Ed. - Michael Kauffeld .... - Handbook On Ice Slurries - Fundamentals and Engineering PDF

Handbook on
Ice Slurries
– Fundamentals and Engineering –
Editors:
Michael Kauffeld, Masahiro Kawaji, Peter W. Egolf
Extensively reviewed by
Åke Melinder, Tom W. Davies
Published by
INTERNATIONAL INSTITUTE OF REFRIGERATION

177 boulevard Malesherbes – 75017 Paris, France
Tel.: +33 1 4227 3235 – Fax: +33 1 4763 1798
E-mail: iifiir@iifiir.org – Web site : www.iifiir.org
ISBN n° 2-913149-42-1
Copyright and Disclaimer

“The information provided in this document is based on the current state of art and is designed to assist
engineers, scientists, companies and other organizations. It is a preliminary source of information that will need
to be complemented prior to any detailed application or project. Whilst all possible care has been taken in the
production of this document, the IIR, its employees, officers and experts cannot accept any liability for the
accuracy or correctness of the information provided nor for the consequences of its use or misuse.
For full or partial reproduction of anything published in this document, proper acknowledgement should be made
to the original source and its author(s). No parts of the book may be commercially reproduced, recorded and
stored in a retrieval system or transmitted in any form or by any means (mechanical, electrostatic, magnetic,
optic photographic, multimedia, Internet-based or otherwise) without permission in writing from the IIR.
© Copyright 2005 International Institute of Refrigeration

IIF-IIR – Handbook on Ice Slurries – 2005
— Table of contents —
Foreword ............................................................................................................................... 9
Preface .................................................................................................................................. 10
1. Summary of Ice Slurry Technology .................................................................... 11

1.1 Introduction ............................................................................................................. 11
1.2 Existing technologies .............................................................................................. 11
1.2.1 Heterogeneous nucleation ............................................................................. 12
1.2.2 Homogeneous or spontaneous nucleation ..................................................... 12
1.3 Modern ice slurry applications ................................................................................ 12
1.4 International Institute of Refrigeration (IIR) Working Party on Ice Slurries ...... 13
1.4.1 Scientific challenges...................................................................................... 13
1.4.2 Technical challenges ..................................................................................... 14
1.4.3 Objectives of the working party .................................................................... 14
1.5 Research institutes.................................................................................................... 15
1.6 Conclusions and outlook on the use of ice slurry ................................................... 15
1.7 Other Phase Change Slurries (PCSs)...................................................................... 15
1.7.1 Dry ice/carbon dioxide slurry........................................................................ 16
1.7.2 Clathrate slurry.............................................................................................. 16
1.7.3 Micro-emulsion slurry................................................................................... 17
1.7.4 Shape-stabilized PCM slurry......................................................................... 17
1.7.5 Microencapsulated PCM slurry..................................................................... 17
1.7.6 Polyethylene pellets....................................................................................... 18
Literature cited in Chapter 1........................................................................................... 18
2. Ice Creation, Thermophysical Properties of Ice Slurries

and Other Characteristics ...................................................................................... 19
2.1 The formation of ice.................................................................................................. 19
2.1.1 Introduction ................................................................................................... 19
2.1.2 Metastable state – supercooled water............................................................ 19
2.1.3 Nucleation ..................................................................................................... 19
2.1.4 Effect of additives ......................................................................................... 24
2.2 Crystal growth ........................................................................................................... 24
2.2.1 Initial crystal growth ..................................................................................... 24
2.2.2 Further processes affecting the crystal size................................................... 26
2.3 Types of ice crystals and ice particles...................................................................... 29
2.3.1 Shape parameters........................................................................................... 30
2.3.2. Ice particle model .......................................................................................... 32
2.3.3 Measurements and statistics .......................................................................... 33
2.4 Properties of aqueous solutions and ice ................................................................... 33
2.4.1 Introduction ................................................................................................... 33
2.4.2 Thermophysical properties of aqueous solutions .......................................... 34
2.4.3 Freezing-point temperature of aqueous solutions ......................................... 36
2.4.4 Thermophysical properties of ice.................................................................. 36
2.5 Physical properties of ice slurry .............................................................................. 37
2.5.1 Freezing point temperature and ice concentration ........................................ 37
3
2.5.2 Viscosity ....................................................................................................... 39

2.5.3 Thermal conductivity ................................................................................... 40
2.5.4 Enthalpy and enthalpy phase diagrams of aqueous solutions ....................... 41
2.6 Safety, ecology and corrosion aspects of aqueous solutions ................................. 48
2.6.1 General characteristics of additives used as aqueous solutions .................... 48
2.6.2 General explanation of symbols ................................................................... 48
2.6.3 Comments on each type of aqueous solution ............................................... 49
2.6.4 Corrosion behaviour and material compatibility ......................................... 50
Literature cited in Chapter 2 ......................................................................................... 50
3. Fluid Dynamics ......................................................................................................... 55

3.1 Rheology .................................................................................................................. 55
3.1.1 State of the art of ice slurry rheology ........................................................... 59
3.2 Flows in tubes and pattern formation .................................................................... 60
3.2.1 State of the art of research on flow patterns.................................................. 61
3.2.2 Velocity profiles ........................................................................................... 64
3.2.3 Concentration profiles ................................................................................... 66
3.2.4 Critical deposition velocity ............................................................................ 67
3.2.5 Theoretical determination of flow patterns .................................................. 70
3.3 Pressure drop .......................................................................................................... 70
3.3.1 Isothermal, stationary, homogeneous ice-slurry flow .................................. 71
3.3.2 State of the art of ice-slurry pressure drops................................................... 73
3.3.3 Friction factor for laminar flow .................................................................... 74
3.3.4 Semi-empirical correlations for the turbulent ice-slurry friction factor........ 74
3.3.5 General empirical determination of the friction factor for ice-slurries ......... 75
3.3.6 Pressure drops of ice-slurry flows in heat exchangers ................................. 77
3.4 Time-dependent behaviour ..................................................................................... 78
Literature cited in Chapter 3........................................................................................... 79
4. Thermodynamics and Heat Transfer ................................................................. 83

4.1 Overview .................................................................................................................. 83
4.2 Basic definitions ...................................................................................................... 83
4.2.1 Effective enthalpy and effective specific heat .............................................. 83
4.2.2 Effective thermal conductivity ..................................................................... 84
4.2.3 Dimensionless numbers ................................................................................ 85
4.2.4 Reference temperatures and heat transfer coefficients ................................. 86
4.3 Measurement techniques ........................................................................................ 87
4.3.1 Introductory remarks .................................................................................... 87
4.3.2 Local ice fraction .......................................................................................... 87
4.3.3 Local ice slurry temperature ......................................................................... 88
4.3.4 Local ice slurry velocity ............................................................................... 88
4.4 Flow phenomena ..................................................................................................... 89
4.4.1 Velocity profile, mean velocity .................................................................... 89
4.4.2 Temperature profile, mean temperature in vertical channels ....................... 92
4.4.3 Local ice fraction profiles ............................................................................ 93
4.5 Heat transfer phenomena ....................................................................................... 94
4.5.1 A brief introduction ...................................................................................... 94
4
4.5.2 Pipes ............................................................................................................. 95

4.5.3 Heat transfer in rectangular channels ......................................................... 105
4.5.4 Heat transfer in rectangular channels with a bend ..................................... 113
4.6 Heat exchangers with stagnant flow .................................................................... 115
4.7 Industrial heat exchangers .................................................................................... 117
4.7.1 Brief introduction ....................................................................................... 117
4.7.2 Heat transfer correlations in compact heat exchangers .............................. 117
4.8 Conclusions and outlook ...................................................................................... 120
Literature cited in Chapter 4 ........................................................................................ 121
5. Ice Slurry Production ........................................................................................... 125

5.1 Fundamentals of ice slurry generation ................................................................. 125
5.1.1 Ice crystallization ........................................................................................... 125
5.1.2 Heat and mass transfer aspects ..................................................................... 126
5.1.3 Categories of ice slurry generator ................................................................. 126
5.2 Current and possible methods of ice slurry production ....................................... 128
5.2.1 Scraped surface ice slurry generator ............................................................. 128
5.2.2 Vacuum ice ................................................................................................... 137
5.2.3 Direct contact generators with immiscible refrigerant ................................. 139
5.2.4 Direct contact generators with immiscible liquid secondary refrigerant ...... 142
5.2.5 Supercooled brine method ............................................................................ 143
5.2.6 Hydro-scraped ice slurry generator .............................................................. 146
5.2.7 Special coating of generator surface to avoid ice sticking to surface ........... 147
5.2.8 Fluidized bed crystallizer .............................................................................. 152
5.2.9 High pressure ice slurry generator ................................................................ 154
5.2.10 Recuperative ice making ............................................................................ 158
6. Transport .................................................................................................................. 167

6.1 Advantage of transport characteristics of ice slurries
over conventional fluids.......................................................................................... 167
6.2 Design of transport systems ................................................................................... 167
6.2.1 General considerations................................................................................... 167
6.2.2 Single-tube versus dual-tube systems ........................................................... 168
6.2.3 Considerations of defrosting ......................................................................... 168
6.2.4 Pipe Materials ............................................................................................... 170
6.2.5 Calculation of pressure drops ........................................................................ 170
6.2.6 Avoidance and influence of phase separation ............................................... 170
6.3 Transport capacity ................................................................................................. 174
6.3.1 Calculation of transport capacity .................................................................. 174
6.3.2 Pipe selection chart ....................................................................................... 175
6.4 Pumping ................................................................................................................. 176
6.4.1 Centrifugal pumps and their application ranges ........................................... 176
6.4.2 Other pumps for ice slurry ............................................................................ 181
6.4.3 Calculation guidelines using manufacturer’s data ........................................ 185
6.4.4 Long term performance test .......................................................................... 189
6.4.5 Internal build up of ice clusters in centrifugal pumps ................................... 190
6.4.6 Corrosion and Erosion .................................................................................. 191
5
6.4.7 Problems at standstill .................................................................................... 191

6.5 Fittings and related pressure loss coefficients ..................................................... 192
6.5.1 Test rig set up ................................................................................................ 192
6.5.2 Design of pressure taps ................................................................................. 193
6.5.3 Data treatment ............................................................................................... 193
6.5.4 Experiments and results ................................................................................ 195
6.5.5 Comparison of measured values of ice slurries and single phase fluids ....... 198
6.6 Valves ..................................................................................................................... 199
6.7 Insulation ............................................................................................................... 199
6.8 Critical conditions ................................................................................................. 199
6.9 Quick guide to ice slurry installation and operation in order
to avoid blockage .................................................................................................... 200
6.9.1 Homogeneously mixed storage ...................................................................... 200
6.9.2 Transport system ........................................................................................... 200
7. Storing/Melting and Mixing ............................................................................... 203

7.1 Storing vessels ....................................................................................................... 203
7.1.1 Storing methods/devices and processes ........................................................ 203
7.1.2 Melting methods/devices and processes ....................................................... 206
7.2 Mixing tanks .......................................................................................................... 207
7.2.1 Mixing methods/devices and processes ........................................................ 207
7.2.2 Stratification in mixing tanks ....................................................................... 208
7.2.3 Power consumption in Newtonian range ...................................................... 209
7.2.4 Power consumption for highly-viscous range .............................................. 209
7.3 Numerical modelling for storage tank design ...................................................... 209
7.3.1 Physical modelling of accumulation/piling and melting ............................... 209
7.3.2 Mathematical modelling ............................................................................... 211
7.3.3 Confirmation of modelling ............................................................................ 215
7.4 Numerical modelling for mixing tank design ...................................................... 219
7.4.1 Physical modelling of flow and stratification ............................................... 219
7.4.2 Mathematical modelling ............................................................................... 220
7.4.3 Confirmation of modelling ............................................................................ 220
8. Melting Ice Slurry in Heat Exchangers ........................................................... 223

8.1 Plate heat exchangers ........................................................................................... 223
8.1.1 Test rig set up ............................................................................................. 225
8.1.2 Measuring equipment ................................................................................. 227
8.1.3 Heat transfer results .................................................................................... 227
8.1.4 Correlation for heat transfer ....................................................................... 229
8.1.5 Pressure drop results .................................................................................. 231
8.1.6 Correlation for pressure drop ..................................................................... 232
8.1.7 Observations ................................................................................................ 233
8.1.8 Minimum flow rate .................................................................................... 234
8.2 Air coolers .............................................................................................................. 235
8.2.1 Test rig set up ............................................................................................. 236
8.2.2 Heat transfer results ................................................................................... 237
6
8.2.3 Correlation for heat transfer ....................................................................... 239

8.2.4 Pressure drop results .................................................................................. 239
8.2.5 Correlation for pressure drop ..................................................................... 240
8.2.6 Comparison between single-phase media and ice slurry ........................... 242
8.2.7 Minimum flow rate .................................................................................... 243
8.3 Observations .......................................................................................................... 244
8.4 Conclusions ........................................................................................................... 248
9. Direct Contact Chilling and Freezing of Foods in Ice Slurries .............. 251
9.1 State of the art and conventional modes of food refrigeration ........................... 251
9.2 Unfreezable liquids and pumpable ice slurries as refrigerating media
and fluidizing agents ............................................................................................. 253
9.3 Cooling of fish with ice slurries ............................................................................. 259
9.4 Ice-slurry-based cooling of fruits and vegetables ................................................ 263
10. The Control of Ice Slurry Systems .................................................................... 273

10.1 Introduction ........................................................................................................ 273
10.2 General organization of an ice slurry system .................................................... 273
10.3 What is to be controlled ...................................................................................... 274
10.3.1 Control of consumer units ..................................................................... 274
10.3.2 Control of the storage tank .................................................................... 276
10.3.3 Control of the ice slurry generator ......................................................... 276
10.3.4 Control of the primary refrigerating unit ............................................... 280
10.4 Influence of the temperature and solute concentration
on the control parameters ................................................................................... 280
10.5 Consequences of the uncertainty about the ice concentration
on the control of the system ................................................................................ 283
10.6 Conclusions ......................................................................................................... 285
Literature cited in Chapter 10 ...................................................................................... 285
11. Optimization of Ice Slurry systems ................................................................. 287

11.1 Criteria ................................................................................................................. 287
11.2 Methods ................................................................................................................ 288
12. Present and Future Applications ...................................................................... 291

12.1 Introduction .......................................................................................................... 291
12.2 Comfort cooling .................................................................................................... 291
12.2.1 Comfort cooling of buildings ..................................................................... 291
12.2.2 Applications to office building .................................................................. 292
12.2.3 Applications to institutional buildings ........................................................ 297
12.2.4 Applications to commercial buildings ....................................................... 299
12.2.5 Mine cooling .............................................................................................. 301
7
12.3 Food processing .................................................................................................... 302

12.3.1 Rapid cooling of vegetables ....................................................................... 302
12.3.2 Fish processing ........................................................................................... 302
12.3.3 Meat processing .......................................................................................... 305
12.3.4 Dairy processing ........................................................................................ 305
12.3.5 Brewery ...................................................................................................... 306
12.3.6 Retail food applications ............................................................................. 308
12.4 Possible future applications ................................................................................. 311
12.4.1 Ice pigging .................................................................................................. 311
12.4.2 Medical applications .................................................................................. 311
12.4.3 Artificial snow ............................................................................................ 311
12.4.4 Transport refrigeration ............................................................................... 312
12.4.5 The use of ice slurry for fire fighting ......................................................... 313
12.5 Conclusions .......................................................................................................... 313
13. Conclusions and recommendations for further R&D ............................... 317
Appendix 1: List of authors .................................................................................... 321

Appendix 2: List of selected plants ....................................................................... 325
Appendix 3: List of symbols .................................................................................... 345
Appendix 4: List of figures ...................................................................................... 351
Appendix 5: List of tables ........................................................................................ 361
8
IIF-IIR –Handbook on Ice Slurries – 2005
FOREWORD
Ice slurries are a really promising technology, because of their advantages regarding the
environment, which is the most important challenge ahead of us, along with safety of course.
The selection of the refrigerant for refrigeration equipment is the key problem, particularly
since the Montreal Protocol and the United Nations Convention for Climate Change: some
refrigerants (CFCs, HCFCs) play a role in the depletion of the ozone layer, while others
(HFCs), which are progressively replacing them, unfortunately also have important global
warming potentials even if they do not deplete the ozone layer. The aim then is to reduce the
refrigerant losses from equipment and then to reduce the refrigerant charge. A very efficient
solution is the use of environmentally friendly secondary refrigerants.
Some other refrigerants, such as ammonia or propane, which are environmentally friendly,
need a lot of precautions for human safety: in a same manner, we then can combine a
protected area with that refrigerant and a safe secondary refrigerant.
Another important issue is the energy efficiency of the refrigeration equipment. Energy
consumption accounts for about 80% (on the average) of the global warming impact of
refrigeration and it uses about 15% of the total electricity produced worldwide. The final
decision concerning the selection of the refrigerant regarding the impact on the environment
has to take into account the Life Cycle Climate Performance (LCCP) of the refrigeration
equipment, including the impact of energy consumption on climate.
Ice slurry as a secondary refrigerant is a very interesting solution thanks to the high cooling
capacity given by the latent heat of phase change. It is simple and environmentally friendly.
We now have a lot of examples of use of this technology. However, we still need a lot of
research, particularly on how to generate ice slurry in an efficient, reliable and economic way
for use in a broader range of applications.
The IIR set up its Working Party on Ice Slurries in 1998; this party has actively organized and
continues to organize workshops in order to better improve and to implement this technology.
This book is the result of years of collaboration and experience. It involved nearly 50 IIR
experts and I want to thank all of them for their work, particularly Peter W. Egolf, Michael
Kauffeld and Masahiro Kawaji, for their efficient coordination of such difficult and useful
work.
I hope that this book will be a really useful guide for all the industrial practitioners and
scientists who want to know more on and to implement this technology in the future.
Didier Coulomb
Director of the IIR
9
PREFACE
The role of secondary refrigerants in industrial and commercial refrigeration is growing as the
focus on the reduction of greenhouse gas emissions increases. The transition from distributed
refrigeration plants with large charges of primary refrigerant (ammonia or HCFC /HFC) is
taking place in some countries and more indirect systems with compact chiller units and
secondary refrigeration loops are now being installed. Based on a simple temperature
analysis, the additional heat exchange introduced when using such indirect systems leads to
increased energy consumption of approximately 10-20%. But such comparisons are not
necessarily valid, as they seldom include inefficiencies due to poor superheat control and
defrosting, pressure losses in refrigerant headers and suction lines of distributed direct-
expansion systems, nor do they take into account the greater flexibility of indirect systems.
An indirect system can be improved if phase-change materials1 are applied instead of

traditional single-phase secondary refrigerants. At temperatures below 0°C ice slurry
facilitates such efficiency improvement as the energy consumption for pumps can be reduced
and the required temperature difference in heat exchangers lowered thanks to some very
attractive thermophysical properties of ice slurry.
Although ice slurry has been known since the ancient Roman times, co-ordinated and focused
research activities were launched.
This Handbook summarizes the current knowledge on ice slurry. Approximately 50

international experts have combined their knowledge and jointly created this unique
document. We hope that it will also be useful to other system designers, engineers or
scientists working with ice slurry or interested in using ice slurry for various applications.
Karlsruhe, Toronto and Yverdon, Michael Kauffeld,

December 2004 Masahiro Kawaji and
Peter W. Egolf
1 )
This publication deals only with ice slurry made from purely liquid phase fluids, i.e. without solid particles
other than ice. Hence the phase change-material described in this Handbook utilizes the phase change of water to
and from ice (energy of fusion with H2O-molecules) solely. The authors are well aware of the multitude of other
phase-change materials used in industry and being subject to research especially in the domain of thermal energy
storage. Examples of other phase-change materials (PCMs) are waxes, polymers (sometimes also employing
water as PCM), salts, etc. Even though this publication has been limited to ice slurry alone, similarities to other
phase-change slurries might be found and hence this book might be helpful to those working with other phase-
change materials as well.
10
CHAPTER 1. SUMMARY OF ICE SLURRY TECHNOLOGY

by Michael Kauffeld, Masahiro Kawaji and Peter W. Egolf
1.1 Introduction
The use of ice for prolonging the storage life of foodstuffs dates back many millennia. Up
until the middle of the last century all ice used for cooling was from natural sources (winter
snow/ice or imported Arctic ice). Sometimes the natural snow was mixed with salt in order to
reach lower temperatures. The production of ice cream is known to have been achieved using
this technnique some thousand years ago in ancient Rome.
With the introduction of a mechanical cold production technology, ice was and is still
produced in different forms, e.g. blocks, cubes, tubes or flake ice. Most of these forms of ice
need a certain degree of manual operation for transportation from one place to another, and
have rather sharp edges that may damage the surfaces of products that need to be kept
refrigerated in direct contact with the ice. Furthermore, they are usually quite coarse and show
poor heat transfer characteristics when releasing the latent heat of fusion.
The technical utilization of ice slurry – a mixture of small (typically 0.1 to 1 mm in diameter)
ice crystals/particles 1 and a carrier fluid — a mixture of water and a freezing point depressant
— allows the phase change material, “ice”, to be pumped to the point of application. Heat
transfer characteristics are enhanced as well as product compatibility in the food industry.
Characteristic of ice slurries is that the particles disappear in the melting process and have to
be created again by a special ice slurry generator. In storage tanks the particles experience
large buoyancy forces, which lead to a high degree of stratification of the ice slurry with a
maximum ice content at the top of the tank. Therefore, usually a mixing device, or some other
special equipment, is necessary to create homogeneous ice particle suspensions, which would
guarantee the safe operation of a system without the occurrence of clogging in the tubes. In
recent years, the development of commercial ice slurry systems has enabled the use of ice
slurry in a wide range of applications. The simplicity of freezing water with an
environmentally friendly additive (alcohol, salts, etc.) and obtaining very high enthalpy
densities makes the application of ice slurries a promising technology for the future.
1.2 Existing technologies
The key issue for using ice slurry is the reliable, energy efficient and cost effective (cheap)
production of the ice slurry. Several variations of ice slurry production technologies exist.
They will be described in detail in Chapter 5, but most of the commercial ice slurry generators
can be classified into two categories depending on the mechanism of ice crystal nucleation:
heterogeneous or homogeneous nucleation.
1
The terms “ice crystal” and “ice particle” are used in parallel throughout this publication. In general the smaller
the particles, the more likely they are to be named “ice crystal”, as they usually consist of a homogeneous crystal
matrix. Especially larger particles above 1 mm diameter are usually called “ice particles”.
11
1.2.1 HETEROGENEOUS NUCLEATION
Systems with mechanical scrapers or brushes produce ice slurry based on a heterogeneous
nucleation process. This process is utilized in a refrigerating machine containing a special
evaporator with a double cylinder or plate in which a part of the water/brine mixture is cooled
at the wall. Then the ice crystals are created with a size of some tenths of millimetres,
mechanically scraped away and accumulate in the suspension. Ice slurry generators of this
type comprise the majority of the installed equipment. First prototypes are known from 1976
with related U.S. patents.
An alternative ice removal technique applies fluidized bed technology (Gun, 2001).
New developments aim at avoiding the scraping elements by alternating between different
evaporators (Davies, 2002), coating the surface of the heat exchanger facing the ice slurry
(Zwieg, 2002) or controlling the flow rate as well as the evaporation temperature (Barth,
2002). All of these new techniques utilise some form of hydro-scraping based on the drag
forces associated with the fluid flow.
1.2.2 HOMOGENEOUS OR SPONTANEOUS NUCLEATION
The ice slurry is created directly by injection and evaporation of a cold refrigerant in the basic
brine solution. The gas expansion causes a homogeneous or spontaneous nucleation. The
direct contact of the cold refrigerant and the secondary fluid leads to a high efficiency of these
systems, because no additional heat exchangers (evaporators) are necessary. Because of the
turbulent jet-like injection of the refrigerant, the ice crystals are finely dispersed in the liquid.
Water itself can also be used as the refrigerant in this process. The system is then commonly
called “vacuum ice”.
An alternative way is to use a water-immiscible single-phase heat transfer fluid, which creates
ice in direct contact when mixed with water. The heat transfer fluid is cooled in a separate
refrigeration unit (chiller). Due to the additional heat transfer between the primary refrigerant,
the heat transfer fluid and finally the water/ice, this process is assumed to be less efficient
than the direct contact evaporation process described above.
Yet another method that utilizes homogeneous nucleation is supercooling of water or brine
and subsequent release of supercooling by ultrasonic or mechanical shock.
1.3 Modern ice slurry applications
Ice slurry is used in many countries in various applications. The following trends can be seen
on an international level.
In China and in other developing countries of the Far East ice slurry is used for cooling of
railway cars, where ice slurry fills the voids surrounding the cargo hold. This technology is
similar to the 100-year-old technique of using block ice, but the ice slurry is much easier to
handle.
A similar system has been developed in Germany for catering vessels, which are used in
passenger trains. The vessels are charged with ice slurry at the beginning of a journey and
12
keep the food refrigerated for many hours. So far the system has only been applied in
Switzerland, but other applications are about to follow.
In Japan, ice slurry is mainly used in connection with air conditioning of large buildings. In
most Japanese installations ice slurry is only used as an energy storage medium and is not
pumped directly to the various air handling coils. Such direct application of ice slurry in air-
conditioning systems might be beneficial when combined with low temperature air
distribution systems (tair ≅ 2°C). The low-temperature air would save fan energy, duct size and
hence building height, but the thermal comfort and air quality perceived by occupants in the
building must of course be ensured. Furthermore, condensation risks must be minimized.
Chilean, Dutch and Icelandic fishermen (to mention just the major ones) use ice slurry for
direct chilling of fish and other catches. The ice slurry is produced from seawater on board in
small ice slurry plants. The ice slurry produced is then stored for up to 10 days in the fish hold
or special ice slurry tanks. When the catch is taken on board it is mixed with the ice slurry
(sometimes, drained ice slurry). Several years of practice have shown a prolonged storage life
of such catch as well as a much better quality, and hence better prices for the catches may be
obtained. Moreover, the application of ice slurry is now moving into fish processing factories
onshore.
Several commercial applications (bread, beer, and sausage production) are known in
Germany. Ice slurry is used in air coils (sometimes air coils designed for direct expansion of
HFC-refrigerants). At least one installation in Germany utilises ice slurry for air conditioning.
Two supermarkets with an ice slurry system for the chilling temperature range are operated in
Switzerland. In addition, two large-scale applications were reported: Zürich-Kloten Airport
and a Pharma Park close to Lugano.
Most installations mentioned in Appendix 2 use ice slurry generators of the scraped surface
type. In many of the installations, initial investment costs are higher due to the type of ice
slurry generator. Often their operating costs are similar to those of other refrigeration systems.
Sometimes cost savings can be obtained by reduced energy costs (e.g. ice slurry storage in
Japan) or in the form of a higher value of the chilled product (e.g. ice slurry for direct cooling
of fish).
1.4 International Institute of Refrigeration (IIR) Working Party on Ice Slurries
1.4.1 SCIENTIFIC CHALLENGES
The basic physical properties of ice slurries have not yet been fully investigated. However,
commercial designs of refrigeration systems require that these properties be well understood,
since they form the basis for most design calculations, e.g. the application of melting/freezing
models, the Navier-Stokes equations with rheological models and also for simpler calculation
methods designed for more practical use.
New additives must be evaluated and tested for low-temperature applications, and the
physical properties of the suspensions containing these additives must also be measured and
published.
13
There is still some slight variation in reported heat transfer characteristics, which have been
experimentally determined by various research groups. Further discussions and detailed
examination of published results are required, but there is also a need for setting up common
standards for presenting new research results. For instance, the work of the various groups
would benefit from the use of common definitions for dimensionless numbers, which may be
calculated either from the well-defined properties of the ice free carrier liquid or from the
actual properties of the ice slurry.
Directly immersing foodstuffs in the ice slurry mixtures is another relevant topic of research,
the results of which will be useful for the fishing industry and certain sectors of agriculture.
1.4.2 TECHNICAL CHALLENGES
There are various methods of producing fine-crystalline ice suspensions and the ice slurry
generators based on those methods should be subjected to long-term testing. Furthermore, the
efficiency of different physical methods to produce the “liquid ice mixtures” should be
evaluated, tested and compared. New methods of ice slurry production need to be continually
developed.
Heat transfer, time dependency (altering particle size distribution as a function of time, see
also Chapter 3.4), pumping characteristics and the pressure drop in piping systems also
require further investigations, while the hydraulic resistances of fittings and control valves,
etc. for ice slurry flow must also be established theoretically and experimentally.
Studies of cold storage behaviour and stratification of ice slurry must be continued. Are
mixing devices in storage tanks necessary to counteract the buoyancy effects caused by the
ice crystals in the aqueous solutions? The use of storage tanks reduces the required size of the
ice slurry generator but increases the running periods of the refrigeration machines. One
major advantage of ice slurries is their dual role as a transportation fluid and storage medium.
Various measurement and control devices are available for ice transport systems, but special
components for ice slurry must be developed and the performance of existing equipment
should be further evaluated by making reliable comparisons.
1.4.3 OBJECTIVES OF THE WORKING PARTY
The main objectives of the working party are to add new results to the currently available
body of knowledge in application and design, and to help improve this new technology by
exchanging information and know-how obtained by various members. Another objective is to
promote positive developments in fine-crystalline ice slurry systems world wide. Positive
efforts are achieved by:
1. Listing and studying current scientific and technical problems.

2. Establishing and continuously updating a reference list of test facilities and experimental
material of research projects concerning ice slurry systems. The current reference list is
available on the World Wide Web at www.ex.ac.uk/ice.
14
3. Establishing and continuously updating a reference list of research publications

concerning ice slurry systems. A reference list and important publications are also
available on the web.
4. Initiating demonstration plants to prove the reliability of ice slurry systems.
5. Initiating further research, e.g. defining experiments involving collaborating group
members.
6. Organising meetings for all members and other persons interested in the topic.
7. Maintaining web pages with continuously updated information about ice slurries and ice
slurry systems.
Meetings are organized, preferably annually, where information is exchanged and latest
findings in the field of ice slurry research are presented.
The working party currently consists of approximately 70 international scientists and

industrial representatives (as of December 2004). New members, who have knowledge of ice
slurry and/or ice slurry systems and are willing to participate in the working party, are always
welcome.
1.5 Research institutes
Several research institutes conducting ice slurry research exist around the world. Please refer
to Appendix 1: “List of Authors” of this handbook to see details of some of those research
institutes/universities/manufacturers.
1.6 Conclusions and outlook on the use of ice slurry
The industrial use of ice slurry began recently (about 25 years ago) but ice slurry has a great
potential for the future. Ice slurries enable the use of indirect refrigeration systems with a
small charge of the primary refrigerant as well as the possibility of thermal storage and
associated money savings. Furthermore, the use of ice slurry enables direct-contact cooling or
freezing of products.
Ice slurry also has better transport properties in tubes and heat exchangers, i.e. an
approximately 8-fold higher heat capacity than the traditional single-phase secondary
refrigerant for similar flow rates. The tube diameter can consequently be reduced by about
50% and the velocity inside the tubes can be reduced by about 50%. As a result, the energy
consumption by the pumps to pump the ice slurry is reduced to only about 1/8th of the energy
consumption necessary to pump the traditional single-phase secondary refrigerant, if the
viscosity is similar. At moderate heat fluxes, the heat transfer coefficient for ice slurries is
increased by a factor of 50 to 100 % as compared to conventional secondary refrigerants.
Therefore, ice slurry is believed to be a very promising secondary refrigerant with potential
for widespread future use.
1.7 Other phase-change slurries (PCSs)
Multifunctional (thermal) fluids and suspensions, also named “intelligent fluids”, yield new
classes of fluids with improved (thermal) properties. These fluids can be designed to
15
optimally fulfil particular objectives. In refrigeration and air conditioning these can be
enhanced thermal conductivity of a fluid, higher heat transfer coefficients, higher thermal
energy storage capacity, temperature stabilization, reduced pressure drop, etc. The use of
phase-change material (PCM) in a dispersed phase in a continuous carrier fluid leads — by its
phase change and latent heat — to very high energy densities, whilst the pumpability of the
fluid still remains. Such substances are named “phase-change slurries (PCS)”. Ice slurry is
the best-known PCS, but restricted to temperatures below zero degrees Celsius. To overcome
this restriction, at present, other types of PCS are being developed, e.g. micro-encapsulated
PCS, clathrates (clathrates utilise a chemical reaction and are therefore not really phase-
change slurries), shape-stabilized paraffins, etc. Nanofluids, e.g. fluids with dispersed metallic
particles on a nano-scale, also show enhanced heat transport properties. An advantage is that
the small particles - as a result of Brownian motion - do not lead to stratification, which is
induced by buoyancy forces.
Phase-change slurries are fluids with dispersed particles, which change phase at the melting
temperature of the dispersed phase. If the fluids are mixtures, very often a temperature glide
occurs, which is seen in the enthalpy density function h(T). Then h alters continuously as a
function of the temperature T. The energy to build up a crystal of a solid in a freezing process
is stored in the material and when the material — in the opposite process — is melted, this
amount of thermal energy (named latent heat) is released again. In a water/ice transition the
stored energy is high, namely 332 kJ/kg. Because the concentration of ice particles in
technical applications usually is less than fifty percent, the enthalpy density can (at most)
reach 166 kJ/kg. Other substances have slightly smaller enthalpy densities, but they are still
of high technical interest. To this latent heat, a smaller fraction of sensible heat may be added.
The energy storage capacity of phase change materials and slurries can be two to ten times
higher compared with the conventional storage technology using water, depending on the
material used and the operational temperature range of the engineering system.
1.7.1 DRY ICE/CARBON DIOXIDE SLURRY
Recently, a Japanese researcher reported on an interesting two-phase fluid system for low
temperature refrigeration, namely carbon dioxide with dry ice content (Inaba, 2003). The
liquid carbon dioxide operates as a boiling secondary refrigerant. It can be cooled for example
by an ammonia system. The dry ice (solid carbon dioxide) can also be mixed with a single-
phase heat transfer fluid as described by Pettersen et al. (1994).
1.7.2 CLATHRATE SLURRY
Clathrates or hydrate slurries are a crystalline compound substance of water (host molecules)
and a low-boiling temperature gas (guest molecules) in a special molecular structure at a
certain temperature and pressure (Inaba, 2000). The particle size is in the order of 5–50 µm.
The clathrate can be separated into liquid water and the gas phase by heating. This is a
chemical reaction with a high reaction enthalpy. Therefore, it is not absolutely correct to list
the clathrates in this chapter on PCS.
16
1.7.3 MICRO-EMULSION SLURRY
Water and liquid paraffin are immiscible. If an emulsifier, which is composed of a hydrophilic
head group and a hydrophobic tail, is injected into water, the paraffin is finely dispersed. Only
small amounts of emulsifier are necessary for this process. The quantity of
emulsifier/surfactant determines the size of the paraffin particles. Different problems have to
be solved by choosing the best additives, namely prevention of:
• Coagulation
• Coalescence
• Ostwald ripening
see e.g. (Hadjieva, 2003).
The process of the growth of larger particles at the expense of smaller ones has to be avoided.
This process, which is also observed in ice slurries, leads to an undesired change of some
physical properties, e.g. shear stress (“viscosity”), effective thermal conductivity, etc. Usually
the density, the enthalpy density and the specific heat are not influenced by such growth
effects.
1.7.4 SHAPE-STABILIZED PCM SLURRY
When a plastic material, e.g. polyethylene, is used as a stabilizing structure for the PCM, the
slurry is referred to as shape-stabilized PCM slurry. It is necessary that the PCM is immiscible
in the carrier fluid and in its liquid state does not leave the stabilizing structure, yielding an
open container. A combination used in practice is water (up to 90% contained in the
polymeric matrix) in oil as carrier fluid. Also gels and aero gels are used for this purpose. The
heat transfer performance is better, compared with microencapsulated PCMs (see Chapter
1.7.5), because there is no additional plastic layer between the carrier fluid and PCM which
presents an additional resistance for the heat flux. In some cases the polyethylene combs are
irregular, e.g. of a dendritic nature.
1.7.5 MICROENCAPSULATED PCM SLURRY
Micro-encapsulation techniques were well developed, mainly by the pharmaceutical indu-

stries, because they were used to produce tablets, copy papers, composites, powders, coatings,
foams and fibres. Furthermore, they were applied in cloth manufacture for heat capacity
enhancement (Colvin, 2003). Plastic microcapsules, containing a PCM, floating in a carrier
fluid for thermal energy transportation (see for example Jahns, 2003) constitutes a newer
application. The encapsulation has significant advantages, e.g. the PCM is completely
encapsulated and, therefore, shows a higher thermal cycling resistance. Important is that the
capsules are sufficiently resistant to the shear stresses occurring in pumps. The smaller the
capsules, the higher is their resistance to destruction.
17
1.7.6 POLYETHYLENE PELLETS
This technology is used rather rarely, but solid-solid transitions, e.g. of pentaelythrytohol
pellets (Inaba, 2000), have the advantage that no encapsulation is necessary, because of their
permanent solid state.
Major research activities in Canada, Europe, Japan and the United States of America are
directed towards the development of multifunctional and “intelligent” fluids. An example is
the clathrate system for a Japanese air-conditioning application reported by Hayashi (2000).
These developments will lead to highly improved fluids compared with the conventional
substances. But so far only ice slurry has made its way to full-scale production. These
applications have been mainly pioneered by the refrigeration and air-conditioning industry.
This handbook is solely devoted to ice slurries and their technical applications.
Literature cited in Chapter 1
1. Barth, M.: Hydro-scraped ice slurry generator. Proceedings of the Fifth IIR Workshop on
Ice Slurries, Stockholm, Sweden, May 2002
2. Colvin, D. P. et al., Proceedings of the Phase Change Material and Slurry Scientific
Conference and Business Forum, 107-121, Yverdon-les-Bains, Switzerland, April 2003
3. Davies, T.: A novel recuperative ice generator. Proceedings of the Fifth IIR Workshop on
Ice Slurries, Stockholm, Sweden, May 2002
4. Gun, M. A. van der; Meewisse, J. W.; Infante Ferreira, C. A.: Ice production in a
fluidized bed crystallizer. Proceedings of the Fourth IIR Workshop on Ice Slurries, Nov.
2001, Osaka, Japan
5. Hadjieva, M.; Gutzow, I.; Vassilev, T.: Proceedings of the Phase Change Material and
Slurry Scientific Conference and Business Forum, 27-32, Yverdon-les-Bains,
Switzerland, April 2003
6. Hayashi, K.; Takao, S.; Ogoshi, H.; Matsumoto, S.: Research and development on high-
density cold latent-heat medium transportation technology. Study based on NKK’s R&D
project: Broad Area Energy Utilization Network System (Eco Energy City Project),
Japan, 2000
7. Inaba, H.: Int. J. Therm. Sci. 39, 991-1003, 2000
8. Inaba, H.: Proceedings of the Phase Change Material and Slurry Scientific Conference
and Business Forum, 3-13, Yverdon-les-Bains, Switzerland, April 2003
9. Jahns, E.: Proceedings of the Phase Change Material and Slurry Scientific Conference
and Business Forum, 45-49, Yverdon-les-Bains, Switzerland, April 2003
10. Pettersen, J.; Jakobsen, A.: A dry ice slurry system for low temperature refrigeration.
SINTEF Report STF11 A94059, Trondheim, Norway, 1994
11. Zwieg, T.; Cucarella, V.; Worch, H.: Novel bio-mimetically based ice-nucleating
coatings for ice generation., Proceedings of the Fifth IIR Workshop on Ice Slurries,
Stockholm, Sweden, May 2002
18
CHAPTER 2. ICE CREATION, THERMOPHYSICAL PROPERTIES OF ICE

SLURRIES AND OTHER CHARACTERISTICS
2.1 The formation of ice

by Thomas Zwieg
2.1.1 INTRODUCTION
Crystallization of ice starts from a small nucleus, on which surface water molecules of the
surrounding liquid phase are incorporated. The temporal and spatial scope of crystallization is
characterized by two processes, nucleation and crystal growth.
2.1.2 METASTABLE STATE – SUPERCOOLED WATER
The phase change from water or an aqueous solution to ice occurs via metastable states; the
supersaturated state in the case of isothermal pressure variations, the supercooled state in the
case of isobaric temperature variations, or a combination of both. In most applications the
metastable liquid state is reached only by cooling. Within the supercooled water, clusters of
water molecules, also called embryos, are formed in an ice-like, hexagonal configuration.
These embryos have only relatively short lifetimes and decay again. Increasing the degree of
super-cooling lowers the total free energy, so that at a certain level of super-cooling, ∆T, the
embryos are more likely to grow than decay. When a critical embryo size is achieved, the
embryo becomes a nucleus. Then the process of ice nucleation or conversion from metastable
supercooled water into stable ice crystals starts (Lock 1990, Vali 1995).
2.1.3 NUCLEATION
For nucleation two different mechanisms are distinguished; the homogeneous and the
heterogeneous nucleation. In homogeneous nucleation embryos are formed internally in the
supercooled liquid containing only water molecules. In heterogeneous nucleation a foreign
substrate initiates nuclei formation (Fletscher 1970).
The nucleation process correlates with a change in Gibbs free energy, which consists of two
counteracting parts, the volume free energy ∆GV and the surface free energy ∆GS :
∆G = −∆GV + ∆GS (2.1)
During the phase transition, energy is released because of the lower free energy in ice. This
energy is proportional to the volume or number of molecules incorporated. The creation of a
new surface, the interface between ice and water, needs energy. This energy demand is
proportional to the surface area, A, and acts against the phase transition. For the very
simplified model of a small nucleus with a spherical shape and the radius, r, the change in
free energy becomes:
4
∆G = −V ∆GV + A∆GS = − π r 3 ( ∆GV ) + 4π r 2γ iw (2.2)
3
where γiw is the surface tension between the ice embryo and the water.
19
For the transformation from liquid to solid the change in volume free energy ∆GV is:
∆GV = ∆H − T∆S = (H L − H S ) − T ( S L − S S ) (2.3)
where the indices L and S represent the liquid and solid phase, respectively. If we assume that
the free energies of the two phases are independent of pressure, and the temperature
dependence is small, then:
L
H L − H S = L f and S L − S S = f (2.4) and (2.5)
Tm
where Lf presents the latent heat of fusion/crystallization and Tm the melting point.
From 2.4 and 2.5, the following expression is obtained:
⎛ T ⎞ L f ∆T
∆GV = L f ⎜ 1 − ⎟ = (2.6)
⎝ Tm ⎠ Tm
Using a molecular approach the volume free energy can be described as:
∆GV = ∆µ = (µ i − µ w ) (2.7)
where µi and µw are the chemical potentials per molecule in the supercooled water and in ice
respectively. Using (2.7), equation (2.2) can be expressed as:
4
∆G = n∆µ ⋅ Aγ iw = πr 3 ρ n ∆µ + 4πr 2 γ iw (2.8)
3
where n is the number of molecules; A is the surface area of the embryo; γiw is the surface
tension between the ice embryo and water and ρn is the density of molecules per unit volume.
Figure 2.1 illustrates the qualitative influence of the radius of the nucleus on the change in
Gibbs free energy.
The radius, r, at which the slope ∂(∆G)/∂r = 0 is

defined as the critical radius, r*. Differentiating
equation (2.2) with respect to r, inserting equation
(2.6) and setting ∂(∆G)/∂r = 0, r* can be expressed
as:
2γ iw 2γ iwTm 2γ iw ⋅ Tm
r* = = = (2.9)
∆GV ∆H∆T Lm ⋅ ∆T
where r* corresponds to the above mentioned

critical embryo size. Table 2.1 illustrates the
relationship between temperature T, critical radius
r* and the critical number of molecules inside the
ice embryo, n*.
Figure 2.1. Critical nucleus radius in homogeneous nucleation
20
Table 2.1. Critical radius and number of molecules per embryo

for homogeneous nucleation of ice (Vali, 1995)
T (°C) r* (nm) n*
–40 0.8 70
–20 1.8 650
–5 7.0 45 000
The energy barrier for a spherically shaped nucleus, also called the work of nucleation, is:
16π γ iw3 ⋅ Tm 2
∆Gn * = (2.10)
3 ( L f ⋅ ∆T ) 2
Above the critical radius, d(∆G)/dr < 0, thus the nucleation process releases energy and
spontaneous growth of the nucleus occurs.
From Equations (2.9) and (2.10) it can be concluded that the critical embryo size and the
energy barrier for spontaneous nucleation decrease significantly with an increasing degree of
supercooling and decreasing equilibrium freezing temperature. At low degrees of
supercooling, few large nuclei form and the nucleation speed is low. On the other hand, at
high degrees of supercooling many small nuclei form and the nucleation speed is high.
The rate of nucleation follows the Arrhenius equation and has the form:
⎛ ∆Gn * −∆G A ⎞
J = K exp⎜ − ⎟ (2.11)
⎝ kT ⎠
where ∆GA is the free energy of activation for diffusion across the boundary separating liquid
and crystal. Turnbull and Fisher (1949) made a formal derivation for the constant K. By
applying (2.10) the rate of nucleation, J, is given by:
nL kT ⎛ ∆GA ⎞ ⎡ 16πγ iw3Tm 2 ⎤

J= exp ⎜ − ⎟ exp ⎢ − ⎥ (2.12)
⎝ kT ⎠ ⎢⎣ 3L f 2 ( ∆T ) kT ⎥⎦
2
h
with nL as the number of water molecules per unit volume. According to Fletcher (1970) [x] in
the specific case of freezing of supercooled water, the term (NkT/h) ≈ 1035 cm-3 s-1. Using the
activation energy for viscous flow of water for ∆GA, we find for the term ∆GA/kT ≈ 10 at
–2°C, ≈ 15 at –22°C and ≈ 18 in the range –30 to –40°C. Thus for freezing of water at –2°C
equation 2.12 can be expressed as:
⎛ 16πγ iw3Tm 2 ⎞
J ≈ 10 exp ⎜ −
40
⎟ cm-3 s-1 (2.13)
⎜ 3L 2 ( ∆T )2 kT ⎟
⎝ f ⎠
In this equation the strong dependence of the nucleation rate upon the degree of supercooling
is apparent.
21
In practice, the majority of nucleation phenomena in liquids take place at degrees of

supercooling significantly less than those predicted by the theory of homogeneous nucleation.
This discrepancy is attributed to the presence of foreign surfaces e.g., cold walls or particles
in contact with the aqueous solution. These surfaces partly offer the necessary embryo surface
for nucleation, thus acting as nucleation catalysts. The volume-to-surface ratio will increase
and the nucleus-liquid interface will be partially replaced with the nucleus-catalyst interface.
In the case of heterogeneous nucleation the work of nucleation can be written as:
∆Ghet * = f (I p β ) ⋅ ∆Gn * , (f ≤ 1) (2.14)
where Ip represents an interface parameter and β a geometrical factor. For the special case of
nucleation at a flat and smooth wall depicted in Figure 2.2, the factor f can be expressed as:
f =
1
(2 + cosθ ) ⋅ (1 − cosθ )2 (2.15)
4
where θ is the contact or wetting angle, which is a function of the surface free energies at the
interface between the phases involved. The contact angle is described by Young's equation:
γ ws − γ is
cosθ = (2.16)
γ wi
All values of θ < π reduce the critical nucleus size and energy barrier for nucleation. Thus,
lower supercooling is necessary in a heterogeneous system.
Homogeneous nucleation is the limiting case when the wetting angle, θ, increases to θ = π.
Figure 2.2. Geometrical and surface energy balances in heterogeneous nucleation
The above treatment of heterogeneous nucleation is based entirely on simplified

thermodynamic considerations. This theory suffers from a number of serious limitations. The
main difficulty is that the factors such as the interaction and structure of the first monolayer of
the liquid in intimate contact with the foreign surface, the crystallographic lattice misfit, the
surface heterogeneity, defects and roughness and the type of bonding, are not taken into
22
consideration. The role of some of these influencing factors in the nucleation process is not
yet completely understood. Fletcher (1970) included these factors in the general interfacial
parameter Ip, used in Equation (2.14). The contact angle in Equation (2.16) as an experimental
measure of the wetting behaviour of the liquid phase at the foreign surface is more a
macroscopic than a microscopic parameter, thus would not be able to fully express
influencing factors at a molecular level.
The factors mentioned above will of course directly influence the surface free energy ∆GS and
the volume free energy ∆GV. In order to visualise these factors they can be summarized in an
additional free energy term ∆GSt, which expresses the strain introduced into the
nucleus/growing crystal by the above mentioned factors. The Equation (2.1) for the Gibbs
free energy can be re-written as:
∆G = ∆GV + ∆G S + ∆G St (2.17)
The crystallographic lattice misfit δ between a nucleating surface and ice may be defined as:
a − a0
δ= (2.18)
a0
where a and a0 are the lattice parameters of the nucleating surface and ice respectively. As
illustrated in Figure 2.3 this misfit will introduce a certain amount of dislocations and elastic
strain ε in the ice.
Figure 2.3. Nucleation of an ice embryo on a crystalline substrate

with a lattice misfit of approximately 10%.
Dislocations (arrows) and elastic strain are introduced into the nucleus.[x, Flet]
An increase in the concentration of dislocations will raise the magnitude of the interfacial
energy γis which implies an increasing contact angle and decreasing nucleation efficiency of
the nucleating surface. The strain in the ice embryo will raise the bulk free energy in the ice,
which also lowers the ice nucleation efficiency of the foreign surface. Consequently the best
nucleating substrate must have a lattice structure like or very similar to the ice.
23
A special case of heterogeneous nucleation is the secondary nucleation, where nucleation is

induced on a seed crystal. Nucleation occurs at the surface of a prior existing crystal of the
material crystallized. Please refer to Nývlt (1985) and Myerson (1992) for a detailed
description.
Similar to the heterogeneous system, nucleation can also be induced at lower supercooling by
dynamic stimulation. The energy required to create the necessary embryo surface is supplied
by friction, vibration or pressure pulses. Ice generation by the application of, for example,
ultrasonic waves is well known.
2.1.4 EFFECT OF ADDITIVES
The equilibrium behaviour of ice and water is changed by the presence of solutes. The
entropy of an aqueous solution is always higher than that of pure water. If the ice formed from
the solution is considered to be pure ice, which requires low freezing rates, the enthalpy
change per mol will be the same as without an additive and therefore the freezing temperature
to fit ∆G = 0 must decrease according to:
Tm’ = Q / ∆S’ = ∆H / ∆S’ , where ∆S’ = Sas – Si (2.19)
Here, Sas represents the entropy of the solution, and Si the entropy of ice. ∆S’ is larger than ∆S
and therefore Tm’< Tm. During solidification the system will suffer an increase in the solute
concentration. For example, sodium chloride (NaCl) is essentially insoluble in ice. At low
freezing rates salt is rejected into the solution as ice forms. At high freezing rates, salt can no
longer diffuse fast enough away from the interface and will be partially trapped in the ice.
Thus, the structure and the properties of the ice formed will be altered.
2.2 Crystal growth

by Thomas Zwieg, Didier Vuarnoz, Torben Hansen and Beat Frei
2.2.1 INITIAL CRYSTAL GROWTH
Under the circumstances that a nucleus larger than the critical size is formed and the
supercooling ∆T is not decreasing, the nucleus will grow further in size due to incorporation
of additional water molecules at the surface. This process, the crystal growth, controls the
final particle size distribution obtained in the mixture.
Several models exist to describe the mechanism of crystal growth. They can be classified into
pure thermodynamic models or models dealing with the actual kinetics of crystal growth. For
a detailed review, refer to Nývlt (1985), Myerson (1992) or Hobbs (1974). The growth of ice
in aqueous solutions may be governed by one or more of the following important factors:
a) the transport of water molecules through the fluid towards the interface,
b) the accommodation of these molecules on the ice surface,
c) the removal of the heat of crystallization.
A model that focuses on the transport of water molecules towards the interface is the diffusion
layer model (Nývlt 1985, Myerson and Ginde 1993).
24
This model divides the crystal growth into two phases:

a) Molecular diffusion through the boundary layer and
b) Incorporation of molecules into the surface layer.
These two phases are combined into one kinetic equation describing the crystal growth,
dmc 1 1 1 D
= K G ⋅ A ⋅ ∆C with = + and k d = (2.20)
dt K G kd ki δ
where A is the surface area of the crystal, KG the overall mass transfer coefficient, kd and ki
the mass transfer coefficients for diffusion and incorporation, ∆C the concentration changes in
the boundary layer, D the diffusion coefficient and δ the boundary layer thickness.
After adsorption of the water molecules on the crystal surface they diffuse two-dimensionally
on the surface before being integrated into the ice lattice. This integration typically occurs at
kink sites, steps, terraces or screw dislocations. Details regarding these phenomena are the
subject of different thermodynamic models, e.g. models of Kossel, Stransky or Burton,
Carbera and Frank (BCF-model). The crystal growth rate may be described as a linear
function and related to the increase in the crystal mass over time, RG, introduced above:
1 dmc α
RG = ⋅ = 3⋅ ⋅ ρ ⋅G (2.21)
A dt β
where A is the surface area of the crystal, α and β volume and area shape factors, ρ the crystal
density and G the linear growth rate (Myerson and Ginde 1992).
The influence of heat of crystallization on crystal growth is

discussed by Pronk et al. (2004).
The ice growth habit or ice growth shape in an aqueous

solution is reported to be dendritic or needle like (Fletcher
1970, Hobbs 1974, Lock 1990). It was shown that the ice
crystals heterogeneously formed in a tube-and-shell ice
slurry generator under flow conditions, were also of
dendritic type (Zwieg 2002).
Figure 2.4. Ice crystals formed heterogeneously in a shell-and-tube ice slurry generator under
forced flow conditions
As illustrated in Figure 2.4 the branches of the dendrites grow under these circumstances
faster on the upstream edge and slower on the downstream edge. The highest growth rates
occur at the branches oriented at an angle of approximately 60° from the flow direction. The
ice dendrites grow flat on the cooled surface, forming a thin, closely packed layer of ice. The
further growth is also of dendritic type, now directed to the centre of the tube, increasing the
layer thickness and decreasing the relative tube diameter.
25
2.2.2 FURTHER PROCESSES AFFECTING THE CRYSTAL SIZE
Once the initial ice crystals are formed in the ice slurry generator, their shape and size are
altered through a number of processes starting immediately after the initial crystallization
process. These processes are called attrition, spheroidization, agglomeration and Ostwald
ripening. All four processes are described in detail in the following paragraphs.
The driving potential for all crystal growth mechanisms is the minimization of the Gibbs' free
energy:
∆G = −SdT −Vdp+ ∑µdn + γdA (2.22)
2.2.2.1 Attrition
Attrition refers to the breaking off of small fragments, e.g. dendrite branches of a crystal. It
occurs when the crystal is subjected to a stress. This can be due to high fluid shear during
flow or mixing, or due to a collision with matter, e.g. other crystals, a mixing device, walls
etc. All kinds of damaging mechanisms are included in attrition, including abrasion. Impact of
attrition on the crystal size distribution can be found in the work of Pronk (2002).
2.2.2.2 Spheroidization
Spheroidization is a crystal forming process, which occurs after the dendritic crystal has left
the ice slurry generator. A sphere has a minimal surface area per unit volume. This is
thermodynamically favourable, therefore, the dendritic ice crystals formed in ice slurry
generators spheroidize during the transport to the storage tank or system, due to Ostwald
ripening in combination with attrition.
2.2.2.3 Agglomeration
When two or more crystals collide in a saturated suspension, they can adhere to each other
and become one crystal. This process is called agglomeration. These crystals can further
coalesce and sinter together, forming a larger stable crystal. The degree of agglomeration
decreases with increasing intensity of agitation and increases due to high supersaturation.
Vuarnoz et al. (2001) and Egolf et al. (2002) further showed the cluster problem in a realistic
tank of about 1000 liters in volume. Kitamoto et al. (2001) proposed the use of chemicals or
biological surfactants to reduce the agglomeration in ice slurry.
2.2.2.4 Ostwald ripening

Ostwald ripening is a crystal coarsening process, which alters the crystal size distribution of a
precipitate even under isothermal conditions. During ripening large crystals tend to grow and
the surrounding small crystals show the tendency to dissolve. This spontaneous process
occurs, because larger crystals (with their smaller surface area to volume ratio) are in a lower
state of free energy ∆G. This phenomenon is described by the Gibbs-Thomson (Ostwald-
Freundlich) equation:
⎛1 1⎞
∆G = 2 ⋅ γ ⋅ Ω ⎜⎜ − ⎟⎟ (2.23)
⎝ 2 r1 ⎠
r
where γ is the crystal interfacial energy and Ω the atomic volume.
26
The difference in curvature is the driving force for the Ostwald ripening. Larger particles,
having a larger radius r are thermodynamically favored during the coarsening. Along the
perimeter of an ice crystal, differences in curvature are a source of atomic interactions through
the ice/water interface and in the course of time lead to spheroidization of the crystals.
If mass transport between the crystals occurs and the growth kinetics is diffusion controlled,
all crystals of radius
β ⋅ Vm ⋅ γ c ⋅ c ∗
r=
(
k ⋅T ⋅ c − c∗ ) (2.24)
are in equilibrium with the bulk solution; here, β is a surface shape factor, Vm the molar
volume of the solute and γc the specific surface energy of the crystal, c the solubility of the
crystal and c* the equilibrium solubility (Nývlt, 1985; Myerson and Ginde, 1993). Crystals
with radii smaller than r tend to dissolve but larger than r tend to grow. The speed of the
ripening process depends largely on the crystal size and the solubility. In the case of diffusion
controlled growth kinetics, the linear growth velocity may be approximately described by:
dr γ c ⋅ Vm2 ⋅ D ⋅ c ∗
≈ (2.25)
dt 3⋅ k ⋅T ⋅ r 2
where D is the diffusion coefficient. Ripening occurs usually at low supersaturation. The
growth kinetics is therefore more likely to be controlled by surface integration effects than by
diffusion. Consequently ripening could be considerably slower than indicated by Equation
2.25.
Pronk et al. (2002) showed experimentally that this Ostwald ripening is the dominant
mechanism. The ripening process appears to be stronger in an ice slurry with a lower
concentration of the freezing point depressant. Similar results were also found by Hansen et
al. (2002). Ostwald ripening was found to be the predominant crystal growth mechanism
during five days of storage in a 1 m³ storage tank. In a storage tank with a mixing device the
crystal growth rate was found to increase due to convection, whereas non-moving layers of
ice crystals grow due to diffusion.
The crystal growth is an inherent property of the ice slurry system and cannot easily be
controlled (Hansen et al., 2002). Ice crystal growth has been thoroughly studied in frozen
food products and additives to decelerate ice crystal growth have been developed for use in,
for example, the ice cream industry. Common crystal growth inhibitors also work by the same
principle based on hydrogen bonding with ice (Hansen et al., 2002).
Polar additives used for freeze protection can form hydrogen bonds with the water molecules.
The freezing point depressing additives, propylene glycol and ethanol used in the Danish
survey, are both polar. With their polar hydrophilic ends they form hydrogen bonds with ice
crystals, while their non-polar hydrophobic ends reject water and ice molecules. The
hydrogen bond is considered as an energy barrier reducing the agglomeration activity. In the
study by Hansen et al. (2002), Tween 85 was only present at a concentration of 0.15% wt. As
both ethanol and propylene glycol form hydrogen bonds with the ice, it is believed that the
concentration of freezing point depressing agent is high in the concentration boundary layer
(Hansen et al., 2002).
27
The re-crystallization rate was always found to depend on time to the power of 1/3 in the case
of solely diffusion driven growth with mixtures of water and polar freezing point depressing
agents at concentrations corresponding to freezing points of –4.5°C and –4.9°C, respectively
(Hansen et al., 2002). Kang (1999) performed fundamental studies on hydrogen bonding
effects on the morphology of ice and Yazaki (1997) investigated and modelled the molecular
motion at the water/ice interface to predict the growth rate, which was found to have a
maximum value close to the freezing point of water.
Figure 2.5 shows a representative example of the development of the size distribution of
crystals with time (Hansen et al., 2002). The results are in good agreement with the finding by
Hagiwara (1996). The initial crystal diameter in the storage tank was distributed in a narrow
band of ±50 µm around the mean crystal size of approximately 100 µm. At the end of the
storage period (94 hours), the mean crystal diameter increased to approximately 480 µm, and
the crystal size ranged from 250 to 700 µm.
Ethanol
14 Propylene glycol
40 12 10 wt.% ice
45 hours
Growth rate ( µm/h)
30 0 hours 22 hours 10
30 wt.% ice - mixed
Frequency (%)
94 hours 8 30 wt.% ice - heterogeneous

20
6 46 wt.% ice
10 4
2
0
0 100 200 300 400 500 600 700 0
Feret diameter (µm) 0 10 20 30 40 50 60 70 80 90 100
Time (h)
Figure 2.5. a) Mean Feret diameter growth dependency on additive, ice concentration and
storage type shows clearly the Ostwald ripening effect.
b) Particle Feret diameter growth rate dependency on additive, ice
concentration and storage type.
In the experiments performed by Hansen et al. (2002), the equation for the average particle
diameter was formulated as:
d Feret = E + B ⋅ τ C (2.26)
where the coefficients E, B, C were determined according to the additive and type of the
storage tank (Table 2.2).
28
Table 2.2. Parameters E, B and C
Additive Storage tank Cice E B C R2

Propylene glycol homogeneous 10% 100.7 28.6 0.53 0.986
Propylene glycol homogeneous 30% 162.6 11.4 0.60 0.892
Propylene glycol stratified 30% 113.2 7.7 0.68 0.938
Propylene glycol stratified 46% 126.5 7.6 0.59 0.916
Ethanol homogeneous 10% 118.9 25.6 0.62 0.988
Ethanol homogeneous 30% 199.7 8.4 0.70 0.937
Ethanol stratified 30% 92.8 21.1 0.49 0.983
Ethanol stratified 46% 144.3 4.6 0.72 0.972
These results reveal three characteristic differences in the parametric influence on the crystal
growth in the case of ice slurry:
1) the growth rate is always higher with ethanol compared to propylene glycol,
2) crystal growth rate is slower at higher ice concentrations,
3) the growth rate is higher in the homogeneously mixed storage.
These findings revealed that contact with air from the surroundings did not influence the
crystal growth rate (Hansen et al., 2002). In the agitated (homogeneous) storage, convection
leads to growth rates which are approximately 60-70% higher than in the heterogeneous
storage.
The size distribution of ice particles influences the operation of ice slurry systems. Ice slurry
with larger ice particles will require a greater stirring effort to obtain a homogeneous mixture.
In heat exchangers the melting characteristics may also change (Hansen et al., 2002).
2.3 Types of ice crystals and ice particles

by Didier Vuarnoz and Torben Hansen
Particle shape and size analysis is important in elementary particle and solid state physics,
chemistry, pharmaceutical studies, biology, etc. (for an overview, see e.g. Kawashima, 1993
or Egolf, 1999). Because of the rather large size of the ice particles in ice slurries (30-300
µm), an ordinary optical microscopic method can be used to study particle characteristics,
although a small heat input into the fluid occurs. The disadvantage of this technique is that
two-dimensional projections of the particles are obtained, which show irregular shapes. The
probability of observing one entire large particle in the field of view of the microscope is
smaller than the probability of detecting a small one.
Two methods have been developed to extract the ice concentration from photography (Sari,
2000) and (Fournaison, 2001). Nevertheless, these methods are not as simple and accurate as
others used to determine the ice concentration.
Numerous microscopy pictures of ice slurry particles have been published, e.g., by Fukusako
et al. (1999), Bel (1996) and Kauffeld et al. (1999). Kawashima (1993) studied ice slurry
flows (snow water mixtures) with a high-speed video camera. Bel (1996) reported that the ice
crystals that have a hexagonal structure, become a little smoother and rounder under the
attrition process. The major difficulty is to interpret the focussed or projected image from
photography to obtain the three-dimensional structure. The first assumption is that the ice
29
particles are flat or round, but the particles may also undergo some metamorphoses, see
section 2.2.2 and (Turian, 1992). Such mechanisms may explain the time behaviour, as
illustrated in Figure 2.6. Many groups are equipped to perform photography of ice crystals
(Hansen et al., 2002; Pronk, 2002 and Vuarnoz, 2001).
a) Only liquid b) Air dissolved in the c) First ice crystals appear

substance leaving it
d) Crystals at the outlet e) Ice particles after 11 f) Ice particles after 10

of a scraper-type ice min of storage hours of storage
slurry generator
g) Ice particles after 22 h) Ice particles during i) Ice particles have

hours of storage their melting process totally disappeared
Figure 2.6. Ice crystal morphology during a freezing/melting cycle,

a), b), c), h), and i) from EIVD-TiS laboratory, Switzerland; d) from DTI, Denmark;
e), f), g) from Delft Univ., Netherlands (Pronk 2002)
2.3.1 SHAPE PARAMETERS
Knowledge of geometrical characteristics of ice slurry particles and their time behaviour is
believed to be an important key to understanding the deviation in the results found in
numerous apparently comparable experimental investigations of thermophysical properties,
pressure drop, heat transfer coefficients, viscosity, etc. (Hansen et al., 2002).
The shape and size of the ice crystals in ice slurries depend on many parameters, and the
shape of ice particles evolves throughout their lifespan.
30
Non-spherical particles and particle size distribution

If the particles are non-spherical, their terminal settling velocity, e.g. in a storage tank, is
lower than that of a spherical particle. Working with projections rather than a full volumetric
analysis leads to a loss of information. The projected image of an individual ice crystal does
not reveal its true shape and dimension.
The more spherical the ice crystals are, the more adequate it is to quantify the crystal
dimensions in terms of the Feret diameter. The Feret diameter is defined as the diameter of a
circle having the same area Ap as the projection of the ice crystal:
4 ⋅ Ap
DFeret = (2.27)
π
Table 2.3. Effective particle diameters (Turian et al., 2002)

dv Diameter of a sphere with the same volume as the 1
⎛ 6 ⋅V ⎞ 3
particle dv = ⎜ ⎟ (2.23)
⎝ π ⎠
ds Diameter of a sphere with the same surface area as ⎛S ⎞
0 .5
the particle ds = ⎜ ⎟ (2.24)

