Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Fuel 281 (2020) 118765

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Effect of design parameters on performance and emissions of DI diesel T


engine running on biodiesel-diesel blends: Taguchi and utility theory
Abhishek Sharmaa, Nagendra Kumar Mauryaa, Yashvir Singhb, , Nishant Kumar Singhc,

Sandeep Kumar Guptad


a
Department of Mechanical Engineering, G L Bajaj Institute of Technology and Management, Greater Noida, Uttar Pradesh, India
b
Department of Mechanical Engineering, Graphic Era Deemed to be University, Dehradun, Uttarakhand, India
c
Department of Mechanical Engineering, Hindustan College of Science and Technology, Mathura, Uttar Pradesh, India
d
Sharda University, Greater Noida, Uttar Pradesh, India

GRAPHICAL ABSTRACT

ARTICLE INFO ABSTRACT

Keywords: In the present context, the use of fossil fuels is rising rapidly, leading to a further rise in the level of air pollution.
Engine Due to all these environmental challenges, there is a need for some innovative cleaner and environmentally
Performance viable fuels. Keeping all these parameters in mind, there is a need for an alternative to fulfill this gap. The
Optimization present study focused on the optimization of a diesel engine operating parameters fuelled with pongamia bio-
Pongamia biodiesel
diesel blend at full load. The optimization was carried out for different input parameters including pongamia
Taguchi
Utility theory
biodiesel blends (0–40%), fuel injection timing (15–31) °bTDC, and fuel injection pressure (16–24) MPa.
Initially, Taguchi was applied, and then it was processed with utility theory. An experiment was performed
according to the combination of Taguchi and utility theory optimization techniques. Further, optimum combi-
nation for engine input parameters (Pongamia biodiesel blending, Fuel injection pressure, and Injection timing)


Corresponding author.
E-mail address: yashvirsingh21@gmail.com (Y. Singh).

https://doi.org/10.1016/j.fuel.2020.118765
Received 10 January 2020; Received in revised form 5 July 2020; Accepted 17 July 2020
0016-2361/ © 2020 Elsevier Ltd. All rights reserved.
A. Sharma, et al. Fuel 281 (2020) 118765

to achieve best engine responses (BTE, UHC, NOx, and Smoke). The blending of pongamia biodiesel (10%) with
injection timing 23°bTDC and injection pressure 22 MPa found the best input engine setting at full engine load.
These optimization results were verified by engine experimentation within the suggested error range.

1. Introduction From the literature, it can be observed that the analysis of variance
(ANOVA) and Taguchi method have been successively applied to
Energy demand in the transport sector is projected to rise by the end achieve optimum levels of operating process parameters at which the
of 2030. Transportation is one of the major sectors that contribute to engine gives the best performance. It has a reduced number of experi-
the rise in air pollution. Nowadays, there are various unconventional ments and allows for good approximation [20]. Ansari et al. [21] im-
sources have been utilized in the transportation sector [1]. However, plemented Taguchi method to evaluate the parameters' combination for
the use of biofuels may play a significant role in the alternative fuel the BTE, BSFC, smoke, and NOX of a diesel engine, and the best com-
sector. It can fulfill both aspects of demand as well as pollution re- bination reduced the BSFC. It has also minimized both smoke emissions
duction [2,3]. The use of biodiesel promotes a reduction in NOx and NOx. It was observed that overall engine performance depends on
emissions as compared to diesel fuel. Excess of oxygen contents in the several input factors. The use of a multi-objective optimization tech-
biodiesel will support burning of fuel and results in less air pollution nique was a great interest to the researchers to achieve the optimum
[4,5]. Improper combustion of fuel results into sulphur content in ex- best levels of process parameters that optimizes all the response vari-
haust gas which is very harmful to the environment. The proper ables.
burning of fuel leads to a high cetane number and also decreases the It has been observed that researchers have applied a hybrid ap-
amount of sulphur content in the exhaust gas [6,7]. Therefore, biodiesel proach for multi-objective optimization i.e. Taguchi with utility theory
may be utilized as an optional source for energy favorable for the en- for optimization of process parameters in manufacturing and aero-
vironment. The use of biodiesel as an alternative fuel for the diesel dynamics research [22,23]. Average utility value is considered for the
engine was the research area for the researchers for more than thirty overall optimization of the process variables. However, utility theory
years [8]. Milano et al. [9] have experimentally evaluated the optimum has not been used for the optimization of engine variables in the pre-
condition for the production of biodiesel from a mixture of Calophyllum vious studies.
inophyllum oil and waste cooking oil. Results showed that microwave The present study aims to optimize the DI diesel engine operating
irradiation helped alkaline-catalyzed transesterification produced su- parameters based on alternative fuel pongamia biodiesel blends with a
perior quality biodiesel. Ong et al. [10] have optimized the process robust optimization process (the combination of Taguchi and utility
variables for biodiesel production of CI40CP60 oil mixture by alkaline- theory). In the analysis, three main engine input factors (blend per-
catalyzed transesterification process. Ant colony and ANN techniques centage, injection timing and injection pressure) have been considered
were used for the optimization of process variables. for optimizing the main engine responses (BTE, UHC, NOx and Smoke).
Many scientists and researchers from various organizations and Five levels have been considered for each factor. Orthogonal array L25
technical institutes have proven that non-edible vegetable oil is a sui- was used to get optimum combination through Taguchi and the best
table source of energy. It was concluded through experimentation that output of Taguchi further optimized using utility theory.
their low quantity blend with conventional energy fuel showed nearly
the same behaviour in comparison to fossil fuel [11]. Silitonga et al.
[12] have evaluated the exhaust emissions and engine performance of 2. Material and methods
blends of bioethanol-diesel- biodiesel through a kernel-based extreme
learning machine. Results revealed that these fuels have a higher value 2.1. Biodiesel preparation
of BTE, and lower value of BSFC, smoke, and carbon monoxide emis-
sions. In the literature, various optimization techniques viz., Taguchi, Raw Pongamia seed oil was taken from the local shop in New Delhi.
RSM, ANN, GA, etc. have been successively used to get optimum During the esterification process of pongamia oil in this work, the free
parameters to run the engine and found suitable for the reduction of fatty acid (FFA) of 4.65% has been attained for the 1.5% p-
time for the process. Optimum levels of process parameters have been Toluenesulfonic acid-PTSA i.e. catalyst at the reaction temperature of
obtained [13,14]. Patowari et al. [15] have used ANN for forecasting 65 °C for 60 min. In the transesterification process at 62.75 °C and
and analyzing the exhaust emissions of diesel engines. 0.88% KOH concentration for 90 min was taken to achieve yields of
Response surface methodology is also a powerful tool useful for the 95.51% using the molar ratio 1:6. The pongamia oil was generated in
analysis of engine operating conditions to optimize the engine re- bulk quantity by considering the above stated optimum conditions. It is
sponses like BTE, exhaust emission, and BSEC [16]. Damodharan et al. seen from the above process that the free fatty achieved was 4.75%, but
[17] have evaluated the optimum levels of process parameters through for higher yield, it should be less than 2%. Therefore, surplus stirring
optimizing the blend condition with IT, IP, and EGR. The optimum for 20 min is required to limit FFA below 2%. Consequently, the oil was
levels of process parameters have been achieved through iso-butanol/ transesterified as stated, the attained optimal conditions of 0.88% KOH
diesel blends of 40%, IP of 240 bar, and IT of 23°CA bTDC and 30% concentration and temperature 63 °C for 1 h 30 min. The received
EGR. Kumar et al. [18] have experimented obtained the optimum levels sample was separated in 12 h by gravity from glycerol. It was seen that
of process parameters for the diesel engine. Results demonstrated that the generated oil has contaminations such as KOH and methanol.
an optimum level of process parameters was found to be 86.3% load, a Further, purification of the pongamia oil was carried out by washing it
compression ratio of 16, a blend of 15 %and injection timing of three times by hot distilled water to eliminate the traces of KOH and
26.24°CA bTDC, which has reduced both BTE and NOx emissions. Gad methanol. To eliminate the moisture from the pongamia oil, heating
et al. [19] have conducted experimental studies to investigate the in- was carried out at 120 °C for 25 min. Finally, the pongamia oil methyl
fluence of blending of nano-additive such as TiB2, CNTs and Al2O3 in a ester was attained as a light-yellow transparent fluid. To see its mis-
blend of diesel, and biodiesel oils on the performance of diesel engine cibility blends were observed for 15 days and reported no symptoms of
and emissions reductions. Results depicted that due to the addition of separation of biodiesel and diesel. For this work, four test samples of 2 L
nano-additive in biodiesel blends, improved results in terms of reducing each were needed (B10, B20, B30, and B40). Test fuel physicochemical
emissions were obtained. properties are illustrated in Table1.

