International Communications in Heat and Mass Transfer 118 (2020) 104819

Contents lists available at ScienceDirect

International Communications in Heat and Mass Transfer

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

Validation of the Gnielinski correlation for evaluation of heat transfer T

coefficient of enhanced tubes by non-linear regression model: An
experimental study of absorption refrigeration system
Syed Muhammad Ammara, Chan Woo Parka,b,
⁎

a
School of Mechanical Design Engineering, Jeonbuk National University, 567, Baekje-daero, Deokjin-Gu, Jeonju-City, Jeonbuk 561-756, Republic of Korea
b
Department of Energy Storage/Conversion Engineering of Graduate School, Jeonbuk National University, 567, Baekje-daero, DeokjinGu, Jeonju 6 City, Jeonbuk 561-756,
Republic of Korea

ARTICLE INFO ABSTRACT

Keywords: The Wilson plot method has widely used to determine general heat transfer correlations, but this technique
Absorber, absorption chiller needs considerable experimental data. The Gnielinski correlation, however, simplifies the evaluation of the
Enhanced tube inside heat transfer coefficient of heating tubes only by using friction factor and hydraulic diameter. Nowadays,
Evaporator enhanced tubes, are used in an absorption refrigeration system to improve the thermal performance of the
Gnielinski correlation
system. As these enhanced tubes have various heat transfer surfaces, the Wilson plot method based on numerous
Wilson plot
measured experimental data has been responsible for the evaluation of the Nusselt correlation; the validation of
the Gnielinski correlation has not been found yet for enhanced tubes. The current study, thus, pursues to find the
validity of the Gnielinski correlation to simplicate the assessment of the inside heat transfer coefficient of en-
hanced tubes over the Wilson plot method. First, a non-linear regression scheme developed by the Wilson plot
method examined the Nusselt correlations of three different types of enhanced tubes including a smooth tube;
friction factors of enhanced tubes, then, were measured to apply the Gnielinski correlation. Gnielinski corre-
lation was modified using 103 data points and a new correlation was obtained, which showed a total mean error
of 2.1%, and an agreement within ± 5%.

1. Introduction from the inside fluid to the tube wall occurs through convection and
conduction. The fluid flow inside the evaporator and absorber of the
Vapor absorption refrigeration system (VARs), also known to reduce VARs is generally turbulent [10]. For turbulent flow, the phenomenon
energy consumption by utilizing low-grade energy sources, comprises of heat transfer from water to the tube wall is mostly through forced
four main components: evaporator, absorber, generator, and condenser convection and a very small amount of heat transfer occurs through
[1]. VARs designing requires the optimization of both outside and tube- conduction, because of the low conductivity of water. On the other
side heat transfer performance. The optimization or heat transfer en- hand, heat transfer within the viscous region occurs through conduc-
hancement can be achieved by augmenting the heat transfer coefficient tion. Thus, it is very important to enhance the inner side of the tube as
[2,3], which results in decreasing the size of thermal systems [4], de- heat transfer is retarded by the film resistance acting on the inside tube.
creasing entropy generation, and increasing the second law efficiency. The inside heat transfer coefficient has been extensively studied.
Heat transfer augmentation is achieved using various techniques such Dittus and Boelter [11] proposed the correlation of the inside heat
as the use of insert devices [5], treated surfaces [6], mechanical aids, transfer coefficient for turbulent flow with the Prandtl number ranging
swirl flow devices, and rough surfaces. [7]. The choice of the technique from 0.6 to 160 for the smooth tube. The correlation is given as:
depends on various parameters such as type of flow (Reynolds number), (1)
Nu = 0.023Re0.8Pr b
chance of fouling, properties of fluids, and the maximum pressure drop
allowed [8]. Where Prandtl number exponent ‘b’ is 0.4 for heating and 0.3 for
Dittus and Boelter [9] explained that the heat transfer phenomenon cooling.
Sieder and Tate [12] modified the general correlation of the inside

⁎
Corresponding author at: School of Mechanical Design Engineering, Jeonbuk National University, 567, Baekje-daero, Deokjin-Gu, Jeonju-City, Jeonbuk 561-756,
Republic of Korea.
E-mail address: cw-park@jbnu.ac.kr (C.W. Park).

https://doi.org/10.1016/j.icheatmasstransfer.2020.104819

0735-1933/ © 2020 Elsevier Ltd. All rights reserved.

S.M. Ammar and C.W. Park International Communications in Heat and Mass Transfer 118 (2020) 104819

