Content uploaded by Ravikumar Ramegouda
Author content
All content in this area was uploaded by Ravikumar Ramegouda on Feb 27, 2023
Content may be subject to copyright.
Fuel 341 (2023) 127575
Available online 10 February 2023
0016-2361/© 2023 Elsevier Ltd. All rights reserved.
Enhancing the performance of renewable biogas powered engine
employing oxyhydrogen: Optimization with desirability and
D-optimal design
Prabhakar Sharma
a
,
*
, Dhinesh Balasubramanian
b
,
*
, Chu Thanh Khai
c
,
Inbanaathan Papla Venugopal
b
, Mansoor Alruqi
d
, Femilda Josephin JS
e
, Ankit Sonthalia
f
,
*
,
Edwin Geo Varuvel
g
,
*
, Esmail Khalife
h
, R. Ravikumar
i
, Makatar Wae-Hayee
j
a
Mechanical Engineering Department, Delhi Skill and Entrepreneurship University, Delhi 110089, India
b
Department of Mechanical Engineering, Mepco Schlenk Engineering College, Tamilnadu, India
c
Faculty of Technology, Dong Nai Technology University, Bien Hoa , Dong Nai, Viet Nam
d
Department of Mechanical Engineering, College of Engineering, Shaqra University, Al Riyadh 11911, Saudi Arabia
e
Department of Computer Engineering, Faculty of Engineering and Natural Sciences, Istinye University, Istanbul, Turkey
f
Department of Automobile Engineering, SRM Institute of Science and Technology, NCR Campus, Modi Nagar, 201204, India
g
Department of Mechanical Engineering, Faculty of Engineering and Natural Sciences, Istinye University, Istanbul, Turkey
h
Department of Civil Engineering, Cihan University-Erbil, Kurdistan Region, Iraq
i
Department of Mechanical and Automobile Engineering, CHRIST University, Bengalore, India
j
Department of Mechanical and Mechatronics Engineering, Faculty of Engineering, Prince of Songkla University, Hatyai, Songkhla 90110, Thailand
ARTICLE INFO
Keywords:
Oxyhydrogen
Renewable
Optimization
Biogas
D-optimal
Desirability
ABSTRACT
The performance and exhaust characteristics of a dual-fuel compression ignition engine were explored, with
biogas as the primary fuel, diesel as the pilot-injected fuel, and oxyhydrogen as the fortifying agent. The trials
were carried out with the use of an RSM-based D-optimal design. ANOVA was used to create the relationship
functions between input and output. Except for nitrogen oxide emissions, oxyhydrogen fortication increased
biogas-diesel engine combustion and decreased carbon-based pollutants. For each result, RSM-ANOVA was
utilized to generate mathematical formulations (models). The output of the models was predicted and compared
to the observed ndings. The prediction models showed robust prediction efciency (R
2
greater than 99.21%).
The optimal engine operating parameters were discovered by desirability approach-based optimization to be 24◦
crank angles before the top dead center, 10.88 kg engine loading, and 1.1 lpm oxyhydrogen ow rate. All
outcomes were within 3.75% of the model’s predicted output when the optimized parameters were tested
experimentally. The current research has the potential to be widely used in compression ignition engine-based
transportation systems.
1. Introduction
The world is undergoing unprecedented levels of human develop-
ment and advancement, which has stretched human activities to
encompass all parts of the ecosystem. To safeguard the environment and
human life, however, human activities and natural resources must both
function ethically and sustainably [1]. The global community has wit-
nessed several signicant monuments of international policy that have
shaped the global environment, especially the 2030 Agenda for Sus-
tainable Development and Paris Agreement on Climate Change (PACC)
[2]. The latter has established seventeen sustainable development goals
(SDGs) as a global platform and a roadmap for peace and prosperity for
people and the planet. The United Nations adopted these goals. The fast
growth of the world’s population, as well as the advancement of in-
dustrial technology, have both contributed to the accumulation of
greenhouse gases (GHGs), which has caused climate change [3].
According to the ndings of the 1.5 ◦C Global Warming Special Study
conducted by the Intergovernmental Panel on Climate Change (IPCC),
the planet must maintain a temperature rise of no more than 1.5 ◦C to
avoid having negative effects on populations and ecosystems. This aim is
still attainable, but it requires the world to reach a level of net-zero
* Corresponding authors: Department of Mechanical Engineering, Mepco Schlenk Engineering College, Tamilnadu, India (D. Balasubramanian).
E-mail addresses: prabhakar.sharma@dseu.ac.in (P. Sharma), dhineshbala91@gmail.com (D. Balasubramanian).
Contents lists available at ScienceDirect
Fuel
journal homepage: www.elsevier.com/locate/fuel
https://doi.org/10.1016/j.fuel.2023.127575
Received 1 August 2022; Received in revised form 27 December 2022; Accepted 20 January 2023
Fuel 341 (2023) 127575
2
emissions by the year 2050 and calls for strong, rapid, and long-term
action on the international climate front [4,5]. The PACC was agreed
upon by 195 nations in 2015, and the framework for the COP24 summit,
which went into force at the beginning of 2019, was the rst step toward
a worldwide revolution in the way energy is generated [6]. To delay or
halt the course of climate change, we must swiftly shift away from fossil
fuels and toward clean, sustainable, and renewable energy sources [7].
Incorporating waste-to-energy (WtoE) into waste management is
widely considered to be a feasible option today, as a result of the many
decades of research and industry activities that led to this conclusion. It
gives a variety of advantages [8]. The use of WtoE to replace fossil fuels
in a variety of settings may qualify it as a source of energy that is only
partially replenishable. WtoE systems refer to any approach that gen-
erates any kind of energy, such as electricity or fuels, from a waste-based
feedstock This includes producing biogas [9]. The development of bio-
energy relies heavily on the use of primary feedstock such as dairy
waste, agricultural waste, and municipal solid waste. Carbon dioxide
(CO
2
) and Methane (CH
4
) are the primary components of biogas, which
are created as a by-product of the digestive process, also known as the
decomposition of organic waste. Because the feedstock that is required
for the generation of biogas is anticipated, these processes are appro-
priate in a great number of nations throughout the world [10].
Municipal waste naturally undergoes the process of biological
dissociation, resulting in the production of biogas. The biogas produced
has the potential to become an alternative energy source that can power
remote communities in areas of the world that are not linked to the grid.
It is also suited for grid decentralization [11]. In response to the rising
cost of fossil-based fuels and the pollution caused by ICEs that utilize
them, researchers from all over the world are focusing their efforts on
fuel additives and alternative fuels [12]. The vast majority of emissions,
including carbon dioxide, carbon monoxide, and hydrocarbons, can be
traced back to carbon, which is present in the bulk of hydrocarbon fuels.
