases due the presence of locally rich regions (as in
diesel engines), and methods for soot characterization and mitiga-
tion are the subjects of much current research also in gasoline-
fueled SI engines. New and impending regulations mandate not
only allowable exhaust soot mass, but also impose soot particle
number density limits. Thus, models to predict particulates in SI
engine combustion are urgently needed to aid engine development
activities.

4. Diesel engines

Diesel engines are attractive because of their superior fuel effi-

ciency and because of the global demand to reduce CO2 emission
from automotive engines, which is directly related to fuel con-
sumption. The modern ‘clean’ automotive diesel engine is 20–
40% more fuel efficient than its gasoline counterpart. The improved
efficiency is partly due to the fact that load control is achieved Fig. 2. Predicted (left) and measured (right) OH distributions in a plane through the
without throttling the intake air flow as is required in the stoichi- sprays for operation in three different diesel combustion regimes. Top: HTC short
ometric SI engine, and thus breathing or pumping losses are re- ignition delay; Middle: HTC, medium ignition delay; bottom: LTC, long ignition
duced. Instead, load control is achieved by varying the amount of delay. White dot at left of images shows location of injector nozzle; white arc at
right is the cylinder liner. Numbers at bottom left indicate crank angle after start of
fuel directly injected into the cylinder. injection, and relative camera gains at top right. White rectangular region in
Much current research is thus aimed at improving the under- experimental LTC image indicates location of OH determined by comparing on and
standing of diesel sprays, such as the Sandia Engine Combustion off-line PLIF data. The blue particles in simulations identify liquid jets.
 R.D. Reitz / Combustion and Flame 160 (2013) 1–8 5

scenarios (details of operating conditions are in [6]) (the presence to meet emissions mandates, such as the use of non-optimal fuel
of OH is an indicator of the main heat release event). The results injection timings (to allow more time for fuel–air mixing), the
show the dramatic dependence of diesel flame structure on the use of large amounts of EGR (to dilute the charge and lower local
time available for fuel–air mixing before combustion is initiated. temperatures), or the use of ultra-high injection pressures (to pro-
The High Temperature Combustion (HTC) short-ignition delay mote vigorous mixing – injection systems with 3000 bar injection
(ID) case is representative of Conventional Diesel Combustion pressure are now commercially available) tend to reduce diesel en-
(CDC). Here a 1–2 mm thick diffusion flame (as visualized by the gine fuel efficiencies, and also increase engine expense.
OH contours) is anchored some distance downstream of the nozzle Most diesel engine manufacturers have elected to use Selective
at the so-called flame lift-off location. In the medium-ignition de- Catalytic Reduction (SCR) exhaust gas after-treatment for NOx
lay case, the diffusion flame thickness increases to 4–6 mm and the reduction. However, with SCR there is also a fuel penalty since a
lift-off length increases. In the Low Temperature Combustion (LTC) reducing agent, such as urea (carbamide) is sprayed into the ex-
long-ignition delay case, the OH spreads throughout the jet, and haust stream at rates (and cost) of about 1% of the fuel flow rate
the combustion represents a Premixed Charge Compression Igni- for every 1 g/kW h of NOx reduction desired. Depending on the en-
tion (PCCI) situation. As can be inferred, the reaction rate depen- gine calibration, this can amount to a 2–4% fuel penalty. Soot con-
dence on injection-generated mixing decreases with increasing trol is achieved using Diesel Particulate Filters (DPF), which require
ignition delay, and the reaction zone thickness increases with in- periodic regeneration by adjusting the fuel–air mixture strength to
creased mixing time. increase exhaust temperatures to burn off accumulated soot. This
The model flame structures in Fig. 2 were predicted without imposes as much as a 3% additional fuel penalty. It is worth noting
considering sub-grid scale turbulence-chemistry interactions. Igni- that the cost of the after-treatment system on a modern truck is
tion occurs in a premixed region and combustion is controlled by roughly the same as the cost of the engine itself, and is thus a ma-
diffusion and large scale (i.e., resolved scale) mixing processes. In- jor expense to the consumer. However, the effectiveness of modern
deed, simulations made using the level set G-Equation model gave diesel exhaust gas after-treatment systems is remarkable [7].
virtually identical results, except in the immediate vicinity of the In conventional diesel engines fuel is injected near Top Dead
flame anchor point [6]. (Note that the flame propagation velocity Center (TDC). As seen in Fig. 2, ignition and combustion occur
is zero in a diffusion flame, except near the lift-off point. This is while the fuel is being injected, and the combustion rate is
captured in the G-Equation model since the flame speed is a func- controlled by the injection and air entrainment rates into the spray
tion of the laminar flame speed, which depends on the local equiv- jets. The mixture equivalence ratio varies within the combusting
alence ratio.) spray and regions of high temperature and relatively low U lead
Conventional diesel combustion is characterized by high NOx to NOx formation, while soot formation is favored in high U,
and soot emissions, since the combusting spray spans a wide range intermediate temperature regions. These trends are also seen in
of local equivalence ratios and temperatures. Measures introduced Fig. 3, which summarizes engine-out CO, HC, NOx and soot

