Energy Conversion and Management 123 (2016) 470–486

Contents lists available at ScienceDirect

Energy Conversion and Management

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

Combined effect of injection timing and exhaust gas recirculation (EGR)

on performance and emissions of a DI diesel engine fuelled with next-
generation advanced biofuel – diesel blends using response surface
methodology
B. Rajesh Kumar a,b,⇑, S. Saravanan c, D. Rana d, A. Nagendran e
a
Research Centre, Department of Mechanical Engineering, Sri Venkateswara College of Engineering, Chennai, TN, India
b
Department of Mechanical Engineering, Jeppiaar Institute of Technology, Chennai, TN, India
c
Engine Research Lab, Department of Mechanical Engineering, Sri Venkateswara College of Engineering, Chennai, TN, India
d
Department of Chemical and Biological Engineering, University of Ottawa, Ottawa, Ontario K1N 6N5, Canada
e
PG & Research Department of Chemistry, Polymeric Materials Research Lab, Alagappa Govt. Arts College, Karaikudi, TN, India

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

Article history: Advanced biofuels like dimethyl-carbonate (DMC), isobutanol and n-pentanol can be derived from non-
Received 4 April 2016 food based biomass feedstock without unsettling food supplies and biodiversity. An experimental and
Received in revised form 22 June 2016 statistical investigation was carried out to analyze the effects of injection-timing and exhaust gas recir-
Accepted 23 June 2016
culation (EGR) on performance and emissions of a DI diesel engine using advanced biofuel/diesel blends
Available online 30 June 2016
(containing 8 wt% oxygen). Engine characteristics were measured under high-load condition using mod-
erate EGR (up to 30%) and injection-timing modification (up to ±2°CA bTDC) for controlling charge-
Keywords:
dilution and combustion-phasing. Multiple regression models developed using response surface method-
Emissions
n-Pentanol
ology (RSM) for measured responses like nitrogen oxides (NOx), smoke opacity and brake specific fuel
Isobutanol consumption (BSFC) were found to be statistically significant by analysis of variance (ANOVA).
DMC Interactive effects between injection timing and EGR for all blends were analyzed using response surface
Diesel engine plots that were fitted using developed models. Optimization was performed using desirability approach
Performance of the RSM to minimize NOx and smoke emissions simultaneously with minimum BSFC. Isobutanol/die-
RSM sel blend injected at 22°CA bTDC without EGR was predicted to be optimum for the tested engine.
Confirmatory tests validated that the models developed using RSM are adequate to describe the effects
of the injection timing and EGR on performance and emissions characteristics using all blends as the error
in prediction is within 5%.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction and smoke remains as a main challenge in modern diesel engine

Diesel engines are the most preferred in major global sectors
like industry, transport, agriculture and power generation because
of its unique combination of higher torque capability, higher fuel
conversion efficiency, higher durability and reliability when com-
pared to gasoline engines [1]. However, diesel engines emit high
NOx and smoke emissions due to their lean burning nature and
fuel-rich local regions in the diffusion controlled combustion. To
add to the woes, the inherent trade-off relation between NOx
⇑ Corresponding author at: Department of Mechanical Engineering, Jeppiaar
Institute of Technology, Chennai, TN, India.
E-mail addresses: rajesh_thermal@yahoo.com (B. Rajesh Kumar), saran@svce.ac.in
(S. Saravanan), rana@uottawa.ca (D. Rana), nagimmm@yahoo.com (A. Nagendran).
research and development. NOx is acknowledged to be the major
cause for smog [2], ground-level ozone [3] and acid rain [4]. Smoke
is found to be carcinogenic [5,6] and its continuous exposure can
cause allergy [7], asthma [8], hypertension [9], cardio-respiratory
diseases [10] and lung cancer [11] to human.
In view of the threats to both environment and human race,
government agencies across the world are piling up pressure by
imposing rigorous emission standards. Diesel engine researchers
often resort to emission reduction by (i) engine design modifica-
tions, (ii) diesel fuel reformulation, (iii) employing new combus-
tion strategies and (iv) using after-treatment devices like diesel
oxidation catalysts, diesel particulate filters and selective catalytic
reduction. Various methods that were specifically employed to
reduce NOx emissions include exhaust gas recirculation (EGR),

http://dx.doi.org/10.1016/j.enconman.2016.06.064
0196-8904/Ó 2016 Elsevier Ltd. All rights reserved.
 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486 471

Nomenclature

Abbreviations ROPR rate of pressure rise

A(B)BDC after (before) bottom dead centre RSM Response Surface Methodology
A(B)TDC after (before) top dead centre SOI start of injection
ANOVA Analysis of Variance TDC top dead centre
BMEP brake mean effective pressure ULSD ultra-low sulfur diesel
ASTM American Society for Testing and Materials
BSFC brake specific fuel consumption Symbols
BTE brake thermal efficiency vol% blending percentage per volume of diesel
CA crank angle wt% blending percentage per weight of diesel
CAS Chemical Abstract Service wt weights assigned for a response
CO carbon monoxide Adj-R2 adjusted R2 value
CO2 carbon dioxide di desirability value of the response
CRDI common rail direct ignition e EGR
DI direct injection F value of Fisher-statistic test
DMC di-methyl carbonate f fuel as a categoric factor
DMC15 15% dimethyl carbonate + 85% ULSD blend high upper limit of a response
EGR exhaust gas recirculation i response
EPA Environmental Protection Agency low lower limit of a response
HC hydrocarbons P percentage contribution
HRR heat release rates Pred-R2 predicted R2 value
ISB38 38% isobutanol + 62% ULSD blend p probability value
NOx nitrogen oxides R2 coefficient of determination
PEN45 45% n-pentanol + 55% ULSD Blend r relative importance of a response
PM particulate matter Ti target value of a response
RFS Renewable Fuel Standard t injection timing
RIN Renewable Identification Numbers Yi value of the response

retarding fuel injection timing, low injection pressure, split injec-
tion, modifying combustion chamber geometry, water injection
and excessive cooling of intake air [12]. Employing EGR suppresses
peak combustion temperatures of fuel/air mixtures which reduces
NOx formation and controls the combustion phasing [13]. As doc-
umented earlier in several studies, the effects of EGR include the
thermal effect (that lowers peak combustion temperature by the
way of recirculating non-reacting high specific heat gases, CO2
and H2O), the dilution effect (as a result of low O2 concentration)
and chemical effect (as a result of dissociation of CO2 and H2O dur-
ing combustion). The combined results of these effects are low
combustion temperature and long ignition delay [14]. On the flip-
side, it can deteriorate combustion and eventually decrease the
performance of the engine due to high levels of HC and CO emis-
sions. Retarding the injection timing also affects the ignition delay
and could result in further reduction of NOx emissions with a slight
penalty in smoke and fuel consumption [15,16].
Diesel reformulation with biofuels is an attractive option
because (i) it reduces fossil diesel dependence and improves the
renewable fraction in the fuel and (ii) it reduces smoke emissions
due to the presence of fuel-bound oxygen. Bio-diesel or bio-
alcohol addition to diesel alters diesel fuel properties like cetane
number, calorific value, viscosity, density and air/fuel ratio which
influences the combustion, emissions and performance character-
istics of the engine. NOx emissions generally increased with bio-
diesel addition to diesel [17]. Bio-alcohols (like ethanol and
methanol) have low energy content and low cetane number [18].
Hence, bio-alcohol addition to diesel lowers the cetane number
of diesel which directly influences NOx emissions and engine per-
formance [19].
This study proposes the use of three advanced biofuels with low
cetane number that includes two higher alcohols (isobutanol and
n-pentanol) and a carbonyl ester (di-methyl carbonate) as blending
components with diesel. Advanced biofuels are derived from non-
food based sources such as stalks, leaves and husks that are left
behind after cultivation of food crops, non-food crops like
switch-grass, water-hyacinth, agave, algae, duckweed and indus-
trial wastes like paper, wood chips and crude glycerine. Currently,
industrial microorganisms (like Escherichia coli and Saccharomyces
cerevisiae) and photosynthetic organisms like cyanobacteria are
engineered to act upon these non-food-based sources and secrete
molecules similar to fossil fuels, called ‘advanced’ or ‘drop-in’ bio-
fuels [20]. Isobutanol and n-pentanol can be both synthesized from
cellulose by modern fermentation processes using new strains of
Clostridium species [21] and by biosynthesis using genetically
engineered micro-organisms like E. Coli [22,23], Cyanobacteria
[24] and S. Cerevisiae [25]. Di-methyl carbonate (DMC) is a green
advanced biofuel produced by using a feedstock of methanol and
waste CO2 obtained from power stations [26,27].
There are some difficulties in using fuels like gasoline, DMC and
alcohols in compression ignition engines owing to their low cetane
number (low reactivity and long ignition delay), low calorific value
(less energy content), high volatility, high latent heat of vaporiza-
tion and high HC, CO emissions [28]. However, these low reactive
biofuels can be suitable for reducing smoke and NOx emissions
simultaneously because when blended with diesel they offer (a)
high resistance to auto-ignition, (b) extend the ignition delay per-
iod (due to low cetane number) for sufficient fuel-air mixing and
(c) faster vaporization (due to high volatility) for faster mixing rate
[29].
In general, low NOx and smoke emissions can be realized by
using moderate EGR rates, early or late injection, and diesel fuel
reformulation [30]. Tornatore et al. [31] used 20% n-butanol/diesel
blend at retarded injection timing coupled with an EGR rate of 50%
in an optically accessible diesel engine and achieved substantial
reductions in smoke and NOx emissions with a slight drop in effi-
ciency. They found that a good compromise between performance
and emissions can be obtained using early injection timing. Valen-
tino et al. [32] reported the effects of injection timing, intake oxy-
gen concentration (via EGR) and injection pressure on the
472 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486

combustion and emission characteristics of a turbo-charged, CRDI Table 1

diesel engine using 40% gasoline/diesel and 40% n-butanol/diesel Fuel propertiesa of diesel, DMC, isobutanol and n-pentanol.

