Ntu LCC Fyp (Submission)

This
document is downloaded from DR‑NTU (https://dr.ntu.edu.sg)
Nanyang Technological University, Singapore.
Optimization design of a micro‑channel heat
exchanger
Low, Choon Chye
2017
http://hdl.handle.net/10356/72276
Nanyang Technological University
Downloaded on 09 Sep 2021 16:45:41 SGT

Acad Year:
15/16
OPTIMIZATION DESIGN OF A MICRO-CHANNEL HEAT
Project No: EXCHANGER
A225

OPTIMIZATION DESIGN OF A MICRO-CHANNEL HEAT EXCHANGER

LOW CHOON CHYE

SCHOOL OF MECHANICAL AND AEROSPACE ENGINEERING
NANYANG TECHNOLOGICAL UNIVERSITY
2017

OPTIMIZATION DESIGN OF A MICRO-CHANNEL HEAT
EXCHANGER

SUBMITTED
BY
LOW CHOON CHYE

SCHOOL OF MECHANICAL AND AEROSPACE ENGINEERING

A final year project report
presented to
In partial fulfilment of the
requirements for the
Degree of Bachelor of Engineering (Aerospace Engineering)

2017
Abstract
Abstract

The application of microchannels in micro-scale systems has been proven to
improve heat transfer performance. However, due to high cost and complexity
of microfabrication, it is not economical to apply in many normally sized
systems in industries today. In 2013, Kong and Ooi came out with the
application of microscale in macro geometry with superior heat transfer
capabilities and with the usage of conventional manufacturing methods. Goh
went on to explore and develop a nature inspired configuration in 2016 to
increase microscale heat transfer in macro geometry utilizing conventional
machining processes and formulate working correlations formulae for practical
implementations.
The present study aims to extend the Ph.D. work of Goh, to make use of the
working correlations proposed by Goh, and use a computerized optimization
method, written in Fortran programming language, to search for a set of design
parameters which produces maximum heat transfer under a given set of
design constraints. The area of research of this study will be focusing on single
phase water cooling to avoid the flow instabilities and complexities of two-
phase flow boiling. This study will also suggest a set of design parameters for
microchannel heat exchanger at large heat duty and competitive costs.
Therefore, it is forecasted that the present study will contribute to more efficient
heat exchangers which require smaller heat transfer area and lower the cost
of materials and fabrication of heat exchangers utilizing conventional method
of fabrication.
Acknowledgement
Acknowledgement
The author would like to express his gratitude to his supervisor, Professor
Ooi Kim Tiow, for the guidance and support for the past year. Professor Ooi
is always very motivating and encouraged him during difficult time. Professor
Ooi is more than just a supervisor, he is a mentor who inspired the author to
be a better person in life. Professor Ooi will always go to the extra mile to
render help to the author when required.
List of Figures
List of Figures
Figure 1-1: Thermal Cooling Unit (TCU) part [3] ............................................. 2
Figure 1-2: Thermal Cooling Unit (TCU) wires insertion [3] ............................ 3
Figure 2-1: Critical Reynolds number ranges for single-phase liquid flow in
microchannels ......................................................................................... 12
Figure 2-2: Flow regimes observed from friction factor data [93] .................. 13
Figure 2-3: Flow regimes observed from Nusselt number data [93] ............. 13
Figure 2-4: Identification of flow transition from the flow and heat transfer
behaviour ................................................................................................ 14
Figure 2-5: Nature-inspired insert profiles ..................................................... 15
Figure 2-6: Geometrical parameters for the IFS, FS, D and Plain profile (top
to bottom) ................................................................................................ 16
Figure 3-1: Laminar, hydrodynamic boundary layer development in a circular
pipe [47] .................................................................................................. 35
Figure 3-2: (a) Laminar and (b) Turbulent boundary layer development in
circular tube [49] ..................................................................................... 36
Figure 3-3: Fully developed turbulent flow in a smooth pipe [14] .................. 37
Figure 3-4: Velocity profiles and friction factor variation for laminar flow in a
circular tube [55] ..................................................................................... 39
Figure 3-5: Velocity profiles and friction factor variation for turbulent flow in a
circular tube [55] ..................................................................................... 40
Figure 3-6: Regimes of turbulent pipe flow as a function of Reynolds number
and relative surface roughness [54] ........................................................ 41
Figure 3-7: Types of laminar flows for constant wall temperature boundary
condition [14] .......................................................................................... 43
Figure 3-8: Thermal boundary layer development in a heated circular tube
[47] .......................................................................................................... 44
Figure 3-9: Velocity and thermal boundary layer development in a circular
tube for Pr > 1 [53] .................................................................................. 45
Figure 3-10: Convective heat transfer coefficients for different flow
configuration [59] .................................................................................... 47
List of Figures
Figure 3-11: Variation of friction factor and convective heat transfer
coefficient with velocity boundary layer and thermal boundary layer
development for Pr > 1 [53] .................................................................... 48
Figure 4-1: Flow Chart for Optimisation Process .......................................... 60
Figure 5-1: Inverted Fish Scale profile’s Reynolds number iteration results . 66
Figure 5-2: Inverted Fish Scale profile’s e/H iteration results ........................ 67
Figure 5-3: Inverted Fish Scale Profile’s P/e iteration results ........................ 67
Figure 5-4: Inverted Fish Scale Profile’s Nu/NuPlain iteration results .............. 68
Figure 5-5: Inverted Fish Scale Profile H=0.3mm ......................................... 69
Figure 5-6: Inverted Fish Scale Profile H=0.45mm ....................................... 69
Figure 5-7: Fish Scale profile’s Reynolds number iteration results ............... 70
Figure 5-8: Fish Scale profile’s e/H iteration results ...................................... 71
Figure 5-9: Fish Scale Profile’s P/e iteration results ...................................... 71
Figure 5-10: Fish Scale Profile’s Nu/NuPlain iteration results .......................... 72
Figure 5-11: Fish Scale Profile with H=0.3mm .............................................. 73
Figure 5-12: Fish Scale Profile with H=0.45mm ............................................ 73
Figure 5-12: Durian Profile’s Reynolds number iteration results ................... 74
Figure 5-13: Durian Profile’s e/H iteration results .......................................... 75
Figure 5-14: Durian Profile’s P/e iteration results .......................................... 75
Figure 5-15: Durian Profile’s Nu/NuPlain iteration results ................................ 76
Figure 5-16: Durian Profile with H=0.3mm .................................................... 77
Figure 5-17: Durian Profile with H=0.45mm .................................................. 77
List of Tables
LIST OF TABLES

Table 2-1: Selected studies from the 1990s to early 2000s on single-phase
flow in microchannels ............................................................................... 9
Table 2-2: Proposed correlations from selected studies for single-phase flow
in microchannels ..................................................................................... 10
Table 2-3: Nusselt number correlation for laminar flow regime ..................... 19
Table 2-4: Applicable Re range of Nusselt number correlation for laminar flow
regime ..................................................................................................... 20
Table 2-5: Applicable Reynolds number range for the NuIFS correlation in
Eqn. (2-3) ................................................................................................ 22
Table 2-6: Applicable Reynolds number range for the NuFS correlation in Eqn.
(2-4) ........................................................................................................ 23
Table 2-7: Applicable Reynolds number range for the NuD correlation in Eqn.
(2-5) ........................................................................................................ 24
Table 2-8: Effect of e/H ratio on Nusselt number at Re≈2,664 .................... 25
Table 2-9: Effect of P/e ratio on Nusselt number at Re≈2,664 .................... 26
Table 2-10: Prediction of Nusselt number for e/H ratio and P/e ratio of 1, at
Re≈2,664 ............................................................................................... 26
Table 2-11: Applicable Re range for the fIFS correlation eq(2-9) .................... 28
Table 2-12: Applicable Re range for the fFS correlation in Eqn. (2-10) .......... 29
Table 2-13: Applicable Re range for the fD correlation in Eqn. (2-11) ........... 30
Table 2-15: Effect of P/e ratio on friction factor at Re≈2,664 ....................... 32
Table 2-16: Prediction of friction factor for e/H ratio and P/e ratio of 1, at Re≈
2,664 ....................................................................................................... 32
Table 3-1: Darcy friction factors for laminar flow in circular pipes ................. 41
Table 3-2: Darcy friction factors for turbulent flow in circular pipes ............... 42
Table 3-3: Typical values of convective heat transfer coefficient [14] ........... 46
Table 3-4: Local and average Nusselt numbers for laminar flow in circular
pipes ....................................................................................................... 49
Table 3-5: Average Nusselt numbers for turbulent flow in circular pipes ...... 50
Table 3-7: Average Nusselt number for laminar flow in parallel plates ......... 54
List of Tables
Table 4-1: Constraints of independent variables ........................................... 62
Table 5-1: Optimization Results of NuNuP for the three different profiles .... 63
Table 5-2: Optimization Results of Nu for the three different profiles ............ 64
Table 5-3: Optimization Results of NuNuP(-‐‑ffP)(13) for the three different
profiles .................................................................................................... 64
Table 5-4: Minimization Results of f for the three different profiles ............... 65
Table 5-5: Minimization Results of ffP for the three different profiles ........... 65

Table of Contents

Table of Contents
Abstract ................................................................................................................. i
List of Figures ......................................................................................................... i
LIST OF TABLES ....................................................................................................... i
Chapter 1: Introduction ......................................................................................... 1

1.1. Motivation for research ........................................................................... 1
1.2. Objective ................................................................................................. 4
1.3. Scope of work .......................................................................................... 4
1.4. Outline of report ..................................................................................... 6
Chapter 2 Literature Review .................................................................................. 8

2.1. Overview of microscale heat transfer .......................................................... 8
2.1.1. Laminar-‐to-‐turbulent Flow Transition ..................................................... 11
From Table 2-‐1, there exist an flow transition from laminar to turbulent flow at
Reynolds number less than 1500 [81,91,95,97,98] from studies from 1994 to
2001. Mala and Li [29] also pointed out that this transition takes place from
Reynolds number of 300 to 900, the flow develops to turbulent flow at Reynolds
number between 1000 to 1500. However, many research results in 2002 and
2003 showed that there was no early transition for Reynolds number less than
2000 [73,82,89,93,99,108]. Sharp and Adrian [30] concluded in 2004 that this
transition flow occurs at the same Reynolds numbers as macroscale channels. 11
2.2. Microchannel Profiles ............................................................................... 14
2.2.1 Performance indicator ............................................................................. 16
2.2.2 Flow loop ................................................................................................ 17
2.2.4 Different Profile Results .......................................................................... 18
2.3.1 Proposed Correlation on Nusselt number ................................................ 19
2.3.2 Proposed Correlation for Average Friction Factor .................................... 26
Table of Contents
2.4 Research gap .............................................................................................. 33
Chapter 3: Heat Transfer ..................................................................................... 34

3.1. Classical theory for Circular Channel .......................................................... 34
3.1.1 Internal flow ........................................................................................... 34
3.2. Hydrodynamic Consideration .................................................................... 35
3.2.1. Velocity Boundary Layer ......................................................................... 35
3.2.2. Friction Factor ........................................................................................ 38
3.2.3 Thermal Consideration ............................................................................ 42
3.2.4. Thermal Boundary Layer ........................................................................ 44
3.2.5 Convective Heat Transfer Coefficient ....................................................... 46
3.3. Classical Theory for Concentric Annular Channel .................................... 52
3.3.1 Hydrodynamic Consideration .................................................................. 52
3.3.2. Thermal Consideration ........................................................................... 54
3.4. Passive Heat Transfer Enhancement Mechanisms ..................................... 55
3.5 Thermo-‐hydraulic Performance .................................................................. 56
Chapter 4: Optimization Method ......................................................................... 57

4.1. Flow chart of Algorithm ......................................................................... 60
4.3 Heat Exchange Performance Optimisation ................................................. 61
Chapter 5: Results and Discussion ....................................................................... 63

5.1. Inverted Fish Scale Profile ......................................................................... 66
5.2. Fish Scale Profile ....................................................................................... 70
5.3. Durian Profile ............................................................................................ 74
6. Conclusion and Future Works .......................................................................... 78

6.1 Conclusion ................................................................................................. 78
6.2 Future Works ............................................................................................. 78
6.2.1 Design of microchannel profiles .............................................................. 78
6.2.2 Extend the range of design constraints for correlations ........................... 79
6.2.3 Different microchannels profile ............................................................... 79
List of References ................................................................................................ 80

Table of Contents

Introduction Chapter 1
Chapter 1: Introduction

This chapter introduces optimization design of a micro-channel heat
exchanger which includes the motivation behind this research study, the
objective and scope of this project and the outline of the report.
1.1. Motivation for research

The utilization of naturally occurring microchannels are commonly found
around us, such as lungs in mammals. There has been a focus in research in
internal flow from macrochannels to microchannels since the introduction of
research in microscale internal flow heat transfer by Tuckerman and Pease
[4]. They showed that the microchannels, those channels with 100 µm or
smaller hydraulic diameters, are capable of removing a heat flux of around 790
W/cm2 in 1981. In 1987, Kishimoto and Sasaki [10] presented the first
successful application of enhanced heat transfer in microchannel. The
advantages of applying microchannel cooling for micro-electronics proofed to
be economical after mass production [5]. To popularize the usage of
microchannel heat transfer, there is a need to increase convective heat
transfer coefficient and at the same time reduces pressure drop.
To further enhance the heat transfer effects, microchannel cooling utilizes two-
phase flow boiling coupled with a single-phase liquid heat transfer was
introduced for the micro geometry of copper or silicon [7]. This technique is
suitable in electronic cooling applications to achieve the high demand of 1000
W/cm2 together with higher transport efficiency. The key difficulty to the usage
of evaporative microscale heat exchangers is the instabilities which arises in
boiling flow [8]. Single-phase liquid flow is also potentially able to compete with
the cooling performance of the two-phase flow systems as it does not involve
as much complexity and high pumping power requirement [9].

