ARTICLE IN PRESS

Energy 32 (2007) 1791–1808

www.elsevier.com/locate/energy

Experimental heat release analysis and emissions of a HSDI diesel

engine fueled with ethanol–diesel fuel blends
C.D. Rakopoulos, K.A. Antonopoulos, D.C. Rakopoulos
Internal Combustion Engines Laboratory, Department of Thermal Engineering, School of Mechanical Engineering,
National Technical University of Athens, Zografou Campus, 9 Heroon Polytechniou St., 15780 Athens, Greece
Received 21 November 2006

Abstract

An experimental study is conducted to evaluate the effects of using blends of ethanol with conventional diesel fuel, with 5%, 10% and
15% (by vol.) ethanol, on the combustion and emissions of a standard, fully instrumented, four-stroke, high-speed, direct injection
(HSDI), ‘Hydra’ diesel engine located at the authors’ laboratory. The tests are conducted using each of the above fuel blends or neat
diesel fuel, with the engine working at a speed of 2000 rpm and at four different loads. In each test, combustion chamber and fuel
injection pressure diagrams are obtained using a specially developed, high-speed, data acquisition and processing system. A heat release
analysis of the experimentally obtained cylinder pressure diagrams is developed and used, with the pertinent application of the energy
and state equations. From the analysis results, plots of the history in the combustion chamber of the gross heat release rate and other
related parameters reveal some very interesting features, which shed light on the combustion mechanism when using these blends.
Moreover, for each test, volumetric fuel consumption, exhaust smokiness and exhaust regulated gas emissions are measured. The
differences in the performance and exhaust emission parameters from the baseline operation of the diesel engine, i.e., when working with
neat diesel fuel, are determined and compared. The heat release analysis results for the relevant combustion mechanism, combined with
the widely differing physical and chemical properties of the ethanol against those for the diesel fuel, are used to aid the correct
interpretation of the observed engine behavior.
r 2007 Elsevier Ltd. All rights reserved.

Keywords: Diesel engine; Ethanol–diesel blends; Heat release analysis; Combustion; Emissions

1. Introduction domain of fuel-related techniques, such as for example the

use of alternative fuels often in fumigated form [3,4], or
Engine manufacturers all over the world have achieved gaseous fuels of renewable nature that are friendly to the
to develop diesel engines with high thermal efficiency and environment [5–7], or oxygenated fuels that show the
specific power output, always trying to keep inside the ability to reduce particulate emissions [8,9] usually with an
limits of the imposed emissions regulations, which every escorting increase of the emitted nitrogen oxides. Some-
day become more stringent. Significant achievements for times, in order to obtain the desirable results, resort is
the development of cleaner diesel engines have been made, being made to the simultaneous use of engine- and fuel-
over the last years, by following various engine-related related techniques.
techniques, such as for example the use of common-rail Dwindling crude oil reserves and their constantly
systems, fuel injection control strategies, exhaust gas increasing prices have placed increasingly sensitive loads
recirculation, exhaust gas after-treatment, etc. [1,2]. on the trade balances of the non-oil producing countries
Furthermore, especially for the reduction of pollutant and, meanwhile, have come to represent a threat to the
emissions, researchers have focused their interest on the existence of the developing and industrialized countries.
Then, considerable attention has been paid on the
Corresponding author. Tel.: +30 210 7723529; fax: +30 210 7723531. development of alternative fuel sources in various coun-
E-mail address: [email protected] (C.D. Rakopoulos). tries, with particular emphasis on the bio-fuels that posses

0360-5442/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2007.03.005
 ARTICLE IN PRESS
1792 C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808

Nomenclature l ratio of crank radius to piston rod length

r density (kg/m3)
bi number of kmol of species ‘i’ j crank angle (degrees)
cv specific heat capacity under constant volume F fuel–air equivalence ratio
(J/kg K)
D cylinder diameter (m) Subscripts
FAst stoichiometric fuel–air ratio (by mass)
h (sensible) specific enthalpy (J/kg) b burned
m mass (kg) c consumption rate
M molecular weight (kg/kmol) e outflow (blowby)
N engine speed (rpm) f fuel
p pressure (Pa) g gross (heat release)
Q heat (J) n net (heat release)
r crank radius (m) w wall
R specific gas constant (J/kg K)
Rmol universal gas constant, 8314.3 J/kmol K Abbreviations
s piston stroke (m)
t time (s) b.m.e.p. brake mean effective pressure (bar)
T absolute temperature (K) BDC bottom dead center
U (sensible) internal energy (J) EVO exhaust valve opening
V volume (m3) E15-D 15% (by volume) ethanol blend in diesel fuel
IVC inlet valve closing
Greek symbols rpm revolutions per minute
TDC top dead center
g specific heat capacities ratio 1CA degrees of crank angle
Y fuel lower calorific value (J/kg)

the added advantage of being renewable fuels that can be by this research group [11], which evaluates and compares
replenished through the growth of plants or production of the performance and environmental behavior of a multi-
livestock, showing an ad hoc advantage in reducing the tude of vegetable oils and their bio-diesels, in blend ratios
emitted carbon dioxide [10,11]. There is a commitment by with diesel fuel of up to 20
80.
the USA government to increase bio-energy three-fold in Because of its high octane number, ethanol is a good
10 years, which has added impetus to the search for viable spark-ignition engine fuel, while vegetable oils and bio-
bio-fuels [12]. diesels are good diesel engine fuels owing to their reason-
Concerning the environmental aspect, rational and ably high cetane number. It is true that alcohols, mainly
efficient end-use technologies are identified as key options ethanol and to a much lesser extent methanol, have been
for the achievement of the Kyoto targets of greenhouse gas considered as alternative fuels for diesel engines too
emissions reduction. For the transport sector of European [12,20,21]. Methanol can be produced from coal or petrol
Union (EU), energy savings of 5–10% in the medium term based fuels with low cost production, but it has a restrictive
and an aggregated of 25% in the long term (2020) are solubility in the diesel fuel. On the other hand, ethanol is a
aimed at, with an expected cut of carbon dioxide (CO2) biomass based renewable fuel, which can be produced by
emissions by 8% by the year 2010. The EU has issued a alcoholic fermentation of sugar from vegetable materials,
directive on the use of bio-fuels accounting to at least 2% such as corn, sugar cane, sugar beets, barley, sweet
of the market for gasoline and diesel fuel sold as transport sorghum, cassava, molasses and the like, and agricultural
fuels by the end of 2005, increasing in stages to a minimum residues, such as straw, feedstock and waste woods by
of 5.75% by the end of 2010 [11,13]. using already improved and demonstrated technologies
Bio-fuels made from agricultural products (oxygenated [19]. Therefore, it has the advantage over methanol of
by nature), may not only offer benefits in terms of exhaust higher miscibility with the diesel fuel and of being of
emissions, but also reduce the world’s dependence on oil renewable nature (bio-ethanol).
imports. Moreover, local agricultural industries can be
supported and farming incomes enhanced, besides provid- 2. Use of ethanol in diesel engines: short literature review
ing a better energy security for many developing countries.
Among these, bio-alcohols and vegetable oils or their The initial investigations into the use of ethanol in diesel
derived bio-diesels (methyl or ethyl esters) are considered engines were carried out in South Africa in the 1970s and
as very promising fuels [14–19]. A work has been presented continued in Germany and the USA during the 1980s.
 ARTICLE IN PRESS
C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808 1793