blends. It was demonstrated that n-butanol/diesel blend showed Properties Diesel Dimethyl Isobutanol n-Pentanol
better efficiency than gasoline/diesel blend with reduced emis- carbonate
sions at moderate injection pressures of 100–120 MPa. In an Molecular formula C12H26–C14H30 CH3–OCO2–CH3 C4H9–OH C5H11–OH
another study, Valentino et al. [33] studied the effects of 20% and Cetane number 48–57 35.5 15 20
40% n-butanol/diesel blends by varying the injection pressure, C (wt%) 86.13 39.96 64.82 68.13
H (wt%) 13.87 0.07 13.49 13.61
injection timing and intake oxygen concentration. The results O (wt%) 0 53.30 21.62 18.15
showed that both blends required advanced injection, moderate Low heating value 43.36 13.50 33.64 34.65
injection pressure (100–120 MPa) and 19% intake oxygen concen- (MJ/kg)
tration to achieve improvement in NOx emissions and drastic Latent heat of 250–290 369 578.4 308.05
vaporization (kJ/kg)
reductions in smoke emissions with 5–7% penalty in specific fuel
Kinematic viscosity 3.52 5.60 2.63 2.89
consumption. Gu et al. [34] showed that a combination of EGR, at 40 °C (mm2/s)
retarded injection timing and butanol addition up to 30% propor- Density (kg/m3) 840 1069.4 802 814.8
tion can simultaneously reduce NOx and soot emissions. Zheng Flash point (°C) 70 18 27.8 49
et al. [35] studied the effects of four butanol isomers/diesel blends Boiling point (°C) 221 90 107.9 138
Self-ignition 254 195 415.6 300
under EGR rates up to 65% and concluded that isobutanol pre-
temperature (°C)
sented the longest ignition delay and the lowest smoke emissions
a
among all isomers. Li et al. [36] studied the combustion and emis- Data have been taken from Refs. [26,35,46,47].

sion characteristics of a diesel engine fuelled with n-pentanol/

diesel and n-pentanol/biodiesel/diesel blends in a single cylinder
diesel engine and concluded that simultaneous reduction in soot
and NOx emissions occurred only at low and middle loads while
NOx emissions increased at high loads when compared to diesel
fuel operation.
Cheng et al. [30] used n-butanol/diesel blends (10 and 30 vol%)
in a four-cylinder light-duty diesel engine and reduced smoke and
NOx emissions through late or advanced injection with moderate
EGR rates up to 26%. The influence of injection pressure on smoke
emissions was found to be weak when higher n-butanol/diesel
blends are used. Recently, the authors of this paper examined
isobutanol/diesel blends up to 40% [37] and n-pentanol/diesel
blends up to 45% [19] in a single cylinder diesel engine at various
SOI’s and EGR conditions and reported simultaneous reduction of
NOx and smoke emissions with a drop in performance. From the
literature reviewed, it can be seen that with better management
of type of fuel used, fuel injection timing, fuel injection pressure,
intake oxygen concentration, NOx, smoke emissions and perfor-
mance could be improved. The present study employs a mathe-
matical modeling and optimization approach which can be used
to study the interactive effects of the considered parameters. Also
quantifying the individual influences of these parameters on
engine characteristics can help engine manufacturers to make
informed decisions over their application. In such multi-objective
problems, the application of non-linear techniques like RSM, Tagu-
chi method, Fuzzy logic, Genetic algorithm and Artificial neural
networks are appropriate to study the combined effect of input fac-
tors [38,39]. RSM has been used earlier for optimization of engine
parameters like load, speed, compression ratio, injection pressure,
injection timing to obtain desired combustion and performance
characteristics [38–45].
The present study utilizes three factors namely, fuel used, EGR
rate and injection timing for optimization. Three levels of each fac-
tor were chosen. Three low reactive advanced biofuels namely
isobutanol, n-pentanol and DMC were used as blend components
with diesel. The properties of these biofuels are presented in
Table 1. Three EGR rates (0, 15 and 30%) and three injection timings
(21, 23 and 25°CA bTDC) were used to control combustion phasing.
A 3
3 full factorial design was used to design the experiments.
Response surface methodology (RSM) was used to obtain optimum
combination of the above factors to simultaneously reduce NOx
and smoke emissions with minimum BSFC.
The objectives of this study can be summarized as follows: (i)
To quantify the individual effects of fuels, EGR and injection timing
on engine performance and emission characteristics. (ii) To predict
NOx, smoke opacity, HC, CO and BSFC for all blends using
regression models for the experimental design. (iii) To determine
an optimum combination of the considered factors to achieve
simultaneous reduction of NOx and smoke with maximum possi-
ble efficiency in a single cylinder DI diesel engine using RSM based
desirability approach. (iv) To demonstrate the use of RSM for effec-
tive management and optimization of factors which would assist
engine manufacturers to make informed decisions over choosing
them for achieving pre-determined emission and performance
objectives.
2. Materials and methods
2.1. Test fuels
This study used three advanced biofuels viz., di-methyl carbon-
ate (CAS No.: 616-38-6), isobutanol (CAS No.: 78-83-1) and n-pen-
tanol (CAS No.: 71-41-0) which are certified with a purity of 98%,
99% and 98% (analytical grade) respectively, procured from a local
supplier. Isobutanol is a four carbon, branched chain alcohol with a
hydroxyl group at the terminal carbon. n-Pentanol is a five carbon,
straight chain alcohol with a hydroxyl group at terminal carbon. All
these biofuels can be used to meet the target (replacing fossil fuel
with 36 billion gallons of advanced biofuels by 2022) set by U.S.
Renewable Fuel Standard (RFS) program, as they qualify an ‘ad-
vanced biofuels’ that can be derived from ligno-cellulosic biomass
feedstocks [48]. Ultra-low sulfur diesel (ULSD) fuel was supplied by
the Indian oil corporation, Chennai. Three advanced biofuel/diesel
blends were prepared by mixing DMC, n-pentanol and isobutanol
with ULSD at a blending ratio of 15/85, 38/62, 45/55 by vol respec-
tively. These blends were prepared to contain the same percentage
of fuel-bound oxygen (8 wt%). Fuel-bound oxygen strongly impacts
the combustion behavior through local air/fuel ratio and influences
emission formation. A similar approach of evaluating combustion,
performance and emission characteristics of a single cylinder die-
sel engine fuelled with vegetable oil, bio-diesel, ethanol, n-butanol
and diethyl ether based on fuel-bound oxygen content was earlier
employed by Rakopoulos et al. [49]. The properties of all test fuels
estimated by ASTM test methods were presented in Table 2.
The methodology of analyzing blends with similar oxygen con-
tent used in this study was also governed by the fact that the US
Environmental Protection Agency (EPA) waiver (211b) allows bio-
fuel blending based on its oxygen content with fossil fuels, for
instance, ethanol is allowed to be blend by up to 3.5 wt% in oxygen
or 10 vol% in gasoline. In this study, the choice of limiting the
oxygen content to ‘8 wt%’ for all the blends was made on the basis
 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486 473

Table 2
Properties of test fuels.

Properties Test method ULSD DMC100 ISB100 PEN100 DMC15 ISB38 PEN45
Oxygen (wt%) – 0 53.3 21.62 18.15 8 8 8
Calculated cetane index ASTM D4737 52 – – – 49.53 38.70 42.80
Low heating value (MJ/kg) ASTM D240 43.36 13.5 33.64 32.16 38.88 39.46 39.44
Kinematic viscosity at 40 °C (mm2/s) ASTM D445 3.52 5.6 2.63 2.89 3.83 3.18 3.24
Density (kg/m3) ASTM D4052 840 1069.4 810 811 874.35 828.6 826.95
Flash point (°C) ASTM D93 70 18 27.8 49 18 27.8 49

of the tests carried out with these biofuel/diesel blends in diesel angle and the position of top dead centre (TDC) were measured
engines. n-Pentanol was tested successfully in diesel engines up using an AVL364 angle encoder. The charge amplifier and the enco-
to 45 vol% with diesel [19]. Similarly, isobutanol was tested by der were connected to an AVL 615 Indimeter A/D card, which con-
up to 40% in diesel engines [35,37]. Addition of DMC above verts analog input to digital output. Al/Cr K-type thermocouple
20 vol% is considered impracticable [26] and hence avoided. There- was used to measure the exhaust gas temperature.
fore the ratios of all blends were suitably selected (45% n-pentanol, AVL 615 Indimeter software was used to analyze the output
38% isobutanol and 15% DMC by vol) to contain ‘8 wt% oxygen’ uni- data from the A/D card. This generates a pressure-crank angle dia-
formly to adhere with these limitations. All the test fuels were gram that indicates the variation of pressure and heat release rate
checked for solubility with diesel and no phase separation was at every crank angle. In addition, this also indicates the crank
found after several days of observation. angles at which the combustion starts and at which the percentage
(10, 50 & 90%) of heat released during the combustion. Measure-
2.2. Test engine and facilities ments of all these parameters were carried out for 100 cycles
and an average value of these cycles was recorded as a measured
Tests were performed in a naturally-aspirated, constant-speed parameter at that load. The fuel consumption rate was measured
(1500 rpm), single-cylinder, four-stroke, direct-injection diesel using an electronic weighing scale of sensitivity 0.1 g and a digital
engine. Specifications of the engine are given in Table 3. This stop watch of resolution 0.01 s. Gaseous emissions including NOx,
engine is predominantly used in the Indian agricultural sector to CO and HC emissions were measured using AVL Di-gas 444 exhaust
drive water pump-sets for irrigation purposes. There are about gas analyzer and smoke opacity was measured using AVL 437C
14.42 million diesel driven pump-sets in the country with the opacimeter. Table 4 shows the range, accuracy and uncertainties
annual growth of production of 7% [50]. In the year 2012–2013, of the instruments used in this study.
these engines consumed about 6 million tonnes of diesel which
is about 8.55% of India’s total diesel consumption [51]. Thus a large 2.2.1. EGR setup
population of farmers in India is severely exposed to toxic diesel In the current study, an external cooled-EGR system was used.
exhaust from these engines. Cooled EGR reduces the intake charge temperature and lowers the
The layout of the engine and its instrumentation is shown in peak in-cylinder temperatures which in-turn inhibits NOx forma-
Fig. 1. A swinging field electrical dynamometer was used to apply tion. Cooling also increases the density of EGR which allows a
the load on the engine. This electrical dynamometer consisted of a greater proportion of EGR to be used. Thus a fraction of exhaust
5-kV A AC alternator (220 V, 1500 rpm) mounted on bearings and gas is sent to the EGR cooler before it is diverted into the intake
on a rigid frame for swinging field type loading. The output power manifold for mixing with the incoming air. EGR cooler is a heat
was obtained by measuring the reaction torque using a strain exchanger in which cooling water is maintained at a constant tem-
gauge type load cell. The pressure inside the combustion chamber perature to absorb heat from the incoming exhaust gas. It has to be
was measured using an AVL GH12D miniature pressure transducer noted that temperature of the exhaust gas re-circulated is kept
connected to an AVL3066A02 Piezo charge amplifier. The crank cooler than the engine exhaust and warmer than the intake air
charge. In this experiment, the exhaust gas is cooled down to
Table 3 35 °C and is kept constant throughout the test. EGR rate is controlled
Engine specifications. by an EGR valve. An orifice is used to measure the flow rate of the
Make and model Kirloskar, TAF1 make exhaust gas. Sufficient mixing of incoming air and re-circulated
Number of cylinders 1
exhaust gas is ensured inside the mixing chamber before it gets
Combustion chamber Hemispherical open type inducted into the combustion chamber. EGR quantity was deter-
Piston Shallow Bowl-in type mined by measuring the concentrations of CO2 in the intake and
Bore, mm 87.5 the exhaust manifolds. The quantity of CO2 in the exhaust was mea-
Stroke, mm 110
sured using the gas analyzer by adjusting the control valve to vary
Connecting rod length, mm 220
Swept volume, cm3 661 the flow rate of the exhaust until the quantity of CO2 in the intake
Clearance volume, cm3 36.87 reaches the desired value. Then the following relation was used,
Compression ratio 17.5:1