As a consequence of this, an alternative fuel that does not include carbon
would be fantastic for lowering these toxic emissions, which have
disastrous repercussions not only for the health of humans but also for
the health of the environment [13]. One example of an alternative fuel
additive is oxy-hydrogen gas. This gas is produced by the electrolysis of
water and has the potential to improve engine performance while
simultaneously lowering exhaust emissions [14]. This gas has tradi-
tionally been used in a number of applications such as air conditioners,
hazardous waste annihilation, fruit drying, welding, and heating. The
fundamental reason for its extensive use is its greater energy level as
compared to a gaseous mixture of hydrogen and oxygen [15,16].
According to a study on the use of biogas in CI engines, it was
determined that various engine operating factors greatly impact engine
emission and performance. As a consequence, adjusting these factors can
result in signicant improvements in both engine efciency and pollu-
tion levels [17,18]. Experiments were conducted by modifying the pa-
rameters one by one; nevertheless, the multivariable assessment may be
an effective strategy for gaining a thorough knowledge of the DF en-
gine’s properties. Non-linear approaches that may be utilized for
multivariate analytic study include machine learning, statistical
methods of optimization, and model prediction [19].
Response surface methodology (RSM) is an excellent tool for design
of experiments (DoE). It helps in investigating the combined impacts of
operational factors on the output responses [41]. It is one of the ap-
proaches that fall under this category. This multi-input/output statistical
strategy concurrently augments the combined impacts of several
different parameters as well as the interacting inuences of the variables
to produce the greatest possible performance from the system [20]. The
key advantage of employing this method over comprehensive factorial
design experimentation is that it includes fewer experiments and takes
less time. Though various studies applying RSM in internal combustion
engine applications have been published, relatively limited study has
been done on the combined impacts of biogas-oxyhydrogen blends and
operational parameters such as load, oxyhydrogen ow rates (OFR), and
diesel injection timings. This provides an opportunity for the current
study.
Several experts have recently reported on the benets of using
oxyhydrogen in internal combustion engines (IC) engines. Paparao and
Murugan [21] explored the working of a CI engine powered with
Jatropha-diesel blends as injected fuel while oxyhydrogen (HHO) as the
main fuel. The study reported improvement in Heat release rate (HRR)
and brake thermal efciency (BTE) 5.2% and 1.1%, respectively. The
unburnt hydrocarbon (UHC), and smoke emissions are reduced by
approximately 33% and 18.5%, respectively. Tuccar [22] employed the
HHO gas and Foeniculum Vulgare oil biodiesel to run a CI engine. The
impact of this novel fuel combination on noise, vibration, and emission
was studied. The ndings demonstrated a reduction in the overall noise
and vibrations produced by the engine. Furthermore, a reduction in CO
emissions and a rise in NOx were reported. The important studies in this
domain are listed in Table 1:
The biogas-powered engine had been gaining increased attention
from researchers owing to its reduced exhaust emission compared to a
diesel engine. Biogas can be employed in the existing CI engine in dual
fuel mode without considerable hardware modication. This preposi-
tion makes it an attractive option considering the circular economy
aspect in terms of waste to energy and reuse concept. Goga et al. [28]
reported that the introduction of biogas in CI engines leads to BTE
reduction of up to 14.68% with increased CO and HC emissions by
45.58% and 17.74%, compared to diesel-only mode. Similar results were
reported by Nayak et al. [29] depicting a reduction in BTE and an in-
crease in the CO and HC emissions. The authors reported that advancing
the biodiesel injection timing could compensate for the reduction in BTE
while improving exhaust emission. In addition to these several other
studies reported the reduction in BTE as well as power derating from
using biogas as fuel [8,30].
Nomenclature
PCP Peak cylinder pressure
OFR Oxyhydrogen ow rate
CO Carbon monoxide
BTE Brake thermal efciency
NOx Nitrogen oxides
CI Compression ignition
DC Direct current
CO2 Carbon monoxide
UHC Unburnt hydrocarbon
ppm Parts per million
SDG Sustainable Development goal
HHO Oxyhydrogen
RSM Response surface methodology
SI Spark ignition
ANOVA Analysis of variance
KOH Potassium hydroxide
Lpm Liters per minaret
HP Horsepower
◦CA bTDC Degree crank angle before top dead center
SS Stainless steel
APC Argon power cycle
OFR Oxyhydrogen ow rate
PIT Pilot injection timing
R2 Coefcient of determination
P. Sharma et al.
Fuel 341 (2023) 127575
3
1.1. Contribution of the present study
The various measures used by researchers to enhance BTE and
braking power have been conned to changing operational factors such
as injection time and pressure, which results in only moderate
improvement. The use of hydrogen as an additive to biogas remains an
expensive endeavour due to storage and safety considerations [42]. The
usage of a low-cost technology known as a dry cell oxyhydrogen
generator has shown to be an appealing choice in a biodiesel engine.
However, it had not been investigated to improve the combustion and
emission of a biogas-powered engine in DF mode. As a result, there is a
research void in this area. The proposed research is an innovative
attempt to solve this research gap by investigating the use of a low-cost
device for manufacturing HHO for blending with biogas in the engine
induction system.
Based on literature in the biogas-powered engine domain, it is
possible to infer that analyzing a large number of input elements pro-
vides more insight into an engine’s operating conditions. These multi-
variable optimizations and regression methods determine the optimal
combination of input elements to achieve the desired engine output.
However, there is only a little number of multivariate analyses of a DF
engine utilizing RSM described in the linked scholarly literature.
Furthermore, no literature on the employment of RSM for analyzing
biogas-diesel fortied with oxyhydrogen as gaseous additive powered
engines with three distinct input parameters, namely engine load,
oxyhydrogen ow rates, and pilot injection timing, was found. Given
this gap, the present study uses D-optimal design-based RSM for
analyzing the combined inuence of independent control factors (engine
load, oxyhydrogen ow rates, and pilot injection timing) on engine
performance and tailpipe emission. Regression models are used to
forecast output responses for all input variable combinations. Using the
desirability method, the research also investigates the optimal combi-
nation of input criteria for reducing pollutants and maximizing engine
efciency.
2. Materials and methods
2.1. Engine setup
Fig. 1 depicts the test engine’s schematic, complete with all in-
strumentations and accessories. Table 2 provides the technical specs for
the test engine and loading units. Apex Innovations, Pune supplied the
whole test engine facilities while the test engine was from M/s Kirloskar
Brothers ltd. The engine load was varied by employing an eddy current-
based dynamometer. The computerized test bench has the provision to
vary the engine load by controlling the electric supply to the dyna-
mometer. A U-tube type manometer is tted beside the airbox right
before the engine’s intake manifold to determine the air supply to the
engine cylinder. To monitor the fuel ow into the engine, a glass burette
with a digital stopwatch was used. The temperature of the exhaust gas
was monitored using a K-type thermocouple inserted at the starting
point of the exhaust pipe. The speed of engine speed is measured using a
non-contact type speed sensor mounted on the dynamometer shaft end.
A Type S bean load sensor was used to measure the engine loads.
