![The historical production measurements of the engine cooling fan The plotting of the histogram with fitting the data to a normal distribution is shown in Fig. 5. The values of mean (µ) and the standard deviation (s) for the normal distribution are: µ = 1429.98[1429.91, 1430.04], and s = 3.1887[3.14511, 3.23352].](https://www.researchgate.net/publication/373523588/figure/fig2/AS:11431281184885716@1693477423371/The-historical-production-measurements-of-the-engine-cooling-fan-The-plotting-of-the_Q320.jpg)
Available via license: CC BY
Content may be subject to copyright.
Engineering technological systems: Reference for Chief Designer at an industrial enterprise
53
Copyright © 2023, Authors. This is an open access article under the Creative Commons CC BY license
PERFORMANCE
OPTIMIZATION OF
RADIATOR ENGINE
PARAMETERS DURING
HARD CONDITIONS
BY CONTROL CHARTS
MONITORING AND
EVALUATING
Ali Fadhil Abduljabbar
Master of Statistics
Department of Medical Laboratory Technologies
Middle Technical University, Kut Technical Institute
Al Za’faraniya, Baghdad, Iraq, 10074
Bashra Kadhim Oleiwi
Doctor of Mechatronics Engineering/Control
and Systems Engineering
Department of Control and Systems Engineering
University of Technology-Iraq
Baghdad, Iraq, 19006
Ahmad H. Sabry
Corresponding author
Doctor of Control and Automation Engineering
Department of Computer Engineering
Al-Nahrain University
Al Jadriyah Bridge, Baghdad, Iraq, 64074
Е-mail: ahs4771384@gmail.com
Recently, engine design and control sys-
tems have been developed using data-driven
modeling techniques to specify the in-cylinder
complicated combustion process. The cooling
fan performance is highly influenced by seve-
ral factors that are determined based on what
is called (DOE) «design of experiments». These
factors include blade tip clearance, pitch angle,
distance from radiator. This work presents
a method to improve a cooling fan performance
of an engine by designing a Six Sigma technique
using Control, Improve, Analyze, Measure, and
Define (CIAMD). First, let’s assess the existing
cooling fan performance and define its problem.
Then, let’s specify the parameters that affect
on fan performance to be optimized. Next, let’s
conduct sensitivity analysis and evaluate manu-
facturing control of the developed cool Fan. The
primary fan does not distribute air enough by the
radiator to maintain the machine cool through-
out hard circumstances. First, the work demon-
strates how to develop an experiment to examine
the influence of three performance elements:
blade pitch angle, blade-tip clearance, and fan
distance from the radiator. In order to improve
the performance of the cooling fan, Box-Behnken
design is adopted for testing quadratic (non-
linear) effects. It then indicates how to predict
optimal quantities for every element, to pro-
duce a technique that makes airflows above the
objective of 1486.6 m3/h when utilizing experi-
mental measurements. Finally, it reveals how to
operate simulations to confirm that this method
creates airflow based on the specifications with
more additional fans manufactured performance
of 99.999 %. The results of S and X-bar control
charts indicate that the manufacturing process
is statistically under control
Keywords: optimization, Six Sigma techni-
que, control chart, Box-Behnken, cooling fan
UDC 621
DOI: 10.15587/1729-4061.2023.285698
How to Cite: Abduljabbar, A. F., Oleiwi, B. K., Sabr y, A. H. (2023). Perfor mance optimization of radiator engine parameters
during ha rd conditions by co ntrol charts monitoring and evaluating. Ea stern-European Journal of Enterpr ise Technologies,
4 (1 (124)), 53–59. doi: ht tps://doi.or g/10.15587/1729-4061.2023.285698
Received date 17. 05.2023
Accepted date 04.08.2023
Publi shed date 31.08.2023
1. Introduction
An essential component of the vehicle cooling system is
the engine cooling fan. The tip of the blade, which performs
the majority of the fan’s work, has a significant impact on
the aerodynamic performance of the fan. Static pressure
and shaft power are crucial indications to evaluate its aero-
dynamic performance. It’s critical that the cooling system
operate with the least amount of input energy possible be-
cause it takes a portion of the engine’s power. Increased fuel
efficiency from a cooling system upgrade can help engines
satisfy legal mobility requirements. A diagram showing the
airflow components of an engine is shown in Fig. 1.
