ABSTRACT

Title of Thesis: ADVANCES TO A COMPUTER MODEL USED IN THE

SIMULATION AND OPTIMIZATION OF HEAT
EXCHANGERS

Robert Andrew Schwentker, Master of Science, 2005

Thesis Directed By: Professor Reinhard Radermacher

Department of Mechanical Engineering

Heat exchangers play an important role in a variety of energy conversion

applications. They have a significant impact on the energy efficiency, cost, size, and

weight of energy conversion systems. CoilDesigner is a software program introduced

by Jiang (2003) for simulating and optimizing heat exchangers. This thesis details

advances that have been made to CoilDesigner to increase its accuracy, flexibility,

and usability.

CoilDesigner now has the capability of modeling wire-and-tube condensers

under both natural and forced convection conditions on the air side. A model for flat

tube heat exchangers of the type used in automotive applications has also been

developed. Void fraction models have been included to aid in the calculation of

charge. In addition, the ability to model oil retention and oil’s effects on fluid flow

and heat transfer has been included. CoilDesigner predictions have been validated

with experimental data and heat exchanger optimization studies have been performed.
 ADVANCES TO A COMPUTER MODEL USED IN THE SIMULATION AND
OPTIMIZATION OF HEAT EXCHANGERS

by

Robert Andrew Schwentker

Thesis submitted to the Faculty of the Graduate School of the

University of Maryland, College Park in partial fulfillment
of the requirements for the degree of
Master of Science
2005

Advisory Committee:

Professor Reinhard Radermacher, Chair

Associate Professor Linda Schmidt
Assistant Professor Bao Yang
 © Copyright by
Robert Andrew Schwentker
2005
Dedication

Dedicated
to
my wife

ii
 Acknowledgements

I would first like to express my deep gratitude to my advisor, Dr. Reinhard

Radermacher, for enabling me to study and conduct research at the University of

Maryland, College Park. His support and faith in my abilities over the past couple of

years are greatly appreciated. I would also like to thank Dr. Linda Schmidt and Dr.

Bao Yang for serving on my thesis committee and for providing valuable comments

regarding my thesis.

I am very grateful to my colleagues I have worked closely with, including

Vikrant Aute, Lorenzo Cremaschi, John Fogle, Amr Gado, Ersin Gerçek, Kai Hübner,

Ahmet Ors, Jon Winkler, and Eric Xuan. They have all been very helpful and have

made the time I spent at the University of Maryland much more enjoyable.

I would also like to thank the companies that support the Center for

Environmental Energy Engineering at the University of Maryland for making the

research presented in this thesis possible.

Finally, I would like to thank my parents, my brother, and my wife. Without

their love and support, this work would not have been possible. I am deeply grateful

to all of them.

iii
 Table of Contents

List of Tables ............................................................................................................... vi

List of Figures ............................................................................................................. vii
Nomenclature.................................................................................................................x
Chapter 1 Introduction................................................................................................1
1.1 Overview........................................................................................................1
1.2 Objectives of Research ..................................................................................3
Chapter 2 Heat Exchanger Modeling .........................................................................5
2.1 Modeling of the Refrigerant Side ..................................................................6
2.2 Modeling of Heat Transfer Between Refrigerant and Air .............................7
2.3 Subdivided Segment Model.........................................................................10
Chapter 3 Wire-and-Tube Condenser Model ...........................................................13
3.1 Fin Efficiency of Wire-and-Tube Condensers.............................................16
3.2 Natural Convection Heat Transfer Model....................................................17
3.3 Forced Convection Heat Transfer Model ....................................................22
Chapter 4 Flat Tube Heat Exchanger Model ............................................................26
4.1 Fluid-Side Modeling ....................................................................................28
4.2 Heat Transfer Between Refrigerant and Air ................................................30
4.3 Air-Side Modeling .......................................................................................31
4.3.1 Fin Types for Flat Tube Heat Exchangers ...............................................31
4.3.2 Tube Configurations for Flat Tube Heat Exchangers ..............................38
Chapter 5 Void Fraction Models and Charge Calculation .......................................40
5.1 Types of Void Fraction Model.....................................................................42
5.1.1 Homogeneous Void Fraction Model........................................................42
5.1.2 Slip-Ratio-Correlated Void Fraction Models...........................................42
5.1.3 Void Fraction Models Correlated With Lockhart-Martinelli Parameter .43
5.1.4 Mass-Flux-Dependent Void Fraction Models .........................................44
5.2 Comparison of Void Fraction Models .........................................................44
Chapter 6 Modeling of Effects of Oil in Heat Exchangers.......................................46
6.1 Oil Mass Fraction and Two-Phase Refrigerant-Oil Mixture Quality ..........47
6.2 Bubble Point Temperature Calculation........................................................49
6.3 Heat Load Calculation and the Heat Release Enthalpy Curve ....................52
6.4 Calculation of Refrigerant-Oil Mixture Properties ......................................54
6.5 Heat Transfer Coefficient Correlations for Refrigerant-Oil Mixture ..........56
6.5.1 Heat Transfer Coefficient for Evaporation ..............................................56
6.5.2 Heat Transfer Coefficients for Condensation ..........................................58
6.6 Pressure Drop Correlation for Two-Phase Refrigerant-Oil Mixture ...........59
6.7 Oil Retention and Void Fraction Models.....................................................61
Chapter 7 Validation and Optimization Studies .......................................................63
7.1 Validation of Microchannel Heat Exchanger Model ...................................63
7.2 Validation of Wire-and-Tube Condenser Model .........................................66
7.3 Validation of Refrigerant-Oil Mixture Model .............................................68
7.4 Optimization Study of Wire-and-Tube Condenser ......................................72

iv
 7.4.1 Optimization of Original Condenser........................................................73
7.4.2 Optimization of Condenser with Larger Face Area.................................75
Chapter 8 Conclusions..............................................................................................79
8.1 New Heat Exchanger Models ......................................................................80
8.1.1 Wire-and-Tube Condenser Model ...........................................................80
8.1.2 Flat Tube Heat Exchanger Model ............................................................81
8.2 Additional Fluid Modeling Capabilities ......................................................82
8.2.1 Void Fraction Models and Charge Calculation .......................................82
8.2.2 Modeling of Oil Effects and Oil Retention..............................................82
8.3 Validation and Optimization Studies ...........................................................83
8.3.1 Validation of Microchannel Heat Exchanger Model ...............................83
8.3.2 Validation of Wire-and-Tube Condenser Model .....................................84
8.3.3 Validation of Oil Retention Model ..........................................................84
8.3.4 Optimization of Wire-and-Tube Condenser ............................................85
Chapter 9 Future Work .............................................................................................86
Appendix......................................................................................................................87
A.1 Air-Side Heat Transfer Coefficient Correlations for Flat Tubes .................87
A.2 Air-Side Pressure Drop Correlations for Flat Tubes....................................88
A.3 Refrigerant-Side Heat Transfer Coefficient Correlations ............................89
A.4 Void Fraction Models ..................................................................................90
A.4.1 Void Fraction Models for Round Tubes ..................................................90
A.4.2 Void Fraction Models for Microchannel Tubes.......................................99
References..................................................................................................................101

v
 List of Tables

Table 3-1. Constants C and m used to calculate the Žukauskas heat transfer

coefficient ................................................................................................. 24

Table 6-1. Empirical constants used in Eqs. 6.6 and 6.7 to calculate bubble point

temperature of refrigerant-oil mixtures..................................................... 51

Table 6-2. Coefficients c and n as a function of the oil mass fraction in the

correlation developed by Chaddock and Mathur (1980) for the heat

transfer coefficient of refrigerant-oil mixtures ......................................... 58

Table 6-3. Constant C used to calculate the two-phase multipliers used in the

Lockhart-Martinelli correlation ................................................................ 61

Table 7-1. Geometric parameters of the microchannel heat exchangers used for

validation................................................................................................... 63

Table 8-1. Summary of modeling capabilities added to CoilDesigner and work

performed in relation to this thesis............................................................ 79

Table A-1. Slip ratios S based on property index P.I. generalized from Thom’s

steam-water data (1964) by Ahrens (1983) .............................................. 91

Table A-2. Coefficients for use with Eq. A.22, the curve-fit equation developed to

calculate the slip ratio for the Thom void fraction model......................... 92

Table A-3. Liquid void fraction (1-α) data presented by Baroczy (1966)................. 94

Table A-4. Hughmark flow parameter KH as a function of Z (1962)......................... 95

Table A-5. Coefficients for use with Eq. A.35, the curve-fit equation developed to

calculate the Hughmark flow parameter KH as a function of Z ................ 96

vi
 List of Figures

Figure 2-1. Drawing of refrigerant undergoing phase changes within segments ...... 11

Figure 3-1. Geometric parameters of wire-and-tube condensers............................... 19

Figure 3-2. Flow chart for iterative scheme to calculate air-side heat transfer

coefficient and heat load for natural convection wire-and-tube condensers

(adapted from Bansal and Chin, 2003) ..................................................... 21

Figure 4-1. Flat tube heat exchanger with plate fins.................................................. 26

Figure 4-2. Flat tube heat exchanger with corrugated fins ........................................ 27

Figure 4-3. Geometric parameters of flat tubes ......................................................... 28

Figure 4-4. Flat tube heat exchanger with serpentine refrigerant flow (airflow into the

page).......................................................................................................... 29

Figure 4-5. Flat tube heat exchanger with parallel refrigerant flow (airflow into the

page).......................................................................................................... 29

Figure 4-6. Flat tube heat exchanger with plate fins (airflow into the page)............. 32

Figure 4-7. Diagram showing the definition of louver pitch ..................................... 33

Figure 4-8. Diagram showing the definition of louver angle and louver height........ 34

Figure 4-9. Flat tube heat exchanger with corrugated fins (airflow into the page) ... 36

Figure 4-10. Flat tube heat exchanger with triangular corrugated fins (airflow into the

page).......................................................................................................... 36

Figure 4-11. Flat tube plate fin heat exchanger with staggered tube configuration .. 38

Figure 4-12. Air-side mass and energy flow from one column of tubes to the next . 39

vii
Figure 5-1. Comparison of charge predictions based on different void fraction

models ....................................................................................................... 45

Figure 6-1. Difference between refrigerant-oil mixture bubble point temperature and

refrigerant saturation temperature, as a function of quality (From Shen and

Groll, 2003, p. 6)....................................................................................... 50

Figure 7-1. Predicted heat load vs. experimentally measured heat load of

microchannel heat exchangers used for validation ................................... 65

Figure 7-2. Predicted refrigerant pressure drop vs. experimentally measured pressure

drop of microchannel heat exchangers used for validation ...................... 65

Figure 7-3. Predicted heat load vs. experimentally measured heat load of wire-and-

tube condensers used for validation.......................................................... 67

Figure 7-4. Predicted pressure drop vs. experimentally measured pressure drop of

wire-and-tube condensers used for validation .......................................... 68

Figure 7-5. Experimentally measured oil retention vs. predicted oil retention in the

evaporator (from Cremaschi, 2004).......................................................... 69

Figure 7-6. Calculated oil retention, mixture quality, and local oil mass fraction in an

evaporator with R-134a/PAG at OMF=2.4% (from Cremaschi, 2004).... 71

Figure 7-7. Experimentally measured oil retention vs. predicted oil retention for the

condenser (from Cremaschi, 2004)........................................................... 72

Figure 7-8. Heat load vs. cost of all test condensers in optimization of baseline

condenser .................................................................................................. 74

Figure 7-9. Heat load vs. cost for all better condensers in optimization of baseline

condenser .................................................................................................. 75

viii
Figure 7-10. Heat load vs. cost of all test condensers in optimization of condensers

with larger face area.................................................................................. 77

Figure 7-11. Heat load vs. cost for all better condensers in optimization of

condensers with larger face area ............................................................... 78

ix
 Nomenclature

A Area (m2)
Afrontal Frontal face area of heat exchanger (m2)
Amin Minimum free flow area (m2)
C Constant
cp Specific heat (J kg-1 K-1)
D Diameter (m)
f Friction factor
Fp Fin pitch (m)
H Heat exchanger height (m)
G Mass flux (kg m-2 s-1)
g Acceleration due to gravity, 9.81 (m s-2)
h Heat transfer coefficient (W m-2 K-1)
j Colburn factor
k Thermal conductivity (W m-1 K-1)
Lθ Louver angle (degrees)
Lh Louver height (m)
Ll Louver length (m)
Lp Louver pitch (m)
m& Mass flow rate (kg s-1)
N Fan rotational speed (rev min-1)
NTU Number of transfer units
Nu Nusselt number
P Pressure (Pa)
p Perimeter (m)
p Pitch (m)
Pr Prandtl number, µ ⋅ c p / k
Q Heat duty (W), Volumetric air flow rate (m3 s-1)
R Heat transfer resistance
Ra Rayleigh number
Re Reynolds number
S Slip ratio
Sl Tube horizontal spacing (m)
St Tube vertical spacing (m)
Sw Wire spacing (m)
St Stanton number
T Temperature (K)
Th Tube height (m)
Tw Tube width (m)
UA Overall heat transfer conductance
v Velocity (m s-1)
W Molecular mass, Fan power consumption (W)
We Weber number

x
Greek
β Thermal expansion coefficient (K-1)
δ Liquid film thickness (m)
ε Heat exchange effectiveness
η Fin efficiency
ηs Surface effectiveness
µ Viscosity (kg m-1 s-1)
ξ Yokozeki factor
ρ Density (kg m-3)
σ Surface tension (N m-1)
σ Stefan-Boltzmann constant, 5.67x10-8 (W K-4 m-2)
Ψ mole fraction

Subscript
a air
c Convective
in Inlet, Inner
liq Liquid
out Outlet, Outer
r Radiative
ref Refrigerant
t Tube
vap Vapor, Gas
w Wire

xi
Chapter 1 Introduction

1.1 Overview

Over the past several years, with energy costs rising and awareness of the

environmental impact of the use of fossil fuels increasing, there has been a greater

focus around the world on energy usage and consumption. Increasing the efficiency

of energy-intensive products and processes is one of the most important methods

available for confronting and reducing the problems associated with energy

consumption. By increasing energy efficiency, traditional energy supplies will last

longer and the harmful effects related to energy consumption, such as global

warming, can be reduced.

Vapor compression systems used in heating, ventilating, air-conditioning, and

refrigerating (HVAC&R) applications are energy-intensive and represent a significant

portion of the total energy consumption of buildings and automobiles. Much progress

has been made over the past couple of decades to improve the energy efficiency of

such systems. However, research continues and more progress can be achieved.

As computer processing power has increased, the ability to use simulation

software for engineering purposes has increased dramatically. This is true of vapor

compression systems, as well. The use of software simulation tools is an increasingly

popular method for improving the efficiency of vapor compression systems. This can

be performed through the use of system-level simulation tools as well as component-

level simulation tools. Heat exchanger simulation is particularly important because

heat exchangers comprise two of the four major components of vapor compression

1
systems. Therefore, it is important to develop a simulations tool that can be used to

design, simulate, and optimize heat exchangers.

The cost associated with designing and manufacturing heat exchangers is also

of major concern. The price of raw materials used in heat exchangers, such as

aluminum and copper, have been rising over the past few years as demand has

increased in countries such as China and India. Moreover, manufacturers now have

to compete with companies from around the globe, making costs a more important

factor than ever. The use of simulation software can reduce the cost and time

required to design heat exchangers for new systems. Instead of building multiple

prototype heat exchangers and testing each one, multiple heat exchanger designs can

be modeled and then a couple of the best designs can be built and tested. This aids in

the design of heat exchangers that will perform as needed on the first try. Heat

exchanger simulation tools can also be used to perform optimization studies in order

to decrease the material and cost necessary to manufacture heat exchangers.

Heat exchangers are also used in a variety of applications beyond HVAC&R

systems. They are also used for thermal management in automobiles as well as in

applications in food processing, petrochemical, textile, and other process industries.

Therefore, a heat exchanger simulation tool can have a variety of applications.

CoilDesigner is a software simulation tool used for the simulation and

optimization of heat exchangers that was first introduced by Jiang (2003). Its most

distinguishing features include its generality, the level of detail, and its user-friendly

graphic interface. At that time, CoilDesigner could be used to model two types of

heat exchanger often used in HVAC&R systems—round tube plate fin (RTPF) heat

2
exchangers and microchannel heat exchangers. Jiang also validated the RTPF model

with experimentally measured data. Over the past couple of years, further research

has been performed to increase the number of applications and the accuracy of

CoilDesigner. In this thesis, advances that have been made to CoilDesigner are

detailed.

1.2 Objectives of Research

The primary objective of this thesis is to detail advances that have been made

to CoilDesigner. The specific objectives of this research include the following:

• Develop two new heat exchanger models in addition to the two pre-existing

heat exchanger models:

o Develop a model that can simulate wire-and-tube condensers of the type

used in refrigerators. Include the ability to model natural convection heat

transfer as well as forced convection heat transfer on the air side.

o Develop a model that can simulate flat tube heat exchangers of the type

used in automotive applications for radiators and charge air coolers.

• Implement new fluid modeling capabilities:

o Research various void fraction models and implement them in

CoilDesigner for the accurate calculation of refrigerant charge.

o Implement the ability to model refrigerant-oil mixtures in heat exchangers

by accomplishing the following:

3
 − Include correlations and procedures to calculate the heat transfer as

well as the thermodynamic and physical properties of refrigerant-oil

mixtures.

− Include correlations developed specifically for refrigerant-oil mixtures

to calculate heat transfer coefficients and pressure drop.

− Calculate oil retention in heat exchangers with the use of void fraction

models.

• Perform validation and optimization studies with CoilDesigner:

o Validate the microchannel heat exchanger model with experimental heat

exchanger performance data.

o Validate the wire-and-tube condenser model with experimental heat

exchanger performance data.

o Use experimental data to validate the refrigerant-oil model and its ability

to calculate oil retention.

o Perform optimization studies of a wire-and-tube condenser to increase

performance and reduce cost.

4
Chapter 2 Heat Exchanger Modeling

The general modeling techniques employed in CoilDesigner have been

detailed by Jiang (2003). This chapter discusses the main concepts and equations in

order to provide the reader with enough background knowledge to be able to

contextualize the advances made to CoilDesigner.

To model a heat exchanger, CoilDesigner uses a segment-by-segment

approach in which each tube is divided into multiple segments and the hydraulic and

heat transfer/energy equations are solved for each segment individually. Dividing

each tube into multiple segments allows two-dimensional non-uniformity of air

distribution to be modeled because different values for air velocity and temperature

can be input for each segment. Dividing tubes into segments also allows

heterogeneous refrigerant flow through a tube to be modeled, increasing the accuracy

of the heat exchanger model.

Each tube is divided into multiple segments, and then each segment is treated

like a small cross-flow heat exchanger. The air-to-refrigerant heat transfer and the

refrigerant pressure drop are calculated for each individual segment. On the

refrigerant side, each segment is provided with an inlet enthalpy, an inlet pressure,

and a mass flow rate. On the air side, the inlet air temperature is provided for each

segment. The inlet air flow rate is also provided for each segment when modeling

heat transfer of heat exchangers undergoing forced convection on the air side.