Nomenclature fitted dependent variables, K2/W2

HTCtube Tube-side heat transfer coefficient, W/m2K
A Heat transfer area, m2 t Tube
a Reynolds number exponent temp. Temperature, °C
b Prandtl number exponent ʋ- Flow velocity, m/s
C1 Nusselt number leading constant
C2 Fixed resistances, K/W Greek symbols
Dh Hydraulic Diameter, m
EF Enhancement factor ρ Density, kg/m3
K Thermal conductivity, W/mK ∑ Sum
L Length, m abs Mean error
LHS Left-hand side ξi Relative error
n Number of data points α Ridge angle, °
N Number of grooves ΔP Pressure drop, kPa
Nu Nusselt number α1, α2 Groove angle, °
PHX Plate heat exchanger
p Inside pitch, m Subscripts
PF Friction penalty factor
Pr Prandtl number f Friction factor
Re Reynold number h Hydraulic
RH Height of the ridge, mm o Outside
RTD Resistance Temperature Detector, °C s Smooth
Rov Overall thermal resistance, K/W w wall
RHS Right-hand side
S Sum of the square of deviation between measured and

heat transfer coefficient in turbulent flow when there were large visc- current study also contains Tube Z, which has a floral shape. The
osity changes owing to a considerable temperature difference between method to measure the inside heat transfer coefficient directly is by
the bulk fluid and wall. Petukhov [13] derived the correlation for a Newton's Law of cooling. The main difficulty is to find the wall tem-
fully developed turbulent pipe flow having a fluid Prandtl number of perature, especially within the heat exchanger. Hence, the Wilson plot
0.5 to 2000. Gnielinski [14] proposed a very accurate correlation for [22] is the only available method to measure the heat transfer coeffi-
turbulent as well as transition regions for a smooth tube. cient. In the past [23–33], mostly inside heat transfer coefficient is
0.11
calculated using Briggs and Young method [34]. In finding the Nusselt
2/3
(f /8)(Re 1000) Pr Dh Pr correlation, the method gives a false impression of convergence of
Nu = 1+
1 + 12.7(f /8)1/2 (Pr 2/3 1) L Prw (2) Reynolds exponent as the author tried for various data sets in this re-
search. The current research used the actual form of the regression
Eq. (2) is valid for 2300 < Re < 106 and 0.6 < Pr < 105. The
model which guaranteed the convergence of Reynolds exponent. The
correlation has an error of ± 20% for 90% of the data based on 800
present study aimed to verify the Gnielinski correlation to simplify the
experimental data points.
evaluation of the inside heat transfer coefficient of enhanced tubes
Many other researchers have also recently studied the inside heat-
because the Wilson plot method requires much time-consuming effort.
transfer coefficient for the transitional flow regime in a smooth hor-
Three types of enhanced tubes, which include a floral tube and two
izontal tube [15–17].
types of ridged tubes, were investigated and tests were conducted for
Thermal performance in enhanced tubes has been extensively stu-
both heating and cooling fluids. Smooth tube was tested as a bench
died. Chen et al. [18] studied tube-side heat-transfer performance in six
mark. The operating conditions simulated evaporator and absorber
types of dimpled tubes, all of which outer performed the smooth tube in
conditions of the VARs. For the current research, thermal performances
terms of heat transfer augmentation. Sivashanmugam and Suresh [19]
of different enhanced tubes were also compared in terms of enhance-
studied heat transfer and friction factor in circular tubes having inserts
ment factor (EF) and friction penalty factor (PF). Additionally, the ef-
and developed a correlation for the Nusselt number and friction factor
fect of property variations were also included in the obtained Nusselt
based on their experimental data. García et al. [8] analyzed the
correlations.
thermal-hydraulic behavior in three types of enhancements: corrugated
tubes, dimpled tubes, and wired coils, and concluded that the type of
enhancement needed depends on the Reynolds number. Smooth tubes 2. Experimental procedure
are better for lower Reynolds numbers, whereas enhanced tubes are
better for high Reynolds numbers. Kim and Go [20] experimentally 2.1. Operation of apparatus
analyzed the tube-side heat transfer and friction performance of various
corrugated tubes and derived correlations for the heat transfer Nusselt An apparatus was developed to obtain the tube-side heat transfer
number and friction factor. coefficient and pressure drop. A similar setup configuration for the
The inside and outside thermal resistance of doubly enhanced tubes evaluation of the heat transfer coefficient has been used by several
are comparable especially for the application involved enhanced out- researchers [23,27,28]. The heat-transfer coefficient was obtained for
side surfaces involving boiling, condensation, and evaporation [21]. the practical operating conditions of the evaporator conditions
Therefore, it is very important to properly investigate the tube-side heat (Cooling) and absorber conditions (Heating). The Wilson plot method,
transfer coefficient. The research on the inside heat transfer coefficient which is based on the non-linear regression scheme [35], was adopted
of enhanced tubes is very rare. Recently, Ji et al. [21] presented an as other methods, such as the direct measurement of the convection
extended Gnielinski correlation for internally grooved tubes. However, coefficient by Newton's cooling law requires measurement of surface
their study was only concerned with internally grooved tubes. The temperature, which is very difficult to obtain practically [26].

