You are on page 1of 10

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/299033657

Single phase heat transfer and pressure drop analysis of a dimpled enhanced
tube

Article  in  Applied Thermal Engineering · March 2016


DOI: 10.1016/j.applthermaleng.2016.03.042

CITATIONS READS

22 419

4 authors:

Ming Li Tariq S. Khan


Khalifa University Higher Colleges of Technology
3 PUBLICATIONS   36 CITATIONS    50 PUBLICATIONS   273 CITATIONS   

SEE PROFILE SEE PROFILE

Ebrahim Al Hajri Zahid H. Ayub


Khalifa University Isotherm, Inc.
38 PUBLICATIONS   206 CITATIONS    129 PUBLICATIONS   1,048 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

The Investigation of Subsurface Heat Transfer in a Set of Different Thermo-geological Conditions for Ground Source Cooling View project

Novel shell and tube exchanger with shell side interstitial turbulators for viscous fluids View project

All content following this page was uploaded by Zahid H. Ayub on 15 July 2016.

The user has requested enhancement of the downloaded file.


Applied Thermal Engineering 101 (2016) 38–46

Contents lists available at ScienceDirect

Applied Thermal Engineering


j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / a p t h e r m e n g

Research Paper

Single phase heat transfer and pressure drop analysis of a dimpled


enhanced tube
Ming Li a, Tariq S. Khan a,*, Ebrahim Al-Hajri a, Zahid H. Ayub b
a Department of Mechanical Engineering, The Petroleum Institute, Abu Dhabi, United Arab Emirates
b Isotherm Inc., Arlington, USA

A R T I C L E I N F O A B S T R A C T

Article history: Thermal-hydraulic performance of an enhanced tube was evaluated using experimental and numerical
Received 26 October 2015 simulation techniques in a pipe-in-pipe heat exchanger. Steady state single phase (liquid-to-liquid) ex-
Accepted 5 March 2016 periments were performed to determine Nusselt number and friction factor. Experiments with water as
Available online 19 March 2016
working fluid were carried out in the Reynolds number range of (500 < Re < 8000), while for water/
glycol solution based experiments the Reynolds number range was kept at (150 < Re < 2000). A non-
Keywords:
dimensional performance evaluation criterion (PEC) was used to assess the thermal-hydraulic performance
Enhanced tube
of heat transfer enhancement achieved with the enhanced tube. Based on the experimental data, Nusselt
Modified Wilson plot method
PEC number and friction factor estimation correlations were proposed for the enhanced tube.
CFD Simulations were carried out to obtain heat transfer and pressure drop characteristics of smooth and
Enhancement mechanism enhanced tubes, using commercial CFD software, Fluent. Realizable k − ε model was employed to assess
the influence of dimples on turbulent flow and velocity field. Simulation results showed that dimples
disturb and mix boundary layers and generate secondary flows that improve turbulence level. Compar-
ison between experimental and numerical simulation results showed good agreement as the data fell
within ±10% error band.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction ferent characteristic parameters in a pipe-in-pipe heat exchanger.


The working fluid was water at Reynolds number 7500–52,000. They
Heat exchangers have extensive use in almost all major indus- proposed correlations for predicting heat transfer coefficient and
tries worldwide. By applying heat enhancement techniques, thermal friction factor. Wang et al. [5] investigated two enhanced tubes with
system efficiency and performance can be greatly improved, thereby ellipsoidal and spherical dimples on the tube surface. Air was used
reducing the size of thermal system and both capital and opera- as internal working fluid and water flowed through annular side of
tional costs. Enhanced heat transfer performance and operational pipe-in-pipe heat exchanger. They reported the ellipsoidal dimple
flexibility are two very important aspects of modern heat exchang- tubes to provide better heat exchange performance compared to
ers. Effective use of heat exchangers in various industries requires spherical dimple tube. Garcia et al. [6] compared three different heat
accurate determination of heat transfer and pressure drop. transfer enhancement techniques, using corrugated tubes, dimpled
Bergles et al. [1] compiled an extensive bibliography on aug- tubes and wire coils. The authors recommended using smooth tube
mentation of convective heat and mass transfer. Rao et al. [2] for Reynolds number less than 200. For higher Reynolds numbers
investigated the rectangular channels with staggered arrays of between 200 and 2000, the wire coil technique was preferred while
dimples/cavities and examined the mechanism of heat transfer en- for Reynolds number beyond 2000, dimpled and corrugated tubes
hancement. Using air as working fluid they suggested that dimples were reported more advantageous.
disturbed the boundary layer thus increasing the near wall turbu- Computational fluid dynamics (CFD) simulations can provide
lence level producing higher convection coefficient. Kukulka and good prediction on thermal-hydraulic performance of enhanced
Smith [3] conducted experiments on dimpled enhanced tubes in tube heat exchangers. Xie and Sundén [7] undertook numerical
a pipe-in-pipe heat exchanger. They also reported a considerable simulation of a rectangular channel with dimples on the bottom
heat transfer enhancement. Chen et al. [4] reported experimental surface. They used realizable k–ɛ turbulence model. Heat transfer
study on thermo-hydraulic performance of dimpled tubes with dif- enhancement with dimpled channel was reported up to 200%
with only 5% extra pressure drop penalty. Similarly Bi et al. [8]
studied a mini-channel with dimples and cylindrical grooves. The
* Corresponding author. Tel.: +971 2 607 5421; fax: +971 2 607 5429. dimples were reported to provide higher heat transfer perfor-
E-mail address: tkhan@pi.ac.ae (T.S. Khan). mance. Fan and Yin [9] reported that the distance between dimples

http://dx.doi.org/10.1016/j.applthermaleng.2016.03.042
1359-4311/© 2016 Elsevier Ltd. All rights reserved.
M. Li et al. / Applied Thermal Engineering 101 (2016) 38–46 39

