CFD Letters 15, Issue 7 (2023) 1-13

CFD Letters
Journal homepage:
https://semarakilmu.com.my/journals/index.php/CFD_Letters/index
ISSN: 2180-1363

Application of Box- Behnken Design with Response Surface Methodology

to Analyse Friction Characteristics for Corrugated Pipe via CFD
Azraf Azman1,2, Mohd Zamri Yusoff1,*, Azfarizal Mukhtar1, Prem Gunnasegaran1, Ng Khai Ching3,
Ahmad Shah Hizam Md Yasir4

1 College of Engineering, Putrajaya Campus, Universiti Tenaga Nasional (UNITEN), Jalan IKRAM-UNITEN 43000 Kajang Selangor, Malaysia
2
Agensi Nuklear Malaysia, Bangi, 43000 Kajang, Selangor, Malaysia
3 University of Nottingham Malaysia Campus, Jalan Broga, 43500 Semenyih, Selangor, Malaysia
4 Rabdan Academy, 65, Al Inshirah, Al Sa’adah, Abu Dhabi, 22401, UAE

ARTICLE INFO ABSTRACT

Article history: Due to the fact that modern technologies are getting smaller and more compact and
Received 6 September 2022 are anticipated to perform better, there has been an increase in interest in heat
Received in revised form 8 October 2022 transfer enhancement employing hybrid nanofluids in tubes and channels in recent
Accepted 7 November 2022 years. This study seeks to establish an outward saw tooth corrugated wall model-based
Available online 1 July 2023
predictive friction characteristic for internal tube flow. Computational Fluid Dynamics
(CFD) was used to simulate the model numerically. The response behaviour study was
then conducted utilising the Design of Experiment (DOE) and the Response Surface
Methodology (RSM). The established surrogate model has led to the consideration of
RSM. In this study, the DOE was executed by utilising the Box-Behnken Design (BBD)
method with two-level factorial by considering three parametric factors which are: (a)
10000 to 30000 axial Reynolds number, Re, (b) 1mm to 4mm wave amplitude, a, and
(c) 5mm to 20mm wavelength, lw. The results showed that the expected response
surface values are consistent with the CFD values and that the predictive model is
therefore reliable. The R-squared (R2) value is 98.25%, indicating that the model can
predict new observations. The wavelength and wave amplitude show significant
Keywords: factors influencing the friction factor which is -0.11235 and 0.14861 respectively and
Hybrid Nanofluids; Corrugated Tube; this is based on the normal plot of the effect from the regression model from RSM.
CFD; Response Surface Method; Friction These results provide data for estimating the geometric characteristics of tube
Factor corrugated wall.

1. Introduction

Various techniques have been developed over the past few decades to improve the heat transfer
rate or efficiency of traditional heat transfer devices in a variety of engineering applications, such as
radiators in automobiles [1], micro channel in micromachining technology [2], cooling electronic
components, and refrigeration systems. These strategies proposed by early scientists fall into two
groups: active technique and passive technique [3]. Active procedures often necessitate the use of

*
Corresponding author.
E-mail address: [email protected] (Mohd Zamri Yusoff)