5
2.1 Modeling of the Refrigerant Side

The heat transfer between the refrigerant and the walls of the tubes through

which the refrigerant flows is governed by the temperature gradient and a heat

transfer coefficient:

q = h ⋅ A ⋅ (Tref − Twall ) (2.1)

In steady state, the heat transfer from the refrigerant to the walls must equal the

overall heat transfer from the refrigerant to the air. Therefore, in order to model the

heat transfer between the refrigerant and the air in heat exchangers, a heat transfer

coefficient must be calculated for the refrigerant as it flows through the tubes.

Multiple theoretical and empirical correlations have been developed over the past

several decades to model heat transfer coefficients for different geometric parameters,

flow regimes, refrigerants, operating conditions, and processes (i.e. if the refrigerant

is undergoing evaporation or condensation). These correlations are typically of the

form

h = ρvc p St (2.2)

in which St is the Stanton Number

j
St = (2.3)
Pr 2 / 3

Correlations included in CoilDesigner are discussed by Jiang (2003). Some

additional correlations that have been included in CoilDesigner over the last couple of

years are discussed in the Appendix.

The pressure drop of the refrigerant as it flows through the heat exchanger

must also be modeled because of the significant impact both on the power

6
consumption of HVAC&R systems and on the thermodynamic and transport

properties of the refrigerant. As discussed by Jiang (2003), the accelerational and

gravitational components of the pressure drop are dominated by the frictional

pressure drop and can therefore be neglected. Thus, the pressure drop can be

expressed by the equation

2L
Pin − Pout = ∆P = f m& 2 (2.4)
πρD 3

where f is a friction factor. Just as for heat transfer coefficients, multiple correlations

have been developed to model the friction factor for different geometric parameters,

flow regimes, refrigerants, and operating conditions. Correlations included in

CoilDesigner are discussed by Jiang (2003) and correlations that have been added are

detailed in the Appendix.

2.2 Modeling of Heat Transfer Between Refrigerant and Air

For each segment of a tube, the energy balance equation for the refrigerant is

described by the following equation:

q = m& ref (href ,out − href ,in ) (2.5)

where href is the enthalpy of the refrigerant. In the case of single-phase refrigerant,

the equation can be approximated as

q = m& ref c p , ref (Tref ,out − Tref ,in ) (2.6)

The energy balance equation for the air side is described by the following equation:

q = m& air (hair ,in − hair ,out ) = m& air c p ,air (Tair ,in − Tair ,out ) (2.7)

7
 In order to use these energy balance equations to calculate the heat transfer

between the air and the refrigerant, a method is needed to calculate the outlet

temperatures. The ε-NTU method for cross-flow configuration with one fluid mixed

and the other fluid unmixed is used for this purpose (Kays and London, 1984). The

refrigerant is modeled as a mixed fluid and the air is modeled as an unmixed fluid.

The heat capacities of each fluid are

C mixed = m& ref c p ,ref (2.8)

C unmixed = m& air c p ,air (2.9)

and the number of heat transfer units, NTU, is defined as:

UA
NTU = (2.10)
C min

UA is an overall heat conductance which is calculated using the idea of a thermal

circuit (Myers, 1998):

1 1 R R R fouling 1
= + thermal + contact + + (2.11)
UA href ⋅ At ,in At ,out At ,out At ,out hair ⋅ η s ⋅ Atotal ,out

where Rthermal, Rcontact, and Rfouling are thermal, contact, and fouling resistances that can

be input by the user, and ηs is the surface effectiveness of the heat exchanger. This

surface effectiveness is a function of the fin efficiency, ηf, as well as of the outer

surface areas of the tubes and fins, and is calculated according to the following

equation:

At ,out + η f A f
ηs = (2.12)
Atotal

Correlations are available to predict refrigerant-side and air-side heat transfer

coefficients as well as fin efficiencies needed for Eq. 2.11. The correlations are

8
typically developed based on empirical measurements and are functions of

geometrical parameters and flow characteristics. Details of correlations that have

been added to CoilDesigner are provided in the Appendix.

The heat exchange effectiveness, ε, is the ratio of the change in temperature,

∆T, to the maximum possible temperature change, based on the inlet temperatures of

the two fluids. ε is calculated for each segment depending on the heat capacities, C,

for each fluid. For the case in which Cmax = Cunmixed (in other words, when the air has

the higher heat capacity),

⎧⎪ Cmax ⎡ ⎛ Cmin ⎞⎤ ⎫⎪
ε = 1 − exp⎨− ⎢1 − exp⎜⎜ − NTU ⎟⎥ ⎬ (2.13)
⎪⎩ Cmin ⎣ ⎝ Cmax ⎟⎠⎦ ⎪⎭

and

Tref ,in − Tref ,out

ε= (2.14)
Tref ,in − Tair ,in

On the other hand, for Cmax = Cmixed,

C max ⎧ ⎡ C min ⎤⎫
ε= ⎨1 − exp ⎢− (1 − exp(− NTU ))⎥ ⎬ (2.15)
C min ⎩ ⎣ C max ⎦⎭

and

Tair ,out − Tair ,in

ε= (2.16)
Tref ,in − Tair ,in

When the refrigerant in a segment is in the two-phase regime,

C min
=0 (2.17)
C max

ε = 1 − exp(− NTU ) (2.18)

9
 Once ε is calculated for the segment, the outlet temperature of either the air or

the refrigerant can be calculated by rearranging Eq. 2.14 or Eq. 2.16. In the case of

two-phase refrigerant flowing through the segment, Eq. 2.16 must be used to

calculate the outlet air temperature because the refrigerant temperature remains

constant during the evaporation or condensation process. After calculating the outlet

air temperature, the heat load of the segment can then be calculated using Eq. 2.7.

2.3 Subdivided Segment Model

As discussed before, in CoilDesigner, tubes are divided into multiple

segments for solving the energy balance and the hydraulic equations. Typically, it

can be assumed that the refrigerant flowing through a segment does not undergo a

change in flow regime. A segment is usually occupied entirely by superheated vapor,

two-phase fluid, or subcooled liquid. However, there may be segments in which the

refrigerant undergoes a flow regime change, as shown in Figure 2-1. In order to

account for the significant changes in the refrigerant properties and the heat transfer

coefficients, a subdivided segment model is included in CoilDesigner in which the

segment is divided further into sub-segments, each of which is occupied entirely by

either single phase or two-phase refrigerant.

10
 segment L1, Tair,out,1 L2, Tair,out,2
gas Pout, href,out
m& ref
Pin, href,in P1, href,1, Tref,1
L2, Tair,out,2 L1, Tair,out,1 two-phase

liquid

vair, Tair,in

Air Flow

Figure 2-1. Drawing of refrigerant undergoing phase changes within segments

The ε-NTU equations presented above are modified as follows. If x is the

fraction of the length of the segment at which a flow regime change takes place, then

the heat capacities and NTU for the sub-segment are calculated according to the

following equations:

C mixed = m& ref c p ,ref (2.19)

C unmixed = xm& air c p ,air (2.20)

UA
NTU = x (2.21)
C min

As before, the heat exchange effectiveness, ε, is calculated using Eq. 2.13,

2.15, or 2.18, depending on which is appropriate according to the relationship

between Cmixed and Cunmixed. The following equations can then be used to calculate

the outlet temperatures of the first sub-segment:

Tref ,in − Tref ,1

ε= (2.22)
Tref ,in − Tair ,in

11
 Tair ,out ,1 − Tair ,in
ε= (2.23)
Tref ,in − Tair ,in

The heat load of the first sub-segment is governed by the equation

q = xm& air c p ,air ⋅ (Tair ,out ,1 ( x ) − Tair ,in ) = m& ref ⋅ (href ,in − href ,1 ) (2.24)

where href,1, the enthalpy at the interface between sub-segments, is the saturated

enthalpy of the refrigerant given the pressure, P1, at the interface:

href ,1 ( x ) = hsat (Pref ,1 ( x )) (2.25)

Pref ,1 (x ) = Pref ,in − ∆P (x, Pref ,in , href ,in ) (2.26)

This set of equations, Eqs. 2.19 through 2.26, reduces to one equation with x

being the unknown. The implicit equation is solved using a numerical iteration

scheme. Once x is calculated, the refrigerant pressure and enthalpy and the air

temperature at the outlet of the sub-segment are known. The ε-NTU method is then

used to calculate the heat load of the next sub-segment. Then the outlet conditions of

the entire segment can be calculated.

12
Chapter 3 Wire-and-Tube Condenser Model

Wire-and-tube condensers have been used in domestic refrigerators for

decades. This type of condenser consists of a single steel tube bent into a serpentine

tube bundle with pairs of steel wires welded onto opposite sides of the tube to serve

as extended surfaces. Wire-and-tube condensers employ either natural convection or

forced convection heat transfer on the air side. Natural convection wire-and-tube

condensers typically consist of a single bank attached to the back of a refrigerator.

They are coated in black paint to increase the emissivity to increase the radiation heat

transfer. As the air around the tubes is heated, its density decreases and it begins to

rise, causing upward-moving turbulent air flow resulting in convective heat transfer.

Forced convection wire-and-tube condensers typically are in the form of a tube

bundle consisting of several rows and several banks, and they have a fan forcing air

circulation through the heat exchanger.

Despite the prevalence of wire-and-tube condensers and their central role in

the efficiency and cost of refrigerators, very little literature has been published about

modeling them. The refrigerant-side heat transfer coefficients and pressure drop can

be modeled using the same equations and correlations as for traditional round tube

plate fin heat exchangers. However, in order to be able to model this type of heat

exchanger and to be able to perform optimization studies, accurate models for heat

transfer to the air, and therefore the air-side heat transfer coefficient and the fin

efficiency are needed.

A few articles have been published in recent years on modeling natural

convection wire-and-tube condensers. Tanda and Tagliafico (1997) developed a

13
correlation to predict the natural convection heat transfer coefficient for wire-and-

tube condensers with vertical wires attached to a single column of tubes. Tanda and

Tagliafico obtained experimental data by measuring the heat transfer from water

flowing through wire-and-tube heat exchangers. In order to minimize the effects of

radiative heat transfer during their experiments, they coated the heat exchangers with

low-emissivity paint. They then calculated theoretically the contribution from

radiative heat transfer and subtracted it from the total heat transfer before calculating

the natural convection heat transfer coefficient. They developed a semi-empirical

heat transfer coefficient correlation as a function of geometric and operating

parameters.

Tagliafico and Tanda (1997) also presented a wire-and-tube condenser model

that accounted for both natural convection and radiation heat transfer. They used the

semi-empirical correlation presented in their previous paper (Tanda and Tagliafico,

1997) to model the natural convection heat transfer coefficient. For radiation heat

transfer, they developed a theoretical model to calculate the average radiation heat

transfer coefficient. The authors then validated their model with a second set of

experimental data from eight wire-and-tube condensers with widely varying

geometric parameters.

Quadir et al. (2002) also developed a wire-and-tube condenser model for

natural convection. They used a finite element method and modeled varying ambient

temperatures and refrigerant flow rates in order to examine the effects on heat

exchanger performance. The authors assumed a constant overall heat transfer

coefficient of 10 W/m2 K.

14
 Bansal and Chin (2003) developed a computer model for natural convection

using a finite element and variable conductance approach written in FORTRAN 90.

The authors used the natural convection heat transfer coefficient correlation

developed by Tagliafico and Tanda (1997). Bansal and Chin also used the same

theoretical approach presented by Tagliafico and Tanda to calculate the radiative heat

transfer coefficient. The authors used these air-side heat transfer coefficients as well

as the thermal conductivity of the tube and the refrigerant-side heat transfer

coefficient to calculate an overall heat transfer conductance value, UA. They used the

UA value in calculating the total heat transfer of each finite element of the heat

exchanger. The paper also contains an iterative computational scheme that the

authors used to calculate the total heat transfer, pressure drop, and outlet conditions of

wire-and-tube condensers. The authors validated their computer model using their

own experimental data. They then used their computer model to optimize a wire-and-

tube condenser.

To the current author’s knowledge, only a couple of articles have been

published in the open literature about modeling wire-and-tube condensers in the

forced convection regime on the air side. Hoke et al. (1997) performed experimental

investigations using single-layer wire-and-tube condensers to measure the air-side

heat transfer coefficients under forced-convection conditions. They measured the

heat transfer coefficient for different angles of attack for two different cases: airflow

perpendicular to the tubes and parallel to the wires as well as airflow parallel to the

tubes and perpendicular to the wires. Using their experimental results, they

developed a model to calculate the air-side heat transfer coefficient. The authors

15
found that correlations developed by Hilpert (1933) and Žukauskas (1972) for single

cylinders overestimated the air-side heat transfer coefficient.

Lee et al. (2001) also performed experiments using single-layer sample wire-

and-tube condensers to measure the air-side heat transfer coefficient under in the

forced convection regime. Based on their experimental results, the authors disagree

with the conclusion by Hoke et al. (1997) that the heat transfer coefficient correlation

developed by Žukauskas (1972) overpredicts the air-side heat transfer coefficient.

Using their experimental results, the authors developed correction factors to use with

the Žukauskas correlation. With the use of their correction factors, the heat transfer

coefficient correlation can be used to model three different airflow configurations—

airflow perpendicular to both the tubes and the wires, airflow perpendicular to the

tubes and parallel to the wires, and airflow parallel to the tubes and perpendicular to

the wires. They define these airflow configurations as all cross, tube cross, and wire

cross, respectively.

3.1 Fin Efficiency of Wire-and-Tube Condensers

The fin efficiency is one of the main parameters affecting heat transfer on the

air side, so an accurate model is needed to be able to calculate the heat transfer from

wire-and-tube condensers accurately. The fin efficiency of a wire can be calculated

using the fin efficiency of a thin rod, which is given by the following equation

(Myers, 1998):

tanh (mL ) hp w 4h
ηw = where m = = (3.1)
mL k w Aw k w Dw

16
where L is half the tube spacing in the direction of the wires, h is the heat transfer

coefficient, pw is the perimeter of the wire (πDw), kw is the thermal conductivity of the

material, and Aw is the cross-sectional area of the wire (πDw2/4). Once the wire

efficiency has been calculated, the total surface effectiveness can be calculated

according to the following equation:

Atube + η w Awire A
ηs = = 1 − (1 − η w ) w (3.2)
Atotal Atotal

3.2 Natural Convection Heat Transfer Model

The heat transfer from the refrigerant to the air can be calculated according to

Fourier’s law (Bansal and Chin, 2003):

q = UA ⋅ ∆T = UA ⋅ (Tref − Tair ) (3.3)

where UA is the overall heat transfer conductance and Tair is the ambient air

temperature. In the model presented in this thesis, UA is, once again, calculated

according to Eq. 2.11.

As can be seen in the equations for UA (Eq. 2.11) and the fin efficiency (Eq.

3.1), these two quantities are dependent on the air-side heat transfer coefficient.

Therefore, a correlation is needed to calculate the heat transfer coefficient of wire-

and-tube condensers undergoing natural convection heat transfer. In typical air-to-

refrigerant heat exchangers, radiative heat transfer is dominated by forced convection

heat transfer caused by air being blown through by a fan. The radiative component of

the total heat transfer can therefore be neglected for modeling purposes. However, in

the natural convection regime, the transfer of heat from condensers is due to both

17
natural convective heat transfer and radiative heat transfer. Therefore, in order to

model wire-and-tube condensers under natural convection conditions, both

convective heat transfer and radiative heat transfer must be accounted for in the air-

side heat transfer coefficient.

As suggested by Tagliafico and Tanda (1997), the air-side heat transfer

coefficient can be modeled as the sum of a natural convection heat transfer coefficient

and a radiative heat transfer coefficient:

hair = hc + hr (3.4)

Now correlations for each of the individual component heat transfer coefficients are

needed.

An empirical correlation can be used to model the natural convection heat

transfer coefficient. To the author’s knowledge, the correlation developed by Tanda

and Tagliafico (1997) is the only natural convection heat transfer coefficient

correlation available in the open literature, so it has been implemented in

CoilDesigner. Their correlation has the following form:

Nu ⋅ k air
hc = (3.5)
H

where

⎛ Ra ⋅ H ⎞
0.25
⎧⎪ ⎡ ⎛ Dt ⎞ ⎤
0.25
⎛ s w ⎞⎫⎪
Nu = 0.66⎜⎜ ⎟⎟ ⎨1 − ⎢1 − 0.45⎜ ⎟ ⎥ exp⎜⎜ − ⎟⎟⎬ (3.6)
⎝ Dt ⎠ ⎪⎩ ⎣⎢ ⎝ H ⎠ ⎦⎥ ⎝ ϕ ⎠⎪⎭

where H is the height of the condenser, the Rayleigh number, Ra, is

⎛ βρ 2 c p ⎞
Ra = ⎜ ⎟ g ⋅ (Tt − Tair ) ⋅ H 3 (3.7)
⎜ µk ⎟
⎝ ⎠ air

and
18
 0. 4 0 .8 0 .5
⎛C ⎞ ⎛C ⎞ ⎛ C2 ⎞ −1.5 −0.5
ϕ = ⎜ 1 ⎟ s w0.9 st−1.0 + ⎜ 1 ⎟ ⎜⎜ ⎟⎟ s w st (3.8)
⎝H⎠ ⎝H⎠ ⎝ Tt − Tair ⎠

where C1=28.2 m, C2=264 K, and st and sw are the following geometric parameters:

S t − Dt
st = (3.9)
Dt

S w − Dw
sw = (3.10)
Dw

Figure 3-1. Geometric parameters of wire-and-tube condensers

The quantity Tt, used in Eqs. 3.7 and 3.8 above and in Eq. 3.13 below is the

temperature of the outside of the tube. It is calculated according to the following

equation:

⎛ 1 R ⎞
Tt = Tref − q ⋅ ⎜ + thermal ⎟ (3.11)
⎜h A At ,out ⎟
⎝ ref t ,in ⎠

This equation is derived from the fact that, in steady state, the heat transfer from the

refrigerant to the air must be equal to the heat transfer from the refrigerant to the outer

19
surface of the tube. As can be seen in Eq. 3.11, Tt is a function of the heat load of a

segment, so it must be guessed initially. As will be explained below, an iterative

solution scheme is used to solve the equations in the natural convection model and to

calculate the temperature of the tube.

As noted before, Tagliafico and Tanda (1997) also developed a theoretical

model to calculate the radiative heat transfer coefficient:

hr = ε appσ
(T 4
ex − Tair4 ) (3.12)
(Tex − Tair )
where εapp is the apparent thermal emittance, which is a function of the thermal

emittance of the heat exchanger surface as well as of geometric parameters such as

the tube and wire diameters and the tube and wire pitches. Bansal and Chin (2003)

found good agreement with experimental results by setting εapp equal to 0.88. σ is the

Stefan-Boltzmann constant, 5.67x10-8 W/(K4 m2). Tex is the mean surface

temperature of the heat exchanger, which can be calculated according to the

following equation:

Tt + GP ⋅ η w ⋅ (Tt − Tair ) + GP ⋅ Tair

Tex = (3.13)
1 + GP

where GP is a geometric parameter dependent on the tube and wire pitches and

diameters, given by the following equation:

⎛S ⎞ ⎛ Dw ⎞
GP = 2⎜⎜ t ⎟⎟ ⋅ ⎜⎜ ⎟⎟ (3.14)
⎝ Dt ⎠ ⎝ S w ⎠

The radiative heat transfer coefficient is a function of the wire efficiency, ηw,

because the mean external temperature is a function of the wire efficiency. The wire

efficiency, in turn, is a function of the heat transfer coefficient, as shown in Eq. 3.1.