Nomenclature

A heat transfer surface area, m2 ΔTm logarithmic mean temperature, K


C constant in Eqs. (8) and (12) u velocity, m s−1
C1 constant in Eqs. (6), (10) and (11) U overall heat transfer coefficient, W m−2 K−1
C2 constant in Eqs. (9)–(12)
cp specific heat, J kg−1 K−1 Greek symbols
d dimple diameter, mm Δ change or difference
di diameter of inner tube, m μ dynamic viscosity, Pa S
Dh hydraulic diameter, m v kinematic viscosity, m2 /s
f friction factor ( 2ΔpDh ρu2L ) ρ density, kg m−3
hi convective heat transfer coefficient, W m−2 K−1 ∀ volume flow rate, m3 s−1
h dimple depth, mm
k thermal conductivity, W m−1 K−1 Subscripts
L length of tube, m a enhanced
m exponent of Reynolds number c cold
m  mass flow rate, kg s−1 corr correlated
Nu Nusselt number ( hi Dh k ) exp experimental
Pr Prandtl number ( c p μ k ) h hot
p dimple pitch, mm i inside
Δp pressure drop, Pa max maximum
Q heat transfer rate, W num numerical
Re Reynolds number ( ρuDh μ ) o outside
R thermal resistance, K/W ov overall
t wall thickness, m s smooth
T temperature, K

can have a considerable effect on heat transfer enhancement. ature difference, heat exchanger was configured in a counter-
They showed that the in-line dimple configuration resulted in current flow arrangement. The desired temperature of water at the
better performance compared to staggered arrangement. Ağra inlet of the heat exchanger was controlled to an accuracy of ±0.1 °C.
et al. [10] determined the heat transfer and pressure drop of The capacity of chiller was 3 kW at −10 °C and the capacity of heater
corrugated and helically finned tubes. Realizable k–ɛ model was was 2.2 kW. The range of Reynolds number for internal flow was
applied and unstructured hybrid mesh was generated near the from 500 to 8000 for water and 150–2000 for glycol/water solu-
wall to resolve the velocity and thermal gradients. The heat tion. The flow rate was controlled through a by-pass valve. The length
transfer coefficient for helically finned tube was reported highest of test section was 1.0 m with an upstream calming length of 2.0 m.
followed by corrugated and smooth tubes. The entire apparatus was insulated with polyurethane and alumi-
Thermal-hydraulic characteristics of enhanced tube heat ex- num foil to minimize heat losses.
changers can be utilized in the industrial applications only if accurate The K-type thermocouples with an accuracy of ±0.15 °C were care-
correlations are available for Nusselt number and frictional pres- fully calibrated with a calibrated reference RTD (resistance thermal
sure drop with all the necessary details. The above literature survey detector). The flow rate was measured by turbine flow meter with
shows that only limited studies are conducted on enhanced tubes an accuracy of ±1% of reading. Pressure drop was measured by a
with discrete dimples/protrusions both experimentally and numer- digital differential pressure transducer with an accuracy of ±0.25%
ically. Therefore, lack of data is still a barrier in the use of such tubes of full scale.
in heat exchangers. Primary aim of current study was to investi-
gate thermal-hydraulic performance of a pipe-in-pipe heat exchanger
2.2. Enhanced tube
built with dimpled tubes. The study focused on reporting heat trans-
fer and pressure drop correlations in low temperature applications.
Fig. 2(a) shows the enhanced tube used in current study. It is
Emphasis was given to describing the heat transfer enhancement
made of stainless steel and is a combination of random sand grain
mechanism.
rough surface and dimples. Discrete dimples are distributed on the
roughed surface of enhanced tube in staggered but in-line arrange-
2. Experimental details ment along the length of the tube. Geometrical parameters are
described in Fig. 2(b), while the values of these parameters are given
2.1. Experimental setup and procedure in Table 1.

Experiments were conducted on a horizontal pipe-in-pipe heat


exchanger. The inner tube was an enhanced tube made of stain- 2.3. Data reduction
less steel. The experimental schematic is presented in Fig. 1. The
experimental apparatus consisted of pipe-in-pipe heat exchanger, The energy balance for heat exchanger is given as Eq. (1). As a
heater, chiller, pumps, turbine flow meters, differential pressure first step to validate experiments, the difference of heat transfer rate
transducer, thermocouples and data acquisition system. Hot water between hot side and cold side were controlled within 5%, Eq. (2).
was pumped through the inner tube while chilled water flowed on
the annular side made of acrylic pipe. To obtain a higher temper- Q = (mC
 p ) (Th1 − Th2 ) = (mC
 p ) (Tc 2 − Tc1 ) (1)
40 M. Li et al. / Applied Thermal Engineering 101 (2016) 38–46

Fig. 1. Experimental schematic.