⎝π ⎠
Diameter of a sphere with the same external surface 2
dsv ⎛ V ⎞ dv
to vol. ratio as the particle d sv = 6 ⋅ ⎜ ⎟ = (2.25)
⎝S⎠ 2
ds
dd Diameter of a sphere with the same resistance to
motion as the particle in the same fluid
df Diameter of a sphere with the same density and free
falling velocity as the particle in the same fluid
dst Free falling diameter of the particle in Stokes law ⎛d 3 ⎞
18 ⋅ η ⋅ w 0
region (Re < 0.2) d st = = ⎜ v ⎟ (2.26)
(ρ f − ρ ice )g ⎜⎝ d d ⎟
⎠
There are a variety of methods for the determination of the effective particle diameter (see e.g.
Turian et al. (1992) and Table 2.3. The ice particles in an ice slurry can vary in size from ~
0.1 to ~ 0.7 mm, as shown by Hansen et al. (2002). Their size depends on the storage time.
The equation for the average particle size determination is given by Turian et al. (1992):
k
dm = ∑n di =1
i i
m
/ ∑n i (2.28)
where ni is the number of particles with a diameter between di and di+1 and m is the mth
moment of the size distribution (Turian et al.,1992):
d length mean diameter (2.29)
1
⎛⎜ d 2 ⎞⎟ 2 “surface” mean diameter (2.30)
⎝ ⎠
1
⎛⎜ d 3 ⎞⎟ 3 “volumetric” mean diameter (2.31)
⎝ ⎠
31
d4
d wm = “weighted-moment” mean diameter (2.32)
d3
The latest findings on the distribution of ice particle sizes are those reported by Hansen et al.
(2002), who arrived at the conclusion that an ice particle size distribution can be
approximated by a normal distribution. The mean value and the standard deviation depend on
the additive used, the storage type and the time.
2.3.2 ICE PARTICLE MODEL
Modelling work facilitates correlation and extrapolation of results. The ice concentration may,
for example, be determined from a few photographs. Another example of an ice particle
model was presented by Egolf (1999) based on his physical properties model for ice slurry.
Geometrical model
Photographs show that in a quiescent (inactive) fluid the solid ice particles are randomly
distributed throughout the carrier fluid. At present, to describe particles, two kinds of models
are discussed. In one model the particle form is assumed to be given by two hemispheres
connected to a small cylinder of length d. The second model assumes disk-shaped particles
with a radius r and a thickness h. (see Figure 2.7a) and b) respectively).
a)
l b)
d
R
Ice particle h
b
Figure 2.7. Ice particle shapes estimated from projected images: a) a spheroidal ice particle
shape composed of two hemispheres connected by a cylinder of the same radius (applied by
Vuarnoz and Egolf [2001]), b) disk-shaped ice particle model proposed by Pronk (2002)
The main objective is to determine the actual volume of the particles from microscopy
images. One method was developed by Vuarnoz and Egolf (2001) for the case of spheroidal
particles. They assumed that the particles are distributed in a homogenous and isotropic
manner. After double space integration, the authors found that the actual distance d is 19%
larger than the optically observed dsphere. This factor is taken into consideration by applying a
correction to the two-dimensional observation to approximate three-dimensional size. Notice
that the quantities d and l are affected by this factor but not the width b. The factor f = 1/1.19
= 0.84 is named the view factor. Based on this method it is possible to estimate the ice
concentration in the ice slurry.
32
2.3.3 MEASUREMENTS AND STATISTICS
Numerous researchers have investigated ice crystals by photographic methods. Sampling

locations and age of the sample are of major importance, as shown by Vuarnoz et al. (2001),
see also Figure 2.8a and b. Measurements were performed on two samples after two hours of
storing the ice slurry with additional mixing. Samples were chosen from four different heights
in the storage tank — two hundred ice particles were used to quantify the ice particle
dimensions. It was observed that the size of the ice particles is not a clearly defined function
of the temperature of the ice slurry (Vuarnoz et al., 2001).
Ice content = 10.3 % Ice content = 21.4 %

300 350
250 300
200 250
150 200
100 150
50 100
0 50
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700
Length l (µ m) Length l (µ m)
Figure 2.8. Ice particle size distribution data obtained for a 10 w/w% water-ethanol ice slurry
after correction with a view factor. Shown are the data for (a) Ci = 10.3%
and b) Ci = 21.4% ice concentration (Vuarnoz, 2001 and Egolf, 2002)
2.4 Properties of aqueous solutions and ice

by Åke Melinder
2.4.1 INTRODUCTION
In some applications, ice slurry is produced with pure water which is an excellent secondary
fluid. However, when temperatures below 0°C are required, one or more freezing point
depressant additives are used. Such binary or ternary aqueous solutions have long been used
in single-phase indirect refrigeration systems to transport energy from the cooling unit to the
evaporator. The same liquid mixtures may be used for ice slurry applications, where small ice
crystals are formed and distributed to the cooling unit from which heat is removed when some
or all of the ice crystals melt.
33
In order to choose a secondary fluid for ice slurry use and to make correct technical
calculations of the ice slurry system there is a need to examine and evaluate thermophysical
properties and other aspects of the aqueous solutions.
There are several requirements that have to be fulfilled by an ideal secondary fluid. It should
possess good thermophysical properties, making it possible to:
• transport the required cooling capacity through small pipes or tubes at a small volumetric
flow rate with a small temperature change over the cooling unit;
• obtain good heat transfer, leading to small temperature differences at the location of heat
transfer in the evaporator or object to be cooled;
• be pumped with an acceptable pressure drop, in order to use a pump with small or
moderate pumping power.
Furthermore, it is important that the fluid does not cause any material corrosion problems. It
should also be non-toxic, environmentally acceptable, non-flammable, safe to handle and
being available at a reasonable price.
2.4.2 THERMOPHYSICAL PROPERTIES OF AQUEOUS SOLUTIONS
Aqueous solutions of ethyl alcohol (ethanol) and sodium chloride (sea water) have been used
in most ice slurry applications in Europe. However, aqueous solutions of glycols, ammonia
and various salts may also be of interest for various applications. From this point of view, the
following seven aqueous solutions are considered for ice slurry use: ethyl alcohol, ethylene
glycol, propylene glycol, ammonia, calcium chloride, sodium chloride and potassium
formate. Alcohol, glycol and salt solutions (sea water) represent the major applications so far.
The other solutions are used thanks to their favourable thermophysical properties. Also, other
substances, e.g. sugars, are of course feasible for ice slurry use, even though they are not
considered here.
The following thermophysical properties are of importance when comparing different

aqueous solutions: the freezing point temperature, density, viscosity, specific heat or enthalpy
change and thermal conductivity. These property values can be obtained from Melinder
(1997). Other thermophysical properties that are useful to know at times are the boiling point
temperature, thermal expansion coefficient and surface tension. Brief comments are given on
each of them here. For most of these aqueous solutions, values of these properties can be
obtained from Melinder (1997).
The freezing point temperature, ϑF, is the temperature at which ice crystals begin to form in
equilibrium if there is no supercooling. In ice slurry applications the slope of the freezing
point curve is important as it determines the ice concentration that can be obtained in
equilibrium (see also 2.4.3).
A low boiling point temperature of the aqueous solution can (even at moderate temperatures)
create problems in a system, especially if the liquid is exposed to atmospheric air. This may
pose a problem with high concentrations of alcohol in water. However, additive
concentrations used in most ice slurry applications are quite low and this is usually not a
problem (see Chart C14, Melinder, 1997).
34
The surface tension, σt, of a liquid is the force per unit length that acts to keep the surface as
small as possible. A low surface tension may increase the tendency of the solution to leak out
and may also increase the risk of foaming in the system and of cavitations in the pump. Ethyl
alcohol has the lowest surface tension of the aqueous solutions followed by the glycols (and
ammonia). The salt solutions have higher surface tensions than water (see Chart C15,
Melinder, 1997).
Knowing the correct mass density, ρ, is important for several reasons. The concentration of a
certain known solution is often measured by checking the density. The density of aqueous
solutions used is higher than the density of ice. In ice slurry systems this high density may
cause a stratification of the fluid caused by the buoyancy forces on the ice particles. The mass
density difference between ice and carrier fluid and hence the tendency to separate is the
lowest with ethyl alcohol (and ammonia), followed by glycols, and most severe with salt
solutions.
It is important to determine the dynamic viscosity, µ, and/or the kinematic viscosity, ν = µ/ρ,
of a liquid. The viscosity should not be too high at the operating temperature if the aqueous
solution is to perform well. The viscosity is also used to calculate the Reynolds number, Re =
(w⋅d)/ν, which determines the type of flow that will prevail, laminar, transitional or turbulent.
Here w is the fluid velocity and d is a characteristic length, i.e. the tube diameter. The
viscosity of (single phase) propylene glycol is the highest, followed by ethyl alcohol and
ethylene glycol, while potassium formate and the chlorides show the lowest viscosities. The
viscosity of ice slurry will, however, depend strongly on the ice content.
A high value of the specific heat, cp, of the fluid is favourable as it affects the heat transfer
coefficient and the volume flow rate needed for a desired cooling capacity. At the low
concentrations usually used in ice slurry, the specific heat of ethyl alcohol is higher than that
of water. The volumetric heat capacity (i.e. specific heat times the fluid density) is inversely
proportional to the secondary fluid flow rate needed for a certain cooling capacity.
A high value of thermal conductivity, k, is desirable as it contributes to good heat transfer and
thereby decreases the temperature difference between the fluid and the tube wall. The
chloride salts and potassium formate have generally a higher thermal conductivity than the
other aqueous solutions. The thermal conductivity of ice slurry will increase with the ice
concentration, but it is very difficult to measure the thermal conductivity of a two-phase
fluid.
The volumetric thermal expansion is of interest to designers of refrigeration systems in

determining the size of expansion vessels. Curves of thermal volume expansion coefficient
from freezing point to a certain temperature for various concentrations are given in Charts
C1-11e in Melinder (1997), for most of the aqueous solutions.
In ice slurry applications, one of the main purposes is to benefit from the latent heat or
enthalpy difference when the ice is melting. The latent heat reduces the volumetric flow rate
required for a desired cooling power. The latent heat available is reduced with increasing
additive concentration. For aqueous solutions the available latent heat reaches its maximum
at temperatures just below the freezing point and decreases at lower temperatures.
35
2.4.3 FREEZING-POINT TEMPERATURE OF AQUEOUS SOLUTIONS
In Figure 2.9 the freezing point temperature is shown as a function of the additive
concentration (cA). Freezing point curves are presented for different aqueous solutions. For
low concentrations these curves are usually based on the values published in the CRC
Handbook of Chemistry and Physics (1989) and for higher concentrations they are based on
the polynomial equations given in Chapter 6 of Melinder (1997). In order to obtain and
maintain ice slurry the operating temperature must be kept below the freezing point. When
the temperature drops below the freezing temperature ice crystals consisting of pure water
may start to freeze out.
0
Ethyl alcohol
Ethylene glycol
-5
Propylene glycol
Ammonia
Freezing point temperature in °C
-10
Calcium chloride
Sodium chloride
-15 Potassium acetate
Potassium formate
-20
-25
-30
-35
0 10 20 30 40 50
Mass concentration of additive, CA in % by weight
Figure 2.9. Freezing point temperature (ϑF) as a function of additive concentration (CA)
2.4.4 THERMOPHYSICAL PROPERTIES OF ICE
In order to predict the thermophysical properties of ice slurries, it is necessary to know the
properties of the pure substances, e.g. ice (water in its solid phase).
The density of ice is lower than the density of water in the liquid phase (ice tends to float on
the liquid surface). The density of ice, ρI, at the temperature, ϑ I (in °C),can be expressed as:
ρI = 917 – 0.13·ϑ I (kg/m3) (2.33)
The thermal conductivity of ice at the temperature, ϑ I, is much higher than that of water in
the liquid phase. This feature has a good influence on the performance of ice slurries. The
thermal conductivity of ice, kI, as a function of temperature, ϑ I, can be described by:
36
kI = 2.21 – 0.012·ϑ I (W/m·K) (2.34)
The equations for the density and the thermal conductivity are based on the data published in
the Handbuch der Kältetechnik, VI/B (1988) and Handbook of Thermodynamic Tables and
Charts (1976).
The specific heat of ice at the temperature, ϑ I, is much lower than the specific heat of water
in the liquid phase. The specific heat of ice, cp,I, as a function of temperature can be described
by:
cp,I = 2.12 + 0.008·ϑ I [kJ/(kg·K)] (2.35)
The enthalpy of ice, hI, at the temperature, ϑ I, can then be calculated with the relation:
hI(ϑ I) = –332.4 + ϑ I ·(2.12 + 0.008·ϑ I) (2.36)
These equations are based on the heat of melting of ice (332.4 kJ/kg) as well as the specific
heat values of ice, presented by Dickinson-Osborne, published in CRC Handbook of
Chemistry and Physics (1989).
2.5 Physical properties of ice slurry

By Åke Melinder and Oleg B. Tsvetkov (see specific list of symbols in Appendix 3)
Most properties of ice slurries can be derived from the individual properties of the aqueous
solution and of ice by combining the respective properties in the correct ratio. The particle or
ice fraction, often called ice concentration is here a crucial parameter. Nonetheless the
viscosity, thermal conductivity and enthalpy of the ice slurry have to be determined more
carefully as described in the following sections.
2.5.1 FREEZING POINT TEMPERATURE AND ICE CONCENTRATION
Figure 2.10 shows how the freezing point is changes with the composition of the mixture.
The diagram can be used for any of the aqueous solutions, but let us use the example of a salt
in water. An increased concentration of the freezing point depressant lowers the freezing point
temperature. This continues until we reach the concentration where salt and water form an
eutectic solution. The freezing point is raised when the concentration becomes higher than the
eutectic. The over-eutectic part of the freezing point curve is generally of no interest for this
application. Note that only certain solutions have well-defined eutectic points.
37
Temp.
A
0
tF F
tS K S
tE E
0 c0 cS cE Conc.
Figure 2.10. Freezing point diagram (schematic)
Let us assume a solution with the concentration c0 and note what happens when it is cooled
down from A to the freezing point F (c0, tF) and below that temperature. If the temperature is
decreased below the freezing point and the process takes place in equilibrium, a separation of
the mixture takes place. Ice crystals that ideally consist of pure water freeze out and are
separated from the rest of the solution and the concentration of the freezing point depressant
in the remaining solution rises as a consequence. When the temperature has been lowered so
that it corresponds to K in Figure 2.10, there are ice crystals with the condition (c = 0, tS) as
well as solution with the condition (cS, tS).
A consequence of this is that the solution does not freeze to a solid mass when the freezing
point is passed, but an ice slurry is formed. Only when the temperature goes down to the
eutectic temperature, would the solution fully solidify. We need also to remember that in
practice sub-cooling phenomena often occur. The first ice crystals are often not formed until
the solution temperature falls to one or two degrees below the freezing point, as indicated by
the freezing point curve. When some ice crystals have been formed, the continuation of the
process seems easier to take place in equilibrium.
The concentration of the remaining solution is given by a mass balance based on the condition
that the mixture has an unchanged total amount of salt. Let us refer to the mass of pure ice by
mI and the mass of solution by mS, the concentration of which is cS. Then the amount of salt is
mS·cS. In the original solution the salt amount was evidently c0·(mI+mS). As the amount of salt
has not changed, it follows that mScS = c0(mI+mS), which also can be written as mS(cS-c0) =
mIc0. The ice concentration, or mass fraction of ice, cI, can then easily be calculated from:
cI = mI/(mI+mS) = (cS-c0)/cS (2.37)
Some extensive studies on ice slurry properties, such as those by Bel et al. (1996) and
Guilpart et al. (1999), present ice concentration as a function of the temperature for various
concentrations of the freezing point depressant substance, usually for ethyl alcohol-water.
Suitable equations for ice slurries are also given in Kauffeld, et al. (1999) and Lottin, et al.
(2000).
38
Figure 2.11 gives in the same manner ice concentration (mass fraction of ice) as a function of
the temperature for various concentrations of the freezing point depressant substance for
aqueous solutions of sodium chloride and ethyl alcohol. Property values are taken from CRC
Handbook of Chemistry and Physics (1989) and an Excel programme based on equations and
coefficients (not valid for low concentrations) in Melinder (1997b).
Sodium chloride-water Ethyl alcohol-water

0,6
0,6
0,55 0,55
Ice concentration [kg ice/ kg tot.] .

0,5 0,5
Ice concentration [kg ice/ kg tot.] .
0,45 0,45
0,4 0,4
0,35 0,35
0,3 0,3
0,25 0,25
0,2
0,2
0,15
0,15
0,1
0,1
0,05
0,05
0
0 -25 -20 -15 -10 -5 0
-25 -20 -15 -10 -5 0 Ice slurry temperature [°C]
Ice slurry temperature [°C]
Initial conc. of additive [kg add./ kg tot.]
Initial conc. of additive [kg add./ kg tot.]
0,03 0,05 0,08 0,10 0,05 0,10 0,15
0,13 0,15 0,18 0,20 0,20 0,25 0,30
Figure 2.11. Ice concentration as function of temperature for two types of aqueous solutions
In 2.5.2 and 2.5.3 the volume fraction of particles, c, is introduced, corresponding to a

volumetric ice fraction, cI,vol = cI/((cI+(1-cI)ρI/ρCS), where cI is the mass ice fraction, ρI is the
density of ice and ρCS is the density of the concentrated solution. The density, ρ, of ice
slurries may be estimated with a mixing equation ρ = 1/[(cI/ρI)+(1-cI)/ρCS].
2.5.2 VISCOSITY
The shear viscosity of a concentrated suspension, taking into account the hydrodynamic
interaction of particles, particle rotation, collision between particles, agglomerate formation,
and so on, can be expressed by a theoretical equation in a power series (Thomas, 1965):
η * / η1 = 1 + a1c + a2c 2 ± a3c 3 ± ... . (2.38)
The constant a1 is generally assumed to have the value determined by Einstein, a1 = 2.5 .
Based on many experimental investigations of the viscosity of suspensions it was shown that
the term 1 + 2.5 ⋅ c + 10.05 ⋅ c 2 accounts for over 97.5% of the values of the relative viscosity
η * / η1 for volume fractions of particles, c, below 25% ( c < 0.25). For values of c > 0.25
satisfactory results were obtained by Thomas (1965):
39
η * / η1 = 1 + 2.5 ⋅ c + 10.05 ⋅ c 2 + ∆ (2.39)
where ∆ = A ⋅ e B⋅c , A = 0.0273, B = 16.6, η * is shear viscosity of the suspension, and η1 is

shear viscosity of a matrix. For two-phase mixtures like ice-slurry the background shear
viscosity η1 is the viscosity of the liquid solution.
Viscosity values of many aqueous solutions can be obtained from Tables D1-11 in Melinder
(1997a, b). Some literature data for the shear viscosity of solutions are also given in Tvsetkov
(2001) and Vargaftic (1972).
The following numerical values for predicting the reduced relative viscosity for c = 0.1 and
c = 0.2 were defined as:
η * / η1 = 1 + 2.5 ⋅ 0.1 + 10.05(0.1) 2 + 0.00273 exp(16.6 ⋅ 0.1) = 1 + 0.25 + 0.1005 + 0.014 = 1.365
for c = 0.1 , and
η * / η1 = 1 + 2.5 ⋅ 0.2 + 10.05(0.2) 2 + 0.00273 exp(16.6 ⋅ 0.2) = 1 + 0.5 + 0.402 + 0.076 = 1.978
for c = 0.2 .
Particularly important are questions pertaining to the non-Newtonian and viscoelastic
behaviour of ice slurry. Many viscosity measurements of ice slurry as a function of the ice
concentration have shown that ice slurry behaves like a Newtonian fluid for ice
concentrations lower than 20% vol. (c < 0.2).
2.5.3 THERMAL CONDUCTIVITY
The theory of the heat conduction phenomena through a random suspension of spherical
particles in a matrix of uniform thermal conductivity has suggested that the volume fraction of
the particles is small enough to make all interactions between the spheres negligible. For this
kind of suspension an effective thermal conductivity k * in the Maxwell approach can be
represented by,
k * k 1 = 1 + 3βc (2.40)
where: β = (α − 1) (α + 2), α = k 2 k1 , c is the volume fraction of the particles (spheres), k1

is the thermal conductivity of a matrix, and k 2 is the thermal conductivity of the spheres.
However, the validity of equation (2.33) is restricted to a very small volume fraction of
particles. To represent the actual behaviour of the thermal conductivity including the effect of
interactions between spheres, the following approximation of k * has been proposed (Jeffrey,
1972).
α −1
k * k 1 = 1 + 3β c + 3β 2 c 2 (1 + 2δ ) (2.41)
α +2
The information about the coefficients δ , β and α is summarized in Table 2.4. For the ice-
slurry models the background conductivity k1 is the thermal conductivity of the liquid water
solution or of another secondary refrigerant, and k 2 is the thermal conductivity of ice. The
equation for ice by (Bobkov, (1977) of the form:
40
k 2 = k I = 2.22(1 − 0.0015ϑ ) (2.42)

has been used for Table 2.4. Here k I is given in Wm-1K-1, and ϑ in °C.
Table 2.4. Coefficients in Eq. (2.40) for k * k 1

α 0.1 0.5 1.0 2.0 5.0 50.0
β -0.429 -0.200 0 0.25 0.571 0.942
δ 0.21 0.21 0.21 0.22 0.22 0.25
Thermal conductivity values of many aqueous solutions can be obtained from Tables D1-11
in Melinder (1997a,b). The thermal conductivities k1 of some actual water solutions are given
in Tvsetkov (2001) and Vargaftic (1972).
Using Table 2.4 it is possible to calculate α (which varies from about 17 to 4) and estimate
k * k 1 according to Eq. (2.41). For example, if α is assumed to be equal to 5, the values of β
and δ from Table 2.4 are as follows: β = 0.571, δ = 0.22. This leads to the relative thermal
conductivity value of:
5 −1
k * k 1 = 1 + 3 ⋅ 0.571 ⋅ 0.1 + 3(0.571) 2 (0.1) 2 (1 + 2 ⋅ 0.22 ) = 1 + 0.1713 + 0.00978 ⋅ 1.25 = 1.183
5+2
where c = 0.1 .
2.5.4 ENTHALPY AND ENTHALPY PHASE DIAGRAMS OF AQUEOUS SOLUTIONS
The secondary fluid flow rate required to achieve a cooling capacity of Q& (in kW) in an
indirect system with a single-phase fluid can be calculated by the following relation:
Q& = m& ⋅ c p ⋅ ∆ϑ = V& ⋅ ρ ⋅ c p ⋅ ∆ϑ (2.43)
For an ice slurry system, it is more suitable to apply the relation:
Q& = m& ⋅ ∆h = V& ⋅ ρ ⋅ ∆h (2.44)
Both sensible heat and latent heat of melting are included in the enthalpy difference, ∆h. In
these relations, m& is the fluid mass flow rate, V& is the volumetric flow rate, ∆ϑ is the
temperature change between the fluid entering and exiting the cooling unit, ρ is the fluid
density and cp is its specific heat.
The “heat transport capability" of ice slurries in equilibrium is proportional to the enthalpy
difference, ∆h, obtained for a certain temperature change between the fluid entering and
exiting the cooling unit. It is much higher for an ice slurry than for mono-phase aqueous
solutions, as there is a benefit from the latent heat of melting of ice crystals, which reduces
the fluid flow rate required for a given cooling capacity and also the temperature change. This
“heat transport capability” depends partly on the specific heat of the fluid, but more on the
amount of ice that will melt, i.e., on the ice concentration of the ice slurry that enters the
cooling unit (Melinder, 2002b).
41
A way to visualize the benefits from the latent heat or enthalpy difference in the melting
process of ice may be achieved by studying an enthalpy-phase diagram of the aqueous
solution used. The amount of heat exchanged in the freezing process can be estimated if we
have access to enthalpy values, for instance from such enthalpy-phase diagrams with some
added features. Such diagrams can help us estimate at what additive concentrations and
temperatures we will get the most benefit. Such enthalpy-phase diagrams are given for seven
aqueous solutions in Figures 2.13 to 2.19. These diagrams are taken from Melinder (2002a)
where also the enthalpy difference between certain temperatures and for various freezing
points is given.
For calculation purposes we choose to view an ice slurry solution as consisting of small ice
crystals of pure water that freeze out, and the rest of the solution with a higher concentration
of the freezing point depressant. In order to estimate enthalpy values of an aqueous solution
with ice particles we need freezing point data, values of specific heat as well as enthalpy
values of ice and heat of mixing of the solution. The heat of mixing can, in the diagrams, be
seen as the variation in enthalpy values with the concentration of a reference isothermal line
(ϑR = 0°C, 10°C or 25°C). The enthalpy of the ice slurry, h, at point K in Figure 2.12, can be
expressed by the following function (see Figure 2.12):
h = h(K) = hI (ϑCS)·cI + hCS(cCS, ϑCS)·(1 – cI) (2.45)
where hI(ϑCS) = hH – hD = hH – hG + (hG – hD) (2.46)
ϑCS
hCS(cCS, ϑCS) = h0,R + ∆hM(cCS, ϑR) + ∫ (c p ·dϑ ) (2.47)
ϑR
The enthalpy difference between D and G, the heat of melting of water, is hD-hG = 332.4
kJ/kg. The enthalpy of ice at various temperatures shown in Figures 2.13-2.19 has been
estimated using Eq. 2.36. In Eq. 2.47, the enthalpy of water at the reference temperature, h0,R
is defined as: h0,R (0°C) = 0 kJ/kg and h0,R (25°C) ≈ 104.8 kJ/kg. Point I in Figure 2.12
corresponds to the heat of mixing value, ∆hM(cCS, ϑR = 0°C). The sources of heat of mixing
values, ∆hM, and the reference temperature, ϑR, for each diagram are as follows:
Figures 2.12 and 2.13 - Sodium chloride (ϑR = 0°C): ∆hM from Bošnjakovic (1961);
Figure 2.14 - Ethylene glycol (ϑR = 25°C): ∆hM from Heats of Mixing Data Collection
(1986);
Figure 2.15 - Propylene glycol (ϑR = 25°C): ∆hM from Heats of Mixing Data Collection
(1986);
Figure 2.16 - Ethyl alcohol (ϑR = 0°C): ∆hM from Landolt-Börnstein, (1976);
Figure 2.17 - Ammonia (ϑR = 10°C): ∆hM from Landolt-Börnstein, (1976);
Figure 2.18 - Potassium formate (ϑR = 25°C): ∆hM from Norsk Hydro Research Center
(2001);
Figure 2.19 - Calcium chloride (ϑR = 0°C): ∆hM from Bošnjakovic (1961).
To illustrate this type of diagrams, let us examine Figure 2.12 for a mixture of sodium
chloride-water. The concentration, cA, refers here to the concentration of pure sodium chloride
in water. The vertical axis gives the enthalpy, h, expressed in kJ/kg mixture. The freezing
point curve is represented in the figure by the bold line, htF, or D – E, where E is the eutectic
42
point. The diagram also shows some (thin) isothermal lines (20°C, 10°C, 0°C, –2°C, –4°C,
etc.). Let us follow as an example the cooling process from point A where ϑA = 0°C for a
mixture with the freezing point temperature ϑF = –5°C. The additive concentration is then
cA ≈ 0.079 (7.9% NaCl by wt). The temperature decreases when heat is removed from the
mixture and the freezing point is passed at point F and it has reached ϑK = –10°C at point K.
Small ice particles can be formed in an ice generator device during such cooling below the
freezing point of a mixture with lower concentration than the eutectic.
As the ice slurry solution can be viewed as consisting of two components: i) ice crystals of
pure water that freeze out and the rest of the solution with a higher concentration of the
additive, the point K thus represents a condition of mixture between pure ice of –10°C (at
point H), and ii) a concentrated solution of the condition at C (cCS ≈ 0.141, ϑF = –10°C). The
amount of heat that has to be removed for each kg of original solution can be seen as the
enthalpy difference hA-hB ≈ 176 kJ/kg. Of this enthalpy difference only 10% is from the
cooling of the liquid from 0°C to –5°C (between A and F), while 90% is from the cooling
from –5°C to –10°C between F and K where the freezing of ice particles takes place (see
dashed line A - F - K). This illustrates well the potential for ice slurries, e.g., for cooling
cabinet applications, and the potential is greater for even lower additive concentrations and at
temperatures just below the freezing point. The lines for ice concentration are obtained by
dividing each two-phase isothermal line in parts between the freezing point curve and the
corresponding enthalpy value of ice. The ice concentration that can be expected at K is given
in the figure as cI ≈ 0.43 (or 43% ice) (see Melinder, 2001).
The enthalpy lines in Figures 2.13-2.19 connecting the temperatures 2, 5 and 10 Kelvin below
the freezing points, are marked htF-2, htF-5, and htF-10. To see how they have been made, note in
Figure 2.12 the dashed lines for htF -htF-5 for the freezing points tF (ϑF) = –3°C,
–5°C, –7°C and –10°C (the line for –5°C continues up to 0°C). These enthalpy lines can help
us estimate the equilibrium potential of performance of the aqueous solution when used as an
ice slurry. A large enthalpy difference indicates a big potential for its “heat transport
capability”.
43
100 100
20°C .20°C 20°C 20°C
Liquid only
50 10°C 10°C 10°C 10°C
50
D 0°C A 0°C
0 0°C 0°C
F
-2°C htF L -10°C 0 -5°C
-4°C -6°C htF
-2°C -4°C -10°C
-10°C -6°C -8°C
-50 -14°C
cI = 0,2 -20°C -10°C
E -50
cI = 0,2 -15°C
Enthalpy, h [kJ/kg] .
cI = Ice conc. E
Enthalpy, h [kJ/kg] .
-100 htF-2
-20°C -21°C
htF-5 -100
cI = 0,4 -21°C
cI = Ice conc.
-150
cI = 0,4 htF-5
K
-150
-200 cI = 0,6
Eutectic area htF-10
≈ -21°C -200 cI = 0,6
-250
Eutectic area
cI = 0,8 ≈ -21°C
-8°C
-300 M -250
cI = 0,8
G -21°C Solid area
-350 -300
H
I Melinder, 2004 -21°C
Melinder, 2004
-400
-350
0 0,05 0,1 0,15 0,2 0,25 0 0,05 0,1 0,15 0,2 0,25
Concentration [kg NaCl / kg solution] Concentration sodium chloride, (cA = cNaCl)
Figure 2.12. Schematic enthalpy phase diagram (NaCl - H2O) Figure 2.13. Enthalpy phase diagram for sodium chloride - water
100 100
20°C Melinder, 2002 Melinder, 2002
20°C
20°C
50 10°C 50
10°C 10°C
10°C
0°C 0°C
0 0°C 0
0°C
-5°C
-2°C htF -2°C -5°C
-4°C -10°C -4°C htF
-50 -6°C -8°C -50 -6°C -10°C
-10°C -8°C
cI=0,2 -5°C cI=0,2 -5°C -10°C
Enthalpy, kJ/kg
Enthalpy, kJ/kg
-15°C
htF-2 -15°C
-100 cI=Ice conc. -100 cI=Ice conc. htF-2
cI=0,4 htF-5 -20°C cI=0,4 -20°C

htF-5
-150 -150
-25°C
-25°C
htF-10 htF-10
-200 -30°C -200
cI=0,6 cI=0,6 -30°C
-35°C
-35°C
-40°C -250 -40°C
-250
cI=0,8 cI=0,8
-300 -300
-350 -350
0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 0 0,1 0,2 0,3 0,4
Concentration ethylene glycol, (cA = cEG) Concentration propylene glycol, (cA = cPG)
Figure 2.14. Enthalpy phase diagram for ethylene glycol – water Figure 2.15. Enthalpy phase diagram for propylene glycol – water
45
100 100
Melinder, 2002 Melinder, 2002
20°C
50 20°C 50
30°C
10°C
10°C
0 10°C 0 20°C
0°C
0°C
-2°C 0°C -2°C 10°C
-50 -4°C -5°C -50 -4°C -5°C
htF
cI=0,2 -6°C -6°C
Enthalpy, kJ/kg
Enthalpy, kJ/kg
-5°C -8°C -10°C cI=0,2 0°C
-10°C -5°C htF
-100 cI=Ice conc. -100 -8°C
htF-2 cI=Ice conc. -10°C
-15°C -10°C
cI=0,4 cI=0,4 htF-2
-150 -150
-20°C -15°C
htF-5
-200 -200
cI=0,6 -25°C cI=0,6 htF-5
htF-10 -20°C
-250 -30°C -250

htF-10 -25°C
cI=0,8 -35°C cI=0,8
-300 -300
-30°C
-35°C
-350 -350
-40°C
0 0,05 0,1 0,15 0,2 0,25 0,3 0 0,02 0,04 0,06 0,08 0,1 0,12 0,14
Concentration ethyl alcohol, (cA = cEA ) Concentration ammonia, (cA = cNH3)
Figure 2.16. Enthalpy phase diagram for ethyl alcohol – water Figure 2.17. Enthalpy phase diagram for ammonia - water
46
100 100
20°C Melinder, 2002 Melinder, 2002
50 10°C 10°C 50
0°C 30°C
0°C
0 0
-5°C 10°C
-2°C htF -10°C
-4°C -6°C 20°C
-8°C-10°C
-50 -50 -2°C 0°C
-5°C -15°C 10°C
cI=0,2 htF-2 cI=0,2
Enthalpy, kJ/kg
Enthalpy, kJ/kg
-4°C
htF
cI=Ice conc. -20°C 0°C
-100 -100 cI=Ice conc. -5°C -6°C
htF-5 -8°C -5°C
-10°C
cI=0,4 -25°C cI=0,4 htF-2 -10°C
-150 -150
htF-10 -30°C
-15°C
-200 cI=0,6 -35°C -200 htF-5
cI=0,6
-40°C
-20°C
-250 -250 htF-10
cI=0,8 -25°C
cI=0,8
30°C
-300 -300
-350 -350
0 0,05 0,1 0,15 0,2 0,25 0,3 0 0,04 0,08 0,12 0,16 0,2
Concentration potassium formate, (cA = cKFo ) Concentration calcium chloride , (cA = cCaCl2)
Figure 2.18. Enthalpy phase diagram for potassium formate - water Figure 2.19. Enthalpy phase diagram for calcium chloride - water
47
2.6 Safety, ecology and corrosion aspects of aqueous solutions

by Cecilia Hägg and Åke Melinder
2.6.1. GENERAL CHARACTERISTICS OF ADDITIVES USED AS AQUEOUS SOLUTIONS
A number of characteristics relating to safety, ecology and corrosion of aqueous solutions

need consideration. Some important aspects are flammability, toxicity, acidity, corrosiveness
and material compatibility. A study on some of these general characteristics of the additives
in aqueous solutions has been done and some results are listed in Table 2.5. A general
explanation of the symbols is given after the table. Comments are then given for each type of
aqueous solution followed by some remarks on corrosion behaviour and material
compatibility. These general characteristics can be viewed as valid for single-phase as well as
two-phase ice slurry applications though the concentrations needed for ice slurry are lower
than for single-phase.
Table 2.5. General characteristics of the listed additives
Potassium
Potassium
Potassium
Carbonate
Propylene
Ammonia
Ethylene
Chloride
Calcium
Formate
Alcohol
Alcohol
Acetate
Methyl
Glycol
Glycol
Ethyl
Health 1 0 0 1 3 - 1 - 2
Flammability 1 1 3 3 1 - 0 - 0
Reactivity 0 0 0 0 0 - 0 - 0
Hazard marks Xn - F F, T C, N Xi - - Xn
Poison Class CH
4 F F 3 - F 5 - 4
WGK 1 1 1 1 2 1 1 0 1
19400-
LD 50 oral rat 4700 6200 5628 350 1000 32050 5500 1870
36000
1.8- 2.4-
Explosion limit 3.5-15 5.5-44 - - - - -
12.8 17.4
Flash point 116 99 12 - - - >250 - -
2.6.2. GENERAL EXPLANATION OF SYMBOLS
The values for Health, Flammability and Reactivity range from 0 to 3. For instance,
flammability 0 means that the chemical is not flammable at all and flammability 3 means
highly flammable.
Descriptions of the Hazard marks are: Xn means harmful, Xi means irritating, F means
flammable, T means toxic, C means corrosive, and N means dangerous for the environment.
Poison class CH is a classification of the chemicals according to Swiss toxicity law.

Descriptions of the symbols are: 3 means strong toxins, 4 means substances and products that
must be considered harmful, 5 means substances and products with a low hazard potential and
F means not subject to toxicity classification.
48
WGK is a classification, according to the national regulation of the Federal Republic of

Germany, of the degree of water pollution. The denominations are: WGK 1 means slightly
water polluting substance and WGK 2 means water polluting substance.
LD 50 oral rat is a toxicological value of lethal dose. It states the dose in mg of substance per
kg (of animal weight) that kills 50% of the laboratory test animals. The greater the LD50
value, the lower the toxicity of a substance. Substances, with an LD50 value greater than 2000
mg/kg are not classified as harmful. The LD50 values given by manufacturers are often those
obtained from literature for pure substances, and do not include the effects caused by the
inhibitors and stabilizers (Lethal dose for human; see 2.6.3).
Explosion limit is given in volume percent and flash point is measured in ºC, both for pure
substances. The above information is mainly based on Merck Chemical Databases and
Hazardous Substances Databank (HSB).
2.6.3. COMMENTS ON EACH TYPE OF AQUEOUS SOLUTION
Ethylene glycol (EG) is highly toxic for humans. The minimum lethal dose for humans is 1-
1.5 ml/kg or approximately 100 ml concentrated glycol in an adult. Short-term exposure can
result in irritation to eyes, skin and respiratory tract. Repeated or long-term exposure can
bring about effects on the central nervous system and eyes. EG is slightly flammable. EG has
good thermo-physical properties for cooling applications.
Propylene glycol (PG) is slightly water polluting and practically non-toxic to humans. The
minimum lethal dose for human adults is more than 15 times greater than that of ethylene
glycol. Propylene glycol has a low fire hazard when exposed to heat or flame. Propylene
glycol without suitable corrosion inhibitors can show a rather low pH value that might affect
the corrosion potential. PG has a very high viscosity at low temperatures.
Ethyl alcohol (ethanol) is highly flammable and is dangerous when exposed to heat and
flame. Vapours may form explosive mixtures with air. Most vapours are heavier than air.
Ethyl alcohol has a very low surface tension, which may cause leakage in sealing devices and
foaming. Mixtures of ethyl alcohol and water are considered non-flammable at concentrations
normally used as ice slurry. Classified as a substance with a fire risk at higher concentrations
because of the low flash point. May cause intoxication if consumed, denaturants added in
many countries. Exhibits high viscosity at low temperatures.
Methyl alcohol (methanol) is listed in Table 2.5 but should not be used in commercial
applications. It is highly toxic and may be fatal if inhaled, ingested or absorbed through skin.
Ammonia has a conspicuous smell, can cause strong irritation to eyes, skin and the
respiratory tract. Death can result if rapid escape is not possible. Swallowing liquid ammonia
is corrosive to mouth, throat and stomach. Exposures to high concentration can cause
temporary blindness and eye damage. Ammonia is highly toxic to aquatic organisms.
Ammonia vapour is lighter than air and there is a flammability risk. In spillage or disposal
most of the quantity evaporates into the atmosphere where it quickly decomposes.
Calcium chloride is not toxic, but causes irritation to nose and throat when inhaled and to
eyes when in contact.
49
Potassium acetate has a low hazard potential. Inhalation may cause mild irritation. The pH
values given for commercial products of potassium acetate are 8-11 (see also 2.6.4).
Potassium formate does not cause any skin, eye or respiratory irritation. The pH values
given for commercial products of potassium formate are 9 – 12 (See also 2.6.4).
2.6.4. CORROSION BEHAVIOUR AND MATERIAL COMPATIBILITY
Ethylene glycol, propylene glycol and ethyl alcohol: glycols seem to cause less general
corrosion on carbon steel than alcohols. The corrosion potential for these solutions is
comparatively the same as for the water the glycol has been mixed with. This means that
almost the same construction materials can be used for these solutions as the system for
water. However, after a long time use, the glycols tend to become slightly acidic because of
oxidation or decomposition, which implies an increase in the corrosion potential.
Ammonia is not compatible with copper. Ammonia is not compatible with elastomers made
from 100% soft rubber, isoprene (IR), natural rubber (GRS), polyurethane (AU) and silicon
rubbers (Schweitzer, 1996).
Calcium and sodium chloride based aqueous solutions show great signs of corrosion on
carbon steel when oxygen is present. They can cause pitting and crevice corrosion, as the
spot-like corrosion of stainless steel in seawater or sodium chloride solution. Chlorides
require buffering and at least two strong inhibitors, which have bad impact on the
environment. Corrosion inhibitors with chromates may cause a health risk. For chloride-based
solutions, it should be sufficient to use the same construction materials as for seawater.
Calcium chloride is not compatible with elastomers made from 100% soft rubber (Schweitzer,
1996).
Potassium acetate, potassium formate (and mixtures of these): parts containing zinc, zinc
alloys (such as galvanised surfaces) or aluminium should be avoided in systems with
potassium formate or potassium acetate solutions that have high pH-values (see 2.6.3).
However, the corrosion of aluminum by these organic salt solutions can be reduced by means
of inhibitors. Potassium acetate is not compatible with elastomers made from 100%
polyether-urethane (EU) (Schweitzer, 1996).
1. Aittomäki, A., Kianta, J, Indirect Refrigeration Systems Design Guide Book, Tampere,
2003.
2. Allen, T.: Particle Size Measurement, Fourth Edition, Chapman and Hall, ISBN 0 412
35 070, 1992.
3. ASHRAE, ASHRAE handbook fundamentals, ISBN 1-883413-88-5, 2001.
4. BASF, Glythermin®, Glycol-based heat- transfer fluids for heating and cooling system.
Brochure.
5. Bel, O.: Contribution à l’étude du comportement thermo-hydraulique d’un mélange
diphasique dans une boucle frigorifique à stockage d’énergie. Ph.D.-thesis, No. 96 ISAL
0088, L’Institut National des Sciences Appliquées de Lyon, France, 1996.
50
6. Bel, O., Lallemand, A. Study of a two-phase secondary refrigerant. 1. Intrinsic

thermophysical properties of an ice slurry, Int. Journal of Refrigeration, IIR/IIF, Paris, ,
22 p.164-174, 1999.
7. Bobkov V.A., Production and Application of Ice, Foods Industry Publ. House, Moscow,
p. 232, 1977.
8. Bošnjaković, F., Technische Thermodynamik, II Teil, Band 12, in Wärmelehre und
Wärmevirtschaft in Einzeldarstellung, Dresden/Leipzig, 1961.
9. CRC Handbook of Chemistry and Physics (edited by Weast, R.C.), 1989.
10. Egolf, P. W., Brühlmeier, J., Özvegyi, F., Abächerli, F., Renold, P.:
Kältespeicherungseigenschaften und Strömungsverhalten von binärem Eis.
Forschungsbericht zuhanden der Stiftung zur Förderung des Zentralschweizerischen
Technikums, June 1996.
11. Egolf, P. W., Frei, B.: The Continuous-prop erties Model for Melting and Freezing
applied to Fine-crystalline Ice Slurries. First IIRWorkshop on Ice Slurries, Yverdon-les-
Bains, Switzerland, p. 25-40, May 1999.
12. Egolf, P. W., Sari, O., Meili, F., Moser, Ph., Vuarnoz, D.: Heat Transfer of Ice Slurries in
Pipes. First IIR Workshop on Ice Slurries, Yverdon-les-Bains, Switzerland, p. 106-123,
May 1999.
13. Egolf, P.W., Vuarnoz, D., Ata-Caesar, D., Kitanovski, A.: Front Propagation of Ice Slurry
Stratification Processes. Fifth IIR Workshop on Ice Slurries, Stockholm, Sweden, May
2002.
14. Fletcher, N.H.: The Chemical Physics of Ice, Cambridge University Press, 1970.
15. Fournaison, L., Chourot, J.M., Guilpart, J.: Different ice mass fraction measurement
methods, Fourth IIR Workshop on Ice Slurries, Osaka, Japan, 2001.
16. Fukusako, S., Kozawa, Y., Yamada, M., Tanino, M.: Research and Development
Activities on Ice Slurries in Japan. First IIR Workshop on Ice Slurries, Yverdon-les-Bains,
Switzerland, p. 83-105, May 1999.
17. Guilpart J., Fournaison L., Ben Lakhdar, M.A. Calculation method of ice slurries
thermophysical properties - application to water/ethanol mixture, Int. Congress of
Refrigeration, IIR/IIF, Sydney, 1999.
18. Hagiwara, T.; Hartel, R. W., Effect of Sweetener, Stabilizer, and Storage Temperature of
Ice Recrystallization in Ice Cream. Journal of Dairy Science, vol. 79: 735-744, 1996.
19. Handbook of Thermodynamic Tables and Charts; Raznjevic, K., 1976.
20. Handbuch der Kältetechnik, VI/B (edited by Steimle, F.), 1988.
21. Hansen, T.M, Radosevic, M.; Kauffeld, M.: Behavior of Ice Slurry in Thermal Storage
Systems, ASHRAE Research Project 1166, published 2002.
22. Hazardous Substances Databank (HSB), Database of the National Library of Medicine’s
TOXINET system, http://toxnet.nlm.nih.gov, 06/10/02.
23. Heats of Mixing Data Collection, 1. Binary Systems, Christensen, J. J, Gmehling, J.,
Rasmussen, P., Weidlich, U., 1986.
24. Hobbs, P.V.: Ice Physics, Oxford University Press, 1974.
25. Inada, T; Lu, S. S.; Grandum, S.; Yabe, A.; Zhang, X.,. Microscale Analysis of Effective
Additives for Inhibiting Recrystallization in Ice Slurries Second IIR .Workshop on Ice
Slurries. Paris, France, May 2000.
26. Jeffrey D.J., Conduction through a random suspension of spheres, Proc. Roy. Soc.,
London A, vol. 335: p. 355-367, 1972.
27. Kang, C.; Okawa, S., Recent Study on Ice Melting and Water Freezing Using Molecular
Dynamics Method.1st report: Recent Study on Molecular Dynamics Method for Water/Ice
Behavior. Transactions of the JSRAE 16(1): 1-10, 1999.
51
28. Kauffeld, M., Christensen, K. G., Lund, S., Hansen, T. M.: Experience with Ice Slurry.
First IIR Workshop on Ice Slurries, Yverdon-les-Bains, Switzerland, p. 42-73, May 1999.
29. Kawashima, T., Saqsaki, M., Takahashi, H.: Experimental Studies of Snow-water Mixture
Flows in Horizontal Pipes. Twelfth Int. Conf. on Slurry Handling and Pipeline Transport,
Hydrotransport 12, Mechanical Engineering Publications Limited, London, 1993.
30. Kossel, Stransky or Burton, Carbera and Frank (BCF-model), Chapter 2 in: Myerson,
A.S.: Handbook of Industrial Crystallization, Butterworth-Heinemann, 1993.
31. Landolt-Börnstein: Numerical Data and Functional Relationships in Science and
Technology, Group IV, Vol. 2. Heats of Mixing and Solution, 1976.
32. List, R.: Private Communication. 2000.
33. Lock, G.S.H.: The growth and decay of ice, Cambridge University Press, 1990.
34. Lottin, O., Epiard, C. Thermodynamic properties of some currently used water - antifreeze
mixtures when used as ice slurries, Eighth Int. Refrigeration, Purdue, USA, p.391-398,
2000
35. Melinder, Å.: Köldbärare för värmepumptillämpningar, Rapport R114:1985,
Byggforskningsrådet, Stockholm 1985.
36. Melinder, Å., Berendson, J., Granryd, E., Köldbärare för värmepumpstillämpningar,
Rapport R18:1989, Byggforskningsrådet, Stockholm, 1989.
37. Melinder, Å., Thermophysical properties of liquid secondary refrigerants, Tables and
Diagrams for the Refrigeration Industry, Paris, IIF/IIR, 1997.
38. Melinder, Å. Thermophysical properties of liquid secondary refrigerants, Charts and
Tables, Stockholm, Swedish Society of Refrigeration, 1997.
39. Melinder, Å., Enthalpy-phase diagrams as an aid in estimating the performance of ice
slurries, Fourth IIR Workshop on Ice Slurries, Osaka, Nov. 2001.
40. Melinder, Å., Enthalpy-phase diagrams of aqueous solutions for ice slurry applications,
Fifth IIR Workshop on Ice Slurries, Stockholm, May 2002a.
41. Melinder, Å., Influence of temperature on latent heat benefits in ice slurry applications,
Zero leakage – Minimum Charge, IIF/IIR, Stockholm, 2002b.
42. Melinder, Å., Estimating the ”transport capability” of ice slurries with the aid of
enhanced enthalpy-phase diagrams, Fifth Gustav Lorenzen Conference on Natural Fluids,
IIF/IIR, Guangzhou, 2002c.
43. Merck, Merck Chemical Database, www.merck.de, 06/10/02.
44. Myerson, A.S; Ginde in Myerson, A.S.: Handbook of Industrial Crystallization.
Butterworth-Heinemann series in chemical engineering, 1992.
45. Norsk Hydro Research Centre, Internal report, 2001.
46. Nývlt: The Kinetics of Industrial Crystallization, Elsevier, 1985.
47. Pronk, P., Infante Ferreira, C.A., Witkamp G.C.: Effects of Long-Term Ice Slurry Storage
on Crystal Size Distribution. Fifth IIR Workshop on Ice Slurries, Stockholm, Sweden,
May 2002.
48. Sari, O., Meili, F., Vuarnoz, D., Moser, Ph., Egolf, P. W.: Thermodynamique d’un
écoulement de coulis de glace dans un tube à flux de chaleur constant. 2000 EUREKA
FIFE, research report No. 6, March 2000.
49. Sari, O., Meili, F., Vuarnoz, D., Moser, Ph., Egolf, P. W.: Thermodynamics of Moving
and Melting Ice Slurries. Second IIR Workshop on Ice Slurries, Paris, France, May 2000.
50. Schweitzer, Philip A., Corrosion engineering handbook, ISBN 0-8247-9709-4, New
York, 1996.
51. Thomas D.G., Transport characteristics of suspension, J. Colloid. Sci., vol. 20: p. 267-
277, 1965.
52. Tsvetkov O.B., Laptev Ju. A., Kolodiaznaia V.S., One-and two-phase refrigerating liquid
media, Kholodilnaja Teknika, no. 10: p. 8-12, 2001.
52
53. Turian, R.M. et al: Settling and Rheology of Suspensions of Narrow-Sized Coal Particles.
AIChE Journal, Vol. 38, No. 7, pp. 969-987, 1992.
54. Turnbull, D.; Fisher, J.C.: Rate of nucleation in condensed systems. Journal of Chemical
Physics 17: p. 71-73, 1949.
55. Vali, G.: Principles of Ice Nucleation, in: Lee, R.E.; Warren, G.J.; Gusta, L.V.: Biological
Ice Nucleation and its Application, APS Press, 1995.
56. Vargaftic, N.B., Thermophysical properties of gases and liquids. Handbook, Nauka Publ.
House, Moscow, 1972.
57. Vuarnoz, D., Sari, O., Egolf, P.W.: Correlations between Temperature and Particle
Distributions of Ice Slurry in a Storage Tank. Fourth IIR Workshop on Ice Slurries,
Osaka, Japan, Nov. 2001.
58. Washington, C.: Particle Size Analysis in the Pharmaceutics and Other Industries, Ellis
Horwood Publishing Company, 1993.
59. Yazaki, T.; Okawa, S.; Saito, Akio: “Molecular Dynamics Simulation Study of Ice
Crystal Growth”, Trans. of the JSRAE, Vol. 14, No. 2, pp. 179-190, 1997.
coatings for ice generation., Fifth IIR Workshop on Ice Slurries, Stockholm, May 2002.
53
30 Years of Leading Slurry Ice Innovation
Filed World’s 1st Slurry Ice Patent in 1976
Operating Systems Between 20 kW and 8000 kW per day in:

Seafood Processing Preservation • HVAC Thermal Storage • Supermarket Display Cooling • Produce
Preservation • Tranportation Cooling • Pharmaceutical Plants • Dairies & Bakeries • Meat & Poultry Cooling…
Additive-Free Slurries with Ice Fractions up to 80%
World’s #1 In Slurry Ice
Sunwell Technologies Inc.

TOLL 1.866.792.4455 TEL 905.856.0400 FAX 905.856.1935
inquiries@sunwell.com www.sunwell.com
CHAPTER 3. FLUID DYNAMICS
The operation of ice slurry systems aimed at the transfer of mass and heat between system
components, depends on the governing principles of fluid dynamics. Without a thorough
understanding of process hydrodynamics, the design of any ice slurry system becomes an art,
specific to the particular application. Fluid dynamics are involved in pumping and
transportation, sedimentation, mixing and separation, filtration and fluidized bed operations.
Unlike gases, which are nature’s simplest substances, liquids are more complex in behaviour,
and hence, cannot be described by a simple thermodynamic relation. If this is true for simple,
single-component liquids, the situation is even more complex for ice slurry. Ice slurry is a
mixture of a liquid phase (which again may consist of several components) and solid particles
in the form of ice, which even grow and decay during operation.
This chapter covers the rheology of ice slurry, i.e. the deformation of the fluid under the
action of stress, flows in tubes and their patterns, pressure drop and the time-dependent
behaviour of ice slurry due to the growth and decay of the ice particles.
3.1 Rheology
by Christian Doetsch, Andrej Kitanovski, Torben Hansen and Beat Frei (see specific list of
symbols in Appendix 3)
Homogeneous suspensions are frequently described as single-phase, isotropic fluids with

modified rheological behaviour. But instead of the viscosity of the liquid (ηL), a modified so-
called effective viscosity of the suspension (ηeff) is used. Many different models have been
used for the suspension viscosity determination. Most of them essentially extend the work of
Einstein on spheres and his equation for viscosity (C < 0.01) (Barnes, 1989):
ηeff = η L (1 + 2.5 ⋅ C ) (3.1)
where C is the concentration (or volume fraction) of the solid phase in a solid-liquid mixture.
In eqn. 3.1, there is no effect of particle size, nor of particle position, because the theory
neglects the effects of other particles. Several other models for viscosity determination have
been developed, as shown in Table 3.1.
The most popular determination of the suspension viscosity, which comprises not only the
concentration of the solid phase, but also the interaction between the solid particles, is based
on the well-known Thomas equation (Thomas 1965):
(
η eff = η L 1 + 2.5C + 10.05C 2 + 0.00273e16.6C ) (3.10)
55
Table 3.1. Formulae for viscosities of Newtonian suspensions as a function of the volume
fraction C of the solid phase (Darby 1986)
Guth and ⎛ 1 + 0.5 ⋅ C − 0.5 ⋅ C 2 ⎞ Predicts η → ∞ for C=0.234 (3.2)

η eff = η L ⎜⎜ ⎟
⎟
Simha ⎝ 1 − 2 ⋅ C − 9 .6 ⋅ C
2
⎠
Vand ⎛ 2. 5 ⋅ C ⎞ No interparticle forces (3.3)

η eff = η L ⋅ exp⎜ ⎟
⎝ 1 − 0.609 ⋅ C ⎠
⎛ 2. 5 ⋅ C + 2. 7 ⋅ C 2 ⎞
η eff = η L ⋅ exp⎜⎜ ⎟
⎟ Includes doublet collisions, (3.4)
⎝ 1 − 0.609 ⋅ C ⎠
but not triplet
Mooney ⎛ 2. 5 ⋅ C ⎞ K depends on the system (3.5)

η eff = η L ⋅ exp⎜ ⎟ 0.75 < K < 1.5
⎝1 − K ⋅C ⎠
Simha ⎛ ⎛ 25 ⋅ C ⎞ ⎞ Dilute suspensions (3.6)

η eff = η L ⎜⎜1 + 1.5 ⋅ C ⎜1 + ⎟.......⎟⎟ 1<f<2
⎝ ⎝ 4f ⎠
3
⎠
C → C max
⎛ 54 ⎛⎜ C ⎞⎞
η eff = η L ⎜1 + 3 ⎜
⎟⎟
⎜ 5 f ⎝ 1 − (C / C max )3 ⎟⎟ Very concentrated (3.7)
⎝ ⎠⎠
−1
Frankel ⎛ 1 ⎞ Concentrated suspensions (3.8)
⎜ ⎛ C ⎞3 ⎟
and η eff = η L ⋅ C ⎜1 − ⎜⎜ ⎟ ⎟ only
⎟
Acrivos ⎜ ⎝ C max ⎠ ⎟
⎝ ⎠
Jeffrey η eff = η L (1 + A ⋅ C ) 2.5<A<10 Ellipsoidal particles (3.9)
This model is valid for concentrations up to C = 0.625 and particle sizes ranging from 0.099
to 435 µm. It considers the flow to be homogeneous. The Thomas equation has been widely
used by researchers investigating ice slurry. However, this equation overpredicts the viscosity
of ice slurry at C > 15 %, as previously shown by Hansen (2000) who used Jeffrey`s equation
(3.9) with the constant A = 4.5 to get the best fit with his experimental results. Simultaneously
Frei et al. (2000) observed time-dependent behaviour of ice slurry, which resulted in different
sizes of ice particles and consequently different viscosities. Later, Hansen et al. (2002)
reported on the ice particle size distribution experiments at the Danish Technological
Institute.
In the past different experiments were performed in order to determine the effective viscosity
of ice slurry. Different measurement methods and different sizes of storage tanks led to quite
large differences in the results obtained (Figure 3.1).
56
Cemagref (Online visc.)

0.175 DTI (Online visc.)
DTI (Model)
0.150 UASCS (Rot. visc. in 1996)
Dynamic Viscosity (Pa s) UASCS (Rot. visc. in small storage)
0.125 UASCS (Online visc. with small stor.)
UASCS (Online visc. with main stor. 1)
0.100
UASCS (Online visc. with main stor. 2)
0.075 UASCS (Online visc. with main stor. 3)
0.050
0.025
0.000
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Ice fraction (-)
Figure 3.1. Eight different series of measurements of the effective viscosity, performed at
CEMAGREF, DTI and UASCS with rotary and online viscometry. The solid curve shows a
model calculation presented by Christensen et al. (1997). (see Frei and Egolf 2000)
The equations (3.1 to 3.10) describe the viscosity of suspensions which are regarded as
Newtonian fluids. The shear stress, τ, is related to the velocity gradient and described by:
dv
τ = η eff ⋅ (3.11)
dy
For higher fractions of the solid phase, suspensions usually show non-Newtonian behaviour,
where the viscosity is a function of the shear rate, as shown in Figure 3.2.
Figure 3.2. Rheogram for typical fluids with and without a minimum yield stress (τ0) and
shear thickening (n > 1) or shear thinning behaviour (n < 1)
As an approximation it is sometimes useful to determine the apparent viscosity:

dv
τ = η app ⋅ (3.12)
dy
57
where ηapp is a function of the shear rate. The rheological characteristics of a suspension are
described by different models, as presented in Table 3.2.
Table 3.2: Different models for laminar suspension flows
(Darby, 1986; Hanks, 1986; Steffe, 1992)
Bingham ⎛ dv ⎞ (3.13)
τ = τ 0 + η B ⋅ ⎜⎜ ⎟⎟
⎝ dy ⎠
Power Law (Ostwald-de-Waele) n1 (3.14)
⎛ dv ⎞
τ = K1 ⋅ ⎜⎜ ⎟⎟
⎝ dy ⎠
Casson (1959) 0.5 (3.15)
⎛ dv ⎞
τ 0.5 = (τ 0 )0.5 + ⎜⎜ηC ⎟
⎝ dy ⎟⎠
Herschel-Bulkley (1987) n2 (3.16)
⎛ dv ⎞
τ n1
= (τ 0 )
n1
+ K1 ⋅ ⎜⎜ ⎟⎟
⎝ dy ⎠
=K 1⋅τ + K 2 ⋅ (τ )n1
Ellis (1927) dv (3.17)
dy
Vocadlo (1968) n1 (3.18)
⎛ 1 ⎛ dv ⎞ ⎞
τ = ⎜⎜ (τ 0 ) n1 + K1 ⋅ ⎜⎜ ⎟⎟ ⎟⎟
⎝ ⎝ dy ⎠ ⎠
Whorlow (1992) 3 5 (3.19)
⎛ dv ⎞ ⎛ dv ⎞ ⎛ dv ⎞
τ = K1 ⋅ ⎜⎜ ⎟⎟ + K 2 ⋅ ⎜⎜ ⎟⎟ + K 3 ⋅ ⎜⎜ ⎟⎟ ......
⎝ dy ⎠ ⎝ dy ⎠ ⎝ dy ⎠
Carreau (1968) (n −1 ) (3.20)
⎡ ⎛ dv ⎞ ⎤
2 2
η = η ∞ + (η 0 − η ∞ ) ⋅ ⎢1 + ⎜⎜ K1 ⋅ ⎟⎟ ⎥
⎢⎣ ⎝ dy ⎠ ⎥
⎦
Cross (1965) η = η∞ +
(η 0 − η ∞ ) (3.21)
n
⎛ dv ⎞
1 + K1 ⋅ ⎜⎜ ⎟⎟
⎝ dy ⎠
Van Wazer (1963) η = η∞ +
(η 0 − η ∞ ) (3.22)
n1
⎛ dv ⎞ ⎛ dv ⎞
1 + K1 ⋅ ⎜⎜ ⎟⎟K 2 ⋅ ⎜⎜ ⎟⎟
⎝ dy ⎠ ⎝ dy ⎠
Rabinowitsch, Steiger/Ory dv (3.23)
= K1 ⋅ τ 3 + K 2 ⋅ τ
dy
Seely η = η ∞ + (η 0 − η ∞ ) ⋅ exp(− K1 ⋅ τ ) (3.24)
Sisko n1 (3.25)
dv ⎛ dv ⎞
τ = η∞ ⋅ + K1 ⋅ ⎜⎜ ⎟⎟
dy ⎝ dy ⎠
Prandtl-Eyring dv ⎛ τ ⎞ (3.26)
= −K1 ⋅ sinh⎜⎜ ⎟⎟
dy ⎝ K2 ⎠
Powell-Eyring (Powell and Eyring, dv ⎛ 1 ⎞ ⎛ dv ⎞ (3.27)
τ = K1 ⋅ + ⎜⎜ ⎟⎟ ⋅ sinh −1 ⎜⎜ K 3 ⋅ ⎟
1944) dy ⎝ K2 ⎠ ⎝ dy ⎟⎠
Reiner-Philippoff (Philippoff, 1935) ⎛ ⎞ (3.28)
⎜ ⎟
⎜
τ = ⎜η ∞ +
(η 0 − η ∞ ) ⎟ dv
⎟⋅
⎜ ⎛ τ 2 ⎞ ⎟ dy
⎜ 1 + ⎜⎜ ⎟⎟
⎟
⎝ ⎝ K1 ⎠ ⎠
58
To construct a rheogram for a particular suspension the shear rate at the wall must be known.
This may be determined by use of experimental data and the Rabinowitch (1929)–Mooney
(1931) equation:
⎡ ⎛ 8v ⎞ ⎤
⎢ d ⎜ ln ⎟ ⎥
⎛ dv ⎞ 8 ⋅ v ⎢3 1 ⎝ D ⎠ ⎥
⎜⎜ ⎟⎟ = ⋅ + ⋅ (3.29)
⎝ dy ⎠ w D ⎢ 4 4 ⎛ ∆p ⋅ D ⎞ ⎥
⎢ d ⎜ ln ⎟⎥
⎣ ⎝ 4 ⋅L ⎠⎦
A plot of the wall shear stress, τw, versus the shear rate at the wall usually shows which
rheological model is more accurate.
3.1.1 STATE OF THE ART OF THE ICE SLURRY RHEOLOGY
In the past different rheological models were used to describe ice slurry behaviour. Sasaki
(1993) considered ice slurry as a dilatant power law fluid. Then Frei and Egolf (2000) have
suggested applying a modified Bingham model. This Bingham model was also considered to
be useful by other authors (e.g. Jensen, 2000; Christensen, 1997). Experiments with ice slurry
have shown that it behaves as a Newtonian fluid at low ice concentrations and a non-
Newtonian fluid at high ice concentrations. Guilpart et al. (1999) have applied the Oswald-
type power law model to describe the rheological behaviour of ice slurry:
n( C )−1
τ ⎛ dv ⎞
η app = = K (C ) ⋅ ⎜⎜ ⎟ (3.30)
⎛ dv ⎞ ⎝ dy ⎟⎠
⎜⎜ ⎟⎟
⎝ dy ⎠
where:
0 ,737
n( C ) = 0.263 + 8.34
0 < C < 0.28 (3.31)
⎛ C ⎞
1+⎜ ⎟
⎝ 0.112 ⎠
K ( C ) = e (−5.441+832.4⋅C )
2.5
0 < C < 0.13 (3.32)
K ( C ) = e (−6.227 +16.487 ⋅C )
0.5
0.13 < C < 0.28 (3.33)
Another model was proposed by Christensen et al. (1997), who treated ice slurry as a
Bingham fluid. They determined the viscosity of the ice slurry by applying the Thomas
equation (eq. 3.10) and a yield stress, determined from experimental data:
τ 0 = 0.00059 ⋅ C 3 − 0.00701 ⋅ C 2 + 0.087 ⋅ C − 0.02498 (3.34)
Later, a similar approach was considered by Jensen et al. (2000):
a1 +b1⋅ x 20 + c1⋅C
τ0 = e (3.35)
where C represents the ice concentration, x0 represents the initial ethanol concentration (5, 10
and 20%) and the coefficients are a1 = -1.47, b1 = 0.0035 and c1 = 0.116.
59
In contrast to other researchers, Doetsch (2001) had a preference for the Casson model. He
used many different kinds of additives in his experiments. Therefore, the model which he
obtained may be taken to describe ice slurries with different kinds of carrier fluids. The
Casson model parameters were determined as a function of the carrier fluid viscosity and the
ice concentration.
( )(
η C = 0.005311 − 2.01656 ⋅ η L + 275.2413 ⋅ η L 2 ⋅ 1 + 0.357403 ⋅ C + 11.38022 ⋅ C 2 ) (3.36)
( )(
τ C = 2.381108 + 487.048 ⋅ η L + 4.53631 ⋅ 10 3 ⋅ η L 2 ⋅ − 1.30982 ⋅ C + 13.25046 ⋅ C 2 ) (3.37)
Doetsch’s (2001) results are shown in Figure 3.3.
Figure 3.3. Casson parameters ηC and τC with a dependence on the viscosity of the carrier
fluid (2 mPas < ηL < 10 mPas) and the ice particle concentration (0-45%) (Doetsch, 2001)
3.2 Flows in tubes and pattern formation

