Development of Efficient Air Conditioning and

Refrigeration System for Service Vehicles

by

Farshid Bagheri

M.Sc., Sharif University of Technology, 2005 B.Sc., Iran University of Science and Technology, 2003

Thesis Submitted in Partial Fulfillment of the

Requirements for the Degree of

Doctor of Philosophy

in the

School of Mechantronic Systems Engineering

Faculty of Applied Science

Farshid Bagheri 2016

SIMON FRASER UNIVERSITY

Spring 2016

ii

Approval

Name: Farshid Bagheri

Degree: Doctor of Philosophy

Title: Development of Efficient Air conditioning and Refrigeration System for Servicer Vehicles

Examining Committee: Chair: Kevin Oldknow Lecturer, School of Mechatronic Systems Engineering

Majid Bahrami Senior Supervisor Professor

John Jones Supervisor Associate Professor

Gary Wang Supervisor Professor

Flavio Firmani Internal Examiner Lecturer School of Mechatronic Systems Engineering

Bidyut Baran Saha External Examiner Professor Department of Energy and Environmental Engineering Kyushu University

Date Defended/Approved: March 31, 2016

iii

Abstract

This research aims to develop a proof-of-concept demonstration of a high efficiency

vehicle air conditioning and refrigeration (VACR) system to be employed in service

vehicles. The work herein is part of a collaborative research project with two local

companies: Cool-It Group, and Saputo Inc., with a main focus on service vehicles. Due

to global energy consumption and the environmental impacts of air conditioning and

refrigeration (A/C-R) systems, the development of a high efficiency system can

significantly contribute to green and sustainable development and environmental

protection.

This research fills a gap in the literature by developing real-time thermal and

performance characteristics of the VACR systems employed in the food transportation

industry. Field data is acquired from pilot refrigerated service vehicles during different

seasons of the year and the duty cycles are established. The acquisition of field data

begins in stationary A/C-R systems and continues in mobile VACR systems. Moreover, a

testbed is built in the Laboratory for Alternative Energy Conversion (LAEC) for more

comprehensive experiments. Mathematical models are developed for thermal and

performance simulations of VACR systems under steady state and transient operating

conditions. The models are validated using the laboratory and field data and employed

for a thermal and performance investigation of VACR systems.

A proof-of-concept demonstration of high efficiency VACR systems is built in LAEC

using variable speed compressor and fans and high efficiency heat exchangers. The

modeling results are validated and used to develop an optimization model. The

optimization model is validated and utilized to determine the optimum compressor and

fans speeds for achievement of the highest coefficient of performance (COP) under real-

time operating conditions. The optimization model is integrated with an existing cooling

demand simulator to develop a proof-of-concept demonstration of a proactive and model

predictive controller (MPC) for the VACR system. The controller is implemented on the

laboratory-built VACR system and a proof-of-concept demonstration of high efficiency

VACR is finalized. The developed concept and platform is expandable to the entire

transportation industry as well as stationary A/C-R systems.

iv

Keywords: Vehicle Air conditioning and Refrigeration, Performance Optimization, Mathematical Modeling, Experimentation, Proactive, Model Predictive Controller

v

Dedication

To my beloved wife, mother, and father

vi

Acknowledgements

I owe my gratitude to a great many people who have made this dissertation possible.

First and foremost, I would like to express my sincere gratitude to my senior supervisor,

Dr. Majid Bahrami, whose expertise, guidance, encouragement, and supports added

considerably to my work. I appreciate his vast knowledge and skill in many areas, and

his assistance in conducting research and writing reports.

I would like to thank my supervisory committee members, Dr. Gary Wang, and Dr. John

Jones for the insightful comments and encouragement they provided at all levels of the

research project. I am grateful to Dr. Flavio Firmani and Dr. Bidyut Baran Saha for their

time reading this thesis and helping to refine it.

I am also indebted to the members of the laboratory for alternative energy conversion

(LAEC) with whom I have interacted during the course of my graduate studies.

Particularly, I would like to acknowledge Mr. Marius Haiducu, Dr. Wendell Huttema,

Dr. Mehran Ahmadi, Dr. Mohammad Fakoor-Pakdaman, Dr. Mohammad Ali

Fayazbakhsh, and undergraduate co-op students Mr. Jit Tian Lim, Mr. Jordan Lam, and

Mr.Tian Lin Yang for their assistance over the past three years.

I recognize that this research would not have been possible without the financial

assistance of natural sciences and engineering research council of Canada (NSERC)

and Simon Fraser University, and supports from Cool-It Group (Mr. Saeed Zaeri) and

Saputo Inc. (Mr. Norman Desilets), and express my gratitude to those institutes.

Most importantly, none of this would have been possible without the love and patience of

my wife and parents.

vii

Table of Contents

Approval .............................................................................................................................ii Abstract ............................................................................................................................. iii Dedication ......................................................................................................................... v Acknowledgements ...........................................................................................................vi Table of Contents ............................................................................................................. vii List of Tables .....................................................................................................................ix List of Figures.................................................................................................................... x List of Acronyms .............................................................................................................. xvi Nomenclature ................................................................................................................. xvii Executive Summary ........................................................................................................ xxi 

Chapter 1.  Introduction ............................................................................................... 1 1.1.  Research Background and Importance .................................................................... 1 1.2.  Research Motivation................................................................................................. 3 1.3.  Research Scope ....................................................................................................... 4 1.4.  Thesis Structure ....................................................................................................... 5 1.5.  Relevant Literature ................................................................................................... 7 

1.5.1.  Thermal and performance studies on A/C-R systems ................................ 7 1.5.2.  Control studies on A/C-R systems ............................................................ 13 1.5.3.  Lack of literature ....................................................................................... 17 

Chapter 2.  Mathematical Model Development ........................................................ 18 2.1.  Steady State Model Development .......................................................................... 19 

2.1.1.  Compressor sub-model ............................................................................. 20 2.1.2.  Condenser and evaporator sub-model ..................................................... 20 2.1.3.  Thermostatic expansion valve sub-model ................................................. 23 2.1.4.  Thermodynamic correlations for the refrigerant ........................................ 23 2.1.5.  Numerical solver ....................................................................................... 23 

2.2.  Quasi-Steady State Model Development ............................................................... 25 2.3.  Transient Model Development ............................................................................... 29 

2.3.1.  Condenser sub-model ............................................................................... 29 2.3.2.  Evaporator sub-model ............................................................................... 33 2.3.3.  Compressor sub-model ............................................................................. 34 2.3.4.  Expansion valve ........................................................................................ 34 2.3.5.  Unknowns of the model ............................................................................ 34 

2.4.  Conclusion ............................................................................................................. 35 

Chapter 3.  Experimental Study and Data Acquisition ............................................ 37 3.1.  Testbed Design and Construction .......................................................................... 37 3.2.  Development of a Proof-of-Concept Demonstration of a High Efficiency

VACR System ........................................................................................................ 42 3.3.  Field Data Collection, Stationary A/C-R System .................................................... 43 

viii

3.3.1.  Data collection and duty cycle definition ................................................... 47 3.3.2.  Effect of on-off cycling on power consumption.......................................... 51 

3.4.  Field Data Collection, VACR System ..................................................................... 52 3.4.1.  Measuring equipment installation and duty cycle definition ...................... 55 3.4.2.  Real-time data acquisition ......................................................................... 60 

3.5.  Conclusion ............................................................................................................. 69 

Chapter 4.  Model Validation, Thermal and Performance Investigation ................ 71 4.1.  Steady State Model Validation, Thermal and Performance Investigation .............. 71 

4.1.1.  Results relevant to the developed testbed in LAEC .................................. 71 4.1.1.1. Effects of condenser air flow rate (condenser fan speed) on COP .......... 78 4.1.1.2. Effects of evaporator air flow rate (evaporator fan speed) on COP.......... 82 

4.1.2.  Results relevant to the studied stationary A/C-R system .......................... 89 4.2.  Quasi Steady State Model Validation ..................................................................... 92 4.3.  Transient Model Validation, Real-time Performance Investigation ......................... 97 4.4.  Conclusion ........................................................................................................... 103 

Chapter 5.  Metamodel Development for Optimization ......................................... 105 5.1.  Metamodel Development Based on Steady State Thermal Model ....................... 105 

5.1.1.  Sample space generation ....................................................................... 110 5.1.2.  Metamodel development ......................................................................... 112 5.1.3.  Metamodel validation .............................................................................. 113 

5.2.  Metamodel Development Based on Transient Thermal Model ............................ 116 5.2.1.  Sample space generation ....................................................................... 118 5.2.2.  Metamodel development ......................................................................... 118 5.2.3.  Metamodel validation .............................................................................. 120 

5.3.  Conclusion ........................................................................................................... 121 

Chapter 6.  Proof-of-concept Demonstration of High Efficiency VACR System for Service Vehicles ................................................................ 122 

6.1.  Development of Optimization Solver and Integration with a Cooling Demand Simulator, Proactive Control ................................................................................. 122 

6.2.  Proactive, Optimization-Based, Model Predictive Control under Steady State Condition ..................................................................................................... 127 

6.3.  Proactive, Optimization-Based, Model Predictive Control under Transient Condition .............................................................................................................. 138 

6.4.  Conclusion ........................................................................................................... 145 

Chapter 7.  Summary and Future Works ................................................................ 147 7.1.  Summary of Thesis .............................................................................................. 147 7.2.  Future work .......................................................................................................... 149 

References.................... ............................................................................................... 151  

ix

List of Tables

Table 1-1: Summary of literature on thermal and performance studies of A/C-R systems ................................................................................................... 11 

Table 1-2: Summary of the literature on control of A/C-R systems ................................. 16 

Table 2-1: Mathematical model correlations ................................................................... 22 

Table 2-2: Output parameters of the mathematical model .............................................. 24 

Table 2-3: Calculated parameters at each time step ...................................................... 35 

Table 3-1: Duty cycle definition for the stationary refrigeration system ........................... 51 

Table 5-1: A few representatives of the generated sample space for metamodel development-steady state ..................................................................... 111 

Table 5-2: Obtained coefficients of metamodel for steady-state condition ................... 113 

Table 5-3: Comparison of obtained accuracies for different NSP ................................. 116 

Table 5-4: Obtained coefficients of metamodel for transient condition ......................... 119 

Table 5-5: Comparison of obtained accuracies for different NSP ................................. 120 

x

List of Figures

Figure 1-1: Schematic of a basic VCR cycle [14] .............................................................. 2

Figure 1-2: Thesis structure .............................................................................................. 6

Figure 2-1: Schematic of a basic A/C-R unit and the main parameters .......................... 19

Figure 2-2: An integral overview of a heat exchanger (evaporator or condenser) in A/C-R systems ..................................................................................... 28

Figure 2-3: Schematic representative of a cross flow condenser in VCR systems ......... 30

Figure 3-1: Typical battery-powered anti-idling VACR system employed in the testbed ..................................................................................................... 38

Figure 3-2: Experimental setup schematic ...................................................................... 39

Figure 3-3: First version of the developed experimental setup ....................................... 40

Figure 3-4: Current experimental setup configuration ..................................................... 41

Figure 3-5: Purchased components for building the proof-of-concept demonstration of a high efficiency VACR system .................................... 43

Figure 3-6: The selected freezer room (left), its condensing unit (mid), dimensions (right) .................................................................................... 45

Figure 3-7: Installation of thermocouples, pressure transducers, and flow meters on the condensing unit ............................................................................ 46

Figure 3-8: Installation of thermocouples on the evaporating unit .................................. 47

Figure 3-9: Sample measured temperature during weekdays - evaporating unit ........... 48

Figure 3-10: Sample measured temperature during weekdays - condensing unit .......... 48

Figure 3-11: Sample measured input electric power to the condensing unit .................. 49

Figure 3-12: A typical power draw of the compressor after start-up ............................... 52

Figure 3-13: Selected typical refrigerated trailers for field data acquisition ..................... 54

Figure 3-14: Sample installed measuring equipment on the condensing unit ................. 55

Figure 3-15: Sample installed measuring equipment on the evaporating unit ................ 56

xi

Figure 3-16: Duty cycle of the VACR unit trailer #1 – Monday to Wednesday ................ 58

Figure 3-17: Duty cycle of the VACR unit trailer #1 – Thursday to Saturday .................. 58

Figure 3-18: Duty cycle of the VACR unit trailer #1 – Saturday to Monday .................... 59

Figure 3-19: Duty cycle of the VACR unit trailer #2 – Weekdays and Weekends ........... 59

Figure 3-20: Sample ambient air (condenser inlet) temperature variations – Jan 2014 ......................................................................................................... 61

Figure 3-21: Sample air temperature variations at evaporator inlet and outlet – trailer # 1, Jan 2014 ................................................................................. 61

Figure 3-22: Sample refrigerant temperature variations at compressor inlet and outlet – trailer # 1, Jan 2014 .................................................................... 62

Figure 3-23: Sample refrigerant temperature variations at evaporating unit – trailer # 1, Jan 2014 ................................................................................. 62

Figure 3-24: Sample ambient air (condenser inlet) temperature variations – Jul 2014 ......................................................................................................... 64

Figure 3-25: Sample ambient air (condenser inlet) temperature variations – Aug 2015 ......................................................................................................... 65

Figure 3-26: Sample refrigerant temperature variations at compressor inlet and outlet – trailer # 1, Mon to Thu, Jul 2014 ................................................. 65

Figure 3-27: Sample refrigerant temperature variations at compressor inlet and outlet – trailer # 1, Thu to Sun, Jul 2014 .................................................. 66

Figure 3-28: Sample air temperature variations at evaporator inlet and outlet – trailer # 1, Mon to Thu, Jul 2014 .............................................................. 66

Figure 3-29: Sample air temperature variations at evaporator inlet and outlet – trailer # 1, Thu to Sun, Jul 2014 .............................................................. 67

Figure 3-30: Sample zoomed air temperature variations at beginning of a cycle “ON”– trailer # 1, Jul 2014 ....................................................................... 67 

Figure 3-31: Sample zoomed air temperature variations at end of a cycle “ON”– trailer # 1, Jul 2014 .................................................................................. 68 

Figure 3-32: Sample refrigerant temperature variations at compressor inlet and outlet – trailer # 2, Aug 2015 ................................................................... 68 

Figure 3-33: Sample air temperature variations at evaporator inlet and outlet – trailer # 2, Aug 2015 ................................................................................ 69 

xii

Figure 4-1: Sample steady state condition achievement - measured refrigerant side temperature ...................................................................................... 72 

Figure 4-2: Validation of the model (lines) with experimental data (symbols), evaporator air outlet temperature , ,( )a evap outT vs. condenser air inlet

temperature , ,( )a cond inT .............................................................................. 73 

Figure 4-3: Validation of the model (lines) with experimental data (symbols), Evaporating temperature ( )eT vs. condenser air inlet temperature

, ,( )a cond inT ................................................................................................. 74 

Figure 4-4: Validation of the model (lines) with experimental data (symbols), refrigerant mass flow rate ( )refm vs. condenser air inlet

temperature , ,( )a cond inT .............................................................................. 76 

Figure 4-5: Validation of the model (lines) with experimental data (symbols), total power consumption vs. condenser air inlet temperature , ,( )a cond inT .......... 76 

Figure 4-6: Validation of the model (lines) with experimental data (symbols), cooling power vs. condenser air inlet temperature , ,( )a cond inT ................... 77 

Figure 4-7: Validation of the model (lines) with experimental data (symbols), COP vs. condenser air inlet temperature , ,( )a cond inT .......................................... 77 

Figure 4-8: Refrigerant flow rate vs. condenser air flow rate ,( )a condm ............................. 80 

Figure 4-9: Total power consumption vs. condenser air flow rate ,( )a condm ..................... 81 

Figure 4-10: Cooling power vs. condenser air flow rate ,( )a condm .................................... 81 

Figure 4-11: COP vs. condenser air flow rate ,( )a condm ................................................... 82 

Figure 4-12: Total power consumption vs. evaporator air flow rate ,( )a evapm ................... 83 

Figure 4-13: Cooling power vs. evaporator air flow rate ,( )a evapm .................................... 84 

Figure 4-14: COP vs. evaporator air flow rate ,( )a evapm ................................................... 84 

Figure 4-15: Total power consumption vs. condenser and evaporator air flow rates , ,( , )a cond a evapm m ................................................................................ 86 

xiii

Figure 4-16: Cooling power vs. condenser and evaporator air flow rates

, ,( , )a cond a evapm m ........................................................................................ 86 

Figure 4-17: COP vs. condenser and evaporator air flow rates , ,( , )a cond a evapm m ............. 87 

Figure 4-18: Comparison of thermodynamic pressure-enthalpy cycle between nominal (red continuous lines) and optimal (green dashed lines) condenser and evaporator air flow rates (ANSI/AHRI 210/240-2008 air-side condition) ........................................................................... 88 

Figure 4-19: Validation of the model (simulation) results with the measured values, walk-in freezer room .................................................................... 89 

Figure 4-20: Present and design COP of the VCR system for different duty cycles ....... 90 

Figure 4-21: Comparison between first quasi-steady model and experimental data during on-off cycling ........................................................................ 93 

Figure 4-22: Comparison between modified quasi-steady model and experimental data during on-off cycling ................................................... 95 

Figure 4-23: Comparison of cooling power variation between modified quasi-steady model and experimental data during combined imposed conditions ................................................................................................ 96 

Figure 4-24: Comparison of power consumption variation between modified quasi-steady model and experimental data during combined imposed conditions .................................................................................. 97 

Figure 4-25: Sample simulated and measured air temperature at evaporator outlet - Trailer #2, Wed to Thu, August 2015 ........................................... 98 

Figure 4-26: Sample simulated and measured cooling power - Trailer #2, Thu to Sat, August 2015 ..................................................................................... 99 

Figure 4-27: Sample simulated power consumption for 10 days - Trailer #2, August 2015 .......................................................................................... 100 

Figure 4-28: Sample simulated COP for 10 days - Trailer #2, August 2015 ................. 102 

Figure 6-1: A sample section of developed LabVIEW interface for acquiring system’s data and implementing the proactive controller ...................... 125 

Figure 6-2: Developed section for executing the cooling demand simulator in LabVIEW ............................................................................................... 126 

Figure 6-3: Developed section in LabVIEW interface for commanding compressor, evaporator and condenser fans based on optimization model results ..................................................................... 127 

xiv

Figure 6-4: Outdoor (ambient) temperature and heater power setting for steady state sample #1 ..................................................................................... 129 

Figure 6-5: Outdoor (ambient) temperature and heater power setting for steady state sample #2 ..................................................................................... 129 

Figure 6-6: Compressor and fans speed ratio variation for steady state sample #1, thermostatic control (low set point = 25.7, high set point = 27.7 oC) 132 

Figure 6-7: Compressor and fans speed ratio variation for steady state sample #2, thermostatic control (low set point = 23, high set point = 25 oC) 132 

Figure 6-8: Indoor (conditioned space) temperature variation for steady state samples, thermostatic control ................................................................ 133 

Figure 6-9: Power consumption of VACR system for steady state sample #1, thermostatic control ............................................................................... 133 

Figure 6-10: Power consumption of VACR system for steady state sample #2, thermostatic control ............................................................................... 134 

Figure 6-11: COP of VACR system for steady state sample #1, thermostatic control .................................................................................................... 134 

Figure 6-12: COP of VACR system for steady state sample #2, thermostatic control .................................................................................................... 135 

Figure 6-13: Compressor and fans speed ratio variation for steady state sample #1, proactive optimal control .................................................................. 135 

Figure 6-14: Compressor and fans speed ratio variation for steady state sample #2, proactive optimal control .................................................................. 136 

Figure 6-15: Indoor (conditioned space) temperature variation for steady state samples, proactive optimal control ........................................................ 136 

Figure 6-16: Power consumption of VACR system for steady state sample #1, proactive optimal control ........................................................................ 137 

Figure 6-17: Power consumption of VACR system for steady state sample #2, proactive optimal control ........................................................................ 137 

Figure 6-18: Heater power setting for transient test ...................................................... 140 

Figure 6-19: Outdoor (ambient) temperature setting for transient test .......................... 140 

Figure 6-20: Compressor and fans speed ratio variation for transient test, thermostatic control (low set point = 25.7, high set point = 27.7 oC) ..... 141 

xv

Figure 6-21: Indoor (conditioned space) temperature variation for transient test, thermostatic control (low set point = 25.7, high set point = 27.7 oC) ..... 141 

Figure 6-22: Power consumption of VACR system for transient test, thermostatic control .................................................................................................... 142 

Figure 6-23: COP of VACR system for transient test, thermostatic control .................. 142 

Figure 6-24: Compressor and fans speed ratio variation for transient test, proactive optimal control ........................................................................ 143 

Figure 6-25: Indoor (conditioned space) temperature variation for transient test, proactive optimal control ........................................................................ 143 

Figure 6-26: Power consumption of VACR system for transient test, proactive optimal control ....................................................................................... 144 

Figure 6-27: COP of VACR system for transient test, proactive optimal control ........... 144 

xvi

List of Acronyms

COP Coefficient of Performance

A/C-R Air Conditioning and Refrigeration

VACR Vehicle Air Conditioning and Refrigeration

VCR Vapor Compression Refrigeration

MPC Model Predictive Controller

GHG Greenhouse Gas

VCC Variable Capacity Compressor

FCC Fixed Capacity Compressor

PCM Phase Change Mater

ANSI American National Standard Institute

AHRI Air Conditioning Heating and Refrigeration Institute

ASHRAE American Society of Heating, Refrigeration, and Air Conditioning Engineers

RAAE Relative Average Absolute Error

NSP Number of Sample Points

xvii

Nomenclature

m Mass flow rate 1( . )kg s

L Length ( )m

t Time ( )s

u Velocity 1( . )m s

x Flow direction

Convection coefficient 2 1. .W m K

Nu Nusselt number

D Pipe diameter ( )m

Density 3.kg m

m Mass ( )kg

P Pressure kPa

h Enthalpy 1.( )kJ kg C

T Temperature K

Re Reynolds number

Pr Prandtl number

Void fraction

Efficiency

Cp Specific heat capacity 1 1. .J kg K

V Compressor displacement 3m per revolution

dC TEV constant

k Thermal conductivity 1 1. .W m K

X Vapor quality

A Area 2( )m

N Compressor rotational speed ( )rps

xviii

W Power consumption kW

0 8 to c c Compressor polynomial constants

0 8 to d d Compressor polynomial constants

0 3 to cf cf Fan power consumption constants

C Thermal capacity W C

*C Thermal capacity ratio ( min maxC C )

1 20 to g g Equation form

Q Heat transfer rate W

NTU Number of transfer units

UA Heat exchanger overall heat transfer coefficient W C

1 20 to x x Variables

E Total energy kJ

Heat exchanger effectiveness

H Equality constraints

G Inequality constraints

Metamodel coefficients

n Number of sample points

e Error of metamodel approximation

TSS Sum of squares-corrected

ESS Sum of squares due to error

xix

Subscripts/Superscripts

i Inside

cond Condenser

in Inlet

s Superheated

2 ph 2-phase

out Outlet

g Gas

c Condensing

a Air

v Volumetric

M Mechanical

e Evaporating

o Outside

ref Refrigerant

1mid Boundary of superheated/2-phase regimes

2mid Boundary of 2-phase/sub-cooled regimes

l Liquid

tot Total

w Wall

cr Critical

comp Compressor

E Electrical

TEV Thermostatic expansion valve

evap Evaporator

f Fan

max Maximum

min Minimum

nom Nominal

n Iteration step number

xx

avg

Average value

Metamodel-formulated

L Lower band

U Upper band

xxi

Executive Summary

Motivation

Air conditioning and refrigeration (A/C-R) systems are used in numerous

stationary and mobile applications to provide comfort for occupants or appropriate

temperature and humidity for refrigerated products. More than 20% of the total energy

consumption in buildings and up to 20% of the total fuel consumption in vehicles is

consumed by A/C-R systems. Among all the energy sectors in the world, the transport

sector is responsible for 22% of CO2 emissions that significantly contribute to global

warming. Approximately 31% of food supply chain includes refrigerated transportation.

There are more than 4 million transport refrigeration systems in the world, of which about

30% are trailers, 30% are large trucks, and 40% are small trucks and vans.

The efficiency of A/C-R systems is defined by the coefficient of performance

(COP), which is the ratio of the cooling output to the input power consumption. Among

all A/C-R applications, vehicle air conditioning and refrigeration (VACR) systems,

especially the refrigeration systems used in trucks and trailers, have some of the lowest

COPs, usually less than 1.5, compared to high efficiency stationary systems with COPs

greater than 3. As such, any improvement in the efficiency of VACR systems can lead to

significant global impact. The relatively low COP of VACR systems is in part due to more

frequent and inefficient on-off control of these systems, which is a result of the lack of

intelligent control modules, the small size of the systems, and intense load variations. As

such, in this research a high efficiency and optimally controlled VACR system is

developed that is equipped with variable speed compressor and fans as well as

optimization-based model predictive controller (MPC). The work herein has been

prompted by a collaborative research project with two local companies: Cool-It Group,

and Saputo Inc. This places the main focus on service vehicles; however, the developed

concept and platform can be expanded to the entire transportation industry.

xxii

Objectives

The research objectives are as follows:

o Establish the duty cycle(s) and investigate the thermal and performance

characteristics of VACR systems under real-time operating conditions by

developing mathematical model, experimental study, and field data acquisition.

o Demonstrate a proof-of-concept for optimization-based, model predictive control

of VACR systems.

o Provide a platform for proactive prediction and optimal control of VACR systems

over the duty cycle.

Methodology

The following highlights the methodology and approach taken for achieving the

objective of this research project:

o Designed and built an experimental setup and performed in-lab data acquisition

to obtain real-time performance characteristics of VACR systems.

o Acquired field data from a stationary commercial refrigeration system as a

starting point to develop the skills for on-site measurements. Established the duty

cycle, and investigated the thermal and energy performance characteristics of

commercial refrigeration systems.

o Performed comprehensive field data collection to establish the duty cycle(s),

thermal behaviour and performance of conventional VACR systems in

collaboration with Saputo.

o Developed new mathematical models for the simulation of thermal and

performance characteristics of VACR systems under realistic operating

conditions and duty cycles.

o Validated accuracy of the developed models using the acquired real-time data.

o Built a lab-scale proof-of-concept demonstration of a high efficiency VACR

system based on variable speed compressor and fans and high efficiency heat

exchangers.

xxiii

o Developed an optimization model to determine the optimum compressor and

fans speeds under real-time operating conditions.

o Developed a proactive controller by integrating the developed optimization model

with an existing cooling demand simulation tool.

o Implemented the proactive optimization-based controller on the built VACR

system, evaluated the performance improvement, and finalized the proof-of-

concept demonstration of high efficiency VACR system to be employed in the

service transportation industry.