2.2. Oxyhydrogen generation
Oxyhydrogen is a mixture of oxygen gas and diatomic hydrogen gas.
An oxyhydrogen gas generator, also known as an electrolytic cell, uses
an electric current to create HHO gas. Water is electrochemically con-
verted into oxygen and hydrogen as a result of ionization interactions,
leading to the formation of HHO. In the present investigation, a 31 plates
HHO gas generator was employed to fortify the biogas. A ashback
arrestor connects the wet cell generator to the engine intake manifold.
During operation, the generator can produce up to 2 lpm of HHO. A
ashback arrestor was tted in the fuel line in the case of a backre to
ensure that the ame didn’t return to the gas supply system. The main
characteristics of the test HHO gas generator and test fuels used in the
present study are listed in Table 2 and Table 3, respectively.
2.3. Soft computing approach
2.3.1. D-optimal design
A scientic approach to the design of experiments (DoE) results in a
fewer number of experimental runs. In the present work, the Design-
Expert was employed for DoE, analysis of variance (ANOVA), develop-
ment of correlation expressions, and optimization. The RSM is a
powerful mathematical program and an outstanding optimization tool
[32]. It may assess the empirical link between many factors. In this
work, the D-optimal modeling technique was used due to its simplicity,
practicality, efciency, and low number of test runs. The factors
considered in the study were denoted by x1,x2,x3,⋯⋯., xn and where
the number of factors is denoted with n. The model can be expressed in
form of main effects and a combination of main effects and interactions
[33]:
Y=βo+∑
n
i=1
βixi+∊(1)
Y=βo+∑
n
i=1
βixi+∑
k−1
i=1
∑
n
j=i+1
βijxixj+∊(2)
herein, the Y denotes the response of the model while the intercept
factor is denoted with βo, the single term of β like β1,β2,β3,⋯,βn shows
the main effect terms (Eq. (1)) while the Eq. (2) shows the second-order
terms depicting the interaction. Given so we also presume that almost all
Table 1
Related studies on biogas powered engine.
Reference Type of fuel Experiential/
Optimization
Outcomes
Kanimozhi
et al. [23]
Moringa Oleifera +
diesel +
Oxyhydrogen
Experimental (i) BSFC reduced
compared to
biodiesel(ii)
6% reduction in
NO
x
emission at low
loads
Kazim et al.
[24]
Diesel +
Oxyhydrogen
Experimental Both torque and
efciency improved
Shahjahan
et al. [25]
Gasoline +
Oxyhydrogen
Experimental (i) 5 to 7%
improvement in BTE
(ii)
Up to 11%
reduction in CO
emission
Bhave et al.
[26]
Diesel +
Oxyhydrogen
Experimental (i) Up to 37.5%
reduction in CO
emission(ii)
7.41% increase in
NO
x
emission
Paparao et al.
[21]
Jatropha biodiesel
+diesel +
oxyhydrogen
Experimental (i) 1.1%
improvement in BTE
(ii)
Up to 19.6%
increase in NO
x
(iii)
29.4% reduction in
CO emission
Kushwaha
and Ismail
[27]
Diesel +
Oxyhydrogen
Experimental Increase in BTE and
NO
x
emission was
reported.
Present study Biogas +diesel +
oxyhydrogen in dual
fuel mode
Experimental +
Optimization
(i) Increase in BTE.
(ii)
Marginal increase in
NO
x
emission.(iii)
Parametric
optimization
P. Sharma et al.
Fuel 341 (2023) 127575
4
n variances are independently uniformly distributed random factors
with zero mean and variance
σ
2 , the optimal linear neutral estimator for
the factors β,is the standard least estimator [34].
β=(XTX)−1XTY and its variance–covariance matrix form is
σ
2(XTX)−1 and its inverse
σ
−2(XTX)is the Fisher information matrix,
which describes the experimental design’s informativeness on the
parameter estimates under study. The independent control variables in
this research were engine load, oxyhydrogen ow rate, and pilot injec-
tion timing. The response parameters were engine performance pa-
rameters (BTE) and tailpipe emission (NO
x
, CO, and HC). The D-optimal
design was employed to carry out 20 experimental runs. The motive for
choosing a D-optimal design was that conventional factorial designs
require a greater number of runs to achieve the same outcomes as D-
optimal ones [35]. Moreover, the design space in conventional factorial
designs is restricted, and it contains some factors that are either
impractical or difcult to implement. D-optimal-based designs are more
viable, simpler to execute, and dependent on linear optimization [36].
The designed array developed following the D-optimal design is shown
in Table 4.
2.3.2. Data analysis using ANOVA
ANOVA is a technique to determine if the average of two data groups
differs. Samples are used in inferential analysis to infer sample size.
ANOVA and other statistical methods help one to nd if sample out-
comes are pertinent to data groups. F-value assesses both, the variation
among groups and variance within the groups. A greater F-value value
shows if the variance in the group dominates the group variation. The p-
value may be used to assess if variances among a few of the data means
are statistically substantial [37,38].
Table 5 reveals that the F-value was 1439.96 for the BTE model. This
high value suggests that it was noteworthy to warrant attention. Simi-
larly, the p-values <0.0500 suggest the model terms of the BTE model
are signicant. The model variables L, P, O, LO, P
2
, and O
2
were crucial
in this condition.
Table 6 shows the outcome of the ANOVA nding in the case of
emission data. The F-value of the CO model was 175.73, indicating that
it is signicant enough. The NO
x
model relies heavily on the model
variables L, P, O, LP, PO, L2, P2, and O2. The F-value of 155.66 in the
instance of the CO emission model indicates that the model was
Fig. 1. Schematics of test set up.
Table 2
Test engine specications of the engine.
Name Specication
Supplier and
Manufacturer
Apex innovations and M/S
Kirloskar
Stated fuel Diesel
Speed (rpm) 1500 ±50, constant speed
Test Engine Bore ×stroke (m) 0.0875 ×0.11
Rated power 7HP, 5.2 kW
Ratio of compression 17.5
Number of strokes &
cylinder
4 & 1
Number of plates 31
Plate material SS 316
Electric current ratings 13 Ampere
Oxyhydrogen
generator
Voltage 12/24 Volts
Peak output 2 L per minute
Gasket Rubber (Neoprene)
End plate material Nylon
Table 3
Main characteristics of the test fuel.
Characteristics Biogas Diesel Oxyhydrogen [26,31]
Kinematic viscosity (@40 ◦C) – 2.67 –
Cetane number – 47 –
Cold-lter plugging index (◦C) – −0.4 –
Flash point (◦C) – 117 –
Lower heating value (MJ/kg) 21.12 42 21.995
Density (kg/m
3
) @ 15 ◦C 0.91 846 0.49115
P. Sharma et al.
Fuel 341 (2023) 127575
5
signicant enough to warrant attention. The CO emission was shown to
be inuenced by all of the primary effects (L, P, O, L
2
, P
2
, and O
2
) as well
as interaction factors (LP, PO, and OL). The F-value of 139.38 in the
instance of the HC model suggests the produced model was notable. The
model parameters L, P, O, LP, OP, OL, L
2
, P
2
, and O
2
were crucial for the
development of the HC model.