1.8 1.8
800.0
.0
1800
1.5 1.5 712.2
1601.1
Equivalence ratio

Equivalence ratio

448.9
1.2 1.2
1004.4
185.6
208.9
0.9 0.9
low .0
10.0 10

0.6
emission CO(g/kg-f) 0.6
HC(g/kg-f)
1601.1 712.2
window 1203.3 536.7
361.1
805.6
0.3 407.8 185.6
10.0
0.3 10.0

1400 1600 1800 2000 2200 1400 1600 1800 2000 2200
Peakcycletemperature (K) Peakcycletemperature (K)
(a)CO (b)HC
1.8 1. 8
4.5
4.0
1.5 1. 5
3.0
Equivalence ratio

Equivalence ratio

1.2 1. 2 1.5
0.5

0.9 0. 9
NO (g/kg-f)
0.5

7.7

.6
.2

Soot (g/kg-f)
36
22

0.6 72.8 0. 6
58.3 4.5
43.9 3.5
29.4 2.5
15.0 1.5
0.3 0.5 0. 3 0.5

1400 1600 1800 2000 2200 1400 1600 1800 2000 2200
Peakcycletemperature (K) Peakcycletemperature (K)

(c)NO (d)Soot
Fig. 3. Predicted HCCI engine-out (a) CO, (b) HC, (c) NO, and (d) soot emissions on equivalence ratio-peak cylinder temperature maps (mfuel = 20 mg, S = 2000 rev/min,
IVC = 142° ATDC) [11]. Ovals indicate location of low emissions window region.
6 R.D. Reitz / Combustion and Flame 160 (2013) 1–8