Recently, engine design and control systems have been
developed using data-driven modeling techniques to specify
the in-cylinder complicated combustion process [1]. Because
uncertainties or stochastic elements would significantly affect
the lower system’s dependability and decoupling state of the
system, robustness optimization to increase the reliability of an
engine mounting system (EMS) is required in the design [2].
Bill Smith introduced a quality-control system known
as Six Sigma in 1986 [3]. It makes use of data-driven exami-
nation to decrease errors or flaws in a business or corporate
operation. The Six Sigma model is a method for working
more accurately and quickly. Six Sigma applies to various
sectors such as a quality approach to robust optimization [4],
for HPLC-UV method optimization [5], Performance opti-
mization of the retail Industry [6], for production process
optimization in a paper production company [7], Improving
online quality control [8], for robust design optimization [9],
Multi-Point Dieless Forming Process [10], Sorbitol pro-
duction optimization in B2B industry [11], in the mould
Eastern-European Journal of Enterprise Technologies ISSN 1729-3774 4/1 ( 124 ) 2023
54
industry [12], for Inherent Issues in Production Shop Floor
Management [13], and others. Therefore, by removing and
locating the root causes of faults and reducing manufacturing
variance, Six Sigma procedures aim to increase manufac-
turing quality. Six Sigma does this by employing Six Sigma
statistical, using experimental and specialist quality control
management techniques.
Therefore, optimizing fan’s parameters has a significant
impact on the aerodynamic performance of the fan and is
essential to improve the performance of radiator engine pa-
rameters during hard conditions.
2. Literature review and problem statements
Using fan’s associated elements such as shroud pro-
vide space for the fan, which in turn distributes air across
the radiator as is required for efficient cooling system of
various engines. The study [14] discussed the use of Six
Sigma technique on the fan shroud as one of the signifi cant
cooling components across the heat exchangers of an engine
that assists to achieve optimum airflow, where the key chal-
lenges are on designing such a shroud that meets harshness,
vibration and noise requirements but with no compromising
on airflow goals. This study used A CFD simulation. Al-
though this study proposed two operation modes pumping
and flexible to optimize parameters such as structural
embossing, number of ribs, and wall thickness as control
factors, the method was not easy to understand with com-
plexity. Under the double heating effects of high-power
electronic apparatus and aerodynamic, the power demand
and heat sink of superior high-speed aircraft have been ris-
ing exponentially, which seriously restricts the aircraft per-
formance. A thermal/power management system (TPMS)
was used to improve system cooling [15]. However, the
optimization variables were not consisting of the fan duct
heat exchanger structure size, cooling air flow rate, and
compressor outlet temperature.
The high-pressure turbine cooling of a fan-shaped hole
was discussed for performance optimization by [16]. A deep
learning architecture was designed to estimate the cooling
airflow rate for the cooling-hole location. Although the
model optimization was conducted to recover the tempera-
ture uniformity and the film cooling effectiveness on the
vane surface with several fan-shaped holes, their results
were only applicable on the film cool-
ing model. The authors of study [17]
also discussed the cooling issue of en-
gines of internal combustion type to
remove waste heat using the radiator
fan(s), radiator, thermostat valve, and
water pump to reject heat to the lo-
cal environment and flow cooling fluid
throughout the engine block. Although
this study utilized compound matrix
of radiator fan to reduce energy usage,
the forced convection heat transmit
process and the mathematical model
was missing. Implementation of cooling
system on diesel engine to predict sys-
tem parameters based on deep learning
network based algorithm was presented
by [18]. However, their experimental
results showed no significant difference
compared to the traditional neural net-
work one. The optimization technique
of blade tip parameters and a calcula-
tion technique of aerodynamic perfor-
mance of the fan such as tip arch height,
tip length, and tip mounting angle were
studied by [19].