Most of these works related to the use in farm equipment proof or lower), which is more economical to produce and
(tractors and combines) or to the reduction in smoke and find, or owing to water absorption by its hygroscopic
particulate emissions [22–26]. nature, an emulsifying agent is required to form the opaque
Nonetheless, there are several critical issues to consider macro-emulsion; this, however, could be separated into the
with the use of ethanol in the diesel fuel. While anhydrous two phases if allowed to stagnate over a long period [22].
ethanol is soluble in gasoline, additives must be used in Blends with up to 15% (by volume) ethanol in diesel fuel
order to ensure solubility of anhydrous ethanol (that is can be considered relatively safe from the engine durability
highly hygroscopic and thus practically impossible to stay viewpoint.
as such) in the diesel fuel under a wide range of conditions. Most of the reported researches with ethanol–diesel fuel
Especially at lower temperatures, the miscibility is limited; refer to higher capacity diesel engines (usually turbo-
it was determined that the aromatic content (having a charged) or smaller stationary ones. The present work
co-solvent action), intermediate distillate temperatures and deals with a standard experimental, Ricardo/Cussons
wax content of diesel fuel had a significant impact on the ‘Hydra’, high-speed, direct injection (DI), naturally aspi-
miscibility limits. Furthermore, adding ethanol to diesel rated engine of the automotive type, which possesses a high
fuel can reduce lubricity and create potential wear versatility and control over the variation of its operating
problems in sensitive fuel pump designs. Ethanol possesses parameters. The experiments were conducted without any
also lower viscosity and lower calorific value, with the modification on the engine. An emulsifying agent by ‘Betz
latter imposing minor changes on the fuel delivery system GE’ was used to satisfy ethanol–diesel fuel mixture
to establish the maximum power delivered by the use of the homogeneity and prevent phase separation, while no
neat diesel fuel [12,27,28]. ignition-improving additives were used.
By all estimates, ethanol has a very low cetane number The work evaluates the effects of using various blends of
that reduces the cetane level of the diesel-ethanol blend, ethanol with normal diesel fuel, with 5%, 10% and 15%
requiring normally the use of cetane enhancing additives, (by volume) ethanol, on the performance and exhaust
which improve ignition delay and mitigate cyclic irregu- emissions at various loads. The differences in the perfor-
larity [26,29–31]. Ethanol has much lower flash point than mance and exhaust emission parameters, from the baseline
the diesel fuel and higher vapor formation potential in operation of the diesel engine (working with diesel fuel),
confined spaces, thus requiring extra precautions to ensure are determined and compared. In a recently appeared
safe handling and use of these blends [12]. publication [42], consideration of theoretical aspects of
Therefore, various techniques involving ethanol–diesel diesel engine combustion and of the widely differing
dual fuel operation have been developed, to make diesel physical and chemical properties of the ethanol against
engine technology compatible with the properties of those for the diesel fuel were used to aid, in a rather
ethanol-based fuels. Broadly speaking, they can be divided qualitative way, the interpretation of the observed engine
into the following three categories: behavior with these blends.
At this point it is stated that there seems to be an obvious
(i) Ethanol fumigation to the intake air charge, by using scarcity of theoretical models scrutinizing the formation
carburetion or manifold injection. There seems to be mechanisms of combustion-generated emissions when
limits to the amount of ethanol that can be used in this using liquid bio-fuels, unlike the advanced models existing
manner, due to incipience of engine knock at high for the study of diesel engines when using conventional
loads, and prevention of flame quenching and misfire diesel fuel [43–47]. Recently, a detailed multi-zone combus-
at low loads [22,32–34]. tion modeling with these ethanol–diesel blends has been
(ii) Dual injection system that is not considered very submitted [48], along the lines of a recent paper by the first
practical, as requiring an extra high-pressure injection three authors [49] applied for the engine in hand when
system for the ethanol and, thus, a related major using vegetable oils or their derived bio-diesels, which has
design change of the cylinder head [22]. in turn expanded on a similar multi-zone one that has dealt
(iii) Blends (emulsions) of ethanol and diesel fuel by using only with the related physical processes [17].
an emulsifier to mix the two fuels in order to prevent As a ‘bridging’ agent for correctly interpreting the
separation; these require no technical modifications on present ethanol–diesel blends results and with a view to
the engine side [22,35–41]. assist the related multi-zone combustion modeling, a heat
release analysis for the relevant combustion mechanism,
It is reminded that stable emulsions usually refer to both combined with the widely differing physical and chemical
solutions of anhydrous ethanol in diesel fuel (transparent) properties of the ethanol against those for the diesel fuel,
and micro-emulsions of ethanol in diesel fuel (translucent). are used in the present study. As known, concerning the
Anhydrous ethanol (200 proof) does not require an study of the combustion process in diesel engines, either of
emulsifying agent or so-called ‘surfactant’ (from, surface the DI or the IDI (indirect injection) type, a very important
active agent) to form a transparent solution in diesel fuel, means to analyze combustion characteristics is the calcula-
but these solutions can tolerate only 0.5% water. Then, in tion and analysis of heat release rates according to actual
the practical cases of using lower proof ethanol (say 190 measurements of pressures in the (main) chamber (and the
 ARTICLE IN PRESS
1794 C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808

pre-chamber, if any) [50–55]. Also, a companion diagram Table 1

of the fuel injection pressure assists towards this side. This Engine and injection system basic data
technique, however, is not devoid of pitfalls, requiring Engine model and type Ricardo–Cussons ‘Hydra’
special care and skill for the correct acquisition and Single-cylinder, compression-
processing of the cylinder pressure signals; much more ignition, direct injection,
care is needed in the case of IDI engines, because of the naturally aspirated, four-stroke,
continuous mass and energy exchange existing between the water cooled, high-speed
two chambers. Speed range 1000–4500 rpm
The experimental cylinder pressure (indicator) diagrams, Bore/stroke 80.26 mm/88.90 mm
from the present DI engine, are directly processed in Compression ratio 19.8:1
connection with the pertinent application of the energy and Inlet valve It opens at 8 1CA before TDC
It closes at 42 1CA after BDC
state equations. Use is made of relevant sub-models for the Exhaust valve It opens at 60 1CA before BDC
gas heat loss to the walls and blowby rate, as well as of It closes at 12 1CA after TDC
appropriate properties of the working medium as a Fuel pump ‘Bosch’ with 11 mm diameter
function of temperature and composition. Attention is plunger
paid to the experimental work related to the specially Injector body and nozzle ‘Bosch’ with four injector nozzle
holes
developed, high-speed, data (signal) acquisition and
Nozzle hole diameter: 0.25 mm
processing system connected to the engine. Opening pressure: 250 bar
The analysis results for the rates of heat release and Advance (at pump spill):
other related parameters in the combustion chamber reveal 0–40 1CA
some very interesting features, which aid the correct
interpretation of the combustion mechanism associated
with the use of ethanol–diesel fuel blends in the diesel control panel and to the controller. Via knobs on the control
engine. Observed phenomena of cyclic irregularity panel, the operator can set the dynamometer to control speed
[30,56,57], especially with the higher ethanol content or load. There is also a capability of setting automatically
blends, will be reported and analyzed in another the static injection timing from a switch on the control
communication. panel.
Electrically driven pumps assure the coolant (water) and
3. Experimental engine and test facility lube-oil circulation, with the temperature controlled by
water-fed heat exchangers. The secondary water-cooling
3.1. Engine description system for these heat exchangers was shop made, compris-
ing a big water tank and a corresponding (secondary) heat
Facilities to monitor and control engine variables such as exchanger. Heaters are used to maintain oil and coolant
engine speed, load, water and lube-oil temperatures, fuel temperatures during warm up and light load conditions.
and air flows, etc., are installed on a fully automated test Thermocouples are located at strategic points in the engine,
bed, single cylinder, four-stroke, water cooled, Ricardo– with their indications shown on a multi-point electronic
Cussons, ‘‘Hydra’’, high speed, experimental standard temperature indicator. The engine exhaust system is
engine located at the first author’s laboratory. This engine connected to a shop made silencer system. A viscous type,
has the ability to operate on the Otto (spark-ignition) or DI ‘Alcock’ laminar flowmeter is used for measuring the air-
diesel or IDI diesel, four-stroke principle, by changing flow aspirated by the engine.
various parts of the crank gear mechanism, cylinder A shop made tank and flow metering system is used for
and head. fuel consumption measurements of the various blend
For the present investigation it is used as a naturally samples as follows. A glass burette of known volume was
aspirated, DI diesel engine having a re-entrant bowl-in- placed in parallel to the 3.5-l tank, and the time was
piston combustion chamber. The engine and injection measured for its complete evacuation of the fuel sample
system basic data are given in Table 1. Fig. 1 gives a full feeding the engine. A system of pipes and valves was
schematic arrangement of the engine test bed, instrumenta- constructed in order to achieve a quick drain of a fuel
tion and data logging system. sample, including the return-fuel from the pump and
injector, and the refill of metering system with the new
3.2. Test installation description fuel sample.

The engine is mounted on a fully automated test bed and 3.3. Exhaust gas analyzers system
coupled to a ‘‘McClure’’ DC motoring dynamometer,
having load absorbing and motoring capabilities, which is The exhaust gas analysis system consists of a group of
equipped with a load cell for the engine torque measure- analyzers for measuring soot (smoke), nitrogen oxides
ments. There is one electric sensor for speed and one for (NOx), carbon monoxide and total unburned hydrocar-
load (torque), with these signals fed to indicators on the bons (HC). The smoke level in the exhaust gas was
 ARTICLE IN PRESS
C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808 1795

Fig. 1. Schematic arrangement of the engine test bed, instrumentation and data logging system.

measured with a ‘Bosch’ RTT-100 opacimeter, the readings signal. The signal from a ‘‘Tektronix’’ TDC magnetic pick-
of which are directly provided as Hartridge units (% up marker is used for time reference.
opacity) or equivalent smoke (soot) density (milligrams of The output signals from the magnetic pick-up and the
soot per cubic meter of exhaust gases). two piezoelectric transducers signals, while they were
The carbon monoxide concentration in ppm (parts per continuously monitored on a dual beam ‘‘Tektronix’’
million, by volume) in the exhaust was measured with a storage oscilloscope, are connected to the input of a
‘Signal’ Series-7200 non-dispersive infrared analyzer ‘‘Keithley’’ DAS-1801ST A/D board installed on an IBM
(NDIR). The nitrogen oxides concentration in ppm in the compatible Pentium III PC. This board can acquire input
exhaust was measured with a ‘Signal’ Series-4000 chemi- data at a total throughput rate of 312.5 ksamples/s from up
luminescent analyzer (CLA). The total unburned hydro- to eight differential analogue inputs, utilizing also dual-
carbons concentration (in ppm) in the exhaust was channel direct memory access (DMA) operation. It is then
measured with a ‘Ratfisch-Instruments’ flame-ionization appropriate for the recording of high frequency engine
detector (FID). The last two analyzers were fitted with signals, with the added advantage of high signal resolution
thermostatically controlled heated lines. (much less than 1 1CA, depending also on the number of
simultaneous recorded signals). This high acquisition rate
capability secures the nearly simultaneous sampling of all
3.4. Transducers measuring set-up and data acquisition input signals too.
system Control of this high-speed data acquisition system is
achieved by developing a computer code based on the
For measuring the pressure in the cylinder, a ‘‘Kistler’’ ‘‘TestPoint’’ control software. The resultant program,
miniature piezoelectric transducer is used, flush mounted to specially designed for steady or transient internal combus-
the cylinder head and connected to a ‘‘Kistler’’ charge tion engine applications, is an object-oriented functional
amplifier. Also, a ‘‘Kistler’’ piezoelectric transducer is code. It allows the direct on-screen selection of the desired
connected on the injector side of the pipe linking the values of all acquisition parameters, namely: engine type
injection pump and injector to provide the fuel pressure (two- or four-stroke); engine speed (in rpm); desired
 ARTICLE IN PRESS
1796 C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808

