Energy Conversion and Management 150 (2017) 222–241

Contents lists available at ScienceDirect

Energy Conversion and Management

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

Response surface methodology (RSM) based multi-objective optimization of MARK

fusel oil -gasoline blends at different water content in SI engine
⁎
Omar I. Awada, , R. Mamata,b, Obed M. Alic, W.H. Azmia, K. Kadirgamaa, I.M. Yusria,
A.M. Lemand, T. Yusafe
a
Faculty of Mechanical Engineering, Universiti Malaysia Pahang, 26600 Pekan, Pahang, Malaysia
b
Automotive Engineering Center, Universiti Malaysia Pahang, Malaysia
c
Technical Institute of Haweeja, Northern Technical University, Kirkuk, Iraq
d
Faculty of Engineering Technology, Universiti Tun Hussein Onn Malaysia, Malaysia
e
Faculty of Engineering and Surveying, University of Southern Queensland, Toowoomba, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: The main objective of this study is to determine the optimal blend ratio of fusel oil–gasoline before and after
RSM water extraction (FBWE10, FBWE20, FAWE10, and FAWE20) regarding the performance and emissions of spark
Fusel oil ignition engine using response surface methodology (RSM). The multi-objective optimization is applied to
Water content maximize the brake power, brake thermal efficiency and minimize the brake specific fuel consumption (BSFC),
Heating value
NOx emission, HC emission and CO emission. The water content of fusel oil has been extracted by employing
Engine performance
Engine emissions
rotary extractor method. The experimental of this study has been carried out with different fusel oil–gasoline
blends, different throttle valve opening position (15%, 30%, 45% and 60%) and different engine speed (1500,
2500, 3500 and 4500 rpm). All the developed models for responses were determined to be statistically sig-
nificant at 95% confidence level. The study results reveal an improvement in heating value of fusel oil after
water extraction with FAWE20 (80 vol% gasoline fuel, 20 vol% fusel oil after water extracted) as the optimally
blended fuel. The best condition of engine parameters with FAWE20 were 55.4% of WOT for load and 4499 rpm
engine speed. In additional of the optimal values with a high desirability of 0.707 were 62.511 kW, 241.139 g/
kW h, 36%, 1895.913 ppm140.829 ppm and % for brake power, BSFC, BTE, NOx, HC and CO emissions re-
spectively. The reduction of water content in fusel oil has a statistical significance influence to increases BTE,
NOx emission and decreases the BSFC, HC and CO emissions.

1. Introduction alcoholic fuel as an additive or blending with gasoline may be an at-

tractive solution, both at present and shortly for improving fuel prop-
The growing demands of fossil fuels due to the rapid development of erties thereby the performance and emissions of spark ignition engine
the industry and automotive society have been linked with environ- improve [12–21]. The octane number rapidly increases when oxyge-
mental pollution issues and encouraged the efforts on exploring alter- nated fuels are blended with gasoline due to the high research and
native fuels for ICE in the past decades [1–3]. Alcohols as alternative motor octane number of it.
fuel have a long history of approximately 100 years as fuels in internal Yucesu et al. [22] studied the impact of gasoline, E10, E20, E40 and
combustion engines (ICE). Alcohol based fuels may have been regarded E60 on the performance of spark ignition engine. It was found that
as one of the renewable solutions, with a potential to be used in a near engine torque and power increased when ethanol was used at higher
CO2-neutral manner through efficient conversion of biomass. Utilizing compression ratio (CR). Butanol that also called butyl alcohol can be
alcohol as a fuel in transportation especially in light cars is not new [4] used in non-modified spark ignition engines. One of the recent experi-
but only lately have such fuels begun to interest significant worldwide mental studies done by Elfasakhany [23] investigated the effects of
attention. Alcohols such as ethanol, butanol, methanol, and fusel oil are adding dual butanol isomers in an unleaded gasoline fuel towards its
used as fuels in ICE [5–7]. Alcohols have higher octane number, and performance, and emissions characteristics spark ignition engine. The
high oxygen content as compared to gasoline and its single boiling point fuels had been prepared according percentages of volume: 3% (1.5%
makes it very suitable for use in a spark-ignition engine [8–11]. The volume of iso-butanol and 1.5% volume of n-butanol), 7% (3.5%

⁎
Corresponding author.
E-mail address: omaribr78@gmail.com (O.I. Awad).

http://dx.doi.org/10.1016/j.enconman.2017.07.047
Received 29 January 2017; Received in revised form 18 June 2017; Accepted 21 July 2017
Available online 17 August 2017
0196-8904/ © 2017 Elsevier Ltd. All rights reserved.
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

volume of iso-butanol and 3.5% volume of n-butanol), and 10% (5% Table 1
volume of iso-butanol and 5% volume of n-butanol) of butanol in ga- Engine specifications.
soline blends. Following the mixture of dual butanol isomers and ga-
Engine type Mitsubishi 4G93 SOHC
soline blends, a slight decrease noted in engine torque, and brake
power. Methanol has not got the equivalent level of use as ethanol and Bore stroke 81.0 mm × 89.0 mm
butanol, but there is a quite substantial experience with methanol from Piston displacement 1.834 L
CR 9.5:1
various studies [24–26]. Turner et al. [27] investigated the impact of
Fuel injection type ECI-Multi (Electronically Controlled Multi-point) fuel
methanol–ethanol–gasoline blends on NOx emission. They used various injection
blend ratios to operate an SI engine and found that dual fuel blends Max power 86 kW@5500 rpm
could reduce NOx emissions to be lesser than that produced by pure Max torque 161 Nm@4500 rpm
gasoline. The experiment performed by Calam et al. 2013 [28] used a
Hydra brand, four cylinders spark ignition engine with a different ga-
soline-fusel oil blends (F0, F5, F10, F20, F30 and F50) under full load methodology (RSM) is a widely-used technique for solving many in-
condition, running at 3500 rpm. It was shown that adding fusel oil into dustrial problems. It is one of the most practical and economical solu-
gasoline lowered the calorific heat value of fusel oil (almost 30% lower tions for evaluating the single and combined factors of experiment
than unleaded gasoline) thereby led to lower calorific heat value of the variables that lead to output responses [46]. The results obtained by
test fuels. Hence, the amount of fuel’s mass taken into the cylinder in- RSM analysis will provide the best system performance for whole op-
creases with increasing amount of fusel oil in the blend. timized sets [47]. The main advantage is that design of experiments
On the other hand, the conventional water (moisture) volumetric based on RSM requires fewer tests and is less time to consume com-
content is about 5% when utilized as dedicated fuel [29,30]. Some al- pared to a real experimental study [48]. This approach is widely used
cohols such hydro-ethanol and fusel oil have high water content that and has been applied in many investigative studies. Studies regarding
considered one of an important property. The high water content and the examination of optimum blend ratios for gasoline - alcohols mix-
ash contents in biomass (biofuel) fuels can cause ignition and com- tures were reported by researchers [40,49–51]. Yusri et al. [40] used
bustion problems [31]. The water content effect on the combustion RSM optimization method to indicate the optimum blending ratio for
efficiency negatively thereby effect on engine performance [32]. gasoline -butanol blends under different engine speeds. Thus, they
Solmaz [33] informed the prolonged were in the flame development showed the optimum mixing ratio of gasoline– butanol regarding the
durations and flame propagation duration of fusel oil due to the higher performance and emission of spark ignition engine it was with GBu15
water content thereby the brake thermal efficiency decreased. Also, he (85% gasoline + 15% butanol) at 3205 rpm. In the study conducted by
stated the torque of SI engine decreased averagely by 6% and 2% when Najafi et al.[49], the engine-operating parameters were optimized uti-
pure and 50% of fuel oil used respectively while brake specific fuel lizing the desirability approach of RSM. E10 (90% gasoline + 10%
consumption (BSFC) increased. The research study by Omar et al. [7] ethanol) was found an optimum blending ratio of gasoline-ethanol at
also found the engine power and torque for F20 (20% fusel oil −80% 3000 rpm as engine speed. The results of this study reported that the
diesel) slightly dropped compared to those with pure diesel while the optimal input parameters, the values of the torque, brake power, BSFC,
fuel consumption increased. CO2, CO, NOx and HC, to be 103.66 Nm, 35.26 kW, 0.25 kg/kW h,
Fusel oil is a by-product obtained through the fermentation of some 12.8%, 3.5%, 1300 ppm and 136.6 ppm respectively.
agricultural products such as beets, cones, grains, potatoes, sweet po- From the literature review, the effects alcoholic fuels on the engine
tatoes, rice, and wheat. Most recently in Turkey, about 12 million tons performance and emission have been investigated. But the relationship
of sugar beets were prepared, and around 550.000 tons of beet molasses among throttle position (%WOT), engine speed, and volume of fusel oil
were taken per year. Almost 30 million liters of ethyl alcohol per year into gasoline, as well as with different water content on performance
were manufactured from that molasses. Approximately 1 Liter of and emissions of a spark ignition engine, has not been studied yet.
acetaldehyde and 5 Liter of fusel oil were obtained for every 1000 Liter Hence, the experimental of this study has been carried out with the
of ethyl alcohol from the distillation [34]. Furthermore, in Brazil, fusel partially opened throttle (%WOT), different engine speed and different
oil is frequently produced in the proportion of 2.5 Liter per 1000 Liter fusel oil- gasoline blends before and after water extraction. The higher
of bioethanol production [35]. Fusel oil has similar properties with water (moisture) content of fusel oil, despite the high-octane number
alcoholic fuel such as high research octane number (RON = 106), and oxygen content played a negative role to limitation the heating
motor octane number (MON = 103), high oxygen content and single value and the engine combustion. In this study, the water content of
boiling point that indicates it can be utilized as an alternative fuel for SI fusel oil has been extracted by employing rotary extractor method. The
engines [32]. On the other hand, the heating value of fusel oil influence of water content reduction on several factors such as the
(29.9 MJ/kg) is less than that of gasoline (40 MJ/kg) by 36% that has a oxygen, hydrogen, carbon, and different properties were provided and
major effect directly on the engine performance and the in particular discussed in addition to the lower heating value of fusel oil and blended
fuel consumption. Furthermore, the high water content of fusel oil fuel. Furthermore, an optimal blending ratio of fusel oil-gasoline based
(10–20%) ceased adverse effect on engine combustion and contributed on the engine performance and emissions has been determined.
to decreasing the heating value of fusel oil [33,36]. Solmaz [33] has
reported the torque of an SI engine has decreased on average by 6% and 2. Methodology and experimental setup
2%, respectively when pure and 50% of fusel oil used while the brake
specific fuel consumption (BSFC) has increased. 2.1. Materials and fuel preparation
To improve the performances and emissions characteristics of in-
ternal combustion engine optimization methods was used to optimize Engine testing was conducted with fusel oil-gasoline blends. The
the operation condition such as ignitions timing, injection timing, fusel oil supplied from Eskişehir sugar refinery, which is producing
speeds, loads, compression ratio and air-fuel ratio especially with al- ethyl alcohol with 99.5% purity. Gasoline was supplied from Shell
ternative fuel [37–45]. The principal technical advantage of optimiza- petrol stations in Pahang-Malaysia with octane number 95 provided
tion for a fraction of bio-origin components in gasoline fuel is enhan- gasoline fuel. The water content of fusel oil was extracted by using
cing engine performance and exhaust emissions as well as using the rotary evaporator (Buchi R-210, Switzerland). Different ratios of fusel
optimized mixture in a spark ignition engine without modification of oil with gasoline before and after extracted water content were pre-
engine. For the accomplishment of this real design of experiments pared (based on volume) through adding fusel oil to the gasoline mixing
(DoE) methods, namely RSM can be utilized. Response surface the blend for about 20 min using electrical magnetic stirrer at 2000 rpm