2
A. Sharma, et al. Fuel 281 (2020) 118765

Nomenclature DI Direct injection


EGT Exhaust gas temperature
ASTM American standard test method FFA Free fatty acid
B10 10% Pongamia biodiesel blending with diesel IT Injection timing
B20 20% Pongamia biodiesel blending with diesel IP Injection pressure
B30 30% Pongamia biodiesel blending with diesel MPa Mega pascal
B40 40%Pongamia biodiesel blending with diesel Pmax Peak cylinder pressure
BTE Brake thermal efficiency RPM Revolution per minute
BSFC Brake specific fuel consumption RSM Response surface methodology
C.I. Compression ignition TDC Top dead centre
CR Compression ratio UHC Unburnt hydrocarbons
CA Crank angle aTDC After the top dead centre
CO Carbon monoxide bTDC Before the top dead centre

2.2. Engine setup freedom required for the analysis of orthogonal array should be greater
than or equal to the required degree of freedom for the design variables.
Kirloskar make single cylinder DI diesel engine that was used as an In Table3, all the decision variables have the same levels and hence,
experimental setup. This engine was attached to a computer employed column 2 represents the blending of biodiesel, column 3 represents fuel
with Legion Brothers designed software, AVL Di gas analyzer and AVL injection timing and column 4 represents fuel injection pressure. Twenty-
444C smoke meter to obtained engine responses BTE, UHC, NOx, and five rows of L25 OA represent twenty-five experiments that are needed for
Smoke. Fig. 1 shows a single cylinder CI engine equipped with the es- each run of the experiment. In this investigation, three runs are per-
sential units which are utilized to measure the BTE and all engine formed corresponding to each experiment and the average value is used
emission with the temperature of the different locations and the applied for the analysis of the data. Analysis of variance (ANOVA) was applied to
load on the engine. The engine on diesel fuel for initial 25 min to sta- analyze the influence of individual process variables on response factors.
bilize with lubricant temperature (around 80 °C), then it was switched First-round engine parameters optimization was carried out using Ta-
to test fuel. The major engine specifications are given in Table 2. guchi, where the findings with the highest desirability were considered
An electronic digital counter measured engine speed (RPM). All as optimum. The engine control factors were injection timing, injection
engine experiments were carried out three times for each Taguchi de- pressure and biodiesel blends. Experimental results are shown in Table4.
fined engine input setting and the average of these engine responses
was used to generate L25 orthogonal matrix for further optimization. 2.4. Statistical analysis of measured data