emission predictions from HCCI engine simulations as a function of high loads. Extended operating range has been demonstrated using
U and the corresponding peak cycle gas temperature. Although spark-assisted HCCI combined with multiple injection/multiple
this analysis does not consider mixing effects, the U–T plot [8] ignition and valve-timing strategies. In Gasoline Direct Injection
has been found to be very useful for understanding engine com- Compression Ignition (GDICI) operation, more control over the
bustion and emissions, and it helps identify the conditions neces- start-of-combustion timing can be achieved with suitably timed
sary for low emissions operation. multiple injections to create partially-premixed (or ‘‘premixed en-
For example, the use of multiple, appropriately timed injections ough’’) in-cylinder U stratification. Since higher U regions are
can reduce the extent of overly fuel rich regions in the combustion more reactive, this leads to sequential auto-ignition at different
chamber, and thus reduce soot formation. ‘‘Lifted flame’’ combus- spatial locations in the chamber and pressure rise rates are
tion is a means to reduce both NOx and soot, in-cylinder. Increas- reduced, with the result that higher loads can be achieved.
ing the ignition delay anchors the diesel flame sufficiently far Greatly improved combustion control has been demonstrated
downstream of the injector nozzle that the air entrained into the at the ERC with dual fuel CI combustion. In-cylinder fuel blending
jet yields local equivalence ratios below about 2 at the lift-off point is arranged using port fuel injection of a lower reactivity fuel
(c.f., HTC Medium ID case, Fig. 2). U < 2 has been found to be (e.g., gasoline), coupled with optimized in-cylinder multiple
sufficiently lean for ‘‘smokeless’’ combustion. Extended ignition injections of more reactive fuel (e.g., diesel). The resulting strat-
delays can be achieved by using a lower Cetane Number (CN) fuel, egy has been called Reactivity Controlled Compression Ignition
together with appropriate injection and thermodynamic condi- (RCCI) and it generates both equivalence ratio and reactivity
tions. The use of oxygenated fuels also reduces the amount of stratifications in the chamber. Combustion progresses sequen-
entrained air needed to reduce the local equivalence ratio, and thus tially from the high reactivity regions to low reactivity regions,
reduces soot emissions. However, oxygenated fuels can also thereby effectively lowering pressure rise rates. RCCI does not re-
increase NOx emissions by moving the combustion closer to quire high fuel injection pressures since the injections occur well
stoichiometric operation, resulting in higher flame temperatures. before TDC, thus providing time for fuel vaporization and mixing.
Indeed, excellent results have been achieved with the diesel
5. Advanced low temperature combustion engines injector replaced by a low pressure GDI injector. RCCI extends
the engine’s operable load range beyond that of HCCI or PCCI
HCCI is a promising combustion strategy because the dilute since fuel blending allows the combustion to be tailored. Indeed,
homogeneous charge avoids locally high temperatures or equiva- gasoline-diesel RCCI has been demonstrated up to 23 bar gIMEP
lence ratios in the chamber. As seen in Fig. 3, this allows operation in a heavy-duty diesel engine with a compression ratio of about
in a low emissions window. Increased thermal efficiency is also 12:1.
seen with low temperature combustion strategies because of lower CFD modeling was a key enabling technology in the develop-
wall heat transfer losses. LTC engines operate fuel lean, which ment of RCCI. CFD, combined with genetic algorithm optimization
yields higher specific heat ratios and hence increased work extrac- techniques, discovered fuel injection strategies and safe operating
tion. However, LTC operation reduces the available exhaust energy, conditions with acceptable pressure rise rates [9]. U.S. EPA 2010
which is needed for turbocharger operation, and also increases heavy-duty diesel engine NOx and soot emission mandates were
combustion irreversibility, due to the need to transfer heat to more met in-cylinder, without the need for after-treatment, while
inert gas. However, the net effect is positive – for example, HCCI is achieving gross indicated thermal efficiencies in excess of 56%. This
a high efficiency, low emissions combustion strategy. efficiency is about 10% higher than conventional diesel combustion
Unfortunately, despite the advantages of HCCI, several prob- (i.e., about 20% less fuel consumed). Further experiments in the
lems limit its adoption. The first difficulty is the achievement of ERC’s single-cylinder heavy-duty diesel engine with 18.7:1 com-
a homogeneous charge in the engine. Many methods of fuel prep- pression ratio have demonstrated gross indicated thermal efficien-
aration have been proposed, such as port injection or early direct cies reaching 60% [10], and pathways to even higher efficiencies
injection with a narrow injected included spray angle to avoid have been identified. These efficiencies rival those of any other
spray-wall impingement. This latter approach is referred to as Par- powerplant, including those of state-of-the-art, MW-level natural
tially Premixed CI, since with late injections the charge may still be gas combined-cycle electrical power stations. The high efficiency
somewhat inhomogeneous at ignition. A second difficulty is HCCI’s of the ERC’s engine was achieved using E85 as the low reactivity
limited operating range. Due to the nature of volumetric combus- fuel (i.e., high ON) and gasoline doped with a cetane improver
tion, the cylinder contents ignite virtually simultaneously and this (ethyl-hexyl-nitrate) as the high reactivity fuel (i.e., high CN) to in-
can produce unacceptable noise and knocking at high loads. At- crease the extent of in-cylinder fuel reactivity gradients. In addi-
tempts have been made to increase load by injecting water, apply- tion, a low-surface-area piston was used to further reduce wall
ing high EGR rates, or by introducing temperature stratification in heat losses, which are already low due to the low chamber gas
the cylinder. temperatures.
An additional difficulty with HCCI is control of the start of com- In contrast to other LTC concepts, the combustion phasing of
bustion timing. In conventional CI engines, start of combustion is RCCI can be altered by adjusting the proportions of the more and
controlled by the fuel injection timing, while in SI engines the less reactive fuels on a same- or next-cycle basis with appropriate
spark initiates combustion. On the other hand, HCCI ignition is combustion feedback control. RCCI also offers considerable fuel
controlled by the charge mixture temperature, composition and flexibility. High-efficiency, low-emissions operation has been dem-
compression history. More control is possible with a low reactivity onstrated using direct-injected diesel with port-injected methanol
fuel, such as gasoline whose high resistance to auto-ignition allows and natural gas, and even with hydrated ethanol, which cannot be
more mixing time prior to combustion, producing lower local U used in either SI or CDC engines due to its poor reactivity. Hydrated
and thus lower NOx and soot emissions. Indeed, the best fuel for ethanol is significantly cheaper to produce, since removing all the
HCCI operation would have auto-ignition qualities between those water from ethanol is extremely costly. In addition, port injection
of diesel and gasoline. High load operation is favored with a low of gasoline, plus a small quantity of cetane improver added to
CN (e.g., high octane number gasoline), while low load operation the direct injected fuel (also gasoline), allows an RCCI engine to
favors a high CN (e.g., diesel). operate on a single fuel (i.e., gasoline), together with a wind-
A practical strategy might use dual mode combustion, with shield-wiper bottle sized supply of cetane improver, which is
HCCI at low and medium loads and conventional combustion at topped-up every oil change interval.
 R.D. Reitz / Combustion and Flame 160 (2013) 1–8 7