223
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

Fig. 1. Schematic of the experimental setup.

Table 2 fusel oil after water extraction) and FAWE 20 (80% gasoline and 20%
Automotive emission analyzer specifications. fusel oil after water extraction).
Exhaust gas Measurements range Measurement Resolution

NOx 0–5000 ppm 1 ppm 2.2. Fuel properties measurement

CO 0–9.99% 0.1%
CO2 0–16% 0.1% Fuel properties of blends were measured according to ASTM stan-
HC 0–5000 ppm 1 ppm
λ 0–2.000 0.001
dard methods. The moisture (water) content of fusel oil in all test
O2 0.00–25.00% 0.01 samples was determined according to ASTM D6304 standard test
method by using Karl Fischer Titration 870 in central laboratory of
University Malaysia Pahang (UMP). The higher heating value (HHV) of
Table 3 the fuel samples were determined according to ASTM D240, using
Parameters and their levels. Oxygen Bomb Calorimeter Model 6772 (Parr Instrument company,
Factors Parameters Level 1 Level 2 Level 3 Level 4
USA). The density of the fuel samples measured at 15 C according to
ASTM D4052 using Density/Specific Gravity Meter, model DA-640. The
Continuous A Load (% of 15 30 45 60 dynamic viscosities of the fuel samples were determined according to
factor WOT) the ASTM D445-01 fuel standards by using a Brookfield DV-II+
Continuous B Speed (rpm) 1500 2500 3500 4500
factor
Programmable Viscometer, and kinematic viscosity was calculated. The
Categorical C Fuel (%Vol) FBWE10 FAWE10 FBWE20 FAWE20 oxygen, carbon, hydrogen, and sulphur were measured by Intertek la-
factor boratories in Kuala Lumper –Malaysia. The carbon, hydrogen, nitrogen
and sulphur according to ASTM D5291, ASTM D5291, ASTM D5291
and ASTM D1552 respectively.
mixing speed. Prepared fule samples include FBWE10 (90% gasoline
and 10% fusel oil before water extraction), FBWE 20 (80% gasoline and
20% fusel oil before water extraction), FAWE10 (90% gasoline and 10%

Table 4
Measured fuel properties.

Property Test method G100 FBWE FAWE FBWE10 FAWE10 FBWE20 FAWE20

Higher heating value (MJ/kg) ASTM D 240 43.5 29.93 33.8 42.217 42.602 40.854 41.624
Boiling point (°C) ASTM D 2887 27–225 [52] 98.4 90.6 – – – –
Moisture content (%) ASTM D6304 0 13.5 6.5 1.35 0.65 2.7 1.3
Density (kg/m3) ASTM D 4052 769 844 843 777 776 785 783
Research octane number (RON) ASTM D 2699 95 106 106 96.1 – 97.2 –
Oxygen (%) 0 30.32 26.1 3.032 2.61 6.064 5.22
Carbon (%) ASTM D5291 87.5 54.2 58.45 84.17 84.595 80.84 81.69
Hydrogen (%) ASTM D5291 12.5 15.1 15.1 12.76 12.76 13.02 13.02
Sulphur ASTM D1552 0.1 0.38 0.28 0.128 0.118 0.156 0.136
Kinematic viscosity(mm2/s) 0.49 4.1588 4.1637 0.86048 0.86097 1.22696 1.22794
Lambda (ʎ) 0.905 – – 0.932 0.932 1.04 1.040

224
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

Fig. 2. Brake power (PW) versus throttle valve

position and engine speeds.

Table 5
ANOVA table for brake power (BP).

Source Sum of Squares Df Mean Square F Value p-value Prob > F PC%

Model 135.45 10 13.54 203.64 < 0.0001 Significant 97.47

A-Load 82.15 1 82.15 1235.1 < 0.0001 Significant 59.11
B-Speed 36.98 1 36.98 555.93 < 0.0001 Significant 26.61
C-Fuel 0.26 3 0.087 1.31 0.2807 Insignificant 0.19
AB 8.05 1 8.05 121.03 < 0.0001 Significant 5.79
A2 3.3 1 3.3 49.62 < 0.0001 Significant 2.37
B2 0.2 1 0.2 2.96 0.0912 Significant 0.14
AB2 3.45 1 3.45 51.83 < 0.0001 Significant 2.48
B3 1.06 1 1.06 16 0.0002 Significant 0.76
Residual 3.53 53 0.067 2.54
Cor Total 138.97 63 100

R-Squared 0.9745
Adj R-Squared 0.9698
Pred R-Squared 0.9630

80 80 80

53.3333 53.3333 53.3333

BP (KW)

BP (KW)
BP (KW)

26.6667 26.6667 26.6667

0 0 0

15 30 45 60 1500 2500 3500 4500 FBWE10 FAWE10 FBWE20 FAWE20

A: Load (% of WOT) B: Speed (rpm) C: Fuel (%)

Fig. 3. Main plots of brake power (BP) versus (A) load (B) speed (C) blended fuels.

2.3. Engine test setup 100 kW Dynalec Controls eddy current dynamometer was utilized in
the tests to measure the brake torque and power. Fuel consumption was
In this study, the tests were done on a Mitsubishi 4G93 SOHC 4- measured using an AIC fuel flow meter with an accuracy of 1%. The
cylinder 4-stroke port fuel injection naturally aspirated SI engine. The ignition timing was optimized by the minimum advance for best torque
engine specifications were given in Table 1. Fig. 1 shows the layout of (MBT) when the fraction of fusel oil was used. Furthermore, the engine
the experimental setup. The experimental of this study has been carried emissions and lambda (ʎ) were measured by the AUTOMOTIVE EMIS-
out with different throttle valve opening position (15%, 30%, 45% and SION ANALYZER QRO–401 with specifications listed in Table 2. For the
60%) and different engine speed (1500, 2500, 3500 and 4500 rpm). A calibration purpose, the gas analyzer was calibrated by the company

225
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

Interaction BP (KW)
B: Speed (rpm) 4500
80 60

Actual Factor
Fuel=Average over
50
53.3333 B- 1500
BP (KW)

B+ 4500
3500

B: Speed (rpm)
40
20
26.6667
30

0 2500

15 30 45 60
A: Load (% of WOT)
Fig. 4. Interaction between load and speed against brake power.