number of recorded cycles; selection of the acquired of 29 1CA before TDC, and four loads corresponding to
channels; and sampling interval (in degrees CA). brake mean effective pressures of 0.00 (no-load), 1.40, 2.57
After the selection of the above values, the resulting and 5.37 bar, respectively. Owing to the differences among
sampling rate per channel is displayed on screen. When the lower calorific values ‘Y’ and oxygen contents of the
acquisition starts, data are initially stored in the computer’s fuels tested, the comparison must be effected at the same
memory. This secures the continuous gap-free acquisition engine brake mean effective pressure, i.e., load, and not at
for the large amount of multi-channel input data. When the the same injected fuel mass or air–fuel ratio. For the no-
desired amount of engine cycles is recorded, the data are load conditions, only pressure measurements were taken.
transferred from the memory to a corresponding file in the In each test, volumetric fuel consumption rate Vfc (in m3/s),
hard disk and directly displayed on-screen (as X–Y graphs). exhaust gas temperature, exhaust smokiness and exhaust
Special warning signs and other secondary options are also regulated gas emissions such as nitrogen oxides, carbon
available. monoxide and total unburned hydrocarbons are measured.
From the first measurement, specific fuel consumption and
4. Properties of fuels thermal efficiency are computed using the fuel sample density
and lower calorific value.
The conventional diesel fuel was supplied by the By knowing the engine brake torque Mt (in N m), i.e.,
Aspropyrgos Refineries of the ‘Hellenic Petroleum SA’ the load, which is set for each experiment and kept
and represents the typical, Greek automotive, low sulphur constant by the engine controllers, the following engine
(0.035%) diesel fuel (gas oil); it forms the baseline fuel of performance quantities were computed, with N the
the present study. The properties of the diesel fuel and the (constant) engine speed (in rpm):
ethanol are summarized in Table 2.
The ethanol (CH3CH2OH) is blended with the normal
5 10
diesel fuel at blend ratios of 95 , 90 and 15
 The brake mean effective pressure, b.m.e.p. (in bar)
85 (by volume). For
the present experiments, no cetane improving additives b:m:e:p: ¼ ð4pM t =V h Þ  105 (1a)
(ignition improvers) were used. An emulsifying agent by 2
with Vh ¼ (p/4)D s the (single-cylinder) engine displace-
‘Betz GE’, a division of General Electric, Inc., was used in
ment volume.
proportions of 1.5% by volume, to satisfy mixture
 The brake power, P (in W)
homogeneity and prevent phase separation.
The emulsifier replaced a corresponding part of the P ¼ M t ð2pN=60Þ. (1b)
diesel fuel. Thus, the 15
85 blend ratio corresponds to 83.5%
diesel fuel, 1.5% emulsifier and 15% ethanol (by volume),  The brake specific fuel consumption, b.s.f.c. (in g/kW h)
and similarly for the other two blends. The mixing protocol
consisted of first blending the emulsifier into the ethanol b:s:f :c ¼ ðV fc rf =PÞ3:6  109 . (1c)
and then blending this mixture into the diesel fuel.
 The brake thermal efficiency, b.t.e.
5. Parameters tested and experimental procedure b:t:e: ¼ P=ðV fc rf YÞ. (1d)

The experiments were performed without any modifica-

tion on the engine. The series of tests are conducted using Table 3 shows the accuracy of the measurements and the
each of the above fuel blends, with the engine working at a uncertainty of the computed results of the various
speed of 2000 rpm, at a static (pump spill) injection timing parameters.
At the same time, during each test, combustion chamber
Table 2
(indicator) and injector pressure diagrams were obtained
Properties of diesel fuel and ethanol using the specially developed data acquisition system
described in Section 3.4. The heat release analysis of these
Fuel properties Diesel fuel Ethanol experimentally obtained combustion chamber pressure
Density at 20 1C, kg/m3 837 788 diagrams is exposed in detail in the next section. The
Cetane number 50 5–8 pressures are measured with accuracy better than within
Kinematic viscocity at 40 1C, mm2/s 2.6 1.2 71% of full-scale output (fso), and the accuracy of the
Surface tension at 20 1C, N/m 0.023 0.015 analogue input readings of the data acquisition system is
Lower calorific value, MJ/kg 43 26.8
within 70.01%. The electric characteristics of the charge
Specific heat capacity, J/kg 1C 1850 2100
Boiling point 180–360 78 amplifiers, such as drift, do not contribute to the total
Oxygen, % weight 0 34.8 error, because the drift and linear trend are remedied via
Latent heat of evaporation, kJ/kg 250 840 software routines during the processing of signals.
Bulk modulus of elasticity, bar 16 000 13 200 The experimental work started with a preliminary
Stoichiometric air–fuel ratio 15.0 9.0
investigation of the engine running on neat diesel fuel, in
Molecular weight 170 46
order to determine the engine’s operating characteristics
 ARTICLE IN PRESS
C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808 1797

Table 3 with a sampling rate corresponding to 0.5 1CA. Initially,

Accuracy of measurements and uncertainty of computed results the indicator diagrams are corrected for any drifting of the
Measurements Accuracy
signals obtained from the charge amplifiers, by taking the
mean value of the pressures in the gas exchange periods
Soot density 71 mg/m3 and ‘setting’ this equal to the atmospheric pressure value.
NOx 75 ppm A signal from a magnetic pick-up, which is simulta-
CO 73 ppm
neously recorded, indicates the position of the TDC in each
HC 71 ppm
Time 70.5% cycle. This is shown in Fig. 2 together with the cylinder
Speed 72 rpm pressure signal, for a certain portion of the whole
Torque 70.1 Nm measurement record. This signal (due to its nature)
Computed results Uncertainty (%) presents a very sharp change from its maximum positive
to a corresponding negative value; by taking the bisection
Fuel volumetric flow rate 71 on these two values, the position of TDC is accurately
Power 71
determined in every cycle. Further, by considering the TDC
Specific fuel consumption 71.5
Efficiency 71.5 position of any two consecutive cycles and knowing the
sampling rate of the measurements (e.g., in the present
case, at 0.5 1CA with a nominal speed of 2000 rpm), the
and exhaust emission levels, constituting the ‘baseline’ that calculation of the accurate engine speed in each cycle from
is compared with the corresponding cases when using the the first to the last one is easily achieved.
ethanol–diesel fuel blends. The same procedure was Pressure data points required in-between the experimen-
repeated for each fuel blend by keeping the same operating tally acquired ones are obtained by interpolation. Then, the
conditions. For every fuel change, the fuel lines were ‘mean’ of the cylinder (indicator) and the fuel pressure
cleaned and the engine was left to run for about 30 min to diagrams are obtained, while a ‘light smoothing’ for the
stabilize at its new condition. pressure signals is applied [51,52,55] that is based on
Owing to the high flexibility of this versatile ‘standard’ performing a four-data points weighted smoothing. This
experimental engine, it was possible to keep constant seems to offer a reasonable compromise between no loss of
(automatically) sensitive operating parameters of the valuable signal information and of relatively smooth values
engine circuits, such as for example the oil and cooling
water temperatures, which can influence mainly the exhaust
emission levels, thus increasing the credibility for compar- Cylinder pressure signal (V)
ing the measured values. The differences in the measured TDC Position signal (V) 6
performance and exhaust emission parameters from the
baseline operation of the engine, i.e., when working with 5
neat diesel fuel, were determined and compared. 4
Engine cycle

6. Experimental heat release analysis 4

2

6.1. General description

3 0
The measured pressure data to be processed for the DI
diesel engine under study concern the closed part of the
thermodynamic cycle. The equations and sub-models for -2
2
developing the thermodynamic analysis using the measured
pressure data for the cylinder (combustion chamber) are -4
described in the following subsections. A spatial uniformity 1
of pressure, temperature and composition in the combus-
tion chamber (single-zone model), at each instant of time, -6
is assumed. The gas medium is assumed to obey the perfect
0
gas law and the diesel fuel to be dodecane (C12H26), which
closely resembles the diesel oil used for high-speed diesel
engines, with a lower calorific value of 43  106 J/kg.
-1

6.2. Preliminary treatment of the experimental pressure 0 500 1000 1500 2000 2500
diagrams and determination of various timings Sampling point number