1

The current study aims to apply microscale heat transfer effects in commonly-
sized heat exchangers, with the replacement of macroscale passages with
microscale passages. There are three possible positive outcomes from the
pursuing of this research area, firstly, a reduction in the physical size, secondly,
a lower material cost and finally, a lower fabrication cost. As microscale heat
transfer is expected to give a better heat transfer performance than the
normally sized heat exchanger, a smaller area for heat transfer is required.
Therefore, the size of the heat exchanger can be greatly reduced. This will be
a great advantage to solve problems due to space constraints faced by
offshore platforms and chemical plants. However, the current cost of
fabricating microchannels is very high due to the usage of costly
microfabrication technologies, including microelectromechanical systems
(MEMS) and photolithographic-based processes [6]. This high cost of
production resulting from the usage of microchannels cooling system
impractical and uneconomical especially in macro-sized applications.
Figure 1-1: Thermal Cooling Unit (TCU) part [3]
2
Figure 1-2: Thermal Cooling Unit (TCU) wires insertion [3]

In 2008, Ooi and Teh [3] presented a design of a thermal control unit (TCU) as
shown in figure 1-1. In this design, the microchannels have been fabricated
using readily available conventional fabrication technique. In the design, 50
mm long channels with one mm diameter were drilled through a copper block.
A wire of diameter slightly smaller than 1 mm is then inserted into these one
mm diameter channels. The gaps between the wire and the channel of 1mm
diameter emulate microchannels and are able to transfer significant heat
similar to those occurring in microchannels. This design mimics microchannel
heat transfers effect but it happens in micro-geometry but the channel was
fabricated using readily available conventional fabrication method which is low
cost. In 2016, Goh [1] explored and developed a nature inspired configuration
to increase microscale heat transfer in macro geometry utilizing conventional
machining processes and formulate working correlations for practical
implementations. Goh’s work encourages further research and studies to apply
microscale passages to conventional sized heat exchangers.
The present study aims to extend the Ph.D. work of Goh, to make use of the
working correlations proposed by Goh, and employing a computerized
optimization method to find a set of design parameters which produces
maximum heat transfer with a given set of design constraints. In this project,
this study will be focusing on single phase water cooling inherited from the
work of Goh [1]. This study will suggest a set of design parameters for
3

microchannel heat transfer effect in normally sized heat exchanger and at
competitive costs. Therefore, it is forecasted that the present study will
contribute to more efficient heat exchangers which help to lower the cost of
materials and fabrication of heat exchangers.
1.2. Objective
The main objective of this project is to apply computerized optimization
approach with given constraints to search for a set of design parameters to
optimize heat transfer coefficient and to reduce pressure drop for practical
applications. This can be achieved by utilizing several formulated working
correlations from Goh’s work.
1.3. Scope of work

The following tasks are to be completed in order to attain the objectives that
were stipulated in section 1.2:
1. To revisit the classical theory for forced convective internal flow in a

conventionally sized flow passage. This provides a good overview of
the concepts on convective heat transfer and enhance understanding
on single phase heat transfer.

2. To understand the enhanced microchannel heat transfer in macro-
geometry. Understand the experimental steps undertaken to obtain
various results on the different profiles.
3. Understand the proposed correlations proposed by Dr Goh.
4

4. Using a computerized optimization approach, and together with
correlations proposed by Dr Goh to search for a set of design
parameters which provides optimized heat transfer coefficient and
minimizes pressure loss.

5

1.4. Outline of report

This report consists of the following:

Chapter 1, the current chapter, is an introduction of this report. It entails the
research motivation, objectives and scope and outline of the project.

Chapter 2 reviews available open literature on microchannel heat transfer in
macro geometry which is relevant to the present study. The main focus is on
the research work carried out by Dr Goh. It reviews the different profiles of
microchannels that were proposed by Goh [1], performance indicator and
results were discussed. The correlations for both Nusselt number and friction
factor proposed by Goh are also presented.

Chapter 3 presents the classical heat transfer theory relevant to the current
study. It includes theory for internal forced convection heat transfer in terms of
hydrodynamic and thermal considerations for circular channels. A brief review
of forced convection in annular channels is also presented. Lastly, enhanced
heat transfer for single phase microchannels with passive heat transfer
enhancement techniques are presented.

Chapter 4 delves into the optimization method adopted in this study. It is a
multi-variable direct search constrained optimization technique, namely
Complex optimization technique attributed to M.J. Box [2]. it is employed to
search for a set of design parameters which maximizes heat transfer and
reduce pressure drop in heat exchanging at the given constraints. The
procedures of the optimization approach is presented together with a flow chart
for better clarity. The details of optimization set up are also shown in this
chapter.

Chapter 5 presents and discuses results obtained in this project.
6

Chapter 6 concludes the work accomplished by the author and summarises
key findings of this project. The recommendations for future work are also
presented in this section.

7
Literature Review Chapter 2
Chapter 2 Literature Review

This chapter presents the current research that is of relevance to this project.
There are a total of four parts to this section. The first part will focus on the
overview of microscale heat transfer while the second part will be on forced
convection for internal flow in both circular and annular channels of
conventional sizing. The third part sums up the research that was carried out
by Goh [1] on enhanced microscale heat transfer in macro geometry. The key
focus with regards to the current study is to suggest a design specification with
a set of parameters that optimizes heat transfer and performance of a
microchannel. Therefore, the last part will be looking into research gaps that
can potentially be filled after the findings from this literature review.

2.1. Overview of microscale heat transfer

There exist an extensive amount of study and research regarding single-phase
liquid flow in microchannels as many review papers are published to date. In
2014, 219 articles on single and two-phase flows in microchannels are
reviewed by Asadi et al [11]. Studies from late 1980s to early 2000s were
mostly on experimental methods. Recent studies and research used numerical
simulations, and proposed correlations were considerably more precise. Obot
[11] also showed that correlations can be utilized to model heat flow in
microchannels that are within the accuracy range.

There are different studies with contradictory conclusions [11]. Some studies
also that reported findings that Reynolds number is lower than classical
predictions. There are also studies that concluded that Nusselt number is
greater than classical predictions. A summary of different research and
from 1990s to early 2000s are presented in
8
Literature review Chapter 2
Table 2-1. Classical predictions include Poiseuille number and Blasius
correlation for friction factor, and Hausen and Dittus-Boelther correlations for
Nusselt number.

Table 2-1: Selected studies from the 1990s to early 2000s on single-phase
flow in microchannels
Study Year Dh (µm) Re range f Nu Critical Re
Wang and Peng [12] 1994 311-747 80-3600 NA < 1000-1500
Peng et al. [13] 1994 133-367 50-4000 > < 200-700
Yu et al. [14] 1995 19-102 250-20 < > ID
000
Harms et al. [15] 1999 404-1923 173-12900 ≈ ≈ 1500
Qu et al. [16] 2000 51-169 6.2-1447 > NA ID
Qu et al. [17] 2000 62-169 94-1491 NA < ID
Xu et al. [18] 2000 30-344 20-4000 ≈ NA 1500
Jiang et al. [19] 2001 300 50-3000 > > 600
Gao et al. [20] 2002 199-1923 100-8000 ≈ < ≈
Qu and Mudawar [21] 2002 349 139-1672 ≈ ≈ ≈
Judy et al. [22] 2002 15-150 8–2300 ≈ NA ≈
Bucci et al. [23] 2003 290 2-5272 ≈ ≈ ≈
Liu and Garimella [24] 2004 244-974 230-6500 ≈ NA ≈
≈ : Experimental results agree with classical predictions
> : Experimental results are higher than classical predictions
< : Experimental results are lower than classical predictions
NA : Not applicable
ID : Insufficient data

9

Table 2-2: Proposed correlations from selected studies for single-phase flow
in microchannels
Study Friction factor / Nusselt number correlation
Peng et al. [13] Laminar: 𝑓 = 𝐶.;0 /𝑅𝑒 4.67
Turbulent: 𝑓 = 𝐶.;8 /𝑅𝑒 4.9:
where 𝐶.;0 and 𝐶.;8 are constants
Wang and Peng 𝑁𝑢 = 0.00805𝑅𝑒 @/A 𝑃𝑟 4/D

[12] Re>1500
Peng et al. [26] Laminar: 𝑁𝑢 = 𝐶E;0 𝑅𝑒 F.G: 𝑃𝑟 4/D
Turbulent:𝑁𝑢 = 𝐶E;8 𝑅𝑒 F.7 𝑃𝑟 4/D
where 𝐶E;0 and 𝐶E;8 are constants
Adams et al. [27] 𝑁𝑢 = 𝑁𝑢HI 1 + 𝐹

𝐹 = 𝐶𝑅𝑒[1 − 𝐷/𝐷O : ]
where 𝐶 and 𝐷O are constants
𝑁𝑢HI refers to Gnielinski correlation
Nguyen et al. [28] Laminar: 𝑁𝑢 = 8.39𝑅𝑒 F.A − 1.33𝑅𝑒 :/D
Turbulent: 𝑁𝑢 = 4.73𝑅𝑒 F.A − 0.22𝑅𝑒 :/D
Jiang et al. [19] 𝑁𝑢 = 0.52(𝑋 V )WF.G: for 𝑋 V < 0.05
𝑁𝑢 = 2.02(𝑋 V )WF.D4 for 𝑋 V > 0.05
where 𝑋 V = 𝑅𝑒𝑃𝑟 𝐷E 𝐿

Correlations for both friction factor and Nusselt number were suggested by
researchers in studies where experimental and studies having a situation
where the experimental results deviate from classical predictions.

10

Table 2-2 presents some proposed correlations from selected studies, with
the focus on water as working fluid.

Wu and Cheng [90] suggested in 2003 that surface roughness will affect heat
transfer in microchannels to resolve discrepancy among earlier microchannel
studies. In summary, there are bound to have discrepancies from earlier
studies as more experimental data are processed.
2.1.1. Laminar-to-turbulent Flow Transition
From
Table 2-1, there exist an flow transition from laminar to turbulent flow at
Reynolds number less than 1500 [81,91,95,97,98] from studies from 1994 to
2001. Mala and Li [29] also pointed out that this transition takes place from
Reynolds number of 300 to 900, the flow develops to turbulent flow at Reynolds
number between 1000 to 1500. However, many research results in 2002 and
2003 showed that there was no early transition for Reynolds number less than
2000 [73,82,89,93,99,108]. Sharp and Adrian [30] concluded in 2004 that this
transition flow occurs at the same Reynolds numbers as macroscale channels.

Morini [3] referenced and reported a number of articles with regards to the
development of flow from laminar to turbulent in microchannels. The critical
Reynolds numbers ranges are presented in
Figure 2-1. The studies took reference from the work of Wang and Peng [12],
Yu et al. [14], Gui and Scaringe [31], Peng and Peterson [111], Nguyen et al.
[28], Harms et al. [15], Mala and Li [29], Ding et al. [32], Celata et al. [74], Yang
et al. [35], Pfund et al. [33], Debray et al. [34], Jiang et al. [19], Gao et al. [20],
Hegab et al. [73] and Bucci et al. [93].
11

Figure 2-1: Critical Reynolds number ranges for single-phase liquid flow in
microchannels

Schlichting [36] explained that flow characteristics which refer to laminar,
laminar to turbulent, turbulent flow regions and possible to be studied using
12
the moody chart. Celata [37] showed two graphs from Bucci et al. [23] in 2004
review which displays the change in flow and heat transfer behavior with
regards to Reynolds number. The datas are depicted in Figure 2-2 and Figure
2-3. The above mentioned flow regime can be identified by the change in data
pattern.

Figure 2-2: Flow regimes observed from friction factor data [23]
13
Figure 2-3: Flow regimes observed from Nusselt number data [23]
From Figure 2-2, for classical theory, friction factor is compared with the
laminar flow prediction by Hagen-Poiseuille, and compared to Colebrook and
Blasius for turbulent flow. From Figure 2-3, Nusselt number is compared with
laminar flow prediction by Hausen, and turbulent flow by Dittus Boelter,
Gnielinski and Adams. It is important to investigate and find out the values of

Reynolds number for laminar to turbulent flow transition. This can be done by
putting the friction and Nusselt number data onto the same graph.
Figure 2-4 displays a graph from Celata [74] in 2002 which concur with the
Reynold number range for transition of flow.

14
Figure 2-4: Identification of flow transition from the flow and heat transfer
behaviour
2.2. Microchannel Profiles

Three series of geometrical profiles, inspired by nature were proposed by Goh
[1] to be used on outer surface to find out their effects on heat transfer and flow
characteristic of microchannel. They are, Inverted Fish Scale (IFS), Fish
Scale(FS) and Durian (D) profile respectively. These profiles are based on
nature’s inspiration. The inverted fish scale profile is thought of as fishes are
naturally sleek to reduce flow resistance as they swim forward. The profile is
such as to oppose flow direction which results in highest flow resistance. As
the penalty for higher heat transfer is the high friction loss, Fish Scale profile
is also studied. The durian profile is used as the copious pointed husk of the
durian may cause flow disturbance and enhance heat transfer.

15
Figure 2-5: Nature-inspired insert profiles
There were 22 microchannel profiles tested by Goh which consisted of 7

profiles for each of the three types of profiles mentioned together with a plain
microchannel. The plain microchannel was set to have no protrusion so as to
be the baseline for cross referencing and comparison. Insert volume, mean
microchannel gap size and volume of fluid flowing through microchannel are
fixed throughout for the study. The microchannel profiles are illustrated in
Figure 2-6 for the ease of understanding.

16
Figure 2-6: Geometrical parameters for the IFS, FS, D and Plain profile (top
to bottom)
For each microchannel series, the effects of two geometric ratios are examined.
They are ratio of scale/thorn height (e) to mean height (H), and ratio of
scale/thorn pitch (P) to scale/thorn height (e).
2.2.1 Performance indicator

A high heat transfer capacity is beneficial in getting a better heat performance
or a decreased average temperature deviation. This can also be known as
thermo-hydraulic performance which is used to differentiate heat transfer
augmentation techniques across [38,39,40]. Thermos-hydraulic performance
is calculated with the normalised Nusselt number and friction factor.

Microchannels enhances heat flow but often resulted in an undesirable high
pumping power required. There is a need for a performance factor in order to
compare the different techniques and to achieve a high heat transfer and
relatively low pumping power required. In 1971, Webb and Eckert [41] came
out with a set of equations to compare the performance between roughened
and smooth tubes in heat exchanger design. The three objectives stipulated
were to obtain decreased area for heat transfer, increased heat exchange
capacity together with a reduction in friction power expenditure.

17

Gee and Webb [38] used the three performance equations in heat exchanger
application. Experiments were conducted to obtain friction factor and Stanton
number data regarding smooth tube. The experiment was repeated for the
case of rough tube. The three performance equations were computed and
results showed that helically rib-roughened tubes using air as working fluid cut
down heat exchange surface area by 25%, increased heat transfer capacity
by 27% and reduced pumping power by 50%.

Kathait and Patil [42] studied the thermo-hydraulic performance of a
corrugated rib roughened tube as of compared to smooth tube in 2014 and
reported that a thermos-hydraulic performance of 95% with Reynolds number
of 7,500. This study was on a macro-scale heat exchange where the Reynolds
number ranges from 7,500 to 50,000.

