EPA - 600/R-96-12 9
November 1996
A Transient and Steady State Study of Pure and Mixed Refrigerants
in a Residential Heat Pump
by
John Judge
Yunho Hwang
Reinhard Radermacher
Center for Environmental Energy Engineering
University of Maryland
College Park, Maryland 20742-3035
EPA Cooperative Agreement CR 822356
EPA Project Officer:
Robert V. Hendriks
Air Pollution Prevention and Control Division
National Risk Management Research Laboratory
Research Triangle Park, North Carolina 27711
Prepared for the
U.S. Environmental Protection Agency
Office of Research and Development
Washington, D.C. 20460
-------
FOREWORD
The U. S. Environmental Protection Agency is charged by Congress with pro-
tecting the Nation's land, air, and water resources. Under a mandate of national
environmental laws, the Agency strives to formulate and implement actions lead-
ing to a compatible balance between human activities and the ability of natural
systems to support and nurture life. To meet this mandate, EPA's research
program is providing data and technical support for solving environmental pro-
blems today and building a science knowledge base necessary to manage our eco-
logical resources wisely, understand how pollutants affect our health, and pre-
vent or reduce environmental risks in the future.
The National Risk Management Research Laboratory is the Agency's center for
investigation of technological and management approaches for reducing risks
from threats to human health and the environment. The focus of the Laboratory's
research program is on methods for the prevention and control of pollution to air,
land, water, and subsurface resources; protection of water quality in public water
systems; remediation of contaminated sites and groundwater; and prevention and
control of indoor air pollution. The goal of this research effort is to catalyze
development and implementation of innovative, cost-effective environmental
technologies; develop scientific and engineering information needed by EPA to
support regulatory and policy decisions; and provide technical support and infor-
mation transfer to ensure effective implementation of environmental regulations
and strategies.
This publication has been produced as part of the Laboratory's strategic long-
term research plan. It is published and made available by EPA's Office of Re-
search and Development to assist the user community and to link researchers
with their clients.
E. Timothy Oppelt, Director
National Risk Management Research Laboratory
EPA REVIEW NOTICE
This report has been peer and administratively reviewed by the U.S. Environmental
Protection Agency, and approved for publication. Mention of trade names or
commercial products does not constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Information
Service, Springfield, Virginia 22161.
ia
-------
Table of Contents
Section Page
List of Tables vi
List of Figures ix
List of Abbreviations xv
Abstract xx
Major Findings xxii
1.0 Introduction 1
1.1 Overview 1
1.2 Motivation 1
1.3 Objectives 3
1.4 References 4
2.0 The Working Fluids Studied . 6
2.1 Overview 6
2.2 R-22 10
2.3 R-32/134a 11
2.4 R-407C 12
2.5 References 13
3.0 Determination of Thermophysical Properties 21
3.1 Introduction 21
3.2 Refrigerant Properties 22
3.2.1 Refrigerant Thermodynamic Property Relations 22
3.2.2 Refrigerant Transport Property Relations 26
3.3 Air Properties 29
3.3.1 Air Thermodynamic Property Relations 29
3.3.2 Air Transport Property Relations 29
3.4 References 29
4.0 The Vapor Compression Systems Studied 36
4.1 Introduction 36
4.2 AC/HP Instrumentation 39
4.3 AC/HP 1 41
4.4 AC/HP 2 42
ii
-------
Table of Contents
Section Page
5,0 The Test Facility 48
5.1 Introduction 48
5.2 Outdoor Chamber 49
5.2.1 Outdoor Chamber Environment Control 49
5.2.2 Outdoor Chamber Instrumentation 50
5.3 Indoor Loop .50
5.3.1 Indoor Loop Environment Control 50
5.3.2 Indoor Loop Instrumentation 51
5.4 References 53
6.0 Instrumentation 56
6.1 Introduction .56
6.2 Gas Chromatograph 56
6.3 References 58
7.0 Experimental Procedure 61
7.1 Introduction 61
7.2 Previous Work 61
7.3 Test Description 64
7.4 Test Procedure 66
7.5 Charge Optimization 67
7.6 Concentration Measurement 68
7.7 References 71
8.0 Data Collection 73
8.1 Introduction 73
8.2 Data Acquisition 73
8.3 Data Reduction 76
8.4 References 80
9.0 Simulation of AC/HP Components 81
9.1 Introduction 81
9.2 Definitions 81
9.3 Compressor Model 83
9.3.1 Previous Compressor Models 84
9.3.2 Description of Compressor Model I 89
9.3.3 Model ! Verification 102
iii
-------
Table of Contents
Section Page
9.3.4 Description of Compressor Model II 104
9.3.5 Model II Verification 105
9.3.6 Compressor References 105
9.4 Heat Exchanger Model 116
9.4.1 Previous Heat Exchanger Models 116
9.4.2 Description of Heat Exchanger Model 118
9.4.3 Heat Exchanger Model Verification 126
9.4.4 The Effect of Heat Exchanger Geometry 132
9.4.5 Heat Exchanger References 133
9.5 Expansion Device Model 146
9.5.1 Previous Expansion Device Models 147
9.5.2 Description of Expansion Device Model 151
9.5.3 Expansion Device References 154
10.0 AC/HP Simulation 156
10.1 Previous AC/HP Simulations 156
10.2 Description of the AC/HP Simulation 161
10.3 Numerical Methods 162
10.3.1 Solving the Non-Linear Equations 162
10.3.2 The First Guess 166
10.3.3 Determining the Time Step Size 167
10.4 AC/HP Simulation References 168
11.0 Experimental Results and Discussion 170
11.1 Introduction 170
11.2 Steady State Test Results 171
11.2.1 Refrigerant Charge Optimization 171
11.2.2 Typical Steady State Test Results 174
11.2.3 The Effect of the Expansion Device on Steady State
Performance 176
11.2.4 Steady State Comparison of Refrigerants 178
11.2.5 The Effect of a VLHX on Steady State Performance . 182
11.3 Cyclic Test Results 183
11.3.1 The Effect of On-Time on Cyclic Performance 186
11.3.2 The Effect of the Expansion Device on Cyclic
Performance 187
11.3.3 Cyclic Comparison of Refrigerants 188
iv
-------
Table of Contents
Section Page
11.3.4 The Effect of a VLHX on Cyclic Performance 190
11.3.5 The Effect of the Mode of Operation on Cyclic
Performance 191
11.4 Seasonal Performance Comparisons 192
11.4.1 Expansion Device Impact 192
11.4.2 Comparison between R-22 and R-407C 193
11.4.3 The Effect of a VLHX on Seasonal Performance .... 195
11.5 Concentration Measurements 195
11.6 References 199
12.0 Simulation Results and Discussion 222
12.1 Introduction 222
12.2 Selection of the Void Fraction Correlation 222
12.3 Steady State Simulation Test Results 225
12.3.1 Charge Optimization Results 225
12.3.2 Typical Steady State Simulation Results 227
12.3.3 Steady State Comparison with Experimental Data . . , 229
12.3.4 Steady State Comparison of Refrigerants 231
12.3.5 The Influence of Connecting Piping on Steady
State Performance 232
12.3.6 The Influence of Charge on Circulated
Concentration 234
12.4 Transient Simulation Results 234
12.4.1 Step Size Independence 234
12.4.2 Typical Transient Simulation Results 236
12.4.3 Transient Comparison of Refrigerants 240
12.4.4 The Influence of Connecting Piping on Transient
Performance 242
12.4.5 Transient Losses 245
12.5 References 250
13.0 Conclusions 269
14.0 Future Work 275
A.1 Experimental Uncertainty Analysis 276
A.2 Heat Exchanger Model Uncertainty Analysis 281
v
-------
List of Tables
Number Page
2.1 Thermodynamic Properties of Pure Refrigerants 14
2.2 Thermodynamic Properties of Refrigerant Mixtures 15
2.3 Transport Properties of Pure Refrigerants at 25 °C 15
2.4 Transport Properties of Refrigerant Mixtures at 25 °C 16
2.5 Environmental Properties 16
3.1 Maximum Permissible Uncertainty in Transport Properties
to Calculate the COP and Capacity within 1 % 32
4.1 Characteristics of Test Units 43
6.1 Baseline Gas Chromatograph Data 59
6.2 Gas Chromatograph Error of R-32/134a (compared to ICI GC) .... 59
6.3 Gas Chromatograph Error of R-32/134a (compared to weighed
concentration) 59
6.4 Gas Chromatograph Error of R-407C (compared to weighed
concentration) 60
7.1 Test Conditions 72
9.1 Heat Transfer Equations 108
9.2 Typical Input Data. 109
9.3 Compressor Model I Verification (low condensing pressure) 110
9.4 Compressor Model I Verification (medium condensing pressure) ..110
9.5 Compressor Model I Verification (high condensing pressure) 110
vi
-------
List of Tables
Number Page
9.6 Refrigerants and Heat Exchangers Tested 136
9.7 The Effect of Heat Exchanger Geometry on Heat Transferred
(Cooling Mode) 136
9.8 The Effect of Heat Exchanger Geometry on Heat Transferred
(Heating Mode) 137
11.1 Charge Optimization Results 201
11.2 R-22 Steady State Results (TXV) 201
11.3 R-22 Steady State Results (STR) 202
11.4 R-407C Steady State Results (TXV) 202
11.5 B Test Cycle Parameters (TXV) 203
11.6 47S Test Cycle Parameters (TXV) 204
11.7 Steady State Results (STR) 204
11.8 C Test Cycle Parameters (STR) 205
11.9 47S Test Cycle Parameters (STR) 206
11.10 R-22 Steady State Results (TXV & VLHX) 206
11.11 R-407C Steady State Results (TXV & VLHX) 207
11.12 Cyclic Tests 208
11.13 Non-Dimensional Cyclic Performance 209
11.14 Overall Seasonal Performance 210
12.1 Cooling Mode Simulation Results 252
vii
-------
List of Tables
Number Page
12.2 Heating Mode Simulation Results 253
12.3 Typical Mass and Pressure Drop Distribution (R-22/C test) 254
12.4 Typical Mass and Pressure Drop Distribution (R-22/47S test) .... 254
12.5 Comparison of Simulated Results with Experimental Results
(Cooling Mode) 255
12.6 Comparison of Simulated Results with Experimental Results
(Heating Mode) 255
12.7 Evaluation of Time Step Independence 256
viii
-------
List of Figures
Number Page
2.1 Pressure - Temperature Diagram 17
2.2 R-134a Pressure - Enthalpy Diagram 17
2.3 R-22 Pressure - Enthalpy Diagram .18
2.4 R-32 Pressure - Enthalpy Diagram 18
2.5 R-125 Pressure - Enthalpy Diagram 19
2.6 R-32/134a Pressure - Enthalpy Diagram 19
2.7 R-407C Pressure - Enthalpy Diagram 20
3.1 Typical Pressure - Volume Diagram 33
3.2 Saturated Liquid Thermal Conductivity of the Refrigerants Studied . 33
3.3 Saturated Liquid Dynamic Viscosity of the Refrigerants Studied .... 34
3.4 Saturated Vapor Thermal Conductivity of the Refrigerants Studied . 34
3.5 Saturated Vapor Dynamic Viscosity of the Refrigerants Studied .... 35
4.1 AC/HP 1 Circuit Schematic 44
4.2 Refrigerant Sampling Port .44
4.3 Refrigerant Sampling Procedure and Apparatus 45
4.4 Indoor Heat Exchanger for AC/HP 1 45
4.5 Outdoor Heat Exchanger for AC/HP1 46
4.6 Indoor Heat Exchanger for AC/HP2 46
4.7 Outdoor Heat Exchanger of AC/HP2 47
ix
-------
List of Figures
Number Page
5.1 Test Facility .54
5.2 Outdoor Chamber 55
5.3 Indoor Loop 55
9.1 Compressor Schematic 111
9.2 Compressor Energy and Mass Exchange 111
9.3 Compressor Detail 112
9.4 Pressure-Volume Diagram for Compression Cycle 112
9.5 Discharge Valve Mass Flow Rate ..113
9.6 Position of Compressor Valves 113
9.7 Compression Cycle with Isentrope 114
9.8 Compressor Model II Comparison with Experimental Power 114
9.9 Compressor Model II Comparison with Experimental Mass
Flow Rate 115
9.10 Heat Exchanger Nodes . 138
9.11 Concentration Shift of R-407C During Phase Change 138
9.12 Flow Boiling Heat Transfer Correlation Results 139
9.13 Evaporation Heat Transfer Coefficient Versus Length (R-22) 140
9.14 Evaporation Heat Transfer Coefficient Versus Length (R-407C)... 140
9.15 Condensation Heat Transfer Correlation Results 141
x
-------
List of Figures
Number Page
9.16 Condensation Heat Transfer Coefficient Versus Length (R-22) ... 142
9.17 Condensation Heat Transfer Coefficient Versus Length (R-407C) . 142
9.18 Baker's Flow Map (R-22) 143
9.19 The Effect of the Number of Nodes on Capacity and Mass 143
9.20 Refrigerant Temperature Profile Along the Condenser 144
9.21 Refrigerant Temperature Profile Along the Evaporator 144
9.22 Condenser Refrigerant and Air Temperature Profiles for Cross
and Parallel Flow Geometries (R-407C) 145
9.23 Condenser Refrigerant and Air Temperature Profiles for Cross
and Counter Flow Geometries (R-407C) 145
11.1 Charge Optimization Curve (R-22, STR, C test) 211
11.2 Compressor Power Versus Charge (R-22, STR, C test) 211
11.3 Suction and Discharge Pressures Versus Charge
(R-22, STR, C test) 212
11.4 Mass Flow Rate Versus Charge (R-22, STR, C test) 212
11.5 Capacity Versus Charge (R-22, STR, C test) 213
11.6 Superheat Versus Charge (R-22, STR, C test) . 213
11.7 Charge Optimization Curves (R-22, STR, C test)
& (R-22, TXV, C test) 214
11.8 Charge Optimization Curves (R-22, TXV, B test)
& (R-407C, TXV, B test) 214
xi
-------
List of Figures
Number Page
11.9 Charge Optimization Curves (R-22, STR, C test)
& (R-32/134a, STR, C test) 215
11.10 Charge Optimization Curves (R-22. STR, 47S test)
& (R-32/134a, STR, 47S test) 215
11.11 Refrigerant Temperatures Versus Time (R-22, STR, D' test) 216
11.12 Suction and Discharge Pressures Versus Charge
(R-22, STR, D' test) 216
11.13 Compressor Power Versus Time (R-22, STR, D' test) 217
11.14 Air Side Cooling Capacity Versus Time
(R-22, STR, D' test) 217
11.15 Refrigerant Temperature Along the Evaporator (R-22) 218
11.16 Refrigerant Temperature Along the Evaporator (R-407C) 218
11.17 Impact of R-22 and R-407C on Global Warming Versus
Life Time of Equipment 219
11.18 R-407C Concentration Versus Time (D' test, Vapor Line) 219
11.19 R-407C Concentration Versus Time (D1 test, Liquid Line) . 220
11.20 R-32/134a Concentration Versus Time (D" test, Vapor Line) 220
11.21 R-32/134a Concentration Versus Time (47C test, Vapor Line) . . . 221
12.1 Charge Optimization Curves 257
12.2 Compressor Power and Capacity Versus Charge 257
12.3 Suction and Discharge Pressures Versus Charge 258
12.4 Mass Flow Rate Versus Charge 258
xii
-------
List of Figures
Number Page
12.5 Superheat and Subcooling Versus Charge 259
12.6 R-22 Cooling Mode Pressure Enthalpy Diagram 259
12.7 R-22 Heating Mode Pressure Enthalpy Diagram 260
12.8 R-32/134a Cooling Mode Pressure - Enthalpy diagram 260
12.9 R-32/134a Heating Mode Pressure Enthalpy Diagram 261
12.10 Compressor Speed Versus Time 261
12.11 Mass Flow Rate Versus Time 262
12.12 Refrigerant Mass in Each Component Versus Time 262
12.13 Suction and Discharge Pressures Versus Time 263
12.14 Refrigerant Temperatures Entering Each Component
Versus Time 263
12.15 Normalized Compressor Power and Evaporator Capacity
Versus Time 264
12.16 Normalized COP Versus Time 264
12.17 Refrigerant Temperature Entering Each Component Versus
Time (R-32/134a) 265
12.18 Concentration Versus Time . 265
12.19 Suction and Discharge Pressures Versus Time
(System With Piping) 266
12.20 Refrigerant Mass Versus Time (System With Piping) 266
xiii
-------
List of Figures
Number Page
12.21 The Power Required to Overcome the Inertia of the Compressor .. 267
12.22 The Storage Terms of the Condenser . 267
12.23 The Storage Terms of the Evaporator 268
12.24 The Energy of the Storage Terms 268
xiv
-------
List of Abbreviations
a - Void Fraction
nv - Volumetric Efficiency
* " [(Pv^Pair)( P/Pwater)]"^
p - Density (kg/m3)
^ " (Owate/oJKM/Mv.aterJCPwaJPl)2]1''3
a) - Angular Velocity (radians/second)
a - Speed of Sound (m/s)
A - Area (m2)
AC/HP - Air-Conditioner and/or Heat Pump
Al - Alefeld's Number (hfgTevap/CP)
AREP - Alternative Refrigerant Evaluation Program
ARI - American Refrigeration Institute
ASHRAE - American Society of Heating, Refrigerating and Air-
Conditioning Engineers
BM - Bisection Method
Cc - Contraction Coefficient
CD - Cyclic Degradation Coeff. or Flow Degradation Coeff.
CFC - Chlorofluorocarbon
CLF - Cooling Load Factor
COP - Coefficient of Performance
Cp - Constant Pressure Specific Heat
xv
-------
List of Abbreviations
CSD - Camahan-Starling-DeSantis EOS
CSPF - Cooling Seasonal Performance Factor
Cv - Clearance Volume Coefficient
CYC - Cyclic
EOS - Equation of State
Evap - Evaporation
g - Acceleration Due to Gravity (9.81 m/s2)
G - Mass Flux (kg/s m2)
GC - Gas Chromatograph
GWP - Global Warming Potential
h - Enthalpy (kJ/kg), Head Loss (m), or Heat Transfer Coeff,
(W/rrfK)
HFC - Hydroflourocarbon
hfg - Latent Heat
HFL - High Flammability Limit
HLF - Heating Load Factor
HSPF - Heating Seasonal Performance Factor
I - Moment of Inertia (N/m3)
k - Isentropic Exponent
k - Thermal Conductivity
I - Liquid
xvi
-------
List of Abbreviations
LFL
Lower Flammability Limit
LKP
Lee-Kesler-Plocker EOS
m
Mass Flow Rate (kg/s)
M
Mass (kg), Mach Number
n
Polytropic Exponent
N1ST
National Institute of Standards and Technology
NRT
Newton-Raphson Technique
Nu
Nusselt Number (h/k d)
ODP
Ozone Depletion Potential
p
Pressure
Pe
Peclet Number (Re Pr)
Pr
Prandtl Number (Cp-p/k)
PR
Peng-Robinson EOS
q
Quality
G
Heat Transfer (kJ)
R
Universal Gas Constant 8.314 (kj/kmol K)
R-32/134a -
R-32/134a (30.0/70.0 wt.%)
Re
Reynold's Number (V-D/v)
RH
Relative Humidity
-------
List of Abbreviations
Redlich-Kwong-Soave EOS
Entropy
Steady State
Short Tube Restrictor
Temperature
Critical Temperature
Tota! Equivalent Global Warming Impact
Thermostatic Expansion Valve
Velocity
Overall Heat Transfer Coefficient (W/m2 K), or Internal
Energy (kJ/kg)
United Nations Environmental Programme
Vapor
Velocity (m/s)
Clearance Volume of Compressor (m3)
Volumetric heat capacity
Vapor to Liquid Line Heat Exchanger
Swept Volume of Compressor (m3)
Work
-------
List of Abbreviations
Concentration (kg/kg) or Quality
Displacement (m)
Height or Length (m)
xix
-------
Abstract
This study is a theoretical and experimental study of the transient and
steady state performance of a residential air-conditioner / heat pump (AC/HP).
Several different refrigerants and system configurations were studied.
Many points were addressed experimentally. The performance of R-407C
relative to R-22 was evaluated in terms of steady state, cyclic, and seasonal
performance. The combination of the steady state and cyclic performances
showed that R-407C has a 4.3% lower cooling seasonal efficiency than R-22 and
up to a 7.0% lower heating seasonal efficiency than R-22. The seasonal
performance was used to show that R-407C poses a greater global warming
threat than R-22. The performance of a vapor to liquid line heat exchanger was
also evaluated with these fluids. Furthermore, the performance of the AC/HP
was quantified with different expansion devices.
The final facet of the experimental work was to measure the concentration
of R-32/134a (30/70 wt.%) and R-407C as a function of time. The results of
these measurements indicated that the steady state circulated concentration of
the refrigerant mixtures shifted away from the less volatile component relative to
the charged concentration. The concentration shift was attributed to the velocity
difference between the phases.
xx
-------
The theoretical aspect of this work was addressed by the development of
a fully implicit, distributed parameter simulation capable of modeling the transient
and steady state aspects of an AC/HP. For steady state, the simulation solves
the complete continuity, species, energy, and momentum equations, while
transiently only the momentum equation is omitted. This simulation is the first
capable of representing the significant transient and steady state physics of an
AC/HP operating with pure and mixed refrigerants while utilizing minimal
empirical data. The simulation was used to study several different system
configurations transiently and at steady state with both R-22 and R-32/134a
(30/70 wt.%).
xxi
-------
Major Findings
The major findings from this work are derived from either experimental or
model results. The experimental findings are addressed first followed by the
model related findings.
The major experimental findings are outlined below.
• The performance of R-407C relative to R-22 was found in terms of steady
state, cyclic and seasonal performance. The combination of the steady
state and cyclic performances showed that R-407C had a 4.3% lower
cooling seasonal efficiency (CSPF) than R-22 and a 1.5 - 7,0% lower
heating seasonal efficiency (HSPF) than R-22. The seasonal
performance was used to show that R-407C is a greater global warming
threat than R-22 in the systems tested.
• The performance of a vapor to liquid line heat exchanger (VLHX) was also
evaluated with R-22 and R-407C. While the VLHX had no impact on the
steady state performance, it did marginally improve the cyclic
performance of both fluids.
• The cyclic and steady state performance of the heat pump was quantified
with short tube restrictors (STRs) and thermostatic expansion valves
xxii
-------
(TXVs). When the AC/HP utilized the STR it had a significantly lower
steady state and cyclic performance. The cumulative effect of the lower
performance associated with the STR relative to the TXV is a 3.6% lower
CSPF and a 0 - 3.9% lower HSPF. Furthermore, the STR significantly
increased the sensitivity of the system performance to the amount of
refrigerant charge.
The concentration of R-32/134a (30/70 wt.%) and R-407C was measured
as a function of time. The results indicate that the concentration of the
refrigerant mixtures changed with time. However, the steady state
circulated concentration was reached rapidly, within 3 minutes. The
results of these measurements also indicate that the circulated
concentration of the refrigerant mixtures was not equal to the charged
concentration. The circulated concentration of both refrigerant mixtures
shifted away from the less volatile component. Specifically, the steady
state circulated concentration of R-134a was 3.0 wt.% less for R-4Q7C
and 0.5 wt.% less for R-32/134a than the charged concentration. In other
words, the circulated concentration had shifted toward the more volatile
component. The concentration shift was attributed to the velocity
difference between the phases in the heat exchangers.
xxiii
-------
The model related findings were derived from a state of the art simulation
developed for this work. The major findings are summarized below.
• The heat exchanger model, a sub-component of the overall simulation,
was used to demonstrate the theoretical performance of different heat
exchanger geometries. The heat exchanger model indicated that R-407C
is more sensitive to counter and parallel flow geometries than is R-22.
Compared to the cross flow capacity, R-407C gained 0.4 - 3.4% for the
counter flow geometry and lost 1.8 - 8,4% for the parallel flow geometry.
• The overall simulation was used to quantify the shift in the circulated
concentration of R-32/134a (30/70 wt.%). The simulation predicted that
the circulated concentration shifted away from the less volatile component
by 0,8 wt.% when the piping which was used to connect the indoor unit
and the outdoor unit was considered. The extent the circulated
concentration could be shifted was evaluated by examining two cases.
The concentration shift increased to 1.4 wt.% when the connecting piping
was removed. To evaluate the maximum theoretical concentration shift,
the simulation was run without connecting piping and with half of the
optimum charge. The results for this scenario indicate that the circulated
concentration is shifted by 2.3 wt.% towards R-32. The simulation was
used transiently to show that the circulated concentration does change as
xxiv
-------
a function of time, but it approaches its steady state value within 30
seconds. At no time was the flammable concentration of this mixture
approached.
The simulation was used in the steady state limit to quantify the penalty
associated with the connecting piping. The simulation showed that the
connecting piping reduced the COP and capacity by 4.1%. The effect of
the connecting piping was also studied transiently. The simulation results
indicate that, for the case studied here, the addition of connecting piping
reduced the normalized COP by 12.2%.
The major losses associated with running an AC/HP transiently were
quantified with the simulation. Reducing the temperature of the
evaporator itself (i.e. the tubes and fins) and redistributing the refrigerant
required the most energy to achieve steady state. The energy required
to overcome the inertia of the compressor was shown to be relatively
small.
Two shortcomings in the literature were found. The flow through the
expansion device is not well understood. As a result, there is no fluid
independent correlation or model for all of the different flow regimes
encountered by the expansion device. The other weakness in the
-------
literature is that there is significant discrepancy among the existing void
fraction correlations which are used to predict the amount of refrigerant
in the two phase region. As a result, the modeling of charge inventory is
compromised.
xxvi
-------
Chapter 1
Introduction
1.1 Overview
This project investigates the transient and steady state performance of a vapor
compression system operating with pure and mixed refrigerants. This is
accomplished both experimentally and theoretically by examining the
performance of a residential air-conditioner/heat pump (AC/HP) operating with
different refrigerants. The theoretical investigation is accomplished through the
use of a computer simulation which is developed and described in this project.
The experimental investigation is conducted by testing an AC/HP in
environmental chambers designed and built for this purpose. In addition, this
project experimentally and theoretically investigates the influence of various
system configurations.
1.2 Motivation
The steady state, transient, and mixture facets of this research have been
motivated by a wide range of factors, These factors range from environmental
to reliability issues.
1
-------
The steady state performance of a vapor compression system is of
interest since it represents the ideal mode of operation. As such it is an upper
limit for the performance of any vapor compression system.
The transients of a vapor compression system are of interest since an
overwhelming majority of these units obtain capacity and temperature control
through cycling the system on and off. Hence, in order to develop control
equipment it is important to understand the transient aspects of these systems.
System reliability is also affected by transient operation. For example, at start
up a system will typically pump some fraction of liquid into the compressor. If too
much liquid is pumped into the compressor the compressor is likely to fail.
Furthermore, as will be shown in Chapter 11, cycling results in capacities and
COPs which are roughly 75% of their steady state values. Hence, real world
performance is actually significantly lower than that predicted by a steady state
system evaluation. This reduction in performance results in a greater
consumption of energy which has obvious negative environmental
consequences.
The mixture aspect of this work is also motivated by environmental
concerns. In accordance with the Montreal protocol, ozone depleting
refrigerants will be phased out (UNEP, 1991). This includes R-22, which is a
2
-------
medium pressure refrigerant used in heating, cooling and refrigeration
applications. Currently, no pure non-flammable refrigerant has been identified
as a acceptable replacement for R-22. However, several suitable zeotropic
mixtures have been identified. Most notably, the mixture of R-32, R-125, and
R-134a in the 23/25/52 wt% concentration, R-407C, is the refrigerant most likely
to replace R-22 in retrofit applications (Geiger, 1993), (Shiflett, 1992), (Spatz and
Zheng, 1992).
1.3 Objectives
There is significant empirical data available on the steady state performance of
R-22 and its replacements. However, there is relatively little data available on
the transient performance of R-22 and no data available on the transient
performance of any zeotropic mixture, including those slated to replace R-22.
There has been some work on the transient performance of R-22 in terms of
modeling. However, the only modeling work on the transient performance of a
zeotropic mixture was a simulation which treated the refrigerant as a pure
component (Sami and Comeau, 1992).
It is the goal of this project to address these data gaps, as well as others.
Specifically, there are six major objectives of this work. They can be divided into
two classes, experimental and modeling. The objectives are listed below.
3
-------
Experimental Objectives
•Determine the steady state, cyclic and seasonal performance of R-407C
relative to R-22
• Evaluate the effect of vapor to liquid line heat exchange on the steady
state, cyclic and seasonal performance of an AC/HP.
• Determine the effect the different expansion devices have on system
performance.
• Measure the circulated concentration of refrigerant mixtures.
Modeling Objectives
• The development of a detailed simulation capable of accurately
modeling the steady state and transient behavior of a AC/HP using pure
and mixed refrigerants.
• Quantify the affects of different system configurations.
Each of these objectives will be discussed in greater detail in the chapters that
follow.
1.4 References
Geiger, K.A., 1993, "DuPont Suva AC9000 A Developmental Alternative to
HCFC-22 for New Equipment or Service", Presented at The 1993 International
CFC and Halon Alternatives Conference, Washington D.C.
Sami, S.M., Comeau, M.A., 1992, "Development of a Simulation Model for
Prediction Dynamic Behavior of Heat Pump with Nonazeotropic Refrigerant
4
-------
4
Mixtures", International Journal of Energy Research, Vol. 16, pp. 431-444.
Shiflett, M.B., 1992, " HCFC-22 Alternatives for Air Conditioning and Heat
Pumps", Proceedings of The 1992 International CFC and Halon Alternatives
Conference, Washington D.C., pp. 55-64.
Spatz, M.W., Zheng, J., 1992, "Performance of HCFC-22 Alternative
Refrigerants in Air Conditioning Equipment", Proceedings of The 1992
International CFC and Halon Conference, Washington D C., pp. 223-232.
UNEP, 1991, "Report of the Refrigeration, Air Conditioning and Heat Pumps
Technical Options Committee", Montreal Protocol.
5
-------
Chapter 2
The Working Fluids Studied
2.1 Overview
All of the fluids studied in this dissertation are halogenated hydrocarbons. They
were first discovered to be excellent refrigerants by Midgley in 1930 (McLinden
and Didion, 1987). At that time the primary constraints on refrigerant selection
were: toxicity, stability, and operating pressure. Due to heightened
environmental concerns, additional constraints have been imposed. Foremost
among these concerns are the depletion of the ozone and the warming of the
earth due to green house gases. It is the ozone issue which has motivated the
study of refrigerant mixtures. This is because no pure component has been
found to replace R-22, which has a high ozone depletion potential (ODP).
The pertinent thermodynamic, transport, and environmental properties
of each fluid investigated will be discussed. Three working fluids are studied.
They are R-22, R-32/134a (30/70 wt.%), and R-32/125/134a (23/25/52 wt.%).
The motivation for choosing these will be discussed later. Before discussing the
different refrigerants it will be useful to point out the importance and usefulness
of different thermophysical properties and diagrams. The thermophysical
6
-------
properties of interest can be subdivided into either thermodynamic and transport
properties.
In this context the thermodynamic properties of a refrigerant determine
how a refrigerant will perform in an ideal system with no pressure drop or heat
transfer resistance. Some of the more important properties can be seen in Table
2.1 for pure refrigerants and Table 2.2 for mixtures. Alefeld's number promotes
an appreciation for the more significant refrigerant properties (Alefeld, 1987).
Alefeld constructed a special number, Al as defined in Equation 2.1, that is
proportional to the COP of a vapor compression system. Since, this research
compares refrigerants for the same application, the evaporation temperature,
Tavap, can be assumed to be the same for each refrigerant considered.
From Alefeld's number it is clear that a refrigerant with a high latent heat and low
specific heat is desired.
The refrigerant's critical temperature is equally important, for two reasons.
The critical temperature is proportional to a fluid's latent heat and inversely
proportional to a refrigerant's operating pressure. From a performance point of
view a high critical temperature is desirable. However, it is more important that
7
-------
a replacement refrigerant have similar operating pressures as R-22. This is
necessary if the replacement is to be used in the same equipment as R-22, The
overall implication of this is that the critical temperature of a replacement
refrigerant should be relatively close to R-22's, if not slightly higher.
Another important thermodynamic parameter is the temperature glide
occurring during phase change. For pure components there is no temperature
glide associated with a change of phase. However, a zeotropic mixture does
have a temperature glide, by definition. Typically, the temperature glide
increases when the difference between the critical temperature of the
components of a mixture increases. The temperature glide affects system
performance through the heat exchanger effectiveness. A temperature glide can
improve performance in a counterflow heat exchanger and degrade performance
in a parallel flow heat exchanger ( Pannock, 1989). This is illustrated in section
9.4.4. Furthermore, the greater the temperature glide the more likely
fractionation may occur.
There are two diagrams which are particularly useful in comparing
different refrigerants and evaluating the thermodynamic quantities previously
discussed. The first of which is the pressure versus temperature (ln(p) vs -1/T)
diagram, Figure 2.1. In Figure 2.1 all of the refrigerants and refrigerant mixtures
8
-------
used in this study are plotted. Given the condensation and evaporation
temperatures, and the ln(p) vs -1/T diagram for a fluid it is possible to determine
the high and low side pressures of a vapor compression system for that fluid.
Also the Clausius-Clapeyron equation shows that the slope of the lines in Figure
2.1 is proportional to the latent heat, which has already been shown to be
important.
The second diagram of interest is the pressure versus enthalpy (p-h)
diagram. The p-h diagram for the refrigerants studied here can be seen in
Figures 2.2 through 2.7, respectively. In addition to the information from the
ln(p) versus -1/T diagram, the p-h diagram also shows the temperature glide of
the mixtures as it changes with pressure. Furthermore, the p-h diagram clearly
shows the latent heat of vaporization. This diagram is also useful for plotting
cycles, which is done in Chapter 12.
In addition to thermodynamic properties, transport properties also
influence the performance of a vapor compression system. The transport
properties play a role in the transfer of momentum and heat within the system.
Among the transport properties, the liquid dynamic viscosity and the liquid
thermal conductivity most significantly influence the system performance
(Domanski and Didion, 1993). Domanski showed that a 10% increase in thermal
9
-------
conductivity represents, roughly a 1% increase in COP while a 10% decrease in
liquid viscosity represents, roughly, a 1% increase in COP. As seen in Table
2.3, the largest difference in transport properties among the pure refrigerants
investigated here exists between R-32 and R-125, In that case the liquid thermal
conductivity of R-32 is 224% higher than that of R-125.
Now that the important thermophysical properties have been discussed,
the refrigerants used in the AC/HP will be examined. R-22 is addressed first
followed by the two replacement refrigerants,
2.2 R-22
R-22, chlorodifluoromethane, has been in use since 1936 in air-conditioners,
heat pumps, and chiller equipment (Calm, 1995). Since then it has become the
most widely used of all refrigerants. The reasons for R-22's popularity becomes
apparent when examining R-22's thermodynamic properties in Table 2.1, and the
pressure - enthalpy diagram displayed in Figure 2.3. Based on Alefeld's number,
it is clear that R-22 is desirable because of its relatively low specific heat and
high latent heat. Based on its interaction with the environment, R-22 is not
attractive. The presence of chlorine causes R-22 to have an appreciable ozone
depletion potential when compared to other refrigerants, as seen in Table 2.5.
This is why R-22 will be phased out in new machines by the year 2000 in
10
-------
Germany (UNEP, 1991) and by the year 2020 in the United States (Didion,
1994). As global warming receives more attention, further restrictions on R-22
may be expected since it is also a greenhouse gas,
2.3 R-32/134a
The mixture of R-32 and R-134a in the 30/70 wt.% concentration (R-32/134a)
was one of the first proposed replacements for R-22 (ARI, 1993). From an
environmental point of view, R-32/134a is more attractive than R-22. R-32/134a
has a lower OOP and GWP as seen in Table 2.5. From the ln(p) versus -1/T
diagram, Figure 2.1, it is clear that R-32/134a's vapor pressure curve is similar
to R-22's, hence it can be used in the same equipment. From the p-h diagram,
Figure 2.6, it is clear that R-32/134a has a higher latent heat and a temperature
glide. As discussed earlier, the higher latent heat tends to indicate a higher
performance. However, the temperature glide may either degrade, improve, or
do nothing to the system performance depending on the system design.
An important concern regarding R-32/134a is its flammability. R-32/134a
is not flammable. However, at a concentration of 56/64 wt.% this mixture does
become flammable (Richard and Shankland, 1992). The potential for this
refrigerant to fractionate does pose a threat. Under some conditions this
refrigerant may theoretically become flammable. An example of this would be
11
-------
if there was a vapor leak in a low pressure receiver that contained both liquid
and vapor. In this case, the concentration of the vapor phase may exceed the
flammable levels. Under these conditions, commonly termed worst case
fractionation, this mixture is considered flammable since it is possible for the
system to leak a flammable fluid. For this reason, R-32/134a is not being
considered in the U.S. as a replacement for R-22, despite its excellent
thermodynamic and transport properties. However, other countries are
continuing to explore the possibility of using R-32/134a as a replacement for R-
22 (Didion, 1994).
2.4 R-407C
R-407C, R-32/125/134a (23/25/53 wt.%), was developed to address the
flammability issues associated with R-32/134a. A third component, R-125, was
added to the R-32/134a mixture to quench the flammability. Although an
excellent flame retardant, R-125 is a poor refrigerant. Table 2.1 shows that R-
125 has a very low latent heat and a moderately high specific heat, both of
which contribute to R-125's poor performance as a refrigerant. Furthermore,
when comparing the transport properties of R-407C to R-32/134a it is clear that
the presence of R-125 lowers the thermal conductivity of the mixture by 10.3%.
In spite of these short comings, R-407C is currently considered the leading
candidate to replace R-22 in the United States.