Fig. 2. Typical cylinder pressure signal with the accompanying signal

A recording is made of the cylinder (indicator) and fuel indicating the TDC position in each cycle, for a certain portion of the
measured pressure data for 10 cycles in a contiguous file, measurement record.
 ARTICLE IN PRESS
1798 C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808

for the first derivative of pressure with respect to crank The difference between the dynamic injection timing and
angle. The sudden changes of the latter one, for the the pump spill timing forms the injection delay.
indicator diagram, would grossly be transferred to the By processing the cylinder pressure diagram, as shown
‘sensitive’ heat release rate diagram (cf. Eq. (6)). for example in Fig. 3b, the ignition timing was located at
By processing the fuel pressure diagram, as shown for the crank angle where the first derivative of this pressure
example in Fig. 3a, the static injection timing (at the with respect to crank angle suddenly changes slope,
injector) was determined at the crank angle where this immediately following the event of the dynamic injection
pressure rises suddenly above the residual (in the connect- timing, going from a negative to a positive value and so
ing pipe) pressure value (point ‘S.inj.’). As no apparatus for presenting a local minimum. The ignition timing was then
measuring the injector needle lift was available, the determined either by using this condition, or by locating
dynamic injection timing was assumed to coincide with the corresponding ‘zeroing’ crank angle of the second
the crank angle where the fuel pressure reaches the value of derivative of pressure with respect to crank angle, assuming
the injector nozzle opening pressure, immediately following that this signal is ‘smooth’ enough; this is shown in Fig. 3b
the event of the static injection timing (point ‘D.inj.’). as point ‘S’. Note that with every differentiation of the
original pressure signal the noise-to-signal ratio increases,
while if over-smoothing is applied this ‘zero’ point might
disappear as being ‘ill conditioned’. The difference between
600 Injection pressure (bar) 100 the ignition timing and the dynamic injection timing forms
Cylinder pressure (bar) the ignition delay.
From the first and second derivatives of the cylinder
80 pressure diagram with respect to crank angle, which are
found numerically, the crank angles of the maximum
400 values of the first derivative of pressure and of the pressure
60 itself can be determined, bearing also in mind that they
immediately follow the ignition timing and in that order.
Injector nozzle
opening pressure 40 6.3. Energy conservation and state equations
200
The heat release rate determination procedure is based
20 on the processing of the cylinder pressure (indicator)
diagram. The properties of the gas content are considered
constant inside each time step, with their values calculated
0 S.inj. D.inj.
0 as described in Section 6.4.
-40 -20 0 20 40 It is understood that all the differentiations to follow
CRANK ANGLE (deg.) with respect to time ‘t’ can be converted to differentiations
Pressure (bar) with respect to degrees crank angle ‘j’, by multiplying by
1st Press. Deriv. (bar/deg) the factor (where the engine speed ‘N’ is expressed in rpm)
100 2nd Press. Deriv. (bar/deg**2) 2 6
dt 1
¼ . (2)
dj 6N
80 4
By applying the first law of thermodynamics in
differential form, for an open system, the net heat release
1
rate is determined as [1,2,58]:
60 2
dQn dme dU dV
 he ¼ þp , (3a)
dt dt dt dt
40 0 where the dme/dt ¼ (dme/dj)6N term stands for the (out-
S
0 flowing) blowby rate, calculated in Section 6.6, having the
specific enthalpy he of the cylinder contents. The terms
20 -2
U and he are defined as the gas sensible internal energy and
sensible specific enthalpy, respectively. The corresponding
sensible term for the fuel, hf (dmf/dt), has been omitted
0 -1 -4
from the above equations as being very small.
-40 -20 0 20 40
Thus, the net heat release rate, so computed, represents
CRANK ANGLE (deg.)
the difference between the combustion energy-release rate
Fig. 3. Determination of static (point ‘S.inj.’) and dynamic (point ‘D.inj.’) (gross) and the rate of heat transferred (loss) to the
injection timings in (a), and of ignition timing (point ‘S’) in (b). chamber walls.
 ARTICLE IN PRESS
C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808 1799

Considering the working medium as a perfect gas, with 6.4. Combustion reaction and working medium properties
m the instantaneous cylinder mass, Eq. (3a) results in
In this subsection only the combustion phase is
dT dQn pdV dme addressed, since the compression and expansion phases
mcv ¼   he . (3b)
dt dt dt dt can be considered as simple sub-cases (absence of chemical
The perfect gas state equation for the cylinder gas reaction and constant composition in time for compres-
content reads sion). Although dissociation could be included
[43,44,58,59], it was opted to neglect it for keeping the
pV ¼ mRT ! T ¼ pV =mR (4a) analysis simple. Moreover, in the light of the inherent
uncertainties involved in experimental heat release analysis
or in differential form as calculations, it was felt that its inclusion would not
dV dp dT dm add much.
p þV ¼ mR þ RT . (4b) The basic stoichiometric chemical equation for the fuel
dt dt dt dt
(with chemical formula CnHmOr)—oxygen reaction is
Combining Eqs. (3b) and (4b), we get the following [2,60],
expression for the first law equation:
Cn Hm Or þ ½n þ ðm=4Þ  ðr=2ÞO2 ! nCO2 þ ðm=2ÞH2 O:
pðdV =dtÞ þ V ðdp=dtÞ  RTðdm=dtÞ
cv (10)
R
dQn dV dme The number of kmoles bi of the constituents (i ¼ 1,2,3,4
¼ p  he . ð5Þ
dt dt dt and 5 for CO2, H2O, O2, N2 and CnHmOr, respectively) at
any time t (or corresponding crank angle j), for complete
Substituting the value of T from Eq. (4a) into Eq. (5), we
combustion in air, are calculated from the following
can calculate the net heat release rate from
expressions:
 
dQn cv dV dp pV dm dV dme
¼ p þV  þp þ he . (6) b1 ¼ nwf ; b2 ¼ ðm=2Þwf ; b3 ¼ b1  b2 =2 þ wf G=F,
dt R dt dt m dt dt dt
b4 ¼ 3:76wf G=F; b5 ¼ w0f  wf , ð11aÞ
The instantaneous mass m in the cylinder, needed in
Eqs. (4a) and (6), at any crank angle j (corresponding ‘t’) where G ¼ n+(m/4)(r/2), and wf ¼ mfb/Mf is the number
is calculated from of kmoles of fuel burned up to that time out of a total of
Z j w0f, with b5 ¼ 0 when Fp1.
dme 1 The fuel–air equivalence ratio F, i.e., the actual fuel–air
mðjÞ ¼ mðjivc Þ  dj, (7)
jivc dt 6N ratio (by mass) divided by its stoichiometric value, is
computed with the cumulative values (up to the crank
where a computation of mass (air and possibly residual angle in question) of the air and fuel burned (calculated in
gases) is accomplished at the point of IVC (inlet valve the previous subsection), as follows:
closure), with known pressure and temperature there for
the engine in hand. F ¼ ½mfb =ðm  mfb Þ=FAst (11b)
Therefore, the corresponding gross heat release rate,
which is the energy released from the combustion of fuel, is with
given by
FAst ¼ ð12n þ m þ 16rÞ=½4:76ðn þ m=4  r=2Þ28:96.
dQg dQn dQw (11c)
¼ þ , (8)
dt dt dt
The specific internal energies (considering only their
where the ‘dQw/dt’ term stands for the rate of heat sensible part) of the components are given as 4th order
transferred to the walls of the combustion chamber, which polynomial expressions of the absolute temperature [60].
is calculated as described in Section 6.5. Similar expressions are then derived for the specific
Obviously, by knowing the fuel lower calorific value Y, enthalpies, specific heat capacities and specific heat
the fuel burned mass rate in the chamber is computed from capacities ratio, by applying the simple thermodynamic
dmfb 1 dQg relations connecting these quantities for a perfect gas
¼ . (9) [1,58]. The mixture properties in the chamber are then
dt Y dt
computed by knowing the prevailing gas composition
If now we integrate Eqs. (8) and (9), from the point of as calculated from Eqs. (11a)–(11c) and the gas tempera-
IVC event up to any crank angle ‘j’ (corresponding ‘t’) of ture T calculated from Eq. (4a). Owing to the small
the closed part of the cycle (along the lines of Eq. (7)), we variation of the mixture molecular weight ‘M’ with
can obtain the respective cumulative values in the chamber composition, the mixture specific constant R ¼ Rmol/M
of the heat released by the fuel Qg and the fuel burned mfb. was considered as constant.
 ARTICLE IN PRESS
1800 C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808