2
S.M. Ammar and C.W. Park International Communications in Heat and Mass Transfer 118 (2020) 104819

The experimental apparatus, shown in Fig. 1, comprises a double- Table 1

pipe heat exchanger having an annular-side diameter of 36 mm and is Experimental conditions.
connected to an annular-side flow meter, a tube-side flow meter, and a Parameters Value
differential pressure gauge. The counter flow configuration is main-
tained by having a turbulent flow on the tube-side as well as the an- Cooling
Tube-Inlet temperature 12 °C
nular-side. A constant annular-side flow rate is maintained, while the
Tube-Reynolds number 5500 - 80,000
tube-side flow rate is varied. The inlet temperatures of the tube and Annular-inlet temperature 5 °C
annular-sides are kept constant using constant temperature baths on
Heating
both sides. The temperatures are measured using four-wire RTD Tube-Inlet temperature 32 °C
( ± 0.1 °C accuracy) at the inlet and outlet of the tube and annular- Tube-Reynolds number 8500 - 12,5000
sides. A differential pressure gauge (0.4% accuracy) is used to measure Annular-inlet temperature 50 °C
the pressure drop across the test tube. Flow meters (Badger meter, ±
0.5% accuracy) are used to calculate the flow in l/min for both sides.
The apparatus is fully insulated and the heat balance is checked. The Table 2
experimental setup is connected to an Agilent 34970A data acquisition Tested tubes specifications.
system, which is provided with a Agilent BenchLink Data Logger 3 to Tubes L [mm] Dh [mm] α [°] p [mm] N [m] RH [mm]
record the data. The data for the steady-state condition is taken and
averaged over five minutes. Smooth 1116 15 – – – –
Tube X 1116 14.9 46.7° 5.13 – 0.048
For the Wilson method, as shown in Fig. 1, a series of experiments Tube Y 1116 14.8 42.4° 5.1 – 0.068
are done by changing the tube-side flow rate. Flow rate (l/min) and Tube Z 1116 14.6 α1 = 137° × 7 3.66 12 –
temperature readings are measured, and tire-some data reduction is α2 = 149° × 5
done.
For the Gnielinski correlation [14], as shown in Fig. 1, Nusselt
number (Eq. (2)) is measured by measuring pressure drop (kPa) and 2.2. Test section
flow rate (l/min) of all enhanced tubes. The friction factor of smooth
and enhanced tubes are measured from the measured pressure drop Microscopic photographs of the enhanced tubes are shown in Fig. 2.
using the differential pressure gauge as: Three enhanced tubes and one smooth tube are tested in the current
study, whose specifications are shown in Table 2. As shown in Fig. 2,
P 2 Dh Tube X and Tube Y have inside ridges of specific pitch and a specific
f=
L v2 (3) angle. Tube X has 38 outside fins per inch of the tube, while Tube Y has
40 outside fins per inch of the tube. Tube X and Tube Y are enhanced by
The complete experimental conditions are given in Table 1, and the machining the outer side of the smooth surface, while Tube Z is en-
specifications of the tested tubes are given in Table 2. All the tubes hanced by machining the outer side of the floral tube. It has notched
specifications are provided by the manufacturing company. fins on the outer surface and twelve grooves on the circumference.
The tubes have a total length of 1116 mm, which includes both

Fig. 1. Schematic of experimental apparatus.

3
S.M. Ammar and C.W. Park International Communications in Heat and Mass Transfer 118 (2020) 104819

Fig. 2. Microscopic photographs of enhanced tubes: (a) Tube X, (b) Tube Y, (c) Tube Z, (a1, b1, c1) inside view, (a2, b2, c2) outside view, and (a3, b3, c3) cross-section
view.

heat-transfer length (1016 mm) and non-heat-transfer length Rov =

1 1
+ C2
(100 mm), which is insulated well and used to connect the test tubes to C1 ReaPr b (Ak /Dh ) (5)
the experimental apparatus.
where the first term represents the inside resistance and C2 re-
presents the fixed resistances. Eq. (5) is a non-linear problem of re-
2.3. Data reduction gression analysis having C1, ‘a’, and C2 as unknowns.
The method of least squares is applied to Eq. (5) as follows:
The Wilson plot [26] technique uses the method of separating the n 2
measured overall thermal resistance into fixed-resistance and variable 1 1
S= (Rov )i C2
thermal resistance. Using the same principle, the current experimental i=1
[C1 ReaPr b (Ak / Dh )] i (6)
study is conducted for two cases. In the first case, the tube-side inlet th
where ‘i'represents the i data point and ‘n’ represents the number
temperature is maintained at 12 °C, with the water being cooled as in
of data points in the current experiment. Eq. (6) can be minimized by
the case of an evaporator of the absorption chiller. The second case
taking the partial derivative of S with respect to the unknowns and
involves the absorber conditions (Heating), where the tube-side inlet
equating the Eqs. to zero, which results in the following:
temperature is maintained at 32 °C. During the tests, the annular-side
velocity and bulk temperature are maintained constant. The complete n
1
n
1
(Rov )i = + nC2
experimental conditions are given in Table 1. C1 Rei a Ji (7)
i=1 i=1
The original Wilson plot [22] is used when the exponent of the
Reynolds number is already known. The current experimental data are n
(Rov )i 1
n
1
n
1
reduced by following the modified Wilson plot scheme, as suggested by = + C2
Rei a Ji C1 Rei 2a Ji 2 Rei a Ji (8)
Khartabil and Christensen [35]. For a single-phase flow, Nusselt cor-
i=1 i=1 i=1

relation can be written as: n

(Rov )iln(Rei) 1
n
ln(Rei)
n
ln(Rei)
= + C2
Nu = C1ReaPr b (4) i=1
Rei a Ji C1 i=1
Rei 2a Ji 2 i=1
Rei a Ji (9)