by Andrej Kitanovski, Didier Vuarnoz, Torben Hansen, Patrick Reghem
In slurry flows the influence of solid particles has to be considered, e.g. the shapes and the
sizes of the particles, their concentration and density, and also the rheological properties of
the mixture. Furthermore, the slurry behaviour depends on the particular system design (pipe
diameter and inclination) and operational conditions (velocity and temperature). The
visualization of ice slurry flows — in order to determine different flow patterns — is difficult,
because of the opacity of the ice slurry and in some cases because of the melting
phenomenon. The characterization of flow patterns is very important, however, in order to
60
obtain stable and safe operational conditions. For example, a flow as a moving bed or
stratification should be avoided. Such conditions may lead to blockage or flow pulsations in
certain parts of the piping system where low velocities occur.
The distinction between different flow patterns is usually made by visual observations.
Various reports on suspensions assign different names to particular flow patterns or the
transition boundaries and velocities, which distinguish flow patterns (Doron, 1996). The most
common classification of suspension flows relates to four flow patterns: “homogeneous”,
“heterogeneous”, “heterogeneous and sliding bed”, “saltation or stationary bed” (Doron,
1996). However, Turian and Yuan (1977) used the term “saltation” also for the flow pattern
of a “moving bed”. According to Doron (1996), some authors are also using “non-deposit
flow regime” and “regime with deposits”, while others refer to “stationary bed”, “partly
stationary bed”, “fully moving bed”, “heterogeneous flow”, “pseudo-homogeneous flow” and
“homogeneous flow”. It is obvious that the determination of the flow patterns depends on the
characteristics of the slurry, which varies as a function of numerous parameters (Figure 3.4).
Some authors are also using two different names for the transition velocity, in order to
classify the flow regimes of an ice slurry. Wasp (1977) denotes the velocity at which the
transition between laminar and turbulent flow occurs as the “transition velocity” and the
critical velocity at which a bed of particles begins to form “deposit velocity”. Roco and Shook
(1991) name this quantity “deposition velocity”. There are also numerous different terms
associated with the velocity at which the settling of particles occurs. Authors relate the
deposition velocity also with the minimal pressure drop (Doron, 1996).
Figure 3.4. Flow patterns of suspensions with S < 1 (see the list of symbols)
3.2.1 STATE OF THE ART OF RESEARCH ON FLOW PATTERNS
Investigations on ice slurry flow patterns began in the middle of the 1980s. The purpose was
to create a design guideline for ice slurry transportation systems. At that time the ice particles
in the mixtures were usually large, e.g. 12 mm (Sellgren, 1986; Takahashi et al., 1993).
61
Nowadays ice slurry generators usually produce very small ice particles (dp= 100 - 300 µm),
which may grow to larger sizes, depending on their life time in the ice slurry system. The ice
particles can also have different shapes and in some cases they agglomerate and form clusters
in the flow. When the flow is heterogeneous it is more difficult to define the properties of ice
slurry. Some researchers developed models for the friction factor of heterogeneous flows
(Takahashi et al., 1993) and evaluated different methods to calculate the velocity
distributions. New investigations are required in order to determine the properties for different
flow regimes. One of the innovations is the research on ice particle shapes, sizes and the
related time-dependant behaviour (Frei and Egolf 2000; Sari et al., 2000; Vuarnoz et al.,
2000; Hansen et al., 2002). Other new investigations are focused on velocity profiles
(Vuarnoz et al., 2000; Kawaji et al., 2001; Stamatiou et al., 2002) determined by different
methods, experimental and theoretical methods for determining the critical deposition
velocities (Kitanovski et al., 2001), concentration profiles (Kitanovski et al., 2002) and of the
pressure drops for heterogeneous flow regimes (Reghem, 2002). Some examples of flow
visualization are shown in Figures 3.5 and 3.6.
Figure 3.5. Visualization of ice slurry flows only allows a qualitative determination of flow
patterns (10% water-ethanol mixture, dp ~ 0.1 mm, Di = 44.6 mm, case A: C = 5%, v = 0.1
m/s, case B: C = 10%, v = 0.2 m/s). Work performed by P. Reghem et al. at Pau University
Figure 3.6. Visualization of ice slurry flows does not allow a very accurate determination of
flow patterns (10 % water-ethanol mixture, dp = 0.2 ~ 0.4 mm, Di = 20 mm, rectangular pipe,
case A: C = 7.6%, v = 0.54 m/s, case B: C = 7.5%, v = 0.23 m/s,
case C: C = 7.5%, v = 0.15 m/s). Work performed by Didier Vuarnoz et al. at EIVD
Up to the present, four flow patterns have been identified for ice slurry consisting of very
small ice particles (100-300 µm). Because some patterns were obtained by visual observation
of the flow, the results are subject to uncertainties. The four flow regimes identified are: a)
homogeneous flow; b) heterogeneous flow; c) moving bed flow; and d) the stationary bed.
Figure 3.7 shows the corresponding flow velocities.
a) Homogeneous flow
This flow pattern occurs at high flow rates, where the solid particles are uniformly distributed.
According to Wasp (1977), the criterion for homogeneity of slurry can be determined
by C/CA > 0.8, which is the ratio of the volumetric concentration of solid particles at 0.08 D
62
from the bottom of the pipe to the value on the pipe axis. The ratio depends on the particle
size.
Figure 3.7. Flow pattern diagrams from three different experiments performed at LTE - Pau
University and the Danish Technological Institute (DN 50, 10% water-ethanol)
b) Heterogeneous flow
At low flow rates the distribution of the ice particles is not uniform. More particles are
transported close to the top of the pipe.
Figure 3.8. Flow pattern “diagrams” performed at the Korean Institute of Energy Research
(Di = 24 mm, 6.5% ethylene glycol-water, d p = 0.27 mm) (Dong Won Lee et al., 2002)
63
c) Moving bed
At a certain flow rate the concentration of ice particles in the upper part of the pipe reaches its
maximum packing concentration and the particles start to accumulate. They form a layer,
which moves along the top of the pipe. In the lower part of the pipe cross section
heterogeneous flow occurs. Sometimes the lower part of the bed shows dune-like forms. In
some experiments large fluctuations have been observed, and the conditions may have been
unstable. Some experimental data are very difficult to interpret.
d) Stationary bed
The flow rate is too small to enable motion of the solid particles. The particle bed increases,
because particles are continuously added. The flow rate on the bottom of the pipe is much
higher than in the flows described in the previous subsections. The flow underneath the
stationary bed is heterogeneous, and in some cases on the lower part of the bed — similar to
the case when a moving bed occurs — wave-like structures also occur.
3.2.2 VELOCITY PROFILES
One way to study flow patterns is via the determination of velocity profiles. The velocity
profile of a laminar homogeneous flow may be determined by applying equations for a
Bingham fluid. Laminar flow conditions prevail in small diameter pipes with a sufficient
mean velocity. Even if the flow is homogeneous a difference between the solid and the liquid
phase velocity may occur. This is related to the so-called slip velocity, which may be
neglected if the flow is homogeneous. When the flow is heterogeneous the difference between
the velocities of the two phases becomes larger, and the velocity profiles become non-
symmetric: now the dynamic axis of the flow is located below the pipe axis. In the past
several authors presented models for the velocity distribution of heterogeneous flow. Most of
these models contain empirical correlations, which cannot be applied to ice slurry flows. Only
one model (Takahashi et al., 1991; Sasaki et al., 1993) refers to ice slurries, but the scale-
model experiments were performed with polystyrene particles of large diameters (d ~ 3 mm).
There are many different experimental methods to determine the velocity profile:
• Ultrasonic Doppler Echography
• Pitot tube
• Hot wire
• Observations using a high speed video camera.
The last method in the list is useless for investigating ice slurry flows, because of the opacity
of the fluid. With other methods one can detect only the velocity profiles of the solid or the
liquid phase, while the Pitot tube can determine the velocity of the mixture. Pitot tube
experiments are, however, very difficult to perform. This measuring technique is only
accurate if the ice concentration is small. Thus, the data obtained with Pitot tubes are usually
not of high quality compared to other methods.
It is obvious that at a high velocity, e.g. vslurry = 0.8 m/s, the flow is homogeneous (see Figure
3.9), because it is observed that the dynamic axis is almost at the same position as the axis of
the pipe. If the velocity of the mixture is lowered, the ice concentration near the top of the
pipe increases, and the dynamic axis approaches the bottom of the tube. The first velocity
profile in Figure 3.9 shows that the velocity at the bottom increases so much that local
64
turbulence is generated, while the ice slurry near the top moves slower and indicates a flow
with a moving bed.
Figure 3.9. Velocity profiles of ice slurry flows (7% ethanol-water mixture, Cr = ~20%, DN
50, dp ~ 0.1 mm, dashed lines are approximate predictions (Kitanovski et al., 2002)
To describe the dispersion of ice particles (e.g. concentration distribution) in an ice slurry
flow, it is more convenient to use the velocity profile of the solid phase. Such experiments,
which provide us with data on the particle velocity profiles (Figure 3.10), were performed at
the University of Applied Sciences of Western Switzerland.
Figure 3.10. Ice particle velocity profiles for ice slurry flow (11% ethanol-water mixture,
Cr = 12, 10, 11%, DN 25, dp ~ 0.1 mm) measured by Vuarnoz et al. (2000)
According to Figure 3.10 homogeneous flow conditions are observed if the slurry velocity is
1.5 m/s. When the velocity is lower, e.g. vslurry = 1.2 m/s, the dynamic axis is a little bit lower
than that of the pipe axis, but the flow can still be considered homogeneous. A further
decrease in velocity leads to a heterogeneous flow, which can obviously be recognized in the
first profile of Figure 3.10. Here the velocity of the ice slurry was 0.25 m/s. The plug velocity
is not constant anymore, because the concentration distribution affects the local viscosity and
local “yield stress”, which increases with a higher local ice concentration.
65
3.2.3 CONCENTRATION PROFILES
The majority of the analytical approaches for the concentration distribution are based on the
concept of turbulent diffusion, originally proposed by Schmidt and Rouse (Doron et al.,
1987):
dC
ES + w ⋅C = 0 (3.38)
dy
where ES is a diffusion coefficient or mass transfer coefficient of solid particles and w is the
terminal settling velocity of the ice particles in a quiescent fluid. In eq. (3.38), only the
equilibrium between the net gravitational force and the lifting force due to turbulence is
considered. Shook (as reported by Nasr-El-Din et al., 1987) found from his experiments with
slurries of higher concentrations that the concentration gradient predicted by eq. (3.38) is
much too low. According to Nasr-El-Din et al. (1987), this may be approximately corrected
using the hindered settling multiplier, which is obtained from the Richardson and Zaki (1954)
correlation. The interactions between the solid particles (dispersive stress) are neglected and
this is important when the concentrations are high or the particles are large. Bagnold (1954)
has even observed a negative slope of the concentration distribution (on the bottom of the
pipe, with the maximum concentration at some distance above) in a laminar flow of a
suspension. A very comprehensive model was proposed by Roco and Shook (1987). The
model is based on the momentum equations, and the differential equation for C(y) is given by:
E S 2 d 2C
+
s ⋅C
⋅
(
∂2 αs ⋅vs2 )
+ C2 = 0 (3.39)
w 2
dy 2 (s − 1) ⋅ g ⋅ tan φ ∂y 2
where vs is the velocity of the solid particles, αs is an experimentally determined turbulence

coefficient and φ is the angle of internal friction of the dispersive stress.
The concentration distribution of the ice slurry depends on many parameters: properties of the
liquid phase, density of the solid particles, the pipe diameter, the diameter and shape of the ice
particle, the slurry velocity, the concentration of the solid phase and the friction during the
transport of the slurry. All these parameters are related to the flow patterns. A very simple
prediction method for the concentration is the following proposed by Doron (1987), and
described by Kitanovski et al. (2002) for ice slurry flows:
w
− ⋅y
C( y ) = C max ⋅ e ES
(3.40)
According to other authors (Roco and Shook, 1987), a similar equation may be applied to ice
slurry flows at low ice concentrations, namely CS < 15 % (Figure 3.11):
CS ⋅ A
C max = (3.41)
⎛ w⎞
⎜ − ⎟⋅y
∫
A
e ⎝ ε ⎠ dA( y )
66
Figure 3.11. Concentration profiles of ice slurry flows in a horizontal pipe calculated
with eq. 3.40 (D = 27.2 mm, dp = 1 mm, 10% water-ethanol) (Kitanovski et al., 2002)
3.2.4 CRITICAL DEPOSITION VELOCITY
The transition between a flow with a moving bed and heterogeneous flow may be
characterized by the critical deposition velocity. Several ice slurry research pioneers (e.g.
Snoek, 1993) had observed an increase in pressure drop, when the ice slurry velocity was
below the critical deposition velocity. The critical velocities were high, because of large
particles, which were produced in these experiments. Newer experiments, with ice slurries
containing smaller ice particles, have shown that the critical deposition velocities are much
lower. In small pipes there is no clear boundary between the moving and the stationary bed.
There are several approaches available to determine the deposition velocity:
a) by evaluation of concentration profiles,
b) by evaluation of velocity profiles,
c) by observation,
d) using empirical functions.
However, all the approaches require experimental data. From the engineering point of view,
the empirical correlation method is one of the simplest approaches, because it permits fast and
simple calculations of the deposition velocity. But the estimation is less accurate than in other
approaches, and the result is only a rough approximation of real conditions. Some authors
report for this method deviations of up to ± 20% (Shook and Roco, 1991).
a) determination by evaluation of concentration profiles:

A boundary condition for C (y = D) may be defined to specify the maximum concentration in
the top region of the pipe. A moving bed forms when a maximum packing fraction occurs at
the top of the pipe, because no further particles can be added to the upper layer. Maximum
packing fractions are mainly determined by the arrangement of the particles, as shown in
Table 3.3.
Researchers usually assume in numerical simulations a maximum packing fraction of 63%.

However, one must note that such a concentration corresponds to steady flow conditions. In
the case of a sedimentation process, the bed is more porous, so a factor for the dynamic
sedimentation process should be implemented (according to Roco and Shook, 1987). This
factor is approximately 0.9. In the case examined here, the maximum ice packing fraction was
assumed to be equal to 0.52. Equation (3.41) was applied to the determination of the
67
concentration distribution. The deposition velocities for ice slurry flow (10% ethanol-water)
predicted by Kitanovski (2002) are shown in Figure 3.12.
Table 3.3. The maximum packing fraction of various arrangements of dispersed spheres
(Barnes, 1989)
Arrangement Maximum packing fraction
Simple cube 0.52
Maximal thermodynamically stable configuration 0.548
Hexagonally packed sheets just touching 0.605
Random close packing 0.637
Face-centred cubic / hexagonal close packed 0.68
Body-centred cubic / hexagonal close packed 0.74
0,7 D (mm)
200
0,6
100
0,5 27.2
Vdeposition (m/s)
0,4
dp (mm)
0,3 1
0,2 0.5
0,1 0.25
0
5 10 15 20 25
CS (%)
Figure 3:12. Deposition velocities for ice slurry flow – theoretical model
(10% ethanol-water) (Kitanovski, 2002)
From Figure 3.12 one can determine the deposition velocities for different sizes of ice
particles and different pipe diameters. This theoretical model is in good agreement with the
results of Hansen (2002). However, the values are too low, if they are compared to the
experimental data of Reghem and Kitanovski (see Figure3.7a). It is possible that because the
ice particles are very small, the boundary between the moving bed and stationary bed is not
very clearly detectable. Thus, additional experiments are required for horizontal pipes and
inclined pipes.
b) determination by evaluation of velocity profiles

The determination of the critical deposition velocity by interpreting velocity profiles is a very
accurate method, however, the procedure and experiments involved need to be further
developed.
68
c) determination by observation
Figure 3.8 shows visualization of ice slurry flow patterns and flows with moving beds.
d) determination using empirical functions

There exist several empirical correlations for the suspension transport (Wasp, 1977; Shook
and Roco, 1991; Darby, 1986), but they cannot be applied to ice slurry flows, because the
constants in the equations fit the conditions only for some specific experiments. One of the
best known empirical functions was developed by Durand (Wasp, 1977):
Vdeposition = FL ⋅ (2 ⋅ g ⋅ D ⋅ (S − 1)) (3.42)
where S represents the ratio between the densities of the solid particles and the carrier fluid,
and FL refers to the specific operating parameters as a function of particle size and
concentration. Using the experimental conditions shown in Figure 3.7a for ice slurry flow
(pipe DN 50, 10% ethanol-water, dp = 0.1 mm, R2 = 0.98), the following empirical correlation
was obtained (Kitanovski et al., 2002):
−0.17833 −0.8383
Vdeposition ⎛D ⎞ ⎛ w ⋅ 10 − 4 ⎞
= 0.397756 ⋅ C r ⋅ (1 − C r ) ⋅⎜ ⎟ ⋅ ⎜⎜ ⎟⎟
0.3 3.756
(3.43)
⎜d ⎟
2 ⋅ g ⋅ D ⋅ (1 − S) ⎝ p ⎠ ⎝ νf ⎠
where w presents the hindered terminal settling velocity. The most applied correlation is the
one proposed by Richardson and Zaki (1954).
Figure 3.13. Deposition velocities obtained from an empirical function

(eq. [3.43]).(Kitanovski et al., 2002)
Equation (3.43) gives the most accurate results for small ice particles and small pipe
diameters (Dpipe < 50 mm), because the correlation was developed for these conditions.
Furthermore, it cannot be used for extrapolated conditions of different ice particle and pipe
diameters. As shown in Figure 3.13, the deposition velocity increases with increasing ice
concentration to some finite value and then decreases again. A reason for this is the influence
of the carrier fluid viscosity, which changes with ice concentration and influences the
dynamic conditions. Similar results (when smaller particles are considered) are presented in
Figure 3.12.
69
3.2.5 THEORETICAL DETERMINATION OF FLOW PATTERNS
Another possibility for the determination of ice slurry flow patterns is through the study of the
ice concentration profile, which can provide us, if the maximum packing fraction is
introduced, with data on flows with a moving bed. Using Wasp’s (1977) criterion for the
occurrence of homogeneous flow, one can also predict the heterogeneous-homogenous flow
transition as shown in Figure 3.14.
Figure 3.14. Flow pattern diagram for ice slurry flows obtained by applying a theoretical
model (DN 50, 10% water-ethanol) (Kitanovski et al., 2002)
3.3 Pressure drop

by Andrej Kitanovski, Christian Doetsch, Torben Hansen, Patrick Reghem, Beat Frei, Didier
Vuarnoz, Issa Chaer
Many different experiments have shown that ice slurry behaves as a Newtonian fluid at low
ice concentrations, and as a non-Newtonian fluid at higher ice concentrations. The most
widely applied rheological model for ice slurries at high ice concentrations is the Bingham
model. For a homogenous flow, the equations in cylindrical coordinates are given by (Sari et
al., 2000):
Continuity equation
∂ρ ∂ρ ∂T ∂v
+ v+ρ =0 (3.44)
∂t ∂T ∂x ∂x
Momentum equation
∂v ∂v ∂p η ∂v 1 dτ ∂T dη ⎛ ∂v ∂T ⎞ ∂ 2v
ρ + ρv + − − τ0 − 0 − ⎜ ⎟ −η 2 = 0 (3.45)
∂t ∂x ∂x r ∂r r dT ∂r dT ⎝ ∂r ∂r ⎠ ∂r
70
3.3.1 ISOTHERMAL, STATIONARY, HOMOGENEOUS ICE-SLURRY FLOW
For isothermal flows the temperature is constant and the energy equation is not considered.
The pressure drop of such a flow may be calculated by the Darcy-Weisbach equation:
L ρv 2
∆p = λ ⋅ (3.46)
D 2
where λ is the friction factor, which may be calculated by some semi-empirical or empirical
model (see Table 3.4 for details).
Table 3.4. Friction factors for Bingham, Power Law and Casson models.
Laminar flow is assumed (Darby, 1986)
Bingham ⎛ ⋅ ⎞
1 ⎛⎜ τ 0 ⎞⎟
4
⎜ 4 ⋅V ⎟ 4
τ w = ηB ⋅ ⎜ 3 ⎟
+ ⋅ τ − ⋅ (3.47)
⎜π ⋅R ⎟ 3
0
3 ⎝τw ⎠ ⎜ 3⎟
⎝ ⎠
64 ⎡ He He 4 ⎤
λ= ⋅ ⎢1 + − ⎥
ReB ⎣⎢ 6 ⋅ ReB 3 ⋅ λ ⋅ ReB ⎦⎥
3 7
(3.48)
D2 ⋅ ρ ⋅τ 0
He =
ηB 2
(3.49)
Power Law 64
λ=
Re0 (3.50)
v (2 −n ) ⋅ D n ⋅ ρ 1
Re0 = ⋅ n
K ⎛ 1 + 3n ⎞
8 n −1 ⋅ ⎜ ⎟
⎝ 4n ⎠
(3.51)
Casson ⎛ ⎞ ⋅
⎜ 4 ⋅V ⎟ 4 τC4
τ w = K1 ⋅ ⎜ 3 ⎟
+ ⋅ τ C −
16
⋅ (τ C ⋅ τ w )0. 5
+ (3.52)
⎜π ⋅R ⎟ 3 7 21 ⋅ τ w 3
⎝ ⎠
λ=
64 ⎡
⋅ ⎢1 −
Ca
+
(2 ⋅ λ ⋅ Ca )0.5 + Ca 4 ⎤
⎥
ReC ⎣⎢ 6 ⋅ ReC 7 21 ⋅ λ3 ReC ⎦⎥
7
(3.53)
D 2 ⋅ ρ ⋅τ C
Ca =
ηC 2
(3.54)
Here, Ca, is the Casson number. For the transition of a Newtonian fluid between laminar and
turbulent flow a criterion is that laminar flow occurs if Re < 2100 and that transition flow
occurs when 2000 < Re < 3000. Hanks (reported by Steffe, 1992) proposed a criterion for this
transition for Bingham fluids:
⎛ ρ ⋅v ⋅D ⎞ ⎡ 4 ⎤
⎟⎟ = 2100 ⋅ ⎢1 − ⋅ α + ⋅ α 3 ⎥ ⋅ (1 − α )−3
1
Rek = ⎜⎜ (3.55)
⎝ η B ⎠ ⎣ 3 3 ⎦
where the coefficient α is determined by:

α He
= (3.56)
(1 − α ) 16800
3
71
This criterion may be applied also to other “Bingham-like models”, such as the Casson model.
The critical Reynolds numbers are shown as a function of Hedstrom number, He, in Figure
3.15.
25000
20000
15000
Re
10000
5000
0
10 100 1000 10000 100000 1000000
He
Figure 3.15. Critical Reynolds number as a function of the Hedstrom number

(Eq. 3.55 + 3.56)
For the friction factor determination of a Power Law fluid in turbulent flow, Dodge and
Metzner (reported by Steffe, 1992) proposed the following empirical correlation:
⎛ (1−0.5 ⋅n ) ⎞
1 2 ⎛λ⎞ ⎟ − 0.2
= log ⎜ Re0 ⋅ ⎜ ⎟ (3.57)
λ n 0.75 ⎜ ⎝4⎠ ⎟ n 1 .2
⎝ ⎠
For a Bingham fluid, the friction factor in turbulent flow may be calculated as given by
Govier and Aziz (also reported by Steffe, 1992):
1 ⎛ τ
= 2.265 ⋅ log10 ⎜⎜1 − 0
⎞
( )
⎟⎟ + 2.265 ⋅ log10 ReB ⋅ λ − 1.832 (3.58)
λ ⎝ τw ⎠
For He > 1000 a relation is given by Hanks (1996):
( )α
1
λ = λαL + λTα (3.59)
where the laminar component, λL, is calculated by using an equation from Table 3.4. The
turbulent component (term), λT, in Eq. (3.59) is calculated by:
4 ⋅ 10 a
λT = 0.193
[ (
a = −1.378 ⋅ 1 + 0.14 ⋅ exp − 2.9 ⋅ 10 −5 ⋅ ReB )] m = 1.7 + 40000
Re
(3.60)
Re B
Calculations by Kitanovski (2002) show that Eq. (3.58) may also be useful in combination
with the Casson model.
72
3.3.2 STATE OF THE ART OF ICE SLURRY PRESSURE DROPS
Earlier investigations did not consider rheological models as a basis for the friction factor
evaluation of laminar and turbulent ice slurry flows. Usually only empirical correlations were
determined. Snoek (1993) did not notice large pressure drop differences in flows with ice
concentrations Cr < 0.1. However, higher ice concentrations resulted in higher pressure drops.
Minimal pressure drops occur at velocities at which the transition between heterogeneous
flow and a flow with a moving bed occurs. Winters and Kooy (as reported by Snoek, 1993)
investigated the pressure drop of ice slurry at velocities above 3 m/s and ice concentrations
over 20%. They did not observe a large difference between the pressure drop of the ice slurry
and that of the carrier fluid (water). However, higher pressure drops are observed with ice
slurry than with pure water, at velocities lower than 1 m/s. Larkin and Young (as reported by
Snoek, 1993) as well as Sellgren (1986) discovered that at velocities in the range 1.0 to 1.5
m/s, the friction factor of the ice slurry (dp = 1.2 cm and Cr < 0.15) is close to that of pure
water. For lower velocities, Sellgren (1986) observed a separation of two phases and an
increase in the pressure losses. Larkin and Young (as reported by Snoek, 1993) performed
experiments at ice concentrations 0 < Cr < 0.25, but did not report large alterations in the
results. At higher velocities the pressure drop rapidly increases. They also observed a
separation of two phases at lower velocities. Knodel et al. in France (as reported by Snoek
1993) reported on reduced pressure drops at higher ice concentrations. Their ice slurry
contained large ice particles (3-6 mm), the velocities varied from v = 1.5 to 3.5 m/s and the
ice concentration was 0 < Cr < 0.15. Takahashi (1993) also conducted experiments on
pressure drops of ice slurry, which contained pure water and crushed ice (dp = 12.5 mm) at a
concentration of Cr = 0.25. The results showed higher pressure drops for ice slurry than for
water at normal operational conditions. At high velocities the pressure drop of ice slurry was
even lower than that of water. This could be explained by a drag reducing effect. Early
investigations showed large deviations in experimental data. The ice particles in these
experiments were rather large, so the results cannot be applied to modern fine-crystalline ice
slurries. Today ice particles in ice slurry systems have a characteristic size between dp = 0.1 to
1 mm. There are also different additives in use nowadays that were not used previously.
Frei and Egolf (2000) observed a change in the pressure drop as a function of time during
pressure drop measurement. Even when the influence of a super-cooled or preheated carrier
fluid was eliminated, the pressure drop decreased with time to an asymptotic value, now
recognised as the time dependent behaviour of the shape and size of ice crystals.
During the last few years numerous researchers performed a large number of experiments to
obtain pressure drop data. Some of their work has been reported in the following references:
Knodel et al. (2000), Chaer et al. (2000/2001), Frei and Egolf (2000), Jensen et al. (2000),
Tassou et al. (2001), Guilpart et al. (1999), Hansen and Kauffeld (2000), Doetsch (2001),
Dong Wong Lee et al. (2002), Reghem (2002), Bel (1996), Christensen et al. (1997), Bellas et
al. (2002), and Egolf et al. (2001). Some of them presented empirical equations, while others
gave semi-empirical equations, which were based on rheological models, mostly the Bingham
or the Casson model. Even in recent experiments on pressure drops, there is a large difference
between the individual pressure drop measurements and the corresponding shear stresses (Frei
et al. 2000). There are many explanations for this phenomenon. One is due to the large
amount of data, which is available. Others are due to the time behaviour (different particle
sizes), different flow patterns (homogeneous-heterogeneous flow), material properties,
geometrical parameters, the effect of a preheated or a superheated carrier fluid, and heat
transfer, i.e. melting of ice particles during measurements.
73
3.3.3 FRICTION FACTOR FOR LAMINAR FLOW
Usually, the friction factors are determined by equations listed in Table 3.4. They depend on
rheological characteristics of the particular fluid or respective model. Therefore, each
parameter of a model has to be determined experimentally. Usually they are obtained by the
application of the Rabinowitch-Mooney equation (see Section 3.1). Several authors have
decided to use such an approach. But it requires a large amount of experimental data;
otherwise the friction factor correlations, determined for the rheological parameters, e.g.
viscosity and yield stress, together with other correlations, lead to a high uncertainty. In such
a case completely empirical models may give better results.
3.3.4 SEMI-EMPIRICAL CORRELATIONS FOR THE TURBULENT ICE-SLURRY FRICTION FACTOR
Turbulent friction factor correlations are based on the rheological properties of a particular
fluid and some experimental data as well. For ice slurry, Doetsch (2001) introduced a new
correlation given by Eq. (3.61). He took the Casson model to describe the rheological
behaviour of ice slurry. His correlation is based on an extended Blasius equation. A similar
equation was also proposed by Thomas and reported by Wasp (1977) for Non-Newtonian
fluids.
λ = 0.34179 ⋅ (Re C )−0.25793 ⋅ (Ca + 1)0.013532 (3.61)
Equation (3.61) is valid for Recritical < ReC < 40 000 (0 < Ca < 100 000) and applies to ice
slurries with any kind of additives. Rheological properties of ice slurry with different
additives are presented in Section 3.1.
Using the experimental data of Hansen (the same data used to develop the equation in Table
3.5), Kitanovski (2002) applied two rheological models, the Casson and Bingham models, in
order to determine which model best fits the experimental data. The yield stress parameter
and viscosity parameter were determined for both models by using the Mooney-Rabinowitch
equation. Jensen et al. (2000) used the same experimental data and Thomas (1965) equation
for the determination of the “Bingham viscosity parameter”, which differs from the one used
by Kitanovski, who obtained both parameters for both models from experimental data. The
procedure of Kitanovski (2002) was in accordance with that of Doetsch (2001) for a Casson
fluid. While Jensen’s data fitted within ±20% for 80% of all experimental data, Kitanovski
has obtained a fit of ±15% for all experimental data for a Bingham fluid, and ±30% for all
experimental data for a Casson fluid. We can conclude that it is very advantageous to
determine both parameters of the Casson or Bingham model from experimental data, not
using known equations for the suspension viscosity of the corresponding Newtonian fluids,
such as the Thomas (1965) equation.
74
Figure 3.16. Darcy´s friction factor dependence on the Reynolds and the Casson numbers
(Doetsch, 2001)
Since Doetsch (2001) obtained a deviation of less than 5% for Ca < 1000, less than 10% for
1000 < Ca < 10 000 and less than 20% for Ca > 10 000, we can conclude that his semi-
empirical equation gives very good results and may be applied to any kind of turbulent ice
slurry flow with any additive. However, it best applies when ice particles are rather small, i.e.,
dp = 0.1-0.5 mm, and of course under homogeneous flow conditions.
3.3.5 GENERAL EMPIRICAL DETERMINATION OF THE FRICTION FACTOR FOR ICE-SLURRIES
Correlations based on a dimensional analysis yield normally better fits to experimental data
than the semi-empirical approaches. However, extrapolation and interpolation using empirical
correlations can lead to large deviations. In this section, only correlations which satisfy both
laminar and turbulent flows are presented.
75
Table 3.5. Empirical correlations for the friction factor of ice slurry flows
Performed Specification of Equation Uncertainty Eq.
by experiment
(year)
Kitanovski ABS plastic pipe b ±15% for all
b4 ⎛⎜ v2 ⎞ 5
⎟
(Experi- inner diameters f is = f f + b1 ⋅ C r ⋅ f f ⋅C D ⋅
b2 b3
experimen- (3.62)
⎜ D ⋅ g ⋅ s −1 ⎟
mental data 12.8, 21, 27.7 mm ⎝ ⎠ tal data
0 . 0791
from Hansen v = 0.5-2 m/s ff =
–DTI ) Cr = 0-30% Re 0.25 (3.63)
(2002) initial ethanol
concentration: L
∆p = 2 ⋅ f ⋅ ρv 2
based on 5, 10, 20% D (3.64)
Turian´s dp ~ 300 µm
work gen.: scraper Coefficients presented in Table 3.6
Reghem Plastic pipe λ is − λf ⎛ v2 ⎞
d ±14% for all
c ⎜ ⎟
(2002) inner diameters ε = = a ⋅ C r ⋅ λf ⋅⋅
b
experimen- (3.65)
λf ⋅ C r ⎜ D ⋅ g ⋅ s −1 ⎟
21.5, 44.6 mm ⎝ ⎠ tal data
0.316
based on v = 0.1 to 3 m/s λf =
Turian´s Cr = 0-30% Ref 0.25 (3.66)
work initial ethanol
concentration: L ρv 2
10% ∆p = λis ⋅
D 2
dp = 100-300 µm Coefficients presented in Table 3.7 (3.67)
gen.: scraper
Reghem Plastic pipe 1 ±6% for all
Re1 = Re⋅
(2002) inner diameter K ⋅ Cr experimen- (3.68)
1+
21.5 mm v tal data in
based on v = 0.1 to 3 m/s 0.316 homoge-
Ogihara´s Cr = 0-30% λis = homogeneous flow for neous flow
Re1 0.25
work initial ethanol regime (3.69)
Re1 > 2100
concentration:
64
10% λis = heterogeneous flow for ±25% for all
Re1
dp = 100-300 µm experi- (3.70)
gen.: scraper Re1 < 2100 mental data
for hetero-
K = 9.75 geneous
L ρv 2 flow regime
∆p = λis ⋅
D 2 (3.71)
Snoek & Pipe diameters λis ±10% for all
Φ2 =
Gupta 38-100 mm λwater experimen- (3.72)
(1993) v = 0.96-3.53 m/s tal data
Φ 2 = 1 + a ⋅ C r ⋅ (Reis ) c + d ⋅ C r e ⋅ D f
b
Cr = 0-31% (3.73)
initial glycol
0.316
concentration λwater = (3.74)
9.6-12.9% Rewater 0.25
dp - not known L ρv 2
gen.: scraper ∆p = λis ⋅ (3.75)
D 2
Coefficients presented in Table 3.7
76
Table 3.6. Coefficients for empirical correlations for initial concentrations

of the additive ethanol (Kitanovski, 2002)
5 wt-% ethanol 10 wt-% ethanol 20 wt-% ethanol
b1 0.00000039332 650.1835 912649.1
b2 0.45034198847 1.147015 0.621567
b3 4.02136479789 3.807498 3.722244
b4 3.84746019112 1.089962 0.092936
b5 –0.67864294281 –0.649095 –0.403264
Table 3.7. Coefficients for empirical correlations for two different additives (from Table 3.5)
10 wt-% ethanol - water 9.6 - 12.9 wt-% glycol - water
(Reghem, 2002) (Snoek, 1993)
a 1640 0.1119
b 1.109 2.151
c 0.376 0.2422
d –0.642 0.02415
e / 0.3996
f / –0.2845
3.3.6 PRESSURE DROPS OF ICE-SLURRY FLOWS IN HEAT EXCHANGERS

Contrary to isothermal flows, the pressure drop of ice slurries in heat exchangers must be
determined considering heat transfer. The reason is that the ice slurry is melting in the heat
exchangers, and the physical properties are therefore changing. Based on this phenomenon, a
valuable theoretical study was performed by Egolf et al. (2001) who proposed an equation for
an average pressure drop. The model applies to the transport of ice slurries through cylindrical
heat exchangers, with small applied heat fluxes,
R in ⎛ R ⎞
R= ⋅ ⎜⎜1 − out ⎟⎟ (3.76)
R ⎝ R in ⎠
ln in
R out
where R denotes the specific pressure drop of the inlet (Rin) and of the outlet (Rout):
2 2
dp λin ρ in ⋅ v in dp λout ρ out ⋅ v out
R in = − = ⋅ and Rout =− = ⋅ (3.77a,b)
dx x =0 D 2 dx x =L D 2
This approach is based on an exponential decrease in the pressure drop due to the transport
and melting through the cylindrical heat exchanger.
x
R( x ) = α exp( − β ) (3.78)
L
77
If one applies the arithmetic mean value between the inlet and the outlet to determine the
average pressure drop, the error may be large. Note that in slow flow, stratification (or flow
with a moving bed) may occur, but this is not the subject of this section.
5
10
R (Pa/m) Example
4 R (Pa/m) M odel
10
Mean specific
R (Pa/m)
pressure drop
1000 1533 Pa/m
100
10
0 0.2 0.4 0.6 0.8 1
x/L (-)
Figure 3.17. The specific pressure drop R in a heat exchanger decreases approximately
exponentially. In this example R changes from the inlet to the outlet by more than a factor of
ten! “Example” denotes a step-by-step downstream calculation through the heat exchanger.
“Model” denotes results obtained by applying Eq. (3.76) (Egolf et al., 2001)
There have been several experiments performed on the pressure drop of ice slurry flows in
heat exchangers, most notably those reported by Bellas et al., 2002; Tassou et al., 2001;
Knodel et al., 2000; and Jensen et al., 2000.
3.4 Time dependent behaviour

by Didier Vuarnoz, Torben Hansen and Beat Frei
In 1999 at the University of Applied Sciences of Central Switzerland a curious observation

was made (Frei and Egolf, 2000). In a storage experiment at a constant ice slurry temperature
(and density), a trough formed on the surface of the ice slurry — shaped around the shaft of
the mixing element — and smoothly altered its form and depth with elapsed time. A first
interpretation was that the viscosity of the fluid may have changed as a function of time. In
further investigations of this phenomenon, the pressure drop in loops outside the storage
vessel was measured, and remarkable changes in specific pressure drop with time and
evaporator temperature were observed. After five cycles of partial melting and generation the
difference in pressure drop in the discharge loop (after storage and mixing) vanished
completely (see Figure 3.18) leading to the conclusion that storage and mixing have an
influence on the behaviour of ice slurry.
Knowledge of the geometrical characteristics of the ice particles in the ice slurry and time
dependent behaviour is now known to be an important key to understanding the deviation in
the results found in numerous apparently comparable experimental investigations of thermo-
physical properties, pressure drop, heat transfer coefficients, etc. The average size, the size
78
distribution and the geometrical characteristics of ice particles are known to affect the thermo-
physical properties of the ice slurry. The geometrical characteristics might depend on the
method of ice generation or the ice crystal history of partial melting and re-crystallization
(Hansen, 2002). Some mechanisms of alteration of the initial shapes of the ice particle are
reviewed in Pronk (2002) and summarized in Chapter 2.2.2 of this handbook.
Figure 3.18. The specific pressure drop in a charge - and discharge - loop alters as a function
of time and evaporator temperature at the same velocity and ice mass fraction
(Frei and Egolf, 2000)
1. Bagnold, R.A., Experiments on a gravity-free dispersion of large solid spheres in a

Newtonian fluid under shear. Proc. R. Soc., A225, pp. 49-63, 1954.
2. Barnes, H.A.,: An Introduction to Rheology, Elsevier, Chapter 7, 1989.
3. Bel, O.: Contribution à létude du comportement thermo-hydraulique dún mélange
diphasique dans une boucle frigorifique à stockage d´énergie. PhD Thesis, LÍnstitut
National des Sciences Appliquees de Lyon, 1996.
4. Bellas, J. et al.: Heat transfer and pressure drop of ice slurries in plate heat exchangers.
Applied Thermal Engineering, Volume 22, Issue 7, pp. 721-732, July 2002.
5. Chaer et al.: Flow Characteristics and Pressure Drop of Ice Slurries in Straight Tubes.
Clima 2000/Napoli 2001 World Congress, Napoli, September 2001.
6. Christensen, K.G.; Kauffeld, M.: Heat Transfer Measurements with Ice Slurry, IIR, Heat
Transfer Issues in Natural Refrigerants, November 1997.
7. Darby, R.: Hydrodynamics of slurries and suspensions. Encyclopedia of Fluid
Mechanics, Vol. 5, Slurry flow technology, Chapter 2, pp. 49-92, 1986.
8. Doetsch, C.: Experimentelle Untersuchung und Modellierung des rheologischen
Verhaltens von Ice Slurries. Ph.D. thesis, Fraunhofer Institut – UMSICHT, Fachbereich
Chemietechnik der Universtät Dortmund, 2001.
9. Dong Won Lee et al.: Experimental study on flow and pressure drop of ice slurry for
various pipes, Fifth IIR Workshop on Ice Slurries, Stockholm, May 2002.
10. Doron, P.; Barnea, D.: Flow pattern maps for solid-liquid flow in pipes, International
Journal of Multiphase Flow, Vol. 22, No. 2, p. 273-283, 1996.
79
11. Doron, P.; Granica, D.; Barnea, D.: Slurry flow in horizontal pipes – experimental and
modeling, International Journal of Multiphase Flow, Vol. 13, No. 4, pp. 535-547, 1987.
12. Egolf, P.W. et al.: Pressure drop and heat transfer in a cylindrical heat exchanger with ice
slurry flow, Third IIR Workshop on Ice Slurries, Lucerne, pp. 77-84, May 2001.
13. Frei, B., Egolf, P.W., Viscometry applied to the Bingham Substance Ice Slurry-
Proceedings of the Second IIR Workshop on Ice Slurries , pp. 48-59, Lucerne,
Switzerland, 2000.
14. Guilpart, J. et al.: Experimental study and calculation method of transport characteristics
of ice slurries, First IIR Workshop on Ice Slurries, Yverdon, pp. 74-82, 1999.
15. Hanks, R.W.: Principles of Slurry Pipeline Hydraulics, Encyclopedia of Fluid Dynamics,
Vol.5, Slurry flow Technology, Gulf Publishing Company, pp. 213-276, 1986.
16. Hansen, T.M.; Radosevic, M.; Kauffeld, M., Behavior of Ice Slurry in Thermal Storage
Systems, ASHRAE Research Project 1166, ASHRAE, 2002.
17. Hansen; T.M., Kauffeld, M.: Viscosity of ice slurry, Second IIR Workshop on Ice
Slurries, Paris, France, May 2000.
18. Jensen, E.N. et al.: Pressure drop and heat transfer with ice slurry, IIR Conference,
Purdue-USA, 2000.
19. Kawaji, M. et al., Ice slurry flow and heat transfer characteristics in vertical rectangular
channels and simulation of mixing in a storage tank, Fourth IIR Workshop on Ice
Slurries, Osaka, Japan, pp. 153-164, Nov. 2001.
20. Kitanovski, A. et al., Flow Patterns of ice slurry flows, Fifth IIR Workshop on Ice
Slurries, Stockholm, May 2002.
21. Kitanovski, A., PhD thesis, Faculty of Mechanical Engineering, University of Ljubljana,
2003.
22. Kitanovski, A., Poredoš A., Concentration distribution and viscosity of ice slurry in
heterogeneous flow, IIR Journal, Vol. 25, Issue 6, pp. 827-835, 2002.
23. Kitanovski, A.; Poredoš, A, Theory on concentration distribution of the ice slurry flow,
Fourth IIR Workshop on Ice Slurries, Osaka, pp. 15-24, 2001.
24. Knodel, B.D. et al.: Heat transfer and pressure drop in ice water slurries. Applied
Thermal Engineering, vol. 20, pp. 671-685, 2000.
25. Mooney, M.: Explicit Formulas for Slip and Fluidity, Journal of Rheology Bd. 2, pp. 210-
215, 1931.
26. Nasr-El-Din, H.; Shook, C.A.: The lateral variation of solids concentration in horizontal
slurry pipeline flow, International Journal of Multiphase Flow, Vol. 13, No. 5, pp. 661-
670, 1987.
27. Pronk, P., Infante Ferreira, C.A., Witkamp G.C., Effects of Long-Term Ice Slurry Storage
on Crystal Size Distribution. Proceedings of the Fifth IIR Workshop on Ice Slurries,
Stockholm, Sweden, May 2002.
28. Rabinowitsch, B.: Über die Elastizität von Solen, Zeitschrift für physikalische Chemie,
Vol. A 145, pp. 1-7, 1929.
29. Reghem, P., PhD thesis, University of Pau, 2002.
30. Richardson, J.F.; Zaki, W.N., Sedimentation and fliudisation: PART 1, Trans. I CHEM.
E., vol. 32, pp. 35-53, 1954.
31. Roco, M.C.; Shook, C.A, New approach to predict concentration distribution in fine
particle slurry flows, PCH, Vol. 8, No. 1, pp. 43-60, 1987.
32. Sari, O.; Meili, F.; Vuarnoz, D; Egolf, P.W., Thermodynamics of moving and melting ice
slurries. Second IIR Workshop on Ice Slurries, Paris, May, pp. 140-153, 2000.
33. Sari, O.; Vuarnoz, D., Meili, F.; Egolf, P.W., Visualization of ice slurries and ice slurry
flows, Second IIR Workshop on Ice Slurries Paris , May, pp. 68-80, 2000.
80
34. Sasaki, M.; Kawashima, T.; Takahashi, H.: Dynamics of snow-water flow in pipelines,
Slurry Handling and Pipeline Transport, Hydrotransport 12, pp. 533-613, 1993.
35. Sellgren, A.: Hydraulic Behaviour of Ice Particles in Water, Hydrotransport 10,
International Conference on Slurry Handling and Pipeline Transport, pp. 213-217, 1986.
36. Shook, C.A., Roco, M.C., Slurry Flow – Principles and Practice, Butterworth –
Heinemann, 1991.
37. Snoek, C.W.: The design and operation of ice slurry based district cooling systems, IEA-
Novem, 1993.
38. Stamatiou, E.; Kawaji, M.; Goldstein, V.: Ice Fraction Measurements in ice slurry flow
through a vertical rectangular channel heated from one side, Fifth IIR Workshop on Ice
Slurries, Stockholm May 2002.
39. Steffe, J.F.: Rheological methods in food process engineering. Second edition, Freeman
press, 1992.
40. Takahashi et al., Flow properties for slurries of particles with densities close to that of
water, ASME, Liquid-Solid flows, vol. 118, pp. 103-108, 1991.
41. Takahashi et al.: A laboratory study of pressure loss and pressure fluctuation for ice water
slurry flows in a horizontal pipe, Hydrotransport 12, pp. 497-511, 1993.
42. Tassou, S. et al.: Comparison Of The Performance Of Ice Slurry And Traditional Primary
And Secondary Refrigerants In Refrigerated Food Display Cabinet Cooling Coils, Fourth
IIR Workshop on Ice Slurries, Osaka, November 2001
43. Thomas, D.G.: Transport characteristics of suspensions. Journal of Colloid Science, no.
20, pp. 267-277, 1965
44. Turian, R.M.; Yuan, T.F.: Flow of slurries in pipelines, AIChE J. 23, p. 232-243, 1977
45. Vuarnoz, D.; Meili, F.; Moser, Ph.; Sari, O.; Egolf, P.W, Fluid and Flow Visualizations
of Ice Slurries, Eureka Project FIFE, research report No. 8, University of Applied
Sciences of Western Switzerland, 2000
46. Wasp, E.J.: Solid-Liquid Flow, Slurry Pipeline Transportation, Series on Bulk Material
Handlings, Vol. 1, No. 4, 1977
81
AXIMA REFRIGERATION WITH TOP COMPETENCE IN NATURAL REFRIGERANTS
Leading edge in refrigeration technology

Our experts cover a wide range of technical and engineering competences. We are acknowledged and welcomed
partner in international research projects and committees. Our expertise covers all aspects of cooling with natural
refrigerants.
Services & innovative products for trouble free operations

Axima Refrigeration focuses on the highest quality standards for both products and services. Whether it involves
industrial refrigeration plants with operating temperatures down to -85°C, ice slurry installations, refrigeration for
sport arenas and ice rinks, centrifugal, water or brine chillers or air conditioners most of our technology is based
on environmental friendly coolants.
All from one source

Our engineers are not only outstanding for their specialist knowledge on ice slurry, but also their wide-ranging
experience in plant construction and the choice of the best and most economic components – with the focus
always on customer benefits.
Our product and service program encompasses development, operation, and maintenance of refrigeration plants,
refrigerators, and cooling towers. Axima Refrigeration not only offers standard products, such as the innovative
chiller QUANTUM, custom-built special HVAC coolers and a wide range of durable components for heavy duty
industrial use, but also customized cooling or refrigeration solutions for your application in agriculture, energy
management, chemistry, oil or gas industry, as well as food processing, plastics, metal & steel work, and
recreation.
Ice slurry References worldwide

Successful applications with ice slurry have been realised in breweries, chemical industry and industrial kitchens
and where huge amounts of periodical cooling demands are required.
Our team of well-experienced engineers and consultants are prepared to provide you with more detailed
information. Contact the German sales headquarters:
www.de.axiref.com
info@de.axiref.com
CHAPTER 4. THERMODYNAMICS AND HEAT TRANSFER

by Peter W. Egolf, Andrej Kitanovski, Derrick Ata-Caesar, Osmann Sari, Alen Sarlah,
Evangelos Stamatiou, Masahiro Kawaji, Jean-Pierre Bedecarrats and Francoise Strub (see
specific list of symbols in Appendix 3)
4.1 Overview
The derivation of a macroscopic theory for ice slurry flows leads to very complex multi-
component and multi-phase mass, momentum and energy conservation equations. Therefore,
most of the existing theoretical work — which is related to heat transfer to and from ice
slurries — is based on rather simple models, which consider ice slurry to be a homogeneous
mixture of a carrier fluid and a large number of small ice particles. Experiments on heat
transfer with ice slurries have shown that simple models usually only roughly describe the
heat transfer phenomena, which are occurring during the melting of ice particles. As in other
cases of fluid dynamics, analytical physical laws are available only for laminar flows. Such
derivations were initiated by Graetz, who studied the thermodynamics of Bingham fluids in
tubes. For the description of transitional flow and turbulent flow, by mathematical
approximation techniques, numerous empirical equations have been derived, in order to
describe heat transfer in heat exchangers (with different geometries) from the hot walls to ice
slurries.
Simultaneous measurements of temperature distributions and ice concentration fields during

melting have shown that a thermal equilibrium between the carrier fluid and the ice particles
is not always obtained. There is considerable lack of agreement between the experimental
results reported by researchers from different laboratories. Qualitative and minor results were
mostly obtained in cases where the concentration of the ice was determined by applying the
equilibrium temperature/ice concentration functions and not by the indirect method of
measuring the density. The latter method shows a high accuracy, especially when the Coriolis
principle is applied. Furthermore, a heterogeneous distribution of ice particles — due to
buoyancy — has in some cases led to experimental results which are difficult to classify.
Despite such indeterminism, numerous comprehensive articles on heat transfer mechanisms
with phase change of various ice slurry flows exist. In this chapter, only the studies in which
the ice slurries were investigated and the related thermodynamic phenomena are clearly
described, are cited or presented.
4.2 Basic definitions
4.2.1 EFFECTIVE ENTHALPY AND EFFECTIVE SPECIFIC HEAT

In order to describe the enthalpy of phase change materials, Egolf and Manz (1994) developed
a macroscopic model to describe the behaviour of melting and freezing mixtures, the
Continuous-Properties Model (CPM), which basically describes a continuous phase transition.
Therefore, it is well suited to describing the overall thermodynamic behaviour of ice slurries.
If in the temperature/specific enthalpy diagram the region of the continuous phase transition is
chosen to be small, the model is also a good approximation for modelling the dynamics of
substances with discontinuous phase transitions. But it may also take into account the
occurrence of a mushy region near the eutectic point of a mixture. In reality, the width of this
region depends on the dynamics of the problem.
83
The enthalpy and the specific heat of ice slurries may be calculated by applying appropriate
equations for mixtures, which are presented in Chapter 2.
4.2 2 EFFECTIVE THERMAL CONDUCTIVITY

For the effective thermal conductivity several models may be found, which are mainly
derived from Jeffrey’s model (1973) (see Chapter 2). A selection of some of the principal
models is presented in Table 4.1.
Table 4.1. Some models for the effective thermal conductivity of suspensions
Performed by Equation Remarks

This is the equation
(
kis = kcf 1 + 3 ⋅ Cv ⋅ β + 3 ⋅ Cv2 ⋅ β 2 ⋅ γ ) which up to the
present has been
β 3β α + 2 applied in most
γ =1+ + ⋅ ,
Jeffrey (1973) 4 16 2α + 3 calculations of the
ki α −1 thermodynamic
α= , β= , 0 ≤ α ≤ ∞. behaviour of ice
kcf α +2
slurries.
3 ⋅ kcf This equation shows

kcf ⋅ (1 − Cv ) + ki ⋅ Cv ⋅ good agreement with
Brailsford and 2 ⋅ kcf + ki experimental results,
kis = .
Major (1964)  3 ⋅ kcf  obtained from
1 − Cv 1 − 
paraffin slurries (Shin
 2⋅k + k 
 cf i  et al., 2002).
ki + 2 ⋅ kcf − 2 ⋅ Cv ⋅ (kcf − ki )
Wasp (1977) kis = kcf ⋅ .
ki + 2 ⋅ kcf + Cv ⋅ (kcf − ki )
kis =
(2 ⋅ k + k ) − 2 ⋅ (k − k )⋅ R
cf i cf i i
,
(2 ⋅ k + k ) + (k − k )⋅ R
cf i cf i i
Maxwell and
Eucken (see 1
Eucken, 1940) Ri = .
 1  ρ
1 +  − 1 i
 Ci  ρ cf
A more comprehensive overview of formulae to calculate the effective thermal conductivities

of suspensions was given by Bel (1996).
All the macroscopic models, shown in Table 4.1, were derived for a homogeneous mixture of
a carrier fluid and particles which are immiscible in the fluid. The effective thermal
conductivity is the thermal conductivity of the mixture. This is a good approximation in the
limit where the particle diameter tends towards zero and the number of particles approaches
infinity, leaving the mass fraction of the particles unaltered. Please note that, depending on
84
the problem studied, a particle here may be a gas bubble, a liquid droplet or a solid particle. In
the case of ice slurries it obviously is an ice particle. Nanofluids yield the best practical
realization of this limit. In several models, as shown in Table 4.1, the effective thermal
conductivity is calculated in analogy with the resistance of electric circuits. This procedure
always leads to models combining the thermal conductivity of a carrier fluid and the thermal
conductivity of the particles. This is performed in such a manner, that in the limit of zero
particle fraction, the thermal conductivity of the carrier fluid results for the effective thermal
conductivity. On the other hand, in the limit of particle fraction equal to unity, the thermal
conductivity of the particles is obtained. In more sophisticated models the geometry of the
particles is also considered. In the case of ice slurries at the melting temperature the particles
do not only conduct heat, they also melt. If the phase transition is discontinuous the particles
are ideal heat absorbers, and in a description of the effective thermal conductivity, the thermal
conductivity of the particles cannot be important. Therefore, the models - developed for non-
melting suspensions - only apply approximately to ice slurries. At present at the University of
Applied Sciences of Western Switzerland further clarifying work on the question of which
situations and how the concept of an effective thermal conductivity applies to phase change
slurries is in preparation. The problem of different thermal relaxation times of the carrier fluid
and the small phase change material regions has to be addressed in future work.
4.2.3 DIMENSIONLESS NUMBERS
In Table 4.2 some definitions of dimensionless numbers are presented. Some authors base
them on the physical properties of the mixture and others on the properties of the carrier fluid.
Table 4.2. Dimensionless numbers relevant to ice slurry fluid dynamics and thermodynamics
All the physical

All the physical
Nusselt h ⋅D properties hcf ⋅ Dh
Nuis = is h correspond to the Nucf = properties
number kis k cf correspond to the
homogeneous ice
carrier fluid.
slurry mixture.
v is ⋅ Dh ⋅ ρ is
Reis = All the physical
ηis properties v cf ⋅ Dh ⋅ ρ cf
All the physical
Reynolds properties
The viscosity should correspond to the Re cf =
number be described by an homogenous ice ηcf correspond to the
carrier fluid.
appropriate rheological slurry mixture.
model.
c pis ⋅ηis
Pris = All the physical
All the physical
Prandtl
k is properties c pcf ⋅ η cf
The viscosity should correspond to the Prcf = properties
number k cf correspond to the
be described by an homogenous ice
carrier fluid.
appropriate rheological slurry mixture.
model.
All the physical
⋅ ⋅ All the physical
Graetz mis ⋅ c pis properties
mcf ⋅ c pcf properties
number Gzis = correspond to the Gz cf = correspond to the
k is ⋅ x homogenous ice k cf ⋅ x carrier fluid.
slurry mixture.
If the Bingham model is applied to describe the rheology of ice slurries, then the Hedstrom
number (containing the critical shear stress) is also important. It is defined in Chapter 3.
85
4.2.4 REFERENCE TEMPERATURES AND HEAT TRANSFER COEFFICIENTS

For the ice slurry flow in the heat exchanger illustrated in Figure 4.1, the local heat transfer
coefficient, his,L, can be determined from knowledge of the thermal conductivity of the fluid,
k, the temperature gradient (∂T ∂y ) y =0 at the wall, and mean temperature difference between
the wall and the fluid (Tw,m − Tis ,m ) x :
k  ∂T 
his , x =   . (4.1)
(Tw ,m − Tis ,m ) x  ∂y  x , y =0
g C
L
Adiabatic Wall
(∂T ∂y ) y =0
x
y
u(y)
Figure 4.1. Heat transfer from the wall to an ice-slurry in a vertical rectangular duct
with a second kind boundary at the wall
The peripheral mean wall temperature, Tw,m, and fluid bulk temperature, Tis,m, of an arbitrary
cross section can be defined as (Shah and London, 1978):
1
℘ ∫Γ
Tw,m = Tw ds . (4.2)
Tis ,m =
∫Ac
ρ ⋅ c p ⋅ u ⋅ T ⋅ dAc
. (4.3)
∫ Ac
ρ ⋅ u ⋅ c p ⋅ dAc
The fluid bulk temperature, Tis,m, also referred to as the “mixing cup” or “flow average
temperature” requires knowledge of the ice slurry density, mixture velocity distribution and
temperature distribution across the flow channel of the heat exchanger.
The average heat transfer coefficient is obtained by integrating the local heat transfer
coefficient over the heat transfer surface in contact with the ice slurry. For cylindrical pipes
and channels it follows that:
L
1
his , L = ∫ h sl , x ⋅ dx . (4.4)
L 0
86
The local Nusselt number is defined by:
his , x ⋅ D
Nu x = , (4.5)
kis , x
where the effective thermal conductivity of the ice slurry is denoted by kis,x. Such relations are
only valid when the ice slurry is considered as an ideal mixture of water with additives and
small ice particles. The average Nusselt number is calculated by substituting the average heat
transfer coefficient and effective thermal conductivity into the definition of the Nusselt
number:
his , L D
Nu L = , (4.6)
k is , L
where the mean effective thermal conductivity is similarly to Eq. (4.4) defined by:
L
1
χ = ∫ χ (x ) dx , χ ∈ {his , x , k is , x }. (4.7)
L 0
4.3 Measurement techniques
4.3.1 INTRODUCTORY REMARKS

Some experimental methods, which were mainly applied by the research group at the
University of Toronto (Kawaji et al., 2001; Stamatiou et al., 2001, 2002; Stamatiou and
Kawaji, 2004a, 2004b), are first described in the following sections and sub-chapters. Other
measurement techniques, which were used by other authors, are described in later sections or
given in related references.
4.3.2 LOCAL ICE FRACTION

The simplest measurement technique for determining the local ice fraction is by removing a
sample through an opening in the channel wall and analyzing the sample using a batch or on-
line calorimetry technique. Other solid fraction measurement techniques available are
reviewed elsewhere (Shook and Rocco, 1991).
In the direct sampling method used by Stamatiou et al. (2002) and Kawaji et al. (2001), the
ice fraction distribution was obtained by manually traversing a 6.35-mm O.D. stainless steel
(0.51 mm wall thickness), L-shaped sampling tube. The probe position was determined using
a calibrated slide-potentiometer. The L-shaped sampling probe was sharpened at its opening,
and was also slightly enlarged (7.6 mm O.D.) to allow for easy passage of ice crystals (see
Figure 4.2). In addition, the ice fraction value was shown to be unaffected by an ice slurry
sampling velocity deviating from its isokinetic sampling value due to the presence of very
fine ice slurry crystals that can follow the carrier fluid’s streamlines.
87
The ice slurry sample removed by the sampling probe was passed through a custom-made on-
line calorimeter unit that was instrumented and properly calibrated to determine the local ice
fraction (see Figure 4.3). The ice fraction distribution near the heated brass wall was obtained
by traversing the sampling probe at an axial elevation of L/Dh ≈ 20 located mid-width
(z/W=0.50) through the rectangular flow channel.
Back Brass Front Acrylic

Plate Wall
25.4 mm
Wall T/C T-type

T/C
Traversing ISP
5.33-mm I.D. S.S. 304 Tube Sample

Copper Coils to calorimeter
I.F. Profile Acrylic rod
Ball Valve
A A’
7.6 mm
I.D.
Unheated or
Heated
Wall
Ice Slurry
Figure 4.2. Local ice fraction distribution Figure 4.3. Calorimeter (680 W) for on-line ice fraction
measurement at L/Dh≈19. measurements
4.3.3 LOCAL ICE SLURRY TEMPERATURE

Local ice slurry temperature measurements were obtained by traversing a 1.56-mm O.D. T-
type thermocouple across the heated flow channel. Temperature recordings were made at six
selected transverse locations between the vertical heated brass plate and an acrylic wall over
the channel gap of 25.4 mm (see Figure 4.4). The thermocouple probe was placed mid-width
(z/W=0.50) through the rectangular channel at an axial elevation of L/Dh ≈ 20 as measured
from the inlet of the heated brass section.
4.3.4 LOCAL ICE SLURRY VELOCITY
Local ice slurry velocity profile measurements were conducted using an L-shaped hot film
probe (TSI, Model: 1231AR) placed at the mid-width (z/W = 0.50) of the rectangular channel
at the axial elevation of L/Dh ≈ 20. The hot film probe was connected to a Dantec CTA bridge
anemometer (Dantec, Model: 56C17) with a fixed bridge ratio of 4:8.
The hot film probe was traversed across the flow channel and local velocity measurements
were made at six different selected locations across the gap between the heated wall and an
opposite wall (see Figure 4.4). The output voltage signal from the hot film probe was
recorded at a sampling frequency of 5 Hz using a data acquisition system.
The hot film probe was initially calibrated at different ice fractions and velocities using a
rotating pool of ice slurry in a circular container placed on a turntable. The hot film
88
anemometer calibration data were used to obtain the velocity profile at specific ice fractions
during the melting experiments. More details of the hot film calibration method are given
elsewhere (Stamatiou, 2003).
Back wall Front Acrylic

brass plate Wall
0.0127 m
Wall T/C T-type
T/C
Traversing probe
Copper Coils Signal to

HWA anemometer
velocity probe
profile
C
L
Unheated or
Heated Wall
Ice Slurry
Figure 4.4. Local temperature and velocity measurements (side view)
4.4 Flow Phenomena
4.4.1 VELOCITY PROFILE, MEAN VELOCITY
4.4.1.1 Introduction
Local measurements of ice slurry velocity profiles were made under adiabatic and non
adiabatic flow conditions in a vertical rectangular channel using hot wire anemometry (HWA)
with an L-shaped hot film probe (Stamatiou and Kawaji, 2004a).
These measurements were carried out in a 0.61 m (L) x 0.31 m (W) x 0.025 m (gap, H)
rectangular channel at relatively low ice slurry velocities of uis ~ 0.10 to 0.15 m/s and volume
ice fractions of up to 10%.
4.4.1.2 Adiabatic ice slurry velocity measurements

Figures 4.5 and 4.6 show the experimental results obtained by traversing the hot film probe in
the vertical rectangular duct under adiabatic conditions at the axial location of L/Dh ~ 19.6. It
is evident that the ice crystals have an effect of producing a blunt velocity profile that deviates
considerably from the profile measured in a single-phase flow (volumetric ice fraction
Φv=0.0%). In these figures, the laminar velocity profile for a power law fluid between two
infinitely long parallel plates (α*=H/W=∞) given by Eq. (4.8) is also plotted (Hartnett and
Kostic, 1989; Skelland, 1967).
89
 n +1

u( y' ) 2n + 1   2 y'  n 
= 1−   (4.8)
V n +1   H  
 
In Eq. (4.8), n is the power law index and y’ is the distance measured from the mid-plane of
the flow channel. The power law index n was determined by best fitting eq. (4.8) to the
experimental data.
Most of the ice slurry HWA results shown were best predicted by n = 0.25, while the single-
phase CFD calculation (PHOENICS code) with Thomas’s viscosity correlation (1965) failed
to predict the present HWA measurements. The power law index of 0.25 suggests that the ice
slurry flow may be interpreted as a pseudoplastic or shear-thinning fluid, which is similar to
the behavior reported by Ben-Lakhdar et al. (1999) based on pressure drop measurements.
For dilute ice slurry mixtures with ice fractions lower than 2% by volume, the non-Newtonian
flow characteristics disappeared.
As Figures 4.5 and 4.6 illustrate, the power law model can predict the ice slurry velocity
profiles for volume ice fractions between 2% and 9%. Although other researchers have
interpreted their ice slurry flows as Bingham plastic or Casson fluids from pressure drop
measurements (Ayel et al., 2003), only the power law model was used to fit the ice slurry
velocity profile data as it is the most widely used and simplest model to apply and can
adequately describe the data. In the work of Stamatiou and Kawaji (2004a), the non-
Newtonian flow characteristics of ice slurries were detected at low ice fractions (Φv ≥ 2%)
since the hot-film anemometer is a much more sensitive instrument than a conventional
pressure-drop measurement equipment.
Although the above results indicate that the non-Newtonian flow characteristics depend on the
ice fraction, various other parameters such as the ice crystal size, shape, distribution, ice
slurry and carrier fluid densities, as well as the surface forces may also affect the deviation
from Newtonian flow (Darby, 1986).
1.50 1.40
1.40
1.30 1.20
1.20 Φv
1.10
1.00
1.00
0.90
0.80
0.80
u/V
u/V
-1 -1
0.70 Φv(%) Resl Recf <u>[m s ] V [m s ]
0.60
0.60 2.4% 1980 2130 0.080 0.110
4.1% 1890 2140 0.081 0.102
0.50
5.9% 1780 2130 0.081 0.095 Φv(%) Resl Recf <u>[m s-1] V[m s-1]
0.40 0.40
0.2% 2140 2160 0.081 0.105 6.0% 3400 4045 0.153 0.160
0.30 0.0% 2180 0.082 0.103 5.9% 1780 2130 0.081 0.095
Eq. (4.4.2-2); n =0.25 0.20 9.0% 2900 3900 0.148 0.147
0.20
PHOENICS k-ε; µsl =2.0 mPa.s (Φv=4.1%), V=0.098 m/s Eq. (4.4.2-2); n =0.25
0.10 PHOENICS k-ε; µsl=2.20 mPa.s, V=0.160 m/s
Single-phase empirical; Sun et al. (2002).
0.00 0.00
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
y/H y/H
Figure 4.5. Velocity profile data for ΦV=0.0-6.0%, Figure 4.6. Velocity profile data for Φv=6.0-9.0%,
<uis>=0.080m/s, Recf,L~2200 Recf,L~2100 & 4000
90
4.4.1.3 Non adiabatic ice slurry velocity measurements near a heated wall
Figure 4.7 compares the dimensionless ice slurry velocity distributions near heated and
unheated walls for runs conducted under similar ice fractions (Φv ~ 5-6%) and Reynolds
numbers (Recf ~ 4000) but different wall heat fluxes, q”w. This figure also compares the HWA
data with the adiabatic laminar velocity profile for a power law non-Newtonian fluid (Harnett
and Kostic, 1989), and the velocity profile predicted using a PHOENICS code with Thomas’s
viscosity correlation (1965) for a flow between two parallel plates.
The local Reynolds number, heat flux and local Nusselt number for the melting runs are also
given in this figure. The definition of the Reynolds number based on the carrier fluid (liquid)
properties, Recf, was adopted. For these data, the average slurry velocity, V, obtained by
integrating the area under the curve, matched the average ice slurry velocity, <uis>, measured
using an electromagnetic flow meter to within 5%.
These velocity profile data suggest that when a wall heat flux is imposed, the flow near the
heated wall decelerates in comparison to adiabatic ice slurry flow, while in the core region the
velocity profile remains similar to that displayed in adiabatic ice slurry flow. The velocity
reduction near the heated wall is due to the formation of an “ice-free” layer near the heated
wall that has a greater density than the bulk ice slurry. When a higher heat flux is further
imposed, the velocity near the heated wall is further decreased by about 10% as Figure 4.7
illustrates. Most of the velocity distribution data obtained displayed an approximately 10%
reduction near the heated wall in comparison to adiabatic ice slurry flows suggesting that the
flow decelerates near the heated wall in a vertical upward flow.
1.40
1.20
1.00
u ( y ') 2n + 1 ⎡
1 − ( 2 y '/ H ) ⎤
n +1/ n
=
0.80
V n ⎣ ⎦
u/V
Φv Resl Recf <usl> V q"w Nucf,L

0.60 [m/s] [m/s] [kW/m2]
6.0% 3400 4045 0.153 0.160 0.00 -
9.0% 2910 3900 0.148 0.147 25.5 234
0.40 7.6% 3090 3920 0.149 0.146 25.0 214
6.3% 3200 3860 0.146 0.157 25.3 206
3.1% 3700 4050 0.153 0.138 24.2 178
0.20
Ref. [10]; n =0.25
PHOENICS, k-ε; µsl=2.2cP, V=0.16 m/s
0.00
0.00 0.20 0.40 0.60 0.80 1.00
y/H
Heated Brass Front Acrylic
Wall Wall
Figure 4.7. Typical ice slurry velocity profiles near a heated wall
91
4.4.2 TEMPERATURE PROFILE, MEAN TEMPERATURE IN VERTICAL CHANNELS

Figure 4.8 shows typical dimensionless ice slurry temperature distributions across a heated
channel (α∗ = 1:12) for Recf ~ 3800 and volume ice fractions up to 26%. In Figure 4.9, the
dimensionless ice slurry temperature distributions obtained in a narrow heated channel (α* =
1:8) are given at turbulent flow velocities (usl = 0.50 m/s), Recf ~ 6600 and volume ice
fractions up to 23%.
These temperature distribution measurements were made at the axial location of x/D h= 7 by
traversing a single T-type thermocouple probe across the flow channel between the vertically
heated brass plate (y/H = 0) and an adiabatic acrylic (y/H = 1.0) wall. The results are
expressed in terms of a dimensionless fluid temperature, θ = T − Tsl Tsl ,in − Tw , as adopted by
Hartnett and Kostic (1989). The temperature data shown in these figures represent the time-
averaged temperatures of 20 data points obtained over a period of approximately 10 sec for a
specified ice fraction. Also, the temperature rise near the acrylic wall (y/H ≥ 0.7) is considered
to be anomalous due to the heat conduction along the thermocouple sheath. Nonetheless, the
temperature data for y/H<0.6 can be considered to be unaffected by axial conduction effects
and satisfactorily represent the ice slurry temperature increase near the heated wall.
The set of experimental results obtained in either wide or narrow channels reveal that the
temperature profiles of melting ice slurries are fairly uniform across the flow channel
regardless of the ice fraction (Xs ~ 2% to 24%). In view of the blunt velocity profiles obtained
and discussed earlier, the bulk ice slurry temperature, Tis,m, can be assumed to be given by the
temperature in the middle of the flow channel, which eliminates the need to integrate the
expression given by Eq. 4.3. Furthermore, these temperature distributions suggest that the
thickness of the thermal boundary layer or “ice-free” zone remains thin even at low ice
fractions.
1.00 1.00
0.95
θ=(T-Tw)/(Tsl,in-Tw)
θ=(T-Tw)/(Tsl,in-Tw)
0.90
0.90 Xs Φv <usl> Resl Recf Thw,m q"w Nucf,L