Contributions

A summary of the contributions of this research project is listed as follows:

o Established the duty cycle and investigated the thermal and performance

characteristics of a pilot commercial A/C-R system. A potential energy saving of

4,600 kWh/yr and COP improvement of 29-39% for the pilot system over the duty

cycle was evaluated [7].

o Established the duty cycle and investigated the thermal and performance

characteristics of VACR systems theoretically and experimentally. A potential

reduction of 3,105 litres/yr of diesel fuel from one pilot truck-trailer equivalent to

8,320 kg CO2/yr (10 million tons of CO2/yr globally) was estimated [3].

o Developed a new steady state model and studied the VACR systems’ thermal

and performance behaviour under steady state conditions. Validated the model

using the collected data in the lab and on-board testing [1].

o Developed and validated quasi-steady state and transient models. The models

were used to investigate real-time thermal and performance characteristics of the

VACR systems [3, 5].

o Built a high efficiency, variable capacity VACR system by employing high

efficiency heat exchangers and variable speed compressor and fans, showed

more than 150% COP enhancement [4, 6, 12, 14-15].

o Developed optimization models based on steady state and transient thermal

models [2, 4].

xxiv

o Developed a proactive and model predictive controller (MPC) for the VACR

system by integrating the optimization model and an existing cooling demand

simulation tool [2, 8-12].

o Developed a proof-of-concept demonstration of a proactively and optimally

controlled high efficiency VACR system that can be integrated in the service

transportation industry. Up to 24% COP enhancement was achieved when

compared to the on-off thermostatic controller systems [2, 8-13].

References

1- F. Bagheri, M. A. Fayazbakhsh, P. C. Thimmaiah, M. Bahrami, “Theoretical and Experimental Investigation into Anti-idling A/C System for Trucks”, Journal of Energy Conversion and Management, Vol. 98, 2015, pp. 173–183.

2- F. Bagheri, M. A. Fayazbakhsh, M. Bahrami, “Proactive and Optimal Control of A/C-R Systems”, In Preparation.

3- F. Bagheri, M. A. Fayazbakhsh, M. Bahrami, “Real-time Performance Evaluation and Potential GHG Reduction in Refrigerated Trailers”, Submitted to International Journal of Refrigeration, Under Review.

4- F. Bagheri, M. A. Fayazbakhsh, M. Bahrami, “Performance Optimization of a Battery-Powered Vehicle Air Conditioning System”, Submitted to Journal of Green Energy, Under Review.

5- F. Bagheri, M. A. Fayazbakhsh, M. Bahrami, “Investigation of Optimum Refrigerant Charge and Fans Speed for a Vehicle Air Conditioning System”, Submitted to ASME Journal of Thermal Science and Engineering Applications, Under Review.

6- F. Bagheri, M. Fakoor-Pakdaman, M. Bahrami, "Utilization of Orthotropic Graphite-Based Plates in Plate Heat Exchangers", International Journal of Heat and Mass Transfer, Vol. 77, 2014, pp. 301-310.

7- F. Bagheri, M. A. Fayazbakhsh, M. Bahrami, “Thermodynamic Investigation of a Typical Commercial Refrigeration System”, IEEE SEMI-THERM 32 Conference, San Jose, March 2016.

8- M. A. Fayazbakhsh, F. Bagheri, M. Bahrami, “An Inverse Method for Calculation of Thermal Inertia and Heat Gain in Air Conditioning and Refrigeration Systems”, Journal of Applied Energy, Vol. 138, 2015, pp. 496–504.

9- M. A. Fayazbakhsh, F. Bagheri, M. Bahrami, “A Resistance-Capacitance (RC) Model for Real-Time Calculation of Cooling Load in HVAC-R Systems”, ASME Journal of Thermal Science and Engineering Applications, Vol. 7, 2015, 041008.

10- M. A. Fayazbakhsh, F. Bagheri, M. Bahrami, “A Self-Adjusting Method for Real-Time Calculation of Thermal Loads in HVAC-R Applications”, ASME Journal of Thermal Science and Engineering Applications, Vol. 7, 2015, 041012.

xxv

11- M. A. Fayazbakhsh, F. Bagheri, M. Bahrami, “Real-Time Thermal Load Calculation by Automatic Estimation of Convection Coefficients”, International Journal of Refrigeration, Vol. 57, 2015, pp. 229–238.

12- M. A. Fayazbakhsh, F. Bagheri, M. Bahrami, "Gray-Box Model for Energy-Efficient Selection of Set Point Hysteresis in HVAC-R Controllers”, Energy Conversion and Management, Vol. 103, 2015, pp. 459-467.

13- Y. Huang, A. Khajepour, F. Bagheri, M. Bahrami, “Modelling and Optimal Energy-Saving Control of Automotive Air-Conditioning and Refrigeration Systems”, Journal of Automobile Engineering 19, 2016, 10.1177/0954407016636978.

14- M. Fakoor-Pakdaman, M. Ahmadi, F. Bagheri, M. Bahrami, " Optimal Time-Varying Heat Transfer in Multilayered Packages with Arbitrary Heat Generations and Contact Resistance”, ASME Journal of Heat Transfer, Vol. 137, 2015, 081401.

15- M. Fakoor-Pakdaman, M. Ahmadi, F. Bagheri, M. Bahrami, "Dynamic Heat Transfer Inside Multilayered Packages with Arbitrary Heat Generations”, AIAA Journal of Thermophysics and Heat Transfer, Vol. 28, 2014, pp. 687-699.

1

Chapter 1. Introduction

1.1. Research Background and Importance

Despite widespread attention during the past decades, there is a significant

potential to reduce energy consumption and greenhouse gas (GHG) emissions from

energy systems all around the globe. Based on the fifth report from Intergovernmental

Panel on Climate Change released in 2013, the average global surface temperature has

increased by 0.85 K during 1880-2012, which accelerates the rate of glacier melt and

sea level rise [1,2]. Among all the energy sectors in the world, the transport sector is

responsible for 22% of global CO2 emissions, which significantly contribute to global

warming [3,4]. Around 31% of the food supply chain includes refrigerated transportation

[5,6]. Furthermore, about 20% of total global refrigerant emissions come from mobile air

conditioning and refrigeration systems [7].

Air conditioning and refrigeration (A/C-R) systems are used in many stationary

and mobile applications to provide either comfort conditions for people or an appropriate

condition for food and other temperature/humidity sensitive creatures/products. These

systems consume a tremendous amount of energy leading to large quantities of GHG

emissions. In the U.S., A/C-R systems consume more than 20% of the total energy used

in residential buildings and more than 25% in commercial buildings [8]. It was reported

that more than 26 billion liters of fuel oil is consumed each year just to run air

conditioning systems in light-duty vehicles in the U.S. [9,10]. Running vehicle air

conditioning and refrigeration (VACR) systems increases higher fuel consumption in

compact-midsize vehicles by 12-17% [11,12].

refrig

expa

In th

comp

pump

cond

secon

from

the e

low-p

coolin

evapo

The e

in ag

Figur

Most exis

eration (VC

nsion valve,

his type of

ponent, suck

ps the high

enser, the g

ndary flow (

the conden

expansion va

pressure, low

ng effect of

orates. This

evaporated

ain by the co

re 1-1: Sche

sting A/C-R

R) cycle and

, and fans. A

f refrigeratio

ks the refrig

h-pressure,

gas is cond

(usually air,

ser then go

alve, the pre

w-temperatu

f an A/C-R

s evaporatio

refrigerant,

ompressor a

ematic of a

R equipmen

d consists o

A schematic

on cycle, t

gerant gas f

high-tempe

densed and

in VACR a

oes to the ex

ssure and te

ure two-pha

system ap

on absorbs h

in the form o

and one cycl

basic VCR

2

t operates

f a compres

c of the basi

he compre

from the ev

rature gas

the latent

pplications).

xpansion va

emperature

ase refrigera

pears in th

heat from a

of saturated

e is complet

cycle [14]

based on

ssor, a cond

c VCR cycle

essor, the m

vaporator an

into the c

and sensib

. The satura

alve. As a re

of refrigeran

ant flows in

he evaporato

a secondary

d or superhe

ted [13,14].

the vapor

enser, an ev

e is shown i

most energ

nd, after co

condenser.

ble heat is r

ated or sub-

esult of thro

nt drops dra

nto the eva

or where th

flow and c

eated gas, is

compressio

vaporator, a

in Figure 1-1

gy-consumin

mpressing

Through th

rejected to

-cooled liqui

ttling throug

astically and

aporator. Th

he refrigeran

cools it down

s then sucke

on

an

1.

ng

it,

he

a

id

gh

a

he

nt

n.

ed

3

The efficiency of A/C-R systems is defined by the coefficient of performance

(COP), which is the ratio of the cooling power output to the input power consumption.

A/C-R systems in transportation applications operate in a harsh environment (intensively

variant) and experience a wide range of cooling demand and constraints imposed by

available space and weight. As such, these systems have lower COP (usually less than

1.5) than stationary systems (usually larger than 3) [15–18]. In addition to the low COP,

the quantity of transported goods, amount of home delivery, and expectations for the

quality of goods are increasing, resulting in a tremendous amount of energy

consumption by the refrigerated transport industry [19,20]. Furthermore, the globalized

nature of food production has led to longer transit distances for many food products [21].

In the U.S. alone, it is estimated that processed foods are, on average, transported over

2,100 km before being consumed and fresh food products 2,400 km [21,22]. In addition,

delivery of food products requires frequent opening of the door, which leads to air

infiltration and a remarkable increase in the cooling demand. It is reported that a food

product can be subjected to as many as 50 door-openings during a delivery [23]. The

low performance, combined with the harsh operating conditions, lead to significant global

impacts and put high pressure on the food industry to find remedies to reduce the

energy consumption of refrigerated transport [19].

1.2. Research Motivation

Due to the low COP of the numerous VACR systems worldwide, any small

improvement can bring a significant global impact. The relatively low COP of VACR

systems is in part due to more frequent and inefficient on-off cycling of these systems,

which is a result of the lack of intelligent control modules, the small size of the systems,

and a more intense load variation. The more intense load variations in VACR systems

are due to poor compartment insulation, the high frequency of opening and closing doors

or windows, sun exposure, and movement of the vehicle.

4

Accordingly, development of an optimally controlled, proactive VACR system with

higher efficiency to replace existing inefficiently controlled VACR systems will have

significant global improvement on energy consumption and the corresponding GHG

emissions. The optimally controlled VACR system is equipped with a variable speed

compressor and variable speed fans that enable the capacity of the system to be

controlled instead of using on-off cycling. The focus of this research is on service

vehicles; however, the developed concept and system could be used in all sectors of the

transportation industry.

The work herein has been prompted by a collaborative research project with two

local companies; Cool-It Group, an Abbotsford, BC based manufacturer of anti-idling

battery-powered VACR systems for truck and vans; and Saputo Inc., Canada’s largest

dairy product processor, located in Burnaby, BC. The low COP of battery-powered

VACR systems leads to a relatively short life for the batteries and highly affects the

reliability of Cool-It’s products. Also, Saputo spends an enormous amount on fuel for

their refrigerated trailers that are used to deliver dairy products every day. Due to the

impact of VACR systems on countrywide fuel consumption and the environment, this

research project is financially supported by the Natural Sciences and Engineering

Research Council of Canada (NSERC) through an Automotive Partnership Canada

(APC) Grant, No: APCPJ 401826-10.

1.3. Research Scope

Based on the reported low performance and significant global impacts of VACR

systems and as a result of collaboration with two local companies; Cool-It Group and

Saputo Inc., this research is focused on developing high efficiency VACR system for

service vehicles. The objectives of the research are to: 1) establish the duty cycle and

actual operating conditions for refrigerated (delivery) service vehicles, 2) investigate the

performance of VACR systems under real-time operating conditions employing

mathematical techniques and experimentation, 3) demonstrate a proof-of-concept for

5

optimization-based, model predictive control of VACR systems, and 4) provide a

platform for proactive prediction and control of VACR systems over the duty cycle. As

such, an experimental setup is developed to perform in-lab data acquisition and obtain

the real-time performance characteristics of VACR systems. In addition, comprehensive

field data is collected to establish the duty cycle(s) and obtain real-time thermal behavior

and performance characteristics of the VACR systems. Mathematical models are

developed and validated to simulate thermal and performance behavior of VACR

systems under steady state and transient conditions. Furthermore, a proof-of-concept

demonstration of a high efficiency VACR system based on variable speed compressor

and fans and high efficiency heat exchangers is developed. A model-based optimization

method is employed to determine the optimum compressor and fans speeds under any

imposed operating condition. Finally, an in-lab proof-of-concept demonstration for

proactive and optimal control of VACR systems by integrating the developed model-

based optimizer with an existing cooling demand simulator is developed. The developed

concept and platform can be expanded to the entire transportation industry and even

stationary systems and will lead to significant global energy savings and GHG reduction.

1.4. Thesis Structure

Based on the scope of research, Figure 1-2 represents the structure of the

Thesis. Chapter 2 covers the development of the mathematical model. In Chapter 3, the

development of the experimental test bed and acquisition of field data are presented.

Model validation as well as an investigation of the thermal and performance behaviour is

presented in Chapter 4. The metamodels for optimization are developed in Chapter 5.

The integration of the model equipped with the optimization solver and an existing

cooling demand simulator is covered in Chapter 6. Chapter 6 also represents a proof-of-

concept demonstration of a high efficiency VACR system equipped with a proactive

optimization-based controller as the expandable solution for the future of the

transportation industry.

6

Figure 1-2: Thesis structure

7

1.5. Relevant Literature

A/C-R systems have been the topic of many studies in the last decades due to

their significant global energy consumption and corresponding environmental impact. In

spite of numerous studies in the literature focusing on the stationary A/C-R systems in

different applications, the number of studies relevant to general category of VACR

systems is limited [24–30]. In addition, most VACR-related studies have concentrated on

engine-driven car A/C-R systems rather than the service vehicle category. Furthermore,

the literature lacks an in-depth study on performance optimization of VACR systems in

service vehicles. The literature review is represented in two separate categories: 1)

thermal and performance studies on A/C-R systems, 2) control studies on A/C-R

systems.

1.5.1. Thermal and performance studies on A/C-R systems

The majority of the studies in the field of VACR systems are related to the

assessment of the effects of refrigerant type on the characteristics and the performance

[30–34]. Yoo et al. [33] studied the performance of an automotive air conditioning system

charged with R152a, and compared it to R134a. The results of this study showed that

the R152a system was slightly better than R134a not only under driving conditions, but

also under idling conditions. Brown et al. [34] compared a VACR system charged with

CO2 and R134a, and showed that the R134a refrigerant led to a 21-34% higher

coefficient of performance (COP). Cho et al. [30] experimentally compared the

characteristics of a VACR system charged alternatively with R134a and R1234yf, and

showed that using R1234yf resulted in a relatively lower compressor power consumption

and cooling capacity by 4% and 7%. They also investigated the effectiveness of

installing an internal heat exchanger for heat transfer between cold refrigerant leaving

the evaporator and warm refrigerant leaving the condenser to increase the amount of

sub-cooling and found a 4.6% improvement in COP.

8

A variety of other important parameters including refrigerant charge, compressor

speed, fan speed, and ambient temperature have also been considered in VACR system

studies. Ratts et al. [35] experimentally analyzed the COP of a passenger car A/C-R

system, focusing on relationships between the COP, the compressor revolution, and the

vehicle speed. Their analysis showed that the performance of the system degraded with

increasing vehicle speed. Macagnan et al. [17] experimentally investigated effects of the

amount of refrigerant charge and compressor speed on the cooling power and

performance of a typical VACR and showed that although the refrigerant charge did not

sensibly affect the VACR system characteristics, the compressor speed caused a

variation in the system COP in a range of 1-1.8. Jabardo et al. [36] studied the effects of

compressor speed and the evaporator and condenser inlet air temperatures on the

performance of a VACR by developing a steady state thermodynamic simulation model.

They experimentally changed the refrigerant charge in a relatively narrow range,

however, could not extract a pattern for the behavior of cooling capacity and COP versus

the refrigerant charge due to the limited range of studied charge.

Lee et al. [29] developed a simulation model for a VACR system based on

component sub-models and reported a 7% discrepancy with experimental data. They

investigated the effects of refrigerant charge and condenser size on the system COP.

Wang et al. [16] measured the vapor quality of refrigerant R134a at the inlet of a

compressor and analyzed the effects of different parameters on the COP of a VACR

system. They showed that the COP of the system decreases with increasing of

refrigerant charge, condensing temperature and compressor speed, and increases with

the increase of evaporator air inlet temperature. They also demonstrated that the total

quality of the refrigerant at the compressor inlet increases with increasing the evaporator

air inlet temperature, and decreases with the increase of the refrigerant charge,

condensing temperature, and compressor speed. The value of COP reported was in the

range of 1.1-2.5.

Since VACR compressors are usually belt-driven by the vehicle engine, the

compressor speed is directly proportional to the engine speed. The compressor can be

either of fixed or variable capacity. Therefore, the cooling capacity of a VACR system

using a fixed capacity compressor (FCC) varies as a function of the engine speed.

9

Consequently, the FCC continually cycles on and off to meet the air conditioning

demand. On the other hand, the cooling capacity of a VACR system using a variable

capacity compressor (VCC) is independent of the compressor speed, because in this

type of compressor the piston displacement varies by speed in order to control the

capacity. Therefore, such a system meets the air conditioning demand at all operating

conditions with much less cycling effects. Alkan et al. [37] experimentally compared

employing VCC or FCC for a VACR system at different engine rotational speeds as well

as evaporator and condenser air stream characteristics. Based on the steady state tests,

they showed that the FCC operations provided a rising cooling capacity with increasing

compressor speed, while the cooling capacity of the VCC remained almost constant

after a certain compressor speed due to the intervention of the capacity control system.

It was shown that although the COP of VCC system during the low engine rpm

operations was slightly poorer, it surpassed the COP of FCC system at the higher rpms.

Moreover, the COP of FCC system decreases persistently with the increase in

compressor speed. Tian et al. [38,39] studied the transient behavior of a VACR system

equipped with a variable capacity compressor experimentally and mathematically using

a steady-state-based compressor model. They assumed that the changes in compressor

rotary speed are very rapid and mainly focused on the transient changes in refrigerant

pressures and piston stroke length after a sudden compressor speed change. They

showed that the variation of a VCC’s piston stroke length usually has a time lag after

changes of the effective parameters (mainly the compressor’s rpm) and some

regulations are required for achieving any desired cooling capacity due to engine’s rpm

variations during the transient operations.

Returning to the category of stationary A/C-R systems, few studies can be found

that focused on performance optimization of these systems. Abdullah et al. [40] studied

the effects of evaporator temperature on the COP of a stationary A/C-R system and

obtained a maximum COP by changing this temperature. Green et al. [41] introduced a

performance function that encompassed food quality, energy efficiency, and system

reliability, to enable performance assessment and optimization of a supermarket

refrigeration system. They studied the effects of thermostatic set point temperature as

the control parameter to optimize the defined performance function. Sahin et al. [42]

presented a thermo-economic performance model for a two-stage refrigerator and

10

determined the optimal operation and design parameters of the refrigerator based on the

thermo-economic criterion. They mainly focused on the effects of investment and energy

costs in designing a high efficiency cascade refrigerator and demonstrated that the

economical parameter had a great influence on the optimal operating and design

condition.

In addition, a variety of innovative technologies to increase the energy efficiency

of VACR systems have been introduced in the literature. Waste-heat-driven air

conditioning systems, which operate based on absorption or adsorption cycles, have

been introduced in several studies as a promising way to make sustainable VACR

systems [9,43,44]. Also, employing microchannel condensers as a performance

enhancement technology for VACR systems is covered in reference [45]. Muller [46]

introduced the feasibility of employing magnetocaloric cooling for transportation industry.

Gould et al. [47] studied the effects of capturing waste heat from the internal combustion

engine exhaust to run air conditioning system using steam-pressure-exchange ejector

for an automobile. Unal et al. [48] investigated the effects of using two-phase ejectors as

the expansion valve in the A/C-R system of buses and showed the potential for a15%

performance improvement.

A summary of the most relevant existing literature on thermal and performance

characteristics of VACR systems is presented in Table 1-1. Because a major part of this

thesis is focused on refrigerated trailers, the literature summary is separated for this type

of VACR systems.

11

Table 1-1: Summary of literature on thermal and performance studies of A/C-R systems

Classification Author Notes

Refrigerated

trailers

Jolly et al. [49]

√ Mathematical modeling of refrigeration unit in a typical

electricity-driven reefer.

√ Measurements of a few thermal parameters under

ambient condition mimicked by an environmental

chamber.

× Did not study the behavior of COP.

× Restricted to steady state condition and full load

operation.

× Did not consider the effects of opening the door,

loading/unloading, or transient ambient conditions on the

behavior of the VCR system.

Tso et al. [50]

√ Mathematical modeling of refrigeration unit in a typical

electricity-driven reefer.

√ Considered effects of hot gas bypass and suction

modulation control on performance in partial load.

× No experimental data or model validation.

× Restricted to steady state condition.

× Did not consider the effects of opening the door,

loading/unloading, or transient ambient conditions on the

behavior of the VCR system.

Moureh et al.

[51,52]

√ Numerical and experimental study of conditioned air

distribution through refrigerated trailers.

√ Considered the effects of using ducts on air distribution.

× No performance evaluation of the refrigeration system.

× Did not consider the real-time performance and

capacity variation of the refrigeration system.

12

Defraeye et al.

[53]

√ Numerical study on conditioned air distribution through

refrigerated trailers.

√ Considered effects of food temperature, direction of

airflow distribution, spacing between pallets, etc., on food

temperature.

× No performance evaluation of refrigeration system.

× Did not consider the real-time performance and

capacity variation of the refrigeration system.

Rodríguez-

Bermejo et al. [54]

√ Experimental study on temperature distribution

throughout refrigerated trailers.

√ Considered effects of being loaded or unloaded and

thermostat commands on temperature distribution.

× No measurements or simulation of a VCR system.

Tassou et al. [19]

√ Estimated average energy intensity, fuel consumption,

and CO2 emission of different food distribution systems in

the UK, using statistical method and literature data.

√ Suggested heat-driven and air-cycle refrigeration

systems instead of VCR for food transport and estimated

the potential energy saving.

× No real-time measurements or modeling for accurate

performance investigation of refrigerated vehicles were

performed.

Spence et al. [55]

√ Designed, built, and tested an air-cycle refrigeration

system instead of VCR for food transport to eliminate the

direct HFC/HCFC-related environmental impacts.

× In general, resulted in lower COP and higher fuel-

related environmental impact.

× Neither modeling was performed nor was the transient

behavior assessed under real-time conditions.

General Yoo et al. [33] √ Evaluated effects of refrigerant type on the COP of a

VACR system.

13

category of

VACR systems

√ Reported a slightly better COP for R152 compared to

R134a.

Brown et al. [34]

√ Evaluated effects of refrigerant type on the COP of a

VACR system.

√ Reported a significantly better COP for R134a

compared to CO2.

Ratts et al. [35]

√ Reported degradation of the COP for an engine-driven

VACR system with an increase in the speed of the

vehicle.

Macagnan et al.

[17]

√ Assessed effects of refrigerant charge and compressor

speed on the COP of a VACR system.

√ Reported that the compressor speed significantly

affects the COP.

× Reported that the refrigerant charge does not

remarkably affect the COP.

Zheng et al. [45] √ Employed microchannel condenser to enhance the

COP of a VACR system

1.5.2. Control studies on A/C-R systems

Controlling the operation of A/C-R systems to achieve a high performance has

been the topic of several studies. A variety of methods have been employed to efficiently

control the performance of A/C-R systems including: PID [56], power law [57], fuzzy logic

[11,58–60], neural networks [61], and model predictive [62–64]. Khayyam et al. [58]

illustrated that it is extremely difficult to control nonlinear and complex A/C-R systems

using a classical PID controller. The power law and fuzzy logic controllers can address

the nonlinearity of an A/C-R system’s behavior if stated in a simplified closed form,

however, the coupled behavior of A/C-R system components does not allow for an

accurate simplification towards a closed-form model. Employing the neural network

14

method requires obtaining a wide range of data for training the multi-input multi-output

model and may not be accurate for conditions not covered during the training period

[58,61]. In the model predictive control (MPC) approach, a model of the A/C-R system is

simulated to predict the operational condition and command the system components

accordingly. Since all the governing equations of the A/C-R system can be in effect in

such a controller, the MPC approach is one of the most accurate and flexible

approaches of controlling and optimizing A/C-R systems [62–64].

Yeh, Lin et al. [65–67] theoretically and experimentally introduced the effects of

speed control for outdoor (condenser) and indoor (evaporator) fans on a split-type

residential A/C-R system’s transient response and efficiency. They designed the control

algorithms based on a low-order, linear model obtained from system identification. Their

first algorithm, which modulated the outdoor fan speed based on the temperature

difference between the condensing and the ambient temperature, showed a reduction in

the steady-state power consumption. Their second algorithm added one more degree of

freedom by modulating the indoor fan’s speed at three levels of low, medium, and high.

It was shown that by controlling the indoor fan, the transient response of the system was

improved; however, the energy efficiency was not considerably affected. They concluded

that the speed of only one fan should be varied while the other one is kept at a constant

speed.

Khayyam et al. [11,58] employed fuzzy logic to develop an intelligent control

system for an engine-driven vehicle air conditioning system, aiming to reduce the total

engine energy consumption during a drive cycle. Their developed controller included

three modules: a fuzzy cruise controller to adapt vehicle cruise speed via prediction of

the road ahead using a look-ahead system, a fuzzy air conditioning controller to produce

a desirable temperature and air quality inside the vehicle cabin via a road information

system, and a fuzzy engine controller to generate the required engine torque to move

the vehicle smoothly on the road. The introduced look-ahead system was an algorithm

that estimated future road slope within a distance of 300-500 m ahead of the vehicle.

The A/C-R system was controlled by setting the blower (evaporator fan) speed based on

feedback from the cabin temperature. The main focus of their study was on engine

15

power control to move the vehicle appropriately, while the A/C-R load was taken into

account as a by-product

Ricker [64] developed a MPC for a supermarket refrigeration system in which all

the actuators were on-off devices. A simplified compressor model employing the cycle’s

high pressure and the cabinet temperature to generate on-off commands were prepared

as the brain of the MPC. They could reduce the frequency of on-off cycling using the

developed MPC controller instead of the existing simple thermostatic controller.

Hovgaard et al. [68] presented the economical optimization of a supermarket

refrigeration system by employing a MPC controller based on the compressor and

refrigerated space thermal models. By optimizing a defined cost function, they showed

that the operational costs of the refrigeration system could be reduced if thermal energy

storage was added to the refrigeration system. Schalbart et al. [62] developed a MPC to

optimize the energy management of an ice-cream warehouse refrigeration system

coupled to a phase-change material (PCM) tank. The MPC model was developed based

on a PCM tank thermal sub-model as well as a simple refrigeration system sub-model.

They showed a potential for energy management by integrating the refrigeration and

thermal storage systems using a MPC approach. Based on the literature, employing an

MPC for performance optimization of A/C-R systems can be a promising methodology.

However, it requires an accurate mathematical simulator of the A/C-R system to operate

satisfactorily. In addition to the mathematical model development, an in-depth

understanding of the A/C-R system’s behavior under real-time operating conditions is

required for an effective design of the controller. Although the performance

characteristics of stationary A/C-R systems have been well attended in the literature,

VACR systems have not been studied in-depth so far. A concise summary of the major

findings from the performed literature review is classified in Table 1-2.

16

Table 1-2: Summary of the literature on control of A/C-R systems

Author Notes

Sørensen et al.

[69]

√ Developed a MPC for the refrigeration unit of reefers to increase

performance.

× Restricted to electrically driven reefers used for sea transportation.

× Employed a simplified linear model for thermal behavior of cargo.

× No comprehensive model development of refrigeration unit for MPC.

× No Real-time study of power consumption and performance behavior.

Khayyam et al.

[11,58]

√ Developed a controller for an engine-driven VACR system.

√ Developed a concept of look-ahead for engine torque prediction

based on road slope.

× Controlled only evaporator fan speed to adjust the cabin temperature.