Marschewski et al. [43] studied the thermo-hydraulic performance of
herringbone-grooved microchannels as of compared to plain microchannels
for Reynolds number ranging from 190 to 510. A thermo-hydraulic
performance of 120% is reported for Reynolds number of 350.

2.2.2 Flow loop

The test loop and by-pass with air-vent loop are part of the flow loop adopted
by Goh [1]. Distilled water flows through the flow loop with the aid of a positive
displacement pump, which is part of the chiller unit. The chiller unit was used
to decrease the temperature of water in order to obtain the targeted test
module’s inlet temperature. Once the distilled water exits the chiller at a fixed
temperature, a needle valve channels a stipulated amount of water to the test
loop, while the remaining water flows back to the reservoir through the by-pass
loop. The air-vent loop solves the issue of higher friction by removing air
bubbles that caused increased flow resistant as the small bubbles formed on
the walls [44].

18

Water flows to a 40 µm filter for the removal of solid particles. After which, it
flows through the microchannel in the test module and electrically-generated
heat is removed from the test module. Coriolis flow meter will then measure
the rate of flow through the test module. Temperature and pressure sensors
are placed at the entrance and exit of module to gather important results. The
test loop flow is then directed back to the chiller for cooling and process repeats
itself.
2.2.4 Different Profile Results

For Goh’s study, heat transfer coefficient result regarding the plain
microchannel is from 10.9 to 27.1 kW/m2·K. The flow rates are between 2 and
7 L/min. This confirms the possibility of attaining microscale heat transfer
performance in macro geometry. Inverted Fish Scale profiles generally
enhance heat flow, for flow rates that are higher than 3.5 L/min, the heat
transfer coefficient by Inverted Fish Scale microchannels are higher than those
of the plain microchannels. Highest Inverted Fish Scale profile’s heat transfer
coefficient observed was 52.8 kW/m2·K, at 6.5 L/min, 2.1 time higher than the
value achieved by plain microchannels, removing heat flux of 375 W/cm2. The
focus is on two parts, scale height(e) to mean channel height (H) ratio and
scale pitch (P) to scale height (e) ratio, to investigate and study the effects
these two ratios have. It was observed that the maximum pressure drop for
Inverted Fish Scale microchannel series was 3.3 bars which can be overcome
by pumps that are available commercially. As for scale height(e) to mean
channel height (H), Inverted Fish Scale profile is the most effective for
improvement of heat transfer as it promotes early flow transition at Reynolds
number smaller than 3400. For Fish Scale profiles, it also enhances heat
transfer, for flow rates that is higher than 3L/min, most of the profiles yield
higher heat convection coefficient as of compared to plain microchannel. The
highest heat transfer coefficient achieved is 47.9 kW/m2·K, at 6.75 L/min. This
confirms the that disruptive Inverted Fish Scale profile will result in more
disturbances in heat transfer and improves heat transfer. For Inverted Fish
Scale microchannel series, a smaller P/e ratio exhibits a higher thermos-
19

hydraulic performance while it is the exact opposite for Fish Scale
microchannel series for Re≲1,400. For Inverted Fish Scale and Fish Scale
profiles microchannels with Re≳3,400 the P/e ratio has an insignificant effect
on the performance. The thorny D microchannel also enhances heat transfer
as well as most of the D microchannels yield a larger heat transfer coefficient
when comparing to the plain microchannel. For the D microchannel series, e/H
ratio on heat transfer is much more significant than P/e ratio as P/e ratio holds
lesser significance at higher flow rates. Maximum pressure drop was 2.7 bars
for D microchannel series.
2.3.1 Proposed Correlation on Nusselt number

For the inverted Fish Scale (IFS) series, it was concluded to have heat transfer
behavior that is identical to plain microchannel when 350≲Re≲2,200. As for
Fish Scale series, it exhibits similar heat transfer properties as the Plain
microchannel in the laminar flow regime. For the Durian (D) series, the heat
transfer properties are similar to the plain microchannel in the laminar flow. We
can then use the Shah and London’s correlation for circular pipes for circular
pipes to find the Nusselt number in the laminar flow region as of shown in the
below table.

Table 2-3: Nusselt number correlation for laminar flow regime

Thermal conditions Correlation Applicable range
Thermally
𝑁𝑢 = 1.953(𝑅𝑒 𝑃𝑟 𝐷E 𝐿)4/D 𝑅𝑒 𝑃𝑟 𝐷E 𝐿 ≥ 33.3
developing,
Shah and London, 1978 [45] 𝑅𝑒 < 2,200
constant qs

20

Table 2-4: Applicable Re range of Nusselt number correlation for laminar flow
regime
Reynolds number range
e/H ratio P/e ratio
IFS series FS series D series
0.1 10 350≲Re≲2,200 350≲Re≲2,200 350≲Re≲2,200
0.3 10 - 350≲Re≲1,850 350≲Re≲1,500
0.5 10 - 350≲Re≲1,300 350≲Re≲1,000
0.7 10 - 350≲Re≲700 350≲Re≲750
0.5 20 - 350≲Re≲1,600 350≲Re≲1,800
0.5 15 - 350≲Re≲1,600 350≲Re≲1,500
0.5 5 - 350≲Re≲800 350≲Re≲800

The Nusselt number correlation for Plain microchannel is proposed as follows:

7
𝑁𝑢]0^_I,8àb`0cI8 = 0.034𝑅𝑒 F.9 𝑃𝑟4/D 1 + (2-1)
𝐿/𝐷E

Applicable for: 3,400≲Re≲4,600;; Pr≈5.5;; L/Dh =50
Type of fluid: Single-phase water
Maximum discrepancy: 5%
Mean discrepancy: 3%

From Goh [1] research, it was found that the new Nusselt number correlation
in order to account for scale/thorn protrusions for enhanced microchannels is
as follows:

21
g
7 𝐸 𝑒/𝐻
𝑁𝑢daO._0c = 0.034𝑅𝑒 F.9 𝑃𝑟4/D 1 + + (2-2)
𝐿/𝐷E 𝑃/𝑒 H

Where e/H – Scale/thorn height (e) to mean channel height (H) ratio,
P/e – Scale/thorn pitch (P) to scale/thorn height (e) ratio,
E/F/G are coefficients to be determined.

Inverted Fish Scale(IFS) microchannel series’ Nusselt number correlation is
as follows:

7 8 𝑒/𝐻 4.4
𝑁𝑢hgi = 0.034𝑅𝑒 F.9 𝑃𝑟4/D 1 + + (2-3)
𝐿/𝐷E 𝑃/𝑒 F.G

Applicable for: Pr≈5.5;; L/Dh=50
The applicable Reynolds number range of the proposed correlation for each
IFS microchannel profile is shown in

Table 2-5.
22

Table 2-5: Applicable Reynolds number range for the NuIFS correlation in
Eqn. (2-3)
e/H ratio P/e ratio Re range
0.1 10 3,400≲Re≲4,600
0.3 10 350≲Re≲4,600
0.5 10 350≲Re≲4,600
0.7 10 350≲Re≲4,600
0.5 20 350≲Re≲4,600
0.5 15 350≲Re≲4,600
0.5 5 350≲Re≲4,600

Fish Scale (FS) microchannel series’ Nusselt number correlation is proposed
as follows:

7 2.75 𝑒/𝐻 4.A
𝑁𝑢gi = 0.034𝑅𝑒 F.9 𝑃𝑟4/D 1 + + (2-4)
𝐿/𝐷E 𝑃/𝑒 F.4A


23

Fish Scale microchannel profile is shown in
Table 2-6.

Table 2-6: Applicable Reynolds number range for the NuFS correlation in Eqn.
(2-4)
P/e
e/H ratio Re range
ratio
0.1 10 3,400≲Re≲4,600
0.3 10 2,500≲Re≲4,600
0.5 10 1,600≲Re≲4,600
0.7 10 1,200≲Re≲4,600
0.5 20 2,200≲Re≲4,600
0.5 15 2,200≲Re≲4,600
0.5 5 1,600≲Re≲4,600

The Nusselt number correlation for the D microchannel series is proposed as
follows:

F.9 7 5 𝑒/𝐻 4.@

𝑁𝑢j = 0.034𝑅 𝑃𝑟4/D 1 + + (2-5)
𝐿/𝐷E 𝑃/𝑒 F.D9A

Applicable for: Pr≈5.5;; L/Dh =50

24
The applicable Reynolds number range of the proposed correlation for each D
microchannel profile is shown in

Table 2-7.

Table 2-7: Applicable Reynolds number range for the NuD correlation in Eqn.
(2-5)
P/e
e/H ratio Re range
ratio
0.1 10 3,400≲Re≲4,600
0.3 10 2,000≲Re≲4,600
0.5 10 1,300≲Re≲4,600
0.7 10 1,000≲Re≲4,600
0.5 20 2,600≲Re≲4,600
0.5 15 2,000≲Re≲4,600
0.5 5 1,300≲Re≲4,600

Coefficient F shows the effect of scale/thorn height (e) on the heat transfer
performance for a fixed mean channel height (H) and P/e ratio. It is 1.1 and
1.5 for IFS and FS respectively. Increasing the scale height results in more
effect on increasing Nusselt number for FS profile microchannel as of
compared to IFS microchannel series. This is because due to the nature of the
fish scale profile, it has the least disruption to the flow. Therefore, the decrease
in minimum channel gap by increasing the scale height plays a more dominant
role in increasing heat transfer by an earlier transition to turbulent flow and the
decreasing size of thermal boundary layer. IFS naturally causes the most
disruption to the flow to the extent that enhanced heat transfer is attained at
lowest e/H ratio for Re≳2,200. Early transition to turbulent flow at Re=600 is
observed for the rest of the IFS profiles. Coefficient F is 1.4 for thorny D
25
microchannel series, by increasing the thorn height, Nusselt number is
increased.

Table 2-8: Effect of e/H ratio on Nusselt number at Re≈2,664
Scale Nusselt number
e/H ratio P/e ratio height
IFS FS D
(mm)
0 (Plain) 10 0 15 15 15
0.1 10 0.03 16 15 16
0.7 10 0.21 37 33 37

E and G contribute to the heat transfer performance. The magnitude of
coefficient G displays the influence of the number of scales/thorns of the same
height(e) on the performance of heat flow given that the channel length(L) and
mean channel height(H) are fixed. The coefficient is 0.6 for Inverted Fish Scale
microchannel series and 0.15 for Fish Scale(FS) microchannel series. By
increasing the number of scales due to dropping the scale pitch(P) will have a
stronger impact in the rise of Nusselt number for Inverted Fish Scale
microchannel series when compared to the Fish Scale microchannel series.
This can be clearly explained as Inverted Fish Scale profile is the most
disordered and Fish Scale profile is the least disordered. The coefficient of G
for thorny D microchannel series is 0.375. There is an implication that by
raising the number of thorns due to decreasing thorn pitch(P) will play a huge
impact in the increase of Nusselt number for D microchannel series. The
coefficient of G alone appears to display the overall effect of the number of
scales/thorns on heat transfer coefficient.

26

Table 2-9: Effect of P/e ratio on Nusselt number at Re≈2,664
P/e Number of Nusselt number
e/H ratio
ratio scales/thorns IFS FS D
20 0.5 10 27 27 25
5 0.5 40 40 29 33

To confirm the applicability of the correlations of Eqns. (2-3) to (2-5) beyond
the measurement boundaries of Goh’s research, coefficient E is studied.
Coefficient E is assumed to be the effect of scale/thorn profile on heat transfer
performance as both ratios e/H and P/e are set to unity.

Table 2-10: Prediction of Nusselt number for e/H ratio and P/e ratio of 1, at
Re≈2,664
Scale Nusselt
Number of
P/e ratio e/H ratio height number
scale/thorns
(mm) IFS FS D
1 1 0.3 100 137 58 92

This seems to concur with logic with the Inverted Fish Scale profile as the most
effective in disturbing flow and increasing heat flow followed by Durian and
Fish Scale profiles.

2.3.2 Proposed Correlation for Average Friction Factor

For Goh’s study [1], the difference of thermo-hydraulic performance with
Reynolds number displays the same relationship where the examined
enhanced microchannels have better properties after critical Reynolds number
27
is exceeded. This is due to the similarity in Nusselt number of these enhanced
microchannels to that of the Plain microchannel when the flow is within laminar
flow region. The friction factor of these enhanced microchannels is larger than
that of Plain microchannel in laminar flow regime. Thus, thermo-hydraulic
performance of these enhanced microchannels can be less than 1.0. This
means that there is no rise in heat exchange capacity for similar heat transfer
area together with the pumping power in laminar flow regime.

There are no benefits to examine the flow in the laminar regime. Therefore,
there is no friction factor correlation in the laminar flow region.
For the turbulent flow regime, the Reynolds number range is 3,400≲Re≲4,600
The correlation is as of:
1 𝑅
𝑓]0^_I,8àb`0cI8 = 𝑄 + (2-6)
𝑅𝑒] 𝐿/𝐷E

The Plain microchannel in the range 3,400≲Re≲4,600. The friction factor
correlation for the Plain microchannels is as follows:
1 2
𝑓]0^_I,8àb`0cI8 = 0.2 +
𝑅𝑒 F.4A 𝐿/𝐷E (2-7)

Applicable for: 3,400≲Re≲4,600;;
L/Dh=50

The friction factor correlation of Plain microchannel in the turbulent flow regime
is used to derive more correlations for the enhanced microchannels. There is
an increase in factor due to the scale/thorn protrusions which is accounted for
by adding another term to the correlation. This will in turn promote early
transition to turbulent flow. The new friction factor correlation for the enhanced
microchannels is of the form:
28

1 𝑅 𝑆 𝑒/𝐻 m
𝑓daO._0c = 𝑄 + + (2-8)
𝑅𝑒] 𝐿/𝐷E 𝑃/𝑒 n

e/H - Scale/thorn height (e) to mean channel height (H) ratio
P/e - Scale/thorn pitch (P) to scale/thorn height (e) ratio
S, T and U are coefficients to be determined.

The friction factor correlation for the Inverted Fish Scale(IFS) microchannel
series:

1 2 15 𝑒/𝐻 4.:
𝑓hgi = 0.2 + + (2-9)
𝑅𝑒 F.4A 𝐿/𝐷E 𝑃/𝑒 4


Inverted Fish Scale microchannel profile is listed below.