12
-------
2.5 References
Alefeld, G., 1987, "Efficiency of Compressor Heat Pumps and Refrigerators
Derived from the Second Law of Thermodynamics", international Journal of
Refrigeration, Vol. 10, No. 11, pp. 331-338.
ARI, 1993, Participant's Handbook: R-22 Alternative Refrigerants Evaluation
Program (AREP).
Calm, J.M., 1995, Refrigerant Data Base, ARTI report DOE/CE23810-59C.
Didion, D.A, 1994, "The Impact of Ozone-Safe Refrigerants on Refrigeration
Machinery Performance and Operation", Proceedings of the Society of Naval
Architects and Marine Engineers, June, pp. 7-12.
Domanski P.A., Didion, D.A., 1993, "Thermodynamic Evaluation of R-22
Alternative Refrigerants and Refrigerant Mixtures", ASHRAE Transactions, Vol.
99, Part 2, pp. 636-648,
McLinden, M.O., Didion, D.A., 1987, "Quest for Alternatives", ASHRAE Journal,
Vol. 24, No. 2, pp. 32-42.
NIST, 1994, "Thermodynamic Properties of Refrigerants and Refrigerant
Mixtures (REFPROP), Version 4.0"; National Institute of Standards and
Technology, Gaithersburg, MD.
Pannock, J., 1989, "A Study of Zeotropic Mixtures", Ph.D. Dissertation,
University of Maryland.
Richard, R., Shankland, I., 1992, "Flammability of Alternative Refrigerants",
ASHRAE Journal, Vol, 24, No. 4, pp. 20-24.
UNEP, 1991, "Report of the Refrigeration, Air Conditioning and Heat Pumps
Technical Options Committee", Montreal Protocol.
13
-------
Table 2.1 Thermodynamic Properties of Pure Refrigerants (NIST, 1994)
Property
R-22
R-32
R-125
R-134a
Chemical
Formula
CHCIF2
CH2F2
CHF2CF3
CH2FCF3
Critical
Temperature
(°C)
96.2
78.2
66.2
101.2
Latent Heat
at 25°C
(kJ/kg)
180.6
271.9
110.5
178.0
Specific Heat
of Liquid at
25°C
(kJ/kg *K)
1.289
1.926
1.426
1.425
LFL
to
HFL
(Richard and
Shankland,
1992)
n/a
12.7%
to
33.5%
n/a
n/a
Molecular
Weight
(kg/kmol)
86.5
55
120
102
Ai/104
at 273 K
3.827
3,856
3.018
3.412
14
-------
Table 2.2 Thermodynamic Properties of Refrigerant Mixtures (NIST, 1994)
Property
R-32/R-134a
(30/70 wt.%)
R-32/R-125/R-134a
(23/25/52 wt.%)
ASHRAE Name
none
R-407C
Critical
Temperature (°C)
101.81
97.5
Latent Heat at 25°C
(kJ/kg)
216.7
189.3
Specific Heat of
Liquid at 25°C
(kJ/kg*K)
1.536
1.481
Temperature
Glide at 1 atm (°C)
7.4
7.1
LFL
to
HFL
n/a
n/a
Molecular
Weight
(kg/kmol)
79.2
86.2
AI/104
at 273 K
3.854
3.491
Table 2.3 Transport Properties of Pure Refrigerants at 25°C (NIST, 1994)
Property
R-22
R-32
R-125
R-134a
Thermal
Conductivity
(W/m*K)
1.15e-2 (v)
8.80e-2 (I)
1.41 e-2 (v)
1.22e-1 (I)
1.50e-2 (v)
5.45e-2 (I)
1.41 e-2 (v)
8.12e-2 (I)
Dynamic
Viscosity
(micropoise)
1.31e+2 (v)
1.71e+3 (I)
1.29e+2 (v)
1.13e+3 (I)
1.45e+2 (v)
1.41e+3 (I)
1 22e+2 (v)
2.12e+3 (I)
Note] v: vapor
I: liquid
15
-------
Table 2.4 Transport Properties of Refrigerant Mixtures at 25°C (NIST, 1994)
Property
R-32/134a
R-407C
Thermal
Conductivity
(W/m*K)
1.31e-2 (v)
9.61 e-2 (I)
1.31 e-2 (v)
8.62e-2 (I)
Dynamic
Viscosity
(micropoise)
1.25e+2 (v)
1.69e+3 (I)
1 28e+2 (v)
1,64e+3 (I)
Note] v: vapor
I: liquid
Table 2.5 Environmental Properties (Calm, 1995)
Property
R-22
30/70
R-407C
O.D.P.
(CFC-11=1.0)
0.050
0
0
G.W.P.
100 yr
(C02 = 1.0)
1700
970
1370
16
-------
ra
0.
p
£
Q.
5000-
2000"
1000^
500"
200-
100
50"
20-
10"
"1 R-22
"2 R-32
3 R-125
4 R-134a
::^:::t::::R.407G:: —1
-80
-40 0
Temperature [°C]
40
80
120
Figure 2.1 Pressure - Temperature Diagram
to
CL
5*
CD
k.
3
to
w
2000
-200
-100
0 100 200
Enthalpy [kJ/kgj
300
400
Figure 2.2 R-134a Pressure - Enthalpy Diagram
17
-------
m
n
£
3
0
OT
-------
ra
Q.
,v
&
%
W
03
-200
-100
0 100 200
Enthalpy [kJ/kg]
300
400
Figure 2.5 R-125 Pressure - Enthalpy Diagram
CO
£L
£
to
CO
03
-200 -100 0 100 200
Enthalpy [kJ/kg]
300
400
Figure 2.6 R-32/134a Pressure - Enthalpy Diagram
19
-------
-200
-100
0 100 200
Enthalpy [kJ/kg]
300
400
Figure 2.7 R-407C Pressure - Enthalpy Diagram
20
-------
Chapter 3
Determination of Thermophysical Properties
3.1 Introduction
Thermophysical properties constitute a broad class of related attributes which
are used to describe a substance. The thermophysical properties of concern in
this project can be subdivided into thermodynamic and transport properties.
The thermodynamic properties of interest are temperature, pressure, volume,
enthalpy, entropy, and concentration. Accurate relationships between these
quantities are necessary for both the experimental and theoretical aspects of this
work. The transport properties of interest are the thermal conductivity and
dynamic viscosity. The viscosity is used in the simulation to determine the
frictional pressure drop and flow regime. The thermal conductivity and viscosity
are necessary for evaluating the heat transfer coefficients required by the
simulation. The property relations of two different types of fluids, refrigerant and
air, are required. Since, there are no parallels between the regions of interest
for these fluids they will be treated separately.
21
-------
3.2 Refrigerant Properties
3.2.1 Refrigerant Thermodynamic Property Relations
The mathematical relationship between various thermodynamic properties is
typically called an equation of state (EOS), Before discussing the merits of
different EOSs it is important to recognize what is required of the EOS for this
work. Since the experimental requirements of the EOS are a subset of the
simulation requirements, the simulation requirements are discussed.
The primary requirement of the EOS is that it be accurate. However, no
EOS is equally accurate at predicting all quantities. An important aspect of the
simulation, that distinguishes it from others, is that it predicts the mass and
concentration of the refrigerant in the system at various locations in time and
space. Therefore, the EOS must accurately represent the liquid refrigerant
volume, as well as the other parameters necessary for determining the COP and
capacity. A secondary requirement is computational speed. Since the EOS is
one of the innermost subroutines of the simulation, the time required for the EOS
plays a large role in the overall speed of the simulation. Furthermore, for the
sake of consistency, the EOS should be suitable for the mixtures as well as for
pure components.
There are several EOS which are exceptionally accurate. Kamei created
22
-------
such an EOS for R-22 which is widely accepted (Kamei et al,, 1992). This EOS
is based on the Helmholtz energy function and utilizes 27 coefficients. Another
EOS of this type which has received attention for pure refrigerants is the
modified Bennedict-Webb-Rubin which has 32 parameters (Bennedict, Webb,
Rubin, 1940). There also exists an accurate EOS tailored for the mixtures which
are examined in this work (Morrison et al., 1994). This EOS is attractive
because it accurately predicts the vapor and liquid volumes as well as the
equilibrium concentrations while maintaining a low computational overhead.
More specifically, it achieves a high level of accuracy by using two different
EOSs for the liquid and the vapor. This is undesirable, since it results in an EOS
that is not thermodynamicalfy consistent. Generally speaking, fluid specific
EOSs rely on a large number of empirical constants. As a result they are
accurate, but they are also computationally expensive which makes these EOSs
undesirable from a simulation point of view. Furthermore, requiring that the
same EOS be used for all refrigerants eliminates these fluid specific EOSs.
There are primarily four other EOSs which have been considered for both
pure components and mixtures. They are:
•Camahan-Starling-DeSantis (CSD) (DeSantis et al., 1976)
• Lee-Kesler-Plocker (LKP) (Plocker et at,, 1978)
-------
•Peng-Robinson (PR) (Peng and Robinson, 1976)
• Redlich-Kwong-Soave (RKS) (Soave, 1980)
Several authors have investigated the performance of some or all of these EOS
(Hogberg and Vamling, 1995), (Gerdsmeyer and Kruse.1988), (McLinden and
Vamling,1993). To summarize, these studies found that no EOS outperformed
the others for all fluids and for all thermodynamic quantities. However, the
following generalizations can be made. The CSD equation better predicted the
specific volume of liquid and vapor for pure components and mixtures
(Gerdsmeyer and Kruse,1988): (McLinden and Vamling,1993). The heat
capacity for pure refrigerants is best predicted by the CSD EOS while the LKP
EOS performs more effectively for mixtures (Gerdsmeyer and Kruse,1988),
(McLinden and Vamling, 1993). When comparing the system capacity and COP
calculated by the EOSs considered here to the results from the Kamei EOS, the
CSD EOS performs the best (Hogberg and Vamling. 1995). It should be
mentioned that the RKS EOS was not quite as accurate as the other EOSs, but
it was three to four times faster cue to the simpler structure of the equation.
Since the CSD EOS performed the best when evaluating COP, capacity,
and liquid volume, the CSD EOS was chosen for this study. The REFPROP
embodiment of the CSD EOS is utilized with some modification (NIST, 1994).
24
-------
The basic form of the CSD EOS is given in Equation 3.1.
p = + y + y—y_) ?— wah y ~ b/Av 31
v(1 - y)3 v(v + b)
Since p is dependent on v nonlinearly, the simulation is written so that T and v
are always the explicit variables. Otherwise, time would be spent iterating for the
volume.
The modifications made to REFPROP are aimed at reducing the time
required to determine the equilibrium properties. This is the most time
consuming process for any EOS. For a pure component, equilibrium between
phases is achieved when the Gibbs free energy of each phase is equivalent at
the temperature of interest. This is the same as requiring that the areas 1 and
2 in the p-v diagram (Figure 3.1) be equal (Bejan, 1988). REFPROP basically
utilizes two steps to find the equilibrium state. First REFPROP starts at very low
and high pressures and marches along the isotherm until there is a change in
slope to find phigh and p,ow, as illustrated in Figure 3.1. These pressures will
bracket the saturation pressure. Then REFPROP starts with values of phigh and
plowand iterates with a secant/reguli-falsi scheme until the Gibbs free energy of
each phase is equal {NIST, 1994). This typically requires between 5 and 15
iterations. This procedure is modified to expedite the calculation of the
saturation conditions.
25
-------
Instead of marching along the isotherm to obtain phigh and pfow an equation
is added which closely approximates the saturation pressure as a function of
temperature. Typically, the equation is accurate enough to only require one or
two iterations of the secant/reguli-falsi scheme. The equation is obtained by
curve fitting the saturated temperature and pressure data from the old scheme.
A similar procedure is followed for the mixtures with the exception that the
saturation pressure curve is now a function of temperature and concentration.
3.2.2 Refrigerant Transport Property Relations
Before discussing the methods used for predicting the transport properties, it will
be useful to establish the sensitivity of COP and capacity of a AC/HP to these
parameters. Table 3.1 shows the deviation in the transport properties that
causes a 1% change in COP and capacity (Hogberg and Vamling, 1995). It is
clear from this table that large errors are acceptable in the vapor transport
properties while moderate errors are acceptable in the liquid transport
properties. It should also be noted that similar values are reported by McLinden
(McLinden and Vamling, 1993).
The transport properties of liquids are strong functions of temperature but
only weak functions of pressure. For example the conductivity of R-22 at 25 °C
only changes by +0.3% when the pressure is changed from 1045 kPa to 1545
26
-------
kPa (NIST, 1994). The absolute viscosity only changes by -0.2% over the same
interval (NIST, 1994). Hence, pressure dependance is not considered important
when predicting liquid transport properties.
The liquid transport properties are represented by curve fitting data from
REFPROP which predicts these properties in addition to the thermodynamic
properties. The data is curve fit because REFPROP utilizes an extended
corresponding states model which requires a significant amount of time to predict
transport properties (Huber et al.s 1992), (Huber and Ely. 1992). Thus the
simulation is made faster by making the transport properties a function of one
variable and one equation. The data from REFPROP can be seen in Figures 3.2
and 3.3 for the liquid thermal conductivity and dynamic viscosity, respectively.
It is clear from Figures 3.2 and 3.3 that these properties behave similarly for
each fluid. Therefore, for simulation purposes, each property was represented
by the same equation with different constants for each fluid. An approach similar
to corresponding states is used to reduce the data to one equation. R-22 is used
as the reference fluid for this procedure. The data in question are generated as
a function of the reduced temperature. Two reference points are taken and the
data are manipulated so that the data reduces to either zero or one at these
reference points. Equation 3.2 is an example of this process for the liquid
conductivity. This results in a function which is independent of the fluid. The
27
-------
maximum deviation in y{Tr) between fluids is less than 0,5% in the temperature
range of interest,
V(T] _ W - VTr ' 085> ,,
'' k,q(Tr = 0.75) - k^Tr = 0.85)
The data from this function, y(Tr)» are now used to generate an equation that
expresses y as a function of Tr. Therefore, to recover the conductivity at any
temperature for any fluid requires only the critical temperature and the values of
the conductivity at the two reference temperatures.
Figures 3.4 and 3.5 show the vapor thermal conductivity and dynamic
viscosity, respectively, and were also generated by REFPROP. It is interesting
to note that, while being more sensitive to pressure than liquid transport
properties, vapor transport properties are only weak functions of pressure. For
R-22 at 25°C the conductivity only changes by -1.1 % when the pressure
changes from 1045 kPa to 545 kPa (NIST, 1994). Similarly, the viscosity only
changes by -2.7 % over the same range (NIST, 1994). This deviation is well
below those in Table 3.1. Therefore, the pressure dependence of the vapor
transport properties will not be considered. Hence, the vapor properties are
predicted using the same algorithm developed for the liquid properties.
28
-------
3.3 Air Properties
3.3.1 Air Thermodynamic Property Relations
Air can be considered an ideal gas since the temperature and pressure range of
interest is sufficiently far from the critical point. This allows many assumptions
which greatly simplify and do not significantly reduce the accuracy of the
equations which relate various thermophysical properties. The thermodynamic
property data of air are obtained using relationships found in ASHRAE
Fundamentals and the interested reader should consult this reference for more
details (ASHRAE, 1993).
3.3.2 Air Transport Property Relations
The thermal conductivity and dynamic viscosity are determined by curve fitting
data in Incropera and DeWitt (Incropera and DeWitt, 1990). The influence of
moisture is neglected as is often the case (ASHRAE, 1993). The error
associated with assuming no moisture dependence on the transport properties
is approximately 2.0% (ASHRAE, 1993).
3.4 References
ASHRAE, 1993, Fundamentals. Atlanta, Georgia.
Bejan, A., 1988, Advanced Engineering Thermodynamics. John Wiley and Sons.
New York.
Bennedict, M., Webb, G.B., Rubin, L.C., 1940, "An Empirical Equation for the
29
-------
Thermodynamic Properties of Light Hydrocarbons and Their Mixtures", Journal
of Chem. Phys., Vol. 8, pp.334-348.
deSantis, R., Gironi, F., Marelli, L., 1976, "Vapor-Liquid Equilibrium from a Hard-
Sphere Equation of State", Ind, Eng. Chem. Fundam., Vol. 15, pp. 183-189.
Gerdsmeyer, K.D., Kruse, H., 1988, "Comparison of Equations of State for
Application to Nonzeotropic Refrigerant Mixtures", IIR, Vol. 2, pp. 53-62.
Hogberg, M., Vamling, L., 1995, "Impact of Uncertainties on Estimations of Heat
Pump Cycle Performance", Ph.D. Dissertation, Chalmers University, Goteborg,
Sweden.
Huber, M.L., Friend, D.G., Ely, J.F., 1992, "Prediction of the thermal conductivity
of refrigerants and refrigerant mixtures", Fluid Phase Equilibria, Vol, 80, pp. 249-
261.
Huber, M.L., Ely, J.F., 1992, "Prediction of the viscosity of refrigerants and
refrigerant mixtures", Fluid Phase Equilibria, Vol. 80, pp. 239-248.
Incropera, F.P., DeWitt, D.P., 1990, Fundamentals of Heat and Mass Transfer.
John Wiley and Sons, New York.
Kamei, A., Beyerlein, S.W., Lemmon, E.W., 1992, "A Fundamental Equation of
State for Chlorodifluoromethane (R-22)", Fluid Phase Equilibria. Vol. 80, pp. 71-
86.
McLinden, M., Vamling, L., 1993, "The Need for, and Availability of, Working
Fluid Property Data: Results from Annexes XIII and XVIII", Heat Pumps for
Energy Efficient Environmental Progress, pp. 115-126.
Morrison, J.D., et a!., 1994, "The Use of an MHV-2 Equation of State for
Modeling the Thermodynamic Properties of Refrigerant Mixtures", IIR
Conference, Padova, Italy, pp. 461 -469.
NIST, 1994, "Thermodynamic Properties of Refrigerants and Refrigerant
Mixtures (REFPROP), Version 4.0"; National Institute of Standards and
Technology, Gaithersburg, MD.
Peng, D.Y., Robinson, D.B., 1976, Ind, Eng. Chem. Fundam,, Vol. 15,
Plocker, U., Knapp, H., Prausnitz, J.M., 1978, "Calculation of High Pressure
30
-------
Vapor-Liquid Equilibrium from a Corresponding States Correlation with Emphasis
on Asymmetric Mixtures", Ind. Chem. Proc. Des, Dev., Vol. 17, pp. 324-332.
Soave, G., 1980, "Rigorous and Simplified Method for Determining the Pure-
Component Parameters in the Redlich-Kwong-Soave Equation of State", Chem.
Eng. Sci., Vol. 35, pp. 1725-1736.
31
-------
Table 3.1 Maximum Permissible Uncertainty in Transport Properties to
calculate the COP and Capacity within 1 % (Hogberg and Vamling,
1995).
Variable
COP
Capacity
Liquid Thermal
Conductivity
11 %
11 %
Liquid Viscosity
20%
16%
Vapor Thermal
Conductivity
43%
67%
Vapor Viscosity
99%
99%
-------
VOLUME
Figure 3.1 Typical Pressure - Volume Diagram
260 270 280 290 300 310 320
TEMPERATURE (K)
R-22 R-32/134a (20/80 wt.%) R-32/134a (30/70 wt.%) R-32/134a (40/50 wt.%) R-407C
Figure 3.2 Saturated Liquid Thermal Conductivity of the Refrigerants
Studied
33
-------
260
270
300
310
280 290
TEMPERATURE (K)
R-22 R-32/134a (20/80 wt.%) R-32/134a (30/70 wt.%) R-32/134a (40/60 wt.%) R-407C
320
Figure 3.3 Saturated Liquid Dynamic Viscosity of the Refrigerants Studied
260
280 290 300
TEMPERATURE (K)
R-22 R-32/134a (20/80 wt.%) R-32/134a (30/70 wt.%) R-32/134a (40/60 wt.%) R-407C
320
Figure 3.4 Saturated Vapor Thermal Conductivity of the Refrigerants Studied
34
-------
1.5E-5
1.4E-5
US
Q
O 1.3E-5
a
1.2E-5
1.1 E-5
1E-5
260
290
TEMPERATURE (K)
R-22 R-32/134a (20/80 wt.%) R-32/134a (30/70 wt.%) R-32/134a (40/60 wt.%) R-407C
Figure 3.5 Saturated Vapor Dynamic Viscosity of the Refrigerants Studied
35
-------
Chapter 4
The Vapor Compression Systems Studied
4.1 Introduction
Heating and cooling can be achieved through numerous processes. The most
popular of these processes is vapor compression. Vapor compression systems
utilize a mechanical device called a compressor to add work to the system so
that heat may be rejected and absorbed, as opposed to other systems which may
require the addition of heat or other forms of energy for this purpose. From a
refrigerator with a capacity of 0.20 kW to chiller packages with capacities in
excess of 2000 kW, there is a wide variety of vapor compression systems that
can be studied, all of which rely on the same principles of operation. The work
here focuses on residential air-conditioners/heat pumps (AC/HP). AC/HPs were
chosen for study because they are one of the most prevalent vapor compression
systems available today. Two AC/HPs are utilized in this study. Both of these
units are used to validate the steady state performance of the simulation
components. Due to the difficulty associated with conducting accurate transient
experiments, only one of these units is used for the transient system analysis.
Before describing each of the units separately, it is instructive to describe the
basic theory of operation.
36
-------
Figure 4,1 illustrates AC/HP 1, which is described in detail in the following
section, but it is used here for the purpose of describing the components and the
theory of operation. The heart of a vapor compression system is the
compressor. The compressor motivates the refrigerant from the low pressure
side to the high pressure side which is achieved through the input of mechanical
work to the compressor. The four-way valve is also a significant component of
a AC/HP. The four-way valve changes the direction of the refrigerant to place
the AC/HP in the heating or cooling mode. The remaining two components are
the indoor and outdoor heat exchangers. The heat exchangers allow heat
transfer between the refrigerant and the air so that the cooling or heating effect
can be utilized. To gain a better understanding of how a AC/HP operates the
refrigerant state throughout the unit is described next. The cooling mode is
considered first.
In the process of describing the operation of an AC/HP some data are
presented from a typical cooling test with R-22, The refrigerant state is
described by following the refrigerant through the cycle, starting and ending at
the compressor discharge. At the discharge of the compressor, the refrigerant
is a warm (77.9°C), high pressure (1440 kPa) vapor. The refrigerant is
discharged from the compressor to the four way valve. In the cooling mode, the
four-way valve directs the refrigerant to the outdoor heat exchanger. There, the
37
-------
refrigerant transfers some of its heat to the outside air (27.8°C), consequently
reducing its temperature to 32.0 °C. In so doing the refrigerant is condensed.
The high pressure liquid, traveling at roughly 0.48 m/s in a pipe with a nominal
diameter of 9.5-10"3 m, is then expanded through an expansion device to a lower
pressure (643 kPa) and temperature (8.3 °C). The refrigerant subsequently
passes through the indoor heat exchanger where the low temperature refrigerant
absorbs heat from the warm air (26.7 °C). In so doing the refrigerant becomes
vapor once again at a temperature of 8.5 °C and a pressure of 618 kPa. The
circuit is completed as the refrigerant vapor flows at approximately 8.8 m/s in a
pipe with a nominal diameter 1.9 10"2 m. The pipe connects with the four-way
valve which directs the refrigerant back to the compressor suction.
For the heating mode the principal of operation is the same but the flow
of refrigerant is reversed through the heat exchangers. The warm high pressure
refrigerant still leaves the compressor discharge but is now routed to the indoor
heat exchanger due to the action of the 4-way valve. There the refrigerant acts
to warm the air stream. In so doing, the refrigerant condenses and becomes a
high pressure liquid. The high pressure liquid now enters the expansion device
in the outdoor unit where the refrigerant's temperature and pressure are lowered.
Then the refrigerant enters the outdoor heat exchanger where it picks up heat
from the ambient. This heat exchange vaporizes the refrigerant. The refrigerant
38
-------
vapor enters the four-way valve which refers the refrigerant to the inlet of the
compressor suction, where the circuit is completed.
4.2 AC/HP Instrumentation
The AC/HP 1 and AC/HP 2 have nearly identical instrumentation. Figure 4.1
shows the position of all of the refrigerant instrumentation for AC/HP 1, and can
be used as an analog for AC/HP 2. Unless otherwise noted, the described
instrumentation applies to both AC/HP1 and AC/HP 2, The purpose of this
section is to describe the iocation of the refrigerant side instrumentation. The
instrumentation accuracy is addressed in Chapter 6.
The temperatures of the refrigerant are measured throughout the system.
The refrigerant temperatures at the inlet and outlet of both the indoor and
outdoor heat exchangers are measured. Each bend of the indoor and outdoor
heat exchanger has a thermocouple attached to it. The temperature of the
refrigerant entering and leaving the compressor and expansion device is
measured. The temperature leaving the mass flow meter is also measured.
The refrigerant pressure is measured in several locations as well. Both
AC/HPs have pressure transducers before and after the compressor, after the
mass flow meter, and on the liquid and vapor lines of the indoor heat exchanger.
39
-------
AC/HP 1 has an additional pressure transducer after the expansion device of the
indoor heat exchanger. This pressure transducer was added so that the
pressure drop through the indoor heat exchanger could be quantified in the
cooling mode.
The refrigerant mass flow rate is also measured. Specifically, a mass flow
meter is located in the liquid line between the indoor and outdoor heat
exchangers.
AC/HP 1 has 10 ports for sampling the refrigerant. Five ports are located
on the liquid line of the indoor heat exchanger and the remaining five are on the
vapor line of the indoor heat exchanger. The refrigerant is sampled so that its
concentration can be determined by using gas chromatography. The sampling
port is depicted in Figure 4.2, A 3.18 mm tube extends down into the middle of
the flow to prevent the sampling of oil that travels along the tube wall. As shown
in Figure 4.3, this tube communicates with a solenoid valve which is activated by
the data acquisition system. The volume between the valve and the AC/HP is
minimized by soldering brass into the dead volume of the valve. This is done to
minimize the effect of the refrigerant that is trapped in the valve prior to being
activated. The trapped volume is three orders of magnitude less than the
metered volume. When the solenoid valve is energized it allows refrigerant to
40
-------
fill the metered volume. After the solenoid valve is closed the manual valve is
opened to allow the refrigerant to expand into the sample bottle. After which the
manual valve is closed and the sample bottle is removed. The sample bottle is
then taken to the gas chromatograph to determine the concentration of the
refrigerant.
The two test units are now described. It should be noted that AC/HP 1 is
used for both simulation verification and detailed system analysis while AC/HP
2 is used only for testing the various component models of the simulation. The
characteristics of these two units are as shown in Table 4.1.
4.3 AC/HP 1
This AC/HP is a nominal 3 ton unit from Lennox. The indoor heat exchanger unit
is model CB19-26 while the outdoor heat exchanger unit is model HP19-411.
The indoor heat exchanger is depicted in Figure 4.4 and the outdoor heat
exchanger is depicted in Figure 4.5. This AC/HP utilizes a scroll compressor
from Copeland (ZR34K1-PFV). This heat pump comes equipped with
thermostatic expansion valves (TXVs) for both heating and cooling modes. A
TXV is used in an effort to maintain the superheat near a constant value
independent of operating conditions. The superheat is maintained by
continuously adjusting the flow area of the TXV in response to conditions at the
41
-------
outlet of the evaporator. The TXVs are replaced with short tube restrictors
(STRs) for some of the experiments for two reasons. One is to evaluate the
effect of the expansion device on system performance and the other is to ensure
that the transients studied were not caused by the TXVs.
Another modification to this system is the addition of a vapor to liquid line
heat exchanger (VLHX). The VLHX transfers heat from the cold vapor line
leaving the evaporator to the warm liquid line entering the expansion device.
This device is installed in an effort to improve the performance of R-407C
relative to R-22 and to examine what affect it has on transient operation. The
VLHX is only used in the cooling mode. This heat exchanger is removed when
it is not being tested.
4.4 AC/HP 2
This AC/HP is manufactured by Trane and has a nominal capacity of two tons.
The indoor heat exchanger unit is a TVW024b14A and is depicted in Figure 4.6.
The outdoor unit is a TWN024c100A and the associated heat exchanger is
shown in Figure 4.7. The compressor is a reciprocating piston type
manufactured by Trane (CP223-BC1-G). As with the previous AC/HP, this model
is designed for single phase 200/230 V. This AC/HP is used only to verify the
different components of the simulation.
42
-------
Table 4.1 Characteristics of Test Units
Unit
AC/HP 1
AC/HP 2
Item
indoor
Outdoor
Indoor
Outdoor
Model
CBH19-26
HP19-411
TWV024b14A
TWN024C100A
Airflow rate [rrf/min]
Fan Motor output [HP]
Fan Motor Speed [RPM]
32.9
1/3
n/a
94.9
1/6
820
22.7
1/4
1075
71.5
1/6
825
Heat Exchanger
Face Area [m2]
Row x Fin Pitch [mm]
Fin Type
0.490
3x2.12
Wavy
1.453
2 x 1.49
Wavy
0.206
4x2.31
Wavy
1.629
1 x1.10
Spine
Compressor
Scroll(ZR34k1 -PFV)
Reciprocating(CP223-BC1 -G)
Expansion Device
TXV
(Cooling)
TXV
(Heating)
S.T.
(Cooling)
TXV
(Heating)
Power [V/Ph/Hz]
208-230/
1/60
208-230/
1/60
200-230/
1/60
200-230/
1/60
Dimension (WxHxD) [cm]
59.1x129.5
x54.6
81.6x78.4
x86.5
59.1x111.9
x45.3
88.3x74.3
x79.4
Net Weight [kg]
80.7
n/a
42.2
97.5
43
-------
INDOOR COIL
Figure 4.1 AC/HP 1 Circuit Schematic
Figure 4.2 Refrigerant Sampling Port
44
-------
Vacuum
Pump
3j Gas-
Chromatography
Figure 4.3 Refrigerant Sampling Procedure and Apparatus
Figure 4.4 Indoor Heat Exchanger for AC/HP1
45
-------
TO3 VIEW
Figure 4.5 Outdoor Heat Exchanger for AC/HP 1
ifiiiii
iMN
f
1§®I
TOP VIEW
o o^\,
AiRFLOW
V&VS OvSvX
> n G/
\r: ox
Figure 4.6 Indoor Heat Exchanger for AC/HP 2
46
-------
Figure 4.7 Outdoor Heat Exchanger for AC/HP 2
47
-------
Chapter 5
The Test Facility
5.1 Introduction
The test facility is constructed to evaluate the performance of two residential air
to air AC/HPs independently. In order to do this, the test facility has two duties.
One is to maintain the environment in which the AC/HP operates and the other
is to measure the indoor air side capacity of the AC/HP. The test facility is
designed to conduct the entire battery of steady state and cyclic tests set forth
in ASHRAE standard 116 (ASHRAE, 1983).
As described in Chapter 4, each heat pump consists of an outdoor unit
and an indoor unit. The outdoor unit is placed inside an environmental control
system which simulates outdoor conditions. This control system is called the
outdoor chamber. The indoor unit is placed inside an environmental control
system which simulates indoor conditions and is termed the indoor loop. The
relative location of the outdoor chambers and indoor loops is shown in Figure
5.1. Both of the indoor loops and outdoor chambers are identical and are
described below. A more detailed description is given in (Judge, 1994).
48
-------
5.2 Outdoor Chamber
5.2.1 Outdoor Chamber Environment Control
The outdoor chamber maintains a dry bulb temperature from -8.33 °C (17 °F) to
35.0 °C (95 °F) over a wide range of wet bulb temperatures as required by
ASHRAE standard 116 (ASHRAE, 1983). The dry bulb temperature does not
vary by more than 0.2 °C (0 4 °F) while the wet bulb temperature does not vary
more than 0.2 °C (0.3 °F). Figure 5.2 shows the configuration of the outdoor
chamber.
In order to remove heat from the chamber each of the outdoor chambers
has its own condensing unit and evaporator which are controlled by a PID
controller. This controller determines whether hot refrigerant from the
compressor or liquid refrigerant from the condenser enters the heat exchanger
in the air handler. This is achieved through the action of solenoid valves. The
refrigerant heat exchanger is followed by a bank of electric heaters which are
also controlled by the same PID unit.
In an effort to maintain control of the humidity, each outdoor chamber has
a desiccant drier and a steam generator which are both controlled by a PID
controller. The humidity is lowered through the use of a desiccant drier which
pulls air from the floor of the chamber and discharges the dry air to the inlet of
49
-------
the air handler. The humidity is increased by a steam generator which injects
steam into the discharge air stream of the air handler.
5.2.2 Outdoor Chamber Instrumentation
The instrumentation in the outdoor chamber is used to maintain the temperature
and humidity levels as required by ASHRAE 116 standards. To this end, there
is a grid of nine thermocouples at the inlet to the AC/HP. The humidity is
measured by two independent methods. One is a relative humidity transducer
and the other is a psychrometer.
5.3 Indoor Loop
5.3.1 Indoor Loop Environment Control
The indoor loop maintains the dry bulb temperature from 21.1 °C (70 °F) to 26.7
°C (80 °F) over a broad range of humidities. The dry bulb temperature does not
vary more than 0.3 °C (0.5 °F) from the set point, while the wet bulb temperature
varies less than 0.2 °C (0,3 °F) from the set point. Figure 5.3 shows the indoor
chamber set-up. The indoor loop has a blower in addition to the one within the
test unit. This blower exists to compensate for the additional pressure drop
caused by the measurement system and the air handler. The speed of the
blower is controlled by an inverter so that the desired flow rate is maintained.
50
-------
The indoor loop and the outdoor chamber utilize identical temperature
control strategies. Furthermore, both the indoor loop and outdoor chamber
utilize steam generators to add moisture to the air. However, the moisture is
removed differently for the indoor loops and the outdoor chambers. While the
outdoor chamber utilizes a desiccant drier to remove water from the air, the
indoor loop makes use of mechanical refrigeration. The indoor loop has an
additional refrigerant heat exchanger. This heat exchanger operates below the
dew point to condense water from the air, thereby lowering the humidity of the
air stream.
5.3.2 Indoor Loop Instrumentation
The instrumentation in the indoor loop is used to ensure that the environmental
conditions are being maintained and to measure the air side capacity of the
indoor unit of AC/HP. Due to its dual duties, the indoor loop has significantly
more instrumentation than the outdoor chamber. Since it is important to
accurately determine the air side capacity of the AC/HP the indoor loop makes
use of redundant measurement devices for key variables. The primary
measurement devices will be described first followed by the secondary device.
The inlet and outlet air temperatures of the AC/HP are each measured by
an equidistant grid of nine thermocouples. Prior to entering each grid of
51
-------
thermocouples, the air moves through two mixers to ensure that the air is
thoroughly mixed. A thermopile is used as the redundant measurement device.
The thermopile consists of nine thermocouples on each leg which are attached
at the same locations as the previous thermocouples.
The humidity at the inlet to the AC/HP is measured with a dew point
meter. If no condensation occurs then the outlet humidity ratio is the same as
the inlet humidity ratio. When condensation does occur the condensate flow rate
is measured. The device used for this task is simply a bucket with the high
pressure side of a differential pressure transducer connected to the bottom of the
bucket. This device is calibrated with a scale to determine the mass of water in
the bucket as a function of the output from the pressure transducer. The data
acquisition system scans the output from this device periodically to determine
how the mass of condensate is changing with time, thereby yielding the
condensate mass flow rate. The outlet humidity ratio can then be calculated
from the inlet humidity ratio, the air flow rate, and the condensate flow rate. The
redundant humidity measurement devices are two psychrometers placed at the
inlet and outlet of the AC/HP.
The last measurement to be described is the measurement of air flow rate.
The airflow rate is determined by measuring the pressure drop across a nozzle
52
-------
specifically designed for this purpose. The redundant air flow rate measurement
device is a pitot tube located at the vena contracta of the nozzle.
5.4 References
ASHRAE, 1983, "Methods of Testing for Seasonal Efficiency of Unitary Air-
Conditioners and Heat Pumps", ASHRAE Standard 116-1983.
Judge, J., 1994, "An Experimental Study of a R-22 Replacement", M.S. Thesis,
University of Maryland.
53
-------
(Q
C
ro
01
Outdoor Chamber
#2
—~j
c/>
*—^
T1
0>
O
5 ft
16.5 ft
T~ P~
Ol
4^
x<£«v2
?:-x«vss
Dehumidifierl
Air Handier
Humidifier
Outdoor
Test
Unit #1
Outdoor Chamber '"door Room with two Indoor
#-) Chambers
16.5 ft
15ft
Dehurnidilier
Humidifier
-------
OUTDOOR TEST UNIT
>T
Refrig. out
AIR HANDLER
Temperature grid
with 9 thermocouples
Wet- and dry-bulb
temperature
FLOW
Code for outdoor unit;
T; Temperature measurement
p: Pressure measurement
Figure 5.2 Outdoor Chamber
Temperature grid
with 9 thermocouples
Refrig, in
Refrig, out
VIE
Discharge
Chamber
Receiving
Chamber
Pressure Two air stream Temperature grid
Taps mixers with 9 thermocouples
Figure 5.3 Indoor Loop
55
-------
Chapter 6
Instrumentation
6.1 Introduction
All of the instrumentation implemented in this work meets or exceeds the
standards set forth by ASHRAE 116 (ASHRAE, 1983). A significant portion of
the instrumentation utilized in this work has been characterized previously in
detail (Judge, 1994), (Hwang et al.. 1996). The following pages describe the
instrumentation not previously discussed.
6.2 Gas Chromatograph
A gas chromatograph is a device which measures the concentration of gas
mixtures. It is used here to determine the concentration of the refrigerant
mixtures at various times during the testing of the AC/HP. The gas
chromatograph is a Shimadzu GC-9A which employs a thermal conductivity
detector to sense the refrigerant. The chromatography column is specifically
designed for hydroflourocarbon (HFC) refrigerants. The column is a Hewlett
Packard series 530 filled with fused silica. The carrier gas is helium.