× Employed a simplified controller model based on fuzzy logic.

Lin et al. [65–67]

√ Developed a MPC for a split A/C-R unit in a residential building.

× Employed a low-order linear model in MPC.

× Considered only variable speed fans and controlled them separately.

Ricker [64]

√ Developed a MPC for a supermarket refrigerator.

× Employed a simplified A/C-R model only based on the cycle’s high

pressure and the refrigerated space temperature.

× Developed an inefficient on-off control for constant speed compressor

and fans.

Hovgaard et al.

[68]

√ Developed a MPC for a supermarket refrigerator.

× Employed a simplified economic-based model for MPC.

× Controlled constant speed components by on-off commands.

17

1.5.3. Lack of literature

The reported literature review indicates that the performance optimization and

proactive control development of VACR systems has not been studied in-depth, and the

pertinent literature lacks the following:

o An in-depth study on the duty cycle, thermal, and performance

characteristics of VACR systems employed in service transportation

under realistic operating conditions.

o Model-based performance optimization and control of variable capacity

VACR systems under transient operating conditions.

o Proactive and model predictive control of the operation of a VACR system

based on cooling demand prediction and VACR behavior simulation.

18

Chapter 2. Mathematical Model Development

Mathematical simulation of A/C-R systems is an effective method for

investigating the thermal behavior and thermodynamic characteristics of these systems

and to obtain the power consumption, cooling power, and COP under any operating

conditions. A mathematical model for a complete A/C-R system is an integration of the

component sub-models, including the compressor, condenser, evaporator, and

expansion valve. Therefore, a detailed analysis of each component in an A/C-R system

is covered during the modeling procedure. The detailed component analysis in the

mathematical modeling results in a complex set of conservation equations that should be

solved numerically [25,26]. Thus, a numerical algorithm and a solver should be

developed to simulate VCR systems.

In Figure 2-1, a schematic of a basic A/C-R system and the main

thermal/thermodynamic parameters used in the mathematical modeling are presented.

In this chapter, the mathematical modeling of such a system is presented in three

sections: 1) steady state model, 2) quasi-steady state model, 3) transient model.

19

Figure 2-1: Schematic of a basic A/C-R unit and the main parameters

2.1. Steady State Model Development

In this section a steady-state mathematical model for investigating the

thermal/thermodynamic behavior of A/C-R systems under various steady state operating

conditions is developed. The steady state model is a basis for the development of the

quasi-steady state and fully transient models and is employed to perform a

comprehensive parametric study of the characteristics of the system. The major output

parameter from the modeling of an A/C-R system is the COP. In addition to the COP,

other salient parameters of an A/C-R system that are obtained from the modeling include

the cooling power, input power (sum of compressor, evaporator and condenser fan

20

powers), and refrigerant mass flow rate. An integration of the component sub-models of

an A/C-R system forms its complete mathematical model [70].

2.1.1. Compressor sub-model

A map-based model is used in this study for thermodynamic simulation of the

compressor under steady state conditions. It has been experimentally shown that this

approach had a good agreement with experimental data [49,71,72]. Second-order

polynomials are developed for the compressor input power and refrigerant mass flow

rate quantities based on the manufacturer's test data, as shown in Eqs. (2.1) and (2.2) in

Table 2-1. These equations yield a maximum of 2.5% discrepancy when compared to

the manufacturer's data. It should be noted that the manufacturer’s data is usually

provided for fixed magnitudes of superheating and subcooling. If the A/C-R system is

equipped with a thermostatic expansion valve, the refrigerant flow rate is controlled to

achieve a pre-set amount of superheating. Accordingly, throughout the duty cycle the

degree of superheating at the outlet of the evaporator is kept nearly constant. However,

the degree of subcooling varies, depending upon the imposed conditions on the system

from the air-stream sides at the condenser (major impact) and evaporator (minor

impact). Therefore, a modification will be required for different degrees of superheating

and subcooling. In the present research, wide ranges of superheating and subcooling (0-

15 oC) are tested and the modified correlations of the compressor model under different

superheating and subcooling values are obtained. The refrigerant side thermodynamic

correlation for the compressor is presented in Eq. (2.3), as seen in Table 2-1. Based on

these correlations, with known inlet conditions at the compressor inlet, the state point of

the refrigerant gas at the compressor discharge can be calculated.

2.1.2. Condenser and evaporator sub-model

Energy balance correlations between the refrigerant and air flow in condensers

and evaporators are used to model heat transfer in these components of the A/C-R

system. In the present study, we follow Shah and Sekulic [73], and an NTU model is

21

used to derive the mathematical model for the condenser and evaporator. In this

approach, the effectiveness is defined as the ratio of the actual heat transfer to the

maximum possible heat transfer; see Eq. (2.4) of Table 2-1. The maximum possible heat

transfer happens when the temperature of one stream at the outlet reaches the inlet

temperature of the other stream and is defined in Eq. (2.5). The actual heat transfer

between the air stream and refrigerant flow can be calculated using Eq. (2.6). To use

this equation, it is required that be found from Eqs. (2.7) and (2.8). For the current

study, the condenser and evaporator specifications are obtained from the manufacturer’s

datasheet and drawings. In addition to the NTU correlations, which represent the

heat transfer between the air and refrigerant streams, correlations of enthalpy change for

both the refrigerant and air streams should be added to complete the heat transfer

model for the evaporator and condenser. Equations (2.9) and (2.10) represent the

enthalpy change for the air and refrigerant streams in the condenser and evaporator.

The cooling capacity of an A/C-R system is obtained using either of these equations (Eq.

(2.9) or (2.10)) when written for the evaporator. Also, the total heat rejection of the

system is calculated using either of these equations for the condenser. For the air

stream in the evaporator, the enthalpy is calculated based on the temperature and

relative humidity to include both the sensible and latent heat transfers. However, for the

air stream in the condenser, the enthalpy of air can be simply found using the

temperature and thermal capacity of air since there is no water condensation on the

condenser coil.

It should be mentioned that two heat transfer regimes exist in the evaporator: 1)

two-phase heat transfer, relevant to the saturated zone; and 2) single-phase heat

transfer, relevant to the superheated zone. Also, the condenser can be divided into three

heat transfer regimes: single-phase superheated; two-phase saturated; and single-

phase subcooled. Due to governance of different convective heat transfer equations for

the mentioned regimes, in this study the evaporator and condenser are divided into two

and three sub-zones, respectively. Thus, the NTU model is applied for each sub-

zone separately to achieve a high level of accuracy. In addition to the heat transfer

model for the condenser and evaporator, map-based correlations are employed to obtain

22

the power consumption of the fans using manufacturer’s datasheets [74], see Eq. (2.11)

in Table 2-1.

Table 2-1: Mathematical model correlations

Compressor sub-model

2 2 2 2 2 20 1 2 3 4 5 6 7 8comp e c e e c c e c e c e cW c c T c T c T c T T c T c T T c T T c T T (2.1)

2 2 2 2 2 20 1 2 3 4 5 6 7 8ref e c e e c c e c e c e cm d d T d T d T d T T d T d T T d T T d T T (2.2)

, , , ,( )ref ref comp out ref comp in comp E Mm h h W (2.3)

Condenser and

evaporator sub-model

max

QQ

(2.4)

max min , ,( )hot in cold inQ C T T (2.5)

min , ,( )hot in cold inQ C T T (2.6)

*

* *1 exp[ (1 )]

1 exp[ (1 )]NTU C

C NTU C

(2.7)

* min

max min,C UAC NTUC C (2.8)

, ,a a in a outQ m h h (2.9)

, ,ref ref out ref inQ m h h (2.10)

2 3

, 0 1 2 3, , ,

a a af f nom

a nom a nom a nom

m m mW W cf cf cf cf

m m m

(2.11)

Expansion valve sub-

model , , , ,ref TEV in ref TEV outh h (2.12)

Refrigerant

thermodynamic

correlations

1 2 2 3( , ), ( ), ( ), ( )i i i e e c c a ah f P T T f P T f P cp f T (2.13)

23

2.1.3. Thermostatic expansion valve sub-model

Following Ref. [49], an isenthalpic model is selected for thermodynamic

simulation of the thermostatic expansion valve. In this model, as a result of the adiabatic

assumption, the inlet and outlet refrigerant enthalpies are considered to be the same,

see Eq. (2.12) in Table 2-1.

2.1.4. Thermodynamic correlations for the refrigerant

In addition to the main component sub-models, a few thermodynamic

correlations are also employed to add the refrigerant thermodynamic properties to the

model. HFC-134a refrigerant is used in most VACR systems, including the main system

studied in this research. Thus, the correlations between the thermodynamic properties of

HFC-134a are obtained and employed [75]; see Eq. (2.13) Table 2-1.

2.1.5. Numerical solver

Developing the mathematical model for an A/C-R system, including the

correlations shown in Table 2-1 and the condenser and evaporator overall heat transfer

relationships, leads to a set of 20 coupled nonlinear equations that have to be solved

simultaneously. The A/C-R system unknown parameters 1 20x x , which are calculated

using the mathematical model, are listed in Table 2-2.

1 1 2 20 , ,

2 1 2 20 , ,

20 1 2 20 , ,

( , , . . ., ) 0

( , , . . ., ) 0

.

.

( , , . . ., ) 0

comp ref TEV in

comp ref TEV in

comp ref TEV in

g x Cooling power x W x h

g x Cooling power x W x h

g x Cooling power x W x h

(2.14)

24

1

1 11

1 20 11 1

120 2020 20 20

1 20

...( )

. . . .

. . . .

. . . .

( )...ni

nn ni

n n ni

x

g g

x x g xx x

g gx x g xx x

(2.15)

Table 2-2: Output parameters of the mathematical model

evapQ

(cooling

power)

compW

(compressor

power)

,f condW

(condenser fan

power)

,f evapW

(evaporator fan

power)

condQ

(heat rejection)

COP

(coefficient of

performance)

evapNTU

(number of

transfer units,

evap.)

condNTU

(number of

transfer units,

cond.)

evap

(effectiveness,

evap.)

cond

(effectiveness, cond.)

Tc

(condensing

temperature)

href,evap,in

(refrigerant

enthalpy @

evap. inlet)

refm

(refrigerant mass

flow rate)

Ta, evap, out

(air temperature

@ evap. outlet)

hg @ Te

(saturated gas

enthalpy @

evaporating

temperature)

href,comp,in

(refrigerant

enthalpy @

comp. inlet)

h ref,cond,in

(refrigerant

enthalpy @

cond. inlet)

hg @ Tc

(saturated gas

enthalpy @

condensing

temperature)

hf @ Tc

(saturated liquid

enthalpy @

condensing

temperature)

h ref,TEV,in

(refrigerant enthalpy

@ expansion valve

inlet)

25

A Newton-Raphson method is employed to develop an iterative numerical code

in C for solving the set of nonlinear correlations presented in Eq. (2.14) [76]. In this

method, the derivatives of all the equations are calculated with respect to all the

unknowns and a Jacobian matrix is formed. During an iterative solution procedure, all

the unknowns are calculated at iteration (n) and used for the next iteration (n+1), see Eq.

(2.15). The iterative procedure is repeated until the maximum relative residual for all the

unknown parameters becomes less than 0.001.

2.2. Quasi-Steady State Model Development

The A/C-R systems undergo a wide range of imposed operating conditions

during the duty cycle. Among all the variables described in the previous section, the

following impose the operational conditions on an existing A/C-R system:

o Evaporator inlet air enthalpy , dependent on both temperature and

humidity

o Condenser inlet air enthalpy , dependent on both temperature and

humidity

o Evaporator air flow rate , dependent on evaporator fan feed power

o Condenser air flow rate ,dependent on condenser fan feed power

o Compressor feed power (on-off or speed/capacity change command)

o Expansion valve orifice area

The rest of the parameters that appear in the described model equations will vary

as the system respond to the imposed conditions. Thus, the developed steady state

model is used to find the response of the system for any imposed operating conditions

and eventually obtain the performance of the system. In real-time, any A/C-R system

experiences a narrow or wide range of imposed operating conditions, depending upon

the application. The imposed operating conditions may change due to one or a

combination of the following reasons:

, ,a evap inh

, ,a cond inh

,fan evapW

,fan condW

26

o Ambient temperature/humidity variations

o Conditioned space cooling demand variations

o Controller commands (usually an on-off command from a thermostatic controller)

Ambient temperature and humidity continuously vary and affect the operation of

almost all components of A/C-R systems. The conditioned space cooling demand varies

due to peripheral heat transfer variations as well as indoor heat generation changes.

Cooling demand variations can happen gradually or suddenly depending upon the

application. For conditioned spaces with either a high rate of door/window

opening/closing or a repetitive loading/unloading of thermal mass, the variations in

cooling demand can be high; however, for most applications the variations are not

significant and do not suddenly affect the A/C-R system, due to the thermal mass of the

conditioned space. In addition to the variations in the ambient or conditioned space

temperatures and cooling demand, all A/C-R systems are equipped with a controller that

commands the components to turn on/off or change their capacity. Accordingly, the

controller command is another imposed parameter that affects the operational

characteristics of an A/C-R system. Due to the existence of such variations in the

operating condition of an A/C-R system, a transient model is required to simulate the

behavior of the system during unsteady conditions. Accordingly, in this research the

transient model is also developed.

In the quasi-steady state approach for modeling A/C-R systems, it is assumed

that at any instant the A/C-R system is at a quasi-equilibrium condition that can be

considered as a steady state condition [77,78]. By this assumption, the modeling of real-

time operation can be condensed into a time-marching steady state approach. Based on

this approach, modeling of the transient behavior of A/C-R systems is divided into two

steps: 1) quasi-steady state modeling, 2) fully transient modeling. The main advantage

of quasi-steady state modeling approach is its lower computational costs with acceptable

accuracy for most of the smooth transitions taking place in reality.

In the quasi-steady state modeling approach, if the rate of variations in the

imposed operating condition of an A/C-R system is small, the quasi-equilibrium

27

assumption will be accurate. The literature review and the experimental data show that

in most VACR systems the ambient temperature and cooling demand vary smoothly with

time. However, the changes due to controller on-off commands are drastic. Since

common VACR systems are thermostatically controlled, the compressor receives an on-

off command based on the air temperature inside the vehicle cabin. Therefore, the

quasi-steady modeling may not accurately predict the behavior of a VACR system

around the time that an on or off command is received.

To evaluate the accuracy of a quasi-steady state model obtained based on the

steady state approach; a time-marching model is first developed in this part of the study.

The accuracy of the first version of the developed quasi-steady model is validated and

presented later in section 4.2. The results show that as expected, there is a relatively

large local discrepancy between the quasi-steady model and the experimental data

around the on-off regimes though the average values are in good agreement. Thus, it is

inferred that the quasi-steady model is appropriate for a time-averaged study of the A/C-

R system behavior; however, it lacks acceptable behavior prediction around the on-off

cycling regimes.

In order to increase the accuracy of the developed quasi-steady model around

the on-off regimes and any other abrupt changes, the behavior of all the system

components has been investigated for a wide range of operating conditions. From this

analysis the response time, the time required to reach steady-state conditions after an

abrupt change, is determined for all of the components. It is determined that:

o The response time of the compressor, fans, and expansion valve is relatively

short: around 10-30 seconds depending upon the operating conditions.

o The response time of the heat exchangers (evaporator and condenser) is

relatively long: around 200-400 seconds depending upon the operating

conditions.

The relatively long response time of the heat exchangers to the abrupt variations

is due to the high thermal inertia of these components. For instance, when the

compressor is turned off, there is still a remarkable amount of refrigerant inside the

28

evaporator that can be evaporated and provide cooling. In addition, when the

compressor starts working again, it takes time for the evaporator to reach the steady

state capacity. A similar behavior exists for the condenser. Therefore, the condensing

and evaporating temperatures and the corresponding pressures reach the magnitude of

steady state condition after a relatively large time interval. Meanwhile, the compressor

and fans reach their specified speeds in a relatively shorter time. Thus, the developed

quasi-steady state model should be modified to capture the effects of thermal inertia

during the transition time.

In Figure 2-2, an overall schematic of a heat exchanger in an A/C-R system is

presented. In the evaporator, the refrigerant absorbs heat and cools down the air

stream. Through the condenser, the refrigerant rejects heat to the air stream. In a typical

A/C-R system, after a signal from the controller to the compressor or fans that can be an

on-off or velocity change command, the refrigerant or air flow rate change rapidly.

However, the rate of heat transfer between the refrigerant and air stream does not

change that quickly because of thermal inertia inside the heat exchanger. In order to add

the effect of thermal inertia to the quasi-steady model, the transient heat transfer

correlation between the refrigerant and air streams is used as presented in Eq. (2.16). It

is assumed that the changes in potential and kinetic energy of the fluid flows through the

heat exchanger are negligible [13,14,73].

Figure 2-2: An integral overview of a heat exchanger (evaporator or condenser) in A/C-R systems

29

, , , , , , ,

heat exchangerin out

ref in ref in ref out ref out a in a in a out a

dEE E

dtd mh

m h m h m h m hdt

(2.16)

Thus, the effect of thermal inertia is taken into account in the quasi-steady model

by adding Eq. (2.16) to each of the condenser and evaporator sub-models. The results

of the modified quasi-steady model are presented later in section 4.2. The comparison

between the model outputs and the experimental data shows a significant improvement

of the quasi-steady model accuracy after the last modification. However, it should be

noted that the included thermal inertia effect comes from the refrigerant only and effects

of thermal inertia of the heat exchanger walls is neglected here to lower the

computational cost of the quasi-steady model. In the next section, the effects of thermal

inertia will be considered in more detail to develop the transient model.

2.3. Transient Model Development

As the final step of model development for the A/C-R systems, a fully transient

model is developed in this section. This model includes all the transient terms of the

energy equation. The aim of developing a fully transient model is to increase the

accuracy of the mathematical model predictions beyond the modified quasi-steady

model and employ it for investigations of the transient thermal and performance

characteristics.

2.3.1. Condenser sub-model

Conservation of mass, momentum, and energy form a complete mathematical

model for the condenser of an A/C-R system. In general, three different refrigerant

regimes including: superheated, 2- phase, and sub-cooled may exist in a condenser;

30

thus, the governing equations are written separately for each of these regimes.

Following [79,80] a schematic representative of the heat transfer regimes in a condenser

is drawn in Figure 2-3.

Figure 2-3: Schematic representative of a cross flow condenser in VCR systems (sections from right to left: superheated, 2-phase, sub-cooled)

- Conservation of mass for refrigerant in superheated section:

, , ,2, , , , 1 , , , ,4

ref cond s cond sref cond in ref cond mid i cond cond s ref cond s

Lm m D L

t t

(2.17)

- Conservation of mass for refrigerant in 2-phase section:

, ,2 ,22, , 1 , , 2 , ,2 , ,24

ref cond ph cond phref cond mid ref cond mid i cond cond ph ref cond ph

Lm m D L

t t

(2.18)

- Conservation of mass for refrigerant in sub-cooled (liquid) section:

31

, , ,2, , 2 , , , , , ,4

ref cond l cond lref cond mid ref cond out i cond cond l ref cond l

Lm m D L

t t

(2.19)

- Conservation of momentum for refrigerant (fully-developed flow):

, , , ,, , , ,

2 2, , , , , ,

2, ,

4 4

4

ref cond in ref cond outcondref cond in ref cond out

ref cond in i cond ref cond out i cond

i cond cond tot

mu m mm m

t D D

PD L Viscous losses

x

(2.20)

- Conservation of energy for refrigerant in superheated section:

, , 1 , , , , , , , ,, ,

2,

, , , ,,, , , , , , , , , ,

4

ref cond mid g i cond cond s ref cond s ref cond s w cond sref cond in

i cond

ref cond s ref cond scond scond s ref cond s ref cond s ref cond s ref cond s cond s

mh m h D L T T

D

hLL h h L

t t t

(2.21)

where, the convective heat transfer coefficient , ,ref cond s is calculable using Dittus-Boelter

correlation [73,81]:

, , , 0.8 0.4

, ,

0.023Re Prref cond s i cond

ref cond s

DNu

k

(2.22)

- Conservation of energy for refrigerant in 2-phase section:

, , 1 , , 2 , ,2 , ,2 , ,2

, ,2 ,22, ,2 , ,2

( )( )

4

ref cond mid g ref cond mid l i cond cond ph ref cond ph c w cond ph

ref cond ph cond phi cond cond ph ref cond ph

m h m h D L T T

h LD L h

t t

(2.23)

The following correlations also govern the 2-phase region [82]:

32

, ,2( ) (1 )ref cond ph g g l l condh h h (2.24)

2

3

1

11 g

l

XX

(2.25)

The convective heat transfer coefficient , ,2ref cond ph is calculated from Kandlikar

correlation [83,84]:

0.040.760.8,0.8 0.4

, ,2 , , 0.38,

3.8 10.023Re Pr 1l cond

ref cond ph l cond l condi cond c

cr

k X XX

D PP

(2.26)

- Conservation of energy for refrigerant in sub-cooled (liquid) section:

, , 2 , , , , , , , ,, ,

2,

, , , ,,, , , , , , , , , ,

4

ref cond mid l i cond cond l ref cond l ref cond l w cond lref cond out

i cond

ref cond l ref cond lcond lcond l ref cond l ref cond l ref cond l ref cond l cond l

m h mh D L T T

D

hLL h h L

t t t

(2.27)

- Conservation of energy for wall in superheated section:

, , , , , , , , , , , ,

, , ,

, , ,2 2, , , , ,4

i cond cond s ref cond s ref cond s w cond s a cond o cond cond s fin fin s

w cond s a cond

w cond s cond so cond i cond cond s w cond sw w

D L T T D L A

T T

T LD D L Cp Cp T

t t

(2.28)

- Conservation of energy for wall in 2-phase section:

33

, ,2 , ,2 , ,2 , , ,2 ,2

, ,2 ,

, ,2 ,22 2, , ,2 , ,24

i cond cond ph ref cond ph c w cond ph a cond o cond cond ph fin fin ph

w cond ph a cond

w cond ph cond pho cond i cond cond ph w cond phw w

D L T T D L A

T T

T LD D L Cp Cp T

t t

(2.29)

- Conservation of energy for wall in sub-cooled (liquid) section:

, , , , , , , , , , , ,

, , ,

, , ,2 2, , , , ,4

i cond cond l ref cond l ref cond l w cond l a cond o cond cond l fin fin l

w cond l a cond

w cond l cond lo cond i cond cond l w cond lw w

D L T T D L A

T T

T LD D L Cp Cp T

t t

(2.30)

- Conservation of energy for air:

, , , , , , , , , ,a cond a cond in a cond out a cond o cond cond tot fin fin a cond w condm h h D L A T T (2.31)

where:

, , , , ,2 ,2 , , ,,

,

w cond s cond s w cond ph cond ph w cond l cond lw cond

cond tot

T L T L T LT

L

(2.32)

2.3.2. Evaporator sub-model

Similar to condenser, writing the conservation of mass, momentum, and energy

forms a complete model for evaporator in an A/C-R system. It should be noted that there

are only two sections in an evaporator: 1) 2-phase, and 2) superheated.

34

2.3.3. Compressor sub-model

For a compressor in A/C-R systems, the refrigerant mass flow rate can be

obtained from the following correlation [85]:

, , ,ref comp ref comp in vm NV (2.33)

where, volumetric efficiency v is a function of evaporating and condensing pressures.

Also, the power input to compressor can be calculated using the following equation:

, , , , ,ref comp ref comp out ref comp in

compE M

m h hW

(2.34)

2.3.4. Expansion valve

Assuming an isenthalpic process, the following correlations can be used for

modelling the expansion valve behavior [81]:

, , ,ref TEV d ref TEV in c em C A P P (2.35)

, , , ,ref TEV in ref TEV outh h (2.36)

2.3.5. Unknowns of the model

All the time-gradient terms appearing in the equations are discretized using the

finite difference approach. For an arbitrary variable A:

1n nA AA

t t

(2.37)

Accordingly, the transient solution is commenced with an initial condition @ 0n and

marches through time by solving all the equations at each time step. The model can be

35

written as a set of 26 nonlinear equations with 26 unknowns as presented in Table 2-3.

The calculated values from each time step are used as the initial condition for the next

time step. A Newton-Raphson based solver is developed in C to numerically solve the

set of equations.

Table 2-3: Calculated parameters at each time step

, ,ref cond inm

, , ,,ref evap out ref compm m

, ,ref cond outm

, , ,,ref evap in ref TEVm m

, ,ref cond inh

, ,ref comp outh

, ,ref cond outh

, ,ref TEV inh

, ,ref evap midm ,cond sL , , 1ref cond midm , , 2ref cond midm

,2evap phL ,evap sL ,2cond phL ,cond lL

, ,ref evap inh

, ,ref TEV outh

, ,ref evap outh

, ,ref comp inh

eP

eT

cP

cT

, ,w cond sT , ,2w cond phT , ,w cond lT , ,w evap sT

, ,2w evap phT , ,a cond outT , ,a evap outT cond

evap compW

2.4. Conclusion

In this chapter, mathematical models were developed for investigation of the

thermal and performance characteristics of VACR systems. A steady state model was

first developed using a map-based technique for the compressor, an NTU method

for the heat exchangers (evaporator and condenser), and an isenthalpic approach for

the thermostatic expansion valve. A numerical solver was developed for simultaneous

solution of the set of nonlinear coupled equations using the Newton-Raphson method. A

quasi-steady state model was developed based on the steady state model to simulate

the VACR systems’ characteristics under transient operating condition with relatively low

36

computational costs. Finally, a transient model was developed to investigate the

dynamic behavior of VACR systems under realistic variable conditions. The transient

model was derived employing the laws of mass, momentum, and energy conservation,

resulting in a set of 26 nonlinear equations that need to be solved numerically to find the

26 unknowns of the VACR system. Model validation is described in Chapter 4.

37

Chapter 3. Experimental Study and Data Acquisition

Obtaining real-time data to verify the accuracy of the mathematical models in a

wide range of operating condition requires an experimental study. Moreover, an

experimental setup should be prepared to implement and test the performance of the

developed proactive controller. For these reasons a testbed is built in the Laboratory for

Alternative Energy Conversion (LAEC) at Simon Fraser University. Furthermore, real-

time field data is collected from stationary and mobile A/C-R units to obtain realistic duty

cycles. Additionally, a new high efficiency and variable capacity VACR system is built

using high efficiency components from well-known manufacturers. In this chapter, first

the development of testbed and then the specifications of high efficiency VACR system

are presented. Thereafter, the field data acquisition from typical stationary A/C-R as well

as VACR units is discussed.

3.1. Testbed Design and Construction

A new testbed was designed and built in LAEC to simulate the real-time

operating condition of a typical VACR system. The employed VACR system in this

testbed is a battery-powered anti-idling system widely installed on trucks and vans.

Figure 3-1 shows a sample of this battery-powered anti-idling VACR installed on the

sleeper berth of a long-haul truck to avoid idling of the truck engine for air conditioning

purposes while the truck is not moving.