Experimentally measured data of Table 4, for the BTE, UHC, NOx,


2.3. Taguchi approach
and Smoke were analyzed by using statistical software MINITAB 14.
Signal to noise (S/N) ratio was used to find out the effect of process
In this research work, the Taguchi method is employed for the op-
parameters caused by each factor. The benefit of using the S/N ratio
timization of engine process parameters. Taguchi design is a robust
was that it uses a mean square deviation (MSD), a single measure that
design that uses an orthogonal array (OAs) for the study of a large
includes the effect of changes in mean as well as the standard deviation
number of process variables with a smaller number of experiments.
with equal priority. Experimental results were linearly varied when it is
Taguchi suggests two different ways for the analysis of results. In the
presented in terms of S/N ratios. The quality characteristic 'larger is
first route results of a single run or average value of multiple runs are
better' was used for the response factor BTE and ‘smaller is the better’
used through the main effect and ANOVA analysis. In the second ap-
was employed for the response factors UHC, NOx, and Smoke. S/N ratio
proach, the S/N ratio is used for the analysis of experimental data.
(ƞ) can be obtained by using Eq. (1) [4,16].
In this investigation three decision variables viz., blending of bio-
diesel, fuel injection pressure and fuel injection timing are used. This = 10log (MSD) (1)
engine set is stationary, primarily used for agriculture and most of the
MSD = 2 (Xave X0 )2 (2)
time runs at full load condition. Hence, the experimentation engine op-
2
erates at a full load condition of 5.2 kW at 1500 RPM. There are five where σ is the variance, MSD is the Mean-Square Deviation, n is the
levels for each decision variable. Every five levels of process parameters data points, Xave is the average value of response factors for each group
have four degrees of freedom (DOF = number of levels-1). The total of experiments, and X0 is the target value (0 in this design). The value of
degree of freedoms required for all three-decision variables at five levels MSD can be determined by using Eq. (2). The S/N ratio for each run of
is 12[DOF = 3(5-1)]. For the Taguchi method, the total degree of the experiments was determined and reported in Table 5.

Table 1
Physicochemical characteristics of the diesel and its blends with Pongamia biodiesel measured according to the respective ASTM standards.
S.No Characteristics ASTM Measuring Equipment Accuracy Diesel B10 B20 B30 B40

1 Chemical formula – C13H24 – – –


2 Density (kg/m3) D6751 Anton Parr density meter 0.1% 832 852 849 851 867
3 Lower heating value(MJ/kg) – Parr 6100 Bomb calorimeter 0.02% 43.99 39.19 40.09 39.2 37.8
4 Kinematic Viscosity at 40 °C(cSt) D445 Petrotest Viscometer 0.1% 2.91 3.32 3.31 3.33 3.71
5 Flashpoint (°C) D93 Pensky-Martens Closed Cup Tester 0.5% 75 81 85 90 101
6 Cetane number D613 Potentiometer titrator 1% 45–55 42 45 43 40
7 Cloud point D2500 Automatic Cloud Point Analyser 0.5% 6.4 8.1 7.3 8 8.9
8 Pour point – Automatic Cloud Point Analyser 0.5% −5.1 −1.2 −2.9 −1.1 3.2
9 Autoignition temperature (°C) – Not-specified – 262 284 292
10 Ash content (% by mass) D874 Ramsbottom Carbon Residue 0.2% 0.58 – –

3
A. Sharma, et al. Fuel 281 (2020) 118765

Fig. 1. Schematic image of the engine setup [30].

2.5. Development of an empirical model for the BTE, UHC, NOx and Smoke Z= f(x1, x2, x3) (3)

The linear equation for the experimental data can be defined by Eq.
An empirical model was developed by using the least square re-
(4).
gression analysis technique. The response function Z for the BTE, UHC,
NOx, and Smoke in terms of three input process parameters (x1, x2 and Z= 0 + 1 x1 + 2 x2 + 3 x3 (4)
x3) can be expressed by Eq. (3).
where, β0, β1, β2, and β3 are the regression coefficient, Zi is the output

Fig. 2. Main effect plot of the SN ratio for BTE.

4
A. Sharma, et al. Fuel 281 (2020) 118765

Table 2 Table 5
Engine technical specifications. Value of S/N ratio for the Responses.
Make Kirloskar S.N S/N ratio BTE S/N ratio UHC S/N ratio Nox S/N ratio Smoke

Type 4-stroke, Single Cylinder diesel engine 1 28.86 −33.62 −55.28 −38.99
Bore and Stroke 87.5 mm and 110 mm 2 29.14 −33.44 −55.37 −38.99
Capacity 0.661 L 3 29.53 −33.26 −56.07 −38.89
Compression ratio 17.5:1 4 29.66 −33.06 −57.13 −38.79
Loading Eddy current dynamometer 5 29.79 −32.87 −59.08 −38.69
Power 5.2 kW @1500 RPM 6 28.78 −32.87 −55.75 −38.99
Temperature sensor Type-K Thermocouple 7 28.96 −32.67 −57.04 −38.89
Injection timing 23°bTDC 8 29.49 −32.47 −57.51 −38.69
Injection pressure 20 MPa 9 29.54 −32.26 −59.10 −38.59
10 29.34 −33.80 −57.36 −38.79
11 28.77 −32.04 −57.12 −38.79
12 28.91 −31.82 −58.76 −38.69
Table 3
13 29.23 −31.60 −59.52 −38.59
Engine Input Parameter Levels Used for L25 Orthogonal Array (OA).
14 28.89 −33.26 −58.14 −38.79
Blending of Biodiesel Fuel injection timing Fuel injection pressure 15 29.20 −33.06 −59.19 −38.69
16 28.61 −31.36 −58.84 −38.69
Symbol B IT FIP 17 28.83 −31.13 −59.71 −38.59
Unit % °bTDC MPa 18 28.59 −32.67 −58.68 −38.69
Level 1 0 15 16 19 28.79 −32.47 −59.70 −38.59
Level 2 10 19 18 20 28.99 −32.26 −60.03 −38.49
Level 3 20 23 20 21 28.47 −31.13 −59.93 −38.49
Level 4 30 27 22 22 28.41 −31.82 −59.08 −38.49
Level 5 40 31 24 23 28.44 −31.60 −59.77 −38.49
24 28.67 −31.36 −60.29 −38.49
25 28.94 −31.36 −60.63 −38.49

response and x1, x2 and x3 are the input parameters. In this study, the
response factors were BTE, UHC, NOx, and smoke having input para-
response factors for any combination of process parameters within the
meters viz. blends (%), I.T (°bTDC), and I.P (MPa). The obtained re-
experimental range. F-test was employed for checking model adequacy.
gression equations for the BTE percentage (Z1), UHC (Z2), NOx (Z3),
and smoke (Z4) are shown in Eqs. (5)–(8).
2.6. Multi-objective optimisation
Z1 = 27.9 0.677x1 + 0.444x2 + 0.319 x3 (5)