The value of the exponent of Prandtl number ‘b’ is 0.4 for the fluid With
being heated, and that of ‘b’ is 0.3 for the fluid being cooled. Pr b Ak
The overall thermal resistance of the double-pipe heat exchanger J=
Dh (10)
can be written as:

4
S.M. Ammar and C.W. Park International Communications in Heat and Mass Transfer 118 (2020) 104819

Eqs. (7–9) are linear for C1 and C2 and non-linear for ‘a’.Thus, C1 found that the divergence rate can become very slow which gives a false
and C2 can be solved using any two equations from Eqs. (7–9). impression of convergence. The same trends have been explained by
Eqs. (7) and (8) can be written in matrix form as: other researcher [35].
n n
1 (Rov )i 4.2. Validation of the method
n
Rei 2a Ji 2 1 1 Rei a Ji
i=1 i=1
Rei a Ji C1 =
n
1 i=1 n
The Wilson plot results obtained using the non-linear regression
n C2 (Rov )i
Rei a Ji scheme for a smooth tube under the absorber conditions (Heating) are
i=1 i=1 (11)
shown in Fig. 3a. During the calculation of Nusselt correlation, it is very
C1 and C2 are obtained as follows. important to include data points where fixed-resistance dominates and
n n n those where the tube-side resistance dominates to improve the accuracy
n
(Rov )i 1
(Rov )i of final results [35].
1 Reia Ji Rei a Ji
= i=1 i=1 i=1 The experimental apparatus is validated by comparing the results of
2
C1 n
1
n
1 smooth tubes with the famous Gnielinski correlation [14]. Fig. 3b
n
i=1
Rei2a Ji2
i=1
Reia Ji
(12) shows the results of experiments with the correlation. The results of the
smooth tube are within a range of ± 10%, which confirms the validity
n n n n
1
(Rov )i
(Rov )i 1 of the experimental apparatus. Fig. 4 compares the overall thermal
Re2ai J 2i Reai Ji Reia Ji
i=1 i=1 i=1 i=1 resistance of the regression model and the experiments. The results
C2 =
n
1
n
1
2
have an error of ± 1%.
n Rei a Ji
Re 2ai J 2i
i=1 i=1 (13)
4.3. Wilson plot results of enhanced tubes
We substitute for Eqs. (12), and (13) into Eq. (9) to obtain the re-
quired non-linear equation in ‘a’, which can be solved numerically by The Wilson plot results obtained using the non-linear regression
using the bisection method (elaborated in results) [35]. scheme for enhanced tubes are shown in Fig. 5. Fig. 5a shows the results
of absorber conditions (Heating), while Fig. 5b shows the results of
3. Uncertainty analysis evaporator conditions (Cooling). Table 5 shows the Nusselt correlations
obtained. The results are drawn for the overall thermal resistance
The method proposed by Kline and McClintock [36] is used in the against the quantity, 1/Reta Prb (Ak/Di). The values of the Reynolds
uncertainty analysis. Kline and McClintock [36] described uncertainty number exponent are found to be different for the evaporator condi-
on parameter uR as follows: tions (Cooling) and absorber conditions (Heating) for the same tube.
Fig. 6 shows the results obtained for Reynolds number exponent ‘a’
2 2 2 0.5
uR =
R
u1 +
R
u2 + .………+
R
un by using the bisection method applied to Eq. (9) by re-arranging it as
X1 X2 Xn (14) follows:

Here, if ‘R' is a function of (X1; X2;…; Xn), the uncertainty on vari- f (a) = LHS RHS (15)
able X1 is u1, that on variable X2 is u2, and so on. The uncertainty The Reynolds exponent ‘a’ is determined using the following pro-
analysis is conducted using the same method and is shown in Table 3. cedure.
The uncertainties in Nusselt numbers are kept to a minimum by
controlling the tube-side resistance and the fixed resistances. As 1. Assuming an arbitrarily small value of ‘a’ and evaluating f(a).
Reynolds number increases, uncertainty in Nusselt number increases 2. Incrementing ‘a’ by a small increase and evaluating the resultant
because of the increasing effect of fixed resistance. However, un- quantity f(a).
certainties in the fixed resistance is directly related to the uncertainties 3. If the sign of Eq. (15) changes, the true value lies between the in-
in the tube side resistance, because they are calculated simultaneously. terval.
The ratio of tube-side resistance to fixed resistance is highest at the
lowest Reynolds number. The results are obtained by choosing the The results can be more clearly illustrated by drawing a graph be-
Reynolds number range such that the tube-side resistance dominates at tween f(a) and a, which crosses the x-axis at the true solution. It is
the lower Reynolds number and the annular-side resistance dominates possible to encounter more than one solution of ‘a’ as Eq. (15) is non-
at a higher Reynolds number. For this purpose, a high-power pump linear in a. However, in this case, the solution will be chosen such that
(Wilo, 7.5 kW) is used to achieve the required higher Reynolds number. it gives the least value of ‘S', as shown in Eq. (6).
The maximum uncertainty in the Nusselt number is 30%.
4.4. Comparison of Nusselt number
4. Results and discussion
The smooth and enhanced tubes were compared in terms of Nusselt
4.1. Briggs and Young method number for both absorber conditions (Heating) and evaporator condi-
tions (Cooling). As shown in Fig. 7, the Nusselt number increases with
The results are first evaluated using the modified Wilson plot the Reynolds number because of having more turbulence at a higher
technique, as recommended by Briggs and Young [34]. Table 4 shows
the application of the Briggs and Young method on a smooth tube under Table 3
the absorber conditions (Heating). Clearly, this method fails to find the Experimental uncertainties.
Reynolds exponent correctly, by converging to the optimum value. In Parameters Uncertainties
fact, the value of ‘a’ is found to diverge. The values are found to be
diverging even at the true value of 0.79 (obtained as shown in Table 5), Temperature ± 0.1 °C
which is obtained in the current solution of Eq. (5) by keeping the Water flow rate 0.5%
Nusselt number 15.2%
actual form of the equation. The authors tried to apply Briggs and
Reynolds number 0.48%
Young method to several data sets, but it was found that the exponent Pressure drop 0.4%
of Reynolds number never converges completely. However, it was