(a) Cross section and outer surface

(b) Geometrical parameters

Fig. 2. Details of enhanced tube.

Table 1 The overall thermal resistance Rov can be written as the sum of
Geometrical characteristics. resistances:
di (mm) t (mm) d (mm) p (mm) h (mm) Configuration
Rov = Ro + R fo + Rt + R fi + Ri (5)
17.272 0.889 5 10 1.2 Staggered

where Ro is the external resistance, R fo is the external fouling re-


Qh − Qc sistance, Rt is the wall resistance, R fi is the internal fouling resistance
 5% (2)
and Ri is the internal resistance.
Qh
Since the internal flow rate is the control variable in this exper-
The overall heat transfer coefficient U and logarithmic mean tem- imentation, so the resistance due to internal and external fouling,
perature difference ΔTm are: external convection and tube wall are considered as constant.

Q Ro + R fo + Rt + R fi = C1 (6)
U= (3)
AΔTm
Eq. (5) may also be written in more elaborate form as following:
ΔT1 − ΔT2
ΔTm = (4) 1 1 1 ln (do di ) 1 1
ln ( ΔT1 ΔT2 ) = + + + + (7)
U o Ao ho Ao h fo Ao 2π kt Lt h fi Ai hi Ai
Modified Wilson plot method was used for heat transfer data
reduction analysis. It provides an excellent way to solve the con- According to the correlation of convective heat transfer coeffi-
vective heat transfer coefficient [11]. The basic calculation procedure cient proposed by Dittus and Boelter [12], Eq. (8), the internal
of this method is described below. convection resistance Ri can be written as Eq. (9).
M. Li et al. / Applied Thermal Engineering 101 (2016) 38–46 41

⎛k ⎞ on the working fluid in simulations. Fig. 3 shows the numerical sche-


hi = CRem Pr 0.4 ⎜ l ⎟ (8) matic. For boundary conditions, velocity range was calculated from
⎝ di ⎠
the range of Reynolds number, 2300–15,000, with inlet condition
as fully developed flow. Constant heat flux 1 × 104 w/m2 which is
1
Ri = C 2 (9) in the range of experimental heat flux 8000–15,000 w/m2 and no
Rem
slip shear condition were imposed on the dimpled surface. The re-
Above equations show that the overall thermal resistance can be alizable k − ε model was employed to investigate influence of dimples
obtained as a linear function of 1 Rem . on thermal-hydraulic performance of the tube.

1
Rov = C1 + C 2 (10)
Rem 3.1. Governing equations

Rearranging and applying logarithms to Eq. (10): The governing equations of fluid flow and heat transfer can be
expressed as follows:
⎛ 1 ⎞ ⎛ 1⎞
ln ⎜ = ln ⎜ ⎟ + m ln ( Re ) (11)
⎝ Rov − C1 ⎟⎠ ⎝ C2 ⎠
• Continuity equation
The following three steps can be used to obtain the values of con- ∂u j
stants m, C1 and C 2: =0 (14)
∂x j
(1) Assume the value of m , then the value of C1 and C 2 can be
obtained from Eq. (10) by linear regression. • Momentum equation
(2) The value of C1 and C 2 are substituted into Eq. (11), then the
∂ ∂p ∂ ∂u ∂u
new value of m can be obtained since it is the slope of line. (ρui u j ) = − + ⎡⎢(μ + μt ) ⎛⎜⎝ i + j ⎞⎟⎠ ⎤⎥ (15)
(3) Repeat steps 1 and 2 until the new value of m does not change ∂x j ∂xi ∂x j ⎣ ∂xi ∂xi ⎦
with respect to previous value.
• Energy equation
Once the values of constants C1 , C 2 and m are determined from
above iteration the convective heat transfer coefficient can be ob- ∂ui T ∂ ⎛ ⎛ μ μt ⎞ ∂T ⎞
= ⎜ + ⎟ (16)
tained from Eq. (8). The value of constant C is calculated as: ∂xi ∂x j ⎜⎝ ⎝ Pr Pt r ⎠ ∂j ⎟⎠

1
C= • Equations for turbulent kinetic energy κ and rate of energy dis-
⎛ k ⎞ 0.4 (12)
C 2 ⎜ ⎟ Pr Ai sipation ε :
⎝ di ⎠
∂ ∂ ⎡ μ ∂κ ⎤
In this study, isothermal friction factor was obtained for smooth (ρκ u j ) = ⎢⎛⎜⎝ μ + t ⎞⎟⎠ ⎥ + Γ − ρε (17)
and enhanced tubes using the Darcy friction factor: ∂x j ∂x j ⎣ σ k ∂x j ⎦