Table 2-11: Applicable Re range for the fIFS correlation eq(2-9)
P/e
e/H ratio Re range
ratio
0.1 10 4,000≲Re≲4,600
0.3 10 900≲Re≲4,600
0.5 10 700≲Re≲4,600
0.7 10 600≲Re≲4,600
0.5 20 1,300≲Re≲4,600
0.5 15 800≲Re≲4,600
29
0.5 5 800≲Re≲4,600

The friction factor correlation for the Fish Scale microchannel series is:

1 2 7 𝑒/𝐻 4.G
𝑓gi = 0.2 + + (2-10)
𝑅𝑒 F.4A 𝐿/𝐷E 𝑃/𝑒 F.9


Fish Scale microchannel profile is shown in Table 2-11.

Table 2-12: Applicable Re range for the fFS correlation in Eqn. (2-10)
P/e
e/H ratio Re range
ratio
0.1 10 3,400≲Re≲4,600
0.3 10 2,400≲Re≲4,600
0.5 10 1,800≲Re≲4,600
0.7 10 700≲Re≲4,600
0.5 20 2,300≲Re≲4,600
0.5 15 2,300≲Re≲4,600
0.5 5 1,800≲Re≲4,600

30
.

The friction factor correlation for the D microchannel series is:

1 2 18 𝑒/𝐻 4.4
𝑓j = 0.2 + + (2-11)
𝑅𝑒 F.4A 𝐿/𝐷E 𝑃/𝑒 @/D


The applicable Reynolds number range of the proposed correlation for each D
microchannel profile is listed in Table 2-12.

Table 2-13: Applicable Re range for the fD correlation in Eqn. (2-11)
P/e
e/H ratio Re range
ratio
0.1 10 Not applicable
0.3 10 2,000≲Re≲4,600
0.5 10 1,000≲Re≲4,600
0.7 10 1,300≲Re≲4,600
0.5 20 2,000≲Re≲4,600
0.5 15 1,800≲Re≲4,600
0.5 5 1,800≲Re≲4,600

31
Coefficient T displays the impact of scale/thorn height(e) on friction losses for
a given fixed mean channel height (H) and P/e ratio. T is 1.2 for IFS and 1.6
for FS microchannel series. This means that the rise in scale height has a
stronger effect in increasing friction factor for FS microchannel series as of
compared to IFS microchannel series. T coefficient is 1.1 for D microchannel

series. This means that raising the thorn height has little impact of increasing
friction factor for D microchannel series.

There is a need to study coefficients S and U as the overall effect of the
scale/thorn height(e) on heat transfer performance cannot be based on
coefficient T alone. This can be seen from the below table where scale height
is increased from 0.08 to 0.31 resulting in a rise in friction factor from 0.08 to
0.31 for Fish Scale microchannel series even when coefficient T is the highest.
This is due to coefficient S being the lowest at 7 compared to 15 for Inverted
Fish Scale and 18 for D microchannel series.

Table 2-14: Effect of e/H ratio on friction factor at Re≈2,664
Scale Friction factor

e/H ratio P/e ratio height
IFS FS D
(mm)
0 (Plain) 10 0 0.08 0.08 0.08
0.1 10 0.03 0.08 0.08 0.08
0.7 10 0.21 0.36 0.31 0.25

The coefficient U shows the effect on friction loss due to the number of
scale/thorn protrusion, given the same height(e), channel length(L) and mean
channel height(H). Results shows that coefficient U is 1 for Inverted Fish Scale
and 0.7 for Fish Scale microchannel series. This means that by raising the
number of scales by reducing the scale pitch(P) has a larger impact on
increasing the friction factor for IFS as of compared to FS microchannel series.
32
This is because the IFS profile is more disruptive as of compared to Fish Scale
profile. For thorny D microchannel series, coefficient of U is 4/3. This means
that the raising of number of thorns due to decrease in thorn pitch(P) has the
largest impact in raising friction factor for D microchannel series.

The overall effect on friction losses can be observed from coefficient G alone.
The increase in number of scales/thorns from 10 to 40 resulted in an increase
of friction factor from 0.12 to 0.36 for D microchannel series. Friction factor
rises from 0.16 to 0.47 for IFS and 0.15 to 0.3 for FS microchannel series.

Table 2-15: Effect of P/e ratio on friction factor at Re≈2,664
P/e e/H Number of Friction factor
ratio ratio scales/thorns IFS FS D
20 0.5 10 0.16 0.15 0.12
5 0.5 40 0.47 0.30 0.36

To examine the suitability of applying the friction factor correlations beyond the
boundaries in present study, coefficient S is examined. It displays the impact
of scale/thorn profile on friction losses, when both ratios of e/H and P/e are set
to unity. The results are as of below:

Table 2-16: Prediction of friction factor for e/H ratio and P/e ratio of 1, at
Re≈2,664
Scale Friction factor
P/e e/H Number of
height
ratio ratio scale/thorns IFS FS D
(mm)
1 1 0.3 100 4.67 2.22 5.59

33
The friction factor values are therefore expected to converge to a single value
for the three different microchannel profiles. Hence, the correlations must be
used with caution beyond the current measurement limits.
34

2.4 Research gap

Although microchannels produces very high heat transfer rates, large-scale
applications have not been viable. This is mainly due to the challenge in
reduction of both the complexity and costly microfabrication processes if they
are to be used in normally sized fabrication, these fabrication approaches are
microelectromechanical systems(MEMS), photolithographic-beased
processes and X-ray micromachining [6]. With regards to this, in 2013, Kong
and Ooi [46] came out with a system of microscale heat transfer application in
macro geometry by inserting a cylindrical insert into a circular channel, creating
a annular microchannel. High heat transfer rate was reported and this adds on
to the viability of low cot macro geometries as it can be produced using
conventional methods.

Macro geometry shows that it is a suitable way of applying heat transfer
enhancement through simple methods like surface modification. In 2016, Goh
came out with a novel nature-inspired configuration to increase heat transfer
in macro geometry using normal machining processes. On top of that, Dr Goh
came out with working correlations and formulations for both means Nusselt
numbers and friction factors. The fish scale, inverted fish scale and durian
profiles were suggested. Moving on, there exists a research gap after
reviewing the findings of the literature review of the current study on microscale
passage. There is a need to come out with a set of design variables that can
result in a highest heat transfer performance within a low pressure loss.

This study hopes to fill up this research gap by making use of the current
correlation to run through an optimization programme, churning out a set of
design variables to maximize heat transfer coefficient and minimizes pressure
loss due to flow. This will potentially result in the implementation of microscale
microchannel heat transfer in normally sized heat exchangers.
35
Heat Transfer Chapter 3
Chapter 3: Heat Transfer

3.1. Classical theory for Circular Channel

3.1.1 Internal flow

Obtaining the total heat transfer rate is one of the key interests in forced
convective internal flow. Newton’s law of cooling equation will be illustrated as
follows:

𝑄 = ℎ𝐴(𝑇r − 𝑇s ) (3-1)
where h is the average convective heat transfer coefficient, A is surface area

where the process of heat flow occurs, Ts is surface temperature and Tm is mean
fluid temperature.

The problem is to calculate the mean convective heat transfer coefficient.
However, the complexity of the problem is that this coefficient depends on surface
geometry and flow conditions, on top of fluid properties [47]. There are many
factors that alters the convective heat transfer coefficient as heat transfer is
connected to the boundary layers that develop on the internal pipe surface.

Velocity and thermal boundary layer, the two layers that are of interest. The
velocity boundary layer is due to a deviation of the free stream velocity from the
zero surface velocity and the thermal boundary layer is due to the deviation
between free stream temperature and surface temperature. Friction factor is often
related to velocity boundary layer and convective heat transfer coefficient is
connected with the thermal boundary layer.
36
3.2. Hydrodynamic Consideration

3.2.1. Velocity Boundary Layer

Boundary layer idea was proposed by Ludwig Prandtl in 1904 [14], where he
gathered evidences to prove the existence of a thin layer near a solid boundary.
At this thin layer, the effects due to viscosity are as important as effects due to
inertia.

Figure 3-1 depicts the change in the velocity boundary layer for laminar flow inside
a circular pipe. The boundary layer will grow in the same direction as fluid flow
located at the hydrodynamic entrance region. Viscous effects are vital in this case
and fluid flow very near to the stationary wall is blocked. Due to mass conservation,
fluid flow is increasing at an increasing rate at the inviscid core. The boundary
layer will then combine at xfd,h if the pipe is long enough. For the developed
flow(fully) region, velocity profile is not dependent of x and viscous effects applies
as the flow is in the fully developed regime.

Figure 3-1: Laminar, hydrodynamic boundary layer development in a circular

pipe [47]
37
For flows that are both incompressible and has constant fluid properties, the fully
developed velocity profile is parabolic. The derivation of this profile is observed
after solving the Navier-Stokes equation in cylindrical coordinates(x-component).
Fully developed velocity profile for turbulent flow is flatter as there is turbulent
mixing in the radial direction. Figure 3-2 illustrates the difference in the velocity
profiles.
Figure 3-2: (a) Laminar and (b) Turbulent boundary layer development in circular
tube [49]
There are three distinct regions due to turbulent mixing: laminar sublayer,
turbulent core and buffer layer. Fluid flows orderly in the laminar sublayer while
fluid move in a chaotic manner in the turbulent core. The eddying motion will
create mixing of the fluid. The buffer layer is a transition between laminar sublayer
and turbulent core. The mean velocity profile also displays a logarithmic behavior
[50].
38
Figure 3-3: Fully developed turbulent flow in a smooth pipe [14]
Reynolds number will decide whether the flow is laminar or turbulent. A turbulent
or laminar flow will determine the extent of the hydrodynamic entrance region.
Reynolds number is determined by the following equation:
𝜌𝑢s 𝐷
𝑅𝑒j ≡ (3-2)
µμ
where um is the mean fluid velocity, and D is the pipe diameter.

Mean fluid velocity is defined from the conservation of mass as:
a}
𝑚 𝜌𝑢 𝑟 𝑑 𝐴x 𝜌𝑢 𝑟 2𝜋𝑟 𝑑𝑟
𝑢s ≡ =
z{
= F
(3-3)
𝜌𝐴x 𝜌𝐴x 𝜌𝜋𝑟O :
where ṁ is mass flow rate, ρ is density of fluid and Ac is the cross-section area.

For a fully developed internal flow, the transition to turbulence Reynolds number
is approximately 2300 [47]. This transition to turbulent starts in the developing
boundary layer of hydrodynamic entrance area. A fully turbulent condition is when
a Reynolds number of 10000 is achieved.

For laminar flow (ReD ≲ 2300), the hydrodynamic entry length xfd,h can be obtained
from [47]:
𝑥.•,E
≈ 0.05 𝑅𝑒j (3-4)
𝐷 0^s
39
There is no expression for entry length for the case of turbulent flow and it is found
to be not dependent of Reynolds number. As a simplified first approximation [51]:

𝑥.•,E
10 ≲ ≲ 60 (3-5)
𝐷 8àb
3.2.2. Friction Factor

It is important to determine the pump power required to maintain the flow in a
forced convective internal flow. This value is dependent on pressure drop across
the length of the pipe which can be calculated by:

𝐿 1 :
∆𝑝 = 𝑓 𝜌 𝑢s (3-6)
𝐷E 2

where f is friction factor,
ρ is density of fluid,
um is mean fluid velocity
L is the length of pipe,
Dh is hydraulic diameter of pipe.

Eqn. (3-6) arises from the definition of friction factor, also known as Moody or
Darcy friction factor:
− 𝑑𝑝 𝑑𝑥 𝐷
𝑓≡ (3-7)
𝜌𝑢s : 2
On the other hand, friction coefficient, also known as Fanning friction factor, is
defined as:
𝜏r
𝐶. ≡ (3-8)
𝜌𝑢s : 2
where τs is the surface shear stress, which is related to the surface velocity
gradient by Eqn. (3-9) for circular pipe:
40
𝑑𝑣ˆ
𝜏r = −𝜇 (3-9)
𝑑𝑟 a‰a}
where μ is dynamic viscosity of fluid.

Figure 3-4: Velocity profiles and friction factor variation for laminar flow in a
circular tube [55]
Figure 3-5 displays the changes in local friction factor along the pipe for turbulent
flow. The development of laminar velocity boundary layer comes be followed by
transition to the turbulent boundary layer. Friction factor becomes independent of
x when flow is fully developed.
41
Figure 3-5: Velocity profiles and friction factor variation for turbulent flow in a
circular tube [55]

Table 3-1 shows the friction factor correlations for laminar flow. It is observed that
for laminar flow, fully developed friction factor depends on Reynolds number only.

42
Table 3-1: Darcy friction factors for laminar flow in circular pipes
Laminar Range and
Correlation
Condition Remarks
∗ W4 ∗ W4/:
∗ W4 : 1.25 𝑧E‹ + 64 − 13.76 𝑧E‹ 1
Developing and 𝑓^dd = [13.76 𝑧E‹ + W: ]∙
∗
1 + 0.00021 𝑧E‹ 𝑅𝑒 (3-10)
fully developed
Shah and London, 1978 [45]
64
Fully 𝑓=
𝑅𝑒j (3-11) 𝑅𝑒 ≤ 2,100
developed
Derivation [47]

For turbulent flow, irregular surface disturbs and affects the flow and affects the
developed friction factor. As illustrated in Figure 3-6, there are three regimes of
turbulent pipe flow, with different friction factor correlations for each regime.
Figure 3-6: Regimes of turbulent pipe flow as a function of Reynolds number
and relative surface roughness [54]

43
Table 3-2 summarizes the Darcy friction factor correlations for turbulent flow. Fully
developed friction factor can also be found using Moody chart [55] and is valid for
both laminar and turbulent flow together with smooth and rough pipes.
Furthermore, the Moody chart is used for both circular and non-circular pipes.