20
Because of this interdependence, an iterative procedure must be used to calculate the

heat transfer coefficients, the wire efficiency, and the external temperature of the heat

exchanger. Bansal and Chin (2003) presented an iterative scheme for these

calculations, shown in Figure 3-2, which has been implemented in CoilDesigner.

Figure 3-2. Flow chart for iterative scheme to calculate air-side heat transfer coefficient and

heat load for natural convection wire-and-tube condensers (adapted from Bansal and Chin,

2003)

21
3.3 Forced Convection Heat Transfer Model

Many wire-and-tube condensers in modern refrigerators employ forced

convection heat transfer on the air side. As noted before, these condensers typically

consist of a long tube bent into several rows and columns to form a tube bundle. A

fan is used to circulate air through the condenser. This type of condenser is very

similar to round tube plate fin (RTPF) heat exchangers except that wires are used as

the extended surface instead of plate fins. Because of this similarity, the heat transfer

from wire-and-tube condensers can be modeled in the same way as for RTPF heat

exchangers. In other words, the ε-NTU method described in Section 2.2 can be used

to calculate the heat transfer from the refrigerant to the air. The only modification

that needs to be made is in the calculation of the air-side heat transfer coefficient.

Correlations developed specifically for wire-and-tube condensers undergoing forced

convection heat transfer are necessary. For this reason, the heat transfer coefficient

correlations developed by Hoke et al. (1997) and Lee et al. (2001) for wire-and-tube

condensers have both been included in CoilDesigner.

Hoke et al. (1997) developed two heat transfer coefficient correlations—one

for airflow perpendicular to the tubes and parallel to the wires and one for airflow

parallel to the tubes and perpendicular to the wires. The heat transfer coefficient for

airflow perpendicular to the tubes and parallel to the wires has been implemented in

CoilDesigner. Their equation for the Nusselt number is

{ (
Nu w = C ⋅ Rewn 1 − C 2 exp − C 3 S w* )} (3.15)

where Rew is the Reynolds number based on the wire diameter, and Sw* is the

dimensionless wire spacing using the wire diameter as the characteristic length:

22
 Sw
S w* = (3.16)
Dw

The constants in Eq. 3.15 are given by the following equations:

(
C = 0.259 − 0.232 cos(α ) exp − 0.00289α 2 ) (3.17)

(
n = 0.55 + 0.269 cos(α ) exp − 0.00597α 2 ) (3.18)

where α is the angle of attack of the airflow, C2 = 100, and C3 = 2.32. Currently,

CoilDesigner only has the capability of modeling airflow at an attack angle of 0o, in

which case Eq. 3.17 and 3.18 reduce to C = 0.027 and n = 0.819.

As mentioned before, Lee et al. (2001) developed correction factors to use

with the Žukauskas (1972) heat transfer coefficient correlation for a single cylinder.

With the use of these correction factors, the heat transfer coefficient correlation can

be used to model three different airflow configurations—all cross, tube cross, and

wire cross. Their correlation has been implemented in CoilDesigner for the cases of

all cross (i.e. perpendicular to the tubes and wires) and tube cross (i.e. perpendicular

to the tubes and parallel to the wires). Their correlation has the following form

K
h= (3.19)
η s Atotal

where ηs is the surface effectiveness as shown in Eq. 3.2. The value K is the air-side

thermal conductance, which is essentially calculated as a weighted combination of the

heat transfer coefficients for the tubes and for the wires. The process for calculating

K is as follows. First, a heat transfer coefficient, denoted hZ, is calculated for both the

tubes and for the wires using the Žukauskas correlation:

k air
hZ ,t = C ⋅ Retm ⋅ Prair
0.37
⋅ (3.20)
Dt

23
 k air
hZ , w = C ⋅ Re wm ⋅ Prair
0.37
⋅ (3.21)
Dw

The constants C and m are dependent on the Reynolds number and are provided in

Table 3-1.

Table 3-1. Constants C and m used to calculate the Žukauskas heat transfer coefficient

Reynolds Number C m
1 - 40 0.75 0.4
40 - 1000 0.52 0.5
5
1000 - 2x10 0.26 0.6
5 6
2x10 - 2x10 0.023 0.8

Once the individual heat transfer coefficients for the tubes and the wires have been

calculated, K is calculated for the case of airflow perpendicular to the tubes and the

wires (all-cross) according to the following equation

K = Fc ⋅ (hZ ,t At + η w hZ , w Aw ) (3.22)

For the case of airflow perpendicular to the tubes and parallel to the wires (tube-

cross), K is calculated according to the following equation

K = Fc hZ ,t At + η w F p hZ , w Aw (3.23)

In both of the preceding equations, the wire efficiency, ηw, is calculated according to

Eq. 3.1, using hZ,w as the heat transfer coefficient. The factors Fc and Fp are the

correction factors developed by Lee et al. based on if the airflow is cross flow or

parallel flow, respectively. For cross flow, they found good agreement with

experimental results by setting the correction factor, Fc, equal to a constant:

Fc = 1.3 (3.24)

For parallel flow, they derived the following equation to calculate the correction

factor:

24
 F p = 0.063 ⋅ Re 0.37 (3.25)

Modeling of forced convection wire-and-tube condensers has been performed

and compared with experimental data as part of a validation study. This modeling

work is detailed in Section 7.2. As part of the study, the predictions of the

correlations developed by Hoke et al. and by Lee et al. were compared. In agreement

with Lee et al., the correlation developed by Hoke et al. was found to underpredict

the air-side heat transfer coefficient. Thus, while both correlations have been

included in CoilDesigner, the correlation developed by Lee et al. is recommended.

25
Chapter 4 Flat Tube Heat Exchanger Model

Flat tube heat exchangers, such as those depicted in Figure 4-1 and Figure 4-2,

are often used for automotive applications such as radiators and charge air coolers.

This fluid-to-air type of heat exchanger usually contains a fluid such as a water/glycol

mixture or some other coolant inside the tubes. The use of flat tubes allows for better

airflow over the tubes compared to round tube plate fin (RTPF) heat exchangers.

Achieving better airflow can help to reduce the fan power consumption as well as the

resistance to heat transfer. In order to model flat tube heat exchangers in

CoilDesigner, a new solver has been created that can account for the unique

geometric and fluid flow characteristics of flat tube heat exchangers.

Figure 4-1. Flat tube heat exchanger with plate fins

26
 Figure 4-2. Flat tube heat exchanger with corrugated fins

The flat tube heat exchanger model must be able to model the following

different options for fluid flow configuration, fin type, and tube configuration:

• Fluid flow configurations

o Serpentine

o Parallel

• Fin types

o Plate fins

o Corrugated fins

• Tube configurations

o Inline

o Staggered

Changes have been made to CoilDesigner to allow for all of these different options.

The changes as well as the modeling equations for the heat transfer and pressure drop

are detailed in the following sections.

27
4.1 Fluid-Side Modeling

On the fluid side, the heat transfer coefficients and pressure drop are

calculated with correlations that were already included in CoilDesigner (Jiang, 2003).

The hydraulic diameter of the flat tube is used in these correlations instead of the

inner diameter of a round tube:

Ac Th ,in ⋅ Tw,in
Dh = 4 ⋅ = 4⋅ (4.1)
Pwet 2 ⋅ (Th ,in + Tw,in )

where Ac is the cross-sectional area of the inside of the tube and Pwet is the wetted

perimeter of the inside of the tube. Th,in and Tw,in are the tube inner height and the

tube inner width, respectively, as shown in Figure 4-3.

Figure 4-3. Geometric parameters of flat tubes

Flat tube heat exchangers can have two different fluid flow configurations.

The first is serpentine flow, which is similar to most RTPF heat exchangers and is

shown in Figure 4-4. Typically in this type of configuration, the fluid flows through

each tube of the heat exchanger in series, and a tube is connected to the next tube in

the tube circuitry by a bend.

28
 Figure 4-4. Flat tube heat exchanger with serpentine refrigerant flow (airflow into the page)

The second type of fluid flow configuration is parallel flow, which is the type

of flow employed in most microchannel heat exchangers and is shown in Figure 4-5.

In this type of configuration, the fluid splits into several streams inside a header and

then flows through multiple tubes in parallel. The fluid enters the tubes from one

header and then is combined at the other end of the tubes in another header before

flowing on to either the next header or to the heat exchanger outlet.

Figure 4-5. Flat tube heat exchanger with parallel refrigerant flow (airflow into the page)

29
 To simulate this type of flow configuration, the fluid flow from upstream

tubes to downstream tubes must be modeled. In order to do this, the concept of a

junction, which was defined by Jiang (2003), is used. A junction is defined as the

intersection where two or more tubes are joined together. In heat exchangers with

parallel flow, a header is considered to be a junction. In steady state, the mass flow

rate into a junction from all of the upstream tubes must equal the mass flow rate

flowing out of the junction through all of the downstream tubes. This is expressed by

the following equation

∑ m&
i
i ,in = ∑ m& i ,out
i
(4.2)

where i represents a tube. The total energy flow entering a junction from all of the

upstream tubes is also equal to the energy flow leaving the junction through the

downstream tubes:

∑ m&
i
h
i ,in i ,in = ∑ m& i ,out hi ,out
i
(4.3)

The enthalpy of the fluid entering each tube downstream of a junction is calculated as

the weighted average of the enthalpy of the fluid entering the junction from the

upstream tubes:

∑ m& h i ,in i ,in

hi ,out = i
(4.4)
∑ m& i
i ,out

4.2 Heat Transfer Between Refrigerant and Air

As in the other CoilDesigner models, each tube is divided into multiple

segments, and the energy and hydraulic calculations are performed for each segment.

30
Also, as in the CoilDesigner models for heat exchanger types that employ forced

convection on the air side, the ε-NTU method for cross-flow configuration with one

fluid mixed and the other fluid unmixed is used to calculate the heat load of each

segment (Kays and London, 1984). Once again, the air side is modeled as an

unmixed fluid and the refrigerant side is modeled as a mixed fluid. Thus, the heat

transfer between the refrigerant and the air for flat tube heat exchangers is modeled

using the same ε-NTU equations given in Chapter 2.

4.3 Air-Side Modeling

4.3.1 Fin Types for Flat Tube Heat Exchangers

Flat tube heat exchangers can have either plate fins, like those in RTPF heat

exchangers, or corrugated fins, which are the same as those found in microchannel

heat exchangers. Both types of fin have been included in the flat tube model in

CoilDesigner. Details about their implementation are included below.

Plate Fins

Flat tube heat exchangers with plate fins are very similar to round tube plate

fin heat exchangers, except for the shape of the tube. After an extensive literature

search, the only correlations that could be found for the air-side heat transfer

coefficient and pressure drop were those developed by Achaichia and Cowell (1988)

for flat tube heat exchangers with louvered plate fins. Louvered fins are often used

because they enhance the air-side heat transfer. The louvers disrupt the path of the

31
airflow, thereby increasing the turbulence of the air and impeding the formation of the

thermal boundary layer, which in turn increases the heat transfer.

No other air-side correlations could be found for flat tube heat exchangers,

even for flat plate fins because this type of fin is apparently rarely used. However, if

modeling flat plate fins is necessary, the air-side heat transfer coefficient and pressure

drop correlations by Kim et al. (1999) developed for RTPF heat exchangers could be

used with correction factors. The tube outer height, Th,out, can be set as the outer

diameter for the purposes of the air-side correlations because this is the amount of the

air stream blocked by the tube. Obviously the airflow around flat tubes and the

turbulence induced will be different than for round tubes. However, the heat transfer

coefficient and pressure drop should exhibit the same trends with respect to changes

in parameters such as air velocity, fin spacing, and tube spacing. Therefore, with

suitable correction factors, the correlations by Kim et al. should predict reasonably

accurate results for the heat transfer coefficient and pressure drop.

Figure 4-6. Flat tube heat exchanger with plate fins (airflow into the page)

32
 Since the air-side heat transfer coefficient and pressure drop correlations by

Achaichia and Cowell (1988) were the only ones that could be found for flat tube

plate fin heat exchangers, they have been implemented in CoilDesigner and are

described below.

Achaichia and Cowell found that they could obtain better correlations using

the Reynolds number based on the louver pitch, shown in Figure 4-7, rather than the

air-side hydraulic diameter. Therefore, their correlations use the Reynolds number

based on the louver pitch:

LpG
Re Lp = (4.5)
µ air

where the mass flux, G, is the mass flux through the minimum free flow area:

G = ρ air ⋅ v max (4.6)

where vmax is the maximum velocity in the core of the heat exchanger:

A frontal
v max = v air ,in ⋅ (4.7)
Amin

where Afrontal is the frontal face area of the heat exchanger and Amin is the minimum

free flow area for the air to pass through.

Figure 4-7. Diagram showing the definition of louver pitch

33
 Figure 4-8. Diagram showing the definition of louver angle and louver height

Achaichia and Cowell developed correlations for the Stanton number. If the

Reynolds number based on the louver pitch is between 150 and 3000, the Stanton

number can be calculated according to the following equation:

−0.19 −0.11 0.15

⎛ Fp ⎞ ⎛ St ⎞ ⎛ Lh ⎞
St = 1.54 ⋅ Re − 0.57 ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ (4.8)
Lp ⎜L ⎟ ⎜L ⎟ ⎜L ⎟
⎝ p ⎠ ⎝ p ⎠ ⎝ p ⎠

where the louver height, Lh, is given by the following equation:

Lh = L p ⋅ sin (Lθ ) (4.9)

If the Reynolds number is between 75 and 150, the Stanton number can instead be

calculated according to the following equation:

−0.09 −0.04
β ⎛ St ⎞ ⎛ Fp ⎞
St = 1.554 ⋅ Re − 0.59 ⎜ ⎟ ⎜ ⎟ (4.10)
Lθ
Lp ⎜L ⎟ ⎜L ⎟
⎝ p ⎠ ⎝ p ⎠

where β is a mean fluid flow angle which the authors have defined by the following

equation:

243 Fp
β = 0.936 − − 1.76 ⋅ + 0.995 ⋅ Lθ (4.11)
Re Lp Lp

Once the Stanton number has been calculated, the heat transfer coefficient is

calculated according to the following equation:

h = St ⋅ G ⋅ c p ,air (4.12)

34
 Based on their experimental measurements, Achaichia and Cowell (1988) also

created correlations to calculate the Fanning friction factor. If the Reynolds number

is between 150 and 3000, then the Fanning friction factor is calculated as follows:

f = 0.895 ⋅ f A1.07 F p−0.22 L0p.25 S t0.26 L0h.33 (4.13)

where

[0.318⋅(ln Re )− 2.25 ]
f A = 596 ⋅ Re Lp (4.14)

If the Reynolds number is less than 150, Achaichia and Cowell found that the friction

factor was best represented by the following equation:

f = 10.4 ⋅ Re Lp
-1.17
F p0.05 L1p.24 L0h.25 S t0.83 (4.15)

Once the Fanning friction factor has been calculated, the air-side pressure drop is

calculated according to the following equation:

Atotal G 2
∆P = f ⋅ ⋅ (4.16)
Amin 2 ρ air

where Atotal is the total surface area of the tubes and the fins.

Corrugated Fins

Corrugated fins, also known as serpentine fins, are depicted in Figure 4-9 and

Figure 4-10. This type of fin is used often in flat tube heat exchangers as well as in

microchannel heat exchangers. Flat tube and microchannel heat exchangers with

corrugated fins actually have the same geometry on the air side. Therefore,

correlations developed for either type of heat exchanger can be used.

35
 Figure 4-9. Flat tube heat exchanger with corrugated fins (airflow into the page)

Figure 4-10. Flat tube heat exchanger with triangular corrugated fins (airflow into the page)

Multiple heat transfer coefficient and pressure drop correlations have been

developed for corrugated fins over the past couple of decades. As a part of this

research on flat tube heat exchangers and as a part of the research on microchannel

heat exchanger simulation, a comprehensive literature search has been performed for

air-side heat transfer coefficient and pressure drop correlations for corrugated fins.

36
 The heat transfer coefficient correlations are typically provided in the form of

the Colburn j factor, which can then be used to calculate the heat transfer coefficient:

j ⋅ G ⋅ c p ,air
h= = St ⋅ G ⋅ c p ,air (4.17)
Prair2 / 3

where the Stanton number is equal to j/Pr2/3. The pressure drop correlations are

typically provided in the form of the Fanning friction factor, f, described above.

Correlations to calculate the j and f factors for plain corrugated fins were

developed by Heun and Dunn (1996) using data provided by Kays and London

(1984). These correlations have been included in CoilDesigner and are detailed in the

Appendix.

The majority of correlations have been developed for louvered corrugated fins

because this is the most common type of fin used in flat tube and microchannel heat

exchangers. Several investigators, including Davenport (1983), Rugh et al. (1992),

Sahnoun and Webb (1992), Sunden and Svantesson (1992), Dillen and Webb (1994),

and Chang and Wang (1996) developed correlations for the Colburn j factor and the

friction factor, f, for louvered fins. Chang and Wang (1997) compiled experimental

data measured by the some of the previous authors as well as by other investigators

and developed a database with 768 heat transfer coefficient measurements and 1109

friction factor measurement from a total of 91 sample heat exchangers. Chang and

Wang then used this database to develop a new generalized correlation for the j

factor. Chang et al. (2000) used the database to develop a generalized correlation for

the friction factor, f. Because their heat transfer coefficient and friction factor

correlations were developed using such an extensive set of data, they are applicable to

a rather wide range of geometries and flow conditions. These correlations have been

37
found to provide very accurate results and have become accepted standard

correlations used in industry. Therefore, the Chang and Wang (1997) and the Chang

et al. (2000) correlations have been included in CoilDesigner and are detailed in the

Appendix.

4.3.2 Tube Configurations for Flat Tube Heat Exchangers

Similar to RTPF heat exchangers, flat tube heat exchangers can have both

inline tube configurations, as shown before in Figure 4-1, and staggered tube

configurations, as shown below in Figure 4-11.

Figure 4-11. Flat tube plate fin heat exchanger with staggered tube configuration

The mass and energy conservation between the neighboring segments are

modeled in the same manner as for round tube plate fin heat exchangers (Jiang,

38
2003). For inline tube arrangements, the inlet air properties of a segment are set

equal to the outlet air properties of the corresponding segment of the tube upstream in

the airflow:

m& air ,k = m& air ,i (4.18)

m& air ,k ⋅ hair ,k ,in = m& air ,i ⋅ hair ,i ,out (4.19)

The subscripts i and k are defined in Figure 4-12. For staggered tube arrangements,

the inlet air properties of a segment are set equal to average of the outlet properties of

the two previous upstream segments:

(
m& air ,k = 0.5 ⋅ m& air ,i + m& air , j ) (4.20)

(
m& air ,k ⋅ hair ,k ,in = 0.5 ⋅ m& air ,i ⋅ hair ,i ,out + m& air , j ⋅ hair , j ,out ) (4.21)

Figure 4-12. Air-side mass and energy flow from one column of tubes to the next

39
Chapter 5 Void Fraction Models and Charge Calculation

One important feature of heat exchanger software modeling tools is the ability

to predict the mass of refrigerant, or the refrigerant charge, in a heat exchanger. The

calculation of refrigerant charge is very important in vapor compression system

simulation for charge management. In single-phase flow, the charge can be

calculated in a straightforward manner by multiplying the density of the refrigerant

times the volume. Previously the charge in a segment was calculated in CoilDesigner

similarly, by multiplying the average density of the two-phase refrigerant by the

volume of the segment. However, in an effort to improve the charge prediction

capabilities of CoilDesigner, the void fraction is now used to calculate the charge.