https://doi.org/10.37934/cfdl.15.7.113

1
CFD Letters
Volume 15, Issue 7 (2023) 1-13

an external power source and may involve mechanical components, whereas passive approaches are
independent of external sources. As a result, passive approaches have been actively investigated
because they can reduce existing system operating costs while maintaining excellent reliability [3].
Numerous passive approaches were utilised, including the incorporation of fins, well-engineered
surface texturing [4] and micro channels. In addition, the application of innovative coolants with
improved thermophysical properties has attracted attention as they appear to be an alternative for
conventional heat transfer fluids with low thermal conductivity, such as water, oil, ethylene glycol,
etc [1]. There was evidence that scientists were attempting to include high thermal conductivity
millimetre and micrometre scaled particles in conventional coolants to create a blend with increased
thermal characteristics. In addition to causing a high-pressure drop, obstructing the flow, and even
corroding the heat exchanger system's components, this strategy was deemed less convincing.
Choi et al., [5] report that the development of powder manufacturing techniques has led to the
synthesis of nanosized particles. The nanofluids are colloidal suspensions created by dispersing solid
nanoparticles (10 to 100 nm) in a base fluid to improve their transport properties. The addition of
solid nanoparticles can also enhance the thermal properties of base fluids [6-8]. These materials have
unique optical, electrical, and chemical properties. Nanofluid is a novel heat transfer fluid created by
dispersing nanometer-sized solid particles in conventional heat transfer fluids such as water or
ethylene glycol to improve thermal conductivity and, consequently, heat transfer performance. It is
known that the effectiveness of heat transfer enhancement is dependent on variables such as the
number of dispersed particles, material type, particle shape, particle volume fraction, and so on.
The primary goal of synthesising hybrid nanofluids is to obtain the properties of their constituent
materials. It is not possible for a single material to possess all the desirable properties required for a
specific application; it may have either good thermal properties or good rheological properties. In
many practical applications, however, it is necessary to compromise between multiple properties,
and this is where hybrid nanofluids come into play. Due to the synergistic effect, the hybrid nanofluid
is anticipated to have a higher thermal conductivity than individual nanofluid [9-12]. Chein and
Chuang [13] found in their investigation that a CuO-H2O nanofluid suspension may absorb more heat
and result in a lower wall temperature than pure liquid. The penalty was a small pressure drop
increase. Nonetheless, a study demonstrates that by distributing composite nanoparticles in the base
fluid (hybrid nanofluid), heat transfer performance is improved compared to nanofluid containing
single nanoparticles [12].
Response Surface Methodology (RSM), however, has been used to increase the design's
effectiveness. The objective of RSM is to discover and study the quantitative assessment of numerous
design parameters that affect the thermal performance of the corrugated tube [6,8,14,15]. This study
seeks to develop a statistical data-based metamodel method that can be used to examine the
association between input variables (factors) and output variables (response) in order to discover the
friction factor utilising the RSM. Similarly, a numerical study on "saw tooth" corrugated wall geometry
turbulent-forced convective heat transfer of Al2O3-CuO/water (Alumina/Copper Oxide-water hybrid
nanofluid with the concentration of 0.05% of 80% of Al2O3 to 20% of CuO (80:20) ratio using CFD
and as most flows in engineering applications such as corrugated pipe are turbulent flow.

2. Methodology
2.1 Physical Model and Assumptions

The two-dimensional corrugated circular pipes used in the simulations are cylindrical pipes with
periodically distributed diametrically symmetric roughness on the wall. According to Kaood et al.,
[16] inward or outward corrugated shapes have higher performance evaluation criterion compared

2
CFD Letters
Volume 15, Issue 7 (2023) 1-13

to smooth tubes. A generic, schematic representation of the corrugated pipe is shown in Figure. 1.
The heat transfer was performed numerically using a corrugated pipe with an inner diameter of 10
mm. The total lengths of the hydrodynamic development section and the test section provide a
maximum length-to-diameter ratio (LT/d) of 80. The length-to-diameter ratio of the investigated
region is L1/d=20, and the length-to-diameter ratio of an upstream section is L/d=40 to ensure a fully
developed flow in the test section ((L/d)>10). The length-to-diameter ratio of the downstream
section (exit section) is L2/d=20 which is used to prevent any reversed flow through the
computational domain. The roughness parameters are determined by the rib height (a), and
wavelength (Lw). In this work the triangular roughness parameters are expressed in the form of
dimensionless roughness parameters: the rib pitch-to-tube diameter ratio (Lw/d) was in the range of
0.5–2.0 and the rib height-to-tube diameter ratio (a/d) was in the range of 0.1–0.4, The following
assumptions are adopted in this work: (i) the ribs and the grooves are periodically distributed in the
axial direction; (ii) the flow is steady, fully developed, turbulent, and two-dimensional; (iii) the tube
material is homogeneous and isotropic; and (iv) the thermal conductivity of the tube wall material
does not change with temperature.

2.2 Governing Equations

The geometry under study is two-dimensional in a rectangular plane. The flow is steady,
incompressible, forced turbulent convection through the straight corrugated circular pipe and
assumed to be axisymmetric along the horizontal plane parallel to the x-axis. The flow and thermal
fields are described by the two-dimensional, steady continuity, momentum, energy equations and
constant thermophysical properties [10,11].
The continuity equation is:

𝜕
(𝜌𝑢𝑖 ) = 0 (1)
𝜕𝑥𝑖

where, ρ is the density of fluid and 𝑢𝑖 is the axial velocity.