5
S.M. Ammar and C.W. Park International Communications in Heat and Mass Transfer 118 (2020) 104819

Table 4 1.2x10
-2

Briggs and Young method to find Reynolds' exponent ‘a’ for

smooth tube for absorber (Heating). -2
1.1x10
Guessed value ‘a’ Obtained value ‘a’

0.2 0.19074475 9.6x10

-3

0.19074475 0.18216383

1/UA [K/W]
0.18216383 0.17352376
No Convergence 8.4x10
-3

0.4 0.37660394
0.37660394 0.35457418
0.35457418 0.33394861 7.2x10
-3

No Convergence
0.6 0.57365892
0.57365892 0.54650734 6.0x10
-3

0.54650734 0.51897742
No Convergence
(a)
0.8 0.79936599 4.8x10
-3
-5 -4 -4 -4 -4 -4 -4
0.79936599 0.79857908 6.0x10 1.2x10 1.8x10 2.4x10 3.0x10 3.6x10 4.2x10
0.79857908 0.79760378
0.79 0.4
0.79 0.78702654 1/Ret Pr (Ak/D) [K/W]
. .
. .
0.78 0.77482319
1024
No Convergence.

Table 5 512
Derived Nusselt Correlations.
+10%
Tubes Heating Cooling
256
NuGnielinski
Smooth Nu = 0.0242Re0.79Pr0.4 Nu = 0.0098Re0.891Pr0.3
Tube X Nu = 0.0146Re0.842Pr0.4 Nu = 0.00718Re0.926Pr0.3
Tube Y Nu = 0.015Re0.843Pr0.4 Nu = 0.00639Re0.944Pr0.3 -10%
Tube Z Nu = 0.00827Re0.906Pr0.4 Nu = 0.01014Re0.91Pr0.3 128

Reynolds number. Tube Z shows the highest Nusselt number obtained, 64

followed by Tube Y, Tube X, and the smooth tube. Tube Z experiences
the highest frictional resistance to the flow, and correspondingly, yields (b)
the highest Nusselt number. The degree of turbulence due to valleys 32
32 64 128 256 512 1024
and mountains on Tube Z causes the Nusselt number to increase. Tube Y
has a higher Nusselt number than Tube X because of the smaller pitch NuWilson
‘p’ and higher ridge height ‘RH’, as shown in Table 2. The same trends
are followed for both evaporator conditions (Cooling) and absorber Fig. 3. Smooth tube results of absorber conditions (Heating) (a) Wilson plot,
conditions (Heating). and (b) Comparison with Gnielinski correlation.
Table 6 shows the EF, as defined in Eq. (16), and PF of the enhanced
tubes. It is clearly shown that EFs increase with PFs; Tube Z has the 1.1x10
-2

highest EF and PF.

(hA)
EF = 9.8x10
-3
(hA)s (16)
+1%
(Rov)regression[K/W]

4.5. Effect of property variations 8.4x10

-3
-1%
The effect of property variations is studied according to Zukauskas
[37], and all correlations are modified as shown in Table 7. The tech- 7.0x10
-3

nique requires to find the Prandtl number at the wall temperature. The
wall temperature is found out using the following procedure:
-3
5.6x10
1. Finding hi using Eq. (4), and calculate the wall temperature as
shown. For heating conditions:
-3
4.2x10
Q 4.2x10
-3
5.6x10
-3
7.0x10
-3
8.4x10
-3
9.8x10
-3
1.1x10
-2
Tw = + Tf
hi As (17)
(Rov)experimental [K/W]
Where Tf is bulk fluid temperature.
Fig. 4. Comparison of overall thermal resistance of smooth absorber tube
2. Calculate the ratio (Pr/Prw), and introduce the ratio into Eqs. (7–9). (Heating).
3. Repeat the non-linear regression scheme to find the unknowns.
4. Repeat the process until the required convergence. The obtained

6
S.M. Ammar and C.W. Park International Communications in Heat and Mass Transfer 118 (2020) 104819

1.2x10
-2
34
Tube X, a= 0.842 Tube X
-2 Tube Y, a=0.843 Tube Y
1.1x10 17 Tube Z
Tube Z, a=0.906
-3
9.2x10 0

f(a) 10 (K /W )
3

2
84
0.