6.5. Gas heat transfer sub-model where atmospheric pressure is assumed to exist. The value
of the discharge coefficient cd is chosen in a way to match
The heat transfer between the chamber trapped mass and the measured blowby rate from the engine. The analysis is
surrounding walls is calculated by using the formula of taken to include also the effect of the mass flow rate
Annand [61], with the heat flux given by leakage through the engine valves (a relatively very
small amount), without making its distinction from the
dQw =dt kgas
¼a ðReÞb ðT  T w Þ þ c  ðT 4  T 4w Þ, (12) blowby one.
A D
where A is the heat transfer surface area of the combustion
6.7. Cylinder volume and its rate of change against crank
chamber walls, Re ¼ rgasDup/vgas is the Reynolds number,
angle
up ¼ 2Ns/60 is the mean piston speed, and kgas, vgas are the
gas thermal conductivity and dynamic viscosity, respec-
The instantaneous volume in the cylinder is given by
tively, which are available as polynomial expressions of the
[1,2,59,63]
gas temperature T. qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
In the above equation, exponent b ¼ 0.75 and constant
V ¼ V c þ Apc r½ð1  cos fÞ þ l  l2  sin2 j, (14a)
1
c¼0 for the compression period, otherwise
c ¼ 3.3  108 W/m2 K4, which are their most usual values where Vc is the cylinder clearance volume, r ( ¼ s/2) the
[44,60]. Finally, constant ‘a’ is evaluated by iteration, in a crank radius, Apc the piston cross section area, and l the
way that the total fuel burned in the chamber, mfbtot, found ratio of crank radius to piston rod length.
by integration, from the IVC event to the EVO event of the Its rate of change against crank angle is,
closed part of the cycle, of Eq. (9) coincides with its qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
corresponding experimental value mfc/(N/120), where mfc is dV =df ¼ Apc r½sin jð1 þ l cos j= 1  l2 sin2 jÞ. (14b)
the measured fuel consumption rate.
Although it is expected that the value of the ‘c’
(radiation) constant is lower for the cases of ethanol
blends due to their lower flame luminosity, it is kept the 6.8. Numerical solution procedure and computer program
same with any differences incorporated in the value of the
‘a’ constant. The equations of the thermodynamic model, presented in
the previous subsections, are solved numerically step-by-
step using a simple time marching technique, with a time
6.6. Blowby rate sub-model
step size corresponding to 0.51CA. The differential
equations are converted to algebraic ones using a simple
For the evaluation of the mass blowby rate (of total
Euler scheme [64].
amount of 1% of the cylinder full charge) a simple model is
Before the start of calculations, the design characteristics
adopted, by assuming an equivalent blowby area the
of the engine are inputted into the program. The
greatest part of which is the piston ring gap. It was felt that
computations start at IVC event and end at EVO event.
the use of more elaborate models, such as for example the
At the IVC event, reasonable estimates must be made for
ones developed by the present research group taking into
the prevailing pressure and temperature in the cylinder.
account the piston rings motion [62], would not add much
The pressure is taken equal to the atmospheric one [2,65].
to this analysis.
This fact was used to ‘reference’ the pressure transducers,
The well-known relation for gas mass flow through a
taking as the value of point at IVC event the average of 5
restriction is used, accounting for one-dimensional, quasi-
points before and 5 after its position, in order to minimize
steady, compressible flow [1,63], as follows:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi the potential error of relying on only one measured point
u "   ðgþ1Þ=g # [51,55]. The corresponding temperature was selected 30 K
dme pu u t 2gR u T u p d
2=g
p
¼ cd F  d higher than the atmospheric one [65]. The associated
dt Ru T u g1 pu pu computer program was written in Fortran V language and
(13a) executed on a Pentium-IV Personal Computer.

for pd/pu4[2/(g+1)]g/(g1), while for the opposite case it

holds 7. Presentation and discussion of the experimental heat
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi release (combustion) analysis results
 
dme pu 2 ðgþ1Þ=ðg1Þ
¼ cd F gRu T u , (13b) It is pointed out that as far as the heat release analysis is
dt Ru T u gþ1
concerned, only the results for the neat diesel fuel and the
where index ‘d’ denotes downstream and index ‘u’ up- blend of 15% (by volume) ethanol in diesel fuel, denoted
stream conditions. hereafter as E15-D, will be shown. This is dictated, on the
In the above equations, F is the geometric flow area of one hand, by the conservation of space of this paper and,
the passageway connecting the cylinder and the crankcase on the other hand, by the differences in the respective
 ARTICLE IN PRESS
C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808 1801

diagrams (of neat diesel fuel against the E15-D blend) that 100
are relatively well discernible in this case.
The results of this section refer to all four loads
considered, corresponding to brake mean effective pres-
80
sures of 0.00 (no-load), 1.40, 2.57 and 5.37 bar, respec-

CYLINDER PRESSURE (bar)

tively. It is understood that all the pressure diagrams to
appear in this section are ‘mean-smooth’, as explained in
Section 6.2, which are then processed to produce the other 60
related parameters.
Fig. 4 shows, at the four loads considered, the fuel
(injection) pressure against crank angle diagrams, for the 40
neat diesel fuel and the E15-D blend. First it can be seen, as
expected, that the injection duration increases with engine
load for both fuels; the same holds true for the injection
pressures. It can be observed that for each load considered, 20
the fuel pressure diagram for the E15-D blend against the
corresponding one for the neat diesel fuel case does not
show any appreciable differences in shape and thus in the 0
injection duration or maximum pressure. The E15-D blend -20 0 20 40
diagram shows, however, a small displacement-delay (that CRANK ANGLE (deg.)
is also translated into a corresponding delay of the dynamic
injection timing, at most of 11), which is attributed to the Fig. 5. Cylinder pressure against crank angle diagrams, at the four loads,
different densities and bulk moduli of elasticity (cf. values for the neat diesel fuel and the 15% ethanol blend.
in Table 2) influencing the whole injection process [17],
while the static injection timing (at pump spill) was kept fuel and the E-15D blend, focusing on their part around
constant. This is why the explanations of the observed TDC. First it can be seen, as expected, that the pressures
behavior of the emissions parameters tested, given in the increase with load (with the compression line remaining the
next section, consider the injection effects of secondary same), while the ignition delay decreases with engine load
importance. for both fuels. The latter is mainly due to the increasing gas
Fig. 5 shows, at the four loads considered, the cylinder temperatures with load. One can observe that for each load
pressure against crank angle diagrams, for the neat diesel considered, the ignition delay for the E15-D blend is higher
(the pressure rise due to combustion starts later) than the
corresponding one for the neat diesel fuel case, while there
is no appreciable difference in the maximum pressure.
600
It is worth explaining this behavior also with reference to
Fig. 6 (composed of (a) and (b) with two loads each, for
clarity), which shows, at the four loads considered, the
corresponding gross heat release rate against crank angle
diagrams, for the neat diesel fuel and the E15-D blend.
First it can be seen, as expected, that the ignition delay
FUEL PRESSURE (bar)

400 decreases with engine load for both fuels, while the heat
release rate values become higher. For the higher loads,
both parts of combustion, i.e., the premixed combustion
(the part under the first ‘abrupt peak’) and the diffusion
combustion (the last part under the second ‘rounded
peak’), are apparent, with the diffusion combustion
200 diminishing with load decrease.
One can again observe, from this figure, that for each
load considered, the ignition delay for the E15-D blend is
higher than the corresponding one for the neat diesel fuel
case, while its premixed combustion peak is much higher
and sharper. It is the lower cetane number of ethanol
0
(cf. values in Table 2) that causes the increase of ignition
-40 -20 0 20
delay and so the increased amount of ‘prepared’ fuel (to
CRANK ANGLE (deg.)
this end may also assist the easier evaporation of ethanol)
Fig. 4. Fuel (injection) pressure against crank angle diagrams, at the four for combustion after the start of ignition. However, as
loads, for the neat diesel fuel and the 15% ethanol blend. noticed in Fig. 5, this is not translated into higher pressures,
 ARTICLE IN PRESS
1802 C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808

50 2000
GROSS HEAT RELEASE RATE (J/deg.)

40
1600

TEMPERATURE (K)
30
1200

20

800

10

400
0

0
-10
-20 0 20 40
-20 0 20 40 60 80
CRANK ANGLE (deg.)
CRANK ANGLE (deg.)
50 Fig. 7. Cylinder temperature against crank angle diagrams, at the four
loads, for the neat diesel fuel and the 15% ethanol blend.
GROSS HEAT RELEASE RATE (J/deg.)

40

a temperature increase with engine load for both fuels. One

30 can observe that for each load considered, the temperatures
for the E15-D blend are slightly lower than the correspond-
ing ones for the neat diesel fuel case. This is due to the
20 engine running overall a little ‘leaner’ with the E15-D
blend, as revealed later in Fig. 10, and possibly to its higher
heat of evaporation and lower calorific value (cf. values in
10 Table 2). It is stated, however, that this is a computed
‘mixed’ temperature due to the inherent single-zone
modeling assumptions of the heat release analysis followed,
0 so that these results must be interpreted with caution.
Actually, as reported in Ref. [40], measured in-cylinder
temperatures during the combustion stage showed much
-10
lower temperatures for the E15-D blend. It must be the
-20 0 20 40 60 80 single-zone modeling assumption that ‘rounds’ these
CRANK ANGLE (deg.) results, although it manages to capture qualitatively
Fig. 6. Gross heat release rate against crank angle diagrams at the four correctly this mechanism.
loads (composed of (a) and (b) with two loads each), for the neat diesel Fig. 8 shows, at the four loads considered, the gross
fuel and the 15% ethanol blend. cumulative heat release against crank angle diagrams, for the
neat diesel fuel and the E15-D blend. One can observe that
for each load considered, the gross cumulative heat release
probably because of the counteracting effect of later curve for the E15-D blend lies a little lower than the
combustion in a lower temperature environment. In Ref. corresponding one for the neat diesel fuel case. Given the
[39] a corresponding decrease of pressure has been same engine load and speed, this is translated into a better
reported, but the injection conditions were over-retarded fuel utilization, i.e., a better thermal efficiency (cf. Eq. (1d)).
in that work (after TDC), so that the effect of later Fig. 9 shows, at the four loads considered, the
combustion in a dropping temperature environment was cumulative heat loss to the combustion chamber walls
actually the dominating one. against crank angle diagrams, for the neat diesel fuel and
Fig. 7 shows, at the four loads considered, the cylinder the E15-D blend. One can observe that for each load
temperature against crank angle diagrams, for the neat considered, the cumulative heat loss curve for the E15-D
diesel fuel and the E15-D blend, focusing on their part blend lies a little lower than the corresponding one for the
around TDC. First it can be seen, as expected, that there is neat diesel fuel case. This is mainly due to the lower
 ARTICLE IN PRESS
C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808 1803

800 0.6
CUMULATIVE GROSS HEAT RELEASE (J)

600

EQUIVALENCE RATIO
0.4
400

200

0.2

-200
0
-20 0 20 40 60 80
-40 -20 0 20 40 60
CRANK ANGLE (deg.)
CRANK ANGLE (deg.)
Fig. 8. Cumulative gross heat release against crank angle diagrams, at the
four loads, for the neat diesel fuel and the 15% ethanol blend. Fig. 10. Fuel–air equivalence ratio against crank angle diagrams, at the
four loads, for the neat diesel fuel and the 15% ethanol blend.