Before the gas chromatograph can be used to determine the
56
-------
concentration of a refrigerant mixture it is necessary to run baseline tests for
each component. Each of the components considered here was tested five times
to ensure that an accurate and reproducible baseline was obtained. The results
of the baseline tests can be seen in Table 6.1. The standard deviation in the
baseline data is two to three orders of magnitude less than the quantities in
question which indicates a high level of reproducibility.
The accuracy of the gas chromatograph was quantified by
chromatographing refrigerant mixtures of known concentration. The first
comparison can be seen in Table 6.2. This comparison is made with a
refrigerant sample from ICI, where the refrigerant was gas chromatographed
prior to shipping. The results from the gas chromatograph differed by 0.27 wt.%
from ICI's data. The second comparison, Table 6.3, is with a mixture made at the
University of Maryland. The concentration was determined by carefully weighing
the amount of each refrigerant charged into the bottle. The gas chromatograph
concentration differed from the weighed concentration by 0.14 wt%. The data
from the third comparison is in Table 6.4. Since it was not possible to either
obtain an accurate sample of R-407C or to make one due to the shortage of R-
125, the manufacturer's refrigerant had to be used with its high uncertainty, ± 1
wt. %. All that can be concluded from the last data set is that the gas
chromatograph measured concentrations within the manufacturer's uncertainty.
57
-------
Based on the worst case from these tests, the gas chromatograph can measure
the concentration to within 0.27 wt.% of the actual concentration.
6.3 References
ASHRAE, 1983, "Methods of Testing for Seasonal Efficiency of Unitary Air-
Conditioners and Heat Pumps", ASHRAE Standard 116-1983.
Hwang, Y., Judge, J., Radermacher, R,, 1995, "Testing of Refrigerant Mixtures
in Residential Heat Pumps", EPRI Final Report, TR-105394.
Judge, J., 1994, "An Experimental Study of a R-22 Replacement", M.S. Thesis,
University of Maryland.
58
-------
Table 6.1: Baseline Gas Chromatograph Data
Refrigerant
Average
Peak Area
Standard
Deviation
Average
Time Delay
Standard
Deviation
R-22
1.7402e7
5.66e4
2.472
1.10e-2
R-125
1.2905e7
6.85e4
5.57387
1.30e-2
R-134a
1.1432e7
5.69e4
7.355
2.28e-2
R-32
8.1053e6
A OOpijdl
» mam mam
4.6246
1.49e-2
Table 6.2: Gas Chromatograph Error of R-32/134a (compared to ICI GC)
Test Run
Mass Concentration of
R-32 (wt.%)
Mass Concentration of
R-134a (wt.%)
Gas
Chromatograph
29.77
70.23
Actual
Concentration
29.50 ± 0.05
70.50 ± 0.05
Error
0.91%
0.37 %
Table 6.3: Gas Chromatograph Error of R-32/134a
(compared to weighed concentration)
Test Run
Mass Concentration of
R-32 (wt.%)
Mass Concentration of
R-134a (wt.%)
Gas
Chromatograph
29.64
70.36
Actual
Concentration
29.78
70.22
Error
0.47 %
0.20%
59
-------
Table 6.4: Gas Chromatograph Error of R-407C
(compared to weighed concentration)
Test Run
Mass
Concentration
of
R-32 (wt.%)
Mass
Concentration
of
R-125 (wt.%)
Mass
Concentration
of
R-134a (wt.%)
Gas
Chromatograph
22.85
25.93
51.22
Actual
Concentration
23.0 ± 1
25.0 ± 1
52.0 ± 1
Error
0.65 %
3.7 %
1.5%
60
-------
Chapter 7
The Experimental Procedure
7.1 introduction
The experimental aspect of this work serves two purposes. One purpose is to
experimentally determine the cyclic and steady state behavior of the refrigerants
and system configurations studied while the other is to experimentally validate
the steady state and transient performance of the AC/HP simulation. Prior to
describing the experimental procedures that were used, the available literature
on the experimental transient performance of heat pumps will be discussed.
7.2 Previous Work
Compared to papers that deal with steady state, there has been relatively little
experimental work published on the transient performance of vapor compression
systems. The following papers are among the few that are relevant to this topic.
Among the early literature on this topic is a paper by Tanaka (Tanaka et
al., 1982). This paper investigated the time dependence of the R-22 mass
distribution in a heat pump. Temperature, pressure, mass, and power data were
all measured as a function of time. Two significant findings were reported. One
61
-------
was that at start up, the evaporator has 50% more refrigerant than at steady
state, while the converse is true of the condenser. Tanaka also found that
between 51 % and 59% of the refrigerant accumulates in the condenser. Finally,
it is concluded that the initial mass distribution is a function of the off-time.
Another experimental study was conducted by Murphy (Murphy and
Goldschmidt, 1984). This study provided temperature, pressure and power data
versus time for a pure component operating in a heat pump. The main
contribution of this study was the recognition of refrigerant migration during the
off-time and how this impacts the performance of the system during the on-time.
The Thermal Machinery Group at NIST has also contributed to the topic
of transient performance of vapor compression systems. Specifically, Mulroy
studied cyclic refrigerant migration as function of on-time (Mulroy and Didion,
1985). This is the first paper to experimentally determine both the capacity and
charge of the system components as functions of time. Mulroy observed that at
start up 56% of the refrigerant was in the evaporator while 11% was in the
condenser. However, at steady state 11% of the refrigerant was in the
evaporator while 46% was in the condenser, which agrees with the results from
Tanaka. In a related paper, Mulroy investigates the effect of control strategies
and refrigerant migration on cyclic performance (Mulroy, 1986). In this paper
62
-------
the capacity, power, arid efficiency are all plotted as functions of time for different
system configurations and different modes of operation, R-22 was the refrigerant
used in both of these papers.
Katipamula was the only researcher found to have studied the effects of
moisture removal on the transient performance of a AC/HP (Katipamula, 1989).
Several observations were made regarding the moisture removal aspects of
transient operation. One such observation was that moisture was actually
added to the air stream at start-up and that dehumidification did not begin for 60
to 150 seconds after start up. Furthermore, it was found that the sensible and
latent capacity reached steady state between 6 and 15 minutes. It was also
noted that the suction and discharge pressures reached steady state within 75
seconds. Katipumala concludes that the transient performance of a heat pump
is sensitive to indoor temperature, outdoor temperature, indoor relative humidity,
cycling rate, and percent on-time.
More recently Votis published transient data on R-22 operating in a
residential heat pump (Votis et al., 1992). Several different system
configurations were examined. For each configuration the temperature,
pressure, power and capacity versus time were presented. The significant
findings of this work were that the transient performance was sensitive to
63
-------
refrigerant migration, the thermal mass of the heat exchangers, and the
expansion device.
To date there are no published experimental studies on the transient
performance of refrigerant mixtures. This is attributed to the recent arrival of
refrigerant mixtures to commercial applications. Hence, one of the goals of this
project is to address this void, since it is not known how a refrigerant mixture will
perform transiently. To this end the temperature, pressure, capacity and power
are measured as functions of time for both pure components and mixtures.
Furthermore, the concentration of a refrigerant mixture is measured as a function
of time during the operation of a AC/HP. Equally important are the steady state
experiments that are conducted so that the steady state performance of these
fluids can be quantified. Also studied experimentally is how the system performs
with different expansion devices and vapor to liquid line heat exchange. The test
procedures and conditions are described next for each of the tests conducted.
7.3 Test Description
Most of the tests conducted come directly from ASHRAE standard 116 and ARI
210/240, which are thorough test procedures developed for rating residential
heat pumps (ASHRAE, 1983), (ARI, 1989). The tests and test conditions are
shown in Table 7.1. All of the tests in Table 7.1 are derived from ASHRAE
64
-------
standard 116 and ARI 210/240 except for D' and 47C. The motivation for all of
the tests is described next.
There are five cooling mode tests which are used to characterize the
performance of the system under a wide range of conditions. Tests A, B, and C
are all steady state cooling mode tests with different ambient conditions. Tests
A and B have a relatively high inlet humidity to the evaporator that causes water
to condense on the heat exchanger. The condensation presents no significant
problems for steady state tests, however the same is not true of cyclic or
transient tests. In order to determine the air side capacity the humidity entering
and leaving the indoor heat exchanger must be measured, it is not practical to
do this transiently. This is the motivation for Test C, which serves as a steady
state reference for tests D and D' which are cyclic. The difference between the
D' test and the D test is the on and off-time. The D test has a 6 minute on-time
with a 24 minute off-time while the D' test has 30 minute on-time and 30 minute
off-time. This test was added to the ASHRAE battery of tests so that the entire
transient is studied and to study the effect of on-time on cyclic performance.
As in cooling, there are five heating mode tests used to evaluate the heat
pump's performance under a wide range of conditions. Tests 47S and 17L are
steady state heating mode tests with different ambient conditions. Tests 47C
65
-------
and 47C are analogous to the cooling mode tests D and D'. The defrost test is
used to quantify the effect of the defrost cycle on the heat pump's performance.
7.4 Test Procedure
For all of the tests the indoor and outdoor chambers are operated at the
prescribed conditions for no less than one hour. If it is the first test for the
refrigerant in question, liquid refrigerant is charged into the system after the
system is evacuated for over eight hours. The tests are performed in the order
they appear in Table 7.1, except for D' and 47C which are preformed last. All
of the data reported here are from tests conducted at either steady state or cyclic
equilibrium. The procedure for the steady state tests is discussed next, followed
by an outline of the procedures utilized for the cyclic tests.
For steady state tests A, B, C, and 47S the AC/HP is operated for
approximately 30 minutes prior to data collection. After which the data are
collected for at least one hour at one minute intervals. For the 17L test the heat
pump is run until a defrost is automatically initiated and terminated. The data are
then collected at one minute intervals following the termination of the defrost
cycle. This is done because frost formation causes the performance of the
AC/HP to change with time. After the defrost in the 17L test, the performance of
the AC/HP improves for approximately 10-15 minutes, as the AC/HP reaches
66
-------
steady state. Then there is a period of time, roughly 20 minutes, where the
performance does not change. It is during this period that the data are used to
evaluate the performance of the AC/HP. Subsequently, the performance of the
AC/HP degrades as frost accumulates on the evaporator.
In terms of conducting the experiments, the primary difference between
the cyclic tests and the steady state tests is the test length. This occurs because
the cyclic tests take longer to reach equilibrium than do the steady state tests.
Typically, two complete on/off cycles are required to reach cyclic equilibrium.
Prior to starting a cyclic test the AC/HP is run continuously for one hour at the
cyclic test conditions. After which the three hour cyclic test begins. The data are
collected in five second intervals as soon as the test is started. For the D and
47C tests this provides information on six complete cycles, while for the D1 and
47C this provides information on three complete cycles.
7.5 Charge Optimization
Charge optimization is the method used to determine the appropriate amount of
refrigerant to put into the system. The initial amount of refrigerant charged into
the system is determined by adding refrigerant into the system while it is running.
This is done until the sight glass in the liquid line shows that the refrigerant is
nearly subcooled. Then a test is run and the COP of the system is measured.
67
-------
After which, roughly 0.1 to 0.3 kg of refrigerant is added to the system and a test
is conducted again to evaluate the COP. This procedure is repeated until the
COP of the system starts to decline. This typically requires 5-15 tests.
Afterwards, a charge optimization curve is generated by plotting the COP versus
the refrigerant charge.
7.6 Concentration Measurement
The R-32/134a and R-407C mixtures are used to determine how the circulated
refrigerant concentration changes with time. After completing the entire battery
of tests in Table 7.1, tests D' and 47C were repeated so that the concentration
could be measured as a function of time. It should be noted that the refrigerant
can only be sampled five times per test, but between 10 and 15 concentration
measurements at one minute intervals are desired. Therefore, it is necessary to
run two or three separate tests to obtain the quantity of data desired. One
approach to obtain the desired amount of data would be to sample the refrigerant
consecutively in time with each subsequent test resuming where the last test
terminated. However, the approach used here is to stagger the sample times for
multiple tests. By taking data in this manner systematic changes in
concentration can not be attributed to experimental error. The procedure for
measuring the concentration for each test is now discussed.
68
-------
The first step for the first test at a given test condition is to clean the
sample bottles which are described in Section 4.2. The sample bottles are
cleaned with acetone to remove any residual oil that might accumulate in the
bottles. To do this the bottles are evacuated and then approximately 100 ml of
acetone is driven in by atmospheric pressure. Then the sample bottle is violently
agitated for approximately 30 seconds. After which, nitrogen is charged into the
bottle to about 350 kPa. The nitrogen is used to rapidly motivate the acetone-oil
solution out of the sample bottle. The bottle is turned upside down and the valve
is opened to allow the acetone-oil solution to flow out. Afterwards the bottles are
attached to a vacuum pump for several hours to remove any remaining acetone.
The bottles are now ready to be attached to the solenoid valves. After the
bottles are attached to the valves they are once again evacuated to remove any
air that is on the bottle side of the valve. The manual valve, which is between
the solenoid valve and the sample bottle, is now closed to establish the metered
volume for the refrigerant. The volume of the sample bottles is sized such that
after the refrigerant expands into the bottle only vapor exists.
The cyclic test is now run according to the procedure described in Section
7.4. However, now the solenoid valves are activated. The data acquisition is
programmed to activate the solenoid valves for the last on-cycle in the staggered
69
-------
manner described previously.
After the solenoid valve has opened and closed the manual valves are
opened to allow the contents of the metered volume to occupy the sample bottle.
Then the manual valves are closed. The sample bottles are removed and the
solenoid valves are capped. The sample bottles are shaken to ensure that there
is not liquid in the bottle.
The sample bottles are now taken to the gas chromatograph where they
are chromatographed in the order in which they were sampled. Prior to gas
chromatographing the samples, the gas chromatograph is run for one hour. After
which a sample of known concentration is gas chromatographed to ensure that
the device is working properly. Each of the sample bottles is gas
chromatographed three times.
This entire process is repeated for each test until the desired number of
data points are obtained, After the concentration is measured for D' and 47C'
the refrigerant in the AC/HP is recovered. This refrigerant is then gas
chromatographed and compared to the charged concentration. This is done to
ensure that the concentration in the system has not changed through the course
of the test.
70
-------
7.7 References
ARI, 1989, "Standard for Unitary Air-Conditioning and Air-Source Heat Pump
Equipment", Standard 210/240.
ASHRAE, 1983, "Methods of Testing for Seasonal Efficiency of Unitary Air-
Conditioners and Heat Pumps", ASHRAE Standard 116-1983,
Katipamula, SM 1989, "A Study of the Transient Behavior During Start-up of
Residential Heat Pumps", Doctoral Dissertation, Texas A&M.
Mulroy, W., Didion, D., 1985, "Refrigerant Migration in a Split Unit Air
Conditioner", ASHRAE Transactions, Vol. 91, pp. 193-206.
Mulroy, W., 1986, "The Effect of Short Cycling and Fan Delay on the Efficiency
of a Modified Heat Pump", ASHRAE Transactions, Vol. 92, pp. 813-826.
Murphy, W.E., Goldschmidt, V.W., 1984, "Transient Response of Air
Conditioners-A Qualitative Interpretation Through a Sample Case", ASHRAE
Transactions, Vol. 90, pp. 997-1008.
Tanaka, N., Ikeuchi, M., Yamanaka, G., 1982, "Experimental Study on the
Dynamic Characteristics of a Heat Pump", ASHRAE Transactions, Vol. 88, pp.
323-331
Votis, P., Tassou, A., Wilson, D., Marquand, C., 1992, "Dynamic Characteristics
of an Air-to-Water Heat-Pump System", International Journal of Refrigeration,
Vol. 15, pp. 89-94.
71
-------
TABLE 7.1 Test Conditions
Test
Indoor dry-
bulb °C (°F)
Indoor wot-
bulb °C (°F)
Outdoor dry-
bulb °C (°F)
Outdoor wat-
bulb "C (°F)
Comments
A
27 (80)
19 (67)
35 (95)
24 (75)
cooling, steady state
B
27 (80)
19 (87)
28 (82)
18 (65)
cooling, steady state
C
27 (80)
<14(57)
28 (82)
18 (65)
cooling, steady state
D
27 (80)
<14 (57)
28 (82)
18 (65)
cooling, cyclic,
6 min on/24 min off
D'
27(80)
<14 (57)
28 (82)
18 (65)
cooling, cyclic,
30 min on/30 min off
47S
21 (70)
<16 (60)
8(47)
6(43)
heating, steady state
47C
21 (70)
<16 (60)
8(47)
6(43)
heating, cyclic,
6 min on/24 min off
47C
21 (70)
<16 (60)
8(47)
6(43)
heating, cyclic,
30 min on/30 min off
Defrost
21 (70)
<16 (60)
1.7(3 5)
0.6 (33)
heating, cyclic
17L
21 (70)
<16 (60)
-8 (17)
-9 (15)
heating, steady state
72
-------
Chapter 8
Data Collection
8.1 Introduction
In the process of running a test there are two distinct phases relevant to data
collection. The first phase is to collect the raw data, data acquisition. The
second phase is to calculate the desired quantities, data reduction. Both the
data collection and the data reduction phases are described in detail below.
8.2 Data Acquisition
The data acquisition system can be divided into two major components, hard-
ware and software. The hardware which collects the data is Hewlett Packard's
HP 7500 series B. The data acquisition system consists of a mainframe unit, two
mainframe extension units, and an IBM compatible computer.
To increase versatility, the data acquisition system has different cards
which are used to accomplish different tasks. For the work here, the cards which
are plugged into the mainframe and extension units are either thermocouple,
voltage, or relay cards. The thermocouple cards are used for measuring the
temperature of T type thermocouples. These cards utilize hardware junction
73
-------
temperature compensation so that an ice bath or other reference temperature is
not required. The voltage cards are used to measure the voltage outputs from
the various transducers (i.e. refrigerant pressure transducers, mass flow meters,
dew point meters, etc.) which are implemented throughout the course of this
work. The relay card allows the data acquisition system to control electric
components. Specifically, the relay card is used to control another set of relays
which require low power and are activated with 24 VAC. This is done to
minimize the wear on the data acquisition system's relays and to keep high
current electric lines away from the measurement system. The second set of
relays are used to control the AC/HPs. The second set of relays also activate
the solenoid valves which sample refrigerant from the AC/HP.
The other aspect of data acquisition is the software. The software
controls the data acquisition hardware. The software was written in a visual
programming language called HPVEE. In HPVEE, programs are written using
flow chart icons which are manipulated to create the code. The demands of this
research are such that two separate programs were written. This is necessary
since the transient experiments require high data collection rates while the
steady state experiments require a high level of real time output.
The program tailored to steady state tests has many features. This
74
-------
program is written so that both AC/HPs can be tested simultaneously.
Obviously, this allows for optimal use of the test facility. This program plots one
of 10 pre-programmed graphs or a user-defined graph while displaying all of the
data collected. This is done for both AC/HP's simultaneously in real time. These
features allow for rapid and easy determination of steady state. Furthermore,
they serve as powerful error detection tools. In addition, this program is also
responsible for turning off and on the AC/HPs and controlling the mode of
operation (i.e. heating, cooling, or defrost). This program also has an emergency
termination algorithm which shuts off a AC/HP if temperatures, pressures or
powers get too high or low. This feature is valuable since it allows the system
to be operated without anyone present. The final aspect of this program is that
it allows the data to be written to a file at a user prescribed collection interval.
Unfortunately, this program requires significant computational effort and as a
result the data collection frequency is compromised. This program can scan
through all of the 252 data points associated with both AC/HP's every 20
seconds. This data collection frequency is not sufficient for the transient
measurements. As a result a second program was created.
The second program is more straightforward than the previous program
outlined above. This program has a narrower focus since it is only used for the
transient tests. The second program simply records all of the data for one
75
-------
AC/HP. This data is displayed on the screen and simultaneously written to the
hard disk. The program also controls the mode of operation of the AC/HP, as
well as the solenoid valves responsible for sampling the refrigerant. The most
rapid data collection interval with this code is 5.0 seconds.
8.3 Data Reduction
The primary goal of data reduction is to determine the AC/HP's efficiency and
capacity for each test conducted. The steady state tests and the cyclic tests
require different techniques to analyze the data. However, for both types of tests
the data undergo a similar analysis to ensure that the data meet or exceed
ASHRAE 116 requirements.
The data from the steady state tests are truncated by a BASIC program
that includes only the data necessary for the calculation of cycle state points,
refrigerant side capacity, and air side capacity. This truncated file is the input
file for a FORTRAN code that utilizes property data, previously discussed in
Chapter 3, to calculate the air and refrigerant side capacities. This code also
calculates the AC/HP's COP and the error between the refrigerant and air side
capacities. This error between capacities is termed the energy balance error.
A similar procedure is used for the defrost test. The only difference is that the
data are averaged over the period between the end of the first defrost and the
76
-------
end of the next defrost.
The data from the cyclic tests are analyzed differently than the other tests.
The data are imported into a spreadsheet file, which contains the algorithm to
calculate variables relevant to the cyclic tests. These variables include the
cooling load factor (CLF) for test D, the heating load factor (HLF) for the 47C
test, and the degradation coefficient (CD) for both tests. All of these quantities
are based on the air enthalpy method. The air enthalpy method is used because
during the transient portion of the test it is not possible, with this test facility, to
measure all of the refrigerant side parameters (i.e. quality) that are necessary for
determining the transient capacity. The CLF and the HLF represent the fraction
of the steady state capacity delivered during the cyclic test and are defined in
Equation 8.1.
The evaluation of Equation 8.1 is not straightforward. Ideally, the
numerator of Equation 8.1 would be evaluated from 0 to 30 minutes, not from 0
to 6 minutes. This is done to account for the effects of thermal energy stored in
the mass of the heat exchanger and the components of the indoor unit. There
lopt
8.1
CLF and HLF = —
Q -At
^ss (on + off)
7?
-------
is a problem with this test procedure when using the closed indoor test loop. The
AC/HP would normally shut the indoor fan off at some point during the
compressor off-time. Unfortunately, in order to maintain precise air temperature
and air flow rate at the beginning of the next on-time the indoor fan must remain
on. Otherwise, it is not possible to accurately reproduce the air temperature and
airflow rate entering the indoor unit for each cycle. Since the fan remains on it
consumes energy and increases the heat transfer rate, which increases the rate
at which the thermal energy stored in the mass of the equipment is yielded to the
air. The increased heat transfer rate is beneficial since it allows more of the
stored thermal energy to be utilized. Unfortunately, the increased fan power is
unwanted since it lowers the system efficiency. Therefore, to compare each fluid
and system configuration on equal terms the limit of integration, topt, used in
Equation 8.1 is that which produces the lowest coefficient of degradation. The
coefficient of degradation quantifies the overall effect of cycling on system
capacity and efficiency. Equation 8.2 shows the definition of CD for the D test.
COP
/a __ v-'^f eyes
" COPss 8.2
D (1 CLF)
The COPcyc is obtained by integrating the capacity and power over the same time
limits as Equation 8.1. When evaluating the cooling CD, the COPss is the value
78
-------
from the C test. The only differences between the heating and cooling CD's are
that the 47S test is used to determine the COPss and that the CLF is replaced by
the HLF.
Once the CD is calculated for tests D and 47C, and the COPs and
capacities are evaluated from the other tests the seasonal performance can be
determined. The cooling seasonal performance factor (CSPF) and the heating
seasonal performance factor (HSPF) are analogous to COP. The difference is
that the seasonal performance factors are the ratio of the cooling or heating
capacity required for an entire season over the power required for an entire
season. In this technique the efficiency and capacity data collected from all of
the cyclic and steady state tests are used to determine the AC/HP seasonal
performance. The seasonal performance is calculated utilizing the bin technique
described in ASHRAE Standard 116 (ASHRAE, 1983). Each bin represents a
2.8°C (5°F) temperature range and the number of hours the climate in question
is within that temperature range. There is one standard climate for the cooling
season which represents an average of U S climates. There are six standard
climates for the heating season (Parker et al.,1980). The least and most severe
heating climates are analyzed to determine the maximum and minimum
performance of the AC/HP.
79
-------
8.4 References
ASHRAE, 1983, "Methods of Testing for Seasonal Efficiency of Unitary Air-
Conditioners and Heat Pumps", ASHRAE Standard 116-1983.
Parker, W., Kelly, G., Didion, DA, 1980, "Method of Testing, Rating and
Estimating the Seasonal Performance of Heat Pumps", NIST NBSIR 80-2002.
80
-------
Chapter 9
Simulation of AC/HP Components
9.1 Introduction
The computer simulation can be thought of as a composite of several distinct
component simulations. In other words, the AC/HP simulation is comprised of
a compressor model, a condenser model, an expansion device model, and an
evaporator model. In this work, the condenser and evaporator models are
grouped into the broader class of heat exchanger models, since they are similar
in every respect except the direction of heat transfer. The AC/HP simulation will
first be described in terms of its components, starting with the compressor. After
the individual components are described, the details of the overall simulation and
the numerical techniques will be addressed in Chapter 10. First, a few terms
relevant to AC/HP modeling will be defined.
9.2 Definitions
The methods used for resolving time dependant phenomena fall into two
computational families. Most time dependent computational methods can be
classified as either explicit or implicit. Explicit methods are methods which
calculate the variables of interest at the next time step directly from information
81
-------
from the previous time steps. On the other hand, implicit methods require
information from the next time step to proceed to that time step. Hence, an
equation solver is required to solve for the variable of interest at the next time
step. If the problem is linear, which is often the case, a diagonal matrix solver
can be used. For the case of non-linear equations an iterative method is
required. Each of these families of methods has its advantages and
disadvantages. The explicit methods are generally easier to code and require
less computational effort per time step. This notwithstanding, explicit methods
suffer from restrictive stability and convergence conditions, which translates into
small steps in time. Small time steps are undesirable because the time required
to compute the domain of interest is inversely proportional to the step size.
Implicit methods can be thought of as the converse of explicit methods in many
respects. Implicit methods are more tedious to program and require significant
computational effort for each time step. However, implicit methods generally
have large stability and convergence regions. The consequence of this is that
they are able to take very large steps in time. In most cases they are able to
calculate the steady state solution directly. For the research conducted here, the
advantages associated with implicit methods outweigh their disadvantages.
Hence, implicit methods are used exclusively in the AC/HP simulation.
In this work there is another distinction made between two types of
82
-------
simulations. A simulation is classified as either a lumped or a distributed
parameter model. A lumped parameter model simply treats the device in
question as a black box and generally utilizes extensive empirical data to correct
for internal effects. A distributed parameter model directly accounts for the
internal physics of the device in question and as a result it is generally more
complex and more accurate. These points can be made more clear by
examining heat exchanger modeling. A lumped parameter heat exchanger
model would consist of simply one control volume with the heat transfer being
approximated through an empirical UA. The output from this model would simply
be the outlet state of the refrigerant and air. A distributed parameter heat
exchanger model would numericaliy solve the continuity, momentum, and energy
equations throughout the heat exchanger. The output from a distributed
parameter model would contain the refrigerant temperatures, pressures, and
qualities throughout the heat exchanger, as well as the air temperature. Hence,
due to the increased accuracy and reduced dependance on empirical data, all
of the component models created for this work are distributed parameter
models.
9.3 The Compressor Model
An accurate compressor model must take into account the many processes and
interactions within a real compressor. In this case, a compressor is a mechanical
83
-------
device which compresses low pressure refrigerant vapor. In so doing the
compressor not only elevates the pressure of the refrigerant but it also elevates
the temperature, which induces heat transfer between various components within
the compressor. In addition to the heat transfer within the compressor, there is
also complex mechanical motion. Specifically, the motion of the suction and
discharge valves as well as the motion of the piston must be accounted for. The
manner in which other researchers handle the relevant physics within a
compressor is discussed next, followed by a detailed outline of the compressor
model used here.
9.3.1 Previous Compressor Models
There are many levels of complexity among the compressor models in the
literature. They range from those that solve the complete set of multi-
dimensional energy, momentum, and continuity equations (Recktenwald et al,,
1986) to those using curves fit to experimental data (Murphy and Goldschmidt,
1985). Compressor models which are potentially suitable for a AC/HP
simulation are discussed. A suitable compressor model is defined as one that
represents the significant physics and is computationally inexpensive. In order
to get a feel for the range of compressor models currently in use. compressor
models from existing AC/HP simulations are discussed first, followed by
compressor models developed solely for the purpose of studying compressors.
84
-------
One of the most simple and most frequently used approaches to
compressor modeling is to curve fit steady state experimental data. One curve
fit method is to define the mass flow rate, discharge state, and compressor power
as functions of the suction conditions and the pressure ratio (Murphy and
Goldschmidt, 1985). Although this method is very accurate for a given fluid at
steady state, it neglects transient effects and is not generally applicable to
refrigerants other than the one it was developed for.
Welsby adopted a more general compressor model for a AC/HP
simulation (Welsby et al., 1988). In this lumped parameter model the
compressor rotates at a constant RPM and does not account for volumetric
efficiency. In other words, the volume flow rate though the compressor is
constant. The compressor power and discharge state are determined by
assuming polytropic compression with a constant polytropic index. This model
neglects transient effects like thermal storage and variable RPM.
Another lumped parameter compressor model worth discussing is the one
developed by Parise for steady state simulations (Parise, 1986) . This model
has seen widespread use in both steady state and transient modeling of vapor
compression heat pumps. The goal of this lumped parameter model can be
broken down into two objectives. One objective is to determine the refrigerant
85
-------
flow rate and the other objective is to determine the outlet state. The refrigerant
flow rate is determined by Equation 9,1.
The terms in the parenthesis represent the theoretical maximum mass flow rate.
This term is multiplied by the volumetric efficiency, nv, which takes into account
some of the volumetric losses associated with the compression process. The
volumetric efficiency is defined as :
The volumetric efficiency accounts for re-expansion of the refrigerant in the
clearance volume and the density change of the refrigerant prior to entering the
compressor. Cv is an empirical volumetric coefficient which accounts for the
heating and pressure drop of the refrigerant prior to entering the compressor,
it is assumed in Equation 9.2 that the compression process is polytropic of
constant index n. The outlet state is determined by using the polytropic
relationship, as seen in Equation 9.3.
9.1
nv = 1 + cvn - {
9.2
Pout
9.3
86
-------
The compressor power can be determined directly by Equation 9.4.
W = ™ ~ 1]
9.4
This compressor mode! is attractive because it accounts for several non-
idealities in a very simple form.
The most complex compressor model used in an AC/HP simulation is that
by MacArthur (MacArthur, 1984). This is a distributed parameter compressor
model with seven different state points. The compression process is assumed
to be polytropic as well as isentropic. Heat transfer between different
compressor components is achieved by utilizing constant heat transfer
coefficients. The effect of thermal storage is accounted for. However, the
compressor is assumed to rotate at a constant angular velocity and the valve
dynamics are not accounted for.
The remaining compressor models were developed solely for the
purpose of studying steady state phenomena occurring within a compressor.
One of the earlier compressor models that did not implement the ideal gas
assumption is written by Ng (Ng et al., 1976). This model has three control
volumes in which the continuity, momentum, and energy equations are solved.
87
-------
The compression process is assumed to be adiabatic. The refrigerant flow is
assumed to be one dimensional. Furthermore, each valve is modeled as a one
dimensional spring, mass, and damper system.
In the Perez-Segarra model the one dimensional continuity, momentum,
and energy equations are solved for 19 different control volumes within the
compressor (Perez-Segarra et al., 1994). The heat transfer between control
volumes is simplified by assuming that the compressor components are at a
constant temperature, which are input variables to the simulation. This model
does not account for valve dynamics. The strength of this mode! is that the state
of the refrigerant can be determined at many locations through out the
compressor. However, the simplified heat transfer is a considerable weakness.
Compared to the other models, Lio's model is quite complex (Lio et
al., 1994). As with the previous compressor model, this model solves the one
dimensional conservation equations for a compressor running at a constant
angular velocity. This model has eight control volumes. The increased
complexity arises from the modeling of the valves. A two dimensional finite
element approach is used to model the motion of the valves. This allows for a
more accurate representation of the valve harmonics which was the focus of the
model.
88
-------
From this literature review it is clear that there is a wide range of methods
available for modeling compressors. The methods used for the compressor
mode! developed here will now be discussed.
In the process of creating the AC/HP simulation two separate compressor
models were developed. The first model was very detailed in that it accounted
for the motion of the valves and piston. Furthermore, experimental validation
showed that the first model was also very accurate. However, as progress was
made on the overall simulation it became apparent that the original compressor
model was too computationally expensive as a consequence of modeling the
internal motion of the compressor components. For this reason, a second model
was developed, which borrowed heavily from the first. However, data from the
first model, Model I, was used to validate assumptions made in the second
compressor model, Model II. Therefore, it will be instructive to describe the first
model and then elaborate on the simplifications which resulted in the second
model.
9.3.2 Description of Compressor Model I
Compressor model I is a distributed parameter model that accounts for valve
dynamics and heat exchange between various components of the compressor.
This model is developed for hermetic reciprocating compressors. Figure 9.1
89
-------
shows the state points associated with this model. The compressor is divided
into nine refrigerant state points. State point 1 is the inlet to the compressor.
State point 2 is shell volume at suction pressure. State point 3 is the iniet to the
suction valve while state point 4 is the outlet to the suction valve. The
compressor swept volume is represented by state point 5. State points 6 and 7
are the inlet and outlet to the discharge valve, respectively. State point 8 is the
refrigerant in the discharge pipe and state point 9 is the refrigerant leaving the
compressor.
From Figure 9.2 it is possible to examine the various heat and mass
exchanges occurring within the compressor. The compressor is comprised of 3
control volumes. The first control volume, the refrigerant in the compressor shell,
is represented by state point 2. The refrigerant typically flows into this control
volume from state point 1 and flows out from state point 3. This control volume
is in heat exchange with the ambient surroundings, the compressor, the motor,
and the discharge pipe. The second control volume is represented by state point
5, the swept volume of the compressor. Typically, refrigerant flows into the
swept volume of the compressor from state point 4 and exits through state point
6. The refrigerant in the compressor swept volume is exclusively in heat
exchange with the compressor. The third control volume is the discharge pipe,
state point 8. The refrigerant generally enters from point 7 and leaves through
90
-------
point 9. The discharge pipe is only in heat exchange with the refrigerant in the
compressor shell.
Swept Volume
The first control volume analyzed is the compressor swept volume. Assuming
well mixed conditions the complete energy equation for the swept volume of the
compressor is:
dE _ 6 SW . K
__ _ g - ___ + mln{hjn + —- + gzm) - m0Jh0Ut + — + gzout) 9.i>
For this case, the potential and kinetic energy terms are negligible. The work for
the control volume is given by:
bW = p dV 9.6
Hence, Equation 9.5 becomes:
dUrv ¦ dVrv
= Ocomp - p~~^- + minhjn - mjiout 9.7
All of the heat transfer terms, including Qcomp, are discussed later. The volume
in equation 9.7, Vcv, represents the instantaneous volume of the compressor,
which is defined by Equations 9.8 and 9.9. L1( L2, and x are depicted in Figure
9.3.
91
-------
V = A ¦, x + V,
cv "piston clearance
9.8
x = U1 - cos(co £)] + LJ1-
N
1 - (—)2sin2((Df)3
^2
9.9
is the distance of the piston from top dead center. L., is the distance from the
center of the fly wheel to the center of the pin that attaches the connecting rod
to the fly wheel. I_2 is the connecting rod length.
In order to evaluate Equation 9.7 the mass flow rates must be determined.
The mass balance will aid in determining these quantities. The mass balance
takes the form:
dM,
CV
dt
= mjn - mout 9.10
In order to determine the inlet and outlet mass flow rates the valve geometry and
physics need to be addressed.
Valves
In the literature two techniques have been commonly used to calculate the mass
92
-------
flow rate through the valves. Both techniques assume that the flow is quasi-
steady and one dimensional but differ in their assumptions regarding the flow
path. Although neither of these assumptions are physically precise they have
produced accurate results when compared to experimental data (Woollatt, 1974),
(Schwerzler and Hamilton, 1972). Other types of analysis are excluded because
they have been found to be too computationally costly for a heat pump model.
" + ain Vi" + gzjn) - (— + aouf Vo^ + gzoul) - htotal 9.11
Pin 2 Pout 2
The first technique assumes that the flow is along a streamline (Perez-
Segarra et al.» 1994). This leads to the use of Bernoulli's equation, Equation
9.11, with pressure drop or head loss, htotal. The terms q and q are kinetic
energy coefficients, which account for the non-uniform velocity distribution in the
pipe cross-section. Since the flows studied here are turbulent the velocity profile
is nearly flat. Hence, a is close to unity and this is the value used here. The
pressure drop takes the form:
y2
^total ~ ^valve~2~ 9.12
Therefore, when neglecting potential energy effects and assuming the upstream
velocity is zero the outlet velocity becomes:
93
-------
2 / Pin Pout
9.13
The second technique often used to determine the outlet velocity is based
on the assumption that the flow follows a polytropic path (Benson and Ucer,
1972), The development starts with the assumption that the flow is reversible
and therefore isentropic. The irreverisibilities will be addressed latter. For these
assumptions, Euler's equation in one dimension with no change in elevation is
the appropriate governing equation, Equation 9.14,
The assumption that the flow is polytropic and isentropic yields the following
relation:
In this equation k is the isentropic expansion exponent. It is often convenient to
+ vdv - o
p
9,14
pvk - constant
9.15
* ¦ -*&). - ^
p pv
9.16
94
-------
express the isentropic expansion exponent in terms of the speed of sound, a.
Equation 9.15 is then differentiated with respect to p and substituted into
Equation 9.14 which is then integrated. The result of this operation is then
solved for the isentropic outlet velocity, V2s, Equation 9.17.