1500
before proceeding to conduct the tests to assure the accuracy of the 15 30 45 60
exhaust gas measurements. After every single test the zero calibration A: Load (% of WOT)
with fresh air was adjusted. To ensure the consistency and the accuracy
Fig. 6. Counter surface plots of brake power against of loads and speeds.
of the measurement data, the tests were conducted under steady state
conditions.
determined during the experiments were the brake power BP (KW),
BSFC (g/KW h), BTE (%), HC emissions (ppm) and NOx emission
2.4. Design of experiments analysis (ppm). Analysis of variance (ANOVA) data used to analyses the dif-
ferences between group means and their interactions regarding im-
The RSM is a useful method for any field of engineering [51,52]. It is portant parameters and square terms. In this analysis, DF describes the
a set of statistical techniques employed the linear or square polynomial degree of freedom value that represents the probability distribution in
functions to describe the relation between responses with its input repeated sampling. When an F-value is high, it presents the persuasive
variables to achieve the objective of maximization or minimization of evidence that the variation is a significant difference in the groups re-
the response properties. presented by each sample. The significance level was set at 95% con-
In the present experiment, there are three (3) influences parameters fidence interval (Prob > F to be maximum at 0.05). The percentage
which are throttle valve position (% of WOT), engine speed (rpm), and contribution (PC%) is often a rough but effective guide to the relative
blended fuels (%Vol). Based on RSM the customer-defined designs were importance of each model term [1]. The percentage contribution can be
used. According to the parameter that used in this experimental, the calculated as Eq. (1):
discrete function was applied. Discrete is defines the factor settings that
available to the experiment for an otherwise constant factor. Using SSd
PC % = ·100%
discrete factor settings can make the experiment more convenient to SST (1)
conduct while having minimal impact on the strength of the analysis.
The required number of these experiments for estimating the three where SSd is the sum of the squared deviations and SST, is the total sum
parameters with four levels should be considered; thereby the total of squared deviations.
experimental runs were 64 runs. The independent variables and their The simplest model which can be used in RSM is based on a linear
related levels and codes are tabulated in Table 3. The responses function presents the following Eq. (2):

Fig. 5. 3D surface plots of brake power against of throttle valve position (load)
and speeds.
80

53.3333
BP (KW)

26.6667

4500
3900 60
3300 45
2700
30
2100
B: Speed (rpm) A: Load (% of WOT)
1500 15

226
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

Fig. 7. Actual and predicted plot for (A) brake power, (B) BSFC, (C) BTE, (D) HC emissions and (E) NOx emission.

k k k
Y = β0 + ∑ βi Xi + ε Y = β0 + ∑ βi Xi + ∑ βij Xi Xj + ε
i (2) i i<j (3)

If this model presents any curvature, a second-order model must be For the present study, quadratic model is suitable to determine a
used as the following Eq. (3): critical function point (maximum, minimum) [3], using Eq. (4):

227
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

Fig. 8. Brake specific fuel consumption (BSFC) versus throttle valve position and engine speeds.

Table 6
ANOVA table for brake specific fuel consumption (BSFC).

Source Sum of Squares Df Mean Square F Value p-value Prob > F PC%

Model 37.01 27 1.37 24.83 < 0.0001 Significant 94.90%

A-Load 17.88 1 17.88 323.79 < 0.0001 Significant 45.85%
B-Speed 0.015 1 0.015 0.27 0.6075 Insignificant 0.04%
C-Fuel 5.77 3 1.92 34.84 < 0.0001 Significant 14.79%
A2 1.77 1 1.77 32.10 < 0.0001 Significant 4.54%
B2 2.33 1 2.33 42.13 < 0.0001 Significant 5.97%
AB2 4.15 1 4.15 75.21 < 0.0001 Significant 10.64%
B3 3.69 1 3.69 66.76 < 0.0001 Significant 9.46%
Residual 1.99 36 0.055 5.10%
Cor Total 39.00 63 100.00%

R-Squared 0.9490
Adj R-Squared 0.9108
Pred R-Squared 0.8094

k k k Experimental−Predicted
Y = β0 + ∑ βi Xi + ∑ βii Xi2 + ∑ ∑ βij Xi Xj + ε Percentage of absolute error(POAE) = ∗100
Experimental
i i i<j (4)
(5)
where k is the number of variables (for this case k = 3), xi, xj and x i2
are represent the variables. β0, βi , βii and βij are the constant term, the 3. Results and discussion
coefficients of the linear terms xi, the coefficients of the quadratic terms
x i2 and the coefficients of the interaction terms xixj respectively. ε is the In this section, the effect of parameters (throttle valve position,
residual associated to the experiments. engine speeds and blended fuels of fusel oil-gasoline) on performance
After that, the selection of optimum parameters condition can be and emissions of SI engine is discussed. Brake power, brake specific fuel
done for multi-responses to improving engine performance and mini- consumption (BSFC), brake thermal efficiency (BTE) and engine emis-
mizing emissions. The expected range of each engine response and the sions (NOx, HC and CO) are plotted by varying plots for the main and
range of parameters can be set and all engine response can be de- interaction effects. Three dimensional surface view plots and their
termined. In order to optimize these parameters, the desirability func- contour plots were obtained based on the three independent factors.
tion approach was carried out by using Design Expert v.10.1.3. The empirical model was fitted to make the relation between responses
To validate the reliability and the conclusion obtained from the and variables. At the end of this section, the multi-objective effects
statistical analysis a comprehensive between the experimental and optimized by using desirability function have been presented.
numerical data and the results obtained from the regression analysis has
been done. the regression equation expresses the relationship between
3.1. Fuel property analysis
the input and output variable. Moreover, the percentage of absolute
error (POAE) between the regression model and the experimental re-
Fuel property is an important indicator for the suitability of the
sults has been measured using Eq. (5):
blend to operate the existence SI engine efficiently. All the measured
properties for fusel oil-gasoline blends (before and after water extrac-
tion), pure fusel oil and pure gasoline are listed in Table 4.

228
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

A B
360 360

340 340

320 320
BSFC (Kg/Kw.h)

BSFC (Kg/Kw.h)
300 300

280 280

260 260

240 240

220 220

15 30 45 60 FBWE10 FAWE10 FBWE20 FAWE20

A: Load (% of WOT) C: Fuel (%)
Fig. 9. Main plots of BSFC versus (A) throttle value position (load) and (B) blended fuels.

Fig. 10. 3D surface plots of BSFC against engine loads and speeds.

353.318

309.208
BSFC (Kg/Kw.h)

265.098

220.988

15

4500
30
3500

A: Load (% of WOT) 45 B: Speed (rpm)

2500

60 1500

The higher water content of fusel oil, despite the high-octane 844 kg/m3 (109.8%) and 843 kg/m3 (109.6%) as shown in Table 4.
number and oxygen content played negatively to the limitation of the Accordingly, the higher fusel oil density at FBWE and FAWE will
heating value and the engine combustion. In this study, the water slightly compensate for their lower heating values compared with ga-
content of fusel oil was extracted by employing rotary extractor soline. Table 4 reveals that fusel oil has a higher viscosity of 4.1588 and
method. By extracting (reducing) the water content of fusel oil from 4.1637 mm2/s before and after water extracted compared to 0.59 mm2/
13.5% to 6.5% the heating value of fusel oil (FAWE) improved by 13% s of gasoline. Therefore, the usage of 10 and 20% fusel oil fraction with
compared with fusel oil before water extracted (FBWE) as shown in gasoline led to the increase of the viscosity of the blends. Almost all the
Table 4. On the correlation with the elemental composition of fusel oil, fusel oil properties changed after the water content reduction, thus the
the oxygen content reduced by 14% from 30.32 to 26.1% while the properties of fusel oil- gasoline blends also affected as illustrated in
carbon content increased by 7.9% from 54.2 to 58.45%, thereby Table 4. Therefore, the engine performance and emission are expected
heating value enhanced which is in agreement with previous studies to be affected by the change in the blended fuel properties.
[36,53]. The density of fusel oil is higher than gasoline, and this causes
a higher pressure drop and a decrease in the rate of fuel mass flow that
injected by a volumetric-operating pump. The density (about gasoline) 3.2. Engine performance optimization
of fusel oil before and after water extracted (FBWE and FAWE) were
The effect of fusel oil–gasoline blends before and after water content

229
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

Fig. 11. Brake thermal efficiency (BTE) versus throttle valve position and engine speeds.

Table 7
ANOVA table of brake thermal efficiency (BTE).

Source Sum of Squares Df Mean Square F Value p-value Prob > F PC%

Model 2.12 8 0.27 205.15 < 0.0001 Significant 96.8%

A-Load 1.41 1 1.41 1086.77 < 0.0001 Significant 64.4%
B-Speed 0.36 1 0.36 276.23 < 0.0001 Significant 16.4%
C-Fuel 0.25 3 0.083 63.83 < 0.0001 Significant 11.4%
AB 0.028 1 0.028 21.89 < 0.0001 Significant 1.3%
A2 0.07 1 0.07 54.41 < 0.0001 Significant 3.2%
B2 0.014 1 0.014 10.45 0.0021 Significant 0.6%
Residual 0.071 55 0.001294 3.2%
Cor Total 2.19 63 100.0%

R-Squared 0.9676
Adj R-Squared 0.9629
Pred R-Squared 0.9565

A B C
38 38 38

34 34 34
BTE (%)

BTE (%)
BTE (%)

30 30 30

26 26 26

15 30 45 60 1500 2500 3500 4500 FBWE10 FAWE10 FBWE20 FAWE20

A: Load (% of WOT) B: Speed (rpm) C: Fuel (%)
Fig. 12. Main plots of BTE versus (A) throttle value position, (B) speeds and (C) blended fuels.