The void fraction is defined as the fraction of a tube occupied by vapor:

Avap
α= (5.1)
Ac

where Avap is the cross-sectional area occupied by vapor and Ac is the total cross-

sectional area of the tube:

Ac = Avap + Aliq (5.2)

In the case of annular flow in a round tube, Eq. 5.1 reduces to

2
⎛ δ⎞
α = ⎜1 − ⎟ (5.3)
⎝ R⎠

where δ is the liquid film thickness and R is the radius of the tube.

The charge in a segment is calculated according to the following equation:

Chargesegment [kg ] = (1 − α ) ⋅ ρ liq ⋅ Lseg ⋅ Ac + α ⋅ ρ vap ⋅ Lseg ⋅ Ac (5.4)

40
By accounting for the actual volume of a segment occupied by each phase of the

refrigerant this equation results in a more accurate calculation of the charge. The

charge in an entire heat exchanger is then calculated by summing the charge in each

segment of each tube:

Chargetotal [kg ] = ∑ ∑ Charge segment (5.5)

tube segment

Accurate void fraction models are needed to predict refrigerant charge.

However, analytical void fraction models typically are not very accurate (Harms et

al., 2003). Therefore, many investigators over the past few decades have developed

empirical models to calculate the void fraction in two-phase flow. An extensive

literature search of these void fraction models was performed, and multiple models

have been included in CoilDesigner. A majority of the models have been developed

for annular two-phase flow because this is the dominant flow regime in evaporators

and condensers. A rather large number of models has been included because the

predictions of different void fraction models can vary greatly (Rice, 1987).

Moreover, obtaining experimental charge data can be difficult, making it difficult to

compare model predictions with the charge of actual heat exchangers. Therefore,

selecting certain “better” correlations to include in CoilDesigner was not very

practical. All of the void fraction models that were researched have been included in

CoilDesigner, and it is left up to the user to decide which models predict charge

better. However, future studies should be performed with experimental charge data to

determine which void fraction models do a better job at predicting charge so that

certain models can be recommended.

41
 Void fraction models can be classified into four main categories—

homogeneous, slip-ratio-correlated, Lockhart-Martinelli parameter correlated, and

mass-flux dependent (Rice, 1987). Each type of model is detailed in the following

sections and correlations based on each type of model are given.

5.1 Types of Void Fraction Model

5.1.1 Homogeneous Void Fraction Model

The homogeneous void fraction models ideal two-phase flow. This model is

the most simplistic and assumes two-phase flow to be a homogeneous mixture with

the liquid and the vapor traveling at the same velocity. The void fraction in this case

can be calculated according to the following equation:

1
α= (5.6)
⎛ 1 − x ⎞ ρ vap
1+ ⎜ ⎟
⎝ x ⎠ ρ liq

Some models simply multiply a constant times the homogeneous void

fraction. Examples of this are models by Armand (1946) and Ali et al. (1993), which

can be used for microchannel tubes and are detailed in the Appendix.

5.1.2 Slip-Ratio-Correlated Void Fraction Models

Slip-ratio-correlated void fraction models build on the homogeneous model,

but the assumption that the liquid and vapor phases travel at the same velocity is

abandoned. The liquid and vapor phases are modeled as two separate streams, each

with its own velocity. The slip ratio is defined as the ratio of the vapor velocity to the

liquid velocity:

42
 v vap
S= (5.7)
vliq

For slip-ratio-correlated models, investigators develop a method to calculate the slip

ratio. The void fraction is then calculated by modifying the homogeneous void

fraction model as follows to in order to account for the slip ratio:

1
α= (5.8)
⎛ 1 − x ⎞ ρ vap
1+ ⎜ ⎟ ⋅S
⎝ x ⎠ ρ liq

Several investigators have developed empirical slip-ratio-correlated void fraction

models, and they were all developed for annular two-phase flow. Models by Thom

(1964), Zivi (1964), Smith (1969), and Rigot (1973) have been included in

CoilDesigner and are detailed in the Appendix.

5.1.3 Void Fraction Models Correlated With Lockhart-Martinelli Parameter

Another group of void fraction models avoids the homogeneous void fraction

model altogether, and instead correlates the void fraction with the Lockhart-Martinelli

parameter. These models are developed for stratified flow. The Lockhart-Martinelli

parameter, which is discussed in further detail in Section 6.6, is calculated according

to the following equation:

0.2
⎛1− x ⎞
0.9
⎛ µ liq ⎞ ρ vap
X tt = ⎜ ⎟ ⎜ ⎟ ⋅ (5.9)
⎝ x ⎠ ⎜µ ⎟ ρ liq
⎝ vap ⎠

Lockhart and Martinelli (1949) and Baroczy (1966) presented void fraction data as a

function of Xtt. Other investigators have since created correlations with their data.

43
These correlations have been implemented in CoilDesigner and are included in the

Appendix.

5.1.4 Mass-Flux-Dependent Void Fraction Models

Mass-flux-dependent void fraction models are typically correlated to the mass

flux through the use of the Reynolds number. Tandon et al. (1985) developed an

analytical model for annular flow. Premoli (1971), Yashar et al. (2001), and Harms

et al. (2003) all developed empirical models for annular flow. Hughmark (1962)

developed an empirical void fraction model for the bubbly flow regime in vertical

upward flow, but found that the correlation worked well for other flow regimes and in

horizontal tubes. Rouhani and Axelsson (1970) developed an empirical model that

can be used for all of the different boiling regions. These void fraction models are all

detailed in the Appendix.

5.2 Comparison of Void Fraction Models

A comparison of the charge predictions based on the different void fraction

models was performed. A round tube plate fin condenser was modeled in

CoilDesigner with an inlet quality of 0.99 and an outlet quality of about 0.06 in order

to cover almost the entire quality range. The charge was calculated with each built-in

void fraction model that was developed for round tubes. The results are presented in

Figure 5-1 and show that there is a wide variation in the predictions of the different

models.

44
 Obtaining experimental data regarding refrigerant charge inventory in heat

exchangers is difficult. Therefore, it is difficult to ascertain which void fraction

models provide accurate predictions. For this reason, all of the correlations that were

researched have been included in CoilDesigner for the user to choose from.

However, as stated before, experimental charge data should be obtained in the future

and studies should be performed to compare the predictions of void fraction models

with actual charge inventory.

0.5
0.45
0.4
0.35
Charge (kg)

0.3
0.25
0.2
0.15
0.1
0.05
0
vi

k
s

t
.

om

.
B a li

li
nd czy

on
ith

.
al

go

al
ou

el

al
o

ar
Zi
em

Sm

ss
tin
et

Th

hm
Ri

et

et
e

ro
en

el
ar
Pr

on

ar

s
g

Ax
rm
og

Hu
sh
rt-

Ha
m

i,
Ya
Ta

an
ha
Ho

ll,

uh
ck

ro
Lo

Ro
G

Void Fraction Model

Figure 5-1. Comparison of charge predictions based on different void fraction models

45
Chapter 6 Modeling of Effects of Oil in Heat Exchangers

In vapor compression systems used in HVAC&R systems, oil is required as a

lubricant and sealant in the compressor. Some of this oil becomes entrained in the

working fluid and is thus circulated along with the refrigerant through the different

components of a vapor compression system. The presence of oil in the working fluid

can have a significant impact on the heat transfer and pressure drop through cycle

components. In order to be able to model HVAC&R systems more accurately, and to

be able to optimize them for variables such as lubricant selection and refrigerant and

oil charge, it is necessary to be able to model the effects of oil on heat transfer and

pressure drop in evaporators and condensers as well as oil retention in these

components.

The presence of oil changes the thermodynamic and physical properties of the

working fluid. Instead of calculating properties such as temperature, density,

viscosity, and surface tension with property calls to Refprop, as is done for pure

refrigerants, methods are necessary to account for the changes due to the presence of

oil. The evaporation and condensation processes are also different when oil is

present. As opposed to evaporation and condensation processes for pure refrigerants,

which occur at a constant temperature, refrigerant-oil mixtures behave similar to

zeotropic mixtures because there is a temperature glide as the mixture quality

changes. This alters the way the heat load must be calculated because there is a

sensible heat load component in addition to the normal latent heat load component.

Correlations are also necessary for modeling the heat transfer coefficient and pressure

drop with oil present.

46
 The additional capabilities necessary to model the presence of oil have been

included in CoilDesigner. This modeling work was performed with Lorenzo

Cremaschi, a former Ph.D. student in the Center for Environmental Energy

Engineering at the University of Maryland, to simulate heat exchanger performance

with oil entrainment (2004). As a part of this modeling work, equations have been

added to calculate refrigerant-oil mixture properties. In addition, heat transfer

coefficient and pressure drop correlations have been implemented that account for the

effects of oil. To calculate the heat load during evaporation and condensation, the

method developed by Thome (1995) has been included. Models for calculating oil

retention in evaporators and condensers have also been added. After making all of

these changes, simulations were performed, and the results have been compared with

experimental results obtained by Cremaschi during his experiments regarding oil

retention in vapor compression systems. The details of the modeling work are

provided in this chapter. In Chapter 7, a comparison between modeled and

experimental results is included.

6.1 Oil Mass Fraction and Two-Phase Refrigerant-Oil Mixture Quality

The local properties of the liquid refrigerant-oil mixture, including the mixture

temperature, are highly dependent on the concentration of oil in the mixture.

However, the concentration of oil in the liquid refrigerant-oil mixture actually

changes throughout a heat exchanger. This is because the oil circulating through

vapor compression systems does not evaporate, so the oil remains concentrated

chiefly in the liquid phase refrigerant. Therefore, the concentration of oil in the liquid

47
refrigerant is dependent on the quality of the refrigerant-oil mixture. As the mixture

quality increases (i.e. more refrigerant evaporates) the concentration of oil in the

remaining liquid refrigerant increases, so it must be calculated in each segment.

In order to calculate the local oil concentration, a baseline oil concentration

for a system must be defined at a point where the refrigerant-oil mixture is completely

in the liquid phase. This occurs between the condenser outlet and the expansion

device. The absolute oil mass fraction for a system is defined at this location

according to the following equation:

m& oil
ω0 = (6.1)
m& oil + m& ref

As the refrigerant-oil mixture travels through the evaporator and condenser,

the quality of the mixture will change as refrigerant evaporates or condenses.

Analogous to the calculation of the quality of a refrigerant in the two-phase region,

the local quality of the refrigerant-oil mixture can be calculated as follows

m& ref ,vap

x mix = (6.2)
m& ref ,vap + m& ref ,liq + m& oil

Once the absolute oil mass fraction and the local mixture quality have been

calculated, the local oil mass fraction can be calculated. Using the conservation of

mass and assuming all of the oil remains in the liquid phase, the local oil mass

fraction is given by the following equation

ω0
ω local = (6.3)
1 − x mix

Because it is assumed that the oil remains in the liquid phase, there exists a

maximum possible quality for the refrigerant-oil mixture, which is less than 1:

48
 m& ref ,vap
x mix ,max = = 1 − ω0 (6.4)
m& ref ,vap + m& oil

If, during an evaporation process, the refrigerant-oil mixture reaches xmix,max, the

temperature of the mixture can increase without the quality of the mixture increasing.

Thus, because the refrigerant has evaporated out of the liquid refrigerant-oil mixture,

the refrigerant-oil mixture enters the so-called superheating region without all of the

mixture being in the vapor phase.

6.2 Bubble Point Temperature Calculation

In order to be able to model heat transfer and the physical properties of

refrigerant-oil mixtures in heat exchangers, a method to calculate the temperature of

such mixtures is necessary. Refrigerant-oil mixtures behave in a manner similar to

zeotropic refrigerants because, at a constant saturation pressure, the temperature

increases as the quality increases, resulting in a temperature glide during the

evaporation and condensation processes. The temperature difference between a

refrigerant-oil mixture and the saturation temperature of the pure refrigerant, as a

function of quality, is depicted in Figure 6-1. The temperature of such mixtures

cannot be evaluated directly with Refprop. However, in Thome’s (1995) work

developing a “thermodynamic approach” to model refrigerant-oil mixtures, he

included a method that can be used to calculate the bubble point temperature of

different refrigerant-oil mixtures using an empirical equation. This method has been

included in CoilDesigner to calculate the temperature of refrigerant-oil mixtures and

is described in the following paragraphs.

49
 Figure 6-1. Difference between refrigerant-oil mixture bubble point temperature and

refrigerant saturation temperature, as a function of quality (From Shen and Groll, 2003, p. 6)

Thome (1995) adopted an equation developed by Takaishi and Oguchi (1987)

to calculate the bubble point temperature of a refrigerant-oil mixture based on the

saturation pressure and the local oil mass fraction:

A(ω local )
Tbub = (6.5)
ln Psat − B(ω local )

where ωlocal is the oil mass fraction in the liquid in a segment and Psat is the local

saturation pressure in MPa. The constants A(ωlocal) and B(ωlocal) can be calculated

according to the following equations

A(ω local ) = a 0 + a1ω local + a 2ω local

3
+ a3ω local
5
+ a 4ω local
7
(6.6)

B(ω local ) = b0 + b1ω local + b2ω local

3
+ b3ω local
5
+ b4ω local
7
(6.7)

50
Takaishi and Oguchi fitted the constants in Eqs. 6.6 and 6.7 for an R-22/synthetic

alkyl benzene oil mixture. The values of the constants a1 – a4 and b1 – b4 are

provided in the following table:

Table 6-1. Empirical constants used in Eqs. 6.6 and 6.7 to calculate bubble point temperature of

refrigerant-oil mixtures

a1 182.52 b1 -0.72212
a2 -724.21 b2 2.3914
a3 3868.0 b3 -13.779
a4 -5268.9 b4 17.066

Thome (1995) extended Eqs. 6.6 and 6.7 to other refrigerant-oil mixtures

beyond R-22/synthetic alkyl benzene oil by suggesting that only a0 and b0 must be

adjusted. He reasoned that the values of the other constants can be considered to

remain constant for different refrigerant-oil pairs and saturation pressures because the

vapor pressure of the oil is negligible. Thome suggested that the values of the

constants a0 and b0 can be correlated depending on the specific pure refrigerant and

the saturation pressure. They are calculated using the following procedure:

- Calculate the saturation temperatures for the pure refrigerant at two different

pressures, one just below the local saturation pressure and one just above

- Set ωlocal equal to zero and evaluate Eq. 6.5 with the two different pairs of Tsat

and Psat

- Solve the resulting system of two equations for the two unknowns, a0 and b0

This method provides a relatively simple, yet accurate way to calculate the bubble

point temperature of different refrigerant-oil mixtures during evaporation and

condensation processes.

51
6.3 Heat Load Calculation and the Heat Release Enthalpy Curve

As in the CoilDesigner solver for pure refrigerants, the ε-NTU method is used

to calculate the heat load in a segment with a refrigerant-oil mixture. For pure

refrigerant undergoing an evaporation or condensation process, the heat load causes

boiling or condensing, resulting in an increase or decrease in the enthalpy and the

quality with no change in temperature. The change in enthalpy is given by the

equation

dh = hLV ⋅ dx (6.8)

On the other hand, for a refrigerant-oil mixture, the heat load causes evaporation or

condensation as well as a change in the temperature of the refrigerant-oil mixture.

The change in enthalpy in a segment consists of three components (Thome, 1995):

- Latent heat due to evaporation or condensation

- Sensible heat due to the change in bubble point temperature of the liquid-

phase refrigerant-oil mixture

- Sensible heat due to the change in bubble point temperature of the vapor-

phase refrigerant

The change in enthalpy in this case is given by the following equation, known as the

heat release enthalpy curve:

dh = hLV ⋅ dx mix + (1 − x mix ) ⋅ c p ,mix ⋅ dTbub + x mix ⋅ c p ,ref ,vapor ⋅ dTbub (6.9)

Once the heat load of a segment is calculated using the ε-NTU method, the

resulting change in quality, dxmix, and the resulting change in the bubble point

temperature, dTbub, must be calculated. Essentially this means dividing the heat load

52
into sensible and latent components. This is necessary so that the outlet temperature

and quality of the refrigerant-oil mixture can be passed as inputs to the next segment.

As described in Section 6.2, the bubble point temperature is dependent on the

quality of the refrigerant-oil mixture. Thus, all three terms on the right-hand side of

Eq. 6.9 are functions of dxmix. The result is that an iterative numerical scheme must

be used to calculate the correct values of dxmix and dTbub. Ridders’ method (Press,

1992) has been implemented to calculate the change in quality that will result in the

correct outlet temperature and the correct heat load.

In the iterative method employed, an outlet quality is guessed. For an

evaporation process, the guessed outlet quality is between the inlet quality and

xmix,max. For a condensation process, the guessed outlet quality is between the inlet

quality and zero. Based on the guessed outlet quality, the outlet oil mass fraction,

ωlocal, is calculated according to Eq. 6.3, and then the outlet temperature is calculated

according to Eq. 6.5.

Once an outlet quality has been guessed and the outlet temperature has been

calculated, the change in quality and temperature through the segment, dxmix and

dTbub, can be calculated. The change in enthalpy can then be calculated according to

Eq. 6.9. This change in enthalpy is then compared to the heat load calculated using

the ε-NTU method. If these two heat load values do not agree within a specified

tolerance, Ridders’ method guesses another value for the outlet quality and the steps

above are repeated.

53
6.4 Calculation of Refrigerant-Oil Mixture Properties

The thermodynamic and transport properties of refrigerant-oil mixtures are

very important for calculating heat transfer and pressure drop, so methods for

calculating these properties were necessary. Shen and Groll (2003) provided a

comprehensive review of the literature regarding refrigerant-oil mixture property

calculations. They suggested methods for calculating properties such as density,

viscosity, and specific heat of refrigerant-oil mixtures. These methods have been

included in CoilDesigner to calculate mixture properties and are summarized below.

Unless specified otherwise, the properties of oil used in the following equations were

obtained from manufacturing data. The properties of liquid and vapor refrigerant are

calculated by making function calls to Refprop using Psat and Tbub as the inputs.