300 proves that the engine runs overall a little ‘leaner’ with the
E15-D blend, for the same engine load and speed
conditions. It is stated that the calculation of F was based
on Eqs. (11b) and (11c) by considering all the fuel-bound
oxygen, hoping this is released in crucial local ‘rich’ areas
CUMULATIVE HEAT LOSS (J)

200
(zones).

8. Discussion of the performance and exhaust emissions

100
results and their interpretation

Based on the very interesting combustion characteristics

of the ethanol–diesel fuel blends revealed by the results of
the previous section, derived after the application of the
0
experimental heat release analysis, the performance and
exhaust emission measurements results, for the present
engine, can be interpreted as exposed in the following.
They refer to measurements only for the three load cases
-100 and not the no-load one.
-80 -40 0 40 80 120 Fig. 11 shows in a bar chart diagram, at the three loads
CRANK ANGLE (deg.) considered, the exhaust smoke (soot) density for the neat
Fig. 9. Cumulative heat loss to the cylinder walls against crank angle
diesel fuel and the various percentages of the ethanol in its
diagrams, at the four loads, for the neat diesel fuel and the 15% ethanol blends with diesel fuel. One can observe that the soot
blend. emitted by the ethanol–diesel fuel blends is significantly
lower than that for the corresponding neat diesel fuel case,
with the reduction being higher the higher the percentage
cylinder temperatures involved with the ethanol blend case. of ethanol in the blend. This may be attributed to the
This contributes to a better thermal efficiency too. engine running overall ‘leaner’, with the combustion being
Fig. 10 shows, at the four loads considered, the fuel–air now assisted by the presence of the fuel-bound oxygen of
equivalence ratio F against crank angle diagrams, for the the ethanol even in locally rich zones (cf. Fig. 10), which
neat diesel fuel and the E15-D blend. One can observe seems to have the dominating influence [8,35]. It was
that for each load considered, the fuel–air equivalence reported in Refs. [40,41], that in-cylinder combustion
ratio curve for the E15-D blend lies a little lower than photography showed indeed lower luminosity flames for
the corresponding one for the neat diesel fuel case. This the ethanol blends, revealing the lower net soot produced.
 ARTICLE IN PRESS
1804 C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808

160 ethanol in its blends with diesel fuel. One can observe that
Diesel
the NOx emitted by the ethanol blends are slightly lower
than those for the corresponding neat diesel fuel case, with
5% Ethanol the reduction being higher the higher the percentage of
10% Ethanol ethanol in the blend. This may be attributed to the engine
120 running overall ‘leaner’ and the temperature lowering effect
15% Ethanol of the ethanol having the dominating influence (cf. Figs. 7
and 10), against the opposing effect of the lower cetane
Soot Density (mg/m3)

number (and thus longer ignition delay) of the ethanol

leading possibly to higher temperatures during the
80 premixed combustion part. This is a delicate balance
weighting more on the one or the other side, depending on
the specific engine and its operating conditions [21], which
for example may be disclosed by the details of a successful
multi-zone model [48].
Fig. 13 shows in a bar chart diagram, at the three loads
40
considered, the carbon monoxide (CO) exhaust emissions
for the neat diesel fuel and the various percentages of the
ethanol in its blends with diesel fuel. One can observe that
the CO emitted by the ethanol–diesel fuel blends is lower
than that for the corresponding neat diesel fuel case, with
0
the reduction being higher the higher the percentage of
b.m.e.p. : 1.40 bar 2.57 bar 5.37 bar
ethanol in the blend. Conclusively, the emitted CO follows
Fig. 11. Bar chart diagram of the emitted soot (smoke) density, at the the same behavior as the emitted soot by the engine, a fact
three loads, for the neat diesel fuel and the 5%, 10% and 15% ethanol collectively attributed to the same physical and chemical
blends.
mechanisms affecting almost in the same way, at least
qualitatively, the net formation of these emissions.
1400 Fig. 14 shows in a bar chart diagram, at the three loads
Diesel considered, the total unburned hydrocarbons (HC) exhaust
5% Ethanol emissions for the neat diesel fuel and the various
1200
10% Ethanol
400
Diesel 10% Ethanol
15% Ethanol
1000
5% Ethanol 15% Ethanol

320
800
NOX (ppm)

600
240
CO (ppm)

400

160
200

0 80
b.m.e.p. : 1.40 bar 2.57 bar 5.37 bar

Fig. 12. Bar chart diagram of the emitted nitrogen oxides (NOx), at the
three loads, for the neat diesel fuel and the 5%, 10% and 15% ethanol
blends.
0
b.m.e.p. : 1.40 bar 2.57 bar 5.37 bar
Fig. 12 shows in a bar chart diagram, at the three loads Fig. 13. Bar chart diagram of the emitted carbon monoxide (CO), at the
considered, the nitrogen oxides (NOx) exhaust emissions three loads, for the neat diesel fuel and the 5%, 10% and 15% ethanol
for the neat diesel fuel and the various percentages of the blends.
 ARTICLE IN PRESS
C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808 1805

120 500
Diesel 10% Ethanol
5% Ethanol 15% Ethanol

Brake specific fuel consumption (g/kW h)

100 400

80 300
HC (ppm)

60 200

40 100

20 0
0 5 10 15
Ethanol in fuel blend (% by vol.)

0 Fig. 15. Brake specific fuel consumption for the neat diesel fuel and the
b.m.e.p. : 1.40 bar 2.57 bar 5.37 bar 5%, 10% and 15% ethanol blends, at the three loads.

Fig. 14. Bar chart diagram of the emitted total unburned hydrocarbons
(HC), at the three loads, for the neat diesel fuel and the 5%, 10% and 15% fuel. The fuel blend mass flow rate is calculated from the
ethanol blends. respective measured volume flow rate value and the blend
density. Since the comparison is made at the same load
percentages of the ethanol in its blends with diesel fuel. One (b.m.e.p.) and speed, which is translated into the same
can observe that the HC emitted by the ethanol–diesel fuel engine power (cf. Eqs. (1a) and (1b)), the brake specific fuel
blends are higher than those for the corresponding neat consumption values are then effectively directly propor-
diesel fuel case, with the increase being higher the higher tional to the fuel mass flow rate (cf. Eq. (1c)). It is observed
the percentage of ethanol in the blend. This is the behavior that for all the ethanol–diesel fuel blends, the specific fuel
reported by almost all investigators on various types of consumption is a little higher than the corresponding diesel
engines and conditions [21]. fuel case, with the increase being higher the higher the
As it is known, the formation of unburned hydrocarbons percentage of ethanol in the blend. This is the expected
originates from various sources in the engine cylinder and behavior due to the lower calorific value of the ethanol
their theoretical study is still at its infancy. It is argued [20] compared to that for the neat diesel fuel, given that the
that, presumably, the increase of HC with the addition of comparison is effected at the same load.
ethanol is due to the higher heat of evaporation of the Fig. 16 shows, at the three loads, the brake thermal
ethanol blends causing slower evaporation and so slower efficiency for the neat diesel fuel and the various
and poorer fuel–air mixing, to the increased spay penetra- percentages of the ethanol in its blends with diesel fuel.
tion causing unwanted fuel impingement on the chamber It is to be noted that the brake thermal efficiency is simply
walls (and so flame quenching) and cushioning in the ring the inverse of the product of the specific fuel consumption
land areas, and to the increase with ethanol of the so-called and the lower calorific value of the fuel (cf. Eqs. (1c)
‘lean outer flame zone’ where flame is unable to exist and (1d)). Given the above-mentioned close interrelation
[1,2,33]. To this end may also contribute the late escape between brake thermal efficiency and specific fuel con-
into the cylinder of the fuel left in the nozzle sac volume, sumption, the results of this figure can be explained. It is
because with the addition of ethanol this is easier observed that for all the ethanol–diesel fuel blends, the
evaporated (as heated) and ‘slipped’ into the cylinder (at brake thermal efficiency is slightly higher than that for the
low velocity, late in the expansion stroke) [1,58]. Further, corresponding neat diesel fuel case, with the increase being
some observed phenomena of cyclic irregularity [30,56,57], higher the higher the percentage of ethanol in the blend.
especially with the higher ethanol content blends (to be This means that the increase of brake specific fuel
reported and analyzed in a following paper), may consumption for the ethanol–diesel blends is lower than
contribute towards this side [1]. the corresponding decrease of the lower calorific value of
Fig. 15 shows, at the three loads considered, the brake the blends. This can be attributed to the higher premixed
specific fuel consumption for the neat diesel fuel and the combustion part possessed by the ethanol blends because
various percentages of the ethanol in its blends with diesel of the lower cetane number of ethanol, leading to higher
 ARTICLE IN PRESS
1806 C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808

0.4 the corresponding neat diesel fuel case (cf. Fig. 7), with this
decrease being higher the higher the percentage of ethanol
in the blend. This is in accordance with the corresponding
trend of slightly higher thermal efficiency, as justified
0.3
above, by denoting a better expansion process; to this side
may also be corroborating the higher latent heat of
Brake thermal efficiency

evaporation (and so lower produced temperatures) of the

ethanol blends against that for the diesel fuel.