38

Figure 3-1: Typical battery-powered anti-idling VACR system employed in the testbed: (a) condensing unit and connections on the rear wall of the truck cabin, (b) evaporating unit inside the truck cabin, (c) batteries

The system is formed from a condensing unit, an evaporating unit, a battery box,

and refrigerant pipelines. The condensing unit, which is installed outside on the rear wall

of the truck cabin, consists of a scroll compressor model YWXD15 with 15 cm3/rev

displacement manufactured by Nanjing Yinmao Compressor Co., a 12 volt axial fan with

0.236 m3/s nominal flow rate, a 2 row 14 fin/inch aluminum finned-tube coil, and a power

control unit. The evaporating unit, which is installed inside the truck cabin, is built using a

3 row 14 fin/inch aluminum finned-tube coil, a centrifugal fan with 0.038 m3/s nominal

flow rate, a thermostatic expansion valve (TEV), and manual temperature set point and

air speed controls. The system is powered by a package of two 12 volt batteries directly

charged by the engine alternator. Normally, 0.79 kg of HFC-134a is charged into the

system as the refrigerant. Since the evaporating and condensing units of the A/C-R

system are installed in separate spaces (outdoor and indoor), the testbed should include

two independent sources of conditioned air supply to impose any desired temperature

39

and humidity at the air inlet of both the evaporator and the condenser. Environmental

chambers can provide air at a wide range of temperature and humidity. Accordingly, for

the testbed two separate environmental chambers are required; one connected to the

evaporator and the other one connected to the condenser. A schematic of the testbed is

presented in Figure 3-2. The schematic shows that a number of temperature, pressure,

and relative humidity sensors as well as velocity and mass flow meters are employed in

the testbed to obtain the required data.

Figure 3-2: Experimental setup schematic (solid lines = air flow; dashed lines = refrigerant flow; T = temperature, P = pressure, RH = relative humidity sensor; M = mass flow rate, V = velocity meter)

A photo of the first testbed is shown in Figure 3-3. The testbed consists of the

anti-idling VACR system installed on a portable base with components connecting it to

the environmental chambers. Booster fans are installed to compensate for pressure drop

in the ducting system and provide nominal air flow rates for both the condenser and

evaporator coils. T-type thermocouples from OMEGA with ±0.5 oC accuracy are installed

at 4 locations on the evaporator and condenser air streams. Moreover, two wind sensors

model MD0550 with accuracy of ±0.15 m/s from Modern Device are installed to measure

the velocity for both the evaporator and condenser air flows. Additionally, two Rotronic

humidity sensors model HC2-S3 with ±0.8% RH accuracy are installed to measure

40

humidity at the inlet and outlet of the evaporator air stream. T-type thermocouples (from

OMEGA with ± 0.5 oC accuracy) and pressure transducers model PX309 manufactured

by OMEGA with 0.25% accuracy are installed at four locations on the VACR unit to

measure both the high and low temperatures/pressures of the refrigerant. A Micro

Motion 2400S transmitter with ± 0.0001 m3.s-1 accuracy from Emerson Electric Co. is

used to log the refrigerant mass flow rate. Three Chroma programmable DC power

supplies model 62012P-40-120 are employed to feed and measure the power

consumption of the compressor and fans. A National Instruments Data Acquisition

System (DAQ) model 9184 is used to collect the input data from the thermocouples,

pressure transducers, refrigerant flow meter, velocity meters, and power supplies, and

send them to a computer. LabVIEW software is employed to obtain all the measured

data from equipment and save them into a file. The measurement rate is 60 samples/s.

Figure 3-3: First version of the developed experimental setup

41

Figure 3-4: Current experimental setup configuration

42

3.2. Development of a Proof-of-Concept Demonstration of a High Efficiency VACR System

Based on the literature review, the COP of VACR systems is generally low. The

experimental and theoretical evaluations of this study also show a relatively low COP for

the current battery-powered anti-idling VACR system (to be discussed later).

Accordingly, the aim is to build a proof-of-concept demonstration of a high efficiency

VACR system equipped with a variable speed compressor and variable speed fans that

operates with the optimization-based proactive controller.

A comprehensive investigation into the available high efficiency two-phase heat

exchangers and variable speed compressor and fans in the market is the first step

towards building the high efficiency system. As such, a high efficiency fan (variable

speed), a high efficiency compressor (variable speed), an evaporator coil, and a

condenser coil are purchased from reputable manufacturers with proven high efficiency

products and the new VACR system is assembled. The new high efficiency VACR

system is then equipped with a proactive optimization-based controller to represent a

proof-of-concept demonstration out of this research. Figure 3-5 shows the new

components used to build this system. The shown fan is connected to the evaporator

side. It should be mentioned that another variable speed fan exists in the LAEC, which is

connected to a wind tunnel, and is utilized for the condenser side of the system. The

variable speed compressor is model “NLV8.4F SECOP” manufactured by Danfoss. Also,

the evaporator and condenser coils are purchased from Madok, to achieve high heat

transfer effectiveness.

43

Figure 3-5: Purchased components for building the proof-of-concept demonstration of a high efficiency VACR system (top left: variable speed fan, top right: condenser coil, bottom left: variable speed compressor, bottom right: evaporator coil)

3.3. Field Data Collection, Stationary A/C-R System

In addition to the testbed development in LAEC for the experimental study, real-

time field data is collected from both stationary A/C-R and VACR systems to have a

more comprehensive understanding of the thermal and performance behavior of these

44

systems during their duty cycle. Preparing the required measuring equipment for the

real-time study of the VACR systems, obtaining the experience of installing and running

the measuring equipment, defining the duty cycle, analyzing the obtained data during the

duty cycle, and validating the developed mathematical model were the main goals of this

part of the study on a stationary A/C-R system. As a starting step, a nearby walk-in

freezer is selected as a pilot stationary A/C-R system on which to perform the real-time

study. Walk-in refrigerators and freezers are widely used for keeping the temperature

and humidity of food products in a safe range during a storage period. Studies have

shown that the world’s demand for commercial refrigeration equipment, including: reach-

in and walk-in coolers, freezers and display cabinets, has increased 4.6% per year

during 2008-2011 [86]. Based on the existing data, refrigeration is the source of 35- 50%

of the total energy consumption in the commercial food sales industry [87–91]. Due to

the significant share of energy consumption by the commercial refrigeration units, the

results obtained from this part of the study can also be employed in future research

around energy conservation in such systems.

The selected walk-in freezer is located in the Central City Pub and Brewery in

Surrey, BC, Canada. Figure 3-6 shows the selected freezer room and its condensing

unit installed on the roof of the room. The compressor of the selected system is model

“CS14K6E-PFV” manufactured by Copeland working with a nominal capacity of 2.3 kW.

HFC-507 is used as the refrigerant in this system. The condenser and evaporator of the

system are manufactured by Emerson Electric Co., and Cancoil Thermal Corporation,

respectively. The installed condensing unit is 10 years old.

45

Figure 3-6: The selected freezer room (left), its condensing unit (mid), dimensions (right)

Measuring equipment is installed on the system to collect the required data, including:

refrigerant temperatures and pressure, air temperature and humidity, refrigerant and air

mass flow rates, and input power to the compressor and fans. T-type thermocouples,

Rotronic humidity sensors (model HC2-S3), pressure transducers (model PX309), and a

data acquisition system (DAQ) is used in this part of the study. Furthermore, an

ultrasonic flow-meter model FLUXUS F601 manufactured by FLEXIM is used to

measure the refrigerant velocity inside the pipes noninvasively with accuracy of ±0.01

m/s. The air flow rate of the condenser and evaporator are measured using an

anemometer model 5725 manufactured by TSI. Additionally, a digital clamp meter model

R5020 manufactured by REED is installed to collect the input power to the compressor

and fans. Figure 3-7 shows some of the installed equipment and the location of

thermocouples installed on the evaporating unit is shown in Figure 3-8.

46

Figure 3-7: Installation of thermocouples, pressure transducers, and flow meters on the condensing unit

The figure shows (a) the entire condensing and compressor unit, (b) piping before the condenser, (c) piping after the condenser, (d) pressure transducer installed before and after the compressor, (e) air inlet side of the condenser heat exchanger, (f) condenser tubing passes, (g) flow meter installation on the piping after the condenser, and (h) air outlet side of the condenser fan. Red dots point at the locations of thermocouple installation. In some locations, more than one thermocouple is used for measurement, the average value of which is incorporated in the study.

47

Figure 3-8: Installation of thermocouples on the evaporating unit

The figure shows (a) the evaporator fans and (b) the piping behind the evaporator system. Red dots point at the locations of thermocouple installation. In some locations, more than one thermocouple is used for measurement, the average value of which is incorporated in the study.

3.3.1. Data collection and duty cycle definition

After installing the equipment, data has been collected onsite for three months.

Real-time behavior of the refrigeration system during different operating hours is

investigated. The collected data provides an in-depth knowledge and a wealth of

detailed information on the operation and the performance of the selected freezer room.

The refrigeration system is then simulated using the developed steady state model

under various duty cycles. The simulation outputs are utilized for the purpose of model

accuracy validation which is presented later. In Figure 3-9 and Figure 3-10 samples of

the collected temperature data for the evaporating and condensing units during

weekdays are illustrated. Moreover, Figure 3-11 shows a sample of the measured input

electrical power to the condensing unit. These plots reflect different types of fluctuations

in the temperature and power as a result of the varying operating conditions, cooling

demand, and control commands.

48

Figure 3-9: Sample measured temperature during weekdays - evaporating unit

Figure 3-10: Sample measured temperature during weekdays - condensing unit

49

Figure 3-11: Sample measured input electric power to the condensing unit

The following can be concluded from the real-time data about the behavior of the

pilot stationary refrigeration system:

o Frost is continuously formed on the evaporator coil of the system due to

the existence of humidity in the freezer room. Therefore, a defrosting

cycle is employed in the system which operates every 6 hours to melt

down the ice that has formed on the evaporator coils. During the

defrosting cycle, the compressor shuts off, and a powerful electrical

heater (~1,600 W) is switched on. As a result of this heat gain, the

refrigerant temperature in the evaporator coils increases, which degrades

the efficiency of the system remarkably. It was noticed that the defrost

process is not really necessary every 6 hours. A more efficient option for

this refrigeration unit could be to replace the timer-based defroster with a

humidity-based defroster that switches on only when necessary. This

replacement could contribute to an overall energy efficiency improvement

for the studied freezer room.

o Irregular compressor on-off cycling due to the freezer room’s thermostat

signals is observed. During night hours, because food is no longer being

50

loaded/unloaded and the door stays closed, the compressor on-off cycles

are fairly regular with a frequency of almost 26 minutes. This is a result of

poor insulation and insulation material degradation over time. In addition,

it is observed that an incandescent light (~100 W) and a door-sealing

heater (~200 W) are continuously turned on. These additional heat loads

could be eliminated by using a more efficient light (e.g. LED) equipped

with motion sensor and a controlled discontinuous heater for the door

seal.

o In spite of employing a timer-based defrosting system, sometimes the

formed frost is not completely cleared from the evaporator coil; see

Figure 3-9. The frost present on the evaporator coil intensely affects the

behavior of the refrigeration system. This irregular long-lasting frost

presence can be as a result of food replenishment or holding the door

open for long periods of time, which result in a significant addition of

moisture into the room.

o The temperature and power input data indicate that the behavior of the

refrigeration system during the day and night hours are different as a

result of door opening/closing, food loading/unloading, and human

activities inside the freezer room.

Discovering the disparate behavior of the selected refrigeration system during the

night and day hours as well as different weekend-weekday operating hours of the

selected restaurant, distinctive duty cycles can be defined for the condensing unit of the

freezer room. The restaurant working hours during the weekdays and weekends are

11:00-23:00 and 11:00-23:59, respectively. As such, four duty cycles are defined in

Table 3-1 and a separate performance evaluation is conducted for each of them using

the mathematical model. Thus, based on the definition of the four duty cycles and due to

the disparate temperature and power behaviors through each cycle, the average values

of the measured parameters for each cycle are obtained and used for modeling

51

purposes. The results of the modeling are presented and compared to the measured

data later in section 4.1.2.

Table 3-1: Duty cycle definition for the stationary refrigeration system

Duty cycle Description Days Start hour End hour

DT1 Weekends work-hours Saturday-Sunday 10 AM 12 AM

DT2 Weekends off-hours Saturday-Sunday 12 AM 10 AM

DT3 Weekdays work-hours Monday-Friday 11 AM 11 PM

DT4 Weekdays off-hours Monday-Friday 11 PM 11 AM

3.3.2. Effect of on-off cycling on power consumption

Investigating the obtained data shows that the on-off cycling of the refrigeration

system due to thermostat commands has a significant impact on the instantaneous

power draw of the system. Also, the literature review indicates that the instantaneous

start-up power draw of electrical compressors is higher than their nominal current [88].

Figure 3-12 shows the effect of system’s compressor start-up on the power draw. The

total input power previously presented in Figure 3-11 was obtained based on data

measurement with frequency of 0.2 Hz. To focus more precisely on the start-up effects,

the input power is measured again with the frequency of 1 kHz and the result for a

typical cycling of the compressor is presented in Figure 3-12. The results show that the

typical start-up power draw of the compressor is almost 6.5 times higher than the steady

state power draw. A similar behavior with less intensity exists for the electrical fans of

the refrigeration system (start-up power: 1.5-2 times the steady power). Based on the

integral of total power draw during different duty cycles, it is concluded that the start-up

of the studied refrigeration system adds an average of 7-8% to the total energy

consumption.

52

Figure 3-12: A typical power draw of the compressor after start-up

Electronically commutated magnet motors (ECM), which are capable of variable

speed operation, have higher efficiencies compared to the conventional constant speed

types at nominal and partial torques, respectively [88]. Analyzing the simulated and

measured cooling power indicates that the studied refrigeration system operates with

partial capacity during more than 65% of its duty cycle. Therefore, there is a remarkable

potential for energy saving that could be achieved if the constant speed compressor and

fans are replaced with variable speed types to reduce the amount of on-off cycling. This

concept will be evaluated later using the in-lab testbed.

3.4. Field Data Collection, VACR System

After obtaining the experience of installing the measuring equipment on a

stationary A/C-R system to collect the field data, pilot VACR systems are selected for the

real-time investigation. Saputo Inc., as the largest Canadian dairy processor, employs

53

numerous refrigerated trailers each day to deliver dairy products throughout the country.

As an industrial partner of the current research, Saputo granted access to their

refrigerated trailers for the purpose of field data collection. Employing this opportunity,

two of the available refrigerated trailers at the Saputo’s Burnaby Plant are selected in

this step. Figure 3-13 shows the selected refrigerated trailers for the field data collection.

A diesel engine, compressor and condenser are installed outdoors on the front wall of

the trailers and the evaporator is located inside. The compressor, condenser and

evaporator fans are directly driven by the diesel engine, which is fed by its own fuel tank.

Therefore, the diesel engine is working only if the refrigeration unit is turned on.

The employed VACR systems are equipped with thermostatic controllers to keep

the dairy products cold (4-5oC set point) during loading and delivery. The selected VACR

units are widely employed in nationwide and worldwide food transportation industries

and manufactured by Carrier Transicold, one of the world’s largest refrigeration systems

manufacturers. In refrigerated trailer #1, a 4 cylinder, 1.9 liter indirect injection diesel

engine with nominal power of 21.6 kW is employed [92]. The engine is directly fed by a

separate aluminum fuel tank with a capacity of 190 litres. R-404A is used as the

refrigerant in this unit. Based on the catalog from the manufacturer, the nominal cooling

capacity of the VACR unit is 9.38 and 14.95 kW at evaporator return air temperatures of

-18 and 2oC, respectively. Trailer #2 is equipped with a 4 cylinder, 2.2 liter direct injection

diesel engine with 18.5 kW nominal power individually fed from a 190 litre aluminum fuel

tank [93]. The corresponding VACR is charged with 7.3 kg of R-404A refrigerant. The

condenser and evaporator coils have the dimensions of 1.940 m × 2.176 m × 0.579 m

and 1.684 m × 1.149 m × 0.280 m , respectively. Based on the data from the

manufacturer, the VACR can deliver nominal cooling powers of 10.26 and 17.59 kW at

evaporator return air temperatures of -18 and 2oC, respectively.

54

(a) Trailer #1; condenser (left), evaporator (right)

(b) Trailer #2; condenser (left), evaporator (right)

Figure 3-13: Selected typical refrigerated trailers for field data acquisition

55

3.4.1. Measuring equipment installation and duty cycle definition

A number of T-type thermocouples and humidity data loggers as well as

anemometers are employed to obtain the real-time field data from the selected VACR

units and investigate the thermal and performance behavior. In Figure 3-14 and

Figure 3-15 samples of the main installed measuring equipment are shown.

Figure 3-14: Sample installed measuring equipment on the condensing unit, (a) air temperature and humidity at condenser inlet, (b) refrigerant temperature at compressor inlet and outlet

(a) (b)

56

Figure 3-15: Sample installed measuring equipment on the evaporating unit, (a) refrigerant temperature at evaporator coil inlet and outlet, (b) air temperature and humidity at evaporator outlet, (c) air temperature and humidity at evaporator inlet

The data collection took place during summer, spring, fall, and winter periods.

Based on the dairy product delivery schedule of the selected trailer and onsite

observations as well as the obtained data from the measuring equipment, the duty cycle

of the VACR systems are determined. The obtained duty cycles for the trailers #1 and #2

are presented in Figure 3-16 to Figure 3-19. The figures indicate that different duty

cycles govern the periods of Monday to Wednesday, Thursday to Saturday, and

Saturday to Monday for trailer #1. From Monday to Wednesday, the refrigeration unit is

turned on around 18:00 and after loading the dairy products into the compartment, the

trailer stays in the parking lot till the next day morning. Around 5:00 AM of the next day,

a truck (tractor) is connected to the trailer to deliver the products which ends around

(a)

(b) (c)

57

10:00 AM. Then, the refrigeration unit is switched off and the trailer is returned back into

the parking area. Therefore, the VACR unit of trailer #1 is “on” between 18:00 to 10:00

during Monday-Wednesday. For Thursday-Saturday, the duty cycle is similar to Monday-

Wednesday with two hours lag in starting the refrigeration unit and loading. During this

period, the refrigeration unit is switched on around 20:00 instead of 18:00 and the rest of

the operating schedule is the same as Monday-Wednesday. There is no delivery during

Sunday for trailer #1. Accordingly, the trailer is not loaded and the refrigeration unit is not

switched on during Saturday evening. Based on this procedure, there is another duty

cycle governing the Saturday-Monday period. Although the refrigeration unit is kept off

during the evening and overnight of Saturday, the Sunday loading for Monday morning’s

delivery is commenced in the morning instead of evening. Therefore, during the third

duty cycle, the refrigeration unit is turned on around 8:00 AM of Sunday, and kept on all

day and overnight. Then, the product delivery is started around 5:00 AM of Monday and

finished around 10:00 AM. The duty cycle of VACR system in trailer #2 is more or less

the same all the weekdays and weekends as represented in Figure 3-19. For this trailer,

the loading and delivery do not take place on Wednesday afternoons and Sunday

mornings. The obtained duty cycles are later employed through the next sections for a

meaningful data analysis.

Figur

Figur

re 3-16: Dut

re 3-17: Dut

ty cycle of t

ty cycle of t

the VACR u

the VACR u

58

unit trailer #

unit trailer #

#1 – Monday

#1 – Thursda

y to Wednes

ay to Saturd

sday

day

Figur

Figur

re 3-18: Dut

re 3-19: Dutde

ty cycle of t

ty cycle of telivery on W

the VACR u

the VACR uWednesday

59

unit trailer #

unit trailer #afternoons

#1 – Saturda

#2 – Weekdas and Sunda

ay to Monda

ays and Weay mornings

ay

eekends (nos)

o

60

3.4.2. Real-time data acquisition

The procedure of data logger installation and data acquisition has been

conducted for periods of several days during different seasons of years 2014 and 2015.

Figure 3-20 to Figure 3-23 represent samples of the acquired temperature data from the

measuring equipment during the cold seasons. The duty cycle intervals are added to

these plots for a better understanding of the system behavior. Based on the acquired

data, the following is concluded for the cold seasons of year:

o The ambient temperature (Lower Mainland, British Columbia, Canada) normally

varies during the day and night with above-zero and sub-zero magnitudes that

directly affect the refrigeration unit’s behavior during the duty cycle; see

Figure 3-20.

o Since the refrigeration unit is not working during the “cycle off” intervals, the air

temperature at evaporator inlet and outlet, which represent the air temperature

inside the trailer, follows the ambient temperature; see Figure 3-21.

o During the “cycle on” intervals (areas not colored in Figure 3-21 to Figure 3-23), a

few short-time compressor operations appear, which change both the refrigerant

and indoor air temperatures. However, due to low ambient temperature, the

demand for operation of the refrigeration unit to reduce the indoor temperature is

quite low. Accordingly, the refrigeration system remains off during most of the

duty cycle.

o The refrigeration unit is completely oversized for the winter period which is longer

than the summer period in Canada. This leads to low efficiency operation of the

unit for most of its duty cycle. Thus, replacing the unit with an adjustable capacity

system equipped with variable speed components could lead to a better

performance throughout the year. In addition, a heavy engine-driven VACR

system with weight of more than 1100 kg, based on manufacturer’s data [94,95],

is transported during most seasons of the year with very little use. This almost

unnecessary weight transport significantly contributes to fuel consumption of the

truck and leads to GHG emissions.

Figur

Figur

re 3-20: Sam20

re 3-21: Samtra

mple ambie014

mple air temailer # 1, Ja

ent air (cond

mperature vn 2014

61

denser inlet

variations at

t) temperatu

t evaporato

ure variatio

or inlet and

ns – Jan

outlet –

Figur

Figur

re 3-22: Samou

re 3-23: Sam#

mple refrigeutlet – traile

mple refrige1, Jan 2014

erant tempeer # 1, Jan 2

erant tempe4

62

erature varia2014

erature varia

ations at co

ations at ev

ompressor i

vaporating u

inlet and

unit – trailer

63

In Figure 3-24 to Figure 3-33, the acquired data from trailers during the warm

seasons (July 2014, Aug 2015) are plotted. The following can be concluded from these

results:

o The ambient temperature varies remarkably during day and night (see

Figure 3-24 and Figure 3-25) and that imposes variable cooling load and

operational conditions on the VACR system. Due to heat gain from trailer wall

and air infiltrations during food loading/unloading, the refrigeration unit frequently

turns on and off during “ON” periods to keep the indoor temperature close to a

pre-set desired temperature (~4oC). The loaded food products do not impose any

additional cooling loads on the VACR system because they have already been

cooled down to the desired temperature before being loaded. The refrigerant

temperature during day and night is a function of duty cycle and ambient

temperature variations.

o The frequency of VACR on-off cycling right after switching the system “ON” from

an “OFF” condition is higher than usual (see Figure 3-27, Figure 3-29, and

Figure 3-30). Especially, whenever the system is switched “ON” after a longer

“OFF” period, the intensity of on-off cycling is more significant. This higher rate of

fluctuations happens due to thermal inertia of the trailer compartment. During the

“OFF” periods, heat is stored throughout the thermal mass (walls) of the trailer.

As such, the VACR system must work harder to overcome the thermal inertia

and keep the temperature desirable, which leads to more intense fluctuations

and higher energy consumption.

o At the end of each “ON” period, which coincides with delivery hours, some

irregular fluctuations with larger amplitudes can be observed on the evaporator

air temperature graphs (see Figure 3-28, Figure 3-29, and Figure 3-31). This

chaotic behavior of the air temperature emerges due to frequent door opening for

unloading products, which causes ambient air infiltration into the trailer. The door

openings and high-amplitude temperature fluctuations during the delivery hours

not only impose extra cooling load on the VACR unit, but also destructively affect

the quality of food products.

o Loading the food products several hours earlier than the delivery leaves a loaded

trailer in the parking area exposed to ambient temperature and sometimes sun

Figur

radiation

trailer wa

(see Figu

Figure 3-2

heat gain

re 3-24: Sam20

that brings t

alls. The freq

ure 3-26 a

29) and mos

and the cor

mple ambie014

o significant

quent on-off

and Figure

st of weekda

rresponding

ent air (cond

64

t energy loss

f cycling of

3-32), spec

ays for traile

energy loss

denser inlet

ses due to in

the VCR u

cially Sund

er #2, indica

es.

t) temperatu

nefficient ins

unit during t

ays for tra

ates the eno

ure variatio

sulation of th

hese period

ailer #1 (se

ormity of suc

ns – Jul

he

ds

ee

ch

Figur

Fig

re 3-25: Sam20

gure 3-26: S

mple ambie015

Sample refri

ent air (cond

gerant temoutlet – tr

65

denser inlet

perature varailer # 1, M

t) temperatu

ariations atMon to Thu,

ure variatio

compressoJul 2014

ns – Aug

or inlet and

Figur

Fig

re 3-27: Samou

gure 3-28: S

mple refrigeutlet – traile

Sample air t

tr

erant tempeer # 1, Thu t

temperature

railer # 1, M

66

erature variao Sun, Jul 2

e variations

Mon to Thu,

ations at co2014

s at evapora

Jul 2014

ompressor i

ator inlet an

inlet and

nd outlet –

Figur

Figur

re 3-29: Samtra

re 3-30: Sam“O

mple air temailer # 1, Th

mple zoomeON”– trailer

mperature vhu to Sun, J

ed air temper # 1, Jul 201

67

variations atul 2014

erature vari14

t evaporato

iations at b

or inlet and

eginning of

outlet –

f a cycle

68

Figure 3-31: Sample zoomed air temperature variations at end of a cycle “ON”– trailer # 1, Jul 2014

Figure 3-32: Sample refrigerant temperature variations at compressor inlet and outlet – trailer # 2, Aug 2015

69

Figure 3-33: Sample air temperature variations at evaporator inlet and outlet – trailer # 2, Aug 2015

3.5. Conclusion

Experimental study and field data acquisition were discussed in this chapter. A

testbed was designed and built in LAEC using an existing VACR system. The testbed

was developed for experimentation and to validate the mathematical models, validation

of the models is presented in Chapter 4. In addition, the testbed was modified to act as a

platform for testing the proof-of-concept demonstration of a high efficiency VACR

system. The most important modification of the testbed was the replacement of all the

existing components of the VACR system with high efficiency ones, including a variable

speed compressor and variable speed fans, as well as high efficiency heat exchangers.

The field data acquisition was initiated on a pilot stationary A/C-R system to gain

experience before moving to the challenging mobile A/C-R systems, which was the

major goal of this project. The required data collecting equipment were determined,

purchased, and tested during the study of stationary system. The data was collected for

a period of three months and analyzed in this chapter. The duty cycle of the pilot

70

stationary A/C-R system was discussed and the thermal and performance behavior of

the system were investigated. Afterwards, field data collection was started on

refrigerated trailers employed by Saputo Inc. for daily refrigerated food products delivery.

The study was performed during different seasons of 2014 and 2015. The duty cycles of

the trailers were established and utilized for data interpretation. The thermal behavior of

the corresponding VACR systems were analyzed, which can fill the existing gap of

literature on these systems. The acquired data from the experimental setup and pilot

stationary A/C-R and VACR systems are used for model validation and performance

analysis in Chapter 4.

71

Chapter 4. Model Validation, Thermal and Performance Investigation

The steady-state, quasi-steady-state and transient mathematical models were

presented in Chapter 2. The developed models are employed in thermal and

performance investigations of the A/C-R systems. In addition, the models are utilized to

develop an optimization-based controller for the proof-of-concept demonstration of high

efficiency VACR. To employ the developed models for these purposes, the accuracy of

their simulations is validated using the obtained experimental and field data. In this

chapter, the model validation is shown and the models are used for thermal and

performance investigations of the stationary and mobile A/C-R systems.