Although the study analyzed the feasibility of the optimi-
zation method, blade velocity vector diagram, blade pressure
diagram and fan performance curve, the optimization and
analysis of fan parameters were not accurate enough for the
engine cooling fan design. These issues have been discussed
by [20], where a vehicle cooling system with front-end intake
air volume was found to be affects directly on the aerody-
namic resistance and heat dissipation performance of engine
compartment, but the computational fluid dynamics (CFD)
used method was not able to optimize the fan speed.
All this allows asserting that it is expedient to conduct
a study on improving a cooling fan performance of an engine
when the primary fan does not distribute air enough by
the radiator to maintain the machine cool throughout hard
circumstances.
3. The aim and objectives of research
The aim of research is to determine the optimal operating
parameters of the radiator engine under hard conditions.
This will make it possible to maintain the machine cool
throughout hard circumstances.
To achieve this aim, the following objectives are accom-
plished:
– to assess the existing cooling fan performance and de-
fine its problem;
– to specify the parameters that affect on fan performance
to be optimized;
– to conduct sensitivity analysis and evaluate manufac-
turing Control of the developed cool Fan.
Cooling
reservoir
Radiator
Fan
Water
pump
Thermostat
Fig. 1. Airflow components of an engine
Engineering technological systems: Reference for Chief Designer at an industrial enterprise
55
4. Methods and materials
4. 1. Object and research hypothesis
This work discusses the radiator circulating flow of
cooling air issue of an engine fan model to maintain the en-
gine cool during variation conditions including hot weather
and stop-and-go traffic. Let’s assume that airflow must be
1486.63 m3/h at least, to maintain engine cooling during
such hard circumstances.
A diagram showing the main me thodology steps is demon-
strated in Fig. 2.
Assess Cooling Fan Performance
Control Manufacturing of the Improved Cooling Fan
Improve the Cooling Fan Performance
Optimize Factor Settings
Sensitivity Analysis
Determine Factors That Affect Fan Performance
End
Start
Define the Problem
Fig. 2. The main methodology steps
Therefore, it is required to assess the existing model
and expand an alternative one that is capable of reaching
the goal of circulating flow of air.
4. 2. Methodology steps
In order to evaluate the performance of the engine
cooling fan, let’s initially load a sample of historical pro-
duction measurements includes 5000 of observations of
the performance of the engine cooling fan. The cooling
fan performance is influenced by several factors that are
determined based on what is called (DOE) «design of
experiments».
These factors include the followings:
– blade tip clearance;
– pitch angle;
– distance from radiator.
Let’s assume that these factors can control and modify.
The airflow rate is the response of the cooling fan (m3/h).
Since, the airflow is a fluid process and has a nonlinear
behavior as usual; let’s adopt a surface design based re-
sponse to predict the nonlinearity relations amongst these
factors. Box-Behnken Design, which is introduced in [21],
is used to create the experimental attempts in normali-
zed (coded) variable range [–1, 0, +1].
The Coded-Value matrix is a (15×3) = [–1–10; –110;
1–10; 110; –10–1; –101; 10–1; 101; 0–1–1; 0–11…
Where the column to the left represents the distance
values of the fan from the radiator, the pitch angle values are
shown in the second column, and the last column includes
the blade clearance tip. Let’s assume that the effects of the
parameters on the following maximum and minimum values
are required to be tested:
– blade tip clearance: 2.5 to 5 cm;
– pitch angle: 14 to 36 degrees;
– distance from radiator: 2.54 to 3.8 cm.