3. Results and discussion
A compression ignition (CI) engine was employed in this study in a
dual-fuel model with biogas-HHO-diesel. To boost efciency and reduce
carbon-based emissions, a low-cost technology for creating oxyhydrogen
gas was used to fortify the biogas-diesel combination. The experimental
investigation was carried out using the D-optimal design, which was
followed by model construction and optimization. The ternary fuel-
driven engine’s thermal performance (BTE) and tailpipe emissions
(NO
x
, HC, and CO) were all tested. The ANOVA was employed initially
to build the connection function among parameters in the form of
algebraic expressions to serve as a prognostic model. The created models
are explained in the following paragraphs:
3.1. Engine performance models
3.1.1. BTE modsel
Fig. 2(a-b) depicts the inuence of oxyhydrogen fortication on
biogas-diesel oil-powered CI engine’s thermal efciency using two sur-
face diagrams. The current research sought to investigate the impacts of
three independent control variables (load on engine, oxyhydrogen ow
rates, and PIT) on the dependent response output i.e., BTE. The surface
diagrams depict the inuence of the control factors on the dependent
response factors in an improved way compared to the single factor at a
time (SFAT) graph. As SFAT illustrates the inuence of one factor on one
output variable and does not reveal the interaction effects of input pa-
rameters. As a consequence, surface diagrams are more suited for dis-
playing the ndings. The ndings must be considered with the outcomes
of the ANOVA in the Table 5. Because oxyhydrogen has a larger caloric
value and a quicker ame speed than biogas, using it in the present test
engine framework improves thermal efciency [39]. When added to the
diesel-biogas mix, HHO gas improves the braking power of the engine.
Among all three control factors, the load on engine showed the
highest inuence (64.85%) on BTE (Fig. 2 and Table 5), followed by PIT
(4.15%) and oxyhydrogen ow rates (2.38%). The remaining contri-
bution is attributed to interactive terms. The governor supplies a greater
quantity of fuel to maintain engine speed as the engine load rises. Higher
BTE with increased engine load is caused by a faster rate of fuel com-
bustion. The FIT given by the manufacturer was 23 ◦CA bTDC. An
advance in PIT toward the higher range led to marginal improvement in
BTE. It allowed extra time for the combustion to complete. The inter-
action effect of PIT and load, as shown in Fig. 2(a), demonstrates that the
maximal BTE could be attained at 13 kg engine loading and 25 ◦CA
bTDC PIT. Likewise, Fig. 2(b) shows that elevated engine load and
greater HHO ow contributed to improved BTE. Eq. (3) depicts the
quadratic equation for calculating BTE (%) using the independent con-
trol factors. Eq. (3) was employed to predict the BTE and the predicted
values are compared in Fig. 2(c). As almost all the data points are near to
the best t line, there is a high degree of correlation between experi-
mental and predicted values (45◦line).
Table 4
D-optimal-based design array.
Exp.
Run
Engine load,
kg
PIT,
CA bTDC
OFR,
lpm
BTE,
%
NO
x
,
ppm
CO,
ppm
HC,
ppm
1 10 26 0.3 21.35 161 120 109
2 10 23 0.3 20.14 164 125 114
3 6 20 0.7 13.75 131 144 136
4 10 26 1.1 24.41 195 115 106
5 10 26 1.1 24.4 195 116 106
6 6 26 0.3 15.45 127 143 135
7 14 23 0.7 27.98 274 142 133
8 14 20 1.1 26.21 225 135 126
9 14 23 0.7 27.95 275 142 133
10 14 23 0.3 26.47 228 144 135
11 10 20 0.7 20.19 165 126 115
12 14 20 0.3 24.78 215 150 141
13 6 23 0.3 14.14 127 147 140
14 14 26 0.3 27.89 224 144 132
15 10 23 0.3 20.12 165 125 114
16 6 26 0.7 17.18 146 135 126
17 6 20 0.3 12.14 118 155 147
18 14 23 0.7 27.96 272 141 132
19 6 23 1.1 17.32 146 125 116
20 10 20 0.7 20.2 161 126 115
Table 5
ANOVA results for performance characteristics.
Parameter BTE
Source SoS Value ’F’ p-value
Model 522.04 1439.96 <0.0001
L 345.22 8570.15 <0.0001
P 27.27 676.9 <0.0001
O 16.16 401.15 <0.0001
LP 0.002 0.0492 0.829
LO 0.6061 15.05 0.0031
PO 0.168 4.17 0.0684
L
2
0 0.0004 0.9846
P
2
0.3753 9.32 0.0122
O
2
0.6562 16.29 0.0024
Error 0.4028
Cor Total 522.44
Table 6
ANOVA results for emission characteristics.
Parameter NO
x
CO HC
Source SoS Value ’F’ p-value SoS Value ’F’ p-value SoS Value ’F’ p-value
Model 48908.95 175.73 <0.0001 2639.16 155.66 <0.0001 2985.15 139.38 <0.0001
L 31721.97 1025.78 <0.0001 45.95 24.39 0.0006 30.92 12.99 0.0048
P 2030.97 65.67 <0.0001 43.15 22.9 0.0007 42.21 17.74 0.0018
O 1164.71 37.66 0.0001 492.64 261.51 <0.0001 469.75 197.39 <0.0001
LP 51.17 1.65 0.2273 22.41 11.9 0.0062 16.08 6.76 0.0265
LO 276.77 8.95 0.0135 64.38 34.17 0.0002 99.35 41.75 <0.0001
PO 543.35 17.57 0.0019 74.21 39.4 <0.0001 112.67 47.35 <0.0001
L
2
1478.76 47.82 <0.0001 1571.5 834.19 <0.0001 1888.53 793.58 <0.0001
P
2
956.4 30.93 0.0002 32.27 17.13 0.002 23.94 10.06 0.01
O
2
1312.01 42.43 <0.0001 14.39 7.64 0.02 12.3 5.17 0.0463
Error 309.25 18.84 23.8
Cor Total 49218.2 2658 3008.95
P. Sharma et al.
Fuel 341 (2023) 127575
6
BTE = − 26.29 +1.67L+1.99P+5.73O−0.0015L×P−0.198L
×O+0.134P×O−0.000123L2−0.033P2−2.59O2(3)
herein, the acronym ’BTE’ stands for brake thermal efciency, ’PIT’ for
the timing of pilot injection, ’o’ for rate of oxyhydrogen ow, and ’l’ for
the load.