Conservation of momentum

𝜕 𝜕𝑃 𝜕 𝜕𝑢 𝜕𝑢 2 𝜕𝑢
(𝜌𝑢𝑖 𝑢𝑗 ) = − 𝜕𝑥 + 𝜕𝑥 [𝜇 (𝜕𝑥 𝑖 + 𝜕𝑥𝑗) − 3 𝜇 𝜕𝑥𝑖 𝛿𝑖𝑗 − ̅̅̅̅̅̅̅̅̅̅
𝜌𝑢′ 𝑖 𝑢′𝑗 ] (2)
𝜕𝑥𝑖 𝑖 𝑗 𝑗 𝑗 𝑖

Conservation of energy

𝜕 𝜕 𝜕𝑇 𝜇𝑡 𝜕(𝐶𝑝 𝑇) 𝜕𝑃 𝜕𝑢𝑖 𝜕𝑢𝑗 2 𝜕𝑢𝑖 𝜕𝑢𝑖

(𝜌𝑢𝑗 𝐶𝑝 𝑇) = − (𝜆 + ) + 𝑢𝑗 + [𝜇 ( + )− 𝜇 𝛿𝑖𝑗 − ̅̅̅̅̅̅̅̅̅
𝜌𝑢′ 𝑖 𝑢′𝑗 ] (3)
𝜕𝑥𝑖 𝜕𝑥𝑖 𝜕𝑥𝑗 𝜎ℎ,𝑡 𝜕𝑥𝑗 𝜕𝑥𝑗 𝜕𝑥𝑗 𝜕𝑥𝑗 3 𝜕𝑥𝑖 𝜕𝑥𝑗

where −𝜌𝑢̅̅̅̅̅̅̅̅̅̅
′ 𝑢 ′ are the Reynolds stress, u and u are the time-averaged velocity for i and j directions.
𝑖 𝑗 i j
Time-averaged temperature, fluid thermal conductivity, density, turbulent Prandtl number for
energy, turbulent viscosity and time-averaged pressure are stated as T, λ, , h,t, µt and P,
respectively.
In the numerical study, the realizable k–ε turbulence model is used in order to give fast and
accurate results [10-12,17,18]. Therefore, turbulent dissipation rates (ε) and transport of turbulence
kinetic energy (k) equations should be considered.

3
CFD Letters
Volume 15, Issue 7 (2023) 1-13

k equation

𝜕 𝜕 𝜇 𝜕𝑘
(𝜌𝑘𝑢𝑗 ) = 𝜕𝑥 [(𝜇 + 𝜎 𝑡 ) 𝜕𝑥 ] + 𝐺𝑘 − 𝜌𝜀 (4)
𝜕𝑥𝑗 𝑖 𝑘 𝑗

ε equation

𝜕 𝜕 𝜇 𝜕𝜀 𝜀2
(𝜌𝜀𝑢𝑗 ) = 𝜕𝑥 [(𝜇 + 𝜎𝑡) 𝜕𝑥 ] + 𝜌𝐶1 𝑆𝜀 − 𝜌𝐶2 𝑘+ (5)
𝜕𝑥𝑗 𝑖 𝜀 𝑗 √𝜐𝜀

Turbulent Prandtl number is expressed as k and ε regarding k and ε in Eq. (14) and Eq. (15). Eq.
(16) represents the turbulent viscosity.

𝑘2
𝜇𝑡 = 𝜌𝐶𝜇 (6)
𝜀

with C1= 1.44, C2 = 1.9, k = 1, ε = 1.2

(a)

(b)
Fig. 1. Schematic diagram of physical model, showing the test section and the boundary conditions

2.3 Thermophysical Properties of Hybrid Nanofluid

These equations are general and can be applied in mono and hybrid nanoparticles with the
suitable modifications, the base fluid is symbolized with (bf), the nanoparticle with (np) and the
nanofluid with (nf), while the volume concentration of the total nanoparticle in the fluid with (∅). It

4
CFD Letters
Volume 15, Issue 7 (2023) 1-13

is important to state that in this study, the nanoparticle “Al 2O3” is symbolized with the number “1”
and the nanoparticle “CuO” with the number “2”. Bellos and Tzivanidis [19] calculate the total
concentration for the hybrid nanofluid (∅) using the following:

∅ = ∅1 + ∅2 (7)

These following equations give the density, the specific heat capacity and the thermal
conductivity of the equivalent nanoparticle of concentration (∅) and aids to the further calculation
of the hybrid nanofluid thermal property definition according to Minea [20] and Sundar et al., [21].