The runs order is randomized, then let’s change the
coded model quantities into actual-world values, next let’s
execute the experimentation in a particular order. In order
to improve the performance of the cooling fan, Box-Behnken
model has been adopted to inspect the quadratic (non-
linear) effect. The nonlinear model can be formed in terms
of the airflow rate (AF) by:
ADPCDP
DC PC DP
F
=+∗+ ∗+ ∗+ ∗∗+
+∗∗+∗∗+ ∗+∗+∗
ββ βββ
βββββ
01 234
567
2
8
2
9
CC2
,
(1)
where D is the distance, P is the pitch angle, and C is the
clearance, while βi represents the formula coefficients that
their normalized magnitudes in a bar chart representation.
To verify the optimal obtained factor settings, it is pos-
sible to make the task of finding the optimum factor settings
automatic by considering the Problem-Based Optimizing
method as demonstrated in the flow diagram of Fig. 3.
The required airflow is met by the upgraded cooling fan
design. Based on the design-tunable factors, the derived
model can accurately predict the fan performance. However,
let’s conduct a sensitivity analysis to make sure that the per-
formance of the fan is resilient to variation in manufacture
and setting up. For sensitivity examination, let’s assume that
Define optimization problem for maximization
Create the objective function to convert
the predictor matrix to a design matrix.
Multiply the result by the model coefficient estimates.
Create an optimization variable bounded between –1 and 1, and
has three components, which represent the three factors.
Convert the objective function to an optimization expression
Set the initial point to be the center of the design of the
experimental test matrix, meaning the vector [0 0 0].
For the problem-based approach, the initial point must be
a structure with the variable name as the name field.
Find the optimal design.
Solving problem using (fmincon) MATLAB function.
Convert the results to real-world units.
End
Start
Fig. 3. The main steps
of Problem-Based Optimizing method
Eastern-European Journal of Enterprise Technologies ISSN 1729-3774 4/1 ( 124 ) 2023
56
the manufacturing uncertainties are as listed in Table 1 based
on historical experience.
Table 1
List of the manufacturing uncertainties
based on historical records
Coded Values Real Values Factor description
–2.5+/–0.25 cm 2.5+/–0.25 cm Blade tip clearance
0.227+/–0.028 degrees 27.3+/–0.25 degrees Blade pitch angle
2.5+/–0.20 cm 2.5+/–0.1 cm Distance from radiator
It is crucial to establish that these variances in com-
ponents will allow for the maintenance of a sturdy design
focusing on the desired airflow. A defect rate of just 3.41
per 1,000,000 fans is the goal of the Six Sigma philosophy.
In other words, the fan must consistently reach the aim of
1486.63 m3/h airflow.
A Monte Carlo simulation can be used to verify the de-
sign by creating 5,000 arbitrary numbers through the given
tolerance for the three parameters. Let’s also incorporate
a noise variable that is inversely correlated with the fitted
model’s RMS error, or noise in the data.
To evaluate the manufacturing control of the developed
cool fan, control charts can be used to track and assess the
creation and installation of the new fan. Analyze the new
cooling fan’s first 30 days of production. At first, five cooling
fans were made each day. Let’s initially load the test data
from the new process. Then, plotted S and X-bar charts.
Next, the data is reshaped to a daily representation.
5. Results of the results of the proposed
optimization process
5. 1. Results of the performance assessment of the cool-
ing fan and define its problem
A Plotting of the historical production measurements
of the engine cooling fan is shown in Fig. 4.
Fig. 4. The historical production measurements
of the engine cooling fan
The plotting of the histogram with fitting the data to
a normal distribution is shown in Fig. 5.
The values of mean (µ) and the standard deviation (s) for
the normal distribution are: µ = 1429.98[1429.91, 1430.04],
and s = 3.1887[3.14511, 3.23352].