4.1. Steady State Model Validation, Thermal and Performance Investigation

4.1.1. Results relevant to the developed testbed in LAEC

Utilizing the developed testbed in LAEC, the accuracy of the steady state model

is validated. To cover all the operating conditions of a real VACR system, the model

validation is carried out in a wide range of temperature, 20-60 oC for the condenser air

inlet temperature (ambient temperature) and 10-50 oC for the evaporator air inlet

temperature (the conditioned space temperature). The experiments show that it takes

around one hour to arrive at steady state conditions for each test, see Figure 4-1. After

achie

the p

Figur

expe

range

humid

cond

respe

cause

beha

the r

eving steady

urpose of m

re 4-1: Samsi

Figure 4-

rimental dat

e of evapora

dity of the in

enser air flo

ectively. The

es a gradu

vior can be

refrigeration

y state cond

model validat

mple steady de tempera

-2 shows a

ta for air te

ator inlet air t

nlet air is kep

ow rates ar

ese results

ual increase

explained b

cycle ,cT T

ditions, the a

ion and syst

state condiature

a compariso

mperature a

temperature

pt at 50% fo

re kept at t

show that e

e in the ev

based on th

eT . By incre

72

average valu

tem perform

ition achiev

on between

at the outlet

e , ,a evap inT . It s

r all the ,a evT

the nominal

even for a c

vaporator ai

he condensin

easing ,a condT

ues of each

ance analys

vement - me

the mathe

t of evapora

should be m

,vap in values.

l capacities

constant ,a eT

ir outlet tem

ng and evap

,d in , cT rises

Steady State

h parameter

sis.

easured refr

ematical mo

ator, , ,a evap ouT

mentioned tha

Also, the ev

of 0.04 an

,evap in , increa

mperature T

porating tem

s to keep

e 

are used fo

rigerant

odel and th

ut , for a wid

at the relativ

vaporator an

nd 0.25 kg/s

asing , ,a condT

, ,a evap outT . Th

mperatures o

the require

or

he

de

ve

nd

s,

in

is

of

ed

temp

The T

comp

low

corre

Figur

obtai

incre

,a evapT

from

erature grad

cT increase

pressor char

pressure (e

espondingly c

re 4-2: Validev

te

Figure 4-

ned experi

asing , ,a condT

,ap in . Increas

the condens

dient betwee

leads to a

racteristics a

evaporating

change.

dation of thvaporator ai

mperature (

-3 presents

mental data

in , eT incre

e in eT is du

sing pressur

en the refrig

higher con

as a turbo-m

pressure eP

e model (linir outlet tem

, ,( )a cond inT

a comparis

a for evap

ases and th

ue to increas

re, cP , enhan

73

gerant and a

ndensing pre

machine and

e ) and the

nes) with exmperature (

son betwee

porating tem

he rate of t

se in the ev

ncement due

air for a suffi

essure cP . T

d the expans

mass flow

xperimental

, ,( )a evap outT vs

n the math

mperature T

this increas

vaporating pr

e to the rise

icient heat r

Therefore, b

sion valve p

w rate of re

l data (symbs. condense

ematical mo

eT and sho

e is higher

ressure eP ,

in cT .

rejection rate

based on th

roperties, th

efrigerant w

bols), er air inlet

odel and th

ows that b

for a highe

which come

e.

he

he

will

he

by

er

es

Figur

total

that b

cT a

incre

reaso

refrig

same

the h

unde

an in

re 4-3: ValidEv

(T

Figure 4-

power cons

by increasing

and the re

asing , ,a evapT

on of mass f

erant increa

e speed, it p

higher densi

r a warmer

crease in ref

dation of thvaporating t

, , )a cond inT

4 to Figure

sumption, co

g , ,a cond inT fo

frigerant m

in leads to

flow rate incr

ase due to th

roduces a h

ty. Based o

ambient con

frigerant ma

e model (lintemperatur

4-7 also sh

ooling powe

or any const

ass flow ra

an increase

rease at hig

he higher cP

igher mass f

on this obse

ndition, whic

ass flow rate

74

nes) with exre ( )eT vs. co

how the beh

er, and COP

tant , ,a evap inT

ate refm . In

e in the ref

her , ,a cond inT

and eP . Be

flow rate of

erved behav

ch increases

.

xperimentalondenser a

havior of ref

P, respective

results in a

n addition,

frigerant ma

and , ,a evap inT

ecause the c

refrigerant a

vior, one ca

s both ,a conT

l data (symbair inlet tem

frigerant ma

ely. Figure

n increase t

for a cons

ass flow rat

is that the d

compressor

at larger cP a

n conclude

,nd in and ,a evT

bols), perature

ass flow rate

4-4 indicate

to both eT an

stant , ,a cond inT

te. The mai

density of th

operates at

and eP due t

that workin

,vap in , leads t

e,

es

nd

n ,

in

he

a

to

ng

to

75

Due to the increase in refrigerant mass flow rate as well as the pressure ratio

changes, the total power consumption increases with increasing , ,a cond inT ; see Figure 4-5.

The same trend exists for increases in , ,a evap inT at any constant , ,a cond inT . This behavior

indicates that in warmer areas, a refrigeration unit draws more full-load power than in

colder areas. Furthermore, as can be observed in Figure 4-6, by increasing , ,a cond inT for

a constant , ,a evap inT , the cooling capacity of the system decreases. By recalling the

behavior of evaporating temperature eT from Figure 4-3, the cooling capacity decrease

can be explained. This decrease is a result of air-refrigerant temperature difference

reduction due to increase in eT from increasing , ,a cond inT for a constant , ,a evap inT . However,

this figure shows that for any constant , ,a cond inT , increasing , ,a evap inT results in a higher

cooling capacity of the system. Therefore, increasing the ambient temperature (higher

, ,a cond inT and , ,a evap inT ) may increase or decrease the total cooling capacity of the system.

The thermal insulation of the conditioned space can be decisive for this behavior.

If the thermal insulation of the conditioned space prevents a large temperature rise

inside the space after a large increase in ambient temperature (increasing , ,a cond inT for

almost a constant , ,a evap inT ), the cooling capacity of the system will drop. However, if poor

insulation lets the indoor temperature increase with increases in ambient temperature,

the cooling provided by the system will increase though the total efficiency degrades due

to higher thermal losses. Furthermore, for any constant , ,a evap inT , the power consumption

increase and cooling capacity decrease of the system with increasing , ,a cond inT leads to a

COP decrease that can be observed in Figure 4-7. However, increasing , ,a evap inT for any

constant , ,a cond inT slightly increases the system COP. Based on this behavior, one can

conclude that increasing the thermostatic set points of the conditioned space (higher

, ,a evap inT ) for any ambient condition will result in a higher COP.

Figur

Figur

re 4-4: Validre

(T

re 4-5: Validpo

dation of thefrigerant m

, , )a cond inT

dation of thower consu

e model (linmass flow ra

e model (linumption vs.

76

nes) with exate ( )refm vs

nes) with excondenser

xperimentals. condense

xperimentalr air inlet tem

l data (symber air inlet te

l data (symbmperature (

bols), emperature

bols), total

, ,( )a cond inT

e

Figur

Figur

re 4-6: Validco

re 4-7: Validvs

dation of thooling powe

dation of ths. condense

e model (liner vs. conde

e model (liner air inlet te

77

nes) with exenser air in

nes) with exemperature

xperimentallet tempera

xperimentale , ,( )a cond inT

l data (symbature , ,( a cond iT

l data (symb

bols), )in

bols), COP

78

The maximum relative discrepancy between the model output and the measured

values is 10.6% emerging from cooling load results. For the rest of the presented

parameters, the maximum relative discrepancy is less than 10%. The main sources of

relative discrepancy between the model and experimental data come from either

measuring equipment (minor error – from experimental data) or sub-models (major error

– from model outputs) inaccuracies. Since there are a variety of characteristic

parameters existing in the sub-model correlations (dimensions of coils and fins, heat

transfer properties of coil/fin material, refrigerant thermodynamics properties, convective

heat transfer coefficients, etc.) that are affected by measurement or round-off errors and

impurities, the model results and experimental data cannot be exactly the same. In this

study, the characteristics of the heat exchangers given by the manufacturer especially

the material property are the main sources of relative difference between the developed

model and the experimental results. It should be mentioned that the uncertainty analysis

for all the experimental data is conducted. Based on this analysis, the uncertainty for the

temperature measurements is ±1oC, for refrigerant flow rate ±1% (relative), for power

consumption 1.5%, for cooling power ±0.5%, and for COP 2%. Employing the validated

model, the thermal and performance behavior of VACR system is investigated in the

following sections.

4.1.1.1. Effects of condenser air flow rate (condenser fan speed) on COP

To investigate the effects of condenser fan speed on the system performance, in

this section a wide range of condenser air flow rate ,a condm is modeled and the results

are presented. In addition to partial flow rates of the existing condenser fan (33%-100%

of the nominal fan capacity 0.25 kg/s), higher flow rates (100%-167%) are also simulated

to assess the possibility of performance improvement by replacing the existing fan with a

higher capacity one. The evaporator air inlet temperature is assumed to be kept at 20oC

and the evaporator fan is set at its nominal speed. Figure 4-8 to Figure 4-11 respectively

represent the variations of refrigerant flow rate, total system power consumption, cooling

power, and COP for the mentioned range of ,a condm based on the steady state

mathematical model. The results show that by increasing ,a condm for any constant

ambient temperature, the following can be inferred:

79

o The refrigerant mass flow rate refm decreases continuously, but with higher rate

at smaller ,a condm (less than 80% of the nominal capacity ~ 0.2 kg/s). This

decrease is due to the higher heat transfer rate at larger ,a condm , which reduces

the condensing temperature cT and the corresponding pressure cP . The cP

reduction results in a decrease in evaporating pressure eP . Accordingly, the total

refrigerant pressure in the cycle decreases and results in lower refrigerant

density, which leads to a lower refm for a constant speed compressor operation.

However, beyond almost 80% of the nominal ,a condm (~ 0.2 kg/s) the variation of

refm becomes less dramatic. This behavior is due to power consumption

increase beyond the mentioned ,a condm that increases the total heat rejection rate

at the condenser. The increase in heat rejection increases cT and negates the

effect of ,a condm enhancement on reducing cT and thus, the refm will no longer

remarkably reduce.

o The total power consumption first decreases to a point of minimum power due to

a reduction in the compressor power, then increases due to the increasing power

of the condenser fan. The reduction in compressor power appears due to the

decrease in both the refrigerant mass flow rate and the high pressure. However,

as the refrigerant flow rate variation becomes negligible after a ,a condm of 0.2

(almost 80% of the nominal ,a condm ), the compressor power consumption also

does not change remarkably. Meanwhile, power consumption of the condenser

fan continuously increases and results in the increase in the total power.

o The cooling power continuously increases, with higher rate at smaller ,a condm less

than 80% of the nominal capacity. This trend is more pronounced at higher

ambient temperature. A more significant gap between the cooling power curves

at higher ambient temperatures indicates a remarkable loss of cooling capacity

for the system in hot zones. The steeper increase of cooling power at smaller

,a condm is due to the greater reduction of eT (as a result of cT reduction), which

increases the refrigerant-air temperature difference at the evaporator. However,

o

Figur

after almo

contribute

The COP

cooling po

The decre

while the

power co

decrease

optimum

points of

installing

leads to a

VACR sy

controlled

the system

re 4-8: Refr

ost 80% of

e to changes

first increas

ower and de

ease of CO

power cons

nsumption i

. Based on

COP can b

optimum CO

and using a

a lower COP

stem with a

d based on

m significant

igerant flow

the nomina

s in eT and t

ses to a poin

ecrease in p

OP is becau

sumption is

ncrease tha

this behav

e found by

OP are conn

a constant s

P operation f

a variable sp

the ambient

tly.

w rate vs. co

80

al ,a condm , inc

hus, the coo

nt of maximu

power consu

use of the n

increasing.

an the coolin

vior, for eac

changing th

nected with a

speed fan fo

for most con

peed fan for

t temperatu

ondenser ai

creasing the

oling power w

um value, be

mption, and

negligible inc

. Therefore,

ng power gr

ch ambient

he speed of

a green line

or the conde

nditions. On

r the conde

re, can imp

ir flow rate

e ,a condm va

will no longe

ecause of th

d then starts

crease of c

due to a h

rowth, the C

temperature

f the conden

e in Fig. 4-11

enser of a V

the flip side

nser, even

rove the pe

,( )a condm

lue does no

er change.

he increase i

to decrease

cooling powe

higher rate o

COP starts t

e, a point o

nser fan (Th

1). Therefore

VACR system

e, equipping

if it is simp

erformance o

ot

in

e.

er

of

to

of

he

e,

m

a

ly

of

Figur

Figur

re 4-9: Tota

re 4-10: Coo

al power con

oling power

nsumption

r vs. conde

81

vs. conden

nser air flow

ser air flow

w rate ,( a com

w rate ,( a condm

)ond

)d

Figur

4.1.1

coolin

ambie

temp

assum

in the

assum

o

re 4-11: CO

1.2. Effects

Figure 4-

ng power, a

ent tempera

erature for a

med to be s

e last sectio

med for the

Both the

for any co

higher he

increase

pressure

P vs. conde

s of evapor

12 to Figure

and COP v

atures based

all of the cas

et at its nom

on, in this

,a evapm . The

power cons

onstant ,a com

eat transfer

in heat trans

will rise, wh

enser air flo

rator air flo

e 4-14 repre

versus evap

d on the stea

ses is assum

minal speed.

section a w

following ca

umption and

ond and ,a coT

r rates for

sfer rate at t

hich leads t

82

ow rate ,( a cm

ow rate (ev

esent the be

orator air fl

ady state m

med to be ke

Similar to th

wide range,

an be conclu

d cooling po

,ond in . The co

larger ,a evapm

the evaporat

to a higher

)cond

vaporator f

ehavior of to

flow rate am

odel results

ept at 20 oC

he condense

33%-167%

uded from th

ower increas

ooling powe

ap values. F

tor, the evap

refrigerant d

fan speed)

otal power c

,a evap for a w

. The evapo

and the con

er air flow ra

of nomina

he results:

se with incre

r increase e

Furthermore

porating tem

density. Acc

on COP

consumption

wide range o

orator air inle

ndenser fan

ate paramete

l capacity,

easing ,a evam

emerged from

e, due to a

mperature an

cordingly, th

n,

of

et

is

er

is

ap

m

an

nd

he

o

Figur

compress

in the pow

The incre

resulting

,a condm an

re 4-12: Tot

sor power co

wer consump

ease in coo

in continuou

nd , ,a cond inT . T

tal power co

onsumption

ption of the s

oling power

us COP enh

The rate of C

onsumption

83

will increas

system.

r dominates

ancement b

COP increas

n vs. evapor

se, which co

s the power

by increasing

se is higher a

rator air flo

ontributes to

r consumpti

g ,a evapm for

at smaller m

ow rate ,( a evm

the increas

ion increase

any constan

,a evapm .

)vap

se

e,

nt

Figur

Figur

re 4-13: Coo

re 4-14: CO

oling power

P vs. evapo

r vs. evapor

orator air flo

84

rator air flow

ow rate ,( a em

w rate ,( a evm

)evap

)vap

85

Comparing the effects of evaporator and condenser air flow rates on COP shows

that although the ,a condm can lead to an optimum COP for the VACR system, changing

,a evapm does not result in optimum values; however, ,a evapm plays an important role in the

magnitude of the system COP. In Figure 4-15 to Figure 4-17, the VACR system’s power

consumption, cooling power, and COP for a wide range of ,a condm and ,a evapm are

presented based on the modeling. The evaporator and condenser air inlet temperatures

( , , , ,,a evap in a cond inT T ) are set at 26.7 and 35 oC respectively, based on ANSI/AHRI standard

210/240-2008. In addition to the mentioned , , , ,,a evap in a cond inT T set points, the simulations

are carried out for other set points and the trends of the results are similar to the

presented ones. Based on Figure 4-15 to Figure 4-17, the following can be concluded:

o Power consumption first decreases to a point of minimum power (compressor

power reduction is predominant), then starts to increase (condenser fan power

increase is predominant).

o Cooling power increases with increasing ,a condm or ,a evapm ; however, the rate of

increase at smaller ,a condm or ,a evapm is more considerable. This behavior emerges

from the variations of eT , cT , and refrigerant mass flow rate.

o For a given , ,a evap inT and , ,a cond inT , which respectively represent the vehicle cabin

temperature and the ambient temperature, and any evaporator fan speed

adjustment by the truck driver, there is a point of optimum COP for the system

that can be achieved by setting the condenser fan speed. The points of optimum

COP are connected through a light-brown line in Fig. 4-17.

o The magnitude of the system’s optimum COP increases with increasing ,a evapm .

Although the rate of this increase is significant at smaller ,a evapm , it becomes

nearly negligible at larger ,a evapm . For the presented ANSI/AHRI 210/240-2008

condition, the maximum achievable COP is at 200% of nominal ,a evapm and 125%

of nominal ,a condm .

Figur

Figur

re 4-15: Tot(m

re 4-16: Coo(m

tal power co

, ,,a cond a evapm m

oling power

, ,,a cond a evapm m

onsumption)p

r vs. conde)p

86

n vs. conde

nser and ev

nser and ev

vaporator a

vaporator a

air flow rates

air flow rates

s

s

Figur

of ch

of the

cycle

only d

by ch

evapo

highe

comp

highe

and a

optim

the s

throu

comp

re 4-17: CO

To give a

anging the f

e system un

es are both f

difference is

hanging the

orating tem

er evaporat

pressor inlet

er for the ca

a larger incr

mal fans’ spe

ystem is lar

gh the com

pressor is hig

P vs. conde

better therm

fans’ speed,

nder the nom

from tests un

s the evapora

fans’ speed

peratures a

ing pressur

, and lower

ase of optim

rease of “en

eed (see the

rger. Further

pression ste

gher due to

enser and e

modynamic r

in Figure 4-

minal and o

nder the sam

ator and con

from the ori

nd the pres

re, which l

cycle pressu

al fans’ spe

nthalpy per u

e green dash

rmore, in sp

ep for the op

the higher r

87

evaporator a

representatio

-18 a compa

optimal fans’

me ANSI/AH

ndenser fan

iginal to the

ssure ratio o

eads to a

ure ratio, the

eed. As a re

unit mass” a

hed line cyc

pite of a lowe

ptimal cycle,

refrigerant flo

air flow rate

on of the sys

arison betwe

’ speeds is

HRI 210/240

speeds. Th

optimal spe

of the cycle

higher ref

e mass flow

esult of a hig

along the ev

cle), the prov

er “enthalpy

, the total po

ow rate. How

es ,( ,a condm m

stem behavi

een thermod

shown. The

-2008 condi

e compariso

eeds, the con

e have chan

frigerant de

rate of refrig

gher refriger

vaporator fo

vided coolin

y per unit ma

ower consum

wever, due t

, )a evapm

ior as a resu

dynamic cycl

e represente

itions and th

on shows tha

ndensing an

nged. Due t

ensity at th

gerant will b

rant flow rat

or the case o

ng capacity o

ass” increas

mption by th

to higher rat

ult

le

ed

he

at

nd

to

he

be

te

of

of

se

he

te

88

of cooling power increase than compressor power increase, the COP of the system is

11% higher for the case of optimal fans’ speed as presented in Figure 4-17.

Figure 4-18: Comparison of thermodynamic pressure-enthalpy cycle between nominal (red continuous lines) and optimal (green dashed lines) condenser and evaporator air flow rates (ANSI/AHRI 210/240-2008 air-side condition)

It can be concluded that for any vehicle’s cabin temperature set point ( , ,a evap inT )

and ambient temperature ( , ,a cond inT ), an optimum COP is achievable by changing the

evaporator and condenser fan speeds. It should be noted that the first VACR system

employed in the testbed (the results presented in this section) was equipped with

constant speed compressor. However, the high efficiency VACR system built in LAEC is

equipped with a variable speed compressor as well, which contributes more significantly

to COP optimization (the results are discussed in Chapter 6). As such, the COP of a fully

variant capacity VACR system can be optimized by developing an intelligent controller to

find the optimum compressor, condenser, and evaporator speeds for any operating

condition.

4.1

walk-

valida

includ

avera

duty

defin

simul

Base

range

durin

refrig

meas

and ±

Figur

.2. Resu

In additio

-in freezer r

ation. Figur

ding: cooling

aged values

cycles are p

ed in Sect

lation and m

ed on the res

e of 1.75-2.0

g the DT1

eration syst

surements s

±0.5% for the

re 4-19: Valwa

ults releva

n to the acq

room (discu

re 4-19 rep

g power, inp

of both the

presented in

tion 3.3.1.

measured pa

sults, the pro

05 kW for th

(weekend

tem varies in

hows an err

e cooling po

idation of talk-in freeze

ant to the

quired data f

ssed in Ch

presents thr

put power, a

model outpu

n this figure

The comp

rameters wi

ovided cooli

e different d

work-hours

n a range of

ror of ±0.5oC

ower.

he model (ser room

89

studied s

rom the test

apter 3) are

ree major p

and evapora

uts and the

. The duty c

arison show

th a relative

ng power by

duty cycles.

s). In addit

f 1.56-1.68

C for the tem

simulation)

stationary

tbed built in

e also used

parameters

ator outlet ai

real-time me

cycles (DT1

ws good a

e difference l

y the refrige

The maximu

tion, the to

kW. The un

mperature, ±1

results wit

A/C-R sys

LAEC, the d

d for steady

of the stu

r temperatu

easurement

– DT4) we

agreement

less than or

eration syste

um cooling p

otal input p

ncertainty an

1.5% for the

h the meas

stem

data from th

y-state mode

udied system

re. The time

s for differen

ere previous

between th

equal to 9%

em varies in

power occur

power to th

nalysis for th

e input powe

ured values

he

el

m

e-

nt

ly

he

%.

a

rs

he

he

er,

s,

contin

strate

these

the d

devia

comm

prese

duty

from

calcu

cond

prese

meas

Figur

day (

main

evapo

syste

A major

nuous perfo

egies to imp

e systems sh

design cond

ation from th

mercial refrig

ent COP of t

cycles. In ad

the manufac

ulated to find

ition. The re

ent COP va

sured COP i

re 4-20: Pre

The resul

(design con

ly i) wear a

orator coils.

em shows a

challenge fo

ormance de

prove the pe

hould first be

ition to find

e design con

geration sys

the studied

ddition to the

cturer’s data

d the deviatio

esults show

lues with m

s calculated

esent and de

ts indicate t

dition). The

and tear; ii)

A compari

a range of 2

or current A

gradation is

erformance o

e obtained.

d the aging-

ndition assis

stems in diff

system bas

e present C

a for the me

on of the cur

a good ag

maximum rela

as 2%.

esign COP

hat currently

reduction i

refrigerant

son betwee

29-39% deg

90

A/C-R syste

s inevitable

of existing A

The obtaine

-related dev

sts in finding

ferent applic

ed on mode

OP, the des

asured oper

rrent perform

reement be

ative differe

of the VCR

y the system

n COP can

leakage; o

en the desig

gradation, w

ems is aging

for all thes

A/C-R syste

ed COPs ca

viation. Eval

g potential en

cations [96]

eling and me

sign COP of

rating condit

mance of the

etween the m

ence of 12%

system for

m features a

be a resul

or iii) fouling

gn and curre

which signific

g. As a res

se systems.

ems, the pre

n be then co

uating the p

nergy saving

. Figure 4-2

easurements

f the system

tions. The d

e system fro

measured a

%. The unce

r different d

lower COP

lt of various

g on the co

ent COP of

cantly affec

sult of aging

. To develo

esent COP o

ompared wit

present CO

gs for existin

20 shows th

s for differen

is calculate

esign COP

om the desig

and simulate

ertainty of th

duty cycles

than the firs

s parameters

ondenser an

the selecte

ts the powe

g,

op

of

th

P

ng

he

nt

ed

is

gn

ed

he

st

s,

nd

ed

er

91

consumption of the system. It should be noted that an electric defroster is installed on

the evaporator coil of the system to melt down the formed ice repeatedly. Based on the

simulation and measurements, the A/C-R system (including its defroster) currently

consumes 16,300 kWh/yr, which is 4,600 kWh/yr more than a new system (brand new of

the same condensing unit). This significantly higher energy consumption indicates the

necessity of either replacing the current system with an efficient A/C-R system or

resolving the performance-related issues of it.

The on-site observations indicate that the system is relatively new (~only 10

years old); fairly well designed; and well maintained. From the calculated COP values,

one can conclude that even for such a relatively new refrigeration system a significant

degradation takes place over years. There are many refrigeration units running around

the world that are older than the studied system. Accordingly, a significant amount of

energy can be saved worldwide by replacing or improving the efficiency of old

commercial refrigeration units.

Based on the calculated relatively high energy consumption of this system and

the large efficiency degradation over its life cycle, implementing performance

improvement technologies can have significant results (see Ref. [87]). During the

measurements, it was observed that the timer-based defrosting system consumed a

large amount of energy and released a frequent and significant heat load into the freezer

room. Utilization of defrosting systems for any freezer room is inevitable due to

continuous ice formation on the evaporator coil, which significantly reduces the thermal

conductivity of the coil. However, timer-based defrosters (employed in the current A/C-R

system) are one of the inefficient types that can be replaced by humidity-based

defrosters. In the current freezer room, the compressor is shut off during the defrost

cycle (every 6 hours) and a 1,600 W heater is turned on for heating the surface of

evaporator coil to melt down the formed ice layer. Based on the on-site observations and

temperature/humidity measurements, more than half of the defrosting cycles, especially

during the restaurant’s “off-hours” duty cycles, are unnecessary. The measurements and

calculations show that more than 1,400 kWh/yr, equal to 9% of the total energy

consumption of the VCR system, can be saved if the timer-based defroster is replaced

by a humidity-based controlled type.

92

In addition, installing a heat recovery unit on the VCR system can have

significant impacts on the performance. Since the freezer room is installed in a

restaurant, heat rejection from the condenser of the A/C-R system can be used for many

purposes such as preheating the service water or fresh air supply to the air conditioning

system during winter. Based on the simulations, the extractable heat rejection from the

condenser of system is more than 30,000 kWh/yr. By considering an effectiveness of

75% for a typical heat recovery unit, 22,500 kWh/year of the restaurant’s heating energy

(equal to 138% of total energy consumption by the A/C-R system) can be provided from

the condenser’s heat rejection. Furthermore, replacing the constant speed compressor

(repeatedly cycles on and off during most of the duty cycles) with a variable speed

compressor that is adjustable for partial loads will have a significant impact on the

energy consumption of the freezer room. Studies have proven that the start-up electric

power draw of electrical compressors is several times the nominal power [88]. The

measured values for the current system were presented in Section 3.3.2.

4.2. Quasi Steady State Model Validation

The accuracy of the first and modified versions of the developed quasi-steady

state model (see section 0) are evaluated under transient operating conditions in this

section. The first version of the quasi-steady state model is developed based on the

validated steady state model. Transient operating conditions can be imposed on an A/C-

R system by variations in evaporator or condenser air inlet conditions as well as the

commands from system controller. Accordingly, any imposed transient condition that

takes into account variations in inlet air condition (temperature/humidity and mass flow

rate) or controller commands (on/off or speed change), can be considered in this section

to verify the accuracy of model.