Table 3-2: Darcy friction factors for turbulent flow in circular pipes
Turbulent Condition Correlation Remarks
𝑓^dd = 𝐴 𝑅𝑒 •
4.06448
Developing and 𝐴 = 0.3716 +
𝑧/𝑑E
fully developed, (3-12) 𝑅𝑒 < 28,000
0.31930
Smooth 𝐵 = −0.268 −
𝑧/𝑑E
Phillips, 1987 [56]
𝑓 = 0.316 𝑅𝑒 W4/@ 3000 ≤ 𝑅𝑒 ≤ 10A
Blasius, 1913 [23]
(3-13)
𝑓 = 0.184 𝑅𝑒 W4/A 𝑅𝑒 ≥ 10A
Fully developed,
First used by Taitel and Dukler [24] f depends only on Reynolds number.
Smooth
3000 ≤ 𝑅𝑒 ≤ 5×10G
𝑓 = (0.790 ln 𝑅𝑒 − 1.64)W:
(3-14) Single correlation valid over a large
First Petukhov equation [58]
Reynolds number range.
f depends on Reynolds number and
relative surface roughness ϵ/D, which
1 𝜀/𝑑E 2.51
= −2log + is the ratio of mean roughness height
𝑓 3.7 𝑅𝑒 𝑓 (3-15)
to pipe diameter.
Fully developed, Implicit Colebrook equation, 1939 [26]

Smooth and rough
Also applicable in transition region.
4.44
1 𝜖/𝐷 6.9 Approximate explicit relation which
≅ −1.8 lg +
𝑓 3.7 𝑅𝑒 (3-16) predicts results within 2% of those
Haaland equation, 1983 [53] from Colebrook equation.
3.2.3 Thermal Consideration

Developing and fully developed flows are two types of flow for hydrodynamic
considerations. With the introduction of thermal effects, type of flow will depend
on the region of introduction of temperature difference. As shown in Figure 3-7,
when temperature changes are after the flow is hydrodynamic developed, there
will be three flow regions. They are hydrodynamic developing flow, thermally
44
developing flow, and fully developed flow where the effects of viscous and heat
transfer apply to the cross-section of the pipe. As illustrated in Figure 3-7(b) and
Figure 3-7(c), if the temperature difference is introduced right at the pipe entrance,
then there are two flow regions. They are simultaneously developing flow, where
the velocity and thermal boundary layers start to develop together, and fully
developed flow, where both boundary layers are completely developed. Within
the simultaneously developing flow region, either one of the velocity and thermal
boundary layers may be fully developed first while the other is still developing,
depending on the Prandtl number.

Figure 3-7: Types of laminar flows for constant wall temperature boundary
condition [14]

45
3.2.4. Thermal Boundary Layer

Thermal boundary layer will be set up in forced convective flow when there is a
deviation in free stream fluid temperature and the surface temperature of the pipe.
For example, there is a cool flow into pipe which is very hot. When the cool fluid
particles contacted the hot surface, thermal equilibrium will be established
between the fluid particles and the hot surface temperature. There is a transfer of
heat energy to the next layer of particles at a lower temperature. This will lead to
the formation of temperature gradients in the fluid. This process will continue until
there is zero heat flow from the surface.

The figure below shows the development of a thermal layer of a heated circular
tube [47]. For fully developed region, the profile of temperature varies with surface
condition which may be constant temperature, constant heat flux and no slip-
surface conditions
Figure 3-8: Thermal boundary layer development in a heated circular tube [47]
The fully developed temperature profile changes as x increases for both of the
conditions of surface. The relative shape remains constant and can be described
by the dimensionless profile of temperature, which is defined below:
𝜕 𝑇r 𝑥 − 𝑇 𝑟, 𝑥
= 0 (3-17)
𝜕𝑥 𝑇r 𝑥 − 𝑇s 𝑥 .•,8

46
where Ts is surface temperature of pipe, T is local fluid temperature, and Tm is
mean temperature of fluid over the cross-section of the tube.

For laminar flow (ReD ≲ 2300), the thermal entry length xfd,t can be obtained from:
𝑥.•,8
≈ 0.05 𝑅𝑒j 𝑃𝑟 (3-18)
𝐷 0^s

For turbulent flow, conditions are nearly independent of Prandtl number and can
be approximated simply by:
𝑥.•,8
= 10 (3-19)
𝐷 8àb

The above equations suggest that Prandtl number is important in the
determination of relative growth in both the velocity and thermal boundary layer.
Prandtl number can be approximated as δ/δt, where δ and δt is the velocity and
thermal boundary layer thickness, respectively. This is further illustrated by the
figure below:

Figure 3-9: Velocity and thermal boundary layer development in a circular tube
for Pr > 1 [53]
47
The hydrodynamic and thermal entrance lengths of turbulent flow are not as long
as those for laminar flow. This means that, for turbulent flow, correlations for fully
developed flow need not consider the two mentioned entrance regions.
3.2.5 Convective Heat Transfer Coefficient

Table 3-3 is the convective heat transfer h of fluid under flow conditions [14]. It
can also be observed that for single-phase liquid water has the h values to be at
most 10 kW/m2·K

Table 3-3: Typical values of convective heat transfer coefficient [14]
h (W/m2·K)
Fluid Free Forced
Convection Convection
Gases 5-30 30-300
Water 30-300 300-10,000
Viscous oils 5-100 30-3,000
Liquid metals 50-500 500-20,000
Boiling water 2,000-20,000 3,000-100,000
Condensing water vapour 3,000-30,000 3,000-200,000

48
Figure 3-10: Convective heat transfer coefficients for different flow configuration
[59]
The local convective heat transfer coefficient h, computed using Nusselt number
with this equation:

𝑁𝑢𝑘.
ℎ= (3-20)
𝐷
where kf is fluid thermal conductivity.

Nusselt number is equal to the dimensionless temperature gradient at the wall
surface:
𝜕𝑇 ∗
𝑁𝑢 = ∗ (3-21)
𝜕𝑟 a ∗ ‰4
49
As mentioned, the thermal boundary layer is developing as x increases. As heat
transfer effects moves towards the pipe centre, the temperature gradient of the
wall decreases along x. Therefore, both Nusselt number and convective heat
transfer coefficient drops as x increases. This will continue till the flow is fully
developed thermally. By definition, the dimensionless temperature profile that is
fully developed does not depend on x where it also means that the Nusselt
number and heat transfer coefficient does not depend on x. Thermal entrance
length is defined as the axial distance to obtain a local Nusselt number that is 1.05
times the fully developed Nusselt number.

Figure 3-11: Variation of friction factor and convective heat transfer coefficient
with velocity boundary layer and thermal boundary layer development for Pr > 1
[53]

Table 3-4 is a summary of all the Nusselt number correlations for the case of
laminar flow. For fully developed region, Nusselt number is not dependent on
50
Reynolds number and is a constant. Table 3-5 is a summary of Nusselt

correlations that can be applied to both transitional and turbulent flow. It is valid
for two cases, constant surface temperature and flux boundary conditions,
Nusselt number correlations can estimate the heat transfer coefficient due to
convection.

Table 3-4: Local and average Nusselt numbers for laminar flow in circular pipes
Laminar Conditions Correlation Remarks
Fully developed, 𝑁𝑢 = 3.66

(3-22)
constant Ts Derivation [60]
Thermally 0.065𝑅𝑒 𝑃𝑟 𝐷E 𝐿
𝑁𝑢 = 3.66 +
developing, 1 + 0.04 𝑅𝑒 𝑃𝑟 𝐷E 𝐿 :/D
(3-23) 𝑃𝑟 ≲ 5
constant Ts Edwards, 1979 [61]
Thermally 0.19 𝑅𝑒 𝑃𝑟 𝐷E 𝐿 F.7
𝑁𝑢 = 3.66 +
developing, 1 + 0.117 𝑅𝑒 𝑃𝑟 𝐷E 𝐿 F.@G9 (3-24) 𝑅𝑒 < 2,200
constant Ts Hausen, 1959 [62]
0.60 ≤ 𝑃𝑟 ≤ 5
0.0044 ≤ 𝜇 𝜇r ≤ 9.75
Simultaneously 𝑅𝑒𝑃𝑟𝐷 4/D
𝜇b F.4@
𝑁𝑢 = 1.86 ≥ 3.66 Re<2200
developing, 𝐿 𝜇r (3-25)
Accounts for viscosity variation
constant Ts Sieder and Tate, 1936 [63]
due to temperature difference
between surface and fluid.
48
Fully developed, 𝑁𝑢 = = 4.36
11 (3-26)
constant qs
Derivation [47]
𝑁𝑢 = 1.953(𝑅𝑒 𝑃𝑟 𝐷E 𝐿)4/D
Thermally for 𝑅𝑒 𝑃𝑟 𝐷E 𝐿 ≥ 33.3
developing, 𝑁𝑢 = 4.364 + 0.0722𝑅𝑒 𝑃𝑟 𝐷E 𝐿 (3-27) 𝑅𝑒 < 2,200
constant qs for 𝑅𝑒 𝑃𝑟 𝐷E 𝐿 < 33.3

51
Table 3-5: Average Nusselt numbers for turbulent flow in circular pipes
Turbulent Conditions Correlation Remarks
Fluid properties computed

4/D
Smooth and 𝑁𝑢 = 0.125 𝑓 𝑅𝑒j 𝑃𝑟 using bulk mean fluid
(3-28)
rough pipes Chilton–Colburn analogy [53] temperature 𝑇b = 1/2(𝑇_I +
𝑇O`8 )
0.7 ≤ 𝑃𝑟 ≤ 160
@/A
𝑅𝑒 > 10,000
𝑁𝑢 = 0.023 𝑅𝑒j 𝑃𝑟 4/D
(3-29) 𝐿/𝑑E ≥ 10
Colburn equation [53]
Substitute Blasius correlation
into Chilton–Colburn analogy
@/A 0.60 ≤ 𝑃𝑟 ≤ 160

𝑁𝑢 = 0.023 𝑅𝑒j 𝑃𝑟 I
𝑅𝑒 ≥ 10,000
𝑛 = 0.4 𝑤ℎ𝑒𝑛 𝑇rà. > 𝑇scÎ (heating)
(3-30) 𝐿/𝑑E ≥ 10
𝑛 = 0.3 𝑤ℎ𝑒𝑛 𝑇rà. < 𝑇scÎ (cooling)
Improved version of Colburn
Dittus–Boelter, 1930 [64]
equation
0.7 ≤ 𝑃𝑟 ≤ 16,700
𝑅𝑒 ≥ 10,000
𝐿/𝑑E ≥ 10
Accounts for viscosity variation
@/A 𝜇 F.4@
𝑁𝑢 = 0.027 𝑅𝑒j 𝑃𝑟 4/D due to large temperature
𝜇r (3-31)
difference between surface
Sieder and Tate, 1936 [63]
Fully developed, and fluid.
Smooth pipe Properties are evaluated at Tb
except μs, which is evaluated

at Ts
𝑓 8 𝑅𝑒j 𝑃𝑟
𝑁𝑢 =
1.07 + 12.7 𝑓 8 F.A (𝑃𝑟 :/D − 1) 1 ≤ 𝑃𝑟 ≤ 2000
Second Petukhov equation, 1970 10@ < 𝑅𝑒j < 5×10A
Slightly complex and
(3-32)
1 applicable for a smaller range,
𝑓=
(1.82 log4F 𝑅𝑒 − 1.64): but improves accuracy to less
Filonenko equation, 1954 [65] than 10% error.

1 ≤ 𝑃𝑟 ≤ 10G
𝑓 8 (𝑅𝑒j − 1000) 𝑃𝑟 2300 ≤ 𝑅𝑒 ≤ 5×10G
𝑁𝑢 =
1 + 12.7 𝑓 8 F.A (𝑃𝑟 :/D − 1) 𝐿/𝑑E ≥ 10
Gnielinski’s equation, 1976 [66] Improves accuracy of second
(3-33)
Petukhov equation at lower
𝑓 = (0.790 ln 𝑅𝑒 − 1.64)W: Reynolds number
First Petukhov equation [58] Generally preferred in
calculations
52
Fluid properties are evaluated

at Tb.
𝑓 8 (𝑅𝑒j − 1000) 𝑃𝑟 𝜇 I
𝜇
𝑁𝑢 = 1≤ ≤ 40
1 + 12.7 𝑓 8 F.A ( 𝑟 :/D − 1) 𝜇r 𝜇r
𝑛 = 0.11 𝑤ℎ𝑒𝑛 𝑇rà. > 𝑇scÎ 0.5 ≤ 𝑃𝑟 ≤ 140
𝑛 = 0.25 𝑤ℎ𝑒𝑛 𝑇rà. < 𝑇scÎ (3-34)
3000 ≤ 𝑅𝑒 ≤ 5×10G
𝑓 = (0.790 ln 𝑅𝑒 − 1.64)W: Accounts for variation in fluid
First Petukhov equation [58] properties

53
3.3. Classical Theory for Concentric Annular Channel

The concentric annular channel is generally quantified by:
𝑟_
𝑟 ∗ = (3-35)
𝑟O

One of the limitation of annular channel is where r*= 1 while the other is the
circular pipe with an infinitesimal core at the centre, where r*=0.

Based on the definition:
𝐷E = 𝐷O − 𝐷_ (3-36)

3.3.1 Hydrodynamic Consideration

Fully developed laminar velocity profile for concentric annular channel is given by
[72]:
1 𝑑𝑝 𝑟 : 𝑟
𝑢 𝑟 =− 𝑟O : 1 − + 2𝑟s∗ : ln (3-37)
4𝜇 𝑑𝑥 𝑟O 𝑟O

4/:
𝑟s 1 − 𝑟∗:
𝑟s∗ = = (3-38)
𝑟O 2 ln(1/𝑟 ∗ )

Here, rm refers to the radius where maximum velocity occurs (𝜕𝑢/𝜕𝑟 = 0).
The mean fluid velocity can be found by:

1 𝑑𝑝
𝑢s = − 𝑟O : (1 + 𝑟 ∗ : − 2𝑟s∗ : ) (3-39)
8𝜇 𝑑𝑥

54
The circumferentially averaged fully developed friction factor for laminar flow is:
64 1 − 𝑟 ∗ :
𝑓𝑅𝑒 = (3-40)
1 + 𝑟 ∗ : − 2𝑟s∗ :

It is noted that Eqn. (3-40) reduces to 𝑓𝑅𝑒 = 64 for r*=0 (circular channel).

Friction factor for the fully developed is given by the equation (3-41) at inner and
outer wall. The latter are given by Eqn. (3-42) and Eqn. (3-43) respectively.