Rated power, kW 4.4 ðCO2 Þintake
EGR % ¼
100
Rated speed, rpm 1500 ðCO2 Þexhaust
Injection type Direct injection
Number of nozzle holes 3 Similar method was used earlier to determine the EGR rates in
Spray-hole diameter, mm 0.25 this engine [18,19,28,29,37].
Injection pressure, bar 210
Cone angle, ° 110
Needle lift, mm 0.25 2.3. Test procedure
Valve timing
Intake valve opening, °CA bTDC 4.5 All the trials were performed under high load conditions that
Intake valve closing, °CA aBDC 35.5 correspond to the brake mean effective pressure (BMEP) of
Exhaust valve opening, °CA bBDC 35.5 5.3 bar at a constant speed of 1500 rpm. The fuel-air mixture
Exhaust valve closing, °CA aTDC 4.5
dwells longer inside the combustion chamber at high load and
474 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486

Fig. 1. Layout of the engine setup and its instrumentation.

Table 4
Range, accuracy and percentage uncertainties of instruments used.

Instrument Measured quantity Range Accuracy Uncertainties (%)

Gas analyzer NOx 0–2000 ppm ±5 ppm ±5
HC 0–20,000 ppm ±0.12 ppm ±5
CO 0–15% ±0.02% ±5
Opacimeter Smoke opacity 0–100% ±1% ±1.0
K-type thermocouple Exhaust gas temperature 0–1000 °C ±1 °C ±0.2
Speed measuring unit Engine speed 0–9999 rpm ±10 rpm ±0.15
Pressure pickup Cylinder pressure 0–250 bar ±0.1 bar ±0.1
Crank angle encoder Crank angle 0–360° ±1° ±0.2
Burette Fuel quantity 0–1000 cc ±0.1 cc ±1.0
Digital stop watch Time – ±0.6 s ±0.2

low speed conditions which characterizes a typical everyday on-
road condition of an automotive diesel engine under part load
and speed. The injection pressure is kept constant (21 MPa)
throughout the study. The manufacturer recommended injection
timing of the test engine is 23°CA bTDC. The injection timing can
be varied by varying the number of shims under the fuel injection
pump. Shims were removed to advance the injection timing while
they were added to retard the injection timing. The thickness of the
shim is 0.4 mm which corresponds to a 2°CA advance or retard of
the injection timing. All tests were conducted under steady state
conditions. The analysis of combustion characteristics was based
on in-cylinder pressure measurements. Heat release rates (HRR)
were calculated from the pressure data that was averaged over
100 cycles in order to minimize the effects of cycle-to-cycle
variations.
The temperature of the lubricating oil was maintained between
85 and 90 °C. Experiments were carried out same day at constant
ambient temperature to improve the reliability of the recordings.
The engine always ran for 10 min to allow for stabilization before
the readings were recorded. Each test was repeated three times
and the results were averaged. The engine was initially run at its
standard injection timing with diesel fuel that forms the reference
fuel for this study. The baseline data was recorded. The engine was
always allowed to run for some time before changing the blends
in-order it to consume the fuel that remained in the fuel system
during the previous trial. Later the engine was run with the test
fuel at the standard injection timing. The EGR rate was varied grad-
ually alongside by opening the EGR valve and the recordings were
made.
2.4. Error analysis
The errors associated with various measurements and calcula-
tions of parameters are computed in this section. The maximum
possible error in calculations was estimated by using the method
proposed by Moffat [52]. Errors were estimated from minimum
values of the output and accuracy of the instrument. If an esti-
mated quantity S, depends on independent variables like
 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486 475

ðx1 ; x2 ; x3 . . . xn Þ, then the error in the value of S is calculated by n-pentanol/diesel blends [19], isobutanol/diesel blends [37] and
using Eq. (1), DMC/diesel blends [53]. EGR rates of up to 30% have been used ear-
( lier which were observed to reduce NOx emissions gradually. How-
2  2  2 )12
@S @x1 @x2 @xn ever when the rates were increased beyond 30%, smoke emissions
¼ þ þ ... þ ð1Þ
S x1 x2 xn increased drastically for these fuel blends. Hence the upper limit of
EGR was set to 30% and three levels of EGR (0, 15 and 30%) were
   
where @x1
; @x 2
, etc. are the errors in the independent variables. used. For the fuel injection timing, a retarded (21°CA bTDC), stan-
x1 x2
dard (23°CA bTDC) and an advanced injection timing (25°CA bTDC)
@x1 is the accuracy of the measuring instrument and x1 is the min-
were chosen. Earlier studies with this engine had shown that fur-
imum value of the output measured during the experiment. As per
ther retardation increased smoke emissions while further advanc-
Eq. (1), the maximum possible error in the calculation of BSFC was
ing increased NOx emissions [16]. The response variables
determined to be 0.32%. Similarly, the errors associated with the
measured were NOx, smoke opacity, HC, CO and BSFC.
measurements of cylinder pressure and crank angle was deter-
mined to be 1.35% and 2% respectively.
2.5.1. Full factorial design
This study adopts a three-factor-three-level full factorial design,
2.5. Experimental design requiring 27 trials to evaluate the effect of considered factors on
measured responses. This experimental design matrix, as shown
Table 5 shows the factors investigated in this study with their in Table 6, is generated using Design Expert [54] software that
chosen levels. The fuel blends used, EGR rate and injection timing has been popularly used for optimizing engine parameters in sev-
were considered as factors. Among these factors, ‘‘fuel used” is eral earlier works [38–40,55–57]. This software is also used for
treated as a nominal categoric factor which has no intrinsic order- regression modeling, generating response surface plots and graph-
ing to the category. DMC/diesel, isobutanol/diesel and n-pentanol/ ical analysis of the measured data in this study.
diesel blends form the levels of this categoric factor. The other two
factors, EGR rates and injection timing are of numeric type with 2.5.2. Response surface methodology (RSM)
discrete values. In earlier studies, EGR has been successfully used RSM is a collection of mathematical and statistical techniques
to reduce NOx emissions in this engine when fuelled with that are useful for modeling and analysis of problems in which
the objective is to optimize a response (like NOx, smoke opacity
Table 5
Factors considered with their chosen levels.
and BSFC) that is influenced by several factors (like fuel, EGR rate
and injection timing) [40]. In this study, RSM is employed to model
Factors Factor Levels
and predict the responses. The experimental data in Table 6 was
type
1 2 3 analyzed using response surface regression and second order poly-
Fuel used (f) Categoric DMC/diesel Isobutanol/ n-Pentanol/ nomial models were developed using Eq. (2),
blend diesel blend diesel blend
EGR rate (e) in% Numeric 0% 15% 30% X
3 X
3 X
n

Injection timing Numeric 21 23 25

Y ¼ bo þ bi X i þ bii X 2i þ bij X i X j þ e ð2Þ
i¼1 i1 i<1
(t) in °CA bTDC

Table 6
Full-factorial experimental design matrix.