230
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

Interaction position (engine load) has higher effects on the brake power, that in-
38 B: Speed (rpm) dicating reliability in the estimation of brake power. A high R-Squared
Actual Factor
Fuel=Average coefficient ensures a satisfactory agreement between the calculated and
over observed data [55]. The percentage contribution “PC%” is often an
B- 1500 effective guide to the relative importance of each model term [55]. It
34
B+ 4500 was seen the engine load has highest contribution effects (59.11%) of
BTE (%)

total variability on the brake power than engine speed (26.61%). While
the blended fuel has insignificant effects on the engine power.
30 The main effects that influence brake power response are engine
load, speed and interaction effect plotted between them, with no effect
by fuel factor. Fig. 3 illustrates the main effects plots for brake power. In
this plot, the greater main effect is depicted by a line with a steeper
26 slope for long range change compared to the effects contributed by less
significant factors [1]. The plot for brake power against its load and
15 30 45 60
speed shows a significant positive effect from the low level to high
A: Load (% of WOT) level. However, Fig. 3(a) shows the value of brake power is a significant
Fig. 13. Interaction between load and speed against BTE. increase from 11.8 to 48.6 kW. The plot of speed Fig. 3(b) shows a less
slope than load. Whereas the plots of brake power against blended fuel
reduction in addition to different throttle valve position and engine showed that there is an increase in the brake power with increased the
speeds on the performance and emissions of SI engine statically studied. ratio of fusel oil in the blended fuel, but it was insignificant effect as
The brake power (PB) is useful power at the output shaft also is the shown in Fig. 3(c). The brake power improved by extracting the water
function of the brake torque and speed of the engine. The brake torque content from fusel oil that could be explained by the water content of
is recorded directly from the dynamometer controller as the engine fusel oil played the main role to limit the combustion and heating value
output. Fig. 2 illustrates the engine power for various gasoline – fusel of fusel oil despite the higher oxygen content of fusel oil, thus a re-
oil blends after and after water content extraction at the different duction in water content led to better engine combustion[53]. Fig. 4
throttle valve position and engine speeds. The brake power increased indicates that the load effect has slight effect when the speed is at the
with the increasing the engine speed and throttle valve position. Fur- low level and very high effects when the speed is at the high level, the
thermore, it observed that the fusel oil after water content extraction best results of brake power obtained with a high level of speed and load.
(FAWE10 and FAWE20) had a slightly higher power when compared to Three-dimensional (3D) surface plotted, in order to understand the
fusel oil before water content extraction (FBWE10 and FBWE20) and it interaction effect between variables and responses. Fig. 5 displays the
was averagely higher by 0.2% and 0.68% respectively. The water 3D surface plot for the relation between the input factors (load, speed,
content may be limited to the engine combustion; thereby, the reduc- and blended fuel) and brake power. It has seen the engine load has
tion in water content led to better engine combustion. Furthermore, the more affected on brake power than blended fuel. The plots show the
improvement in power with FAWE10 and FAWE20 may be due to the maximum value of the brake power was when the highest value of the
higher heating value of FAWE when compared to the heating value of load and the speed which is agreed with previous study conducted by
FBWE. Najafi et al. [44]. It has seen the clearly significant change on the
Table 5 illustrations the ANOVA data of brake power. The F-value of surface plot when the load change from low to high value while slightly
this model 203.64 and p-value was 0.0001 less than 0.05 thereby the change for speed. The counter surface Fig. 6 shows the desired regain
model is significant. A high R2 value, close to 1, is desirable, and rea- which represents high power obtained at the condition speed is within
sonable agreement with adjusted R2 is necessary [54]. Also, R2 value the range of 3900–4500 rpm, load in the range of 51–60. Similar results
indicates the total variability of responses after considering the sig- were obtained by [33,56] that indicated the brake power of SI engine
nificant factors and the value account for the number of predictors in increased by increasing the engine loads and speeds. As a conclusion,
the model. According to the P-value, ANOVA table indicated that the statically the highest increasing in brake power linked with engine
load and speed factors were significant terms, while the fuel was not loads and speed. The fusel oil blends statistically have an insignificant
significant. The statistical result in Table 4 also shows the throttle valve impact on brake power.

Fig. 14. 3D surface plots of BTE against engine loads and speeds.

38

34
BTE (%)

30

26

4500 60

3500 45

2500 30
B: Speed (rpm) A: Load (% of WOT)
1500 15

231
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

Fig. 15. NOx emissions versus throttle valve position and engine speeds.

Table 8
ANOVA table of NOx emission.

Source Sum of Squares Df Mean Square F Value p-value Prob > F PC%

Model 2749.35 11 249.94 252.15 < 0.0001 Significant 98.19%

A-Load 2331.74 1 2331.74 2352.39 < 0.0001 Significant 83.28%
B-Speed 76.36 1 76.36 77.04 < 0.0001 Significant 2.73%
C-Fuel 51.2 3 17.07 17.22 < 0.0001 Significant 1.83%
A2 202.97 1 202.97 204.77 < 0.0001 Significant 7.25%
B2 21.04 1 21.04 21.23 < 0.0001 Significant 0.75%
A2B 24.35 1 24.35 24.57 < 0.0001 Significant 0.87%
AB2 7.67 1 7.67 7.74 0.0075 Significant 0.27%
A3 28.02 1 28.02 28.27 < 0.0001 Significant 1.00%
Residual 51.54 52 0.99 1.84%
Cor Total 2800.89 63 100.03%

R-Squared 0.9816
Adj R-Squared 0.9777
Pred R-Squared 0.9732

A B C
2481.11 2481.11 2481.11

1832.96 1832.96 1832.96

NOx (PMM)

NOx (PMM)
NOx (PMM)

1184.81 1184.81 1184.81

536.664 536.664 536.664

16.139 30.7224 45.3057 59.889 1500 2500 3500 4500 FBWE10 FAWE10 FBWE20 FAWE20
A: Load (% of WOT) B: Speed (rpm) C: Fuel (%)

Fig. 16. Main plots of NOx versus (A) throttle value position, (B) speeds and (C) blended fuels.

232
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

Fig. 17. 3D surface plots of NOx against engine loads and speeds.

2500
NOx (PMM)

1833.33

1166.67
60
500

45
4500

3500 30
B: Speed (rpm) A: Load (% of WOT)
2500

1500 15

Fig. 18. HC emissions versus throttle valve position and engine speeds.

Table 9 The difference between the actual and predicted values is called
ANOVA table of HC emission. residuals [57]. Fig. 7 showed actual and predicted plot for brake power,
BSFC, BTE, HC emissions and NOx emission. The general impression
Source Sum of Df Mean F Value p-value PC%
Squares Square Prob > F
from actual and predicted data display is that the error value is ap-
proximately acceptable and there no large deviation between them.
Model 175.41 6 29.23 168.68 < 0.0001 Significant 95% Also, Fig. 7 showed the regression models for all responses were de-
A-Load 123.22 1 123.22 710.95 < 0.0001 Significant 67% termined to be statistically significant under 95% confidence level.
B-Speed 37.65 1 37.65 217.22 < 0.0001 Significant 20%
C-Fuel 3.45 3 1.15 6.63 0.0006 Significant 2%
The energy content of a fuel has a direct influence on brake specific
AB 11.1 1 11.1 64.03 < 0.0001 Significant 6% fuel consumption (BSFC) [31,58]. The brake specific fuel consumption
Residual 9.88 57 0.17 5% (BSFC) is the ratio between fuel mass consumption and the brake power
Cor Total 185.29 63 100% [59]. BSFC is one of the significant parameters when the engine is
R-Squared 0.9467 working with different fuels, which exhibits the performance of fuels
Adj R- 0.9411 [60]. It is an indicator of how efficiently the engine is producing work
Squar-
with different fuels. In theory, BSFC is bound to increase upon the in-
ed
Pred R- 0.9329
clusion of an alcohols fuel additive owing to the lower heating value of
Squar- the fuel blend demanding more fuel to be injected to obtain the same
ed power output [61].
Fig. 8 shows BSFC of fusel oil–gasoline blends before and after water
extracted at different throttle value position and speeds. The BSFC was
decreasing as the engine loads increased. Logically, fuel amount con-
sumption (mass of fuel that taken by the engine cylinder) increased by

233
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

A B C
250 250 250

183.333 183.333 183.333

HC (PMM)
HC (PMM)

HC (PMM)
116.667 116.667 116.667

50 50 50

15 30 45 60 1500 2500 3500 4500 FBWE10 FAWE10 FBWE20 FAWE20

A: Load (% of WOT) B: Speed (rpm) C: Fuel (%)
Fig. 19. Main plots of HCversus (A) throttle value position, (B) speeds and (C) blended fuels.

Fig. 20. 3D surface plots of HC emissions against engine loads and

speeds.

250

183.333
HC (PMM)

116.667

50

60

1500
45
2500

A: Load (% of WOT) 30 3500 B: Speed (rpm)

15 4500

increasing the engine load; however, the BSFC decreased with load decreeing in BSFC was significant with load and highest contribution
engine decreased. Besides, the BSFC of fusel oil–gasoline blends after effects (34.6%) on BSFC with the load compared with other parameters.
water extraction (FAWE10 and FAWE20) was lower than the blends Fig. 9(b) shows the effects of blended fuels on BSFC. The BSFC of fusel
before water extraction (FBWE10 and FBWE20). This can be explain by oil–gasoline blends after water extraction (FAWE10 and FAWE20) was
the improvement in the heating value of fusel oil. lower than the blends before water extraction (FBWE10 and FBWE20).
Table 6 displays the ANOVA data for BSFC. The throttle value po- This could be explained by the lean and rich mixture condition ac-
sition (load) and blended fuel have significant impacts on the BSFC cording to the lambda (λ) that taken by the gas analyzer as shown in
while the engine speeds insignificant. The P-value of the model is less Table 1. The λ for 10% fusel oil concentration (FBWE10 and FAWE10)
than 0.0001% indicates the model is significant. The value of R-Squared and 20% concentration (FBWE20 and FAWE20) is determined to be
74% implies the model is acceptable. The difference between Pred R- 0.932 and 1.04, respectively. The existence of lambda different from
Squared and Adj R-Squared is less than 0.2, the main effect of throttle one could be explained by the variation of combustion between the lean
value position dominates in this process with 45.85% of the total and rich mixture condition. When the engine operated with lean fuel
variability, compared to 14.78% for blended fuel. means the amount of fuel injection becomes small thereby the BSFC
Fig. 9 illustrations effect of engine loads, speeds and blended fuels decreased. As a conclusion, the BSFC has improved with the extraction
on BSFC. Fig. 9(a) reveals that there has been decreasing in the BSFC of water content and the higher BSFC is recorded with FAWE10 under
with increased the engine loads. It is possible to understand the BSFC rich.
affected by the brake power thereby the BSFC decrease with the in- Fig. 10 displays the 3D surface plot for the relation between the
crease of engine load [62,63]. According to Table 5 and Fig. 9(a), the input factors (load, speed, and blended fuel) and BSFC. It has seen the

234
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

Fig. 21. CO emissions versus throttle valve position and engine speeds.