2Δp Dh ∂ ∂ ⎡ μ ∂ε ⎤ ε2
f=
ρui2 L
(13)
(ρε u j ) = ⎢⎛⎜⎝ μ + t ⎞⎟⎠ ⎥ + C 3Γε − C 4 (18)
∂x j ∂x j ⎣ σ ε ∂x j ⎦ κ + νε

where Γ is production rate of κ


2.4. Uncertainty analysis
∂ui μt ⎛ ∂ui ∂u j ⎞ ∂ui
Uncertainty analysis of experimental data was conducted ac- Γ = −ui u j = + (19)
∂x j ρ ⎜⎝ ∂x j ∂xi ⎟⎠ ∂x j
cording to procedure presented by Figliola and Beasley [13]. Based
on data reduction and instrumentation errors, error propagation was
conducted to estimate the uncertainty in heat transfer rate Q , Nusselt k2
μt = ρC μ (20)
number Nu and friction factor f .Table 2 gives determined uncer- ε
tainties in these parameters.
The coefficients appeared in above realizable κ − ε equations are as
following [14]
3. Numerical details
⎡ μt ⎤
C 3 = max ⎢0.43, , C 4 = 1.0, σ k = 1.0, σ ε = 1.2
Simulations were carried out using water as the working fluid. ⎣ μt + 5 ⎥⎦
Since the range of operating temperatures in current study was small,
the properties of water were assumed to be constant. Additional-
ly, in order to further minimize the effect of this assumption, 3.2. Parameter definition
properties obtained based on experimental outcome were imposed
The y + is the indicator of near wall mesh resolution which is
dependent on the Reynolds number and size of wall-adjacent cell.
Table 2 ⎛u ⎞
Uncertainty analysis. y+ = y ⎜ * ⎟ (21)
⎝ v⎠
Parameter Uncertainty

Q ±5% where y is the distance from the wall to the center of wall-adjacent
Nu ±10.5% cell and v is the kinematic viscosity. Friction velocity u can be cal-
f ±8% *
culated as following [15]:
42 M. Li et al. / Applied Thermal Engineering 101 (2016) 38–46

Fig. 3. Schematic for numerical simulations.

Table 3
12 Mesh independence test.
⎛ f⎞
u =⎜ ⎟ u (22)
* ⎝ 8⎠ Number Mass averaged Friction Friction Nusselt Nusselt
of cells fluid factor factor number number
temperature difference difference
The convective heat transfer coefficient is defined as [14] (°C) % %

Q 1,100,916 296.28 0.185 3.6% 190.7 1.8%


hi = (23) 1,495,576 296.28 0.189 1.6% 193.8 0.45%
Twall − Tref 1,935,620 296.30 0.190 1.0% 193.8 0.25%
2,974,920 296.31 0.192 Baseline 194.6 Baseline
where Tref is the mass-weighted average temperature of the fluid
and Twall is the local wall temperature.
The average Nusselt number was calculated by using the area-
weighted average of all local Nusselt numbers for the heating wall. 4. Results
The local Nusselt number of each cell was determined as:
4.1. Validation
hD
Nui = i h (24)
k 4.1.1. Validation of experiment
Experimental results for smooth tube were validated with em-
To determine the pressure drop of internal flow, the Darcy friction pirical correlations of Gnielinski [16] and Petukhov [17] for Nusselt
factor, Eq. (13) was applied. number and friction factor, respectively. These two correlations along
with their valid ranges are presented in Eqs. (25) and (26) below.
3.3. Mesh independence test
( f 8) (Re − 1000) Pr
Nu = 1
f 2⎛ 2 ⎞
1 + 12.7 ⎛⎜ ⎞⎟ ⎜ Pr 3 − 1⎟
As shown in Fig. 4, unstructured hybrid grid system was em-
ployed in the numerical model. Fully developed flow can be divided ⎝ 8⎠ ⎝ ⎠ (25)
into three regions; the viscous sublayer, the overlap region and the
⎡ 0.5  Pr  2000 ⎤
outer turbulent layer throughout the center portion of the flow. To ⎢3000  Re  5 × 106 ⎥
resolve the sublayer, near wall meshes should be fine enough, usually ⎣ ⎦
indicator of near wall mesh resolution y + ≈ 1. Thus the height of wall-
f = (0.790 ln Re − 1.64 )
−2
adjacent cell can be determined by Eqs. (21) and (22). In the current (26)
study, value of y was set to 0.03 mm while the growth factor of cell [3000  Re  5 × 10 ] 6

in boundary layer was set as 1.1, i.e., height of first cell y1 = 0.03 mm ,
height of second cell y2 = 0.033 mm and y10 = 0.03 × 1.19 mm . Fig. 5(a) and (b) shows the comparison of experimental Nusselt
Table 3 shows results of some simulations run with different number number and friction factor with Gnielinski and Petukhov correla-
of cells to check for grid independence. Approximately two million tions, respectively. Since the valid range of Reynolds number for these
grids were found to give good estimation. two correlations starts from Re equal to 3000, the comparison was

Fig. 4. Grid system of enhanced tube.