The density () of the equivalent nanoparticle is given as:

∅1 𝜌𝑛𝑝−1 +∅2 𝜌𝑛𝑝−2

𝜌𝑛𝑝 = (8)
∅

The specific heat capacity (Cpnp) of the equivalent nanoparticles is given as:

∅1 𝜌𝑛𝑝−1 𝐶𝑝𝑛𝑝−1 +∅2 𝜌𝑛𝑝−2 𝐶𝑝𝑛𝑝−2

𝐶𝑝𝑛𝑝 = (9)
𝜌𝑛𝑝 ∅

The thermal conductivity (knp) of the equivalent nanoparticles is given as:

∅1 .𝑘𝑛𝑝−1 +∅2 .𝑘𝑛𝑝−2

𝑘𝑛𝑝 = (10)
∅

The density (nf) of the hybrid nanofluid is given as:

𝜌𝑛𝑓 = 𝜌𝑏𝑓 . (1 − ∅) + 𝜌𝑛𝑝 . ∅ (11)

The specific thermal capacity (𝐶𝑝𝑛𝑓 )is given as:

𝜌𝑏𝑓 .(1−∅) 𝜌𝑛𝑝 .(1−∅)

𝐶𝑝𝑛𝑓 = . 𝐶𝑝𝑏𝑓 + . 𝐶𝑝𝑛𝑝 (12)
𝜌𝑛𝑓 𝜌𝑛𝑓

The thermal conductivity of the nanofluid (knf) is calculated according to the Maxwell model [22]:

𝑘𝑛𝑝 +2.𝑘𝑏𝑓 +2.(𝑘𝑛𝑝 −𝑘𝑏𝑓 ).∅

𝑘𝑛𝑓 = 𝑘𝑏𝑓 (13)
𝑘𝑛𝑝 +2.𝑘𝑏𝑓 −(𝑘𝑛𝑝 −𝑘𝑏𝑓 ).∅

The nanofluid dynamic viscosity (μ) can be calculated according to the Brinkman model [19]:
𝜇𝑏𝑓
𝜇𝑛𝑓 = (1−∅)2.5
(14)

Table 1
Water and nanoparticles properties
Materials  (kg/m3) Cp (J/kg.K) k (W/m.K)  (Pa s)
Water 998 4182 0.597 0.000998
Alumina, Al2O3 3880 765 40 -
Copper Oxide, CuO 6350 535 69 -

5
CFD Letters
Volume 15, Issue 7 (2023) 1-13

3. Results

The present computations are performed for a 2-D turbulent flow of the Al2O3/CuO -water
nanofluid over the corrugated saw tooth pipe. Various forms of quantitative and qualitative results
can be displayed from the output of simulations, but due to the space restriction, friction factor result
is presented. As presented in Table 2, hybrid nanofluids parameters are considered.

Table 2
Hybrid nanofluids properties
Nanofluids Volume fraction,  % (mixing ratio)  (kg/m3) Cp (J/kg.K) k (W/m.K)  (Pa s) Pr
Al2O3-CuO 0.5% (80:20) 1013.885 4104.04 0.621852 0.000861 5.681

3.1 Verification and Validation

CFD results were compared with previous experimental results for validation. Grid independence
studies were performed using different mesh sizes, such as 0.3mm, 0.2mm, 0.1mm, 0.09mm, and
0.08mm. Figure.2 shows the grid size for the 2D turbulent flow simulation, and Figure. 3 indicates the
heat transfer Nusselt number and friction factor values obtained using different mesh sizes.
Therefore, a mesh size of 0.1 mm size was selected. The 0.1 mm mesh size will be used for subsequent
flow analysis. The percentage difference between 0.1mm, 0.09mm, and 0.08mm is 1-2%, which
indicates that the current CFD model is reliable.

Fig. 2. Grid size of 0.1mm used for the 2D flow analysis

Fig. 3. Grid independence test

6
CFD Letters
Volume 15, Issue 7 (2023) 1-13

To assure the reliability and validity of the numerical algorithm procedure presented in the
current study, the numerical results for fully developed turbulent water flow in the smooth tube were
compared to the well-known empirical correlations proposed by Dittus-Boelter, Eq. (15), Petukov,
Eq. (16), and Glieninski, Eq. (17) for the average Nusselt number and the correlations proposed by
Filonenko, Eq. (18), McAdams, Eq. (19), and Petukhov, Eq. (20), for the friction factor.