A remaining challenge with LTC concepts is UHC and CO emis- picted by large arrows in Fig. 4) and, unless high EGR levels are
sions, which increase unless combustion temperatures exceed used, it features high local temperatures, leading to high NOx. Also,
about 1600 K under lean conditions (see Fig. 3). However, recent because the charge is stratified, fuel rich regions can lead to high
RCCI experiments show that the use of a Diesel Oxidation Catalyst soot. Meeting the simultaneous constraints of high-efficiency and
(similar to the SI engine’s 3-way catalyst) can be effective for ex- clean combustion, in-cylinder, has been extremely challenging.
haust gas UHC/CO reduction. The use of the DOC also decreases This has necessitated the development of costly exhaust gas
the already extremely low amount of engine-out Particulate Mat- after-treatment systems.
ter (PM), which is primarily comprised of organic carbon (i.e., con- The most fuel efficient and clean engines exploit compression
densed heavy HCs). Little is known about the formation ignition after the fuel–air mixing process is more-or-less complete,
mechanisms of this type of PM. Also, further research on low tem- and the role of turbulent mixing is of secondary importance com-
perature oxidation catalysts would be very helpful. pared to chemical kinetics. Low temperature combustion regimes
include the Premixed Charge, Reactivity Controlled and Homoge-
6. Closure neous Charge Compression conditions depicted in Fig. 4. LTC oper-
ation is largely controlled by the details of the fuel’s ignition
This article has attempted to identify research areas that, in this chemistry. Indeed, the best fuel efficiency is achieved when the
author’s opinion, will help address the urgent fuel usage and pollu- in-cylinder flow field is quiescent (i.e., turbulence and organized
tant emission problems associated with transportation and other charge motion are small) to further reduce convective wall heat
internal combustion engine applications. Three classes of IC en- transfer losses. Thus, when trying to tame fluid mechanics and
gines were discussed and contrasted – Spark-Ignited, Conventional combustion in an IC engine, the lesson is ‘‘the simpler the better!’’.
Diesel and Low Temperature Combustion, and their operating re- Advanced combustion strategies such as Gasoline-Direct-Injec-
gimes are summarized conceptually in Fig. 4. It is noted that the tion (GDICI) and Reactivity-Controlled Compression Ignition (RCCI)
ubiquitous gasoline SI engine, which relies on high temperature have demonstrated up to 20% improved fuel efficiency over stan-
turbulent flame propagation, is controlled by complex physical dard diesel, and 40–50% over conventional SI gasoline engines.
and chemical kinetic processes, including both large- and small- These technologies also offer significant cost advantages because
scale mixing (indicated in Fig. 4 by bent arrows in the flame brush). expensive NOx and soot after-treatment and costly ultra-high
The SI engine is also the least fuel efficient, partly because its high injection pressure systems are not required. Although the combus-
flame temperatures lead to high wall heat transfer losses. High tion process may be ‘‘simpler’’ in advanced engines, advanced feed-
flame temperatures also lead to high NOx emissions, but tail-pipe back control and high-performance air-handling systems will be
emissions are low, thanks to the low cost 3-way catalyst. However, required for optimal performance. However, the implications of
a remaining concern is emissions of N2O and other UHC pollutants the promising improvements in fuel efficiency of LTC engines are
during cold-start (i.e., before catalyst ‘‘light-off’’). N2O has a long very significant. It has been estimated that if RCCI was adopted
atmospheric lifetime (approximately 120 years) and is about 300 to replace the relatively inefficient SI engine, U.S. transportation
times more effective than CO2 as a GHG. In stratified charge GDI oil usage would be reduced by 34%, which is 100% of current U.S.
engines, the combustion spans a region between the rich and lean oil imports from the Persian Gulf. Even greater reductions in oil
limits (also depicted in Fig. 4), and this necessitates advanced and usage are possible if the efficiency improvements are combined
costly after-treatment measures for both NOx and soot control. with electric hybrid technologies in the vehicle.
The conventional diesel engine is 20–40% more fuel efficient Finally, the dual fuel RCCI engine has the further advantage of
than the SI engine. The combustion process is also less complex, fuel flexibility. As long as the two fuels have significantly different
since it does not rely on flame propagation. Instead, a diffusion reactivities they can be blended appropriately, in-cylinder, for opti-
flame is formed and the rich and lean flammability limits deter- mal performance as a function of speed and load. RCCI can operate
mine the thickness of the flame surrounding the fuel spray (de- with a wide range of fuels, including gasoline, natural gas and die-
picted by the broad horizontal region around stoichiometry in sel. It can also be used with a single fuel where one stream is add-
Fig. 4). Combustion is controlled by large-scale fuel–air mixing (de- itized with a small amount of cetane improver, which is
replenished at oil change intervals. In addition, RCCI can be used
with alternative or renewable fuels, such as hydrated ethanol,
methanol, and biodiesel. It is fair to say that the IC engine is a
remarkably efficient device, and when matched with technologies
that allow fuel flexibility, it provides a path away from the monop-
oly of fossil oil for our future energy security.