𝑓_ 𝑟_ + 𝑓O 𝑟O
𝑓= (3-41)
𝑟_ + 𝑟O
4 𝑑𝑝 𝐷E 𝑟s: − 𝑟_:
𝑓_ 𝑅𝑒 = − (3-42)
𝜇 𝑑𝑥 𝑢s 𝑟_
4 𝑑𝑝 𝐷E 𝑟O: − 𝑟s:
𝑓O 𝑅𝑒 = − (3-43)
𝜇 𝑑𝑥 𝑢s 𝑟O

The hydrodynamic entrance lengths for laminar flow in annular channels of
various r* are reviewed by Shah and London [45]. The dimensionless
hydrodynamic entrance length for r* = 0.97 is given by:

𝑥.•,E
≈ 0.0108 (3-44)
𝐷E 𝑅𝑒 0^s

Reynolds number for annular channel terms of mass flow rate:
𝑚 𝐷E 4𝑚
𝑅𝑒 = = (3-45)
𝐴x µμ 𝜋(𝐷O + 𝐷_ )µμ
𝐷E = 𝐷O − 𝐷_ (3-46)
𝜋
𝐴x = (𝐷 : − 𝐷_ : ) (3-47)
4 O

55
Table 3-6: Darcy friction factor for laminar flow in parallel plates
Laminar
Correlation Remarks
Condition
∗
𝑧
𝑧E‹ =
𝐷E 𝑅𝑒
∗ W4 ∗ W4/:
13.76 0.674 𝑧E‹ + 96 − 13.76 𝑧E‹
Developing and 𝑓^dd 𝑅𝑒 = + Reference for
4 : ∗ W:
∗
𝑧E‹ 1 + 0.000029 𝑧E‹ (3-48)
fully developed the annular
Shah and London, 1978 [45] channel in the
present study
3.3.2. Thermal Consideration

The thermal entrance lengths for laminar flow in annular channels of various r*
are also reviewed by Shah and London [45]. The dimensionless thermal entrance
length for r* = 0.97 is given by:

𝑥.•,8
≈ 0.04101 (3-49)
𝐷E 𝑅𝑒𝑃𝑟 0^s

Table 3-7 shows the Nusselt number correlation for laminar flow in parallel plates.
Table 3-7: Average Nusselt number for laminar flow in parallel plates
Laminar Condition Correlation Remarks
4/D
𝑁𝑢 = 2.236(𝑅𝑒𝑃𝑟 𝐷E 𝐿)
for 𝑅𝑒𝑃𝑟 𝐷E 𝐿 ≥ 1000
Thermally 𝑁𝑢 = 2.236(𝑅𝑒𝑃𝑟 𝐷E 𝐿)4/D + 0.9 𝑅𝑒 < 2,200
developing, for 100 < 𝑅𝑒𝑃𝑟 𝐷E 𝐿 < 1000 (3-50) Reference for the annular channel
constant qs 𝑁𝑢 = 8.235 + 0.0364𝑅𝑒𝑃𝑟 𝐷E 𝐿 in the present study
for 𝑅𝑒𝑃𝑟 𝐷E 𝐿 ≤ 100
56
3.4. Passive Heat Transfer Enhancement Mechanisms

According to Bergles et al. [67] in 1996, there are two classifications of heat
transfer improvement techniques for single phase liquid flow are, passive and
active. The passive techniques do not require any external power and hence, they
are preferred for many applications of engineering [68]. These techniques include
flow disruption, secondary flows, channel curvature, surface treatments, and
entrance effects. Three heat transfer enhancement mechanism which are
plausible have been suggested by Tao et al. [69] in 2002. These three techniques
are by thinning of the thermal boundary layer, raising flow interruption, and to
increase velocity gradient near the heated surface.

The re-developing thermal boundary layer so as to increase and optimize heat
transfer has been proven by conducting experiments by Xu et al. [70] in 2005. It
was conducted by using single-phase deionized water with Reynolds number that
is between 100 and 2000 while the heat flux is up to 100 W/cm2. The results
proved that the Nusselt number of 155 µm triangular channels can be increased
by 26% relative to channel without transverse passages. There were also
evidences that show that these enhanced liquid mixing can promote early
transition of flow from laminar to turbulent flow at critical Reynolds number of 1000
to 1100. A similar experiment was conducted using oblique fins in microchannel
by Lee et al. [46] in 2015. The usage of deionized water with Reynolds number
between 180 and 680 yielded an optimized heat transfer enhancement of 47%. In
the same year, Deng et al. [71] also came out with a special re-entrant
microchannel heat sing and concluded that it can increase Nusselt number up to
39% when compared to rectangular microchannels.

The concept of Dean vortices, a form of secondary flow that improves fluid mixing
was substantiated by Sui et al. [72] where sinusoidal channels were used to
57
improve performance of microchannel heat sinks due to centrifugal forces

generated when liquid flows through the curved paths with high Reynolds number.

3.5 Thermo-hydraulic Performance

The thermo-hydraulic performance is given by:
4
𝑁𝑢 𝑓 D
(3-1)
𝜂 =
𝑁𝑢]0^_I 𝑓]0^_I

58
Optimization Method Chapter 4
Chapter 4: Optimization Method

The objective of this study is to find a set of design parameters to maximize heat
transfer coefficient. A direct search multi-variable constrained optimization
approach attributed to M.J. Box [2] namely Complex optimization method was
used to obtain the set of design parameters for the intended heat exchanger
design.

Generally, the complex optimization method used in this project can be presented
mathematically as shown below:
1. To maximize/minimize the general objective function: 𝐹 𝑥 =

𝑓(𝑥4 , 𝑥: , … , 𝑥Ÿ ) with Explicit, geometrical and implicit constraints:
i. 𝐿 (𝑥)_ ≤ 𝐸 𝑥 _ ≤ 𝐻 𝑥 _ , 𝑖 = 1,2, … , 𝑁
ii. 𝐿H (𝑥)¢ ≤ 𝐺 𝑥 ¢ ≤ 𝐻H 𝑥 ¢ , 𝑗 = 1,2, … , 𝑀
iii. 𝐿h (𝑥)¦ ≤ 𝐼 𝑥 ¦ ≤ 𝐻h ¦ , 𝑘 = 1,2, … , 𝐿
where E, G and I stand for the explicit, geometrical and implicit constraints,
L and H stand for the lower and higher limit, i, j and k stand for the number
of explicit, geometrical and implicit constraints respectively.

An initial point that satisfies all constraints was chosen. The pseudo-random
numbers generator then generates K-1 extra points where each independent
variable satisfies the given constraints.

𝑥_,¢ = 𝐿 𝑥_ + 𝑟_.¢ × 𝐻 (𝑥_ − 𝐿 𝑥_ ] (4-1)
𝑖 = 1,2, … , 𝑁;
𝑗 = 1,2, … , 𝐾 − 1.

59
where the extra points produced together with the initial point are termed
as the original complex.

N is the number of explicit free variable constraints and K = N + 1.
𝐿 𝑥_ and 𝐻 (𝑥_ ) are the lower and higher limit of the free variables.
𝑟_.¢ is the pseudo-random number between 0 and 1.
xi,j is the generated variable that satisfies the given constraint that is
bounded by the upper and lower limits.

The points generated satisfy explicit constraints but may in turn not satisfying
other constraints. In the case where geometric or implicit constraint is not satisfied,
the point will be reflected by half the distance to the centroid of the original point
reflected as shown in the formula depicted below. The centroid of the point is
determined by the average sum of all the points.

ˆ©,ª O0• Vˆ©,{
𝑥_,¢ 𝑛𝑒𝑤 = , 𝑖 = 1,2, … , 𝑁 (4-2)
:
with the centroid 𝑥_,x defined as

4 «
𝑥_,x = ¢‰4 𝑥_,¢ − 𝑥_,¢ ( 𝑙𝑑) , 𝑖 = 1,2, … , 𝑁 (4-3)
«W4

The process is iterated to a point where no more constraints are violated.
For the set of points, the objective function will be evaluated and the point that
has the lowest objective function value will be reflected by a stipulated factor for
reflection along the line that connects the previous point and the centroid. The
new point is:
𝑥_,¢ 𝑛𝑒𝑤 = 1 + 𝛼 𝑥_,x − 𝛼𝑥_,¢ 𝑜𝑙𝑑 (4-4)
𝑥_,¢ (𝑜𝑙𝑑) is the point with the lowest objective function value.
𝑥_,x is the calculated centroid of the remaining points.

60
To improve the efficiencies of optimization search, the centroid is calculated by
assigning weightage to each point according to their objective function value. The
method weighted arithmetic mean where the centroid is calculated by weighing
each point to their objective function value. The equation for determining the
centroid is as follows:
· ¯ª °¯±}²³´
ª¸¹ ˆ©,ª ¯
µ¶³´ °¯±}²³´
𝑥_,x = ¯ª °¯±}²³´ , 𝑖 = 1,2, … , 𝑁 (4-5)
·
ª¸¹ ¯
µ¶³´ °¯±}²³´
2. In the case where the point violates an explicit constraint after it is reflected,
the point is then determined by moving it to the upper/lower limit by a factor
of 𝛽=0.0001:
𝑥_,¢ 𝑛𝑒𝑤 = 𝐻 𝑥 − 𝛽 𝑖𝑓 ℎ𝑖𝑔ℎ𝑒𝑟 𝑏𝑜𝑢𝑛𝑑𝑎𝑟𝑦 𝑖𝑠 𝑣𝑖𝑜𝑙𝑎𝑡𝑒𝑑 (4-6)

𝑥_,¢ 𝑛𝑒𝑤 = 𝐿 𝑥 + 𝛽 𝑖𝑓 𝑙𝑜𝑤𝑒𝑟 𝑏𝑜𝑢𝑛𝑑𝑎𝑟𝑦 𝑖𝑠 𝑣𝑖𝑜𝑙𝑎𝑡𝑒𝑑 (4-7)

3. The last step is to check if the objective function value has improved and
converged. If the deviation between the highest objective function value
and the lowest objective function value given that all constraints are
satisfied lies within a tolerance, the particular point where the highest
objective function value will be selected.
This set of procedure will be repeated to come out with a few set of initial
variables to find a few local optimum points.
The local optimum points are then compared to find the global optimum points
and they will be the optimized design variables of the heat exchangers.
61

4.1. Flow chart of Algorithm

The set of procedures of the algorithm is shown in the form of a flow chart for
better understanding of the entire optimization procedure.

Figure 4-1: Flow Chart for Optimisation Process
62

4.3 Heat Exchange Performance Optimisation

In this project, the objective is to find a set of parameters of a microchannel heat
exchanger to maximize different objective functions which are given as
Ÿ`
𝐹Ob¢4 = (4-8)
Ÿ`Á
𝐹Ob¢: = 𝑁𝑢 (4-9)
ÂÃ
ÂÃÁ
𝐹Ob¢D = ¹
°Ä ( )
(4-10)
( ) Å
ÄÁ
𝐹Ob¢@ = −𝑓 (4-11)
W.
𝐹Ob¢A = (4-12)
.Á

Subjected to constraints:
𝑥0 𝑖 < 𝑥 𝑖 < 𝑥E 𝑖 , 𝑖 = 1,2,3. (4-13)

Eq. (4-8) to Eq. (4-12), the objective function which is performance which is the
ratio of Nusselt number for the enhanced profile to Nusselt number for the plain
channel. This accounts for rate of heat transfer, since variation in the design
variables will result in different flow rates and heat transfer which in turn affect the
final result of objective function. By looking at the different objective function value,
the highest objective function value will produce the best design variables that will
be shown by the program.

In Eq. (4-13), 𝑥 𝑖 , 𝑖 = 1,2,3 are design independent variables e/H, P/e and Re.
Here, e/H, P/e and Re are scale height(e) to mean channel height (H) ratio, scale
pitch (P) to scale height (e) ratio and Reynolds number. Furthermore, the values
of 𝑥0 𝑖 and 𝑥E 𝑖 are the minimum and maximum cap of the design variable.

63

Table 1 shows the limits of the constraints for each variable. The final design’s
range will be within the stipulated constraints.
Table 4-1: Constraints of independent variables

No. Lower Limit Design Variables Upper Limit Unit
1 350 Re 4600 -
2 0.1 e/H 0.7 -
3 5.0 𝑃/𝑒 20.0 -
64
Results and Discussion Chapter 5
Chapter 5: Results and Discussion

The optimization algorithm has been implemented using Fortran programming
language. A set of variables which optimizes heat transfer while keeping pressure
drop low was obtained. This set of variables include e/H, P/e and Re which are
scale height(e) to mean channel height (H) ratio, scale pitch (P) to scale height (e)
ratio and Reynolds number respectively. The tables below show the results of the
ratios for the different profiles (Inverted Fish Scale, Fish Scale and Durian) and
objective functions.

Ÿ`
Table 5-1: Optimization Results of for the three different profiles
Ÿ`Á
Nusselt Number/
Reynolds
Profile Nusselt Plain e/H P/e
number
Ratio

Inverted Fish
2.8 4600 0.7 5.0
Scale

Fish Scale 2.1 4600 0.7 5.0

Durian 2.5 350 0.7 5.0

65

Table 5-2: Optimization Results of Nu for the three different profiles
Reynolds
Profile Nusselt Number e/H P/e
number

Inverted Fish
70 4600 0.7 5.0
Scale

Fish Scale 50 4240 0.7 5.0

Durian 61 4600 0.7 5.0

ÂÃ
ÂÃÁ
Table 5-3: Optimization Results of ¹
°Ä ( )
for the three different profiles
( ) Å
ÄÁ
Hydraulic Reynolds
Profile e/H P/e
Performance number

Inverted Fish
1.34 350 0.7 5.0
Scale

Fish Scale 1.31 350 0.7 20

Durian 1.29 350 0.7 20

66

Table 5-4: Minimization Results of f for the three different profiles
Reynolds
Profile Friction Factor e/H P/e
number

Inverted Fish
0.08 4600 0.1 20
Scale

Fish Scale 0.08 4600 0.1 20

Durian 0.08 4600 0.1 20

.
Table 5-5: Minimization Results of for the three different profiles
.Á
Friction
Reynolds
Profile factor/Friction e/H P/e
number
Plain Ratio

Inverted Fish
1.2 4600 0.1 20
Scale

Fish Scale 1.2 4600 0.1 20

Durian 1.2 4600 0.1 20

67

5.1. Inverted Fish Scale Profile

The results from this study showed that there exist a set of parameters to
maximize heat transfer as shown in the above table. This study will analyze and
interpret the results for Nusselt number to Nusselt Plain number ratio. The
maximum Nusselt number to Nusselt Plain number ratio is 2.8 by inverted fish
scale which is the highest value out of the three profiles. The scale height(e) to
mean channel height(H) ratio and scale pitch(P) to scale height ratio are 0.7 and
5.0 respectively. Below are the plots of the changes in Reynolds number, e/H ratio,
P/e ratio and Nusselt number to Nusselt Plain number ratio over several iterations
as exhibited by the Inverted Fish Scale Profile.