𝑁𝑢 = 0.024𝑅𝑒 0.8 𝑃𝑟 0.4 (15)

𝑓
𝑅𝑒𝑃𝑟( )
8
𝑁𝑢 = 𝑓
2 (16)
1.07+12.7( )0.5 (𝑃𝑟 3 −1)
8

𝑓
( )(𝑅𝑒−1000)𝑃𝑟
8
𝑁𝑢 = 𝑓
2 (17)
1.0+12.7( )0.5 (𝑃𝑟 3 −1)
8

𝑓 = (1.84𝑙𝑜𝑔10 𝑅𝑒 − 1.64)−2 (18)

𝑓 = 0.18𝑅𝑒 −2 (19)

𝑓 = (0.79𝑙𝑛𝑅𝑒 − 1.64)−2 (20)

Figure. 4 and Figure. 5 shows the comparison of the numerical Nu and ƒ with empirical
correlations, respectively. Obviously, the numerical results are in good agreement with the existing
correlations. It is noted that the average discrepancy of Nusselt number with Dittus-Boelter is within
11.8%, Petukov is within 4%, and Gnielinski it is within 7%, while the average discrepancy of friction
factor was within 3.6%, 6.87%, and 7.9% for Filonenko, McAdams, and Petukhov correlations,
respectively.

Fig. 4. Validation for smooth tube for Average Nusselt

number

7
CFD Letters
Volume 15, Issue 7 (2023) 1-13

Fig. 5. Validation for smooth tube for friction factor

number

As illustrated in Figure 6, the comparison shows good trend agreement between the presented
numerical results and empirical correlations data. The average discrepancy of average Nusselt
number with Pak-Cho is within 6.3% and Maiga et al., is within 24.6%. The deviation may be related
to different properties of Al2O3 used as an input data in the simulation.

Fig. 6. Validation for smooth tube with Al2O3 nanofluids

for Average Nusselt number

8
CFD Letters
Volume 15, Issue 7 (2023) 1-13

3.2 Response Surface Analysis (RSM)

RSM is known to be straightforward and computationally efficient. It can be easily constructed.

RSM is a statistical modelling technique that uses regression analysis to determine 𝑓(𝑥)in the
observed response value (𝑦̂) and then estimates the effect of the independent parameters (𝑥). A
standard Design of Experiment (DOE) embedded in RSM is called Box Behnken design (BBD). BBD is
an independent three-level quadratic experimental design method which does not contain an
embedded factorial design. BBD method is being applied in this simulation plan. BBD is specially
designed to fit a second-order model. The resulting value from RSM analysis is used to determine the
polynomial equation and simplify the equation according to the influence of the factors in the final
response. The three factors, which are the wavelength (Lw), amplitude (a) and Reynolds number (Re),
are chosen based on Sundar et al., [21] and Amani et al., [22]. There are two levels of the factors
listed in Table 2. The lower and Upper bounds represent the factors' low and high levels, respectively.
The simulation design based on BBD for 32 factorials can be seen in Table 3.

Table 3
Factors and level of the BBD
No Factors Level
Lower bound Upper bound
1 Lw 5 20
2 a 1 4
3 Re 10000 30000

In the current study, from three variables, 15 design points have been generated from the BBD
model and then 15 CFD models must be simulated. According to the result of BBD, friction analysis
based on the finite volume method (FVM) is performed to obtain the values of the different response
variables. The simulation results are presented in Table 4. The response variables obtained from the
simulations are then analysed by multiple quadratic regression to determine the mathematical
models with the best fits. The adequacy and reliability of the regression models were also tested by
using the analysis of variance (ANOVA).

Table 4
Factors and corresponding response as per CFD
design used
No. Parameters Response
Lw a Re CFD RSM
1 5 4 20000 0.291983 0.298955
2 5 2.5 30000 0.276350 0.263656
3 12.5 2.5 20000 0.160232 0.160291
4 12.5 4 10000 0.207289 0.211745
5 5 1 20000 0.150803 0.167953
6 12.5 2.5 20000 0.160320 0.160291
7 5 2.5 10000 0.306728 0.295301
8 12.5 1 30000 0.079924 0.075468
9 12.5 2.5 20000 0.160320 0.160291
10 20 2.5 30000 0.151389 0.162816
11 12.5 1 10000 0.081810 0.076087
12 20 4 20000 0.188894 0.171744
13 20 1 20000 0.053222 0.046250
14 20 2.5 10000 0.134533 0.147227
15 12.5 4 30000 0.190583 0196306