Z2 = 50.9 2.12x1 + 0.580x2 1.52 x3 (6) In the present study, engine operating process variables viz.,
blending of biodiesel percentage, injection timing, and injection pres-
Z3 = 359 + 81.2x1 + 41.9x2 + 37.6x3 (7) sure for the desired response variables i.e. brake thermal efficiency,
UHC, smoke and NOx were optimized. Multi-objective optimization
Z4 = 89.7 0.46x1 0.32x2 0.38x3 (8)
problems involve more than one objective function. In multi-objective
where x1 is the blends percentage (0%, 10%, 20%, 30% and 40%), x2 is optimization problems, objective functions are maximized or mini-
the I.T°bTDC (15, 19, 23, 27 and 31) and x3 is the I.P in MPa (16, 18, mized [29]. The overall performance analysis of any machine is eval-
20, 22 and 24). The above developed empirical model predicts the uated based on the number of output responses. Optimal conditions for

Table 4
Orthogonal array (L25) and responses.
Exp. run Blends (%) I.T (°bTDC) I.P (MPa) BTE (%) UHC (ppm Vol.) NOx (ppm Vol.) Smoke (%)

1 0 15 16 27.72 48 581 89
2 0 19 18 28.65 47 587 87
3 0 23 20 29.94 46 636 87
4 0 27 22 30.4 45 719 86
5 0 31 24 30.85 44 899 87
6 10 15 18 27.49 44 613 88
7 10 19 20 28.05 43 711 87
8 10 23 22 29.82 42 751 86
9 10 27 24 29.99 41 902 86
10 10 31 16 29.3 49 738 86
11 20 15 20 27.45 40 718 87
12 20 19 22 27.89 39 867 86
13 20 23 24 28.94 38 946 85
14 20 27 16 27.834 46 807 87
15 20 31 18 28.83 45 911 86
16 30 15 22 26.96 37 875 86
17 30 19 24 27.65 36 967 85
18 30 23 16 26.87 43 859 86
19 30 27 18 27.5 42 966 86
20 30 31 20 28.16 41 1004 85
21 40 15 24 26.53 36 992 85
22 40 19 16 26.34 39 899 87
23 40 23 18 26.43 38 974 86
24 40 27 20 27.12 37 1034 85
25 40 31 22 27.98 37 1075 84

5
A. Sharma, et al. Fuel 281 (2020) 118765

Table 6 where Wi is the weight function. The logarithmic scale is used for
Optimum level of process parameter. evaluating the performance scale. The performance scale for the re-
Response Factor Blends level IT level IP level Optimum model value sponse factor can be calculated by using Eq. (12).

BTE (%) 0 31 24 31.057 Xi


Pi = A log
UHC 40 19 24 33.835 X1 (12)
NOx 0 15 16 520.4
th
Smoke 40 31 24 82.6 where Xi is the value of quality characteristics of an i attribute, A is the
constant and ‘X1’ is the minimum acceptable value. In maximization
problem, the minimum acceptable value is the peak value of the ex-
response variables, i.e. BTE, UHC, NOx, and smoke are shown in perimental result and for maximization problem, the minimum accep-
Table 6. The optimal conditions of process variables for the response table value is the lowest value of the experimental result. The value of
variables were different. the constant term can be evaluated with the help of optimal condition.
In this study, four response variables were used to optimize the If Xi = X* (where X* is the optimal value) and (Pi = 25), for this case
performance of the diesel engine. The multi-objective optimization performance scale is represented in Eq. (13).
technique ‘utility theory’ was used to optimize the engine performance
parameters. Utility theory is a powerful tool for the optimization of X
Pi = A log
multi-response factors. However, it can be applied if the range of op- X1 (13)
timum solution is known. It can be applied for maximization as well as
The overall utility value can be calculated using Eq. (14).
a minimization problem. It may also apply where objective functions
are mixed type, as some objective functions are maximization and other n

objective functions are minimized. Nevertheless, the use of Taguchi U= Wi Pi


i= 1 (14)
with utility theory cannot reach global maxima or minima. The utility
theory assumed that any decision was made based on the maximum The overall utility value can be used as a single objection function.
utilization of the utility. According to the utility theory, if Xi is the Optimization was carried out by using the ‘larger is the better’ quality
measure of effectiveness for the quality characteristics ‘i' and n are the characteristics.
attributes for evaluating in the outcome space then the joint utility can
be express as below in Eq. (9) [24].
2.6.1. Construction of the performance scale.
U(X1, X2, X3, Xn) = f(U1 (X1), U2 (X2), U3 (X3)….Un (Xn)) (9)
where Ui(Xi) is the utility of the ith attribute. For independent attri- • Performance scale for BTE (%)
butes, the utility function is presented in Eq. (10).
X* = 31.057% optimum value of BTE from Table 6.
X1 = 26.34% minimum acceptable value of BTE from Table 4
n
U(X1,X2,X3. ...Xn) = Ui(Xi)
(10) A = 349.43
i= 1
Performance scale for the response factor BTE can be constructed by
The overall utility in terms of weight function can be expressed by using Eqs. (12) and (13). Eq. (15) shows the performance scale for the
Eq. (11). BTE.
n
U(X1,X2,X3. ...Xn) = Wi Ui(Xi) XZ1
PZ1 = 349.43log
i= 1 (11) 26.34 (15)