06
42
1/UA [K/W]

a=

0.9
0.8
-3
7.9x10 -17

a=

a=
-10
-3
6.6x10 -34

-3
5.3x10 -51

-3
4.0x10 -68

-3
(a) (a)
2.6x10
0.0 1.7x10
-5
3.4x10
-5
5.0x10
-5
6.7x10
-5
8.4x10
-5 -85
0.78 0.81 0.84 0.87 0.90 0.93 0.96 0.99
a 0.4
1/Ret Pr (Ak/Dh) [K/W] a
-2
1.7x10 52.2
Tube X, a= 0.926 Tube X
Tube Y, a=0.944 Tube Y
Tube Z, a=0.91 34.8
1.4x10
-2 Tube Z

17.4
1
1/UA [K/W]

f(a) 10 (K /W )
0.9
2
-2
1.1x10
0.0 a=
26 44
2

0.9 0.9
8.4x10
-3
-17.4 a= a=
-10

-34.8
-3
5.6x10

(b) -52.2

2.8x10
-3 (b)
0.0 2.2x10
-5
4.5x10
-5
6.7x10
-5
9.0x10
-5
1.1x10
-4
-69.6
0.87 0.88 0.90 0.92 0.94 0.95 0.97 0.99
a 0.3
1/Ret Pr (Ak/Dh) [K/W] a
Fig. 5. Wilson plot results for enhanced tubes: (a) Heating and (b) Cooling. Fig. 6. Bisection method to find Reynolds exponent ‘a’: (a) Heating and (b)
Cooling.

value is the required wall temperature.

NuGnielinski, new
The modified correlations do not have a significant change in the (f /8)(Re 1000) Pr Di 2/3
Pr
0.11 0.77

results, as the Nusselt correlation results are changed by about only 6% = 0.3Re0.22 1+
1 + 12.7(f /8)1/2 (Pr 2/3 1) L Prw
due to the small temperature difference between the wall and the bulk Gnielinski

fluid. (20)
The correlation is obtained by introducing an extra Reynolds
4.6. New correlation number term in the Gnielinski correlation. As shown in Fig. 9, the
correlation suggested the enhanced tube results well with an agreement
The obtained Wilson plot Nusselt number is compared with the fa- within ± 5%, having a total mean error of 2.1%.
mous Gnielinski correlation. The absolute and mean errors are calcu- The new Gnielinski correlation is based on the current shapes of
lated as: fabricated tubes. The correlation is valid for the present operating
NuGnielinski NuWilson conditions, as shown in Table 1, of heating and cooling fluids.
=
i
NuWilson (18)
5. Conclusions
N
1
abs = i, abs The current study used a non-linear regression scheme to investigate
N (19)
i=1
the tube-side heat-transfer coefficient of three newly fabricated tubes
As shown in Fig. 8, the results of fabricated tubes are satisfied with under practical operating conditions of the absorber (Heating) and
an error of ± 20%. The total mean error as calculated using Eq. (18) is evaporator (Cooling). The important findings are listed as below:
14.7%. As the original Gnielinski correlation is based on a smooth tube,
we suggested a new Nusselt correlation based on 103 data points of 1. Gnielinski correlation, which is famous to investigate the Nusselt
both heating and cooling conditions for enhanced tubes and is shown correlation of a smooth tube, is used to validate the results of the
as: enhanced tube for both heating and cooling fluids. The results were

7
S.M. Ammar and C.W. Park International Communications in Heat and Mass Transfer 118 (2020) 104819

749 1024
Smooth Smooth
Tube X Tube X
642
Tube Y Tube Y +20%
Tube Z 512 Tube Z
535

-20%

NuGnielinski
428
NuWilson

256
321

214
128

107
(a) (a)
0 64
0 21428 42856 64284 85712 107140 128568 149996 64 128 256 512 1024
NuWilson
Re
644
1024
Smooth
Smooth
Tube X Tube X
552
Tube Y
512
Tube Y +20%
Tube Z Tube Z
460
NuGnielinski
368 256
NuWilson

-20%
276
128

184

64
92

(b) (b)
0 32
0 12857 25714 38571 51428 64285 77142 89999 32 64 128 256 512 1024

Re NuWilson
Fig. 7. Comparison of tubes in terms of Nusselt number: (a) Heating and (b) Fig. 8. Comparison with Gnielinski correlation: (a) Heating and (b) Cooling.
Cooling.