M. Li et al. / Applied Thermal Engineering 101 (2016) 38–46 43

70 160

60
Smooth tube, Exp Enhanced tube, Exp +10.5%
140
Gnielinski [16] Smooth tube, Exp
50 120
Enhanced tube, Num
100 -10.5%
40
Nu

Nu
80
30 +10.5%
60
20
40 -10.5%
10 20

0 0
0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000
Re Re
(a) Nusselt number (a) Comparison of Nusselt number for enhanced
and smooth tubes (Tube side: water)
0.14

0.12
Smooth tube, Exp
Petukhov [17]
0.1
64/Re
0.08
f

0.06

0.04

0.02

0 (b) Velocity distribution at cross section


0 2000 4000 6000 8000 10000
Re
(b) Friction factor Dimple

Fig. 5. Comparison of current experimental results with previous correlations.

conducted in the range of 3000  Re  8000 . Fig. 5(a) shows Nusselt


number to increase with increasing Reynolds number due to im-
proved turbulent effect with increasing Reynolds number. Fig. 5(b)
compares smooth tube experimental results with Petukhov corre-
lation. For completeness, friction factor results for laminar region (c) Velocity vectors around a dimple
are also included. The friction factor in this region is determined
Fig. 6. Variation of Nusselt number with Reynolds number and enhancement
by: mechanism.

64
f=
Re (27)
[0  Re  2300] results were further validated by comparing them with a previous
study by Chen et al. [4]. The deviation of Nusselt number is found
It can be seen that there is rapid friction factor drop in laminar to lie within ±20% and that of friction factor is within ±15%. Further
region while it reduces gradually with Reynolds number in the tur- details of this validation are given in [18].
bulent region. These figures illustrate that the deviations between
experimental data and empirical correlations for both Nusselt 4.2. Heat transfer
number and friction factor were within 10%, which means the ac-
curacy of current experimental approach for smooth tube is Fig. 6(a) shows the comparison of Nusselt number for en-
acceptable, hence validating the current experimental approach. hanced and smooth tubes. Hot water (32 °C) was used as internal
fluid and cold water (21.5 °C) flowed in the annular section. It is
4.1.2. Validation of simulation visible that the Nusselt number increases with Reynolds number.
Validation of simulation was conducted by comparing the pre- Enhanced tube shows a clear heat transfer advantage compared to
dicted thermal-hydraulic performance of smooth tube and enhanced smooth tube, especially at higher Reynolds numbers beyond 3000.
tube with empirical correlations and experimental results. The de- Enhancement of more than 200% is observed for the enhanced
viation between predicted Nusselt number and Gnielinski correlation tube at Re = 5000 . The trend continues with Reynolds number. In
[16] was within 20% and the deviation for friction factor was less laminar and transient region ( 500 < Re < 4000 ), enhancement ratio
than 10% in the range of Reynolds number 8000–14,000. Addition- increases faster with Reynolds number than in turbulence region
ally, with increasing Reynolds number, the accuracy of both Nusselt ( 4000 < Re < 8000) where maximum yet stable enhancement is
number and friction factor improved. It may be noted that the k − ε observed.
model applied in current simulations is a turbulent model which Heat transfer characteristics of enhanced tube were also simu-
over-predicts thermal performance in transient flow. Simulation lated numerically. Fig. 6(a) also shows comparison between
44 M. Li et al. / Applied Thermal Engineering 101 (2016) 38–46