^2s ~
2k ^ Pin _ Pout ^ g
k - 1 Pin Pout
To account for non-ideal flow, Equation 9.17 is multiplied by a degradation
coefficient. Equation 9.18 is used to calculate the degradation coefficient, Cd,
and is taken from (Shapiro, 1953).
Cd = 0.1375sin[—(1 -1.5152-^)] + 0.7025 g is
2 Pout
Although Equations 9.13 and 9.16 start with two different assumptions the
results for both take a similar form. The major difference between the two
approaches is that Equation 9.17 has a greater dependance on fluid properties
and therefore is more computationally expensive. Although both techniques
provide similar results, since Equation 9.17 represents the theoretical maximum
velocity it is used in the model in conjunction with the degradation coefficient.
It should be pointed out that the greatest Mach number encountered for flow in
the compressor piping is 0.27. Generally, flows with M < 0.3 are considered
95
-------
incompressible (Fox and McDonald, 1985). However, in this case the Mach
number in the immediate vicinity of the valve could exceed 0.3. Hence the
assumption that the flow is incompressible is questionable.
The mass flow rate is simply the product of the outlet velocity, the flow
area, and the density. The flow area is a function of valve position which must
be known. The valve position is determined by solving the force balance,
Equation 9.19.
mvaive 2 ~ ~ ^v— ~ ky + mvgjveg ~ FpreioaCj + Aforce(&p) 9.19
The valve is assumed to be a one degree of freedom spring mass system. For
the purpose of determining refrigerant state and mass flow rate through the
valves this is a reasonable assumption (Doria and Bucciarelli, 1994). The force
area, A^, has been shown to be independent of valve position and hence not
equal to the flow area, Anow (Schwerzler and Hamilton, 1972). The flow area
must be related to the valve position. There is a plethora of diverse data on this
topic (Maclaren and Kerr, 1967), (Woollatt, 1974), (Davis, 1970), For the work
presented here the following correlation is used (Parise and Cartwright, 1985).
-V = ®in [H( >L_)i 9.20
Z yMAX
96
-------
Hence, the mass flow rate through a valve is simply Equation 9.21.
m2 = p A„0WV2sCd 9.21
Compressor Shell
The refrigerant in the compressor shell, state point 2, exchanges heat with the
compressor workings (T^p), the discharge port (Ts), and the ambient (Tamb). The
refrigerant in the compressor shell also receives heat from mechanical and
electrical losses. Assuming well mixed conditions, no work, and negligible
kinetic and potential energies the energy balance becomes Equation 9.22,
dU.
2 = Qshell ~ ™-h\ ~ 9.22
dt
Qm*rh Q r.nmn ^dis S.23
The volume of the shell, VsM, remains constant; therefore, the mass balance for
state point 2 is
d p9
Kmdf = m, - m3 9.24
The pressure drop in the compressor shell is not considered since it is negligible
97
-------
when compared to the pressure drop across the valves.
Discharge Pipe
The energy and mass balance for state point 8, the discharge pipe, are similar
to the those for state point 2,
dUs
dt
m7h7
mQhQ
Q.
dis
9.25
dp8
dt
m7 - m9
9.26
As in the compressor shell, the pressure drop in the discharge pipe is considered
negligible when compared to the pressure drop across the valves.
Compressor
it is assumed that the entire internal workings of the compressor are at one
temperature. This assumption is reasonable since the Biot number, Bi, is
approximately 0.02. Typically, conditions in which the Bi is less than 0.1 can be
evaluated using the lumped capacitance method (Ozisik, 1993). The internal
workings of the compressor are thermally communicating with two other control
98
-------
volumes. One is the refrigerant in the compressor swept volume, state point 5,
and the other is the refrigerant inside the shell of the compressor, state point 2.
The governing equation takes the form:
dT
Q comp = Q Q g 27
comp p ^ ^swept ^comp v'*"'
The temperature of the internal workings , Tcomp, can be determined analytically
in terms of the temperatures at state points 2 and 5, T2 and T5 respectively.
Motor and Compressor Speed
The compressor speed is determined by solving the torque balance around the
compressor shaft, Equation 9.28. The motor torque and efficiency of the motor
are determined by curve fitting data supplied by the manufacturer. Hence,
curves for efficiency and RPM are generated as a function of motor torque.
I dCO -j~ ~r~ .
~JT ~ motor ~ compressor 9.2o
99
-------
Heat Transfer Terms
There are six different heat transfer terms, all of which are diagramed in Figure
9.2. The heat transfer equations are given in Table 9.1. However, in order to
evaluate some of these equations, heat transfer correlations must be used. The
relevant correlations are discussed next. It should be noted that most of the heat
transfer correlations used in this work are developed for steady state, fully
developed flow. This is done in large part because correlations do not exist for
partially developed flow. The implication in using these equations is that at any
instant the flow in the compressor is assumed to be fully developed.
The heat transfer coefficient between a gas and the inside of a piston and
cylinder has received much attention (Recktenwald et a!., 1986), (Woschni,
1967). The literature review by {Fagotti et al., 1994) determined that the heat
transfer in the compressor is best modeled by an equation developed by Annand
(Annand, 1967).
Nu = OJORe070 9.29
In this correlation Re is based on the mean piston speed, Vp. It is worth noting
that for most cylinder heat transfer correlations there is some Prandtl number
dependance. For example, the correlation by Brok (Brok et al., 1980):
Nu = 0.053RemPr03 9.30
100
-------
The heat transfer in the discharge pipe is modeled with the Dittus-Boelter
equation, Equation 9.31 (Dittus and Boelter, 1930).
Nud = 0.023RefPr°3 9.31
Unlike the heat transfer between the refrigerant in the compressor's swept
volume and the compressor itself, the heat transfer between the refrigerant in the
shell and compressor has not been studied. The heat transfer coefficient in the
shell is idealized by treating the refrigerant as if it were flowing in an annulus.
In the absence of a more appropriate correlation, Equation 9.31 is used with the
hydraulic diameter.
The heat generated by mechanical losses is a result of bearing friction.
The power required to overcome the bearing friction is approximated by Petroffs
law. The result of this can be seen in Table 9.1. The first term is a sole function
of bearing geometry; r is the bearing radius, I is bearing length, and c is bearing
clearance. Since the compressor has many different bearing surfaces, this term
is replaced by a constant that represents the mean bearing geometry. The
second term is the absolute viscosity of the oil. Initially the viscosity was
calculated as a function of the shell temperature and pressure, since the shell
acts as the oil sump. However, it was found that better agreement with
experimental data over a wide range of conditions was obtained when assuming
101
-------
a constant viscosity. This may occur because the temperature of the oii between
the bearing surfaces does not change as appreciably as the oil in the sump.
Since the first two terms are constants they are lumped together and multiplied
by the last term. Therefore, the mechanical energy rejected to the compressor
shell is proportional to the square of the angular velocity of the compressor and
is independent of the bearing load.
The energy rejected to the compressor shell due to the electric motor is
also shown in Table 9.1. The difference between the electrical power into the
motor and the shaft power is the power rejected to the refrigerant in the
compressor shell. At steady state the shaft power is the sum of the compressor
work and the work to overcome the mechanical friction. However, in the
transient model of operation there is an additional term from the inertia of the
compressor.
9.3.3 Model I Verification
Some of the input data required by the model are considered either proprietary
or empirical and as such were not directly available. Therefore the input
parameters that were not available were adjusted within a reasonable range so
as to best reproduce the calorimeter data (Trane, 1994). Examples of input data
not available include, spring stiffness, spring mass, and valve damping
102
-------
coeffecient. The input data can be seen in Table 9.2.
Tables 9.3, 9.4 and 9.5 compare the steady-state results of Model 1 with
the calorimeter data for the compressor from AC/HP 2 with R-22 as the working
fluid. The compressor model appears to accurately represent the physics of the
compressor since the model does an excellent job of reproducing the results
over a wide range of conditions. However, it is interesting to note that the error
in the compressor power increases as the condensing temperature decreases.
The advantage of this model is that it enables the user to examine the
details of each revolution. For example, Figure 9.4 is the pressure-volume
diagram generated by this model. From this diagram one can get a sense of the
pressure drop associated with the valves and the amount of additional work due
to this pressure drop. Another item of interest is the refrigerant flow rate through
the valves. Figure 9.5 shows the mass flow rate through the discharge valve.
In this figure it is clear that there exists a momentary negative mass flow rate
through the valve. This has the effect of reducing the compressor's efficiency.
The cause of the negative mass flow rate can be seen in Figure 9,6, which is a
graph of the valve position. The results of the model indicate that a finite amount
of time is required to open and close the valves. The mass of the valves prohibit
them from closing immediately, hence refrigerant temporarily flows backwards.
103
-------
9.3.4 Description of Compressor Model II
As previously mentioned, Model I was abandoned in favor of a slightly less
complex model, Model II, Since Model I resolves the valve and piston position,
it requires exceptionally small time steps (At ~ 5e-5 seconds) which is the reason
the model is not suitable for the AC/HP simulation. Model II was developed so
that much larger time steps could be taken (At ~ 5e-2 seconds). Two
simplifications had to be made to achieve this increase in step size.
The first simplification is in the compression process. In Model II the
compression process is assumed to be isentropic. This assumption is validated
by examining the results from Model I where this assumption was not made.
Figure 9.7 shows the p-v diagram for the AC/HP2's compressor with an isentrope
plotted along the compression process. From Figure 9,7 it is clear that the
compression process is nearly isentropic. This occurs because once the piston
reaches bottom dead center the primary irreversibility is caused by heat transfer,
since there is no mixing or further pressure drop. From Model I the heat
transferred between the refrigerant in the cylinder and the cylinder itself is less
than 1% of the total energy required to compress the refrigerant. This low heat
transfer is the product of two terms. One is the low heat transfer coefficient (h^p
~ 200 W/m2K) and the other is the relatively small surface area (Acomp « 2,5 e-3
m2). The second simplification was to replace the valve assemblies (i.e. valve,
104
-------
spring, mass, and damper) with an equivalent orifice. Hence, Equation 9.21 is
still used, but instead of calculating the instantaneous flow area, an effective flow
area is given. These two simplifications eliminate the need for small time steps
since the processes which occur on the smallest time scales are not modeled.
9.3.5 Model 11 Verification
The input parameters for model II were adjusted to obtain the best results for a
wide range of conditions and working fluids. The compressor model was
evaluated by comparing it to experimental data from AC/HP 1 and AC/HP2. The
experimental data used for this purpose is from the C and 47S tests described
in Chapter 7. Figure 9.8 is a plot of the measured compressor power versus the
compressor power predicted by the model. Figure 9.9 is similar to Figure 9.8
except that it is a graph of refrigerant flow rate. The compressor model I predicts
both of these quantities within ±5%.
9.3.6 Compressor References
Annand, W.J.D., 1967, "Heat Transfer in the Cylinders of Reciprocating Internal
Combustion Engines", Proc. Mech, Engrs., Vol. 117, pp. 973-996.
Benson, R.S., Ucer, A.S, 1972, "A Theoretical and Experimental Investigation of
a Gas Dynamic Model for a Single Stage Reciprocating Compressor v/ith Intake
and Delivery Pipe Systems", JNL Mech Eng. Science 14(4), pp. 264-279.
Brok, S.W., Touber, S., van der Meer. J.S., 1980, "Modeling of Cylinder Heat
Transfer-Large Effort Little Effect?", Purdue Compressor Conference
Proceedings, pp 43-50.
105
-------
Davis, H., 1970, "Effects of Reciprocating Valve Design on Performance and
Reliability", Proceedings, Instn. Mech. Engrs. Conf. on 'Industrial Reciprocating
and Rotary Compressors, Design and Operational Problems', pp. 9-23, London,
October.
Dittus, F.W., Boelter, L.M.K., 1930, University of California Publications on
Engineering, Vol. 2, p. 443, Berkeley.
Doria, M., Bucciarelli, M., 1994, "Numerical Analysis of the Dynamics of Reed
Valves Taking into Account Acoustic Coupling with the Fluid. Application to
Compressors for Domestic Refrigeration", Purdue Compressor Conference
Proceedings, pp. 229-234.
Fagotti, F., Todescat, M.L., Ferreira, R.T., Prata, A.T., 1994, "Heat Transfer
Modeling in a Reciprocating Compressor", Purdue Compressor Conference
Proceedings, pp.605-610.
Fox, R.W., McDonald, AT., 1985, "Introduction to Fluid Mechanics", John Wiley
and Sons, New York.
Lio, M., Doria, A, Bucciarelli, M.,1994, "Numerical Analysis of the Dynamics of
Reed Valves Taking into Account the Acoustic Coupling with the Fluid", Purdue
Compressor Conference, Vol. 2, pp. 229-234.
MacArthur, J.W., 1984, "Theoretical analysis of the dynamic interactions of vapor
compression heat pumps", Energy Conversion Management, Vol. 24, pp. 49-66.
Maclaren, J.F T., Kerr, S.V., 1967, "Analysis of Valve Behavior in Reciprocating
Compressors", paper 3.39, Proc. Xli, International Congress of Refrigeration,
Madrid.
Murphy, W.E., Goldschmidt, V.W., 1985, "Cyclic Characteristics of a Typical
Residential Air Conditioner-Modeling of Siart-Up Transients", ASHRAE
Transactions, Vol. 92, pp. 427-444.
Ng, E,, Tramschek, A., Maclaren, J., 1976, "Computer Simulation of
Reciprocating Compressor Using a Real Gas Equation of State", International
Journal of Refrigeration, Vol. 15, pp. 33-42.
Ozisik, M.N., 1993, "Heat Conduction", John Wiley and Sons, New York.
Parise, JAR., Cartwright, W.G., 1985, "Simulation of reciprocating
106
-------
compressors", Re vista Braesileiria de Ciencias Mecamicas, VII (2), pp. 129-152.
Parise, J.A.R., 1986, "Simulation of Vapour Compression Heat Pumps",
Simulation, Vol. 46, pp. 71-76.
Perez-Segarra, C.D., Escanes, F., Oiiva, A., 1994, "Numerical Study of the
Thermal and Fluid-Dynamic Behavior of Reciprocating Compressors", Purdue
Compressor Conference Proceedings, pp 145-150.
Recktenwald, G.W., Ramsey, J.W., Patankar, S.V., 1986, "Predictions of Heat
Transfer in Compressor Cylinders", Purdue Compressor Conference
Proceedings, pp. 159-174.
Schwerzler, D.D., Hamilton, J.F., 1972, "An Analytical Method for Determining
Effective Flow and Force Areas for Refrigeration Compressor Valving Systems",
Purdue Compressor Conference Proceedings, pp. 30-37.
Shapiro, A.H., 1953, "The Dynamics and Thermodynamics of Compressible Fluid
Flow", Vol. I, John Wiley and Sons, New York.
Trane Co., 1994, "Compressor Calorimeter Test Data Sheet".
Welsby, P., Devotta, $., Diggory, P.J., 1988, "Steady- and Dynamic-State
Simulations of Heat-Pumps. Part II: Modeling of a Motor Driven Water-to-Water
Heat-Pump", Applied Energy, Vol. 31, pp. 239-262.
Woollatt, D., 1974, "A Simple Numerical Solution for Compressor Valves with
One Degree of Freedom", Purdue Compressor Technical Conference
Proceedings, pp. 159-165.
Woschni, G., 1967, "A Universally Applicable Equation for the Instantaneous
Heat Transfer Coefficient in the Internal Combustion Engine", SAE Transactions,
pp. 3065-3083.
107
-------
Table 9.1 Heat Transfer Equations
Heat Transfer Term
Equation
Heat transfer from swept
volume to compressor
O = a h (T - T )
^swept swepr swept* 5 camp>
Heat transfer from
discharge pipe to shell
= t^T(T8 " ^
\ * h~2
Heat transfer from the
ambient to shell
A_
n 2 (T ~ t )
^amb ^ ^ v' a/rib 2>
hgmb
Heat transfer from
compressor to shell
Qcomp ^comfPcomp^comp
Heat generated from
mechanical losses
c
Heat generated from
electrical losses
Q,„c = (^sln/lK—L - D
'e/ec
108
-------
Table 9.2 Typical Input Data for the Compressor Model
Variable
Value
Motor Torque at Design RPM (N m)
7.46
Motor Effeciency at Design RPM
0.88
Motor Design RPM
3550
Piston Diameter (m)
4.7625e-2
Flywheel Radius (m)
1,0087e-2
Connecting Rod Length (m)
2.1000e-2
Clearance Volume (rn3)
1.08e-6
Suction Valve Spring Pre-load (N)
2.00
Suction Valve Spring Stifness (kN/m)
5.00
Suction Valve Damping Factor
1,0e-2
Suction Valve Maximum Displacement (m)
3,00e-3
Suction Valve Mass (kg)
3,00e-3
Suction Flow Area
1.85e-4
Discharge Valve Spring Pre-load (N)
2.00
Discharge Valve Spring Stifness (kN/m)
5.00
Discharge Valve Damping Factor
1,00e-2
Discharge Valve Maximum Displacement (m)
3,00e-3
Discharge Valve Mass (kg)
2.00e-3
Discharge Flow Area
1.85e-4
Heat Generated by Friction at Design RPM (W)
180
Mass of Compressor (kg)
7.75
Diameter of Shell (m)
0.22
Length of Shell (m)
0.15
Diameter of Discharge Pipe (m)
6.35e-3
Length of Discharge Pipe (m)
0.45
Compressor Inertia (kg«m2)
7.0e-2
109
-------
Table 9.3 Model I Verification
"37.8°C (100°F) Condensing arid 7.2°C (45°F) Evaporating Temperatures]
Parameter
Model I
Calorimeter
Error
Power (kW)
1.47
1.54
4.5 %
Mass Flow Rate (kg/s)
0.0423
0.04
3.6 %
Discharge Temperature
(K)
346.9
347
0.1 K
Table 9.4 Model I Verification
48.9°C (120°F) Condensing and 7.2°C (45°F) Evaporating Temperatures]
Parameter
Model I
Calorimeter
Error
Power (kW)
1.833
1.810
1.3%
Mass Flow Rate (kg/s)
4.03e-2
4.05e-2
0.5 %
Discharge Temperature
-------
Figure 9.1 Compressor Schematic
AMBIENT
1 : Compressor Suction
3: Cylinder
nlet
4;
Cylinder
Inlet Port
MECHANICAL AND
ELECTRICAL
LOSSES
SWEPT
6; Cylinder
VOLUME
Outlet Port
5: Cylinder
7: Cylinder
inside
Outlet
W
PIPE
18; Discharge
Pipe
9: Compressor Discharge
~
PISTON &
CYLINDER
¦
MASS
FLOW
ENERGY
FLOW
Figure 9.2 Compressor Energy and Mass Exchange
111
-------
Lt: Discrepancy
Figure 9.3 Compressor Detail
VOLUME (m3)
(Times 1E-5)
Figure 9.4 Pressure - Volume Diagram for Compression Cycle
112
-------
Top Dead Center angle (Degrees) j0p Dead Center
Figure 9.5 Discharge Valve Mass Flow Rate
O 60 120 180 240 300 360
Top Dead Center angle (Degrees) Top Dead Center
Figure 9.6 Position of Compressor Valves
113
-------
VOLUME (m3)
(Times 1E-5)
Figure 9.7 Compression Cycle with Isentrope
MEASURED POWER (W)
Figure 9.8 Compressor Model II Comparison with Experimental Power
114
-------
MEASURED MASS FLOW RATE (kg/s)
Figure 9.9 Compressor Model II Comparison with Experimental
Mass Flow Rate
115
-------
9.4 The Heat Exchanger Model
In order to develop a high quality cycle simulation for a vapor compression
system it is essential to have an accurate heat exchanger model. The heat
exchanger model not only significantly impacts the heat transferred but also the
amount of refrigerant in the system. The model developed here is a distributed
parameter heat exchanger simulation which solves the continuity, species,
momentum, and energy equations simultaneously for air to refrigerant heat
exchangers. The simulation is described and then it is compared to steady state
experimental results from four heat exchangers, each of which is tested with
three different refrigerants.
9.4.1 Previous Heat Exchanger Models
Prominent among the previous work on this subject is the tube by tube analysis
developed by Domanski (DomanskL 1989), (Domanski. 1991). In this analysis
a wet or dry evaporator was modeled by assuming the air passes directly from
one tube to another. The heat and mass transfer of each tube was evaluated
based on its own air and refrigerant inlet states. This allowed investigation of air
mal-distribution. The simulation, which was restricted to pure refrigerants, was
experimentally verified for wet coils and found to predict the overall capacity
within -8.8% and +23.3% (Chawalowski et al., 1989).
116
-------
Using the work by Domanski, Conde (Conde and Sutter, 1992) dev&loped
a similar simulation with several modifications. One of the improvements
included the ability to determine the refrigerant inventory within the evaporator.
This code was part of an overall steady state cycle simulation and was
experimentally verified with HCFC-22. The verification showed that the
simulation frequently over predicted the capacity by more than 10%.
Oskarsson et al. developed three steady state models for dry, wet, and
frosted evaporators (Oskarsson et al., 1990a), (Oskarsson et al., 1990b). AN of
the models assumed that the entire heat exchanger could be simulated with one
equivalent tube in cross flow. Furthermore, these models were developed for
pure refrigerants. One of the models divided the heat exchanger into three
different regions, the two phase region, the transitional region, and the
superheated region. The most complex model divided the heat exchanger into
finite elements. The simplest mode! was parametric and required one of the
other models to generate constants for the model. The three region model was
experimentally verified and found to predict the capacity of a dry coil between
+11.3% and -17.7%, Furthermore, the finite element model agreed with the
three region model within 2%.
More recently, Nyers has developed a transient evaporator model (Nyers
117
-------
and Stoyan, 1994). This model was developed for water to pure refrigerant
applications. This model was fully distributed and simultaneously solved the
continuity, momentum, and energy equations. The model was not compared to
any transient or steady state experimental data.
9.4.2 Description of the Heat Exchanger Model
The model is based on several assumptions which simplify the problem to a
manageable level. Since the heat exchanger simulation will be applied to a
cycle simulation, the main concern of the heat exchanger simulation is predicting
the refrigerant outlet conditions given refrigerant and air inlet conditions. The
intermediate air and refrigerant conditions are of interest, but are not the primary
focus of the simulation. A heuristic test of these assumptions will be the level of
agreement between the simulation and the experiment. With this in mind, the
assumptions are discussed next.
One assumption is that the rows of parallel refrigerant circuits behave as
one integrated circuit with an equivalent heat transfer area. This scenario is
approached when the refrigerant is uniformly distributed to each circuit in the
heat exchanger and the same heat flux exists on each tube. The error
introduced by assuming uniform heat flux on each circuit is minimized in
conventional cross-flow heat exchangers since the refrigerant flow is often
118
-------
circuited so that each circuit is exposed to a similar amount of inlet and outlet air.
A more detailed analysis would model the distribution system and use a tube by
tube analysis, however for a cycle simulation this approach would be too
computationally costly. In addition, the tube by tube technique assumes that the
air incident on each row does not mix or exchange energy with the air from other
rows. This assumption is questionable since the air is turbulent downstream of
the tube ( 1500 < Redalr< 2500). Furthermore, visualization with smoke
confirmed that the air mixes as it goes through the heat exchanger.
For the heat exchangers studied here, the air is assumed to be uniformly
distributed to each tube. This assumption is appropriate when the air-side
momentum change due to friction through the tube bank is significantly greater
than the momentum change associated with changing the direction of the air
flow. In other words, a large pressure drop across a tube bank that is inclined
to the incoming air will force the air to enter normal to the tube bank thereby
minimizing air distribution problems. This argument holds when dp/dx » pV2/r.
When one assumes the radius of curvature of the air is approximately equal to
the depth of the heat exchanger this expression becomes Ap » pV2. Air side
pressure drops are on the order of 40 Pa and the velocities are approximately
2-3 m/s. Hence, the pressure drop across the heat exchanger is in fact much
larger than the pressure drop associated with turning the air stream .
119
-------
These assumptions allow an inclined multi-circuit heat exchanger to be
treated as one tube in cross flow relative to the air. This is the approach taken
here. It should be noted that these assumptions while simplistic do not introduce
significant errors in the mean for heat exchangers that are not drastically
different from a single circuit cross flow heat exchanger. For example, let us
examine the refrigerant distribution. If there exists a refrigerant distribution
problem, where one tube receives more refrigerant than another, the two tubes'
heat transfer coefficients may be significantly different. However, the mean heat
transfer coefficient will most probably be somewhere between the two.
Therefore, the error in the overall heat transfer coefficient may not be greatly
affected. Furthermore, a more detailed analysis, which would inevitably use
empirical data, may provide additional insight into the heat exchanger but will
probably be no more accurate. This is due to the fact that the existing refrigerant
side empirical correlations (i.e. two phase heat transfer and pressure drop
correlations) have relatively large uncertainty.
The reduced problem, a tube in cross flow, has three aspects which must
be dealt with. These are the refrigerant, the tube and fins, and the air. These
are now considered.
120
-------
Refrigerant Side
There is a minimum of three conservation equations which must be solved for the
refrigerant side. They are continuity, momentum, and energy (Equations 9.32,
9.33, and 9.34 respectively) alf of which assume the flow to be one dimensional.
Since this simulation is developed for refrigerant mixtures, the continuity
equation is written in terms of each component.
= o 932
dt dZ
dG 3r x2G2 (1 -x)2G2n BP , dPs
+ [ + ] - + ( )friction 9.33
dt dz pa p.(1 a) 8z 8z fnc&0"
A d(pu) + d(rhh) _ q
cs dt dz
The momentum equation is written in terms of mass flux, G, and void fraction, a.
The void fraction represents the fraction of the cross-sectional area that is
occupied by vapor. The void fraction plays a large role in determining the
amount of refrigerant in the heat exchangers and in the system. Since there are
numerous correlations for the void fraction, the void fraction correlation which
best reproduces the charge optimization curve will be used. This is discussed
in more detail in Chapter 12. In order to solve the momentum equation empirical
121
-------
correlations are also required for the pressure drop due to friction. In the
literature there is no clear preference for the frictional pressure drop term. This
term is discussed later in this chapter. It should be noted that for the transient
simulation results presented in Chapter 12 the pressures throughout the heat
exchangers were assumed not to vary spatially. This was necessary because
resolution of the momentum equation would have required time steps 10 to 100
times smaller than those used, which would have caused the time required for
one test to exceed 300 hours. The exceptionally small step size is due to the
rapid acceleration of the refrigerant when the system is first started.
The energy equation, Equation 9.34, assumes axial conduction to be
negligible ( Pe,iq« 38,000, Pevap» 160,000 ). It is also assumed that dissipation
is negligible ( Pr,iq - 2.49 ,Prvap- 1.05 ). The thermal energy into the refrigerant
from the heat exchanger, Q, is determined from Equation 9.35.
0 = hr0fPhTref 9,35
Equation 9.35 describes the heat transfer between the heat exchanger
and the refrigerant. The href is the refrigerant side heat transfer coefficient. A
correlation is required for this term. As with the frictional pressure drop term,
there are many heat transfer correlations, none of which have received
significant praise in the literature. Consequently, several correlations are
122
-------
investigated. The heat transfer correlation will be determined by using the one
which provides the best agreement with steady state experimental heat
exchanger capacity data. Often, pressure drop correlations are developed
simultaneously with heat transfer correlations. Therefore, the pressure drop
correlation associated with the best heat transfer correlation will be used.
Tubes and Fins
The energy stored in the mass of the heat exchanger during transient processes
is governed by Equation 9,36. Equation 9.36 is developed for each node of the
discretized heat exchanger. It is assumed that each node is at one temperature
and that the axial conduction is negligible.
- mhTlm)air - {hAAT)ref g,36
(J • '
"r ~ MMrJr,) + + (AhMiTihZ) 937
Several heat transfer resistances associated with the tube and fin are
encountered in Equation 9.37. The first resistance is that of the tube itself. The
second resistance is the contact resistance between the fin and the tube. The
contact resistance is estimated from Sheffield's correlation (Sheffield et al.,
1989).
123
-------
The fin efficiency is calculated using the method described by Incropera
(Incropera and Dewitt, 1976).
Air Side
The air side is assumed to be incompressible and hydraulically fully developed.
Since latent loads are not considered, the only relevant conservation equation
is the energy equation, Equation 9.38. All of the terms relevant to Equation 9.38
have been explained except for the air side heat transfer coefficient, hair. Most
air side heat transfer correlations for different fin geometries are developed for
a single tube in cross flow, not a bank of tubes. Therefore, to overcome this
problem a method used by Domanski is adopted (Domanski, 1989). The air side
heat transfer coefficient is calculated for the given heat exchanger geometry
except that the fins are initially assumed to be flat. The correlation used for this
purpose is by Gray and Webb (Gray and Webb, 1986). A fin multiplier is then
used to account for non-flat or enhanced fins. The fin multiplier is simply the
heat transfer coefficient for one row of a given fin divided by the heat transfer
coefficient for one row of flat fins. For wavy fins, the fin multiplier is determined
using correlations developed by Trauger (Domanski, 1989). For spine fins, the
fin multiplier is determined using correlations developed by Rabas (Rabas and
Eckels, 1985).
124
-------
p/LC— + mCa — = - Q - 9.38
dt P dz
where Qair = (UA&Tlm)ajr
Discretization
The heat exchangers are represented by 100 nodes that are depicted in Figure
9.10. The number of nodes is determined by conducting a node size error
analysis. That analysis is presented later,
The differential equations are discretized using an implicit second order
accurate scheme (Nyers and Stoyan, 1994). Equations 9.39 and 9.40 are
examples of this approach. The advantage of this technique is that the steady
state solution can be determined directly. The disadvantage is that a system of
non-linear equations must be solved. The Newton-Raphson method is used to
solve this set of equations. The Newton-Raphson routine is considered to
converge when all of the residuals are less than 10 9 and is described in more
detail in Chapter 10.
3(py) = g(pV)H ipV)^ +(1 (plOy ^p^/+Q[Af(0-5-a )+A(2| 9,39
dt M At
125
-------
m = a;^-'pr,(i-aj(M.OMO.S-a^Az'l 9.40
dz At Az
9.4.3 Model Verification
The heat exchanger model is verified by comparing the simulation data to
experimental data taken at the heat pump test facility. The indoor and outdoor
heat exchangers of AC/HP1 and AC/HP2 are used to verify the simulation. The
configuration of these heat exchangers was described in Chapter 4. The inlet
air conditions were chosen so that neither condensate nor frost was formed on
the heat exchanger. The test conditions were the C and 47S tests described in
Chapter 7. The refrigerant and heat exchanger combinations that were tested
are shown in Table 9.6. Since the heat exchangers are components of a heat
pump, oil circulates with the refrigerant. The oil concentration was measured to
be less than 1.0% by mass. Therefore, it is not considered significant.
For the experimental data presented here the refrigerant and air enthalpy
methods agreed within 0.29-5.4%. The air flow rates for the outdoor coils are
assumed to be those reported by the manufacturers. To evaluate the accuracy
of this assumption, two steps were taken. The RPM of the fans were compared
to manufacturers data and found to agree within 2.4%. The other step taken was
to calculate the airflow rate based on the measured refrigerant side capacity and
126
-------
air side temperature difference. The reported flow rate agreed with the
calculated within 3.7%.
Five flow bailing and five condensation routines were used in the
simulation to determine which one works best with the simulation to predict the
overall heat transferred. The flow boiling heat transfer correlations, in order of
increasing simulation accuracy, were; (Gungor and Winterton, 1986), (Shah,
1982), (Kandlikar, 1991), (Chen, 1966) and (Jung, 1989) . Jung's and
Kandlikar's correlations were the only ones developed for mixtures. These
correlations are generally applicable to any binary mixture of halogenated
refrigerants. Since ternary mixtures are studied here two of the three
components with the most similar vapor pressures are treated as one component
for heat transfer calculations. In the case of mixtures of R-32/R-125/R-134a R-
32 and R-125 were grouped together. This does not introduce a significant error
since R-32 and R-125 have very similar vapor pressures in the region studied.
This conclusion is supported by Figure 9.11. The relative concentration of R-32
and R-125 remains nearly constant in the liquid and the vapor during a phase
change. Therefore, there is a negligible mass transfer resistance between these
two refrigerants to hinder heat transfer.
The simulation results are presented by plotting the measured capacity
127
-------
versus the capacity predicted by the model. The evaporation results are shown
in Figure 9.12, When using Jung's heat transfer correlation, which provided the
best results, the maximum difference between the simulation and experimental
results ranges from -5.0% to +3.8%. Figures 9.13 and 9.14 show the flow boiling
heat transfer coefficient versus position for the different heat transfer correlations
for R-22 and R-407C, respectively. It is interesting to note that in Figure 9.14 all
but one of the pure component heat transfer correlations predict lower heat
transfer coefficients than does Jung. This was not anticipated, since the pure
component heat transfer correlations do not take into account the degradation
in heat transfer associated with mixtures The other correlation developed for
mixtures, Kandlikar's, predict's heat transfer coefficients that are significantly
lower than that of Jung's. This is particularly interesting since Kandlikar used
Jung's data to develop his correlation.
The condensation correlations used, in order of increasing simulation
accuracy, were; (Tandon et al., 1986 ), (Chen et al,, 1987), (Traviss et al., 1973),
(Shah, 1982) and (Dobson et al., 1994). Dobson's correlation is developed for
R-32/R-125 (50%/50%) and Tandon's correlation was developed from R-22/R-12
data for several different concentrations. The remaining correlations were
developed for pure fluids. The condensation results are presented in a similar
format as the evaporation results and can be seen in Figure 9.15. For
128
-------
condensation Dobson's correlation worked the best with the current heat
exchanger model. The maximum difference between the simulation and
experimental results ranges from -3.3% to +7.2%. Typical condensation heat
transfer coefficients versus position in the heat exchanger are presented in
Figures 9.16 and 9,17. The increased error in condensation relative to
evaporation occurs for two reasons. One is that none of the condensation heat
transfer correlations were developed to be generally applicable for any mixture.
The other reason is that most of the correlations considered were developed for
the annular flow regime, which is discussed next.
The heat exchanger simulation was used to generate points on Baker's
flow map {Carey, 1992). Figure 9.18 is a flow map of R-22 in evaporation and
condensation, respectively. It is clear that for evaporation a large fraction of the
heat is absorbed in the annular regime. Furthermore, the other region, wavy
flow, is somewhat similar to annular flow in that the entire circumference of the
tube can be covered with liquid. Hence, annular flow correlations are often used
to approximate wavy flow. In any event, it should be noted that Jung's flow
boiling correlation is not restricted to the annular flow regime. At low mass fluxes
Jung's correlation reduces to pool boiling, which is attractive since pool boiling
represents the lower limit for all other boiling regimes. For condensation the flow
map shows that a significant portion of the heat is rejected in the slug and plug
129
-------
flow regimes. This is unfortunate since there exists very little heat transfer data
in these regimes. However, Dobson's condensation correlation for the annular
flow regime reduces to a single phase flow correlation as the quality is reduced.
Hence, Dobson's correlation is accurate in the limits of single phase flow and
annular flow. Therefore, it is used to approximate the heat transfer coefficient
for the flow regimes in between. Furthermore, it should be pointed out that the
sample size is not sufficiently large to state that the heat exchanger model will
predict the capacity for any heat exchanger configuration within the errors
reported in here. An estimate of the uncertainty associated with the heat
exchanger model is presented in Appendix A2.
The effect of increasing the number of nodes is seen in Figure 9.19. The
percent change in capacity and refrigerant mass are shown for R-22 using the
indoor heat exchanger from AC/HP 2 as a condenser and an evaporator. These
results are typical of other configurations, regardless of refrigerant or heat
exchanger. The results in Figure 10 show that 100 nodes is sufficient to
accurately model the processes occurring in the heat exchanger. The graph
shows that the mass of refrigerant in the heat exchanger is more sensitive to the
number of nodes than is the capacity of the heat exchanger. This occurs
because the void fraction equation is a function of the thermodynamic state of
130
-------
the refrigerant which in turn is a function of both the momentum and energy
equations. The lack of smoothness in the mass deviation curve for condensation
is the result of subcooling and a small number of nodes. The overall mass of
refrigerant in the heat exchanger is dramatically affected when a node is
determined to be liquid and there are only a few nodes.
The simulated refrigerant temperature profile of a condenser and an
evaporator are shown in Figures 9.20 and 9.21, respectively. In these figures
the temperature profiles of a pure component, R-22, and a mixture , R-407C, are
plotted. The temperature profile plots, Figures 9.20 and 9.21, show the
consequences of using zeotropic mixtures as compared to pure fluids. During
evaporation the pure fluid's temperature decreases by 1.18 °C owing to the
frictional pressure drop, while the mixture's temperature increases by 3.92 °C
because of the temperature glide associated with a zeotropic mixture. The
frictional pressure drop minimizes the temperature glide of the mixture for
evaporation. However, when condensing the pressure drop augments the
temperature glide of the mixture as seen in Figure 9.20. The temperature of R-
22 and R-407C changes by 2.33 °C and 6.86 °C, respectively, during
condensation. The temperature glide of the mixture represents a relatively large
fraction of the total temperature difference possible between the air and
refrigerant. As a result of this, if care is not taken, the temperature glide of the
131
-------
mixture can be either detrimental or beneficial to system performance. This point
is addressed next,
9.4.4 The Effect of Heat Exchanger Geometry
The simulation was run in counter and parallel flow modes, in addition to the
cross flow mode, to quantify the potential degradation or improvement
associated with R-407C's temperature glide. The air side values of area and
heat transfer coefficient were those determined from the cross flow case. This
was done so that the interaction of refrigerant properties and heat exchanger
geometry could be studied directly. The results can be seen in Tables 9.7 and
9.8 and Figures 9.22 and 9.23. Figures 9.22 and 9.23 are graphs of the
refrigerant and air temperatures as they vary throughout the length of the heat
exchanger for the different flow geometries. It should be noted that the kink in
the cross flow air temperature is due to the rapid change in the heat transfer
coefficient between the single and two phase regions. As expected, the mixture
suffers and gains the most from the parallel and counter flow modes,
respectively. It is interesting to note that the penalty of parallel flow is more
severe than the benefit of counter flow. This is a result of the large heat transfer
area of the cross flow baseline case. This translates into a high heat exchanger
efficiency which is difficult to improve upon with counter flow. However, this
large area allows for a pinch point to be approached in the parallel flow heat
132
-------
exchanger causing a significant degradation in heat exchanger performance.