5000
4500
4000
3500
Reynolds Number
3000
2500
2000
1500
1000
500
0
1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97
Number of iterations

Figure 5-1: Inverted Fish Scale profile’s Reynolds number iteration results

68
0.8
0.7
0.6
0.5
e/H ratio
0.4
0.3
0.2
0.1
0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101

Figure 5-2: Inverted Fish Scale profile’s e/H iteration results

25
20
15
P/e ratio
10
0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101

Figure 5-3: Inverted Fish Scale Profile’s P/e iteration results

69
3
2.5
Nusselt/Nusselt Plain Ratio
1.5
0.5
0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101

Figure 5-4: Inverted Fish Scale Profile’s Nu/NuPlain iteration results

With the optimized e/H and P/e as 0.7 and 5.0 respectively, this study shall
examine the reasons behind why different scale height(e) give similar heat
transfer at the same e/H, P/e and Reynolds number. The values of scale height(e)
can be altered together with the mean height of channel and scale pitch to give a
constant e/H of 0.7 and P/e of 5.0 in order to achieve maximum heat transfer.

An increase in scale height(e) will mean an increase in mean channel height (H)
and scale pitch (P) in order to keep e/H and P/e constant. The narrower the gap
where fluid flows, the higher the Nusselt number, at an optimum flow rate, the
incompressible fluid is to flow through the small gap which resulted in a rise in
fluid velocity due to mass conservation. This large velocity of the fluid will produce
a thinner thermal boundary later along the copper wall. For a constant mean
channel height, thermal boundary layer will keep growing along the channel until
fully grown state. At higher e/H of 0.7, it is evident that there is significant thinning
of thermal boundary layer which resulted in a higher Nusselt number. Figure 5-5
and 5-6 show two microchannels with the same e/H and P/e ratios but different
scale height(e), mean channel height(H) and scale pitch(P). Mean channel
70

height(H) of the figure 5-6 is 1.5 times the mean channel height of figure 5-6. With
the same e/H and P/e ratios, the two microchannels will display maximum heat
transfer.
Direction of
flow H=0.3mm
e
P

Figure 5-5: Inverted Fish Scale Profile H=0.3mm
Direction of flow
H=0.45mm

e
P

Figure 5-6: Inverted Fish Scale Profile H=0.45mm

71

5.2. Fish Scale Profile

The results from this study showed that the optimized set of parameters for a
maximum heat transfer as shown in table 5-1. The maximum Nusselt number to
Nusselt Plain number ratio is 2.1 by Fish Scale Profile. The scale height(e) to
mean channel height(H) ratio and scale pitch(P) to scale height ratio are 0.7 and
5.0 respectively. Below are the plots of the changes in Reynolds number, e/H ratio,
as exhibited by the Fish Scale Profile.

5000
4500
4000
3500
Reynolds Number
3000
2500
2000
1500
1000
500
0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101

Figure 5-7: Fish Scale profile’s Reynolds number iteration results

72
0.8
0.7
0.6
0.5
e/H ratio
0.4
0.3
0.2
0.1
0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101

Figure 5-8: Fish Scale profile’s e/H iteration results

25
20
15
P/e ratio
10
0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101

Figure 5-9: Fish Scale Profile’s P/e iteration results

73
2.5
Nusselt/Nusselt Plain ratio
2
1.5
0.5
0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101

Figure 5-10: Fish Scale Profile’s Nu/NuPlain iteration results

The heat transfer mechanism for Fish Scale microchannel profiles is very similar to
Inverted Fish Scale profiles which has been discussed in the previous section. It was
shown that the optimum e/H ratio is 0.7. Inverted Fish Scale profile has a higher
Nusselt number as of compared to Fish Scale profile at e/H of 0.7. This is because
the maximum fluid velocity for Inverted Fish Scale profile is higher than that of the
Fish Scale Profile which caused a thinner boundary layer as there is an immediate
narrowing of channel gap in Inverted Fish Scale microchannel after the scale tip while
a sudden widening of channel gap for Fish Scale microchannel after the scale tip. .
Figure 5-11 and 5-12 show two microchannels with the same e/H and P/e ratios
but different scale height(e), mean channel height(H) and scale pitch(P). Mean
channel height(H) of the figure 5-12 is 1.5 times the mean channel height of figure
5-11. With the same e/H and P/e ratios, the two microchannels will display
maximum heat transfer.

74
Direction of
flow H=0.3mm
e

P

Figure 5-11: Fish Scale Profile with H=0.3mm
Direction of flow
H=0.45mm

e
P

Figure 5-12: Fish Scale Profile with H=0.45mm

75

5.3. Durian Profile

The results from this study showed that the optimized set of parameters for a
maximum heat transfer as shown in Table 5-1. The maximum Nusselt number to
Nusselt Plain number ratio is 2.5 by Durian Profile. The scale height(e) to mean
channel height(H) ratio and scale pitch(P) to scale height ratio are 0.7 and 5.0
respectively. Below are the plots of the changes in Reynolds number, e/H ratio,
as exhibited by the Inverted Durian Profile.

5000
4500
4000
3500
Reynolds number
3000
2500
2000
1500
1000
500
0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101

Figure 5-12: Durian Profile’s Reynolds number iteration results

76
Chart Title
0.8
0.7
0.6
0.5
e/H ratio
0.4
0.3
0.2
0.1
0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101

Figure 5-13: Durian Profile’s e/H iteration results

25
20
15
P/e ratio
10
0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101

Figure 5-14: Durian Profile’s P/e iteration results

77
3
2.5
Nusselt/ Nusselt Plain ratio
1.5
0.5
0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 101

Figure 5-15: Durian Profile’s Nu/NuPlain iteration results

The heat transfer mechanism for Durian microchannel profiles is very similar to
Inverted Fish Scale profiles which has been discussed in section 5.1. It was shown
by optimization that the optimum e/H ratio is 0.7. Inverted Fish Scale profile has a
higher Nusselt number as of compared to Durian profile at e/H of 0.7. The maximum
fluid velocity of Fish Scale channel and Durian Channel is similar. The wavy durian
profile promotes a progressive contraction and expansion which resulted in the flow
for Durian microchannel having fluid velocity for both upstream and downstream flow
for a larger region. This causes the decrease in size of the thermal boundary layer
and a larger Nusselt number for Durian microchannel as of compared to Fish Scale
microchannel which has a smaller region of thinner thermal boundary. Figure 5-16
and 5-17 show two microchannels with the same e/H and P/e ratios but different
scale height(e), mean channel height(H) and scale pitch(P). Mean channel
height(H) of the figure 5-17 is 1.5 times the mean channel height of figure 5-16.
With the same e/H and P/e ratios, the two microchannels will display maximum
heat transfer.

78

Direction of

flow H=0.3mm

e

P

Figure 5-16: Durian Profile with H=0.3mm
Direction of flow
H=0.45mm

e

P

Figure 5-17: Durian Profile with H=0.45mm

79
Conclusion and Future Works Chapter 6
6. Conclusion and Future Works

6.1 Conclusion
In essence, the current study aims to make use of working correlations proposed
by Goh, and run a computerized optimization method to suggest a set of design
parameters to optimize heat transfer with a given set of constraints for the three
different microchannel profile namely, Inverted Fish Scale, Fish Scale and Durian
profile. The optimization programme was written in Fortran programming
language. The design parameters suggested were Reynolds number, scale
height to mean channel height ratio (e/H), scale pitch to scale height ratio (P/e)
for the three different profiles. A total of five objective functions were studied which
included, Nusselt number to Nuplain number ratio, Nusselt number, hydraulic heat
performance, friction factor and friction factor to frictionplain ratio. For Nusselt
number to Nuplain number ratio, the optimal values of Reynolds number, e/H, P/e
were 4600, 0.7 and 5.0 for all the three microchannel profiles except for the
Reynolds number for durian profile which was 350. The results prove that it is
plausible to apply microscale heat transfer in today’s context due to the
enhancement of heat transfer that microscale systems can offer.

6.2 Future Works
6.2.1 Design of microchannel profiles
The current study suggested the various design parameters that can be
actualized in real-life context. The next step is bringing this study forward to
designing and fabricating a maro heat exchanger with these microchannels to
test the feasibility of this concept in large scale applications. This will be
beneficial to many industries that depend on the use of heat exchangers.
80
Conclusion and Future Works Chapter 6
6.2.2 Extend the range of design constraints for correlations

This study is limited by the range of design constraints for both e/H and P/e
which are from 0 to 0.7 and 5 to 20 respectively. There is a need for further
research in this area to test and expand the applicability of correlations to
greater range for e/H and P/e ratios. This will result in more flexibility in the
design of microchannel profiles which is definitely useful in the design phrase of
the microchannel.

6.2.3 Different microchannels profile

Another possible future work is to explore and experiment new microchannel
profiles that can help to enhance heat transfer rate further and yet reduces
pressure drop by applying constructal and/or synergy theories. More research
and design efforts can also be put into this area to come out with more
applicable and appreciable design of microchannels to ensure a better heat
macro scale exchange performance.
81
LIST OF REFERENCE
List of References

[1] A.L. Goh, K.T. Ooi, Nature-inspired Inverted Fish Scale microscale
passages for enhanced heat transfer, Int. J. Therm. Sci. 106 (2016) 18–
31

[2] M. J. Box;; A New Method of Constrained Optimization and a Comparison
with Other Methods. Comput J 1965;; 8 (1): 42-52. doi:
10.1093/comjnl/8.1.42

[3] Ooi, K.T., Teh,Y.L., Ng, P.F., Phay, T.T., 2008, A Novel Thermal Solution
for Electronics Cooling, Part I: Design of a Multi-Evaporator Refrigeration
System, Int’l Refrigeration and Air Conditioning Conf at Purdue, No. 2381

[4] D.B. Tuckerman, R.F.W. Pease, High-performance heat sinking for VSLI,
Electron Device Lett. 2 (1981) 126–129.
[5] T. Bayraktar, S.B. Pidugu, Characterization of liquid flows in microfluidic
systems, Int. J. Heat Mass Transf. 49 (2006) 815–824.
[6] N.T. Obot, Toward a better understanding of friction and heat/mass
transfer in microchannels — a literature review, Microscale Thermophys.
Eng. 6 (2002) 155–173.
[7] B. Agostini, M. Fabbri, J.E. Park, L. Wojtan, J.R. Thome, B. Michel, State
of the Art of High Heat Flux Cooling Technologies, Heat Transf. Eng. 28
(2007) 258–281.
[8] S.G. Kandlikar, S. Colin, Y. Peles, S. Garimella, R.F. Pease, J.J.
Brandner, D.B. Tuckerman, Heat Transfer in Microchannels — 2012
Status and Research Needs, ASME J. Heat Transf. 135 (2013) 091001.
[9] M.E. Steinke, S.G. Kandlikar, Single-phase heat transfer enhancement

techniques in microchannel and minichannel flows, in: ASME 2004 2nd
Int. Conf. Microchannels Minichannels, ASME, Rochester, NY, USA,
2004: pp. 141–148.
[10] T. Kishimoto, S. Sasaki, Cooling characteristics of diamond-shaped

interrupted cooling fin for high-power LSI devices, Electron. Lett. 23 (1987)
456–457.
[11] M. Asadi, G. Xie, B. Sunden, A review of heat transfer and pressure drop
82
LIST OF REFERENCE
characteristics of single and two-phase microchannels, Int. J. Heat Mass
Transf. 79 (2014) 34–53.
[12] B.X. Wang, X.F. Peng, Experimental investigation on liquid forced-
convection heat transfer through microchannels, Int. J. Heat Mass Transf.
37 (1994) 73–82.
[13] X.F. Peng, G.P. Peterson, B.X. Wang, Frictional Flow Characteristics of
water flowing through rectangular microchannels, Exp. Heat Transf. 7
(1994) 249–264.
[14] D. Yu, R.O. Warrington, R. Barron, T. Ameel, An experimental and
theoretical investigation of fluid flow and heat transfer in microtubes, in:
ASME/JSME Therm. Eng. Jt. Conf., ASME/JSME, Maui, 1995: pp. 523–
530.
[15] T.M. Harms, M.J. Kazmierczak, F.M. Gerner, Developing convective heat
transfer in deep rectangular microchannels, Int. J. Heat Fluid Flow. 20
(1999) 149–157.
[16] W. Qu, G.M. Mala, D. Li, Pressure-driven water flows in trapezoidal silicon
microchannels, Int. J. Heat Mass Transf. 43 (2000) 353–364.
[17] W. Qu, G.M. Mala, D. Li, Heat transfer for water flow in trapezoidal silicon
[18] B. Xu, K.T. Ooi, N.T. Wong, W.K. Choi, Experimental investigation of flow
friction for liquid flow in microchannels, Int. Commun. Heat Mass Transf.
27 (2000) 1165–1176.
[19] P.-X. Jiang, M.-H. Fan, G.-S. Si, Z.-P. Ren, Thermal–hydraulic
performance of small scale micro-channel and porous-media heat-
exchangers, Int. J. Heat Mass Transf. 44 (2001) 1039–1051.
[20] P. Gao, S. Le Person, M. Favre-Marinet, Scale effects on hydrodynamics
and heat transfer in two-dimensional mini and microchannels, Int. J.
Therm. Sci. 41 (2002) 1017–1027.
[21] W. Qu, I. Mudawar, Experimental and numerical study of pressure drop
and heat transfer in a single-phase micro-channel heat sink, Int. J. Heat
Mass Transf. 45 (2002) 2549–2565.
[22] J. Judy, D. Maynes, B.W. Webb, Characterization of frictional pressure
drop for liquid flows through microchannels, Int. J. Heat Mass Transf. 45
(2002) 3477–3489.
83
LIST OF REFERENCE
[23] A. Bucci, G.P. Celata, M. Cumo, E. Serra, G. Zummo, Water Single-Phase
Fluid Flow and Heat Transfer in Capillary Tubes, in: 1st Int. Conf.
Microchannels Minichannels, ASME, 2003: pp. 319–326.
[24] D. Liu, S. V. Garimella, Investigation of Liquid Flow in Microchannels, J.
Thermophys. Heat Transf. 18 (2004) 65–72.
[25] B.X. Wang, X.F. Peng, Experimental investigation on liquid forced-
convection heat transfer through microchannels, Int. J. Heat Mass Transf.
37 (1994) 73–82.
[26] X.F. Peng, G.P. Peterson, B.X. Wang, Heat Transfer Characteristics of
Water Flowing Through Microchannels, Exp. Heat Transf. 7 (1994) 265–
283.
[27] T.M. Adams, S.I. Abdel-Khalik, S.M. Jeter, Z.H. Qureshi, An experimental
investigation of single-phase forced convection in microchannels, Int. J.
Heat Mass Transf. 41 (1998) 851–857.
[28] N.T. Nguyen, D. Bochnia, R. Kiehnscherf, W. Dötzel, Investigation of
forced convection in microfluid systems, Sensors Actuators A Phys. 55
(1996) 49–55.
[29] G. Mohiuddin Mala, D. Li, Flow characteristics of water in microtubes, Int.
J. Heat Fluid Flow. 20 (1999) 142–148.
[30] K. V. Sharp, R.J. Adrian, Transition from laminar to turbulent flow in liquid
filled microtubes, Exp. Fluids. 36 (2004) 741–747.
[31] F.J. Gui, R.P. Scaringe, Enhanced heat transfer in the entrance region of
microchannels, in: 30th Intersoc. Energy Convers. Eng. Conf., 1995: pp.
289–294.
[32] L.S. Ding, H. Sun, X.L. Sheng, B.D. Lee, Measurement of Friction Factors
for R134A and R12 Through Microchannels, in: Proc. Symp. Energy Eng.
21st Century, 2000: pp. 650–657.
[33] D. Pfund, D. Rector, A. Shekarriz, A. Popescu, J. Welty, Pressure drop
measurements in a microchannel, AIChE J. 46 (2000) 1496–1507.
[34] F. Debray, J.P. Franc, T. Maı̂;tre, S. Reynaud, Mesure des
coefficients de transfert thermique par convection forcée en mini-canaux,
Mécanique Ind. 2 (2001) 443–454.
[35] C.Y. Yang, J.C. Wu, H.T. Chien, S.R. Lu, Friction characteristics of water,
84
LIST OF REFERENCE
R-134a and air in small tubes, Microscale Thermophys. Eng. 7 (2003)
335–348.
[36] H. Schlichting, Boundary Layer Theory, McGraw-Hill, New York, 1979.
[37] G.P. Celata, Single-Phase Heat Transfer and Fluid Flow in Micropipes,
Heat Transf. Eng. 25 (2004) 13–22.
[38] R.. Webb, E.R.. Eckert, Application of rough surfaces to heat exchanger
design, Int. J. Heat Mass Transf. 15 (1972) 1647–1658.
[39] P.M. Ligrani, M.M. Oliveira, T. Blaskovich, Comparison of Heat Transfer
Augmentation Techniques, AIAA J. 41 (2003) 337–362.
[40] S.W. Chang, T.-M. Liou, M.H. Lu, Heat transfer of rectangular narrow
channel with two opposite scale-roughened walls, Int. J. Heat Mass
Transf. 48 (2005) 3921–3931.
[41] R.. Webb, E.R.. Eckert, Application of rough surfaces to heat exchanger
design, Int. J. Heat Mass Transf. 15 (1972) 1647–1658.
[42] P.S. Kathait, A.K. Patil, Thermo-hydraulic performance of a heat