- Liquid mixture density

The density of the mixture of liquid refrigerant and oil is calculated using the

method given by Jensen and Jackman (1984):

1 ω local 1 − ω local
= + (6.10)
ρ mix ρ oil ρ ref ,liq

- Liquid mixture viscosity

The viscosity of the mixture of liquid refrigerant and oil is calculated using

the method provided by Yokozeki (1994):

ln µ mix = ∑ ξ i ⋅ ln µ i = ξ ref ,liq ⋅ ln µ ref ,liq + ξ oil ⋅ ln µ oil (6.11)

i

where ξi is the Yokozeki factor

Wi kΨ i
ξi = (6.12)
∑ W jkΨ j
j

54
 and Wi and Ψi are the molecular mass and the mole fraction, respectively, of

component i. The exponent k is an empirical constant that is specific to

different refrigerant-oil pairs. However, Yokozeki found that setting k = 0.58

provided accurate results for most refrigerant-oil pairs. The mole fraction of

the oil can be obtained from the equation

⎛ Wref ⎞
ω local ⎜⎜ ⎟⎟
Ψ oil = ⎝ Woil ⎠ (6.13)
⎛ Wref ⎞
1 − ω local + ω local ⎜⎜ ⎟⎟
⎝ Woil ⎠

- Liquid mixture surface tension

The surface tension of the mixture of liquid refrigerant and oil is also

calculated according to the method provided by Jensen and Jackman (1984):

σ mix = σ ref ,liq + (σ oil − σ ref ,liq ) ω local (6.14)

- Liquid mixture thermal conductivity

The thermal conductivity of the mixture of liquid refrigerant and oil is

calculated according to the method provided by Filippov and Novoselova

(1955):

k mix = k ref ,liq (1 − ω local ) + k oil ω local − 0.72(k oil − k ref ,liq )(1 − ω local )ω local (6.15)

- Liquid mixture specific heat

The specific heat of the mixture of liquid refrigerant and oil is calculated

according to the method provided by Jensen and Jackman (1984):

c p ,mix = ω local c p ,oil + (1 − ω local )c p ,ref ,liq (6.16)

The specific heat of the oil is calculated according to the following equation

recommended by Thome (1995)

55
 ( [ ]
⎛ 0.338 + 0.00045 ⋅ 1.8 ⋅ T o C + 32
c p ,oil = 4.186⎜
) ⎞⎟ (6.17)
⎜ sg ⎟
⎝ ⎠

where sg is the specific gravity of the oil, sg = ρoil/ρwater, at 15.6oC.

- Liquid and vapor phase Reynolds numbers

The Reynolds numbers for the liquid and vapor phases are calculated using

the quality of the refrigerant-oil mixture, xmix:

Gtotal ⋅ (1 − x mix ) ⋅ D
Reliq = (6.18)
µ mix ,liq

Gtotal ⋅ x mix ⋅ D
Revap = (6.19)
µ ref ,vap

6.5 Heat Transfer Coefficient Correlations for Refrigerant-Oil Mixture

6.5.1 Heat Transfer Coefficient for Evaporation

As mentioned before, the ε-NTU method was used to calculate the heat load

for refrigerant-oil mixtures. This method requires the calculation of a heat transfer

coefficient on the refrigerant side. Thus, in order to be able to model refrigerant-oil

mixtures accurately, correlations that account for the presence of oil are needed.

While some previous investigators have used mixture properties directly in

correlations developed for pure refrigerants to predict the heat transfer coefficient of

refrigerant-oil mixtures, other investigators have created semi-empirical heat transfer

coefficient correlations based on pure refrigerant properties (Shen and Groll, 2003).

However, as discussed by Shen and Groll (2003), these correlations do not generally

provide very accurate predictions because either they were developed for pure

56
refrigerants or because they use pure refrigerant properties. Therefore, they do not

capture the true behavior of refrigerant-oil mixtures during evaporation.

Chaddock and Mathur (1980) developed the only semi-empirical correlation

that is based on refrigerant-oil mixture properties. They developed their correlation

using experimental results for flow boiling of R-22/Suniso 3 GS oil inside a smooth

tube. Shen and Groll (2003) recommended extending the semi-empirical correlation

developed by Chaddock and Mathur to other refrigerant-oil mixture pairs. Therefore,

the correlation, which has the following form, has been implemented in CoilDesigner:
n
htwo − phase,mix ⎛ 1 ⎞
= c⎜⎜ ⎟⎟ (6.20)
hliq ⎝ X tt ⎠

where Xtt is the Lockhart-Martinelli parameter of the refrigerant-oil mixture. The

calculation of the Lockhart-Martinelli parameter is described in Section 6.6. The

constants c and n are semi-empirical constants that depend on the oil mass fraction

and are given in Table 6-2. The liquid heat transfer coefficient is calculated using the

Dittus-Boelter correlation (1930) with the refrigerant-oil mixture properties:

⎛ k ref ,liq ⎞ 0.8

hliq = 0.023⎜⎜ ⎟⎟ Remix ,liq ⋅ Prmix
0.4
,liq
⎝ D ⎠
0.8 0.4 (6.21)
⎛ k ref ,liq ⎞⎛ G ⋅ (1 − x mix ) ⋅ D ⎞ ⎛ c p ,ref ,liq ⋅ µ mix ,liq ⎞
= 0.023⎜⎜ ⎟⎟⎜ ⎟ ⎜ ⎟
⎜ µ ⎟ ⎜ ⎟
⎝ D ⎠⎝ mix ,liq ⎠ ⎝ k ref ,liq ⎠

where G is the total mass flux of the refrigerant and oil. The correlation was

developed by Chaddock and Mathur for mass fluxes varying from 149.0 to 908.5

kg/(m2 s), heat fluxes varying from 7.73 to 40.54 kW/m2, and vapor quality varying

from 0.0 to 1.0.

57
Table 6-2. Coefficients c and n as a function of the oil mass fraction in the correlation developed

by Chaddock and Mathur (1980) for the heat transfer coefficient of refrigerant-oil mixtures

Oil Mass Fraction (wt. %) c n

0.0 3.90 0.62
1.0 4.72 0.59
2.9 4.36 0.60
5.7 4.97 0.59

6.5.2 Heat Transfer Coefficients for Condensation

The heat transfer coefficient for single-phase refrigerant-oil mixtures in the

subcooled region can be calculated using traditional single-phase correlations such as

the Dittus-Boelter (1930) and Gnielinski (1976) correlations with the refrigerant-oil

mixture properties.

As with evaporating heat transfer coefficients, previous researchers have

modeled two-phase convective condensing heat transfer coefficients of refrigerant-oil

mixtures using pure refrigerant correlations with the mixture properties or by

developing semi-empirical correlations based on either pure refrigerant properties or

refrigerant-oil mixture properties (Shen and Groll, 2003). The presence of oil has

been shown to have less of an adverse effect on convective condensation compared to

convective evaporation. Therefore, efforts to calculate the condensation heat transfer

coefficient using existing correlations with refrigerant-oil mixture properties have met

with greater success than for evaporation. Shah’s condensation correlation (1979)

using mixture properties has been shown to predict the heat transfer coefficient

accurately (Shen and Groll, 2003), so it has been implemented in CoilDesigner.

Shah’s method first calculates the liquid-phase heat transfer coefficient using the

58
Dittus-Boelter correlation, given in Eq. 6.21, and then multiplies this quantity by the

following two-phase multiplier:

0.76
3.8 ⎛ x mix ⎞
C = 1 + 0.38 ⎜⎜ ⎟⎟ (6.22)
Pr ⎝ 1 − x mix ⎠

where Pr is the reduced pressure of the refrigerant (P/Pcrit).

6.6 Pressure Drop Correlation for Two-Phase Refrigerant-Oil Mixture

The presence of oil can reduce the frictional pressure drop as the refrigerant

flows through the tubes of a heat exchanger (Shen and Groll, 2003). The main reason

for this phenomenon is that the oil increases the viscosity of the fluid, which can

cause a transition to laminar flow from what would normally be turbulent flow. This

can be illustrated with the equation for the liquid Reynolds number, Eq. 6.18, in

which increasing the viscosity decreases the Reynolds number. The liquid Reynolds

number can drop below the transition point between laminar and turbulent, especially

at higher qualities, since it is proportional to 1–xmix.

The Lockhart-Martinelli correlation accounts for the flow state (laminar or

turbulent) when calculating the pressure drop. Thus, the influence of oil on the flow

state and hence on the pressure drop can be captured by using the Lockhart-Martinelli

correlation with mixture properties (Shen and Groll, 2003). Therefore, it has been

recommended by Shen and Groll and implemented in CoilDesigner to predict the

pressure drop of refrigerant-oil mixtures. The pressure drop is calculated according

to the following procedure. First, the Lockhart-Martinelli parameter is calculated

according to the following equation:

59
 0.5
⎡ − ∆Pmix ,liq ⎤
X tt = ⎢ ⎥ (6.23)
⎣⎢ − ∆Pvap ⎦⎥

where the liquid-phase and vapor-phase frictional pressure drops are calculated

according to the following two equations:

Lseg G 2 (1 − x mix )
2

∆Pmix ,liq = f mix ,liq (6.24)

2 Dρ mix ,liq

Lseg G 2 x mix
2

∆Pvap = f vap (6.25)

2 Dρ vap

where Lseg is the length of the segment. The friction factor for each phase is

calculated according to the following equations, depending on whether the flow is

laminar or turbulent:

16
fk = for Rek < 2000 (6.26)
Rek

0.046
fk = for Rek > 2000 (6.27)
Rek0.2

where the liquid and vapor Reynolds numbers are calculated as in Eqs. 6.18 and 6.19.

When both the liquid phase and the vapor phase are turbulent, the Lockhart-Martinelli

parameter reduces to the following rather well-known equation:

0.9 0.2
⎛ 1 − x mix ⎞ ⎛ µ mix ,liq ⎞ ρ vap
X tt = ⎜⎜ ⎟⎟ ⎜ ⎟ ⋅ (6.28)
⎜ µ ⎟ ρ mix ,liq
⎝ x mix ⎠ ⎝ vap ⎠

After the Lockhart-Martinelli parameter has been calculated, the two-phase

multiplier for each phase must be calculated according to the following equations:

C 1
φ mix
2
,liq = 1 + + 2 (6.29)
X tt X tt

60
 φvap
2
= 1 + CX tt + X tt2 (6.30)

where the constant C is determined by the flow states of the vapor and the liquid, as

shown in Table 6-3.

Table 6-3. Constant C used to calculate the two-phase multipliers used in the Lockhart-

Martinelli correlation

Liquid Flow State Vapor Flow State C

Laminar Laminar 5
Laminar Turbulent 10
Turbulent Laminar 12
Turbulent Turbulent 20

The two-phase frictional pressure drop is then calculated as the maximum of

the pressure drop multiplied by the two-phase multiplier for each phase:

(
∆Pfrictional = Max ∆Pmix ,liq ⋅ φ mix
2
,liq ; ∆Pvap ⋅ φ vap
2
) (6.31)

6.7 Oil Retention and Void Fraction Models

One of the motivations for modeling oil in heat exchangers is to model the oil

retention for purposes of oil charge management. In order to calculate the oil

retention, the void fraction of the two-phase refrigerant-oil mixture must be

calculated. The fraction of a segment occupied by liquid is equal to (1 – α), and the

fraction of the liquid mixture comprised of oil is ωlocal. Therefore, the oil retention in

a segment can be calculated using a fairly simple modification to Eq. 5.4, the

equation for calculating the charge in a segment:

ORM segment [kg ] = ω local ⋅ Vsegment ⋅ (1 − α ) ⋅ ρ mix.liq (6.32)

The total oil retention in an entire heat exchanger is then calculated by summing up

the oil retention in each segment of every tube:

61
 ORM total [kg ] = ∑ ∑ ORM segment (6.33)
tube segment

Several void fraction models, such as those by Lockhart-Martinelli (1949),

Hughmark (1962), and Premoli et al. (1971), have been recommended in the recent

literature for the purposes of calculating the void fraction of refrigerant-oil mixtures

(Shen and Groll, 2003). In order to use these void fraction models for refrigerant-oil

mixtures, the liquid properties such as density are simply replaced with the mixture

properties. All of these void fraction models are included in the Appendix. They

were all tested with experimental data, and the Premoli void fraction model was

found to provide the most accurate results.

62
Chapter 7 Validation and Optimization Studies

7.1 Validation of Microchannel Heat Exchanger Model

A study was performed to determine the accuracy of the microchannel heat

exchanger model. The model was validated with experimental data from 8

microchannel condensers using the refrigerant R-134a. Seven of the heat exchangers

comprised of 48 tubes, with the refrigerant making a first pass through 32 tubes and a

second pass through 16 tubes. The other heat exchanger comprised of 24 tubes, with

the refrigerant making a first pass through 18 tubes and a second pass through 6

tubes. The fins used were louvered, corrugated fins. The geometric parameters of the

heat exchangers are provided in Table 7-1. All of the heat exchangers used in the

validation had a tube height of 1.9 mm and a fin thickness of 0.114 mm. The

experimental data was measured for varying air and refrigerant flow rates and varying

inlet conditions. Results from a total of 35 tests were used in the model validation.

Table 7-1. Geometric parameters of the microchannel heat exchangers used for validation

No. Hydraulic Tube Tube Tube Fin Fin Fin Louver Louver No.
of Diameter Depth Length Pitch Density Depth Height Length Pitch Data
Tubes (mm) (mm) (mm) (mm) (fpi) (mm) (mm) (mm) (mm) Points
48 0.775 18.77 479.6 10.16 19.6 18.8 7.9 6.6 1.4 4
48 0.775 18.77 479.6 10.16 11.8 21.1 7.9 6.6 1.4 4
48 0.775 18.77 479.6 10.16 19.6 21.1 7.9 6.6 1.4 4
48 0.775 18.77 479.6 10.16 23.6 21.1 7.9 6.6 1.4 4
48 0.771 13.54 479.6 10.16 19.9 15.9 12.7 10.9 1.14 5
48 0.771 13.54 479.6 10.16 19.9 15.9 10.8 9.5 1.4 5
48 0.771 13.54 479.6 10.16 20 15.9 10.0 8.7 1.4 5
24 0.775 18.77 1069 10.48 24 21.1 7.92 6.6 1.4 4

For the air-side heat transfer coefficient and pressure drop calculations,

correlations developed specifically for microchannel or flat tube heat exchangers with

louvered fins were necessary. The correlations presented by Chang and Wang (1997)

63
and Chang et al. (2000) were used for the air side Colburn j factor and the fraction

factor, f. These correlations are detailed in the Appendix. On the refrigerant side, the

Gnielinski correlation (1976) was used for the calculation of single-phase heat

transfer coefficients. The correlation developed by Dobson and Chato (1998) for heat

transfer during condensation was used in the two-phase region. Good results were

obtained using this correlation with a correction factor of 1.5. For the refrigerant-side

pressure drop calculations, the correlation developed by Churchill (Chisholm, 1983)

was used in the single-phase region, and the correlation developed by Jung and

Radermacher (1989) was used for the two-phase region.

The results of the validation are presented in Figure 7-1 and Figure 7-2. As

can be seen in Figure 7-1, the heat load was predicted very accurately. 80% of the

heat load data points were predicted within 2.25%. All of the heat load data points

were predicted with an average error of -0.84%, an average absolute error of 1.6%,

and a maximum error of 4.6%. For the refrigerant pressure drop, shown in Figure

7-2, 54.3% of the data points were predicted within 25%, and the average error,

average absolute error, and maximum error were -12.1%, 25.9%, and 62.8%,

respectively. The refrigerant pressure drop was generally underpredicted. This could

be due in part to the fact that the pressure drop through the inlet tube to the heat

exchanger was not modeled.

64
 13000
±2.25%
12000

11000
Predicted Heat Load (W)
10000

9000

8000

7000

6000

5000

4000
4000 5000 6000 7000 8000 9000 10000 11000 12000 13000
Measured Heat Load (W)

Figure 7-1. Predicted heat load vs. experimentally measured heat load of microchannel heat

exchangers used for validation

40
±25%
35
Calculated Pressure Drop (kPa)

30

25

20

15

10

0
0 5 10 15 20 25 30 35 40
Measured Pressure Drop (kPa)

Figure 7-2. Predicted refrigerant pressure drop vs. experimentally measured pressure drop of

microchannel heat exchangers used for validation

65
7.2 Validation of Wire-and-Tube Condenser Model

An initial validation of the wire-and-tube condenser model for forced

convection heat transfer on the air side was performed using experimentally

determined data. The data was measured for 3 different condensers—one condenser

with airflow perpendicular to the tubes and parallel to the wires and two condensers

with airflow perpendicular to both the tubes and the wires. Two test cases were

measured for each condenser.

The experimental results were compared to simulations using both of the

forced convection air-side heat transfer coefficients—namely the correlations

developed by Hoke et al. (1997) and by Lee et al. (2001). The correlation developed

by Lee et al. provided more accurate results. The correlation developed by Hoke et

al. was found to underpredict the heat transfer coefficient. This result is in agreement

with the findings by Lee et al. Therefore, the air-side heat transfer coefficient from

Lee et al. was used for the validation study.

On the refrigerant side, the Gnielinski (1976) correlation was used to calculate

the single-phase heat transfer coefficient. The correlation developed by Cavallini and

Zecchin (1974), given in the Appendix, was used to calculate the heat transfer

coefficient during condensation. For the calculation of the friction factor the

correlation developed by Churchill (Chisholm, 1983) was used for single-phase

refrigerant and the Lockhart-Martinelli (1949) correlation was used for two-phase

refrigerant.

The normalized results of the validation are presented in Figure 7-3 and

Figure 7-4. The results have been normalized in order to prevent the disclosure of

66
any condenser specifications. As shown in Figure 7-3, the heat load was predicted

very accurately, with all of the heat load data points being predicted within 2.0%,

with an average error of 1.0%. The normalized refrigerant pressure drop results are

shown in Figure 7-4. All of the pressure drop data points were predicted within 20%

with an average error of 3.8% and an average absolute error of 7.0%.

The accuracy of the model, as exemplified by these two figures, is very

promising. However, as mentioned previously, only six data points were used in the

validation. In the future, more validation studies of the wire-and-tube condenser

model should be performed with more data points over a wider range of geometries

and flow parameters.

1.5
Predicted Heat Load (Normalized)

1.4

±2%
1.3

1.2

1.1

1
1 1.1 1.2 1.3 1.4 1.5
Measured Heat Load (Normalized)

Figure 7-3. Predicted heat load vs. experimentally measured heat load of wire-and-tube

condensers used for validation

67
 3

Predicted Pressure Drop (Normalized)

2.5
±20%
2

1.5

0.5

0
0 0.5 1 1.5 2 2.5 3
Measured Pressure Drop (Normalized)

Figure 7-4. Predicted pressure drop vs. experimentally measured pressure drop of wire-

and-tube condensers used for validation

7.3 Validation of Refrigerant-Oil Mixture Model

The refrigerant-oil mixture model described in Chapter 6 was used to model

the heat load and oil retention of evaporators and condensers. In order to validate the

model, the predicted results were then compared with experimental results obtained

by Cremaschi (2004).

As a part of his experimental work, Cremaschi measured oil retention in round

tube plate fin evaporators and condensers by injecting oil into the refrigerant flow at

the inlet of a heat exchanger and then collecting the oil in an oil accumulator at the

outlet of the heat exchanger. The oil retention in a heat exchanger could then be

calculated as the difference between the oil injection flow rate and the rate of

increasing oil volume in the oil accumulator. The experiments focused on three

68
different oils: mineral oil (MO) as well as the synthetic lubricants polyolester (POE)

and polyalkylene glycole (PAG). Oil retention tests were performed with R-22/MO,

R-410A/MO, R-410A/ POE, R-134a/POE, and R-134a/PAG mixtures. The oil mass

fraction was varied between 1 and 8% by weight of the total mass flow rate of the

refrigerant-oil mixture. The refrigerant mass flow rates were also varied between 42

and 60 g/s.