Table 7
Utility value of response factors and S/N ratio for the combined utility.
Exp. No UZ1 (BTE) UZ2 (UHC) UZ3 (NOx) UZ4 (Smoke) Uoverall S/N ratio

1 7.74 1.39 21.20 0.00 7.58 17.59


2 12.74 2.81 20.85 9.63 11.51 21.22
3 19.42 4.25 18.09 9.63 12.85 22.18
4 21.73 5.73 13.86 14.53 13.96 22.90
5 23.96 7.25 6.16 9.63 11.75 21.40
6 6.48 7.25 19.35 4.79 9.47 19.53
7 9.53 8.79 14.24 9.63 10.55 20.47
8 18.81 10.38 12.36 14.53 14.02 22.94
9 19.67 12.00 6.05 14.53 13.06 22.32
10 16.14 0.00 12.96 14.53 10.91 20.76
11 6.26 13.66 13.91 9.63 10.87 20.72
12 8.67 15.37 7.41 14.53 11.50 21.21
13 14.27 17.11 4.40 19.48 13.82 22.81
14 8.36 4.25 9.88 9.63 8.03 18.09
15 13.69 5.73 5.70 14.53 9.91 19.92
16 3.53 18.91 7.09 14.53 11.02 20.84
17 7.36 20.75 3.65 19.48 12.81 22.15
18 3.02 8.79 7.73 14.53 8.52 18.61
19 6.53 10.38 3.68 14.53 8.78 18.87
20 10.13 12.00 2.35 19.48 10.99 20.82
21 1.09 20.75 2.77 19.48 11.02 20.84
22 0.00 15.37 6.16 9.63 7.79 17.83
23 0.52 17.11 3.40 14.53 8.89 18.98
24 4.42 18.91 1.34 19.48 11.04 20.86
25 9.15 18.91 0.00 24.50 13.14 22.37

6
A. Sharma, et al. Fuel 281 (2020) 118765

• Performance scale for UHC Table 8


Analysis of Variance for BTE (%), using Adjusted SS for Tests (BTE).
X* = 33.835 optimum value of UHC from Table 6 Source DF Seq SS Adj MS F P Remarks
X1 = 49 minimum acceptable value of UHC from Table 4
A = −155.44 Regression 3 37.833 12.611 178.27 F0.05,3,12 = 8.7446
Blends (%) 4 22.9600 5.7400 68.85 0.00 F > F0.05,3,12
Performance scale for the response factor UHC can be constructed
I.T (obTDC) 4 10.1470 2.5367 30.43 0.00 Model is adequate
by using Eqs. (12) and (13). Eq. (16) shows the performance scale for I.P (MPa) 4 5.2108 1.3027 15.63 0.00
the UHC. Error 12 1.0004 0.0834
Total 24 39.3182
XZ2
PZ2 = 155log
49 (16)

• Performance scale for the NOx Table 9


Analysis of variance for UHC, using adjusted SS for tests.
X* = 520.4optimum value of NOx from Table 6
X1 = 1075 minimum acceptable value of NOx from Table 4 Source DF Seq SS Adj MS F P Remarks

A = −35.73 Regression 3 357.06 119.02 178.79 0.000 F0.05,3,12 = 8.7446


Performance scale for the response factor NOx can be constructed Blends 4 225.040 56.260 70.92 0.000 F > F0.05,3,12
by using Eqs. (12) and (13). Eq. (17) shows the performance scale for I.T 4 19.440 4.860 6.13 0.006 Model is adequate
the NOx. I.P 4 117.040 29.260 36.88 0.000
Error 12 9.520 0.793
XZ3 Total 24 371.040
PZ3 = 79.34log
1075 (17)

• Performance scale for the smoke developed model for brake thermal efficiency is statistically correct
X* = 83.9optimum value of Smoke from Table 6. [25]. It was observed that the Pongamia biodiesel blend, fuel injection
X1 = 89 minimum acceptable value of Smoke from Table 4. timing, and fuel injection pressure were significant process parameters
A = −975.494 for the BTE. However, process parameters blending of Pongamia bio-
Performance scale for the response factor Smoke can be constructed diesel has the highest influence on BTE.
by using Eqs. (12) and (13). Eq. (18) shows the performance scale for From Fig. 2 it can be observed that BTE of the diesel engine is de-
the SMOKE. creasing by increasing the biodiesel blend percentage [5]. The BTE was
increasing by increasing the injecting timing. Break thermal efficiency
XZ4
PZ4 = 975.4942log was increasing due to the availability of more time for the combustion
89 (18)
of fuel. It was also observed that the brake thermal efficiency was in-
creased by increasing the injection pressure. At higher fuel injection
2.6.2. Utility value calculation pressure, better atomization of fuel occurs because biodiesel blends
The utility value for the responses BTE, UHC, NOx, and smoke were have more viscosity as compared to diesel. With an increase in injection
calculated by using Eqs. (15)–(18). The combined utility values of all pressure, the finer diameter of the fuel obtained which deeply pene-
the responses were calculated by the weighted average method. The trated the combustion chamber resulting in more BTE [26,27].
combined utility is considered a single objective function, which opti-
mizes all the responses. The overall utility can be calculated by using
Eq. (19). 3.2. Effect of process variables on UHC
U= PZ1 W1 + PZ2 W2 + PZ3 W3 + PZ4 W4 (19)
Fig. 3 shows the main effect plot of the SN ratio of response variable
In this work, NOx was considered for the highest weight (0.4). unburnt hydrocarbon (UHC). The lowest value of UHC was obtained at
However, the other response factor was considered as equal weight Pongamia biodiesel blending percentage of 40%, injection timing of 19
function (0.2). Table 7 shows the summary of utility value and the S/N °bTDC, and a fuel injection pressure of 24 MPa. A regression model in
ratio corresponding to the overall utility value. The quality character- terms of process variables for the forecasting of UHC value was devel-
istics 'larger is the better' was implied for the calculation of the S/N oped and presented in Eq. (6). This equation was tested for the ade-
ratio. Fig. 6 shows the best conditions of process parameters, which quacy of the model by using ANOVA analysis and F-test, and findings of
optimize all the response variables. The best level of process variables the test were presented in Table 9. The probability value of process
was the blending of Pongamia biodiesel of 40%, fuel injection timing of parameters was less than 0.05, which shows that model was significant.
23°bTDC, and a fuel injection pressure of 24 MPa. The blending of Pongamia biodiesel, injection timing, and injection
pressure were significant process parameters. Out of these process
3. Results and discussion parameters, the blending of Pongamia biodiesel and injection pressure
has the highest contribution. The value of R2 was 0.962 closed to 1. The
3.1. Effect of process variables on BTE R2(0.962) shows good agreement to R2-Adj (0.957).
From Fig. 3 it can be observed that UHC emission is decreasing by
It is well-known fact that the efficiency of the diesel engine could be increasing the biodiesel blending percentage. UHC emission tends to
measured in terms of brake thermal efficiency [25]. Fig. 2 shows the decreases due to additional oxygen present in Pongamia biodiesel. Be-
main effect plot of S/N ratio of response BTE. The peak value of BTE sides this, fuel injection pressure also has the same impact on UHC
(30.85%) was obtained at the blending of Pongamia biodiesel 0%, in- emissions due to better atomization resulting in ineffective combustion.
jection timing of 31° bTDC, and an injection pressure of 24 MPa. The same reason for the decrease in UHC was mentioned in the lit-
Table 8 shows the analysis of variance (ANOVA) of BTE. Eq. (5) shows erature [28]. However, UHC is increasing slightly by increasing the
the mathematical model for the BTE in terms of process variables. The injection timing.
value of R2 is 0.961 and R2-Adj is 0.957 close to 1, which shows that the