numerical and experimental results for Nusselt number. Since tur- ding from the dimple. The average near wall region is higher than other
bulence model was used in the simulation part, it was regions. These characteristics along with flow disturbance caused by
understandable to observe an over-prediction in the range of dimples enhance the overall heat transfer performance.
500 < Re < 2300 . For this reason the comparison between Experiments were also conducted to investigate the perfor-
numerical and experimental results was made in the range of mance of enhanced tube at low temperatures. For this purpose, 25%
2300 < Re < 9000 . The difference between geometries of enhanced glycol/water solution was used in the annular section to obtain low
tube used in current experimental study and numerical model temperature ( Tc,inlet = −6.7 °C ). Different hot water inlet tempera-
applied in simulation is that the enhanced tube used in experi- tures with Prandtl number between 5.2 and 8.8 were used. Fig. 7(a)
mentation is a combination of dimples and roughed surface while shows experimental Nusselt number versus Reynolds number for
the numerical model takes into account only the dimples with several inlet temperature difference, ΔT = Th,inlet − Tc ,inlet . Increasing
smooth tube surface. Owing to the complexity of generating random trend in Nusselt number is observed with Reynolds number, similar
sand grain roughness, the tube surface has been assumed as smooth to water experiments.
for simulations. The comparison presented in these figures show Another set of low temperature experiments were carried out
a good agreement between experimental and numerical results. Like- with 20% glycol/water solution on the tube side and 25% glycol/
wise, comparison of enhanced tube results with equivalent size water solution ( Tc,inlet = −7.6 °C) on annular side. Three sets of
smooth tube show clear advantage of enhancement. This enhance- experiments with ΔT , 14.2 °C, 17.2 °C and 20.2 °C, were con-
ment implies that dimples play a dominant role in heat transfer ducted. Fig. 7(b) shows the experimental results in terms of Nusselt
augmentation that can be attributed to the fact that dimples disturb number and Reynolds number. It is obvious from Fig. 7(a) and (b)
streamlines of flow, change equivalent diameter of the tube and that for different ΔT, there is insignificant effect on Nusselt number;
augment heat transfer area. indicating higher temperature efficiency. This aspect is highly de-
The mechanism of heat transfer enhancement is further elabo- sirable in an efficient heat exchanger.
rated in Fig. 6(b) and (c). It is clear that the dimples disturb and mix
boundary layer and generates secondary flows, therefore, improving 4.3. Pressure drop
turbulence. Fig. 6(b) shows that the velocity distribution in radial di-
rection is altered by the dimples due to space occupation within the Pressure drop is an important evaluation parameter in heat ex-
tube. Fig. 6(c) shows the streamlines around dimples. In axial direc- changers because it relates to pumping power requirement. It is
tion, local near wall velocity is reduced upstream of dimples and known that flow restrictors such as dimples, ribs and fins help
accelerated at the crown of dimples. Meanwhile, reverse flow is gen- improve turbulence. Isothermal pressure drop for both smooth and
erated in the immediate downstream of dimples. Due to accelerated enhanced tubes was measured at room temperature (20 °C). Fig. 8
fluid near crown of dimples and recirculation zone at the dimple down- presents friction factor as a function of Reynolds number. Reyn-
stream, the average wall temperature on dimple upstream is lower than olds number ranging from 500 to 8000 can be divided into three
that of dimple downstream. Similarly, the dimple downstream region regions, laminar, transient and turbulent. As expected, in laminar
represents the highest turbulent kinetic energy due to vortex shed- region the friction factor reduced rapidly with Reynolds number.
Data shows an early transition of flow for enhanced tube (Re < 2000)
where the friction factor increased sharply. In turbulent region
(Re > 3000) friction factor decreases slowly with Reynolds number.
160
Smooth tube (ΔT=24.7°C) Enhanced tube (ΔT = 16.7°C) A good agreement between friction factor of water and glycol so-
140 Enhanced tube (ΔT = 20.7°C) Enhanced tube (ΔT = 24.7°C) lution was observed. The pressure drop enhancement, increased with
Enhanced tube (ΔT = 29.7°C) Enhanced tube (ΔT = 36.7°C)
120 Enhanced tube (ΔT = 41.7°C) Reynolds number in 500 < Re < 2000 range and remained almost
100 constant (1.8) for Re > 2000 . Numerically determined friction factor
for enhanced tube is also included in Fig. 8. A good agreement (±10%)
Nu

80
was achieved between the numerically predicted friction factor and
60 experimental friction factor.
40

20 4.4. Correlations

0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Nusselt number is commonly expressed as a function of Prandtl
Re (Pr) and Reynolds (Re) numbers:
(a) Variation of Nusselt number with Reynolds number for various
heat exchanger inlet temperature differences (Tube side: water)
110 0.16
Smooth tube (ΔT = 20.2 °C) Enhanced tube (ΔT = 20.2 °C)
Smooth tube, Exp, water
0.14 Enhanced tube, Exp, water-glycol
90 Enhanced tube (ΔT = 17.2 °C) Enhanced tube (ΔT = 14.2 °C)
0.12 Enhanced tube, Exp, water
70 Enhanced tube, Num, water
0.1
Nu

50
0.08 +8%
f

30 0.06
-8%
10 0.04

0.02
-10
0 500 1000 1500 2000 2500 0
Re 0 2000 4000 6000 8000 10000
(b) Nusselt number versus Reynolds number (Tube side: glycol/water) Re

Fig. 7. Variation of Nusselt number with Reynolds number. Fig. 8. Friction factor versus Reynolds number.
M. Li et al. / Applied Thermal Engineering 101 (2016) 38–46 45

2.5 5.2 < Pr < 6.6 5.9 < Pr < 7.1 6.8 < Pr < 7.7
2.3 Pr = 6.0 Pr = 6.5 Pr = 7.3 Pr = 7.6 Pr = 8.3 7.3 < Pr < 8.0 8.0 < Pr < 8.5 8.6 < Pr < 8.8
Pr = 8.7 Pr = 23.5 Pr = 26.5 Pr = 29.1 22.6 < Pr < 25.0 25.4 < Pr < 28.5 28.2 < Pr < 30.7
2.1 120
1.9
100
1.7
Log Nu

1.5 80
1.3

Nu corr
+10%
1.1 60

0.9 -10%
40
0.7
0.5 20
1.5 2 2.5 3 3.5 4 4.5
Log Re 0
(a) 0 20 40 60 80 100 120
Nu exp
2.5 (a) Correlated Nusselt number versus
Pr = 6.0 Pr = 6.5 Pr = 7.3 Pr = 7.6 Pr = 8.3 experimental Nusselt number
2 Pr = 8.7 Pr = 23.5 Pr = 26.5 Pr = 29.1 0.16
Pr = 7.1
0.14
Log (Nu/Prn)

1.5
0.12 Pr = 19.2 +10%

1 0.1

f corr 0.08 -
0.5 10%
0.06

0.04
0
1.5 2 2.5 3 3.5 4 4.5 0.02
Log Re
(b) 0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
f exp
Fig. 9. Dimensionless representation of convective heat transfer results.
(b) Correlated friction factor versus experimental
friction factor
Nu = CRem Pr n (28)
Fig. 10. Comparison of proposed correlations with experimental results.