3.2. Exhaust emission models
3.2.1. No
x
emissions model
The oxides of Nitrogen (NO
x
) have been considered harmful airborne
contaminants, and the diesel combustion contributes signicantly to
NO
x
emissions. Because of increasingly rigorous emission rules, great
effort is being made to eliminate or reduce NO
x
emissions from com-
bustion engines. Owing to the complexity and unknown chemical
species involved in its synthesis, the development of kinetic model of
NO
x
emission remains a complex phenomenon [43]. The parametric
modeling described in this paper provides a simpler machine-specic
option. It does, however, provide a foundation for developing a model
that may then be utilized for optimization. This reduces NO
x
emissions
by adjusting the operating settings to a need-based optimal range. Eq.
(4) was derived in this work to estimate NO
x
emissions.
NOx = − 685.04 −17.95L+73.26P−28.54O+0.24L×P+4.24
×L×O+7.64P×O+1.19L2−1.66P2−115.85O2
(4)
Herein, the acronym ’NO
x
’ stands for nitrogen oxide monoxide, ’PIT’ for
the timing of pilot injection, ’o’ for rate of oxyhydrogen ow, and ’l’
denotes load on the engine.
Fig. 3(a) shows variation in the NO
x
emissions vis-`
a-vis engine
Fig. 2. Model of BTE; (a) 3-D surface plots for PIT vs load; (b) 3-D surface plots for rate of oxyhydrogen ow vs load; (c) Model predicted vs observed.
P. Sharma et al.
Fuel 341 (2023) 127575
7
loading and PIT. Among the control factors, the load on engine had the
most signicant inuence on NO
x
emissions. This was validated by
ANOVA results as shown in Table 6. The data analysis shows the engine
load had almost 64.86% inuence over NO
x-
type emission. On higher
load, the greater cylinder pressure and temperature led to elevated NO
x
generation. This was attributed to the oxidation of nitrogen at elevated
cylinder temperatures, forming oxides of nitrogen oxides [44]. The
addition of oxyhydrogen and increase in NO
x
were shown to be directly
linked in the Fig. 3(b). The impact of PIT variation on NO
x
emissions was
quite modest. As shown in Fig. 3(a), the highest NO
x
emissions were
recorded at advanced PIT (25◦CA bTDC) and 100% load. The surface
diagrams show the cumulative inuence of operational settings. The
lowest NO
x
emission levels were reported at 6 kg engine loading, 20◦
injection advance, and 0.3 lpm rate of oxyhydrogen ow. Eq. (4) was
used to predict the emission of NO
x
emissions values over different
settings of engine operation. The comparison of model forecasted and
actual NO
x
values is depicted in Fig. 3(c). A robust prediction efciency
was observed in the results.
3.2.2. CO emission model
The variations in CO emission values with engine loading and PIT
change are depicted in Fig. 4(a). The data acquired in the experimental
phase shows that the load on the engine had the most substantial in-
uence on CO emission. The ANOVA outcomes (ref Table 6) show that
engine load and its quadratic component contributed to almost 61.28%
of CO emissions. However, the effects of engine load were not
Fig. 3. Model of NO
x
; (a) 3-D surface plots for PIT vs load; (b) 3-D surface plots for rate of oxyhydrogen ow vs load; (c) Model predicted vs observed.
P. Sharma et al.
Fuel 341 (2023) 127575
8
proportional, as shown by ANOVA ndings, because the second order
component was signicantly bigger compared to the main effect term. It
was observed that CO emission rst fell up to 10 kg engine load before
rising again. Similar tendencies were seen in the case of PIT. At 23.5 ◦CA
bTDC, the CO emission was the lowest. The joint effects of load in engine
and PIT, as shown in Fig. 4(a) suggest that the least CO was observed
around 10.5 kg engine load and injection advance at 23 ◦CA bTDC. The
change in rated PIT affected the delay in ignition, resulting in poor
combustion. Aside from the late initiation of fuel mixture burning in
cracks, incomplete combustion is also a signicant reason of CO emis-
sion [45]. Because more oxygen had been made available in the cylinder
during combustion, the levels of CO decreased, making the mixture lean.
The fortication of biogas with oxyhydrogen resulted in decreased CO
emissions overall (Fig. 4(b)). Additionally, the presence OH radical in
HHO is a potent oxidizer, leading to higher combustion temperature,
promoting CO to CO
2
conversion in the presence of oxyhydrogen.
Hence, the overall CO emission was decreased.
As indicated in Eq. (5), the correlation between CO and independent
control variables is shown as:
CO =516.96 −29.034L−18.26P−86.1O+0.16L×P+2.04L
×O+2.83P×O+1.23L2+0.31P2−12.13O2
(5)
herein, the letters ’CO’ stand for carbon monoxide, ’PIT’ for the timing
of pilot injection, ’o’ for rate of oxyhydrogen ow, and ’l’ denotes load
Fig. 4. Model of CO; (a) 3-D surface plots for PIT vs load; (b) 3-D surface plots for rate of oxyhydrogen ow vs load; (c) Model predicted vs observed.
P. Sharma et al.
Fuel 341 (2023) 127575
9
on the engine.
Fig. 4(c) shows a comparative analysis between model-predicted and
observed CO levels. The ndings revealed a strong predictive efcacy.
3.2.3. HC emissions model
Based on the design matrix, the observed data of HC emissions was
used to construct a quadratic equation, as depicted in Eq. (6):
HC =505.34 −31.28*L−16.53*P−106.98*O+0.135*L*P
+2.54*L*O+3.48*P*O+1.343*L2+0.262*P2−11.22*O2
(6)
wherein, the abbreviation ’HC’ stand for unburned hydrocarbon emis-
sion, ’p’ represents pilot injection timings, ’o’ rate of oxyhydrogen ow,
and ’l’ denotes load on engine
Fig. 5(a) shows the impact of PIT and load changes on the HC
emissions. A signicant of engine load over HC emissions were
observed. The ANOVA outcomes listed in Table 5 also show that the
second order term of engine load contributed 63.26%, and the same is
reected in the parabolic nature of the surfaces diagrams shown in Fig. 5
(a) and Fig. 5(b). The inuence of engine load, on the other hand,
resulted in a drop in HC emissions only up to 10 kg engine load. On
increasing the engine load, the HC emissions again augmented up to the
full load position. The lowest emission has been observed at a 10 kg load
with 25 ◦CA bTDC as shown in Fig. 5(a). Any alterations to the PIT were
found to change the ignition timing and thus causing combustion delays,
resulting in poor combustion [46,47]. Incomplete combustion owing to
delayed fuel combustion is a major cause of HC emissions [48,49].
Fig. 5. Model of HC; (a) 3-D surface plots for PIT vs load; (b) 3-D surface plots for rate of oxyhydrogen ow vs load; (c) Model predicted vs observed.
P. Sharma et al.
Fuel 341 (2023) 127575
10
HC levels decreased when higher amount of oxygen was available in
the cylinder as a result of oxyhydrogen induction. The presence of
higher-quality fuel and more oxygen particles aids in the decrease of HC
during operation of engine [50]. The interaction impacts of load and rate
of oxyhydrogen ow is depicted in the Fig. 5(b). Enhancing load on
engine with rate of oxyhydrogen ow helped to reduce HC emissions.