Fig. 5. The histogram based on normal distribution fit
for the measurements of the engine cooling fan
The plot fits the measured data into a normal distribu-
tion to estimate the data parameters. The predicted 95 %
confidence interval of the mean speed of the fan airflow
is (1429.91, 1430) and the mean value of the airflow speed
is 1429.976 m3/h. This prediction provides clear insight that
the current airflow value of the fan is not reach to the target
value 1486.63 m3/h. Therefore, it is required to develop
a design for the fan to fulfill the goal airflow. This algorithm
produces the response and design parameter values that are
listed in Table 2.
Table 2
Design parameter values
Run_Number Airflow Clearance Pitch Distance
6 1422 0 –1 –1
3 1468 0 1 –1
11 1408.5 0 –1 1
7 1454.3 0 1 1
14 1495 –1 0 –1
8 1493.4 1 0 –1
5 1481.5 –1 0 1
15 1484.9 1 0 1
1 1417 –1 –1 0
2 1415.3 1 –1 0
4 1461 –1 1 0
13 1459.4 1 1 0
9 1485 0 0 0
10 1488 0 0 0
12 1486.6 0 0 0
According to the mode design test outcomes, the chang-
ing factors values are directly affect airflow rate. Further-
more, it is unclear how resilient the design is to changes in
the factors. Therefore, using the existing experimental data,
let’s built a model to optimize the settings of its factors.
5. 2. Results of specifying parameters that affect on fan
performance to be optimized
The formula coefficients that their normalized magni-
tudes in a bar chart representation are shown in Fig. 6.
Engineering technological systems: Reference for Chief Designer at an industrial enterprise
57
Fig. 6. Equation (AF) coefficients’ normalized magnitudes
in a bar chart representation
The dominant factors in the bar chart representation are
the Pitch2 and Pitch and the association between one out-
put variable and multiple input variables is represented by
creating a surface response plot. Let’s employ the MATLAB
function (plotSlice) to produce the model surface response
plots as shown in Fig. 7.
The results here demonstrates that the relation bet-
ween the pitch and airflow has a nonlinear form, and the
dashed blue bound lines show the result the other factors
over the airflow.
Let’s perform extra investigation to confirm that
a 27.27-degree pitch angle is the ideal value because pitch
angle has a major impact on airflow. The R square value of
the fitting model is found to be 0.9963, which show that
the nonlinear/quadratic design clarifies the well influence
of airflow by pitch angle.
The result of plotting the airflow with pitch angle and
developed proposed fitting Quadratic model is shown
in Fig. 8.
With a simple substitution, the pitch angle value that
equivalents to the highest airflow was found to be 27 degree,
and the further analysis supports the conclusion that the
ideal pitch angle is 27.3 degrees.
Fig. 8. The measured/tested data against airflow
with the developed fitting model
5. 3. Results of sensitivity analysis and manufacturing
Control evaluation
A histogram data representation is used to evaluate the
model deviation for the predicted airflow, and a normal dis-
tribution fitting of the data is used to estimate the standard
deviation and the mean. Fig. 9 shows Monte Carlo model rep-
resentation of the data includ-
ing the related mean and stan-
dard deviation.
The S and X-bar control
charts are shown in Fig. 10.
The predicted value of 1.73
and the Cp value of 1.75 are
extremely similar. The Cp va-
lue is greater than the Cpk
value, which is 1.66. Merely
the Cpk value lower than 1.35,
which shows that the me-
thod drastically deviated in
the direction of the limits of
the process, raises suspicions,
though. The procedure opera-
tes fine within the parameters
and delivers the desired air-
flow 1486.6 (m3/h) more than
99.999 % of the time.
Fig. 7. The parameters of the model surface response plots
Fig. 9. Monte Carlo simulation representation of the data
with the related mean and standard deviation
Eastern-European Journal of Enterprise Technologies ISSN 1729-3774 4/1 ( 124 ) 2023
58
Fig. 10. S and X-bar control charts
6. Discussion of the results of optimization
process and evaluation for the radiator
engine parameters
Based on the historical production measurements of
the engine cooling fan shown in Fig. 4, the majority values
are clearly fall inside the range of about 13.59 m3/h and
the data is centered on 1430.57 m3/h. However, the plot
does not reveal a lot about the original data distribution.