0.2 9. Conclusions

An extended experimental study is conducted to evaluate

and compare the use of ethanol as supplement to the
5 10
0.1 conventional diesel fuel at blend ratios (by volume) of 95 , 90
15
and 85 in a high-speed, DI diesel engine located at the
authors’ laboratory.
The series of tests are conducted using each of the above
0 fuel blends, with the engine working at a speed of 2000 rpm
and at four loads. In each test, exhaust smokiness and
0 5 10 15
exhaust regulated gas emissions such as nitrogen oxides,
Ethanol in fuel blend (% by vol.)
carbon monoxide and total unburned hydrocarbons are
Fig. 16. Brake thermal efficiency for the neat diesel fuel and the 5%, 10% measured. Brake specific fuel consumption and thermal
and 15% ethanol blends, at the three loads. efficiency are computed from the measured fuel volumetric
flow rate, and densities and calorific values.
A heat release analysis of the experimentally obtained
400 combustion chamber pressure diagrams is developed and
used, with the pertinent application of the energy and state
equations. From the analysis results, plots of the history in
360 the combustion chamber of the gross heat release rate and
other related parameters reveal some very interesting
Exhaust gas temperature (°C)

features, which shed light on the combustion mechanism

320 when using these blends. It is revealed that with the use of
the ethanol blends against the neat diesel fuel, the ignition
delay is increased, fuel injection pressure diagrams are
280
slightly displaced (delayed), maximum cylinder pressures
are hardly affected, cylinder temperatures are a little
240
decreased, while the engine runs overall a little ‘leaner’.
The differences in the measured performance and
exhaust emission parameters from the baseline operation
200 of the engine, i.e., when working with neat diesel fuel, are
determined and compared. The heat release analysis results
for the relevant combustion mechanism, combined with the
160 widely differing physical and chemical properties of the
0 5 10 15 ethanol against the ones for the diesel fuel, are used to aid
Ethanol in fuel blend (% by vol.) the correct interpretation of the observed engine behavior.
The smoke density was significantly reduced with the use
Fig. 17. Exhaust gas temperature for the neat diesel fuel and the 5%, 10% of the ethanol–diesel fuel blends with respect to that of the
and 15% ethanol blends, at the three loads.
neat diesel fuel, with this reduction being higher the higher
the percentage of ethanol in the blend.
percentage of ‘constant volume’ combustion [36], and to The NOx emissions were slightly reduced with the use of
the lower heat losses and ‘leaner’ combustion discussed in the ethanol–diesel fuel blends with respect to that of the
the previous section. neat diesel fuel, with this reduction being higher the higher
Lastly, Fig. 17 shows, at the three loads considered, the the percentage of ethanol in the blend.
exhaust gas temperature for the neat diesel fuel and the The CO emissions were reduced with the use of the
various percentages of the ethanol in its blends with diesel ethanol–diesel fuel blends with respect to that of the neat
fuel. It is observed that for all the ethanol–diesel fuel diesel fuel, with this reduction being higher the higher the
blends, the temperature is very slightly lower than that for percentage of ethanol in the blend.
 ARTICLE IN PRESS
C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808 1807

On the contrary, the unburned hydrocarbons (HC) [14] Graboski MS, McCormick RL. Combustion of fat and vegetable oil
emissions were increased with the use of the ethanol–diesel derived fuels in diesel engines. Prog Energy Combust Sci 1998;24:
fuel blends with respect to that of the neat diesel fuel, with 125–64.
[15] Crookes RJ, Bob-Manuel KDH. RME or DME: a preferred
this increase being higher the higher the percentage of alternative fuel option for future diesel engine operation. In:
ethanol in the blend. Frangopoulos C, Rakopoulos C, Tsatsaronis G, editors. Proceedings
Concerning the engine performance with the ethanol– of 19th international conference ‘ECOS 2006’, vol. 2, Crete, Greece,
diesel fuel blends against the neat diesel fuel case, with July 12–14, 2006. p. 1105–12.
increasing percentage of the ethanol in the blends, a little [16] Rakopoulos CD. Comparative performance and emission studies
when using olive oil as a fuel supplement in DI and IDI diesel
higher specific fuel consumption was observed with engines. Renew Energy 1992;2:327–31.
corresponding slight increase of brake thermal efficiency [17] Rakopoulos CD, Antonopoulos KA, Rakopoulos DC. Multi-zone
and little lower exhaust gas temperatures. modeling of diesel engine fuel spray development with vegetable
The interesting heat release analysis results, for the oil, bio-diesel or diesel fuels. Energy Convers Manage 2006;47:
relevant combustion mechanism of the ethanol–diesel fuel 1550–73.
[18] Rakopoulos CD, Antonopoulos KA, Rakopoulos DC, Kakaras EC,
blends, will assist to the construction of a reliable multi- Pariotis EG. Characteristics of the performance and emissions of a
zone combustion model (to be reported soon) for further HSDI diesel engine running with cottonseed oil or its methyl ester
explaining the behavior of these blends when used in and their blends with diesel fuel. Int J Vehicle Design 2007, in press.
engines. [19] Demirbas A. Progress and recent trends in bio-fuels. Prog Energy
Combust Sci 2007;33:1–18.
[20] Ecklund EE, Bechtold RL, Timbario TJ, McCallum PW. State-of-
References the-art report on the use of alcohols in diesel engines. Warrendale
(PA): Society of Automotive Engineers Inc; 1984 [SAE Paper no.
[1] Heywood JB. Internal combustion engine fundamentals. New York: 840118].
McGraw-Hill; 1988. [21] Corkwell KC, Jackson MM, Daly DT. Review of exhaust emissions
[2] Stone R. Introduction to internal combustion engines. 3rd ed. of compression ignition engines operating on E diesel fuel blends.
London: Macmillan; 1999. Warrendale (PA): Society of Automotive Engineers Inc; 2003 [SAE
[3] Kouremenos DA, Rakopoulos CD. The operation of a turbulence Paper no. 2003-01-3283].
chamber diesel engine with LPG fumigation for exhaust emissions [22] Likos B, Callahan TJ, Moses CA. Performance and emissions of
control. Verein Deutscher Ingenieure—Forschung im Ingenieurwesen ethanol and ethanol–diesel blends in direct-injected and pre-chamber
1986;52:185–90. diesel engines. Warrendale (PA): Society of Automotive Engineers
[4] Kouremenos DA, Rakopoulos CD, Kotsiopoulos PN. Comparative Inc; 1982 [SAE Paper no. 821039].
performance and emission studies for vapourized diesel fuel and [23] Meiring P, Allan RS, Hansen AC, Lyne PWL. Tractor performance
gasoline as supplements in swirl-chamber diesel engines. Energy and durability with ethanol–diesel fuel. Trans ASAE 1983;26(1):
1990;15:1153–60. 59–62.
[5] Rakopoulos CD, Kyritsis DC. Comparative second-law analysis of [24] Meiring P, Hansen AC, Vosloo AP, Lyne PWL. High concentration
internal combustion engine operation for methane, methanol and ethanol–diesel blends for compression-ignition engines. Warrendale
dodecane fuels. Energy 2001;26:705–22. (PA): Society of Automotive Engineers Inc; 1983 [SAE Paper no.
[6] Rakopoulos CD, Kyritsis DC. Hydrogen enrichment effects on the 831360].
second law analysis of natural and landfill gas combustion in engine [25] Goering CE, Shirvani H, Sorenson SC. Alcohol-heavy distillate
cylinders. Hydrogen Energy 2006;31:1384–93. blends as diesel fuel. Trans ASAE 1983;26(1):2–5.
[7] Rakopoulos CD, Scott MA, Kyritsis DC, Giakoumis EG. [26] Satge de Caro P, Mouloungui Z, Vaitilingom G, Berge JCh. Interest
Availability analysis of hydrogen/natural gas blends combustion of combining an additive with diesel–ethanol blends for use in diesel
in internal combustion engines. In: Frangopoulos C, Rakopoulos C, engines. Fuel 2001;80:565–74.
Tsatsaronis G, editors. Proceedings of 19th international [27] Wrage KW, Goering CE. Technical feasibility of diesohol. Trans
conference ‘ECOS 2006’, vol. 2, Crete, Greece, July 12–14, 2006. ASAE 1980;23(6):1338–43.
p. 983–90. [28] Li D, Zhen H, Xingcai L, Wugao Z, Jianguang Y. Physico-chemical
[8] Xiao Z, Ladommatos N, Zhao H. The effect of aromatic properties of ethanol–diesel blend fuel and its effect on performance
hydrocarbons and oxygenates on diesel engine emissions. Proc Inst and emissions of diesel engines. Renew Energy 2005;30:67–76.
Mech Eng, J Automobile Eng 2000;214:307–32. [29] Schaefer AJ, Hardenberg HO. Ignition improvers for ethanol fuels.
[9] Rakopoulos CD, Hountalas DT, Zannis TC, Levendis YA. Opera- Warrendale (PA): Society of Automotive Engineers Inc; 1981 [SAE
tional and environmental evaluation of diesel engines burning Paper no. 810249].
oxygen-enriched intake air or oxygen-enriched fuels: a review. [30] Saeed MN, Henein NA. Combustion phenomena of alcohols
Trans SAE, J Fuels Lubr 2004;113:1723–43 [SAE Paper no. 2004- in C.I. engines. Trans ASME, J Eng Gas Turbines Power 1989;111:
01-2924]. 439–44.
[10] Rickeard DJ, Thompson ND. A review of the potential for bio-fuels [31] Simonsen H, Chomiak J. Testing and evaluation of ignition
as transportation fuels. Warrendale (PA): Society of Automotive improvers for ethanol in a DI diesel engine. Warrendale (PA):
Engineers Inc; 1993 [SAE Paper no. 932778]. Society of Automotive Engineers Inc; 1995 [SAE Paper no. 952512].
[11] Rakopoulos CD, Antonopoulos KA, Rakopoulos DC, Hountalas [32] Shropshire GJ, Goering CE. Ethanol injection into a diesel engine.
DT, Giakoumis EG. Comparative performance and emissions study Trans ASAE 1982;25(3):570–5.
of a direct injection diesel engine using blends of diesel fuel with [33] Hayes TK, Savage LD, White RA, Sorenson SC. The effect of
vegetable oils or bio-diesels of various origins. Energy Convers fumigation of different ethanol proofs on a turbocharged diesel
Manage 2006;47:3272–87. engine. Warrendale (PA): Society of Automotive Engineers Inc; 1988
[12] Hansen AC, Zhang Q, Lyne PWL. Ethanol–diesel fuel blends—a [SAE Paper no. 880497].
review. Bioresource Technol 2005;96:277–85. [34] Noguchi N, Terao H, Sakata C. Performance improvement by
[13] Koerbitz W. Biodiesel production in Europe and North America, an control of flow rates and diesel injection timing on dual-fuel engine
encouraging prospect. Renew Energy 1999;16:1078–83. with ethanol. Bioresource Technol 1996;56:35–9.
 ARTICLE IN PRESS
1808 C.D. Rakopoulos et al. / Energy 32 (2007) 1791–1808