Greater HC reduction was observed during incidence of elevated cylin-
der temperature, which stimulates better combustion. The HC emissions
were predicted using Eq. (6). The observed UHC emission levels are
compared to the predicted HC values in Fig. 5(c). The D-optimal based
HC model was shown to be successful in forecasting HC emission levels.
3.3. Statistical evaluation of the models
Three statistical indices R
2
, adjusted R
2
, and predicted R
2
were
employed in the present study to evaluate the developed models. The
coefcient of determination (R
2
) measures the amount of variation
explained by a regression model. A regression model’s R
2
is positive if
the model’s forecast is better than the mean of the given ’y’ values;
otherwise, it is negative. When we add new predictors to a multiple
regression model, R
2
increases or remains constant, even if the new
predictors are unrelated to the target variable and provide no value to
the model’s predicting power. R- square’s disadvantage is reduced with
adjusted R-squared. It increases only if the extra predictor improves the
model’s prediction power. The adjusted R
2
drops as independent and
irrelevant components are added to a regression model. The predicted
R
2
evaluates how well a regression model predicts the responses of new
data. When the model ts the original data but is less capable of pro-
ducing effective predictions for further observations, this value is
higher. The R
2
, adjusted R
2
, and predicted R
2
of all ve models devel-
oped in the present study are shown in Fig. 6. The high values (greater
than 0.9258) of these indices in all cases show the robust forecasting
capability of the RSM-based D-optimal design.
3.4. Desirability-based optimization
A desirability technique was employed in the present research to
optimize independent control parameters (namely engine load, oxyhy-
drogen ow rates, and pilot injection timing) for the best combination of
output parameters (viz. BTE, NO
x
, carbon mono oxide, and HC). Using
Design-Expert commercial software, all inputs were converted to a
dimensionless value of desirability (z) ranging from 0 to 1. The value of
desirability as z =0 implies an unacceptable result, while z =1
represents an exceedingly good response. Depending on the criteria of
the issue, the goal of each result is to minimize or maximize, the range,
goal, and/or equal to. In the present study, all three control factors were
kept within range while the BTE was desired to be maximum. The
exhaust emission indices were dened to be minimum. Table 7 lists the
desirability determined for each variable in the current study.
The implementation of the multi-input/output optimizations tech-
nique employing the desirability technique entails combining many
outputs into a dimensionless entity of performance known as the func-
tion of desirability (FD). Fig. 7(a) displays the desirability values ach-
ieved in this investigation.
The desirability-based optimized engine operating parameters ach-
ieved were 10.88 kg engine load, 24 ◦CA bTDC PIT, and 1.1 lpm rate of
oxyhydrogen ow. The performance output of the study was 24.47%
BTE, 204.32 ppm NO
x
, 113.63 ppm CO, and 103.83 ppm HC emissions,
at the optimized setting.
3.5. Experimental validation
After the optimal parameters of engine operation were determined
using ANOVA and desirability, the model output was veried using lab-
based testing. The engine was operated at an ideal operating congu-
ration, and the performance was recorded. Table 8 displays the optimum
operating parameters, model optimized output, test results, and errors in
the model projected output. The predicted values of the model were
determined to be within 3.75% of the actual value, demonstrating a
reliable modelling strategy.
4. Conclusions
The combustion and emission characteristics of a dual-fuel CI engine
with biogas as the primary fuel, diesel as the pilot-injected fuel, and
oxyhydrogen as a fortifying agent were investigated. The trials were
conducted using an RSM-based D-optimal design. The correlation
function between control variables and output was created using
Fig. 6. Statistical indices.
Table 7
Desirability.
Control factor Response variables
Parameter Engine
load
OFR PIT BTE NO
x
CO HC
Desire In range In range In range Max. Min. Min. Min.
P. Sharma et al.
Fuel 341 (2023) 127575
11
ANOVA. The key ndings of the study were as follows:
o The inclusion of oxyhydrogen enhanced the combustion in biogas-
diesel fuelled engines. Improved fuel economy and reduced
carbon-based emission levels were observed with a marginal in-
crease in NO
x
levels.
o RSM-ANOVA was used to create mathematical expressions (models)
for each outcome. The models were used to forecast output and were
compared to observed results. All of the models showed robust pre-
diction efciency (R
2
greater than 99.21%).
o The optimal engine operating parameters achieved through
desirability-based optimization were 24◦crank angles advance,
10.88 kg engine loading, and 1.1 lpm rate of oxyhydrogen supply.
o Experiment testing corroborated the optimized settings, and all
outcomes were within 3.75% of the model-forecasted output.
Present study has the prospects to be widely utilised in diesel engine-
fueled transportation systems. The economic elements should be
examined as a future scope for the present study.
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Data availability
No data was used for the research described in the article.
Acknowledgements
The authors would like to thank the Deanship of Scientic Research
at Shaqra University for supporting this work.
References
[1] Aghbashlo M, Peng W, Tabatabaei M, Kalogirou SA, Soltanian S, Hosseinzadeh-
Bandbafha H, et al. Machine learning technology in biodiesel research: A review.
Prog Energy Combust Sci 2021;85:100904. https://doi.org/10.1016/j.
pecs.2021.100904.
[2] Fekete H, Kuramochi T, Roelfsema M, den Elzen M, Forsell N, H¨
ohne N, et al.
A review of successful climate change mitigation policies in major emitting
economies and the potential of global replication. Renew Sustain Energy Rev 2021;
137:110602. https://doi.org/10.1016/J.RSER.2020.110602.
[3] Bwapwa JK, Anandraj A, Trois C. Possibilities for conversion of microalgae oil into
aviation fuel: A review. Renew Sustain Energy Rev 2017;80:1345–54. https://doi.
org/10.1016/J.RSER.2017.05.224.
[4] Hansen J, Sato M, et al. Regional climate change and national responsibilities.
Environ Res Lett 2016;11. https://doi.org/10.1088/1748-9326/11/3/034009.
[5] Thiyagarajan Subramanian, Edwin Geo Varuvel, Jesu Martin Leenus, Nagalingam
Beddhannan. Effect of electrochemical conversion of biofuels using ionization
system on CO2 emission mitigation in CI engine along with post-combustion
system. Fuel Processing Technology, 173, 21-29, 2018.
[6] Obaideen K, Abdelkareem MA, Wilberforce T, Elsaid K, Sayed ET, Maghrabie HM,
et al. Biogas role in achievement of the sustainable development goals: Evaluation,
Challenges, and Guidelines. J Taiwan Inst Chem Eng 2022;131:104207. https://
doi.org/10.1016/J.JTICE.2022.104207.