Table 10 The BTE of fusel oil–gasoline blends after water extraction (FAWE10
ANOVA table of CO emission. and FAWE20) were slightly higher than the blends before water ex-
traction (FBWE10 and FBWE20). This is attributed to the higher flame
Source Sum of Df Mean F Value p-value PC%
Squares Square Prob > F
speed for a blend that drives to a higher degree of constant volume heat
addition process for the ideal cycle. Moreover, the improvement in the
Model 1.03 13 0.079 25.41 < 0.0001 Significant 87% BTE with fusel oil may be attributed to the higher reaction activity with
A-Load 0.54 1 0.54 173.36 < 0.0001 Significant 45% the increasing additive fraction of fusel oil under fuel rich conditions
B-Speed 0.25 1 0.25 81.43 < 0.0001 Significant 21%
C-Fuel 0.05 3 0.017 5.29 0.003 Significant 4%
thereby may lead to shorter combustion duration. Also this could be
AB 0.046 1 0.046 14.84 0.0003 Significant 4% explained by the improvement in the heating value of fusel oil.
AC 1.75E- 3 5.85E- 0.19 0.9047 Insignificant 0% Table 7 displays ANOVA data for brake thermal efficiency (BTE). It
03 04 is shown that the impact of the parameter on BTE were the engine
A2 8.34E- 1 8.34E- 2.67 0.1087 Insignificant 1%
loads, engine speeds, and blended fuels by the contribution effects 64.4,
03 03
B2 0.055 1 0.055 17.47 0.0001 Significant 5% 16.4 and 11.4 respectively. The model F-value of 205.15 implies that
A2B 0.042 1 0.042 13.29 0.0006 Significant 4% the model is significant. Additionally, the value of R-Squared 0.96%
AB2 0.034 1 0.034 10.82 0.0018 3% that indicates the model is satisfactory.
Residual 0.16 50 3.12E- 13% Fig. 12 illustrates the main effects plots for BTE. In this plot, the
03
Cor Total 1.19 63 100%
greater main effect is depicted by a line with a steeper slope for long
range change compared to the effects contributed by less significant
R-Squared 0.8685
factors [1]. However, Fig. 12(a) shows the value of BTE is a significant
Adj R- 0.8344
Squar- increase from 31 to 43(%). The plot of speed Fig. 12(b) shows a less
ed slope than load. The plots for BTE against blended fuel showed that
Pred R- 0.8185 there is an increase BTE with increased the ratio of fusel oil in the
Squar- blended as shown in Fig. 12(c). The AB interaction Fig. 13 indicates that
ed
the load effect is very small when the speed is at the low level and very
high effects when the speed is at the high level, the best results of BTE
engine load has more affected on BSFC than blended fuel. The plots obtained with a high level of speed and load.
show the minimum value of the BSFC was when the highest value of Fig. 14 displays the 3D surface plot for the relation between the
load and 2500 rpm engine speeds. It has seen the clearly significant input factors (load, speed, and blended fuel) and BTE. The plots show
change on the surface plot when the load change from low to high value the maximized value of the BTE was when the highest value of the load
while slightly change for speed. and the speed. It has seen the clearly significant change on the surface
The engine brake thermal efficiency (BTE) is described as the brake plot when the load change from low to high value while slightly change
power of a heat engine as a function of the heat input from the fuel. It is for speed. Similar results that obtained by [33,56] indicated the BTE of
used to estimate how well an engine converts the heat from fuel to SI engine increased by increasing the engine loads and speeds.
mechanical energy [64]. The variations of brake thermal efficiency
(BTE) for fusel oil–gasoline blends before and after water extracted at 3.3. Engine emissions optimization
different throttle value position (load) and speeds are shown in Fig. 11.
Engine emissions such as nitrogen oxides (NOx) and hydrocarbon

235
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

A B C
3 3 3

2.5 2.5 2.5

CO emisssion (%)

CO emisssion (%)

CO emisssion (%)
2 2 2

1.5 1.5 1.5

1 1 1

15 30 45 60 1500 2500 3500 4500 FBWE10 FAWE10 FBWE20 FAWE20

A: Load (%of WOT) B: Speed (rpm) C: Fuel (% )
Fig. 22. Main plots of CO versus (A) throttle value position, (B) speeds and (C) blended fuels.

Fig. 23. 3D surface plots of CO emissions against engine loads and speeds.

2.5
CO emisssion (%)

1.5

1500
60
2500
45

30 3500
A: Load (%of WOT) B: Speed (rpm)
15 4500

Table 11
constraints for optimization of cutting parameters and responses.

Name Goal Lower Limit Upper Limit Lower Weight Upper Weight Importance

A: Load Is in range 15 60 1 1 3
B: Speed Is in range 1500 4500 1 1 3
C: Fuel Is in range FBWE10 FAWE20 1 1 3
BP Maximize 15 70.45 1 1 3
BSFC Minimize 230.875 350 1 1 3
TE Maximize 29 36 1 1 3
NOx Minimize 1000 2240 1 1 3
HC Minimize 110 240.029 1 1 3
CO Minimize 1.535 2.813 1 1 3

(HC in spark ignition engines are related to the engine operating con- of NOx formation as a by-product [67]. Many kinds of literature had
dition, the homogeneity of the fuel with air and the fuel properties. The shown that NOx emission decreased with the increase in the content of
nitrogen oxides (NOx) are a mixture of such compounds: nitrogen di- alcoholic fuel [16,33,68,69]. The variations of NOx emissions of fusel
oxide (NO2), nitric oxide (NO), dinitrogen trioxide (N2O3), dinitrogen oil–gasoline blends before and after water extracted at different throttle
tetroxide (N2O4), dinitrogen pentoxide (N2O5) and nitrous oxide (N2O) value position (load) and speeds are shown in Fig. 15. In general; NOx
[65]. The NO and NO2 are the most famous of the nitrogen oxides emissions were increased as the engine loads increased from 15 to 45
because the others are in very small quantities [66]. Oxidation of ni- (% of WOT). It was seen the NOx emission of fusel oil–gasoline blends
trogen molecules at a high temperature inside the cylinder is the cause after water extraction (FAWE10 and FAWE20) were slightly higher than

236
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

Table 12
Solution of optimized cutting parameters for work material.

No Load Speed Fuel BP BSFC TE NOx HC CO Desirability

1 55.444 4499.999 FAWE20 62.511 241.139 36.000 1895.913 140.829 2.337 0.707
2 55.680 4499.996 FAWE20 62.681 241.067 36.015 1896.884 141.105 2.337 0.707
3 56.119 4500.000 FAWE20 62.993 240.922 36.042 1898.803 141.621 2.338 0.707
4 56.715 4499.999 FAWE20 63.409 240.701 36.077 1901.669 142.322 2.338 0.707
5 56.926 4499.997 FAWE20 63.554 240.616 36.089 1902.760 142.571 2.336 0.707
6 54.549 4500.000 FAWE20 61.850 241.374 35.942 1892.579 139.780 2.339 0.706
7 58.185 4500.000 FAWE20 64.392 240.027 36.159 1910.210 144.060 2.333 0.705
8 58.499 4499.999 FAWE20 64.593 239.858 36.175 1912.337 144.432 2.331 0.705
9 53.800 4500.000 FAWE20 61.282 241.529 35.892 1890.159 138.908 2.32 0.705
10 53.115 4500.000 FAWE20 60.748 241.639 35.844 1888.176 138.111 2.134 0.703
11 58.104 4500.000 FBWE20 64.904 240.069 35.153 1862.001 151.556 2.133 0.697
12 57.932 4499.998 FBWE20 64.791 240.157 35.144 1860.912 151.347 2.135 0.697
13 57.712 4499.995 FBWE20 64.646 240.265 35.132 1859.567 151.079 2.136 0.697
14 58.817 4499.996 FBWE20 65.361 239.677 35.190 1866.880 152.425 2.132 0.697
15 57.340 4499.997 FBWE20 64.397 240.438 35.112 1857.411 150.628 2.137 0.697