Run Injection timing EGR Fuel NOx Smoke opacity HC CO BSFC

°CA bTDC % – ppm % ppm vol% kg/kW h
1 23 0 DMC/diesel 1162 0.3 33 0.01 0.2603
2 23 15 DMC/diesel 1081 3.6 42 0.02 0.2688
3 23 30 DMC/diesel 839 5 46 0.03 0.2704
4 21 0 DMC/diesel 1037 2.3 35 0.02 0.2642
5 21 15 DMC/diesel 953 4.4 39 0.03 0.2819
6 21 30 DMC/diesel 692 18.4 40 0.05 0.2834
7 25 0 DMC/diesel 1515 0 26 0.02 0.2618
8 25 15 DMC/diesel 1447 0 29 0.02 0.2774
9 25 30 DMC/diesel 1094 1.2 30 0.04 0.2770
10 23 0 Isobutanol/diesel 1226 0 42 0.02 0.2461
11 23 15 Isobutanol/diesel 1109 0 54 0.01 0.2538
12 23 30 Isobutanol/diesel 866 1.6 69 0.02 0.2553
13 25 0 Isobutanol/diesel 1406 0 35 0.01 0.2475
14 25 15 Isobutanol/diesel 1365 0 31 0.02 0.2616
15 25 30 Isobutanol/diesel 1036 0 24 0.03 0.2612
16 21 0 Isobutanol/diesel 1102 0 45 0.02 0.2497
17 21 15 Isobutanol/diesel 990 0 46 0.02 0.2656
18 21 30 Isobutanol/diesel 749 1.9 40 0.03 0.2670
19 23 0 n-Pentanol/diesel 1378 12.7 35 0.03 0.2531
20 23 15 n-Pentanol/diesel 1281 13 37 0.03 0.2612
21 23 30 n-Pentanol/diesel 913 17.5 44 0.05 0.2628
22 21 0 n-Pentanol/diesel 1197 13.5 36 0.03 0.2568
23 21 15 n-Pentanol/diesel 1128 15.5 36 0.04 0.2737
24 21 30 n-Pentanol/diesel 779 17.6 37 0.06 0.2752
25 25 0 n-Pentanol/diesel 1841 7.8 30 0.01 0.2545
26 25 15 n-Pentanol/diesel 1564 12.1 33 0.03 0.2694
27 25 30 n-Pentanol/diesel 1136 16.4 35 0.05 0.2691
476 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486
where Y is the response, Xi are numeric values of the factors, terms
b0, bi, bii and bij are regression coefficients, i and j are linear and
quadratic coefficients, and e is the experimental error [43]. Factors
can be numeric or categoric. In case of a categoric factor, second
order polynomial models for each response variable would be
developed for each categoric level, in terms of other numeric fac-
tors. In this study, ‘fuel’ is treated as a categoric factor with three
different fuels comprising DMC/diesel, isobutanol/diesel and n-
pentanol/diesel blends forming three categoric levels with no
intrinsic ordering. Hence, second order polynomial models for each
response variable will be developed for each fuel category used. The
developed equations, if statistically significant, present a correlation
between factors and measured responses and can be used to predict
the responses. Response surface plots were constructed using these
fitted models. The optimal combination of biofuel blends, EGR rates
and injection timing can be finally obtained by using the desirability
approach of RSM.
2.5.3. Desirability approach
Desirability approach is a statistical technique used for solving
multi-response problems. It is widely used for its simplicity, avail-
ability across software platforms and flexibility in assigning
weighting and importance for individual responses. This involves
combining multiple responses like NOx, smoke and BSFC into a sin-
gle dimensionless measure of performance called the desirability
function. The approach as such involves transforming each esti-
mated response Y i that varies over the range 0 < di < 1, where di
value indicates the response value Y i to be more desirable. For
instance a value of di ¼ 1 indicates a completely acceptable
response while a value of di ¼ 0 indicates a completely undesired
response. The objective of each response is to ‘minimize’, ‘maxi-
mize’, ‘target’, ‘in range’ or ‘equal to’ based on the nature of the
optimization. In the present work, the objective is to minimize
NOx, smoke opacity and BSFC simultaneously.
For an objective of minimizing a response, the desirability di
will be defined by,
di ¼ 1; when Y i 6 Lowi
 wti
Highi  Y i
di ¼ ; when Lowi < Y i < Highi
Highi  Lowi
di ¼ 0; when Y i P Highi
For achieving a set target for a response, the desirability di will
be defined by,
di ¼ 0; when Y i < Lowi ; Y i > Highi
 wti
Y i  Lowi
di ¼ ; when Lowi < Y i < T i
T i  Lowi
 wti
Y i  highi
di ¼ ; when T i < Y i < Highi
T i  highi
Similarly the definitions of di for other objectives like ‘maxi-
mize’, ‘in range’ and ‘equal to’ can be found in Refs. [38,58]. Here
i indicates the response, Y the value of the response, low and high
represents the lower and upper limits of the response respectively;
T indicates the target value of the response. The weights wti are
used to give more emphasis to the upper/lower bounds. Weights
varies over the range 0:1 < wti < 10; a weight greater than 1 gives
more emphasis to the objective, while weights less than 1 give less
emphasis. Individual desirability of each responses are then com-
bined using geometric mean to obtain an overall desirability objec-
tive function D that varies over 0 < D < 1, which is calculated by,
!P1
Y
n ri
r
D¼ di i
i¼1
In the desirability objective function (D), each response is
assigned an importance (r) that is relative to other responses.
Importance varies from the least important value of 1 (indicated
by +) and the most important value of 5 (indicated by +++++). A
high value of D indicates a desirable and optimal solution. The opti-
mum value of factors are then determined from the values of indi-
vidual desired function that maximizes D [38,39,58]. The various
solutions obtained using desirability approach is then validated
by conducting confirmatory experimental trials based on the set
optimization criterion.
3. Results and discussion
3.1. Analysis and evaluation of the model
The quality of the models developed using RSM for describing
the experimental data of all the responses (NOx, smoke opacity,
HC, CO and BSFC) are validated by analysis of variance (ANOVA)
which is also used to identify the most significant factors and pro-
vides a means for quantifying (as a percentage) the individual
effects of factors on responses. A probability ‘p-value’ of less than
0.05 in ANOVA indicates that the factors have significant effect at
95% confidence level [59]. All models are found to be statistically
significant with all p-values being less than 0.05, as seen from
Tables 7 and 8. The coefficient of determination (R2) indicates
how well the experimental data fit the developed statistical mod-
els. A value of 1 indicates a perfect fit while a value of 0 indicates
that the regression line does not fit the data [60]. It can be inferred
that the correlation between the experimental and predicted val-
ues were found to be excellent for NOx (R2 = 0.9671), smoke opac-
ity (R2 = 0.9349), CO (R2 = 0.9089) and BSFC (R2 = 0.9727). The
adjusted R2 is a modified version of R2 that has been adjusted for
the number of predictors in the model. The adjusted R2 values
are high for NOx (adj.R2 = 0.9777), smoke opacity (adj.
R2 = 0.8871), CO (adj.R2 = 0.8421) and BSFC (adj.R2 = 0.9527) which
indicates that these models are highly accurate [61]. However, HC
model was found to correlate only by up to 35.20% (R2 = 0.3520) to
the experimental data and hence this model would be unable to
predict the engine behavior with a meaningful accuracy (Adj.R2
being as low as 0.2342 too).
The predicted R2 indicates how well a regression model predicts
the responses for new observations. All the ‘Pred-R2’ values are in
reasonable agreement with the ‘‘Adj-R2” as the difference between
them is less than 0.2 [62]. ‘‘Adeq Precision” measures the signal to
noise ratio and a value greater than 4 is always desirable. All the
values for ‘‘Adeq Precision” greater than 4, indicate an adequate
signal implying that these models can be used to navigate the
design space [40]. The visualization of the predicted models can
be attained by a response surface plot and it is also possible to find
the optimum region through visual inspection of these surfaces
[63].
3.2. NOx emissions
From the results of ANOVA in Table 7, it can be seen that all the
considered factors and their interactions have significant influence
on NOx emissions under 99% confidence level. All the square terms
(e2 and t2) in the model were found to be significant which implies
that there is a curved-line relationship between the factors and the
responses [41]. EGR and injection timing exert similar levels of
influence (41%) on NOx emissions which could be attributed to
the low injection pressure (210 bar). The addition of a low reactive
advanced biofuel to diesel has the least influence (8%) on NOx
emissions when compared to the other factors.
 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486 477

Table 7
Results of ANOVA – Emissions.

Source NOx Smoke opacity HC emissions CO emissions

Quadratic model Quadratic model Linear model Quadratic model
F p-Value P% F p-Value P% F p-Value P% F p-Value P%
Model 104.61 <0.0001 – 19.58 <0.0001 – 2.99 0.0412 – 13.61 <0.0001
t 485.51 <0.0001 41.64 13.36 0.0023 5.80 5.72 0.0258 16.84 9.61 0.0073 5.83
e 481.15 <0.0001 41.27 18.96 0.0006 8.23 1.39 – 4.11 70.78 <0.0001 42.98
f 43.38 <0.0001 7.44 83.06 <0.0001 72.11 2.42 – 14.26 22.35 <0.0001 27.14
te 7.37 0.0160 – 2.33 – – – – – 0.29 – –
tf 6.68 0.0084 – 3.79 0.0465 – – – – 1.37 – –
ef 8.58 0.0033 – 2.86 – – – – – 5.49 0.0162 –
t2 19.80 0.0005 – 0.047 – – – – – 5.29 0.0362 –
e2 39.59 <0.0001 – 1.23 – – – – – 5.29 0.0362 –
R2 0.9671 0.9349 0.3520 0.9089
Adj-R2 0.9777 0.8871 0.2342 0.8421
Pred-R2 0.9483 0.7005 0.0345 0.6652
Adequate precision 39.721 14.425 6.047 13.386

F – value of F-statistic; p-value – Probability value; P – percentage contribution; f- fuel used; e – EGR; t – fuel injection timing.
R2 – coefficient of determination; adj-R2 – adjusted R2; pred-R2 – predicted R2.

Table 8 The in-cylinder pressure and heat release rate profiles of the
Results of ANOVA – Performance. baseline diesel engine and the engine fuelled with DMC/diesel
Source BSFC blend injected at 21°CA bTDC under 30% EGR conditions is pre-
Quadratic model sented in Fig. 3. It can be seen that both the peak pressure and heat
F p-Value P% release rates (HRR) decreased due to the combined ‘temperature
Model 48.63 <0.0001 –
lowering’ effect of retarded injection timing and EGR. The rate of
pressure rise (ROPR) is low for DMC/blend (5.45 bar/°) when com-
t 16.47 0.0010 2.99
e 185.31 <0.0001 33.70
pared to ULSD (6.29 bar/°). NOx emissions of the engine fuelled
f 107.96 <0.0001 39.26 with DMC/diesel blends injected at 21°CA bTDC under 30% EGR
te 2.20 – – were found to be reduced by 46.11%, when compared to diesel
tf 0.014 – – operation at standard conditions (naturally aspirated engine with
ef 0.16 – –
standard injection timing of 23°CA bTDC).
t2 67.53 <0.0001 12.28
e2 47.21 <0.0001 8.58 Regression equations were developed for NOx emissions for all
the blends using the second-order response surface models that
R2 0.9727
Adj-R2 0.9527 includes full quadratic terms, interaction terms and linear terms
Pred-R2 0.9154 of the two input factors i.e., injection timing (t in °CA) and EGR
Adequate precision 27.178 rates (e in %). ANOVA has already showed that the predicted NOx
F – value of F-statistic; p-value – Probability value; P – percentage contribution; emission model is significant under 99% confidence levels. They
f- fuel used; e – EGR; t – fuel injection timing. are given as Eqs. (3)–(5) below,
R2 – coefficient of determination; adj-R2 – adjusted R2; pred-R2 – predicted R2.
DMC/diesel blend

NOx ¼ 7897:10648  713:63889
t þ 26:01481
e

The interactive effect of EGR and injection timing for all three
blends are depicted in Fig. 2. The best advantageous conditions
for NOx emissions occurred at regions where the EGR rate is high
and the injection timing is latest. NOx emission decreases as the
injection timing is retarded from 25°CA to 21°CA bTDC with
increasing EGR rates due to low combustion temperatures for all
blends. During late injection, the combustion continues to the
expansion stroke causing low peak pressures accompanied with
heat losses due to expansion. This lowers the peak combustion
temperature and discourages NOx formation. Thus NOx emissions
are lowest at the highest EGR rate (30%) and the latest injection
timing (21°CA). It can also be noted that NOx emissions were high
for n-pentanol/diesel blends. DMC/diesel blends and isobutanol/
diesel blends seem to have comparable NOx reduction capabilities.
High latent heat of vaporization of fuel blends causes a ‘‘cooling
effect” which reduces in-cylinder temperatures leading to less
thermal NOx formation [19]. The latent heat of vaporization of n-
pentanol is lowest among the three (From Table 1) which indicates
that NOx emission is high due to high in-cylinder temperatures
when compared to DMC/diesel and isobutanol/diesel blends.
 1:05556
t
e þ 18:34722
t2  0:46123
e2 ð3Þ
Isobutanol/diesel blend
NOx ¼ 8681:32870  747:63889
t þ 26:08148
e
 1:05556
t
e þ 18:34722
t2  0:46123
e2 ð4Þ
n-Pentanol/diesel blend
NOx ¼ 8014:74537  708:38889
t þ 20:47037
e
 1:05556
t
e þ 18:34722
t2  0:46123
e2 ð5Þ
3.3. Smoke opacity
The small p-values of all considered factors indicate that their
contribution is significant to the smoke opacity model as evi-
denced from ANOVA in Table 7. It can be inferred that the addition
of an advanced biofuel to diesel has a statistical significance on
smoke opacity. Earlier Liu et al. [64] who investigated soot
478 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486

Fig. 2. Interactive effect of Injection timing and EGR on NOx emissions for (a) DMC/diesel blend, (b) isobutanol/diesel blend and (c) n-pentanol/diesel blend.