The histogram result that is based on normal distribution
fit for the measurements provides clear insight that the
current airflow value of the fan is not reach to the target
value (1486.63 m3/h). Therefore, it is required to develop
a design for the fan to fulfill the goal airflow. This algo-
rithm produces the response and design parameter values
that are listed in Table 2. According to the mode design
test outcomes, the changing factors values are directly
affect airflow rate. In addition, there is four experiment
attempts exceed or meet the goal airflow rate with a value
of 1486.63 m3/h (runs 14, 12, 4, and 2). Yet, which of these
runs, if any, is the best is unclear. Furthermore, it is unclear
how resilient the design is to changes in the factors. There-
fore, using the existing experimental data, a model to opti-
mize the settings of its factors is built.
As per Equation (AF), which is plotted in Fig. 6, the dom-
inant factors in the bar chart representation are the Pitch2
and Pitch and the association between one output variable
and multiple input variables is represented by creating
a surface response plot. The result of applying Problem-Based
Optimizing method shows the following indicators:
1. When the constraints are satisfied and the objective
function is non-reducing in practicable directions inside
the optimality tolerance value, the optimization is said to
be complete.
2. Finding a local minimum that complies with the re-
strictions.
3. A lower value for the objective function that has fea-
sible point.
4. The optimum values of the design factors were 1499
for the airflow, 2.5 for the clearance, 27.2747 for the pitch
and 2.5 for the distance.
According to the optimization outcome, the developed
fan should be installed 2.5 cm away from the radiator, with
a pitch angle of 27.3, and with 2.5 cm between the fan blade
tips and the shroud. The result of the Monte Carlo simu-
lation (Fig. 9) seems promising, where it shows that the
airflow is better than 1486.63 m3/h for the majority of data
and average airflow is 1498.528 m3/h. The probability of
1486.63 m3/h or less of the airflow can be found using the
MATLAB function (cdf (1486.63)), which was 1.454e-07.
Then, computing of (1–1.454e-07)*100 = 99.999 indicates
that the airflow of at least 1486.63 m3/h appears to be at-
tained by the design 99.99 % of the time. Therefore, the
process capability factors can be estimated using the simu-
lation results, where Cpk was with a value of 1.709768, Cpu
was 1.804268, Cpl was 1.709768, Cp was 1.757018, Pu was
3.1022444e-08, Pl was 1.45407686e-07, P was 0.99999982,
s was 1.42286513, and µ was 8.82298307e+02.
Cp is valued at 1.75. When Cp is equal to or greater than
1.62, a process is deemed to be of excellent quality. The pro-
cess is centered because the Cpk value resembles the Cp value.
Now it is possible to put this plan into action and Check to
make sure the cooling fan performs to a high standard and
to confirm the design process.
The results of S and X-bar control charts, shown in
Fig. 10, indicate that the manufacturing process is statis-
tically under control, as shown by non-random patterns in
the data or the absence of violations of control limits over
time. For further process assessment, let’s also perform a data
capacity analysis, which generates the values such that Cpk
was 1.663, Cpu was 1.8479, Cpl was 1.6635, Cp was 1.75575,
Pu was 1.47906e-08, Pl was 3.00893e-07, P was 0.9999, s was
1.423887, and µ was 8.821061e+02.
The presented approach was capable to specify and op-
timize the parameters that affect engine fan performance as
well as conduct sensitivity analysis and evaluate manufactur-
ing control of the developed cool fan, which is not presented
previously in the literature in such an application.
The only limitation of this approach is that the optimum
values of the design factors are obtained according to the
assumed airflow particular value to maintain engine cooling
during such hard circumstances. The use of Box-Behnken
model to test the quadratic (nonlinear) effects was time
consume. This disadvantage may require more analytical at-
tempts to select the appropriate nonlinear model for solving
the effects of the physical system nonlinearity.