[%] [%] [m/s]
o
[ C] [kW/m ]
2 Xs Φv <usl> Resl Recf Thw,m q"w Nucf,L
[%] [%] [m/s] [oC] [kW/m2]
2.2% 2.5% 0.147 3620 3900 39.3 23.6 123
4.5% 5.1% 0.147 3330 3880 40.1 24.3 127 0.80 5.9% 6.6% 0.519 6170 6650 41.6 39.7 109
9.9% 11.2% 0.147 2630 3800 40.4 25.3 128 9.3% 10.5% 0.530 4800 6700 40.1 38.8 114
0.85 14.0% 15.7% 0.141 2170 3700 40.4 25.7 137 12.7% 14.2% 0.520 4070 6600 39.9 38.9 123
18.4% 20.4% 0.142 1800 3720 40.2 25.8 130 14.8% 16.5% 0.504 3600 6350 41.0 41.7 138
20.1% 22.2% 0.144 1700 3770 40.1 26.6 153 20.5% 22.7% 0.517 2850 6500 39.8 45.5 171
24.0% 26.4% 0.143 1400 3700 38.7 26.3 144
0.80 0.70
0.00 0.20 0.40 0.60 0.80 1.00 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
y/H
y/H
Heated Front
Brass Wall Acrylic Wall Heated Front
Brass Wall Acrylic Wall
Figure 4.8. Ice slurry temperature distributions for Figure 4.9. Ice slurry temperature distributions
Recf=3800 and α*=H/W=1:12 (H=25.4 mm) for Recf=6600 and α*=H/W=1:8 (H=12.7 mm)
92
4.4.3 LOCAL ICE FRACTION PROFILES
4.4.3.1 Adiabatic ice fraction distributions

Figure 4.10 shows typical ice fraction profiles (Φ/Φm values) measured by Stamatiou et al.
(2002) using an on-line ice fraction sampling probe as described in section 4.3, for mean
volume ice fractions ranging from Φv ~ 6.0 to 11%. These ice fraction measurements were
conducted at similar ice slurry sampling velocities, Uo. The transverse location, y/H,
represents the distance measured from the brass wall to the centre of the sampling probe.
Here, Φm represents the average ice fraction of the local measurements. In Figure 4.10, a
small variation in the ice fraction distribution is noted across the flow channel for dilute
slurries (Φv<8%), with peak values appearing near both adiabatic walls. Similar peaking of
the volume fraction profiles of the disperse phase has been obtained in vertical bubbly flows
(Class et al., 1991) as well as in the higher density sand-water flows (Nasr-El-Din et al.,
1984). In ice slurries, the ice crystals are less dense than the carrier fluid so that the ice
fraction peaking phenomena near the solid boundaries are less pronounced compared to other
systems with greater density differences.
Figure 4.11 shows that at high volume ice fractions, Φv > 10%, the ice fraction distribution
becomes flatter than that displayed for the lower ice fraction range, 3% <Φ v< 8%. This
phenomenon may be attributed to several factors such as the increase in the drag force exerted
by the fluid on the ice crystals and greater interactions between adjacent ice crystals at higher
ice fractions (Nasr-El-Din et al., 1984).
1.10 1.10
1.00 1.00
Φ/Φm
Φ/Φm
0.90 0.90
ΦV[%]
o -1
Tsl,L [ C] usl [m s ] Uo/<usl> Recf,L
10.8 -4.27 0.146 1.92 3840
ΦV[%] Tsl,L [oC] usl [m s-1] Uo/<usl> Recf,L
9.1 -4.40 0.147 1.80 3850
0.80 0.80 16.1 -4.07 0.146 1.35 3806
7.0 -4.01 0.149 1.85 3930
6.3 -4.01 0.149 1.75 3910 14.5 -4.40 0.146 1.32 3820
5.9 -4.05 0.148 1.75 3890 13.0 -4.43 0.146 1.23 3825
0.70 0.70
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
y/H y/H
Figure 4.10. Effect of Φv (6-11%) on the ice Figure 4.11. Effect of Φv on the ice fraction
fraction distribution at Uo/<uis > ~ 1.8 distribution at Uo/<uis > ~ 1.3
4.4.3.2 Ice fraction distributions near a heated wall

It is of further interest to examine the ice fraction distributions near a heated wall since it has
been previously observed that the local heat transfer characteristics could be markedly
affected by local ice fraction variations (Horibe et al., 2001; Kawanami et al., 1998).
Figure 4.12 shows the effect of average ice fraction on the normalized local ice fraction
distribution (Φ/Φm) near a heated wall at Recf ~ 3800 and q”w ~ 24 kW/m2 (Stamatiou et al.,
2002; Stamatiou and Kawaji, 2004b). It can be observed that the ice fraction in the vicinity of
the heated wall is approximately 5% to 20% lower than in the bulk. Furthermore, the wall
93
peaking phenomenon previously shown to occur near the adiabatic walls in dilute ice slurries
(Φv < 8%), disappears near the heated boundary (y/H = 0) but remains near the other adiabatic
wall (y/H ~ 0.75). At the nearly adiabatic wall, the ice fraction profile is similar to that
obtained under adiabatic conditions, i.e., wall peaking appears in runs with small average ice
fractions (3% < Φv < 8%). At higher mean ice fractions (i.e., Φv > 10%), however, the peaking
phenomenon disappears.
As seen in Figure 4.12, the local ice fraction near the heated wall does not show any
systematic variation with increasing ice fraction in the free-stream. The latter result may
suggest that even at low ice fractions and mean velocities, the “ice-free” zone remains thin as
the ice crystals in vertical flows disperse uniformly across the heated channel. In comparison
to adiabatic ice fraction profiles, however, the ice fraction reduction near the heated wall in
dilute slurries (Φv < 8%) is approximately 30% to 40% greater than in the more concentrated
ice slurries (Φv > 10%). In the more concentrated ice slurries (Φv > 10%), the ice crystals are
more closely packed so that the ice fraction reduction near the heated wall is only 10 to 20%
in comparison to the adiabatic ice fraction profiles. This, in turn, would impact the local heat
transfer coefficient.
1.4
1.2
1.0
0.8
Φv
Φ/Φm
Recf <usl > Uo/<usl> q"w Nucf,L

[ m/s ] [ kW/m2 ]
0.6 3.3% 3900 0.149 2.3 22.5 139

5.1% 3800 0.143 1.8 23.7 153
7.6% 3830 0.145 1.2 24.4 157
0.4 8.6% 3820 0.145 2.0 24.8 158
8.8% 3840 0.146 1.4 24.6 153
10.8% 3810 0.145 1.7 25.4 167
0.2 11.7% 3840 0.145 1.5 24.0 171
15.9% 3710 0.141 1.5 25.5 174
18.1% 3760 0.144 1.3 25.1 161
0.0
0.00 0.20 0.40 0.60 0.80 1.00
y/H
Heated Front
Brass Wall Acrylic Wall
Figure 4.12. Effect of Φv on ice fraction distribution at Recf ~ 3800 near a heated wall
4.5 Heat transfer phenomena

4.5.1 A BRIEF INTRODUCTION
Numerous experiments have been performed to determine the local or averaged heat transfer
coefficient and the corresponding Nusselt numbers for ice slurry flows through heat
exchangers of different types. The following types of heat exchangers are discussed in further
sub chapters:
94
• Pipes
• Channels
• Channels with a bend
• Heat exchangers with stagnant flow
• Industrial heat exchangers.
4.5.2 PIPES
Christensen and Kauffeld (1997) experimentally determined the average Nusselt number in
flows through a cylindrical pipe (see Figure 4.13). A constant heat flux was provided to the
inner pipe by condensing refrigerant R134a between two concentric tubes. The heat flux was
varied between 6 and 14 kW/m2. The ice slurry was produced with an aqueous ethanol
solution of initially 10 mass percent. The test section, constructed from a stainless steel tube,
had an outer diameter of 26.9 mm, an inner diameter of 21.6 mm and a length of 1 m. The
heat transfer coefficient, his,x, was determined by considering the temperature difference
between the refrigerant R134a and the ice slurry. The Nusselt number for the ice slurry was
determined as follows:
A) For an ice concentration, Cv, above 5%:
−0.192⋅
(30−Cv )
Nuis 0.339⋅Reis ⋅10− 4
= 1 + 0.103 ⋅ Cv − 2.003 ⋅ Reis 30
⋅ Cv . (4.9)
Nucf
B) For an ice concentration, Cv, below 5%:
Nuis
=1. (4.10)
Nucf
250
200
150
Nuis
100
v=1,0 m/s; D=0,018 m

50 v=2,3 m/s; D=0,018 m
v=1,0 m/s; D=0,023 m
v=2,3 m/s; D=0,023 m
0
0 5 10 15 20 25 30
Cv (%)
[%]
Figure 4.13. Average Nusselt number for ice slurry flows through a pipe
95
The equations (4.9) and (4.10) are valid for the following parameter ranges:
• 0,7 < uis < 2.5 m/s
• 0 < Cv < 30%
• 18 < Dh < 25 mm
• 6 < q& < 14 kW/m2
• Initial ethanol concentration ξe = 10%.
The authors found that the average Nusselt number increases with increasing ice mass
fraction and Reynolds number. The increasing heat transfer coefficient with increasing ice
mass fraction is explained by the melting of ice particles near the wall, and therefore, a
decrease in the temperature gradient within the boundary layer. The dynamic viscosity of the
ice slurry was determined by applying the Thomas model (see Chapter 3 or Thomas [1965]),
and the ice slurry was considered as a Bingham fluid. The yield stress was determined from
the data obtained in pressure drop experiments. The measured heat transfer coefficients
increased from 3000 W/m2K for an ice mass fraction of 5% to 7000 W/m2K for a mass
fraction of 25%.
Jensen et al. (2000) performed a very similar experiment to that of Christensen and Kauffeld
using condensing NH3 instead of the refrigerant R134a in the outer pipe. In Table 4.3 the
parameters of their experiments are presented.
Table 4.3. Test parameters (after Jensen et al., 2000)

Parameter Pressure loss Heat transfer
Concentration of ethanol 5, 10, 20 mass-% 10 mass-%
Ice concentration 5, 10, 15, 20, 25, 30 mass-% 5, 10, 20, 30 mass-%
Pipe diameters (inner diam.) 12.8, 21, 27.7 mm 12, 16, 20 mm
Velocities 0.5, 1, 1.5, 2 m/s 0.5, 1, 1.5 m/s
Heating power - 500, 1000, 1500 W
The average heat transfer coefficient, his, was determined using the temperature differences
between the refrigerant NH3 and the ice slurry. The average Nusselt number was determined
as follows:
0.14
Nu is( A) ⎛ D 2 ⋅ ρ cf ⋅ τ 0 ⎞
For X > 0.01: = 10.42 ⋅ Re −0.5
⋅⎜ ⎟ , (4.11)
Nucf
cf ⎜ η ⎟
⎝ cf ⎠
where
τ0
X= . (4.12)
v ⋅ ρ cf
2
For X < 0.0015 it was assumed that the ice slurry behaves as a Newtonian fluid. Thus, it
follows that:
Nu is( B )
= 1. (4.13)
Nucf
96
For 0.0015 < X < 0.01 the following interpolation formula must be applied:
Nuis = Nu is( A) ⋅ A + (1 − A) ⋅ Nu is( B ) , (4.14)
where:
X − 0.0015
A= and Nucf = 0.0237 ⋅ Re cf ⋅ Prcf .
0.81 0.39
(4.15)
0.0085
The uncertainty in the experimentally determined Nusselt number is ±10%. These empirical
formulae are valid for:
• 0.5 < u is < 1.5m/s
• 5 < Cv< 30%
• 12 < Dh < 20mm
• 1.6 < q& < 13kW/m2
• Initial ethanol concentration ξe = 10%.
Knodel et al. (2000) measured the heat transfer rate in an ice slurry flow with a constant heat
flux imposed by electrical heaters on a stainless steel tube of 24 mm inner diameter and
4596 mm length. The velocity of the ice slurry varied between 2.8 and 5 m/s. The
experimental device was set up with two heaters, each having a 30 kW heating power. The ice
slurry was produced with pure water. During each operation of the experimental loop, at
different downstream positions, the local heat transfer coefficients were determined by
evaluating the temperature difference between the pipe wall and the bulk temperature of the
ice slurry which was kept constant at Tb = 0°C. For each cross section, the inner tube wall
temperature was calculated using the energy conservation equation and the measured outer
wall temperature. Then the heat transfer coefficient data were obtained as shown in Figure
4.14 and compared with Petukhov’s (1970) relation for the average Nusselt number:
⎛f⎞
⎜ ⎟ Re cf Prcf
Nucf = ⎝8⎠ , K1 = 1 + 3.4 ⋅ f cf , K 2 = 11.7 + 1.8 ⋅ Prcf
−
1
3 . (4.16)
1
⎛f⎞ 2
(
K1 + K 2 ⎜ ⎟ Prcf − 1
2/3
)
⎝8⎠
In Equation (4.16) all the quantities must be determined using the physical properties of the
carrier fluid. The coefficient f stands for the friction factor and is calculated using:
f = (1.82 log10 Re cf − 1.64) .

−2
(4.17)
For pipes of small diameters, the authors determined the following relation between the
experimentally determined Nusselt number for the ice slurry flow and the one calculated
using Eq.(4.15) as :
Nuis
= 0.885 . (4.18)
Nucf
The empirical equation above is valid for the following parameter ranges:
• 4 < Cv < 11%
• 3.8·104 < Re < 7.4·104
• Dh = 24 mm.
97
These equations are only valid for “pure water” as the carrier fluid. When the Nusselt number
is evaluated using the properties of a 10 mass-% ethanol aqueous solution, it decreases with
increasing ice mass fraction.
Figure 4.14. Nusselt number data for an ice slurry flow heated with a heat flux
of 15 kW/m2 and 40 kW/m2 (Knodel et al., 2000)
Snoek and Bellamy (1997) performed heat transfer experiments with a 2 m long pipe of 15.74
mm internal diameter under the conditions summarized in Table 4.4.
Table 4.4. Parameters of the experiments performed by Snoek and Bellamy (1997)
Variable Range Units

Heat flux 28-114 kW/m2
Velocity 5-12 m/s
Glycol concentration 8-10 %
Ice concentration 4-33 %
The pipe was heated by electrical heaters to provide a constant heat flux. To describe the heat
transfer rates Snoek and Bellamy applied the Dittus-Boelter (1930) equation for turbulent
flow in a smooth pipe, where the ice slurry properties are used instead of the carrier fluid:
kis
hD− B ,is = 0.023
0 ,8
Reis Pris0, 4 . (4.19)
D
After comparing the experimental results with the heat transfer coefficients calculated using
equation (4.19), Snoek and Bellamy (1997) proposed the following correlation:
98
Cv 0.562
⎛C ⎞ −5
Θ = 0.924 + 0.076 ⋅ e − 6.43 ⋅ 10 ⎜ v ⎟
100 0.827
Reis , (4.20a)
⎝ 100 ⎠
his
Θ= . (4.20b)
hD − B ,is
The predicted heat transfer coefficients are shown against the ice mass fraction in Figure 4.15.
The accuracy of the proposed empirical model is ±10% when compared with their
experimental results. They found that the heat transfer coefficient increases with increasing
ice mass fraction (increase of 19% for Cv = 20%).
16000
12000
K]
(W/m K)
22
αhis [W/m
8000
v=5 m/s
4000 v=8 m/s
0
0 10 20 30
Cv [%]
(%)
Figure 4.15. Calculated heat transfer coefficients obtained by using Equations (4.19) and (4.20)
Guilpart et al. (1999) worked with an aqueous ethanol solution (with an initial ethanol
concentration of 11%) to produce ice slurry in their experimental device. The heat transfer
section consisted of a copper tube of 30 mm inner diameter and 0.5 m length. The applied
constant heat flux was varied between 0-12 kW/m2. The wall temperature profile in the
downstream direction was measured with nine thermocouples, positioned at a distance of 2
mm from the inner wall. The heat transfer coefficients were determined by evaluating the
difference between the wall temperature and the bulk fluid temperature measured at different
equidistant downstream positions. From their experimental results, an equation describing the
heat transfer for laminar ice slurry flow was proposed. It is of a Graetz-Nusselt type equation
with an accuracy of approximately ±13%:
0.52
⎛C ⎞
Nu(z ) = 38.3 ⋅ Gz 0.15
⋅⎜ v ⎟ . (4.21)
⎝ 100 ⎠
The Graetz number is calculated from:

⋅
mis ⋅ c pis π D
Gz = = ⋅ ⋅ Re is ⋅ Pris . (4.22)
k is ⋅ z 4 z
99
The model equations are valid over the following parameter ranges:
• 0 < Cv < 35%

• 3 < Re < 2000.
The authors also introduced a dimensionless Cameron number, X+. Its value is related to the
downstream position z by the following formula:
z
2
X+ = D = π . (4.23)
Re is ⋅ Pris 2 ⋅ Gzis
The predictions of Equations (4.21) – (4.23) are shown in Figure 4.16.
3000
Cv=5%
2500
[W/mK)K]
Cv=15%
2 2
2000 Cv=25%
islocal(W/m
1500
αhis,local
1000
500
0
0 0.0001 0.0002 0.0003 0.0004 0.0005
+
X [-]
Figure 4.16. The local heat transfer coefficient calculated by applying Eq. (4.21) - (4.23)
Sari et al. (2000) performed experiments to determine the heat transfer rates in ice slurry
flowing through an aluminium tube in a heat exchanger as shown in Figure 4.17. The
cylindrical heat exchanger was equipped with four electrical heating elements (with a total
power of 1.02 kW), which yielded a constant heat flux boundary condition. Between the
heating elements the probes could be continuously driven into the fluid to measure
temperature profiles in cross sections perpendicular to the tube axis. Temperature profiles
were measured at the inlet, the outlet and at three axial locations in the intermediate region of
the heat exchanger. In the cross sections at the inlet and outlet (sections 1 and 5, respectively),
more measurements could be performed as shown in two cross sections each. In Figure 4.17,
the possible positions of the temperature probes and the actual positions are clearly
distinguished.
The ice slurry was produced using an aqueous Talin solution with an initial Talin
concentration of 10 mass-percent. In these experiments a constant heat flux of 14.1 kW/m2
was applied to the outer wall of the pipe. The heated section had a length of 1-meter and a
100
diameter of 23 mm. Experimental data on wall surface temperatures, fluid temperatures, mass
flows, densities and inlet and outlet pressures were recorded.
Rheological data were taken from experimental work, which had been published by Frei and
Egolf (2000), whereas the ice concentration, the density and the overall specific heat were
determined in accordance with the measurements and model calculations of Egolf and Frei
(1999). The effective thermal conductivity was determined by applying Jeffrey’s model (see
Table 4.1). In the paper of Sari et al. (2000), a perturbation analysis was performed to
numerically determine the heat transfer rate in an ice slurry flow. The applicability of this
model is limited to small heat fluxes. Even though the heat flux in the experiment was chosen
to be low, the basic requirements for the validity of the perturbation model could be only
approximately fulfilled. The reason is that a high temperature rise in a thermal boundary layer
occurs, due to the low effective thermal conductivity of the ice slurry.
Aluminium tube: L= 1m, φ = 23mm, φ = 60mm

i e
Electrical heaters: q&
1 2 3 4 5
m& in ρin pin m& out ρout pout
Section 1 Section 2 Section 3 Section 4 Section 5
Surface temperature measurements
Possibility of fluid temperature measurements
Performed fluid temperature measurements
Figure 4.17. Test section schematic and temperature measurement locations (Sari et al., 2000)
The numerical model was developed for laminar and turbulent flows by applying a simple
turbulent Prandtl number concept for turbulent flow (Cebeci and Bradshaw, 1984). The
numerical results for laminar flow showed that the heat transfer rate is rather small, because
the thermal penetration depth is low. The reason is a small diffusivity perpendicular to the
pipe wall compared to the high convection velocity in the flow direction in the heat
exchanger. From these results, it was concluded that tube-like heat exchangers with laminar
flow will not work very efficiently. The numerical programme to calculate turbulent ice slurry
flows was based on an average turbulent eddy viscosity of εh = 6.4 mm2/s in order to get the
best agreement between the calculated and measured temperature profiles. By multiplying εh
by the specific heat and density, a “turbulent thermal conductivity” of 250 W/(m K) is
obtained, which is 360 times higher than in the case of laminar flow (e.g. 0.7 W/(m K)). It is
101
not surprising that the “turbulent” convection heat transfer in a turbulent ice slurry flow is
much more efficient than the molecular heat diffusion in a laminar flow.
The mean temperature in the radial direction at a certain downstream location was determined
by:
R
2π ∫ T (r )rdr
1
< T >r = Ac = πR 2 . (4.24)
Ac 0
The local heat transfer coefficient was then determined using the following formula:
q& ( x )
h( x ) = . (4.25)
Tw ( x )− < T > r ( x )
Numerically calculated temperature profiles (solid curves) are compared to measured profiles
(black dots) in Figure 4.18. The cross sections 1 (x = 0 mm) to 5 (x = 1000 mm) correspond
to the five sections shown in Figure 4.17. The flow direction is from the left to the right. At
the inlet of the heat exchanger a constant temperature boundary condition was obtained. The
surface temperatures of sections 2 to 4 are found to be higher and are not shown in these
figures.
x = 0 mm x = 250 mm x = 500 mm x = 750 mm x = 1000 mm
10
Radial distance from center
-5
-10
-5.1 -4.9 -4.7 -4.5 -5.1 -4.9 -4.7 -4.5 -5.1 -4.9 -4.7 -4.5 -5.1 -4.9 -4.7 -4.5 -5.1 -4.9 -4.7 -4.5
Figure 4.18. Numerically calculated temperature profiles compared to measured profiles
The Nusselt number as a function of the Hedström number, characterizing the Bingham effect,
is shown in Figure 4.19 for laminar flow. On the left, presented by a rectangular symbol, the
Nusselt number for a water/ethanol system without any ice fraction, is shown.
The Nusselt number as a function of the Hedström number is shown in Figure 4.20 for
turbulent flow. The heat transfer coefficient is higher in turbulent flow than in laminar flow.
102
This can be seen by comparison of the results of Figures 4.19 and 4.20. This corresponds to
the physical fact that turbulent fluctuations lead to an additional heat transport mechanism.
250
200
Nusselt number
150
100
Laminar
Re=780-1620
50
0
0 5000 1000 1500 2000 2500
Hedström number
Figure 4.19. The Nusselt number as a function of the Hedström number for laminar flow
350
300
250
Nusselt number
200
Turbulent
150
Re =3890-4840
100
50
0
0 5000 1000 1500 2000
Hedström number
Figure 4.20. The Nusselt number as a function of the Hedström number for turbulent flow
In Equation (4.25), Tw(x) represents the wall temperature at position x. Figure 4.18 shows a
comparison between the experimental and numerical results for turbulent flow. In Figures
4.19 and 4.20 the Nusselt numbers are presented as a function of the Hedstrom and Reynolds
numbers. The circular symbols show the results obtained at different sections of the heat
103
exchanger. They are not specifically distinguished in this presentation. The black symbols
show the mean values of each experimental series for a particular Reynolds number. The
range of Reynolds numbers investigated is given in the figure. On the left, characterized by a
large rectangular symbol, the Nusselt number for water/ethanol (without ice) is shown. This
value is Nu = 4.3. For higher Hedström numbers, the Nusselt number increases, as expected
by a physical explanation of an increasing melting process.
The authors emphasized the importance of the effective thermal conductivity, the only
relevant physical property that has not been measured yet. The laser flash method or the hot
wire method with a Dirac heating pulse are being developed for such applications.
In 2002, Bedecarrats studied the behaviour of polypropylene particles of 3 mm diameter and

ice slurry with an aqueous ethanol solution of initially10 mass-% ethanol. The outer side of
the heat exchanger pipe was heated with hot water. The temperature of the hot water was
controlled at the location of the pipe entrance. The analysis of the experimental data was
based on a modified Wilson method (Shah, 1990; Wilson, 1915), which was used to
determine the values of heat transfer coefficients from the determination of the overall
thermal resistance of the heat exchanger:
1 1
R= = = ∑ R j = Ri + Rw + Rd + Ro (4.26)
U o Ao U i Ai
and:
⎛r ⎞
ln⎜⎜ 2 ⎟⎟
R=
1
+ ⎝ r1 ⎠ + ⎛⎜ 1 + 1 ⎞⎟ + 1 , (4.27)
hi Ai 2 ⋅ π ⋅ L ⋅ k ⎜⎝ hdi Ai hdo Ao ⎟⎠ ho Ao
where Ri denotes the inside film resistance, Rw the wall resistance, Rd the combined inside and
outside fouling resistance, and R0 the outside film resistance. Wilson (1915) proposed his
method, which was also described by Dittus and Boelter in 1930 and by Sieder and Tate in
1936. From these equations it may be shown that Ri is proportional to v-0.8. The effect of the
velocity, v, on R0 occurs only indirectly (by an alteration of the average outside temperature)
and is generally negligible. The effect on Rw and Rd is also negligible. Therefore, it follows
that:
≈ C1 ⋅ (v ) + C 2 ,
1 −0,8
R = (4.28)
U o ⋅ Ao
and:
C2 = R0 + Rd + Rw . (4.29)
The linear equation (4.29) has to be adapted to the outside film resistance and velocity by
adjusting the two “constants”, C1 and C2. When the velocity approaches infinity, C1 represents
the slope of the function and C2 is the intercept where v -0.8 = 0. Assuming a clean tube (Rd =
0) and taking the known value of Rw into consideration, ho can be defined by determining C2
from the graph. Once C2 has been determined, for a given velocity Ri is then determined by:
Ri = R − C2 . (4.30)
104
Measuring the heat flow and the logarithmic temperature difference between the fluid at the
inlet and outlet, the overall resistance is determined by:
•
Q
U 0 ⋅ A0 = . (4.31)
∆Tln
The authors have emphasized that an increase in ice mass fraction, while keeping the mass
flow rate unaltered, increases the heat transfer coefficient, and an increase in the mass flow
rate (at an unaltered ice concentration) also leads to an increase.
4.5.3 HEAT TRANSFER IN RECTANGULAR CHANNELS
The thermal-hydraulic behavior of ice slurries in laminar and turbulent flows through non-
circular channels is of special interest due to the wide application of such geometries in
compact heat exchangers. Nonetheless, very few experimental and numerical investigations of
ice slurries have been carried out in such geometries since the associated hydrodynamic and
heat transfer phenomena are generally more complex than in circular pipes, and require two-
dimensional analysis. The following sections aim at presenting some fundamental aspects of
the fluid mechanics and heat transfer properties of ice slurries in vertical rectangular channels.
In 1998, Kawanami et al. performed an experiment with a horizontal rectangular duct of 1 m

length and 60 mm width. It was designed with variable wall heights of 20, 40 and 60 mm. The
top and bottom walls were electrically heated to guarantee a constant heat flux. An initial 20
mass-percent ethylene-glycol aqueous solution and a 20 mass-% ice concentration were
chosen for these experiments. Four different heat fluxes of 2, 4, 6 and 8 kW/m2 were used and
kept constant over the duration of each experiment. The average heat transfer coefficient was
determined using the following equation:
q&
hx = , (4.32)
Tw, x − Tis ,i
where Tw,x represents the wall temperature at the location x and Tis,i is the initial ice slurry
temperature at the entrance to the heat exchanger, which is considered to be constant.
Based on the results shown in Figure 4.21, the authors drew the following conclusions:
• At the lower wall the variation of the heat flux had less effect on the heat transfer
coefficient than on the top wall. This was thought to be due to the stratification in the flow
(heterogeneous flow) and a resulting smaller latent heat potential at the bottom of the
channel. The applied heat served to increase the carrier fluid temperature in regions with
mainly sensible heat absorption.
• At the top wall the variation of the heat flux resulted in an increased heat transfer
coefficient. This was due to a higher concentration of ice particles close to the top, and
therefore, the temperature difference — as defined in Equation (4.32) — was small.
According to Eq. (4.33) this results in an increase in the heat transfer coefficient.
105
Figure 4.21. Local heat transfer coefficients measured in a horizontal rectangular channel
of height 60 mm. Other parameters are given in the figures (from Kawanami et al., 1998)
At the top wall, an increase in the ice slurry flow rate, while maintaining a certain heat flux,
had less effect on the heat transfer coefficient than at the lower wall. The authors explained
this by referring to less stratification in the flow at a higher ice slurry velocity. More ice
particles would flow closer to the lower wall and so the heat transfer coefficient increases at
higher velocities.
Figure 4.22. Flow patterns according to Kawanami et al. (1998) and some predictions of the IPF
(ice packing factor) for different channel heights: a: H = 60 mm, b: H = 40 mm, c: H = 20 mm
Kawanami et al. (1998) also studied the influence of the channel height on the heat transfer
coefficient, while keeping the heat flux and the ice slurry flow rate constant. On both the
bottom and top walls of the channel, the heat transfer coefficient decreases with the channel
106
height. For the same ice slurry flow rate, a higher channel wall leads to greater stratification
of the ice slurry (see Figure 4.22). The authors explained this by assuming that a decreasing
channel height increases the local velocity near the upper wall of the channel, inducing
stronger mixing, and therefore, leading to a better heat transfer rate in this region. At the
bottom of the channel, an increasing number of ice particles with the decreasing channel
height is observed and this increases the heat transfer coefficient, because of the latent heat of
the ice particles.
Kawanami et al. (1998) determined the average heat transfer coefficient by applying Eq. (4.32).
For different heat fluxes, they observed a significant difference in the heat transfer
coefficients at the top and bottom walls of the channel. The higher curve, representing the
greater heat transfer coefficient at the top wall, was related to the close contact of the ice
particles and melting at the top wall in this region. For different velocities at a given heat flux
and channel height, larger heat transfer coefficients were determined at the top of the channel.
However, the slope of the curve representing the heat transfer coefficient versus velocity was
higher for the lower wall of the channel. Increasing the velocity reduced the difference
between the mean heat transfer coefficients at the top and the bottom walls. The variation in
the channel height showed that the mean heat transfer coefficient was higher at the top
channel wall, but both heat transfer coefficients decreased with the increasing channel height.
By considering the difference in the heat transfer coefficients between the top and bottom
channel walls, the authors proposed the following formula, which contains a dimensionless
height and a dimensionless heat flux:
0.36
⎛H⎞
Num = 23.7 ⋅ Reis ⋅⎜ ⎟ ⋅ q *0.05 ,
0.15
(4.33)
⎝W ⎠
where H is the height of the channel, W is its width and q* is the dimensionless heat flux
defined using the latent heat of the ice, hm, in J/kg as follows:
⋅
q
q* = , (4.34)
hm ⋅ ρ is ⋅ vis
The predicted mean heat transfer coefficients are shown in Figure 4.23.
650
600
550
2 K]
K)
500
2
[W/m
his (W/m
450
v=0,2 m/s; q*=3 kW/m2
αis
400 v=0,7 m/s; q*=3 kW/m2

v=0,2 m/s; q*=7 kW/m2
350 v=0,7 m/s; q*=7 kW/m2
300
0.02 0.03 0.04 0.05 0.06
H [m]
H (m)
Figure 4.23. Mean heat transfer coefficients calculated using Eqs. (4.33) and (4.34)
107
The accuracy of this empirical equation is ±20%, and it is valid for the following parameter
values and ranges:
• Cv = 20%
• 0.1 < vis < 0.8 m/s
• H = 20/40/60 mm
• 2 < q& < 8 kW/m2.
At the University of Applied Sciences of Western Switzerland, Kitanovski et al. (2003)

studied heat transfer to an ice slurry of a 10% propylene-glycol solution. This ice slurry
flowed through a rectangular, horizontal aluminium heat exchanger of 1-meter length. The
hydraulic diameter was 23 mm, which is identical to the inner diameter of the tube heat
exchanger used in the previous work of this research team. Two mass flow meters were
installed, one before and the other after the heat exchanger. Temperatures at the inlet and
outlet were measured with Pt-1000 sensors, while the vertical temperature profiles at discrete
cross sections of the heat exchanger were measured with thermocouples. Low (1800 W/m2)
and high (7200 W/m2) heat fluxes were used.
For the low heat flux case, temperature distributions in the channel cross section were
measured in the heat exchanger at five equidistant downstream locations (the first cross
section at the inlet, and subsequent downstream locations in steps of 0.25 m). For the higher
heat flux of 7200 W/m2, the temperature distribution was measured only at three axial
locations (the first cross section at the inlet and in steps of 0.5 m). Simultaneously with the
temperature measurements, the pressure drop was determined at four equidistant axial
locations. The experimental data (temperature profiles) revealed a turbulent flow through the
test section. Therefore, a thermal eddy diffusivity model was introduced in the numerical
simulations. It is known that such local turbulence modelling cannot give an accurate
description (see e.g. Egolf and Weiss, 2000). Nevertheless, in such a case, it yields a simple
and valuable first approach. Figure 4.24 shows the temperature distributions in the
downstream direction of this rectangular heat exchanger for the low heat flux case. The
authors assumed that a perturbation analysis only approximately applies to such small heat
fluxes, but they have not derived a criterion for this. Each temperature profile measurement
was performed in a separate experiment and since the initial temperature conditions at the
inlet were not always identical, they are specified in Figure 4.24 on the right-hand side of the
curves.
Figure 4.24. Experimentally and numerically determined temperature profiles for a heat flux of 1800
W/m2 in a 1-meter long rectangular heat exchanger at a mass flow rate of 1.2 kg/s. The measurement
locations were: x = 0.25 m, 0.5 m, 0.75 m and 1 m. Trapezoidal symbols denote experimental data,
while small triangles and the fitted solid curves represent numerically calculated profiles
108
In Figure 4.25 the temperature profiles for the higher heat flux case are presented. A
significant deviation between numerical and experimental temperature profiles is observed.
The authors have explained that this is due to a failure of the perturbation analysis model at
very high heat flux densities. In all cases the temperature distributions were not exactly
axisymmetric. The reason is a stratification of the ice particle field. Ice particles have a lower
density than the carrier fluid. This leads to higher ice concentrations at the top of the channel.
Therefore, lower temperatures were observed in the higher layers. In all cases the flow was
heterogeneous with an average velocity of approximately 0.75 m/s. As seen in Figures 4.24
and 4.25, there are low gradients of the temperature profiles T(y,x=const) towards the wall.
The temperature measurements close to the wall are very difficult to perform. Therefore, these
measurements have to be interpreted very carefully; as the authors have stated, they could
even be erroneous.
Figure 4.25. Temperature profiles for a heat flux of 7200 W/m2 in a 1-meter long rectangular heat
exchanger. The axial locations of the measurements are: x = 0.5 m and 1 m. Trapezoidal symbols
denote experimental data and the solid curves represent numerically calculated profiles
Figure 4.26 shows the ratio of the thermal eddy diffusivity to thermal diffusivity related to
molecular diffusion. The turbulent diffusivities were determined by fitting the numerical
calculation results to the measured temperature profiles. Surprisingly these ratios decrease in
downstream direction. The results for the low and high heat flux cases show qualitatively the
same behaviour.
Figure 4.26. Ratios between the thermal eddy diffusivity and the thermal diffusivity at different axial
locations in a rectangular heat exchanger for different heat fluxes. The data correspond to the results
shown in Figures 4.24 and 4.25. The dotted line was drawn in analogy to the solid line. It must be
confirmed by more measurements
109
The ratio of the thermal eddy diffusivity to the molecular diffusivity is large near the inlet of
the heat exchanger and decreases continuously toward the outlet. A diffuser was mounted at
1-meter upstream of the heat exchanger inlet, with an inclination angle of 20°. It was assumed
that in this diffuser strong turbulent fluctuations (eddies) were created, which then decayed
when transported downstream. As a result even in isothermal flows the turbulent
characteristics were not constant but varied as a function of the downstream location x. This
explains the decreasing ratio of diffusivities. This observation was made by numerical
investigations by altering the eddy diffusivity and applying curve fitting to the measured and
calculated temperature profiles.
A second important observation was a deviation between the measured temperatures at the
outlet and the equilibrium ice fractions. The density measurements, which are a reliable
measure of the ice concentration, showed higher ice fractions at the outlet than expected by
temperature measurements. Temperatures above 0oC were observed, while the corresponding
density measurements clearly showed a non-negligible ice fraction in the fluid. Therefore,
overheating of the carrier fluid appears to have occurred, which may have influenced the
measurements of the temperature sensors. This phenomenon has also been observed in other
laboratories. It has also occurred in plate heat exchangers (e.g. see Frei and Boyman, 2003).
Stamatiou and Kawaji (2004b) also studied heat transfer in rectangular channels with a
constant heat flux. For the cross section of a rectangular duct shown in Figure 4.27, the
dimensions H and W define the height and width, the ratio of which is the channel aspect ratio,
α* = H/W. If α*→ ∞, it corresponds to a channel between two parallel plates, in which case
the side walls will have an insignificant effect on the flow field.
For flows in non-circular channels, the hydraulic diameter Dh of the duct is commonly used as
the characteristic length (Cho et al., 1998). The hydraulic diameter for the rectangular duct of
Figure 4.27 is defined as follows:
2( H ⋅ W )
Dh = 4 Ac P = (4.35)
(H + W )
where Ac is the flow cross sectional area and P is the wetted perimeter of the channel.
H α*=H/W
x
W
Figure 4.27. Geometrical dimensions of a rectangular duct
Heat Transfer Parameters

The circumferentially averaged but axially local ice slurry heat transfer coefficient hsl,L for a
constant wall heat flux, q w" , is defined by:
110
constant wall heat flux, q w" , is defined by:
qw'' = his ,L (Tw,m − Tis ,m ) , (4.36)
where Tw,m is the mean wall temperature, and Tis,m is the bulk ice slurry temperature. The ice
slurry heat transfer results are presented in terms of a dimensionless heat transfer coefficient
known as the Nusselt number, Nu.
The local ice slurry Nusselt number, Nucf,L, was evaluated based on the local heat transfer
coefficient (his,L), the thermal conductivity of the carrier fluid (kcf), and the hydraulic diameter,
Dh:
his , L ⋅ Dh
Nu cf , L = . (4.37)
k cf
The average ice slurry Nusselt number, Nucf,m, was defined based on the carrier fluid
properties, and was computed using the arithmetic mean of the local ice slurry Nusselt
numbers (Stamatiou and Kawaji, 2004b).
Thermal Boundary Condition

The thermal boundary condition H was investigated in which a uniform wall heat flux is
applied axially at one heated wall, while the other walls are maintained at nearly adiabatic
condition.
Comparison with Single-Phase Heat Transfer Data

The Nusselt numbers can be compared with well known single-phase heat transfer
correlations to examine the effect of ice crystals on the average heat transfer coefficient.
These single-phase heat transfer correlations are summarized below.
Turbulent Single-Phase Convection

The average ice slurry Nusselt numbers can be compared with well known single-phase
convective heat transfer correlations for fully developed turbulent tube flow in the thermal
entrance region given by Eq. (4.38) (Holman, 1998; Knudsen, 1958),
0.055
⎛D ⎞ L
Nu cf = 0.036 Recf0.8 ⋅ Prcf1 / 3 ⎜ h ⎟ 10< <400. (4.38)
⎝ L ⎠ Dh
Transition Single-Phase Convection

The single-phase heat transfer coefficients in the transition region, Recf =2300 ~ 104, can be
predicted by the correlation given by Gnielinski (1976, 1983):
( f / 2) ⋅ ( Recf − 1000) ⋅ Prcf
Nu cf = , (4.39)
1 + 12.7 ⋅ ( f / 2) ⋅ ( Prcf2/3 − 1)
where f is the Fanning friction factor, and Prcf is the Prandtl number.
111
Laminar Single-Phase Convection

The laminar heat transfer results are compared with the single-phase correlation proposed by
Sieder and Tate (1936) for the fully developed laminar convective heat transfer in tubes:
⎛D ⎞ ⎛µ ⎞
1/ 3 0.14
Nucf = 1.86(Recf ⋅ Prcf ) ⎜ h ⎟ ⎜ cf ⎟ 1/ 3

, (4.40)
⎝ L ⎠ ⎝ µw ⎠
where all the fluid properties are evaluated at the mean bulk temperature of the carrier fluid
except for µw, which is evaluated at the wall temperature.
Average Ice Slurry Heat Transfer Coefficient Data

Figure 4.28 shows the effects of average ice fraction and Reynolds number (Recf,m) on the
mean Nusselt number, Nucf,m, for runs conducted at nearly the same wall heat flux of ~ 40
kW/m2 and heating water temperature of about 41.0oC. The average single-phase Nusselt
numbers measured at Tcf = –2.8oC (Tcf>Tfp) are also shown in this figure. A fair agreement
was found between the measured single-phase average Nusselt numbers and those predicted
by Eqs. (4.39) and (4.40), respectively.
Figure 4.28 suggests that the average ice slurry heat transfer coefficients are greater than
those achieved in single-phase turbulent flows at similar mean velocities (or Reynolds
numbers, Recf). Furthermore, the average Nusselt number increases as the Reynolds number
and ice fraction increase. This is also seen in Figure 4.28 for the heat transfer data obtained at
a lower heat flux, q”w ~ 15 kW/m2 (at the average heated wall temperature Thw,m ~ 14.8oC).
The average ice slurry heat transfer coefficient data shown in these figures obey the turbulent
convection relationship (i.e., Nu ~ Re0.8) suggesting the turbulent character of ice slurries at
these Reynolds numbers and ice fractions. Thus, these ice slurry heat transfer results can be
correlated using conventional turbulent convection correlations; any enhancement is due to
the ice fraction and/or heat flux.
300
Xs,m Thw,m q"w
Thw,m=41.0oC; α*=1:8
[%] [oC] [kW/m2]
0.0% 41.5oC 35.0 (22.0 ~ 45.7)
250
1.9% 41.5oC 40.3
5.6% 41.5oC 40.0 Xs,m=19.8%
Nucf,m=hsl,mDh/kcf [ - ]
10.1% 41.0oC 40.6

200
13.2% 40.6oC 40.9
15.8% 41.0oC 43.5
19.8% 40.4oC 44.3
150
100
Gnielinski (1976)
50
NuDh=0.036Re0.80Pr1/3(Dh/L)0.055
0
0 2000 4000 6000 8000 10000 12000
Recf,m [ - ]
Figure 4.28. Nucf,m as a function of Recf,m and Xs,m at Thw,m ~ 41.0oC and α*=1:8 (Stamatiou, 2003)
112
300
Xs,m Thw,m q"w
Thw,m=14.8oC; α*=1:8
[%] [oC] [kW/m2]
250 0.0% 14.7oC 12.2 (8.2 ~ 15.9)
5.7% 14.9oC 14.2
11.6% 14.8oC 15.4
Nucf,m =hsl,mDh/kcf [ - ]
200 14.7% 14.8oC 16.3

21.6% 14.9oC 17.8
150
Xs,m=21.6%
100
Gnielinski (1976)
50
NuDh=0.036Re0.80Pr1/3(Dh/L)0.055
0
0 2000 4000 6000 8000 10000 12000
Recf,m [ - ]
Figure 4.29. Nucf,m as a function of Recf,m and Xs,m at Thw,m ~ 14.8oC and α*=1:8 (Stamatiou, 2003)
The present heat transfer results given in Figures 4.28 and 4.29 also suggest that there is a
greater enhancement of the average Nusselt number at low Reynolds numbers with increasing
ice fraction in comparison to single-phase fluid flow. In particular, for the laminar and slightly
turbulent velocity ranges investigated (Recf,m < 4000), there is approximately a two fold
increase in the average heat transfer coefficient data compared to single-phase convection,
whereas in more turbulent flows this increase is only about 40% to 60%. This improvement in
the ice slurry Nusselt number at low Reynolds numbers may be attributed to the mixed-
convection effects and non-Newtonian flow characteristics of ice slurries as well as the
thermally developing flow conditions. The free-convection and thermal entry effects are
inherent in short compact heat exchangers, so that improved heat transfer coefficients would
be obtained in practice.
4.5.4 HEAT TRANSFER IN RECTANGULAR CHANNELS WITH A BEND

Kawanami et al. (2001) studied the heat transfer rates in a rectangular curved channel
(vertically positioned return bend) of 80 mm width and 190 mm mid-channel bend radius (see
Figure 4.30). The ice slurry was produced with an ethylene-glycol solution (20% initial
concentration) and the ice slurry at the entrance to the curved heat exchanger was step-wise
altered up to 20 mass-percent. A heat flux of 8 kW/m2 was applied with electrical heaters
mounted on the wall. For the experimental data, the local heat transfer coefficient was
determined according to the angle of the azimuthal position and was therefore defined as:
q&
hφ = , (4.41)
Tφ − Tin
113
where the index of Tφ denotes the angle at which the wall temperature was measured, and Tin
denotes the bulk temperature of the ice slurry at the entrance (in=inlet) to the curved heat
exchanger (at φ = 0). The average heat transfer coefficient was determined to be:
π
1
hm =
π ∫ hφ ⋅ dφ .
0
(4.42)
The experiments were performed with different flow velocities (e.g. 0.1, 0.4 and 0.8m/s) and
the results are shown in Figure 4.31, where the angle φ is denoted by θ.
Figure 4.30. Schematic of the flow model (Kawanami et al., 2001)
Figure 4.31. Local (above) and mean (below) heat transfer coefficients for the concave and convex
curved walls (Kawanami et al., 2001)
114
4.6 Heat exchangers with stagnant flow
Various experiments have been performed to study the melting of ice slurries. One can also
find numerous papers applying melting models to ice slurries. In most cases they were also
developed by applying certain numerical methods and then some numerical simulations led to
the final results. The inverse process of freezing usually does not occur. As an example, here
the work of Kawanami et al. (1999) is presented since a thermal camera and a shadowgraph
method were used in their experiments. The visualisation results are helpful in understanding
some characteristics of the ice slurry melting phenomenon. The authors performed the
experiments with ice slurry in a transparent (Lucite) tube of 100 mm inner diameter and 69
mm length. The details of the experimental loop are explained in the above cited reference.
Three different heat fluxes, 800, 1600 and 2400 W/m2, generated by electrical heaters, were
applied. The ice slurry was produced with an ethylene-glycol solution and the average
diameter of the ice particles obtained was rather large, about 0.5 mm.
Because of their lower average density, ice particles rise to the top of the cylindrical tube. Due
to the buoyancy force on the particles, a good thermal contact with the heated wall results,
leading to an intense melting process. The upward motion of the ice particles causes the
melted substance - with a lower viscosity - to be squeezed out of the close contact region and
to flow into the outer region, containing an ice slurry, which is situated below. Figure 4.32
reveals that stratified layers form between this domain and the wall. This is a result of the
coupled heat and mass transfer, which is discussed in greater detail later. A more detailed
observation of Figure 4.32 also shows how the ice slurry content of the channel separates, due
to the plume containing liquid, flowing away from the diffuse mushy melting region (see the
lower section of the melting region in Figures 4.32c and 4.34).
Figure 4.32. Melting behaviour of an ice slurry with an ice fraction of Ci=20 mass-%
and heat flux of q& =2.4 kW/m2 after: a) 5 min, b) 15 min, c) 25 min
In Figure 4.33a it is shown that the ice slurry, by buoyancy, is pressed against the top of the
heated channel wall. However, a direct contact between the upper region of the ice slurry,
containing ice particles, and the wall does not occur, due to the continuous production of pure
fluid by the melting process. Therefore, a thin liquid gap between the ice particle containing
region and the upper wall is observed. The energy, necessary for the melting of the ice
particles, is transported by thermal conduction across the fluid gap. It should also be noted
that a fraction of the melted ice is squeezed back into the ice slurry region. This consequently
115
results in a jet-like flow out of the liquid in the bottom region containing melting ice slurry
(see Figure 4.34). As indicated in Figure 4.33b,
Figure 4.33. Melting mechanisms as described and explained by Kawanami et al. (1999)
Figure 4.34. Melting behaviour and temperature distribution for Ci=20 mass%, a) q·t = 2.88x106 J/m2;
d) q = 800 W/m2; b and e) 1600 W/m2, c and f) 2400 W/m2 (Kawanami et al., 1999)
in the initial stage of the melting process, the heat transport is dominated by thermal
conduction. After a mixing process between the water — produced by the melting of some ice
crystals — and the ethylene glycol aqueous solution, a stable ice particle distribution is
formed in the melting region. As illustrated in Figure 4.33c, free convection occurs as time
116
elapses. Then a number of stratified layers, which are based on diffusion, are formed in the
melting region between the ice particle containing region and the heated wall (see Figures
4.32 and 4.34).
It is observed that the number of stratified layers increases with an increasing initial
concentration of the aqueous binary solution as well as with a decreasing heat flux at the wall.
As it was already described above, liquid is pressed out of the lower layer of the domain with
a higher ice fraction (see also Figures 4.32b and 4.32c). Hence the temperature at the exit of
the “chimney” decreases below the phase change temperature of the suspension close to the
bottom of the mushy melting interface. This causes the formation of a dendritic frozen layer
(see Figure 4.33d). Figure 4.34 shows the effect of the heat flux on the melting behaviour and
the temperature distribution in the vessel. The pictures were taken with a shadowgraph
method and a thermo-camera. Here one can clearly see that the number of stratified layers
decreases with an increase in heat flux. This results in a variety of shapes of the melting
interface. This is because the thermal buoyancy force increases with increasing heat flux, due
to an increase in the temperature of the liquid layer in the vicinity of the heated wall. It also
appears as if the flow rate of the fluid flowing out of the bottom of the ice particle rich
domain and the rate of dendritic ice formation increases with an increasing heat flux. It is seen
from the temperature distribution (see Figures 4.34d, e, and f), that the temperature in the
liquid region changes significantly with an increasing heat flux. This may be a result of the
increasing heat flux, which causes a decrease in the number of stratified layers, an increase in
the wall temperature, and also an increase in the intensity of the free convection. Contrary to
this the temperature in the ice slurry decreases with an increasing heat flux. This may be
explained by considering that increasing the heat flux results in enhanced melting and
decreased particle diffusion in the ice slurry. In the ice slurry this diffusion is reduced with
increasing melting velocity due to an increasing heat flux. The authors used their results to
determine local and mean heat transfer coefficients.
4.7 Industrial heat exchangers
4.7.1 BRIEF INTRODUCTION
A comprehensive study was performed by Nørgaard et al. (2001). This work is presented in
detail in Chapter 8. Reference must also be made to the experimental results obtained at
Brunel University (UK) on pressure drop and heat transfer with plate heat exchangers (Bellas
et al., 2002).
4.7.2 HEAT TRANSFER CORRELATIONS IN COMPACT HEAT EXCHANGERS

Turbulent heat transfer correlations
A dimensional analysis is employed to derive a new heat transfer correlation for turbulent ice
slurry flows in compact heat exchangers. In this procedure, the following dimensionless
variables are used to account for the effect of ice-crystals on convective heat transfer: the
average Nusselt number (Nucf,m), Reynolds number (Recf), Prandtl number (Prcf), average ice
fraction (Xs) and the ratio of viscosities evaluated at the bulk carrier fluid and the wall
temperatures, µcf/µw. A multiple, linear regression analysis was performed by Stamatiou and
Kawaji (2004b) on their average heat transfer coefficient data as shown in Figure 4.35. In
117
their correlation, only heat transfer data from the narrow channel were used as they were free
of natural convection effects. The coefficients of the proposed regression model were
determined using the method of least squares, which yielded the following correlation giving
the best fit to the present experimental data for 3300 < Recf < 11,000 and 0.0 < Xs < 0.25:
Nu sl ,m =
hsl ,m ⋅ Dh
k cf
[
= Nu gn 1 + 1.85 × 10 5 ⋅ X s0.72 ⋅ Recf-1.30 (µ cf / µ w )
2.47
] (4.43)
In Eq. (4.43), Nugn is the single-phase heat transfer coefficient given by Eq. 4.39 and Xs is the
weight ice fraction. Figure 4.35 compares the experimental heat transfer results (symbols ○
and ●) with those predicted by Eq. (4.43), and also shows the predictions of the heat transfer
correlations proposed by Jensen et al. (2000) and Horibe et al. (2001), which covered similar
ice fraction and Reynolds number ranges. The discrepancy of the previously proposed heat
transfer correlations with Stamatiou and Kawaji’s heat transfer results was attributed to the
shorter thermal entry lengths involved in the experiments. The shorter thermal entry lengths
common in compact heat exchangers should yield higher heat transfer coefficients. In
addition, the previously proposed heat transfer correlations were based on a log-mean-
temperature difference, so one needs to properly determine the resistances on the warm fluid-
side and the heat transfer surface area to extract the ice slurry-side heat transfer coefficient.
220
Horibe et al. (2001)

150
Nusl,predicted
10 2
Eq. (5.9)
80 Eq. (4.43)
+15%
-15%
Jensen et al. (2000)
80 150 220
10 2
Nu sl,m =h sl,m D h/kcf [ - ]
o
T hw,m=39.6 C
T hw,m=14.8oC
} α*=1:8, Experimental data
T hw,m=39.6oC; Jensen et al. (2000)
T hw,m=14.8oC; Jensen et al. (2000)
T hw,m=14.8oC; Horibe (2001)
Figure 4.35. Comparison of experimental results and proposed empirical correlations for convective
heat transfer to ice slurries (Stamatiou and Kawaji, 2004b)
An alternative heat transfer correlation for turbulent convection to ice slurry flow is shown in
Figure 4.36. This figure shows that the relationship between Nusl ,m X s−0.216 ( µ cf / µ w )
−0.723
and
Recf for turbulent ice slurry flow is proportional on a logarithmic graph. By evaluating the
exponent of Recf and the coefficient from the experimental data, the following empirical
correlation is proposed for the average ice slurry-side heat transfer coefficient (Stamatiou and
Kawaji, 2004b):
118
Nusl ,m X s−0.216 ( µ cf / µ w )
−0.723
= 2.10 Re
Recf0.511 (4.44)
which is valid for the range 3300<Recf<11,000 and 0.01<Xs<0.25. Unlike the correlation
proposed by Horibe et al. (2001) and the numerical predictions of Roy and Avanic (2001), the
ice slurry heat transfer coefficient data were only weakly dependent on the Stefan number, so
the Stefan number was eliminated from the correlation.
400
o
Thw,m=39.6 C
o
Thw,m=14.8 C } α*=1:8
[-]
. -0.216. µ µ -0.723
200
( cf / w)
+15%
Nusl ,m X s−0.216 ( µ cf / µ w )
−0.723
102
Nusl,m Xs
= 2.10 Re0.511
cf
70 -15%
103 2x103 5x103 8x103 104 2x104

Recf [ - ]
Figure 4.36. An alternative heat transfer correlation for turbulent ice slurry flow
(Stamatiou and Kawaji, 2004b)
Laminar heat transfer correlation

Figure 4.37 shows a local heat transfer correlation developed for laminar ice slurry flow in a
wide channel, α*=1:12 by Stamatiou and Kawaji (2004b). For the low velocity ranges
involved (usl<0.15 m/s), the mixed-convection effect and non-Newtonian flow characteristics
would become more prevalent as previously discussed. For the laminar convection
correlation, the dimensionless variables used, including the effect of ice-crystals on heat
transfer are: the local Nusselt number (Nucf,L), Graetz number (Gz = RecfPrcfDh/x), the average
ice fraction (Xs) and the viscosity ratio between the bulk fluid and the wall, µcf /µw. The
coefficients of the proposed regression model were determined using the method of least
squares and 180 independent experimental data points, which yielded the following
correlation giving the best fit to the experimental data for 2100 < Recf < 4,000, 0.01 < Xs <
0.25 and GrPr(Dh/L) > 106:
Nucf , L = 4.0Gz 0.486 X s0.30 ( µ cf / µ w )

0.24
(4.45)
Figure 4.37 compares the heat transfer results (symbols ● and ○) with those predicted by Eq.
(4.45). All the data can be predicted by Eq. (4.45) to within ±15%. A similar heat transfer
correlation was first formulated by Ben-Lakhdar et al. (1999) for ice slurry flows in a short
119
tube without the viscosity ratio, µcf /µw. The results predicted by Ben-Lakhdar et al. (1999) are
also plotted in Figure 4.37 (□ and ■). As seen in this figure, the correlation by Ben-Lakhdar et
al. (1999) does not collapse the current data into a single line. Instead, the data are divided
into two regions: the high heat flux (symbols ■) and low heat flux (□).
The discrepancy between the two proposed correlations is mainly attributed to a different
definition of the Graetz number used by Ben-Lakhdar et al. (1999). Their Graetz number was
based on the non-Newtonian Reynolds and Prandtl numbers, which also depend on the
consistency index (K) and power law index (n) for a pseudoplastic fluid. Although the
consistency index definition of Ben-Lakhdar et al. (1999) cannot be applied, their correlation
for K was employed to predict the data in Figure 4.37 in the absence of any other information.
The proposed correlation given by Eq. (4.45) eliminates this problem as the Graetz number is
defined based on the carrier fluid properties.
400
Thw,m=39.6oC
} α*=1:12
0.24
Thw,m=14.0oC
(µcf/µw)
o Ben Lakhdar et al. (1999)

Thw,m=39.6 C
Thw,m =14.0 C
o
}
200
0.30
Xs
150
0.486
Nucf,L=4.0Gz
102
+15%
80
Re Pr
-15%
Gz =
x / Dh
80 150 200 400

102
Nusl,L= hsl,xDh /kcf
Figure 4.37. Comparison of local Nusselt number data with predictions for laminar Non-Newtonian
heat transfer to ice slurries (Stamatiou and Kawaji, 2004b)
4.8 Conclusions and outlook
Chapter 4 gives a comprehensive review of the work performed on the determination of heat
transfer in pipes and rectangular channels with and without bends. Because a convenient
theoretical treatment of turbulent ice slurry flows and convection is still missing, only
empirical equations for the Nusselt number can be determined. Several first approximation
functions are now available for correlation purposes.
Some experiments were performed in the rather early stages of ice slurry research. Studies
actually started not much earlier than the date of the creation of the Working Party of the
International Institute of Refrigeration on Ice Slurries, which organized its first workshop in
spring 1999. At this time it was not known that ice slurry exhibits a time-dependent behaviour
on the shape and number of the ice particles. This also results in a time dependent behaviour
of some physical properties. The density, enthalpy and specific heat capacity are not
influenced by this phenomenon, because the ice concentration remains constant. On the other
120
hand the rheological data such as shear stress, apparent viscosity, critical shear stress and the
effective thermal conductivity, etc. are dependent on this effect and change as a function of
time. Therefore, a question arises as to how precise the investigations can be without taking
conscious consideration of this time dependence. There surely is a potential for repeating and
improving some existing results.
The superheating of the continuous phase compared to the dispersed phase in all kinds of heat
exchangers is another recently discovered phenomenon, which causes difficulties and calls for
further systematic theoretical and experimental investigations.
1. Ayel, V.; Lottin, O.; Peerhossaini, H.: Rheology, Flow Behaviour and Heat Transfer of
Ice Slurries: a Review of the State of the Art, Int. J. Refrig. 26, no. 1, pp. 95-107, 2003.
2. Bédécarrats J.-P. ; Strub, F. ; Stutz, B.: Heat transfer for different slurries flowing in heat
exchanger. Fifth Workshop on Ice Slurries of the IIR, Stockholm, Sweden, pp. 94-101,
2002.
3. Bellas, J.; Chaer, I.; Tassou, S. A.: Heat transfer and pressure drop of ice slurries in plate
heat exchangers, Applied Thermal Engineering 22, pp.721-732, 2002.
4. Bel, O.: Contribution a l'étude du comportement thermo hydraulique d'un mélange
diphasique dans une boucle frigorifique a stockage d'énergie, PhD Thesis, L'Institut
National des Sciences Appliquées de Lyon, Lyon, 1996.
5. Ben Lakhdar, M.A.; Guilpart, J.; Lallemand, A.: Experimental study and calculation
method of heat transfer coefficient when using ice slurries as secondary refrigerant, Heat
and Technology 17, no. 2, pp. 49-55, 1999.
6. Brailsford, A.D.; Major, K.G.: The thermal conductivity of aggregates of several phases,
including porous materials, Brit. J. Appl. Phys. 15, pp. 313-318, 1964.
7. Cho, Y.I.; Ganic, E.N.; Harnett, J.P.; Rohsenow, W.M.: Basic concepts of heat transfer,
In: Rohsenow, W.M., Hartnett, J.P. and Cho, Y.I. eds., Handbook of Heat Transfer, Third
Edition, McGraw-Hill Publishing Company, New York: Chapter 1, p. 1.9, 1998.
8. Christensen, K.G..; Kauffeld, M.: Heat transfer measurements with ice slurry, IIR/IIF
International Conference on Heat Transfer Issues on Natural Refrigerants, 1997.
9. Cebeci, T.; Bradshaw, P.: Physical and computational aspects of convective heat transfer.
Springer-Verlag, New York, 1984.
10. Class, G.; Meyder, R.; Sengpiel, W.: Measurements of spatial gas distribution and
turbulence structure in developing bubbly two-phase flow in vertical channels, Proc. Int
Conf. Multiphase Flows ’91, Tsukuba, Japan, vol. 1, pp. 473-477, 1991.
11. Dittus, F.W.; Boelter, L.M.K.: Heat transfer in automobile radiators of the tubular type.
University of California (Berkeley) Pub. Eng. 2, p. 443, 1930.
12. Darby, R.: Hydrodynamics of Slurries and Suspensions, Cheremissinoff, ed.: in
Encyclopedia of Fluid Mechanics, Vol. 5, Slurry Flow and Technology, Chapter 2, Gulf
Pub., Houston, pp. 49-91, 1986.
13. Egolf, P.W.; Manz, H.: Theory and modelling of phase change materials with and without
mushy regions, Int. J. Heat and Mass Transfer 37, No. 18, pp. 2917-2924, 1994.
14. Egolf, P.W.; Frei, B.: The continuous-properties model for melting and freezing applied to
fine-crystalline ice slurries, First Workshop on Ice Slurries of the IIR, Yverdon-les-Bains,
Switzerland, pp. 25-40, 1999.
15. Egolf, P.W.: Weiss, D.: Difference-quotient turbulence model: Analytical solutions for the
core region of plane Poiseuille flow. Phys. Rev. E 62, 1-A: pp. 553-563, 2000.
121
16. Eucken, A.: Allgemeine Gesetzmässigkeiten für das Wärmeleitvermögen verschiedener

Stoffarten und Aggregatzustände. Forsch. Arb. Ing. Wes. 11, No. 1, pp. 6-20, 1940.
17. Frei, B.; Egolf, P. W.: Viscometry applied to the Bingham substance ice slurry. 2nd IIR
Workshop on Ice Slurries, pp. 48-59, 2000.
18. Frei, B.; Boyman, T.: Plate heat exchanger operating with ice slurry. First International
Conference and Business Forum on Phase Change Materials and Phase Change Slurries,
23-26 April, Yverdon-les-Bains, Switzerland, 2003.
19. Gnielinski, V.: New equations for heat and mass transfer for turbulent pipe and channel
flow, Int. Chem. Eng. 16, no. 2, pp. 359-368, 1976.
20. Gnielinski, V.: Forced convection ducts, in E.U. Schlunder (Ed.), Heat exchanger design
handbook, Hemisphere, D.C., pp. 2.5.1-2.5.3, 1983.
21. Guilpart, J.; Fournaison, L.; Lakhdar, B.M.A.: Calculation Method of Ice Slurries
Thermophysical Properties-Application to Water/Ethanol Mixture, 20th International
Congress of Refrigeration, II/IIF, Sydney, 1999.
22. Guilpart, J.; Fournaison, L.; Ben Lakhdar, M.A.; Flick, D.; Lallemand. A.: Experimental
study and calculation method of transport characteristics of ice slurries, 1st IIR Workshop
on Ice Slurries, Yverdon-les-Bains, Switzerland, pp. 74-82, 1999.
23. Hartnett, J.P.; Kostic M.: Heat Transfer to Newtonian and Non-Newtonian Fluids in
Rectangular Ducts, In: Hartnett JP and Irvine TF-Jr. eds, Advances in Heat Transfer,
Academic Press Inc., San Diego, vol. 19, pp. 247-366, 1989.
24. Holman, J.P.: Heat Transfer, McGraw-Hill, Inc., New York, pp. 320-321, 1998.
25. Horibe, A.; Inaba, H.; Haruki, N.: 2001, Melting heat transfer of flowing ice slurry in a
pipe, in: S. Fukusako (Ed.), 4th Workshop on ice slurries, Osaka, Japan, pp. 145-152,
2001.
26. Jeffrey, D. J.: Conduction through a random suspensions of spheres, Proc. R. Soc. Lond.,
A335, pp. 355-367, 1973.
27. Jensen, E.N.; Christensen, K.G.; Hansen, T.M.; Schneider, P.; Kauffeld, M.: Pressure drop
and heat transfer with ice slurry, in: Final Proceedings of the IIR-Gustav Lorentzen
Conference on Natural Working Fluids at Purdue, Ray W. Herric Laboratories, West
Lafayette, IN, July 25-28, pp. 572-580, 2000.
28. Kawaji, M., Stamatiou, E.; Hong, R.; Goldstein, V.: Ice-slurry flow and heat transfer
characteristics in vertical rectangular channels and simulation of mixing in a storage tank,
Proceedings of the 4th IIR Workshop on Ice Slurries, Osaka, Japan, 2001.
29. Kawanami, T.; Fukusako, S.; Yamada, M.: Cold Heat Removal Characteristics from
Slurry Ice as a New Phase Change Material, Natural working fluids '98, IIR-Gustav
Lorentzen Conference: Proceedings of the conference of Commission B2 with B1, E1 &
E2, June 2-5 Oslo, Norway, pp. 146-156, 1998.
30. Kawanami, T.; Fukusako, S.; Yamada, M.; Itoh, K.: Experiments on melting of slush ice
in a horizontal cylindrical capsule, Int. J. Heat and Mass Transfer, 42, pp. 2981-2990,
1999.
31. Kawanami, T.; Yamada, M.; Fukusako, S.: Melting characteristics of fine particle ice
slurry at the return bend of flow path, 3rd IIR Workshop on Ice Slurries, Lucerne, pp. 69-
75, 2001.
32. Kitanovski, A.; Sarlah, A.; Poredoš, A.; Egolf, P.W.; Sari, O.; Vuarnoz, D.; Sletta, J.P.:
Thermodynamics and fluid dynamics of phase change slurries in rectangular channels,
International Congress of Refrigeration, Washington, D.C., 2003.
33. Knodel, B.D.; France, D.M.; Choi, U.S.; Wambsganss, M.W.: Heat transfer and pressure
drop in ice-water slurries, J. App. Thermal Engineering 20, pp. 671-685, 2000.
34. Knudsen, J.G.; Katz, D.L.: Fluid Dynamics and Heat Transfer, McGraw-Hill Book
Company Inc, New York, p. 241, 1958.
122
35. Nasr-El-Din, H.; Shook, C.A.; Esmail, M.N.: Isokinetic Probe Sampling from Slurry
Pipelines, Canad. J. of Chem. Eng. 63, October, pp. 743-746, 1984.
36. Nørgaard E.; Sørensen, T.A.; Hansen, T.M.; Kauffeld, M.: Performance of components in
ice slurry systems, plate heat exchanger, fittings. Third Workshop on Ice Slurries of the
IIR, Horw/Lucerne, Switzerland, pp.129-136, 2001.
37. Petukhov, B.S.: Heat transfer and friction in turbulent pipe flow with variable physical
properties, Advances in Heat Transfer, Vol. 6, Academic Press, New York, pp. 503-564,
1970.
38. Roy, S.K.; Avanic, B.L.: Turbulent heat transfer with phase change suspensions, Int.
J.Heat Mass Transfer 44, pp. 2277-2285, 2001.
39. Sari, O.; Meili, F.; Vuarnoz, D.; Egolf, P.W.: Thermodynamics of moving and melting ice
slurries, Second Workshop on Ice Slurries of the IIR, Paris, May, pp.140-153, 2000.
40. Shah, R.K.: Assessment of modified Wilson plot techniques for obtaining heat exchanger
design data, Proc. 9, Int. Heat Transfer Conference, Vol. 5, pp. 51-56, 1990.
41. Shah, R.K.; London, A.L.: Laminar Flow Forced Convection in Ducts: A Source Book for
Compact Heat Exchanger Analytical Data, Academic Press, New York, pp. 196-222,
1978.
42. Shin, S. et al., Viscosity and conductivity measurements for dilute dispersions of rod like
paraffin particles in silicone oil, Int. Comm. Heat Mass Transfer 29, pp. 203-211, 2002.
43. Shook, C.A.; Rocco, M.C.: Slurry Flow: Principles and Practice, Butterworh-Heinemann
Series in Chemical Engineering, Boston, pp. 27-33, 1991.
44. Sieder, E.N.; Tate, G. E.: Heat transfer and pressure drop of liquids in tubes, Industrial
and Engineering Chemistry 28, No. 12, pp. 1429-1435, 1936.
45. Skelland, A.H.P.: Non-Newtonian Flow and Heat Transfer, John Wiley and Sons, Inc.,
New York, p. 405, 1967.
46. Snoek, C. W.; Bellamy, J.: Heat transfer measurements of ice slurry in tube flow, ExHFT
4, Brussels, Belgium, 1997.
47. Stamatiou, E.; Kawaji, M.; Lee, B.; Goldstein, V.: Experimental investigation of ice-slurry
flow and heat transfer in a plate-type heat exchanger, Proceedings of the Third IIR
Workshop on Ice Slurries, Lucerne, Switzerland, May, pp. 61-68, 2001.
48. Stamatiou, E.; Kawaji, M.; Goldstein, V.: Ice fraction measurements in ice slurry flow
through a vertical rectangular channel heated from one side, Proceedings of the Fifth IIR
Workshop on Ice Slurries, Stockholm, Sweden, May 30-31, 2002.
49. Stamatiou, E.; Kawaji, M.: Thermal and flow behavior of ice slurries in a vertical
rectangular channel – Part I: Local distribution measurements in adiabatic flow, Int. J
Heat Mass Transfer, 48 (17), pp. 3527-3543, 2005.
50. Stamatiou, E.; Kawaji, M.: Thermal and flow behavior of ice slurries in a vertical
rectangular channel – Part II. Forced convective melting heat transfer, Int. J Heat Mass
Transfer, 48 (17), pp. 3544-3559, 2005.
51. Stamatiou, E.: Experimental Study of the Ice Slurry Thermal-Hydraulic Characteristics in
Compact Plate Heat Exchangers, Ph.D. Dissertation at the University of Toronto, 2003.
52. Thomas, D.G.: Transport Characteristics of Suspension: VIII. A note on the viscosity of
Newtonian suspensions of uniform spherical particles. J. of Colloid Science 20, pp. 267-
277, 1965.
53. Wasp, E. J.: Solid-liquid flow. Slurry Pipeline Transportation, Series on Bulk Material
Handlings, Vol. 1, No. 4, 1977.
54. Wilson, R. E.: A basis for rational design of heat transfer apparatus, ASME Transactions
37, pp. 47-70, 1915.
123
HFC conversion
Absorption
Cold Storage
Secondary Refrigerants
Ice Slurry
Floating HP
Speed Variation
R&D competencies to enhance your energy efficiency !