Hence, the heat exchanger geometry must be considered when using R-407C
or any fluid with a significant temperature glide.
9.4.5 Heat Exchanger References
Carey, V.P., 1992, "Liquid Vapor Phase Change Phenomena", Hemisphere
Publishing Co., Washington, pp. 457.
Chawalowski, M., Didion, D.A., Domanski, P.A., 1989, "Verification of evaporator
models and analysis of performance of an evaporator coil", ASHRAE
Transactions, Vol. 95, pp. 1229-1236.
Chen, J.C., 1966, "Correlation for Boiling Heat Transfer to Saturated Fluids in
Convective Flow", Ind. Eng. Chem. Proc. Design and Dev.. Vol. 5, No.3, pp.322-
339.
Chen, S.L., Gerner, F.M., Tein, C.L., 1987, "General Film Condensation
Correlations", Experimental Heat Transfer, Vol. 1, pp.93-107.
Conde, M.R., Sutter, P., 1992, "The Simulation of Direct Expansion Evaporator
Coils for Air-Source Heat Pumps", International Congress of Refrigeration,
Montreal, pp. 1459-1463
Dobson, M.K., Chato, J.C., Wattelet, J.P., GaibeL J.A., Ponchner, M., Kenney,
P.J., Shimon, R.L., Villaneuva, T.C., Rhines, N.L., Sweeney, K.A., Allen, D.G.,
Hershberger, T.T., 1994, "Heat Transfer and Flow Regimes During
Condensation in Horizontal Tubes"., ACRC Project 37.
Domanski, P.A., 1989, "Evsim - An evaporator simulation model accounting for
refrigerant and one dimensional air distribution", NISTIR 89-4133.
Domanski, P.A., 1991, "Simulation of an evaporator with nonuniform one-
dimension air distribution", ASHRAE Transactions, Vol. 97, pp. 793-802.
Gray, D.L., Webb, R.L, 1986, "Heat Transfer and Friction Correlations for Plate
Finned-Tube Heat Exchangers Having Plain Fins, Proc. of Eighth Int. Heat
Transfer Conference, San Francisco, pp. 123-131.
133
-------
Gungor, K.E., Winterton, R.H.S., 1986, "A General Correlation for Flow Boiling
in Tubes and Annuli", International Journal of Heat and Mass Transfer, vol. 29,
pp. 351-358.
Incropera, F.P., Dewitt, DP., 1976, "Fundamentals of Heat and Mass Transfer",
John Wiley and Sons, New York.
Jung, D.S., 1989, "Horizontal-flow boiling heat transfer using refrigerant
mixtures", EPRI ER-6364, Project 8006-2.
Kandlikar, S.G., 1991, "Correlating Flow Boiling Heat Transfer Data in Binary
Systems", ASME National Heat Transfer Conference, Minneapolis, July.
Nyers, J., Stoyan, G., 1994, "A dynamical model adequate for controlling the
evaporator of a heat pump", International Journal of Refrigeration, Vol. 17, pp.
101-108.
Oskarsson, S.P., Krakow, K.L Lin, S., 1990a, "Evaporator models for operation
with dry , wet, and frosted finned surfaces Part I: Heat Transfer and Fluid Flow
Theory", ASHRAE Transactions, Vol. 96, pp. 373-380.
Oskarsson, S.P., Krakow, K.I., Lin, S., 1990b, "Evaporator models for operation
with dry , wet, and frosted finned surfaces Part II: Evaporator Models and
Verification", ASHRAE Transactions, Vol. 96, pp. 373-380.
Rabas, T.J., Eckels, P.W., 1985, "Heat Transfer and Pressure Drop Performance
of Segmented Extended Surface Tube Bundles", ASME Journal of Heat Transfer,
Vol. 75, pp. 1-8.
Shah, M.M., 1982, "Chart Correlation for Saturated Boiling Heat Transfer:
Equations and Further Study", ASHRAE Transactions, Vol. 88, pp. 185-196.
Sheffield, J.W., Wood, R.A., Sauer, H.J., 1989, "Experimental Investigation of
Thermal Conductance of Finned Tube Contacts", Experimental Thermal and
Fluid Science, Elsevier Science Publishing Co. Inc, New York, pp. 107-121.
Tandon, T.N., Varma, H.K., Gupta, C.P., 1986, "Generalized Correlation for
Condensation of Binary Mixtures Inside a Horizontal Tube", international Journal
of Refrigeration, Vol. 9, pp. 134-136.
Traviss, D.P., Rohsenow, W.M., Baron, A.B., 1973, "Forced convection
condensation inside tubes: A heat transfer equation for condenser design",
134
-------
ASHRAE Transactions, Vol. 79, pp. 157-165,
135
-------
Table 9.6 Refrigerants and Heat Exchangers Tested
Refrigerants
Heat Exchangers Tested with
Refrigerant
R-22
AC/HP1 Indoor, AC/HP1 Outdoor,
AC/HP2 Indoor, AC/HP2 Outdoor
R-407C
R-32/125/134a (23/25/52 wt%)
AC/HP1 Indoor, AC/HP1 Outdoor,
AC/HP2 Indoor, AC/HP2 Outdoor
R-32/125/134a (30/10/60 wt%)
AC/HP 1 Indoor, AC/HP1 Outdoor
Atochem
AC/HP2 Indoor, AC/HP2 Outdoor
Table 9,7 The Effect of Heat Exchanger Geometry on Heat Transferred1
(Cooling Mode)
Condensation
Evaporation
Flow
Capacity2
Capacity3
Geometry
R-22
R-407C
R-22
R-407C
% (kW)
% (kW)
% (kW)
% (kW)
Cross
100(11.38)
100(12.13)
100 (9.67)
100 (9.88)
Parallel
96.1 (10.94)
91.7 (11.12)
97.7 (9.45)
96.4 (9.52)
Counter
102.5
102.8
100.7
103.3
(11.67)
(12.47)
(10.21)
(10.21)
1 Air and refrigerant boundary conditions are based on experimental values. The air side heat
transfer coefficient, area and fin efficiency are based on the cross flow values.
2 Using unit 1 outdoor heat exchanger
3 Using unit 1 indoor heat exchanger
136
-------
Table 9.8 The Effect of Heat Exchanger Geometry on Heat Transferred1
(Heating Mode)
Flow
Geometry
Condensation
Capacity2
Evaporation
Capacity3
R-22
% (kW)
R-407C
% (kW)
R-22
% (kW)
R-407C
% (kW)
Cross
100 (9.787)
100(10.400)
100 (8.328)
100 (8.508)
Parallel
95.8 (9.378)
94,6 (9.834)
98.7 (8.217)
98.2 (8.352)
Counter
103.4
(10.120)
104.4
(10.853)
100.4
(8.360)
100.4
(8.543)
1 Air and refrigerant boundary conditions are based on experimental values. The air side heat
transfer coefficient, area and fin efficiency are based on the cross flow values.
2 Using unit 1 indoor heat exchanger
3 Using unit 1 outdoor heat exchanger
137
-------
Qair,
mj-r
m i
mair, in """air, in
(tiout, Pout, ^out
W/////////&//////////.VSS3ZZ
I *
H
m
]
j+1
i+2
"tout, Pout, ^out
mair, out Tair, out
Figure 9.10 Heat Exchanger Nodes
Figure 9.11 Concentration Shift of R-407C During Phase Change
138
-------
14
12"
^1°[
8^ ^
F Ci-
9 O
0
1
S 4
2-
*P"y
0 2 4 6 8 10 12 14
MEASURED CAPACMY (kW)
Figure 9.12 Flow Boiling Heat Transfer Correlation Results
139
-------
R-22, AC/HP 1, Indoor Heat Exchanger
JUNG, 1989 GUNGOR, 1986 SHAH, 1982 CHEN, 1987 KANDLIKAR, 1991
0,2 0.4 0.6 0,8
RELATIVE HEAT EXCHANGER LENGTH
t *
t"
t \
Figure 9.13 Evaporation Heat Transfer Coefficient Versus Length (R-22)
10
R-407C, AC/HP 1, Indoor Heat Exchanger
2*
2
UJ
o
ll. »
Li- ]2
"J 8
O »
« S
ft
Ui
U-
co
z
<
LU
X
4 -
JUNG, 1989 GUNGOR, 1986 SHAH, 1982 CHEN, 1987 KANDLIKAR, 1991
0,2 0,4 0.6
RELATIVE HEAT EXCHANGER LENGTH
0.8
Figure 9.14 Evaporation Heat Transfer Coefficient Versus Length (R-4Q7C)
140
-------
2 4 6 8 10 12 14
MEASURED CAPACITY
14
12"
'•10
U 6
tz
~ 10*
//&'/ -10*
TRAVISS, 1873
*
0 2 4 8 8 10 12 14
MEASURED CAPACITY (kW)
MEASURED CAPACITY
-------
10
2
E
z
Ui
o
LL «
u. "Q
U S
o «
a g
oc -c
LU I—
LL
CO
2
2
<
LU
I
R-22, AC/HP 1, Indoor Heat Exchanger
TANDON, 1986
TRAVISS, 1973
SHAH, 1982
CHEN, 1966
DOBSON, 1994
0.2 0.4 0.6
RELATIVE HEAT EXCHANGER LENGTH
0.8
Figure 9.16 Condensation Heat Transfer Coefficient Versus Length (R-22)
10
£•
E
LU
o
LL «
LL 12
LU §
o «
° o
a: £
tu t-
u.
to
"Z,
fcr
<
X
R-4G7C, AC/HP 1, Indoor Heal Exchanger
TANDON, 1986
TRAVISS. 1973
SHAH, 1982
CHEN. 1966
DOBSON, 1994
0.2 0,4 0,6
RELATIVE HEAT EXCHANGER LENGTH
0.8
Figure 9.17 Condensation Heat Transfer Coefficient Versus Length (R-407C)
142
-------
100
Gx
(kg/mfe)
10
1.C:
.1
: R-22, Indoor HTX from AC/HP1
Evaporation
Evaporation
(Outdoor Htx)
(Indoor Htx)
Annular
\ Dispersed
\
¦¦¦
*
Wavy v f
* "i
*. Condensation
* * (indoor Hl
-------
Length
Figure 9.20 Refrigerant Temperature Profile Along the Condenser
Figure 9.21 Refrigerant Temperature Profile Along the Evaporator
144
-------
Length
Figure 9.22 Condenser Refrigerant and Air Temperature Profiles for Cross
Parallel Flow Geometries (R-407C)
Length
Figure 9.23 Condenser Refrigerant and Air Temperature Profiles for Cross
Counter Flow Geometries (R-407C)
145
-------
9.5 Expansion Device Model
The purpose of the expansion device is to reduce the pressure of the refrigerant.
By doing this the temperature of the refrigerant is also reduced, so that the
refrigerant can absorb heat in the evaporator. The short tube restrictor is the
particular type of expansion device used in the AC/HP simulation. The short
tube restrictor is simple in construction It is roughly a 1.25 mm diameter hole
in a piece of brass 10.0 mm long . Often the hole is chamfered at the inlet and/or
the outlet.
Although simple in design, it has proven difficult to develop correlations
which accurately relate the refrigerant flow rate to the boundary conditions. This
is due to the complex and not well understood processes occurring within and
around the short tube restrictor. During normal operation of the heat pump, the
inlet of the expansion device receives sub-cooled liquid. If the downstream
pressure is greater than the saturation pressure the refrigerant remains sub-
cooled. In this case the short tube restrictor behaves as a simple orifice.
However, the down stream pressure is below the saturation pressure during the
normal operation of the AC/HP. At pressures below saturation, a portion of the
refrigerant begins to flash to vapor. At this point, the mass flow rate through
the short tube becomes less dependent on the downstream pressure. This flow
regime is commonly referred to as choked, even though there is still a
146
-------
measurable downstream pressure dependence on mass flow rate. At some
distance after the short tube restrictor, the expansion process stops.
Furthermore, at various positions in and/or downstream of the short tube
restrictor the refrigerant is in a metastable state. Hence, there are a host of
complex flow characteristics which make the short tube restrictor difficult to
model,
9.5.1 Previous Expansion Device Models
Due to the complex processes occurring within a short tube restrictor, there has
been uncertainty on how best to correlate experimental data. While there are
many empirical correlations, to date, no one has developed a fundamental
relationship which accurately handles various refrigerants and various boundary
conditions. The existing literature on the topic is now discussed, starting with
studies devoted to constant area expansion devices. This summary will be
followed by a review of how previous system simulations approached the
modeling of the expansion device.
The most commonly cited work on the subject of short tube restrictors is
by Aaron and Domanski. (Aaron and Domanski, 1990). This work is well
referenced because of the large and accurate experimental data base which was
used to develop a short tube flow model. Several important observations were
147
-------
made by the authors of this paper. One was that in the so called 'choked' flow
region the mass flow rate can vary from 3 - 8 % depending on downstream
pressure for the conditions tested. Another relevant observation was that
increasing the sub-cooling increases the mass flow rate through the short tube
restrictor. Specifically, increasing the sub-cooling by 8.3 °C increased the mass
flow rate by approximately 23%. Conversely, the lower the sub-cooling the
greater the pressure drop. The reason for this is that at a lower sub-cooling
more of the short tube restrictor is exposed to two phase fluid. The effect the
length has on the mass flow rate was also investigated. The authors found that
doubling the length of the short tube reduced the mass flow rate by 5%. Another
observation was that inlet chamfering increased the mass flow rate by 5 - 25 %
while exit chamfering had no impact on the flow rate.
The flow model developed to correlate this experimental data was almost
entirely empirical and was restricted to R-22, The flow model was based on the
orifice equation, Equation 9.41, which was modified to better represent the
experimental data. The correction was accomplished by calculating P2 as a
function of the sub-cooling, L/D, the saturation pressure, and the actual
downstream pressure.
m = CcAJ2p(P, - P2)f2
9.41
-------
For short tubes with chamfered inlets, Cc in Equation 9.41 was given as function
of chamfer depth and L/D. Several other researchers have used Aaron's
equation to correlate their data while modifying the form the P2 equation (Kuehl
and Goldschmidt, 1992), (Kim and O'Neal, 1993), (Kim and O'Neal, 1994).
Using the experimental data from Aaron, a more physically based model
was developed by Kornhauser (Kornhauser, 1993). This model considers the
metastable conditions occurring within the short tube restrictor. Using only two
empirical constants this model fits experimental data well. However, this model
was developed for the 'choked' flow region only.
Kuehl and Goldschmidt through two companion papers, created a
substantial experimental data set and two models for R-22 flowing in capillary
tubes (Kuehl and Goldschmidt, 1990), (Kuehl and Goldschmidt, 1991). Several
observations were made from the experimental data. It was observed that the
mass flow rate increased by approximately 12 % for a 5.5 °C increase in sub-
cooling, which is the same order of magnitude as Aaron found for a short tube
restrictor (Kuehl and Goldschmidt, 1990). Another significant finding was that
doubling the capillary tube length decreased the mass flow rate by approximately
29%. The first mode! developed by Kuehl and Goldschmidt was empirical, and
as such was not generally applicable. The second model had a more solid
149
-------
theoretical basis. For a given capillary tube and inlet conditions, the theoretical
model could be used to determine the pressure at the outlet of the capillary tube.
This was achieved by dividing the capillary tube into a sub-cooled liquid region
and a two phase region. The subcooled region was handled by using the moody
friction factor to calculate where the flow becomes two phase. In the same region
the refrigerant was assumed to follow an isotherm. The two phase region was
modeled as homogenous, adiabatic, and isenthalpic flow. This method utilizied
an equivalent two phase viscosity and friction factor. The mass flow was
checked to ensure that it was equal to or below the critical mass flux as defined
by Lahey (Lahey, 1977). Kuehl noted that it was necessary to introduce an
empirical constant into the theoretical model to account for the effects of
metastable flow. Hence, even a theoretical model of this complexity needed to
be manipulated to fit the data. This is evidence of the complex processes
occurring within an expansion device.
One of the more complex theoretical approaches to modeling a capillary
tube was conducted by Li (Li et al., 1990). This model considered the effects of
thermodynamic non-equilibrium and the relative velocity between the liquid and
vapor phases. Furthermore, several different flow regimes within the capillary
tube were accounted for. The mode! assumed that the flow was adiabatic.
Hence, the governing equations were reduced to continuity and momentum,
150
-------
which were solved through discretization. Among the literature examined here,
this model is the most complex. However, this work suffers from minimal
experimental verification. Therefore, it is difficult to assess the general
performance of this model.
Among the vapor compression simulations there are two trains of thought
on how to model a fixed area expansion device. Most simulations utilize the
orifice equation in conjunction with the assumption that the flow is isenthalpic
(MacArthur, 1984a), (MacArthur, 1984b), (Vargas and Parise, 1995), (Welsby et
al., 1988). However, for some simulations where a capillary tube is being
modeled a different approach is used. The capillary tube is broken down into
two regions, the single phase flow region and the two phase flow region. The
pressure drop in the single phase flow region is handled using friction factors
while the two phase regions is handled by assuming Fanno flow through the rest
of the capillary tube. The following simulations utilized different variations on
this approach: (Murphy et al., 1985), (Sami and Duong. 1987), (Melo et al.,
1990).
9.5.2 Description of Expansion Device Model
Several approaches were explored before settling on the method used to model
the short tube. What follows is a discussion of the expansion device model
151
-------
selection process.
The theoretical expansion device models were considered first. However,
it became evident during the development of the overall AC/HP simulation that
it would be beneficial for the short tube model to have a downstream pressure
dependence. Without this there is no connection, other than mass flow rate,
between the high and low pressure sides. An expansion device model with no
downstream pressure dependence would require additional assumptions to
determine the downstream pressure. Therefore, it was decided that the
expansion device model must have a downstream pressure dependence. This
restriction eliminates some of the approaches which result in a 'choked' flow
condition. This includes the Fanno flow assumption (Moran and Shapiro, 1988),
two phase critical flow models, and the Kornhauser model among others.
Finally, models which are more theoretically based, like the one developed by
Li, are too expensive to be considered for a transient simulation. This leads to
the consideration of empirical correlations.
Additional validity was given to Aaron's model when it was compared to
data collected for this dissertation. Aaron's model was compared to R-22 data
from AC/HP2. The model was used to predict the low side pressure. The results
indicate errors in low side pressure of less than 5.7%, To assess the generality
152
-------
of Aaron's correlation it was compared to R-407C experimental data. For R-
407C Aaron's model resulted in errors in the low side pressure that were less
than or equal to 10.5%. This crudely demonstrates that while being accurate for
R-22, the refrigerant for which it was developed, Aaron's correlation is not
generally applicable. However, since Kim developed a similar correlation that
was applicable over a broader range of inlet conditions that correlation was also
compared to the R-22 experimental data (Kim and O'Neal, 1994). Kim's model
produced errors that were nearly twice as high as the errors from Aaron's
correlation and, as a result, Kim's model was abandoned.
Due to the absence of one correlation that is applicable for a wide range
of inlet conditions and refrigerants, the orifice equation (Equation 9.41) is used
here to model the refrigerant flow through the short tube restrictor. In using
Equation 9.41, it is assumed that the inertial effects are negligible. This is
appropriate since the mass of the refrigerant in the short tube is at least 6 orders
of magnitude less than the mass of refrigerant in the rest of the system.
Furthermore, the flow through the short tube restrictor is assumed to be
adiabatic. It should be noted that the kinetic energy is not assumed to be
negligible. Given the inlet state and mass flow rate, the outlet state of expansion
device can be determined with these assumptions and the orifice equation.
153
-------
9.5.3 Expansion Device References
Aaron, DA, Domanski, P.A., 1990, "Experimentation, analysis, and correlation
of Refrigerant 22 flow through short-tube restrictors" ASHRAE Transactions, Vol.
96, pp. 729-742.
Kim, Y., O'Neal D.L., 1993, "An experimental study of two-phase flow of HFG-
134a through short tube orifices". ASHRAE Transactions, Vol. 98 , pp. 122-124.
Kim, Y., O'Neal, D.L., 1994, "Two-phase flow of Refrigerant-22 through short
tube orifices", ASHRAE Transactions, Vol. 100, pp. 323-333.
Kornhauser, A., 1993, "Physically Realistic Closed Form Model for Flashing Flow
In Short Tubes", Presented at The 28th Inter-society Energy Conversion
Engineering Conference.
Kuehl, S.J., Goldschmidt, V.W., 1990, "Transient response of fixed -area
refrigerant expansion devices", ASHRAE Transactions, Vol. 96, pp. 743-747.
Kuehl, S.J., Goldschmidt, V.W., 1991, "Modeling of steady flows of R-22 through
capillary tubes", ASHRAE Transactions, Vol. 97, pp. 139-148.
Kuehl, S., Goldschmidt, V.W., 1992, "Flow of R-22 through short tube
restrictors," ASHRAE Transactions, Vol. 78, pp. 59-64,
Lahey, R.T., 1977, 'The Thermal Hydraulics of a Boiling Water Nuclear Reactor",
American Nuclear Society Monograph, pp. 306-315.
Li, R.Y., Lin, S., Chen, Z.H., 1990, "Numerical modeling of thermodynamic non-
equilibrium flow of refrigerant through capillary tubes". ASHRAE Transactions,
Vol. 96, pp. 542-549.
MacArthur, J.W., 1984a, "Analytical Representation of the Transient Energy
Interactions in Vapor Compression Heat Pumps", ASHRAE Transactions, Vol.
90, pp. 982-996.
MacArthur, J.W., 1984b, "Theoretical analysis of the dynamic interactions of
vapor compression heat pumps", Energy Conversion Management, Vol. 24, pp.
49-66.
Melo, C., Ferreira, R.T.S., Pereira, R.H., Aranda, A.L.M., 1990, "Impact of the
capillary tube and condenser modeling approach on the performance of a
154
-------
dynamic simulation program for domestic refrigerators", Purdue CFC Conference
Proceedings, pp. 120-129.
Moran, M.J., Shapiro, H.N., 1988, "Fundamentals of Engineering
Thermodynamics", John Wiley & Sons, New York,
Murphy, W.E., Goldschmidt, V.W., 1985, "Cyclic Characteristics of a Typical
Residential Air Conditioner-Modeling of Start-Up Transients", ASHRAE
Transactions, Vol. 92, pp. 427-444.
Parise, JAR., 1986, "Simulation of Vapour Compression Heat Pumps",
Simulation, Vol. 46, pp. 71-76.
Recktenwald, G.W., Ramsey, J.W., Patankar, S.V., 1986, "Predictions of Heat
Transfer in Compressor Cylinders", Purdue Compressor Conference
Proceedings, pp. 159-174.
Sami, S.M., Duong, T., 1987, "An improved model for predicting refrigerant flow
characteristics in capillary tubes", ASHRAE Transactions, Vol. 93, pp. 682-699.
Vargas, J.V.C., Parise, J.A.R., 1995. "Simulation in Transient Regime of a Heat
Pump with Closed-Loop and On-Off Control", International Journal of
Refrigeration, Vol. 18, No. 4 pp. 235-243.
Welsby, P., Devotta, S., Diggory, P.J., 1988, "Steady- and Dynamic-State
Simulations of Heat-Pumps. Part II: Modeling of a Motor Driven Water-to-Water
Heat-Pump", Applied Energy, Vol. 31, pp. 239-262.
155
-------
Chapter 10
The AC/HP Simulation
10.1 Previous AC/HP Simulations
There exists a wide range of complexity among the previously published AC/HP
simulations. In this context, steady state simulations represent a subset of
transient simulations. Hence, the focus of this work is on transient simulations.
Overall, the complexity of transient AC/HP simulations has increased with time.
Therefore, the literature review will be organized in chronological order, starting
with the earliest reference found on this topic.
Strader wrote the first digital computer transient simulation of an AC/HP
that did not require significant amounts of experimental data (Strader, 1976).
This simulation was a simple explicit first order lumped parameter model and as
a consequence it had many shortcomings. Since the evaporator control volume
was combined with the compressor shell this simulation had only two control
volumes. Furthermore, the simulation did not include the thermal capacity of the
tubing and compressor components. The result was a simulation that modeled
part of the significant physics and only yielded inlet and outlet data to the
evaporator and condenser. Although an excellent start, much work needed to
156
-------
be done.
Manmohan worked to address some of the shortcomings of the Strader
simulation (Manmohan, 1978). This simulation was also an explicit first order
lumped parameter model of a AC/HP. The inclusion of the heat exchanger
thermal mass and the addition of more state points were some of the
improvements in the Manmohan simulation. However, there still remained many
weaknesses. Although experimental data was not used to verify the simulation,
steady state experimental data was used to determine the overall heat transfer
coefficients in the heat exchangers. The compression process was modeled as
polytropic and internal heat exchange within the compressor was neglected. The
equation of state was a series of curve fits for the regions and properties of
interest.
Chi further advanced the state of the art by including more of the relevant
physics and minimizing the dependence on empirical data (Chi and Didion,
1982). In this simulation the compression process was considered polytropic but
heat exchange within the compressor was included. The heat transfer was
handled through the use of heat transfer correlations. The simulation was
compared to experimental data. However, the experimental data collection rate
was every 30 seconds which prevented the rapid transients from being resolved.
157
-------
It is worth noting that the largest time step this explicit first order simulation could
take was approximately 5e-3 seconds.
Murphy also developed an explicit first order lumped parameter
simulation which took a 5e-3 second time step (Murphy and Goldschmidt, 1985).
However, Murphy's simulation was different in many respects. The compressor
was modeled by curve fitting steady state experimental data. Furthermore, the
evaporator was not modeled, instead transient experimental data was utilized to
represent the evaporator. As one would expect, this simulation was found to
agree very well with experimental data, since many of the inputs to the simulation
came from the experiments themselves. The simulation was used to evaluate
the sensitivity of the transient performance to the size of the liquid line and the
condenser. It was found that the larger the condenser or the liquid line the
poorer the transient performance of the AC/HP
In a series of three papers Welsby developed and tested a transient
AC/HP simulation that is similar in many respects to Chi's (Welsby et al,, 1988a),
(Welsby et al., 1988b), (Welsby et al., 1989). This lumped parameter simulation
required some empirical data, primarily for the expansion device and
compressor. The simulation agreed marginally well with experimental data taken
at 50 second intervals.
158
-------
Katipamula, whose work was discussed in Chapter 7, also worked on the
transient modeling of a AC/HP (Katipamula, 1989). Katipamula spent his efforts
modifying Chi's simulation to handle a latent load. This simulation is the only
one found that is capable of modeling a transient latent load. He experimentally
verified the simulation against data collected at seven second intervals. He
found that the model was most accurate for long on times and high relative
humidities.
Through a five year evolution, MacArthur has significantly furthered the
transient modeling of AC/HPs (MacArthur, 1984a, 1984b), (MacArthur and Grald,
1987, 1989). Similar to other simulations, the compressor was modeled with a
polytropic compression process and internal heat exchange. However,
MacArthur significantly improved upon existing heat exchanger models used in
transient and steady state AC/HP simulations. The heat exchangers were
modeled by solving the discretized energy and continuity equations. This first
order accurate simulation is unique in that some conservation equations are
explicit while others are implicit. Experimental verification showed that the
simulation did an excellent job of predicting the performance of a water cooled
AC/HP. The major shortcoming of this simulation is that it cannot directly
compute the steady state performance of a AC/HP
159
-------
In order to study the transients of a distillation system Salim developed a
heat pump simulation (Salim et al., 1991). Salim assumed that the compression
process is polytropic and that the compressor does not exchange heat.
Furthermore, the continuity equation for the heat exchangers did not include any
storage terms {i.e. the mass flow rate does not change throughout the heat
exchanger). The author used this lumped parameter simulation to show that
increasing the temperature lift decreases the efficiency of the system.
The only researcher to model the transients of a AC/HP operating with a
refrigerant mixture is Sami (Sami and Zhou, 1995), (Sami and Comeau, 1992).
Al! of the components of this simulation were modeled as lumped parameters.
Nominally the simulation was developed for zeotropic mixtures but it does not
have a species balance. Hence, this model assumed that the zeotropic mixture
behaved as a pure refrigerant.
The most recent transient simulation is by Vargas (Vargas and Parise,
1995). The simulation was developed to evaluate the effects of different control
strategies. Vargas utilized one of the more advanced explicit time marching
techniques. Specifically, he employed a Runge-Kutta Fehlberg fourth-fifth order
method with controlled step size. While being computationally complex, the
simulation is theoretically simple. This lumped parameter simulation relies on
160
-------
many simplifying assumptions, including assumptions regarding heat transfer
and the equation of state. Despite these simplifications, this simulation is well
suited for its purpose. Vargas showed that a closed loop control algorithm
utilizing a variable speed compressor can save 11% in energy consumption as
compared to the typical on-off control system used in residential AC/HPs.
Hence, the literature review revealed no simulation capable of modeling
the potential concentration shifts of a refrigerant mixture either at steady state
or transiently. Additionally, most of the transient simulations were not capable
of directly determining the steady state performance. Furthermore, no simulation
was found that resolved the continuity, energy, and momentum equations for any
fluid in either the steady state or transient modes of operation. It is the goal of
the simulation developed here to address these weaknesses.
10.2 Description of the AC/HP Simulation
The overall simulation is comprised of several component simulations which
have been described earlier. The component simulations work independent of
the overall simulation and are distinct from each other. Each component has it's
own subroutine and when supplied with the inlet conditions it is able to calculate
the outlet conditions along with many other variables of interest. The overall
simulation starts with a guess of the compressor inlet conditions, which include
161
-------
the temperature, pressure, concentration, and mass flow rate. The compressor
model then proceeds by calculating, among other quantities, the outlet
conditions. The outlet quantities are then used as inputs to the next component,
the condenser, which in turn calculates the pertinent outlet conditions. This
process is repeated for the expansion device and evaporator. Once the outlet
conditions are calculated for the last component, the overall simulation checks
to see if these conditions are equal to the inlet conditions of the first component.
If this is not the case, the simulation iterates to find the appropriate inlet
conditions that satisfy this condition. At this point one step in time has been
calculated. The variables are now updated and the process is repeated for the
next time step, starting with the guess of the inlet conditions to the compressor.
10.3 Numerical Methods
The numerical methods used to solve the non-linear equations are divided into
two parts. The first part that is described is the actual solution of the non-linear
equations. Equally important, and often neglected, is the method used to
provide the first guess which initiates the solution process. This is discussed in
Section 10.3.2.
10.3.1 Solving the Non-Linear Equations
The non-linear equations associated with the compressor, heat exchanger, and
162
-------
the overall simulation are all solved with the same non-linear equation solver.
Specifically, the non-linear equations are solved using a modified Newton-
Raphson technique. The standard Newton-Raphson technique (NRT) is outlined
first and is followed by a description of the enhancements to that technique used
for this research.
The standard Newton-Raphson process starts with a guess of the
variables in question. The iterative process begins with the evaluation of the
residual equations to determine if the current value of the variables satisfies
these equations within some small tolerance, e. In this work, the residual
equations are the conservation equations structured so that the sum of the terms
equals zero. Hence, the residuals represent the extent to which the conservation
equations are not satisfied. If the residual equations are not satisfied two steps
are taken. First, the partial derivatives of the residual equations, the Jacobian,
are evaluated through the use of the secant method in a subroutine called
'pardif. Secondly, through the use of gaussian elimination, the correction vector
is determined in the subroutine 'gausel'. The previous variables are now
corrected by the addition of the correction vector. This process is repeated until
each residual equation is less than e.
Several modifications are implemented to improve the convergence and
163
-------
robustness of the NRT, One of the modifications deals with the convergence
criteria. As previously stated, the solution to the set of non-linear equations is
obtained when each residual is less than e. For the work conducted here e = 10"
B. Occasionally, after numerous iterations the residual of one equation would not
satisfy this requirement by roughly an order of magnitude. It was later
determined that the magnitude of the correction for that residual equation was
smaller than e and the variable in question was alternating between two values.
In other words, the NRT could not resolve the change in the variable required to
satisfy the residual equation. Hence, the variable found by the NRT was within
±e of the variable which satisfies the residual equation. Therefore, the code was
modified to incorporate this scenario as an additional convergence criteria .
The remaining enhancements to the NRT involve the modification of the
correction vector or the corrected variables. One such modification is that
corrected values of the variables are constrained to reasonable values in an
absolute sense. An example of this modification would be that negative absolute
pressures and temperatures are not tolerated.
Another modification is the implementation of a region of confidence.
When the error in the residuals is high, it is possible for the NRT to diverge from
the solution. This is mitigated by defining a region of confidence, In other
164
-------
words, it is known, through experience, that the solution to the residual equations
is in a predefined region. For example, when marching in time one can be
confident that the variables at the next time step are within 1% of the variables
of the last time step. Therefore, intermediate solutions to the non-linear
equations are not allowed outside of this confidence region.
The magnitude of the correction vector is also adjusted to enhance
convergence. This is done when the solution is not obtained within a certain
number of iterations and the residual error is relatively low, within two orders of
magnitude of e. Usually this occurs when NRT is continuously and marginally
over shooting the solution. In this case, the magnitude of the correction vector
is reduced by a factor of two.
A second method is used for solving the non-linear equations associated
with the expansion device. The pressure at the outlet of the short tube restrictor
is solved by using the bisection method (BM). This method was chosen because
of its robustness, it essentially never diverges. This method is often maligned
because of it's poor rate of convergence. However, in this case the BM typically
required a similar number of function evaluations as did the NRT. There are two
reasons for this. One reason is that the BM is supplied with an excellent guess
from the previous time step. The second reason is that the BM does not have
165
-------
the overhead of derivative evaluation as does the NRT.
10.3.2 The First Guess
Before either the NRT or the BM can get started a guess value of the variables
in question must be provided. The closeness of the guess values to actual
solution is directly proportional to the number of iterations required by the
solution method to find the actual solution. In order to develop a simulation
which requires as little computational time as possible it is necessary to use
accurate guess values.
The method developed here is only suitable for equations developed from
time marching or analogous problems. The first step in time uses a very simple
guessing routine which other simulations use at all time steps. Specifically, the
guessed value is simply the value at the previous time step. While suitable, this
is not always the optimal technique. A more complex technique is employed at
time steps beyond the first. The backwards zeroth through third order derivative
with respect to time of each variable to be guessed is calculated for the previous
time step. One of these four derivatives will be used to calculate the guess value
at the next time step. This is similar to an explicit time marching technique (i.e.
explicit Euler or Leap Frog). The order of the derivative chosen to calculate the
guess value is determined by evaluating which derivative would have come
166
-------
closest to predicting the value at the last time step. For example, the second
order derivative will be used if the value at the last time step would have been
best predicted by the second order derivative form the previous time step.
Compared to using the value at the last time step as the guess value, this
technique reduces the time required to run the simulation by more than a factor
of two. Furthermore, when convergence difficulties arise or the guess produces
extraordinarily high residuals the program cycles through the other derivatives
to find one that performs the best.
10.3.3 Determining the Time Step Size
There are primarily two constraints which determine the size of the time step.
The most important constraint is the accuracy of the solution. The solution
should be independent of step size. It will be shown in Chapter 12 that to satisfy
this requirement the step size should be less than 5.0e-2 seconds. This
constraint is only on the early transients which occur in the first 30 seconds. The
other and more rigorous constraint, in this case, is the timely convergence of the
NRT, If the step size is too large it can take an excessive amount of iterations
for the NRT to find the solution to the non-linear equations in question. Since
the time marching technique only uses information from the previous step it is
possible to continuously change the time step size without adversely affecting
the solution. The method used here continuously monitors the number of
167
-------
iterations required to solve for the next time step. If the average iterations of the
last five time steps exceeds 8 then the time step is divided by 1.05. If the
average number of iterations for the last five time steps is less than 4, the time
step is multiplied by 1.05. However, a maximum of 2e-2 seconds is imposed on
the time step during the early transients.
10.4 Simulation References
Chi, J., Didion, D.,1982. "A Simulation Model of a Heat Pump's Transient
Performance", International Journal of Refrigeration, Vol. 5, No. 3 pp. 176-184.
Katipamula, S., 1989, A Study of the Transient Behavior During Start-Up of
Residential Heat Pumps, Ph.D. Dissertation, Texas A&M University.
MacArthur, J.W., 1984a, "Analytical Representation of the Transient Energy
Interactions in Vapor Compression Heat Pumps", ASHRAE Transactions, Vol.
90, pp. 982-996.
MacArthur, J.W., 1984b, "Theoretical analysis of the dynamic interactions of
vapor compression heat pumps", Energy Conversion Management, Vol. 24, pp.
49-66.
MacArthur, J.W., Graid, E.W., 1987, "Prediction of cyclic heat pump performance
with a fully distributed model and comparison with experimental data", ASHRAE
Transactions, Vol. 93, pp. 1159-1178.
MacArthur, J.W., Grald, E.W., 1989, "Unsteady compressible two-phase flow
model for predicting cyclic heat pump performance and comparison with
experimental data", Rev. Int. Froid, Vol. 12, pp. 29-41.
Manmohan, D., 1978, "Transient Analysis of Refrigeration System", Ph.D.
Dissertation, Purdue University.
Murphy, W.E., Goldschmidt, V.W., 1985, "Cyclic Characteristics of a typical
Residential Air Conditioner-Modeling of Start-Up Transients", ASHRAE
Transactions, Vol. 92, pp. 427-444.