[7] Said Z, Nguyen TH, Sharma P, Li C, Ali HM, Nguyen VN, et al. Multi-attribute
optimization of sustainable aviation fuel production-process from microalgae
source. Fuel 2022;324:124759. https://doi.org/10.1016/J.FUEL.2022.124759.
[8] Mrad N. Maria Paraschiv, Fethi Aloui, Edwin Geo Varuvel, Mohand Tazerout, Sassi
Ben Nasrallah. Liquid hydrocarbon fuels from sh oil industrial residues by
catalytic cracking. Int J Energy Res 2013;37(9):1036–43.
[9] Naresh Kumar A, Sarkar O, Chandrasekhar K, Raj T, Narisetty V, Mohan SV, et al.
Upgrading the value of anaerobic fermentation via renewable chemicals
production: A sustainable integration for circular bioeconomy. Sci Total Environ
2022;806:150312. https://doi.org/10.1016/J.SCITOTENV.2021.150312.
Fig. 7. Desirability values.
Table 8
Optimum operational conditions, results, lab-based assessment output, and model.
Engine load, kg PIT, ◦CA bTDC OFR, lpm BTE, % NO
x
, ppm CO, ppm HC, ppm
10.88 24 1.1 Predicted value 24.47 204.32 113.63 103.83
Test value 25.1 212 111 100
% Error 2.57 3.75 2.31 3.69
P. Sharma et al.
Fuel 341 (2023) 127575
12
[10] Leonzio G. Studies of Mixing Systems in Anaerobic Digesters using CFD and the
Future Applications of Nanotechnologies. Waste Biomass Valoriz 2020;11:
5925–55. https://doi.org/10.1007/s12649-019-00828-1.
[11] Ramos A, Monteiro E, Silva V, Rouboa A. Co-gasication and recent developments
on waste-to-energy conversion: A review. Renew Sustain Energy Rev 2018;81:
380–98. https://doi.org/10.1016/j.rser.2017.07.025.
[12] Praveena V, Martin M, Edwin V, Geo.. Experimental characterization of CI engine
performance, combustion and emission parameters using various metal oxide
nanoemulsion of grapeseed oil methyl ester. J Therm Anal Calorim 2020;139(6):
3441–56.
[13] Varuvel EG, Sonthalia A, Subramanian T, Aloui F. NOx-smoke trade-off
characteristics of minor vegetable oil blends synergy with oxygenate in a
commercial CI engine. Environ Sci Pollut Res 2018;25(35):35715–24.
[14] Said Z, Le DTN, Sharma P, Dang VH, Le HS, Nguyen DT, et al. Optimization of
combustion, performance, and emission characteristics of a dual-fuel diesel engine
powered with microalgae-based biodiesel/diesel blends and oxyhydrogen. Fuel
2022;326:124987. https://doi.org/10.1016/J.FUEL.2022.124987.
[15] Kanimozhi B, Kumar G, Alsehli M, Elfasakhany A, Veeman D, Balaji S, et al. Effects
of oxyhydrogen on the CI engine fueled with the biodiesel blends: A performance,
combustion and emission characteristics study. Int J Hydrogen Energy 2021.
https://doi.org/10.1016/J.IJHYDENE.2021.08.054.
[16] Usman M, Hussain H, Riaz F, Irshad M, Bashir R, Shah MH, et al. Articial neural
network led optimization of oxyhydrogen hybridized diesel operated engine.
Sustainability (Switzerland) 2021;13:1–23. https://doi.org/10.3390/su13169373.
[17] Sharma P, Sahoo BB. An ANFIS-RSM based modeling and multi-objective
optimization of syngas powered dual-fuel engine. Int J Hydrogen Energy 2022;47:
19298–318. https://doi.org/10.1016/j.ijhydene.2022.04.093.
[18] Sharma P, Le MP, Chhillar A, Said Z, Deepanraj B, Cao DN, et al. Using response
surface methodology approach for optimizing performance and emission
parameters of diesel engine powered with ternary blend of Solketal-biodiesel-
diesel. Sustainable Energy Technol Assess 2022;52:102343. https://doi.org/
10.1016/J.SETA.2022.102343.
[19] Sharma P, Said Z, Kumar A, Niˇ
zeti´
c S, Pandey A, Tuan Hoang A, et al. Recent
Advances in Machine Learning Research for Nanouid-Based Heat Transfer in
Renewable Energy System. Energy Fuel 2022. https://doi.org/10.1021/acs.
energyfuels.2c01006.
[20] Sharma P, Sharma AK. Application of response surface methodology for
optimization of fuel injection parameters of a dual fuel engine fuelled with
producer gas-biodiesel blends. Energy Sources Part A 2021;00:1–18. https://doi.
org/10.1080/15567036.2021.1892883.
[21] Paparao J, Murugan S. Dual-fuel diesel engine run with injected pilot biodiesel-
diesel fuel blend with inducted oxy-hydrogen (HHO) gas. Int J Hydrogen Energy
2022;47:17788–807. https://doi.org/10.1016/j.ijhydene.2022.03.235.
[22] Tüccar G. Effect of hydroxy gas enrichment on vibration, noise and combustion
characteristics of a diesel engine fueled with Foeniculum vulgare oil biodiesel and
diesel fuel. Energy Sources Part A 2018;40:1257–65. https://doi.org/10.1080/
15567036.2018.1476622.
[23] Kanimozhi B, Kumar G, Alsehli M, Elfasakhany A, Veeman D, Balaji S, et al. Effects
of oxyhydrogen on the CI engine fueled with the biodiesel blends: A performance,
combustion and emission characteristics study. Int J Hydrogen Energy 2022;47:
37668–76. https://doi.org/10.1016/J.IJHYDENE.2021.08.054.
[24] Kazim AH, Khan MB, Nazir R, Shabbir A, Abbasi MS, Abdul Rab H, et al. Effects of
oxyhydrogen gas induction on the performance of a small-capacity diesel engine.
Sci Prog 2020;103:0036. https://doi.org/10.1177/0036850420921685.
[25] Shajahan MI, Sambandam P, Michael JJ, Abdelmoneam Hussein HM.
Environmental impact of oxyhydrogen addition on high-speed gasoline engine
characteristics. Energy Sources Part A 2020;1–14. https://doi.org/10.1080/
15567036.2020.1812768.
[26] Bhave NA, Gupta MM, Joshi SS. Effect of oxy hydrogen gas addition on
combustion, performance, and emissions of premixed charge compression ignition
engine. Fuel Process Technol 2022;227:107098. https://doi.org/10.1016/J.
FUPROC.2021.107098.
[27] Kushwaha P, Ismail S. Investigation of the effects of oxyhydrogen gas (HHO) ow
rate and injector penetration on the performance of a dual fuel CI engine. Indian
Chem Eng 2022:1–13. https://doi.org/10.1080/00194506.2022.2048096.