Fig. 3. Comparison of in-cylinder pressure and HRR profiles between the baseline diesel engine and the engine with lowest NOx emissions.

formation, oxidation, and distribution for various oxygenated bio- emissions. The interaction between low reactivity fuel addition
fuels in a constant volume combustion chamber have asserted that to diesel and injection timing also had a significant effect on smoke
a proper choice of a low reactivity fuel can be crucial for smoke opacity. The interactive effects of EGR and injection timing for
 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486
Fig. 4. Interactive effect of Injection timing and EGR on Smoke opacity for (a) DMC/diesel blend, (b) isobutanol/diesel blend and (c) n-pentanol/diesel blend.
smoke opacity are portrayed as response surface plots for all
blends in Fig. 4. Smoke opacity was best pronounced at the regions
where the injection timing is advanced and the EGR rate is low.
Smoke emissions aggravate for all blends due to deterioration of
normal combustion process as a result of EGR.
Equally from Fig. 4, it is evident that isobutanol/diesel blends
decreases smoke opacity to a greater extent than DMC/diesel and
n-pentanol/diesel blends. The response surface plot is almost flat
for isobutanol/diesel blends with smokeless emissions observed
across major areas. The modification of injection timing had no
effect on smoke emissions in case of isobutanol/diesel blend due
to its already longer ignition delay that promotes fuel-air mixing.
There is a slight slope upwards when the EGR rates approach the
maximum and the injection timing is delayed to 21°CA. There is
evidence in the literature that isobutanol/diesel blends have better
smoke suppression capabilities than n-pentanol/diesel blends [18].
In case of DMC/diesel and n-pentanol/diesel blends, the perfor-
mance characteristic for smoke opacity is best pronounced at the
injection timing of 25°CA bTDC and it wanes when the timing is
retarded towards 21°CA. n-Pentanol/diesel blends presents higher
soot emissions when compared to DMC/diesel and isobutanol/
diesel blends. This is because the sooting tendency of longer
carbon-chain alcohols (like n-pentanol) is larger than the direct
479
chemical effects of its oxygen, and hence isobutanol and dimethyl
carbonate with smaller carbon numbers are effective for soot
reduction [65].
In summary, isobutanol/diesel blend injected at 25°CA bTDC
under naturally aspirated conditions (0% EGR) produces lowest
smoke emissions among the blends. Cylinder pressure and HRR
profiles as seen in Fig. 5, indicate that this engine showed high
peak pressures and higher premixed combustion phasing than die-
sel due to its longer ignition delay time. The rate of pressure rise
(ROPR) was found to be very high for isobutanol/diesel blend
(15.4 bar/°) when compared to ULSD (6.29 bar/°). The peak HRR
occurred at a crank angle, which is shifted to a later stage. Pre-
mixed combustion phasing mainly depends on the time available
for fuel-air mixing before ignition. Longer ignition delays promote
fuel-air mixing accompanied with higher premixed combustion
fractions [19,35]. This fuel blend decreased the smoke opacity by
99.52% when compared to the baseline diesel engine.
The experimental data of smoke opacity for all blends were
modeled by multiple regression to fit the second-order polynomial
equations. The quadratic model for smoke opacity was found to be
significant at 99% confidence level using ANOVA from Table 7. The
regression models of smoke opacity for individual blends are
expressed in Eqs. (6)–(8),
480 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486
Fig. 5. Comparison of In-cylinder pressure and HRR profiles between the baseline diesel engine and the engine with lowest smoke opacity.
DMC/diesel blend
Smoke opacity ¼ 61:66481  3:84306
t þ 0:88954
e
 0:034167
t
e þ 0:051389
t 2
þ 4:69136
103
Isobutanol/diesel blend
Smoke opacity ¼ 19:05926  2:00972
t þ 0:68398
e
 0:034167
t
e þ 0:051389
t 2
þ 4:69136
103
n-Pentanol/diesel blend
Smoke opacity ¼ 46:44815  2:70972
t þ 0:83954
e
 0:034167
t
e þ 0:051389
t 2
þ 4:69136
103
3.4. BSFC
The BSFC model developed is quadratic in nature. Table 8 shows
that all the main effects of the factors considered for this model are
significant. The influence of adding an advanced biofuel to diesel
(39.26%) has a dominant influence on BSFC than EGR (33.70%).
There is negligible interaction between the factors for BSFC with
both the square terms being statistically significant. The injection
timing had least effect on BSFC. BSFC is generally low at the
engine’s standard injection timing because the combustion phase
approaches the TDC. There is a decrease in combustion efficiency
during advanced and late injection timing with all the three blends,
because of the substantial deviation of combustion phasing from
the TDC [30]. EGR affects the engine performance due to the dete-
rioration of normal combustion process due to its chemical effect,
presence of high specific heat gases and oxygen-deprived condi-
tions. It can be deduced from Fig. 6 that BSFC is at its best when
the blends are injected at the engine’s standard timing of 23°CA
bTDC under naturally aspirated conditions (denoted by a bluish1
area which intensifies to green and then red). It should be noted that
all the blends have similar oxygen content and incidentally, similar
calorific values (from Table 2) which implies that BSFC is predomi-
nantly dependent on the combustion efficiency.
1
For interpretation of color in Figs. 6 and 8, the reader is referred to the web
version of this article.
From the interactive effect plot of Fig. 6, it could be seen that
DMC/diesel blends have high BSFC among the three blends. The
bluish fuel economy region is best pronounced for isobutanol/die-
sel blends at the engine’s standard injection timing and naturally
aspirated conditions. The pressure and heat release rate profiles
ð6Þ
of the engine at this particular condition in comparison with ULSD
are presented in Fig. 7. Isobutanol/diesel blends show longer igni-
tion delay due to its lower cetane index when compared to ULSD.
The LHV of isobutanol/diesel blend is lower than ULSD (from
Table 2) and hence more amount of fuel blend would be required
to produce the same power output resulting in higher BSFC than
ð7Þ
ULSD. Lower viscosity (from Table 2) of isobutanol/diesel blend
when compared to other blends favors fuel atomization and
enhances spray characteristics resulting in high combustion effi-
ciency. A more enhanced premixed combustion phase with higher
pressure and HRR peaks with improvement in the diffusion com-
bustion (due to the presence of fuel-bound oxygen) is observed
ð8Þ
with isobutanol/diesel blends when compared to diesel.
Table 9 lists the combustion parameters of all test fuels at
SOI = 23°CA bTDC and EGR = 0%, isobutanol/diesel blend has long-
est ignition delay due to its lowest cetane index (from Table 2)
when compared to other test fuels. This long ignition delay pro-
motes fuel-air mixing that leads to higher premixed combustion
fraction. The combustion duration of isobutanol/diesel blend is
thus shortest because of faster burning of more premixed charge
after the long ignition delay [28,35].
Regression equations developed for BSFC for all the blends
using the second-order response surface models includes full
quadratic terms, interaction terms and linear terms of the two
input factors i.e., injection timing (t in °CA) and EGR rates (e in
%). This model is significant under 99% confidence level as per
ANOVA from Table 8. They are given as Eqs. (9)–(11) below,
DMC/diesel blend
BSFC ¼ 1:25128  0:085800
t þ 2:04569
103
e
 3:15257
105
t
e þ 1:85141
103
t2
 2:75207
105
e2 ð9Þ
Isobutanol/diesel blend
BSFC ¼ 1:23422  0:085692
t þ 1:99834
103
e
 3:15257
105
t
e þ 1:85141
103
t 2
 2:75207
105
e2 ð10Þ
 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486 481

Fig. 6. Interactive effect of Injection timing and EGR on BSFC for (a) DMC/diesel blend, (b) isobutanol/diesel blend and (c) n-pentanol/diesel blend.

Fig. 7. Comparison of in-cylinder pressure and HRR profiles between the baseline diesel engine and the engine with lowest BSFC.
482 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486

Table 9 Graphical optimization displays the area of feasible response

Combustion parameters for all test fuels at SOI = 23°CA bTDC, EGR = 0%, BMEP = 5.3 values in the factor space and the regions that fit the optimization
bar and speed = 1500 rpm.
criteria [66]. Fig. 8 shows the overlay plots obtained by superim-
Fuel ID CD ROPR Pmax BTE BSFC posing the contours of all the response surfaces of the developed
° ° bar/° bar % kg/kW h regression models of NOx, smoke opacity and BSFC for all blends.
ULSD 20.10 57.60 6.69 73.05 36.23 0.2230 This plot reveals some ‘‘sweet spots” (shown in yellow) where
Isobutanol/diesel blend 24.35 51.70 13.57 77.86 37.28 0.2455 optimization can be met according to the limits presented in
n-Pentanol/diesel blend 22.30 52.60 12.89 77.05 37.06 0.2461
DMC/diesel blend 21.85 54.70 8.48 75.11 36.06 0.2618
Table 10. Fig. 8(a) includes the overlay plot obtained for DMC/die-
sel blend within the upper and lower limits considered in this
study. Similarly Fig. 8(b) and (c) portray the plots for isobutanol/
diesel and n-pentanol/diesel blends.
n-Pentanol/diesel blend