168
-------
Salim, M.A., Sadasivam, M., Balakrishnan, A.R., 1991, "Transient Analysis of
Heat Pumps Assisted Distillation Systems", International Journal of Energy
Research, Vol. 15, pp. 123-135.
Sami, S.M., Comeau, MA, 1992, "Development of a Simulation Model for
Prediction Dynamic Behavior of Heat Pump with Nonazeotropic Refrigerant
Mixtures", International Journal of Energy Research, Vol. 16, pp. 431-444.
Sami, S.M., Zhou, Y., 1995, "Numerical Prediction of Heat Pump Dynamic
Behavior Using Nonazeotropic Refrigerant Mixtures", International Journal of
Energy Research, Vol. 19, pp. 19-35.
Strader, D.L., 1976, "Computer Simulation of the Dynamics of Refrigeration
Cycles", M.S. Thesis, Purdue University.
Vargas, J.V.C., Parise, J.A.R., 1995, "Simulation in Transient Regime of a Heat
Pump with Closed-Loop and On-Off Control", International Journal of
Refrigeration, Vol. 18, No. 4 pp. 235-243.
Welsby, P., Devotta, S., Diggory, P.J., 1988a, "Steady- and Dynamic-State
Simulations of Heat-Pumps. Part I: Literature Review", Applied Energy, Vol. 31,
pp. 189-203.
Welsby, P., Devotta, S., Diggory, P.J , 1988b, "Steady- and Dynamic-State
Simulations of Heat-Pumps. Part II: Modeling of a Motor Driven Water-to-Water
Heat-Pump", Applied Energy, Vol. 31, pp. 239-262.
Welsby, P., Devotta, S., Diggory, P.J., 1989, "Steady- and Dynamic-State
Simulations of Heat-Pumps. Part III: Comparison Between Predictions and
Measurements", Applied Energy, Vol. 32, pp. 1-18.
169
-------
Chapter 11
Experimental Results and Discussion
11.1 Introduction
Several issues are addressed by the experimental aspects of this research. One
of the primary objectives is to evaluate the performance of R-407C relative to R-
22, Another experimental goal is to study the effect the expansion device has
on system performance. Furthermore, the effect of vapor to liquid line heat
exchange is also evaluated. These comparisons are conducted at steady state
and cyclic conditions. The refrigerants and system configurations are also
compared in terms of seasonal performance. Furthermore, R-32/134a is
compared to R-22. These refrigerants are compared for model verification
purposes and to study mixture related phenomena. The steady state data is
presented and described first, fallowed by the cyclic and seasonal performance
data. Beyond the objectives previously mentioned, the circulated concentration
of the refrigerant mixtures is investigated. This final facet of the experimental
work is presented last.
170
-------
11.2 Steady State Test Results
11.2.1 Refrigerant Charge Optimization
Before a battery of tests could be conducted, it was necessary to determine the
appropriate amount of refrigerant to charge into the system. Charge optimization
tests were conducted to make this determination. As outlined in Section 7.5, the
amount of refrigerant in the system was varied and the system performance
measured until an optimum COP was revealed. At the point of highest COP the
charge was considered to be optimum. The optimum charge and COP for each
charge optimization are displayed in Table 11.1 along with the uncertainty in
each quantity. The uncertainty analysis is presented in Appendix A1 and A2.
Figure 11.1 shows a typical charge optimization curve that is non-
dimensionalized with its optimal values. This particular plot is of R-22 in the
cooling mode with a short tube restrictor (STR) being used as the expansion
device.
To understand the shape of the charge optimization curve it is necessary
to examine how the amount of refrigerant in the system effects the power and
capacity that constitute the COP. The influence of refrigerant charge on
compressor power will be described first followed by the influence of refrigerant
charge on capacity.
171
-------
The relationship between refrigerant charge and compressor power for the
data presented in Figure 11.1 can be seen in Figure 11,2. Clearly, as refrigerant
is added to the system the compressor power increases. Equations 9.1 - 9,4 will
be useful to illustrate why compressor power increases with an increasing
refrigerant charge. A common misperception is that the pressure ratio and the
pressure difference increase as refrigerant is added, and this increases the
amount of work required by the compressor. Figure 11.3 contests this view.
Figure 11.3 is a graph of suction and discharge pressures versus refrigerant
charge. This figure indicates that the pressure difference remains essentially
constant and that the pressure ratio decreases with refrigerant charge. Based
on Equation 9.4, this would imply that the compressor work should decrease
since the pressure ratio decreases. However, there are competing effects that
act to increase the work of the compressor. Nonetheless, the pressures do hold
the key to why the compressor power increases with refrigerant charge, but with
a different line of reasoning than proposed earlier. It is clear from Figure 11.3
that the suction pressure increases with charge. Since the compressor used
here is a positive displacement compressor, the volumetric capacity of the
compressor is nearly constant. In other words, the compressor moves the same
volume flow rate of refrigerant independent of operating conditions. This is
equivalent to assuming that both the clearance volume, Vc earanee, is zero in
Equation 9.2 and the speed of the compressor in Equation 9.1 is constant.
172
-------
Therefore, increasing the suction pressure increases the suction density, which
increases the refrigerant mass flow rate. This conclusion can be verified by
examining Figure 11.4, which is a graph of refrigerant mass flow rate versus
charge. This figure supports the conclusion that the compressor power
increases monotonically because the mass flow rate increases as result of the
increased suction gas density. On the other hand, the capacity, which is the
other component of COP has a different character.
The capacity increases to a maximum value, and then declines, as
depicted in Figure 11.5. The shape of the capacity curve is related to the mass
flow rate as was the compressor power curve. This relationship exists because
most of the evaporator capacity comes from the latent heat of the refrigerant,
which does not change appreciably in the range considered here. Assuming all
of the refrigerant is evaporated in the evaporator, the capacity is approximately
equal to the mass flow rate of the refrigerant multiplied by the latent heat. As the
mass flow rate increases so does the capacity, to a point. The sharp decrease
in capacity with increasing refrigerant charge is a result of having an evaporator
with a finite area. Before explaining further, it should be noted that the
evaporator can be divided into two regions, the two phase region and the single
phase region. The single phase region is occupied by superheated vapor.
When the refrigerant charge and mass flow rate increase, the area of the
173
-------
evaporator occupied by the two phase refrigerant increases at the expense of
the single phase region. This occurs because at higher mass flow rates a
greater area is required for the increasing amount of heat being transferred in
the two phase region. The net result is that the superheat leaving the evaporator
decreases with refrigerant charge, as shown in Figure 11.6. The capacity starts
to decrease dramatically when there is not enough heat exchange area to reject
all of the latent heat, and as a result, a two phase condition exists at the outlet
of the heat exchanger. The refrigerant that is not evaporated represents lost
cooling capacity causing the negative slope on the capacity versus refrigerant
charge figure.
11.2.2 Typical Steady State Test Results
Once the optimal charge is determined, the entire battery of tests is conducted.
Typical results are displayed in Table 11.2. The data in Table 11.2 are for R-22
with TXVs in the heating and cooling modes. The COP and capacity in Table
11.2 change in response to the different ambient conditions associated with each
test. From the B to A test the outdoor dry-bulb temperature changes from 27.8°C
(82.0°F) to 35.0°C (95.0°F) while the indoor conditions remain unchanged.
Thus the temperature lift has changed from 1.1 °C (2.0°F) to 7 2°C (13.0°F). This
causes a 4.5% decrease in capacity and a 15% decrease in COP. This occurs
for the following reasons. For the condenser to reject heat to the warmer outside
174
-------
air the condensing temperature must increase which in turn requires that the
condensing pressure must increase. In this case, the discharge pressure
increases from 1460 kPa to 1719 kPa. The increase in pressure increases the
amount of work done by the compressor, thereby decreasing the COP,
Furthermore, the warmer refrigerant entering the expansion device decreases
the cooling capacity by increasing the enthalpy entering the evaporator. The
effect of changing evaporator conditions can be seen by examining tests B and
C. Tests B and C have the same dry-bulb temperatures for both the indoor and
outdoor environments, but the indoor relative humidity changes from 51.5% in
the B test to less than 20.0% in the C test. The low relative humidity of the air
in the C test corresponds to a dew point temperature that is far below the
temperature of the indoor heat exchanger. As a result, water does not condense
from the air stream onto the evaporator. Therefore, the difference between the
B and C tests is in the air side latent capacity. Since the G test has no air side
latent capacity the total capacity of the evaporator is reduced, in this case by
3.0%. It should be noted that without a TXV this reduction in capacity would be
more significant as will be explained in Section 11.3 2.
As compared to the cooling mode, the performance of the system in the
heating mode is relatively poor. For example, the COP for the 47S test is 10%
less than COP for the B test. The temperature lift is again the major reason for
175
-------
this performance drop. The 47S test has a 12.8°C (23.0°F) temperature lift while
the B test has a 1.1°C (2.0°F) temperature lift. A more significant reflection of
the temperature lift is seen in the 17L test. The temperature lift for the 17L test
is 29.4°C (53 o°F) and as a result the COP is 40% less than the B test.
Furthermore, the capacity is lower for the heating tests because of the lower
suction pressures that are responsible for the lower densities at the compressor
inlet, which negatively affects capacity.
11.2.3 The Impact of the Expansion Device on Steady State Performance
The steady state performance of a AC/HP operating with short tube restrictor
(STR) and a thermostatic expansion valve (TXV) is now investigated. The size
of the STR for the cooling mode was chosen so the system with the STR would
have the same performance as the system with the TXV at the same refrigerant
charge. This was done at the B test conditions . The size of the STR for the
heating mode was determined similarly, although the 47S test conditions were
used. The closest matching STR's had diameters of 1.65 mm (0.065 in) and
1.91 mm (0.075 in) for the heating and cooling modes, respectively. The results
of the tests can be seen in Tables 11.2 and 11.3. At the B test conditions the
STR produced a COP that was 2.8% less than the COP of the system with the
TXV. For the 47S test conditions, the STR produced a COP that was 3.9% less
than the COP obtained when the TXV was utilized. The difference between the
176
-------
system performance with the TXV and the STR at these two test conditions is a
result of the discrete sizes of STR's available. Given an infinite variation in STR
sizes, there is no fundamental reason why identical results for the TXV and STR
could not be obtained at the two test conditions.
In the cooling mode the performance of the system is significantly affected
by the STR. Compared with the TXV results the STR restrictor had a 4.8% lower
COP for the A test and 8.0% lower COP for the C test. When discounting for the
less than perfect size of the STR, the degradation at the A test conditions is
approximately 2.0% and the degradation at the C test conditions is approximately
5.2%. The degradation in performance for the A test is small because only the
condenser test conditions change from test A to test B. As expected, the
performance of the system with different expansion devices is not sensitive to
condenser conditions. The same would not be expected for changes in
evaporator conditions. Point in case, from test B to test C the evaporator
conditions change while the condenser conditions remain unchanged. In this
case, the TXV is able to adjust to the new conditions and maintain system
performance while the STR can not. In the heating mode the degradation in
performance associated with the STR is not as severe. The TXV does not work
as well in the heating mode because of the exceptionally broad range of
evaporator conditions which it must accommodate [ -8,3°C (17.0°F) - 18.3
177
-------
(65.0°F)]. Hence, at the conditions tested here the TXV is in the lower limits of
its control range. Therefore, there is not a significant difference between the
STR and the TXV.
The refrigerant charge sensitivity of the STR and the TXV can be seen in
Figure 11,7. This figure shows the charge optimization curves for the system
with a TXV and a STR in the cooling mode. It is clear that STR is far more
sensitive to charge than is the TXV. This is expected since the STR is unable
to forestall the onset of two phase flow at the exit of the evaporator.
11.2.4 Steady State Comparison of Refrigerants
Since the experimental work serves to evaluate R-407C relative to R-22 and to
verify the AC/HP simulation through the comparison of R-32/134a to R-22 the
refrigerant comparison is made in two parts. First, R-407C is compared to R-22
with the system outfitted with TXVs. Afterwards R-32/134a is compared to R-22
with the system outfitted with STRs.
The charge optimization curves for R-22 and R-407C can be seen in
Figure 11.8. These tests were conducted at the B test conditions. In terms of
charge sensitivity these refrigerants are practically identical. The optimum
charge for R-22 and R-407C was 6.54 kg (14 40 lb) and 6.81 kg (15.00 lb),
178
-------
respectively. These refrigerants have similar optimum charges because they
have similar molecular weights and vapor pressure curves, as shown in Section
2.
The results from the steady state battery of tests can be seen in Table
11.2 for R-22 and in Table 11.4 for R-407C. The steady state results show that
R-407C has anywhere from a 4.6% to a 9.8% lower COP than R-22. The
capacity of R-407C ranges from the same as R-22 to 1.5% higher than R-22.
These results agree with those presented elsewhere (Godwin, 1993). Some
typical values of cycle parameters are demonstrated in Tables 11.5 and 11.6.
There are noteworthy differences between the fluids in terms of the pressure
ratio and discharge temperature. The pressure ratio for R-407C is generally
higher than the pressure ratio for R-22. This is due to the steeper vapor-
pressure curve which is also related to the, approximately, 9% higher latent heat
of R-407C relative to R-22. Since the constant pressure specific heat of R-407C
is approximately 15% higher than the specific heat of R-22, the compressor
discharge temperature is typically lower for R-407C. It should be noted that a
lower discharge temperature is beneficial since it is known that compressor wear
increases with discharge temperature (Richardson and Gatecliff, 1992).
R-22 is now compared to R-32/134a. As already mentioned the purpose
179
-------
of this element of the experimental work is to provide data to compare to the
AC/HP simulation. The simulation is not capable of modeling air side latent
loads (i.e. condensation and frost accumulation). The two tests which do not
have latent loads are tests C and 47S. These tests were used to evaluate each
fluid. Charge optimization tests were conducted for each of the fluids at each of
these test conditions.
Figure 11,9 displays the cooling mode charge optimization curves for both
refrigerants. From this graph it is clear that in the cooling mode both fluids
behave similarly with respect to the amount of refrigerant in the system. In the
cooling mode, the optimum charge for R-32/134a is 8.2% less than the optimum
charge for R-22. This is consistent with the 8.4% lower molecular weight of R-
32/134a. The COP and capacity are compared with and without fan power in
Table 11.7 so that comparisons can be made with other experimental data and
the data from the AC/HP simulation which does not include fan power. There
are two fans, one for each heat exchanger. The indoor fan represents a power
source in the air stream which reduces the capacity in the cooling mode but in
the heating mode the indoor fan increases the capacity. In both modes the fan's
power increases the total power in the calculation of the COP. When fan power
is not included R-32/134a has a 3.4% higher cooling COP when compared to R-
22 but when fan power is included R-32/134a has a cooling COP that is only
180
-------
2.9% higher. The cooling capacity of R-32/134a excluding fan power is 1.3%
higher than that of R-22 which is barely greater than the estimated random error
of 1.2%. These moderate increases in performance are expected based on the
slightly more desirable thermodynamic properties of R-32/134a described in
Chapter 2. Table 11.8 is a table of system parameters for R-22 and R-32/134a
at the C test conditions. The similarity of the various parameters for the two
fluids is noteworthy. The major difference between them is that R-32/134a has
a lower discharge temperature which, once again, is due to its higher specific
heat.
The relative performance of R-32/134a changes in the heating mode.
Figure 11.10 is a graph of R-22's and R-32/134a's charge optimization curves
for the heating mode. From this figure it is clear that R-32/134a is less sensitive
to refrigerant charge than is R-22. As was the case in the cooling mode, the
optimum charge of R-32/134a is less than the optimum charge of R-22.
Specifically, the optimum charge for R-32/134a is 0.23 kg or 4% lower than that
of R-22. Relative to R-22, the optimum COP of R-32/134a is 7.4% and 8.5%
less, with and without fan power, respectively. Table 11.9 displays some cycle
parameters of interest for R-22 and R-32/134a at the 47S test conditions and
sheds some light on the lower performance of R-32/134a relative to R-22. In
Table 11.9 the most noteworthy difference between R-22 and R-32/134a is the
181
-------
two phase condition existing at the outlet of the evaporator. Since R-32/134a
has a steeper vapor pressure curve it requires a greater pressure difference
between the evaporator and the condenser. The short tube restrictor acts to
maintain this difference. However, since the short tube restrictor was sized for
R-22 it does not provide enough pressure difference between the low and high
pressure sides. As a result the temperature of the R-32/134a entering the
evaporator is slightly higher than desirable and is not sufficient to boil all of the
refrigerant. This trend was also present in the cooling mode but to a lesser
extent. The two phase condition leaving the evaporator and entering the
condenser has two consequences. One is that it lowers the capacity of the
system and the other is that the compressor work increases significantly. Both
of which act to reduce the COP. Furthermore, if this compressor was not a scroll
it would have probably been damaged as a result of attempting to compress a
two phase fluid.
11.2.5 The Impact of a VLHX on Steady State Performance
The suction line to liquid line heat exchanger (VLHX) is implemented in an effort
to improve the performance of R-407C relative to R-22. The VLHX was installed
with the intent that it would be used in the cooling mode only. The tests were
performed with a TXV as the expansion device. The previously determined
optimum charge without the VLHX was used for the VLHX tests. The COP and
182
-------
capacity for R-22 and R-407C with the VLHX are outlined in Tables 11.10 and
11.11, respectively. The effect of the suction line heat exchanger on R-22's
steady state system performance is not surprising, as it is well known that R-22
gains little to nothing in terms of performance from a VLHX (Domanski, 1994).
The measured steady state benefit of the suction line heat exchanger with R-
407C is within experimental uncertainty. Among the parameters presented in
Table 11.5, the only significant difference caused by the VLHX is the increase
in compressor discharge temperature. This increase is a result of the increased
compressor suction temperature caused by the heat exchange with the warm
liquid line. It was already mentioned that an increased compressor discharge
temperature correlates with decreased compressor life. Hence, the suction line
heat exchanger is of no use in the limit of steady state operation and may be
detrimental to compressor life.
11.3 Cyclic Test Results
All of the cyclic tests conducted are presented in Table 11.12. This table
contains information on each of the variables changed from test to test (e.g.,
refrigerant, expansion device, on-time). The cyclic data is normalized with
steady state data so that phenomena that can be explained from a steady state
viewpoint are not addressed in this section. Table 1113 also contains the
normalized data of interest (power, capacity and COP) for each cyclic test.
183
-------
Before discussing the normalized data it will be useful to examine some cycle
parameters as they change with time from a typical data set,
The data presented in this section are from the D' test with R-22. The
expansion device used for this test was a STR. Figure 11,11 is a graph of
refrigerant temperatures at different points in the cycle versus compressor on-
time. It is important to note that the evaporator inlet and outlet temperature
measurements were made in situ while the condenser inlet and outlet
temperature measurements were made on the tube. In situ temperature
measurements in the outdoor unit were not possible due to lack of space for the
additional tubing associated with in situ measurements. Several observations
can be made from Figure 11.11. One observation is that for the first 54 seconds
the temperature of refrigerant leaving the evaporator is lower than the
temperature of the refrigerant entering the evaporator. This is caused by two
factors, namely, two phase flow leaving the evaporator and pressure drop
through the evaporator. In the absence of pressure drop a two phase or
saturated vapor state would be indicated by a refrigerant temperature leaving the
evaporator that is equal to the refrigerant temperature entering the evaporator.
Another observation dealing with the evaporator is the minimum refrigerant inlet
temperature to the evaporator. The minimum is 2,46°C below the steady state
temperature and occurs at 132 seconds. An explanation for this can be gleaned
184
-------
from Figure 11.12. This figure is a graph of compressor suction and discharge
pressure versus compressor on-time. It is clear from this figure that the
compressor suction pressure follows the same trends as the evaporator inlet
pressure. The suction pressure reaches its minimum, 48 kPa below steady
state, at 138 seconds. The evaporator pressure and temperature drop below
their steady state values at 138 seconds because more refrigerant has been
removed from the evaporator by the compressor than is allowed in by the STR.
This is a consequence of the relatively high suction gas density and the relatively
low density at the inlet to the STR. The same arguments explain the existence
of the maximum discharge pressure which is 116 kPa above steady state and
occurs at 30 seconds. This point will be made more clear in Chapter 12 through
the discussion of the theoretical results. Related to the minima and maxima in
suction and discharge pressures is the compressor power which is displayed in
Figure 11.13. The increased pressure difference and suction gas density, in
addition to the inertia of the compressor, cause the compressor power to initially
exceed the steady state compressor power. The maximum compressor power
is 27% higher than the steady state power and occurs at 12 seconds.
Figure 11.14 is of air side capacity and shows a significantly longer
transient period than the refrigerant side quantities. For this work, steady state
is defined as the point in time when the value in question reaches 95% of its final
185
-------
value. Using this definition the air side capacity reaches steady state in 294
seconds. This is considerably slower than the results observed on the
refrigerant side. Using the same definition and the refrigerant suction pressure,
the refrigerant side reaches steady state in 192 seconds. The large discrepancy
in time is caused by the thermal mass of the indoor unit (blower, housing, etc.)
which must be brought to its equilibrium temperature before the air side capacity
can reach steady state.
11.3.1 The Impact of On-Time on Cyclic Performance
Although instantaneous quantities provide useful information, the COP, capacity
and power integrated over the on-time are particularly valuable for evaluating
different system modifications. Among the different variables studied, the
compressor on-time is investigated first. The comparison is made between
experiments with a six minute on-time and a 30 minute on-time. This is done
with R-22 and R-32/134a operating with STRs. All of the comparisons are made
relative to the six minute on-time data. This is done because as systems are
currently designed they are more likely to have a 6 minute on-time than a 30
minute on-time. Examination of the data in Table 11.13 reveals that having a
longer compressor on-time invariably improves performance. This is expected
since most of the losses occur in the first few minutes. As on-time increases
these losses become a smaller fraction of the total on-time. The data reveal that
186
-------
the improvement in performance is independent of refrigerant and the mode of
operation. In the cooling mode, R-22's capacity increased by 28% while R-
32/134a's capacity increased by 29%, when comparing 6 minute compressor on-
time data to 30 minute compressor on-time data. The same data sets showed
that the normalized power was insensitive to on-time for both fluids. However,
the normalized COP increased by 29% and 28% for R-22 and R-32/134a,
respectively. The improvement in performance in the heating mode, due to a
longer on-time, is even more significant than the improvement observed in the
cooling mode. For R-22 the capacity increased by 53% whereas the normalized
power and COP increased by 6.7% and 43%, respectively. Similarly, for R-
32/134a the capacity was 52% higher, the power was 7.3% higher, and the
normalized COP was 42% higher. The increased benefit of a longer on-time in
the heating mode is a result of the greater losses associated with cyclic
operation in the heating mode which is explained in Section 11.3.5.
11.3.2 The Impact of the Expansion Device on Cyclic Performance
Just as the expansion device impacts steady state performance it also impacts
cyclic performance. The comparison between a STR and a TXV is made with R-
22 in both the cooling and heating modes. The cooling mode data indicates that
when the TXV was used, as opposed to the STR, the normalized capacity was
5.4% higher, the normalized power was 3.0% higher and the normalized COP
187
-------
was 2.6% higher. Similar results were obtained in the heating mode. The
normalized capacity increased by 5.7%, the normalized power increased by
4.0%, and the normalized COP increased by 1.6%. These cyclic gains in
performance stem from the steady state purpose of the TXV. The TXV is
designed to maintain a specific superheat leaving the evaporator. When the
superheat is less than desired the TXV reduces the flow area. When the
compressor is turned off the refrigerant in the evaporator approaches a two
phase condition, no superheat. In response to this the TXV reduces the flow
area, which then inhibits the migration of refrigerant during off-time. Therefore,
less time is required to re-establish steady state conditions.
11.3.3 Cyclic Performance Comparison of Refrigerants
The cyclic performance of the different refrigerants is now investigated. In this
vein, R-407C and R-32/134a are compared to R-22. The performance of each
of the mixtures is compared to R-22 in both heating and cooling modes. R-407C
is addressed first, followed by R-32/134a.
The cyclic performance comparison of R-407C to R-22 was conducted
with TXV's in both the heating and cooling modes. In the cooling mode, the
combination of a 1.8% increase in normalized capacity and a 2.9% decrease in
normalized power resulted in a 4 5% increase in normalized COP for R-407C.
188
-------
as compared to R-22. Similar trends were exhibited in the heating mode, where
the 5.4% higher capacity of R-407C and a 5.1 % lower power contribute to a 11 %
higher COP relative to R-22. This data indicates that R-407C attains its steady
state performance faster than R-22. R-407C reaches its steady state
performance faster than R-22 because it attains it's steady state mass flow rate
more rapidly than R-22. This occurs because of the temperature glide of the
mixture. The temperature glide of the mixture results in a higher refrigerant
temperature leaving the evaporator during the early transients. This is made
clear in Figures 11.15 and 11.16. These figures show how the temperatures,
measured at the U bends along the evaporator, vary with compressor on-time.
The action of the temperature sensing bulb and the relatively high temperature
leaving the evaporator cause the thermostatic expansion valve to have a greater
flow area for R-407C than R-22. This translates into a higher refrigerant flow
rate for R-407C. Figures 11.15 and 11.16 also directly support the conclusion
that R-407C attains steady state faster than R-22. As these figures show, there
is a significant difference between the temperature profiles at four and six
minutes for R-22 but for R-407C the difference is less pronounced.
The comparison of R-32/134a to R-22 was conducted with STR's in both
the heating and cooling modes. In the cooling mode the normalized COP,
capacity, and power of R-22 and R-32/134a are essentially identical. This is not
189
-------
the case in the heating mode. The performance of the R-32/134a mixture is
appreciably better than that of R-22. The normalized capacity of R-32/134a is
9.0% better than that of R-22 while it's normalized power is essentially the same
as R-22. The result is a normalized cyclic COP that is 8.3% higher than R-22's,
Since all of these values are normalized with their steady state counterparts, the
large improvement of R-32/134a relative to R-22 is caused by the poor steady
state performance of R-32/134a in the heating mode. This indicates that in some
cases poor performance at steady state may not worsen in the transient mode
of operation.
11.3.4 The Impact of a VLHX on Cyclic Performance
The VLHX was evaluated with R-22 and R-407C. The system was outfitted with
a TXV. As previously mentioned, the VLHX was connected so that it could be
used in the cooling mode only. It is interesting that the cyclic performance of
both fluids improves with the use of the VLHX. The VLHX increased the
normalized capacity of R-22 by 1.5% and it lowered the normalized power by
1.0%. These changes cause the normalized COP to increase by 2.4%. Similar
results were obtained with R-407C. With the VLHX the normalized capacity of
R-407C rose by 1.8% while its normalized power remained unchanged resulting
in a 1.8% increase in normalized COP. Possible explanations for this
phenomena follow. During the initial transients a two phase condition exists at
190
-------
the evaporator outlet because of the high evaporating pressure and temperature
(see section 11.3.1 as well as Figures 11.15 and 11.16). The two phase
condition leaving the evaporator normally results in lost capacity. However, the
suction line heat exchanger recoups some of this lost capacity by removing heat
from the incoming high pressure stream which results in an improved cyclic
performance. It should be noted that the suction line heat exchanger had a
mean effectiveness of 0.33. Accordingly, some of the trends exhibited here may
be more pronounced with a more efficient suction line heat exchanger.
11.3.5 The Impact of the Mode of Operation on Cyclic Performance
The impact that the mode of operation has on cyclic system performance is now
addressed. As with the steady state tests, the increased temperature lift
associated with the heating mode of operation causes the performance of the
system to be lower for all refrigerants. For example, when comparing the heating
mode performance to the cooling mode performance for R-22 the normalized
capacity decreased by 22% for both TXV and STR tests. Furthermore, the same
data showed that the normalized COP decreased by 16% and 15% for the TXV
and STR tests, respectively. These trends are not restricted to R-22, R-407C
and R-32/134a exhibit similar behavior. Since the data is normalized with the
steady state performance, this indicates that the cyclic performance is even more
sensitive to the temperature lift than is the steady state performance.
191
-------
11.4 Seasonal Performance Comparisons
The seasonal performance is used to evaluate the net impact of the cyclic and
steady state performances. The seasonal performance, as described in Chapter
8,0, is used to evaluate the three different system configurations and two
different refrigerants. The refrigerants investigated are R-22 and R-407C. The
system configurations studied include comparisons between STR's and TXVs
and comparisons of system performance with and without a VLHX.
11,4.1 The Impact of the Expansion Device on Seasonal Performance
The overall seasonal performance is used to quantify the combined effect of the
lower cyclic and steady state performance of the STR with respect to the TXV.
Table 11.14 shows the cooling seasonal performance and two heating seasonal
performances for several system configurations and refrigerants. Since there is
not one standard heating climate, as there is for cooling, the extreme heating
cases are evaluated. In the cooling mode the STR causes a 3.6% reduction in
performance as compared to the TXV. In the heating mode the degradation
associated with the STR ranges from less than a percent to 3.9%.
Practically speaking, the TXV is the superior expansion device. Not only
does it produce higher efficiencies and capacities over a range of conditions it
also guards against compressor flooding. The disadvantage of the TXV is its
192
-------
cost. A TXV costs approximately 2 to 4 times that of a STR (Nungsser, 1996).
11.4.2 Comparison Between R-22 and R-407C
When comparing the cooling seasonal performance of R-407C and R-22, the
replacement refrigerant shows a 4.3% degradation in performance. The heating
seasonal performance of the mixture under performs the pure refrigerant by 1.5%
to 7.0%. The degradation is less for the more severe climate because a greater
fraction of the heating load is handled by electric resistance heaters for both
fluids.
As seen in Table 2.4, R-407C has no direct impact on the ozone layer
when compared to R-22. However, the total effect R-407C has on global
warming is not so obvious. The GWP of R-407C is lower than R-22's, but R-
407C requires more energy to cool and heat, as evidenced by the lower CSPF
and HSPF. The total impact of R-22 and R-407C on global warming is evaluated
by summing the direct and indirect contribution of each refrigerant to global
warming. The direct impact is evaluated in terms of the equivalent mass of C02
which would be needed to produce the same amount of global warming as each
refrigerant. If it were to assumed that all of the refrigerant in the system would
be released to the environment over the life of the system then the direct impact
is the amount of refrigerant in the system multiplied by the GWP in Table 2.4.
193
-------
The indirect impact of the refrigerant on global warming is evaluated by
estimating the amount of C02 produced to generate electricity to power the
AC/HP. In order to determine the indirect global warming, the power plant
generation efficiency, and the amount of C02 produced per unit of primary
energy must be determined. The power generation efficiency used was 0,343.
This number was calculated by taking the total amount of electrical energy
consumed by end users and dividing that by the total amount of primary energy
consumed by electrical energy producers in the U.S. (Energy Information
Administration, 1992). Hence, the calculated efficiency includes all of the losses
associated with the supply of electricity (i.e. transmission, distribution,
conversion, etc.). Coal is the primary fuel source for producing energy (Energy
Information Administration, 1992). Based on an average coal composition and
the heat of combustion, the amount of C02 produced per unit of primary energy
for coal was calculated to be 0.66 kg C02/kWh. These results are used to
calculate the total amount of C02 produced by both fluids as a function of system
life, as seen in Figure 11.17. The figure shows that R-22 has a lower total
impact on global warming than R-407C for a system life that exceeds 5.5 years.
Since AC/HP systems typically last much longer than 5 5 years, R-22 has a lower
overall impact on global warming than does R-407C.
194
-------
11.4.3 The Impact of a VLHX on Seasonal Performance
The benefit of the VLHX for both R-22 and R-407C, as quantified by the
seasonal performance, is less than one percent. As such, the measured benefit
is within the uncertainty of the seasonal performance calculation, Therefore it
can be concluded that the VLHX tested is of no benefit when considering
seasonal performance. As mentioned earlier, the particular VLHX tested had an
effectiveness of 0.33, which implies that a more effective VLHX may provide
some benefit.
11.5 Concentration Measurements
The circulated concentration of R-407C and R-32/134a was measured as
function of time as outlined in Section 7.6. The experimental data is presented
first for R-407C and then for R-32/134a. This is followed by an explanation of
the observed trends.
Concentration measurements were made with R-407C at the D" test
condition using the methods described in Section 7.6. A typical data set is
shown in Figures 11.18 and 11.19. These figures represent the concentration
of R-134a in the liquid and vapor lines of the evaporator as a function of
compressor on-time. The concentration of R-32 and R-125 are not displayed in
these figures since their concentration relative to each other did not change. It
195
-------
should be noted that during the first 150 seconds a two phase condition exists
at the sampling ports. Refrigerant is not sampled during this time since it is
unlikely that the sample would represent the bulk flow. With this in mind, there
are several interesting aspects of these figures. The first point is that the vapor
concentration of R-134a starts significantly lower than the charged and steady
state concentrations. This indicates that initially there is vapor in the vapor line
since the equilibrium vapor concentration is relatively low in R-134a. This
implies that the fluid initially entering the compressor is significantly richer in the
more volatile and flammable R-32. A second observation is that the steady state
R-134a concentration of the liquid and vapor lines is less than the charged
concentration. Specifically, the steady state circulated concentration of R-134a
is 3.0 wt.% less than the charged concentration. In other words, the circulated
concentration has shifted towards the more volatile component. This could have
occurred if the oil which circulates with the refrigerant preferentially absorbed R-
134a. However, mineral oil, which does not absorb R-134a, was used for this
experiment. The reason behind the steady state concentration shift is addressed
after presenting the data from R-32/134a.
Similar results were obtained with R-32/134a. R-32/134a's concentration
was measured at the D' test condition and the 47C test condition. The D' test
results are quite similar to R-407C's results, as can be seen from Figure 11.20.
196
-------
In these equations, G is the mass flux, q is the quality, a is the void fraction, and
p is the density. Recall from Section 9,3 that the void fraction is the cross
sectional area of the pipe occupied by vapor divided by the total cross sectional
area of the pipe. The slip between the phases arises from the force balance
between the phases and the no-slip criteria at the interface. The no-slip criteria
requires that the velocity of the liquid must equal the velocity of the vapor at the
interface. Equation 11.3 represents the force balance between the liquid and
vapor,
u dUvaP = n dUIl 11 3
dr dr
Since the absolute viscosity of the liquid is more than an order of magnitude
greater than the absolute viscosity of the vapor, the velocity gradient in the vapor
must be greater. To satisfy this criteria and the no-slip condition at the interface,
the bulk velocity of the vapor must be greater than the bulk velocity of the liquid.
Therefore, the slip is greater than unity when the refrigerant is flowing and equal
to unity when the refrigerant is not flowing. It is possible to show that as the slip
increases so does the fraction of the pipe's cross section that is occupied by
liquid. This is achieved by manipulating the definition of slip and Equations 11.1
and 11,2 to express the void fraction as a function of slip. The result is Equation
11.4.
198
-------
The major difference is that the circulated concentration did not differ as much
from the charged concentration. In this case, the steady state circulated
concentration shifted towards the more volatile component by 0.5 wt.%, which
is approaching the experimental uncertainty of 0.27 wt.%.The heating mode
results can be seen in Figure 11.21. In this case, the heat exchanger where the
sampling occurs acts as a condenser. Relative to the cooling mode, there is
more scatter in the heating mode data. Regardless, the steady state circulated
concentration of R-134a still remains below the charged concentration. The
explanation for this phenomena, which is related to the difference in the vapor
and liquid velocities, follows.
The velocity difference between the bulk of the liquid and the bulk of the
vapor is often called the vapor-liquid slip. The slip, S, is defined as the ratio of
the bulk vapor velocity to the bulk liquid velocity. The bulk velocities of the liquid
and vapor are defined in Equations 11.1 and 11.2.
7T _ G(1 - g)
* " PnP ~ «) 11,1
U,
vap
Gq
P vap a
11.2
197
-------
1 + £z2E(jLjL5)s WA
@iiq
From Equation 11.4 it is clear that as the slip increases the area occupied by
vapor decreases while the area occupied by liquid increases. Therefore, at a
given quality the volume occupied by liquid increases with an increasing slip.
Hence, when the refrigerant is flowing and there is slip, there is a greater volume
occupied by liquid as compared to when refrigerant is not flowing. The
concentration shift at steady state occurs because a greater fraction of the
refrigerant is in the liquid phase and the liquid phase contains a greater fraction
of the less volatile component in order to satisfy thermodynamic equilibrium.
Therefore, a greater fraction of the less volatile component is stored in the liquid
phase of the refrigerant when the system is running. The net effect is that the
circulated concentration is richer in the more volatile component. This effect has
also been documented by Chen (Chen and Kruse, 1995).
11.6 References
Chen, J., Kruse, H., 1995, "Calculating Circulation Concentration of Zeotropic
Refrigerant Mixtures", HVAC&R Research, Vol. 1, No. 3, pp. 219-231.
Domanski, P., 1994, "Theoretical Evaluation of the Vapor Compression Cycle
With a Liquid-Line/Suction-Line Heat Exchanger, Economizer, and Ejector",
NISTIR 5606.
Godwin, D.S., 1993, "Results of Soft-Optimized System Tests in ARl's R-22
199
-------
Godwin, D.S., 1993, "Results of Soft-Optimized System Tests in ARI's R-22
Alternative Refrigerants Evaluation Program", Proceedings of the 1993
international CFC and Halon Alternatives Conference, Washington, D.C., pp. 7-
12.
Nungsser, Roy, 1996, Engineer at Parker-Hannifin, Personal Conversation.
Richardson, H., Gatecliff, G.,1992, "Comparison of the High Side vs. Low Side
Scroll Compressor Design", International Purdue Compressor Conference, Vol.2,
pp. 14-17.
U.S. Energy Administration, 1992, "Annual Energy Review", p. 3.