[35] Boruff PA, Schwab AW, Goering CE, Pryde EH. Evaluation of diesel formation in a direct injection diesel engine running with vegetable oil
fuel—ethanol microemulsions. Trans ASAE 1982;25(1):47–53. or its bio-diesel. Energy Convers Manage 2007;48 in press.
[36] Hansen AC, Taylor AB, Lyne PWL, Meiring P. Heat release in the [50] Rakopoulos CD, Hountalas DT. Net and gross heat release rate
compression-ignition combustion of ethanol. Trans ASAE 1989; calculations in a DI diesel engine using various heat transfer models.
32(5):1507–11. In: Proceedings of ASME-WA meeting, vol. 33, Chicago, Illinois,
[37] Ajav EA, Singh B, Bhattacharya TK. Experimental study of some AES, 1994. p. 251–62.
performance parameters of a constant speed stationary diesel engine [51] Homsy SC, Atreya A. An experimental heat release rate analysis of a
using ethanol–diesel blends as fuel. Biomass Bioenergy 1999;17: diesel engine operating under steady state conditions. Warrendale
357–65. (PA): Society of Automotive Engineers Inc; 1997 [SAE Paper no.
[38] Can O, Celikten I, Usta N. Effects of ethanol addition on 970889].
performance and emissions of a turbocharged indirect injection [52] Theobald MA, Alkidas AC. On the heat-release analysis of diesel
diesel engine running at different injection pressures. Energy Convers engines: effects of filtering of pressure data. Warrendale (PA): Society
Manage 2004;45:2429–40. of Automotive Engineers Inc; 1987 [SAE Paper no. 872059].
[39] Xingcai L, Jianguang Y, Wugao Z, Zhen H. Effect of cetane number [53] Kort RT, Mansouri SH, Heywood JB, Ekchian A. Divided-chamber
improver on heat release rate and emissions of high speed diesel diesel engine, Part II: experimental validation of a predictive cycle—
engine fuelled with ethanol–diesel blend fuel. Fuel 2004;83:2013–20. simulation and heat release analysis. Warrendale (PA): Society of
[40] Xingcai L, Zhen H, Wugao Z, Degang L. The influence of ethanol Automotive Engineers Inc; 1982 [SAE Paper no. 820274].
additives on the performance and combustion characteristics of diesel [54] Kouremenos DA, Rakopoulos CD, Hountalas DT. Experimental
engines. Combust Sci Technol 2004;176:1309–29. and computational determination of heat release rates of an IDI
[41] Chen H, Shuai S, Wang J. Study on combustion characteristics and diesel engine. In: Conference ‘Autotech 91’, Birmingham, UK,
PM emission of diesel engines using ester–ethanol–diesel blended November 12–15, 1991, Institution of Mechanical Engineers,
fuels. Proc Combust Inst 2007;31:2981–9. p. 9–15. [Paper no. C427/28/123].
[42] Rakopoulos CD, Antonopoulos KA, Rakopoulos DC, Hountalas [55] Rakopoulos CD, Hountalas DT, Rakopoulos DC, Giakoumis EG.
DT, Andritsakis EC. Study of the performance and emissions of a Experimental heat release rate analysis in both chambers of an
high-speed direct injection diesel engine operating on ethanol–diesel indirect injection turbocharged diesel engine at various load and
fuel blends. Int J Alternative Propul 2007;1:309–24. speed conditions. Trans SAE, J Engines 2005;114:867–82 [SAE Paper
[43] Rakopoulos CD, Hountalas DT. Development and validation of a no. 2005-01-0926].
3-D multi-zone combustion model for the prediction of DI diesel [56] Sczomak DP, Henein NA. Cycle-to-cycle variation with low ignition
engines performance and pollutants emissions. Trans SAE, J Engines quality fuels in a CFR diesel engine. Trans SAE 1979;88:3124–44
1998;107:1413–29 [SAE Paper no. 981021]. [SAE Paper no. 790924].
[44] Rakopoulos CD, Rakopoulos DC, Kyritsis DC. Development and [57] Kouremenos DA, Rakopoulos CD, Kotsos KG. A stochastic-
validation of a comprehensive two-zone model for combustion and experimental investigation of the cyclic pressure variation in a DI
emissions formation in a DI diesel engine. Energy Res 2003;27: single-cylinder diesel engine. Energy Res 1992;16:865–77.
1221–49. [58] Ferguson CR. Internal combustion engines. New York: Wiley; 1986.
[45] Rakopoulos CD, Rakopoulos DC, Giakoumis EG, Kyritsis DC. [59] Way RJB. Methods for determination of composition and thermo-
Validation and sensitivity analysis of a two-zone diesel engine model dynamic properties of combustion products for internal combustion
for combustion and emissions prediction. Energy Convers Manage engine calculations. Proc Inst Mech Eng 1977;190(60/76):687–97.
2004;45:1471–95. [60] Benson RS, Whitehouse ND. Internal combustion engines. Oxford:
[46] Rakopoulos CD, Giakoumis EG. Second-Law analyses applied to Pergamon; 1979.
internal combustion engines operation. Prog Energy Combust Sci [61] Annand WJD. Heat transfer in the cylinders of reciprocating internal
2006;32:2–47. combustion engines. Proc Inst Mech Eng 1963;177:973–90.
[47] Rakopoulos CD, Giakoumis EG. Review of thermodynamic diesel [62] Kouremenos DA, Rakopoulos CD, Kotsos KG. A mathematical
engine simulations under transient operating conditions. Trans SAE, simulation of the piston crevice flow and blowby in internal
J Engines 2006;115 [SAE Paper no. 2006-01-0884]. combustion diesel engines. Entropie 1993;174/175:51–60 [in French].
[48] Rakopoulos CD, Antonopoulos KA, Rakopoulos DC, Hountalas [63] Watson N, Janota MS. Turbocharging the internal combustion
DT. Multi-zone modeling of combustion and emissions formation in engine. London: MacMillan; 1982.
a DI diesel engine operating on ethanol–diesel fuel blends. Energy [64] Conte SD, DeBoor C. Elementary numerical analysis: an algorithmic
Convers Manage 2007;48 submitted for publication. approach. 2nd ed. New York: McGraw-Hill; 1972.
[49] Rakopoulos CD, Antonopoulos KA, Rakopoulos DC. Development [65] Zinner K. Supercharging of internal combustion engines. Berlin:
and application of a multi-zone model for combustion and pollutants Springer; 1978.