EDF R&D
Eco-efficiency and industrial processes
Les Renardières – 77 250 Moret-sur-Loing, France
Tel : (33) 160 73 65 41 – Web Site: www.edf.fr
CHAPTER 5. ICE-SLURRY PRODUCTION
The production of ice slurry is the key technology in any ice slurry system. It is also the area
of most research in the field of ice slurry as the cost of current ice slurry generators prohibits
the wide use of this promising technology. This chapter gives an overview of all currently
known methods of ice slurry production.
5.1 Fundamentals of ice slurry generation

by Jeroen Meewisse
5.1.1 ICE CRYSTALLIZATION
Ice slurries consist of water solutions in which small ice crystals are present. The creation of
these small ice crystals is the most difficult part in ice slurry generation (List, 2000). A
number of steps are required for making small ice crystals in a solution. First a driving force
needs to be applied to create supersaturation in the solution, making it possible to start the
formation of fresh ice crystals, a process called nucleation. Finally there is a growth phase,
during which the ice crystals grow larger until the required size is reached. Further reading:
Meyerson (1993), Arkenbout (1995) and Leloux (1999).
Supersaturation of solutions can be achieved in a number of ways: the temperature can be

decreased and the pressure or the concentration of additive can be changed. In most types of
ice slurry generator, a temperature difference is applied, but in some generators pressure
methods are applied.
Nucleation can occur either homogeneously, requiring a pure solution, or heterogeneously,

aided by foreign material present in the solution. This material can be dirt particles, walls of
tanks or other foreign objects. Ice crystals already present can also form fresh nuclei, called
secondary nucleation. Primary nucleation is the direct formation of ice crystals from a
solution; secondary nucleation is the formation of new ice crystals from ice crystals already
present in the solution. Control of homogeneous ice nucleation in a continuously operating
crystallizer is rather difficult. The freezing temperature can fall to low values and the initial
ice formation rate might become so high that the apparatus will become covered by ice and
malfunction. Nucleation in ice slurry generators therefore is engineered to occur
heterogeneously.
The fact that ice crystals can be used to start ice formation is often utilized in ice slurry
generation. Seeding, or the addition of some ice crystals to a mildly supercooled solution, can
effectively initiate ice formation and control the temperature at which this happens. Seeding
also improves operating stability, as initial ice formation rates are moderate.
During the growth phase ice crystals can grow at various rates, depending on conditions in the
crystallizer such as the driving force, the remaining supersaturation of the solution and the
residence times. Some ice slurry generators produce large ice particles or even ice in large
chunks (plate ice), so that some form of crushing or grinding is required to achieve the
required ice slurry properties. It may also be necessary to install an ice concentrator, to
increase ice fraction in the ice slurry to more economic levels (Kauffeld et al., 1999).
It is important to realize that ice slurry is a dynamic fluid. The properties of ice slurry are not
only determined by its generation process, but also by its handling history including transport,
125
storage, concentration, utilization and other processes. Due to the effects involved with ice
crystals such as attrition, agglomeration and ripening, the size and shape of ice crystals are not
fixed once ice slurry is removed from an ice slurry generator (Pronk et al., 2002). Stabilizers
added to the slurry may reduce possible effects (Inada et al., 2000). These processes are
discussed in detail in other chapters.
5.1.2 HEAT AND MASS TRANSFER ASPECTS
Two requirements are of decisive importance for ice slurry generation. First a generator must
make ice crystals of a correct size and shape, to ensure the fluidity of the slurry. Rounded and
sufficiently small ice crystals are most suited for this. Secondly a high heat transfer rate is
required, so that ice slurry production rates can be high and energy and investment costs are
low.
Ice easily sticks to metallic walls at lower temperatures. This is disadvantageous if heat must
be transferred through these walls as ice has low thermal conductivity compared to metals and
the overall heat transfer rate decreases. As the ice sticking to the walls forms an increasingly
thick layer, heat transfer rates may drop drastically. Defrosting is then required to restore heat
transfer capacity. This cycle is time consuming and may add heat into the ice slurry thereby
further reducing efficiency of the ice slurry generation. Prevention of build-up of ice layers on
heat exchanging surfaces is therefore required. Mechanical solutions may increase investment
and maintenance costs. On the other hand, a number of ideas have been proposed that remove
ice from walls and simultaneously increase heat transfer rates and ice production rates.
In summary, an ideal ice slurry generator should produce ice crystals of desired crystal size at
high heat transfer rates, without ice sticking to heat exchanger walls.
5.1.3 CATEGORIES OF ICE SLURRY GENERATOR
No single type of ice slurry generator is the most suitable for all situations. Different ice slurry
generators are used for different applications. At present some systems are already applied
commercially while many systems are in development. Each method has its particular merits
and disadvantages and a detailed discussion of the most promising ice slurry generator types
is given in section 5.2. Topics covered for each system include: operating principle, working
media, economic and energetic performance, practical machines, operating ranges and
reliability.
Traditionally, ice used in cooling applications is produced in relatively large sizes, for
example as flake or plate ice. To obtain ice in slurry form with acceptable fluidity, additional
equipment is required to crush the ice chunks. Defrosting cycles are also common in these
systems either as an integral part of the ice generation process, or to periodically remove
build-up of unwanted ice stuck to heat exchangers. Although the traditional methods do not
appear to be efficient, these can still be used in some situations where demands on ice slurry
crystals are not as strict or where pure fresh water ice slurry is required. Also investment costs
can be low for these systems.
A number of technical solutions are available to solve the problem of ice sticking to heat
exchanging surfaces. Additionally, some methods actively increase heat transfer rates and
126
consequently ice slurry production rates. Here the basic principles of the ice slurry generators
are briefly introduced. A more detailed description of the individual techniques is given in
Chapter 5.2:
– Continuous ice removal from heat exchanging surfaces with mechanical equipment.
Ice forms on or near (tubular) heat exchanger surfaces, where it is removed using
scrapers, brushes, helical screws or rotating rods. (Kauffeld et al., 1999) (see 5.2.1).
– Water as refrigerant. A direct contact process with evaporation of the primary
refrigerant water which occurs under triple point conditions, with pressures below 6
mbar. Condensation of the water vapour occurs through compression and subsequent
cooling (Paul, 1996) or on a heat exchanger in the ice slurry generating vessel (Zakeri,
1997) (see 5.2.2).
– Direct contact evaporation of an immiscible primary refrigerant. This makes the heat
exchanger walls obsolete and leads to increased heat transfer rates. The method is
similar to the vacuum ice generation method, except that a primary refrigerant other
than water is used (Wobst, 1999) (see 5.2.3).
– Direct contact of an immiscible secondary refrigerant. This makes the heat exchanger
walls in the ice slurry generating vessel obsolete, but compared with direct contact
evaporation, a heat exchanger (evaporator) is needed in the primary refrigeration unit
(Watanabe et al., 1995) (see 5.2.4).
– Supercooling effect. The supercooling effect can be used to cause the ice to crystallize
at a different location in the system, for example in a tank past the actual heat
exchanger cooling the carrier fluid. Nucleation can be triggered by various methods,
for example, by seeding or by acoustic methods (Fukusako, 1999) (see 5.2.5).
– Removal of ice particles from the evaporator surface by increased ice slurry flow
velocity and increased evaporation temperature upon occurrence of ice crystals (see
5.2.6).
– Special coatings or materials. There exist a number of materials that prevent adhesion
of ice to metallic surfaces. Some may be suitable for ice slurry generators, so that ice
slurry can be produced in a basic heat exchanger/evaporator. (Zwieg et al., 2002) (see
5.2.7).
– Continuous ice removal from the heat exchanging surfaces with a fluidized bed. The
impacts of solid particles (e.g. steel) prevent build-up of an ice layer on the (tubular)
wall surface of the ice slurry making heat exchanger and increase heat transfer rates
(Meewisse, 2001) (see 5.2.8).
– High pressure methods. Raising water to a high pressure causes the freezing point to
decrease. The high pressure solution is cooled and after release of the pressure, ice
crystals are formed (Sanz et al., 2001) (see 5.2.9).
– Cyclic removal of ice from ice banks by heat (defrosting) or with mechanical means,
where ice crushing devices may be used. Most traditional ice cooling systems are
based on similar methods (Fukusako, 1999) (see 5.2.10).
Some methods also increase heat transfer rates compared to regular heat exchangers. This
mostly happens by increased turbulence near heat exchanger walls, induced for example by
mechanical devices. If no heat exchanger walls are required, as in triple point systems
(vacuum ice) or direct contact systems, heat transfer rates are also augmented.
127
5.2. Current and possible methods of ice slurry production
The different methods briefly mentioned in section 5.1 are described in more detail in the
following sub-sections.
5.2.1 SCRAPED-SURFACE ICE SLURRY GENERATOR

by E. Stamatiou and M. Kawaji
Initially all scraped-surface type ice slurry generators were of tubular type with blades
scraping the inside of a refrigerated cylinder, similar to those applied for decades in the food
processing industry (e.g. for juice concentration). Recently plate type scraped-surface ice
slurry generators have been developed to production state.
5.2.1.1 Scraped-surface ice slurry generator of a tube type

The scraped-surface ice slurry generator is the most technologically developed and widely
accepted ice slurry generation process. It consists of a shell-and-tube type heat exchanger that
has a rotator/blade assembly housed inside the tube where the ice slurry is generated, and a
refrigerant stream that evaporates on the outer shell-side of the heat exchanger. The ice slurry
generator is capable of producing a pumpable mixture of small ice crystals and water from a
binary super-cooled solution. The high agitation speeds and phase change involved offer
exceptionally high heat transfer rates, which in turn result in rapid product cooling and a good
end product.
5.2.1.1.1 Principles
The scraped-surface ice slurry generator consists of a circular shell-and-tube type heat
exchanger, cooled on its outer shell side by an evaporating refrigerant, and scraped on its
inner side by spring loaded rotating blades, orbital rods, brushes or helical screws to prevent
any crystal deposits on the cooled surface (see Figures 5.1a to 5.1d). This scraping action is
required to prevent formation of an ice layer on the ice slurry generator walls, which would
otherwise introduce an additional thermal resistance and could seriously hinder the heat
transfer rate. The continuous accumulation of the ice layer on the ice slurry generator walls
would eventually block rotation of the scraper blades, and cause freezing up of the ice slurry
generator. To prevent the freeze-up of the ice slurry generator walls, solutes are added to the
water and these depress the freezing point of the solution but also affect the heat transfer rates.
In addition, turbulence is mechanically induced into the ice slurry by the action of the rotating
scraper blades mounted in the centre of the heat exchanger, thus greatly increasing the heat
transfer rates facilitating the production of a homogeneous ice slurry mixture.
A condensing refrigeration unit, consisting of a compressor, condenser and an expansion

device (such as a thermal expansion valve or capillary tubes), normally supplies the
refrigerant to the shell-side of the ice slurry generator, which is also known as the evaporator
in the refrigeration cycle (see Figure 5.2). As the refrigerant evaporates at a low pressure
through the outer pipe of the ice slurry generator, it withdraws heat from its surroundings,
cooling off the incoming binary solution (a dilute glycol or inorganic brine solution) flowing
through the inner pipe of the ice slurry generator. Using this indirect cooling process, ice
slurry is generated on the tube side of the ice slurry generator. The refrigeration unit may
employ a liquid overfeed or flooded type scheme depending upon the application; however, it
has been reported that flooded type refrigeration systems could cause the accumulation of oil
lubricants in the evaporator side due to low temperatures involved and thus reduce heat
transfer rates (Briley, 2002). At the exit of the evaporator, a refrigerant vapour with enough
128
superheat is generated from this indirect heat exchange process that is recompressed and
recondensed at high pressures to continue the refrigeration cycle.
Liquid Ice Inlet
Refrigerant
Outlet
Refrigerant
Inlet
Liquid Ice Outlet
Figure 5.1b. Orbital rod ice slurry generator

(Courtesy of Paul Muller Co.)
Figure 5.1a. Scraped-surface ice slurry
generator
(Courtesy of Sunwell)
Mixture Refrigerant
Evaporator
Refrigerant Ice-
inside Slurry
Refrigerant
Scraper
Figure 5.1d. Scraped-surface ice slurry generator

with helical screw – originally designed for the
Figure 5.1c. Scraped-surface ice slurry production of flake ice
generator (Courtesy of Ziegra)
(Courtesy of Integral)
129
Compressor
Ice
Slurry
Condenser
Ice slurry generator

Receiver
Expansion
Valve Brine in
Figure 5.2. Schematic of an ice slurry generation system
Very little information is available regarding the ice crystallization process in scraped-surface
ice slurry generators and most theories of ice crystallization mechanisms largely depend on
anecdotal evidence and are somewhat speculative. Wang and Kusumoto (2001), Wang and
Goldstein (1996), Snoek (1993) and Gladis et al. (1999) have claimed that during the ice
slurry generation process, as the binary liquid is super cooled below its freezing point,
spontaneous nucleation initiates the growth of ice crystals in the bulk liquid; and the rotating
wiper blades or orbital rods are hypothesized to continuously disturb the thermal boundary
layer preventing the formation of ice crystal deposits on the refrigerated surface, and transfer
the cold fluid from the vicinity of the chilled wall to the bulk of the solution. On the other
hand, Russell et al. (1999), Schwartzberg and Liu (1990), Schwartzberg (1990) and Hartel
(1996), who studied scraped-surface crystallizers in relation to ice cream making, and
Armstrong (1979) and Patience et al. (2001) who studied scraped-surface crystallizers for the
production of paraxylene, all rather suggest that ice crystals are formed near the refrigerated
barrel wall and are dispersed into the centre of the ice slurry generator tube by the action of
the scraper blades. To date the only known experimental investigations that support the
heterogeneous ice crystal formation at the chilled wall are those of Schwartzberg (1990) and
Sodawala and Garside (1997), who used video microscopy to study the formation and growth
of ice crystals in an aqueous sucrose-water solution following an exposure to large
temperature gradients in a scraped-surface freezer and in a scraped-surface flow cavity,
respectively. Both of these studies rather supported the idea that scraper blades shear off
dendritic ice crystals that are growing on the chilled surface, and disperse them into the bulk
phase.
Regardless of the mechanism by which nucleation occurs, the rate of nucleation largely
depends on the degree of local supercooling at the wall, which is primarily dependent on the
refrigerant temperature (Russell et al., 1999). Thus, to make the best use of the scraped-
surface ice slurry generator, it is essential to operate the units at high log mean temperature
130
differences between the refrigerant and ice slurry streams so as to obtain increased heat
transfer rates. Nonetheless, the addition of a freezing point depressing agent inherently
reduces this temperature driving force and thus reduces the overall heat transfer rates. To
compensate for this loss, ice slurry generators must be optimally designed to have very high
heat transfer coefficients on the refrigerant and ice slurry sides. This requires accurate
determination of the refrigerant and ice slurry heat transfer coefficients from laboratory tests.
5.2.1.1.2 Practical systems

Most scraped-surface ice slurry generators are made of 316-grade stainless steel with plastic
spring-loaded scraper blades, although efforts have been made to use extruded materials to
minimize production costs. In the food industry, tubes are nearly always manufactured from
nickel with a hard chrome-plated finish on the product contact side, because nickel has a far
superior thermal conductivity than stainless steel and offers a good wear resistance when
application of stainless steel scraper blades is necessary (Smith, 1972). In addition, ice slurry
generators are insulated on their outside perimeter with a material that is compatible with the
temperature of the refrigerant.
The scraped-surface ice slurry generators use an electric speed motor to drive the scraper
blades at a typical rotational speed of about 450 rpm. Using higher rotational speeds will
generally result in higher heat transfer rates but lead to disproportionate power requirements.
In food plants, a hydraulic motor normally drives the shaft to avoid grease leakage from
mechanical parts into the product side.
Orbital rod evaporators are falling-film type ice slurry machines with vertical spinning rods
and thus are inherently manufactured in a vertical configuration. The falling film liquid
solution acts as a lubricant minimizing wear and ensuring that the orbital rods do not contact
the tube walls. In this system, component wear is reported to be minimal due to the
lubricating and cooling effects of the low temperature solution that continuously wets all
moving parts (Gladis et al., 1999).
5.2.1.1.3 Capacity and size

Scraped-surface ice slurry generators are manufactured with a wide range of cooling
capacities and can typically produce about 3 to 400 tons of ice per day (10 to 1400 kW
refrigerating capacity). Some typical sizes and capacities for both the scraped-surface and
orbital rod evaporators are listed in Table 5.1.
Scraped-surface units are available in both vertical and horizontal arrangements. Horizontal
arrangements become attractive when space and ceiling requirements are limited. On the other
hand, orbital rod evaporators are manufactured exclusively in a vertical orientation.
Scraped-surface ice slurry generator tubes are 1.8 to 2.4 metres long with a typical inner ice
slurry side diameter of 0.15 m. The inlet and outlet ice slurry streams to the ice slurry
generator are often made tangential to enhance turbulence and mixing.
131
Table 5.1. Typical specifications for ice slurry generators

Scraped-surface Orbital rod
Ice slurry side

Tube material 304 grade stainless steel 304 grade stainless steel for NH3
Copper 122 for HCFC
Freezing point depressant NaCl, ethanol, glycol NaCl, ethanol, glycol, urea
Crystal sizes 250 to 500 µm 50 to 100 µm
Heat transfer area 0.85 m2 per tube 0.13 m2 per tube
Tube length 1.8 – 2.4 m 1.20 m
Flow rate per tube 10-23 litres/min 6 litres/min
Ice fraction change per tube 15% 6-8%
Nominal cooling capacity 21-85 kW 10-1800 kW
Agitation mechanism Plastic scraper blades Metal orbital rods
Agitation speed 450 rpm 850 rpm
Power requirements per heat transfer area 1.2-1.8 kW/m2 0.22 kW/m2
Refrigerant Side
Refrigerant type R22, R404, R717 NH3, R22, R717, R134a
Evaporating temperature –10 and –19oC –10 to –8oC for stainless steel
–10 to –4.4oC for copper tubes
Typical refrigerant flow rate 0.15 kg/s matches instantaneous
evaporator capacity
Economics
Cost US$300 - $600/kW US$160/kW
5.2.1.1.4 Costs and energy consumption (performance characteristics)

Scraped-surface ice slurry generators typically require 1 to 2 kW/m2 of scraper driving power
to generate the ice slurry. Due to a lower ice content in the ice slurry generator, ice slurry
generators typically require lower scraper driving power than ice cream and freeze
concentrate maker machines; the latter use about 4-12 kW/m2 (Schwartzberg, 1990).
5.2.1.1.5 Economics, investments, operating costs and reliability

Scraped-surface ice slurry generators typically cost about US$1500-2000 per ton of ice
produced. (Producing 1000 kg of ice in 24 hours requires about 3.5 kW refrigeration
capacity). Orbital-rod evaporators are cheaper costing about US$550 per ton of ice produced,
which makes them the lowest cost dynamic ice-slurry generator type (Gladis et al., 1999).
5.2.1.1.6 Applications
Traditionally, scraped-surface ice slurry generators have been used by process chemical
industries for the separation of organic mixtures such as paraxylene from its isomers largely
performed by crystallization. These crystallizers are typically longer (6 to 12 metres long)
than ice slurry generators with inner tube diameters ranging from 0.15 to 0.30 metres, and are
normally installed as a series of several parallel tubes to achieve plug flow conditions. The
scraper blades rotate at moderate speeds (15 to 30 rpm) necessary for separating the ice
crystals from the solution. Using higher agitation speeds will promote heat transfer rates and
produce a more homogeneous, stable ice crystal mixture.
In the food industry, scraped-surface crystallizers are used for the production of viscous,
sticky fluids such as ice cream, frozen fruit concentrated products, slush-ice cooler beverages,
margarine, butter, process cheese, chilli, baked beans, pet foods, and marshmallow toppings.
Ice cream plants normally use scraped-surface crystallizers to harden ice cream to –7°C
before packaging and directing the ice cream to a hardening process where it is frozen down
to –29°C (Briley, 2002). Recently, slush ice cooler beverages are becoming increasingly
132
popular among consumers compared with crushed and blended ice drinks due to the fine ice
crystal texture, choice of different flavours and colours, and the increased cooling effect
provided by these beverages.
On the other hand, ice slurry generators have a more limited market and are currently used for
directly cooling meat, poultry, vegetable and fish products as well as for the supply of ice in
supermarket displays, onboard trawlers and thermal energy storage (TES) systems. More
recently ice slurry generators have been used in food processing plants for cooling milk and
dairy products where the use of HydroChloroFluoroCarbon (HCFC) refrigerants will pose
environmental hazards. Furthermore, many scraped-surface ice slurry generators have already
been installed worldwide for building air-conditioning applications and have been shown to
provide tremendous energy cost and power savings (Wang and Kusumoto, 2001; Nelson et
al., 1999).
5.2.1.1.7 Advantages
Scraped-surface ice slurry generators offer much smaller heat transfer units than comparable
HVAC cooling technologies, with reduced space, weight, and power requirements which
counterbalance their higher capital costs. The massive savings in the power requirements can
result in relatively short payback periods (Gladis et al., 1999; Wang and Goldstein, 1996;
Wang and Kusumoto, 2001).
Another major advantage of existing scraped-surface ice slurry generators over other ice
slurry generation technologies is that the mechanical agitation results in exceptionally high
heat transfer rates that translate into rapid cooling rates providing an excellent end product.
This is especially advantageous when immediate treatment of a product is required such as in
the direct spraying of seafood with a nozzle to prevent hot spot formation. Furthermore,
scraped-surface ice slurry generators offer a modular design, which allows for easy expansion
with growth in demand.
5.2.1.1.8 Disadvantages and limitations

One major disadvantage of scraped-surface ice slurry generators is the high capital cost of the
ice slurry generator units (Wang and Goldstein, 1996) due to the custom evaporator design.
To make the scraped-surface ice slurry generators commercially viable and capable of
competing with existing HVAC technologies, the cost of ice slurry generators must be
substantially reduced. Current ice slurry generators cost from US$160 to US$600 per kW of
refrigeration capacity (see Table 5.1).
Another limitation of existing scraped-surface ice slurry generators is the minimum

concentration of the freezing point depressant that can be used on the ice slurry side to
generate the ice. At very low depressant concentrations the continuous accumulation of ice
layers on the ice slurry generator walls cannot be prevented which would eventually block the
rotation of the scraper-blades and cause the freeze-up of the ice slurry generator. Increasing
the additive concentration prevents freeze-up, however, it would reduce the heat transfer rates
and immediately affect the ice slurry temperature. Existing scraped-surface ice slurry
generators are reported to work very reliably with minimum additive concentrations that have
a corresponding freezing point temperature of –2°C.
One additional drawback of existing scraped-surface ice slurry generators is that the rotating
scrapers, brushes and orbital rods wear over time and have to be replaced at a given time
interval. The replacement period varies depending upon the length of operation of the unit.
133
5.2.1.1.9 Working fluids, type of additive, ice fraction range

The scraped-surface ice slurry generators can use a broad range of organic and inorganic
additives such as ethanol, ethylene and propylene glycol or sodium chloride in a water
solution (salt solutions are generally referred to as brines).
The ice slurry generator is capable of producing ice slurry with an exit ice fraction ranging
from 0% (as a conventional chiller) to 35% (Wang and Goldstein, 1996; Wang and
Kusumoto, 2001). To achieve higher ice fractions, recycle streams, multiple passes, ice
concentrators or several modular ice slurry generators can be used in series. The high
agitation speeds involved produce homogeneous ice slurry with fine crystals ranging from 25
to 250 µm in size (Wang and Goldstein, 1996; Wang and Kusumoto, 2001). The crystal size
depends upon the operating conditions, the type of additive and its concentration. The orbital
rod evaporator produces ice slurry with comparable ice crystal sizes ranging from 50 to
100 µm (Yundt-Jr., 2002).
5.2.1.2 Scraped-surface ice slurry generator of disk type

by Adrien Laude-Bousquet
There are two types of disk type scraped-surface ice slurry generators:
– disk type scraped-surface ice slurry generators with an envelope,
– immersed disk type scraped-surface ice slurry generators.
5.2.1.2.1 Disk type scraped-surface ice slurry generators with an envelope
Figure 5.3. Double-walled tubular ice slurry generators with an envelope
Double-walled tubular ice slurry generators resemble conventional scraped-surface ice slurry
generators, but brushes are used instead of scrapers (Figure 5.3). In most ice slurry
generators, the brush is fitted onto a wiper shaft in a helicoidal manner, thus ensuring not only
scraping but also ice slurry circulation.
134
The supercooling phenomenon is controlled by regulating the rotation speed of the brushes.
The refrigerating capacity of ice slurry generators of this tubular type is generally quite low,
usually under 10 kW. In order to increase the refrigeration capacity, a circular plate design
can be used for the evaporator. This disk type scraped-surface ice slurry generator is equipped
with parallel refrigerating disks in an insulated jacket (Figure 5.4). The refrigerating disks are
hollow and the expansion of the refrigerant takes place within this hollow area (Figures 5.5
and 5.6). The refrigerants currently used are: R134a, R404A and R717 (NH3).
Figure 5.4. Disk type scraped-surface ice slurry generator with an envelope
Figure 5.5. The internal circuits of a Figure 5.6. Photo of an 80-kW ice
refrigerating disk slurry generator with
staggered disks
135
In order to obtain ice slurry at a temperature of –5.5°C, the evaporating temperature is

generally kept within a range of –11°C to –12°C.
The brushes are fitted onto a wiper shaft driven by a geared motor. The brushes are staggered
as shown in Figure 5.7.
Figure 5.7. Staggered brushes
The refrigerating capacity of this type of ice slurry generator is within a range of 7.5 to 120
kW. The capacity of the geared motor is 1.5 kW for an ice slurry generator capacity of 80 kW.
5.2.1.2.2 Immersed disk type scraped-surface ice slurry generator

The technology used for immersed disk type scraped-surface ice slurry generators is more
complex. It is used in a packaged unit comprising: 1) the ice slurry generator; 2) storage
vessel; and 3) reactivation.
Here again, the ice-slurry generator section consists of evaporating disks and brushes, but the
circulation of the ice slurry in the storage vessel is ensured by the wiper shaft and brush
orientation that in turn keeps the fluid moving (Figure 5.8).
Figure 5.8. Immersed disk type scraped-surface ice slurry generator
136
Brush scraped-surface ice slurry generators are very suitable for ice slurry production. The
scraping power required is roughly 1 to 2% of the refrigerating capacity. These ice slurry
generators can be operated at very low temperature differences, and also at very high
temperature differences if necessary. These ice slurry generators have been patented including
the components for storage, regeneration and circulation of ice slurries.
5.2.2 VACUUM ICE

by Michael Kauffeld
5.2.2.1 Principles
The most efficient means of producing ice slurry employs a direct contact heat transfer
vacuum freeze process where water is used as refrigerant. A schematic drawing of such a
cycle is shown in Figure 5.9. In the evaporator (1) which contains a water/salt solution, water
is evaporated and compressed to the condenser pressure (3). The pressure inside the
evaporator is close to the triple-point of water i.e. slightly below 6 mbar vapour pressure,
hence the common name vacuum ice arises for such systems.
make-up water
9 10
8
atm. pressure
7 6
1 5
ice slurry @
4 atm. pressure
Vacuum
Figure 5.9. Ice slurry generation with water as refrigerant (also known as vacuum ice)
The compressor has to handle a very large volume due to the low vapour pressures of water.
Most plants installed in the field use centrifugal compressors. To avoid the compression of
large volumes of water vapour, an alternative method has been investigated. Here a normal
evaporator cooled by a conventional refrigerant freezes the water vapour necessary to remove
the heat of fusion during the ice generation process in the liquid. In order to allow for
continuous operation, two evaporators are installed in the ice generating vessel. One works as
freezer for the flash vapour, building up a thin ice layer while the other evaporator is de-iced
by warm condensate of the primary refrigerant. Only a small displacement volume vacuum
pump is needed to maintain the triple point pressure in the ice slurry generator, see Figure
5.10 (Zakeri, 1996). This system has until now only been investigated at a research institute.
No practical applications are known so far.
137
Defrosting cycles may be required for this type of heat exchangers, as they are likely to freeze
if pure water is used. If a volatile freezing point depressant such as ethanol is added, the
vapour will be relatively rich in the freezing point depressant and ice formation during
condensation is unlikely.
Figure 5.10. Vacuum ice system with water vapour removal by freeze-out (Zakeri, 1996)
5.2.2.2 Practical systems

Most existing vacuum ice systems are installed in South Africa for the cooling of deep mines
(Paul, 1996). Most of these systems use centrifugal compressors, although ejector technology
is also applied in some cases. Depending on the temperature available for the condensation of
the water vapour, some systems are built as cascade systems using conventional refrigeration
equipment with CFCs, HCFCs, HFCs or ammonia as the upper stage. Basically all centrifugal
compressors used on the water side up to now are modified desalination compressors.
The system has also been used as a heat pump in Denmark since 1986. Seawater is partly
frozen using a vacuum ice system with a centrifugal compressor. This leads to larger
enthalpies removed from the sea water compared to normal sensible cooling of sea water in
traditional heat pumps (Madsbøll et al., 1994). Originally the upper stage of the cascade
system used R12 as refrigerant.
5.2.2.3 Capacity and size

The physical size of ice slurry generation equipment using water as refrigerant is quite large
when compared to other refrigerants due to the very low operating pressures and consequent
large vapour volumes. Also pressure drops have to be carefully avoided, since even very
small pressure drops of 1 mbar represent 25% of the evaporation pressure. Therefore
refrigerant (water vapour) pipes have to be very large in diameter and are kept as short as
possible.
138
The systems installed in the field using centrifugal compressors have refrigeration capacities
in the range of 150 kW to 3 MW. Systems with a vapour injector technology have been
reported to start at 50 kW refrigeration capacity (Malter, 1996).
Developments towards smaller capacities with mechanically driven compressors have been
performed at the Danish Technological Institute using a cycloid compressor. A mixture of
water and potassium carbonate with a freezing point of around –0.2°C was applied (Madsbøll
et al., 1994).
Other research work is known from Canmet, Ottawa Canada. The researchers there used
screw compressors and rotary vane compressors.
So far none of these research type systems have been developed to production level.
5.2.2.4 Advantages and limitations

The obvious advantage of using water as refrigerant in ice slurry production is the complete
environmental safety of this refrigerant, due to a very low GWP, its non-flammability and
non-toxic behaviour. Another advantage is the direct contact heat exchange resulting in lower
inefficiencies due to temperature differences found in traditional heat exchange equipment.
When compared to systems with direct evaporation of other refrigerants (described in section
5.2.3), water has the advantage of not affecting the ice slurry properties. In addition any
refrigerant water carried over in the ice slurry system will just mix perfectly with the carrier
fluid water and return safely to the ice slurry generator.
Nonetheless, the production of ice slurry using water as refrigerant (vacuum ice technology)
is limited to temperatures between the freezing point of water, i.e. 0°C and approx. –4°C due
to the tremendous increase in water vapour volume to be compressed at lower temperatures
(Malter, 1996).
Also the design of the individual vessels and pipes has to be done very carefully in order to
avoid pressure drops as these harm the low-pressure system much more than conventional
refrigeration systems.
5.2.3 DIRECT CONTACT GENERATORS WITH IMMISCIBLE REFRIGERANT

by Jeroen Meewisse
Some industrial applications use a primary refrigerant that directly evaporates in a fluid with
the goal of ice formation in this fluid. For desalination of sea water such a process has been
applied successfully (Byrd, 1986; Wiegandt, 1987). Application of a directly evaporating
refrigerant has also attracted attention for production of ice slurries for secondary cooling
systems (Knodel et al., 1986). Two basically different principles are observed: one where a
single fluid (water) is used both as the primary refrigerant and as the secondary refrigerant,
and the second where an immiscible primary refrigerant is used. For the first method low
pressure conditions are required, but for the second method higher pressures are often
required in the evaporator, which is also the tank in which the ice crystals form. In this section
the immiscible refrigerant method is discussed (the immiscible liquid system is described in
section 5.2.4, and the low-pressure methods are discussed in section 5.2.2.)
139
5.2.3.1 Principle
In a direct contact ice slurry generator the immiscible primary refrigerant is expanded and
then injected into a tank where it evaporates. The evaporation cools and supersaturates the
water and small dispersed ice particles are formed. Injection devices are required in the
evaporating tank in order to distribute the primary
refrigerant so that ice slurry will be formed evenly
throughout the tank. The injectors should be
designed so that there is no risk of ice formation on
the injectors themselves. Also, injection and
evaporation have to induce enough turbulence to
make sure that finely dispersed ice slurry will form.
Withdrawal of the ice slurry formed inside the tank

may be a problem, because operating pressures are
higher than atmospheric and because the primary
refrigerant must fully remain in the evaporator.
Another option is to use the tank similarly to an ice
bank, and install heat exchanging surfaces inside
the tank. Additional agitation by a stirrer may be
required to keep the ice slurry homogeneous. A
schematic diagram is shown in Figure 5.11,
(Wobst, 1999).
The main advantage of the direct contact system is Figure 5.11. Schematic diagram of
that no physical boundary exists between the direct contact ice slurry generator
primary refrigerant and the ice slurry, reducing (Wobst, 1999)
investments and increasing heat transfer rates. This,
however, can also be seen as a disadvantage if the main reason to install an ice slurry system
was to create a physical boundary between the primary refrigerant and the customer’s heat
exchanger.
5.2.3.2 Practical systems

A few research groups have worked on ice slurry generators based on direct contact
evaporation. These are the Chicago Bridge & Iron Company, USA (Knodel, 1986), ILK,
Dresden Germany (Wobst, 1999) and the University of Applied Sciences of Western
Switzerland in Yverdon-les-Bains (Hansen et al., 2001). Coldeco of France also owns patents
and aims for commercial manufacturing of direct contact ice slurry generators: Figure 5.12
(Chuard and Fortuin, 1999). So far, however, only experimental installations exist.
from storage tanks

1 and/or consumers
2 Ice-Slurry Circuit
5 Figure 5.12. Ice slurry
Refrigerant Circuit
generator based on direct
3 contact evaporation
towards storage tanks (Coldeco®)
4 and/or consumers
140

The ILK group in Dresden reported a heat transfer rate per unit volume in the direct contact
tank of approximately 1000 W/m3K (Wobst, 1999). This implies that the size of installations
will be rather large, but the tank(s) may also be used as storage tanks, eliminating the need for
additional tanks. For high capacity ice slurry generators several ice slurry generators have to
be used in parallel, reducing economic benefits of increasing scale in large scale applications.
For example, a tank with a volume of 10 m3 would be required for a cooling capacity of 50
kW if a temperature difference of 5 K can be achieved between the evaporating refrigerant
and the ice slurry.
5.2.3.4 Cost and energy consumption (performance characteristics)

As there is no heat transfer wall between the primary refrigerant and ice slurry, no extra
temperature difference is required and the primary cooling cycle uses less power than ice
slurry generators that have a solid heat transfer wall. The efficiency of the primary cooling
cycle may be slightly reduced by the injection nozzles; also some power may be required for
homogenisation of the ice slurry. No data for exact power consumption have been reported for
ice slurry applications.
5.2.3.5 Economics, investments, operating costs and reliability

As no heat exchanging surface is required, investment costs are relatively low for direct
contact systems compared with systems with heat exchanging surfaces. However, injection
nozzles, installed in the ice slurry storage tank, will add to the investment cost of direct
contact systems. The design of the generator tank should ensure that there are no areas in the
tank without agitation, which might cause non-uniform ice formation and therefore blockages.
Operating costs will be low, because there are few parts requiring maintenance and also
energy consumption is low, due to high heat transfer rates.
Reliability is associated with the quality of injection nozzles. If these are designed so that they
can remain free of ice and keep conditions inside the generator sufficiently turbulent,
continuous operation should be possible. Further reliability depends on how well primary
refrigerant and oil can be kept out of the secondary system, either by insolubility or by
additional separators.
5.2.3.6 Disadvantages and limitations

Mal-distribution of the evaporating refrigerant in the ice slurry generator will lead to
operational problems. This will cause non-uniform ice formation and perhaps blockages of
injection nozzles by ice freezing onto them.
Operating problems are encountered if the refrigerant is even slightly soluble in water, as the
primary refrigerant might leak into the ice slurry cooling system. This can cause safety
problems in the case of flammable or toxic refrigerants or environmental hazards if the
refrigerant has a high global warming potential, which is the case for the fully fluorinated
substances used by some research groups. A small amount of refrigerant always seems to get
trapped in the ice crystals, no matter how insoluble the fluid is. Similar problems are
associated with the lubrication oil of the primary refrigerant. Problems can be reduced by
installation of additional oil separators, which would, however, increase the cost of the
system.
5.2.3.7 Working fluids, type of additive, ice fraction range

Any refrigerant that is insoluble in water can be utilized, but since complete insolubility is
rare, a primary refrigerant should be as insoluble as possible. Hydrocarbons and most HFCs
141
can therefore be used, but ammonia cannot be used. For the case of ammonia as primary
refrigerant, an additional step can be included using an immiscible liquid, completely
insoluble in water, which also has a lower freezing point, for example, some kind of paraffin
(see section 5.2.4 for a more thorough description of this technique).
In the ice slurry itself, only additives can be used which do not evaporate easily. Ethanol
cannot be used, while inorganic additives may be acceptable. Direct contact systems have
been investigated extensively for applications in sea water desalination; sodium chloride can
therefore be used. Also the use of glycol has been reported in test installations.
The maximum ice fraction achievable in the system reported by Fukusako et al. (1999) is
40%. At these high concentrations it will be difficult to extract the ice slurry from the tank. It
is not known how the evaporation process is affected by higher ice fraction slurries.
5.2.4 DIRECT CONTACT GENERATORS WITH IMMISCIBLE LIQUID SECONDARY REFRIGERANT

by Pepijn Pronk, Jeroen Meewisse and Michael Kauffeld
An alternative arrangement for a direct contact heat exchanger tested by a Japanese company
was described in Fukusako et al. (1999). Here the primary refrigerant is used to cool a non-
soluble liquid, which is then sprayed into the ice slurry feed water, using similar injection
nozzles as for direct evaporation. The advantage of this method is that the primary
evaporating refrigerants can be used that do mix with water, for example, ammonia. However,
demands on the non-soluble liquid are high, because it should have higher density than water
and also a much lower freezing point. No sample fluids are mentioned, except that they are
organic compounds. Prior to the Japanese publication a similar system was worked on by the
Chicago Bridge & Iron Company in the USA.
The schematic of a typical direct contact ice slurry generator is shown in Figure 5.13. The
generator contains an extra cycle in which a heavy, non-miscible liquid provides the heat
exchange between the ice slurry loop and the primary refrigeration cycle. The liquid is cooled
by the evaporator of the primary cycle and is mixed with an aqueous solution in the ejector.
Because the temperature of the liquid is below the freezing temperature of the aqueous
solution, the formation of ice crystals occurs. In the freeze tank, ice crystals rise upward and
the heavy liquid sinks to the bottom from where it returns to the pump and the evaporator.
Water / Ice slurry

Ice slurry outlet
Water inlet
Ejector
Primary
refrigerant loop
Heavy, non-
Pump miscible liquid
Figure 5.13. Direct contact liquid coolant type ice slurry generator
142
It is difficult to find a suitable liquid for this ice slurry generator, because the requirements for
this heavy liquid are strict. The liquid must have a higher density than water and must be non-
miscible with it. Furthermore, the fluid must have a lower freezing point compared with
water. Another disadvantage of the direct contact liquid coolant type ice slurry generator is
the extra cycle, whereby both the investment costs and the energy consumption are relatively
high.
This type of ice slurry generator has been globally described by Rivet et al. (1998) and by Ure
(1997). Recently, a consortium of three companies introduced a system based on this
technology and was awarded the ASERCOM Energy Efficiency Award in 2003 (HK 2003).
Another system was investigated in Japan. Fragile ice uses the injection of cold air into a
water layer to create ice slurry (Fukusako et al., 1999; Inaba, 2001).
5.2.5 SUPERCOOLED BRINE METHOD

by Tanino Masayuki, Mito D., Kozawa Yoshiyuki
Ice generator systems using the supercooled water method have been developed and
introduced into air-conditioning systems by several companies in Japan (Tanino et al., 1997,
2000, 2001 a, b, c; Kozawa and Tanino, 1999; Kozawa et al., 2001; Mito et al., 2001, 2002;
Nagato, 2001; Kurihara and Kawashima, 2001; Kema et al., 1997; Bedecarrats et al., 2000).
These systems are shown in Figure 5.14 (A to G). The differences in these systems are mainly
the type of supercooler (heat exchanger for producing supercooled water) and the method of
supercooling release.
System (A) (Tanino et al., 1997, 2000, 2001 a, b, c; Kozawa and Tanino, 1999; Kozawa et al.,
2001; Mito et al., 2001, 2002) of Figure 5.14 consists mainly of a refrigerator (conventional
chiller, which is omitted in the figure), an ice slurry storage tank, a supercooler, a releaser
(releasing the supercooling of water), a heat exchanger, and pumps.
The water in the ice slurry storage tank is sent to the supercooler where it is supercooled to
–2°C. This supercooled water runs against the inner wall of the pipe in the releaser and
undergoes a phase change to become ice slurry containing very small ice particles. The ice
content within the ice slurry is 2.5 wt%. The ice slurry is fed through a piping system to the
ice slurry storage tank where it is separated by the difference in density between ice and
water.
The water sent to the supercooler is preheated to 0.5°C in order to melt the small particles of
ice completely prior to entering the supercooler. This measure ensures stable production of
supercooled water. Even the smallest ice crystal or other impurity may act as a seed for ice
crystal growth and would prevent supercooling of water (see chapter 2.1). Moreover, the other
systems, System (B) and System (C) in Figure 5.14 (Nagato, 2001), have a filter for small ice
particles for stable production of supercooled water, which is the key to the industrial
application of this system (Bedecarrats et al., 2000).
Figure 5.15 depicts the water circulation cycle and the production and storage process of ice
slurry. In this water circulation cycle, sherbet-like ice is stored in the ice storage tank after the
release of the supercooled state (production of ice slurry) and separation between ice and
water.
143
System Construction/Company/Characteristics System Construction/Company/Characteristics
(A) (A’)
(Tanino (Tanino
et al., et al.,
1997, 2000;
2001 a, Mito et
b, c;
Kozawa
al.,
et al., 2002)
1999;
2001, Takasago Thermal Engineering Co., Ltd. Takasago Thermal Engineering Co., Ltd.
Mito et Supercooler Shell & tube type Supercooler Shell & tube type
al., (Supercooled heat exchanger) (Supercooled heat exchanger) / Plate type
2001) Releaser Falling to pipe Releaser Ultrasonic waves
(Method of releasing supercooling) (Method of releasing supercooling)
(B) (C)
(Nagato,
2001)
Nippon Steel Corporation Shinryo Corporation

Supercooler Shell & tube type Supercooler Plate-type
Supercooled heat exchanger (Supercooled heat exchanger)
Releaser Falling to tank Releaser Mechanical Vibration
Method of releasing supercooling (Method of releasing supercooling)
(D) (E)
(Kuri-
hara et
al.,
2001) Shimizu Corporation Mitsubishi Heavy Industries, Ltd.
Supercooler Plate-type Supercooler Shell & tube-type
(Supercooled heat exchanger) (Supercooled heat exchanger)
Releaser Additional cooling Releaser Falling to tank
(F) (G)
(Kema
et al., Daikin Industries, Ltd. Toshiba Corporation
1997) Supercooler Shell & tube-type Supercooler Plate-type
(Supercooled heat exchanger) (Supercooled heat exchanger)
Releaser Additional cooling Releaser Falling to tank
(A),(B),(C),(D): large-scale on-site-type system. (A’): Newer version of system (A).
Figure 5.14. Current and possible systems of supercooled brine method in Japan
144
This supercooled method uses solely pure water for the production of ice slurry. It is suitable
for large-scale on-site-type ice storage systems for air conditioning applications. Figure 5.16
shows the total number and the total capacity of the actual equipment installed in Japan for
type (A), (B) and (C) systems shown in Figure 5.14. The total number of (A) type systems
installed is 24 sets and the total refrigeration capacity of the system (A) type is 35 MW or
more. As shown in the figure, it can be estimated that many types of equipment are going to
be introduced into actual air-conditioning systems and cooling systems for the peak power
shift and/or the peak power cut in Japan.
Supercooled Water
Phase change
Cooling
(Production of ice slurry)
Water Water & Ice Ice
Separate Store
Figure 5.15. Ice slurry production process using the supercooled water method
in an ice storage system
12,000 30
Capacity of System
10,000 Capacity of System 25
Capacity of System Nu
Total Capacity USRt
Number of System mb
8,000 Number of System 20 er
Number of System of
6,000 15 Sy
ste
m
4,000 10
2,000 5
0 0
'88 '89 '90 '91 '92 '93 '94 '95 '96 '97 '98 '99 '00 '01 '02
Year
Figure 5.16. Actual numbers of equipment and installed refrigeration capacity of ice storage
systems using the supercooled method in Japan
(System (A), system (B) and system (C) of Figure 5.14)
145
5.2.6 HYDRO-SCRAPED ICE SLURRY GENERATOR

by M. Barth and M. Leprieur
Ice slurry, especially that made from carrier fluids with high concentrations of freezing point
depressing additives, can be produced in conventional heat exchangers, if the evaporation
temperature of the primary refrigerant and the flow rate of the ice slurry are carefully
controlled. In the Hydro-Scraped Ice Slurry Generators, ice particles are flushed away by the
fluid flow itself, hence the name “hydro-scraped ice slurry generator”. A prototype has been
developed at the Technical Centre of Dinan (Pole Cristal) in France.
5.2.6.1 Principle
A schematic of the hydro-scraped ice slurry generator is shown in Figure 5.17. It consists of a
refrigeration loop and a secondary refrigerant or ice slurry flow loop.
Figure 5.17: Schematic of a hydro-scraped ice slurry generator
The standard refrigeration loop consists of a compressor (20), a condenser (22), an evaporator
(1) and an expansion valve (25).
In the suction line (21), the evaporating pressure is regulated by a valve (23). In addition, the
refrigeration circuit consists of a hot gas line (26) and another regulation valve (27).
The ice slurry flow loop consists of an evaporator (1), a pump (8), a variable speed motor
(13), a storage tank (12), pressure and temperature gauges (10 and 11) and a control box (14).
The distribution of the ice slurry to the point of use is through a pump (8b), distribution line
(5), return line (7) and heat exchangers for melting the ice slurry (6).
5.2.6.2 Practical system

Using a standard heat exchanger (shell-and-tube evaporator or plate heat exchanger), the
hydro-scraped ice slurry generator operates according to the following procedure.
146
The pressure and temperature measurements in the ice-slurry flow loop are used to determine
the formation of ice crystals in the ice slurry generator (evaporator (1). When the pressure
difference between the inlet and outlet of the refrigerant stream increases and the temperature
difference decreases, it indicates that ice crystals have started to form in the ice slurry
generator.
The electronic controller then initiates two actions which have three consequences (C):
Action 1 Increase very quickly the flow rate of ice slurry

C1 – Increasing the ice slurry flow rate generates turbulence which causes the ice
crystals to detach from the evaporator surface.
C2 – The velocity increase due to the flow rate increase modifies the heat transfer
helping to remove ice crystals in formation.
Action 2 Decrease at the same time the refrigeration capacity

C3 – The decreasing refrigeration capacity boosts consequence C2.
As result of these actions, the ice crystals formed on the evaporator surface detach from the
surface and are removed in the secondary refrigerant flow forming ice slurry. The pressure
difference in the refrigerant loop decreases, the temperature difference increases and the flow
rate returns to nominal conditions.
5.2.7 SPECIAL COATING OF GENERATOR SURFACE TO AVOID ICE STICKING TO SURFACE

by Thomas Zwieg, Viktor Cucarella, Hartmut Worch, Michael Kauffeld
In order to simplify the very careful operation of the hydro-scraped ice slurry generator
described in section 5.2.6, the heat exchanger surface on the ice slurry side can be coated with
special substances preventing the growth of ice crystals on the surface.
5.2.7.1 Principle
On the typical metallic heat exchanger surfaces, the crystallized ice sticks with strong
adhesion forces. Depending on the temperature gradients just above the cooled surface in the
flowing aqueous solution, a flat and compact ice layer is created. If not removed by, for
example, mechanical scraping devices, this ice layer may increase in thickness with time and
result in a decreasing flow rate and finally a flow blockage.
Existing developments based on the fluorinated organic coatings cannot overcome this
problem. The typical water repelling properties of fluorinated organic surfaces at room
temperature are generally not transferable to ice repelling properties in an actual application.
Fluorinated coatings such as commercial types of PTFE, FEP but also fluorinated
alkoxsilanes from different sol gel coating systems have been tested but were not convincing
in their icephobicity. In the tests, these fluorinated coatings showed the highest
hydrophobicity at contact angles of approximately 95 up to 115 degrees in water and
water/freezing-depressant solutions. But this higher hydrophobicity had no visible influence
on the surface-icing behaviour. Based on these tests, it can be concluded that hydrophobic
surfaces are not consequently also icephobic. Obviously, the distinct hydrophobicity of
fluorinated coating systems is no longer effective when ice crystals with their strong polarity
and directed dipole moments are formed at the surface. A dipole seems to be induced in the
147
highly electronegative fluorine resulting in secondary forces and sticking of the ice.
Furthermore, mechanical interlocking of the ice crystals formed at the PTFE surface occurs
because of the well-known porosity of the sintered PTFE surface.
Though existing silicone based lacquer type coatings show good ice repelling properties, the
insufficient mechanical stability and very low thermal conductivity of these coatings makes
their use in practical applications impossible. The relatively abrasive ice particles destroy
these coatings after a short time. Also, the dramatically decreased ice nucleation rate is a
further highly negative factor.
According to the disclosure in International Patent Application WO 00/06958, a sol gel
technology has been used for the production of a corrosion resistant hydrophobic coating
which also appears to prevent ice formation on the surface of an evaporator. This condition is,
however, only valid for surface temperatures that are close to the freezing point. If the
temperature of the surface is substantially below the freezing point, ice will form on the
surface despite the hydrophobic conditions.
The ice generating characteristic is achieved by a heterogeneous surface having nucleation
sites which cause precipitation of ice in a fluid at these nucleation sites under suitable
physical conditions. The heterogeneous surface can be configured such that the Gibbs free
energy at the nucleation sites, when the surface is in contact with the fluid and the ice, is
lower at the nucleation sites than between the nucleation sites (see also Chapter 2.1 of this
handbook).
This new type of ice slurry generator without mechanically moving parts mimics a peculiar
mechanism used by some organisms adapted to cold regions that can survive in extremely
cold climates. These organisms show an ice nucleating activity (INA) through a specific
membrane protein (INP) in order to slow down the freezing rate and thus survive extremely
cold conditions (Figure 5.18). They protect their cell tissue by controlled ice generation.
Novel ice nucleating coatings (INC) have been designed based on the cell membrane of INA
organism. Chemical compounds with similar estimated size and properties to that of the key
residues of INPs were used to achieve a coating network able to nucleate and release ice
(Figure 5.19). In addition, the newly developed INCs could be further improved. The tests
using several successful INCs showed a unique combination of high ice nucleation rate and
ice repelling results.
5.2.7.2 Practical system

The sol-gel technology was applied to produce the ice nucleation coatings (INCs) (Figure
5.19). Nanoparticles with both hydrophobic and hydrophilic properties were synthesised from
organic modified silicone alkoxides. These nanoparticles were thermally cross-linked to the
pre-treated surface of the aluminium half pipes. Smooth and very thin nanostructured hybrid
layers were created at the metal surface, formed by an inorganic-organic network. Due to the
selection of the reactive functional groups of the organosilane, the nanostructured layers
contain both the hydrophobic anti-adhesive sections as well as hydrophilic segments. These
very small hydrophilic segments are considered to act as local separated nucleation sites. As
is known from the INA organism, and resembling a chessboard with its black and white
fields, the local ice nucleation sites are surrounded by anti adhesive ice repelling segments or
areas.
The actual ice slurry generator made with the special INCs is basically a shell-and-tube type
heat exchanger where the evaporating refrigerant is on the shell side and the ice slurry is
formed in the coated tubes of the heat exchanger. The Danish Technological Institute together
148
with a Danish refrigeration company has built a first prototype with 10 kW refrigeration
capacity working with ammonia in the flooded shell-and-tube evaporator.
Figure 5.18. Representative scheme of the INP Figure 5.19. Aluminium oxide
arrangement at the cell membrane of INA- modified by applying the sol-gel
bacteria. Hydrophilic residues act as nucleating technology. Red colour (wavy line
sites, matching an ice-like structure. in the centre) represents hydrophilic
Neighbouring hydrophobic residues surround the residues and blue colour (two outer
nucleation site, preventing ice sticking on the wavy lines on either side of the red
cell membrane curve) represents hydrophobic
residues

So far all the development work has concentrated on shell-and-tube heat exchangers. The
capacity and size follow those known from conventional shell-and-tube evaporators, although
the ice slurry flow pattern might have to be modified in order to ensure the minimum flow
speed of 1 m/s in all passes of the heat exchanger.
The same surface coating technique should also be applicable to plate heat exchangers, but
trials have not been performed yet.
5.2.7.4 Costs and energy consumption (performance characteristics)

The cost of the coating is the only addition to the price of the shell-and-tube or plate type heat
exchanger for ice slurry production. The entire refrigeration system is standard. The cost for
the coating material is currently 2 euros/m2 for the prototype production and is expected to
fall to 0.2 euro/m2 for full scale production.
The coating does not decrease heat conduction through the heat exchanger walls due to its
extreme thinness of less than 10 µm. Hence energy consumption is expected to be the same as
for conventional chillers operating at the same temperature range.
5.2.7.5 Economics, operating costs and reliability

Initial results have been very promising and the ice slurry generated is very similar to the ice
slurry generated in a generator with moving blades/scrapers, except that there is no need for
moving parts in the specially coated ice slurry generator. Therefore, this type of ice slurry
generator can be made much cheaper, both in terms of initial costs and in operating costs.
149
The lifetime of the developed INCs is an important factor with regard to reliability in practical
applications. The ice nucleation process and the ice crystals themselves hardly affect the
surface stability of the coatings. Nonetheless, a time-dependent ageing process was observed
in several coatings tested, resulting in reduced ability to release the ice formed. This effect
was evident especially after strong freezing of the surfaces due to very fast supercooling
(Zwieg et al., 2002). The coating properties of the INCs were improved by increasing the
density of the inorganic network and selecting other functional groups of nanoparticles.
Furthermore, the content of inorganic compounds was varied in the hybrid network to achieve
the necessary mechanical and chemical stability of the coating. In addition organic bridging
polymers were anchored within the coating network. Due to these improvements, the
properties of the INCs have been visibly improved and a life time of 15 years should be
possible to achieve, although tests of this long duration are still lacking. Special corrosion
tests designed to give indications of life span of surfaces were successfully passed.
5.2.7.6 Applications
So far this type of ice slurry generator only exists at the laboratory scale, but a German
refrigeration equipment manufacturer work is conducting ongoing work in order to bring this
very promising technology to a commercial application (Behnert and König, 2002).
5.2.7.7 Advantages
The new biomimetric Ice Nucleating Coatings (INC’s) are able to nucleate ice properly and
release small crystals (0.1-2 mm) under flow conditions, see Figure 5.20. Two different ice
formation mechanisms exist for the INC’s investigated so far. The first consists of local
nucleation and local growth of ice at specific nucleation points (NPs). When the ice crystals
reach a certain size, they are released by the secondary fluid flow (INC-014). Ice crystals of a
few millimetres are formed and up to 5% of the cooled surface area is covered with ice at the
maximum stage of the ice production. In the second mechanism, ice nucleates locally at
specific NPs producing very small ice crystals (INC-021). However, the small ice crystals
grow further by moving above the cooled surface. This mechanism is also named the
“snowball” effect (Zwieg et al., 2002). This mechanism shows a clear efficiency
improvement in the rate of ice production. Ice crystals cover up to 10% of the cooled surface
area. Further approaches based on this effect led to multi-local ice nucleation (INC-023) by
increasing the amount of hydrophilic nanoparticles in the coating. This appeared to increase
the ice production rate, and the results have shown ice formation on up to 40% of the cooled
surface area (see Figure 5.20C).
For a single nucleation point (NP) a strong dependence on flow speed can be noticed. All
tested INCs, except for INC-024, showed similar rates of ice nucleation. At the working flow
speed of 1.3 m/s, the NPs produce ice crystals approximately every five seconds. Due to
further improvements in the coating composition and processing parameters, the nucleation
rate of the NPs was increased, i.e. the modified INC-024 clearly increased the ice nucleation
rate, and some specific points were found to produce ice continuously (see Figure 5.20D).
Working temperatures and blending composition had slight effects on the results of the
different coatings. All tested INCs seemed to operate considerably better when using
propylene-glycol aqueous solutions, which freeze at –5°C or lower temperatures. Some of
them also worked well with less concentrated solutions that freeze at –2.5°C.
150
Figure 5.20. Sketches of the ice formation process over different coating materials
The images were captured from a 7.7 × 5.8 mm selected area on the coated surface every
second. Circles delimit the nucleation sites and arrows follow the route of the crystals. Four
different coatings are presented A: INC-014, B: INC-021, C: INC-023, D: INC-024
5.2.7.8 Limitations
Until now certain requirements regarding the concentration of the freezing point depressing
agent have to be met in order for this ice slurry generator to work properly. Future work will
enable the ice slurry generator to work over a wider range of operating conditions and at
higher temperatures.
A minimum flow velocity is required on the ice slurry side, i.e. above 1 m/s. Below this speed
icing of the coated surface may occur depending on the type of additive used.
5.2.7.9 Working fluids, type of additive, ice fraction range

Best results have so far been obtained with propylene glycol and ethanol. As in the case of the
hydro-scraped ice slurry generator (see section 5.2.6) the coated ice slurry generator works
better as the additive concentration increases. So far propylene glycol solutions with a
freezing point below –2.5°C can be used safely as well as ethanol solutions with a freezing
point below –5°C.
Salt-based brines for ice slurry production tend to exhibit higher adhesion forces, and are
consequently less suitable for this type of ice slurry generator, although this might change as
development of the coating material matures.
The addition of small nuclei in the ice slurry flow increases the operating range as well as the
reliability of this type of ice slurry generator. Ice particles formed elsewhere in the system or
fed back from the exit of the ice slurry generator to the entrance reduce the supercooling of
the water and reduce the risk of freeze-up.
151
5.2.8 FLUIDIZED BED CRYSTALLIZER

by Jeroen Meewisse
Fluidized bed heat exchangers have been known for some years and are nowadays used
mainly for heat exchange with severely fouling process fluids (Klaren, 2001). Earlier,
fluidized bed systems were used for water purification by freezing, an application similar to
ice slurry generation. The fluidized bed system was first suggested as an ice slurry production
method by Klaren (1991).
5.2.8.1 Principle
Fluidized bed heat exchangers are shell-and-tube or tube-
in-tube type heat exchangers. On the shell side, a
primary refrigerant is evaporated, for example, ammonia,
a hydrocarbon or a chemical refrigerant. Ice is formed on
or near the inside surface of the tubes mounted in the
shell of the heat exchanger. Inside the tubes a fluidized
bed is contained, consisting of small particles made of
steel or glass, with a diameter of 1 to 5 mm. The particle
beds are fluidized by the upward flowing liquid phase,
which is the ice slurry feed flow. When fluidized, the
solid particles continuously impact on the inside walls of Figure 5.21. Particle/Ice
the tubes. Build-up of an ice layer on the heat Removal Mechanism
exchanging surface is prevented in this way, as
is displayed in Figure 5.21. Fluidized particles
also continuously disturb the heat exchanging
boundary layer. The thickness of this layer
therefore becomes small and heat transfer rates
are enhanced.
Fluidized beds in heat exchangers with

multiple tubes can be fluidized simultaneously
by installing flow distribution devices at the
inlets of the tubes. These parallel fluidized beds
can then be operated by a single pump.
Multiple fluidized beds can be arranged easily
in a shell-and-tube configuration, which allows
for significant benefits of larger scale. These
benefits can also be obtained by increasing the
height of the tubes. In fluidized bed heat
exchangers for other than cooling purposes,
column heights of up to 10 metres have been
applied successfully. Figure 5.22. Fluidized bed heat
A fluidized bed heat exchanger can be operated with or without circulation of particles,
depending on the superficial velocity used. In a circulation mode, particles move continuously
out of the top of the beds and need to be recycled to the bottom of the heat exchanger to re-enter
the fluidized bed. For this operation mode, the heat exchanger is equipped with a zone where
the particles can be separated from the ice slurry flow. Also a downcomer tube is required,
where the particles are stored in the form of a packed bed that moves slowly downward.
152
In Figure 5.22, a schematic of a fluidized bed heat exchanger is shown in which there are
several parallel fluidized beds. Particles are circulated through a downcomer tube on the outside
of the shell. For different types of particles, especially small or light particles, a separation
device like a cyclone might be installed to separate the particles from the ice slurry.
5.2.8.2 Practical machines

No commercial installations have yet been realized for ice slurry applications. At the
laboratory scale test set-ups are operating satisfactorily. Controlling the solid particles in ice
slurry flows and temperature differences allowed in the system is essential for a continuous
and stable operation. These factors need to be explored further. The fluidized bed heat
exchange technique is well known from various applications outside the ice slurries and is
used extensively (Klaren, 2000).
5.2.8.3 Energy requirement

Wall to bed heat transfer rates in the fluidized bed ice slurry process are relatively high (2500-
3500 W/m2K) because of three effects: the laminar boundary layer, which creates heat
transfer resistance, is reduced in size by the particles; the particles also transfer heat by
conduction during contact; and finally the heat exchanger walls are kept free of ice by particle
collision (Meewisse, 2001).
The temperature difference between the primary refrigerant and ice slurry must be low;
therefore energy consumption will be low as well. Additional power is only required for the
ice slurry circulation pump, that also supplies the energy for fluidization. With a careful
selection of process conditions, pumping power requirement can be kept reasonable, 1 to 5%
of the capacity of the ice slurry generator.