Acknowledgments

The support of the Engine Research Center’s major research

sponsors, including the U.S. Department of Energy and the Sandia,
Oak Ridge and Argonne National Laboratories, the U.S. Navy, GM,
Ford, Caterpillar and the ERC’s Direct-injection Engine Research
Consortium (DERC) members and the Princeton CEFRC is gratefully
acknowledged. In addition, the contributions of ERC Faculty, Staff
and Students are very much appreciated.

Fig. 4. Conceptual diagram of Conventional Diesel (CDC), stoichiometric and References

stratified Spark Ignition (SI), and Low Temperature Compression Ignition (P-R-H-
CCI) regime locations on the U–T map. The rich and lean limits determine the [1] Energy Technology Perspectives, International Energy Agency, Paris,
turbulent flame propagation limits of stratified SI operation. Use of EGR lowers peak 2012. Available from <http://www.iea.org/publications/freepublications/
temperatures. publication/name,31269,en.html>.
8 R.D. Reitz / Combustion and Flame 160 (2013) 1–8

[2] Assessment of Fuel Economy Technologies for Light-Duty Vehicles, National [10] Splitter, D.A., High Efficiency RCCI Combustion, Ph.D. Thesis, University of
Research Council, 2011, ISBN 978-0-309-15607-3. Wisconsin-Madison, 2012.
[3] Y. Shi, H.-W. Ge, R.D. Reitz, Computational Optimization of Internal [11] S.W. Park, R.D. Reitz, Combust. Sci. Technol. 179 (11) (2007) 2279–2307.
Combustion Engines, 2011. Available from <http://www.springer.com/
engineering/mechanical+eng/book/978-0-85729-618-4>.
Rolf Reitz is Wisconsin Distinguished Professor of Mechanical Engineering at the
[4] DOE Vehicle Technologies Multi-Year Program Plan 2011-2015. Available from
University of Wisconsin-Madison where he is currently director of the Engine
<http://www1.eere.energy.gov/vehiclesandfuels/pdfs/program/vt_mypp_2011-
2015.pdf>. Research Center. He also directs the ERC’s Direct-injection Engine Research Con-
[5] http://www.sandia.gov/ecn/workshop/ECN1.php. sortium. He received the Ph.D. degree from Princeton University in 1978. Before
[6] S.L. Kokjohn, R.D. Reitz, ASME, J. Eng. Gas Turbines Power 133 (2011) 102805- joining UW-Madison in 1989, he spent six years at the GM Research Laboratories
1–10. and several years on the staff at Princeton University and at the Courant Institute of
[7] Today’s 18-wheeler diesel-engine truck would have to drive 143 miles on the Mathematical Sciences, New York University. His research includes the develop-
freeway to put out the same mass of particulates as a single charbroiled ment of advanced fuel injection and combustion strategies, as well as computer
hamburger patty! University of California Riverside Center for Environmental models for IC engines. He is consultant to many industries and has won major
Research and Technology. Available from <http://ucrtoday.ucr.edu/8896>. research awards, including the 2012 DOE Vehicle Technologies R&D Award, SAE
[8] T. Kamimoto, M. Bae, High Combustion Temperature for the Reduction of Horning Memorial awards, and the ASME Soichiro Honda Medal and Internal
Particulate in Diesel Engines, SAE paper 880423, 1988. Combustion Engine award. He co-founded the International Journal of Engine
[9] S.L. Kokjohn, R.M. Hanson, D.A. Splitter, R.D. Reitz, Int. J. Eng. Res. Special Issue Research (IJER) in 1999, and serves as Editor, Americas. Professor Reitz has more
on Fuel Efficiency 12 (2011) 209–226, http://jer.sagepub.com/content/12/3/
than 400 publications and 5 patents – http://reitz.me.wisc.edu.
209.full.pdf+html.