7
A. Sharma, et al. Fuel 281 (2020) 118765

Fig. 3. Main effect plot of the SN ratio for UHC.

Fig. 4. Main effect plot of the SN ratio for NOX.

Table 10 Table 11
Analysis of Variance for NOx, using Adjusted SS for Tests. Analysis of Variance for smoke, using Adjusted SS for Tests.
Source DF Seq SS Adj MS F P Remarks Source DF Seq SS Adj MS F P Remarks

Regression 3 487,973 162,658 142.19 0.000 F0.05,3,12 = 8.7446 Regression 3 22.9200 7.6400 31.58 0.000 F0.05,3,12 = 8.7446
Blends 4 332,727 83,182 82.64 0.000 F > F0.05,3,12 Blends 4 10.8000 2.7000 8.10 0.002 F > F0.05,3,12
I.T 4 88,216 22,054 21.91 0.000 Model is adequate I.T 4 5.6000 1.4000 4.20 0.024 Model is adequate
I.P 4 78,976 19,744 19.62 0.000 I.P 4 7.6000 1.9000 5.70 0.008
Error 12 12,078 1007 Error 12 4.0000 0.3333
Total 24 511,997 Total 24 28.0000

8
A. Sharma, et al. Fuel 281 (2020) 118765

Fig. 5. Main effect plot of the SN ratio for Smoke.

3.3. Effect of process variables on NOx and injection pressure were significant. The value of R2 is 0.953 which
was closed to 1. The R2(0.953) shows good agreement with R2-Adj
The main effect plot for the NOx concerning three variables is (0.946).
shown in Fig. 4. The value of S/N ratio is presented in Table 5. The From Fig. 4 it can be observed that on the increase of Pongamia
value of NOx was lowest at 0% blending of Pongamia biodiesel, fuel biodiesel blending, engine response factor NOx increases very rapidly
injection timing of 15°bTDC, and a fuel injection pressure of 16 MPa. that is an attribute of more oxygen content present in biodiesel due to
For calculation of the impact of the individual parameters, ANOVA that in-cylinder temperature during combustion were higher. On the
analyses were performed. Table 10 shows the result of the ANOVA increase of injection timing and injection pressure, NOx increases at a
analysis of NOx. The probability value regression model and process slower rate that is due to better combustion and slightly more exhaust
parameters were less than 0.05, which shows that the developed model gas temperature during combustion. However, the impact of process
of NOx was statically significant. It was observed that process para- parameters injection timing and injection pressure was less as com-
meters blending of Pongamia biodiesel percentage, injection timing, pared to the blending of Pongamia biodiesel. Many studies have

Fig. 6. Main effect plot of the SN ratio for the combined response factor overall utility.