7. Conclusions
1. The assessment for the existing cooling fan perfor-
mance and definition of its problem has been performed by
analyzing the historical measurements and its histogram
based on normal distribution fit for the engine cooling fan.
The predicted 95 % confidence interval provides clear insight
that the current airflow value of the fan was not reach to the
target value 1486.63 m3/h.
2. The cooling fan parameters (Blade tip clearance, Blade
pitch angle, Distance from the radiator) have been identified
and shown their affect. This method created airflow based on
the specifications with more additional fans manufactured
performance of 99.999 %.
3. The proposed approach analyzed the cooling fan sen-
sitivity and evaluated manufacturing control to show that
the process operates fine under the optimized parameters
and delivers the desired airflow. Furthermore, the predicted
value (1.73) and the value due to the evaluation process
of Cp (1.75) were extremely similar, where the process is sup-
posed high quality when Cp is equal to or greater than 1.6.
Engineering technological systems: Reference for Chief Designer at an industrial enterprise
59
Conflict of interest
The authors affirm that they do not have any conflicts of
interest with this research financial, authorship, personal, or
otherwise that would have
Financing
The study was performed without financial support.
Data availability
Data will be made available on reasonable request
Acknowledgments
All authors are acknowledging the Middle Technical
University and University of Technology-Iraq for their assis-
tance and support.
References
1. Kaleli, A., Akola , H. . (2021). The design and development of a diesel engine electromechanical EGR cooling system based on machine
learning-genetic algorithm prediction models to reduce emission and fuel consumption. Proceedings of the Institution of Mechanical
Engineers, Part C: Journal of Mechanical Engineering Science, 236 (3), 1888–1902. doi: https://doi.org/10.1177/09544062211020045
2. Wu, J., Liu, X., Shan, Y., He, T. (2018). Robustness optimization of engine mounting system based on Six Sigma and torque roll axis
decoupling method. Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, 233 (4),
1047–1060. doi: https://doi.org/10.1177/0954407018755247
3. Nejlaoui, M., Alghafis, A., Sadig, H. (2022). Six sigma robust multi-objective design optimization of flat plate collector system under
uncertain design parameters. Energy, 239, 121883. doi: https://doi.org/10.1016/j.energy.2021.121883
4. Xiao, S., Li, Y., Rotaru, M., Sykulski, J. K. (2015). Six Sigma Quality Approach to Robust Optimization. IEEE Transactions on
Magnetics, 51 (3), 1–4. doi: https://doi.org/10.1109/tmag.2014.2360435
5. Ibrahim, A. M., Hendawy, H. A. M., Hassan, W. S., Shalaby, A., El-sayed, H. M. (2019). Six Sigma quality approach for HPLC-UV
method optimization. Microchemical Journal, 144, 303–308. doi: https://doi.org/10.1016/j.microc.2018.09.023
6. Madhani, P. M. (2020). Performance optimisation of retail Industry: Lean six sigma approach. ASBM Journal of Management,
13 (1), 74–91. Available at: https://ssrn.com/abstract=4002475
7. Adeodu, A., Kanakana-Katumba, M. G., Rendani, M. (2021). Implementation of Lean Six Sigma for production process optimization in
a paper production company. Journal of Industrial Engineering and Management, 14 (3), 661. doi: https://doi.org/10.3926/jiem.3479
8. Abualsauod, E. H., Othman, A. M. (2019). Improving Online Quality Control in the Six Sigma Methodology Using a Multiobjective Op-
timization Approach. Industrial Engineering & Management Systems, 18 (2), 195–209. doi: https://doi.org/10.7232/iems.2019.18.2.195
9. Shimoyama, K., Oyama, A., Fujii, K. (2008). Development of Multi-Objective Six Sigma Approach for Robust Design Optimization.