The results from modeling the oil retention in the evaporator are shown in

Figure 7-5. As discussed before, the void fraction model by Premoli et al. (1971)

provided the most accurate predictions for the oil retention. The average absolute

error of all of the oil retention predictions was about 21%, and 72% of the data points

were predicted within ±29%.

Figure 7-5. Experimentally measured oil retention vs. predicted oil retention in the evaporator

(from Cremaschi, 2004)

69
 CoilDesigner provides segment-by-segment output data, so detailed results for

the oil retention, mixture quality, and local oil mass fraction can be obtained. Figure

7-6 shows these three quantities with respect to the dimensionless length of the

evaporator. The local oil retention is plotted in dimensionless form, which is

calculated as the ratio of the local oil retention volume over the maximum oil

retention in a segment in the evaporator. The results shown in this figure are for the

evaporator modeled with R-134a/PAG and an oil mass fraction of 2.4%. The tube

circuitry of the evaporator is shown at the top of the figure. Of considerable note are

the two kinks in the mixture quality curve corresponding to the points where the

refrigerant-oil mixture flows from one column of tubes to the next. CoilDesigner

predicts a change in the rate of increase of the mixture quality at these two points.

This is to be expected because, as the fluid approaches the warmer inlet air, the

refrigerant should evaporate at a greater rate. As can also be seen in Figure 7-6, the

local oil mass fraction in the liquid mixture starts with a low value at the inlet to the

evaporator because most of the refrigerant is in the liquid phase. As the refrigerant

evaporates along the length of the heat exchanger, the oil remains in the liquid phase,

so the local oil mass fraction increases. Closer to the outlet of the evaporator, most of

the refrigerant has evaporated, so the local oil mass fraction is much higher. As a

result, the viscosity of the liquid refrigerant-oil mixture is much higher towards the

outlet of the heat exchanger, so most of the oil was retained in the last section of the

evaporator.

70
Figure 7-6. Calculated oil retention, mixture quality, and local oil mass fraction in an evaporator

with R-134a/PAG at OMF=2.4% (from Cremaschi, 2004)

In addition to the oil retention, the heat load results of the evaporator were

also compared with the experimental results. The heat load was predicted accurately

with all of the data points predicted with an average absolute error of 12%.

The modeling results for the condenser are presented in Figure 7-7. The

average absolute error for all of the oil retention data points was 23%, and about 70%

of the data points were predicted within ±30%.

71
Figure 7-7. Experimentally measured oil retention vs. predicted oil retention for the condenser

(from Cremaschi, 2004)

7.4 Optimization Study of Wire-and-Tube Condenser

One of the major motivations for creating heat exchanger models is to be able

to perform optimization studies on heat exchangers in order to improve their capacity,

efficiency, cost, and volume. The wire-and-tube condenser model was used to

perform an optimization study to improve the cost and heat load of a wire-and-tube

condenser that is currently used by a refrigerator manufacturer. Geometric

parameters of the condenser were varied according to data provided by the

manufacturer, but the number of different design possibilities was somewhat limited.

Therefore, an optimization could be performed by searching the entire design space

72
for the best condenser designs. For this reason, more sophisticated techniques such as

genetic algorithms were not necessary to perform the optimization study.

The following parameters were varied during the optimization study:

• Wire diameter

• Number of wires

• Tube diameter

• Horizontal and vertical distances between tubes

Each of these parameters was varied over a certain range of values, and every

possible combination of parameters in those ranges was modeled. The cost for each

test condenser was calculated according to cost information provided by the sponsor.

Two different optimization studies were performed. The first study used the

same tube configuration as the existing condenser. The other optimization study

involved modeling condensers with larger frontal face areas. This was accomplished

by decreasing the number of columns of tubes in the direction of the airflow and

increasing the number of rows of tubes in the vertical direction.

7.4.1 Optimization of Original Condenser

As noted above, the first optimization was performed using the same tube

configuration as the existing condenser. Therefore, the number of rows and the

number of columns of were kept constant while the parameters listed above were

varied. In all, a total of 448 possible condensers were modeled. The heat load and

the cost of the test condensers were compared with those of the original condenser.

Figure 7-8 shows the heat load results with respect to the cost for all of the possible

73
condensers. The heat load and the cost have been normalized to prevent the

disclosure of any proprietary data, so the original condenser has a value of 1 for both

values.

Test Coils Baseline Coil

1.1

1.05
Heat Load (Normalized)

Baseline Coil
0.95

0.9

0.85

0.8

0.75
0.65 0.75 0.85 0.95 1.05 1.15 1.25 1.35
Cost (Normalized with respect to baseline coil)

Figure 7-8. Heat load vs. cost of all test condensers in optimization of baseline condenser

In the optimization study, any condenser that was modeled as having a heat

load greater than or equal to the baseline condenser and a cost less than or equal to

the baseline condenser was considered to be better than the baseline condenser. Thus,

in Figure 7-8, any point to left of and above the baseline condenser is considered to

be an improvement. Of the 448 different condensers modeled, 147 of them fit these

criteria. These better condensers are shown in Figure 7-9. Of particular note is the

condenser at the far left of the graph. This condenser is modeled as having a heat

load about 6.7% greater than the baseline condenser with a cost reduction of about

15%.

74
 Better Coils Baseline Coil
1.035

1.03

1.025
Heat Load (Normalized)

1.02

1.015
15% Cost Reduction
1.01

1.005

1
Baseline Coil
0.995
0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 1 1.02
Cost (Normalized with respect to baseline coil)

Figure 7-9. Heat load vs. cost for all better condensers in optimization of baseline condenser

7.4.2 Optimization of Condenser with Larger Face Area

Another optimization study was performed in which wire-and-tube condensers

with larger frontal face areas were modeled, and the results were compared to the

original condenser. The motivation behind modeling condensers with larger face

areas was to design a condenser that could utilize more efficiently the space available

to the condenser underneath the refrigerator. By filling more of the space available,

better airflow through the condenser can be achieved, thereby increasing the heat

transfer.

In order to model condensers with larger face areas, the number of columns in

the direction of the airflow was reduced and the number of rows in the vertical

direction was increased. It was assumed that for a larger face area, a larger fan would

75
be necessary. However, larger fans require more power to rotate at the same speed as

a smaller fan. Therefore, in order to maintain or improve the efficiency of the entire

refrigeration system, the speed of the new fan should be decreased to maintain the

same power consumption as the original fan. The fan speed and the air velocity for

the larger fan were calculated using the following procedure with the ASHRAE Fan

Laws (ASHRAE, 2000). First, the power of the new fan, W1, was assumed to be

equal to the power of the original fan, W2. Fan Law 1c relates the power of one fan to

the power of another fan:

5 3
⎛D ⎞ ⎛ N1 ⎞ ⎛ ρ1 ⎞
W1 = W2 ⎜⎜ 1 ⎟⎟ ⎜⎜ ⎟⎟ ⎜⎜ ⎟⎟ (7.1)
⎝ D2 ⎠ ⎝ N2 ⎠ ⎝ ρ2 ⎠

Assuming constant fan power and constant air density, Eq. 7.1 reduces to the

following equation for the speed of the new fan, N1:

5/3
⎛D ⎞
N 1 = N 2 ⎜⎜ 2 ⎟⎟ (7.2)
⎝ D1 ⎠

Once the fan speed of the new condenser is calculated, the volumetric flow rate of the

new fan, Q1, can be calculated according to Fan Law 1a:

3
⎛D ⎞ ⎛ N1 ⎞
Q1 = Q2 ⎜⎜ 1 ⎟⎟ ⎜⎜ ⎟⎟ (7.3)
⎝ D2 ⎠ ⎝ N2 ⎠

which reduces to the following equation when Eq. 7.2 is substituted into Eq. 7.3:

4/3
⎛D ⎞
Q1 = Q2 ⎜⎜ 1 ⎟⎟ (7.4)
⎝ D2 ⎠

Once the volume flow rate of the air has been calculated, the average air velocity

through the new condenser can be calculated according to the following equation:

76
 Q1
v avg = (7.5)
A1

This average air velocity is then set as the input to the new condenser under

consideration.

In addition to varying the number of rows and columns of tubes and the fan

diameter in this optimization study, the parameters mentioned previously were also

varied. Thus, for each tube configuration, several different values were modeled for

the wire diameter, the number of wires, the tube diameter, and the horizontal and

vertical distances between tubes. A total of 1,344 different condensers were modeled,

of which 219 were found to be better than the original condenser. The normalized

heat load and cost results of all the condensers are shown in Figure 7-10.

Test Coils Baseline Coil

170

165

160

155
Baseline Coil
Heat Load (W)

150

145

140

135

130

125

120
0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4
Cost (Normalized with respect to baseline coil)

Figure 7-10. Heat load vs. cost of all test condensers in optimization of condensers with larger

face area

77
 Better Coils Baseline Coil
1.05

1.045

1.04
Heat Load (Normalized)

1.035

1.03

1.025

1.02

1.015
14.5% Cost
1.01

1.005

1
Baseline Coil
0.995
0.85 0.87 0.89 0.91 0.93 0.95 0.97 0.99 1.01
Cost (Normalized with respect to baseline coil)

Figure 7-11. Heat load vs. cost for all better condensers in optimization of condensers with

larger face area

Figure 7-11 shows the condensers that were modeled to be better than the

original condenser. Of particular note in this optimization study is the point at the far

left of Figure 7-11. This condenser was modeled to have a heat load slightly greater

than the original condenser with a cost 14.5% less than the original condenser.

The company for which this optimization work was performed has since built

a few condenser prototypes based on the results of the optimization. The company

has reportedly obtained good results with a couple of the condenser designs. Though

no details have been provided as far as which condensers provided better results, this

news from the company is encouraging and provides motivation to perform other heat

exchanger optimization studies with CoilDesigner.

78
Chapter 8 Conclusions

Several advances have been made to CoilDesigner to increase its modeling

capabilities. The main motivation behind the work detailed in this thesis has been to

improve the simulation and optimization capabilities of the heat exchanger design

tool. The major advances and conclusions of this thesis are summarized in the Table

8-1 and then in the sections that follow.

Table 8-1. Summary of modeling capabilities added to CoilDesigner and work performed in

relation to this thesis

Models Included/Work Comments

Performed
Heat Exchanger o Wire-and-tube condenser model The models existing prior
Models - Forced convection model to the current research
- Natural convection model were:
o Flat tube heat exchanger model o Round tube plate fin
o Microchannel
Void Fraction o Homogeneous model Void fraction models were
Models o Slip-ratio-correlated models not previously available in
o Lockhart-Martinelli parameter CoilDesigner. Multiple
correlated models correlations of each type
o Mass-flux-dependent models of void fraction model
have been included.
Refrigerant-Oil o Property calculations of
Mixture Model refrigerant-oil mixtures
o Heat load model accounting for
temperature glide
o Heat transfer coefficient and
pressure drop correlations
accounting for refrigerant-oil
mixture properties
Validation and o Validation of microchannel heat Previously, the round tube
Optimization exchanger model plate fin model was the
Studies o Validation of wire-and-tube only model that had been
condenser model validated.
o Validation of refrigerant-oil
mixture model
o Optimization of wire-and-tube

79
 condenser
Additional o Louvered fins for RTPF, For the most part, these
Modeling microchannel, and flat tube heat modeling capabilities
Capabilities exchangers have not been described in
o Slit fins for RTPF heat detail in this thesis.
exchangers Extensive literature
o Serpentine microchannel heat searches have been
exchanger model performed to find
o Bend pressure drop for RTPF correlations appropriate
and serpentine microchannel for these additional
and flat tube heat exchangers modeling capabilities.
o Header pressure drop for
microchannel heat exchangers
with parallel flow
o Carbon dioxide heat transfer
coefficient and pressure drop
correlations

8.1 New Heat Exchanger Models

Prior to the work presented in this thesis, CoilDesigner was capable of

modeling two types of heat exchanger, namely round tube plate fin (RTPF) and

microchannel heat exchangers. As a part of the research detailed in this thesis,

models for wire-and-tube condensers and flat tube heat exchangers have been

developed, thereby doubling the total number of heat exchanger models.

8.1.1 Wire-and-Tube Condenser Model

Wire-and-tube condensers are often used in domestic refrigerators. This type

of heat exchanger employs either natural convection or forced convection heat

transfer on the air side. A literature search has been performed to find correlations to

calculate the air-side heat transfer coefficient, and models for both types of

convection have been incorporated into CoilDesigner.

80
Natural Convection Wire-and-Tube Condenser Model

The natural convection model for wire-and-tube condensers includes the heat

transfer due to natural convection as well as the heat transfer due to radiation.

Correlations by Tanda and Tagliafico (1997) have been implemented to calculate

convective and radiative heat transfer coefficients on the air side. An iterative

method suggested by Bansal and Chin (2003) has been implemented to calculate the

heat load.

Forced Convection Wire-and-Tube Condenser Model

The forced convection model includes the capability of modeling two

different airflow configurations—airflow perpendicular to the tubes and to the wires

as well as airflow perpendicular to the tubes and parallel to the wires. Correlations

developed by Hoke et al. (1997) and by Lee et al. (2001) have been included to

calculate the air-side heat transfer coefficient. Just as in the round tube plate fin heat

exchanger model, the ε-NTU method is used to calculate the heat load in the forced

convection model.

8.1.2 Flat Tube Heat Exchanger Model

The ability to model flat tube heat exchangers has also been incorporated into

CoilDesigner. Flat tube heat exchangers are often used for automotive applications

such as radiators and charge air coolers. The new solver allows for the following

different geometric and fluid flow characteristics: serpentine and parallel fluid flow,

81
plate fins and corrugated fins, and inline and staggered tube configurations. Again,

the ε-NTU is used to calculate the heat load. However, new air-side correlations for

heat transfer and pressure drop were researched and implemented.

8.2 Additional Fluid Modeling Capabilities

8.2.1 Void Fraction Models and Charge Calculation

Accurate void fraction models are necessary to provide accurate charge

calculations for a heat exchanger. Therefore, a comprehensive literature search has

been performed for void fraction models. The models can be grouped into four

different categories—homogeneous models, slip-ratio-correlated models, Lockhart-

Martinelli parameter correlated models, and mass-flux-dependent models. A rather

large number of void fraction models has been included in CoilDesigner because it is

difficult to obtain experimental charge data with which to compare predicted results.

8.2.2 Modeling of Oil Effects and Oil Retention

Oil is an integral part of vapor compression systems and it can have a

significant impact on the heat transfer and pressure drop in heat exchangers. The

ability to model the effects of oil has been included in CoilDesigner. In order to

model refrigerant-oil mixtures accurately, methods have been included to calculate

the properties of refrigerant-oil mixtures as a function of the oil mass concentration.

Refrigerant-oil mixtures behave like zeotropic mixtures because there is a

temperature glide during the evaporation and condensation processes. Thus, the heat

transfer with the air results in a latent heat load component and a sensible heat load

82
component in the refrigerant-oil mixture. A model introduced by Thome (1995) that

accounts for the temperature glide is used to model the change in enthalpy of the

refrigerant-oil mixture. Heat transfer coefficient and pressure drop correlations have

also been included that account for the presence of oil.

8.3 Validation and Optimization Studies

Validation studies of the microchannel heat exchanger model and the wire-

and-tube condenser model have been performed. A study has also been performed to

determine the predictive capabilities of the model accounting for the effects of oil on

heat exchanger performance. Finally, CoilDesigner has been used to perform

optimization studies to design a more efficient and cost-effective wire-and-tube

condenser. The results from these validation and optimization studies are

summarized below.

8.3.1 Validation of Microchannel Heat Exchanger Model

A validation study of the microchannel heat exchanger has been performed.

Experimental results from 35 tests on 8 different heat exchangers were compared with

predicted results. The predictions were quite encouraging as 80% of the heat load

data points were predicted within 2.25%, and the average absolute error for all of the

heat load data points was 1.6%. For the refrigerant pressure drop, 54.3% of the data

points were predicted within 25%, and the average absolute error for all data points

was 25.9%.

83
8.3.2 Validation of Wire-and-Tube Condenser Model

A validation of the forced convection wire-and-tube condenser model has also

been performed. Experimental data was collected from 3 different condensers—one

condenser with airflow perpendicular to the tubes and parallel to the wires and two

condensers with airflow perpendicular to both the tubes and the wires. During the

validation study, a comparison was made between the air-side heat transfer

coefficient correlations developed by Hoke et al. (1997) and Lee et al. (2001). The

Lee et al. correlation was found to predict the heat transfer coefficient more

accurately, so it was used for the validation study. The heat load data points of the

wire-and-tube condensers were predicted within 2.0%, and the average error was only

1.0%. All of the pressure drop data points were predicted within 20% with an

average absolute error of 7.0%.

8.3.3 Validation of Oil Retention Model

Experimental data obtained by Cremaschi (2004) was used to perform a

validation of the oil retention model that was implemented in CoilDesigner. Oil

retention was modeled in an evaporator and a condenser for R-22/MO, R-410A/MO,

R-410A/ POE, R-134a/POE, and R-134a/PAG mixtures. For the evaporator, the

average absolute error of all of the oil retention predictions was about 21%, and 72%

of the data points were predicted within ±29%. Meanwhile, for the condenser, the

average absolute error for all of the oil retention data points was 23%, and about 70%

of the data points were predicted within ±30%.

84
8.3.4 Optimization of Wire-and-Tube Condenser

Two optimization studies were performed for a wire-and-tube condenser that

is currently in use in domestic refrigerators. In the optimization studies, condensers

were modeled using the entire design space specified by the manufacturer in order to

find the best possible condensers given certain geometric constraints. In the first

optimization study, condensers with the same tube circuitry as the original heat

exchanger were modeled, but the wire diameter, the number of wires, the tube

diameter, and the distances between the tubes were varied. Of the condensers

modeled, 147 were found to provide a heat load greater than or equal to the original

condenser with a cost less than or equal to the original condenser. One condenser in

particular was modeled to provide a heat load about 6.7% greater than the original

condenser with a cost reduction of about 15%

In the second optimization study, condensers with larger frontal face areas

were modeled in order to enhance the airflow through the condensers. The number of

tube columns in the direction of the airflow was reduced and the number of tube rows

in the vertical direction was increased. Of the condensers modeled in this study, 219

were found to provide a higher heat load at a lesser cost compared to the original

condenser. One condenser in particular was modeled to have a heat load slightly

greater than the original condenser with a cost 14.5% less than the original condenser.

85
Chapter 9 Future Work

• Implement two-phase pressure drop and heat transfer coefficient

correlations based on flow pattern maps

• Research and implement refrigerant-side heat transfer coefficient and

pressure drop correlations developed specifically for flat tubes

• Obtain experimental data and perform validation studies for the natural

convection wire-and-tube condenser model and for the flat tube model

• Obtain experimental charge data and compare the prediction capabilities

of the multiple void fraction models that have been included in

CoilDesigner

86
 Appendix

A.1 Air-Side Heat Transfer Coefficient Correlations for Flat Tubes

The air-side heat transfer coefficient correlations that follow are in the form of

the Colburn factor, j. Once the Colburn factor has been calculated, the air-side heat

transfer coefficient can be calculated according to the following equation:

j ⋅ G ⋅ c p ,air
h= 2/3
(A.1)
Prair

Kays and London (1984)

Heun and Dunn (1996) used data from Kays and London (1984) to develop a

correlation to calculate the Colburn factor for flat tube heat exchangers with

corrugated fins:

−0.401
j = 0.0538 ⋅ Re Dh (A.2)

where ReDh is the Reynolds number based on the hydraulic diameter on the air side:

G ⋅ Dh
Re Dh = (A.3)
µ air

The air-side hydraulic diameter is calculated according to the following equation:

4 Amin L flow
Dh = (A.4)
Atotal

where Amin is the minimum free flow area for the air, Lflow is the heat exchanger depth

in the airflow direction, and Atotal is the total surface area of the tubes and fins.