BSFC ¼ 1:24277  0:085752
t þ 2:02472
103
e
 3:15257
105
t
e þ 1:85141
103
t2
5
 2:75207
10
e 2
ð11Þ
3.5. Optimization
There is a trade-off relation between NOx, smoke opacity and
BSFC. Hence, it is necessary to optimize injection timing, EGR and
the advanced biofuel to be added to diesel with the objective of
minimizing NOx and smoke emissions simultaneously with mini-
mum BSFC. Table 10 shows the criteria of optimization for the
responses based on the objective stated above with their lower
and upper limits, the weightages used and their importance.
Weights range from 0.1 to 10. A weight greater than 1 gives more
emphasis to the goal while a weight less than 1 gives less empha-
sis. The desirability value varies in a linear mode with weight equal
to 1. All the parameters were given high and equal relative impor-
tance (r = 5) among each other.
Table 11 presents the top five solutions as per the set criteria of
optimization (as described in Table 10) generated by the Design
Expert software using desirability approach. Solutions with high
desirability are close to the set criteria and are preferred. From this
approach, isobutanol/diesel blend injected at 22°CA bTDC under
naturally aspirated condition is considered to be optimum for this
particular engine because it has a maximum desirability of 0.988.
The other blends that are closer to the optimization criteria are also
shown in Table 11.
3.6. Validation
It is important to validate the solutions obtained using desir-
ability approach by confirmatory experiments. Three trials were
taken and averaged at the optimum factor levels. Table 12 shows
the results of the confirmatory experiments with the predicted val-
ues from RSM models and the averaged actual values from the con-
firmatory trials. The percentages of error in predicting the
optimized parameters (NOx, smoke opacity and BSFC) were given.
It can be seen that the models developed using RSM for NOx,
Smoke opacity and BSFC were found to be adequate to describe
the effect of the injection timing and EGR on performance and
emissions for all blends and the error in prediction was found to
be within 5%.
Table 13 shows the response values at optimum factor levels
obtained using the desirability approach when compared to base-
line diesel operation. It could be seen that isobutanol/diesel blend
injected at 22°CA bTDC under naturally aspirated condition which
has a maximum desirability of 0.988, presented a simultaneous
reduction of NOx (12.4%.) and smoke opacity (100%.) with a
slight increase in BSFC (2.9%N).
It could be noted that three of five solutions with high desirabil-
ity in Table 11 occurs at ‘Zero’ EGR conditions due to the positive
effects of injection timing and biofuel blends used. Nevertheless,
‘Zero’ EGR conditions may not be relevant if the NOx emission level
is excessive compared to the regulations. Hence for the optimiza-
tion to approach a more balanced solution, a target level for NOx
could be chosen. From Table 10, it can be seen that NOx emission
ranges between 692 and 1841 ppm for all the three biofuel/diesel
blends. Hence a target level closer to the lower limit (692 ppm)

Table 10
Optimization criteria of emission and performance parameters.

Constraints name Target Limits Weight Importance

Lower Upper Lower Upper
Injection timing (°CA) In range 21 25 1 1 5
EGR (%) In range 0 30 1 1 5
Fuel In range DMC n-Pentanol 1 1 5
NOx (ppm) Minimize 692 1841 1 0.1 5
Smoke opacity (%) Minimize 0 18.4 1 0.1 5
BSFC (kg/kW h) Minimize 0.246144 0.283438 1 0.1 5
HC (ppm) None 24 69 – – –
CO (vol%) None 0.01 0.06 – – –

Table 11
Predicted solutions from desirability approach closer to optimization criteria (smoke: minimize, NOx: minimize and BSFC: minimize).

# Injection timing EGR Fuel NOx Smoke opacity BSFC HC CO BTE Desirability
°CA bTDC % ppm % kg/kW h ppm vol% %
1 22 0 Isobutanol/diesel 1105.537 0 0.246 47.668 0.016 37.064 0.988
2 23 30 Isobutanol/diesel 826.222 1.231 0.256 40.812 0.023 35.570 0.976
3 23 0 DMC/diesel 1176.389 0.606 0.257 38.000 0.014 35.955 0.968
4 23 0 n-Pentanol/diesel 1407.632 11.362 0.250 38.447 0.021 36.449 0.949
5 23 30 DMC/diesel 877.581 6.605 0.272 32.255 0.036 34.118 0.929
 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486 483

Fig. 8. Overlay plots showing regions of optimum performance and emission conditions based on the optimization criteria (smoke: minimize, NOx: minimize and BSFC:
minimize) for (a) DMC/diesel blend, (b) isobutanol/diesel blend and (c) n-pentanol/diesel blend.

within the range was selected. In Table 10, the optimization crite-
rion for NOx alone is set to have a target value, for instance,
700 ppm. The top five solutions predicted by the Design Expert
software using desirability approach as per this new criterion of
optimization are shown in Table 14. Confirmatory trials for these
solutions were not conducted since it was already established that
the RSM models developed for NOx, Smoke opacity and BSFC were
found to be adequate with a prediction error under 5%. When com-
pared to baseline diesel operation, it could be seen that isobutanol/
diesel blend injected at 23°CA bTDC under 30% EGR rate has a max-
imum desirability of 0.976 and presented a simultaneous reduction
of NOx from 1284 ppm to 826 ppm (35.67%.) and smoke opacity
from 20.7% to 1.231% (94.05%.) with a slight increase in BSFC from
0.243 kg/kW h to 0.256 kg/kW h (5.35%N). When compared to the
set criteria in Table 10, this balanced solution (with a set target for
NOx emissions) achieved a further reduction in NOx emissions
(from 12.4 to 35.67%) with slight penalty in smoke opacity (from
100% to 94.05%) and BSFC (from 2.9% to 5.35%).
4. Conclusion
An experimental and statistical investigation was carried out to
analyze the combined effect of injection timing and EGR on the
combustion and performance characteristics of a stationary diesel
engine fuelled with three advanced biofuel/diesel blends, namely
484 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486

Table 12
Confirmatory test results at BMEP = 5.3 bar and speed = 1500 rpm with % error (optimization criteria: smoke: minimize, NOx: minimize and BSFC: minimize).

# Injection timing EGR Fuel NOx Smoke opacity BSFC BTE HC CO

°CA bTDC % ppm % kg/kW h % ppm vol%
Ref 23 0 ULSD 1284 20.7 0.223 23 0.04
Model 1105.537 0 0.246 37.064 47.668 0.016
1 22 0 Isobutanol/diesel Actual 1125 0 0.250 36.543 55 0.02
% Error 1.76 0 1.63 1.41 15.38 25
Model 826.222 1.231 0.256 35.570 40.812 0.023
2 23 30 Isobutanol/diesel Actual 854 1.2 0.254 35.867 45 0.02
% Error 3.36 2.52 0.78 0.83 10.26 13.04
Model 1176.389 0.606 0.257 35.955 38.000 0.014
3 23 0 DMC/diesel Actual 1165 0.6 0.253 36.064 44 0.01
% Error 0.97 0.99 1.56 0.30 15.79 28.57
Model 1407.632 11.362 0.250 36.449 38.447 0.021
4 23 0 n-Pentanol/diesel Actual 1390 11.8 0.253 36.064 42 0.03
% Error 1.25 3.85 1.20 1.06 9.24 42.86
Model 877.581 6.605 0.272 34.118 32.255 0.036
5 23 30 DMC/diesel Actual 845 6.4 0.261 34.945 34 0.03
% Error 3.71 3.1 4.04 2.42 5.41 16.67

Table 13
Optimized responses in comparison with baseline diesel operation at BMEP = 5.3 bar and speed = 1500 rpm (optimization criteria: smoke: minimize, NOx: minimize and BSFC:
minimize).

# Injection timing EGR Fuel NOx Smoke opacity BSFC BTE HC CO

°CA bTDC % – ppm % kg/kW h % ppm vol%
23 0 ULSD 1284 20.7 0.243 37.233 23 0.04
1 22 0 Isobutanol/diesel 1125 0 0.250 37.064 55 0.02
(12.4%.) (100%.) (2.9%N) (0.45%.) (139.1%N) (50.0%.)
2 23 30 Isobutanol/diesel 854 1.2 0.254 35.570 45 0.02
(33.5%.) (94.2%.) (4.5%N) (4.47%.) (95.7%N) (50.0%.)
3 23 0 DMC/diesel 1165 0.6 0.253 35.955 44 0.01
(9.3%.) (97.1%.) (4.1%N) (3.43%.) (91.3%N) (25.0%.)
4 23 0 n-Pentanol/diesel 1390 11.8 0.253 36.449 42 0.03
(8.3%N) (43.0%.) (4.1%N) (2.11%.) (82.6%N) (33.3%.)
5 23 30 DMC/diesel 845 6.4 0.261 34.118 34 0.03
(34.2%.) (69.1%.) (7.4%N) (8.37%.) (47.8%N) (33.3%.)

Table 14
Predicted solutions from desirability approach closer to modified optimization criteria (smoke: minimize, NOx: target = 700 ppm and BSFC: minimize).

# Injection timing EGR Fuel NOx Smoke opacity BSFC BTE HC CO Desirability
°CA bTDC % ppm % kg/kW h % ppm vol%
1 23 30 Isobutanol/diesel 826.222 1.231 0.256 35.570 40.812 0.023 0.976
2 23 0 DMC/diesel 1176.389 0.606 0.257 35.955 38.000 0.014 0.968
3 24 30 DMC/diesel 943.106 5.074 0.272 34.031 35.972 0.036 0.943
4 22 0 n-Pentanol/diesel 1310.250 11.440 0.252 36.145 35.472 0.025 0.938
5 24 30 n-Pentanol/diesel 1020.410 15.557 0.264 34.531 36.306 0.048 0.910