the blends before water extraction (FBWE10 and FBWE20), which contribution on HC emissions were the engine loads, engine speeds, and
could explain the higher water content of fusel oil. blended fuels by 67%, 20% and 2% respectively.
Table 8 illustrated ANOVA data for NOx emission. The significant Fig. 19 shows the individual influence of parameters (load, speed,
models, in this case, are load, speed, and blended fuels. The F-value of and blended fuel) on HC emission. The load was significant factor in-
this model 252.15 and p-value was 0.0001 less than 0.05 thereby the fluence on the HC as shown in Fig. 19(a) also it can see the HC in-
model is significant. The value of R-Squared 98% implies the model is creased from 15–45% engine load while at high load the HC decreased.
suitable. The difference between Pred R-Squared (0.977) and Adj R- The curve of HC versus speed shows that there is a dropped in the HC
Squared (0.97) is less than 0.2, the main effect of load and its square are emission curve by increasing the speed as shown in Fig. 19(b). More-
dominated in this process with 83.28% and 2.7%, respectively. It is over, the HC slightly increased as blended fuel (percentage of fusel oil)
shown that the impact of parameters contribution on NOx emissions increased as shown in Fig. 19(c). Also, the fusel oil blends after water
were the engine loads, engine speeds, and blended fuels by 83.28, 2.73 content extracted have lower HC emission.
and 1.83 respectively. Fig. 20 shows the 3D surface plots of HC emission against engine
Fig. 16 shows the individual influence of parameters (load, speed speed, load. It was seen from 3D surface plot the engine load has a
and blended fuel) on NOx emission. The load was significant factor significant effect on increasing the value of HC emission. Also, the plots
influence on the NOx as shown in Fig. 16(a) also it can see the NOx showed the mixumum value of the HC emission was at higher engine
increased from 15 to 45% engine load while at high load the NOx de- load and lower engine speed.
creased. The curve of NOx versus speed shows that there is slightly Fig. 21 shows the variations of CO emissions of fusel oil -gasoline
change in the curve by increase the speed as shown in Fig. 16(b). blends before and after water content extraction at different throttle
Moreover, the NOx decreased as blended fuel (percentage of fusel oil) valve position and engine speeds. It was seen the CO emissions in-
increased as shown in Fig. 16(c). Also, the fusel oil blends after water creased with increasing the blends of fusel oil -gasoline and engine
content extracted have higher NOx emission. loads. While the CO emissions decreased as engine speeds increased.
Fig. 17 present the 3D surface plots of NOx emission against engine Furthermore, the CO emission of fusel oil –gasoline blends after water
speed, load. It clearly from 3D surface plot the engine load has a sig- extraction (FAWE10 and FAWE20) were slightly lower than the blends
nificant effect on increasing the value of NOx emission. The plots show before water extraction (FBWE10 and FBWE20).
the maximized value of the NOx emission was when the highest value of Table 10 illustrations the ANOVA data of CO emissions of fusel oil
the load and the speed. It has seen the clearly significant change on the -gasoline blends before and after water content extraction at different
surface plot when the load change from low to high value while slightly throttle valve position and engine speeds. The F-value of this model
change for speed which is agreed with previous study conducted by 25.41and p-value was 0.0001 less than 0.05 thereby the model is sig-
Yusri et al. [40]. nificant. Also, the R2 value was 86%. According to the P-value, ANOVA
HC emissions represented the unburnt fuel present in the in-cylinder table indicated that the throttle valve position (engine load) and speed
combustion, and this reduced the thermal efficiency and produced air factors were significant terms, while the fuel has less effects. The per-
pollution [70]. HC emissions were small due to a complete combustion centage contribution it was seen the engine load has highest contribu-
process [71]. Fig. 18 shows the variations of HC emissions of fusel oil- tion effects (45%) of total variability on the CO emissions than engine
gasoline blends and pure gasoline. It was seen the HC emissions in- speed (21%). While the blended fuel has 4%.
creased with increasing the blends of fusel oil-gasoline and engine Fig. 22 shows the individual influence of parameters (load, speed
loads. While the HC emissions decreased as engine speeds increased. and blended fuel). The plot of CO emission against its load, speed and
Furthermore, the HC emission of fusel oil–gasoline blends after water fuel shows the effect from high level to low level. Fig. 22(a) shows the
extraction (FAWE10 and FAWE20) were slightly lower than the blends CO emission increased as engine load increased. While CO emissions
before water extraction (FBWE10 and FBWE20) which was around 5%. decreased with increasing engine speeds as shown in Fig. 22(b). Fur-
Due to the reduced water content, the engine combustion temperature thermore, Fig. 22(c) shows the CO emission increased as the fraction of
had increased; thereby the HC emission decreased. fusel oil increased. While it has noted that slightly reduction in the CO
Table 9 illustrations ANOVA data for HC emission. The significant emission after fusel oil after water extraction (FAWE10 and FAWE20).
model, in this case, are load, speed and blended fuels. The F-value of Over all, the water content of fusel oil led to increased CO emissions
this model 168.68 and p-value was 0.0001 less than 0.05 thereby the also that could be explained the slightly reduction in CO after water
model is significant. The value of R-Squared 95% implies the model is extraction.
suitable. The difference between Pred R-Squared (94%) and Adj R- Fig. 23 presents the 3D surface plots of CO emission against engine
Squared (93%) is less than 0.2. It is shown that the impact of parameter speed and load. It clearly from 3D surface plot the engine loads has a

237
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

Desirability BP (KW) Fig. 24. Counter Plot of loads and speeds for
4500 4500 60 (A) desirability, (B) brake power, (C) BTE,
Desirability 0.707063 Prediction 62.5111
0.6 (D) BSFC, (E) NOx emission, (F) HC emis-
3900 3900 50 sions and (G) CO emissions.
0
B: Speed (rpm)

B: Speed (rpm)
40
3300 3300
0 20
0.2 0.4
30

2700 2700
0

2100 0 2100
0.4
0

1500 1500
15 24 33 42 51 60 15 24 33 42 51 60
A: Load (% of WOT) A: Load (% of WOT)
A B
BSFC (Kg/Kw.h)
TE (%) 4500
4500 Prediction 241.138
Prediction 36.0001 36

3900
3900 B: Speed (rpm)
B: Speed (rpm)

300
260
3300 280
3300
34

32
2700
2700
240

2100 2100
30

1500 1500
15 24 33 42 51 60 15 24 33 42 51 60

C A: Load (% of WOT) A: Load (% of WOT)

D
NOx (PMM) HC (PMM)
4500 4500
100 Prediction 140.829
Prediction 1895.92

3900 3900
B: Speed (rpm)

B: Speed (rpm)

3300 3300 150

2000
1000
1500
2700 2700

200
2100 2100

1500 1500
15 24 33 42 51 60 15 24 33 42 51 60

E A: Load (% of WOT) A: Load (% of WOT)

F
CO emisssion (%)
4500
Prediction 2.33732

3900
1.6
B: Speed (rpm)

3300 1.8

2700 2.2

2.4
2100 2.6

1500
15 24 33 42 51 60
A: Load (%of WOT) G

significant effect on increasing the value of CO emission. While the 3.4. Multi-objective
engine speeds, has less effect on decrease the CO emission. The plots
show the minimum value of the CO emission was when the highest Desirability function approach was carried out to obtain optimum
value of speeds and lower engine loads parameters for multi-responses as well as to achievement of objectives
which maximize the brake power and BTE as well as to minimize the
brake specific fuel consumption (BSFC), NOx emission and HC emission.

238
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

Table 13
Validation of test results.

No Engine load (% of WOT) Engine speed (rpm) Fuel type Brake power (kW) BSFC (g/kW h) BTE (%) NOx (ppm) HC (ppm) CO (%)

1 53.84 4500 FAWE20 Predicted 61.315 238.9 36.4 1925.35 138.7 2.27
Experimental 62.7 244.5 35.5 1960 140.9 2.38
POAE (%) 2% 2% −3% 2% 2% 5%

2 53.63 4499.9 FAWE20 Predicted 61.153 238 36.9 1924.28 138.4 2.27
Experimental 63.47 245.7 36.1 1980.3 142.3 2.4
POAE (%) 4% 3% −2% 3% 3% 5%

3 53.39 4499.9 FAWE20 Predicted 60.97 238.3 36.2 1923 138 2.27
Experimental 62.34 246.9 36.4 1985.6 143.7 2.32
POAE (%) 2% 3% 1% 3% 4% 2%

4 52.83 4499.9 FAWE20 Predicted 60.528 238.1 36.1 1920.24 137.8 2.27
Experimental 62.01 248.1 36.8 1970.9 140.1 2.34
POAE (%) 2% 4% 2% 3% 2% 3%