9
CFD Letters
Volume 15, Issue 7 (2023) 1-13

After simulating the 15 design points via Ansys Fluent, the simulation friction factor was used to
find the quadratic regression model. Interestingly, the predicted value obtained by the quadratic
regression model agreed well with the CFD values (see Table 5). Likewise, in Figure 6, the coefficient
of determination (R2) of 98.25% is in reasonable agreement with the adjusted coefficient of
determination (R2) of 95.11%, thus indicating a well-represented response surface. The large F value
is 31.27 indicates the great significance of the regression model. The associate P-values less than 0.05
for the model indicate that the model terms are statistically significant, and the effects of the model
terms with a P-value greater than 0.05 are insignificant. Esfe et al., [23], show R2 and P values are
99.73% and 0.0001, respectively, in their study on optimising nanofluid flow in a double tube heat
exchanger, which indicates the accuracy and great significance important of the regression model.
Referring to Montgomery et al., [24] and Khalid et al., [25], the R2 is called the coefficient of
determination and is often used to check the adequacy of a regression model. The large value R 2
indicates that the model has successfully explained the variability in the response. From Figure 7, the
predicted values obtained from the linear model agree well with the simulated values. The
mathematical model for predicting the friction factor can be expressed as follows:

FF = 0.2700 - 0.03076 Lw + 0.1230 a - 0.000007 Re + 0.000785 Lw*Lw - 0.01476 a*a + 0.000000 Re*Re
- 0.000122 Lw*a + 0.000000 Lw*Re - 0.000000 a*Re

Where Lw is the wavelength, a is the amplitude and Re is the Reynolds number. This equation
can be used to study the response of friction factor by varying the involved parameters. For example,
the positive coefficients associated with the factors a, Lw2, Re2, and Lw*Re reveal an increased
friction factor if the factors are increased. Conversely, the negative coefficients associated with
factors Lw, Re, a2, Lw*a, and a*Re indicate decreased friction factor when these factors are increased.
According to Chiang et al., [26], the increase in friction factor is apparently related to pitch height
and pitch-to-pitch distance and this concurs with the results of this study.

Table 5
Regression result using BBD
Source DF Adj SS Adj MS F-Value P-Value
Model 9 0.077334 0.008593 31.27 0.001
Linear 3 0.064003 0.021334 77.63 0
Lw 1 0.030979 0.030979 112.72 0
a 1 0.032895 0.032895 119.7 0
Re 1 0.000129 0.000129 0.47 0.524
Square 3 0.012711 0.004237 15.42 0.006
Lw*Lw 1 0.007194 0.007194 26.18 0.004
a*a 1 0.004071 0.004071 14.82 0.012
Re*Re 1 0.000607 0.000607 2.21 0.197
2-Way Interaction 3 0.00062 0.000207 0.75 0.566
Lw*a 1 0.000008 0.000008 0.03 0.875
Lw*Re 1 0.000558 0.000558 2.03 0.214
a*Re 1 0.000055 0.000055 0.2 0.674
Error 5 0.001374 0.000275
Lack-of-Fit 3 0.001374 0.000458 177441.35 0
Pure Error 2 0 0
Total 14 0.078708

10
CFD Letters
Volume 15, Issue 7 (2023) 1-13

Fig. 7. Comparison of predicted value using regression

model and actual value from CFD

While assessing the points on the normal probability plot in Figure 8, the plotted points fall
roughly along the straight line. This observation supports the claim that the residuals are normally
distributed, and the assumption of normality can be satisfied. Also, it shows that this model is
appropriate for this analysis. The histogram plot follows a symmetric distribution, indicating that it is
an appropriate model for the data. As shown by the versus fits plot, the residual has about the same
amount of variation at all tiers of the fitted values. Therefore, it can be considered that the residuals
are homoscedasticity. By performing subsequent analysis on the versus order plot, it shows no
obvious and drastic pattern change. Therefore, the residual is probably mutually independent and
homoscedastic with respect to run order. From the above analysis, all the assumptions required to
validate the quadratic regression for this model appear to be satisfied, fit and adequate.

Fig. 8. Residual plot for friction factor

11
CFD Letters
Volume 15, Issue 7 (2023) 1-13

4. Conclusion

A CFD friction factor model of internal corrugated pipe flow has been verified and validated with
the results obtained from the regression model. The response surface methodology has been used
to determine three design factors' effects on the friction factor (response variable). The quadratic
model obtained from the Box Behnken design (BBD) approach has been successfully developed. It
reveals that three factors (i.e., Lw, a, Re) affect the friction factor in a significant manner.