Journal of Aerospace Computing, Information, and Communication, 5 (8), 215–233. doi: https://doi.org/10.2514/1.30310
10. Abebe, M., Yoon, J., Kang, B.-S. (2020). Multi-Objective Six-Sigma Approach for Robust Optimization of Multi-Point Dieless
Forming Process. International Journal of Precision Engineering and Manufacturing, 21 (10), 1791–1806. doi: https://doi.org/
10.1007/s12541-020-00373-1
11. Pariasa, I. I., Anam, M. A. S., Hardana, A. E. (2021). Sorbitol production optimization in B2B industry with six sigma approach.
IOP Conference Series: Earth and Environmental Science, 733 (1), 012056. doi: https://doi.org/10.1088/1755-1315/733/1/012056
12. Pereira, A. M. H., Silva, M. R., Domingues, M. A. G., S , J. C. (2019). Lean Six Sigma Approach to Improve the Production Process
in the Mould Industry: a Case Study. Quality Innovation Prosperity, 23 (3), 103. doi: https://doi.org/10.12776/qip.v23i3.1334
13. Tripathi, V., Chattopadhyaya, S., Mukhopadhyay, A. K., Sharma, S., Li, C., Singh, S., Saleem, W. et al. (2022). Recent Progression
Developments on Process Optimization Approach for Inherent Issues in Production Shop Floor Management for Industry 4.0.
Processes, 10 (8), 1587. doi: https://doi.org/10.3390/pr10081587
14. Konikineni, P., Sundaram, V., Sathish, K., Thirukkotti, S. (2016). Optimization Solutions for Fan Shroud. SAE Technical Paper
Series. doi: https://doi.org/10.4271/2016-01-1393
15. A, R., Pang, L., Jiang, X., Qi, B., Shi, Y. (2021). Analysis and comparison of potential power and thermal management systems for high-speed
aircraft with an optimization method. Energy and Built Environment, 2 (1), 13–20. doi: https://doi.org/10.1016/j.enbenv.2020.06.006
16. Jun, S., Rhee, D.-H., Kang, Y. S., Chung, H., Kim, J.-H. (2022). Application of the Source Term Method and Fan-Shaped Hole for
Cooling Performance Improvement in a High-Pressure Turbine. Energies, 15 (19), 6943. doi: https://doi.org/10.3390/en15196943
17. Wang, T., Jagarwal, A., Wagner, J. R., Fadel, G. (2015). Optimization of an automotive radiator fan array operation to reduce power
consumption. IEEE/ASME Transactions on Mechatronics, 20 (5), 2359–2369. doi: https://doi.org/10.1109/tmech.2014.2377655
18. Akola , H. ., Kaleli, A., Bakirci, K. (2020). Design and implementation of an autonomous EGR cooling system using deep neural
network prediction to reduce NOx emission and fuel consumption of diesel engine. Neural Computing and Applications, 33 (5),
1655–1670. doi: https://doi.org/10.1007/s00521-020-05104-1
19. Wan, L., Duan, L., Guo, Q., Duan, Y., Yu, N., Shangguan, W. (2019). Optimization of Engine Cooling Fan Tip Parameters. Huanan
Ligong Daxue Xuebao/Journal of South China University of Technology (Natural Science).
20. Liu, C., Zhang, R., Duan, M., Liu, K., Wang, W. (2019). A Study on the Matching of Air Inlet Grille Angle and Cooling Fan Speed
in a Car. Automotive Engineering, 4, 388–394. doi: https://doi.org/10.19562/j.chinasae.qcgc.2019.04.005
21. Ferreira, S. L. C., Bruns, R. E., Ferreira, H. S., Matos, G. D., David, J. M., Brand o, G. C. et al. (2007). Box-Behnken design: An alternative
for the optimization of analytical methods. Analytica Chimica Acta, 597 (2), 179–186. doi: https://doi.org/10.1016/j.aca.2007.07.011