200
-------
Table 11.1 Charge Optimization Results
Refrigerant
Expansion
Device
Mode
(test)
Optimum
Charge
kg (lb)
COP
fan power
included
COP
fan power
excluded
Random Error
/Total Error-*
+0.1%
/±0.2%
±1.5%
/±3.0%
+1.5%
/± 3.0%
R-22
TXV
cool
(B)
6.54
(14.40)
3.96
5.11
R-407C
TXV
cool
(B)
6.81
(15.00)
3.71
4.68
R-22
STR
cool
©
6.92
(15.25)
3.83
4.91
R-32/134a
STR
cool
©
6.36
(14.00)
3.79
5.07
R-22
STR
heat
(¦47 S)
5.68
(12.50)
3.50
4.24
R-32/134a
STR
heat
(47S)
5.45
(12.00)
3.24
3.88
Table 11.2 R-22 Steady State Results (TXV)
Test
COP
Capacity
(kW)
Energy Balance
Error (%)
Random Error
/Total Error -4
±1.5%
/±3.0%
±1.2%
(±2.7%
n/a
A
3.36
9.52
0.6
B
3.96
9.97
2.0
C
3.87
9.67
2.1
47S
3.56
9.83
1.5
17L
2.39
6.02
-0.6
note: COP and Capacity are calculated with fan power
201
-------
Table 11.3 R-22 Steady State Results (STR)
Test
COP
Capacity
(kW)
Energy Balance
Error (%)
Random Error
/Total Error ->
+1.5%
/+3.0%
±1.2%
/+2.7%
n/a
A
3.20
9.49
1.6
B
3.85
9.91
0.6
C
3.56
9.08
2.3
47 S
3.42
9.68
0.0
17L
2.28
5.94
-1.4
note: COP and Capacity are calculated with fan power
Table 11.4 R-4Q7C Steady State Results (TXV)
Test
COP
Capacity
(kW)
Energy Balance
Error (%)
Random Error
/Total Error
±1.5%
/+ 3.0%
±1.2%
H2.7%
n/a
A
3.06
9.52
1.3
B
3.71
10.06
1.4
C
3.60
9.74
1.5
47 S
3.23
9.85
-0.4
17L
2.28
6.11
-1.4
note: COP and Capacity are calculated with fan power
202
-------
Table 11.5 B Test Cycle Parameters (TXV)
ITEM
R-22
R-22 with
VLHX
R-407C
R-407C
with VLHX
Charge (kg)
6.54
6.54
6.81
6.81
COP
3.96
3.93
3.71
3.73
Capacity (kW)
9.97
9.90
10.06
10.08
Compressor Power (kW)
2.04
2.04
2.24
2.23
Mass Flow Rate (kg/s)
5.56e-2
5.46e-2
5.66e~2
5.53e-2
Suction Pressure (kPa)
572.5
568.4
573.6
572.5
Discharge Pressure (kPa)
1460
1455
1606
1603
Pressure Ratio
2.55
2.56
2.80
2.80
Discharge Temperature
(°C)
80.7
84.0
74.7
78.2
Superheat (°C)
12.3
12.0
10.3
10.1
Subcoolina (°C)
4,8
7 8
2 8
5 3
note: COP and Capacity are calculated with fan power
203
-------
Table 11.6 47S Test Cycle Parameters (TXV)
ITEM
R-22
R-407C
Charge (kg)
6.54
6,81
COP
3.56
3.23
Capacity (kW)
9.83
9.85
Compressor Power (kW)
2.30
2.58
Mass Flow Rate (kg/s)
49.7
50.3
Suction Pressure (kPa)
490.1
494.6
Discharge Pressure (kPa)
1637
1830
Pressure Ratio
3,34
3.70
Discharge Temperature (°C)
77.4
72.8
Superheat (°C)
2.8
0.7
Subcooling (°C)
5.5
3.9
note: COP and Capacity are calculated with fan power
Table 11.7 Steady State Test Results (STR)
Test
Refrigerant
Optimum
Charge
kg (lb)
COP
Capacity
(kW)
Energy
Balance
Error (%)
Random Error
/Total Error -*
±0.1%
/±0.2%
±1.5%
I± 3,0%
±1.2%
/±2.7%
n/a
C
R-22
6.92
(15.25)
3.83
(4.91*)
9.75
(10.07*)
0.1
47 S
R-22
5.68
(12.50)
3.50
(4.29*)
9.62
(9.22*)
0.7
C
R-32/134a
6.36
(14.00)
3.94
(5.07*)
9.88
(10.20*)
0.6
47 S
R-32/134a
5.45
f12.00)
3.24
C3 88*1
9.35
(8.95*)
2.1
note: * Quantities are calculated without fan power while others are calculated with fan power
204
-------
Table 11.8 C Test Cycle Parameters (STR)
ITEM
R-22
R-32/134a
Charge (kg)
6.92
6.36
COP
o oo
o.oo
3.94
Capacity (kW)
9.75
9.88
Compressor Power (kW)
2.05
2.01
Mass Flow Rate
5.79e-2
5.20e-2
Suction Pressure (kPa)
584.8
564.2
Discharge Pressure (kPa)
1497
1467
Pressure Ratio
2.56
2.60
Discharge Temperature (°C)
70.0
66.7
Superheat (°C)
1.7
0.4
Subcooling (°C)
5.3
1.2
note: COP arid Capacity are calculated with fan power
205
-------
Table 11.9 47S Test Cycle Parameters (STR)
ITEM
R-22
R-32/134a
Charge (kg)
5.68
5.45
COP
3.50
3.24
Capacity (kW)
9.62
9.35
Compressor Power (kW)
2.15
2.31
Mass Flow Rate (kg/s)
4.87e-2
4.25e-2
Suction Pressure (kPa)
497.7
461.4
Discharge Pressure (kPa)
1548
1615
Pressure Ratio
3.11
3.50
Discharge Temperature (°C)
72.8
68.0
Superheat (°C)
6.3
2 Phase
Subcooling (°C)
2.4
2.5
note: COP and Capacity are calculated with fan power
Table 11.10 R-22 Steady State Results (TXV & VLHX)
Test
COP
Capacity
(kW)
Energy Balance
Error (%)
Random Error
/Total Error-#
±1.5%
/±3.0%
±1.2%
i+2.7%
n/a
A
3.39
9.54
1.4
B
3.93
9.90
1.1
C
3.85
9.61
1.9
47S
17L
-
note: COP and Capacity are calculated with fan power
206
-------
Table 11.11 R-407C Steady State Results (TXV & VLHX)
Test
COP
Capacity
(kW)
Energy Balance
Error {%)
Random Error
/Total Error-*
±1.5%
/±3.Q%
±1.2%
/±2,7%
n/a
A
3.10
9.52
1.6
B
3.73
10.08
1.9
C
3.61
9.72
1.7
47S
-
17L
-
-
note: COP arid Capacity are calculated with fan power
207
-------
Table 11.12 Cyclic Tests
Data
Set
Refrigerant
Expansion
Device
On Time/
Off Time
Mode of
Operation
1
R-22
TXV
6 min./
24 min.
Cooling Mode
2
R-22
TXV
6 min./
24 min.
Cooling Mode
3
R-22
STR
6 min./
24 min.
Cooling Mode
4
R-22
STR
30 min./
30 min.
Cooling Mode
5
R-407C
TXV
6 min./
24 min.
Cooling Mode
6
R-407C
TXV
6 min./
24 min.
Cooling Mode
7
R-32/134a
STR
6 min./
24 min.
Cooling Mode
8
R-32/134a
STR
30 min./
30 min.
Cooling Mode
9
R-22
TXV
6 min./
24 min.
Heating Mode
10
R-22
STR
6 min./
24 min.
Heating Mode
11
R-22
STR
30 min./
30 min.
Heating Mode
12
R-407C
TXV
6 min./
24 min.
Heating Mode
13
R-32/134a
STR
6 min./
24 min,
Heating Mode
14
R-32/134a
STR
30 min./
30 min.
Heating Mode
208
-------
Table 11.13 Non-Dimensional Cyciic Performance
Data
Set
Q , /Q
^on-time' **3.3,
^orc-time^s.s.
rnp
V-fWI on_tjme
/COP,s.
Random Error
/Total Error -»
±0.4%
/±2.5%
±0.1%
I± 0.3%
±1.5%
/+3.6%
1
0.782
1.04
0.752
2
0.794
1.03
0.770
3
0.742
1.01
0.733
4
0.949
1.01
0.942
5
0.796
1.01
0.788
6
0.810
1.01
0.802
7
0.740
1.01
0.737
8
0.943
0.997
0.945
9
0.609
0.961
0.634
10
0.576
0.924
0.624
11
0.880
0.986
0.892
12
0.642
0.912
0.704
13
0.629
0.930
0.676
14
0.956
0.998
0.958
note: COP and Capacity are calculated without fan power
209
-------
Table 11.14 Overall Seasonal Performance
R-22
R-22
R-22
R-407C
R-407C
PARAMETER
w/o
w/o
w/
w/o
w/
VLHX
VLHX
VLHX
VLHX
VLHX
(STR)
(TXV)
(TXV)
(TXV)
(TXV)
CSPF
3.46
3.59
3.58
3.44
3.48
(Btu/W-hr)
(11.81)
(12.25)
(12.22)
(11.74)
(11.88)
HSPF
2.73
2.84
2.64
REGION I
(9.32)
(9.69)
(10.03)
(Btu/W-hr)
HSPF
1.31
1.32
1.30
_
REGION V
(4.47)
(4.51)
(4.44)
(Btu/W-hr)
note: CSPF and HSPF are calculated with fan power
210
-------
Charge/Optimum Charge
Figure 11.1 Charge Optimization Curve (R-22. STR, C test)
0,6 0.8 1 1.2 1.4
Charge/Optimum Charge
Figure 11.2 Compressor Power Versus Charge (R-22, STR, C test)
211
-------
1,600
1,400
1,200
§ 1,000
800
600
400
Compressor Suction
Pressure
Compressor Discharge
Pressure
0,6
0.8 1
Charge/Optimum Charge
1.2
1.4
Figure 11.3 Suction arid Discharge Pressures Versus Charge
(R-22, STR, C test)
0.8 1
Charge/Optimum Charge
Figure 11.4 Mass Flow Rate Versus Charge (R-22, STR, C test)
212
-------
7,5
0.6
0.8 1 1.2
Charge/Optimum Charge
1.4
Figure 11.5 Capacity Versus Charge (R-22, STR, C test)
Charge/Optimum Charge
Figure 11,6 Superheat Versus Charge (R-22, STR, C test)
213
-------
0,6 0.8 1 1.2 1.4
Charge/Optimum Charge
Figure 11.7 Charge Optimization Curves (R-22 SIR, C test) &
(R-22, TXV, C test)
1.05
of ^
¦
_
R-22 R-407C
TXV (cool) TXV (cool)
i.l,
j . i
Q.
o
o 0.95
E
Q. 0.9
o
0.85
0.6
0.6 0.8 1 1.2 1.4
Charge/Optimum Charge
Figure 11,8 Charge Optimization Curves (R-22, TXV, B test) &
(R-407C, TXV, B test)
214
-------
1,05
a.
O
o 0,95
£
3
E
H3
Q.
o
a 0.9
O
o
0,85
0.8
R-22 R-32/134a
STR (cool) STR (cool)
0,6
0,8 1
Charge/Optimum Charge
1,2
1,4
Figure 11.9 Charge Optimization Curves (R-22, STR, C test) &
(R-32/134a, STR, C test)
1,05
CL
o
O 0,95
a.
O
3l 0,9
O
o
0.85
0.8
R-22 R-32/134a
STR (heal) STR (heat)
0.6
0,8
1,2
Charge/Optimum Charge
1.4
Figure 11.10 Charge Optimization Curves (R-22, STR, 47S test) &
(R-32/134a, STR, 47S test)
215
-------
Time (minutes)
Figure 11.11 Refrigerant Temperatures Versus Time
(R-22, STR, D' test)
Time (minutes)
Figure 11.12 Suction and Discharge Pressures Versus Charge
(R-22, STR, D' test)
216
-------
1 -
0 5 10 15 20 25 30
Time (minutes)
Figure 11.13 Compressor Power Versus Time (R-22, STR, D' test)
15
£10
10
15
Time (minutes)
20
25
30
Figure 11.14 Air Side Cooling Capacity Versus Time
(R-22, STR, D' test)
217
-------
30
25
o
3
20
15
10
2 3
Position in Heat Exchanger
0 min
6 min
OUTLET
Figure 11.15 Refrigerant Temperature Along Evaporator
(R-22, TXV, D test)
30
25
0
1
20
15
10
1
INLET
0 min t
— -a— —
*-—
4 & 6 min
-
_
2 min
—
i
i
2 3 4
Position in Heat Exchanger
5
OUTLET
Figure 11.16 Refrigerant Temperature Along Evaporator
(R-407C, TXV, D test)
218
-------
Figure 11.17 Impact of R-22 and R-407C on Global Warming Versus
Life Time of Equipment
100
80
• R-407C
• D1 test
• Vapor
Line
1 60
C
-------
100
80
I 60
• R-407C
• D' test
• Liquid
Line
. R-134a Charged Concentration
40
20
R-134a Measured Concentration
10
20
Time (minutes)
30
40
Figure 11.19 R-407C Concentration Versus Time (Liquid Line)
Figure 11.20 R-32/134a Concentration Versus Time (D' Test, Vapor Line)
220
-------
100
80
60
• R-32/134a
• 47C test
• Vapor
C
-------
Chapter 12
Simulation Results and Discussion
12.1 introduction
As previously stated, the primary goal of the simulation is to develop a better
understanding of the processes occurring during the steady state and transient
operation of the AC/HP. This goal is achieved by studying the simulation results
from standard system configurations and by examining the results of several
simulation runs in which different parameters are varied. However, before doing
this the void fraction correlation must be chosen. The void fraction correlation
is selected based on how well the simulation predicts the charge optimization
curve when using that void fraction correlation. This point is addressed in the
following section.
12.2 Selection of the Void Fraction Correlation
The simulation was used to generate a charge optimization curve for each of the
void fraction correlations considered. This was done to find the void fraction
correlation that worked the best with the simulation to predict the experimental
charge optimization curve, This process is necessary since there are numerous
void fraction correlations currently available, but none are designed specifically
222
-------
for this application. In addition to the homogeneous void fraction model, the void
fraction correlations investigated in this work are those by Anasova (Anosova et
al., 1990), Budrick (Budrick et al., 1990), Hughmark (Hughmark, 1962), and Zivi
(Zivi,1964). These correlations were chosen based on a review of the literature
and the recommendations of Farazad, Orth, and Rice (Farazad and O'Neal,
1994), (Orth et al., 1995), (Rice, 1987). The charge optimization data was
obtained for the C test conditions. Some of the charge optimization curves
generated by the simulation can be seen in Figure 12.1 and are plotted
alongside the experimental data. The best simulation results were obtained with
the Zivi correlation, which resulted in a -4.7% error in the optimum COP and a
-15.4% error in the optimum charge. Al! of the other correlations produced
similar COPs but varied significantly in the optimum charge. The homogenous
model and the Anasova correlation produced optimum charges that were -49%
and -23% different from the experimental results, respectively. On the other
hand, the Hughmark and Budrick correlations produced optimum charges that
were +38% and +43% different from the experimental results. Since the Zivi
correlation performed the best it is used for the remainder of simulation data
presented here.
There are several pertinent observations regarding Figure 12 1.
Specifically, Figure 12.1 shows that the void fraction correlation has a
223
-------
significantly greater impact on the optimum charge than on the optimum COP.
An additional observation is that the simulation does not accurately predict the
optimum charge utilizing any of these correlations. There are two probable
sources for this discrepancy. One is an error in calculating the system volume
and the other is errors associated with the void fraction correlation. The
refrigerant charge is very sensitive to the calculated system volume. Hence,
errors in calculating the system volume may be responsible for the errors in the
refrigerant charge, This concern was put to rest by indirectly measuring the
volume of the system and comparing it to the calculated volume. The volume
was measured by charging the system with a known amount of nitrogen and then
measuring the system pressure and temperature. The temperature and pressure
are used to calculate the density of nitrogen, which together with the mass of
nitrogen, is used to determine the volume of the system. This procedure showed
that the calculated volume was in error by less than 2.5%. Hence, it is unlikely
that this is the source of the discrepancy between the theoretical and
experimental results. The other probable source of the error is the void fraction
correlation. Like any correlation for a quantity in the two phase region, the void
fraction correlations have relatively high errors. Comparisons by Chexal show
that void fraction correlations routinely have errors in excess of 20% (Chexal et
a!., 1992). Therefore, it is likely that the error in the predicted optimum charge
stems from errors in the void fraction correlation.
224
-------
12.3 Steady State Simulation Test Results
12.3.1 Charge Optimization Results
Several system parameters generated by the simulation are compared to
experimental data as the charge in the system is varied, The parameters of
interest include subcooling, superheat, compressor power, and cooling capacity.
Figures 12.2 through 12.5 are plots of these quantities versus refrigerant charge.
These figures are obtained from simulated and experimental tests with R-22 at
the C test conditions.
The compressor power and the evaporator cooling capacity are depicted
in Figure 12.2. When compared with the experimental data from Chapter 11,
Figure 12.2 indicates that the simulated trends associated with compressor
power are accurately represented. However, the same does not hold true for the
cooling capacity. Specifically, the simulation results show that the cooling
capacity asymptotically increases while the experimental results indicate a sharp
maximum. The difference between the two arises because the simulation does
not predict the optimum COP to be at low evaporator outlet superheats, as was
the case experimentally. Experimentally, a small increase in charge beyond the
optimum causes a two phase condition to be present at the outlet of the
evaporator. This represents lost cooling capacity which produces steep negative
slope in the experimental cooling capacity.
225
-------
The suction and discharge pressures are depicted in Figure 12.3. The
simulation does a good job of reproducing the trends and values associated with
the experimental data. This level of accuracy is not obtained for the mass flow
rate data as shown by Figure 12.4. The predicted values are significantly lower,
10.4% at the optimum, and are not as sensitive to charge. This discrepancy,
which is caused by the STR model, is addressed in Section 12.3.3. However,
the trend of increasing mass flow rate with increasing refrigerant charge is
correct.
It is interesting to observe the influence that subcooling has on superheat
for the simulation data, as seen in Figure 12.5. Figures 12.3 and 12.4
demonstrate that this influence also extends to the refrigerant pressures and
mass flow rate. All of these parameters have a distinct change in slope at the
inception of subcooling. The reason for the sensitivity of these parameters to the
existence of subcooling is that the density at the inlet to the STR is sensitive to
subcooling. When there is a two phase condition at the inlet of the STR, the
density of the fluid at the inlet is very sensitive to refrigerant charge. This is in
contrast to when liquid enters the STR. Therefore, the increased density
sensitivity translates into a greater mass flow rate sensitivity, this in turn is
responsible for the change in slope of the other parameters. The relationship
226
-------
between density and mass flow rate through the STR can be clearly seen in
Equation 9.41.
12.3.2 Typical Steady State Simulation Results
Typical steady state cooling and heating tests are now examined. The data
discussed in this context utilizes R-22 as the refrigerant. As with the
experimental data, numerous system parameters can be obtained from the
simulation. Those of interest are outlined in Tables 12.1 through 12.4. Several
points can be made regarding the steady state data from the simulation.
Throughout the AC/HP the state and amount of the refrigerant is known
and therefore, virtually any quantity of interest can be determined from the
simulation. This is the primary advantage the simulation has over experiments.
Examples of this are in Tables 12.3 and 12.4, which contain the pressure drop
and amount of refrigerant associated with each component for the cooling and
heating modes, respectively. For the cooling mode, over 56% of the refrigerant
is in the condenser, the outdoor heat exchanger. On the other hand, the indoor
heat exchanger, which is the evaporator, only contains 10.5% of the refrigerant
charge. In the heating mode the trend is reversed in that the indoor heat
exchanger has more refrigerant than the outdoor. Specifically, 34.4% of the
refrigerant is in the indoor heat exchanger and 28.5% is in the outdoor heat
227
-------
exchanger, in this mode of operation the indoor heat exchanger is the
condenser and the outdoor heat exchanger is the evaporator. Therefore,
regardless of the mode of operation, the condenser contains more refrigerant
than any other component in the system This is because the condenser
contains high pressure refrigerant which has a significantly greater density. It
is interesting to note that in this particular system in the cooling mode there is
more refrigerant in the liquid line than in either the evaporator, compressor, or
vapor line. This is a result of the high density of liquid and the exceptionally long
length of pipe which was necessary for experimental reasons. The long length
of liquid line is also responsible for the relatively high pressure drop of this
component, as seen in Table 12.3. In cooling and heating modes, the pressure
drop of the evaporator is several times larger than that of the condenser. This
is expected during the evaporation process since the refrigerant is accelerated
which augments the frictional pressure drop. The converse occurs during flow
condensation. Specifically, during condensation the refrigerant is decelerated
causing a pressure recovery which offsets some of the frictional pressure drop.
The points regarding pressure drop, as well as others, can also be
illuminated through the use of the p-h diagram. The simulation data was plotted
on the p-h diagram in Figures 12.6 and 12.7 for the cooling and heating modes,
228
-------
respectively. These figures represent more than just the inlet and outlet of each
component. In fact there are over 350 intermediate points plotted on each figure.
However, the path of the expansion in the STR is not known. Only the inlet and
outlet state are known, but dashed lines are drawn to connect these points to
complete the cycle. The relative magnitude of the pressure drop associated with
each component can be better appreciated in Figures 12.6 and 12.7. The
amount of subcooling and superheat at various locations throughout the cycle
is also better appreciated through the use of these figures . Furthermore, the
considerable differences between the ideal vapor compression cycle (i.e. no
pressure drop, no subcooling, no superheat, perfect heat exchange, etc.) and the
actual cycle can readily be determined from the p-h diagram. Since the ideal
cycle would operate between the saturation pressures of the source and sink
temperatures, the ideal cycle would greatly under predict the pressure levels of
the actual system. As a result, an idealized cycle analysis would greatly over
predict the COP of the real system.
12.3.3 Steady State Comparison with Experimental Data
The performance of the simulation is evaluated by comparing it to the
experimental results discussed in Chapter 11. Tables 12.5 and 12.6 show the
deviation between the simulation and the experimental results for the cooling and
heating test, respectively. The simulation consistently under predicts the optimal
229
-------
refrigerant charge for both fluids and both modes of operation. As stated earlier,
this error is attributed to the lack of an appropriate void fraction model. Another
systematic error which occurs throughout the data is that the simulation
consistently over predicts the pressure ratio. This is a consequence of using the
orifice equation to model the expansion process. The STR model consistently
over predicts the pressure drop through the device at the optimal charge. This
has several effects. The obvious impact is that it lowers the suction pressure of
the compressor. This in turn causes the flow rate through the system to be
reduced due to the lower suction vapor density. Furthermore, the lower mass
flow rate tends to reduce the capacity of the system. Hence, the lack of a
generally valid STR correlation is responsible for a large fraction of the errors
associated with the simulation. However, this does not take away from the
simulation's ability to accurately predict most trends. There is one notable
exception, however, and that is in regard to the performance of R-32/134a
relative to R-22 in the heating mode. The cause of this error is also routed in the
short tube restrictor model, in Chapter 11 it was pointed out that the unexpected
performance degradation of R-32/134a was due to the large pressure ratio
associated with this refrigerant and the relatively large STR These factors
combine to cause a two phase condition at the inlet to the compressor. Since
the current STR model is over predicting the pressure drop through the STR, a
two phase condition did not exist at the inlet to the compressor in the simulated
230
-------
results. Hence the negative consequences of a two phase fluid entering the
compressor are not encountered. However, the simulation accurately predicts
the lower superheat of R-32/134a relative to R-22 in both the heating and cooling
modes.
12.3,4 Steady State Comparison of the Refrigerants
The trends examined in this section are between R-32/134a and R-22 and the
parameters of interest are presented in Tables 12.1 and 12.2 for the cooling and
heating modes, respectively. While the simulation reproduces most of the trends
already discussed in Chapter 11, it also provides information regarding other
points. One such point is the circulated concentration. Obviously, for R-22 the
circulated concentration does not change between modes of operation.
However, for R-32/134a the same cannot be said. In the heating mode there is
a greater shift in the circulated concentration than there is in the cooling mode.
There are two reasons for this. One is that there is less charge in the heating
mode and as a result a greater fraction of the less volatile component is in the
heat exchangers. The other reason is that at the lower pressure associated with
the heating mode there is a greater concentration difference between the phases
at equilibrium. This also causes more of the less volatile R-134a to be held in
the liquid phase of the heat exchangers. It is also interesting to compare the p-h
diagrams of R-32/134a, Figures 12,8. and 12.9, to those for R-22. The most
231
-------
striking point regarding these figures is the slanted isotherms associated with the
zeotropio mixture, R-32/134a. However, based on the close proximity of the
evaporator outlet to the source temperature there appears to be no ill effects due
to the temperature glide. In other words, the approach temperature of the
evaporator is nearly the same for both fluids. Another observation from these
figures is that R-32/134a has a lower pressure drop through the various system
components than does R-22. This primarily has to do with the lower mass flow
rate of R-32/134a, but a second order effect is the lower viscosity of the mixture.
12.3.5 The Influence of Connecting Piping on Steady State Performance
The connecting piping is the tubing which allows the refrigerant to flow to and
from the indoor and outdoor units of an AC/HP. The connecting piping is
detrimental for system performance because it causes additional pressure drop.
The additional pressure drop has two effects. Firstly, it increases the work of the
compressor by increasing pressure difference which it must overcome.
Secondly, the density of the refrigerant entering the compressor is reduced by
the pressure drop from the evaporator to the compressor. This causes the mass
flow rate through the system to be reduced which in turn results in a lower
cooling capacity. To quantify these effects the simulation was run for the C test
conditions with and without connecting piping. The refrigerant charge used for
the case without piping was the optimum charge with connecting piping less the
232
-------
mass of refrigerant in the vapor and liquid lines. The simulation results indicate
that the capacity increases by 4.2% when the connecting piping is removed.
Furthermore, the compressor power remained essentially unchanged even
though the compressor was moving 5.3% more refrigerant. The net result is that
removing the connecting piping causes a 4.1 % increase in COP.
There is another steady state effect of the connecting piping that pertains
to zeotropic mixtures. As already discussed, zeotropic refrigerant mixtures have
a circulating concentration that is shifted towards the more volatile component.
This is caused by the accumulation of the less volatile component in the two
phase regions. Hence, without the connecting piping, a greater fraction of the
refrigerant is held in the two phase regions. This results in a greater
concentration shift. To evaluate the magnitude of this effect the simulation was
run at the C test conditions with and without connecting piping. As before, the
refrigerant charge is reduced for the case without connecting piping. The results
indicate that the circulated concentration changes from 30.8/69.2 wt.% (R-
32/134a) with connecting piping to 31.4/68.6 wt.% (R-32/134a) without
connecting piping. Therefore, the concentration shift of the circulated refrigerant
is more significant in systems that have short lengths of connecting piping.
Examples of such systems include window air-conditioners, dehumidifiers, and
car air-conditioners.
233
-------
12.3.6 The Influence of Charge on the Circulated Concentration
Based on arguments from the preceding section, decreasing the amount of
refrigerant in the system will decrease the circulated concentration of the less
volatile refrigerant. The simulation was used to quantify a worst case scenario
where a system with no connecting piping was charged with only half the
optimum amount of refrigerant. This was done at the C test conditions. As the
charge changed from 100% to 50% the circulated concentration changed from
31.4/68.6 wt.% (R-32/134a) to 32.3/67.7 wt.% (R-32/134a). Hence, even if a
system with no connecting piping was only half charged, the circulated
concentration of this mixture would only be 2.3% wt.% richer in the more volatile
component. It is worth noting that the circulated concentration of this worst case
scenario is significantly far away from the flammable range.
12.4 Transient Simulation Results
12.4.1 Step Size Independence
Before any transient data is considered, it is important to show that the data is
independent of step size. Otherwise, the effects studied could be attributed to
discretization errors. The data was shown to be independent of step size by
comparing the integrated refrigerant capacity, compressor power and system
COP for several different time steps. The refrigerant capacity, compressor
power, and system COP were chosen since changes in other variables would be
234
-------
reflected in these. These refrigerant quantities were integrated from 0 to 60
seconds. This time interval was chosen because most of the changes with
respect to time occur in this time frame. In addition, a time frame greater than
a 100 seconds would dampen the effect of the time step since all time steps
approach the same steady state value of COP. The data, which is normalized
with the data at the smallest time step, is shown in Table 12.7. This data was
obtained for R-22 operating in the cooling mode. It is clear from the data that the
step size has minimal influence on the accuracy of the solution. In fact, the
primary constraint on step size is the ability of the equation solver to rapidly
solve the non-linear equations which result from the discretization of the heat
exchanger.
It should be noted that the initial conditions for all of the transient tests,
including the data in Table 11.7 are the same. The initial conditions are
determined through two assumptions. The first assumption is that initially the
refrigerant is assumed to be in thermal equilibrium with the air entering each
heat exchanger. The second assumption is that the pressure is assumed to be
the same throughout the system, since the high and low sides of the system are
able to communicate. As a result of these assumptions, the refrigerant initially
in the evaporator is in the two phase region while the refrigerant initially in the
condenser is in the vapor region.
235
-------
12.4.2 Typical Transient Simulation Results
Several parameters are investigated as a function of time from a typical R-22
cooling test. The connecting piping between the indoor and outdoor heat
exchangers is not included in this analysis. The effect of connecting piping is not
included at this time since it significantly increases the computational time. The
longer computation time is clearly undesirable since, without connecting piping
it takes on the order of 36 hours for a single R-22 run when the IBM compatible
personal computer with a 100 MHz Pentium CPU is used.
The first quantity examined is the compressor's angular velocity. Figure
12.10 is a graph of the compressor's angular velocity in RPM versus time. It is
clear from this figure that the compressor rapidly achieves its steady state value.
The angular velocity reaches 95% of its steady state value in 0.94 seconds. The
rapid acceleration of the compressor is due to the relatively high torque of the
motor and the low inertia of the compressor components.
The rapid acceleration of the compressor and the initially high suction gas
density combine to cause a significant acceleration of the refrigerant at the
compressor inlet and outlet. Figure 12.11 shows the mass flow rate of refrigerant
entering the compressor, condenser, expansion device, and evaporator as a
function of time. In this figure the refrigerant flow rate is normalized with respect
236
-------
to the steady state mass flow rate. It is clear that the refrigerant is rapidly
removed from the evaporator arid deposited in the condenser. The mass flow
rate at its highest point is 2.95 times its steady state value. The mass rate
quickly declines from its maximum when the quality of the refrigerant entering the
compressor shell starts to decrease.
Directly related to the mass flow rate into each component is the mass of
the refrigerant in each component. The mass of the refrigerant in each
component normalized with the total mass in the system is plotted as a function
of time in Figure 12.12. The expansion device represents an exceptionally small
volume and as a result there is essentially no refrigerant stored in the expansion
device. Initially the evaporator and compressor both contain more refrigerant in
them than the condenser. However, the compressor quickly redistributes the
refrigerant so that the condenser has 95% of its steady state charge within 38.14
seconds.
The discharge and suction refrigerant pressures of the compressor are
plotted in Figure 12.13. This figure has several interesting attributes. One of the
more noteworthy features is that the suction side pressure drops 105 kPa below
its steady state value prior to reaching steady state. The suction pressure is
primarily dictated by the inlet conditions of the expansion device. At the point in
237
-------
time where the minimum in low side pressure occurs, the inlet mass flow rate and
pressure are close to their steady state values, but the fluid entering the
expansion device is in the two phase region. This significantly increases the
pressure drop of the expansion device which causes the depression in low side
pressure. Another interesting attribute of Figure 12.13 is that there exists a local
maximum in the discharge pressure which occurs at 25 seconds. The graph of
mass flow rate versus time, Figure 12.11, is useful in explaining this phenomena.
The mass flow rate leaving the condenser, the STR flow rate, monotonically
increases with time but the mass flow rate entering the condenser rises sharply
and falls in response to the suction density. Hence, the peak in pressure occurs
because of the high rate at which refrigerant is being pumped into the condenser
from the compressor and the low rate at which it is being removed by the
expansion device. In other words, refrigerant is accumulating in the condenser
faster than it can be removed and this is what causes the refrigerant pressure to
rise. The refrigerant pressure begins to drop when the inlet mass flow rate of the
condenser begins to approach the outlet flow rate.
The refrigerant temperatures entering each component are plotted versus
time in Figure 12.14. The temperature entering the evaporator closely
reproduces the trends of the low side pressure since it is in the two phase
region. The temperature leaving the evaporator follows the inlet temperature of
238
-------
the evaporator until the exiting refrigerant becomes superheated vapor. The
temperature of the refrigerant leaving the compressor abruptly increases after
it has leveled off. This occurs because in the beginning there is liquid refrigerant
in the compressor shell which supplies the compressor with saturated vapor.
The abrupt increase occurs when there is no longer any liquid in the compressor
shell, and as a result, the compressor receives superheated vapor. The
temperature of the refrigerant leaving the condenser and entering the expansion
device changes little when compared to the other temperatures. This is primarily
due to the large heat transfer area of the condenser which allows the outlet
refrigerant temperature to closely approach the outdoor temperature which
remains constant.
The heat loads of the evaporator and condenser, in conjunction with the
compressor power, are now examined. Each of these quantities are normalized
with their respective steady state values and are presented in Figure 12.15. The
initial sharp peak in power is a consequence of the finite inertia of the
compressor. Additional power is required to get the compressor to full speed.
This additional power requirement is compounded by the low efficiency of the
motor at low RPMs. Also plotted in Figure 12.15 are the air side heat capacities.
Since the air side evaporator capacity slowly rises while the compressor power
rapidly exceeds it's steady state value, the COP of the system is poor when the
239
-------
unit is first turned on. This is depicted in Figure 12.16. The COP rapidly
increases as the refrigerant initially leaves the evaporator. However, the COP
decreases shortly after the compressor is started because refrigerant
accumulates in the compressor shell. This causes the system to behave as if it
were undercharged. More precisely, the evaporation of the refrigerant in the
compressor shell serves no useful purpose. Hence, the capacity and COP drop
off as refrigerant is evaporated in the compressor shell.
12,4.3 Transient Comparison of Refrigerants
The simulation was run with the mixture R-32/134a for the identical conditions
described in Section 12.4.2. The time required for the simulation to complete
this test was nearly 90 hours. There are two reasons for the exceptionally long
computation time. One reason is that the species balance must be solved to
ensure conservation of mass. This additional equation adds roughly 20% to the
time required. However, the major increase in time arises from the EOS. The
added complication of determining the equilibrium properties of a mixture more
than doubles the time required by the EOS. Since the EOS is one of the
innermost routines, doubling the time required by EOS doubles the time required
by the overall simulation.
As with the experimental results, there was essentially no difference in the
240
-------
integrated and normalized power, capacity, and COP between R-32/134a and
R-22. Furthermore, most of the other R-32/134a quantities (mass, mass flow
rate, pressure, etc.) were nearly identical to R-22, which have already been
presented. However, with regard to the refrigerant temperature and the
refrigerant concentration appreciable differences were found to exist.
The refrigerant temperature of R-32/134a at the inlet of the various system
components is plotted versus time in Figure 12.17. Although very similar to the
plot for R-22 there are a few points which make it unique. The most notable
difference is at the compressor inlet. The temperature at the compressor inlet
does not follow the temperature of the evaporator inlet as did R-22 in the early
transients. This is a result of the temperature glide that occurs during the phase
change of a zeotropic mixture. Another difference between Figure 12.17 and the
R-22 plot, Figure 12.14, is the considerably lower condenser inlet temperature.
This is a consequence of R-32/134a's higher specific heat as was discussed in
Chapter 11. Also noteworthy is the slightly lower evaporator inlet temperature
of R-32/134a as compared to R-22.
The refrigerant concentration is the other quantity which makes R-
32/134a unique from R-22. Figure 12 18 is a plot of the R-32 concentration
versus time for the R-32/134a mixture, Initially, the refrigerant leaving the
241
-------
evaporator and entering the compressor is depleted of the more volatile
component, R-32. This is because at start-up the refrigerant in the evaporator
is in the two phase region, and contains less R-32 than the charged
concentration. On the other hand, the vapor which is initially in equilibrium with
the refrigerant in the evaporator is rich in R-32 and occupies the condenser.
Hence, the condenser acts like a reservoir of R-32 and the evaporator acts like
a reservoir of R-134a. As time proceeds the two phase, R-134a rich, refrigerant
enters the compressor. There the compressor shell separates the liquid from the
vapor so that the compressor compresses only R-32 rich vapor. As a
consequence the refrigerant leaving the compressor is still relatively rich in R-32.
However, as the liquid leaves the evaporator and the amount of liquid in the
compressor shell decreases, the R-32 concentration leaving the compressor
starts to decline. This subsequently causes the decline in the R-32 concentration
leaving the condenser. These processes continue until the circulated
concentration is reached at the inlet to each component. It is interesting to note
that at no point in time is the flammable concentration of R-32/134a reached.
12.4.4 The Influence of Connecting Piping on Transient Performance
The computational time required to run the simulation under the same conditions
described in Section 12.4.2 except with the addition of connecting piping
exceeds 90 hours. There are two reasons for this. One is that the piping is
242
-------
discretized like the heat exchangers. This increases the total number of nodes
by 30%, which in turn increases the computational time for each time step by
30%. The other reason is that additional piping doubles the time required to
reach steady state. Therefore, it takes nearly twice as many steps in time to
reach steady state and thereby doubles the already lengthened computational
time.
The refrigerant discharge and suction pressures are plotted versus time
in Figure 12.19. This figure is similar to the data presented without piping except
that the time required to reach steady state has significantly increased. It takes
98.53 seconds to reach the minimum suction pressure with connecting piping
while it only takes 42.06 seconds without piping. The primary reason for this
increase in time is the additional refrigerant charge associated with the
connecting piping. The refrigerant charge with connecting piping has increased
from 4.80 kg to 5.86 kg. Figure 12.20 displays how this charge is distributed
over time. As can be seen in Figure 12.20, the piping adds a significant amount
of refrigerant in the low pressure side of the system (i.e. vapor line, evaporator,
and compressor shell), and as a result a longer amount of time is required to
distribute it.