dimethyl carbonate (DMC)/diesel, isobutanol/diesel and n-
pentanol/diesel that were prepared to contain 8 wt% of oxygen.
Response surface methodology (RSM) coupled with desirability
approach was used to model, predict and optimize the response
data measured from experiments conducted as per the full-
factorial experimental design matrix. The following conclusions
were drawn from the analysis:
1. As per ANOVA,
(i) All the developed regression models for NOx, smoke opac-
ity, HC, CO and BSFC were found to be statistically signifi-
cant under 99% confidence level.
(ii) EGR and injection timing exerted similar levels of influ-
ence (41%) on NOx emissions. The addition of a low reac-
tive advanced biofuel to diesel had the least influence
(8%) on NOx emissions when compared to the other
factors.
(iii) Addition of an advanced biofuel to diesel has a statistical
significance on smoke opacity.
(iv) The influence of adding an advanced biofuel to diesel
(39.26%) has a dominant influence on BSFC than EGR
(33.70%).
2. From the response surface plots, it can be concluded that
(i) The best advantageous conditions for NOx emissions
occurred at regions where the EGR rate is high and the
injection timing is latest. NOx emissions were found to
be high for n-pentanol/diesel blends.
(ii) Smoke opacity was best pronounced at the regions where the
injection timing is advanced and the EGR rate is low. isobu-
tanol/diesel blends decreased smoke opacity to a greater
extent than DMC/diesel and n-pentanol/diesel blends.
(iii) BSFC was generally low for all fuel blends at the engine’s
standard injection timing at naturally aspirated condi-
tions. BSFC of the engine operating with isobutanol/diesel
 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486
blend was found to be least. DMC/diesel blend was the
least favorable in terms of performance.
3. The optimization criteria was to minimize smoke and NOx
emissions with minimum BSFC. Solutions that are closer to this
criteria were obtained using desirability approach. From this
approach, isobutanol/diesel blend injected at 22°CA bTDC under
naturally aspirated condition was found to be optimum for this
particular engine. This optimum combination presented a
simultaneous reduction of NOx (12.4%.) and smoke opacity
(100%.) with a slight increase in BSFC (2.9%N), when compared
to baseline diesel operation.
4. As most of the solutions for the optimization criteria (NOx:
minimize, Smoke: minimize and BSFC: minimize) occurred at
‘‘Zero” EGR conditions, NOx emission levels could be excessive
compared to the regulations. Hence for the optimization to
approach a more balanced solution, the criterion for NOx alone
was modified from ‘‘minimize” to a new criterion with a target
level of 700 ppm. This new optimization criteria generated a
solution that achieved a further reduction in NOx emissions
(from 12.4 to 35.67%) with a slight penalty in smoke opacity
(from 100% to 94.05%) and BSFC (from 2.9% to 5.35%).
5. The solutions obtained using desirability approach were vali-
dated by confirmatory experiments. Results indicated that the
models developed using RSM for NOx, smoke opacity and BSFC
for all blends were adequate to describe the effect of the injec-
tion timing and EGR and the error in prediction was found to be
within 5%.
Thus RSM coupled with desirability approach can be effectively
used to determine optimum factor levels to achieve desired emis-
sion and performance characteristics in a single cylinder diesel
engine fuelled with advanced biofuel/diesel blends. This method
is robust and can also be extended to other engines for developing
models that can predict engine characteristics with reasonable
accuracy.
Acknowledgements
The authors appreciate the support of the management of Sri
Venkateswara College of Engineering, Sriperumbudur, Chennai
and Jeppiaar institute of Technology for facilitating this research.
The authors also acknowledge the assistance rendered by Poly-
meric Materials Research Lab at Alagappa College, Karaikudi for
estimating fuel properties.
References
[1] Kumar B Rajesh, Saravanan S. Use of higher alcohol biofuels in diesel engines: a
review. Renew Sustain Energy Rev 2016;60:84–115. http://dx.doi.org/10.1016/
j.rser.2016.01.085.
[2] Hong Y-K, Lee D-W, Ko Y-C, Yinghua L, Han H-S, Lee K-Y. Passive NOx
reduction with CO using Pd/TiO2/Al2O3 + WGSR catalysts under simulated
post-Euro IV diesel exhaust conditions. Catal Lett 2010;136(1):106–15. http://
dx.doi.org/10.1007/s10562-010-0312-5.
[3] Innes WB. Effect of nitrogen oxide emissions on ozone levels in metropolitan
regions. Environ Sci Technol 1981;15(8):904–12. http://dx.doi.org/10.1021/
es00090a003.
[4] Mills A, Elouali S. The nitric oxide ISO photocatalytic reactor system:
measurement of NOx removal activity and capacity. J Photochem Photobiol,
A 2015;305:29–36. http://dx.doi.org/10.1016/j.jphotochem.2015.03.002.
[5] Payne RL, Alaves VM, Larson RR, Sleeth DK. An evaluation of diesel particulate
matter in fire station vehicle garages and living quarters. J Chem Health Saf
2015. http://dx.doi.org/10.1016/j.jchas.2015.10.020 [in press].
[6] Benbrahim-Tallaa L, Baan RA, Grosse Y, Lauby-Secretan B, El-Ghissassi F,
Bouvard V, et al. Carcinogenicity of diesel-engine and gasoline-engine exhausts
and some nitroarenes. Lancet Oncol 2012;13(7):663–4. http://dx.doi.org/
10.1016/ s1470-2045(12)70280-2.
[7] Macchiarulo S, Gant T, Leonard M. Transcriptome profiling reveals allergy
associated gene expression in human bronchial epithelial cells and dendritic
cells exposed to diesel exhaust particulate chemicals. Toxicol Lett 2015;238
(2):S233. http://dx.doi.org/10.1016/j.toxlet.2015.08.688.
485
[8] Leonard M, Macchiarulo S, Gant T. Diesel exhaust particulate associated
chemicals elicit a pattern of asthma associated gene expression in human
primary bronchial epithelial cells. Toxicol Lett 2014;229:S206. http://dx.doi.
org/10.1016/j.toxlet.2014.06.694.
[9] Bangia KS, Symanski E, Strom SS, Bondy M. A cross-sectional analysis of
polycyclic aromatic hydrocarbons and diesel particulate matter exposures and
hypertension among individuals of Mexican origin. Environ Health 2015;14(1).
http://dx.doi.org/10.1186/s12940-015-0039-2. 51 [11p].
[10] Giles LV, Carlsten C, Koehle MS. The effect of pre-exercise diesel exhaust
exposure on cycling performance and cardio-respiratory variables. Inhalation
Toxicol 2012;24(12):783–9. http://dx.doi.org/10.3109/08958378.2012.
717649.
[11] Gamble JF. IARC evaluations of cancer hazards: comment on the process with
specific examples from volume 105 on diesel engine exhaust. J Clin Toxicol
2012;02(06). http://dx.doi.org/10.4172/2161-0495.1000e106.
[12] Jothi NK Miller, Nagarajan G, Renganarayanan S. LPG fueled diesel engine
using diethyl ether with exhaust gas recirculation. Int J Therm Sci 2008;47
(4):450–7. http://dx.doi.org/10.1016/j.ijthermalsci.2006.06.012.
[13] Kook S, Bae C, Miles PC, Choi D, Pickett LM. The influence of charge dilution and
injection timing on low-temperature diesel combustion and emissions. SAE
technical paper # 2015-01-3837; 2005. http://dx.doi.org/10.4271/2005-01-3837.
[14] Chen Z, Wu Z, Liu J, Lee C. Combustion and emissions characteristics of high n-
butanol/diesel ratio blend in a heavy-duty diesel engine and EGR impact.
Energy Convers Manage 2014;78:787–95. http://dx.doi.org/10.1016/j.
enconman.2013.11.037.
[15] Qi D, Leick M, Liu Y, Lee CF. Effect of EGR and injection timing on combustion
and emission characteristics of split injection strategy DI-diesel engine fueled
with biodiesel. Fuel 2011;90(5):1884–91. http://dx.doi.org/10.1016/
j.fuel.2011.01.016.
[16] Saravanan S, Nagarajan G, Sampath S. Combined effect of injection timing, EGR
and injection pressure in NOx control of a stationary diesel engine fuelled with
crude rice bran oil methyl ester. Fuel 2013;104:409–16. http://dx.doi.org/
10.1016/j.fuel.2012.10.038.
[17] McCormick R. Effects of Biodiesel on NOx Emissions 2005. http://dx.doi.org/
10.2172/15016824.
[18] Kumar B Rajesh, Saravanan S. Effects of isobutanol/diesel and n-pentanol/
diesel blends on performance and emissions of a DI diesel engine under
premixed LTC (low temperature combustion) mode. Fuel 2016;170:49–59.
http://dx.doi.org/10.1016/j.fuel.2015.12.029.
[19] Kumar B Rajesh, Saravanan S. Effect of exhaust gas recirculation (EGR) on
performance and emissions of a constant speed DI diesel engine fueled with
pentanol/diesel blends. Fuel 2015;160:217–26. http://dx.doi.org/10.1016/
j.fuel.2015.07.089.
[20] Lali A. Biofuels for India: what, when and how. Curr Sci 2016;110(4):552–5.
http://dx.doi.org/10.18520/cs/v110/i4/552-555.
[21] Higashide W, Li Y, Yang Y. Metabolic engineering of Clostridium cellulolyticum
for production of isobutanol from cellulose. Appl Environ Microbiol 2011;2077
(8):2727–33. http://dx.doi.org/10.1128/aem.02454-10.
[22] Cann AF, Liao JC. Pentanol isomer synthesis in engineered microorganisms.
Appl Microbiol Biotechnol 2009;85(4):893–9. http://dx.doi.org/10.1007/
s00253-009-2262-7.
[23] Desai SH, Rabinovitch-Deere CA, Fan Z, Atsumi S. Isobutanol production from
cellobionic acid in Escherichia coli. Microb Cell Fact 2015;14(1). http://dx.doi.
org/10.1186/s12934-015-0232-6. 52 [10p].
[24] Formighieri C. Cyanobacteria as a platform for direct photosynthesis-to-fuel
conversion. In: Solar-to-fuel conversion in algae and cyanobacteria,
SpringerBriefs in environmental science. Cham (Switzerland): Springer;
2015. p. 31–8. http://dx.doi.org/10.1007/978-3-319-16730-5_7.
[25] Ofuonye E, Kutin K, Stuart DT. Engineering Saccharomyces cerevisiae
fermentative pathways for the production of isobutanol. Biofuels 2013;4
(2):185–201. http://dx.doi.org/10.4155/bfs.12.85.
[26] Rounce P, Tsolakis A, Leung P, York APE. A comparison of diesel and biodiesel
emissions using dimethyl carbonate as an oxygenated additive. Energy Fuels
2010;24(9):4812–9. http://dx.doi.org/10.1021/ef100103z.
[27] Wu XL, Shu D, Meng YZ. Direct synthesis of dimethyl carbonate from CO2 and