Acknowledgement
The authors are grateful for the support from Universiti Tenaga Nasional (UNITEN), Agensi Nuklear
Malaysia (ANM), Jabatan Perkhidmatan Awam Malaysia (JPA) and Rabdan Academy for their
supports and helps in completing this work.

References
[1] Bigdeli, Masoud Bozorg, Matteo Fasano, Annalisa Cardellini, Eliodoro Chiavazzo, and Pietro Asinari. "A review on
the heat and mass transfer phenomena in nanofluid coolants with special focus on automotive
applications." Renewable and Sustainable Energy Reviews 60 (2016): 1615-1633.
https://doi.org/10.1016/j.rser.2016.03.027
[2] Sidik, Nor Azwadi Che, M. M. Yassin, and M. N. Musa. "Turbulent-forced convective heat transfer and pressure drop
analysis of Fe3O4 magnetic nanofluid in a circular microchannel." Jurnal Teknologi 75, no. 11 (2015).
https://doi.org/10.11113/jt.v75.5293
[3] Wen, Dongsheng, Guiping Lin, Saeid Vafaei, and Kai Zhang. "Review of nanofluids for heat transfer
applications." Particuology 7, no. 2 (2009): 141-150. https://doi.org/10.1016/j.partic.2009.01.007
[4] Ventola, Luigi, Francesco Robotti, Masoud Dialameh, Flaviana Calignano, Diego Manfredi, Eliodoro Chiavazzo, and
Pietro Asinari. "Rough surfaces with enhanced heat transfer for electronics cooling by direct metal laser
sintering." International Journal of Heat and Mass Transfer 75 (2014): 58-74.
https://doi.org/10.1016/j.ijheatmasstransfer.2014.03.037
[5] Choi, S. US, and Jeffrey A. Eastman. Enhancing thermal conductivity of fluids with nanoparticles. No. ANL/MSD/CP-
84938; CONF-951135-29. Argonne National Lab.(ANL), Argonne, IL (United States), 1995.
[6] Han, Huai-Zhi, Bing-Xi Li, Hao Wu, and Wei Shao. "Multi-objective shape optimization of double pipe heat exchanger
with inner corrugated tube using RSM method." International Journal of Thermal Sciences 90 (2015): 173-186.
https://doi.org/10.1016/j.ijthermalsci.2014.12.010
[7] Ny, G., N. Barom, S. Noraziman, and S. Yeow. "Numerical study on turbulent-forced convective heat transfer of
Ag/Heg water nanofluid in pipe." J. Adv. Res. Mater. Sci 22, no. 1 (2016): 11-27.
[8] Permanasari, Avita Ayu, Muhammad Taufiq Affandi, Poppy Puspitasari, and Mirza Abdillah. "Optimization of
manganese ferrite/distilled water parameter design on heat exchanger using RSM and CFD." In IOP Conference
Series: Materials Science and Engineering, vol. 1034, no. 1, p. 012056. IOP Publishing, 2021.
https://doi.org/10.1088/1757-899X/1034/1/012056
[9] Sajid, Muhammad Usman, and Hafiz Muhammad Ali. "Thermal conductivity of hybrid nanofluids: a critical
review." International Journal of Heat and Mass Transfer 126 (2018): 211-234.
https://doi.org/10.1016/j.ijheatmasstransfer.2018.05.021
[10] Bahiraei, Mehdi, Mohamad Berahmand, and Amin Shahsavar. "Irreversibility analysis for flow of a non-Newtonian
hybrid nanofluid containing coated CNT/Fe3O4 nanoparticles in a minichannel heat exchanger." Applied Thermal
Engineering 125 (2017): 1083-1093. https://doi.org/10.1016/j.applthermaleng.2017.07.100
[11] Bahiraei, Mehdi, Reza Rahmani, Ali Yaghoobi, Erfan Khodabandeh, Ramin Mashayekhi, and Mohammad Amani.
"Recent research contributions concerning use of nanofluids in heat exchangers: a critical review." Applied Thermal
Engineering 133 (2018): 137-159. https://doi.org/10.1016/j.applthermaleng.2018.01.041
[12] Gürbüz, Emine Yağız, Halil İbrahim Variyenli, Adnan Sözen, Ataollah Khanlari, and Mert Ökten. "Experimental and
numerical analysis on using CuO-Al2O3/water hybrid nanofluid in a U-type tubular heat exchanger." International
Journal of Numerical Methods for Heat & Fluid Flow 31, no. 1 (2021): 519-540. https://doi.org/10.1108/HFF-04-
2020-0195
[13] Chein, Reiyu, and Jason Chuang. "Experimental microchannel heat sink performance studies using
nanofluids." International journal of thermal sciences 46, no. 1 (2007): 57-66.
https://doi.org/10.1016/j.ijthermalsci.2006.03.009