In addition to delaying steady state, the connecting piping also
243
-------
dramatically effects the performance of the system. To quantify this effect, the
COP, compressor energy and cooling energy over the 6 minute on-time are
compared for the system with and without connecting piping. These terms are
normalized with their respective steady state values. The normalized
compressor energy is not appreciably effected by the presence of connecting
pipes. The normalized compressor energy without connecting piping is 0.9994
and with piping it is 0.9997. However, the cooling energy is significantly
effected. The normalized cooling energy with connecting pipes is 0.8078 and
without connecting pipes it is 0.9197. Hence, the normalized COP is 12.2%
less with connecting piping than without. This is a significantly higher penalty
than the 4.1% incurred at steady state as a result of the connecting piping.
Figure 12.20 also contains the experimental data from Chapter 11. While
the trends in pressure are well represented, the duration of the transients is not.
The experimental suction pressure reached its minimum at 138.0 seconds as
compared to the 98.5 seconds required by the simulation. There are several
items which may account for this discrepancy. One is the compressor
performance during the initial phases of start up. The compressor used for this
research is a scroll type compressor. Before a scroll compressor can efficiently
pump refrigerant there needs to be a certain amount of back pressure on the
scroll set. The back pressure helps seal the surfaces between the orbiting and
244
-------
fixed scrolls. When the compressor is first turned on there is no back pressure
and as a result the volumetric efficiency of the scroll may be reduced. Since the
simulation does not account for this effect, the simulation may be over predicting
the mass flow rates during the early transients. A more significant factor deals
with the amount of refrigerant in the system. As already discussed, the
simulation uses roughly 15% less refrigerant than does the real system. The
energy and time required to distribute the actual refrigerant charge is not
accounted for in the simulation. Hence, it is reasonable to expect that the
simulation would under predict the time and energy associated with achieving
steady state.
12.4.5 The Transient Losses
In this section the losses associated with running a vapor compression system
transiently are analyzed. This is accomplished by evaluating the power and
energy associated with the storage terms of the energy equation. The control
volume used for this purpose encompasses the entire component in question.
All of the system components are examined except the short tube restrictor. The
short tube restrictor is omitted from this analysis because of its inconsequential
mass and volume.
The first component analyzed is the compressor. The control volume
245
-------
encompasses the compressor and compressor motor. The only significant term
in this analysis is the power of the compressor motor that is required to
overcome the inertia of the compressor. The instantaneous power of this term
divided by the steady state compressor power is plotted in Figure 12.21. At the
highest point the power required to overcome the inertia of the compressor is 6.3
times the power required at steady state. The additional power required to
overcome inertia is compounded by the relatively low efficiency of the motor at
such a high torque.
The next component analyzed is the condenser. Figure 12.22 shows the
power associated with the different storage terms associated with the condenser.
To make these terms more clear, consider Equation 12,1. This equation
represents
jr
^condenser = + (™h)airjn ~ i^ref.out ~ air,out 12.1
the energy balance over the entire heat exchanger. The change of interna!
energy in the condenser, (dE/dt)condcnser, can be represented as the sum of four
terms, which are outlined in Equation 12.2.
246
-------
,dEs _ ,dus . cfi/> .c/a, _ ,,dm,
^ .. ^condenser ( \htx + ( ^air + ( j( hef ( ^ )ref 12.2
at dt dt dt dt
The first term of Equation 12.2, m(du/dt)htx, represents the rate at which energy
is stored in the mass of the heat exchanger. This includes the copper tubes and
aluminum fins. In Figure 12.22 the m(du/dt)htx term has the second largest peak,
and is positive. This implies that as time proceeds thermal energy is being
stored in the mass of the heat exchanger. The second term on the right hand
side of Equation 12.2, m(du/dt)air, is the rate at which energy is being stored in
the air. This term is barely perceivable in Figure 12.22 because of the low heat
capacity and low mass of the air. The third and fourth terms in the equation
embody the rate at which energy is being stored in the refrigerant. The third
term represents the rate at which energy must be added or removed from the
condenser to change the state of the refrigerant. In this case, the m(du/dt)ref term
is negative which means that energy is being removed from the refrigerant. This
occurs because the condenser initially contains vapor which has a relatively high
internal energy and as time proceeds the vapor becomes liquid with a relatively
low internal energy. The last term in Equation 12.2, u(dm/dt).ef, is the rate at
which energy is added to the control volume through the accumulation of
refrigerant. The u(dm/dt)ref term has the highest instantaneous power of all those
in Figure 12.22. The reason that the magnitude of this term is so large is the
same reason why the last term was negative. Since the condenser initially
247
-------
contains low density vapor which rapidly becomes liquid there is a significant and
rapid change in the mass of the control volume. Hence, the u(dm/dt)ref term
overwhelms the others in Figure 12,22.
The last component analyzed is the evaporator, which is similar to the
condenser in many respects. As was the case for the condenser, the rate at
which energy is stored in the air, m(du/dt)ajn is insignificant due to the low mass
and low heat capacity of the air in the control volume. As a result, this term can
not be resolved in Figure 12.23, which is a plot of all the transient storage terms
associated with the evaporator, The rate of energy devoted to changing the
state of the refrigerant, m(du/dt)ref, is also similar in magnitude to the same term
for the condenser. However, the m(du/dt)ref term is positive for the evaporator.
This is because the evaporator initially contains low quality refrigerant which has
a low internal energy relative to the refrigerant at steady state. The u(dm/dt)ref
term is negative which is also due to the presence of the low quality and
relatively high density refrigerant at the beginning of on-time. However, unlike
the condenser, the u(dm/dt)ref term is not the largest among the other evaporator
storage terms. The largest instantaneous power among the evaporator storage
terms is that associated with changing the energy of the heat exchanger itself.
This is because the temperature of the evaporator changes significantly over the
course of time.
248
-------
The transient power terms discussed in the preceding paragraphs were
integrated over time to assess the absolute amount of energy associated with
each of the terms. The results of the integration can be seen in Figure 12.24. It
should be kept in mind that all of the transient energy terms take away energy
that would otherwise be used for cooling or heating. Hence, these terms
represent the losses associated with establishing steady state.
The energy required to change the state of the air is not included in
Figure 12.24. This is because the energy required to change the state of the air
in the evaporator and the condenser is three orders of magnitude lower than the
other energies, and as a result is insignificant. Although the power required to
overcome the inertia of the compressor was several times higher than the other
transient power terms, the energy associated with this term is the lowest of those
plotted in Figure 12.24. This is because of the short duration of the power
required to establish the steady state velocity of the compressor, which in turn
is a function of the low inertia of the compressor.
On the other hand, the energy consumed by the evaporator (i.e. the tubes
and fins) is the largest of all the transient energy terms. Hence, efforts at
improving the transient performance of a AC/HP should focus on reducing the
heat capacity of the evaporator. This could be achieved by simply reducing the
249
-------
amount of material used to construct the heat exchangers.
The energy required to establish the steady state charge of the condenser
is practically equal in magnitude to the energy stored in the evaporator. Hence,
efforts at improving the transient efficiency of a vapor compression system
should also focus on reducing the amount of refrigerant that needs to be
distributed at start up. This is consistent with the findings of Chapter 11 where
the TXV improved the cyclic performance by up to 2.7% because it minimized
charge migration during the off-time. Reducing the amount of refrigerant
required by the system is one way of reducing the energy required to establish
the steady state charge distribution. Alternately, a more elaborate approach
would be the use of valves throughout the system that would prohibit the
refrigerant from migrating when the compressor was shut off.
12.5 References
Anosova, G.M., Dashevskil, Y.M., Miropolskili, Z.L., 1990, "The void fraction with
adabatic flow of vapor-liquid mixtures in horizontal tubes", Thermal Engineering.
Vol. 37, No. 7, pp. 373-374.
Budrik, V.V., Eliseev, A.B., Tonchak, I.N., 1990, "Method of Void Fraction
Calculation for Vapor- or Gas-Liquid Flow Conditions", Soviet Journal of Low
Temperature Physics, Vol. 16, No. 4, pp.236-239.
Chexal, B., Lellouche, G., Horowitz, J., Healzer, J., 1992, " A void fraction
correlation for generalized applications", Progress in Nuclear Energy, Vol. 27,
No. 4, pp. 255-295.
250
-------
Farzad, M., O'Neai, D,, 1994, "The effect of void fraction model on estimation of
air conditioner system performance variables under a range of refrigerant
charging conditions", International Journal of Refrigeration, Vol. 17, pp. 85-93.
Hughmark, G.A., 1962, "Holdup in gas-liquid flow", Chemical Engineering
Progress, Vol. 58, pp. 62-65.
Orth, L.A., Zietlow, D.C., Pedersen, C.O., 1995, "Prediction refrigerant inventory
of R-134a in air-cooled condensers", ASHRAE Transactions, Vol. 101.
Rice, C.K., 1987, "The effect of void fraction correlation and heat flux
assumption on refrigerant charge inventory predictions", ASHRAE Transactions,
Vol. 93, pp. 341-367.
Zivi, S.M., 1964, "Estimation of steady state steam void fraction by means of the
principle of minimum entropy production", Journal of Heat Transfer, pp. 247-252.
251
-------
Table 12.1 Cooling Mode Simulation Results
System Parameter
R-22
R-32/134a
Optimum Refrigerant
Charge (kg)
5.86
5.41
COP
4.72
4.83
Capacity (kW)
9.62
9.78
Compressor Power (kW)
2.04
2.02
Mass Flow Rate (kg/s)
5.19e-2
4.60e-2
Angular Velocity of
Compressor (RPM)
3538.6
3538.9
Suction Pressure (kPa)
532.4
509.9
Discharge Pressure (kPa)
1468.7
1476.5
Pressure Ratio
2.76
2.90
Compressor Discharge
Temperature (°C)
86.6
77.9
Superheat (°C)
15.5
11.5
Subcooling (°C)
8.1
12.0
Circulated Concentration
(kg/kg)
1.00
0.308/0.692
252
-------
Table 12.2 Heating Mode Simulation Results
System Parameter
R-22
R-32/134a
Optimum Refrigerant
Charge (kg)
4.82
4.12
COP
4,66
4.75
Capacity (kW)
10.14
10.22
Compressor Power (kW)
2,18
2.15
Mass Flow Rate (kg/s)
4.68e-2
3.96e-2
Angular Velocity of
Compressor (RPM)
3535.6
3536.2
Suction Pressure (kPa)
455,5
432.6
Discharge Pressure (kPa)
1625.8
1653.1
Pressure Ratio
3.57
3 82
Compressor Discharge
Temperature (°C)
86.5
83.4
Superheat (°C)
4.48
3.8
Subcooling (°C)
9.35
15.8
Circulated Concentration
(kg/kg)
1.00
0.312/0.688
253
-------
Table 12.3 Typical Mass and Pressure Drop Distribution (R-22/C test)
System Parameter
Pressure Drop
(kPa)
Mass
(kg)/(%)
Compressor
n/a
6.85e-1/11.7%
Condenser
(Outdoor Heat Exchanger)
21.02
3.54/60.4%
Liquid Line
85.5
9.83e-1/16.8%
Short Tube Restrictor
678.3
7.11e-6/1.2e-4%
Evaporator
(Indoor Heat Exchanger)
125.0
5.700-1/9.7%
Vapor Line
26.5
7.80e-2/1.3%
Table 12.4 Typical Mass and Pressure Drop Distribution (R-22/47S test)
System Parameter
Pressure Drop
(kPa)
Mass
(kg)/(%)
Compressor
n/a
6.13e-1/12.8%
Condenser
(Indoor Heat Exchanger)
9.7
1.66/34.4%
Liquid Line
65.7
9.74e-1/20.2%
Short Tube Restrictor
1000.8
4.39e-6/9.1e-4%
Evaporator
(Outdoor Heat Exchanger)
151.9
1.37/28.5%
Vapor Line
11.0
1.98e-1/4.1%
254
-------
Table 12.5 Comparison of Simulated Results with Experimental Results
(Cooling Mode)
System Parameter
R-22
R-32/134a
Optimum Refrigerant
Charge (kg)
-15.35%
-14.88%
COP
-3.87%
-4.73%
Capacity (kW)
-4.47%
-4.12%
Compressor Power (kW)
-0.54%
0.40%
Mass Flow Rate (kg/s)
-10.36%
-11.54%
Suction Pressure (kPa)
-8.96%
-9.62%
Discharge Pressure (kPa)
-1.89%
0.65%
Pressure Ratio
7.81%
11.54%
Table 12.6 Comparison of Simulated Results with Experimental Results
(Heating Mode)
System Parameter
R-22
R-32/134a
Optimum Refrigerant
Charge (kg)
-15.07%
-24.38%
COP
8.62%
22,42%
Capacity (kW)
9.98%
14.19%
Compressor Power (kW)
1.87%
-6.81 %
Mass Flow Rate (kg/s)
-3.90%
-3.65%
Suction Pressure (kPa)
-8.48%
-5.32%
Discharge Pressure (kPa)
5.03%
-1.95%
Pressure Ratio
14.79%
3.52%
255
-------
Table 12.7 Evaluation of Time Step Independence
Variable
At= 5e-3 s
At = 1 e-2 s
At = 2e-2 s
At = 4e-2 s
COP/COP&t=5e.3s
1.00
1.00
1.00
1.01
Q/Q/\t =5e-3 s
1.00
1.00
1.00
1.01
P/Pftt =5c 3 s
1.00
1.00
1.00
0.99
256
-------
3,5
Experiment Zivi Homogeneous Void Fraction Model Hughmark
Charge (kg)
Figure 12.1 Charge Optimization Curves
11
10
12
2.4
10
O 8
2,2
Capacity Power Power Capacity
(simulation) (simulation) (experiment) (experiment)
1.6
0.6
0.8 1 1.2
Charge/Optimum Charge
1.4
Figure 12.2 Compressor Power and Capacity Versus Charge
257
-------
2000
1500
1000
»
9>
500
—Suction
(simulation)
Discharge
(simulation)
-
Discharge
(experiment)
Suction
, -r-
I. . I , I
: . !
0,6
0.8 1 1,2
Charge/Optimum Charge
1,4
Figure 12.3 Suction and Discharge Pressures Versus Charge
0,065
0.06
0.055
0.05
0,045
Mass Flow Rate Mass Flow Rale
(simulation) (experiment)
0,04
0.6
0.8 1 1.2
Charge/Optimum Charge
1.4
Figure 12.4 Mass Flow Rate Versus Charge
258
-------
Charge/Optimum Charge
Figure 12,5 Superheat and Subcooling Versus Charge
Enthalpy (kJ/kg)
Figure 12.6 R-22 Cooling Mode Pressure Enthalpy Diagram
259
-------
Enthalpy (kJ/kg)
Figure 12.7 R-22 Heating Mode Pressure Enthalpy Diagram
Enthalpy (kJ/kg)
Figure 12.8 R-32/134a Cooling Mode Pressure - Enthalpy Diagram
260
-------
Enthalpy (kJ/kg)
Figure 12.9 R-32/134a Heating Mode Pressure Enthalpy Diagram
Time (seconds)
Figure 12.10 Compressor Speed Versus Time
261
-------
Time (seconds)
Figure 12.11 Mass Flow Rate Entering Each Component Versus Time
-
Compressor Condenser STR Evaporator
"\ '
K
! \
STR (almost zero)
; , . t
0 60 120 180
Time (seconds)
Figure 12.12 Refrigerant Mass in Each Component Versus Time
262
-------
180
Time (seconds)
Figure 12.13 Suction and Discharge Pressure Versus Time
380
60 120
Time {seconds}
180
Figure 12.14 Refrigerant Temperatures Entering Each Component Versus Time
263
-------
7
®
! 4
0 111
Compressor Condenser Evaporator
Power Capacity Capacity
60
120
Time (seconds)
180
Figure 12.15 Normalized Compressor Power and Evaporator Capacity
Versus Time
60 120
Time (seconds)
180
Figure 12.16 Normalized COP versus Time
264
-------
380
60 120
Time (seconds)
Figure 12.17 Refrigerant Temperature Entering Each Component Versus
Time (R-32/134a)
Compressor Condenser STR and Evaporator
60 120 180
Time (seconds)
Figure 12.18 Concentration Versus Time
265
-------
Time (seconds)
Figure 12,19 Suction and Discharge Pressures Versus Time
(system with piping)
Time (seconds)
Figure 12.20 Refrigerant Mass Versus Time (system with piping)
266
-------
to
E
30
Time (seconds)
Figure 12.21 The Power Required to Overcome the Inertia of the
Compressor
a
s
m
£
20 30 40
Time (seconds)
Figure 12.22 The Storage Terms of the Condenser
267
-------
30
Time (seconds)
Figure 12.23 The Storage Terms of the Evaporator
400
200
-200
-400
Figure 12.24 The Energy of the Storage Terms
268
-------
Chapter 13
Conclusions
An experimental and modeling analysis was conducted on a residential heat
pump. This was done to evaluate the transient and steady state performance of
several system configurations. The major conclusions derived from this project
can be divided into two categories, experimental and modeling.
Experimental Conclusions
The performance of R-407C relative to R-22 was evaluated. In light of the
±1.5% random experimental uncertainty, R-407C had a 4.6 - 9.8% lower steady
state COP. The steady state capacities of the two refrigerants were within the
random experimental uncertainty of ±1.2%. The cyclic tests revealed that R-
407C has a 4.5 - 11.0% higher normalized cyclic COP than R-22 and a 1.8 -
5.4% higher normalized cyclic capacity than R-22. These are significant
improvements in performance since the random experimental uncertainty for the
normalized cyclic COP and the normalized cyclic capacity are ±1.8% and ±1.3%,
respectively. The combination of the steady state and cyclic performances
269
-------
showed that R-407C has a 4,3% lower CSPF than R-22 and a 1.5 - 7.0% lower
HSPF than R-22. The random uncertainty in the seasonal performance
calculation is ±1.8%. The seasonal performances were used to show that R-
407C is a greater global warming threat than R-22 in systems that have lifetimes
that exceed 5.5 years.
The performance of a VLHX with a measured mean effectiveness of 33%
was evaluated. The VLHX had no impact on the steady state performance of the
AC/HP operating with either R-22 or R-407C. However, the normalized cyclic
COP, which is measured with a random uncertainty of ±1.8%, improved by 2.4%
and 1.8% for R-22 and R-407C, respectively. These improvements were not
sufficient to significantly effect the CSPF. It was noted that the VLHX caused the
compressor discharge temperature to rise by 3.3 - 3.5°C for both fluids, and as
result may adversely influence the compressor reliability.
The AC/HP performance was also quantified with STRs and TXVs
Relative to the TXV, the STR had a up to a 5.2±1.5% lower steady state COP
and up to a 6.1 ±1.2% lower steady state capacity. The cyclic performance of the
AC/HP was also adversely effected by the STR. The STR was responsible for
a 1.6 - 2.6% lower normalized cyclic COP, which is measured with a random
experimental uncertainty of ±1.8%. The STR also caused a 5.4 - 5.7% reduction
270
-------
in the normalized cyclic capacity, which is measured within ±1.3%, The
cumulative effect of the lower performance of the STR is a 3.6±1.8% lower CSPF
and up to a 3.9±1.8% lower HSPF as compared to the TXV. Furthermore, the
STR significantly increased the sensitivity of the system performance to the
refrigerant charge. Hence, the TXV is a significantly superior expansion device
with respect to all facets of AC/HP performance.
The final facet of the experimental work was to measure the concentration
of R-32/134a and R-407C as a function of time. The results indicate that the
concentration of the refrigerant mixtures changed with time. However, the
circulated concentration was reached rapidly, within 3 minutes. The results of
these measurements also indicate that the circulated concentration of the
refrigerant mixtures was not equal to the charged concentration. The circulated
concentration of both refrigerant mixtures shifted away from the less volatile
component. R-407C shifted away from the less volatile component by 3.0 wt.%
while R-32/134a shifted by 0.5 wt.%. These measurements are relevant in the
context of the ± 0.27 wt.% experimental uncertainty. The concentration shift was
attributed to the velocity difference between the phases in the heat exchangers.
Modeling Conclusions
A fully implicit, distributed parameter simulation capable of modeling the
271
-------
transient and steady state aspects of AC/HP was developed for this aspect of the
project. For steady state, the simulation solves the complete continuity, species,
energy and momentum equations, while transiently only the momentum equation
is omitted. This simulation is the first capable of representing the significant
transient and steady state physics of an AC/HP operating with pure and mixed
refrigerants while utilizing minimal empirical data.
First the heat exchanger model, a sub-component of the overall
simulation, was used to demonstrate the theoretical performance of different heat
exchanger geometries. The heat exchanger model indicated that R-407C is
more sensitive to counter and parallel flow geometries than R-22. Relative to the
cross flow capacity, R-407C gained 0.4 - 3.4% for the counter flow geometry and
lost 1.8- 8.4% for the parallel flow geometry. This indicates that the temperature
glide of R-407C is not sufficient for large performance gains but is sufficient to
cause moderate performance degradation. This also indicates that the heat
exchanger geometry must be considered when using R-407C as a drop-in
replacement for R-22.
The AC/HP simulation was used to quantify the shift in the circulated
concentration of R-32/134a, At steady state the simulation predicted that the
circulated concentration shifted away from the less volatile component by 0.8
272
-------
wt.% when the connecting piping was considered. The extent the circulated
concentration could be shifted was evaluated by examining two cases. The
concentration shift increased to 1.4 wt.% when the connecting piping was
removed. To evaluate the maximum theoretical concentration shift, the
simulation was run without connecting piping and with half the optimum charge.
The results for this scenario indicate that the circulated concentration is shifted
by 2.3 wt.% towards R-32. Simulation was also used to model the transient
concentration shift of R-32/134a. The simulation showed that during the early
transients the circulated composition of R-32/134a varies with time and location,
but the concentration approaches its steady state value within 30 seconds.
Hence, for refrigerants with a moderate temperature glide the shift in the
circulated concentration due to the velocity difference between the phases is not
significant. It should also be noted that at no point in time or space does the
concentration of R-32/134a approach flammable levels.
The simulation was also used in the steady state limit to quantify the
penalty associated with connecting piping. The simulation showed that the
connecting piping reduces the COP and capacity by 4.1%. The effect of
connecting piping was also studied transiently. The simulation results indicate
that for the case studied here the addition of connecting piping reduced the
normalized COP by 12.2%.
273
-------
Finally, the simulation was used to quantify the major losses associated
with running an AC/HP transiently. Reducing the temperature of the evaporator
itself (i.e. the tubes and fins) and redistributing the refrigerant required the most
energy to achieve steady state. The energy required to overcome the inertia of
the compressor was shown to be relatively small.
274
-------
Chapter 14
Future Work
Through the course of this work several research needs were identified. One
such need is the development of physically grounded short tube restrictor model.
Another area of potential research is the development of an accurate void
fraction correlation applicable for refrigerants. It would also be worth while to
experimentally study the key transient loss mechanisms identified, Since R-
407C poses a greater global warming hazard than R-22, it is suggested that
other refrigerants, such as R-410a and R-410b, be investigated. Lastly, the
study of a more effective VLHX may yield interesting results.
275
-------
Appendix A1
Experimental Uncertainty Analysis
The experimental performance of the AC/HP is characterized by parameters
such as the capacity, the steady state and cyclic coefficients of performance
(COP), the part load factors (CLF/HLF), and the cyclic degradation coefficient
(CD). Each of these parameters has a finite uncertainty which is discussed in this
section and in detail by Hwang (Hwang et a!., 1995).
The total experimental uncertainty is composed of the bias uncertainty
(systematic uncertainty) and the random uncertainty (precision uncertainty). The
bias uncertainty is determined using the Pythagorean summation of the discrete
uncertainties as shown in Equation A1.1 (Kline and McClintock. 1953).
uf
m(u" 12 + (u« S2 + -+ lu-if
Where: f Quantity in question (COP, capacity, etc.)
X; Variables used to calculate f
ux, Bias uncertainty in Xj
uf Bias uncertainty in f
276
-------
uf is the overall bias of function f resulting from the individual uncertainties
of Xj... x,,. The individual uncertainties, u xl, are quantified by Hwang (Hwang et
al., 1995).
For example, consider the air side capacity. The variables which are used
to calculate the air side capacity are the inlet and outlet dry-bulb temperatures
(Tjni Tout), the inlet and outlet humidity ratios (coiw wout), the inlet pressure (Pin) and
the mass flow rate of the air (mair). Hence, Equation A1.1 becomes Equation
A1.2 for the air side capacity.
c?0ry dQA!r ~ cOp.y «
uc = [(uT —^ 2 + (uT ^ 2 + (u, —- 2
Qair T,n 3Tin r-' dT0Ut dm!n
(u»„ + ("p ^>2 + <"« A1-2
" in dm
Most of the uncertainties in Equation A1.2 are known directly, except for the
uncertainty in mass flow rate. Therefore, the mass flow rate uncertainty must
be calculated with Equation A1.1. Since the mass flow rate is a function of the
pressure difference across the nozzle, the inlet pressure, and the inlet
277
-------
temperature, all of these variables must be accounted for in the mass flow rate
uncertainty calculation. The remaining unknowns in Equation A1.2 are the
partial derivatives. Some of the partial derivatives in Equation A1.2 can be
calculated in closed form. However, for consistency all of the partial derivatives
used in the uncertainty calculations are approximated using a first order finite
difference. Once the partial derivatives are evaluated it is possible to calculate
the overall bias in the air side capacity.
All of the bias uncertainties are shown in Table A1.1. Table A1.1 shows
a relatively large bias uncertainty for CD, which is due to the form of the equation
which defines CD.
The random uncertainty is analyzed using a statistical methodology. The
Gaussian probability distribution is assumed for the data. The standard
deviation, o, is calculated using Equation A1.3 (Kline and McClintock, 1953).
A1.3
Where: o Standard deviation
n Number of data
x, The magnitude of the measured quantity
xm The arithmetic mean value
278
-------
The three-sigma confidence interval (± 3a), which has 99.7% certainty if the data
has a normal distribution, is used to define the random uncertainty. Twelve sets
of data are used for calculating the random uncertainties of the steady state and
cyclic parameters. Each set of data contains at least six discrete data points for
the variable in question. The random uncertainties are shown in Table A1.2.
After evaluating the bias and random uncertainties, the total uncertainties are
calculated. The total uncertainties are presented in Table A1.3. It should be
noted that the steady state uncertainties are calculated for the B test conditions
while the cyclic uncertainties are calculated for the D test conditions.
A1.1 References
Hwang, Y., Judge, J., Radermacher, R. 1995, "Testing of Refrigerant Mixtures
in Residential Heat Pumps", EPRI Report.
Kline, S.J., McClintock, F.A., 1953, "Describing Uncertainties in Single Sample
Experiments", Mechanical Enaineerina, d. 3.
i f - j |
279
-------
Table A1.1 Estimated Bias Uncertainty of Characteristic Parameters
Parameters
Capacity
COP
CLF&
HLF
COPcyc
/COPss
CD
CSPF&
HSPF
Mode
Steady
State
Steady
State
Cyclic
Cyclic
Cyclic
n/a
Bias
Uncertainty
±1.7%
± 1.7%
± 2.4%
± 2.4%
± 13.0%
±2.1%
Uncertainty of Key Variables:
[T] = + 0.17°c, [TdJ = ± 0.20°C, [PJ = ±54 Pa, [AP] = ± 2.5 Pa, [m] = ± 0.02 kg
Table A1.2 Estimated Random Uncertainty of Characteristic Parameters
Parameters
Capacity
COP
CLF &
HLF
COPcyc
/COPss
CD
CSPF&
HSPF
Mode
Steady
State
Steady
State
Cyclic
Cyclic
Cyclic
n/a
Random
Uncertainty
+ 1.2%
± 1.5%
± 1.3%
+ 1.8%
±11.4%
± 1.8%
Table A1.3 Estimated Total Uncertainty of Characteristic Parameters
Parameters
Capacity
COP
CLF &
HLF
COPcyc
/COPss
CSPF&
HSPF
Mode
Steady
State
Steady
State
Cyclic
Cyclic
Cyclic
n/a
Total
Uncertainty
± 2.9%
± 3.2%
± 3.7%
± 4.2%
± 24.4%
± 3.9%
280
-------
Appendix A2
Heat Exchanger Model Uncertainty Analysis
It is the goal of this section to estimate the uncertainty of the heat exchanger
model as it pertains to the predicted capacity of the heat exchanger. This is
done to explain the relatively high level of accuracy (+10%) reported in Section
9.4.
The uncertainty is estimated by utilizing the Pythagorean summation of
the discrete uncertainties as shown in Equation A1.1 (Kline and McClintock,
1953). One approach would be to utilize Equation A1.1 at every one of the 150
nodes of the heat exchanger and account for the uncertainty of each quantity
which contributes to the calculation of the capacity. This is impractical for
several reasons. One reason is that it is not feasible to evaluate the all of the
necessary partial derivatives for a!! of the nodes and for all of the quantities
which influence the capacity. This is due to the large number of variables which
impact the overall capacity of the heat exchanger. The second reason is that not
all of the discrete uncertainties associated with the quantities used to calculate
281
-------
the capacity are known (e.g., fin efficiency, void fraction correlation).
Furthermore, it is difficult to accurately quantify the uncertainties of the various
assumptions implemented in modeling the heat exchanger (e.g., cross flow,
uniform air and refrigerant distribution). Therefore, a simplified approach is used
to quantify the major contributors to the uncertainty in the predicted capacity.
Examination of the heat transfer resistances associated with an air to
refrigerant heat exchanger shows that the air heat transfer resistance (Rair =
2.2-10"2 m2KAAI) is an order of magnitude large than the refrigerant heat transfer
resistance (Rref = 3.8-10"3 m2K/W), which itself is several orders of magnitude
larger than the other heat transfer resistances in the heat exchanger (Rcontoc, =
1.10-5
m2K/W & Rtabe = 8-10"7 m2K/W). Hence, in estimating the uncertainty of the
heat exchanger model only the air and refrigerant heat transfer resistances are
considered. With this simplification, Equation A1.1 becomes Equation A2.1.
UQ =
(uh
dQ s.2
dh
f + (u„
ref
dQ)
dh ¦
c 'air
A2.1
The uncertainties in Equation A2.1 are found in the literature for the heat transfer
correlations used. However, the task remains to evaluate the partial derivatives
in Equation A2.1. The partial derivatives are approximated with a first order
282
-------
accurate finite difference. The finite difference is calculated by varying the heat
transfer coefficient in question by 1.0 % and evaluating the resultant change in
heat transferred.
Results and Discussion
For evaporation heat transfer, which had the lowest error in heat transferred in
Section 9.4, the estimated uncertainty is 1.9% as calculated by Equation A2.1.
This result is based on the evaporation of R-22 which resulted in 7.2 kW of
cooling capacity, and is typical of the other configurations. The reasons for this
surprisingly low uncertainty are now addressed.
The uncertainties of the air and refrigerant heat transfer correlations are
relatively low, which contributes to the low overall uncertainty of the heat
transferred. The uncertainty of the refrigerant heat transfer coefficient is +7.2%
for R-22, which is low relative to other, more general evaporation heat transfer
correlations (Jung, 1989). The uncertainty used for the air side heat transfer
coefficient is ±5.0% (Webb, 1990). This value is utilized since the air heat
transfer coefficient correlation implemented in the simulation predicted 98% of
data within ±5.0% (Webb, 1990), This correlation achieves a high level of
accuracy by focusing on a relatively narrow region of interest (5 < Re Pr Dh/L
<180). It should be noted that the correlation was developed for the same
283
-------
geometry arid flow rates used in the experiment.
Another point which contributes to the low overall uncertainty is the
insensitivity of the heat transferred to changes in the overall heat transfer
coefficient. This is demonstrated in Figure A2.1, which was generated by the
heat exchanger simulation program. This figure shows the normalized capacity
of an evaporator as it varies with the normalized overall heat transfer coefficient
(U) multiplied by the normalized heat exchange area (A). As UA is increased,
the sensitivity of the capacity to UA decreases. Specifically, increasing the UA
by 25% from the original UA increases the capacity by 4.0%, while decreasing
the UA by 25% from the original UA reduces the capacity by 9.6%. This is
expected because as the UA increases the refrigerant temperature approaches
the air inlet temperature asymptotically, thereby increasing the refrigerant
superheat. However, increasing the amount of superheat adds little to the
overall capacity since the fraction of heat rejected in the superheated vapor
region is relatively small compared to the total. For example, consider the
evaporation of R-22. If the heat absorbed in the superheated region was
neglected altogether then the error in capacity would only be 5,4%. Conversely,
if the heat exchanger was assumed to be infinitely long, and as a result the
refrigerant outlet temperature equaled the air inlet temperature, then the error
in capacity would only be 4.2%. Therefore, the low sensitivity of the heat
284
-------
transferred to changes in UA is a consequence of the relatively large area of the
heat exchangers studied, which ensures that the outlet state is superheated.
The relatively large area of the heat exchanger is demonstrated experimentally
as shown by Figure A2.2. Figure A2.2 shows the experimental temperature
profile of R-22 during evaporation. It is clear in this figure that 50% to 75% of the
heat exchanger is used to reject heat in the two phase region. This leaves a
large fraction of the heat exchanger to superheat the refrigerant, which adds little
to the overall capacity.
The way the air and refrigerant heat transfer coefficients impact the
overall heat transfer coefficient also contributes to the low uncertainty in the
predicted capacity of the heat exchanger. Since the uncertainties in air heat
transfer coefficients are usually about 5%, while the uncertainties in refrigerant
heat transfer coefficients are usually about 10-25%, it is beneficial that the air
side heat transfer coefficient contributes more significantly to the overall heat
transfer coefficient. Specifically, a 25% change in the air heat transfer coefficient
results in a 22.4% change in the overall heat transfer coefficient while a 25%
change in the refrigerant heat transfer coefficient translates into only a 4.8%
change in the overall heat transfer coefficient. Therefore, relatively large errors
can be tolerated in the refrigerant side heat transfer coefficient.
285
-------
The low error (±5% for evaporators and ±10% for condensers) obtained
in Section 9.4 for the heat exchanger model is a product of the previously
mentioned points as well as some others. For example, any systematic bias that
may have existed in the simulation was minimized or eliminated by trying many
different refrigerant heat transfer correlations and using the one which best
reproduced the experimental results. Furthermore, the small sample size used
in Section 9.4 may facilitate this type of bias reduction.
A2.1 References
Jung, D.S., 1989, "Horizontal-flow boiling heat transfer using refrigerant
mixtures", EPRI ER-6364, Project 8006-2.
Kline, S.J., McClintock, FA, 1953, "Describing Uncertainties in Single Sample
Experiments", Mechanical Engineering, p. 3.
Webb, R.L., 1990, "Air-side Heat Transfer Correlations for Flat and Wavy Plate
Fin and Tube Geometries", ASHRAE Transactions, Vol. 1, pp. 445-449.
286
-------
1,25
.0.75
0.25
0.25
0.5 0.75 1
Normalized UA
1,25
Figure A2.1 The Effect of UA on Capacity (R-22, evaporation)
.25 ,5 ,75
Normalized Heat Exchanger Length
Figure A.2.2 Experimental Refrigerant Temperature Profile
(R-22, evaporation)
287
-------
TECHNICAL REPORT DATA
(Please read Insimctions on the reverse before compl || | |||| || |||||| 1111||| 11| | || |j|
1, REPORT NO, 2.
EPA-600/R-96-129
PB97-117741¦
4. TITLE AND SUBTITLE
A Transient and Steady State Study of Pure and Mixed
Refrigerants in a Residential Heat Pump
5. REPORT DATE
November 1996
6. PERFORMING ORGANIZATION CODE
7. AUTHORfSS
John Judge and Reinhard Radermacher
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
The University of Maryland
The Center for Environmental Energy Engineering
College Park, Maryland 20742-3035
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
CR822356
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Air Pollution Prevention and Control Division
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final; 10/94 - 10/95
14. SPONSORING AGENCY CODE
EPA/600/13
16. supplementary notes /\PPCD project officer is Robert V. Ilendriks, Mail Drop 63, 919/
541-3928.
i6. abstract rj-^g rep0rt gives results of an experimental and theoretical investigation
of the transient and steady state performance of a residential air-conditioning/heat
pump (AC/HP) operating with different refrigerants. (NOTE: The project was moti-
vated by environmental concerns related to the replacement of stratospheric ozone
depleting refrigerants as required by international agreement and U. S. law, Hydro-
chlorofluorocarbon (HCFC)_22, a medium pressure refrigerant, is scheduled to be
phased out of production and must be replaced. Significant empirical data are avail-
able on lICFC-22, but relatively little data exist on the transient performance of any
of the zeotropic mixtures being considered as HCFC-22 replacements.) The experi-
mental work,conducted by testing an AC/HP- in environmental chambers, documented
refrigerant performance for steady state, cyclic, and seasonal performance, evalu-
ated various equipment modifications, and measured changes in the concentrations of
refrigerant mixtures as a function of time. A computer model was developed, cap-
able of modeling the transient and steady state performance of an AC/HP. This mo-
del is the first capable of representing the significant transient and steady state phy-
sics of an AC/HP operating with pure and mixed refrigerants while utilizing minimal
empirical data.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ended terms
c. COSATI Field/Group
Pollution
Refrigerants
Heat Pumps
Residential Buildings
Ozone
Greenhouse Effect
Mathematical Models
Pollution Prevention
Stationary Sources
Stratospheric Ozone
Zeotropes
13 B
13 A
13 M
07B
04A
12 A
18. distribution statement
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21, NO. OF PAGES
314
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
------- |