2. Briggs and Young method failed to predict the solution of Reynolds

Table 6
exponent ‘a’. In fact, it was observed that the value of ‘a’ diverges
Heat transfer EF and friction PF of enhanced tubes.
even at the true value.
Tubes Heating Cooling 3. The obtained experimental Nusselt number as calculated by the
Wilson plot method showed a reasonable thermal comparison be-
EF PF EF PF
tween different enhanced tubes and the smooth tube. The EFs in-
Tube X 1.08 1.19 1.09 1.19 crease with PFs.
Tube Y 1.11 1.37 1.19 1.43
Tube Z 1.28 1.61 1.28 1.67

Declaration of Competing Interest

satisfied with a total mean error of 14.7% and within ± 20%. We

developed a new Nusselt correlation for the current fabricated tubes The authors declare that they have no known competing financial
using the original Gnielisnki correlation, and comparison of a new interests or personal relationships that could have appeared to influ-
correlation with Wilson plot data showed an agreement of ± 5%. ence the work reported in this paper.

Table 7
Modified Nusselt correlations including the effect of property variations.
Tubes Heating Cooling

Smooth Nu = 0.0213Re0.8Pr0.4(Pr/Prw)0.25 Nu = 0.0112Re0.88Pr0.3(Pr/Prw)0.25

Tube X Nu = 0.0126Re0.853Pr0.4(Pr/Prw)0.25 Nu = 0.00834Re0.914Pr0.3(Pr/Prw)0.25
Tube Y Nu = 0.013Re0.853Pr0.4(Pr/Prw)0.25 Nu = 0.00743Re0.932Pr0.3(Pr/Prw)0.25
Tube Z Nu = 0.0071Re0.917Pr0.4(Pr/Prw)0.25 Nu = 0.0118Re0.898Pr0.3(Pr/Prw)0.25

8
S.M. Ammar and C.W. Park International Communications in Heat and Mass Transfer 118 (2020) 104819

1024 [10] C.W. Park, S.M. Ammar, Personal Communication with Samjung Tech Co., Ltd in,
(2019).
New Correlation [11] F.W. Dittus, L.M.K. Boelter, Heat Transfer in Automobile Radiators of the Tubular

512
+5% Type, University of California Press, Berkeley, Calif, 1930.
[12] E.N. Sieder, G.E. Tate, Heat transfer and pressure drop of liquids in tubes, Industrial
& Engineering Chemistry 28 (1936) 1429–1435.
[13] B. Petukhov, Heat transfer and friction in turbulent pipe flow with variable physical
properties, Advances in Heat Transfer, Elsevier, 1970, pp. 503–564.
-5%
NuGnielinski,new