[28] Goga G, Singla V, Mahla SK, Chauhan BS, Dhir A, Balasubramaniam D, et al. Effects
of ternary fuel blends (diesel-biodiesel-n-butanol) on emission and performance
characteristics of diesel engine using varying mass ow rates of biogas. Energy
Sources Part A 2021. https://doi.org/10.1080/15567036.2021.1910754.
[29] Nayak SK, Hoang AT, Niˇ
zeti´
c S, Nguyen XP, Le TH. Effects of advanced injection
timing and inducted gaseous fuel on performance, combustion and emission
characteristics of a diesel engine operated in dual-fuel mode. Fuel 2022;310:
122232. https://doi.org/10.1016/J.FUEL.2021.122232.
[30] Sharma A, Singh Y, Ahmad Ansari N, Pal A, Lalhriatpuia S. Experimental
investigation of the behaviour of a DI diesel engine fuelled with biodiesel/diesel
blends having effect of raw biogas at different operating responses. Fuel 2020;279:
118460. https://doi.org/10.1016/J.FUEL.2020.118460.
[31] Sharma PK, Sharma D, Soni SL, Jhalani A, Singh D, Sharma S. Characterization of
the hydroxy fueled compression ignition engine under dual fuel mode:
Experimental and numerical simulation. Int J Hydrogen Energy 2020;45:8067–81.
https://doi.org/10.1016/J.IJHYDENE.2020.01.061.
[32] Sharma P, Sharma AK. AI-based prognostic modeling and performance
optimization of CI engine using biodiesel-diesel blends. Internat J Renewable
Energy Resour 2021;11:701–8.
[33] Nirmala N, Dawn SS. Optimization of Chlorella variabilis. MK039712.1 lipid
transesterication using Response Surface Methodology and analytical
characterization of biodiesel. Renew Energy 2021;179:1663–73. https://doi.org/
10.1016/j.renene.2021.07.123.
[34] Sharma P. Prediction-optimization of the effects of Di-tert butyl peroxide-biodiesel
blends on engine performance and emissions using multi-objective response
surface methodology (MORSM). J Energy Res Technol 2021:1–26. https://doi.org/
10.1115/1.4052237.
[35] Fidaleo M, Miele NA, Armini V, Cavella S. Design space of the formulation process
of a food suspension by D-optimal mixture experiment and functional data
analysis. Food Bioprod Process 2021;127:128–38. https://doi.org/10.1016/J.
FBP.2021.02.007.
[36] Rahman SMA, Sharma P, Said Z. Application of response surface methodology
based D-optimal design for modeling and optimisation of osmotic dehydration of
Zucchini. Digital Chem Eng 2022;4:100039. https://doi.org/10.1016/J.
DCHE.2022.100039.
[37] Muhammad G, Potchamyou Ngatcha AD, Lv Y, Xiong W, El-Badry YA, Asmatulu E,
et al. Enhanced biodiesel production from wet microalgae biomass optimized via
response surface methodology and articial neural network. Renew Energy 2022;
184:753–64. https://doi.org/10.1016/J.RENENE.2021.11.091.
[38] Moreira MO, Balestrassi PP, Paiva AP, Ribeiro PF, Bonatto BD. Design of
experiments using articial neural network ensemble for photovoltaic generation
forecasting. Renew Sustain Energy Rev 2021:135. https://doi.org/10.1016/j.
rser.2020.110450.
[39] Mahla SK, Saeddin Ardebili SM, Sharma H, Dhir A, Goga G, Solmaz H.
Determination and utilization of optimal diesel/n-butanol/biogas derivation for
small utility dual fuel diesel engine. Fuel 2021;289:119913. https://doi.org/
10.1016/j.fuel.2020.119913.
[41] Bhowmik S, Paul A, Panua R, Ghosh SK, Debroy D. Performance-exhaust emission
prediction of diesosenol fueled diesel engine: An ANN coupled MORSM based
optimization 2018;vol. 153. https://doi.org/10.1016/j.energy.2018.04.053.
[42] Das S, Kashyap D, Bora BJ, Kalita P, Kulkarni V. Thermo-economic optimization of
a biogas-diesel dual fuel engine as remote power generating unit using response
surface methodology. Thermal Science and Engineering Progress 2021;24:100935.
https://doi.org/10.1016/J.TSEP.2021.100935.
[43] Elkelawy M, Alm-Eldin Bastawissi H, El Shenawy EA, Taha M, Panchal H,
Sadasivuni KK. Study of performance, combustion, and emissions parameters of DI-
diesel engine fueled with algae biodiesel/diesel/n-pentane blends. Energy Convers
Manage: X 2021;10:100058. https://doi.org/10.1016/J.ECMX.2020.100058.
[44] Iverach D, Basden KS, Kirov NY. Formation of nitric oxide in fuel-lean and fuel-rich
ames. Symp (Int) Combust 1973;14:767–75. https://doi.org/10.1016/S0082-
0784(73)80071-2.
[45] Bora BJ, Dai Tran T, Prasad Shadangi K, Sharma P, Said Z, Kalita P, et al.
Improving combustion and emission characteristics of a biogas/biodiesel-powered
dual-fuel diesel engine through trade-off analysis of operation parameters using
response surface methodology. Sustainable Energy Technol Assess 2022;53:
102455. https://doi.org/10.1016/J.SETA.2022.102455.
[46] Lingesan S, Annamalai K, Parthasarathy M, Ramalingam KM, Dhinesh B,
Lalvani JIJ. Production of Garcinia gummi-gutta Methyl Ester (GGME) as a
potential alternative feedstock for existing unmodied DI diesel engine:
combustion, performance, and emission characteristics. J Test Eval 2018;46(6):
2661–78.
[47] Veza I, Afzal A, Mujtaba MA, Hoang AT, Balasubramanian D, Sekar M, et al.
Review of articial neural networks for gasoline, diesel and homogeneous charge
compression ignition engine. Alex Eng J 2022.
[48] Karthickeyan V, Ashok B, Thiyagarajan S, Nanthagopal K, Geo VE, Dhinesh B.
Comparative analysis on the inuence of antioxidants role with Pistacia khinjuk oil
biodiesel to reduce emission in diesel engine. Heat Mass Transf 2020;56(4):
1275–92.
[49] Bui TT, Balasubramanian D, Hoang AT, Konur O, Nguyen DC, Tran VN.
Characteristics of PM and soot emissions of internal combustion engines running
on biomass-derived DMF biofuel: a review. Energy Sources Part A 2021:1–22.
[50] Jeyakumar N, Narayanasamy B, Balasubramanian D, Viswanathan K.
Characterization and effect of Moringa Oleifera Lam. antioxidant additive on the
storage stability of Jatropha biodiesel. Fuel 2020;281:118614.
Further reading
[40] Sharma P. Articial intelligence-based model prediction of biodiesel-fueled engine
performance and emission characteristics: A comparative evaluation of gene
expression programming and articial neural network. Heat Transfer 2021;50:
5563–87. https://doi.org/10.1002/htj.22138.
P. Sharma et al.