Fluidized bed systems can be built small and capacities start from a few kilowatts. Benefits of
scaling however will make systems more attractive in relatively large installations. The
fluidized beds are essentially vertical, therefore these systems will not require a lot of floor
space, but larger installations cannot be installed in rooms with low ceilings. Ice production in
a single pass through a fluidized bed is limited to a few % increase in ice mass fraction; so an
ice storage vessel is therefore almost always required.
5.2.8.5 Economics
Investments are low because there is no need for complicated mechanical moving parts
(Meewisse, 1999). However, installations at relatively high temperatures require a large heat
exchanging area, as there is a maximum allowable temperature difference between the
primary refrigerant and ice slurry. Additional equipment consists of a pump, which can be any
type capable of pumping ice slurries, and a basic storage vessel. This storage vessel can also
be used to gain economic benefits by load shifting and peak load shaving.
Operating and maintenance costs are low, as there are no mechanical parts apart from the
pump. The fluidized bed particles are inexpensive and basically all types of freezing point
depressants can be used in the system, as long as the hardware materials used are corrosion
resistant to the substances used.
5.2.8.6 Limitations
One limit already mentioned is the maximum allowable temperature difference between the
primary refrigerant and ice slurry. At high temperature differences, the fluidization particles
cannot prevent freezing of heat exchanger walls. At low freezing point depressant
153
concentrations, which means at higher temperature ice slurries, the maximum allowable
temperature difference is low (Meewisse, 2001).
Other limits of the process are associated with the fluidization of the particles. The bed
porosity cannot be too high, because then the particles will not fully cover the entire heat
exchanging area. Also the fluidization velocity should be high enough; otherwise the particles
will not impact the heat exchanging walls hard enough to prevent build-up of an ice layer.
Due to this minimum fluid velocity, the residence time of ice slurry in the fluidized bed
system is limited. The ice fraction achievable upon a single pass is therefore limited; an
increase in the ice fraction of 2-3% is expected for a single pass in a medium sized
installation.
Another limitation is the inherent vertical lay-out of the fluidized beds. Limited vertical space
in an application may prevent installation of the ice slurry generator with optimal dimensions.
Applications to on-board systems in trawlers is doubtful.
5.2.8.7 Reliability
Long term reliability has not been tested yet for this fluidized bed ice slurry generation
technology. Other fluidized bed heat exchanging systems have been operating for many years
(Rautenbach, 1996; Klaren, 2000). Wear of components by abrasion of particles has been
noted, however, the rates are sufficiently low so that the components can operate for several
years without any maintenance.
5.2.8.8 Working fluids

Any additive in the ice slurry can be used as a freezing point depressant, as long as the
construction material is resistant to this additive. A fluidized bed heat exchanger can, for
example, be made of stainless steel, which may give problems if NaCl based ice slurries are
used. Different types of steel or even copper can be used, or corrosion inhibitors can be added
to the ice slurry.
Production of ice slurries with high ice fractions is possible (up to 30%), since heat transfer
rates during the production are relatively unaffected by the higher ice fractions. Multiple
passes through the generator or ice concentrators will, however, be required to obtain higher
ice fractions.
5.2.9 HIGH-PRESSURE ICE SLURRY GENERATOR

By Pedro D. Sanz
Practically all “traditional” freezing processes rely on the thermal conductivity of the sample
and its interface to dissipate its heat (to reduce temperature first, and then to induce the phase
change). This has the effect of originating a thermal gradient that gives rise to a supercooling
gradient, with a higher degree of supercooling in the region closer to the heat exchange
surface. The result is that ice nucleation takes place only in this area close to the interface
with the heat-dissipating medium. Other regions are unable to reach a satisfactory degree of
supercooling, because, once freezing has started, the latent heat released during phase change
increases the temperature of the surrounding area, thus reducing the degree of supercooling.
Reducing the thickness of the sample or increasing the heat exchange rate in the interface area
can attenuate this effect. Additionally, in samples with a heterogeneous composition (even
microscopically heterogeneous), ice will originate only in the micro-regions with a higher
nucleation temperature, which will depend on their freezing point temperature and a number
154
of factors related to ice nucleation. In these cases, ice nuclei will occur in regions as a
function of local composition and of supercooling degree, which will smoothly decrease away
from the dissipating interface (Sanz et al., 2001). These few originally formed nuclei grow,
without competition, to reach a large crystal size. Consequently, in aqueous solutions ice
grows very often in a dendrite or needle form from regions at the interface (or the easily-
freezing heterogeneities) towards the centre, to include the whole sample volume.
Thanks to general scientific and technological advances, some promising technologies are
being revisited. This is the case with the high-pressure processes for different applications,
which include ceramics, semiconductors or ice-water phase transitions. The application of
high-pressures to ice formation in ice slurry generators requires a number of technical
problems to be solved. Among them is the high economic cost of high-pressure equipment,
when compared with other ice slurry generating systems. Also the current concept of high-
pressure equipment is scarcely compatible with continuous processes.
Two kinds of high-pressure freezing processes can be distinguished, although they are
frequently confused in the literature (Otero and Sanz, 2000). In both, once the operating
pressure has been achieved, no more mechanical but only refrigeration energy consumption is
required.
5.2.9.1 High-pressure assisted freezing

In this process, phase transition is induced under constant high-pressure, by lowering the
temperature to the corresponding freezing point. In this way, ice (or other known ice
polymorphs, depending on pressure and temperature conditions) can be obtained. Cooling of
the sample proceeds from the surface to the centre. The freezing process is governed by
thermal gradients. Ice nucleation only occurs in the outer zone of the sample being frozen and
needle-shaped ice crystals grow radially towards the centre.
5.2.9.2 High-pressure shift freezing (HPSF)

Phase transition occurs due to a pressure change that promotes metastable conditions and
instantaneous ice production. The pressure can be released slowly, over several minutes, or
quickly, in 1-2 seconds. In each case, ice nucleation is produced through different levels of
supercooling reached at different pressures, and hence freezing kinetics are different.
Table 5.2 gives the theoretically obtained percentage of ice (Chizhov and Nagornov, 1991)
for some adiabatic and isentropic expansions from the water melting curve. Initially 100% of
liquid water is considered. From Table 5.2, 35.2% of ice could be obtained by considering
210 MPa as the initial pressure when the expansion is carried out along the melting curve as is
in all cases represented in Table 5.2.
155
Table 5.2. Percentage of ice instantaneously produced in different adiabatic

and isentropic HPSF processes, obtained by mathematical modelling
Initial Initial T Theoretical

Pressure (ºC) instantaneous ice after
(MPa) expansion (%)
210 –22.00 35.2
200 –20.70 32.7
190 –19.40 30.3
180 –18.12 27.9
170 –16.86 25.6
160 –15.62 23.5
150 –14.41 21.4
140 –13.23 19.4
130 –12.08 17.5
120 –10.96 15.7
110 –9.87 14.0
100 –8.81 12.4
90 –7.79 10.8
80 –6.80 9.3
70 –5.84 7.9
60 –4.92 6.6
50 –4.02 5.3
40 –3.16 4.1
30 –2.33 3.0
20 –1.53 1.9
10 –0.75 0.9
It is not possible to follow the melting curve experimentally. This is due to the supercooling
effect appearing after the expansion as represented at the bottom of Figure 5.23.
Table 5.3 shows the instantaneous ice percentage obtained from quasi-adiabatic expansions
by using a heat balance calculation (Otero and Sanz, 2000). In this calculation the metastable
conditions of the liquid water have been considered, as well as the average value of the
specific heat corresponding to the liquid water and to the ice conditions, the latent heat, etc.
Pressure release takes place practically instantaneously upon expansion throughout the entire
sample volume (after the Pascal principle) and, subsequently, a decrease in its temperature
value associated with the pressure reduction is produced. A high degree of supercooling
occurs throughout the entire sample, implying high ice nucleation velocities. It has been
reported (Burke et al., 1975) that for each degree of supercooling, the ice-nucleation rate
increased about tenfold. Small ice crystals of granular shape dispersed throughout the sample
have been found by different authors experimenting with different samples frozen by high-
pressure shift, proving the homogeneous (not heterogeneous) character of ice nucleation.
Nevertheless, ice crystal diameters grow from the surface to the centre of the sample due to
the thermal gradients that are established at atmospheric pressure after expansion, in which
conditions most of the freezing process takes place.
156
Figure 5.23. High-pressure shift freezing of an agar gel cylinder from 210 MPa and –21ºC
Figure 5.23 shows temperature and pressure data for an HPSF experiment from a start-point
of –21ºC/210 MPa, on a sample of agar gel. At the bottom a water phase diagram has been
superimposed. The upper part of the figure shows the complete process, while the lower part
is a detail of the pre-cooling and expansion phases. The experiment starts with sample
compression up to 210 MPa (associated with a temperature increase). Subsequently, it is pre-
cooled under pressure until a temperature near the freezing point at this pressure is reached,
while the sample remains always in the liquid state. Immediately after a rapid expansion (1-2
seconds) the sample temperature decreases due to pressure release. No phase-change occurs
for a short period (1-2 seconds) and the sample can be considered to be in a high supercooling
degree and to be metastable in a liquid state. This supercooling gives rise to uniform
nucleation throughout the sample, with the corresponding latent heat release. This latent heat,
suddenly released, cannot be removed quickly enough by the cooling system and is therefore
absorbed by the sample, which increases its temperature up to the equilibrium phase-change
temperature. Therefore, in an HPSF process, when the freezing plateau is initiated at
atmospheric pressure, there is already a percentage of ice uniformly distributed. The amount
depends on the pressure and temperature values before the expansion.
157
Table 5.3. Approach to the experimental case. Supercooling and percentage of ice
instantaneously obtained by a heat balance calculation
Initial Initial T Theoretical Theoretical

Pressure (ºC) supercooling (ºC) instantaneous ice
(MPa) (%)
210 –22.00 26.36 29.1
200 –20.70 24.68 27.5
190 –19.40 23.03 25.8
180 –18.12 21.41 24.1
170 –16.86 19.81 22.5
160 –15.62 18.25 20.9
150 –14.41 16.73 19.3
140 –13.23 15.25 17.7
130 –12.08 13.82 16.2
120 –10.96 12.43 14.7
110 –9.87 11.09 13.2
100 –8.81 9.80 11.7
90 –7.79 8.56 10.3
80 –6.80 7.38 9.0
70 –5.84 6.25 7.6
60 –4.92 5.19 6.4
50 –4.02 4.17 5.2
40 –3.16 3.22 4.0
30 –2.33 2.33 2.9
20 –1.53 1.50 1.9
10 –0.75 0.69 0.9
5.2.10 RECUPERATIVE ICE MAKING

by T.W. Davies
5.2.10.1 The evaporator head

Almost all commercially available ice slurry generators are essentially direct-expansion
refrigerators with an evaporator section designed to supercool a secondary fluid, with ice
build up on the chilled surfaces being critically limited by mechanical removal, such as
scraping.
The novel feature of the new recuperative ice slurry system is that ice build-up on the chilled
surfaces is reduced and periodic removal is accomplished by means of recuperative heat
exchange rather than scraping.
De-icing is achieved by using two identical heat exchangers through which the direction of
refrigerant flow can be reversed at optimal frequency. The evaporator head therefore has
twice the heat transfer surface of a scraped surface device of similar capacity but has the great
advantage of operating without any motorized moving parts and with almost free defrost.
The pair of heat exchangers (4a and 4b in Figure 5.24) form the interface between the
refrigeration circuit and the ice making circuit. The heat exchangers can be of any appropriate
design best suited to the type of ice required. In a prototype system recently built and tested
by the author (Davies, 2003a) they were simple vertical copper tubes, with refrigerant on the
inside, immersed in the flow of secondary fluid, but in the latest design they are now
158
commercially available corrugated stainless steel plates (sold as turbo-chillers, see Figure
5.25), with refrigerant on the inside, and wetted on the outside by a falling film of secondary
fluid.
Figure 5.24. Refrigeration circuit for recuperative ice slurry generator
Figure 5.25: A turbo-chiller plate of about 10 kW capacity, 1.5 m wide, 1 m high

made from two sheets of 1 mm thick stainless steel plate welded together,
with refrigerant entry/exit at the top and at the bottom to the right
5.2.10.2 Mode of operation of the refrigeration circuit

Referring to Figure 5.24, the flow of refrigerant from the condenser pack [compressor (1) and
air-cooled condenser (2)] is fed to a reversible four-way valve (3), which is used to direct the
refrigerant stream to either one of the pair of exchangers (4a or 4b). Figure 5.24 shows the
flow being directed by the four-way valve (3) through heat exchanger (4a), which will have
previously been in ice-making mode and so coated with ice. Initially hot compressor exhaust
gas will pass through the air-cooled condenser (2) and actually condense in (4a), so rapidly
releasing the ice. As the temperature in unit (4a) rises towards the ambient air temperature,
condensation begins to occur in the condenser (2) with warm liquid from (2) now passing to
(4a), whose temperature will therefore not rise above the condensing temperature set by the
fan speed controller on (2).
159
Sub-cooled liquid refrigerant leaving the heat exchanger (4a) is then directed by check valves
(5a,b) [which are operated by the differential pressure between the suction and delivery lines]
to an efficiency booster (7) (a liquid accumulator fitted with a heat exchange coil), and then to
an expansion valve (8). The resulting wet vapour is then directed via check valves (9a,b) to
the second heat exchanger (4b), which, acting as an evaporator, now enters the ice-build
mode. Spent vapour from (4b) then exits to the four-way valve (3) and is directed to the
efficiency booster (7) and then to the compressor suction port, (1). Since the temperature
change of the refrigerant in passing from heat exchanger (4a) to heat exchanger (4b) is small
when both plates are continuously wetted by the falling film of chilled secondary fluid, the
amount of flash gas formed at the expansion valve is also small and the cooling capacity of
the evaporator plate therefore remains reasonable. Continuous sub-cooling of the liquid fed to
the expansion valve by chilled secondary fluid is almost energy neutral, since the energy
returned to the circulating fluid in the form of heat is almost matched by the improvement in
the evaporative effect when feeding sub-cooled liquid to the expansion valve.
5.2.10.3 Flow reversal

After an optimal time it becomes necessary to de-ice the heat exchanger (4b) and this is
simply achieved by reversing the refrigerant flow direction using solenoid operated valve (3)
so that the sub-cooler becomes the evaporator and vice versa. During this flow reversal any
liquid in the exchanger (4a) and associated lines is sucked back to the unit (7) where it is
evaporated by the sub-cooling of liquid now entering the coil in (7) from exchanger (4b),
again in an almost energy neutral process. Thus compressor work used in the production of
this unused liquid refrigerant is recovered via enhanced evaporator efficiency brought about
by sub-cooling. The cyclical operation is therefore referred to as recuperative.
5.2.10.4 Ice production

The turbo-chiller plate has been developed to give ultra-high heat transfer efficiencies when
used for chilling water, the corrugated surface creating highly turbulent flow in the falling
liquid film. Most of the slush ice formed on the surface is sloughed off under gravity by the
agitated falling liquid but a more persistent layer of ice builds up on the surface, which needs
periodic de-icing, as described above. Ice precipitation within the falling film may vary with
position and time during the half-period of the operating cycle.
5.2.10.5 Advantages and drawbacks

The cyclic system described above is more efficient than existing cyclic harvester devices
which use hot gas defrost because the metal components will never rise above the condensing
temperature set by the fan speed controller on the air-cooled condenser. This temperature will
be about 10 K above local ambient air temperature. Using hot gas defrost the metal
components may rise to 50°C or above. The unit will also produce slurry ice pseudo-
continuously via a series of transients without the need for the moving parts usually employed
in such devices (scrapers, orbital rods or augers) and a wide range of freezing point depressant
additives can be used. With existing scraped surface devices making slurry ice a change in the
composition of the secondary fluid can cause severe problems with the operation of such
devices. The new system may be operated at higher evaporator temperatures than scraped
surface devices, so less additive is required to produce slurry ice. Additive costs can be an
important component of the total production cost of slurry ice. Freedom of choice of
evaporator temperature and the absence of an expensive scraping mechanism and drive motor
are significant advantages of the system.
The system can be used with equal ease to produce either sheet ice when pure water is used,
or slurry ice when a water/freezing point depressant mixture is used. Only simple adjustments
160
to the evaporator temperature and cycle reversal frequency are needed to effect a change in
the type of ice produced.
Because the heat exchanger system does not involve moving parts it is much cheaper to
construct, operate and maintain than any existing systems. Copper rather than stainless steel
can be the material of construction for non-food applications.
Stabilisation of the refrigerant temperature entering the evaporator section of the heat
exchanger removes one of the difficulties of maintaining cooling capacity at the carefully
controlled levels needed for the generation of liquid ice.
The use of multiple pairs of heat exchangers (modular design) operating in parallel and
independently creates additional degrees of freedom in the management of ice production rate
and liquid refrigerant return to the accumulator during cycle reversals.
The additional evaporator surface adds to the initial cost of the device.
5.2.10.6 Environmental implications

McCloskey et al. (1995), Ure (1996) and Davies (2003b) describe refrigeration plant which
combine potential energy savings and reduction in CO2 emissions with load shifting
capabilities, by using stored pumpable ice integrated with a direct expansion system. The key
to the implementation of these new processes utilising thermal storage is the availability of a
cheap, reliable and efficient ice slurry generator such as that described here.
1. Arkenbout, G.F.: Melt Crystallization Technology, Technomic Publishing Co.,

Lancaster, Pennsylvania USA, 1995.
2. Armstrong, A.J.: Cooling crystallization and flow patterns in scraped-surface
crystallizers, Institution of Chemical Engineers, pp. 685-687, 1979.
3. Bedecarrats, J.P.; Strub, F.; Dumas, J.P.; Roset, G.: Supercooled Ice Slurry Production,
First results from a test plant, Second IIR Workshop on Ice Slurries, Paris, France, pp.
110-117, 2000.
4. Behnert, T.; König, H.: Erzeugung von Ice Slurry in einem Eisgenerator ohne
bewegliche Teile. KK 55, 9, 2002.
5. Bellas, J.; Chaer, I.; Tassou, S.A. : Heat transfer and pressure drop of ice slurries in plate
heat exchangers, Applied Thermal Engineering Science, vol. 22, pp. 49-55, 2002.
6. Briley, G.C.: Scraped surface heat exchangers, ASHRAE Journal, vol. 44 (5), p. 52,
2002.
7. Burke, M.J.; George, M.F.; Bryant, R.G.: In Water Relations of Foods, Food Science
and Technology Monographs, Duckworth RB, Eds.; Academic Press; pp. 111-135,
1975.
8. Byrd, L.W.; Mulligan, J.C.: A Population Balance Approach to Direct-Contact
Secondary Refrigerant Freezing, AIChE Journal, Vol. 32, no. 11, pp. 1881-1888, Nov.
1986.
9. Chizhov, V.E.; Nagornov, O.V.: Thermodynamic properties of ice, water and their
mixture under high pressure. Glaciers Ocean Atmosphere Interactions (Proceedings of
the International Symposium held at St. Petersburg, September 1990). IAHS Publ. nº
208: pp. 463-470, 1991.
10. Chuard, M.; Fortuin, J.P.: Coldeco - A New Technology System for Production and
Storage of Ice. First IIR Workshop on Ice Slurries, Yverdon-les-Bains Switzerland,
161
published by IIR, pp. 139-145, May 1999.

11. Davies, T.W.: An efficient and environmentally friendly refrigeration system.
Eurotherm Seminar No 72, Thermodynamics, Heat and Mass Transfer of Refrigeration
Machines and Heat Pumps, ISBN 84-931209-80-7, pp. 307-312, 2003b.
12. Davies, T.W.: An efficient ice slurry generator suitable for thermal storage
applications. Eurotherm Seminar No 72, Thermodynamics, Heat and Mass
Transfer of Refrigeration Machines and Heat Pumps, ISBN 84-931209-80-7,
pp. 241-245, 2003a.
13. Fukusako, S.; Kozawa, Y.; Yamada, M.; Tanino, M.: Research and Development
Activities on Ice Slurries in Japan. Second IIR Workshop on Ice Slurries, Yverdon-les-
Bains, Switzerland, pp. 83-105, May 1999.
14. Gladis, S.P.; Marciniak, M.J.; Joseph, B.; O’Hanlon, J.E.; Brad Yundt, P.E.: Ice crystal
slurry TES system using the orbital rod evaporator, Paul Mueller Company technical
paper, 1999: http://www.muel.com/products/thermalstorage/docs/ore/default.htm.
15. Hansen, T.; Kauffeld, M.; Sari, O.; Egolf, P.W.; Pasche, F.: Research, development and
applications of ice slurry in Europe, Fourth IIR Workshop on Ice Slurries, Osaka Japan,
Nov. 2001.
16. Hartel, R.W.: Ice crystallization during the manufacture of ice cream, Treads in Food
Science and Technology, vol. 7, pp. 315-321, 1996.
17. HK 2003: Heizung, Klima, Gebäudetechnik: Der ASERCOM Energy Efficiency Award.
HK 12, pp. 54-56, 2003.
18. Inaba, H.: Fundamental Research and Development of Ice Slurry for its Cooling System
Design in Japan, Fourth IIR Workshop on Ice Slurries, Osaka Japan, Nov. 2001.
19. Inada, T.; Lu, S.; Grandum, S.; Yabe, A.; Zhang, X.: Microscale analysis of effective
additives for inhibiting recrystallization in ice slurries, Second IIR Workshop on Ice
Slurries, Paris, p. 84, 2000.
20. Kauffeld, M.; Christensen, K.G.; Lund, S.; Hansen, T.M.: Experience with ice slurry,
First IIR Workshop on Ice Slurries, Yverdon-les-Bains, Switzerland, p. 42, 1999.
21. Kema, D.; Kondou, I.; Shigenaga, Y.; Torikoshi, K.; Katayama, Y.: Development of a
Dynamic-type Ice Thermal Storage Air Conditioning System of Individual Control for
Building Use, Proc. of 7th International committee on thermal energy storage
MEGASTOCK’97, Sapporo, Japan, p. 739-744, 1997.
22. Klaren, D.G., in: Handbook of heat exchanger fouling : mitigation and cleaning
technologies. ed. Muller-Steinhagen, H., Rugby: Institution of Chemical Engineers,
2000.
23. Klaren, D.G.; van der Meer, J.S.: A fluidized bed Chiller: A new approach in making
slush ice, Industrial Energy Technology Conference Houston Proc., 1991.
24. Knodel, B.D.; Ludwigsen, J.S.; Ludwigsen, J.L.; Gallagher, T.A.: Apparatus and
method for cold aqueous liquid and/or ice production, storage and use for cooling and
refrigeration. US patent no. 4,596,120, June 24, 1986.
25. Kozawa, Y.; Aizawa, N.; Tanino, M.: Study On Ice Storing Characteristics in Dynamic-
type Ice Storage System by using Supercooled Water. Effects of the supplying
conditions of ice-slurry at deployment to District Cooling and Heating System, Third
IIR Workshop on Ice Slurries,, Lucerne, Switzerland, p. 87-96, 2001.
26. Kozawa, Y.; Tanino, M.: Ice-water Two-phase Flow Behavior in Ice Heat Storage
System, First IIR Workshop on Ice Slurries, Yverdon-les-Bains Switzerland, pp. 146-
156, 2001.
27. Kurihara, T.; Kawashima, M.: Dynamic Ice Storage System using Super Cooled Water,
Introduction of heat pump and thermal storage unit, Fourth IIR Workshop on Ice
162
Slurries, Osaka Japan, pp. 61-69, Nov. 2001.

28. Leloux, M.S.: The Influence of Macromolecules on the Freezing of Water, Journal of
Macromolecular sciences, reviews in macromolecular chemistry and physics, C39(1),
pp. 1-16, 1999.
29. List, R.: Ice nucleation and freezing of water, Second IIR Workshop on Ice Slurries,
Paris, p. 61, 2000.
30. Madsbøll, H.; Minds, G.; Nyvad, J.; Elefsen, F.: The state of the art for water vapour
compressors and cooling plants using water as refrigerant. Proceedings of the IIR Conf.
New Applications of Natural Working Fluids in Refrigeration and Air Conditioning.
Hannover, Germany, pp. 743-754, May 1994.
31. Malter, L.: Binary ice-generation and applications of pumpable ice slurries for indirect
cooling. IIR Applications for Natural Refrigerants, Aarhus, Denmark, pp. 527-534, Sept.
1996.
32. McCloskey, W.D. and Brady, T.W.: Supplemental cooling system for coupling to
refrigeration cooled apparatus. US Patent 5383339, 1995.
33. Meewisse, J.W.; Infante Ferreira, C.A.: Comparing Alternative Ice Slurry Production
Methods, Proc. Sydney Conf., IIR, paper no.580, 1999.
34. Meewisse, J.W., Infante Ferreira, C.A.: Experiments on Fluidized Bed Ice Slurry
Production, Third IIR Workshop on Ice Slurries, Lucerne, Switzerland, pp. 105-112,
2001.
35. Meewisse, J.W.; Infante Ferreira, C.A.: Fluidized bed ice slurry generator: Operating
Range, Fifth IIR Workshop on Ice Slurries, Stockholm, May 2002.
36. Meyerson, A.S.; Ginde, R.: Crystals. 1993, Crystal Growth and nucleation, In: A.S.
Meyerson, Handbook of Industrial Crystallization, Butterworth-Heinemann, Stoneham,
pp. 33-62, 1993.
37. Mito, D.; Mikami, Y.; Tanino, M.; Kozawa, Y.: a New Ice-slurry Generator by using
Actively Thermal-hydraulic Controlling both Supercooling and Releasing of Water,
Fifth IIR Workshop on Ice Slurries, Stockholm, May 2002.
38. Mito, D.; Tanino, M.; Kozawa, Y.; Okamura, A.: Application of a Dynamic-type Ice
Storage System to the Intermittent Cooling Process in the Food Industry, Fourth IIR
Workshop on Ice Slurries, Osaka Japan, pp. 105-114, Nov 2001.
39. Nagato H.: a Dynamic Ice Storage System with a closed Ice-making Device Using
Supercooled Water, Fourth IIR Workshop on Ice Slurries, Osaka Japan, pp. 97-103,
Nov. 2001.
40. Otero, L.; Sanz, P.D.: High-Pressure Shift Freezing. Part 1. Amount of Ice
Instantaneously Formed in the Process, Biotechnol. Prog., 16, pp. 1030-1036, 2000.
41. Patience, D.B.; Rawlings J.B.; Mohameed H.A.: Crystallization of para-xylene in
scraped-surface crystallizers, AIChE Journal, 47 (11), pp. 2441-2451, 2001.
42. Paul, J.: Compressors for refrigeration plants and ice makers with ‘water as refrigerant’,
Proc. Aarhus Conf., IIF/IIR, pp. 577-584, 1996.
43. Paul, J.; Jahn, E.; Lausen, D: Cooling of mines with vacuum ice. Conf. Proc. FRIGAIR
’96, Johannesburg/South Africa, 1996.
44. Pronk, P.; Infante Ferreira, C.A.: Effect of long term ice slurry storage on crystal size
distribution, Fifth IIR Workshop on Ice Slurries, Stockholm, May 2002.
45. Rautenbach, R.; Katz, T.: Survey of Long Time Behavior and Costs of Industrial
Fluidized Bed Heat Exchangers. Desalination, vol. 108, pp. 335-344, 1996.
46. Rivet, P.; Koelet, P.C.: Koeling met tweefasige mengsels, vast en vloeibaar (Cooling by
two phase mixtures, solid and liquid), Koude & Luchtbehandeling, no.2 (February),
1998.
163
47. Russel, A.B.; Cheney, P.E.; Wantling: Influence of freezing conditions on ice
crystallisation in ice cream, Journal of Food Engineering, vol. 39, pp. 179-191, 1999.
48. Sanz, P.D.; Molina-Gracia, A.D.; Diaz, J.M.; Elvira, C. de; Otero, L.: Obtaining
homogeneously dispersed ice crystals by high-pressure-shift-freezing, Third IIR
Workshop on Ice Slurries, Lucerne, Switzerland, pp. 113-118, 2001.
49. Schwartzberg, H.G.: Food freeze concentration, In H.G. Schwartzberg, and M.A. Rao,
Biotechnology and Food Process Engineering, (Chap. 5, pp. 127-202). New York:
Marcel Dekker, 1990.
50. Schwartzberg, H.G.; and Liu, Y.: Ice crystal growth on chilled scraped surfaces, Paper
No. 2g, American Institution of Chemical Engineers Summer National Meeting, San
Diego, CA., 19-22 August, 1990.
51. Smith R.W.: Applications of plate and scraped surface heat exchangers, Food
Manufacture, vol. 47 (10), pp. 37-40, 1972.
52. Snoek, C.W.: The design and operation of ice slurry based district cooling systems”,
Energy Research Laboratories/CANMET, Report submitted to NOVEM, BV Sittard,
The Netherlands, 1993.
53. Sodawala, S.; Garside, J.: Ice nucleation on cold surfaces: application to scraped surface
heat exchangers, The 1997 Jubilee Research Event: a two day symposium held at the
East Midlands Conference Centre, Nottingham, 8-9 April 1997, Institution of Chemical
Engineers, pp. 477-480, 1997.
54. Tanino, M.; Kozawa, Y.: Ice-water Two-phase Flow Behavior in Ice Heat Storage
Systems. Int. Jour. of Refrigeration of IIR, 24(7); pp. 638-652, 2001.
55. Tanino, M.; Kozawa, Y.: Performance Evaluation of an On-site Type Ice Storage
System by using Ice Slurries Made of Supercooled Water. Jour. of AIRAH, 2001: 55(9);
pp. 28-31, 2001.
56. Tanino, M.; Kozawa, Y.; Hijikata, K.; Nakabeppu, O.: Prediction of Ice Storage Process
in Dynamic-type Ice Storage System, Proc. of 10th International Conference on Thermal
Engineering and Thermogrammetry, Budapest, Hungary, pp. 321-326, 1997.
57. Tanino, M.; Kozawa, Y.; Mito, D.; Inada, T.: Development Of Active Control Method
for Supercooling Releasing of Water. Second IIR Workshop on Ice Slurries, Paris, pp.
127-138, 2000.
58. Tanino, M.; Mito, D.; Kozawa, Y.: Recent Study on Ice Slurries. Jour. of AIRAH,
55(8); pp. 17-18, 2001.
59. Ure, Z.: Slurry Iced Based Cooling Systems, In: CIBSE 97 Virtual Conference “Quality
for People”. Chartered Institution of Building Services Engineers, 1997.
60. Ure, Z.: Thermal energy storage for supermarkets, Proc. Intl. Sustainable TES
Conference, Minneapolis, 1996.
61. Wang, M.J.; Goldstein, V.: 1996, A novel ice slurry generator system and its
applications, Refrigeration Science and Technology Proceedings-Applications for
natural refrigerants: proceedings of meeting of, commissions B1, B2, E1, E2, pp. 543-
551, Sept. 1996.
62. Wang, M.J.; Kumusoto, N.: Ice slurry based thermal storage in multifunctional
buildings, Int. J. Heat and Mass Transfer, vol. 37, pp. 597-604, 2001.
63. Watanabe, Y.; Katsuya, Y.; Sekita, S.; Okazaki, T.; Hashiguchi, M.: Development of an
ice storage system using direct-contact heat transfer between water and water-insoluble
antifreeze. Proc. of Autumn Symposium SHASE G-17, 1995.
64. Wiegandt, H.F.; Madani, A.; Harriott, P.: Ice Crystallization Developments for the
Butane Direct Contact Process, Desalination, 67, pp. 107-126, 1987.
164
65. Wobst, E.; Vollmer, D.: Ice Slurry Generation by Direct Evaporating of Refrigerant,
First IIR Workshop on Ice Slurries, Yverdon-les-Bains Switzerland, pp. 125-131, May
2001.
66. Yundt-Jr., B.: Personal communication with E. Stamatiou, 2002.
67. Zakeri, G.R.: A new vacuum freeze system design for energy effective production of ice
slurry. IIR Applications for Natural Refrigerants, Aarhus, Denmark, pp. 585-591, Sept.
1996.
68. Zakeri, G.R.: Vacuum freeze refrigerated circuit, a new system design for energy
effective heat pumping applications, Proc. Linz (Austria) Conf, IIF/IIR, comm.
E1,E2,B2, pp. 182-190, 1997.
coating for ice generation. Fifth IIR Workshop on Ice Slurries, Stockholm, May 2002.
165
IIF-IIR – Handbook on Ice Slurries - 2005
CHAPTER 6. TRANSPORT
by Torben M. Hansen, Ebbe Noergaard and Beat Frei
The transport properties of ice slurry and other slurries in particular have been studied
extensively. Most early studies focussed on frictional pressure drop with drag force reducing
additives, friction factor measurements in various flow regimes and pressure loss in single
components (Snoek, 1993; Larkin, 1988; Liu, 1988; Nørgaard et al., 2000; and many others).
Concern is often expressed about the problems of ice blockage and agglomeration in transport
systems, but only a few authors have actually reported such incidents. Christensen and
Kauffeld (1998) have reported blockage problems from storage extraction when the outlet
velocity is too high and flow fluctuations due to separation of phases if the velocity is too low.
No crystal agglomeration growth is believed to occur in flowing ice slurry due to the very
short time of contact between ice crystals caused by turbulence. However, in combination
with laminar Bingham flow (plug flow), little or no freezing point depressing agent and long
contact time in long transport systems, some agglomeration can be expected. These effects are
not very likely to be seen in most laboratory experiments, where flow is disturbed by bends
and other fittings.
Agglomeration can be expected to occur in closed pipe sections with no flow. Fine crystalline
ice slurry plugs are normally easily dissolved after standing still. Furthermore, in small pipes,
the ice will normally melt off prior to agglomeration. However, in large insulated pipes, it is
much more unlikely that the ice can be removed by heat input from the surroundings before
agglomeration takes place, especially in low additive concentration ice slurries close to 0°C.
6.1 Advantage of transport characteristics of ice slurries over conventional fluids
As can be seen from the thermophysical properties, a high ice content allows for very high
density of cooling capacity per unit volume compared with ice-free coolant. The higher the
ice concentration that can be achieved in the transport system the more saving can be obtained
on pipe dimensions and pump capacity. However, several investigations report that pressure
drop increases progressively at ice concentrations above approximately 25 wt-% (or % by
weight, ref. Chapter 3.3). Also bearing in mind the operation of pumps and extraction
reliability from storage tanks, an ice content of 30 wt-% in the transport system is considered
to be the upper practical limit.
With an upper limit of 30 wt-% of ice, the pipe diameter can be reduced by 40% and savings
on the pipe installation are estimated to be 70% (Choi and Knodel, 1992). The outstanding
transport properties are clear from Figure 6.9 showing the range of cooling capacity that can
be obtained in pipe diameters from 16 to 50 mm.
6.2 Design of transport systems
6.2.1 GENERAL CONSIDERATIONS
Transport systems for ice slurry may be designed in a similar way to traditional water
systems; however, in some cases there is concern that the flow rate in branches may drop
below the point of separation of ice and carrier fluid. In such cases, the transport loop and the
167
individual branches must be designed to maintain a given minimum flow during varying
cooling demands (e.g. between 0-100% of the branch design).
6.2.2 SINGLE-TUBE VERSUS DUAL-TUBE SYSTEMS
Due to the high enthalpy of ice compared with the single-phase fluid the temperature of ice
slurry remains almost constant during melting of ice. Note that the temperature glide of ice
slurry varies with the chosen freezing point depressant and with the concentration of the
additive.
In contrast to single-phase fluids ice slurry can exchange heat with multiple serially connected
heat loads without introducing a significant reduction in the temperature difference in the
downstream equipment. However, while the temperature remains fairly constant in the entire
transport system, the heat exchange performance of the fluid decreases as the ice content
decreases. This means that the benefits from installing less piping must be compared to the
drawbacks of having to install larger heat exchangers. Figure 6.1 shows an example of
reduced piping length by applying a single pipe system.
Mono pipe system: 80 metres Dual pipe system: 110 metres
10 m 10 m 10 m 10 m
25 m 15 m 15 m
25 m
10 m 10 m 10 m 10 m
Figure 6.1. Example of reduced total pipe length by using a single pipe system
6.2.3 CONSIDERATIONS OF DEFROSTING
Any air coil operating with ice slurry below –4°C occasionally requires defrosting to ensure
optimal performance. Among typical installations where defrosting must be considered are
supermarkets (ice slurry at –8°C), cold storage rooms (ice slurry at –8°C to –6°C), abattoirs
and slaughterhouses, restaurants/catering equipment and low supply air temperature air
conditioning systems (ice slurry at 0°C to –10°C).
Various methods can be applied to defrost ice slurry air coils. In applications with a high
product heat capacity or in systems with long part-load periods the most beneficial defrosting
method is by recirculation of air through the air coil (i.e. applying fan heat to melt the ice),
which at the same time ensures air movement and maintains a uniform air temperature in the
cabinet or in the room. In the case when an air coil requires cleaning, hot water spray systems
are suitable for defrosting. Electrical rods for defrosting are also commonly used, especially
168
in supermarket installations where low initial cost is considered more important than the
energy consumption during operation. Regardless of which of the above-mentioned methods
is applied, the coil must be shut off from the ice slurry supply line.
Equivalent to the hot gas defrost applied with refrigerants, warm liquid circulation defrost
may be applied in ice slurry systems. The main criterion for using this technique is that the
warm liquid has the same overall additive concentration as the ice slurry.
The warm liquid can be heated with waste heat sources (e.g. condensing heat or compressor
cooling oil). The location of heaters can be installed in either a central (Figure 6.2) or a local
hot water system (Figure 6.3). If the defrost demand is large and the number of heat
exchangers that require defrost is large, then centralized systems are most beneficial, while
local defrost systems are suitable to cover dispersed defrost duties.
Centralized warm liquid circulation systems can be installed with a 3- or 4-pipe system as
shown in Figure 6.2 while various layouts for local defrost configurations are shown in
Figure 6.3.
Figure 6.2. Centralized warm liquid defrosting systems
169
T T
T
T
T
T T T
Figure 6.3. Local warm liquid defrosting systems
6.2.4 PIPE MATERIALS
ABS pipes (a thermo plastic) can be used at temperatures down to –40°C and have proved
mechanical durability and chemical stability with various additives, sodium chloride, ethanol
and propylene glycol. Due to their low weight and easy assembly the ABS pipes are highly
suitable for cooling installation and the pipes offer very flexible mounting possibilities. Ice
slurry systems are, in the majority of systems, open to the ambient air to a certain degree and
the non-corroding ABS pipes therefore are preferable to copper or steel piping.
6.2.5 CALCULATION OF PRESSURE DROPS
Pressure drop calculation for ice slurry flow is described in Chapter 3.3.
6.2.6 AVOIDANCE AND INFLUENCE OF PHASE SEPARATION
Experience with ice slurry has revealed a tendency of ice settling in zones with low flow rate
or recirculation swirls. A major concern is that once the ice settlement is initiated it will
continue as the settled ice in some cases acts as a filter to other particles.
The slurry flow behaviour of gypsum- and laterite-water slurries at concentrations up to 12.7
and 30.6% vol. respectively has been observed in 90° and 180° bends, gate valves,
contractions, expansions and Venturis (Turian et al., 1998). Pipe diameters were 25 and 50
mm. No blockage was reported for flow with Reynolds numbers in the range of 103 to 105.
Single-phase pressure loss coefficients in the fully turbulent range were identical to water
flow.
Kawada et al. (1996) investigated ice and snow slurry flows at concentrations of 5 to 20 wt-%
in 90° branches (T-junctions). In the experiment, a 78 mm main pipe and a 50 mm branch
were used to investigate flow conditions and single-phase pressure loss coefficients in
horizontal and vertically upward and downward flows. Kawada et al. also measured the ratio
of ice concentration in the 2 branches in the horizontal and vertical T-junctions. The results
revealed that with horizontal branching, the ice concentration was identical in the main pipe
and the branch if the main pipe velocity exceeded 1.5 m/s. At the lower velocity of 1 m/s, the
ratio of branch and main pipe volume flow rates must exceed 0.5 to ensure identical ice
fractions. In vertically upward and downward branches, Kawada et al. found fractionation of
ice concentrations in the main pipe and branch in all practical cases.
170
Due to the density difference of the phases, a certain slip will always occur in the velocity of
the phases. If the drag force on particles drops below a certain value, a significant separation
of phases can be observed. The problem is well-known in the mining industry and a lot of
literature on transport of mineral slurries is believed to apply to ice slurry as well.
In horizontal pipes, the authors have defined 3 flow regimes as moving bed, stratified flow
and homogeneous flow (Figure 6.4). The results shown in Figure 6.4 have recently been
validated from the experiments reported by Kitanovski and Poredos (2001). In vertical pipes,
the ice will form a plug around the centre line of the pipe, most noticeable in the downward
flow direction (Fukusako et al., 1999). Separated flow results in increased pressure drop
(additional friction between phases) and may also lead to undesirable flow fluctuations.
0,5
0,4
Velocity [m/s]
0,3 Moving bed

Stratisfied
0,2 Homogeneous
0,1
0
0% 5% 10% 15% 20% 25% 30%
Ice concentration [wt.%]
Figure 6.4. Separated flow regimes. Left: measurement of separation velocities

in a horizontal 50 mm pipe based on visual inspection. Right:
ice concentration profiles (dashed lines) and velocity profiles (solid lines) for moving bed,
stratified and homogeneous flow (top to bottom). (Hansen and Noergaard).
(This reference is missing in the List of References)
Normally in a horizontally mounted pipe with flow velocities higher than 1 m/s, the
turbulence of the flow will be sufficient to suppress separation of flow due to the difference in
densities of carrier fluid and ice. When the flow velocity is lowered, the ice concentration will
be richer at the top of the pipe and eventually a moving bed of ice will form in a fully
separated flow (cf. Figure 6.4). Parameters of importance to ice separation have been found to
be:
1. Ice concentration: the lower the ice concentration, the earlier the onset of phase
separation.
2. Pipe diameter: the bigger the pipe, the earlier the onset of phase separation.
3. Density ratio: the bigger the ratio between carrier fluid and ice density, the earlier
separation occurs.
4. Additive type and concentration: influence due to the density ratio, but also due to the
viscosity. The more viscous the carrier fluid, the later the onset of phase separation.
5. Mean size of ice crystals: the bigger the ice crystals, the earlier the onset of phase
separation.
171
6. Age of ice in the storage system: the flow separates more readily when the ice crystals are
old (large).
Problems caused by phase separation have not been investigated in this survey.
In order to avoid phase separation in horizontal pipes, the minimum velocity can be calculated
from Equation 6.1. For additional safety, it is recommended to dimension the piping so as to
avoid velocities lower than twice the minimum velocity, umin. For better prediction, the more
accurate multiple parameter model suggested by Kitanovski and Poredos (2001) is
recommended.
⎛ ρ ⎞
u min = 1.4 ⋅ g ⋅ D ⋅ ⎜⎜1 − ice ⎟
⎟ (6.1)
⎝ ρ cf ⎠
In vertical pipes, the upward flow is homogeneous even at quite low velocity. In downward
flow, however, separation of ice in a slow flowing ice core around the centre line of the pipe
is likely to occur if the flow velocity is kept below the minimum flow velocity determined for
horizontal flow.
Fittings and single components

No problems have been identified in long and short bends, expansions or reductions. Even
large plugs of agglomerated ice can flow easily through horizontal and vertical bends,
reductions and expansions.
Branching of ice slurry in T-junctions calls for attention to proper mounting and orientation.
The most obvious situation is the T-junction having a vertical branch facing upwards as
shown in Figure 6.5. When the valve is open and the horizontal flow is above the phase
separation velocity the operation is safe. When the valve on the upward branch is closed, the
turbulence induced from the flow passing the branch and the gravity difference of ice and
carrier fluid will create a growing plug starting at the position of the valve. Due to the
presence of carrier fluid the plug remains relatively soft, but when the valve is opened again
and the plug is released it does cause a significant flow pulsation. However, it has not been
possible to force a blockage of the flow, not even in valves (ball and diaphragm) with 10%
opening angle further downstream from the valve directly above the T-junction.
Figure 6.5. T-junction with a vertical upward facing branch without flow,
i.e. the valve is closed
172
Experiments with an insulated vertical branch of 25 mm inside diameter did not change this
conclusion. Due to the small diameter, the ice is more likely to melt than to agglomerate into
a harder block. It is believed that this conclusion is not necessarily valid for large pipe
diameters where the amount of ice to melt is much larger compared to the heat input from the
surroundings, and thus the experiment was repeated in an insulated 100 mm pipe system,
where no problems were found. Note that all experience described here was gained with ice
slurry containing a freezing point depressant, i.e. ethanol or glycol, resulting in a freezing
point of the mixture below –4°C. Results might be different for pure water or very low
additive concentration ice slurry, due to a higher rate of agglomeration.
Figure 6.6 shows a vertically mounted T-junction without flow in the left-hand pipe section.
Again the turbulence induced from the sudden change in the direction of flow results in a
positive carry-over of ice to form a slab of ice deposit in the top of the left-hand pipe. The
length of the slap seems to reach a stable size of 4-6 times the pipe diameter. The slab is soft
and does not cause problems in the downstream section of the left-hand pipe section.
The same situation will occur if the closed pipe (here left-hand side) is facing downwards.
The situation was also investigated with insulation on the left side part, but the conclusion is
similar to the previous case, i.e. no blockage could be forced. The case was re-investigated in
a larger 100 mm pipe system with insulation, where no problems occurred.
Figure 6.6. Vertical T-junction without flow in the left branch
Other flow directions of vertical and horizontal orientation of T-junctions (including 45°)
have not revealed potential agglomeration.
During a test with the 100 mm inside diameter ABS pipe arrangement shown in Figure 6.7,
the ball valve in the furthest downstream parallel branch was fully open, the butterfly valve
was closed and the three valves closest to the pump, located to the right of the shown pipe
section, were fully or partially open. The flow rates in the diaphragm, gate and globe valve
were relatively high and a relatively lower flow was distributed downstream to the ball valve.
During the first four hours of operation the flow through the ball valve fluctuated and
eventually a sudden pipe blockage occurred. After approximately six hours of blockage the
flow revived and the flow was highly unstable for another ten hours of operation until the pipe
was finally blocked again. There is no straightforward explanation for the flow scenario
shown in Figure 6.7. It is believed that the flow in the supply pipe downstream of the three
open valves was in the stratified region and even a moving bed type of flow might have
173
occurred. Since the fully open ball valve represents no flow restriction in the parallel branch it
is probable that the entire vertical pipe section was filled with ice and blocked.
The fact that the flow revived after six hours is believed to be due to melting of the ice in the
pipe. In the measuring period from approximately 800 to 950 minutes the pipe seemed to be
slowly filling up with ice and then suddenly the flow recovered to normal level, after which
the pipe was blocked after another 3½ hours. This unstable behaviour observed in the
experiment remains unexplained.
After the sudden opening of the butterfly valve after 24 hours, the flow through the section
was immediately established. Since the flow had been blocked in the downstream ball valve
section there had been no transport of ice into the vertical branch containing the butterfly
valve for four hours. In this case, any ice that might have been deposited in the vertical branch
with the butterfly valve was most probably melted when the valve was opened.
Butterfly Gate valve

Ball valve Globe valve Diapraghm valve
valve 100% open
100% open 100% open 50% open
closed
40 5
35
Distribution [% of total flow]
4
30
25
[gallons/s]
3
20
2
15
10
1
5
0 0
0 200 400 600 800 1000 1200 1400 1600
Time [minutes]
ball butterfly globe gate diaphragm flow (2nd axis)
Figure 6.7. Flow in parallel valve sections, experiment 14A.

(Hansen, Radosevic, Kauffeld, 2002)
6.3 Transport capacity
6.3.1 CALCULATION OF TRANSPORT CAPACITY
The transport capacity can be calculated as fluid mass flow multiplied by enthalpy change
between the supply and return lines. The enthalpy of ice slurry, h, at a temperature (T) can be
written as:
h = hice (T , xice ) ⋅ xice + hcf (T ) ⋅ (1 − xice ) (6.2)
174
where hice is the enthalpy of ice and hcf the enthalpy of the carrier fluid. In the case of non-
ideal mixture, the heat of mixing must be included in the enthalpy term for the carrier fluid.
(See Eq. 2.40 in Chapter 2.5.3).
Based on the thermophysical properties of a pure substance at a reference temperature of 0°C,
the enthalpy can be calculated as:
h=((1-xice)·cp,cf+xicecp,ice) (Tslurry ) – hls· xice (6.3)
where hls is the latent melting heat of ice and cp is the specific heat capacity of the 2 phases,
ice and carrier fluid (cf), respectively.
Figure 6.8 shows a temperature enthalpy diagram of an ice slurry made from ethanol/water.
For comparison, plain water ice slurry is also shown. It should be noted that usually
supercooling of water and water/ethanol mixture occurs prior to the formation of the first ice
particles. This supercooling can be as high as 3 to 4 K and is not shown in Figure 6.8.
40 wt% ice 30 wt% ice 20 wt% ice 10 wt% ice

0 water
-2 5wt% ethanol
-4
10wt% ethanol
Temperature (°C)
-6
-8
-10
-12 20wt% ethanol
-14
-16
-18
-20
-175 -150 -125 -100 -75 -50 -25 0
Enthalpy (kJ/kg)
Figure 6.8. Enthalpy of ice slurry depends on ice fraction and additive concentration
If one compares the large enthalpy content due to the latent heat stored in the ice particles of
the 10% ethanol ice slurry to a conventional liquid-only 20 wt-% ethanol/water mixture at the
same temperature range, one can see the great difference in heat capacity or enthalpy between
ice slurry and a traditional secondary refrigerant.
6.3.2 PIPE SELECTION CHART
Transport capacity, Q, (in kW of cooling at full melt off) in pipes of inner diameter, Drør, is
shown in Figure 6.9. The increase in transport capacity with the ice slurry flow rate, ice
concentration and pipe diameter is clearly shown.
175
500 500
fullafsmeltning
Fuld melt off
Max. flow = 1.8 m/s
Drør = 50 mm
Min. flow = 0.6 m/s
100 Max. flow = 1.6 m/s 100

Drør = 32 mm
Min. flow = 0.5 m/s
Q [kW]
Q [kW]
Max. flow = 1.4 m/s
Drør = 25 mm
Min. flow = 0.4 m/s
10 10
Max. flow = 1.2 m/s
Drør = 16 mm
Min. flow = 0.3 m/s
1 1
0 5 10 15 20 25 30 35 40
Ice concentration in %
Iskoncentration [%]
Figure 6.9. Transport capacity (kW of cooling at full melt off) in pipes
(Drør : Inner pipe diameter)
6.4 Pumping
As in other hydraulic energy transport systems, pumps are essential in ice slurry systems
where ice is distributed between the storage tank and the cooling consumers. The scope of this
description includes screen tests of pump types suitable for ice slurry and established
knowledge of pump performance with ice slurry.
6.4.1 CENTRIFUGAL PUMPS AND THEIR APPLICATION RANGES
Ice slurry differs significantly from traditional single-phase coolants due to the high energy
density and the presence of solid particles. Compared with e.g. cold water systems ice slurry
requires four to eight times lower flow rate but two to three times higher pressure lift. Positive
displacement pumps, such as gear pumps, lobe pumps, screw pumps and mono pumps, are
often used in industrial applications involving slurry flow, due to their ability to pump highly
viscous media. Compared with centrifugal pumps, these pumps are more expensive and,
therefore, it has been of great interest also to investigate the operational range of centrifugal
pumps with ice slurry.
Single and multiple stage centrifugal pumps have proved to work well with ice slurry at ice
concentrations up to 35 wt-% and fluid kinematic viscosity less than 50.10-6 m2s-1. Above ice
concentrations of 35 wt-%, screw or mono pumps are recommended.
176
Six different multiple stage centrifugal pumps (manufactured by Grundfos) have been tested
with 16 % propylene glycol and two pumps have been tested with 10 % ethanol (results from
the Danish Technological Institute) as listed in Table 6.1.
Table 6.1: Centrifugal pumps tested with ice slurry;

the manufacturer’s (Grundfos) product name/number is used
No. of stages Test fluid

CRE 2-50 5 16% Propylene glycol
CRE 4-20/1 1 16% Propylene glycol
CRE 3-110 11 10% Ethanol
UPE 80-2 2 10% Ethanol
The experimental set-up is shown in Figure 6.10. The ice slurry flowed from an agitated 1 m3
accumulation tank into a feeding pump and transported to the test pump. On both sides of the
test pump, pressure transmitters were installed to measure the head of the test pump. The ice
concentration was determined from the density measured with a Coriolis mass flow meter.
Since the mass flow meter generated a relatively large pressure drop, the loop with the mass
flow meter was only momentarily active before each measurement. Thereafter, the flow rate
was measured with an electromagnetic volume flow meter to allow determination of the entire
performance curve.
Figure 6.10. The experimental set-up at the Danish Technological Institute

and a sketch of a Grundfos CR pump
The pump efficiency was calculated as shown by Equation (6.4), where Q , ∆p and
Pconsumption , total are the measured flow rate, pressure difference, and power consumption,
respectively. The power loss in the motor, Ploss , motor was determined from manufacturer’s
data:
Q ⋅ ∆p
η pump = (6.4)
Pconsumption , total − Ploss , motor
The performance curves shown in Figures 6.11 and 6.12 are representative of all the pumps
tested.
177
The pump capacity of the pump decreases when the ice concentration increases. At constant
pump speed the delivered pressure head of the pump decreases progressively with increasing
ice concentration especially at higher flow rates.
50 Performance curves
2800 rpm
45
40 0% ice
35 10% ice
30 2200 rpm 20% ice
H (m)
30% ice
25
20
15 1450 rpm
10
5
0
0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5
3
Flow rate (m /hr)
Figure 6.11. Performance curves of a Grundfos CR 2-50 pump working with ice slurry
The performance loss with increased ice content is also evident from the efficiency curves in
Figure 6.12. The power consumption increases at larger ice concentrations and the efficiency
decreases. The results can be explained by the change in viscosity and density and possibly
also the presence of solid particles.
1,0 1000
30 wt.% ice
0,9 900
0,8 800
0,7 700
0 wt.% ice
0,6 600
P [watt]
η (−)
0,5 500
0 wt.% ice
0,4 400
0,3 300
0,2 200
0,1 30 wt.% ice 100
0,0 0
0 1 2 3 4 5
3
Flow rate (m /hr)
Figure 6.12. Stage efficiency and power consumption curves

of a Grundfos CR 2-50 pump working at 2800 rpm with ice slurry
178
A new series of centrifugal pumps with modified rotors and houses (CR5, but also CR1 and
CR3) revealed significantly improved performance with ice slurry as shown in Figure 6.13,
especially in the series with high ice concentration.
0,70
0,60
0,50
0,40 0 wt.% ice

η (-)
10 wt.% ice
20 wt.% ice
0,30
30 wt.% ice
0,20
0,10
0,00
0,00 1,00 2,00 3,00 4,00 5,00 6,00 7,00 8,00 9,00
Flow rate (m3/hr)
Figure 6.13: Stage efficiency of a Grundfos CR5-11 pump

working at 2800 rpm with ice slurry
6.4.1.1 Special purpose centrifugal pumps
6.4.1.1.1 Immersed pump

A special centrifugal pump to mount directly into the top of the storage tank was designed and
tested. The Grundfos pump was of the type CRK-4-330/11, with 11 impeller wheels and 22
empty flow chambers. At the end of the impeller wheel two screws or snails are mounted to
create mixing zones near the pump inlet and the inlet to the first impeller wheel.
The idea of using an immersed pump was first tested to investigate if the normal stirring
device (propeller type) could be substituted by the flow conditions induced by the immersed
pump. The suction inlet of the pump creates a downward directed vortex which together with
a tangential bypass on the pressure side should facilitate proper mixing. Secondly the design
should eliminate any potential problems induced by the storage tank outlet.
The mixing was tested at varying ice fractions and with different jet nozzles at the pressure
outlet – but proper mixing was not achieved under any conditions.
179
Extension pipe formed by

empty flow chambers
Flow chamber with

impeller wheels
Inlet feeding snail
Figure 6.14. Design of centrifugal pump to be immersed in to the ice storage
The specific pump design revealed severe start-up problems after a long time of immersion in
the ice slurry storage tank without operation. The problem is believed to be due to ice that
packs and drains inside the empty flow chamber and thereby causing an internal flow
blockage. Further, at high ice fractions the pump showed the most significant performance
loss due to the flow resistance through the empty upstream flow chambers. These flow
problems can be solved through alternative design, however, the use of the pump as a
replacement for a propeller stirring device is not promising.
6.4.1.1.2 Control loop circulation pump

A special pump has been designed for use in heat exchanger control loops. The purpose of the
pump is to maintain a constant flow rate through the heat exchanger while the cooling load is
varying. The high energy density of ice slurry and the pressure drop characteristics of ice
slurry affect the specifications for the pump. Compared to traditional single-phase fluids the
pump must have a combination of low volume flow and increased pump head, e.g. 0.5
m³/hour at 10 metres pressure head for a cooling capacity between 0 and 12 kW.
These specifications were met by combining the pump housing of a Grundfos UPE pump with
three impeller stages of a Grundfos SP pump. The SP stages provide a higher pump efficiency
(0.2 to 0.4) compared to traditional circulation pumps (0.1 to 0.2) and thus the energy
consumption for the internal circulation loop is reduced by four to eight times compared to
common single-phase fluid systems. The flow chamber and the number of stages can be
adjusted to meet any other specifications regarding flow and pressure.
180
Figure 6.15. Control loop circulation pump

(combined UPE and a SP pump: UPESP1A-3)
6.4.2 OTHER PUMPS FOR ICE SLURRY
6.4.2.1 Screw pump

A commercially available screw pump with three spindles was tested with ice slurry made
with 10 wt-% ethanol. The pump investigated (see Figure 6.16) operated in a range of
volumetric flow rates between 1-2.5 m³/h. For a defined volumetric flow rate, the
manufacturer decided which pump size should be built into the experimental device.
175
ice fraction=0% / -3.85°C screw pump
Pump total head H (m)
150 ice fraction=5% / -4.15°C

ice fraction=10% / -4.48°C S=1500 rpm
125 ice fraction=15% / -4.86°C 10 w% talin
ice fraction=25% / -5.85°C
100
75
50
25
0
0.0 0.5 1.0 1.5 2.0 2.5
Volume rate of flow Q (m³/h)
Figure 6.16. Screw pump Figure 6.17. Pump characteristics

for the screw pump in Fig. 6.16
The total head of this screw pump is shown in Figure 6.17. For a given flow rate the head of
the pump increases with the increasing ice fraction. The results without ice differ significantly
from those with ice. This can be explained by the fact that the pump was especially designed
for fine crystalline ice slurry. The distance between the three spindles is extended.
181
In those cases where a screw pump is used it should be ensured that the ice fraction is not
lower than 5%. This should be considered in the layout of an ice slurry application.
Figure 6.18 shows the measured electrical power input, P (denoted “E” in the figure) against
.
the volumetric flow rate V (denoted “Q” in Figures 6.17-19, 6.21-23) for the screw pump
shown in Figure 6.16. These results show an increasing slope of the curves obtained with
increasing ice fraction. A considerable difference in the behaviour is observed in the presence
of ice particles. Even for low ice fractions, i.e. 5%, a significant increase of the slope is
observable. Again the reason lies in the construction principle of this pump.
1500
Electrical power input E (W)

1200 ice fraction=10% / -4.48°C
600 S=1500 rpm

10 w% talin
300
screw pump
0
0.0 0.5 1.0 1.5 2.0 2.5
Figure 6.18.. Electrical power input, P (denoted “E” in the figure) against volumetric
flow rate V (denoted “Q” in the figure) for the screw pump shown in Figure 6.16
The overall efficiency. was calculated as given by Equation (6.5), using the measured
volumetric flow rate V (m³ s-1), pump’s total head H (m), density ρ (kg m-3), and electrical
power input P (W):
V& ⋅ H ⋅ ρ ⋅ g
η= (6.5)
P
In contrast to centrifugal pumps, the screw pump indicates that the overall efficiency is
influenced by the viscosity (Figure 6.19). Increased ice fraction, i.e. a higher viscosity value,
means an increased overall efficiency for the screw pump. The overall efficiency is
significantly lower without ice than with ice fractions of 5 to 25%.
182
0.70
Overall efficiency η (-)

0.60 ice fraction=5% / -4.15°C
0.50
0.40 ice fraction=25% / -5.85°C
0.30 S=1500 rpm

0.20 10 w% talin
0.10 screw pump
0.00
0.0 0.5 1.0 1.5 2.0 2.5
Figure 6.19. Overall efficiency for the screw pump shown in Figure 6.16
6.4.2.2 Side channel pump

A commercially available side channel pump was chosen (see Figure 6.20) operating in a
range of volumetric flow rates between 1-2.5 m³/h. The ice slurry was made using a 10 wt-%
ethanol solution. For a given volumetric flow rate the manufacturer decided which pump size
should be built into the experimental device.
Figure 6.20. Side channel pump
For the side channel pump the values of the pump total head for ice fractions between 0 and
15% are within the range of measurement uncertainty (Figure 6.21). Markedly lower values
for the pump total head were measured for an ice fraction of 25%. The slope of the curve
characteristic of a side channel pump is greater than the slope of a centrifugal pump (cf.
Figure 6.11).
183
80
side channel pump
Pump total head H (m)

60 S=1450 rpm
10 w% talin
40
0
0.0 0.5 1.0 1.5 2.0 2.5
Figure 6.21. Pump total head for the side channel pump shown in Figure 6.20
The electrical power input to the side channel pump was measured and is shown in Figure
6.22. Scatter of individual measuring points varies within the range of measurement
uncertainty. The ice fraction, even for values of 25% has no observable influence. The values
for the electrical power input increase with decreasing volumetric flow rate.
1500
Electrical power input E (W)
1200
900
600
ice fraction=5% / -4.15°C S=1450 rpm
ice fraction=10% / -4.48°C 10 w% talin
ice fraction=25% / -5.85°C side channel pump
0
0.0 0.5 1.0 1.5 2.0 2.5
Figure 6.22. Pump total head for the side channel pump shown in Figure 6.20
The overall efficiency of the side channel pump was calculated using Equation (6.5) and is
shown in Figure 6.23. For ice fractions of 5 to 15%, the differences vary within the range of
measurement uncertainty. At an ice fraction of 25% the overall efficiency is significantly
lower.
184
0.25
side channel pump
Overall efficiency η (-)

0.20
S=1450 rpm
10 w% talin
0.15
0.10 ice fraction=0% / -3.85°C

0.05 ice fraction=10% / -4.48°C
0.00
0.0 0.5 1.0 1.5 2.0 2.5
Figure 6.23. Overall efficiency of the side channel pump show in Figure 6.20
6.4.2.3 Lobe pump

The lobe pump tested revealed flow pulsations at ice concentrations above approximately 25
wt-% of ice. Apparently the small inlet diameter of the pump relative to the swept volume
chamber caused problems in the feed supply into the pump. Due to this no further
investigations were carried out with lobe pumps.
6.4.2.4 Mono pumps

Mono pumps have been tested successfully with ice concentrations up to 60 wt-%. Above 30
wt-% a special inlet design with an internal feeding screw must be used to ensure the supply
of ice into the pump rotor. In systems where ice slurry must be distributed at ice
concentrations higher than 30 wt-%, mono pumps are recommended. If the allowed pump
outlet pressure is higher than the design pressure of the piping systems a pressure switch
should be installed for protection purposes.
6.4.3 CALCULATION GUIDELINES USING MANUFACTURER’S DATA
Apart from applicability another important issue in this context is whether the standard
manufacturer’s data on centrifugal pumps for ice-free liquid can be converted into data which
can predict the pump performance for applications with ice slurry.
6.4.3.1 Prediction of pump performance using viscosity models

If the pump performance for ice slurry is mainly influenced by the increasing viscosity as the
ice concentration is increased then pump manufacturer’s data should provide a conversion
approach to predict ice slurry operation. Figure 6.24 shows the performance curves predicted
from the manufacturer’s data (Grundfos Win Caps 7.1) using the Jeffrey viscosity model
For comparison, the measured values of pump head and flow rate are plotted in the same
figure. Among other viscosity models, the pump curves based on the Jeffrey viscosity model
proved the best fit to the measured values, especially at high ice concentrations. The results
obtained are not the same for all pumps. In some cases, the manufacturer’s data are above the
185
measured values, and in some cases they are below. The deviation between the
manufacturer’s data and the measured values might be explained to some extent by
manufacturer’s tolerances, see Figure 6.25.
50
45 Viscosity, 10 wt.% ice
Viscosity, 20 wt.% ice
40 Viscosity, 30 wt.% ice
35 Measured, 10 wt.% ice
Measured, 20 wt.% ice
30 Measured, 30 wt.% ice
H (m)
25
20
15
10
5
0
0 1 2 3 4
Flow rate (m 3 /h)

Figure 6.24. Measured performance curves compared to manufacturer’s data based
on Jeffrey’s viscosity model (Grundfos pump CR 2-50 at 2800 rpm)
40,0
30 wt.% ice
35,0 2750 rpm
30,0
25,0
2200 rpm
H (m)
20,0
15,0
1450 rpm
10,0
5,0
0,0
0,00 0,50 1,00 1,50 2,00 2,50 3,00 3,50
Flow rate (m³/h)
Figure 6.25. Measured performance curves compared to manufacturer’s data based

on Jeffrey’s viscosity model (30 wt-% ice, Grundfos pump CR 2-50 with variable speed).
Cross bars indicate manufacturer’s tolerance
186
Conversion factors
To present the change in the performance at different flow rates and ice concentrations, it was
suggested by Langgaard (2001) to plot the change in relative head, H/H0, with the ice
.
concentration for different flow rates relative to V 0 (denoted “Qo” in Figures 6.26-27). The
.
rated head H0 and flow rate V 0 are the values at 0% ice found at the optimum pump
efficiency, η pump . The overall idea is to enable the prediction of the performance of one
specific impeller geometry from only one normalised diagram when the performance data is
. . .
known, e.g., for water. In Figure 6.26, the curves for 0.6 x V 0 , 1.0 x V 0 and 1.4 x V 0 are
shown for the pumps CR2-50, CR2-70, CR4-40 and CR5-11.
The internal flow pattern for the CR4-40/1 pump with only one stage differs significantly
from those of the other pumps tested and is therefore not represented in Figure 6.26.
From Figure 6.26, it appears that the two pumps CR2-50 and CR2-70 with the same impeller
wheel size group around the same level of relative head change (H/H0). However, from the
curves of the pumps CR2-50 and CR2-70 with identical impeller dimensions, it can be seen
that the internal flow losses between the stages have a significant effect on the performance
above approximately 20 wt-% of ice, especially at high relative flow rates. At this operational
point it appears that the higher the number of stages the higher the internal losses.
The drop in performance is less for pumps with the larger impellers CR4-40 and CR5-11
because it leads to a higher specific speed number. The CR5 pump losses are maximum 20-
25% of pressure head when the flow rate and ice concentration are increased, whereas the
CR2 pumps have much more obvious performance losses.
1.30
1.20
1.10
1.00
0.6 x Qo
0.90
0.80
0.70
H/H0
0.60
1.0 x Qo
0.50
0.40
0.30 CR 5-11
CRE 2-50
0.20 1.4 x Qo
CRE 2-70
0.10 CRE 4-40
0.00
0 5 10 15 20 25 30
Ice concentration wt.%
Figure 6.26. Relative performance curves of the tested centrifugal pumps at 2800 rpm
187
To reveal the influence of ice particles on the pump performance, measurements were carried
out on the CR5 pump with a single-phase viscous fluid with a kinematic viscosity varied from
0 to 15 cSt, which corresponds to the viscosity of ice slurries with 0 to 30 wt-% ice. The
relative performance curves at varying viscosity and flow rates are shown in Figure 6.27. At
low flow rates the pump performance for ice slurry and single-phase fluid is remarkably
similar. At the nominal flow rate the performance remains almost identical up to 20 wt-% ice,
but thereafter the relative pump head is degraded approximately by 10% when the ice
concentration is increased to 30 wt-%.
The internal flow behaviour of a non-Newtonian fluid such as ice slurry apparently increases
the effect of losses. The presence of internal recirculation zones caused by the shape of the
inlet section and the channels in the pump may account for some of the deviation.
From the results in Figure 6.27 there is a clear indication that at high flow rates the presence
of solid particles has an influence on the pump performance, beyond the fluid viscosity.
1.2
0.6 x Qo
1.0 x Qo
0.8
H/H_0
0.6 0 wt.% ice

1.4 x Qo
10 wt.% ice
20 wt.% ice
0.4
30 wt.% ice
0.2
Viscous single phase fluid
Ice slurry
0
0 5 10 15
Dynamic viscosity [cSt]
Figure 6.27. Comparison of measured relative performance of the CR5 centrifugal pump
(2800 rpm) with and without ice particles
The results shown also indicate that the manufacturer’s performance data together with a
viscosity model may be applied for pump selection in most cases. It is notable, however, that
the selection of pumps with larger impeller wheels show less deviation from single-phase
viscous fluid performance. Furthermore, it should be noted, that system design and operation
.
of the pump near or below the optimum efficiency flow rate V 0 provides better accuracy of
the predicted performance.
188
6.4.4 LONG TERM PERFORMANCE TEST
The properties of ice slurry change over time, even though the additive concentration and the
ice concentration are held constant (see Chapter 2.2). To test the effects of both the change in
ice slurry properties and potential particle deposition, a Gundfos CR5–11 pump was subjected
to 24-hour tests. The pump had a modified housing that allowed a visual inspection of the
flow pattern inside the housing. The inspection was focused on detecting any location behind
the housing where the ice slurry might tend to cluster and generate solid ice blocks.
The performance was monitored in tests with different ice concentrations in the range 0-30
wt-% ice. During each 24-hour test, the ice concentration was controlled by a continuous
density measurement, where a signal (on/off) was sent to the ice generator, when the density
deviated more than 1 kg/m3 from the set point.
During the 24-hour run the pump performance was monitored by measuring the flow rate, the
power consumption and the pressures at the pump inlet and outlet. A smaller pump fed the
test pump. In this way, the inlet pressure was kept above 1 bar.
.
The pump was tested at its optimum flow rate V 0 and at a lower flow rate of approximately
.
33% of V 0 .
At the optimum flow rate the pump operation remained stable in all tests with 0 to 30 wt-%
ice. However, at the lower flow rate range the tests revealed a gradual increase in the flow rate
and a corresponding drop in the average pump head. As an illustration of this behaviour, the
test results with 20 wt-% ice are shown in Figures 6.28 and 6.29.
Density 20% ice