A variety of tests have been carried out to verify the accuracy of the developed

quasi-steady state model. Based on the comparisons, the accuracy of the first version of

model to predict the behavior of system under transient conditions degrades as the

intensity of changes increases. The worst case scenario, with highest discrepancy

betwe

comm

stead

range

discre

figure

stead

and a

by a

the ti

mode

repre

Figur

een the me

mands from

dy model ou

e of 2% -

epancy betw

e shows a co

dy model an

ambient tem

thermostat l

ime. In this

el is not zer

esent the “off

re 4-21: Comdu

easurements

the control

utputs and th

100%. In F

ween the mo

omparison o

d measured

mperature of

ocated in th

figure, the i

ro represent

f” cycles.

mparison buring on-off

s and the s

ler. The ran

he measurem

Figure 4-21

odel predictio

of the cooling

d in the testb

30oC. The c

e conditione

intervals thr

t the “on” cy

between firsf cycling

93

simulation re

nge of relati

ments throu

, one of th

ons and the

g load calcul

bed for a con

compressor

ed space. Ho

rough which

ycles and th

st quasi-stea

esults, appe

ive discrepa

gh the cond

he results w

experimenta

lated by the

nditioned sp

and conden

owever, the

the predict

he intervals

ady model a

eared aroun

ancy betwee

ducted tests

with the hig

al data is pre

first version

pace temper

nser fan are

evaporator

ted cooling p

with zero c

and experim

nd the on-o

en the quas

varies in th

ghest relativ

esented. Th

n of the quas

ature of 40o

cycled on-o

fan is “on” a

power by th

cooling powe

mental data

off

si-

he

ve

is

si-

C

off

all

he

er

a

94

During the “on” conditions, the model predicts a constant cooling power of the

system due to steady state air-side temperature and compressor/fans speeds. However,

based on the measurements, the cooling power increases slowly after each start-up

before reaching a steady state condition. During the “off” intervals, the model does not

predict any deliverable cooling power by the system because the compressor does not

create any refrigerant flow. However, the measurements show that the system still

provides cooling during the compressor “off” intervals due to the existence of cold

refrigerant in the evaporator (the evaporator fan is kept “on”). Accordingly, during the

“off” intervals the model predicts the system cooling power with 100% relative difference.

The relative differences between the model and experimental data for the other

parameters of the system are less than the presented cooling power; however, the

model accuracy is not acceptable. As described in Section 2.2, the main source of this

high relative discrepancy is a lack of considering the thermal inertia of the condenser

and evaporator in the model. Therefore, a modified version of the quasi-steady model is

developed and discussed in that section. Figure 4-22 shows a comparison between the

modified quasi-steady model and the experimental data for the same testing conditions

as Figure 4-21. The modified model predicts the behavior of the A/C-R system

remarkably better than the first quasi-steady model. The maximum relative difference

between the model and the measured values is less than 24%. However, the maximum

relative difference between the time-averaged values of model predictions and the

measurements is less than 8%.

Figur

of th

There

evapo

(cond

temp

simul

comm

evapo

Figur

for th

are

predi

instan

For

instan

re 4-22: Comda

To achiev

e A/C-R sy

efore, a com

orator fan

denser air in

erature (eva

ltaneously o

manded in a

orator and

re 4-23 and

he cooling p

presented.

cted by the

ntaneous re

the power

ntaneous va

mparison bata during o

ve more com

ystem, more

mbination of

in a range

nlet temperat

aporator air i

on the syst

a random wa

condenser

Figure 4-24,

ower and po

Based on

e model is

elative differe

consumptio

alues are les

between moon-off cyclin

mprehensive

e complex d

on-off comm

of 25-100

ture) change

inlet temper

tem. The c

ay and in th

are varied

, the compar

ower consu

the obtaine

4% differe

ence betwee

on, the av

ss than 26%

95

dified quasng

model valid

dynamic cha

mands to th

0% of nomi

e in a range

ature) variat

compressor

e meantime

using the e

rison of mod

mption of th

ed results,

nt from the

en the mode

verage valu

% off. The ca

si-steady mo

dation for the

anges are i

he compress

inal capacit

e of 20-60 oC

tion in a ran

and evapo

e, the air tem

environment

del prediction

he system d

the time-a

e measured

el and expe

es are diff

lculated ave

odel and ex

e real opera

mposed on

sor, speed c

ty, ambient

C, and cond

ge of 10-50

orator fan a

mperatures a

tal chamber

ns and expe

uring the m

averaged co

d values. Th

eriment is les

ferent by 8

erage COP o

xperimental

ting conditio

the system

change to th

temperatur

itioned spacoC is applie

are manual

at the inlet o

r controls. I

erimental dat

mentioned tes

ooling powe

he maximum

ss than 22%

8% and th

of the system

on

m.

he

re

ce

ed

ly

of

In

ta

st

er

m

%.

he

m

durin

The

avera

consi

mode

not b

the a

Figur

g this test is

obtained ac

aged perform

idering the

el. However,

be satisfacto

ccuracy of th

re 4-23: Comstco

s 1.50 which

ccuracy from

mance study

relatively low

, for dynamic

ry although

he develope

mparison oeady mode

onditions

h is less than

m the quasi

y of the A/C-

wer comput

c behavior in

the computa

ed fully trans

of cooling pol and exper

96

n 5% differe

-steady stat

-R systems,

tational cost

nvestigation

ational cost

sient model w

ower variatrimental dat

ent from the

te model pr

this model c

ts in compa

of A/C-R sy

is relatively

will be valida

ion betweeta during co

measured v

roves that f

can be relia

arison to a f

ystems the a

y low. In the

ated.

n modified ombined im

value of 1.57

for any time

bly employe

fully transien

accuracy ma

next section

quasi-mposed

7.

e-

ed

nt

ay

n,

Figur

4.3.

valida

traile

from

valida

perio

show

are p

evapo

comp

(Figu

re 4-24: Comstco

TransieInvesti

In this sec

ated and th

rs of Saputo

Saputo trai

ation and rea

d and its ou

wn in the res

presented in

orator and

pared on th

re 4-26) is d

mparison oeady mode

onditions

ent Modigation

ction, the ac

he performa

o Inc. (see S

lers (see Se

al-time perfo

utputs are co

sults. Two sa

Figure 4-25

the instanta

hese graphs

directly calcu

of power conl and exper

el Valida

ccuracy of th

nce behavio

Section 3.4)

ection 3.4) i

ormance inv

ompared wit

ample comp

5 and Figur

aneous cool

s. The insta

ulated from

97

nsumption rimental dat

ation, Re

e developed

or of VACR

is investigat

s utilized in

estigation. T

th the meas

parisons of t

re 4-26. The

ing power d

antaneous c

air enthalpy

variation beta during co

eal-time P

d transient m

R systems e

ted. The coll

this section

The model is

ured values

the model r

e temperatur

delivered by

cooling pow

y entering an

etween modombined im

Performa

model (see S

employed in

ected real-ti

n for the tra

s executed o

s, with a goo

results with t

re of cold a

y the VACR

wer from m

nd leaving th

dified quasmposed

ance

Section 2.3)

n refrigerate

ime field dat

ansient mode

over the stud

od agreemen

the field dat

ir leaving th

R system ar

measurement

he evaporato

i-

is

ed

ta

el

dy

nt

ta

he

re

ts

or

(obta

and t

maxim

unce

for th

Figur

coolin

(discu

concl

(end

beha

infiltra

the V

the m

16.9

Figur

ained from p

the air veloc

mum relative

rtainty analy

he measure

re 4-25 and

ng power d

ussion and e

lude that at

of each op

vior appear

ation (due to

VACR system

maximum de

and 17.8 kW

re 4-25: Sam- T

psychrometri

city data (ob

e difference

ysis of the e

ed temperat

Figure 4-26

during the r

explanations

the beginnin

perating cycl

rs due to th

o door open

m. The simu

elivered coo

W that happe

mple simulaTrailer #2, W

ic data base

btained from

of 8.5% bet

experimenta

ture and ±0

6 show the

regular and

s can be fou

ng of each “

le) the cooli

hermal inerti

ning during d

ulation and m

oling power

ens at 19:50

ated and meWed to Thu,

98

ed on the m

air velocity

tween the m

l data show

0.5% for the

fluctuations

d chaotic on

und in Sectio

“ON” period

ing power o

ia of trailer

delivery) tha

measuremen

by the VAC

and 18:47 o

easured air August 20

measured te

y sensors). T

model and m

ws a maximu

e measured

s of conditio

n-off cycling

on 3.4.2). Fr

as well as d

of the VACR

(at the beg

at impose hi

nts over the

CR system o

of a warm da

r temperatur15

mperature a

The compar

measuremen

um uncertain

d cooling p

ned air tem

g of the de

rom Figure 4

during the d

R system is

ginning) and

igher cooling

study perio

of trailers #1

ay in July, re

re at evapo

and humidity

isons show

t results. Th

nty of ±0.5o

ower values

mperature an

elivery hour

4-26 one ca

delivery hour

s higher. Th

d ambient a

g demand o

ods show tha

1 and #2 ar

espectively.

rator outlet

y)

a

he

C

s.

nd

rs

an

rs

is

air

on

at

re

t

99

Figure 4-26: Sample simulated and measured cooling power - Trailer #2, Thu to Sat, August 2015

One of the major sought-after outputs of this study is the real-time COP of the

VACR systems. To obtain the real-time COP, both instantaneous cooling power and

input power to the VACR system are required. Because the VACR system is directly

driven by a diesel engine, the instantaneous mechanical power input to the compressor

and fans of the system cannot be directly measured. As such, the only way of obtaining

the instantaneous power consumption and COP is through the simulation using the

developed and validated model. However, the measurement of overall energy

consumption by the refrigerated trailer is also done over the study period to achieve

comprehensive and more reliable results. The most detailed overall energy consumption

achieved in this study is the amount of fuel needed to refill the trailer. To achieve the

highest accuracy for the fuel consumption data, the fuel tank is fully filled to capacity at

each refill and thus, the amount of fuel consumption during each interval is equal to

amount of fuel needed to refill the tank after that interval.

Figure 4-27 shows a sample of simulated transient power input to the VACR

system of trailer #2 over a 10 day period. Based on the transient simulation results over

100

day and nights of mild and warm seasons, the power consumption of the VACR system

(trailer #2) varies between 11.2 and 18.4 kW over the “ON” cycles depending upon the

imposed cooling load as well as the ambient and indoor temperatures. In general, the

power consumption of the VACR system is higher at the beginning and end of each

“ON” cycle due to higher imposed cooling load because of thermal inertia in the

beginning and door opening during the delivery hours. Based on the simulations over the

year, the maximum power consumption by the VACR system of trailers #1 and #2 are

18.0 and 18.4 kW that appear at 18:02 and 17:36 of a warm day in July, respectively.

Furthermore, the average power consumption over an “ON” cycle is higher in days of the

warm seasons and lower during nights of mild seasons. Based on the obtained results,

the maximum power consumption of the studied VACR system of the trailers during a

complete “ON” cycle is 44.5 kWh, which occurs on a warm day in July. The average

yearly energy consumption of the simulated VCR systems is 11,800 kWh.

Figure 4-27: Sample simulated power consumption for 10 days - Trailer #2, August 2015

Finally, the real-time COP of the refrigerated trailers was obtained using the

simulated transient cooling power and power consumption over the study period.

101

Figure 4-28 shows a sample of the obtained COP results. Based on all the obtained

results during mild and warm seasons for both trailers, the following can be concluded:

o The real-time COP of the refrigerated trailers varies in a range of 0.001-1.33

over the “ON” periods. This wide range of COP indicates sensitivity of the

VACR systems to the operating condition. The significantly small values of

COP (less than 0.1) exist over a short period of time at most of the on-off

fluctuations right after the VACR starts working. Over such periods, the

VACR immediately starts consuming significant power; however, it takes time

(several seconds up to two minutes) to deliver a considerable cooling effect.

o The larger COP values appear over the beginning and final hours of most

“ON” periods, even though the power consumption is a bit higher over those

hours (see Figure 4-27). These higher COP values exist due to higher cooling

power of the VACR system over those hours (see Figure 4-26). However, in

the middle of most “ON” periods the COP is smaller due to smaller cooling

power even though the power consumption is also a bit lower.

o The average simulated COP of the trailers over the year is 0.58. This number

is significantly smaller than COP of stationary systems and highlights the

importance of studying the food transportation industry, as it has an

enormous impact on global energy consumption and the environment.

o Using the measured fuel consumption of the trailer, the experimental COP is

also calculated. As such, the measured cooling power is integrated over time

and used along with the measured fuel consumption over the same period of

time, which results in an overall time-averaged COP of 0.185 for the engine-

VACR combination. Based on the manufacturer’s data, the average thermal

efficiency of the employed Diesel engines is nearly 30%. As such, the time-

averaged COP of the VACR systems will be 0.62 based on the

measurements. Thus, there is only a 6.5% relative difference between the

simulated and measured average COP values of the studied VACR systems,

which validates the accuracy of the obtained results.

102

Figure 4-28: Sample simulated COP for 10 days - Trailer #2, August 2015

The obtained results from simulations and measurements shows that a typical

VACR of a refrigerated trailer that works in a relatively average climatic condition (British

Columbia, Canada) consumes 11,800 kWh/yr. Assuming that the average diesel engine

has a thermal efficiency of 30%, the typical refrigerated trailer consumes more than

3,000 liter/yr of Diesel fuel. The maximum required power and energy during an “ON”

period throughout the year (including loading and delivery) are calculated as 18.4 kW

and 44.5 kWh, respectively. Based on the obtained data from the GPS of the trucks that

move the studied trailers, more than 60,000 km/yr is travelled by each trailer. Therefore,

a heavy and low-efficiency engine-driven VACR (total of 1,100 kg including the fuel tank)

is transported more than 60,000 km/yr, significantly contributing to GHG emissions. As a

key conclusion of this part of the study regarding the required power and energy and

performance variations of the VACR systems, it is highly recommended that all the

engine-driven refrigerated trailers around the world be replaced by battery-powered

refrigerated trailers to eliminate the relevant GHG emissions. Based on the

manufacturer’s data [94,95], eliminating the engine and fuel tank will reduce the total

weight of trailer up to 600 kg. Assuming an average specific power and energy of 300

103

W/kg and 0.2 kWh/kg for Li-ion batteries to replace the engine and fuel tank [97,98], the

maximum weight of required electromotor and batteries will be 250 kg. Thus, 350 kg of

total trailer weight can be removed by replacing the engine-driven VACR with a battery-

driven VACR system. This weight reduction equals 1.5% of total truck-full trailer weight

based on the field measurements and manufacturer’s data.

Utilizing the available data on average fuel consumption versus weight of freight

in transportation industry, the total fuel consumption of the truck that moves the trailer

will be reduced by more than 0.5% as a result of such a weight reduction [99,100]. Thus,

assuming an average truck-trailer combination fuel consumption of 35 L/100 km

[101,102], at least 105 liters of Diesel fuel can be saved per year per truck-trailer. In

addition, all the fuel consumption (3,000 liter/yr based on the obtained results) and GHG

emissions from the refrigerated trailer’s Diesel engine will be eliminated. As such, a total

of 3105 liters of diesel fuel, which is equivalent to 8,320 kg CO2/yr [103,104] can be

reduced per truck-trailer combination by replacing the engine-driven VACR with battery-

powered type. In some areas around the globe, electricity is produced using clean

techniques. Eventually, expanding the obtained results from the studied trailers shows

that if all the engine-driven trailers in the world (estimated to be 1.2 million [7]) are

replaced by battery-powered types fed from clean tech-generated electricity, more than

3.7 billion liters of diesel fuel will be saved and CO2 production can be reduced by up to

10 million tons annually.

4.4. Conclusion

In this chapter, the developed steady state, quasi-steady state, and transient

models were validated and employed for comprehensive thermal and performance

investigation of stationary A/C-R and VACR systems. The results of the steady state

model were validated over a wide range of operating conditions using the testbed and

showed less than 10.6% relative difference from experimental data. In addition, the

accuracy of the steady state model was validated using the filed data from a stationary

A/C-R system and showed a maximum of 9% relative difference. The quasi-steady state

104

model was validated using the experimental data from testbed and showed a maximum

of 26% relative difference on the dynamic parameters; however, a good accuracy with

less than 8% relative difference was achieved on the time-averaged parameters. The

accuracy of the transient model was also validated using the field data from real VACR

systems. The results showed a good agreement with less than 8.5% maximum relative

difference on the transient parameters of VACR systems.

The validated steady state model was employed for a comprehensive

investigation of thermal and performance characteristics of VACR systems under

different operating conditions. The results showed that for any imposed ambient and

conditioned space temperatures, a point of maximum COP can be achieved by properly

adjusting the evaporator and condenser fans speeds. The obtained results showed 11%

potential improvement of COP under ANSI/AHRI 210/240-2008 conditions. The obtained

results will be utilized for optimization model development in the next chapter. The

steady state model was also employed for a performance evaluation of the pilot

stationary A/C-R system and showed a remarkable degradation of the COP of the

system from the design condition. Areas of potential energy savings were discussed for

the studied stationary system. Finally, the results of the transient modelling of the real

VACR system over the duty cycles were discussed and employed for an estimation of

the global impact. By expanding the obtained results, it was recommended that all the

existing engine-driven VACR systems employed in the refrigerated trailers all around the

world be replaced by battery-powered VACR systems, which could lead to saving more

than 3.7 billion liters of diesel fuel and reducing CO2 production by up to 10 million tons

per annum. The results obtained in this chapter will be employed to develop an

optimization model that will be implemented on the proof-of-concept high efficiency

VACR system.

105

Chapter 5. Metamodel Development for Optimization

In Section 4.1.1 of Chapter 4, it was concluded that for any imposed operating

condition on a VACR system, a point of maximum COP can be achieved by adjusting

the capacity of controllable components of the system. As such, taking advantage of

optimization techniques can properly lead the project toward achieving the final goal:

development of an optimally controlled high-efficiency VACR system. A variety of

techniques have been developed during the last decades for optimization of engineering

systems in the steps of design or operation [105]. The literature review shows that

numerous studies have generally covered the optimization of A/C-R systems in a variety

of applications; however, the literature lacks a systematic optimization model

development for VACR systems equipped with variable speed compressor and fans

under realistic operating condition [106–116]. As such, in this chapter, a metamodel is

developed for the new high efficiency VACR system built in LAEC to implement an

optimization method on it and achieve the highest COP under any operating condition.

The metamodel will be used in an optimization solver to be implemented on the proof-of-

concept demonstration of high efficiency VACR system in Chapter 6.

5.1. Metamodel Development Based on Steady State Thermal Model

The goal of the optimization in this research is to maximize the COP of the VACR

system under any operating conditions by adjusting the speed of the compressor and

fans. As such, the so-called cost (objective) function to be minimized in the standard

106

form of optimization problem definition is -COP [105,117]. The COP is calculated using

the following equation:

, ,

evap

comp f cond f evap

QCooling powerCOP

Input Power W W W

(5.1)

To achieve the maximum COP of the VACR system in reality, all the appeared

variables ( evapQ , , ,, ,comp f cond f evapW W W ) should be calculable and controllable. However,

these variables are not directly controllable since they are functions of other system

variables. The only controllable variables are the speed of compressor and fans. Based

on the model developed in Chapter 2, the cooling power evapQ is calculated using:

, , , ,evap ref ref evap in ref evap outQ m h h (5.2)

The appeared variables on the right hand side are not still directly controllable.

These three variables are also functions of other VACR system’s variables as follows:

2 2 2 2 2 20 1 2 3 4 5 6 7 8ref e c e e c c e c e c e cm d d T d T d T d T T d T d T T d T T d T T

(5.3)

, , 1 , , 2( , ); ( )ref evap out ref evap out e e eh f T P P f T (5.4)

min, , , , ,, , , ,

( )evap evap ref evap in ref evap outref evap in ref evap out

ref

C T Th h

m

(5.5)

Again, none of the appeared variables are controllable since all of them are

functions of other variables of a VACR system. Accordingly, other correlations should be

employed for calculating these variables [70]. In addition to the abovementioned

correlations for calculating the cooling power, evapQ , the power consumption (input

107

power) variables appeared in the denominator of Eq. (5.1) should also be correlated

based on the controllable variables (see Chapter 2):

2 2 2 2 2 20 1 2 3 4 5 6 7 8comp e c e e c c e c e c e cW c c T c T c T c T T c T c T T c T T c T T (5.6)

2 3

, , ,, , , 0 1 2 3

, , , , , ,

a cond a cond a condf cond f cond nom

a cond nom a cond nom a cond nom

m m mW W cf cf cf cf

m m m

(5.7)

2 3

, , ,, , , 0 1 2 3

, , , , , ,

a evap a evap a evapf evap f evap nom

a evap nom a evap nom a evap nom

m m mW W cf cf cf cf

m m m

(5.8)

From the above equations, one can conclude that the power consumptions of the

compressor and fans are correlated to variables that are not directly controllable (similar

to the cooling power). Accordingly, these formulae should also be broken into other

formulae to have relationships between controllable variables and the power

consumptions [70].

According to the abovementioned governing equations for COP calculation, it

can be concluded that the cost function, which is -COP, cannot be explicitly expressed

based on the VACR system’s controllable variables. In other words, all the mentioned

coupled equations should be solved together in order to achieve a value for COP. As

such, the full steady state mathematical model obtained in Chapter 2, which is a set of

20 coupled nonlinear equations, should be solved to calculate the COP of the VACR

system. In order to achieve optimization for problems like this, metamodeling has been

intensively used. The followings are limitations of the classical gradient-based

optimization methods that hinder the direct application of these methods for solving such

problems [117]:

1- Explicitly formulated and/or computationally cheap models are required; however,

engineering simulation involves implicit and computation-intensive models.

108

2- The output of gradient-based methods is usually a single optimum solution, while

obtaining multiple design alternatives is desirable.

3- The gradient-based methods provide inadequate insight since they are

sequential and non-transparent.

4- High level of expertise in optimization is required to apply the gradient-based

optimization methods.

On the flip side, the metamodeling approach has the following advantages [117]:

1- Optimization techniques for metamodels have higher efficiencies.

2- Parallel computations can be done to generate sample points since the

metamodels are obtained from separate independent points.

3- Engineering insight can be more efficiently achieved since the sensitivity of

effective variables is easily assessable.

4- The metamodeling approach can deal with both continuous and discrete

variables.

Metamodel-based optimization can be performed with different strategies. In this

research, the “sequential approach” is employed that is formed from: 1) sample space

generation, 2) metamodel development, 3) metamodel validation, 4) optimization on

metamodel [117]. For the current problem, which is maximization of the COP, the

sample space is generated by executing the steady state mathematical model

developed in Chapter 2. Then, the metamodel is obtained by using “Least Square

Regression” for the generated space and after validation, it is employed for optimization.

A standard optimization problem is usually formulated as [105]:

1 2

min ( )

. .

0; 1:

0; 1:

, ,..., ; ,

i

j

T

n L U

f x

S T

H i p

G j m

x x x x x x x

x

x (5.9)

109

where, ( )f x COP in this study. Therefore, a typical metamodel formulation for

optimization is [117]:

1 2

min ( )

. .

0; 1:

0; 1:

, ,..., ; ,

i

j

T

n L U

f x

S T

H i p

G j m

x x x x x x x

x

x

(5.10)

For the current problem, the speed of compressor, condenser and evaporator

fans are controllable and independent. In addition, the metamodel-based optimization

solver is supposed to be integrated with a cooling demand simulator in the next chapter

to proactively and optimally control the operation of VACR system under any imposed

ambient condition. As such, the cooling demand is an important input to the metamodel

for the purpose of COP maximization. In addition, based on the model characteristics

and the results of thermal study presented in Chapter 3 and Chapter 4, the air

temperature at evaporator and condenser entries directly and independently affect the

thermal and performance characteristics of the VACR system. The rest of available

parameters and variables appeared in the mathematical model are dependent on the

abovementioned variables. Accordingly, the following is defined as the standard form of

the so-called design variables vector x for the metamodel:

1

2

3

4 , ,

5 , ,

:

:

:

: ( @ )

: ( @

oa evap in

oa cond in

x Compressor rotational speed ratio

x Condenser fan rotational speed ratio

x Evaporator fan rotational speed ratiox

x T Air temperature evaporator entry C

x T Air temperature condenser entry C

6

)

: ( )x Cooling demand kW

(5.11)

110

Therefore, the metamodel 1 2 6( ) ( , ,..., )f x COP x x x will be obtained for the

optimization purpose. Due to capacity and range of operation, the following linear

constraints govern the optimization problem:

531

51 3

2 64

2 4 6

15 000

60 01 0 1 0

0 010 0

1 0 50 0 2.5 0

xxx

xx x

x xx

x x x

(5.12)

Based on the developed formulation, the metamodel is obtained in this chapter

and will be used along with a cooling demand simulator in the next chapter to find the

global optimum point of operation for any imposed cooling demand and temperatures on

the VACR system. The three steps of sequential metamodel development for this

problem including: sample space generation, metamodel development, and metamodel

validation, are discussed in this section. In the next chapter, the validated metamodel will

be used for the purpose of proactive and optimal control of the built high efficiency

VACR system.

5.1.1. Sample space generation

The previously developed steady state model is executed for random design

variables within the feasible ranges of application formulated in Eq. (5.12). Different

numbers of sample points (NSP) between 500-5000 are tested to achieve the highest

accuracy; a comparison between the results is presented in the next section. A few

representative of the generated sample points are shown in Table 5-1.

111

Table 5-1: A few representatives of the generated sample space for metamodel development-steady state

f COP 1x 2x 3x 4x 5x 6x

-3.001 0.4 0.625 0.75 26.7 35 1.1

-2.404 0.35 0.625 0.25 25 33 0.5

-1.929 0.3 0.875 0.25 28 35 0.5

-3.269 0.3 0.375 0.75 26.7 30 1.1

-2.358 0.65 0.875 0.5 26 35 1.1

-2.181 0.3 0.625 0.5 21.5 35 0.8

-1.044 1 0.375 0.5 25 20 0.9

-1.543 1 0.875 0.75 22.5 28 1.4

-1.460 1 0.875 0.5 15 40 1.1

-3.648 0.65 0.375 0.5 10 20 0.9

-3.478 0.3 0.625 0.25 25 60 0.6

-0.967 0.3 0.375 0.25 10 60 0.3

-2.943 0.3 0.875 0.25 25 20 0.6

112

5.1.2. Metamodel development

Utilizing the entire generated sample points, the metamodel is developed. Based

on the COP behavior studied in section 4.1.1, a second-order response model is defined

for the metamodel [117,118]:

6 6 6 62

01 1 1 1

( )i iii i ij i j

i i i j i j

f x x x x x

(5.13)

Using the least square approach:

1 0

2 111 12 1

3 221 22 2

1

1 ...

1 ...; . , , .

. . . ... .. .

1 . .... .

k

k

n nk

n ij

f

fx x x

fx x x

x x

f

f x β e f x β (5.14)

where, e represents the error, k is the number of coefficients and n is the number

of sample points, the coefficients of model can be obtained. Therefore, following [105]:

2

1

Tn

ii

L e

f x β f x β (5.15)

2 2 0T TL

x f x x ββ

(5.16)

113

1T Tβ X X X f (5.17)

a sample of the obtained elements of β for NSP = 5000 are shown in Table 5-2.

Table 5-2: Obtained coefficients of metamodel for steady-state condition

0 1 2 3 4 5 6

-2.8219 0.0584 0.0143 0.0842 0.0805 0.0054 -11.4991

11 22 33 44 55 66 12

-0.0003 -0.0001 -0.0009 -0.0008 -0.0001 -4.0872 -0.0001

13 14 15 16 23 24 25

-0.0003 -0.0006 -0.0000 0.0530 -0.0000 -0.0002 -0.0000

26 34 35 36 45 46 56

0.0050 -0.0009 -0.0000 0.1219 -0.0000 0.1255 -0.0009

5.1.3. Metamodel validation

For validating the accuracy of metamodel, two different methods including: R-

squared and relative average absolute error (RAAE), are employed for the developed

metamodel using the sample and testing points [119]. This section explains the

114

calculation of these parameters and validation of the metamodel. The corrected sum of

squares TSS is used as a measure of overall variability in the data [117,119].