Since the value of constants C, m and n are independent of fluid prop-


erties; the straight lines in Fig. 9(a) corresponding to different Prandtl 3
Smooth tube, Exp, water
numbers are collapsed to a single line by plotting the results in terms
Enhanced tube, Exp, water
of the ratio, Nu Pr n , as shown in Fig. 9(b). 2.5 Enhanced tube, Exp, water glycol
Based on the experimental data, Nusselt number for the en- Enhanced tube, Num
hanced tube can be correlated as: 2 Raj et al. [22]
PEC

Sivashanmugam [23]
Nu = 0.033Re Pr 0.82 0.4

⎡ 5.2  Pr  30.7 ⎤ (29) 1.5

⎢500  Re  8000 ⎥
⎣ ⎦
1
where the values of lead constants and exponent of Reynolds number
were obtained from Fig. 9(b). Fig. 10(a) shows the comparison 0.5
between correlated and experimental Nusselt number. The error of 0 2000 4000 6000 8000 10000
Nusselt number correlation is within ±10% for all the experimen- Re
tal data.
Fig. 11. PEC versus Re for different enhanced tubes.
Separate correlations of friction factor for the following three
Reynolds number regions are proposed:

f = 5.3176Re −0.614 (500 < Re < 1300)) (30) 4.5. Performance evaluation

f = 0.0003Re0.7033 (1300 < Re < 3000) (31) Performance Evaluation Criteria (PEC) was used to assess the
overall performance of heat transfer and pressure drop enhance-
ment. These criteria proposed by Gee and Webb [19] is defined as
f = 0.3486Re −0.18 (3000 < Re < 8000) (32)
follows:
Fig. 10(b) shows comparison between correlated friction factor and
Nu Nus
experimental friction factor. Most of the data fall within an error PEC = (33)
range of ±5%. ( f f s )1 3
46 M. Li et al. / Applied Thermal Engineering 101 (2016) 38–46