Table 11 shows a summary all the criteria applied to find the optimum efficiency (BTE), NOx emission and decreases BSFC, HC and CO emis-
settings and constraints imposed used to perform the multi-objective sions as well as has insignificant effects on brake power.
optimization. The optimum was setting for input factors (engine para-
meters and blended fuel) in the range. In this study, assumed that the 3.5. Validation of optimization results
five response features are equally important (i.e. weight w = 1:1:1:1).
Also, shows the desirable setting for parameters to achieve the goals of In order to validate the optimized results, the experiments were
responses. Desirability analysis was performed on the response values performed best 4 solutions that obtained from multi-optimization based
with the-larger-the-better desirability function, and the highest desir- on RSM. The optimal conditions of engine load, engine speed and
ability value was selected as the optimal condition [70]. The results of blende fuel were listed in Table 13. There was a significant positive
the multi-optimization analysis labeled in Table 12. A total of 15 de- correlation between all of the output responses at the optimum condi-
sirable results were obtained, and the solutions with desirability values tions, which indicates a combined desirability value of 0.703. It was
close to 1 are the best options. observed that the percentage of absolute error in prediction of devel-
Thus, the best solution was with 20% of fusel oil after water ex- oped models was in acceptable agreement. The validation of results
tracted (FAWE20) at higher load and speed. Solution (underline values) indicated that the models developed were quite accurate for perfor-
1 was selected with 0.707 desirability value as shown in Table 12. mance characteristics with a percentage of absolute error lower than
Moreover, this selection predicted the highest possible combined values 4%. However, for exhaust emissions characteristics, the percentage of
of brake power, BTE and lowest combined values of BSFC, NOx and HC absolute error less than 5%. This was perhaps due to the lack of com-
emissions. The best condition of parameters with FAWE20 were 55.4 bined desirability value, which was contributed by lack of experimental
(% of WOT) for load and 4499.99 (rpm) engine speed as listed in numbers. Finally, the validation results show good agreement with the
Table 10. Also, the predicted values were 62.511 (kW), 241.139 (g/ experimental data with maximum percentage of error which is less than
kW h), 36 (%) 1895.913 (ppm),140.829 (ppm) and 2.334 for brake 6%.
power, BTE, BSFC, NOx,HC and CO emissions respectively. It is inter-
esting to note that, with FAWE20, the engine would benefit regarding 4. Conclusion
the performance and emissions characteristics, especially at high engine
load and speed. This is principally due to the desirable fuel properties of The DoE based on RSM was highly helpful to design the experiment
fusel oil specifically the higher-octane rating, oxygen content and and the statistical analysis to identify the significant parameters that are
higher heat of vaporization [33,71]. most impacting on the engine performance and emission. The aim of
Fig. 24 shows the contour graphs of brake power, BTE, BSFC, NOx this study was to determine the optimum blending ratios of fusel oil
and HC at maximum desirability value, for selected solution (FAWE20) gasoline blends regarding performance and exhaust emissions of SI
after the multi-optimization has been done. The desirability value as engine based on various throttle position (%WOT), engine speeds, and
shown in Fig. 24(a) is good to expect the values of parameters for multi- amount of fusel oil into gasoline in order to minimize the brake specific
objectives optimization. Fig. 24(b) and (c) shows the brake power and fuel consumption (BSFC) nitrogen oxide (NOx), hydrocarbon (HC),
BTE increased as engine loads and speeds increase so the highest brake carbon monoxide (CO) as well as to maximize brake power (BP) and
power and BTE were with highest load and speeds. The minimum value brake thermal efficiency (BTE). From the outcome of this research, the
of the BSFC was when the highest value of the load and 2500 rpm following conclusions are drawn:
engine speeds as shown in Fig. 24(d). The minimum value of the NOx
and HC emissions were at lowest engine load as shown in • All the developed regression models for responses were determined
Fig. 24(e) and (f). The desired regain which represents lower CO to be statistically significant under 95% confidence level.
emissions obtained at the condition speed is within the range of • FAWE20 (80 vol% gasoline fuel, 20 vol% fusel oil) was the opti-
3500–4500 rpm, load in the range of 15–30 as shown in Fig. 24(g). It mally blended fuel.
has seen the clearly significant change on the surface plot when the load • The best condition of engine parameters with FAWE20 were 55.4 (%
changes from low to high. of WOT) for load and 4499.99 rpm. In additional the optimal values
Overall, it could be concluded based on the result and statistical with a high desirability of 0.707 were 62.511 kW, 241.139 g/kW h,
analysis the throttle value position (load (is significant impact para- 36% 1895.913 ppm, 140.829 ppm and% for brake power, BTE,
meter on all outputs response. The engine speed has fewer impacts on BSFC, NOx, HC and CO emissions respectively.
outputs response from the load. The blended fuel has insignificant on • The extracted water content from fusel oil has a statistical sig-
brake power while highest impacts of blended fuel were on BSFC re- nificance of increases the brake thermal efficiency (BTE), NOx
sponse. The extracted water content from fusel oil (FAWE10 and emission and decrease the brake specific fuel consumption (BSFC),
FAWE20) has a statistical significance of increases the brake thermal HC and CO emissions as well as has insignificant effects on brake