2

1

n

iiT

T

f

SSn

f f (5.18)

In addition, the sum of squares due to error is:

TT TESS f f β x f (5.19)

As such, the R-squared is defined as follows:

2

2 12

1

1

ns s

tT E

ns sT

t

f x f xSS SS

RSS

f x f x

(5.20)

where, sf x and sf x are the so-called black-box function (previously developed

mathematical model of VACR system, see section 2.1) and the metamodel function at a

sample point sx , respectively. Moreover, sf x denotes the mean of black-box

function over the n sample points in approximating the black-box function. The R-

squared indicates the overall accuracy of the approximated metamodel with given

sample points. The closer the value of R-squared approaches one, the more accurate is

the approximation model. However, R-squared is not sufficient for comparing accuracies

of two models because it only shows the accuracy with respect to modeling points, not

the model’s extrapolation ability. As such, the RAAE is defined as:

115

1

nt t

t

f x f xRAAE

n STD

(5.21)

2

1

1/n

t t

t

STD n f x f x

(5.22)

where, STD denotes the standard deviation of the black-box function prediction at the

test points. The defined RAAE indicates the overall accuracy of the approximated model

in the entire design range, computed at random test points. The smaller the RAAE value,

the more accurate is the approximation. Both R-squared and RAAE show the average

accuracy over the entire space.

As such, the R-squared and RAAE for all the NSP values from 500 to 5000 are

calculated and presented in Table 5-3. For the selected NSP = 5000, the R-squared is

0.820 and the RAAE is 0.184. The results show that by increasing the NSP, the R-

squared decreases; however, the RAAE also decreases. The R-squared value is

decreasing with the increasing number of sample points. It is expected because the

model tries to find the coefficients in a way that the model becomes close to all of the

sample points. By comparing the RAAE values, it can be concluded that the derived

model is getting more accurate by increasing the number of sample points. This

happens due to the fact that the RAAE is getting lower as the NSP increases [119]. In

fact, by increasing the NSP, the accuracy of obtained metamodel increases since a

wider range of design variable values are engaged in development of the model. A

comparison between the NSP of 2000 and 5000 shows that the relative difference

between the R-squared values and RAAE values on these NSPs are less than 4%,

which shows a good validation for such complex engineering problem [117–119].

Accordingly, the NSP of 5000 and the corresponding metamodel are selected for the

purpose of global optimization in this study.

116

Table 5-3: Comparison of obtained accuracies for different NSP

NSP R-squared RAAE

500 0.912 0.977

1000 0.851 0.484

2000 0.834 0.191

5000 0.820 0.184

5.2. Metamodel Development Based on Transient Thermal Model

Based on the presented results in sections 3.3, 3.4, and 4.3, the VACR systems

undergo a dynamic load and operating condition during the duty cycles. As such, for

optimizing the COP under transient operating condition, a metamodel based on the

validated transient mathematical model of the VACR system is required. However, the

previously developed metamodel in the last section can be employed for the steady

state condition. Similar to the optimization methodology discussed for the steady state

condition in the last section, due to complexity, non-linearity, and coupling of the

transient model equations, a metamodel-based optimization is used for the transient

condition.

Equation (5.10) governs the metamodel for transient condition. Similar to the

steady state condition, the compressor and fans’ speed are the main controllable

variables of the VACR system. In addition, the proactive control design that is based on

the integration of metamodel-based optimization solver with an existing cooling demand

simulator in the next chapter requires the cooling demand to be an input to the

117

metamodel. Furthermore, the ambient (outdoor) and conditioned space (indoor)

temperatures directly affect the system operation and behavior. Moreover, as discussed

in Chapter 3 and Chapter 4, the transient variation of ambient and conditioned space

temperatures directly or indirectly affect the thermal and performance characteristics of a

VACR system. The remaining variables and parameters in the developed transient

mathematical model (see section 2.3) are dependent on the abovementioned variables.

Accordingly, the following is defined as the so-called design variables vector x for the

metamodel under transient condition:

1

2

3

4 , ,

5 , ,

:

:

:

: ( @ )

: ( @

oa evap in

oa cond in

x Compressor rotational speed ratio

x Condenser fan rotational speed ratio

x Evaporator fan rotational speed ratio

x T Air temperature evaporator entry C

x T Air temperature condenser entry Cx

6

, , 17

, , 18

)

: ( )

: ( . )

: ( . )

a evap in o

a cond in o

x Cooling demand kW

dTx C s

dtdT

x C sdt

(5.23)

Using the field and experimental data on the range of variation of defined

variables, the following constraints govern the optimization problem:

5 731

5 71 3

2 6 84

2 4 6 8

15 0 0.07 000

60 0 0.07 01 0 1 0

0 0 0.03 010 0

1 0 50 0 2.5 0 0.03 0

x xxx

x xx x

x x xx

x x x x

(5.24)

Following same procedure applied for the steady state model, the three steps of

metamodel development for the transient condition are reported in this section as:

sample space generation, metamodel development, and metamodel validation.

118

5.2.1. Sample space generation

The developed and validated transient model in section 2.3 is executed for

random design variables within the feasible ranges of application formulated in Eq.

(5.24). Similar to the steady state condition, a variety of sample space sizes between

500 and 10000 are tested to achieve the highest accuracy.

5.2.2. Metamodel development

Based on the system behavior and similar to the steady state model, a second-

order response model is defined for the metamodel of transient condition:

8 8 8 82

01 1 1 1

( )i iii i ij i j

i i i j i j

f x x x x x

(5.25)

Using the least square approach, the highest accuracy metamodel is obtained.

Table 5-4 shows the calculated elements of β for NSP = 10000.

119

Table 5-4: Obtained coefficients of metamodel for transient condition

0 1 2 3 4 5 6

-4.3211 0.1097 0.0169 0.1986 0.1985 0.0034 -22.1227

7 8 11 22 33 44 55

-0.6255 -0.3349 -0.0005 -0.0000 -0.0019 -0.0020 -0.0002

66 77 88 12 13 14 15

-5.6818 -0.1378 0.1009 -0.0002 -0.0006 -0.0018 0.0000

16 17 18 23 24 25 26

0.0934 0.0020 0.0065 -0.0001 -0.0005 -0.0001 0.0248

27 28 34 35 36 37 38

0.0090 0.0003 -0.0027 -0.0000 0.2039 0.0004 0.0017

45 46 47 48 56 57 58

0.0001 0.2781 0.0098 0.0055 0.0195 -0.0079 -0.0083

67 68 78

-0.0001 -0.0005 -0.0003

120

5.2.3. Metamodel validation

Following same procedure of steady state model for the transient condition, the

accuracy of developed metamodel is validated. The R-squared and RAAE for all the

NSP values from 500 to 10000 are calculated and presented in Table 5-5. For the

selected NSP = 10000, the R-squared is 0.751 and the RAAE is 0.217. A comparison

between the R-squared and RAAE values for NSP of 5000 and 10000 shows a

maximum relative difference of less than 5% that represents a good validation for the

metamodel. Therefore, the NSP of 10000 and the corresponding metamodel are

selected to be integrated with the cooling demand simulator in the next chapter.

Table 5-5: Comparison of obtained accuracies for different NSP

NSP R-squared RAAE

500 0.876 1.068

1000 0.818 0.574

2000 0.781 0.310

5000 0.764 0.228

10000 0.751 0.217

121

5.3. Conclusion

The metamodels for steady state and transient operating conditions were

obtained in this chapter. Due to complexity of previously developed mathematical

models, which appeared from the nonlinear and coupled governing equations, the

metamodel-based optimization was selected in this study. The objective function was

defined and the metamodel was obtained for steady state and transient conditions

separately. The so-called design variables vector for steady state condition was formed

from speeds of compressor and fans, cooling demand, ambient temperature, and

conditioned space temperature. The ambient and conditioned space temperature

gradients (time derivative) were added to the mentioned variables to form the set of

design variables for the transient metamodel. Sample spaces with NSP of 500-5000 for

the steady state and 500-10000 for the transient condition were generated by executing

the original mathematical models for random design variables. The metamodels were

then developed using second-order response model approach and validated using R-

squared and RAAE methods. The values of 0.820 for R-squared and 0.184 for RAAE of

steady state metamodel were achieved using the NSP of 5000. In addition, the R-

squared of 0.751 and the RAAE of 0.217 were achieved for the transient model using

the NSP of 10000. The obtained metamodels will be integrated with a cooling demand

simulator in the next chapter to develop the proactive, optimization-based model

predictive controller for achieving the highest COP of the built VACR system in the

LAEC.

122

Chapter 6. Proof-of-concept Demonstration of High Efficiency VACR System for Service Vehicles

The main goal of this research is to develop a proof-of-concept demonstration of

high efficiency VACR system to be used for service vehicles in transportation industry.

Thus far, the comprehensive theoretical and experimental study on thermal and

performance characteristics of pilot stationary A/C-R and VACR systems have been

performed and metamodels are developed based on the validated results. Moreover, a

high efficiency VACR system is built in the lab using the variable speed compressor and

fans. The final step to achieve the main goal is to develop an optimization solver for the

metamodels and integrate it with a validated cooling load simulation tool and develop a

proactive and optimal controller to be added to the built VACR system. In this chapter,

the optimization solver is developed and integrated with a cooling demand simulator and

implemented on the VACR system. The system is then tested under both steady state

and transient conditions to evaluate the achieved efficiency enhancement in this project.

6.1. Development of Optimization Solver and Integration with a Cooling Demand Simulator, Proactive Control

The reported literature review in section 1.5 and observations on the existing

A/C-R systems show that most existing systems are equipped with simple thermostatic

controllers to keep the desired temperature in the conditioned spaces. Although the

usage of thermostatic controllers reduces the initial cost of A/C-R systems and has

shown a reliable functionality of these systems, the inevitable on-off cycling over the

123

partial cooling demands (as described in the previous chapters) significantly reduces the

life time and performance of the systems. The thermostatic controllers are usually

designed to keep the conditioned space temperature between adjustable low and high

set points. As such, the A/C-R systems receive an “ON” command from the

corresponding thermostat whenever the temperature is above the high set point and an

“OFF” command whenever the temperature drops below the low set point. The

difference between high and low set points varies depending upon the application;

usually around 2-3 oC. The existence of such temperature range in the controller of A/C-

R systems means that the conditioned space temperature can be at most kept within a

range of ±1-1.5 oC from a desired temperature. Therefore, there is always inevitable

temperature fluctuations in the conditioned spaces as reported and discussed in Chapter

3 and Chapter 4.

According to inefficient on-off cycling of common stationary A/C-R and VACR

systems, an idea of “proactive and optimal control” development for these systems

emerged in this research. The proactive feature enables the controller to predict the

effects of current cooling demand in a close future and command the system

accordingly. In other words, if the upcoming effects of cooling demand in a conditioned

space is accurately predicted using a cooling demand simulator, the capacity of

corresponding A/C-R system can be adjusted in advance to compensate the upcoming

effects. This proactive action prevents the system from more or less than enough cooling

power generation and leads to keeping a constant temperature in the conditioned space.

The proactive control concept development is an important step toward the efficient

usage of A/C-R systems. If the proactive controller is also equipped with an optimization

solver of the developed metamodels, the final product can predict the upcoming effects

of current cooling demand, find the optimum speeds of compressor and fans for

compensating the demand, and command the system to operate optimally. As such, the

system can keep the temperature on the desired set point and operate optimally.

To develop a proactive and optimization-based controller for the built VACR

system, an optimization solver of the developed metamodels in Chapter 5 is integrated

with a validated high accuracy cooling demand simulation model [120]. The cooling

demand simulator has been recently developed and validated through a PhD thesis at

124

Simon Fraser University. This simulation model calculates the instantaneous cooling

demand of a conditioned space by solving the energy equation. It requires the

temperature measurement on different spots of the conditioned space. As such, a

LabVIEW interface is designed to collect the required parameters of the demand

simulator and execute a “.dll” version of it. Figure 6-1 to Figure 6-3 represent sample

sections of the developed LabVIEW interface for acquiring the required data, execution

of the cooling demand simulator, and commanding the compressor and fans based on

the optimization solver calculations.

The developed metamodels in Chapter 5 are used in MATLAB to develop an

optimization solver for finding the optimum component speeds. The optimization solver

in MATLAB is developed based on Genetic Algorithm (GA) to find the global optimum of

the metamodels. At each time step, the LabVIEW interface writes the cooling demand

and air temperature at evaporator and condenser inlets into a “.txt” file. The MATLAB

code reads these numbers from the file and executes the GA to find the optimum speeds

and writes them into another “.txt” file. The LabVIEW interface then reads the speeds

from this file and commands the compressor and fans accordingly. As such, the

proactive and optimization-based controller is developed and implemented on the built

VACR system to enhance the performance of it. This controller also belongs to the

category of model predictive controllers (MPCs) as it executes realistic models to control

the operation of an energy system.

It should be mentioned that the developed GA-based solver in MATLAB is

executed for a variety of settings during the tests to obtain the most appropriate setting

before finalizing the controller. As such, a variety of important settings parameters of GA

including the population size, the reproduction crossover fraction (RCF), the crossover

ratio (CR), the creation function, the crossover function, and the mutation function are

tested over the experiment. The most appropriate setting that returns the smallest

minimum value for the cost function (metamodel developed in Chapter 5) is found as

RCF = 0.6 and CR = 0.6 and the rest of the GA parameters do not sensibly affect the

optimum design conditions. Therefore, the optimization model solver is finalized and

used along with the LabVIEW interface to control the VACR system proactively and

optimally.

125

Figure 6-1: A sample section of developed LabVIEW interface for acquiring system’s data and implementing the proactive controller

126

Figure 6-2: Developed section for executing the cooling demand simulator in LabVIEW

127

Figure 6-3: Developed section in LabVIEW interface for commanding compressor, evaporator and condenser fans based on optimization model results

6.2. Proactive, Optimization-Based, Model Predictive Control under Steady State Condition

Utilizing the developed proactive and optimal controller, a variety of tests is

performed on the built VACR system to evaluate the performance characteristics. In

addition, to evaluate the achieved performance enhancement due to development of the

proactive and optimal controller, the results are compared with a thermostatic based on-

off controller on the same system. To perform the thermostatic on-off tests, the cooling

demand simulator is bypassed in the LabVIEW interface and the optimization code is

128

replaced by a simple thermostatic command code. This code just compares the average

temperature in the conditioned space with arbitrary and pre-defined high and low set

points and commands the compressor and fans to either turn on (with nominal speeds)

or turn off. As such, all the tests are done once using the optimal and once using the on-

off controllers. All the operating conditions are kept the same for accurate comparisons.

The condenser of VACR system (see Figure 3-4, Chapter 3) is connected to an

environmental chamber to adjust any desired temperature and humidity for the entering

air stream (represents the ambient air). The evaporator is connected to a chamber in

which an adjustable electric heater is installed. The chamber represents a conditioned

space for doing the tests. By changing the heater power, the cooling demand imposed

on the VACR system will change. As such, a variety of realistic operating condition is

tested for the built VACR system using both the proactive, optimal controller and

thermostatic on-off controller. In this section two sample tests for steady state condition

are presented and discussed. Furthermore, a variety of other conditions are tested and

similar results are obtained.

Figure 6-4 and Figure 6-5 show the settings for steady state #1 and #2 tests. The

heater power is set at 0.8 and 0.6 kW respectively for tests #1 and #2. The air

temperature at environmental chamber (represent the outdoor temperature) is set as 35 oC (based on ANSI/AHRI 210/240-2008) and 30 oC respectively for tests #1 and #2. The

results are obtained after around half an hour from start of the tests to achieve the

steady state condition. The desired set point for the conditioned space (chamber with

electric heater) is set at 26.7 oC (based on ANSI/AHRI 210/240-2008) for test #1 and 24 oC for test #2 with ±1 oC for high and low set points on thermostatic controller (based on

ASHRAE comfort zone table [121]).

129

Figure 6-4: Outdoor (ambient) temperature and heater power setting for steady state sample #1

Figure 6-5: Outdoor (ambient) temperature and heater power setting for steady state sample #2

130

The results of these tests are presented in Figure 6-6 to Figure 6-12 for the

thermostatic (on-off) controller and in Figure 6-13 to Figure 6-17 for the developed

proactive, optimization-based, model predictive controller. All the presented results

except the speed ratios are from measurements. The speed ratios are generated by the

proactive and optimal controller and sent to the compressor and fans. It should be

mentioned that in conditioned space temperature graph (Figure 6-8), the error bars are

not shown to have a clearer vision of the variations. The uncertainty of temperature

measurements is ±0.5 oC. The following are concluded from the presented results:

o The VACR system equipped with thermostatic controller frequently cycles on-off

(see Figure 6-6 and Figure 6-7) to keep the conditioned space temperature within

the desired ranges (see Figure 6-8). However, the system equipped with

proactive and optimal controller adjusts the speed of compressor and fans (see

Figure 6-13 and Figure 6-14) to overcome the imposed cooling load and keep the

temperature constant in the conditioned space (see Figure 6-15). Therefore, the

proactive and optimal controller functions satisfactorily (as per its design) and

eliminates any inefficient on-off cycling of the VACR system. This is one of the

most important achievements of this study that can also solve many issues in

industries with high sensitivity to temperature variations.

o The steady state tests are repeated for a variety of temperature and loads and

similar trends are obtained. The proactive optimal controller functions

consistently with no observed incoherent and unreasonable outputs under

different test conditions. From the results one can conclude that a same

command is reproduced through a steady state condition that confirms stability of

the model predictive controller (see Figure 6-13 and Figure 6-14). The

consistency and reproducibility of commands is a major element of controller’s

reliability.

o The frequent on-off cycling is reflected in the power consumption and COP of the

thermostatically controlled VACR system (see Figure 6-9 to Figure 6-12). The

frequent cycling of the VACR system’s components reduces their lifetime and

performance. However, a consistent and uniform behavior is observed for the

proactively and optimally controlled VACR system.

131

o The COP of thermostatically controlled VACR system for the presented tests

varies between 0 (for “OFF” cycles) and 3.3 (during the “ON” cycles) depending

upon the operating condition and cycling. However, the COP of proactively and

optimally controlled VACR system is nearly constant over time for each of the

tests. The average COP of system during the steady state test #1 is 3.0 and 3.7

respectively for the thermostatic (averaged only for the “ON” cycles) and optimal

controllers. Also, for the test #2 the average COP is 2.8 and 3.4 respectively for

the thermostatic and optimal controllers. These numbers indicate 23.3% and

21.4% higher COP for the optimal controller in comparison with the thermostatic

controller.

o A comparison of the achieved COP from the proactively and optimally controlled

VACR system built in LAEC with the original VACR system (the first VACR

system from industrial partner, see section 3.1) under same steady state

condition of ANSI/AHRI 210/240-2008 shows an improvement from 1.3 to 3.7

equivalent to 185%. From this 185% COP improvement, 162% is achieved by

replacing the VACR elements including compressor, condenser, evaporator, and

fans with high efficiency products in the market. The rest of COP enhancement,

which is 23%, is achieved by developing the proactive and optimal model

predictive controller. The comparisons for other tests over different operating

conditions show a COP enhancement in a range of 179-187%. The share of

proactive and optimal controller in COP enhancement varies in a range of 19-

24% depending upon the operating condition.

o The 19-24% COP enhancement is achieved using the developed proactive and

optimal controller on a VACR system that is built from high efficiency

components in the market. It is predicted that utilizing the developed controller for

any existing high efficiency VACR system (or even any A/C-R system) can lead

to similar remarkable COP enhancement. Furthermore, for any low efficiency

A/C-R system, the components can be replaced with high efficiency and variable

speed types and the new system be integrated with the proactive and optimal

controller to achieve a significant COP enhancement.

132

Figure 6-6: Compressor and fans speed ratio variation for steady state sample #1, thermostatic control (low set point = 25.7, high set point = 27.7 oC)

Figure 6-7: Compressor and fans speed ratio variation for steady state sample #2, thermostatic control (low set point = 23, high set point = 25 oC)

133

Figure 6-8: Indoor (conditioned space) temperature variation for steady state samples, thermostatic control

Figure 6-9: Power consumption of VACR system for steady state sample #1, thermostatic control

134

Figure 6-10: Power consumption of VACR system for steady state sample #2, thermostatic control

Figure 6-11: COP of VACR system for steady state sample #1, thermostatic control

135

Figure 6-12: COP of VACR system for steady state sample #2, thermostatic control

Figure 6-13: Compressor and fans speed ratio variation for steady state sample #1, proactive optimal control

136

Figure 6-14: Compressor and fans speed ratio variation for steady state sample #2, proactive optimal control

Figure 6-15: Indoor (conditioned space) temperature variation for steady state samples, proactive optimal control

137

Figure 6-16: Power consumption of VACR system for steady state sample #1, proactive optimal control

Figure 6-17: Power consumption of VACR system for steady state sample #2, proactive optimal control

138

6.3. Proactive, Optimization-Based, Model Predictive Control under Transient Condition

Similar to the steady state condition, in this section the performance

characteristics of the developed proactively and optimally controlled VACR system are

investigated under transient condition. Different transient heater power patterns that

represent transient cooling demands are imposed on the VACR system during the in-lab

tests. In the meantime, different variable outdoor temperature patterns are imposed on

the VACR system using the environmental chamber. As such, the functionality of

proactive and optimal model predictive controller is evaluated and compared to the

common thermostatic-based controller. In this section, the results of a sample test are

presented and discussed. The repeated transient tests for other conditions and patterns

show similar trends and potentials.

Figure 6-18 and Figure 6-19 show the imposed time-dependent heater power

and outdoor temperature variations for the sample transient test. The results of this test

for thermostatically controlled VACR system are presented in Figure 6-20 to Figure 6-23.

The low and high set points for conditioned space temperature during the thermostatic-

based test are 25.7 and 27.7 oC respectively (based on ANSI/AHRI 210/240-2008 with ±

1 oC allowance). Figure 6-24 to Figure 6-27 represent the results of same transient test

on proactively and optimally controlled system. It should be mentioned that all the results

except the speed ratios (Figure 6-20 and Figure 6-24) are from measurements.

However, due to large number of data points in each graph the error bars are not shown

for a clearer vision. The following are concluded from the obtained results:

o The thermostatically controlled system cycles on-off with higher frequencies at

lower partial loads and keeps the temperature of conditioned space within the ± 1 oC range of the desired temperature (see Figure 6-20 and Figure 6-21). However,

the proactively and optimally controlled system keeps the temperature nearly

constant (less than 0.3 oC variation) under the imposed transient condition (see

Figure 6-25). The optimal controller adjusts the compressor and fans speed in a

way to compensate the cooling demand and prevent any temperature rise or

drop (see Figure 6-24). Similar to the steady state study, the repeated tests with

139

different transient conditions show reproducibility and consistency of the

proactive and optimal controller. Therefore, this controller can be reliably

integrated with the VACR system to function desirably under realistic operating

conditions.

o For the presented results, the COP of thermostatically controlled VACR system

varies in a range of 0-3.4 depending upon the imposed instantaneous condition.

The time-averaged COP for this test is 3.1 (averaged only over the “ON” cycles).

However, the COP of proactively and optimally controlled VACR system varies

within 3.4-3.9 with the time-averaged value of 3.7. Accordingly, the proactively

and optimally controlled VACR system shows more than 19% higher COP than

the thermostatically controlled system under same transient condition. Repeating

the tests for other transient conditions shows a range of 18-23% enhancement in

COP of the VACR system when equipped with the proactive and optimal

controller rather than the thermostatic controller.

o Considering the COP enhancement due to replacing the components of original

VACR system (received from the industrial partner of project) with high efficiency

products beside the utilization of proactive and optimal controller, up to 185%

enhancement in the COP under transient operating conditions is achieved in this

project. As such, the in-lab developed proof-of-concept demonstration of high

efficiency VACR system equipped with variable capacity components and

proactive, optimal, model predictive controller can be expanded to transportation

industry (and even all A/C-R systems in different applications) to achieve

significant global energy saving and GHG reduction.

140

Figure 6-18: Heater power setting for transient test

Figure 6-19: Outdoor (ambient) temperature setting for transient test

141

Figure 6-20: Compressor and fans speed ratio variation for transient test, thermostatic control (low set point = 25.7, high set point = 27.7 oC)

Figure 6-21: Indoor (conditioned space) temperature variation for transient test, thermostatic control (low set point = 25.7, high set point = 27.7 oC)

142

Figure 6-22: Power consumption of VACR system for transient test, thermostatic control

Figure 6-23: COP of VACR system for transient test, thermostatic control

143

Figure 6-24: Compressor and fans speed ratio variation for transient test, proactive optimal control

Figure 6-25: Indoor (conditioned space) temperature variation for transient test, proactive optimal control

144

Figure 6-26: Power consumption of VACR system for transient test, proactive optimal control

Figure 6-27: COP of VACR system for transient test, proactive optimal control

145

6.4. Conclusion

In this chapter, the optimization models from previous chapter were integrated

with an existing validated cooling demand simulator and a proactive, optimization-based,

model predictive controller was developed. A LabVIEW interface was designed to collect

all required data for the cooling demand simulator and execute a “.dll” version of it. A

MATLAB code was developed to find the global maximum COP based on genetic

algorithm, which was coupled with the LabVIEW interface later on. At each time step, the

LabVIEW interface executed the cooling demand simulator and wrote the cooling

demand and air temperatures into a “.txt” file. The MATLAB code read these numbers

from the file, executed the GA, found the optimum speed of compressor and fans, and

wrote them into another “.txt” file. The LabVIEW interface then read the speeds from this

file and commanded the compressor and fans accordingly. As such, the performance of

VACR system was proactively and optimally controlled based on model predictions. The

functionality of developed controller was tested under different steady state and transient

conditions that showed a consistent and reliable operation. The performance

characteristics of the built VACR system equipped with the proactive and optimal

controller was tested under different operating conditions. In addition, a thermostatic-

based controller as the most commonly used controller for A/C-R systems was

developed to evaluate the COP enhancement of proactive and optimal controller. The

steady state and transient tests were performed and comparisons showed significant

COP enhancement for proactive and optimal controller in comparison with the

thermostatic controller. The proactive and optimal controller could eliminate the

inefficient on-off cycling of the system’s components and keep the conditioned space

temperature nearly constant under steady or transient working conditions. The results

showed up to 187% COP enhancement of the developed proactively and optimally

controlled VACR system (equipped with high efficiency and variable capacity

components) in comparison with the original thermostatically controlled constant

capacity VACR system received from the industrial partner of the project. Up to 24% of

the COP enhancement was achieved from the developed proactive and optimal

controller. The developed concept and platform can be expanded to any VACR or even

146

stationary A/C-R application to significantly reduce the global energy consumption and

environmental impacts.

147

Chapter 7. Summary and Future Works

7.1. Summary of Thesis

In this thesis, a lab-scale proof-of-concept demonstration of high efficiency

vehicle air conditioning and refrigeration (VACR) system is developed. The developed

high efficiency system is scalable and can be expanded to all transportation and

stationary applications. Due to tremendous global energy consumption and

environmental impacts of air conditioning and refrigeration (A/C-R) systems, any

improvement of these systems’ performance significantly contributes to green and

sustainable development and environmental protection. The work is initiated as

collaboration with two local companies: Cool-It Group, and Saputo Inc., with main focus

on service vehicles; however, the developed concept and platform is expandable to any

size and application.