256 [14] V. Gnielinski, New equations for heat and mass transfer in turbulent pipe and
channel flow, Int. Chem. Eng. 16 (1976) 359–368.
[15] L.M. Tam, A.J. Ghajar, Transitional heat transfer in plain horizontal tubes, Heat
Transfer Engineering 27 (2006) 23–38.
128 [16] J. Olivier, J.P. Meyer, Single-phase heat transfer and pressure drop of the cooling of
water inside smooth tubes for transitional flow with different inlet geometries (RP-
1280), HVAC&R Research 16 (2010) 471–496.
[17] J.P. Meyer, J.A. Olivier, Heat transfer and pressure drop characteristics of smooth
64 horizontal tubes in the transitional flow regime, Heat Transfer Engineering 35
(2014) 1246–1253.
[18] J. Chen, H. Müller-Steinhagen, G.G. Duffy, Heat transfer enhancement in dimpled
tubes, Appl. Therm. Eng. 21 (2001) 535–547.
32 [19] P. Sivashanmugam, S. Suresh, Experimental studies on heat transfer and friction
32 64 128 256 512 1024 factor characteristics of laminar flow through a circular tube fitted with helical
screw-tape inserts, Appl. Therm. Eng. 26 (2006) 1990–1997.
[20] N.-H. Kim, M.-G. Go, Tube-side heat transfer and friction characteristics of titanium
NuWilson corrugated tubes used for steam condensers, J. Mech. Sci. Technol. 32 (2018)
4535–4543.
[21] W.-T. Ji, D.-C. Zhang, Y.-L. He, W.-Q. Tao, Prediction of fully developed turbulent
Fig. 9. Comparision of new Nusselt correlation with Gnielinski correlation.
heat transfer of internal helically ribbed tubes–an extension of Gnielinski equation,
Int. J. Heat Mass Transf. 55 (2012) 1375–1384.
[22] E.E. Wilson, A basis for rational design of heat transfer apparatus, The J. Am. Soc.
Acknowledgments
Mech. Engrs. 37 (1915) 546–551.
[23] S.M. Ammar, N. Abbas, S. Abbas, H.M. Ali, I. Hussain, M.M. Janjua, U. Sajjad,
This work was supported by the Korea Institute of Energy A. Dahiya, Condensing heat transfer coefficients of R134a in smooth and grooved
multiport flat tubes of automotive heat exchanger: an experimental investigation,
Technology Evaluation and Planning (KETEP) and the Ministry of
Int. J. Heat Mass Transf. 134 (2019) 366–376.
Trade, Industry, and Energy (MOTIE) of the Republic of Korea [24] S.M. Ammar, N. Abbas, S. Abbas, H.M. Ali, I. Hussain, M.M. Janjua, Experimental
(20192010107020, 20172010105860, P0002131), by the National investigation of condensation pressure drop of R134a in smooth and grooved
Research Foundation of Korea funded by the Korean Government multiport flat tubes of automotive heat exchanger, Int. J. Heat Mass Transf. 130
(2019) 1087–1095.
(2019R1A2C2010607), and by “Leaders in Industry-university [25] J. Fernández-Seara, F.J. Uhía, J. Sieres, A. Campo, Experimental apparatus for
Cooperation +” Project (2001000783) supported by the Ministry of measuring heat transfer coefficients by the Wilson plot method, Eur. J. Phys. 26
Education and National Research Foundation of Korea. (2005) N1.
[26] J. Fernández-Seara, F.J. Uhía, J. Sieres, Laboratory practices with the Wilson plot
method, Experimental Heat Transfer 20 (2007) 123–135.
References [27] N.-H. Kim, J.-P. Cho, J.-O. Kim, B. Youn, Condensation heat transfer of R-22 and R-
410A in flat aluminum multi-channel tubes with or without micro-fins, Int. J.
Refrig. 26 (2003) 830–839.
[1] J.H. Lee, D.H. Kim, S.M. Kim, M.S. Kim, I.G. Kim, S.M. Woo, S.J. Hong, C.W. Park,
[28] N.-H. Kim, E.-J. Lee, H.-W. Byun, Evaporation heat transfer and pressure drop of R-
Heat transfer characteristics of a falling film generator for various configurations of
410A in flattened smooth tubes having different aspect ratios, Int. J. Refrig. 36
heating tubes in an absorption chiller, Appl. Therm. Eng. 148 (2019) 1407–1415.
(2013) 363–374.
[2] M. Attalla, H.M. Maghrabie, E. Specht, An experimental investigation on fluid flow
[29] L.J. Brognaux, R.L. Webb, L.M. Chamra, B.Y. Chung, Single-phase heat transfer in
and heat transfer of rough mini-channel with rectangular cross section, Exp.
micro-fin tubes, Int. J. Heat Mass Transf. 40 (1997) 4345–4357.
Thermal Fluid Sci. 75 (2016) 199–210.
[30] G.J. Zdaniuk, L.M. Chamra, P.J. Mago, Experimental determination of heat transfer
[3] Z. Hao, Z. Lan, Q. Wang, Y. Zhao, X. Ma, Heat and mass transfer enhancement for
and friction in helically-finned tubes, Exp. Thermal Fluid Sci. 32 (2008) 761–775.
falling film absorption with coated distribution tubes at high temperature, Exp.
[31] L. Herbert, U. Sterns, Heat transfer in vertical tubes—interaction of forced and free
Thermal Fluid Sci. 53 (2014) 147–153.
convection, The Chemical Engineering Journal 4 (1972) 46–52.
[4] A.O. Adelaja, D.R. Ewim, J. Dirker, J.P. Meyer, Heat transfer, void fraction and
[32] E. Gorgy, S. Eckels, Convective boiling of R-134a on enhanced-tube bundles, Int. J.
pressure drop during condensation inside inclined smooth and microfin tubes, Exp.
Refrig. 68 (2016) 145–160.
Thermal Fluid Sci. 109 (2019) 109905.
[33] A. Jokar, S.J. Eckels, M.H. Hosni, Single-phase flow in meso-channel compact heat
[5] H. Bas, V. Ozceyhan, Heat transfer enhancement in a tube with twisted tape inserts
exchangers for air conditioning applications, Heat Transfer Engineering 31 (2010)
placed separately from the tube wall, Exp. Thermal Fluid Sci. 41 (2012) 51–58.
3–16.
[6] S.K. Gupta, R.D. Misra, An experimental investigation on flow boiling heat transfer
[34] D. Briggs, E.H. Young, Modified Wilson plot techniques for obtaining heat transfer
enhancement using cu-TiO2 nanocomposite coating on copper substrate, Exp.
correlations for shell and tube heat exchangers, Chemical Engineering Progress
Thermal Fluid Sci. 98 (2018) 406–419.
Symposium Series, AIChE, New York, NY, 1969, pp. 35–45.
[7] K. Nilpueng, S. Wongwises, Experimental study of single-phase heat transfer and
[35] H. Khartabil, R. Christensen, An improved scheme for determining heat transfer
pressure drop inside a plate heat exchanger with a rough surface, Exp. Thermal
correlations from heat exchanger regression models with three unknowns, Exp.
Fluid Sci. 68 (2015) 268–275.
Thermal Fluid Sci. 5 (1992) 808–819.
[8] A. Garcia, J. Solano, P. Vicente, A. Viedma, The influence of artificial roughness
[36] S.J. Kline, Describing uncertainty in single sample experiments, Mech, Engineering
shape on heat transfer enhancement: corrugated tubes, dimpled tubes and wire
75 (1953) 3–8.
coils, Appl. Therm. Eng. 35 (2012) 196–201.
[37] A. Žukauskas, Heat transfer from tubes in Crossflow, in: J.P. Hartnett, T.F. Irvine
[9] F. Dittus, L. Boelter, Heat transfer in automobile radiators of the tubular type,
(Eds.), Advances in Heat Transfer, Elsevier, 1972, pp. 93–160.
International Communications in Heat and Mass Transfer 12 (1985) 3–22.