239
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

power 1996.
[25] Brinkman N, Ecklund E, Nichols R. Fuel methanol: a decade of progress.
Warrendale, USA: Society of Automotive Engineers. Inc; 1975.
This experimental design and ANOVA analysis proved the extracted [26] Brandberg A, Hedbom A, Isaksson K, Laveskog A. Project M100-a test with me-
water content from fusel oil enhanced the performance and emission of thanol-fueled vehicles in Sweden; 1987.
[27] Turner J, Pearson R, Dekker E, Iosefa B, Johansson K, Ac K. Bergström. Extending
spark ignition engine. This could be explained by the improving in the role of alcohols as transport fuels using iso-stoichiometric ternary blends of
heating value of fusel oil after extracted the water content by 13% gasoline, ethanol and methanol. Appl Energy 2013;102:72–86.
compared with that before water extraction. This research has thrown [28] Calam A, İçingür Y, Solmaz H, Yamık H. A comparison of engine performance and
emissions of fusel oil and gasoline mixtures at different ignition timings. Int J Green
up many questions in need of further investigation such as the effects Energy 2013.
higher compression ratio by using the fusel oil blending on engine [29] Lanzanova TDM, Dalla Nora M, Zhao H. Performance and economic analysis of a
performance and emission due to the higher-octane rating and oxygen direct injection spark ignition engine fueled with wet ethanol. Appl Energy
2016;169:230–9.
content of fusel oil.
[30] Jeuland N, Montagne X, Gautrot X. Potentiel de l'éthanol en tant que carburant pour
un moteur dédié. Oil Gas Sci Technol - Rev IFP 2004;59:559–70.
Acknowledgments [31] Demirbas A. Combustion characteristics of different biomass fuels. Prog Energy
Combust Sci 2004;30:219–30.
[32] Calam A, İçingür Y, Solmaz H, Yamık H. A comparison of engine performance and
The financial support offered by the Universiti Malaysia Pahang the emission of fusel oil and gasoline mixtures at different ignition timings. Int J
under RDU150369 is gratefully acknowledged. Green Energy 2015;12:767–72.
[33] Solmaz H. Combustion, performance and emission characteristics of fusel oil in a
spark ignition engine. Fuel Process Technol 2015;133:20–8.
References [34] Anonymous. Market supply Information in Ethanol Sector 2013. Turkish Tobacco
and Alcohol Market Regulatory Authority; 2013.
[1] Thangavelu SK, Ahmed AS, Ani FN. Review on bioethanol as alternative fuel for [35] Ferreira MC, Meirelles AJ, Batista EA. Study of the fusel oil distillation process. Ind
spark ignition engines. Renew Sustain Energy Rev 2016;56:820–35. Eng Chem Res 2013;52:2336–51.
[2] Demirbas A. Progress and recent trends in biofuels. Prog Energy Combust Sci [36] Demirbas A. Relationships between heating value and lignin, moisture, ash and
2007;33:1–18. extractive contents of biomass fuels. Energy Explor Exploit 2002;20:105–11.
[3] Ou X, Yan X, Zhang X, Liu Z. Life-cycle analysis on energy consumption and GHG [37] Nakayasu T, Yamada H, Suda T, Iwase N, Takahashi K. Intake and exhaust systems
emission intensities of alternative vehicle fuels in China. Appl Energy equipped with a variable valve control device for enhancing of engine power. SAE
2012;90:218–24. Technical Paper; 2001.
[4] White TL. Alcohol as a fuel for the automobile motor. SAE Technical Paper; 1907. [38] Atashkari K, Nariman-Zadeh N, Gölcü M, Khalkhali A, Jamali A. Modelling and
[5] Pourkhesalian AM, Shamekhi AH, Salimi F. Alternative fuel and gasoline in an SI multi-objective optimization of a variable valve-timing spark-ignition engine using
engine: a comparative study of performance and emissions characteristics. Fuel polynomial neural networks and evolutionary algorithms. Energy Convers Manage
2010;89:1056–63. 2007;48:1029–41.
[6] Semelsberger TA, Borup RL, Greene HL. Dimethyl ether (DME) as an alternative [39] Gölcü M, Sekmen Y, Erduranlı P, Salman MS. Artificial neural-network based
fuel. J Power Sources 2006;156:497–511. modeling of variable valve-timing in a spark-ignition engine. Appl Energy
[7] Awad OI, Mamat R, Ali OM, Yusri IM, Abdullah AA, Yusop AF, et al., The effect of 2005;81:187–97.
adding fusel oil to diesel on the performance and the emissions characteristics in a [40] Yusri IM, Mamat R, Azmi WH, Omar AI, Obed MA, Shaiful AIM. Application of
single cylinder CI engine. J Energy Inst. doi: http://dx.doi.org/10.1016/j.joei.2016. response surface methodology in optimization of performance and exhaust emis-
04.004. sions of secondary butyl alcohol-gasoline blends in SI engine. Energy Convers
[8] Agathou MS, Kyritsis DC. Electrostatic atomization of hydrocarbon fuels and bio- Manage 2017;133:178–95.
alcohols for engine applications. Energy Convers Manage 2012;60:10–7. [41] Jun L, Chao G, Wen Bin W, Wu ZJ. Optimal mixture ratios of biodiesel ethanol
[9] Ali OM, Mamat R, Masjuki HH, Abdullah AA. Analysis of blended fuel properties diesel for diesel engines. Energy Power Eng 2011;3:625.
and cycle-to-cycle variation in a diesel engine with a diethyl ether additive. Energy [42] Datta Bharadwaz Y, Govinda Rao B, Dharma Rao V, Anusha C. Improvement of
Convers Manage 2016;108:511–9. biodiesel methanol blends performance in a variable compression ratio engine using
[10] Bae C, Kim J. Alternative fuels for internal combustion engines. Proc Combust Inst response surface methodology. Alexandria Eng J 2016;55:1201–9.
2015. [43] Fang W, Kittelson DB, Northrop WF. Optimization of reactivity-controlled com-
[11] Awad OI, Ali OM, Mamat R, Abdullah A, Najafi G, Kamarulzaman M, et al. Using pression ignition combustion fueled with diesel and hydrous ethanol using response
fusel oil as a blend in gasoline to improve SI engine efficiencies: a comprehensive surface methodology. Fuel 2015;160:446–57.
review. Renew Sustain Energy Rev 2016. [44] Khoobbakht G, Najafi G, Karimi M, Akram A. Optimization of operating factors and
[12] Brinkman N, Gallopoulos N, Jackson M. Exhaust emissions, fuel economy, and blended levels of diesel, biodiesel and ethanol fuels to minimize exhaust emissions
driveability of vehicles fueled with alcohol-gasoline blends. SAE Technical Paper; of diesel engine using response surface methodology. Appl Therm Eng
1975. 2016;99:1006–17.
[13] Cairns A, Stansfield P, Fraser N, Blaxill H, Gold M, Rogerson J, et al. A study of [45] Pandian M, Sivapirakasam SP, Udayakumar M. Investigation on the effect of in-
gasoline-alcohol blended fuels in an advanced turbocharged DISI engine. SAE Int J jection system parameters on performance and emission characteristics of a twin
Fuels Lubr. 2009;2:41–57. cylinder compression ignition direct injection engine fuelled with pongamia bio-
[14] Calam A, Solmaz H, Uyumaz A, Polat S, Yilmaz E, İçingür Y. Investigation of us- diesel–diesel blend using response surface methodology. Appl Energy
ability of the fusel oil in a single cylinder spark ignition engine. J Energy Inst 2011;88:2663–76.
2015;88:258–65. [46] Asghar A, Abdul Raman AA, Daud WMAW. A comparison of central composite
[15] Dimitroff E, Vitkovits JA. Blend of gasoline and vapors of dissociated alcohol. design and Taguchi method for optimizing Fenton process. Sci World J 2014;2014.
Google Patents; 1980. [47] Bezerra MA, Santelli RE, Oliveira EP, Villar LS, Escaleira LA. Response surface
[16] Ghazikhani M, Hatami M, Safari B, Ganji DD. Experimental investigation of per- methodology (RSM) as a tool for optimization in analytical chemistry. Talanta
formance improving and emissions reducing in a two stroke SI engine by using 2008;76:965–77.
ethanol additives. Propul Power Res 2013;2:276–83. [48] Ma L, Han Y, Sun K, Lu J, Ding J. Optimization of acidified oil esterification cat-
[17] Karaosmanoğlu F, Işığıgür A, Soruşbay C, Aksoy H. Methanol-gasoline blends as alyzed by sulfonated cation exchange resin using response surface methodology.
alternative engine fuel. In: IX International symposium on alcohol fuel (ISAF). Energy Convers Manage 2015;98:46–53.
Firenze 1991. [49] Najafi G, Ghobadian B, Yusaf T, Safieddin Ardebili SM, Mamat R. Optimization of
[18] Ozsezen AN, Canakci M. Performance and combustion characteristics of alco- performance and exhaust emission parameters of a SI (spark ignition) engine with
hol–gasoline blends at wide-open throttle. Energy 2011;36:2747–52. gasoline–ethanol blended fuels using response surface methodology. Energy
[19] Yüksel F, Yüksel B. The use of ethanol–gasoline blend as a fuel in an SI engine. 2015;90(Part 2):1815–29.
Renewable Energy 2004;29:1181–91. [50] Prajapati NB, Patel PR, Patel TM, Rathod GP. Optimization of SFC using mathe-
[20] Yusri I, Mamat R, Azmi W, Najafi G, Sidik N, Awad OI. Experimental investigation matical model based on RSM for SI engine fueled with petrol-ethanol blend.
of combustion, emissions and thermal balance of secondary butyl alcohol-gasoline [51] Chen Y-L, Chen S, Tsai J-M, Tsai C-Y, Fang H-H, Yang IC, et al. Optimization of
blends in a spark ignition engine. Energy Convers Manage 2016;123:1–14. suitable ethanol blend ratio for motorcycle engine using response surface method.
[21] Zhang Z, Wang T, Jia M, Wei Q, Meng X, Shu G. Combustion and particle number J. Environ. Sci. Health Part A 2012;47:101–8.
emissions of a direct injection spark ignition engine operating on ethanol/gasoline [52] Agarwal AK, Karare H, Dhar A. Combustion, performance, emissions and particu-
and n-butanol/gasoline blends with exhaust gas recirculation. Fuel late characterization of a methanol–gasoline blend (gasohol) fuelled medium duty
2014;130:177–88. spark ignition transportation engine. Fuel Process Technol 2014;121:16–24.
[22] Yücesu HS, Topgül T, Çinar C, Okur M. Effect of ethanol–gasoline blends on engine [53] Demirbas A. Effects of moisture and hydrogen content on the heating value of fuels.
performance and exhaust emissions in different compression ratios. Appl Therm Energy Sources Part A: Recov Util Environ Eff 2007;29:649–55.
Eng 2006;26:2272–8. [54] Nithyanandan K, Wu H, Huo M, Lee C-F. A preliminary investigation of the per-
[23] Elfasakhany A. Experimental study of dual n-butanol and iso-butanol additives on formance and emissions of a port-fuel injected SI engine fueled with Acetone-
spark-ignition engine performance and emissions. Fuel 2016;163:166–74. Butanol-Ethanol (ABE) and Gasoline. SAE Technical Paper; 2014.
[24] Egebäck K-E, Pettersson E. Research on a Scania 11 liter ethanol fueled bus engine; [55] Noordin MY, Venkatesh VC, Sharif S, Elting S, Abdullah A. Application of response

240
O.I. Awad et al. Energy Conversion and Management 150 (2017) 222–241

surface methodology in describing the performance of coated carbide tools when and Calophyllum inophyllum based biodiesel and investigation of blend perfor-
turning AISI 1045 steel. J Mater Process Technol 2004;145:46–58. mance and exhaust emission in an unmodified diesel engine at high idling condi-
[56] Calam A, Solmaz H, Uyumaz A, Polat S, Yilmaz E, Içingür Y. Investigation of us- tions. Energy Convers Manage 2013;76:362–7.
ability of the fusel oil in a single cylinder spark ignition engine. J Energy Inst 2014. [65] Cooper CD, Alley FC. Air pollution control: a design approach. Waveland Press;
[57] Montgomery DC. Design and analysis of experiments. John Wiley & Sons; 2008. 2002.
[58] Awad OI, Mamat R, Ibrahim TK, Hagos FY, Noor MM, Yusri IM, et al. Calorific value [66] Normann F, Andersson K, Leckner B, Johnsson F. Emission control of nitrogen
enhancement of fusel oil by moisture removal and its effect on the performance and oxides in the oxy-fuel process. Prog Energy Combust Sci 2009;35:385–97.
combustion of a spark ignition engine. Energy Convers Manage 2017;137:86–96. [67] Chong J, Tsolakis A, Gill S, Theinnoi K, Golunski SE. Enhancing the NO2/NOx ratio
[59] Awad OI, Mamat R, Noor MM, Ibrahim TK, Yusri IM, Yusop AF. The impacts of in compression ignition engines by hydrogen and reformate combustion, for im-
compression ratio on the performance and emissions of ice powered by oxygenated proved aftertreatment performance. Int J Hydrogen Energy 2010;35:8723–32.
fuels: a review. J Energy Inst. doi: http://dx.doi.org/10.1016/j.joei.2016.09.003. [68] Gu X, Huang Z, Cai J, Gong J, Wu X, Lee C-F. Emission characteristics of a spark-
[60] Ali OM, Mamat R, Najafi G, Yusaf T, Safieddin Ardebili SM. Optimization of bio- ignition engine fuelled with gasoline-n-butanol blends in combination with EGR.
diesel-diesel blended fuel properties and engine performance with ether additive Fuel 2012;93:611–7.
using statistical analysis and response surface methods. Energies 2015;8:14136–50. [69] He B-Q, Liu M-B, Yuan J, Zhao H. Combustion and emission characteristics of a
[61] Fang Q, Fang J, Zhuang J, Huang Z. Effects of ethanol–diesel–biodiesel blends on HCCI engine fuelled with n-butanol–gasoline blends. Fuel 2013;108:668–74.
combustion and emissions in premixed low temperature combustion. Appl Therm [70] Dhamodaran G, Esakkimuthu GS, Pochareddy YK. Experimental study on perfor-
Eng 2013;54:541–8. mance, combustion, and emission behaviour of diisopropyl ether blends in MPFI SI
[62] Automated engine for determining the combustion quality of a fuel. Google Patents; engine. Fuel 2016;173:37–44.
1967. [71] Poulopoulos S, Samaras D, Philippopoulos C. Regulated and unregulated emissions
[63] Lapuerta M, Armas O, García-Contreras R. Effect of ethanol on blending stability from an internal combustion engine operating on ethanol-containing fuels. Atmos
and diesel engine emissions. Energy Fuels 2009;23:4343–54. Environ 2001;35:4399–406.
[64] Rahman SA, Masjuki H, Kalam M, Abedin M, Sanjid A, Sajjad H. Production of palm

241