980
975
[kg/m ]
3
970
965
960
0 200 400 600 800 1000 1200 1400
[min]
Figure 6.28. Density of the ice slurry during the 24-hour test, 20% ice = 969 kg/m3
189
2 68
1.9
1.8
[m3/hr]
H [m]
1.7
1.6
1.5 67.5
0 200 400 600 800 1000 1200 1400
[min]
Flow rate Head
Figure 6.29. 24 hour performance test of the CR5-11(Grundfos)

at 0.33 x V 0 with 20 wt-% ice
Whether the pump curve or the system pressure loss curve change over time cannot be
concluded from these experiments. An investigation by Frei and Egolf (2000) also found that
ice slurry becomes easier to pump over time due to a change in crystal size and shape.
6.4.5 INTERNAL BUILD UP OF ICE CLUSTERS IN CENTRIFUGAL PUMPS
The visual inspection of the flow pattern inside the housing of the pump showed some
clustering of the ice slurry during the 24 hours test (cf. Chapter 6.4.4). Figure 6.30 shows the
typical locations of the clustering. The figure indicates that the clustering forms a cone shaped
area that grows over time up to a certain amount.
Flow direction
Clustering area
Figure 6.30: Typical locations of the ice slurry clustering, photo taken from inlet side
During the long term tests the clustering did not actually cause any blockage of the pump. It
was not possible to detect any decrease in flow rate or head, and the power consumption
increased only moderately. However, the flow rate was more unstable at 30 wt-% ice than
with 10 wt-% ice (Figure 6.31). This might indicate that ice concentrations higher than 30 wt-
190
% will cause an even more unstable flow and perhaps more serious clustering and actual
blockage in the pump.
5,8
Flow rate (m³/h)
5,6 10 wt.% of ice
5,4
30 wt.% of ice
5,2
5
0 500 1000 1500
Time (min)
Figure 6.31. Flow rate over time of the pump shown in Figure 6.30
with 10 and 30 wt-% ice fraction
The higher the ice concentration and the lower the flow rate the bigger the size of the cone. At
.
30 wt-% of ice and a flow rate of 0.33 x V 0 the cone takes up almost half of the visible space
behind the housing.
6.4.6 CORROSION AND EROSION
Centrifugal pumps were tested in systems using ice slurry of ethanol (10 to 20% wt.) and
propylene glycol (16 wt-%) in water. All systems were charged with tap water. During a total
of 4 years of test period the systems have experienced a broad variety of operating conditions
including long periods of stand still charged with the water/freezing point depressant mixture
with corrosion inhibitor at ambient temperature.
The shaft seal of one pump in an ethanol system had to be replaced after approximately three
years of operation.
In most of the pumps working in the ethanol systems the impeller wheels are covered with a
thin white deposit layer. Chemical analysis revealed that the layer mainly consisted of
calcium carbonate (chalk). The layer does not appear to have any measurable effect on the
pump performance.
In storage tanks with ethanol/water mixture the precipitation of calcium carbonate also could
be observed during stand still. Therefore, it is recommended to use demineralized water to
charge ice slurry systems based on ethanol.
6.4.7 PROBLEMS AT STANDSTILL
The pumps have been tested for start-up after standstill with ice slurry. After a long standstill
period the ice inside the pump melts thus there is no start-up problem related to the pump.
191
Start-up after a short standstill (without melting of ice) was tested to reveal if separated ice
inside the pump housing would lead to increased start torque or cause internal blockage and
thereby hinder the pump to establish flow conditions. All the in-line pumps tested revealed no
start-up problems after a short standstill period.
6.5 Fittings and related pressure loss coefficients
A survey has been carried out on selected ABS fittings (manufactured by George Fischer) to
gather knowledge on pressure loss coefficients in specific geometries with single-phase
secondary fluids and ice slurry. This investigation on pressure loss coefficients of fittings was
not meant to offer a thorough scientific explanation for all types of fittings. The overall
intention of the investigation was to clarify whether or not special precautions should be
taken, and to give an overview of pressure losses in fittings when dimensioning an ice slurry
system. Generalising the results to other dimensions, media and flow conditions is not
recommended, as the experimental material is not adequate for such purposes.
6.5.1 TEST RIG SET-UP
A schematic of the test section is shown in Figure 6.32. Table 6.2 contains a list of the single
components tested. All piping components made of ABS (a thermo plastic) were
manufactured by George Fischer and assembled in accordance with the manufacturer’s
recommendations. The numbers in Table 6.2 correspond to the numbers shown in Figure
6.32.
Table 6.2. List of components
4 No Description Dimension (inner) Accuracy
1 Long bend 27.7 mm -
2 Tee (Distribution) 27.7 mm -
3 Elbow 27.7 mm -
5 6 4 Reduction 27.7 – 21 mm -
5 Elbow 21 mm -
1 7 6 Long bend 21 mm -
A 7 Expansion 21 – 27.7 mm -
8 Tee (Branching) 27.7- 27.7- 27.7 mm -
2 3 9 Tee (Branching) 27.7- 21- 27.7 mm -
10 Straight pipe (1m) 21 mm -
11 Straight pipe (1m) 27.7 mm -
A Coriolis mass flow DI 15 (280-5600 ±0.15% of
11 8
meter (Danfoss) l/hr) reading
B Electromagnetic flow DI 10 (0-2000 l/hr) ±0.13% of
10 meter (Danfoss) reading
C Differential pressure 0 - 100 mbar ±2% FS
9 transmitter
B (Yokogawa)
D Differential pressure 0 - 26 mbar ±2% FS
transmitter
(Yokogawa)
Figure 6.32. Test section used in the

experiments. The dots show the pressure
measurement locations
192
All pieces of piping were installed in one horizontal plane in order to eliminate any changes in
elevation and potential energy. In addition, the parameters related to buoyancy forces on ice
particles are also minimised when the pipes are in one horizontal plane.
6.5.2 DESIGN OF PRESSURE TAPS
The design of the pressure taps was carried out on the basis of a European standard for testing
of valves (EN 60534-2-3). The recommendations were not followed completely as it would
demand an unreasonable effort in the context of the screening type of study mentioned earlier.
The distance upstream of the fitting to the pressure tap was five times the inside pipe
diameter, di, and the corresponding distance downstream was 10 x di. The design of the
pressure taps is shown in Figure 6.33. However, the pressure taps were placed on the sides of
the plastic pipes and not on top as shown in Figure 6.33.
Figure 6.33. Pressure taps (Dimensions in millimetres.)
6.5.3 DATA TREATMENT
The following explanation gives the theoretical background for the calculation of pressure
loss coefficients. The reference for calculating the friction losses in the test section built is the
steady-state macroscopic mechanical energy balance, which under isothermal conditions and
with no work involved is given by:
p1 v1 2 p v 2
+ + g ⋅ z1 = 2 + 2 + g ⋅ z 2 + hf (6.6)
ρ1 2 ρ2 2
where subscripts 1 and 2 designate the positions of the pressure taps upstream and
downstream of the fitting, respectively.
The total friction loss, hf, was calculated as:

⎛ λ ⋅L ⎞
∑ ζ ⎟⎠
1
hf = ⋅ ρ ⋅v 2 ⋅⎜ + (6.7)
2 ⎝ d
193
The total pressure loss caused by a fitting is given by:

∆pt = ∆pu + ∆po + ∆pd (6.8)
where ∆pt given by Equation (6.8) includes the actual length of the fitting lf.
The pressure profiles upstream and downstream of the fitting is shown in Figure 6.34.
According to this figure, the total pressure loss ∆pt caused by a disturbance consists of three
contributions:
1. A small contribution from flow disturbance upstream ∆pu
2. The main contribution from flow disturbance within the disturbance itself ∆po
3. Flow disturbance downstream ∆pd.
Figure 6.34. Typical axial pressure profile along a piping system containing a fitting
As already mentioned, all pieces of piping in the test rig were laid in one horizontal plane
ensuring that there is no change in potential energy. Therefore the term g ⋅ z can be excluded
from Equation (6.6).
Rearranging Equations (6.6) and (6.7) and substituting for hf:
(p1 − p2 ) + 1 ⋅ ρ ⋅ (v1 2 − v 2 2 ) = 1 ⋅ ρ ⋅ v ref 2 ⋅ ⎛⎜ λ ⋅ L + ζ ref ⎞⎟ ⇔

2 2 ⎝ d ⎠
(p1 − p2 ) + 1 ⋅ ρ ⋅ (v1 2 − v 2 2 ) − 1 ⋅ ρ ⋅ v1 2
⎛ λ ⋅L ⎞ 1 ⎛ λ ⋅L ⎞
⋅ ⎜⎜ 1 1 ⎟⎟ − ⋅ ρ ⋅ v 2 2 ⋅ ⎜⎜ 2 2 ⎟⎟ (6.9)
2 2 ⎝ d1 ⎠ 2 ⎝ d2 ⎠
ζ ref =
1
⋅ ρ ⋅ v ref 2
2
or:
∆pm + (pdyn ,1 − pdyn ,2 ) − (∆pL1 + ∆pL2 )

ζ ref = (6.10)
pdyn ,ref
where ∆pL1 and ∆pL2 were measured in test sections 10 and 11 (see Figure 6.32) and
corrected for the actual length between the pressure taps upstream and downstream of the
fitting being examined. Equation (6.10) was used for calculating the pressure loss coefficients
shown in the figures below.
194
6.5.4 EXPERIMENTS AND RESULTS
Preliminary experiments were carried out with 15 wt-% propylene glycol in order to have a
reference for comparison of results with and without ice crystals. The results are listed in
Table 6.3.
Table 6.3. Pressure loss coefficients calculated with reference to the dynamic pressure in the
actual cross section including the length of the fitting. For the reducer, the reference dynamic
pressure is downstream of the fitting. For the expansion, the reference dynamic pressure is
upstream of the fitting. Danvak (1992) and VDI (1991) refer to the respective Danish and
German engineering reference books
Results Danvak VDI
90° bend d32 0.4 0.3

90° bend d25 0.4 0.3
90° elbow d32 0.9 .1.1
90° elbow d25 0.7 1.1
Reduction d32-d25 0.2 0.2
Expansion d25-d32 0.2 0.2
Flow rate ratio
ζ
Branch d32 with reduction (d25) c2/c1 2
Distribution d32 1 0.9 1.4

0.8 1.6 2.1
2 0.3 5.8 6.4
1 0.6 2.5 3
0.9 1.4 1.9
Branch d32 1 1.4 2.1
0.9 1.6 2.4
1
2 0.7 2.2 3.3
0.4 6 6.5
1.7 0.8 1.1
1.4 0.9 1.4
1 1 1.4 2.2
2 1.1 1.2 1.9
0.5 4.6 5.3
The measured pressure loss coefficients are generally lower than the theories suggest. Only
the measurements on the long bends seem to be somewhat higher than the theoretical values.
Two series of measurements have been carried out with ice slurry, i.e. one series with 10 wt-
% ice and the other with 30 wt-%. For each fitting tested, the flow velocities were varied
between 0.5 and 2 m/s. The measured values are compared with the values for turbulent water
flow from the literature in Figure 6.35 a-h. Note the figure captions and different scales of the
axis.
195
90° Bend 90° Elbow

3,5
Brine 27mm
10% ice 27mm
3 30% ice 27mm
Brine 21mm
2,5 10% ice 21mm
30% ice 21mm
2 Theory
[-]
1,5 1,1
1
0,3
0,5
0
0 0,5 1 1,5 2 0 0,5 1 1,5 2
Flow velocity [m /s] Flow velocity [m /s]
a) b)
Reduction d32-
Contraction 27-21 Expansion d25-
Expansion 21-27
1,0
Brine
0,8 10% ice
30% ice
0,6 0,2
[-]
0,2
0,4
0,2
0,0
0 0,5 1 1,5 2 2,5 3 0 0,5 1 1,5 2 2,5 3
Flow velocity outlet [m /s] Flow velocity inlet [m /s]
c) With ref. to the dynamic pressure dow nstream d) With ref. to the dynamic pressure upstream
Tee
16 4
Brine 27mm
2 2
12 1 10% ice 27mm 3 1
30% ice 27mm 1,4
[-]
[-]
8 2
4 1
0 0
0,0 0,2 0,4 0,6 0,8 1,0 1,2 0 0,5 1 1,5 2
Flow velocity ratio c2/c1 Flow velocity outlet [m /s]
e) Flow velocity upstream 1,5 m/s constantly f) c 1/c 2 = 1
Tee
12 4
Brine 27-27-27
9 10% 27-27-27 3 1
30% 27-27-27 2
1,1
[-]
[-]
6 Brine 27-21-27 2
10% ice 27-21-27
3 30% ice 27-21-27 1
0 0
0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8 0 0,5 1 1,5 2 2,5 3
Flow velocity ratio c1/c2 Flow velocity outlet [m /s]
g) Flow velocity upstream 1,5 m/s reduction, and 1,7 w ith reduction
h) c 1/c 2 = 1 w ithout contraction, contr.
Figure 6.35 a-h. Experimental results on pressure loss measurements in ABS-fittings
196
In general, the brine results are very close to the values found in the literature and relatively
independent of the flow velocity as expected. The loss coefficients for ice slurry are the
highest for the combination of high ice concentration and low flow velocity. The largest
deviation from turbulent flow theory occurs for the 90° bend, for which the loss coefficient
for 30 wt-% ice is approximately eight times higher than that for 0% ice.
When the flow velocity is increased, the values of the loss coefficients for ice slurry
asymptotically approach the tabulated values. This phenomenon is related to flow
disturbance, turbulence in general and reforming of the flow profile.
Figure 6.35 d shows the results for the experiments carried out on an expansion. It appears
that the loss coefficient can be neglected when the ice concentration is high. The explanation
might be that the ice crystals form a venturi-like diffuser from the small pipe diameter to the
larger and thereby minimise the disturbance. See Figure 6.36.
Ice deposit
Figure 6.36 The left figure shows a sketch of the possible flow pattern
with ice slurry when the flow is highly turbulent, while the right sketch shows
a possible flow pattern when the flow is laminar or of Bingham type
Theory indicates that particularly in the laminar case the loss coefficients depend on the
Reynolds number. This relation is investigated and the result is shown in Figure 6.37.
3.5
3.0 Bend d32; 10%

2.5 Bend d32; 30%
Bend d25; 10%
2.0
Bend d25; 30%
1.5
Propylen glycol
1.0
0.5
0.0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Re [-]
Figure 6.37. Measured pressure loss coefficients for 90° bends with 10 and 30 wt-% ice
fraction in a water/glycol carrier fluid. The diagram shows a clear relation
between the loss coefficients and the Reynolds number.
As the Reynolds number is increased the value of the loss coefficient approaches
the tabulated value. Even the experiments without ice slightly show this trend
197
In the Reynolds number range of about 1500 – 4000 the experimental data deviate relatively
more from the trend line than the rest. Analysing the data more deeply reveals that the effect
of the ice concentration on the loss coefficients more significant than those caused by
viscosity change and subsequent change in Reynolds number. Figure 6.38 shows a 3-
dimensional plot of the loss coefficients with the Reynolds number on the x-axis, the ice
concentration on the y-axis and the pressure loss coefficient ζ on the z-axis.
Bend
Rank 22 Eqn 3568 z=a+b/x^(1.5)+cy^2
r^2=0.96472678 DF Adj r^2=0.96283715 FitStdErr=0.14338074 Fstat=779.47847
a=0.25076052 b=37639.105
c=0.00074990389
3
3
2.5
2.5
2
2
Zeta
1.5
Zeta
1.5
1
1
0.5
0.5
0
0 25
0 20
1 0 02 0 0 0 0 0 0 15
trat
ion
3 0 4 00 0 0 0 10 en
50 6 0 0 00 0 5 onc
Re ic 7 0 8 00 0 0 0 0 Ic ec
e 9 0 0 00
1
Figure 6.38. The loss coefficients plotted against the Reynolds number and the ice
concentration. This correlation has a R2 value of 0.96 compared to 0.84 for the correlation
shown in Figure 6.37. Therefore it is concluded that the presence of solid particles in the fluid
influences the loss coefficient negatively by a factor of 2 to 3
6.5.5 COMPARISON OF MEASURED VALUES FOR ICE SLURRIES AND SINGLE-PHASE FLUIDS
A number of selected fittings have been investigated with a focus on pressure loss
coefficients. Experiments with single-phase fluids showed good agreement with the literature.
With ice slurry, the picture is somewhat different. In some of the fittings tested, the pressure
loss coefficients showed a clear dependency on the ice concentration, but also a great
dependency on the flow velocity. The largest deviation between the single-phase and ice
slurry results occurs for a combination of high ice concentration and low flow velocity. In this
case, the pressure loss coefficient can be up to eight times larger than that for the
corresponding single-phase fluid. If, however, the loss coefficients are plotted against the
Reynolds number this increase is not so noticeable. The increase is instead a factor of two to
three.
The main aim of the investigation was to reveal how much attention should be paid to
pressure drop in fittings with ice slurry. Several of the most frequently used fittings have been
investigated and the conclusion is, that in a widely branched distribution system the pressure
loss caused by fittings might be important and a detailed analysis should therefore be carried
198
out. The correlation is implemented in IceProgTM (Danish Technological Institute’s

calculation software package) for all the investigated fittings, and this piece of software is a
useful tool in the analysis of pressure drop in an ice slurry distribution system.
6.6 Valves
Based on operating experience, fittings and valves can be either the source of agglomeration
problems or the recipient of ice formations originating from upstream or downstream
agglomeration sources in the transport system. All traditional valves (butterfly, ball,
diaphragm, etc.) are known to cause flow pulsation problems at low opening angles where ice
can build up and gets eventually released as the shear stress increases. Valves with low
opening angles are also ideal traps for agglomerated ice.
6.7 Insulation
Insulation should be designed and applied as with conventional evaporating refrigerants. Note
that the heat transfer from the melting ice slurry is 2 to 3 times higher than from a
corresponding single-phase liquid.
6.8 Critical conditions
Improper operation/design of homogeneously mixed storage tanks can lead to the formation
of ice slabs and in the course of time, large ice blocks may develop in the storage tank
especially with access of air to the top. The presence of ice blocks implies a potential risk of
transport system failure, e.g. if an ice block is induced into the pump suction line.
In the transport system, ice slurry does not cause blockage due to moderate flow obstacles as
long as a certain (sufficient) pressure head is available. If the flow is stopped and the ice is
allowed to separate, then the nominal pump head in the scenarios investigated was sufficient
to allow re-establishment of the flow.
One flow condition that must always be fulfilled in order to reduce the potential blockage is
that the flow velocity in every line is above the separation velocity. All the incidences of flow
pulsation and blockages observed were found in the furthest downstream pipe sections, i.e.
the pipe section with the lowest flow rate. The problematic flow behaviour in all these
scenarios were caused by low flow rate (separated, moving bed flow conditions) and in most
cases in combination with partially closed valves.
It can be concluded that improper design of ice slurry transport systems together with lack of
consideration of all possible flow conditions during operation can lead to unpredicted
pluggages even in unused lines. Once plugging occurs it might not be easy to unplug the
system again.
The findings of unstable flow and blockages in downstream branches are important to
consider in the design of ice slurry transport systems and heat exchanger control loops. It
must be ensured that the flow rate under no condition of part load drops below a lower limit.
Individual circulation pumps might be desirable to provide sufficient pressure to maintain the
flow rate through partially closed control valves in parallel branches. Alternatively, variable
speed pumps can replace some of the valves and eventually lead to a valveless system. Such a
199
system will also reduce energy consumption because throttling losses in valves are reduced or
eliminated.
The larger the pipe diameter, the more attention must be paid to the potential risk of blockage,
due to the longer time needed for melting blockages.
The findings also raise questions about the application of speed controlled pumps in ice slurry
systems together with valves. Variable flow capacity control should be considered carefully if
applied and it seems that constant pressure control of pumps offers a more reliable and safe
alternative in conventional systems.
From the results it cannot be concluded if any valve types are more suitable in ice slurry
installations than others. However, flow fluctuations and blockages were observed in
branches containing diaphragm, ball and butterfly valve.
In general, the valve types investigated should be used only as open/shut valves and not as
control valves.
6.9 Quick guide to ice slurry installation and operation in order to avoid blockage
To minimise potential risk of operational failure, the following non-exhaustive checklist

might serve as a guide: The checklist has been divided into operational problems related to
the storage system and the transport system. The checklist is cited from ASHRAE RP 1166
(Hansen, Radosevic, Kauffeld, 2002).
6.9.1 HOMOGENEOUSLY MIXED STORAGE
o The outlet should not be located at the bottom.

o Precaution must be taken to prevent air uptake if stirring devices are used.
o Foaming on free surfaces can cause problems when using glycol.
o Do not apply mixing baffles in the storage vessels.
o Avoid pipes and other objects projecting through the open surface of a storage vessel.
o Examine surface flow conditions carefully and eliminate stationary zones.
o Consider a coarse filter net with a mesh size of 0.15 to 0.2 inch (4-5 mm) on the outlet
pipe.
6.9.2 TRANSPORT SYSTEM
o Avoid low flow velocity (below the separation velocity) in all pipe sections at all
times.
o Use caution if applying speed controlled pumps – ensure the flow and the pressure at
any possible point of operation are always well above the specified minimum values
determined by the flow separation velocity and the specific valve pressure drop
characteristic.
o Avoid direct vertical branches with low flow rate – if possible branch off sideways or
downward before bending to a vertical riser.
o Flows in reductions, expansions, bends and T-junctions are seldom critical.
o Mount shut off valves at the lowest possible location in vertical pipes.
200
o Ball valves and butterfly valves should only be used in fully open and fully shut
positions.
o Use caution if employing diaphragm, gate and globe valves as control valves.
o Clam-shell valves might be the safest choice for a control valve.
Literature cited in chapter 6
1. Choi, D.M.; Knodel, B.D.: Impact of Advanced Fluids on Costs of District Cooling
Systems, 83rd Annual Conference of the Int. District Heating and Cooling Association,
Boston, MA, 17 pages, June 1992.
2. Christensen, K.G.; Kauffeld, M.: Ice Slurry Accumulation, IIR Conference: Natural
Working Fluids ’98, Oslo, 6/1998.
3. Danvak: Varme og klimateknik, Grundbog, pp. 422-425, 1st edition, Danvak ApS,
Copenhagen, 1992.
4. European standard, EN 60534-2-3.
5. Frei, B; Egolf, P.W.: Viscometry applied to the Bingham Substance Ice Slurry, Second
IIR Workshop on Ice Slurries, Paris , May, 2000.
6. Fukusako, S.; Kozawa, Y.; Yamada, M.; Tanino, M.: Research and Development
Activities on Ice Slurries in Japan, First IIR Workshop on Ice Slurries, Yverdon-les-
Bains, Switzerland, 1999.
7. Hansen, T.M.; Kauffeld, M.; Grosser, K.; Zimmermann, R.: Viscosity of Ice Slurry,
Second IIR Workshop on Ice Slurries, Paris, France, May 2000.
8. Hansen, T.M.; Radosevic, M.; Kauffeld, M.: Behaviour of ice slurry in thermal storage
systems, ASHRAE Research Project - RP 1166, February 2002.
9. Kawada, Y.; Shirakashi, M.; Takizawa, S.: Characteristics of Ice/Water Mixture Flow in
a Branching Pipe and Development of an Ice Fraction Control Technique, J. of Japanese
Society of Snow and Ice, Vol. 58, No. 5, pp. 405-415, 1996.
10. Kitanovski, A.; Poredos, A.: Theory on concentration distribution of the ice slurry flow,
Fourth IIR Workshop on Ice Slurries, Osaka, Japan, Nov. 2001.
11. Langgard, G.: Grundfos Management, Personal Communication, 2001.
12. Larkin, B.; Young, John C. O’C.: Influences of Ice Slurry Characteristics on Hydraulic
Behaviour, pp. 340-351, 1989, Proc. 80th Annual Conf. IDHCA. 340-351.
13. Liu, K.V.; Choi, U.S.; Kasza, K.E.: Measurements of Pressure Drop and Heat Transfer in
Turbulent Pipe Flows of Particle Slurries, Argonne National Laboratory, May 1988.
14. Nørgaard, E.; Christensen, K.C.; Hansen, T. M. et al.: Heat transfer and pressure drop
with ice slurry, 4th IIR Gustav Lorenzen Conf., Purdue, July 2000.
15. Snoek, C.: The Design and Operation of Ice-Slurry Based District Cooling Systems,
International Energy Agency Programme of Research, Development and Demonstration
District Heating and Cooling, November 1993.
16. Turian, R.M. et al.: Flow of concentrated Non-Newtonian Slurries: 2. Friction losses in
Bends, Valves, and Venturi Meters, Int. J. Multiphase Flow, Vol. 24, No. 2, pp. 243-269,
1998.
17. VDI Wärmeatlas, chap. L, VDI Verlag, Düsseldorf, 1991.
201
CHAPTER 7. STORING/MELTING AND MIXING

Yoshiyuki Kozawa, Robert Huhn, Torben M. Hansen, Michael Kauffeld
Storage of ice slurry is very important technology, since the latent heat of the ice particles
greatly increases the capacity of thermal storage systems. Basically all ice slurry systems
contain an ice storage vessel of one form or another. The general system operation is to
charge a storage vessel by the ice slurry generator and to discharge the same vessel by
pumping ice slurry, dry ice crystals or liquid without ice particles to the point of use and
returning the melted/heated solution. In certain food applications ice slurry is used with the
product to be cooled and is not returned to the storage vessel. Both mixed (agitated) and
unmixed storage vessels are used in ice slurry installations.
7.1 Storing vessels
7.1.1 STORING METHODS/DEVICES AND PROCESSES
One way of categorizing ice slurry thermal storage systems is to distinguish between stores
with heterogeneous and homogeneous compositions of ice particles. Another possibility is to
refer to the possible extraction of ice slurry or chilled water from the storage.
The choice of storage needs to consider:

• High net useable storage capacity or discharge capacity
• High nominal storage capacity or overall ice concentration (wt-%)
• Low installation cost and minimum required maintenance
• Operational reliability (Hansen et al., 2002)
None of the combinations shown in Figure 7.1 offers superiority in all of these criteria.
Whether a heterogeneous or homogeneous storage principle should be chosen depends on the
specific application, available storage volume, desired temperature and type of ice slurry
generator (Hansen et al., 2002).
The shapes of the storage tanks are mostly cylindrical or cubic. Often existing ice water stores
are retrofitted as ice slurry stores (Huhn, 2003).
203
Important criterias
Crunched ice Vacuum, Scraped
Supercooled Cooling capacity
Ice generation from absorption & surface
water Reliabilty
harvesters others generators
Control
Stability
Economy
Storage inlet Middle of Bottom of

(return from transport Spray Reliabilty
Top of tank tank tank
system) Control
Stability
Heterogeneous Homogeneous
storage storage Max. ice content pr.
up to 50 wt.% ice up to 30 wt.% ice storage volume
Discharge rate
Reliability
Control
None Pump Propeller Maintainance
Agitation Economy
Control
Pure ice harvest Reliability
Chilled water with scraper at Ice slurry Economy
Extraction
top or bottom Maintainance
from tank
Ice slurry properties

Chilled water Ice slurry Reliability
Transport media Maintainance
Economy (reduced
pipe diameter)
Figure 7.1 Overview of different categories of storage strategies (Hansen et al., 2002)
7.1.1.1 Heterogeneous storage

Ice particles will separate from the carrier liquid because of the density difference. If no
special means of mixing is applied, the ice composition in the storage tank will be non-
uniform. In the upper part of the storage tank, pure ice without any carrier liquid can be
obtained. In the bottom part of the storage tank, a zone with low or zero ice concentration can
exist (Figure 7.2 b, d).
The overall ice concentration in these types of stores may be as high as 60 wt-%. The main
drawback of high concentration partial storage is the means of ice slurry extraction. If the
initially produced ice consists of small particles, e.g. 0.1 mm, then ice particles in a size range
of 1 to 2 mm can be harvested from the top (Figure 7.2 b). The harvested pure ice must
afterwards be mixed with pre-cooled coolant in an agitated mixing tank before distribution by
the transport system. Alternatively, a warm water return or spray may be applied (Figure 7.2
d, e, f) at a lower discharge rate. Because only chilled water can be extracted, there is no
benefit of savings in pipe dimensions. The level of floating plugs of ice can be controlled by
fluidising the bottom part of storage tank and ice slurry can be extracted with a bottom
harvester (Figure 7.2 c).
7.1.1.2 Homogeneous storage

In a homogeneous storage tank, the ice particles are continuously stirred by mechanical
agitation (propeller, pumps, paddles, etc.), as shown in Figure 7.2 a). Applying air-injection
204
has proved to be inefficient and can also lead to severe corrosion problems in the piping
systems. Stabilizing the fluid with homogenising additives has been tested at the Technical
University of Denmark (DTU) (Riber, 1999). The major drawback of these additives is a
significantly increased viscosity, and hence the required pumping power in the distribution
system by far exceeds the power consumption of the agitation devices. The details of agitation
devices are described in section 7.2.1.
d) Partial storage, supercooled water, warm spray

a) Homogeneous stirred storage
return. Maximum ice content approx 40 wt.%
Maximum ice content approx 30-35 wt.%
e) Partial storage, immiscible cold fluid, warm spray

b) Partial storage, top harvester and mix tank return. Maximum ice content approx 40 wt.%
Maximum ice content approx 50-60 wt.%
Warm water Ice slurry supply
Chilled water
extraction
c) Partial storage, bottom harvest and fluidiser f) Pressurized flooded storage, warm water top return
Maximum ice content approx 45 wt.% Maximum ice content: Unknown
Figure 7.2. Examples of some typical storage layout principles (Hansen et al., 2002)
The ice storage process is shown in Figure 7.3, taking as an example a heterogeneous storage
system. When the temperature of the water inside the tank reaches about 0°C, the ice slurry is
supplied and fine ice particles float in the tank as pictured in the left drawing (1). Because
they have different densities from water, these ice particles form a sherbet-like ice-rich layer
(the shaded portion in the figure) which floats in the tank as shown in picture (2). On the
upper surface of the ice-rich layer, while liquid water permeates into the layer, the ice slurry
flows from a supply opening in the centre of the tank toward the side walls of the tank, and
205
the ice-rich layer spreads out horizontally. At the same time, with the water permeating
through the ice-rich layer, ice accumulates on the upper surface of the ice-rich layer,
successively increasing the thickness of the ice-rich layer, which grows toward the bottom of
the tank. After the ice-rich layer reaches the bottom of the tank as shown in picture (3), ice
gradually piles up above the initial water level without the ice-rich layer floating, and the
thickness of the ice-rich layer increases further.
Suply Slurry Ice Slurry Ice

Opening Ice Storage
Tank Side Wall
Initial
Water revel
Ice Granules Ice-rich Ice-rich
Layer Bottom
Layer
Inlet of Inlet of of tank Inlet of
Slurry Ice Water 0 Ž Slurry Ice Slurry Ice
Water Outlet of Outlet of
Outlet of Water
Water Water Water
(1) ⇒ (2) ⇒ (3)
Figure 7.3. Ice storage process in a tank (seen from the side)
7.1.2 MELTING METHODS/DEVICES AND PROCESSES
One method to melt the stored ice and utilise the stored thermal energy, in order to supply
low-temperature cold water to, say, an air-conditioning load, is the spray method in which
spray nozzles are arranged in the top part of the ice storage tank.
A model of the ice melting process is shown in Figure 7.4 a, where η is the thermal storage
energy utilisation rate. Water, whose temperature has been raised by the air-conditioning
load, is uniformly sprayed from spray nozzles onto the stored ice-rich layer in a tank, as
shown in picture (1). The ice is melted mainly inside the ice-rich layer near the sprayed water
surface, and the apparent porosity of this part of the ice-rich layer increases. The water
generated by the melting of this ice permeates through the ice-rich layer and melts the ice
farther down in the ice-rich layer. As a result, as time passes, the thickness of the ice-rich
layer decreases, as shown in picture (2). When the ice-rich layer thins to a thickness of about
0.1m as shown in picture (3), the ice-rich layer splits into a number of lumps of ice, the
apparent porosity of the ice-rich layer further increases, and the ice melts rapidly. A melting
process of this type proceeds while generally maintaining one-dimensionality in the height
direction. The other method used to melt the stored ice is the jet method as shown in Figure
7.4 b. The stored ice melts due to a jet of water and mixing of water and ice in the tank.
The choice between these two methods depends upon consideration of heat transfer
characteristics in the tank and required pumping power, etc.
206
Spray
Water Nozzle Water Water
Inlet Inlet Inlet
Moving
Initial Permeating Water
Bottom Ice-rich Ice Storage
Water Ice-
of Tank Layer Tank
Level Layer
Ice-rich Water
Layer Water
Water Water Water
Outlet Outlet Outlet
(1)η < 0.3 (2)η = 0.6 (3)η > 0.9
(a) Ice melting process in the spray method (lateral view)
Outlet
Ice-rich
Layer
Ice-rich
Layer
Jet
Water Flow
Moving
Inlet
Jet Nozzle
(1)η < 0.3 (2)η = 0.6 (3)η > 0.9
(b) Ice melting process in the jet method (seen from above)
Figure 7.4. Ice melting process in a tank
7.2 Mixing tanks
7.2.1 MIXING METHODS/DEVICES AND PROCESSES
Mixing is applied in homogeneous storage tanks to avoid separation of the ice from the liquid
phase. In heterogeneous storage tanks with harvested pure ice, external mixing of the ice with
pre-cooled coolant in an agitated mixing tank is required before distribution by the transport
system.
Some examples of common mechanical agitation devices are:
– Propellers with flat inclined blades or 3D-shaped blades
– Agitators with cross bars
– Agitators with a special shape
Agitators can be installed vertically from above or horizontally through a side wall, or in the
case of an immersed agitator, they are directly located in the storage medium and hence the
driving unit must be totally sealed. A disadvantage of the latter option is that all the driving
power including waste heat is dissipated as heat into the ice slurry.
207
A complete circulation of the tank contents is necessary for homogenous storage. That can be
achieved with large diameter agitators at low rotational speeds and moderate power input. The
mean flow in the tank can be supported by jet nozzles. There is no application known which
exclusively uses jet nozzles for mixing in a homogeneous tank.
The specific power input decreases with an increasing diameter ratio, Dagitator / Dtank , whereby
the torque of the agitator shaft increases at the same time. Also the agitator location (height
and distance from the centre of the tank) is important for its effectiveness.
In cylindrical storage tanks with a centrally arranged agitator, special baffles might be
necessary in order to achieve complete mixing. Tanks of cubic shape do not need these
baffles, since the flow direction will be sufficiently disturbed by the tank walls resulting in
good mixing.
An upper limit on the average ice fraction in propeller-mixed storages is found to be 25 wt-%
in available literature (Kakutami and Noburo, 1999). Up to 35 wt-% ice fraction has been
reached by optimisation of propeller design and location as well as the propeller speed
The main advantages of homogeneous ice slurry storage are reduced agglomeration of ice
crystals and the easy discharge of the slurry. Good discharge performance and very good heat
transfer rates are achieved due to the large specific surface area of the small ice crystals.
Because of the limitation on the average ice fraction, homogenous ice storage systems have to
be designed with a larger storage volume than heterogeneous storage systems.
7.2.2 STRATIFICATION IN MIXING TANKS
In the following, stratification is defined as the separation of ice crystals and liquid as a result
of different densities of the components. The ascent of the lighter ice crystals within the liquid
proceeds all the time until stable stratification has been reached. This steady state condition
implies that all ice particles are at the top of the tank and the maximum packing factor has
been reached. That means each particle is in direct contact with the adjacent particles and not
enough space exists between the particles to get any closer. This maximum packing factor
depends on the crystal shape and size distribution. For spherical particles of uniform size, it is
74%. For ice slurry the value might be higher due to the dish-like shape and non-uniform
size-distribution.
There is a lack of knowledge on whether the particles and their shape in this packed layer are
affected by the compressive force, which arises from the buoyancy force on the single
particles.
Stratification occurs in ice slurry tanks if no efforts are undertaken to keep the slurry mixed.
That might be the case for a stationary agitator. The density driven process of separation of
ice crystals and liquid depends on the following parameters:
– density difference between ice and liquid
– ice crystal size and shape
– viscosity of the liquid
– overall ice concentration
208
For ascending ice particles in an unmixed storage tank laminar flow prevails. In an agitated
storage tank, stratification can occur if the agitator is operated at too low a speed and power.
Then usually a three-dimensional turbulent flow exists. For this case (Re > 104) the shape of
the particles has only very little influence on the rate of ascent (Schempp, 1996).
Particle distributions in a tank also have been calculated by Egolf et al. (2001) using a multi-
component approach.
7.2.3 POWER CONSUMPTION IN NEWTONIAN RANGE
Propellers have been tested at the Danish Technological Institute (Hansen et al., 2002) and the
power consumption was approximately 70 W per m³ storage volume. The required mixing
power rises significantly above 30 wt-% ice and the practical limit for most applications
therefore is considered to be between 30 and 35 wt-%.
Values for specific power consumption in actual ice slurry applications are rather higher than
the 70 W per m³ storage volume stated above (Huhn, 2003).
7.2.4 POWER CONSUMPTION FOR THE HIGHLY-VISCOUS RANGE
Although the transition from the Newtonian to the Bingham behaviour has been reported to
depend on both ice concentration and additive concentration (Nørgaard et al., 2000; Frei and
Egolf, 2000), a widely applicable map of the transition from Newtonian to Bingham
behaviour, and the magnitude of shear stress required are still lacking (Hansen et al., 2002). In
the available literature, different values for the transition from the Newtonian range to “high-
viscous range” are found, varying between 20 and 30 wt-%.
Higher ice fractions up to 52 wt-% in agitated homogeneous storage systems have been
achieved at the Danish Technological Institute by applying a special spiral-shaped stirring
device. Non-Newtonian behaviour can be assumed for this case. Mixing was possible, but a
strongly increased power input was required compared to the case of ice fractions within the
Newtonian range (up to 30 wt-%) and mixing with a propeller. An increase by a factor of 10
was reported.
7.3 Numerical modelling of storage tank design
A numerical model has been applied to an actual design of supercooling-type ice storage
systems (Tanino et al., 1995, 1997, 1999 and 2001).
7.3.1 PHYSICAL MODELLING OF ACCUMULATION/PILING AND MELTING
The accumulation and piling of ice was modelled, as shown in Figure 7.5, in order to
determine the overall properties of the ice storage process (the contour of ice-rich layer) and
to make use of them in the actual design (Tanino et al., 1997; Kozawa and Tanino, 1999;
Kozawa et al., 2001; Tanino and Kozawa, 2001). A two-dimensional cylindrical coordinate
system was adopted, and the following assumptions were made:
209
(a) There is no heat transfer

(b) Flow in the water region inside the tank is not considered.
The time history of the contour of the ice-rich layer can be calculated, considering the mass
conservation in the upper part of the ice-rich layer. The calculation procedure is as follows;
(1) Calculate the velocity distribution of the water permeating through the ice-rich layer.
(2) Estimate the quantity of ice accumulated in a given time interval.
(3) Consider the deformation and increasing compactness of the ice-rich layer.
(4) Determine the contour of the ice-rich layer, and repeat steps (1) to (4) at a new time
step.
Center axis Center axis

z z
Supply opening Water velocity Ice piling
Ice piling on upper on upper surface
u
surface Mz(r)
Ice growing
v v
Ice toward perimeter
slurry v Mr Ice growing up
moving Ice granules bed toward the upper space
Ice
content Initial
αice=2.5% Ice-rich Water level
layer Shear stress Packing by self-weight
G τi
Side Side
Water wall wall
Deformation by buoyancy
and self-weight
Ice-rich layer
In the only water region,

the effect of flow is not Radius
considered Radius
axis axis
r r
(1) (2)
Figure 7.5. Overview of ice storage model in a tank
In the ice melting process, stored ice particles in the ice-rich layer melt, and the ice layer is
rebuilt by breaking up and re-agglomerating, causing non-uniform thermohydraulic behaviour
within the ice-rich layer. It is most important for the actual design to determine how the
temperature of the cold water changes with time at the outlet of the tank.
A one-dimensional model (Kozawa and Tanino, 1999; Tanino and Kozawa, 2001; Tanino and
Kozawa, 2001) was adopted in the water depth direction as shown in Figure 7.6. In this model
of the ice melting process, it was assumed that:
(a) The ice particles in the ice layer are spherical in shape
(b) Within each calculation element, the diameter of the ice particles changes with time,
but they are spatially uniform
(c) The number of ice particles within each calculation element is temporally constant
(d) Water is completely mixed in the water region at the bottom of the ice layer.
The time history of the cold water temperature at the outlet of the tank, in a computational
210
element, can be calculated using relationships between length, porosity, and ice particle
radius. The calculation procedure is as follows:
(1) Calculate the amount of ice melting within the ice-rich layer.
(2) Calculate the spatial distribution of the temperature of the water permeating through
the ice-rich layer, using the mass and energy conservation equations within the ice-
rich layer.
(3) Estimate the spatial distribution of porosity and ice particles radius within the ice-
rich layer.
(4) Calculate the temperature at the outlet of the tank.
(5) Determine the water temperature at the inlet of the tank, and repeat steps (1) to (5) at
a new time step.
Water Inlet
Ice Particle Spray
Ice-rich Layer Tin Tin nozzle
( 0 Ž)
u r,ε m Tl
l
Thickness of ice-rich layer
r0 u
Ice melting m Tl
lo
ε0
l
Tl0 in ice-rich layer
u
Thickness of ice-rich layer
Water permeating m Tl
u
l
r0 in ice-rich layer
ε0 Tl0
lo
Water cooling in u
ice-rich layer m
l
Tl
u r,ε
r0
lo
ε0 Tl0
Only-water region
r : Radius of u(b) T (b)
an ice particle Tl : Water temperature
ε : porosity of u : Water velocity
u ice-richlayer
r0
lo
ε0 Tl0
Water mixing in
Water
only-water region
l : Calucration Tout Water Outlet Tout m : Mass rate of
element ice melting
Time = 0 Time = t
Figure7.6. Overview of the ice melting model in a water-spraying system
7.3.2 MATHEMATICAL MODELLING
For the accumulation and piling model of the ice storage process in Figure 7.5, Equation (7.1)
expresses the mass conservation in a time interval (∆t) at the upper surface of the ice-rich
layer, when the ice slurry supplied at the upper part of the centre axis flows in the radial
direction (Figure 7.5) along the upper surface of the layer:
αiceG∆t = ∑ rr =0 Mz (r ) + Mr (7.1)
Here, Mz(r) is the amount of ice accumulation in the vertical direction (the z direction in
Figure 7.5) during time ∆t on the upper surface of the ice-rich layer. Mr is the amount of ice
growing in the r direction during time ∆t along the perimeter of the ice-rich layer. The symbol
211
α denotes the content of the ice slurry and G is the mass flow rate of the ice slurry supplied
from the supply opening.
If the velocity of water permeating downward from the upper surface through the ice-rich
layer is determined, the amounts of ice Mz(r) and Mr can be calculated. A continuity equation
for water permeating through the ice-rich layer can be written as follows:
∂u u ∂v
+ + =0 (7.2)
∂r r ∂z
Here, u and v are permeated water velocities through the ice-rich layer in r and z directions,
respectively.
The permeating water flow through the ice-rich layer is assumed to obey Darcy’s law, thus
the velocities are expressed as follows:
κ ∂p κ ∂p κ
u=− (7.3a), v=− − ρl g (7.3b)
µ l ∂r µ l ∂z µ l
where p is the pressure, g is the gravitational acceleration, and κ is the permeability of water
in the ice-rich layer, and µl and ρl are the viscosity and the density of water, respectively.
The boundary conditions for the pressure can de determined as follows:
p = p0 (interface between the ice-rich layer and the atmosphere)

p = p0 – ρlgh (interface between the ice-rich layer and the water)
dp/dr = 0 (central axis of symmetry)
where h is the water depth and p0 is the atmospheric pressure.
The permeability of water in the floating ice-rich layer was determined from experiments by
using a test cylinder (Tanino et al., 1997). In this model a constant value for the permeability
(κ = 1.4 × 10-9m2) was adopted from the experimental data.
A balance of forces in the floating ice-rich layer is considered to yield the deformation of the
ice-rich layer due to the buoyancy force and the weight of the floating ice-rich layer. This
means that the floating ice-rich layer is not solid but deforms by the shear stress due to the
buoyancy and the weight of the cross section in the z direction. Separate cylindrical shells of
the floating ice-rich layer are assumed to allow the movement in the z direction, where its
buoyancy and its own weight are in balance. Equation (7.4) is used to calculate the shear
stress.
τ i = Fi / 2πrili (7.4)
( )
Fi = ∫ r p 2π rl a (1 − ε )ρ i gdr − ∫ r p 2π rl b (1 − ε ) ρ l − ρ s gdr
ri ri
where τi is the shear stress. The quantity ε is the porosity in the floating ice-rich layer (ε =
0.85), ρs is the density of ice, a subscript i denotes a position in the r direction, subscripts a
and b represent the heights of the floating ice-rich layer above and below the water level, and
a subscript p denotes the perimeter of the ice-rich layer.
212
It is assumed that the shear stress of 4.9 Pa can divide the floating ice-rich layer, which then
leads to rearrangement and marching of the ice-rich layer.
In the ice storage process, after the floating ice-rich layer reaches the bottom of the tank, the
layer is compressed due to its own weight of accumulated ice. From ice packing experiments
using test cylinders, the change in porosity due to the pressure by the weight of the
accumulated ice-rich layer is determined as follows:
ε = 0.85 − 5 × 10 −5 ⋅ p press (ppress <10 × 103 Pa) (7.5 a)

ε = 0.35 (ppress >10 × 10 Pa)
3
(7.5 b)
where ε is the porosity of the ice-rich layer and ppress is the pressure given by the weight of the
accumulated ice-rich layer.
It is assumed that the relationship between the porosity and permeability of the compressed
ice-rich layer can be described by the following equation:
a
⎧⎛ ε ⎞
3
⎛1 − ε f ⎞
2
⎫
⎪ ⎪
κ = κ f ⋅ ⎨⎜⎜ ⎟
⎟
⎜
⎜ 1− ε
⎟
⎟ ⎬ , a = 0.1 (7.6)
⎪⎩⎝ ε f ⎠ ⎝ ⎠ ⎪⎭
where εf is the porosity of the floating ice-rich layer (εf =0.85) and κf is the permeability of
water in the floating ice-rich layer (κf =1.4 × 10-9 m2).
For the model describing ice melting in Figure 7.6, the changes in the length and porosity of
the ice-rich layer and radius of the ice particle are calculated using the mass conservation
equation (7.7). In this manner the reduction in the amount of ice in a computational element
of the melting ice-rich layer is calculated.
M S = ρ S ⋅ l ⋅ A ⋅ (1 − ε ) = πr 3 nρ S
3
(7.7)
4
Here Ms denotes the mass of ice, l is the length of the computational element, ε is the porosity,
r is the radius of the ice particles, n is the number of ice particles of radius r in the
computational element, which is assumed to be constant in the ice melting process, A is the
cross-sectional area of the tank and ρs denotes the density of ice.
The relationship between the length and the porosity in the computational element is as
follows:
1 − ε0
ε = 1− l (7.8)
l0
Here, a subscript 0 denotes the initial condition, and n / (l•A ) = 1.6×105 for ε0 = 0.9.
Because the temperature of ice is 0 °C in the supercooling-type ice storage system, the energy
conservation in the ice-rich layer can be described by Equation (7.9), in which the thermal
conduction term is neglected:
∂Tl ∂T
εcl ρl + εcl ρ l u l =
m
(clT + hL ) (7.9)
∂t ∂z εAl
213
Here Tl and u are the temperature and velocity of the permeating water, respectively, m is the
quantity of water produced by melting of ice in the ice-rich layer, cl is the specific heat of
water, ρl is the density of water and hL denotes the latent heat of ice.
The mass conservation in the ice-rich layer can be written as follows:
∂u m ερl dz
ερl = + (7.10)
∂z Al l dt
Here dz/dt denotes the change in height due to the melting of ice.
The time history of the temperature of the cold water taken from the ice storage tank can be
calculated according to a heat balance in the water region below the ice layer. It is assumed
that the water is completely mixed in the region. Then it follows:
∂H
(Tout − Ts ) + H ∂Tout = u( b )Tl ( b ) − uoutTout (7.11)
∂t ∂t
where Tout is the water temperature below the ice-rich layer at the outlet of the tank, H is the
height of the water region below the ice-rich layer, Ts is the temperature of ice (0 °C), and a
subscript (b) denotes a position in the boundary between the ice-rich layer and the water
region.
In the model the most important step is to estimate the amount of melting ice in the ice-rich
layer correctly. The amount of melting ice is calculated as follows:
dM S α p Ap ( Tl − TS )
m=− = (7.12a)
dt hL
λl
αp = ⋅ Nu (7.12b)
2r
Here αP is the heat transfer coefficient between the ice particles and the water, Ap is the
surface area of ice particles in the calculation element, λl is the thermal conductivity of water
and Nu is the Nusselt number for a sphere.
Microscopically the interior of the ice-rich layer is not uniform at all positions; there exist
"water paths" along the side wall of the tank and within the ice-rich layer, and the flow of
permeating water within the ice-rich layer varies with time and space. Thus, how the area of
effective heat transfer surface changes with respect to the quantity of water produced in the
ice-rich layer was modelled as follows:
Ap = C ⋅ 4πr 2 n (7.12c)
C = C1 (ε ) ⋅ C2 (uavg ) ⋅ C3 (Lice ) (7.12d)
Here C is the product of the effective heat transfer area affected by the porosity within the ice-
rich layer C1(ε), the velocity of the permeating water within the ice-rich layer C2(uavg) and the
side wall surface of the ice storage tank C3(Lice). Formulae for these quantities are shown in
Table 7.1.
214
Table 7.1. Ratio of effective heat transfer area
Effect of porosity within ice-rich layer C1 (ε ) = 0.2 (ε<0.7)

C1 (ε) β
⎛ ε − 0.7 ⎞
C1 (ε ) = 0.2 + ⎜ 0.8 ⋅ ⎟ ,β=2 (0.7≤ε≤0.9)
⎝ 0.2 ⎠
C1 (ε ) = 1 (ε>0.9)
Effect of velocity of permeating water ⎛ u avg ⎞
2
C2 (uavg) (
C 2 u avg ) =⎜
⎜ u avg 0
⎟
⎟
, uavg0=0.87mm/s
⎝ ⎠
Effect of side wall surface of tank C3 ( Lice ) = (Lice / L0 )−2 , L0 = 0.7m
C3 (Lice)
Lice: Initial
7.3.3 CONFIRMATION OF MODELLING
The time history of the floating ice-rich layer was investigated experimentally with two mass
flow rates, G = 0.33 and 0.92 kg/s, with a comparison of calculation results shown in Figure
7.7 (1). When G is small, (G = 0.33 kg/s), the contour of the floating ice-rich layer did not
spread to the side wall and the water region was formed in the tank, as shown in picture (a).
But when G is large, (G = 0.92 kg/s), as shown in picture (b), then the contour of the floating
ice-rich layer maintains one-dimensionality in the direction of the water depth, with almost no
formation of a water region. Thus it was possible to demonstrate with this calculation model
the difference in the contours of the floating ice-rich layer due to differences in the mass flow
rate G, as obtained in experiments.
Figure 7.7 (2) presents a comparison of calculated and experimental results for the evolution
in the contour of a compact ice-rich layer (the piled ice-rich layer) after the floating ice-rich
layer reaches the bottom wall of the tank. The mass flow rate G = 1.11 kg/s is the condition
under which a floating ice-rich layer is formed whose one-dimensionality in the direction of
water depth is maintained (G > 0.92 kg/s). In this figure some difference is seen between the
measured and calculated contours of the piled ice-rich layer. There was no big difference
between the experimental and predicted results for the maximum thickness of the piled ice-
rich layer in the center of the tank 80 minutes later (the time after the ice-rich layer reached
the bottom of the tank).
215
0.8 0.8 0.8 0.8

Initial Initial Initial
0.7 Water level Water level 0.7 0.7 Water level 0.7
0.6 0.6 0.6 0.6

10min 10min
10min 10min
0.5 0.5 0.5
z,m
0.5
z,m
z,m
z,m
Ice-rich Ice-rich
20min
0.4 layer 0.4 0.4 layer 20min 0.4
Ice-rich 20min
20min Ice-rich
layer
0.3 0.3 0.3 30min layer 0.3
30min
30min
0.2 30min 0.2 0.2 0.2
Water 40min
40min
0.1 Water 0.1 0.1 40min 0.1
36min Water 46min
Water
0 0 0 0
0.5 0.4 0.3 0.2 0.1 0 0.1 0.2 0.3 0.4 0.5 0.5 0.4 0.3 0.2 0.1 0 0.1 0.2 0.3 0.4 0.5
r,m r,m
Experiment Prediction Experiment Prediction
(a) G= 0.33 kg/s (vavg = 4.2×10-4 m/s ) (b) G= 0.92 kg/s (vavg = 1.17×10-3 m/s)
(1) Comparison of contours of floating ice-rich layer between experimental

and calculation results
Side wall Center axis Side wall

1.2 1.2
Suplly of Ice Slurry
G =1.11kg/s
G =1.11kg/s
α ice =2.5%
1 Operatiom time=80min 80m in 1
70m in
60min
60m in
70min
0.8 Ice-rich layer 50m in 0.8
40min 30min
40m in
0min 30m in
z ,m
,m
z z,m
0.6 Initial water level 0.6

Ice-rich layer
-30min
0.4 Operation time= 0.4
-30min Ice-rich layer
-15min
0.2 0.2
Operation time=0min
-15min Ice-rich layer reaching
the bottom of tank
0min Bottom
0 0 wall
0.5 0.4 0.3 0.2 0.1 0 0.1 0.2 0.3 0.4 0.5
r ,m rr r,m
, ,mm
Experimant Prediction
(2) Comparison of contours of piled ice-rich layer between experimental
and calculation results
Figure 7.7. Comparison of prediction and measurement in an ice storage process
The design conditions of an ice storage tank for storing a prescribed amount of ice (thermal
energy) can be determined by calculating the contour of the ice-rich layer based on this
model.
At first the ice slurry flow rate (for the effective use of the volume of the ice storage tank) and
216
the number and positions of the ice slurry supply openings are determined from the horizontal
distribution of the floating ice-rich layer. Figure 7.8 shows both the calculation and
experimental results on the relationship among the average downward superficial velocity of
water inside the tank vavg, the ice packing factor (IPF), and the thickness of the ice-rich layer
at the side wall in relation to the central axis of the tank Lmin/Lmax. The radius of the ice
storage tank R is 0.5 m and the water depth h is 0.62m. As evident in the figure, vavg > 0.8
mm/s can be determined as the condition that allows the tank volume to be used effectively
with the ice-rich layer spreading to the side wall and the water region remaining small
(Lmin/Lmax>>0).
20 1
Lmin/Lmax(exp)
R=0.5m
16 G=0.33 0.92kg/s 0,8

h=0.62m
IPF (%)
12 Lmin/Lmax(exp) 0,6
Lmin/Lmax (-)
8 0,4
IPF(exp)
4 0,2
0 IPF(exp) IPF(cal) 0
Lmin/Lmax(cal)
0,2 0,4 0,6 0,8 1 1,2
v avg (m/s)
-3
×10
Figure 7.8. Rate of utilisation of the volume of the ice storage tank
Moreover, the height of the space above the initial water level in the tank (freeboard) is
determined from the distribution in the height direction of the piled ice-rich layer. Figure 7.9
shows the predicted contour of the ice-rich layer for the actual design study in a large scale ice
storage tank of size 8.8 × 8.5 × 4 m (length x width x height). Here, R is the equivalent radius
of the cross-sectional area of the rectangular tank (1/2 the distance between the ice slurry
supply openings). In the figure, predicted results are shown when the ice-rich layer reaches
the side wall of the tank, and in 480 minutes of ice storage operation, approximately 0.4 m of
ice piles up in the vertical direction from the initial water level. Through this calculation, the
number of ice slurry supply openings required to store the prescribed quantity of ice and
perform ice storage operation for 480 minutes could be determined. From such predicted
contours of the ice-rich layer and the structure inside the tank (placement of girders and
communicating openings to other tanks), the height of the available space at the top of the
tank could be calculated as 1.2 m (water depth = 2.8 m).
217
4
Supply of
Initial
Slurry Ice 480min
Water Level
3
60min
z,m
2 120min
180min
Ice-rich Layer
240min
1
Water
0
0 1 2 3 4 R
=4.9m
r,m
(a) Predicted contours of an ice-rich layer in a large scale tank
Supply of Ice Slurry
Ice-rich layer
(b) Ice-rich layer in an actual large scale tank

Figure 7.9. Design conditions for ice storage
A comparison between the experimental and predicted results for the cold water temperature
at the outlet of a flat tank in the ice melting process is presented in Figure 7.10. Tout(exp) is
the experimental value, Tout(cal) is the calculated value according to this model, Lice is the
calculated value of the thickness of the ice-rich layer, W is the flow rate of sprayed water, and
IPF is the ice packing factor under charging conditions, which is averaged over an initial
water volume. The experimental results can be well predicted by the model. Also shown in
figure 7.10 is Tout(emp-rel) which is the value obtained from a simple empirical correlation
developed assuming the temperature within the ice-rich layer to be at the freezing-point
everywhere, and using the ice-rich layer cooling capacity obtained from an analysis of the
experimental data (Tanino et al., 1995, Tanino et al., 1997 and Mito et al., 2001).
These models of storing (accumulating/baling) and melting ice have been used for the design
in over 20 actual supercooling-type ice storage systems in Japan. The sizes of the systems
vary from 25 to 1120 m3 tank capacity and 25 to 1060 tons of refrigeration cooling capacity.
Applications cover office buildings, colleges, commercial buildings, shopping centres, food
distribution centres and factories.
218
14 1.8
Tank:2.5×2.0×2.4mH (Initial water level=1.6 mH)
IPF=34% W=15 m 3/h
Heat
Lice Ice manufacturing =3
Heat load=71kw
T , •C
Tin
Lice ,m
7 0.9
Tout(cal)
Tout(exp
Tout(emp-rel)
0 0
0 120 240
t , min
Figure 7.10. Comparison of experimental and predicted results

for cold water exit temperature from a flat water tank
7.4 Numerical modelling for mixing tank design
7.4.1 PHYSICAL MODELLING OF FLOW AND STRATIFICATION
In modelling the stratification process, the forces acting on the ice particles have to be in
equilibrium. Then the velocity and the distribution of ice particles can be calculated.
The forces acting on the ice particle are (Westphal, 1947):
– gravity,
– buoyancy,
– drag force on the particle moving in the liquid.
The rise velocity of an isolated particle can then be derived. It corresponds to the required
local downward flow velocity of the liquid to keep the particle in balance and thus prevent
stratification. To determine the drag force, a spherical particle shape and laminar flow at low
Reynolds numbers are assumed.
The ice particles are in practice not isolated but interact with each other. A downward flow of
the liquid will occur, balancing the volume in the tank if the particles ascend. The effective
rise velocity is then different from the local velocity between the particles and liquid.
Therefore a correction for the rise velocity of a particle swarm dependent on the average ice
concentration has to be applied.
In agitated tanks with turbulent flow the rise velocity of the ice crystals is superposed with the
mean flow of the liquid. A clear reduction in the local rise velocity of the particles in the
liquid can result from this, but reliable quantitative results are rare even for particle loaded
flows in general (Schempp, 1996).
219
7.4.2 MATHEMATICAL MODELLING
The following three equations define mathematically the storage problem with buoyancy-
induced motion of the ice particles in an ice slurry with diffusion, but without the occurrence
of melting (Egolf et al., 2001):
∂ r
(ρˆ ice ) + div (ρˆ ice ⋅ v ice ) − D ⋅ ρ ⋅ ∆c ice = 0 (7.13)
∂t
r
Here, D is the diffusion constant, v the velocity vector and ρ̂ the density of a component of
the ice slurry.
The initial condition is defined by a homogeneous ice fraction cice throughout the fluid
domain Γ in the slurry tank. Thus the density is constant at time zero:
r r
ρˆ ice (t = 0, x ) = ρˆ ice,0 = const, ∀x ∈ Γ (7.14)
A further boundary condition is that no ice particles cross the bounding walls of the storage
tank and the free surface of the ice slurry, which are denoted by ∂Γ .
[ρˆ ice ⋅ vrice − D ⋅ ∇ρˆ ice ] ⋅ γr(t, xr ) = 0, r r

∀x, γ ∈ ∂Γ (7.15)
Starting from laminar flow around a spherical particle at low Reynolds numbers the following
r
rise velocity vice is found. Here, vice is the vertical component of v . The complete derivation
is given in (Egolf et al., 2001). The equation is valid for ice fractions up to 15-20%. For
higher ice fractions the Bingham behaviour of ice slurry has to be taken into consideration.
Results for that range are not known so far.
⎧ g ⋅ Dpart
2
(ρ slurry (T , c ice ) − ρ ice (T ))
⎪ ⋅ , c ice < c ice,max
v ice (T , c ice ) = ⎨ 18 η slurry (T , c ice ) (7.16)
⎪
⎩0, c ice ≥ c ice,max
ρ ice Vice,max
c ice,max = ⋅
ρ V
Here c ice,max is the maximum ice concentration resulting from the maximum packing factor
as described in section 7.2.2.
7.4.3 CONFIRMATION OF MODELLING
Vuarnoz et al. (2001) performed experimental investigations on the stratification process with
initial homogeneous ice particle fields of 10.3% and 21.4% ice fraction. Samples were taken
from four different levels of an ice slurry tank with 1000 litre volume and investigated under a
light-transmittance microscope. The experiments provided information on ice fraction
distribution in the tank as a function of time and space.
220
The initial condition is obtained by stopping the mixer suddenly. The constant temperature
profile in the tank remained during the stratification process. Thus, stratification cannot be
determined by temperature measurements alone.
The observation chamber had a height of 1mm. A frame of 2384 µm by 1603 µm enlarged
with a microscope was photographed. Ice fraction was determined by evaluating the number
and size of ice particles in the photograph.
A front is defined as the interface between the mushy ice slurry and the ice-free liquid, which
moves upward in the tank according to the level of extraction. After some time ice has
accumulated at the top and the lower part of the tank contains pure liquid.
At the top, particles cannot rise further and therefore ice fraction increases until maximum
packing is reached. A second front in the ice concentration is the level where maximum
packing has been just reached. This front propagates from the top of the tank downwards. Not
much is known about this steady-state ice layer with the maximum ice fraction. As the
buoyancy force continues to act, this layer may be further compressed. It might be that the ice
crystals freeze slightly together in this layer.
Another effect is documented by Egolf et al. (2002). From the photographs in (Vuarnoz et al.,
2001) it was found that the ice particles had approximately an elliptical shape with a mean
width to mean length ratio of 0.7. Assuming spherical ice particles with an identical volume
in the numerical modelling, an equivalent diameter of 200 µm was obtained. The observed
front propagation times could not be explained by the simulations with this diameter.
Adjustment of the results is obtained by setting the diameter to 500 µm in the simulation. This
assumption, that a small number of particles stick together resulting in clusters with a larger
equivalent diameter, could be confirmed by analysing the photographs.
1. Egolf, P.W.; Vuarnoz, D.; Sari, O.: A model to calculate dynamical and steady-state
behaviour of ice particles in ice slurry storage tanks. Fourth IIR Workshop on Ice Slurries,
Osaka, Japan, November 2001.
2. Egolf, P.W.; Vuarnoz, D.; Sari, O.; Ata-Caezar, D.; Kitanovski, A.: Front propagation of
ice slurry stratification process. Fifth IIR Workshop on Ice Slurries, Stockholm, Sweden,
May 2002.
3. Frei, B.; Egolf, P.W.: Viscometry applied to the Bingham substance ice slurry. Second IIR
Workshop on Ice Slurries, Paris, France, May 2000.
4. Hansen, T.M.; Radosevic, M.; Kauffeld, M.: Behaviour of ice slurry in thermal storage
systems. ASHRAE Research Project - RP 1166, February 2002.
5. Huhn, R.: Untersuchung und Modellierung der Speicherung von pumpfähigen Sole-
Eisgemischen in Pufferspeichern sowie der Eisbildung an Verdampferoberflächen.
Forschungsrat Kältetechnik e.V., Bericht Nr. 88/02, Frankfurt (Main) 2003.
6. Kakutami, S.; Noburo, O.: Development of High Density Heat Transportation System.
Refrigeration, Vol. 74, No. 856, pp. 28-35, 1999.
7. Kozawa, Y.; Aizawa, N.; Tanino, M.: Study on Ice Storing Characteristics in Dynamic-
type Ice Storage System by Using Supercooled Water, Effects of the supplying conditions
of ice-slurry at deployment to District Cooling and Heating System, Third IIR Workshop
on Ice Slurries, pp. 87-96, Lucerne 2001.
8. Kozawa, Y.; Tanino, M.: Ice-water Two-phase Flow Behavior in Ice Heat Storage System,
221
First IIR Workshop on Ice Slurries, pp. 146-156,Yverdon, Switzerland, 1999.