12
CFD Letters
Volume 15, Issue 7 (2023) 1-13

[14] Mukhtar, A., K. C. Ng, and M. Z. Yusoff. "Passive thermal performance prediction and multi-objective optimization
of naturally-ventilated underground shelter in Malaysia." Renewable Energy 123 (2018): 342-352.
https://doi.org/10.1016/j.renene.2018.02.022
[15] Yusoff, Mohd Zamri, Azfarizal Mukhtar, Khai Ching Ng, and Mohamad Fariz Mohamed Nasir. "Application of Box-
Behnken design with response surface to optimize ventilation system in underground shelter." Journal of Advanced
Research in Fluid Mechanics and Thermal Sciences 52, no. 2 (2018): 161-173.
[16] Kaood, A., T. Abou-Deif, H. Eltahan, M. A. Yehia, and E. E. Khalil. "Numerical investigation of heat transfer and
friction characteristics for turbulent flow in various corrugated tubes." Proceedings of the Institution of Mechanical
Engineers, Part A: Journal of Power and Energy 233, no. 4 (2019): 457-475.
https://doi.org/10.1177/0957650918806407
[17] Ekiciler, Recep, Kamil Arslan, Oğuz Turgut, and Burak Kurşun. "Effect of hybrid nanofluid on heat transfer
performance of parabolic trough solar collector receiver." Journal of Thermal Analysis and Calorimetry 143 (2021):
1637-1654. https://doi.org/10.1007/s10973-020-09717-5
[18] Tan, C., S. Zainal, C. J. Sian, and T. J. Siang. "ANSYS simulation for Ag/HEG hybrid nanofluid in turbulent circular
pipe." Journal of Advanced Research in Applied Mechanics 23, no. 1 (2016): 20-35.
[19] Bellos, Evangelos, and Christos Tzivanidis. "Thermal analysis of parabolic trough collector operating with mono and
hybrid nanofluids." Sustainable Energy Technologies and Assessments 26 (2018): 105-115.
https://doi.org/10.1016/j.seta.2017.10.005
[20] Minea, Alina Adriana. "Hybrid nanofluids based on Al2O3, TiO2 and SiO2: numerical evaluation of different
approaches." International Journal of Heat and Mass Transfer 104 (2017): 852-860.
https://doi.org/10.1016/j.ijheatmasstransfer.2016.09.012
[21] Sundar, L. Syam, Korada Viswanatha Sharma, Manoj K. Singh, and A. C. M. Sousa. "Hybrid nanofluids preparation,
thermal properties, heat transfer and friction factor–a review." Renewable and Sustainable Energy Reviews 68
(2017): 185-198. https://doi.org/10.1016/j.rser.2016.09.108
[22] Amani, Mohammad, Pouria Amani, Alibakhsh Kasaeian, Omid Mahian, and Somchai Wongwises. "Thermal
conductivity measurement of spinel-type ferrite MnFe2O4 nanofluids in the presence of a uniform magnetic
field." Journal of Molecular Liquids 230 (2017): 121-128. https://doi.org/10.1016/j.molliq.2016.12.013
[23] Esfe, Mohammad Hemmat, Hadi Hajmohammad, Davood Toghraie, Hadi Rostamian, Omid Mahian, and Somchai
Wongwises. "Multi-objective optimization of nanofluid flow in double tube heat exchangers for applications in
energy systems." Energy 137 (2017): 160-171. https://doi.org/10.1016/j.energy.2017.06.104
[24] Montgomery, Douglas C., George C. Runger, and Norma F. Hubele. Engineering statistics. John Wiley & Sons, 2009.
[25] Zarina Mohd Khalid, Norazlina Ismail, Norhaiza Ahmad, Noraslinda Mohamed Ismail, Arifah Bahar, Ismail
Mohamad, Muhammad Hisyam Lee, and Muhammad Fauzi Hamdan, Statistics For Engineers. 2019, Johor,
Malaysia: UTM PRESS. 187.
[26] Chiang, Ko-Ta, and Fu-Ping Chang. "Application of response surface methodology in the parametric optimization of
a pin-fin type heat sink." International communications in heat and mass transfer 33, no. 7 (2006): 836-845.
https://doi.org/10.1016/j.icheatmasstransfer.2006.04.011

13