9
A. Sharma, et al. Fuel 281 (2020) 118765

reported that the emission of NOx depends on various factors. This in- where, F (1, DOFe) is the fisher value, which is equal to the DOFe of error
cludes the higher fuel density, higher fuel viscosity, and delay period terms. Ve is the variance of the error terms and Ne is the effective
which were dependent on the chemical structure of the fuel [2,7,29]. number of replication. DOFm is the degree of freedom of mean term,
which is always equal to one, DOFi is the degree of freedom of ith
3.4. Effect of process variables on smoke significant process parameters and np is the number of design para-
meters that are affecting the quality characteristic. The best values of
Nowadays, smoke emission is one of the important issues in the the diesel engine operating process variables within the experimental
diesel engine. In this study, Pongamia biodiesel is blended with diesel domain are presented in Table 12. Confirmation test results have shown
to minimize the smoke in the exhaust of the diesel engine. A gas ana- that the hybrid approach Taguchi with utility theory is a power full tool
lyzer was used for the measurement of smoke. The solid soot particle in for the optimization of multi-objective optimization problems.
the exhaust emission is known as smoke emission. The emission of the
smoke depends on the presence of air in the mixture during combustion
4. Conclusion
and atomization of fuel. The lowest value of smoke (84%) was obtained
at the blending of Pongamia biodiesel of 40%, injection timing of
In this study, an experimental and statistical investigation has been
31°bTDC, and a fuel injection pressure of 24 MPa. Table11 shows the
conducted into single-cylinder DI diesel engines using pongamia bio-
results of the ANOVA analysis of experimental data. The probability
diesel blended fuels. Also, the effects of the blending percentage of
value of process parameters was less than 0.05, which shows that model
Pongamia biodiesel, fuel injection timing, and fuel injection pressure on
was significant. The blending of Pongamia biodiesel, fuel injection
engine responses such as BTE, UHC, NOx, and smoke emission were
timing, and fuel injection pressure was significant. The value of R2 is
assessed. The engine experiments were accompanied by the factorial
0.819 and R2-Adj was 0.793, which implies that there was a better
design matrix of Taguchi, coupled with the desirability approach, pre-
agreement between R2 and R2-Adj.
dicting and optimizing the input engine output response. Following
From Fig. 5 it can be observed that smoke was decreasing very ra-
conclusions have been drawn from the work;
pidly by increasing the biodiesel blending percentage due to the pre-
sence of more oxygen content in the mixture. Smoke content is de-
creasing at a slower rate by increasing the injection timing and injection • BTE was found to be maximum (31.057%) at process parameters i.e.
blending of Pongamia biodiesel 0%, fuel injection timing 31 °bTDC,
pressure. With an increase in fuel injection pressure, fuel gets easily
and an injection pressure of 24 MPa.
atomized resulting in proper mixing of the mixture during the com-
bustion while the increase in injection timing promotes delay in com- • The value of UHC was found to be minimum (33.835 ppm vol.) at
40% blending of Pongamia biodiesel, injection timing of 19 °bTDC,
bustion resulting in more smoke. The same trend with the effect of
and an injection pressure of 24 MPa.
injection pressure and injection timing including biodiesel was reported
by Agarwal et al. [2]. However, the impact of process parameters in- • NOx was found to be 520.4 ppm vol. at the level of process para-
meters with diesel, fuel injection timing 15 °bTDC, and an injection
jection timing and fuel injection pressure was less as compared to the
pressure 16 MPa.
blending of biodiesel.
• The value of Smoke was observed minimum at the level of process
parameter having 40% blending of Pongamia biodiesel, fuel injec-
3.5. Confirmation test tion timing 31 °bTDC, and injection pressure 24 MPa.

The confirmation test was performed at the best level of engine


• The value of R2 for the developed models for the response factors
viz., BTE, UHC, NOx, and Smoke were found to be 0.962, 0.962,
operating process condition to check the acceptability of the established 0.953, and 0.819. It indicates that the developed model was statis-
models of BTE, UHC, NOx, and Smoke. The average values of three tically accurate.
experiments were used for the comparison of model value and experi-
mental value. Results revealed that experimental values of response
• The best levels of engine operating process parameters optimized for
all the response factors were blending of Pongamia biodiesel 10%,
factors were within the range of 95% CI of the model values. The es- fuel injection timing 23°bTDC, and injection pressure 22 MPa. At
timated value of BTE, UHC, NOx, and Smoke can be obtained from Eqs. this engine setting engine responses were (25.32% BTE, UHC
(20)–(22). 42 ppm volume, NOx 761.34 ppm Vol. and 85% Smoke)
Expected response = Zopt ± CI (20) • The results of the ANOVA analysis indicated that the model F-value
of the response variables is greater than the F-tabulated and p-value
CI may be calculated by using equation20 of regression models and process variables are less than 0.05, which
proves that developed models are statistically correct.

F(1, DoFe)×Ve
CI = Confirmation experiments at the best levels of engine operating
Ne (21)
parameters have shown that model value and experimental values
n are with the range of confidence interval level of 95%.
Ne = np
DOFm+ DOFi
i= 1 (22) CRediT authorship contribution statement

Table 12 Abhishek Sharma: Data curation, Writing - original draft.


Validation of the results. Nagendra Kumar Maurya: Writing - original draft. Yashvir Singh:
Response Process parameter Model Value Experimental Writing - original draft. Nishant Kumar Singh: Writing - review &
Factor Value editing. Sandeep Kumar Gupta: Writing - review & editing.
Blending Fuel Injection
of injection pressure
biodiesel timing Declaration of Competing Interest
BTE (%) 10% 23 22 MPa 24.6336 25.32
UHC (°bTDC) 42.32 42.00 The authors declare that they have no known competing financial
NOX 797.5 761.34 interests or personal relationships that could have appeared to influ-
Smoke 86.3 85.00
ence the work reported in this paper.

10
A. Sharma, et al. Fuel 281 (2020) 118765

References J Adv Manuf Technol 2010;51(5–8):627–38.