87
Chang and Wang (1997)

Chang and Wang (1997) developed an empirical correlation to calculate the

Colburn factor for corrugated louvered fins for flat tube heat exchangers:

−0.14 −0.29 −0.23 0.68 −0.28 −0.05

⎛L ⎞
0.27
⎛ Fp ⎞ ⎛ Fl ⎞ ⎛ Tw,out ⎞ ⎛ Ll ⎞ ⎛ St ⎞ ⎛ Ft ⎞
j = Re − 0.49
⋅⎜ θ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ (A.5)
Lp
⎝ 90 ⎠ ⎜L ⎟ ⎜L ⎟ ⎜ L ⎟ ⎜L ⎟ ⎜L ⎟ ⎜L ⎟
⎝ p ⎠ ⎝ p ⎠ ⎝ p ⎠ ⎝ p ⎠ ⎝ p ⎠ ⎝ p ⎠

A.2 Air-Side Pressure Drop Correlations for Flat Tubes

The air-side pressure drop correlations that follow are in the form of the

Fanning friction factor, f. Once the friction factor has been calculated, the air-side

pressure drop can be calculated according to the following equation:

Atotal G 2
∆P = f ⋅ ⋅ (A.6)
Amin 2 ρ air

Kays and London (1984)

Heun and Dunn (1996) also used data from Kays and London (1984) to

develop a correlation to calculate the friction factor for flat tube heat exchangers with

corrugated fins:

−0.429
f = 1.298 ⋅ Re Dh (A.7)

where ReDh is the Reynolds number based on the hydraulic diameter of the air side,

given by Eq. A.3.

Chang et al. (2000)

f = f1 ⋅ f 2 ⋅ f 3 (A.8)

88
where, if the Reynolds number based on the louver pitch is less than 150, the

quantities f1, f2, and f3 are calculated according to the following equations:

3.04
⎡ ⎛ F p ⎞⎤
⋅ ⎢ln⎜1 + ⎟⎥
− 0.805⋅ F p / Fl
f 1 = 14.39 ⋅ Re (A.9)
Lp ⎜
⎢⎣ ⎝ L ⎟
p ⎠⎥⎦
−1.435
⎡ ⎛⎛ ⎞
0.48
⎞⎤ ⎛D ⎞
− 3.01
F
f 2 = ⎢ln⎜⎜ ⎜ t
⎢ ⎜⎝ F p
⎟
⎟
+ 0.9 ⎟⎟⎥
⎥
⋅⎜ h
⎜L
⎟
⎟
[
⋅ ln (0.5 ⋅ Re Lp ) ]
−3.01
(A.10)
⎣ ⎝ ⎠ ⎠⎦ ⎝ p ⎠

−0.308 −0.308
⎛ Fp ⎞ ⎛F ⎞ ⎛ S ⎞ 0.35
f 3 = ⎜⎜ ⎟⎟ ⋅ ⎜⎜ d ⎟⎟ ⋅ exp⎜⎜ − 0.1167 ⋅ t ⎟ ⋅ Lθ
⎟ (A.11)
⎝ Ll ⎠ ⎝ Ll ⎠ ⎝ Th ,out ⎠

On the other hand, if the Reynolds number is between 150 and 5,000, then f1, f2, and

f3 are calculated according to the following equations:

−0.527
⎡ ⎛⎛ ⎞
0 .5
⎞⎤
F
⋅ ⎢ln⎜⎜ ⎜ t + 0.9 ⎟⎟⎥
0.6049 −1.064 / Lθ0.2
f 1 = 4.97 ⋅ Re Lp ⎟ (A.12)
⎢ ⎜⎝ F p ⎟
⎠ ⎥
⎣ ⎝ ⎠⎦

S
− 2.966 − 0.7931⋅ t
⎡D ⎤ ⎛ Fp ⎞
f 2 = ⎢ h ⋅ ln (0.3 ⋅ Re Lp )⎥
H
⋅ ⎜⎜ ⎟⎟ (A.13)
⎢⎣ L p ⎥⎦ ⎝ Ll ⎠

-3.553
⎛ S ⎞
-0.0446 ⎡ ⎛ ⎛ Lp ⎞
1 .4
⎞⎤
f 3 = ⎜⎜ t ⎟
⎟ ⋅ ⎢ln⎜⎜1.2 + ⎜ ⎟ ⎟⎥
⎟⎥ ⋅ Lθ−0.477 (A.14)
⎢ ⎜F ⎟
⎝ Th ,out ⎠ ⎣ ⎝ ⎝ p ⎠ ⎠⎦

A.3 Refrigerant-Side Heat Transfer Coefficient Correlations

Cavallini and Zecchin (1974)

0.33
⎛ µ liq ⋅ c p ,liq ⎞ k liq
h = 0.05 ⋅ Re ⋅ ⎜ 0.8 ⎟ ⋅ (A.15)
⎜ k eq ⎟ D
⎝ liq ⎠

89
where

0.5
µ vap ⎛ ρ liq ⎞
Reeq = Reliq + Revap ⋅ ⋅⎜ ⎟ (A.16)
µ liq ⎜ρ ⎟
⎝ vap ⎠

where the liquid and vapor Reynolds numbers are defined as follows:

4 ⋅ m& ⋅ (1 − x )
Reliq = (A.17)
π ⋅ D ⋅ µ liq

4 ⋅ m& ⋅ x
Revap = (A.18)
π ⋅ D ⋅ µ vap

A.4 Void Fraction Models

The void fraction models implemented in CoilDesigner are included in this

section. If more than one reference is included for a model, then the original

reference could not be found, and so the model was taken from another reference.

A.4.1 Void Fraction Models for Round Tubes

Homogenous Void Fraction Model

As discussed in Chapter 5, the homogeneous void fraction is calculated

according to the following equation:

1
α= (A.19)
⎛ 1 − x ⎞ ρ vap
1+ ⎜ ⎟
⎝ x ⎠ ρ liq

Slip-Ratio-Correlated Void Fraction Models

90
 As also discussed in Chapter 5, the slip-ratio-correlated models use the

following equation to calculate the void fraction:

1
α= (A.20)
⎛ 1 − x ⎞ ρ vap
1+ ⎜ ⎟ ⋅S
⎝ x ⎠ ρ liq

All of the slip-ratio-correlated void fraction models were developed for annular flow.

Thom (1964)

Thom (1964) developed slip ratios for steam-water flow at different pressures

based on experimental data. Ahrens (1983) generalized the slip ratios to use with

other fluids through the use of the following property index:

0.2
⎛ µ liq ⎞ ρ vap
P.I . = ⎜ ⎟ ⋅ (A.21)
⎜µ ⎟ ρ liq
⎝ vap ⎠

Table A-1. Slip ratios S based on property index P.I. generalized from Thom’s steam-water data

(1964) by Ahrens (1983)

P.I. 0.00116 0.0154 0.0375 0.0878 0.187 0.446 1.0

S 6.45 2.48 1.92 1.57 1.35 1.15 1.00

In order to use the Thom slip ratio in CoilDesigner, a curve-fit equation was

developed based on the data in Table A-1. The curve-fit equation was generated

using the software tool TableCurve 2D® (SYSTAT, 2002) and has the following

form:

c4 c5
S = c1 + c 2 ⋅ P.I . + c3 ⋅ P.I .3 + + (A.22)
P.I . P.I .2

where the coefficients c1 through c2 are shown in Table A-2.

91
 Table A-2. Coefficients for use with Eq. A.22, the curve-fit equation developed to calculate the

slip ratio for the Thom void fraction model

c1 0.938959893346381
c2 -0.178509024795173
c3 0.0479370689134024
c4 0.191578314482982
c5 -1.52972456792532E-07

Zivi (1964)

Zivi (1964) developed an analytical model for the slip ratio based on the

assumption of minimum entropy production under the ideal conditions of zero wall

friction and zero entrainment. The slip ratio is given by the following equation:

−1 / 3
⎛ ρ vap ⎞
S =⎜ ⎟ (A.23)
⎜ρ ⎟
⎝ liq ⎠

Based on experimental data, Zivi found that the actual void fraction is bracketed

between the predictions using this slip ratio and using the homogeneous void fraction

model.

Smith (1969; Tandon et al., 1985)

1/ 2
⎡ ρ liq ⎛1− x ⎞⎤
⎢ + K⎜ ⎟⎥
⎢ ρ vap ⎝ x ⎠⎥
S = K + (1 − K ) ⋅ (A.24)
⎢ ⎛1− x ⎞ ⎥
⎢ 1+ K⎜ ⎟ ⎥
⎣⎢ ⎝ x ⎠ ⎦⎥

Smith found good agreement with experimental data by setting K = 0.4.

Rigot (1973; Rice, 1987)

92
 Rigot (1973) made the assumption of a constant slip ratio:

S=2 (A.25)

Lockhart-Martinelli Parameter Correlated Void Fraction Models

In the following void fraction models, semi-empirical correlations have been

developed to describe the void fraction as a function of the Lockhart-Martinelli

parameter:

0.2
⎛1− x ⎞
0.9
⎛ µ liq ⎞ ρ vap
X tt = ⎜ ⎟ ⎜ ⎟ ⋅ (A.26)
⎝ x ⎠ ⎜µ ⎟ ρ liq
⎝ vap ⎠

Lockhart-Martinelli (1949; Rice, 1987)

Lockhart and Martinelli (1949) did not actually develop a void fraction

correlation, but they presented void fraction data as a function of Xtt. Based on their

data, Wallis (1969) and, later, Domanski and Didion (1983) developed the following

equations to calculate the void fraction (Rice, 1987):

α = (1 + X tt0.8 )
−0.378
for Xtt ≤ 10 (A.27)

α = 0.823 − 0.157 ln X tt for Xtt > 10 (A.28)

Baroczy (1966; Rice, 1987; Koyama et al., 2004)

Baroczy (1966) presented void fraction data as a function of both Xtt and the

property index given in Eq. A.21. The liquid void fraction data (i.e., 1-α) Baroczy

measured is presented in the Table A-3:

93
 Table A-3. Liquid void fraction (1-α) data presented by Baroczy (1966)

X tt
0.01 0.04 0.1 0.2 0.5 1 3 5 10 30 100
0.00002 0.0012 0.009 0.068 0.17 0.22 0.30 0.47 0.71
0.0001 0.0015 0.0054 0.030 0.104 0.23 0.29 0.38 0.57 0.79
0.0004 0.0022 0.0072 0.1800 0.066 0.142 0.28 0.35 0.45 0.67 0.85
0.001 0.0018 0.0066 0.0170 0.0345 0.091 0.170 0.32 0.40 0.50 0.72 0.88
PI 0.004 0.0043 0.0165 0.0370 0.0650 0.134 0.222 0.39 0.48 0.58 0.80 0.92
0.01 0.0050 0.0210 0.0475 0.0840 0.165 0.262 0.44 0.53 0.63 0.84 0.94
0.04 0.0056 0.0250 0.0590 0.1050 0.215 0.330 0.53 0.63 0.72 0.90 0.96
0.1 0.0058 0.0268 0.0640 0.1170 0.242 0.380 0.60 0.70 0.78 0.92 0.98
1 0.0060 0.0280 0.0720 0.1400 0.320 0.500 0.75 0.85 0.90 0.94 0.994

Based on the data presented by Baroczy, the following correlation has since been

developed to calculate the void fraction (Koyama et al., 2004):

−1
⎡ −
0.74
⎛ ρ vap ⎞
0.65
⎛ µ liq ⎞
0.13
⎤
⎛ 1 x ⎞
α = ⎢1 + ⎜ ⎟ ⎜⎜ ⎟
⎟
⎜
⎜µ
⎟
⎟
⎥ (A.29)
⎢ ⎝ x ⎠ ⎝ ρ liq ⎠ ⎝ vap ⎠ ⎥
⎣ ⎦

Mass-Flux-Dependent Void Fraction Models

Hughmark (1962)

The Hughmark void fraction model (1962) was developed for the bubble flow

regime in vertical upward flow. However, Hughmark found the correlation to work

well for other flow regimes in horizontal tubes. The void fraction is calculated

according to the following equation:

KH
α= = KH β (A.30)
⎛ 1 − x ⎞ ρ vap
1+ ⎜ ⎟
⎝ x ⎠ ρ liq

where β is simply the homogeneous void fraction given by Eq. A. and KH is a

function of the parameter Z, which is defined as

94
 Reα1 / 6 Fr 1 / 8
Z= (A.31)
yl1 / 4

where Reα is a void-fraction-weighted Reynolds number,

D ⋅G
Reα = (A.32)
µ liq + α ⋅ (µ vap − µ liq )

Fr is the Froude number,

2
v2 1 ⎛⎜ Gx ⎞
⎟
Fr = = (A.33)
gD gD ⎜⎝ βρ vap ⎟
⎠

where g is the acceleration due to gravity and yliq is the liquid volume fraction:

1
y liq = = 1− β (A.34)
⎛ x ⎞ ρ liq
1+ ⎜ ⎟
⎝ 1 − x ⎠ ρ vap

Hughmark provided data for KH as a function of Z, reproduced here in Table A-4.

Table A-4. Hughmark flow parameter KH as a function of Z (1962)

Z KH
1.3 0.185
1.5 0.225
2 0.325
3 0.490
4 0.605
5 0.675
6 0.720
8 0.767
10 0.780
15 0.808
20 0.830
40 0.880
70 0.930
130 0.980

For the purposes of calculating KH in CoilDesigner, Hughmark’s data was used to

develop a curve-fit equation. The curve-fit equation was developed using TableCurve

2D® (SYSTAT, 2002) and has the following form:

95
 c1 + c 2 ⋅ Z + c3 ⋅ Z 2 + c 4 ⋅ Z 3 + c5 ⋅ Z 4
KH = (A.35)
1 + c 6 ⋅ Z + c 7 ⋅ Z 2 + c8 ⋅ Z 3 + c 9 ⋅ Z 4

Table A-5. Coefficients for use with Eq. A.35, the curve-fit equation developed to calculate the

Hughmark flow parameter KH as a function of Z

c1 -0.102620597649546
c2 0.221766018541533
c3 0.00426608255825076
c4 0.00716611986632982
c5 9.0414594961056E-05
c6 0.0532071133347019
c7 0.0267848812904344
c8 0.00852408911446845
c9 8.1662804078423E-05

Because the Reynolds number is a function of the void fraction, α, and the

void fraction is, in turn, a function of the Reynolds number, the void fraction equation

is an implicit function. Therefore, the set of equations defined by Eq. A.30 through

Eq. A.35 must be solved iteratively. Ridders’ method (Press et al., 1992) with

brackets 0 and 1 was chosen to perform the numerical iteration and solve for the void

fraction. Since the Hughmark void fraction model must be solved using an iterative

process, this model requires a larger amount of computation time than the other void

fraction models and can, therefore, slow down the modeling of a heat exchanger.

Rouhani and Axelsson (1970)

Rouhani and Axelsson (1970) developed their void fraction model to be valid

throughout the different boiling regions, without discontinuities during the transitions

between different boiling regions.

96
 −1

x ⎧⎪ ⎡ x 1 − x ⎤ 1.18 ⎡ σ ⋅ g (ρ liq − ρ vap )⎤ ⎫⎪

1/ 4

α= ⎨C ⎢ + ⎥+ ⋅⎢ ⎥ ⎬ (A.36)
ρ vap ⎪ ⎣⎢ ρ vap ρ liq ⎦⎥ G ⎣⎢ ρ liq2 ⎥⎦ ⎪⎭
⎩

where C = 1.12 for G > 200 kg/(m2 s), C = 1.54 for G < 200 kg/(m2 s), σ is the surface

tension of the liquid, and g is the acceleration due to gravity.

Premoli et al. (1971; Kakaç, 1991)

The void fraction model developed by Premoli et al. (1971) is also known as

the CISE correlation. The Premoli void fraction model is a slip-ratio-correlated

model for annular flow, so the void fraction is calculated by substituting the slip ratio

into Eq. A.20. However, as opposed to the other slip ratio models, the Premoli slip

ratio is dependent on the mass flux and is calculated according to the following

equations:
0.5
⎛ y ⎞
S = 1 + E1 ⋅ ⎜⎜ − yE 2 ⎟⎟ (A.37)
⎝ 1 + yE 2 ⎠

0.22
⎛ ρ liq ⎞
E1 = 1.578 ⋅ Re − 0.19 ⎜ ⎟ (A.38)
liq ⎜ρ ⎟
⎝ vap ⎠

−0.08
⎛ ρ liq ⎞
E 2 = 0.0273 ⋅ Weliq ⋅ Re − 0.51 ⎜ ⎟ (A.39)
liq ⎜ρ ⎟
⎝ vap ⎠

β
y= (A.40)
1− β

where, again, β is the homogeneous void fraction, given in Eq. A.. In this case, the

liquid Reynolds and Weber numbers are calculated according to the following

equations:

97
 GD
Reliq = (A.41)
µ liq

G2D
Weliq = (A.42)
σρ liq

where σ is the surface tension of the liquid.

Tandon, Varma, and Gupta (1985)

The Tandon et al. void fraction model (1985) is an analytical model for two-

phase annular flow. The model accounts for the effect of wall friction through the use

of the Lockhart-Martinelli parameter. The liquid Reynolds number is used and is

calculated according to Eq. A.41 above. The void fraction is calculated according to

the following equation for 50 < Reliq < 1125:

⋅ [F ( X tt )] + 0.9293 ⋅ Reliq ⋅ [F ( X tt )]
−1 −2
α = 1 − 1.928 ⋅ Reliq
- 0 .315 - 0 .63
(A.43)

and according to the following equation for Reliq > 1125:

⋅ [F ( X tt )] + 0.0361 ⋅ Re liq ⋅ [F ( X tt )]
−1 −2
α = 1 − 0.38 ⋅ Re liq
- 0 .088 - 0 .176
(A.44)

where F(Xtt) is defined by the following equation:

F ( X tt ) = 0.15(X tt−1 + 2.85 X tt−0.476 ) (A.45)

Yashar et al. (2001)

The Yashar et al. (2001) void fraction model was developed for stratified and

annular flow. It accounts for gravitational dominated effects through the use of the

Froude rate parameter and for the viscous drag effects through the use the Lockhart-

Martinelli parameter:

98
 −0.321
⎛ 1 ⎞
α = ⎜1 + + X tt ⎟ (A.46)
⎝ Ft ⎠

where Ft is the Froude rate

0.5
⎡ G 2 x3 ⎤
Ft = ⎢ ⎥ (A.47)
⎣⎢ (1 − x )ρ vap gD ⎦⎥
2

Harms, et al. (2003)

The void fraction model developed by Harms et al. (2003) for annular flow is

an empirical model that accounts for the effect of momentum eddy diffusivity

damping at the interface between the liquid and vapor. The authors obtained good

agreement with experimental results at moderate mass fluxes.