An in-lab testbed is built in Laboratory for Alternative Energy Conversion (LAEC)

for performance investigation of VACR systems under different operating conditions. The

testbed is first built using the components of a VACR system received from one of the

industrial partners. A number of measuring equipment, an environmental chamber, and

power supplies are utilized in the testbed. A LabVIEW interface is developed for data

collection that is featured with compressor and fans speed control later. The

performance of VACR system is comprehensively investigated using the test results.

The testbed is later upgraded using high efficiency heat exchangers and variable speed

compressor and fans.

148

In addition to the in-lab experimental study, real-time field data is acquired from a

stationary A/C-R system used in commercial refrigeration and VACR systems used in

trailers of food transportation. The real-time thermal and performance characteristics of

these systems are investigated under different duty cycles. The potential energy saving

and GHG reduction is estimated.

Mathematical models are developed for thermal and performance simulation of

VACR systems under steady state and transient operating conditions. The models are

validated using the experimental and field data and used for performance investigations.

The validated results are utilized for development of performance optimization model.

A high efficiency VACR system is built using variable speed compressor and fans

and high efficiency heat exchangers as a platform for proof-of-concept demonstration.

Two metamodels are developed, validated, and used for optimization of COP under

realistic operating conditions. An optimization solver is coded in MATLAB and integrated

with the LabVIEW interface to form an optimal model predictive controller (MPC) for the

built VACR system. A feature of “proactive” response is added to the controller using a

validated existing cooling demand simulator. The proactive and optimal controller is then

tested under different steady state and transient operating conditions that show

significant COP enhancements. The developed proof-of-concept demonstration is

scalable and expandable to any stationary and mobile air conditioning and refrigeration

system. The contributions resulted from this study are listed below:

o Establish the duty cycle and investigate the thermal and performance

characteristics of a pilot commercial A/C-R system, “SEMI-THERM 32

Conference, San Jose, 2016”.

o Establish the duty cycle and investigate the thermal and performance

characteristics of VACR systems theoretically and experimentally, “Submitted to

International Journal of Refrigeration, under review”.

o Develop a new steady state model and study the VACR systems’ thermal and

performance behaviour under steady state conditions. Validate the model using

the collected data in the lab and on-board testing, “Published in Journal of

Energy Conversion and Management, Vol. 98, 2015, pp. 173–183”.

149

o Develop and validate quasi-steady state and transient models for VACR

systems, “Submitted to ASME Journal of Thermal Science and Engineering

Applications, under review”.

o Build a high efficiency, variable capacity VACR system by employing high

efficiency heat exchangers and variable speed compressor and fans, “Published

in International Journal of Heat and Mass Transfer, Vol. 77, 2014, pp. 301-310”,

“Published in Journal of Energy Conversion and Management, Vol. 103, 2015,

pp. 459-467”, “Published in ASME Journal of Heat Transfer, Vol. 137, 2015,

081401”, “Published in AIAA Journal of Thermophysics and Heat Transfer, Vol.

28, 2014, pp. 687-699”, “Submitted to Journal of Green Energy, under review”.

o Develop optimization models based on steady state and transient thermal

models “In preparation to be submitted to Journal of Applied Energy”.

o Develop a proactive and model predictive controller (MPC) for the VACR system

by integrating the optimization model and an existing cooling demand simulator,

“Published in Journal of Applied Energy, Vol. 138, 2015, pp. 496–504”,

“Published in ASME Journal of Thermal Science and Engineering Applications,

Vol. 7, 2015, 041008”, “Published in ASME Journal of Thermal Science and

Engineering Applications, Vol. 7, 2015, 041012”, “Published in International

Journal of Refrigeration, Vol. 57, 2015, pp. 229–238”.

o Develop a proof-of-concept demonstration of a proactively and optimally

controlled high efficiency VACR system that can be integrated in the service

transportation industry, “Published in Journal of Automobile Engineering 19,

2016, 10.1177/0954407016636978”, “In preparation to be submitted to Journal of

Applied Energy”.

7.2. Future works

The following research directions can be considered as the continuation of this

study:

o Implement the developed proactive, optimal, and model predictive concept on

engine-driven VACR systems of trailers and other service vehicles and

150

investigate the real-time performance improvement in such applications. Reduce

the computational costs of model predictive controller by simplifying the models

(e.g. find the dominant terms of optimization model and drop the insignificant

terms).

o Study the effects of using different types of refrigerant on the thermal and

performance characteristics of the VACR systems.

o Develop similar concept and platform for cascade refrigeration used in

commercial systems.

o Extend the developed concept to air conditioning and refrigeration systems in

buildings and integrate it with building energy management systems to achieve

the highest overall energy efficiency.

o Equip the developed controller with additional features to enhance its

comprehensiveness; for instance, defrost module can be added to the controller

to avoid evaporator coil freeze in low temperature applications.

o Develop a similar concept and platform for thermally-driven air conditioning and

refrigeration systems or even heating systems in buildings and industries.

151

References

[1] Wu X., Hu S., and Mo S., 2013, “Carbon footprint model for evaluating the global warming impact of food transport refrigeration systems,” J. Clean. Prod., 54, pp. 115–124.

[2] WMO, and UNEP, 2013, Report on climate change 2013; the physical science basis.

[3] International Energy Agency, 2013, Report on CO2 emissions from fuel combustion.

[4] Haass R., Dittmer P., Veigt M., and Lütjen M., 2015, “Reducing food losses and carbon emission by using autonomous control – A simulation study of the intelligent container,” Int. J. Prod. Econ., 164, pp. 400–408.

[5] Fitzgerald W. B., Howitt O. J. a., Smith I. J., and Hume A., 2011, “Energy use of integral refrigerated containers in maritime transportation,” Energy Policy, 39(4), pp. 1885–1896.

[6] Jul M., 1985, “Refrigeration and world food requirements,” Int. J. Refrig., 8, pp. 6–9.

[7] UNEP, 2010, Report on the refrigeration, air conditioning and heat pumps.

[8] 2014, U.S. Energy Information Administration, Annual energy outlook 2014 with projections to 2040.

[9] Li J., and Xu S., 2013, “The performance of absorption–compression hybrid refrigeration driven by waste heat and power from coach engine,” Appl. Therm. Eng., 61(2), pp. 747–755.

[10] Johnson V. H., 2002, “Fuel used for vehicles air conditioning: a state by state thermal comfort approach,” SAE Paper 2002-01-1957.

152

[11] Khayyam H., Nahavandi S., Hu E., Kouzani A., Chonka A., Abawajy J., Marano V., and Davis S., 2011, “Intelligent energy management control of vehicle air conditioning via look-ahead system,” Appl. Therm. Eng., 31(16), pp. 3147–3160.

[12] Lambert M. A., and Jones B. J., 2006, “Automotive adsorption air conditioner powered by exhaust heat. Part 1: conceptual and embodiment design,” Automob. Eng., 220, pp. 959–972.

[13] Bejan A., 2006, Advanced Engineering Thermodynamics, John Wiley & Sons.

[14] Cengel Y., and Boles M., 2014, Thermodynamics: An Engineering Approach, McGraw-Hill.

[15] Park K.-J., and Jung D., 2007, “Thermodynamic performance of HCFC22 alternative refrigerants for residential air-conditioning applications,” Energy Build., 39(6), pp. 675–680.

[16] Wang S., Gu J., Dickson T., Dexter J., and McGregor I., 2005, “Vapor quality and performance of an automotive air conditioning system,” Exp. Therm. Fluid Sci., 30(1), pp. 59–66.

[17] Macagnan M. H., Copetti J. B., Souza R. B., Reichert R. K., and Amaro M., 2013, “Analysis of the influence of refrigerant charge and compressor duty cycle in an automotive air conditioning system,” 22nd International Congress of Mechanical Engineering (COBEM 2013), pp. 6151–6161.

[18] Alklaibi a. M., 2008, “Evaluating the possible configurations of incorporating the loop heat pipe into the air-conditioning systems,” Int. J. Refrig., 31(5), pp. 807–815.

[19] Tassou S. a., De-Lille G., and Ge Y. T., 2009, “Food transport refrigeration – Approaches to reduce energy consumption and environmental impacts of road transport,” Appl. Therm. Eng., 29(8-9), pp. 1467–1477.

[20] Ahmed M., Meade O., and Medina M. a., 2010, “Reducing heat transfer across the insulated walls of refrigerated truck trailers by the application of phase change materials,” Energy Convers. Manag., 51(3), pp. 383–392.

[21] Abban S., 2013, “Microbial behaviour and cross contamination between cargoes in containerized transportation of food,” PhD Thesis, University of Copenhagen.

[22] Hill H., 2008, Report on food miles : background and marketing.

153

[23] James S. J., James C., and Evans J. A., 2006, “Modelling of food transportation systems – a review,” Int. J. Refrig., 29(6), pp. 947–957.

[24] Afram A., and Janabi-Sharifi F., 2014, “Review of modeling methods for HVAC systems,” Appl. Therm. Eng., 67(1-2), pp. 507–519.

[25] Anand S., Gupta a., and Tyagi S. K., 2013, “Simulation studies of refrigeration cycles: A review,” Renew. Sustain. Energy Rev., 17, pp. 260–277.

[26] Ding G., 2007, “Recent developments in simulation techniques for vapour-compression refrigeration systems,” Int. J. Refrig., 30(7), pp. 1119–1133.

[27] Hermes C. J. L., Melo C., Knabben F. T., and Gonçalves J. M., 2009, “Prediction of the energy consumption of household refrigerators and freezers via steady-state simulation,” Appl. Energy, 86(7-8), pp. 1311–1319.

[28] Boeng J., and Melo C., 2014, “Mapping the energy consumption of household refrigerators by varying the refrigerant charge and the expansion restriction,” Int. J. Refrig., 41, pp. 37–44.

[29] Lee G. H., and Yoo J. Y., 2000, “Performance analysis and simulation of automobile air conditioning system,” Int. J. Refrig., 23, pp. 243–254.

[30] Cho H., Lee H., and Park C., 2013, “Performance characteristics of an automobile air conditioning system with internal heat exchanger using refrigerant R1234yf,” Appl. Therm. Eng., 61(2), pp. 563–569.

[31] Joudi K. a., Mohammed A. S. K., and Aljanabi M. K., 2003, “Experimental and computer performance study of an automotive air conditioning system with alternative refrigerants,” Energy Convers. Manag., 44(18), pp. 2959–2976.

[32] Han X. H., Li P., Xu Y. J., Zhang Y. J., Wang Q., and Chen G. M., 2013, “Cycle performances of the mixture HFC-161 + HFC-134a as the substitution of HFC-134a in automotive air conditioning systems,” Int. J. Refrig., 36(3), pp. 913–920.

[33] Yoo S. Y., and Lee D. W., 2009, “Experimental study on performance of automotive air conditioning system using R-152a refrigerant,” Int. J. Automot. Technol., 10(3), pp. 313–320.

[34] Brown J. S., Yana-motta S. F., and Domanski P. A., 2002, “Comparitive analysis of an automotive air conditioning systems operating with CO2 and R134a,” Int. J. Refrig., 25, pp. 19–32.

154

[35] Ratts E. B., and Brown J. S., 2000, “An experimental analysis of cycling in an automotive air conditioning system,” Appl. Therm. Eng., 20(11), pp. 1039–1058.

[36] Jabardo J. M. S., Mamani W. G., and Ianella M. R., 2002, “Modeling and experimental evaluation of an automotive air conditioning system with a variable capacity compressor,” Int. J. Refrig., 25, pp. 1157–1172.

[37] Alkan A., and Hosoz M., 2010, “Comparative performance of an automotive air conditioning system using fixed and variable capacity compressors,” Int. J. Refrig., 33(3), pp. 487–495.

[38] Tian C., and Li X., 2005, “Numerical simulation on performance band of automotive air conditioning system with a variable displacement compressor,” Energy Convers. Manag., 46(17), pp. 2718–2738.

[39] Tian C., and Li X., 2005, “Transient behavior evaluation of an automotive air conditioning system with a variable displacement compressor,” Appl. Therm. Eng., 25(13), pp. 1922–1948.

[40] Al-Rashed A., 2011, “Effect of evaporator temperature on vapor compression refrigeration system,” Alexandria Eng. J., 50(4), pp. 283–290.

[41] Green T., Izadi-Zamanabadi R., Razavi-Far R., and Niemann H., 2014, “Plant-wide dynamic and static optimisation of supermarket refrigeration systems,” Int. J. Refrig., 38, pp. 106–117.

[42] Sahin B., and Kodal A., 2002, “Thermoeconomic optimization of a two stage combined refrigeration system: a finite-time approach,” Int. J. Refrig., 25, pp. 872–877.

[43] Sharafian A., and Bahrami M., 2014, “Assessment of adsorber bed designs in waste-heat driven adsorption cooling systems for vehicle air conditioning and refrigeration,” Renew. Sustain. Energy Rev., 30, pp. 440–451.

[44] Rêgo a. T., Hanriot S. M., Oliveira a. F., Brito P., and Rêgo T. F. U., 2014, “Automotive exhaust gas flow control for an ammonia–water absorption refrigeration system,” Appl. Therm. Eng., 64(1-2), pp. 101–107.

[45] Zheng W., Chen Y., Hua N., Zhong T., and Gong Y., 2014, “Comparative performance of an automotive air conditioning system using micro-channel condensers with and cithout liquid-vapor separation,” Energy Procedia, 61, pp. 1646–1649.

155

[46] Muller C., “New reversible air-conditioning magnetocaloric system, environmentally friendly and highly energy efficient,” SAE Tech. Pap. no. 2009-01-0313.

[47] D G., 2009, “Theoretical analysis of the steam pressure exchange ejector for an automotive air conditioning application,” University of Vermont.

[48] Ünal Ş., and Yilmaz T., 2015, “Thermodynamic analysis of the two-phase ejector air-conditioning system for buses,” Appl. Therm. Eng.

[49] Jolly P. G., Tso C. P., Wong Y. W., and Ng S. M., 2000, “Simulation and measurement on the full-load performance of a refrigeration system in a shipping container,” Int. J. Refrig., 23, pp. 112–126.

[50] Tso C. P., Wong Y. W., Jolly P. G., and Ng S. M., 2001, “A comparison of hot-gas by-pass and suction modulation method for partial load control in refrigerated shipping containers,” Int. J. Refrig., 24, pp. 544–553.

[51] Moureh J., Tapsoba S., Derens E., and Flick D., 2009, “Air velocity characteristics within vented pallets loaded in a refrigerated vehicle with and without air ducts,” Int. J. Refrig., 32(2), pp. 220–234.

[52] Moureh J., and Flick D., 2004, “Airflow pattern and temperature distribution in a typical refrigerated truck configuration loaded with pallets,” Int. J. Refrig., 27(5), pp. 464–474.

[53] Defraeye T., Cronjé P., Verboven P., Linus U., and Nicolai B., 2015, “Exploring ambient loading of citrus fruit into reefer containers for cooling during marine transport using computational fluid dynamics,” Postharvest Biol. Technol., 108, pp. 91–101.

[54] Rodríguez-Bermejo J., Barreiro P., Robla J. I., and Ruiz-García L., 2007, “Thermal study of a transport container,” J. Food Eng., 80(2), pp. 517–527.

[55] Spence S. W. T., John Doran W., and Artt D. W., 2004, “Design, construction and testing of an air-cycle refrigeration system for road transport,” Int. J. Refrig., 27(5), pp. 503–510.

[56] Khayyam H., Kouzani A. Z., Hu E. J., and Nahavandi S., 2011, “Coordinated energy management of vehicle air conditioning system,” Appl. Therm. Eng., 31(5), pp. 750–764.

156

[57] Buzelin L. O. S., Amico S. C., Vargas J. V. C., and Parise J. a. R., 2005, “Experimental development of an intelligent refrigeration system,” Int. J. Refrig., 28(2), pp. 165–175.

[58] Khayyam H., Abawajy J., and Jazar R. N., 2012, “Intelligent energy management control of vehicle air conditioning system coupled with engine,” Appl. Therm. Eng., 48, pp. 211–224.

[59] Thompson R., and Dexter A., 2005, “A fuzzy decision-making approach to temperature control in air-conditioning systems,” Control Eng. Pract., 13(6), pp. 689–698.

[60] Calvino F., La Gennusa M., Rizzo G., and Scaccianoce G., 2004, “The control of indoor thermal comfort conditions: introducing a fuzzy adaptive controller,” Energy Build., 36(2), pp. 97–102.

[61] Durovic Z. M., and Kovadevic B. D., 2004, “Control of heating, ventilation and air conditioning system based on neural network,” 7th Seminar on Neural Network Applications in Electrical Engineering, Ieee, pp. 37–38.

[62] Schalbart P., Leducq D., and Alvarez G., 2014, “Ice-cream storage energy efficiency with model predictive control of a refrigeration system coupled to a PCM tank,” Int. J. Refrig., pp. 1–11.

[63] Hovgaard T. G., Larsen L. F. S., Edlund K., and Jørgensen J. B., 2012, “Model predictive control technologies for efficient and flexible power consumption in refrigeration systems,” Energy, 44(1), pp. 105–116.

[64] Lawrence Ricker N., 2010, “Predictive hybrid control of the supermarket refrigeration benchmark process,” Control Eng. Pract., 18(6), pp. 608–617.

[65] Yeh T.-J., Chen Y.-J., Hwang W.-Y., and Lin J.-L., 2009, “Incorporating fan control into air-conditioning systems to improve energy efficiency and transient response,” Appl. Therm. Eng., 29(10), pp. 1955–1964.

[66] Lin J.-L., and Yeh T.-J., 2007, “Modeling, identification and control of air-conditioning systems,” Int. J. Refrig., 30(2), pp. 209–220.

[67] Lin J.-L., and Yeh T.-J., 2007, “Identification and control of multi-evaporator air-conditioning systems,” Int. J. Refrig., 30(8), pp. 1374–1385.

157

[68] Hovgaard T. G., Larsen L. F. S., Skovrup M. J., and Jørgensen J. B., 2011, “Power consumption in refrigeration systems - modeling for optimization,” International Symposium on Advanced Control of Industrial Processes, pp. 234–239.

[69] Sørensen K. K., Stoustrup J., and Bak T., 2015, “Adaptive MPC for a reefer container,” Control Eng. Pract., 44, pp. 55–64.

[70] Bagheri F., Fayazbakhsh M. A., Cheppudira Thimmaiah P., and Bahrami M., 2015, “Theoretical and experimental investigation into anti-idling A/C system for trucks,” Energy Convers. Manag., 98, pp. 173–183.

[71] Ge Y. T., and Tassou S. A., 2011, “Performance evaluation and optimal design of supermarket refrigeration systems with supermarket model ‘SuperSim’, Part I: Model description and validation,” Int. J. Refrig., 34(2), pp. 527–539.

[72] Sanaye S., and Malekmohammadi H. ., 2004, “Thermal and economical optimization of air conditioning units with vapor compression refrigeration system,” Appl. Therm. Eng., 24(13), pp. 1807–1825.

[73] Shah R. K., and Sekulic D. P., 2003, Fundamentals of heat exchanger design, John Wiley & Sons.

[74] Zhao L., Cai W., Ding X., and Chang W., 2013, “Model-based optimization for vapor compression refrigeration cycle,” Energy, 55, pp. 392–402.

[75] Dupont, 2013, Report on thermodynamic properties of HFC-134a.

[76] Abbasbandy S., 2003, “Improving Newton–Raphson method for nonlinear equations by modified Adomian decomposition method,” Appl. Math. Comput., 145(2-3), pp. 887–893.

[77] Corberan J. M., Finn D. P., Montagud C. M., Murphy F. T., and Edwards K. C., 2011, “A quasi-steady state mathematical model of an integrated ground source heat pump for building space control,” Energy Build., 43(1), pp. 82–92.

[78] Borges B. N., Hermes C. J. L., Gonçalves J. M., and Melo C., 2011, “Transient simulation of household refrigerators: A semi-empirical quasi-steady approach,” Appl. Energy, 88(3), pp. 748–754.

[79] Li B., and Alleyne A. G., 2010, “A dynamic model of a vapor compression cycle with shut-down and start-up operations,” Int. J. Refrig., 33(3), pp. 538–552.

158

[80] Liang N., Shao S., Tian C., and Yan Y., 2010, “Dynamic simulation of variable capacity refrigeration systems under abnormal conditions,” Appl. Therm. Eng., 30(10), pp. 1205–1214.

[81] Liang N., Shao S., Tian C., and Yan Y., 2010, “Dynamic simulation of variable capacity refrigeration systems under abnormal conditions,” Appl. Therm. Eng., 30(10), pp. 1205–1214.

[82] Wang F. Q., Maidment G. G., Missenden J. F., and Tozer R. M., 2007, “A novel special distributed method for dynamic refrigeration system simulation,” Int. J. Refrig., 30(5), pp. 887–903.

[83] Browne M. W., and Bansal P. K., 2002, “Transient simulation of vapour-compression packaged liquid chillers,” Int. J. Refrig., 25, pp. 597–610.

[84] Beatty K., and Katz D., 1948, “Condensation of vapours on the outside of finned tube,” Chemical Engineering Progress.

[85] Koury R. N. N., Machado L., and Ismail K. A. R., 2001, “Numerical simulation of a variable speed refrigeration system,” Int. J. Refrig., 24, pp. 192–200.

[86] Gaspar P. D., Carrilho Gonçalves L. C., and Pitarma R. a., 2011, “Experimental analysis of the thermal entrainment factor of air curtains in vertical open display cabinets for different ambient air conditions,” Appl. Therm. Eng., 31(5), pp. 961–969.

[87] Yang L., and Zhang C.-L., 2010, “Analysis on energy saving potential of integrated supermarket HVAC and refrigeration systems using multiple subcoolers,” Energy Build., 42(2), pp. 251–258.

[88] Pasific Gas and Electric Company, 2013, Supermarket refrigeration; 2013 California building energy efficiency standards.

[89] D. Cowan, J. Gartshore, I. Chaer, C. Francis G. M., 2009, “REAL Zero – reducing refrigerant emissions and leakage - feedback from the IOR Project,” Proc. Inst. Refrig.

[90] Tassou S. a., Ge Y., Hadawey a., and Marriott D., 2011, “Energy consumption and conservation in food retailing,” Appl. Therm. Eng., 31(2-3), pp. 147–156.

159

[91] Kullheim J., 2011, “Field measurement and evaluation of CO2 refrigeration systems for supermarket,” KTH, School of Industrial Engineering and Management.

[92] Carrier Transicold, 2010, Ultra XT/Extra XT unit catalog.

[93] Carrier Transicold, 2013, Vector 8600 MT unit catalog.

[94] Carrier Transicold, 2010, Trailer and rail refrigeration-service parts list.

[95] Carrier Transicold, 2012, Truck regrigeration unit-operation and service.

[96] Zhao L., Cai W.-J., Ding X., and Chang W., 2013, “Decentralized optimization for vapor compression refrigeration cycle,” Appl. Therm. Eng., 51(1-2), pp. 753–763.

[97] Brodd R. J., 2012, Batteries for Sustainability, Springer.

[98] Tahil W., 2010, Report: "How much lithium does a LiIon EV battery really need ?"

[99] Galos J., Sutcliffe M., Cebon D., Piecyk M., and Greening P., 2015, “Reducing the energy consumption of heavy goods vehicles through the application of lightweight trailers: Fleet case studies,” Transp. Res. Part D Transp. Environ., 41, pp. 40–49.

[100] Casadei A., and Broda R., 2008, Report on impact of vehicle weight reduction on fuel economy for various vehicle architectures.

[101] “http://www.inboundlogistics.com/cms/article/smarter-trucking-saves-fuel-over-the-long-haul.”

[102] “http://www.nrcan.gc.ca/energy/efficiency/transportation/commercial-vehicles/reports/7607.”

[103] “http://www.eia.gov/tools/faqs/faq.cfm?id=307&t=11.”

[104] “http://people.exeter.ac.uk/TWDavies/energy_conversion/Calculation%20of%20CO2%20emissions%20from%20fuels.htm.”

[105] Arora J., 2004, Introduction to Optimum Design, Elsevier Press.

160

[106] Green T., Izadi-zamanabadi R., and Razavi-far R., 2014, “Plant-wide dynamic and static optimisation of supermarket refrigeration systems,” Int. J. Refrig., 38, pp. 106–117.

[107] Wang B., Shi W., Han L., and Li X., 2009, “Optimization of refrigeration system with gas-injected scroll compressor,” Int. J. Refrig., 32(7), pp. 1544–1554.

[108] Yen R. H., Huang B. J., Chen C. Y., Shiu T. Y., Cheng C. W., Chen S. S., and Shestopalov K., 2013, “Performance optimization for a variable throat ejector in a solar refrigeration system,” Int. J. Refrig., 36, pp. 1512–1520.

[109] Turkakar Goker, and Tuba O.-O., 2015, “Entropy generation analysis and dimensional optimization of an evaporator for use in a microscale refrigeration cycle,” Int. J. Refrig., 56, pp. 140–153.

[110] Ge Y. T., Tassou S. a., Santosa I. D., and Tsamos K., 2015, “Design optimisation of CO2 gas cooler/condenser in a refrigeration system,” Appl. Energy, 160, pp. 973–981.

[111] Selbas R., Kizilkan O., and Sencan a, 2006, “Thermoeconomic optimization of subcooled and superheated vapor compression refrigeration cycle,” Energy, 31(12), pp. 2108–2128.

[112] Zhao L., Cai W., Ding X., and Chang W., 2013, “Model-based optimization for vapor compression refrigeration cycle,” Energy, 55, pp. 392–402.

[113] Xu Y.-C., and Chen Q., 2013, “A theoretical global optimization method for vapor-compression refrigeration systems based on entransy theory,” Energy, 60, pp. 464–473.

[114] Aminyavari M., Najafi B., Shirazi A., and Rinaldi F., 2014, “Exergetic , economic and environmental ( 3E ) analyses , and multi- objective optimization of a CO 2 / NH 3 cascade refrigeration system,” Appl. Therm. Eng., 65, pp. 42–50.

[115] Hao J.-H., Chen Q., and Xu Y.-C., 2014, “A global optimization method for regenerative air refrigeration systems,” Appl. Therm. Eng., 65(1-2), pp. 255–261.

[116] Ghorbani B., Ma M., and Shirmohammadi R., 2014, “Optimization of operation parameters of refrigeration cycle using particle swarm and NLP techniques,” Nat. Gas Sci. Eng., 21, pp. 779–790.

161

[117] Wang G., and Shan S., 2007, “Review of metamodeling techniques in support of engineering design optimization,” J. Mech. Des., 129, pp. 370–380.

[118] Simpson T. W., Peplinski J. D., Koch P. N., and Allen J. K., 1997, “On the use of statistics in design and the implications for deterministic computer experiments,” ASME Design Engineering, pp. 1–14.

[119] Haji Hajikolaei K., and Wang G., 2014, “High dimensional model representation with principal component analysis,” ASME J. Mech. Des., 136.

[120] Fayazbakhsh M. A., 2015, “Intelligent Calculation of Thermal Loads in Mobile and Stationary Heating, Ventilation, Air Conditioning, and Refrigeration Systems,” PhD Thesis, Simon Fraser University.

[121] ASHRAE, 2009, ASHRAE Handbook-Fundamentals.