exchanger tube with discrete corrugations, Appl. Therm. Eng. 66 (2014)
162–170.
[43] J. Marschewski, R. Brechbühler, S. Jung, P. Ruch, B. Michel, D.
Poulikakos, Significant heat transfer enhancement in microchannels with
herringbone-inspired microstructures, Int. J. Heat Mass Transf. 95 (2016)
755–764.
[44] S.G. Kandlikar, History, Advances, and Challenges in Liquid Flow and
Flow Boiling Heat Transfer in Microchannels: A Critical Review, ASME J.
Heat Transf. 134 (2012) 034001–1.
[45] R.K. Shah, A.L. London, Laminar flow forced convection in ducts -
Supplement 1 to Advances in Heat Transfer, Academic Press, New York,
1978.
[46] K.S. Kong, K.T. Ooi, A numerical and experimental investigation on
microscale heat transfer effect in the combined entry region in macro
geometries, Int. J. Therm. Sci. 68 (2013) 8–19.
[47] C.Y. Yang, J.C. Wu, H.T. Chien, S.R. Lu, Friction characteristics of water,
R-134a and air in small tubes, Microscale Thermophys. Eng. 7 (2003)
335–348.
85
LIST OF REFERENCE
[48] D. Yu, R.O. Warrington, R. Barron, T. Ameel, An experimental and
theoretical investigation of fluid flow and heat transfer in microtubes, in:
ASME/JSME Therm. Eng. Jt. Conf., ASME/JSME, Maui, 1995: pp. 523–
530.
[49] J.F. Douglas, J.M. Gasiorek, J.A. Swaffield, Fluid Mechanics, 4th ed.,
Prentice Hall, 2001.
[50] W. George, Is there a universal log law for turbulent wall-bounded flows?,
Trans. R. Soc. 365 (2007) 789–806.
[51] W.M. Kays, M.E. Crawford, B. Weigand, Convective heat and mass
transfer, 4th ed., McGraw-Hill Higher Education, Boston, 2005.
[52] R.K. Shah, A.L. London, Laminar flow forced convection in ducts -
Supplement 1 to Advances in Heat Transfer, Academic Press, New York,
1978.
[53] A. Faghri, Y. Zhang, J. Howell, Advanced Heat and Mass Transfer, Global
Digital Press, 2010.
[54] R.D. Blevins, Applied Fluid Dynamics Handbook, Van Nostrand Reinhold
Company Inc, New York, N.Y., 1984.
[55] Y.A. Çengel, Heat and Mass Transfer: A Practical Approach, 3rd ed.,
Mcgraw-Hill Education, 2006.
[56] R.J. Phillips, Forced-Convection, Liquid-Cooled, Microchannel Heat Sinks,

M.S. Thesis, Massachusetts Institute of Technology, 1987.
[57] Y. Taitel, A.E. Dukler, A model for predicting flow regime transition in
horizontal and near horizontal gas-liquid flow, AIChE J. 22 (1976) 47–55.
[58] B.S. Petukhov, Heat Transfer and Friction in Turbulent Pipe Flow with
Variable Physical Properties, Adv. Heat Transf. 6 (1970) 503–564.
[59] A. Bejan, Heat Transfer, Wiley, New York, 1993.
[60] W.M. Kays, M.E. Crawford, Convective Heat and Mass Transfer, McGraw-
Hill, New York, 1993.
[61] D.K. Edwards, V.E. Denny, A.F. Mills, Transfer Processes, 2nd ed.,
Hemisphere, Washington, DC, 1979.
[62] H. Hausen, Neue Gleichungen fur dieWarmeubertragung bei freier oder
86
LIST OF REFERENCE
erzwungener Stromung, Allg. Waermetechnik. 9 (1959) 75–79.
[63] E.N. Sieder, G.E. Tate, Heat Transfer and Pressure Drop of Liquids in
Tubes, Ind. Eng. Chem. 28 (1936) 1429–1435.
[64] W.H. McAdams, Heat Transmission, 2nd ed., McGraw-Hill, New York,
1942.
[65] G. Filonenko, Hydraulic Resistance in Pipes, Teploenergetika. 1 (1954)

40–44.
[66] V. Gnielinski, New Equations for Heat and Mass Transfer in Turbulent
Pipe and Channel Flow, Int. Chem. Eng. 16 (1976) 359–368.
[67] A.E. Bergles, M.K. Jensen, B. Shome, The Literature on Enhancement of
Convective Heat and Mass Transfer, J. Enhanc. Heat Transf. 4 (1996) 1–
6.
[68] Z.S. Kareem, M.N. Mohd Jaafar, T.M. Lazim, S. Abdullah, A.F.
Abdulwahid, Passive heat transfer enhancement review in corrugation,
Exp. Therm. Fluid Sci. 68 (2015) 22–38.
[69] W.Q. Tao, Y.L. He, Q.W. Wang, Z.G. Qu, F.Q. Song, A unified analysis on
enhancing single phase convective heat transfer with field synergy
principle, Int. J. Heat Mass Transf. 45 (2002) 4871–4879.
[70] J. Xu, Y. Gan, D. Zhang, X. Li, Microscale heat transfer enhancement
using thermal boundary layer redeveloping concept, Int. J. Heat Mass
Transf. 48 (2005) 1662–1674.
[71] D. Deng, W. Wan, Y. Tang, H. Shao, Y. Huang, Experimental and
numerical study of thermal enhancement in reentrant copper
[72] Y. Sui, C.J. Teo, P.S. Lee, Y.T. Chew, C. Shu, Fluid flow and heat transfer
in wavy microchannels, Int. J. Heat Mass Transf. 53 (2010) 2760–2772.
[73] H.E. Hegab, A. Bari, T. Ameel, Friction and Convection Studies of R-134a
in Microchannels within the Transition and Turbulent Flow Regimes, Exp.
Heat Transf. 15 (2002) 245–259.
[74] G.P. Celata, M. Cumo, M. Guglielmi, G. Zummo, Experimental
Investigation of Hydraulic and Single Phase Heat Transfer in 0.130μm
CapillaryTube, Microscale Thermophys. Eng. 6 (2002) 85–97.
87

Ntu LCC Fyp (Submission)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ntu LCC Fyp (Submission)

Uploaded by

Copyright:

Available Formats

This

Downloaded on 09 Sep 2021 16:45:41 SGT

LIST OF TABLES

List of Figures ......................................................................................................... i

LIST OF TABLES ....................................................................................................... i

Chapter 1: Introduction ......................................................................................... 1

Chapter 2 Literature Review .................................................................................. 8

Chapter 3: Heat Transfer ..................................................................................... 34

Chapter 4: Optimization Method ......................................................................... 57

Chapter 5: Results and Discussion ....................................................................... 63

6. Conclusion and Future Works .......................................................................... 78

List of References ................................................................................................ 80

Chapter 1: Introduction

1.1. Motivation for research

Figure 1-­1: Thermal Cooling Unit (TCU) part [3]

Figure 1-­2: Thermal Cooling Unit (TCU) wires insertion [3]

1.3. Scope of work

1. To revisit the classical theory for forced convective internal flow in a

3. Understand the proposed correlations proposed by Dr Goh.

1.4. Outline of report

Chapter 2 Literature Review

2.1. Overview of microscale heat transfer

Wang and Peng 𝑁𝑢 = 0.00805𝑅𝑒 @/A 𝑃𝑟 4/D

Adams et al. [27] 𝑁𝑢 = 𝑁𝑢HI 1 + 𝐹

2.1.1. Laminar-­to-­turbulent Flow Transition

2.2. Microchannel Profiles

Figure 2-­5: Nature-­inspired insert profiles

There were 22 microchannel profiles tested by Goh which consisted of 7

2.2.1 Performance indicator

2.2.2 Flow loop

2.2.4 Different Profile Results

2.3.1 Proposed Correlation on Nusselt number

Table 2-­3: Nusselt number correlation for laminar flow regime

0.3 10 -­ 350≲Re≲1,850 350≲Re≲1,500

0.5 10 -­ 350≲Re≲1,300 350≲Re≲1,000

0.7 10 -­ 350≲Re≲700 350≲Re≲750

0.5 20 -­ 350≲Re≲1,600 350≲Re≲1,800

0.5 15 -­ 350≲Re≲1,600 350≲Re≲1,500

0.5 5 -­ 350≲Re≲800 350≲Re≲800

F.9 7 5 𝑒/𝐻 4.@

2.3.2 Proposed Correlation for Average Friction Factor

Scale Friction factor

2.4 Research gap

Chapter 3: Heat Transfer

3.1.1 Internal flow

where h is the average convective heat transfer coefficient, A is surface area

3.2. Hydrodynamic Consideration

3.2.1. Velocity Boundary Layer

Figure 3-­1: Laminar, hydrodynamic boundary layer development in a circular

3.2.2. Friction Factor

where μ is dynamic viscosity of fluid.

3.2.3 Thermal Consideration

3.2.4. Thermal Boundary Layer

3.2.5 Convective Heat Transfer Coefficient

Reynolds number and is a constant. Table 3-­5 is a summary of Nusselt

Fully developed, 𝑁𝑢 = 3.66

Fluid properties computed

@/A 0.60 ≤ 𝑃𝑟 ≤ 160

Smooth pipe Properties are evaluated at Tb

except μs, which is evaluated

Fluid properties are evaluated

3.3. Classical Theory for Concentric Annular Channel

3.3.1 Hydrodynamic Consideration

Figure 1-1: Thermal Cooling Unit (TCU) part [3]

Figure 1-2: Thermal Cooling Unit (TCU) wires insertion [3]

2.1.1. Laminar-to-turbulent Flow Transition

Figure 2-5: Nature-inspired insert profiles

Table 2-3: Nusselt number correlation for laminar flow regime

0.3 10 - 350≲Re≲1,850 350≲Re≲1,500

0.5 10 - 350≲Re≲1,300 350≲Re≲1,000

0.7 10 - 350≲Re≲700 350≲Re≲750

0.5 20 - 350≲Re≲1,600 350≲Re≲1,800

0.5 15 - 350≲Re≲1,600 350≲Re≲1,500

0.5 5 - 350≲Re≲800 350≲Re≲800

Figure 3-1: Laminar, hydrodynamic boundary layer development in a circular

Reynolds number and is a constant. Table 3-5 is a summary of Nusselt

3.5 Thermo-hydraulic Performance

𝑥_,¢ 𝑛𝑒𝑤 = 𝐻 𝑥 − 𝛽 𝑖𝑓 ℎ𝑖𝑔ℎ𝑒𝑟 𝑏𝑜𝑢𝑛𝑑𝑎𝑟𝑦 𝑖𝑠 𝑣𝑖𝑜𝑙𝑎𝑡𝑒𝑑 (4-6)

Figure 4-1: Flow Chart for Optimisation Process

Table 4-1: Constraints of independent variables

Figure 5-2: Inverted Fish Scale profile’s e/H iteration results

Figure 5-3: Inverted Fish Scale Profile’s P/e iteration results

Figure 5-4: Inverted Fish Scale Profile’s Nu/NuPlain iteration results

Figure 5-7: Fish Scale profile’s Reynolds number iteration results

Figure 5-8: Fish Scale profile’s e/H iteration results

Figure 5-9: Fish Scale Profile’s P/e iteration results

Figure 5-10: Fish Scale Profile’s Nu/NuPlain iteration results

Figure 5-12: Durian Profile’s Reynolds number iteration results

Figure 5-13: Durian Profile’s e/H iteration results

Figure 5-14: Durian Profile’s P/e iteration results

Figure 5-15: Durian Profile’s Nu/NuPlain iteration results

Figure 5-16: Durian Profile with H=0.3mm

Figure 5-17: Durian Profile with H=0.45mm

[9] M.E. Steinke, S.G. Kandlikar, Single-phase heat transfer enhancement

[10] T. Kishimoto, S. Sasaki, Cooling characteristics of diamond-shaped

[36] H. Schlichting, Boundary Layer Theory, McGraw-Hill, New York, 1979.

[42] P.S. Kathait, A.K. Patil, Thermo-hydraulic performance of a heat

[56] R.J. Phillips, Forced-Convection, Liquid-Cooled, Microchannel Heat Sinks,