2
⎡ 7.242 ⎞ ⎤
−1 / 2
0.5 2 ⎛
α = ⎢1 − 10.06 ⋅ Reliq (1.74 + 0.104 ⋅ Reliq ) ⋅ ⎜⎜1.376 + 1.655 ⎟⎟ ⎥
− 0.875
(A.48)
⎢⎣ ⎝ X tt ⎠ ⎥
⎦

where Xtt is again calculated according to Eq. A.26, and Reliq, in this case, is

calculated according to the following equation:

Reliq =
(1 − x )GD (A.49)
µ liq

A.4.2 Void Fraction Models for Microchannel Tubes

Armand (1946)

Although the Armand void fraction model (1946) was developed for round

tubes, it was found to work well for microchannel tubes with a hydraulic diameter of

about 1 mm by Chung and Kawaji (2004). The correlation simply multiplies a

constant times the homogeneous void fraction.

99
 0.833
α= (A.50)
⎛ 1 − x ⎞ ρ vap
1+ ⎜ ⎟
⎝ x ⎠ ρ liq

Ali, Sadatomi, and Kawaji (1993)

Ali et al. (1993) also found the void fraction in narrow channels to be

predicted well by multiplying a constant times the homogeneous void fraction:

0 .8
α= (A.51)
⎛ 1 − x ⎞ ρ vap
1+ ⎜ ⎟
⎝ x ⎠ ρ liq

Their model was developed using data from experimental investigations with

channels of width 0.778 mm and 1.465 mm between two flat plates.

100
 References

Achaichia, A., and Cowell, T.A., 1988, Heat Transfer and Pressure Drop
Characteristics of Flat Tube and Louvered Plate Fin Surfaces, Experimental Thermal
and Fluid Science, Vol. 1, No., 2, pp. 147-157.

Ali, M.I., Sadatomi, M., and Kawaji, M., 1993, Adiabatic Two-Phase Flow in Narrow
Channels Between Two Flat Plates, The Canadian Journal of Chemical Engineering,
Vol. 71, pp. 657-666.

Ahrens, F.W., 1983, Heat pump modeling, simulation and design, Heat Pump
Fundamentals: Proceedings of the NATO Advanced Study Institute on Heat Pump
Fundamentals, Espinho, Spain (i.e. Portugal), 1980, ed. J. Berghmans, Martinus
Nijhoff Publishers, The Hague, Netherlands.

Armand, A.A., 1946, The resistance during the movement of a two-phase system in
horizontal pipes, Izv. Vses. Teplotekh, Inst. 1, pp. 16-23, AERA-Lib/Trans 828.

ASHRAE, 2000, ASHRAE Handbook—HVAC Systems and Equipment, Atlanta, GA,

pp. 18.4-18.5.

Bansal, P.K., and Chin, T.C., 2003, Modelling and optimisation of wire-and-tube
condenser, International Journal of Refrigeration, Vol. 26, pp. 601-613.

Baroczy, C.J., 1966, A systematic correlation for two-phase pressure drop, Chemical
Engineering Progress Symposium Series, Vol. 62, No. 62, pp. 232-249.

Cavallini, A., and Zecchin, R., A dimensionless correlation for heat transfer in forced
convection condensation, Proceedings of the Sixth International Heat Transfer
Conference, Vol. 3, pp. 309-313.

Chaddock, J.B., and Mathur, A.P., 1980, Heat transfer to oil-refrigerant mixtures
evaporating in tubes, Proceedings of the Second Multi-Phase Flow and Heat Transfer
Symposium-Workshop, Vol. 2, pp. 861-884, Miami Beach, FL.

Chang, Y.J., Hsu, K.C., Lin, Y.T., and Wang, C.C., 2000, A generalized friction
correlation for louver fin geometry, International Journal of Heat and Mass Transfer,
Vol. 43, No. 12, pp. 2237-2243.

Chang, Y.J., and Wang, C.C., 1996, Air side performance of brazed aluminum heat
exchangers, Journal of Enhanced Heat Transfer, Vol. 3, No. 1, pp. 15-28.

Chang, Y.J., and Wang, C.C., 1997, A generalized heat transfer correlation for louver
fin geometry, International Journal of Heat and Mass Transfer, Vol. 40, No. 3, pp.
533-544.

101
Chisholm, D., 1983, Two-Phase Flow in Pipelines and Heat Exchangers, Longman
Inc., New York.

Chung, P.M.Y., and Kawaji, M., 2004, The effect of channel diameter on adiabatic
two-phase flow characteristics in microchannels, International Journal of Multiphase
Flow, Vol. 30, pp. 735-761.

Cremaschi, L., 2004, Experimental and Theoretical Investigation of Oil Retention in

Vapor Compression Systems, Ph.D. Dissertation, Department of Mechanical
Engineering, University of Maryland, College Park, MD.

Cremaschi, L., Schwentker, R.A., and Radermacher, R., In Press, Modeling of Oil
Retention in the Suction Line and Evaporator of Air Conditioning Systems, HVAC&R
Research Journal.

Davenport, D.J., 1983, Correlation for heat transfer and flow friction characteristics
of louvered fin, AIChE Symposium Series, Vol. 79, No. 225, pp. 19-27.

Dillen, E.L., and Webb, R.L., 1994, Rationally Based Heat Transfer and Friction
Correlations for the Louver Fin Geometry, 1994 SAE International Congress &
Exposition, Detroit, MI, Paper 940504.

Dittus, F.W., and Boelter, L.M.K., 1930, Heat Transfer in Automobile Radiators of
the Tubular Type, University of California Publications on Engineering, Vol. 2, p.
443, Berkeley, CA. Reprinted in International Communications in Heat and Mass
Transfer, 1985, Vol. 12, No. 1, pp. 3-22.

Dobson, M.K., and Chato, J.C., 1998, Condensation in Smooth Horizontal Tubes,
Transactions of the ASME, Journal of Heat Transfer, Vol. 120, pp. 193-213.

Fang, X., Bullard, C.W., and Hrnjak, P.S., 2001, Modeling and Analysis of Gas
Coolers, ASHRAE Transactions, Vol. 107, No. 1, Report 4411.

Filippov, L.P., and Novoselova, N.S., 1955, Thermal conductivity of normal liquid
solutions, Vestnik Moskov. Gos. Univ., Ser. Fiz., Vol. 3, pp. 37-40.

Gnielinski, V., 1976, New equations for heat and mass transfer in turbulent pipe and
channel flow, International Chemical Engineering, Vol. 16, No. 2, pp. 359-368.

Harms, T.M., Li, D., Groll, E.A., and Braun, J.E., 2003, A void fraction model for
annular flow in horizontal tubes, International Journal of Heat and Mass Transfer,
Vol. 46, pp. 4051-4057.

Heun, M.K., and Dunn, W.E., 1996, Principles of Refrigerant Circuiting with
Application to Microchannel Condensers: Part 2—The Pressure Drop Effect and the
Cross-Flow Heat Exchanger, ASHRAE Transactions, Vol. 102, No. 2, pp. 382-393.

102
Hilpert, R., 1933, Wärmeabgabe von geheizten Drähten und Rohren im Luftstrom,
Forsch. Gebiete Ingenieurw, Vol. 4, pp. 215-224.

Hoke, J.L., Clausing, A.M., and Swofford, T.D., 1997, An Experimental Investigation
of Convective Heat Transfer From Wire-On-Tube Heat Exchangers, Transactions of
the ASME, Journal of Heat Transfer, Vol. 119, pp. 348-356.

Hughmark, G.A., 1962, Holdup in gas-liquid flow, Chemical Engineering Progress,

Vol. 58, No. 4, pp. 62-65.

Jensen, M.K., and Jackman, D.L., 1984, Predictions of Nucleate Pool Boiling Heat
Transfer Coefficients of Refrigerant-Oil Mixtures, Transactions of the ASME,
Journal of Heat Transfer, Vol. 106, pp. 184-190.

Jiang, H., Aute, V., and Radermacher, R., 2002, A User-Friendly Simulation and
Optimization Tool for Design of Coils, Proceedings of the 9th International
Refrigeration and Air Conditioning Conference at Purdue, Purdue University, West
Lafayette, IN, Paper R5-1.

Jiang, H., 2003, Development of a Simulation and Optimization Tool for Heat
Exchanger Design, Ph.D. Dissertation, Department of Mechanical Engineering,
University of Maryland, College Park, MD.

Jung, D.S., and Radermacher, R., 1989, Prediction of pressure drop during horizontal
annular flow boiling of pure and mixed refrigerants, International Journal of Heat
and Mass Transfer, Vol. 32, No. 12, pp. 2435-2446.

Kakaç, S., 1991, Boilers, Evaporators, and Condensers, John Wiley & Sons, Inc.,
New York.

Kakaç, S., and Liu, H., 2002, Heat Exchangers: Selection Rating and Thermal
Design, 2nd edition, CRC Press, Boca Raton, FL.

Kays, W.M., and Crawford, M.E., 1993, Convective Heat and Mass Transfer, 3rd
edition, McGraw-Hill, Inc., New York.

Kays, W. M., and London, A. L., 1984, Compact Heat Exchangers, 3rd edition,
McGraw-Hill, New York.

Kim, M. H., and Bullard, C. W., 2001, Development of a microchannel evaporator

model for a CO2 air-conditioning system, Energy, Vol. 26, No. 10, pp. 931-948.

Kim, N.H., Youn, B., and Webb, R.L., 1999, Air-Side Heat Transfer and Friction
Correlations for Plain Fin-and-Tube Heat Exchangers With Staggered Tube

103
Arrangements, Transactions of the ASME, Journal of Heat Transfer, Vol. 121, pp.
662-667.

Koyama, S., Lee, J., and Yonemoto, R., 2004, An investigation on void fraction of
vapor-liquid two-phase flow for smooth and microfin tubes with R134a at adiabatic
condition, International Journal of Multiphase Flow, Vol. 30, pp. 291-310.

Lee, T.H., Yun, J.Y., Lee, J.S., Park, J.J., and Lee, K.W., 2001, Determination of
airside heat transfer coefficient on wire-on-tube type heat exchanger, International
Journal of Heat and Mass Transfer, Vol. 44, pp. 1767-1776.

Lemmon, E.W., McLinden, M.O., and Huber, M.L., 2002, REFPROP – Reference
Fluid Thermodynamic and Transport Properties, National Institute of Standards and
Technology, Boulder, CO.

Lockhart, R.W., Martinelli, R.C., 1949, Proposed correlation of data for isothermal
two-phase two-component flow in pipes, Chemical Engineering Progress, Vol. 45,
No. 1, pp. 39-48.

McQuiston, F., and Parker, J., 1994, Heating, Ventilating, and Air Conditioning:
Analysis and Design, 4th edition, John Wiley & Sons, Inc, New York.

Myers, G.E., 1998, Analytical Methods in Conduction Heat Transfer, 2nd edition,
AMCHT Publications, Madison, WI.

Ortiz, T.M., and Groll, E.A., 2002, Validation of a New Model for Predicting the
Performance of Carbon Dioxide as a Refrigerant for Residential Air Conditioners,
Proceedings of the 9th International Refrigeration and Air Conditioning Conference
at Purdue, Purdue University, West Lafayette, IN, Paper R11-2.

Ortiz, T.M., Li, D., and Groll, E.A., 2003, Evaluation of the Performance Potential of
CO2 as a Refrigerant in Air-to-Air Conditioners and Heat Pumps: System Modeling
and Analysis, Air-Conditioning and Refrigeration Technology Institute Report,
ARTI-21CR/610-10030.

Pettersen, J., 2004, Flow vaporization of CO2 in microchannel tubes, Experimental

Thermal and Fluid Science, Vol. 28, pp. 111-121.

Pettersen, J., Rieberer, R., and Leister, A., 2000, Heat Transfer and Pressure Drop
Characteristics of Super-Critical Carbon Dioxide in Microchannel Tubes Under
Cooling, Proceedings of the 4th IIR Gustav Lorentzen Conference on Natural
Working Fluids, IIF-IIR Commission B1, B2, E1, and E2, Purdue University, West
Lafayette, IN.

Pettersen, J., Rieberer, R., and Munkejord, S.T., 2000a, Heat Transfer and Pressure
Drop Characteristics of Evaporating Carbon Dioxide in Microchannel Tubes,

104
Proceedings of the 4th IIR Gustav Lorentzen Conference on Natural Working Fluids,
IIF-IIR Commission B1, B2, E1, and E2, Purdue University, West Lafayette, IN.

Pettersen, J., Rieberer, R., and Munkejord, S.T., 2000b, Heat Transfer and Pressure
Drop for Flow of Supercritical and Subcritical CO2 in Microchannel Tubes, Final
Technical Report, United States Army, Contract Number N-68171-99-M-5674,
Issued by SINTEF Energy Research and Norwegian University of Science and
Technology.

Premoli, A., Francesco, D., and Prina, A., 1971, A dimensionless correlation for
determining the density of two-phase mixtures, La Termotecnica, Vol. 25, pp. 17-26.

Press, W.H., Teukolsky, S.A., Vetterling, W.T., and Flannery, B.P., 1992, Numerical
Recipes in C: The Art of Scientific Computing, 2nd edition, Cambridge University
Press, New York, online at http://www.library.cornell.edu/nr/cbookcpdf.html.

Quadir, G.A., Krishnan, G.M., and Seetharamu, K.N., 2002, Modeling of wire-on-
tube heat exchangers using finite element method, Finite Elements in Analysis and
Design, Vol. 38, pp. 417-434.

Rice, C.K., 1987, The Effect of Void Fraction Correlation and Heat Flux Assumption
on Refrigerant Charge Inventory Predictions, ASHRAE Transactions, Vol. 93, No. 1,
Report 3035.

Rigot, G., 1973, Fluid capacity of an evaporator in direct expansion, Chaud-Froid-

Plomberie, No. 328, pp. 133-144 (English translation, ORNL-tr-5217, Oak Ridge
National Laboratory, Oak Ridge, TN).

Rouhani, S.Z., and Axelsson, E., 1970, Calculation of Void Volume Fraction in the
Subcooled and Quality Boiling Regions, International Journal of Heat and Mass
Transfer, Vol. 13, pp. 383-393.

Rugh, J.P., Pearson, J.T., and Ramadhyani, S., 1992, A study of a very compact heat
exchanger used for passenger compartment heating in automobiles, Compact Heat
Exchangers for Power and Process Industries, ASME Symposium Series, ASME
HTD Vol. 201, pp. 15-24.

Sahnoun, A., and Webb, R.L., 1992, Prediction of Heat Transfer and Friction for the
Louver Fin Geometry, Transactions of the ASME, Journal of Heat Transfer, Vol.
114, pp. 893-900.

Schwentker, R.A., Aute, V.C., and Radermacher, R., 2005, Simulation and Design
Tool for Wire-and-Tube Condensers, IIR Conferences: Thermophysical Properties
and Transfer Processes of Refrigerants, Vicenza, Italy.

105
Schwentker, R.A., Aute, V.C., Radermacher, R., and Mercer, K.B., 2005, Simulation
and Design Tool for Microchannel Heat Exchangers, Fifth International Conference
on Enhanced, Compact and Ultra-Compact Heat Exchangers: Science, Engineering
and Technology, Whistler, British Columbia, Canada.

Shah, M.M., 1979, A general correlation for heat transfer during film condensation
inside pipes, International Journal of Heat and Mass Transfer, Vol. 22, No. 4, pp.
547-556.

Shah, R.K., and Sekuliç, D.P., 2003, Fundamentals of Heat Exchanger Design, John
Wiley & Sons, Inc., Hoboken, NJ.

Shen, B., and Groll, E.A., 2003, Critical Literature Review of Lubricant Influence on
Refrigerant Heat Transfer and Pressure Drop, Air-Conditioning and Refrigeration
Institute Report, ARTI-21CR/611-20080.

Smith, S.L., 1969, Void fractions in two-phase flow: a correlation based upon an
equal velocity head model, Proceedings of the Institution of Mechanical Engineers,
Vol. 184, Pt. 1, No. 36, pp. 647-664.

Sunden, B., and Svantesson, J., Correlation of j- and f-factors for multilouvered heat
transfer surfaces, Proceedings of the Third UK National Conference on Heat
Transfer, pp. 805-811.

SYSTAT Software Inc., 2002, TableCurve 2D® Version 5.01, Port Richmond, CA.

Tagliafico, L., and Tanda, G., 1997, Radiation and natural convection heat transfer
from wire-and-tube heat exchangers in refrigeration appliances, International Journal
of Refrigeration, Vol. 20, No. 7, pp. 461-469.

Tanda, G., and Tagliafico, L., 1997, Free Convection Heat Transfer From Wire-and-
Tube Heat Exchangers, Transactions of the ASME, Journal of Heat Transfer, Vol.
119, pp. 370-372.

Tandon, T.N., Varma, H.K., and Gupta, C.P., 1985, A void fraction model for annular
two-phase flow, International Journal of Heat and Mass Transfer, Vol. 28, No. 1, pp.
191-198.

Thom, J.R.S., 1964, Prediction of Pressure Drop During Forced Circulation Boiling
of Water, International Journal of Heat and Mass Transfer, Vol. 7, pp. 709-724.

Thome, J.R., 1995, Comprehensive Thermodynamic Approach to Modeling

Refrigerant-Lubricating Oil Mixtures, HVAC&R Research, Vol. 1, No. 2, pp. 110-
126.

106
Thome, J.R., 2003, On Recent Advances in Modeling of Two-Phase Flow and Heat
Transfer, Heat Transfer Engineering, Vol. 24, No. 6, pp. 46-59.

Wang, C.C., Lee, W.S., and Sheu, W.J., 2001, A comparative study of compact
enhanced fin-and-tube heat exchangers, International Journal of Heat and Mass
Transfer, Vol. 44, pp. 3565-3573.

Yashar, D.A., Wilson, M.J., Kopke, H.R., Graham, D.M., Chato, J.C., and Newell,
T.A., 2001, An Investigation of Refrigerant Void Fraction in Horizontal, Microfin
Tubes, HVAC&R Research, Vol. 7, No. 1, pp. 67-82.

Yin, J.M., Bullard, C.W., and Hrnjak, P.S., 2001, R-744 gas cooler model
development and validation, International Journal of Refrigeration, Vol. 24 No. 7,
pp. 692-701.

Yokozeki, M.A., 1994, Solubility and viscosity of refrigerant-oil mixtures,

Proceedings of the 1994 International Refrigeration Conference at Purdue, Purdue
University, West Lafayette, IN, pp. 335-340.

Zivi, S.M., 1964, Estimation of Steady-State Steam Void-Fraction by Means of the

Principle of Minimum Entropy Production, Transactions of the ASME, Journal of
Heat Transfer, Vol. 86, pp. 247-252.

Žukauskas, A., 1972, Heat Transfer from Tubes in Crossflow, in Advances in Heat
Transfer, Vol. 8, eds. J.P. Hartnett and T.F. Irvine, Jr., Academic Press, New York,
pp.93-160.

107