[16] Singh Y, Sharma A, Tiwari S, Singla A. Optimization of diesel engine performance
and emission parameters employing cassia tora methyl esters-response surface
[1] Ong HC, Masjuki HH, Mahlia TMI, Silitonga AS, Chong WT, Yusaf T. Engine per- methodology approach. Energy 2019;168:909–18.
formance and emissions using Jatropha curcas, Ceiba pentandra and Calophyllum [17] Damodharan D, Sathiyagnanam AP, Rana D, Kumar BR, Saravanan S. Combined
inophyllum biodiesel in a CI diesel engine. Energy 2014;69:427–45. influence of injection timing and EGR on combustion, performance and emissions of
[2] Agarwal AK, Dhar A, Gupta JG, Kim WI, Choi K, Lee CS, et al. Effect of fuel injection DI diesel engine fueled with neat waste plastic oil. Energy Convers Manage
pressure and injection timing of Karanja biodiesel blends on fuel spray, engine 2018;161:294–305.
performance, emissions and combustion characteristics. Energy Convers Manage [18] Kumar BR, Saravanan S, Rana D, Nagendran A. Combined effect of injection timing
2015;91:302–14. and exhaust gas recirculation (EGR) on performance and emissions of a DI diesel
[3] Patel C, Tiwari N, Agarwal AK. Experimental investigations of Soyabean and engine fuelled with next-generation advanced biofuel–diesel blends using response
Rapeseed SVO and biodiesels on engine noise, vibrations, and engine character- surface methodology. Energy Convers Manage 2016;123:470–86.
istics. Fuel 2019;238:86–97. [19] Gad M, Jayaraj S. A comparative study on the effect of nano-additives on the
[4] Sharma A, Singh Y, Singh GK, Habte AT, Singh N. Production of polanga methyl performance and emissions of a diesel engine run on Jatropha biodiesel. Fuel
esters and optimization of diesel engine parameters through response surface 2020;267:117168.
methodology approach. Process Saf Environ Prot 2019;121:94–102. [20] Balki MK, Sayin C, Sarıkaya M. Optimization of the operating parameters based on
[5] Singh Y, Sharma A, Kumar Singh G, Singla A, Kumar SN. Optimization of perfor- Taguchi method in an SI engine used pure gasoline, ethanol and methanol. Fuel
mance and emission parameters of direct injection diesel engine fuelled with pon- 2016;180:630–7.
gamia methyl esters-response surface methodology approach. Ind Crops Prod [21] Ansari NA, Sharma A, Singh Y. Performance and emission analysis of a diesel engine
2018;126:218–26. implementing polanga biodiesel and optimization using Taguchi method. Process
[6] Patel H, Rajai V, Das P, Charola S, Mudgal A, Maiti S. Study of Jatropha curcas shell Saf Environ Prot 2018;120:146–54.
bio-oil-diesel blend in VCR CI engine using RSM. Renew Energy 2018;122:310–22. [22] Rao DB, Rao KV, Krishna AG. A hybrid approach to multi response optimization of
[7] Reddy SM, Sharma N, Gupta N, Agarwal AK. Effect of non-edible oil and its bio- micro milling process parameters using Taguchi method based graph theory and
diesel on wear of fuel injection equipment components of a genset engine. Fuel matrix approach (GTMA) and utility concept. Measurement 2018;120:43–51.
2018;222:841–51. [23] Du X, Leifsson L. Optimum aerodynamic shape design under uncertainty by utility
[8] Abed KA, Gad MS, El Morsi AK, Sayed MM, Elyazeed SA. Effect of biodiesel fuels on theory and metamodeling. Aerosp Sci Technol 2019;95:105464.
diesel engine emissions. Egypt J Pet 2019. [24] Yusri I, Majeed AA, Mamat R, Ghazali M, Awad OI, Azmi W. A review on the ap-
[9] Milano J, Ong HC, Masjuki HH, Silitonga AS, Chen W-H, Kusumo F, et al. plication of response surface method and artificial neural network in engine per-
Optimization of biodiesel production by microwave irradiation-assisted transester- formance and exhaust emissions characteristics in alternative fuel. Renew Sustain
ification for waste cooking oil-Calophyllum inophyllum oil via response surface Energy Rev 2018;90:665–86.
methodology. Energy Convers Manage 2018;158:400–15. [25] Jagtap SP, Pawar AN, Lahane S. Improving the usability of biodiesel blend in low
[10] Ong HC, Milano J, Silitonga AS, Hassan MH, Shamsuddin AH, Wang C-T, et al. heat rejection diesel engine through combustion, performance and emission ana-
Biodiesel production from Calophyllum inophyllum-Ceiba pentandra oil mixture: lysis. Renew Energy 2020.
optimization and characterization. J Cleaner Prod 2019;219:183–98. [26] Hirkude BJ, Padalkar AS. Performance optimization of CI engine fuelled with waste
[11] Juan JC, Kartika DA, Wu TY, Hin T-Y-Y. Biodiesel production from jatropha oil by fried oil methyl ester-diesel blend using response surface methodology. Fuel
catalytic and non-catalytic approaches: an overview. Bioresour Technol 2014;119:266–73.
2011;102(2):452–60. [27] Silitonga A, Masjuki H, Mahlia T, Ong H, Chong W, Boosroh M. Overview properties
[12] Silitonga AS, Masjuki HH, Ong HC, Sebayang AH, Dharma S, Kusumo F, et al. of biodiesel diesel blends from edible and non-edible feedstock. Renew Sustain
Evaluation of the engine performance and exhaust emissions of biodiesel-bioe- Energy Rev 2013;22:346–60.
thanol-diesel blends using kernel-based extreme learning machine. Energy [28] Adepoju TF. Optimization processes of biodiesel production from pig and neem
2018;159:1075–87. (Azadirachta indica a.Juss) seeds blend oil using alternative catalysts from waste
[13] Dharma S, Masjuki H, Ong HC, Sebayang A, Silitonga A, Kusumo F, et al. biomass. Ind Crops Prod 2020;149:112334.
Optimization of biodiesel production process for mixed Jatropha curcas–Ceiba [29] Jain A, Singh AP, Agarwal AK. Effect of split fuel injection and EGR on NOx and PM
pentandra biodiesel using response surface methodology. Energy Convers Manage emission reduction in a low temperature combustion (LTC) mode diesel engine.
2016;115:178–90. Energy 2017;122:249–64.
[14] Domínguez-Sáez A, Rattá GA, Barrios CC. Prediction of exhaust emission in tran- [30] Sharma A, Singh Y, Kumar Singh N, Singla A, Chyuan Ong H, Chen W-H. Effective
sient conditions of a diesel engine fueled with animal fat using Artificial Neural utilization of tobacco (Nicotiana Tabaccum) for biodiesel production and its ap-
Network and Symbolic Regression. Energy 2018;149:675–83. plication on diesel engine using response surface methodology approach. Fuel
[15] Patowari PK, Saha P, Mishra P. Artificial neural network model in surface mod- 2020;273:117793.
ification by EDM using tungsten–copper powder metallurgy sintered electrodes. Int

11

You might also like