Fig. 11 shows PEC for current study and other related work. The range • The best performance (PEC = 1.55) was obtained at Reynolds
of PEC for enhanced tube is from 1.1 to 1.55 for water based ex- number 3500–4500 for water.
periments. It may be noted that at low Reynolds number range, PEC • Glycol/water solution showed higher PEC in the Reynolds number
increases with Reynolds number starting from a low value of 1.1; range 150–2000.
meaning that the pressure drop penalty caused by enhanced tube • Numerical simulations adequately predicted the experimental
dominates the performance in this region. This is comparable with data within ±15%.
Guo et al. [20] who also reported the increase of PEC as a function • The enhanced tube is a combination of random sand grain rough-
of Reynolds number in laminar region. The best performance of en- ness and dimples. However, the dimples are found to have a
hanced tube is obtained for Reynolds number 3500–4500. Bergles dominant effect on performance enhancement.
and Morton [21] have mentioned transition region to provide • Non-dimensional performance evaluation criteria showed clear
maximum heat transfer enhancement. This phenomenon can be ex- advantage of enhanced tube over the smooth tube.
plained from the perspective of transient flow. In this region the
flow inter-switches between laminar and turbulent in an appar-
References
ently random fashion. The introduction of vortex generators
(dimples) apparently enhances this process, thus significantly im-
[1] A.E. Bergles, V. Nirmalan, G.H. Junkhan, R.L. Webb, Bibliography on
proving the convective heat transfer coefficient with relatively lower Augmentation of Convective Heat and Mass Transfer-II, Iowa State University,
pressure drop penalty. PEC determined based on simulation results ISU-ERI-Ames-84221, 1983.
is also included in Fig. 11. Results show close correspondence to ex- [2] Y. Rao, C. Wan, Y. Xu, An experimental study of pressure loss and heat transfer
in the pin fin-dimple channels with various dimple depths, Int. J. Heat Mass
perimental results.
Transf. 55 (2012) 6723–6733.
Fig. 11 also shows comparison of current experimental results [3] Kukulka, R. Smith, Thermal-hydraulic performance of Vipertex 1EHT enhanced
with results from Raj et al. [22] and Sivashanmugam [23]. Raj et al. heat transfer tubes, Appl. Therm. Eng. 61 (2013) 60–66.
[4] J. Chen, H. Müller-Steinhagen, G.G. Duffy, Heat transfer enhancement in dimpled
[22] performed the experiments on doubly enhanced tube in pipe-
tubes, Appl. Therm. Eng. 21 (2001) 535–547.
in-pipe heat exchanger. Sivashanmugam [23] conducted the [5] Y. Wang, Y.-L. He, Y.-G. Lei, J. Zhang, Heat transfer and hydrodynamics analysis
experiments on circular tube fitted with helical screw-tape inserts of a novel dimpled tube, Exp. Thermal Fluid Sci. 34 (2010) 1273–1281.
in Reynolds number range 200–3000. The helical screw-tape insert [6] A. Garcia, J. Solano, P. Vicente, A. Viedma, The influence of artificial roughness
shape on heat transfer enhancement: corrugated tubes, dimpled tubes and wire
showed superior performance in Reynolds number range of 500– coils, Appl. Therm. Eng. 35 (2012) 196–201.
1800. For Re > 1800, the enhanced tube used in current study gives [7] G. Xie, B. Sundén, Numerical predictions of augmented heat transfer of an
the best performance. On the other hand the doubly enhanced tube internal blade tip-wall by hemispherical dimples, Int. J. Heat Mass Transf. 53
(2010) 5639–5650.
of Raj et al. [22] relatively showed lower performance due to lower [8] C. Bi, G.H. Tang, W.Q. Tao, Heat transfer enhancement in mini-channel heat sinks
heat transfer enhancement. It is worth noting that in this study the with dimples and cylindrical grooves, Appl. Therm. Eng. 55 (2013) 121–132.
PEC for glycol/water based experiments showed clear advantage over [9] Q. Fan, X. Yin, 3-D numerical study on the effect of geometrical parameters on
thermal behavior of dimple jacket in thin-film evaporator, Appl. Therm. Eng.
water based experiments and other referenced previous studies in 28 (2008) 1875–1881.
low Reynolds range. This performance aspect gives clear advan- [10] Ö. Ağra, A. Aara, H. Demir, S.A. Atayilmaz, F. KantaA, Numerical investigation
tage of current tube over other available tubes in applications with of heat transfer and pressure drop in enhanced tubes, in: vol. 38, International
Communications in Heat and Mass Transfer, Elsevier, 2011, pp. 1384–1391.
high viscosity fluids. [11] J. Fernández-Seara, F.J. Uhía, J. Sieres, A. Campo, A general review of the Wilson
plot method and its modifications to determine convection coefficients in heat
5. Conclusions exchange devices, Appl. Therm. Eng. 27 (2007) 2745–2757.
[12] F. Dittus, L. Boelter, vol. 2, Publications on Engineering, University of California
at Berkeley, Berkeley, CA, 1930, pp. 433–461.
This paper presents investigation of heat transfer and pressure [13] R.S. Figliola, D. Beasley, Theory and Design for Mechanical Measurements, John
drop characteristics of an enhanced tube using experimental and Wiley & Sons, 2015.
numerical simulation techniques in a double pipe heat exchanger. [14] A.F. Ansys, 14.0 Theory Guide, ANSYS Inc., 2011.
[15] Liu, J.K. Agarwal, Experimental observation of aerosol deposition in turbulent
For experimentation, the heat exchanger was configured in a counter flow, J. Aerosol Sci. 5 (1974) 145–155.
flow arrangement. Reynolds number was varied from approximate- [16] V. Gnielinski, New equations for heat and mass-transfer in turbulent pipe and
ly 500–8000 for water and 150–2000 for glycol/water solution. channel flow, Int. Chem. Eng. 16 (1976) 359–368.
[17] B. Petukhov, Heat transfer and friction in turbulent pipe flow with variable
Modified Wilson plot method was used to obtain the inside con- physical properties, Adv. Heat Transfer 6 (1970) i565.
vective heat transfer coefficient. Based on experimental data, [18] T.S.K. Ming Li, Ebrahim Al-Hajri, Zahid H. Ayub, Heat transfer and pressure drop
correlations to estimate Nusselt number and friction factor were pro- characteristics of a single enhanced tube for single phase flow, in: ASME-ATI-UIT
2015 Conference on Thermal Energy Systems: Production, Storage, Utilization
posed. Simulation results showed good agreement with experimental and the Environment, Napoli, 2015.
data. Major conclusions can be summarized as follows. [19] D.L. Gee, R.L. Webb, Forced convection heat transfer in helically rib-roughened
tubes, Int. J. Heat Mass Transf. 23 (1980) 1127–1136.
[20] J. Guo, A. Fan, X. Zhang, W. Liu, A numerical study on heat transfer and friction
• Heat transfer coefficient increased with Reynolds number.
factor characteristics of laminar flow in a circular tube fitted with center-cleared
• Heat transfer enhancement was found to be in excess of 200% twisted tape, Int. J. Therm. Sci. 50 (2011) 1263–1270.
as compared to an equivalent smooth tube. [21] A.E. Bergles, H.L. Morton, Survey and Evaluation of Techniques to Augment
Convective Heat Transfer, MIT Dept. of Mechanical Engineering, Cambridge,
• Pressure drop enhancement ratio increased with Reynolds
Mass, 1965.
number in 500 < Re < 2000 region while remained constant at [22] R. Raj, N.S. Lakshman, Y. Mukkamala, Single phase flow heat transfer and
1.8 in Re > 2000 region. pressure drop measurements in doubly enhanced tubes, Int. J. Therm. Sci. 88
(2015) 215–227.
• Heat transfer enhancement was higher than pressure drop en-
[23] P. Sivashanmugam, Experimental studies on heat transfer and friction factor
hancement at any given operating condition, therefore resulting characteristics of laminar flow through a circular tube fitted with helical
in higher PEC. screw-tape inserts, Appl. Therm. Eng. 26 (2006) 1990–1997.

View publication stats

You might also like