methanol. J Chem Environ 2006;9(1):74–9. http://dx.doi.org/10.1002/
chin.200609267.
[28] Kumar B Rajesh, Saravanan S, Rana D, Anish V, Nagendran A. Effect of a
sustainable biofuel – n-octanol – on the combustion, performance and
emissions of a DI diesel engine under naturally aspirated and exhaust gas
recirculation (EGR) modes. Energy Convers Manage 2016;118:275–86. http://
dx.doi.org/10.1016/j.enconman.2016.04.001.
[29] Kumar B Rajesh, Saravanan S, Rana D, Nagendran A. Use of some advanced
biofuels for overcoming smoke/NOx trade-off in a light-duty DI diesel engine.
Renewable Energy 2016;96:687–99. http://dx.doi.org/10.1016/j.renene.2016.
05.029.
[30] Cheng X, Li S, Yang J, Liu B. Investigation into partially premixed combustion
fueled with n-butanol-diesel blends. Renewable Energy 2016;86:723–32.
http://dx.doi.org/10.1016/j.renene.2015.08.067.
[31] Tornatore C, Marchitto L, Mazzei A, Valentine G, Corcione FE, Merola SS. Effect
of butanol blend on in-cylinder combustion process, Part 2: compression
ignition engine. J KONES Powertrain Transp 2011;18(2):473–83.
[32] Valentino G, Corcione FE, Iannuzzi SE. Effects of gasoline-diesel and n-butanol-
diesel blends on performance and emissions of an automotive direct-injection
diesel engine. Int J Engine Res 2012;13(3):199–215. http://dx.doi.org/10.1177/
1468087412441879.
486 B. Rajesh Kumar et al. / Energy Conversion and Management 123 (2016) 470–486
[33] Valentino G, Corcione FE, Iannuzzi SE, Serra S. Experimental study on
performance and emissions of a high speed diesel engine fuelled with n-
butanol diesel blends under premixed low temperature combustion. Fuel
2012;92(1):295–307. http://dx.doi.org/10.1016/j.fuel.2011.07.035.
[34] Gu X, Li G, Jiang X, Huang Z, Lee C. Experimental study on the performance of
and emissions from a low-speed light-duty diesel engine fueled with n-
butanol–diesel and isobutanol–diesel blends. Proc Inst Mech Eng, Part D: J
Automobile Eng 2013;227(2):261–71. http://dx.doi.org/10.4271/2012-28-
0011.
[35] Zheng Z, Li C, Liu H, Zhang Y, Zhong X, Yao M. Experimental study on diesel
conventional and low temperature combustion by fueling four isomers of
butanol. Fuel 2015;141:109–19. http://dx.doi.org/10.1016/j.fuel.2014.10.053.
[36] Li L, Wang J, Wang Z, Xiao J. Combustion and emission characteristics of diesel
engine fueled with diesel/biodiesel/pentanol fuel blends. Fuel
2015;156:211–8. http://dx.doi.org/10.1016/j.fuel.2015.04.048.
[37] Kumar B Rajesh, Saravanan S. Effect of isobutanol addition to diesel fuel on
performance and emissions of a DI diesel engine with exhaust gas
recirculation. Proc Inst Mech Eng, Part A: J Power Energy 2015;230
(1):112–25. http://dx.doi.org/10.1177/0957650915617107.
[38] Pandian M, Sivapirakasam SP, Udayakumar M. Investigation on the effect of
injection system parameters on performance and emission characteristics of a
twin cylinder compression ignition direct injection engine fuelled with
pongamia biodiesel–diesel blend using response surface methodology. Appl
Energy 2011;88(8):2663–76. http://dx.doi.org/10.1016/j.apenergy.2011.01.
069.
[39] Hirkude BJ, Padalkar AS. Performance optimization of CI engine fuelled with
waste fried oil methyl ester-diesel blend using response surface methodology.
Fuel 2014;119:266–73. http://dx.doi.org/10.1016/j.fuel.2013.11.039.
[40] Ganapathy T, Gakkhar RP, Murugesan K. Optimization of performance
parameters of diesel engine with Jatropha biodiesel using response surface
methodology. Int J Sustain Energ 2011;30(sup1):S76–90. http://dx.doi.org/
10.1080/14786451.2011.594889.
[41] Najafi G, Ghobadian B, Yusaf T, Ardebili SM Safieddin, Mamat R. Optimization
of performance and exhaust emission parameters of a SI (spark ignition)
engine with gasoline–ethanol blended fuels using response surface
methodology. Energy 2015;90:1815–29. http://dx.doi.org/10.1016/j.energy.
2015.07.004.
[42] Ileri E, Karaoglan AD, Atmanli A. Response surface methodology based
prediction of engine performance and exhaust emissions of a diesel engine
fuelled with canola oil methyl ester. J Renew Sustain Energy 2013;5(3). http://
dx.doi.org/10.1063/1.4811801. 033132 [20p].
_
[43] Atmanlı A, Yüksel B, Ileri E, Karaoglan A Deniz. Response surface methodology
based optimization of diesel–n-butanol–cotton oil ternary blend ratios to
improve engine performance and exhaust emission characteristics. Energy
Convers Manage 2015;90:383–94. http://dx.doi.org/10.1016/j.enconman.
2014.11.029.
[44] Dhingra S, Bhushan G, Dubey KK. Performance and emission parameters
optimization of mahua (Madhuca indica) based biodiesel in direct injection
diesel engine using response surface methodology. J Renew Sustain Energy
2013;5(6). http://dx.doi.org/10.1063/1.4840155. 063117 [15p].
[45] Win Z, Gakkhar RP, Jain SC, Bhattacharya M. Parameter optimization of a diesel
engine to reduce noise, fuel consumption, and exhaust emissions using
response surface methodology. Proc Inst Mech Eng, Part D: J Automobile Eng
2005;219(10):1181–92. http://dx.doi.org/10.1243/095440705x34919.
[46] Ozsezen AN, Turkcan A, Sayin C, Canakci M. Comparison of performance and
combustion parameters in a heavy-duty diesel engine fueled with isobutanol/
diesel fuel blends. Energy Explor Exploit 2011;29(5):525–41. http://dx.doi.org/
10.1260/0144-5987.29.5.525.
[47] Campos-Fernandez J, Arnal JM, Gomez J, Lacalle N, Dorado MP. Performance
tests of a diesel engine fueled with pentanol/diesel fuel blends. Fuel
2013;107:866–72. http://dx.doi.org/10.1016/j.fuel.2013.01.066.
[48] Lawyer K, Ickes A, Wallner T, Ertl D, Williamson R, Miers S, Naber J. Blend ratio
optimization of fuels containing gasoline blend stock, ethanol, and higher
alcohols (C3–C6): Part I – methodology and scenario definition. SAE technical
paper # 2013-01-1144; 2013. http://dx.doi.org/10.4271/2013-01-1144.
[49] Rakopoulos DC, Rakopoulos CD, Giakoumis EG, Papagiannakis RG, Kyritsis DC.
Influence of properties of various common bio-fuels on the combustion and
emission characteristics of high-speed DI (direct injection) diesel engine:
vegetable oil, bio-diesel, ethanol, n-butanol, diethyl ether. Energy
2014;73:354–66. http://dx.doi.org/10.1016/j.energy.2014.06.032.
[50] GOI (Government of India). Annual report (2013–2014), Ministry of Petroleum
& Natural Gas, New Delhi, IndiaAvailable from: http://petroleum.nic.in/docs/
Annual_Report/AR13-14.pdf2014 [accessed 28 March 2016].
[51] GOI (Government of India). Indian Petroleum and Natural Gas Statistics,
Economics and Statistics Division, Ministry of Petroleum & Natural Gas, New
Delhi, IndiaAvailable from: http://petroleum.nic.in/docs/pngstat.pdf2014
[accessed 28 March 2016].
[52] Moffat RJ. Using uncertainty analysis in the planning of an experiment. J Fluids
Eng 1985;107(2):173–8. http://dx.doi.org/10.1115/1.3242452.
[53] Kumar B Rajesh, Saravanan S. Partially premixed low temperature combustion
using dimethyl carbonate (DMC) in a DI diesel engine for favorable smoke/NOx
emissions. Fuel 2016;180:396–406. http://dx.doi.org/10.1016/j.fuel.2016.04.060.
[54] Design-Expert 10.0. 2021 E. Hennepin Avenue (Ste 480, Minneapolis, MN
55413-2726, USA): Stat-Ease Inc.
[55] Prashant GK, Lata DB, Joshi PC. Investigations on the effect of ethanol blend on
the combustion parameters of dual fuel diesel engine. Appl Therm Eng
2016;96:623–31. http://dx.doi.org/10.1016/j.applthermaleng.2015.11.051.
[56] Abuhabaya A, Fieldhouse J, Brown D. The effects of using biodiesel on CI
(compression ignition) engine and optimization of its production by using
response surface methodology. Energy 2013;59:56–62. http://dx.doi.org/
10.1016/j.energy.2013.06.056.
[57] Fayyazbakhsh A, Pirouzfar V. Investigating the influence of additives-fuel on
diesel engine performance and emissions: analytical modeling and
experimental validation. Fuel 2016;171:167–77. http://dx.doi.org/10.1016/
j.fuel.2015.12.028.
[58] Benyounis KY, Olabi AG, Hashmi MSJ. Multi-response optimization of CO2
laser-welding process of austenitic stainless steel. Opt Laser Technol 2008;40
(1):76–87. http://dx.doi.org/10.1016/j.optlastec.2007.03.009.
[59] Myers RH, Montgomery DC, Anderson-Cook CM. Response surface
methodology: process and product optimization using designed
experiments. 4th ed. Hoboken (NJ): John Wiley & Sons; 2016.
[60] Gangadharan D, Sivaramakrishnan S, Nampoothiri KM, Sukumaran RK, Pandey
A. Response surface methodology for the optimization of alpha amylase
production by Bacillus amyloliquefaciens. Bioresour Technol 2008;99
(11):4597–602. http://dx.doi.org/10.1016/j.biortech.2007.07.028.
[61] Mutuk T, Mesci B. Analysis of mechanical properties of cement containing
boron waste and rice husk ash using full factorial design. J Clean Prod
2014;69:128–32. http://dx.doi.org/10.1016/j.jclepro.2014.01.051.
[62] Pamnani R, Vasudevan M, Vasantharaja P, Jayakumar T. Optimization of A-
GTAW welding parameters for naval steel (DMR 249 A) by design of
experiments approach. Proc Inst Mech Eng, Part L: J Mater: Des Appl 2015.
http://dx.doi.org/10.1177/1464420715596455 [in press].
[63] Bezerra MA, Santelli RE, Oliveira EP, Villar LS, Escaleira LA. Response surface
methodology (RSM) as a tool for optimization in analytical chemistry. Talanta
2008;76(5):965–77. http://dx.doi.org/10.1016/j.talanta.2008.05.019.
[64] Liu H, Bi X, Huo M, Lee CF, Yao M. Soot emissions of various oxygenated
biofuels in conventional diesel combustion and low-temperature combustion
conditions. Energy Fuels 2012;26(3):1900–11. http://dx.doi.org/10.1021/
ef201720d.
[65] McEnally CS, Pfefferle LD. Sooting tendencies of oxygenated hydrocarbons in
laboratory-scale flames. Environ Sci Technol 2011;45(6):2498–503. http://dx.
doi.org/10.1021/es103733q.
[66] Ghafari S, Aziz HA, Isa MH, Zinatizadeh AA. Application of response surface
methodology (RSM) to optimize coagulation–flocculation treatment of
leachate using poly-aluminum chloride (PAC) and alum. J Hazard Mater
2009;163(2–3):650–6. http://dx.doi.org/10.1016/j.jhazmat.2008.07.090.