9. Mito, D.; Mikami, Y.; Tanino, M.; Kozawa, Y.: A New Ice-slurry Generator by using
Actively Thermal-hydraulic Controlling both Supercooling and Releasing of Water, Fifth
IIR Workshop on Ice Slurries, Stockholm, Sweden, May 2002.
10. Mito, D.; Tanino, M.; Kozawa, Y.; Okamura, A.: Application of a Dynamic –type Ice
Storage System to the Intermittent Cooling Process in the Food Industry, Fourth IIR
Workshop on Ice Slurries, pp.105-114, Osaka, Japan, November 2001.
11. Nørgaard, E.; Christensen, K.C.; Hansen, T.M. et al.: Heat transfer and pressure drop with
ice slurry. 4th IIR Gustav Lorenzen Conference, Purdue, July 2000.
12. Riber, C.: Akkumumlering af sjapis. Danmarks Tekniske Universitet, January 1999.
13. Schempp, A.: Modellierung von Feststoffverteilung und Wärmeübergang in gerührten
Fest/Flüssig-Systemen. Fortschritt-Berichte / VDI : Reihe 3, Verfahrenstechnik Nr. 430,
VDI-Verl., Düsseldorf 1996.
14. Tanino, M.; Kozawa, Y.; Hijikata, K.; Nakabeppu, O.: Prediction of Ice Storage Process in
Dynamic-type Ice Storage System, Proc. of 10th International Conference on Thermal
Engineering and Thermogrammetry, pp. 321-326, Budapest, Hungary 1997.
15. Tanino, M.; Kozawa, Y.: Ice-water Two-phase Flow Behavior in Ice Heat Storage
Systems, Int. Jour. of Refrigeration of IIR, 24(7); 638-652, 2001.
16. Tanino, M.; Kozawa, Y.; Mito, D.; Inada, T.: Development of Active Control Method for
Supercooling Releasing of Water. Second IIR Workshop on Ice Slurries, pp. 127-138,
Paris, France, May 2000.
17. Tanino, M.; Kozawa ,Y.: Performance Evaluation of an On-site Type Ice Storage System
by using Ice Slurries Made of Super-cooled Water, Jour. of AIRAH, : 55(9); pp. 28-31,
2001.
18. Tanino, M.; Mito, D.; Kozawa, Y.: Recent Study on Ice Slurries, Jour. of AIRAH, 55(8);
17-18, 2001.
19. Tanino, M.; Moriya, M.; Kikuchi, S.; Shiraushi, H.; Okonogi, T.; Kozawa, Y.: An Ice
Storage System Using Supercooled Water, 2st Report; Ice Storage and Melting
Characteristics. Trans. of the Japanese Association of Refrigeration, 12(3), pp. 39-49 (in
Japanese), 1995.
20. Tanino, M.; Moriya, M.; Okamura, A.; Yamazaki, K.; Okonogi, T.; Seki, Y.; Kozawa, Y.;
Miyata, Y.; Ohta, M.: Development of Large-scaled Ice Storage System by Using Super-
cooled Water, Jour. of the Society of Heating, Air-conditioning and Sanitary Engineers of
Japan, 71(11), pp. 73-83 (in Japanese), 1997.
21. Vuarnoz, D.; Sari, O.; Egolf, P.W.: Correlation between temperature and particle
distribution of ice slurry in a storage tank. Fourth IIR Workshop on Ice Slurries, Osaka,
Japan, November 2001.
22. Westphal, W.H.: Physik – Ein Lehrbuch von Wilhelm H. Westphal. 12. Auflage,
Springer-Verlag, Berlin 1947.
222
CHAPTER 8. MELTING ICE SLURRY IN HEAT EXCHANGERS

by Ebbe Nørgaard
Ice slurry is often used in indirect cooling applications where ice slurry is a refrigerant (in
contrast to the direct contact applications especially known from the food and fish industry,
see Chapter 9). When ice slurry is exchanging heat with another liquid, plate heat exchangers
are a very common choice in current technology. But, because of their small flow channels,
ice slurry flow may present a challenge for their use. Nevertheless, experience has shown that
compact plate heat exchangers can be used with ice slurry. Actual experience and calculation
methods are described in this chapter.
Another topic of heat exchange is the cooling of air in fin-and-tube heat exchangers (also
known as air coils). This chapter also covers these types of heat exchangers.
8.1 Plate heat exchangers
A very compact and highly effective type of heat exchanger is the Compact Brazed Plate Heat
Exchanger (CBE). Plate heat exchangers have found uses in many different applications. This
section focuses on applications where CBE´s are used to cool secondary media with ice
slurry, hence involving melting of ice particles.
Because ice slurry has the latent heat stored in the media itself, a relatively high Log Mean
Temperature Difference (LMTD) is achievable compared to single-phase media (see Figure
8.1). This means that the volumetric flow rate can be very low when the ice concentration is
high compared to a single-phase medium. Furthermore ice slurry has good heat transfer
characteristics.
Applications where the use of plate heat exchangers and ice slurry as a coolant is considered
feasible, are generally as intermediate heat exchangers. This could be in a cooling system with
many different temperature demands or in a large distribution system where the energy
consumption for transport is an important issue. Figures 8.2 and 8.3 illustrate some of these
applications.
In order to gain knowledge about the thermodynamic and hydraulic behaviour of ice slurry in
plate heat exchangers and to create a foundation for dimensioning, three small scale copper
brazed heat exchangers have been tested in a laboratory with ice slurry. The CBE´s tested had
different geometry in order to find the optimal heat exchanger for ice slurry.
223
Figure 8.1. Illustration of a theoretical temperature difference between the two media that
exchange energy. All the ice is melted in the heat exchanger
Ice generator
Chiller
Ice Storage
Tank
CBE CBE
CBE CBE
Building
Figure 8.2. A district cooling system where CBE´s are used as intermediate heat exchangers
between the central refrigeration unit and the buildings to be cooled
224
S
Air cooled
R condensor
R S K1
Propane S
C1
K2
Ice slurry
R
Icegenerator
K3
Ice creme (-32°C)
R
CBE R F1
S
S
S
C2 F2
S CO2 (-8°C)
P2
P1/P2
F3
R
Figure 8.3. Sketch of a laboratory test plant where a CBE was used with ice slurry as a
condenser in a cascade system with propane in the top cycle and carbon dioxide in the bottom
cycle. The installation is designed for a supermarket application. The condensing pressure of
the carbon dioxide was two bar lower than with the original single-phase media
8.1.1 TEST RIG SET-UP
A test rig for the examination of pressure drop and heat transfer in a plate heat exchanger was
constructed. Three heat exchangers were tested:
• Ⓐ with 10 plates. Called B15 in Figure 8.6 and thereafter (connections: 4 channels with
ice slurry and 5 channels with glycol 30 wt-%);
• Ⓑwith 10 plates. Called B10 in Figure 8.6 and thereafter (connections: 4 channels with
ice slurry and 5 channels with glycol 30 wt-% and vice versa);
• Ⓒ with 28 plates Called B5 in Figure 8.6 and thereafter (connections: 13 channels with
ice slurry and 14 channels with glycol 30 wt-%).
The physical dimensions of the three heat exchangers are illustrated in Figure 8.4.
225
35
70
283
35
113
183
70 462
70
• Heat transfer area: • Heat transfer area: 0.26 m2 • Heat transfer area: 0.29
0.336 m2 m2
Ⓒ Ⓑ Ⓐ
Figure 8.4. Dimensions of the three different heat exchangers tested
The ice slurry was generated by a standard ice slurry system with a scraped surface ice slurry
generator and stored in a homogeneous storage facility. The ice concentration was controlled
by on/off regulation of the ice generator, which allowed the ice content to vary within ± 2 wt-
%. A solution of 10 wt-% ethanol in water was used throughout the experiments (freezing
point temperature = –4.5°C). A schematic of the test rig is shown in Figure 8.5.
Thermocouple chain
Return ice slurry
dp
Controller for Buffer tank Massflow meter
heater Electrical heater
Supply ice slurry
Propylene glycol
Volumeflow meter
Figure 8.5. Sketch of test rig
On the ice slurry side (primary side), two pieces of transparent plastic pipe were placed just
before and after the heat exchanger in order to visualise the flow of ice slurry through the heat
exchanger.
On the secondary fluid side an open storage tank was installed containing approximately 70
litres of 30 wt-% propylene glycol. A 6 kW electrical heater installed in the tank was
activated by a relay, which was controlled by the temperature of the fluid in the tank. In that
way the inlet temperature of the glycol could be kept at the desired level within ± 0.5 K.
226
8.1.2 MEASURING EQUIPMENT
The list of measuring equipment used in the experiments is shown in Table 8.1.
Table 8.1. Measuring equipment
Instrumentation Type/make Range Accuracy

Mass flow meter / Sensor: Danfoss 2100 280-5600 kg/h ± 0.1% (of reading) or
density (DI15)/ Converter: 100-2900 kg/m3 ± 3 kg/m3
Danfoss 3000
Volume flow meter Sensor: Danfoss Mag 0-1330 L/hr ± 0.15% fs (full scale)
1100 (DN 10)/
Converter: Danfoss
Mag 5000
Differential pressure Yokogawa Dpharp 0 – 400 kPa ± 0.075% fs
transmitter dp-transmitter
Absolute temperature Thermocouple Type T -200°C – 350°C ± 0.3 K
ice slurry and brine
Temperature Difference Thermocouple chain 0–3K ± 0.05 K
(ice slurry) Type T / calibrated
Temperature Difference Thermocouple chain 0–5K ± 0.05 K
(Brine) Type T / calibrated
8.1.3 HEAT TRANSFER RESULTS
In order to achieve equal flow conditions on the glycol side and thereby rule out that
parameter in the data analysis, the brine flow rate (or Reynolds number) was kept almost
constant throughout the experiments. This resulted in heat transfer coefficients on the glycol
side in the range of 5000 – 8000 W/m2K. The variation was also caused by small variations in
the flow rate, but it was mainly caused by the temperature dependent fluid properties.
The experiments were planned so that all the ice was melted before leaving the heat
exchanger, so as to get full benefit from the latent heat of the ice slurry. This means that the
heat flux was also increased in line with increased flow rate and thereby the Reynolds
number.
The inlet temperature of the glycol was adjusted to the level where all the ice was melted. The
heat transfer coefficients obtained are shown in Figure 8.6. Correlations are given in section
8.1.4.
227
5000 B10 30% 4ch

4500 B10 20% 4ch
B10 10% 4ch
4000
B10 30% 5ch
3500
B10 20% 5ch
h (W/m2K)
3000 B10 10% 5ch

2500 B5 30% 13ch
2000 B5 20% 13ch
1500 B5 10% 13ch
B15 30% 4ch
1000
B15 20% 4ch
500
B15 10% 4ch
0 25 50 75 100 125 150 175
Re (-)
Figure 8.6. Heat transfer coefficients on the ice slurry side obtained for all three CBE´s and
plotted against the Reynolds number based on average ice slurry properties
between the inlet and outlet conditions. The ice concentration shown in the legend is the
inlet condition – the outlet concentration is always 0 wt-%.
The channel number, “ch”, denotes channels with ice slurry
The heat transfer coefficients show an almost linear increase when plotted against the
Reynolds number, which indicates that the flow is laminar. The experiments with type
(B15) show clearly that the heat transfer coefficient increases with increasing ice
concentration. This tendency is not so clear with the other two exchangers (B5) and
(B10). Unknown parameters in the experiments are the time dependent crystal size
distribution and viscosity, and how much these factors influence the heat transfer
characteristics of ice slurry.
Heat transfer coefficients for two-phase fluids are known to show a dependency on the heat
flux [kW/m2]. Figure 8.7 shows the relations among heat flux, Reynolds number and heat
transfer coefficients obtained in the experiments.
228
7000
6000
7000
5000
6000
h [W/m2K]
4000
5000
h [W/m2K]
3000
4000
2000
3000
1000
2000
0
1000 150
125
0 100
20 75 [-]
15 50 Re
10 5
25
Heat flux [kW/m2K] 0 0
Figure 8.7. Heat transfer coefficients plotted against the heat flux and Reynolds number.
R2 = 0.97 for the grid shown
The plot shows that the heat flux has a positive effect on the heat transfer coefficient up to 10-
15 kW/m2. Above 15 kW/m2 there is only a slight increase in the heat transfer coefficient-heat
flux slope, which indicates that the heat transfer process on the ice slurry side is dominated by
the carrier fluid properties and constrained by the heat transfer between the carrier fluid and
the ice particles.
8.1.4 CORRELATION FOR HEAT TRANSFER
All fluid properties used in the following correlations are average ice slurry properties
between the inlet and outlet condition unless otherwise stated.
Q = m& 1 ⋅ C p1 ⋅ ∆t1 = m& 2 ⋅ C p 2 ⋅ ∆t 2 = K ⋅ A ⋅ LMTD (8.1)
Q = Heat capacity [kW]

&
m = Mass flow rate [kg/s]
Cp = Specific heat capacity [kW/(kg K)]
∆t = Temperature change [K]
A = Area [m²]
K = Heat transfer coefficient [W/(m² K)]
LMTDsh = Logarithmic Mean Temperature Difference [K]. (Superheated - see section 8.3)
Fouling was not taken into account because the heat exchangers were new so the overall heat
transfer coefficient, K, is defined as:
229
1
K= (8.2)
1 1t
+ +
h1 λ h2
h = film coefficient [W/(m² · K)]

t = Wall thickness [m]
λ = Wall thermal conductivity [W/(m · K)]
The film coefficient, h , is defined as:
Nu ⋅ λ
h= ⋅ψ ice (8.3)
dh
0.3
⎛η ⎞ (Re + 6)0.125
ψice = Two-phase correction factor, ψ ice = ⎜ bulk ⎟
⎜η ⎟
⎝ cf ⎠
Nu = Nusselt number [-]
λ = Fluid thermal conductivity [W/(m K)] •
dh = Hydraulic diameter [m]

µ bulk = Ice slurry viscosity [Pa s]•
µ cf = Carrier fluid viscosity [Pa s] •
The correlation for the Nusselt number is given by the expression:
Nu = C ⋅ Re n ⋅ Pr y (8.4)
C = Empirical constant [-]

n = Empirical exponent [-]
m& channel ⋅ 2
Re = Reynolds number [-] Re = where w is the channel width in m
w ⋅ η bulk
C p ,cf ⋅ µ
Pr = Prandtl number [-] Pr =
λ
6.4
e Pr + 30
y = Prandtl number exponent [-], y = (about 1/3 for laminar flow and 0.4 for
3
turbulent flow)
From the experimental data the empirical values shown in Table 8.2 were found:
Table 8.2. Empirical values obtained to correlate heat transfer results
Ⓒ Ⓑ Ⓐ
C 0.73 0.31 0.28
n 0.42 0.61 0.57
The relation between the measured and calculated Nusselt numbers is shown in Figure 8.8.
230
35
+20%
30
Calculated Nusselt number (-)

25 -20%
20
15
B10
B15
10
B5
0
0 5 10 15 20 25 30 35
Measured Nusselt number (-)
Figure 8.8. Accuracy of heat transfer correlation
8.1.5 PRESSURE DROP RESULTS
In parallel with the heat transfer measurements the pressure drop was measured on the ice
slurry side. The measurements are shown in Figure 8.9.
20 B10 30% 4ch
18 B10 20% 4ch
16 B10 10% 4ch
14 B10 30% 5ch
12 B10 20% 5ch

∆p (Kpa)
10 B10 10% 5ch
8 B5 30% 13ch
6 B5 20% 13ch
4 B5 10% 13ch
B15 30% 4ch

2
B15 20% 4ch
0
0 25 50 75 100 125 150 175 B15 10% 4ch
Re (-)
Figure 8.9. Pressure difference plotted against the Reynolds number. The ice concentration on
the label shows the inlet condition – the outlet concentration is always 0 wt-% - and “ch” is
short for channels with ice slurry
231
The pressure drop increased with increasing Reynolds number and ice concentration. The heat
exchanger of type Ⓒ (B5) (short, narrow and many parallel channels) had the lowest pressure
drop of the three heat exchangers tested and Ⓐ (B15) the highest (long, narrow and a few
channels). In practical installations for a given capacity, the optimal heat exchanger with
regard to energy consumption by distribution pumps and installation cost, is the heat
exchanger that is the cheapest, most compact, has the lowest pressure drop and the highest
heat transfer properties for a given capacity. The heat transfer and pressure drop results
indicates that the optimal heat exchanger for ice slurry must be short and have many parallel
channels like the typeⒸ (B5) or alternatively wide and short like the type Ⓑ(B10).
8.1.6 CORRELATION FOR PRESSURE DROP
All fluid properties used in the following correlations are average ice slurry properties
between the inlet and outlet conditions unless otherwise stated.
f (Re) ⋅ 1000
∆Pchannel = ⋅m
& channel
2
⋅ ϕ ice [Pa] (8.5)
ρ
m& channel = Channel mass flow rate [kg/s]
ρ = Density [kg/m3]
⎛ 1. 2 ⎞
⎜ ⎟⎟
⎛η ⎞ ⎜⎝ Re +6 ⎠
φice = Two-phase correction factor, ϕ ice = ⎜⎜ wall ⎟
⎟
⎝ η bulk ⎠
⎛ b ⎞
⎜ a+ ⎟
f(Re) = e⎝ Re1.5 ⎠
[1/m4]
& channel ⋅ 2
m
Re = Reynolds number, Re =
w ⋅ η bulk
w = Channel width [m]
ηbulk = Ice slurry viscosity [Pa s] •
From the experimental data the empirical values for the constants a and b were found for the
three CBE’s, Ⓐ, Ⓑ and Ⓒ as shown in Table 8.3.
Table 8.3. Empirical values obtained to correlate pressure drop results
Ⓒ Ⓑ Ⓐ
a 17 16 17.6
b 110 221 181
The relation between the measured and calculated pressure drop values is shown in Figure
8.10.
232
20
+25%
B10
Calculated pressure drop (kPa)

B15
15
B5
-25%
10
0
0 5 10 15 20
Measured pressure drop (kPa)
Figure 8.10. Accuracy of pressure drop correlation
Although not all data were well correlated, the correlation can predict the data within ± 25%.
The viscosity of ice slurry is affected by the age of the ice slurry, which was not measured in
the experiments. This might explain some of the larger deviations. However, additional
experiments were carried out subsequently to investigate the relation between pressure drop
and time. See section 8.3.
8.1.7 OBSERVATIONS
In order to illustrate the effect of the latent heat stored in the solid ice particles a comparison
between the measured pressure drop for ice slurry and calculated pressure drops for a 30 wt-%
solution of propylene glycol in water has been carried out, see Figure 8.11.
60
B10 30 % ice
50
B10 30% pg
∆p (kPa)
40
30
20
10
0
0 1 2 3 4 5 6 7 8
Q (kW)
Figure 8.11. Pressure drop data plotted against the heat exchanger capacity. The ice
concentration in the legend shows the inlet condition – the outlet concentration is 0 wt-%. The
comparison is made with equal temperature differences on both primary and secondary sides
and equal mass flow rates on the secondary side
233
Figure 8.11 shows that the pressure drop is 5 to 10 times higher with propylene glycol
compared with ice slurry when the heat exchanger capacity is the basis for comparison.
A simple analysis has been made where the goal was to investigate how to optimally operate a
CBE with regards to ice concentration, pressure drop and heat exchanger capacity. Figure
8.12 shows a plot of the power consumption required to transport the ice slurry through the
heat exchanger against the transferred heat capacity.
25 B10 30% 4ch
B10 20% 4ch
20 B10 10% 4ch
B10 30% 5ch

Ptransport [W]
15 B10 20% 5ch
B10 10% 5ch
10 B5 30% 13ch
B5 20% 13ch
5 B5 10% 13ch
B15 30% 4ch
0 B15 20% 4ch
0 2 4 6 B15 10% 4ch
Qperformance [kW]
Figure 8.12. Pumping power (Ptransport) required to transport the ice slurry through the CBE
against the transferred heat capacity (Qperformance). The pump efficiency coefficient used to
calculate the transport effect is assumed to be constant at 0.1
From the analysis it is evident that the optimal point of operation is when the inlet ice
concentration is high. This becomes clearer as the capacity increases. The largest benefit is
when the ice concentration is increased from 10 to 20 wt-% ice at the inlet. If the criterion for
optimal operation is not low pressure drop but instead high performance, the flow rate should
be increased so that the ice is not melted off completely at the outlet. In addition the inlet ice
concentration should be as high as possible.
8.1.8 MINIMUM FLOW RATE
Besides the heat transfer and pressure drop experiments another series of experiments was
carried out, in order to create a map useful for operation because flow pulsation was observed
in some cases. This happened when the flow rate was low in combination with high ice
concentration.
Each experiment was carried out at a constant flow rate and with increasing ice concentration
from 0 to 30 wt-%. It was then observed at which ice concentration the flow began to pulsate
234
carried out for each heat exchanger with different flow rates. The results are shown in Figure
8.13.
50
45 B5
Reynolds number (-)
40 B15 Stable flow conditions

35 B10
30
25
20
15
10
5 Unstable flow conditions
0
0 5 10 15 20 25 30
Ice concentration (% wt.)
Figure 8.13: The relation between the Reynolds number and ice concentration where the flow
begins to pulsate. If stable flow conditions are desired the point of operation
should be above the dotted line. The experiments were carried out
without any heat flux from the secondary side
For plate heat exchangers Ⓒ (B5) and Ⓐ (B15) there is a constant lower limit at a Reynolds
number of approximately 20, which should be always exceeded if stable flow conditions are
desired. The results for heat exchanger Ⓑ (B10) showed a linearly increasing tendency,
which indicates that when the ice concentration is high, higher Reynolds numbers are required
in order to obtain stable flow conditions. Actual flow blockages were not observed at any
time.
8.2 Air coolers
The use of ice slurry to cool air is applicable to many different industries. The main constraint
when using ice slurry in air coolers is the temperature level, which is below 0°C. This means,
for example, that the use of ice slurry in air conditioning systems is not straight forward
because the cold air is generated at a lower temperature than required for this type of
application. This results in larger energy consumption by the primary refrigeration plant than
what is required to cool air with water at 7°C. On the other hand, the lower air temperature
reduces the amount of air required for a given cooling load and hence air ducts and fans can
be reduced in size resulting in lower investment and operating costs of the air handling
system.
Other applications where the use of ice slurry to cool air is more attractive are:
• Supermarket display cabinets.
• Cold storage for food products (meat, beer, dairy products, vegetables, etc.).
235
• Industrial air conditioning – such as providing cabin cooling for parked aircraft during
static turn-around periods.
• Retro-fitting and refurbishment of existing pumped or DX cooling coils and plants,
which for environmental or other reasons, are being replaced with a secondary cooling
system.
In order to gain some fundamental knowledge about the behaviour of ice slurry in air coolers
and to establish a foundation for sizing, some experiments have been conducted on two
standard (single-phase) air coolers with different geometry.
8.2.1. TEST RIG SET UP
The experiments were conducted in a closed air circuit loop containing air conditioning
equipment, fan and an orifice to measure the air flow rate. Figure 8.14 shows a schematic of
the air circuit.
Figure 8.14. Sketch of test rig. The ice slurry air coil is labelled 1
The temperatures of air entering and leaving the heat exchanger were measured with nine
evenly distributed thermocouples in order to detect if there were any dead zones and if a
blockage would occur. The relative humidity was measured at the inlet and outlet of the heat
exchanger. On the ice slurry side the mass flow rate, inlet temperature and ice concentration
were measured at the inlet using a Coriolis mass flow meter. The temperature and pressure
differences on the ice slurry side were also measured.
The capacity of the air cooler was calculated on the airside, which also by calculation led to
the outlet ice concentration. In order to find the accuracy of this method a preliminary energy
balance between the air and liquid streams was conducted with a single-phase fluid on the
liquid side which resulted in a deviation of less than 10%.
The list of measuring equipment used in the experiments is given in Table 8.4, while Table
8.5 gives the dimensions of the two tested coils.
236
Table 8.4. Measuring equipment
Instrumentation Type/make Range Accuracy

Mass flow meter / Sensor: Danfoss 2100 280-5600 kg/hr ± 0.1 % (of reading) or
density/temperature (DI15)/ Converter: 960-1020 kg/m3 ± 0.2 % or
Danfoss 6000 –30-100°C ± 0.1 K
Differential Pressure Yokogawa Dpharp 0-100 kpa ± 0.075 % of full scale
transmitter dp-transmitter (fs)
Temperature difference Thermocouple chain 0-3 K ± 0.05 K
Type T / calibrated
Air volumetric flow rate Orifice (DIN1952) 0.08-2 m3/s ± 0.5%
Relative humidity Vaisala 0-100 % RH ±2%
Air temperature Thermocouple –5-30°C ± 0.3 K
Type T / calibrated
Table 8.5 Coil dimensions
Parameter 3/8” tube coil 12 mm tube coil

No. of circuits 14 11
Fin pitch 2.5 mm 2.5 mm
Inner tube diameter 9.25 mm 11.6 mm
Outer surface area 32 m2 54 m2
Inner surface area 2.4 m2 2.3 m2
Inner header diameter 18 mm 39 mm
Rows deep 4 4
Face area 0.7 x 0.72 = 0.50 m2 0.72 x 0.72 = 0.52 m2
No. of Tubes 112 88
8.2.2. HEAT TRANSFER RESULTS
Throughout the experiments the airflow was kept at a constant flow rate of 1.3 m3/s ≈ 2.7 m/s
mean face velocity. The humidity of the air was only measured entering the 12 mm tube coil
because dry cooling was assumed in the closed air circuit. It turned out that the circuit was not
perfectly closed resulting in ice build-up on the air side of the coil. The heat transfer results
for the 12 mm coil were corrected with the contribution of rim formation with comparable
results on the 3/8” tube coil where the relative humidity was measured both entering and
leaving the coil.
Figure 8.15 shows the heat transfer results for both coils. The heat transfer coefficient is seen
to increase with increasing Reynolds number and ice concentration.
237
Heat transfer coefficient

30 % ice 3/8
20 % ice 3/8
5000
4500 10 % ice 3/8
4000 10 % ice 12
h (W/m2K)
3500 20 % ice 12
3000 30 % ice 12
2500
2000
1500
1000
500
0 250 500 750 1000 1250 1500 1750 2000 2250 2500
Re (-)
Figure 8.15. Heat transfer results for both coils. The ice concentration shown in the legend is
the inlet ice concentration. The outlet ice concentration varied from 0-25 wt-%. The Reynolds
number was calculated based on the average ice slurry properties between the inlet and outlet
conditions. Furthermore the heat flux also varied between the experiments
In order to investigate the influence of heat flux on the internal heat transfer coefficients a
diagram was developed where the internal heat transfer coefficients are plotted against the
heat flux and Nusselt number (Figure 8.16).
4500
4000
4500
3500
HTC [W/m^2K]
4000
3000
3500
HTC [W/m^2K]
2500
3000
2000
2500
1500
2000
1000
1500 5
5 5
45 0
1000 4
35 0 -]
.5 3 [
12 10 2 0 Nu
20 5
7 .5 1
Heat f
lux [kW
5
5 10 5
/m^2] 2.
Figure 8.16. Heat transfer results for both coils. The area refers to the inner surface area.
The Nusselt number is the average of the inlet and outlet conditions
Because the correlation coefficient (R2 value) is relatively low (0.71), unambiguous
conclusions are difficult to draw. However, the plot indicates that the heat transfer coefficient
is unaffected by the heat flux, although the experiments using the CBE´s showed some
238
relation between the heat flux and heat transfer coefficients. In the plate heat exchanger
experiments, there was an effect of the heat flux up to 15 kW/m2 while the present results
showed none. A likely explanation for this difference is that the Reynolds numbers in the air
cooler were much higher than in the plate heat exchanger. In the air cooler experiments the
boundary layer could have been dominated by the ice slurry properties, resulting in relatively
high heat transfer coefficients.
8.2.3. CORRELATION FOR HEAT TRANSFER
Using the correlation described in section 4.5.1 “Heat transfer in straight tubes” and average
fluid properties between the inlet and outlet, it is possible to calculate the inner heat transfer
coefficients within ±30 % deviation. See Figure 8.17.
Accuracy of heat transfer correlation
+30%
5000
Calculated HTC (W/m2k)
4000
-30%
3000
2000 3/8"
12mm
1000
0
0 1000 2000 3000 4000 5000
Measured HTC (W/m2k)
Figure 8.17. Ice slurry side heat transfer results for both coils. The calculated values are based
on the correlation described in section 8.1.4 and average fluid properties between the inlet and
outlet. The air-side HTC is calculated from a correlation supplied by the manufacturer
Because the inner heat transfer coefficient is so much higher than the air-side heat transfer
coefficient, the overall heat transfer is more dependent on the air-side heat transfer coefficient.
8.2.4. PRESSURE DROP RESULTS
In parallel with the heat transfer measurements the pressure drop on the ice slurry side was
also measured as shown in Figure 8.18. The pressure drop increased with increasing
Reynolds number and ice concentration. In general the pressure drop was higher in the 3/8”
coil compared with the 12 mm coil due to the smaller tube diameter. The length of each pass
was similar for both coils.
239
Pressure difference
30 % ice 3/8
80 20 % ice 3/8
10 % ice 3/8
60 10 % ice 12
20 % ice 12
∆p (kPa)
40 30 % ice 12
20
0
0 250 500 750 1000 1250 1500 1750 2000 2250 2500
Re average (-)
Figure 8.18. Pressure drop results. The pressure loss in the connections is at maximum 8 kPa
and included in the values shown. The ice concentration shown in the legend is the inlet ice
concentration. The outlet ice concentration varied from 0-25 wt-%. The Reynolds number was
calculated based on the average ice slurry properties between the inlet and outlet conditions.
Furthermore the heat flux varied between the experiments
8.2.5. CORRELATION FOR PRESSURE DROP
By correlating the experimental pressure drop data during melting, it was found that a
standard Newtonian theory would be appropriate in combination with the coil manufacturer’s
data. So, in contrast with pressure loss in transport tubes, where the flow is Bingham and
Newtonian – the flow during melting is solely correlated as Newtonian. The length of the
tubes was 0.7 m, which is short relative to the transport tubes, therefore, the velocity profile
was not fully developed.
The following information was provided by the manufacturer:
When Re < 1500, the flow is considered laminar and the friction factor is λ = 64/Re.
When Re > 1500, the flow is transitional/turbulent and λ = 0.2 Re0.2. •
The pressure loss in the return bends was turned into an equivalent length of a straight tube.
The data given by the manufacturer did not perfectly match the measured pressure losses for
ice slurry. From experiments on pressure loss in single components it was found that the
friction coefficients for ice slurry are higher than the values tabulated for turbulent water flow.
However, as the Reynolds number increases the friction factor for ice slurry flow decreases
and approaches the tabulated values for turbulent flow of water.
Because the friction loss in return bends can be higher for ice slurry than for a single-phase
fluid, the pressure drop correlation was fitted by adjusting the equivalent length of the return
bends. This was done independently of the ice concentration – which is a simplification.
Table 8.6 shows the fitted values for ice slurry.
240
Table 8.6. Equivalent lengths of return bends.

The values represent the equivalent length of each bend
12mm coil 3/8” coil
Ice slurry (fitted value) 1.7 m 0.6 m
The pressure drop in the supply and return headers was calculated solely from manufacturer’s
correlation:
1 ⎛ λ ⋅ Lh ⎞
dp h = ⋅ ρ ⋅ ch2 ⋅ ⎜⎜ 0.85 + ⎟⎟ (8.6)
2 ⎝ dh ⎠
where the velocity and friction factor are calculated at the inlet (highest velocity).
The pressure drop decreases through the coil as the ice is melted off. Therefore the pressure
drop in the coils was calculated using the average of the inlet and outlet fluid properties and
according to:
1 ⎛ λ ⋅ L + Leq ,bend ⎞
dp in ,out = ⋅ ρ in ,out ⋅ c in ,out 2 ⋅ ⎜⎜ ⎟
⎟ (8.7)
2 ⎝ d ⎠
which results in:

dp in + dp out
dp = + dp hin + dp h ,out (8.8)
2
Using the above correlation the measured pressure drop data were predicted theoretically
within ± 15 %. See Figure 8.19.
70
Calculated pressure drop (kPa)
12 mm +15%
60
3/8"
50 -15%
40
30
20
10
0
0 10 20 30 40 50 60 70
Measured pressure drop (kPa)
Figure 8.19. Comparison of measured and calculated pressure drop
241
8.2.6. COMPARISON BETWEEN SINGLE-PHASE MEDIA AND ICE SLURRY
Using the manufacturer’s dimensioning software a comparison was made between the
measured ice slurry results and a single-phase fluid as shown in Figure 8.20.
Fluid Velocity v's Internal HTC´s
3500
3000
Internal HTC´s (W/m2K)
12 mm Tube
2500 10 mm Tube
10% Ethanol/Water
2000
20% Ice Concentration
1500
1000
500 30% Propylene glycol

0
0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0
Velocity (m/s)
Figure 8.20. Comparison of internal heat transfer coefficients between a single-phase fluid
and ice slurry. The outlet ice concentration varied from 0-15 wt-%
Ice slurry heat transfer coefficients were 3 to 4 times higher than those of the single-phase
liquid. This difference means that the capacity of the air cooler is increased approximately by
5-10 % or the heat transfer area can be reduced by 5-10% (largest improvement when the
flow velocity is low). In addition the average temperature difference is increased as a result of
the phase change of the ice slurry. Consequently, the heat transfer area can be reduced by up
to 30 % with ice slurry compared with a single-phase fluid.
Furthermore, the circuitry on the liquid side does not have to be cross-counter because the
temperature glide of ice slurry is low. This gives the designer an extra degree of freedom,
which opens up the possibility for minimising production cost.
A similar comparison was also carried out with respect to pressure drop as shown in Figure
8.21. At low velocities the pressure drop for ice slurry is almost equal to that of the single-
phase fluid, but at higher velocities the pressure drop for ice slurry becomes twice as high as
for the single-phase fluid.
242
Fluid Velocity v's Pressure Drop
70
60 12 mm Tube 10% Ethanol/Water

Pressure Drop (kPa)
50 10 mm Tube 20% Ice Concentration
40
30
20
10
30% Propylene Glycol
0
0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0
Velocity (m/s)
Figure 8.21. Comparison of pressure drop between a singlephase fluid and ice slurry.
The outlet ice concentration varied from 0-15 wt-%
8.2.7. MINIMUM FLOW RATE
Tube blockages were not observed at any time during the trials, but the flow can pulsate if the
flow velocity is very low. At flow velocities higher than 0.5 m/s the flow is stable. This works
also in the headers.
Experiments carried out on a transparent air cooler showed that the ice crystals tend to
agglomerate in the headers if the flow velocity is relatively low. Therefore, it is important to
design the headers so that dead spots are avoided. Some examples are shown in Figure 8.22 to
illustrate how to – and how not to – design the headers.
Figure 8.22. Illustration of appropriate/inappropriate header design
243
8.3 Observations
Through the experiments carried out on both compact brazed heat exchangers (CBE´s) and air
coolers it was discovered that there were still ice particles left in the fluid at the outlet from
the heat exchanger although the temperature at the outlet was higher than the freezing point
temperature.
This phenomenon is defined as superheating of the carrier fluid (SH) and in reality it means
that the liquid is in thermodynamic non-equilibrium with the ice particles. This was visually
observed through transparent sections of pipe right after the heat exchangers. Further
downstream of the return tubes, the ice crystals were melted and the temperature of the liquid
decreased again and reached the freezing point temperature, which means that the liquid
reached the equilibrium condition. See Figure 8.23.
Figure 8.23. Schematic of temperature development through a CBE

and further downstream when superheating (SH) occurs
For operating heat exchangers with multiple parallel passes, it is important that the
distribution of ice slurry is uniform. If the distribution is non-uniform in the inlet
distributor/header, the melt off rate will vary in the parallel passes which can create a
temperature gradient at the outlet. This can affect a temperature measurement at the outlet if
the probe is located immediately after the heat exchangers.
In the CBE experiments it was deemed too difficult to measure the temperature at the outlet of
each pass. It was, however, possible to conduct this measurement in the air cooler
experiments. Therefore, the surface temperature was measured on six different tubes at the
outlet of the pass just before the return header on the 3/8” coil, see Figure 8.24.
244
Figure 8.24. Location of thermocouples at the outlet of 6 parallel passes, just before the return
header. Thermocouple no. 3 was not used. There were 14 passes in all
The results are shown in Figure 8.25.
1 2.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29
0 1.8
-1 1.5
-2 1.3
T [K]
T [°C]
-3 1.0
-4 0.8
-5 0.5
-6 0.3
Tube_1 Tube_2 Tube_4 Tube_5 Tube_6 Tube_7
-7Ice_on Ice off SH 0.0
Figure 8.25: Surface temperatures of six outlet tubes in 29 experiments.
The ice slurry temperature was measured with a Pt 100 probe in the liquid both at the inlet
and outlet. Those two measurements are labelled as “Ice On” and “Ice Off” respectively. SH
stands for superheat and is the only measurement that refers to the right y-axis. The number at
the top of the diagram identifies the number of experiments. Hence the data points in the
columns under each number (1-29) correspond to one experiment
If the degree of super-heating is related to non-uniform distributions between the passes, the
degree of superheating should be most severe when the difference between the surface
temperatures of the outlet tubes is the largest. In experiment number 4, the difference between
the highest and lowest tube surface temperatures was 3.0 K and the degree of superheating
was 1.6 K (third highest), which is consistent with the theory. In experiment number 18,
however, the tube temperature difference was 0.8 K and the degree of superheating was 1.8 K,
245
which does not correspond very well to this theory. Consequently, the degree of superheating
is believed to be not solely dependent on the distribution of ice slurry in the
header/distributor.
Plotting the degree of superheat against the melt off rate of ice slurry in the air cooler
experiments shows some tendencies, see Figure 8.26.
30 % ice 3/8
2,0 20 % ice 3/8
1,8 10 % ice 3/8
1,6 10 % ice 12
1,4 20 % ice 12
30 % ice 12
SH (K)
1,2
1,0
0,8
0,6
0,4
0,2
0,0
0 5 10 15 20 25 30 35 40
dxice (% wt.)
Figure 8.26. Super-heating of carrier fluid at the outlet. The parameter on the x-axis is the
melt off rate, and the ice concentration in the legend is the inlet concentration. The numbers
3/8 and 12 identify the two different air coils tested and refer to the diameter of the tubes
The results showed that it is relatively difficult to melt off the last 5-10 wt-% ice. This is
related to the heat transfer rate between the ice crystals and carrier fluid plus the thermal
conductivity of the mixture. When the ice concentration was high, for example 30 wt-% vs. 0
wt-%, the heat conduction ability of the mixture is almost twice as high, which will lead to a
quicker equalisation of any temperature differences in the mixture.
Furthermore, the crystal size also influences the temperature difference required to melt off an
ice crystal. The amount of time required to melt off ice crystals increases with the size of the
crystals. The crystal sizes were not measured in the experiments but they probably varied
between 50 and 1000 microns or more, because the ice slurry was stored over many days.
According to Hansen et al. (ASHRAE, 2002) the sizes of ice crystals can increase from 50 to
700 microns over a 4-day period. In some of the experiments the ice slurry had been stored
for at least a week or more. In addition the distribution of crystal sizes also increases over
time (ASHRAE, 2002).
When melting ice slurry, smaller crystals vanish before the larger ones because the ratio of
surface area to volume increases, as the crystals become smaller. So, the crystals that were
visible at the outlet of the heat exchangers were initially the largest.
In the plate heat exchanger experiments the degree of superheating was generally a little
higher than in the air coolers. These experiments were all planned so that all the ice would be
melted at the outlet. This is consistent with the findings in the air cooler experiments that it
246
requires a relatively much larger heat transfer area to melt off the last 5-10 wt-% ice, see
Figure 8.27.
3,0 B10 30% 4ch

B10 20% 4ch
2,5 B10 10% 4ch
B10 30% 5ch
2,0
B10 20% 5ch
SH (K)
1,5 B10 10% 5ch

B5 30% 13ch
1,0 B5 20% 13ch
B5 10% 13ch
0,5 B15 30% 4ch
B15 20% 4ch
0,0
B15 10% 4ch
0 25 50 75 100 125 150 175
Re (-)
Figure 8.27. Superheating (SH) of the carrier fluid in all the CBE experiments plotted
against the average Reynolds number
The experiments carried out on the A-type (B15) plate heat exchanger showed the lowest
superheat and those carried out on the C-type (B5) the highest. There is no clear relation
between the Reynolds number and the degree of superheating. Furthermore there is no clear
relation among superheat and ice content, heat flux or the duration of ice particles inside the
heat exchanger. At this point the only two likely explanations left are: 1) the particle size,
which varied throughout the experiments, again in the range from 50 to 1000 microns; and 2)
flow distribution problems between the parallel passes.
Thus, a series of measurements were carried out on the C-type (B15) heat exchanger where
the time dependency of heat transfer coefficient, pressure drop and degree of superheating
was investigated. The duration of each experiment was 10-15 minutes after which the heat
source was turned off leaving the ice crystals to grow in the accumulation tank. After a certain
period another experiment was initiated and so on. The results are shown in Figure 8.28.
247
1800 4
1700 3
SH (K)
h (W/m2K)
1600 2
1500 HTC 1
SH
1400 0
0 10 20 30 40 50 60
Time (h)
Figure 8.28. The experimental results on time dependency in the C-type (B5) heat exchanger
(CBE) with constant mass flow rate and full melt off from an inlet concentration of 20 wt-%.
Heat transfer coefficient (HTC) refers to the left y-axis labelled “h”
and the superheat refers to the right y-axis labelled “SH”
Even with small ice crystals superheat can occur. There is, however, a tendency which shows
an increase in the superheat of 30 % and a decrease of 10 % in the heat transfer coefficient as
the crystals increase in size. Because the superheat is 2.5 K after only half an hour it can be
concluded that the superheating phenomenon is not only caused by the delayed melt-off of
relatively large ice crystals. The C-type (B5) heat exchanger showed the highest degree of
superheat. It had 28 plates compared with 10 for both the A-type (B15) and B-type (B10).
Hence it is likely that the superheat in plate heat exchangers is caused mainly by poor flow
distribution between the parallel passes, although this was not actually measured in the
experiments.
8.4 Conclusions
Three different plate heat exchanger models were tested with ice slurry. The results showed
very low pressure drop and high heat transfer capacity compared to single-phase media. The
optimal ice concentration was found to be between 20-30 wt-% at the inlet. The most efficient
geometry of the three plate heat exchangers was the short/slim and the one that had the
highest number of parallel passes. The heat transfer coefficients were not higher than what is
obtainable with a single-phase fluid, because the Reynolds numbers were very low.
The carrier fluid properties dominate the heat transfer characteristics at low Reynolds
numbers, because the heat flux is relatively high in a plate heat exchanger compared with
other types of heat exchangers. This means that the ice particles near the wall are quickly
melted off leaving only the carrier fluid to dominate the heat transfer. A break point for the
heat transfer coefficient exists around a heat flux value of 10-15 kW/m2 in the tested range of
Reynolds numbers. The heat flux should not be lower than this value if optimal heat transfer
performance on the ice slurry side is desired.
248
Correlations have been developed to calculate heat transfer coefficients and pressure drop in
three different compact brazed heat exchangers. The analysis showed that it is impossible to
convert constants obtained for a specific CBE geometry with single-phase fluids to those for
an ice slurry. Hence it is necessary to find new constants for every new heat exchanger
geometry if reasonable accuracy is required.
Two air coolers designed for single-phase media have been tested with ice slurry and
correlations have been developed. Analyses of the experimental results showed that at a
constant flow velocity, the internal heat transfer coefficient was improved by a factor of three
to four and the corresponding pressure drop was increased by a factor of up to two compared
to a single-phase fluid.
It was also found feasible to simplify the fluid circuitry in the air coil investigated and thereby
reduce the production cost.
1. ASHRAE Final Reports 1166: Behaviour of Ice Slurries in Thermal Storage Systems.
Principal Investigator: Michael Kauffeld, Danish Technological Institute; Conducted:
April 2000-Oct. 2001, published in 2002.
249
CHAPTER 9. DIRECT CONTACT CHILLING AND FREEZING OF FOODS IN

ICE SLURRIES
by Kostadin Fikiin, Ming-Jian Wang, Michael Kauffeld and Torben M. Hansen (see specific
list of symbols in Appendix 3)
Ice slurry has received increasing attention and demand for refrigeration processes in different
industries because of the widespread concerns over product quality, process efficiency and
environmental friendliness. This chapter deals with the application of ice slurry technologies
to direct contact chilling and freezing of foods. As a rule, direct contact cooling in ice slurries
improves the product quality. To date this technology has mainly been used in the fish
industry (by employing sea-water-based ice slurry) but its recent applications to the fruit and
vegetable processing sectors revealed a very promising potential as well. Both laboratory and
industrial trials demonstrated convincingly the superiority of the ice-slurry-based immersion
methods over conventional modes of food refrigeration.
Direct contact cooling by dried ice slurries has lately been applied in supermarket display
cabinets. Ice slurry is thereby used similarly to flake ice, but the slurry handling is much
easier because it can be pumped to the display cases and released via a hose, whereas the ice
flakes have to be shovelled. Moreover, there are no traumatic effects on the food surface,
which may occur when flakes are involved. Although the cooling ability of the ice slurry is
very similar to that of flake ice in this particular application of dried ice slurry, ice slurry
shows heat transfer benefits for the direct contact/immersion applications described below.
9.1. State of the art and conventional modes of food refrigeration
Let us, for instance, illustrate the ice slurry capabilities for food freezing applications as
compared with the most common techniques known so far. In the early 1900s, many people
were experimenting with mechanical and chemical methods to preserve food. As an industrial
process, quick freezing began its history some 70 years ago when Clarence Birdseye found a
way to flash-freeze foods and deliver it to the public – one of the most important steps
forward ever taken in the food industry. During his stay in the Arctic, Birdseye observed that
the combination of ice, wind and low temperature almost instantly froze just-caught fish.
More importantly, he also found that when such quick-frozen fish were cooked and eaten,
they were scarcely different from the fresh fish in taste and texture. After years of work,
Birdseye invented a system that packed dressed fish, meat or vegetables into waxed-
cardboard boxes, which were flash-frozen under pressure (US Patent No. 1,773,079, 1930).
Then he turned to marketing and a number of ventures were initiated to manufacture,
transport and sell frozen foods (e.g. construction of double-plate freezers and grocery display
cases; lease of refrigerated boxcars for railway transport; and retail of frozen products in
Springfield, Massachusetts, in 1930). These technological achievements constituted the
world's first cold chain for frozen foods, which became shortly a legend (Fikiin, 2003).
Thus, quick freezing has further been adopted as a widespread commercial method for long-
term preservation of perishable foods, which improved both the health and convenience of
virtually everyone in the industrialised countries. Freezing rate affects strongly the quality of
frozen foods, in which the predominant water content should quickly be frozen in a fine-grain
crystal structure in order to prevent damages to the cellular tissues and to inhibit rapidly the
spoiling microbiological and enzymatic processes.
251
Basic heat transfer considerations (Fikiin, 2003) clearly suggest that the desired shortening of
freezing duration and a resulting high throughput of refrigerating equipment could be
achieved by means of: (i) lower refrigerating medium temperature (which generally requires
greater investment and running costs for the refrigeration machines to be employed), (ii)
enhanced surface heat transfer coefficients (by increased refrigerating medium velocity and
boundary layer turbulence, involvement of surface phase-change effects and less packaging),
and (iii) reduced size of the refrigerated objects (by freezing small products individually or
appropriately cutting the large ones into small pieces).
Air-blast and multiplate freezers are most widespread, while air fluidizing systems are used
for individual quick freezing (IQF) of small products. The cryogenic IQF is still very
restricted because of the high prices of the liquefied gases used.
Fluidized-bed freezing systems

Air fluidization has been studied extensively and used commercially, with increasing
popularity, over the last forty years (Fikiin et al., 1965, 1966, 1970; Fikiin, 1969, 1979, 1980).
This freezing principle possesses many attractive features, including:
• High freezing rate due to the small sizes and thermal resistance of the IQF products, large
overall heat transfer surface of the fluidized foods and high surface heat transfer
coefficients.
• Good quality of the frozen products, that have an attractive appearance and do not stick
together.
• Continuity and possibilities for complete automation of the freezing process.
In spite of these advantages, fluidization freezing by air has some drawbacks, such as:
• Necessity of two-stage refrigerating plants (often using large quantities of CFC-, HCFC-
or HFC-based refrigerants with significant ozone depletion or global warming potentials)
to hold an evaporation temperature of about –45°C, which results in high investment and
power costs.
• Lower surface heat transfer coefficients and freezing rates in comparison with the
immersion methods described below.
• Need for a high speed and pressurized airflow, that results in large fan power
consumption.
• Some moisture losses from the product surface and rapid frosting of the air coolers,
caused by the large temperature difference between the products and the evaporating
refrigerant.
• Excessive sensitivity of the process parameters to the product shape, mass and size, that
requires careful control specific to every separate food commodity.
Freezing by immersion
The immersion freezing in non-boiling liquid refrigerating media is a well-known method
having several important advantages: high heat transfer rate, fine ice crystals in foods, high
throughput, low investments and operational costs (Woolrich, 1966; Tressler, 1968; Fleshland
and Magnussen, 1990; Lucas and Raoult-Wack, 1998). The immersion applications have been
limited because of the uncontrolled solute uptake by the refrigerated products and operational
problems with the immersion liquids (high viscosity at low temperatures, difficulty in
maintaining the medium at a definite constant concentration and free from organic
contaminants). Recent achievements in heat and mass transfer, physical chemistry, fluid
252
dynamics and automatic process control make it possible to solve these problems and to
develop advanced immersion individual quick freezing (immersion IQF) systems (Fikiin and
Fikiin, 1998, 1999a, 2002, 2003a,b; Fikiin, 2003).
9.2.　Unfreezable liquids and pumpable ice slurries as refrigerating media and

fluidizing agents
The Hydro Fluidization Method (HFM) for fast freezing of foods was suggested and patented
recently to overcome the drawbacks and to bring together the advantages of both air
fluidization and immersion food freezing techniques (Fikiin, 1985, 1992, 1994). The HFM
uses a circulating system that pumps the refrigerating liquid upwards, through orifices or
nozzles, in a refrigerating vessel, thereby creating agitating jets. These form a fluidized bed of
highly turbulent liquid and moving products, and thus evoke extremely high surface heat
transfer coefficients. The principle of operation of an HFM freezing system is illustrated in
Figure 9.1.
253
Figure 9.1. Possible arrangements of a HFM-based freezing system combining the advantages
of both air fluidization and immersion food freezing techniques (Fikiin and Fikiin, 1998,
1999a): (1) charging funnel; (2) sprinkling tubular system; (3) refrigerating cylinder; (4)
perforated screw; (5) double bottom; (6) perforated grate for draining; (8) sprinkling device
for glazing; (7 and 9) netlike conveyor belt; (10 and 11) collector vats; (12) pump; (13 and
14) rough and fine filters; (15) cooler of refrigerating medium; (16) refrigeration plant
Unfreezable liquid refrigerating media as fluidizing agents

Although various immersion techniques have been known for a long time, until now
hydrofluidization principles have not been used for chilling and freezing of foods.
Experiments on HFM freezing of small fish and some vegetables through an aqueous solution
254
of sodium chloride showed a much higher freezing rate when compared with other IQF
techniques (Fikiin, 1992, 1994). The maximum surface heat transfer coefficient achieved
exceeded 900 W/(m2K), while this was 378 W/(m2K) when immersing in a flowing liquid,
432 W/(m2K) for sprinkling and 475 W/(m2K) for immersion with bubbling through (Fikiin
and Pham, 1985). Even at a slight or moderate jet agitation and a comparatively high
refrigerating medium temperature of about –16°C, scad fish were frozen from 25°C down to
–10°C in the centre within 6-7 minutes, sprat fish and green beans within 3-4 minutes and
green peas within 1-2 minutes. As an illustration Figure 9.2 shows recorded temperature
histories during hydrofluidization freezing of scad and sprat fish, green beans and peppers.
a b
30 30
25 25
Ambient temperature = –15/–16 C o Ambient temperature = –15/–16 o C
20 20
Average jet velocity = 2 m s–1 Average jet velocity = 2 m s–1
15 15
Temperature ( C)
Temperature ( C)
o
o
10 10
Core
5 5
Centre
Peppers (30 g)
0 0
-5 -5
Scad (28 g)
-10 -10
Surface
-15 Surface -15
Sprat (7g) Green beans (11 g)
-20 -20
0 100 200 300 400 500 600 700 800 0 50 100 150 200 250 300 350
Time (s) Time (s)
Figure 9.2. Experimental temperature histories during HFM freezing of some kinds of (a) fish
and (b) vegetables, when using sodium chloride solution (without ice slurry) as a fluidizing
agent (Fikiin, 1992; Fikiin and Fikiin, 1998, 1999a)
Two-phase ice slurries as fluidizing agents

Pumpable ice slurries (known under different trade names, such as FLO-ICE, BINARY ICE,
Slurry-ICE, Liquid ICE, Pumpable ICE or Fluid ICE) were proposed recently as
environmentally benign secondary coolants circulated to the heat transfer equipment of
refrigeration plants, instead of the traditional ozone-depleting CFC- or HCFC-based
refrigerants (Paul, 1995; Ure, 1998, Egolf et al., 1996; Bel and Lallemand, 1999; Pearson and
Brown, 1998). Promising attempts to refrigerate foods by immersion in such slurries have
already been carried out. As already discussed, fish chilling in sea water-based ice slurries has
good potential to replace the traditional use of ice flakes (Fikiin et al., 2002). A number of
foods immersed in slurries with various ice contents are shown in Figure 9.3.
255
Figure 9.3. Different foods immersed in slurries with various ice concentrations: (a) fruits; (b)
vegetables; (c) chickens; (d), (e) and (f) fish (Fikiin et al., 2002)
Fikiin and Fikiin (1998, 1999a) launched, therefore, the idea to enhance the advantages of
hydrofluidization (described above) by employing two-phase ice suspensions as fluidizing
media. Ice slurries possess a large energy potential as HFM refrigerating media whose small
ice particles absorb latent heat when thawing on the product surface. Hence, the goal of ice
slurry usage is to provide a high surface heat transfer coefficient (of the order of 1000-2000
W/(m2K) or more), shortened freezing time and uniform temperature distribution in the whole
volume of the freezing apparatus. The combination of the HFM with the high heat transfer
efficiency of the ice-slurry-based refrigerating media represents a new interdisciplinary
research field whose development would advance essentially the refrigerated processing of
foods. The HFM freezing with ice slurries can acquire a process rate approaching that of the
cryogenic flash freezing modes. For instance, at a refrigerating ice-slurry temperature of
–25°C and a heat transfer coefficient of 1000 W/(m2K), strawberries, apricots and plums can
be frozen from 25°C down to an average final temperature of –18°C within 8-9 minutes;
raspberries, cherries and morellos within 1.5 to 3 minutes; and green peas, blueberries and
cranberries within about 1 minute only. The general layout of an ice-slurry-based system for
hydrofluidization freezing is shown in Figure 9.4.
Advantages of hydrofluidization freezing with ice slurries

As described above, the novelty of the hydrofluidization method lies in the involvement of
unfreezable liquids or pumpable ice slurries as fluidizing agents. It is well-known that the
immersion freezing history began with the use of brines to freeze fish, vegetables and meat.
Binary or ternary aqueous solutions containing soluble carbohydrates (e.g. sucrose, invert
sugar, glucose [dextrose], fructose and other mono- and disaccharides) with additions of
ethanol, salts, glycerol, etc., have been studied as possible immersion media. There are
practically unlimited possibilities to combine constituents and to formulate appropriate
256
Figure 9.4. Schematic diagram of an ice-slurry-based hydrofluidization system HyFloFreeze®

(Fikiin and Fikiin, 1998, 1999a)
multicomponent HFM refrigerating media based on single-phase liquids or two-phase ice

slurries, which have to be both product- and environment-friendly and to possess a low
enough viscosity in terms of pumpability and good hydrofluidization.
The main advantages of hydrofluidization over the conventional freezing modes can be
summarized as follows:
• The HFM facilitates a very high heat transfer rate with a small temperature difference
(product to cooling medium). The evaporation temperature can be maintained much
higher (at –25/–30°C) achievable by a single-stage refrigerating machine with much
higher COP and nearly half the investment and energy costs as compared with the
conventional air fluidization. Cold dissipation through the freezer walls is subsequently
also lower. The water flow rate or fan power consumption for cooling the condenser
decreases as well, due to the reduced mechanical work of the refrigeration unit running at
higher evaporation temperature.
• The critical zone of water crystallization (from –1 to –8°C) is quickly passed through,
which ensures a fine ice crystal structure in foods preventing the cellular tissues from
perceptible damage.
257
• The product surface freezes immediately in a solid crust that hampers the osmotic
transfer and gives an excellent appearance. The water losses tend to zero, while in air
freezing tunnels the moisture losses are usually 2-3%.
• Delicious new products can easily be formulated by using some selected product-friendly
HFM media (for example, fruits frozen in syrup-type sugar solutions turn into dessert
products with beneficial effects on colour, flavour and texture). Such media can also
include appropriate antioxidants, flavourings and micronutrients to extend the shelf life
of the products and to improve their nutritional value and sensory properties.
• The HFM freezers use environmentally friendly secondary coolants (for instance, syrup-
type aqueous solutions and ice slurries) and the primary refrigerant is closed in a small
isolated system, in contrast to the common air fluidization freezers where large quantities
of ozone depleting and/or high global warming potential CFCs, HCFCs or HFCs are
circulated to remote evaporators with a much greater risk for emission to the
environment.
• Fluidized state is acquired with low velocity and pressure of the fluid jets due to the
Archimedes forces and buoyancy of the products, that lead to both energy savings and
minimum mechanical action on the foods.
• The operation is continuous, easy to maintain, convenient for automation and the labour
costs are substantially reduced. Further processing or packaging of the HFM-frozen
products is considerably easier since they emerge from the freezer in a “free-flowing”
state, i.e. do not stick together.
• Ice-slurry-based HFM agents may easily be integrated into systems for thermal energy
storage, accumulating ice-slurry during the night at lower electricity costs.
The top view photos on Figure 9.5 show how a hydrofluidized bed of highly turbulent ice
slurry is formed inside the HyFloFreeze® prototype's freezing compartment.
Figure 9.5. HyFloFreeze® prototype: hydrofluidized bed of highly turbulent ice slurry
(Fikiin, 2003).
International research co-operation

Two main applications of the suggested HFM freezing technique can clearly be distinguished:
(i) employment of unfreezable liquids as fluidizing agents and (ii) use of pumpable ice
slurries as fluidizing media. This freezing principle provides an extremely high heat transfer
rate, short freezing times, great throughput and better product quality at higher refrigerating
temperatures. Thus, only about half the investment and power costs are necessary as
compared with the popular individual quick freezing methods. Moreover, such
258
hydrofluidization freezing systems are less hazardous from an environmental viewpoint, since
the primary refrigerant is limited to a small isolated circuit.
The emerging HFM technology has attracted the attention of a number of academics and
industrialists. The identification of optimal design specifications for HFM freezing systems
requires an interdisciplinary approach by researchers with complementary skills. The
HyFloFreeze project was, therefore, funded by the European Commission and performed by
an international research consortium of six participating organisations (four universities and
two SMEs) from Belgium, Bulgaria, Russia and the UK (Fikiin, 2003).
9.3. Cooling of fish with ice slurries
The use of ice for extending the storage life of fish dates back many millennia. Up to the
middle of the last century all ice used for fish cooling was from natural sources (winter snow
or imported arctic ice). With the introduction of mechanical cold production, ice was and is
produced in different forms, e.g. block, cube, tube or flake ice. Most of these forms need a
certain degree of manual operation for transportation from one place to another, and have
rather sharp edges capable of damaging the fish surface. Furthermore, they are usually quite
coarse, resulting in poor heat transfer. The introduction of ice slurry for direct contact cooling
of fish (Figure 9.6) presents several attractive features to the process operation over other
forms of ice (Wang et al., 2000).
For quality assurance it is essential that effective product cooling be provided throughout the
entire production chain, from catch at sea, to storage, transportation, and processing. Cooling
technologies employed often use refrigerated seawater and/or different types of fresh ice. Sea
water systems placed on board large vessels maintain fish, by sea water circulation, at a mean
temperature of about 2-4°C. Disadvantages of such systems are the volume taken up by the
refrigeration machinery, difficult working conditions and salt uptake by the fish, especially in
smaller species like sardine. Onboard smaller ships, fresh ice is normally loaded from
harbour, stored and eventually mixed with the catch. The tendency of fresh ice to re-
crystallize and agglomerate in many cases limits the quality of icing, since the ice has to be
loosened by manual means before mixing. Often large blocks of ice mixed with the fish result
in uneven cooling rates.
Bacterial growth is highly temperature-dependent. As a rule of thumb, the shelf life of fish
kept at 0°C is more than doubled compared to +5°C; the closer to the freezing point of the
fish, the better the product quality. Bearing in mind that bacterial growth rates can not be
stopped, but only decelerated, it is also important to minimize the bacterial level by a rapid
temperature drop in each part of the process. Possible freezing of the fish anticipated for
chilling may result in decreased sensory perception (Sørensen, 1999) and should therefore be
avoided. Intuitively, the use of ice slurry could improve cooling of fish by exploiting
properties such as (Wang et al., 2000):
• narrow approach to the freezing point of fish;
• close to isothermal heat exchange during melt-off;
• smooth surface preventing fish from damage;
• easy dosing, mixing, handling and pumping;
• possible drainage of salty water to minimize salt uptake.
259
Figure 9.6. Fish chilling in ice slurry (Fikiin et al., 2002): (a) land-based systems and
(b) on-board systems (Liquid IceTM, Brontec, Iceland)
Because of its tiny ice crystals, ice slurry is soft and flexible for the chilled product. It
effectively avoids any hot spots in the fish container, and provides excellent contact with fish
without bruising. As fish are surrounded by numerous ice crystals, high cooling rate is
achieved. This significantly retards bacterial growth and reduces fish tissue degradation.
Analysis of cooling methods and process simulation

Ice slurry for fish cooling and controlling temperature during storage can be implemented in
different ways. In order to verify the aforementioned advantages of ice slurry a simulation
model has been developed and related experiments have been carried out (Hansen et al.,
1999). To simulate the cooling process and to check whether freezing occurs near the fish
260
surface, a differential heat conduction equation was solved by using finite-element-based

CFD software:
div(k grad T)= ρ ⋅c p ∂T (9.1)

∂τ
Thermal properties are calculated by considering that phase change of water will occur over a
wide temperature interval (see Figure 9.7). The majority of water (70-90%) is mechanically
trapped in and between cells, and is relatively easy to freeze. During freezing the salt content
increases, resulting in a temperature glide of phase change. Individual freezing curves may
vary among species, although the characteristics remain the same, i.e. thermo-physical
properties are strongly influenced down to a temperature of approximately –10°C. The
behaviour of phase change described has a large effect on the thermo-physical properties that
vary with the temperature throughout a frozen fish. The composition of fish changes a lot
among species and depends on the place and season of harvesting, which may also affect
thermo-physical properties. White fish may be considered relatively stable containing about
80% water and 20% dry materials. For fatty fish like horse mackerel and sand eel, the content
of dry materials may remain approximately 20% during the season whereas the oil content
varies from 1 to 20%. Temperature-dependent thermo-physical properties for white fish (80%
water) and fatty fish (70% water, 10% oil) are shown in Figure 9.7.
100 50 80 1.6
White fish
Thermal conductivity (W/m/K)

90 15
70 1.4
Frozen water content (%)
80 -20
Specific enthalpy (kJ/kg)
Specific heat capacity
60 1.2
70 -55 Fat fish
50 1
(kJ/kg/K)
60 -90
50 -125 40 0.8
40 -160 30 0.6
30 -195
20 0.4
20 -230
10 Fat fish -265
10 0.2
0 White fish -300 0 0

-30 -25 -20 -15 -10 -5 0 5 -30 -20 -10 0
Temperature (°C) Temperature (°C)
Figure 9.7. Enthalpy-temperature curve from differential scanning calorimetry of cod fillets
(Sørensen and Spange, 1999) and calculated thermophysical properties
Based on the freezing curve, the enthalpy can be calculated as a function of temperature by
assuming constant heat capacities of constitutient substances. For fat fish neglecting the effect
of oil on crystallization may cause some errors. The specific heat capacity of fish is calculated
as follows:
∆h = ∆T ⋅ (X d c p ,d + X oil c p ,oil + ( X w − X i ) c p , w + X i c p ,i ) + X i L (9.2)
261
⎛ ∂h ⎞
c p = ⎜⎜ ⎟⎟ (9.3)
⎝ ∂T ⎠P
The thermal conductivity, k, is determined by the assumption that the overall resistance to
conduction is due to the resistance of parallel layers of each substance in parallel (=) or
perpendicular (⊥) to the heat flow direction. It is further assumed that both cases are equally
represented, i.e.:
k = ½(k = + k ⊥) (9.4)
Experiments with artificial and real fish samples

To eliminate biological and seasonal variation of fish properties when comparing cooling
methods, artificial fish samples with thermophysical properties close to those of white fish
have been made out of Karlsruhe food simulator (a mixture of methylcellulose, sodium
chloride and water). The aim of the experiments was to compare cooling of fish in containers
by using ice slurry or flake ice. The distribution of fish is assumed to be in perfect layers of
fish and ice respectively. Three different cooling scenarios were set up, two of them
representing an ideal distribution of either flake ice or ice slurry on top of each fish layer, and
the third assuming larger re-crystallized/agglomerated blocks of ice covering only 25% of the
total surface area, while liquid film heat transfer is present on the remaining surface.
Measurements were made for an artificial fish model with thermocouples placed internally.
The fish was placed between artificial dummies in the middle of the container. A total of 20%
pure ice was employed for each experiment. Results obtained were compared with the data
for horse mackerel. For modelling purposes, fish were assumed to be cylinders of infinite
length with adiabatic contact lines between fish. Temperatures of flake ice and ice slurry were
set to 0 and –3°C, respectively. The heat transfer coefficient between the ice particles and fish
was estimated to be 750 W/(m²K), a value affected by the resistance of the melting liquid
layer between the fish surface and ice. Liquid film heat transfer coefficient was assumed to be
50 W/(m²K).
Figure 9.8 shows a reasonable agreement between calculated and measured temperature
histories. However, it is also evident that a simple set of boundary conditions is not sufficient
to describe the real situation in the container and some deviation may exist, especially for the
time interval when almost all the ice has melted (1200 to 1600 seconds). Furthermore, it
appears that the centre temperature decreases at a higher rate when using ice slurry at –3°C.
This is even more obvious at the surface. The calculated temperature histories indicate that
freezing does not occur at the outer shell if the ice slurry is kept at above –3°C. The final
equilibrium temperature of the fish was approximately 1 K lower when using the same
amount of ice slurry versus using flake ice. The difference is caused by the larger enthalpy
change of the melting ice slurry.
262
20 0.0
20
18 0.0
18 Horse mackerel, 15% ice
0.0
16 Flake ice blocks 16
-0.0
14 14
Temperature (°C)
Temperature (°C)
-0.0
12 -0.0 -0.0 0.0 0.0 0.0 12
10 Measurements
10
8 FEM model 8 Flake ice
Flake ice
6 6
4 Ice slurry 4
Ice slurry
2 2
0 0
0 1200 2400 3600 0 1200 2400
Time (s) Time (s)
Figure 9.8. Variation in centre temperature (left graph) measured in an artificial fish model
(43 mm diameter) cooled in a container with ice slurry at –3°C and with flake ice at 0°C,
together with calculated temperatures (initially 20% ice in the container). Temperature
measurement (right graph) in horse mackerel cooled in a container with ice slurry and with
flake ice (initially 15% ice in the container)
The results from the experiments with the artificial fish model were comparable with those
with horse mackerel. A similar tendency was found, i.e., faster temperature drop and about 1
K lower final temperature when using ice slurry. The end cooling temperature was, however,
different because of the smaller amount of ice. If the distribution of flake ice is not ideal
because of re-crystallized blocks, Figure 9.8 reveals that cooling with flake ice is less
effective than with ice slurry.
Cooling experiments with a fish model immersed into a flowing stream of ice slurry with
different velocities and ice concentrations were also conducted to identify any advantages of
pre-cooling before container storage. Measurements were performed for flow velocities of 0.1
and 0.2 m/s and at ice concentrations of 10, 20 and 30 wt-%. Results showed that there is no
major difference in terms of cooling time for different combinations of these parameters.
Compared with the immersed fish model, no significant difference was detected when cooling
a container with perfect distribution.
9.4. Ice-slurry-based cooling of fruits and vegetables
The ice-slurry cooling method for fish can also be employed for fruit and vegetables, along
with other food commodities to be chilled or frozen. In many cases the uncontrolled uptake of
solutes (salt or alcohol) on the product surface is more undesirable than for fish (unless
osmotic phenomena are exploited to give added value to the product). Similar to the fish
cooling, much faster heat transfer rates can be obtained. Most notable is the improvement
when comparing previously air cooled products versus ice slurry cooled products. For
instance, the Danish Technological Institute investigated the chilling and freezing of pig
263
carcasses in ice slurry with good results, while a number of researchers around the world
performed promising experiments with various fruits and vegetables.
When using suitable product-friendly media, the osmotic phenomena on the surface can be
exploited to give added value to the products. In particular, by freezing of fruits in sugar-
ethanol-based aqueous solutions or pumpable ice slurries new delicious dessert products can
be formulated with beneficial effect on colour, flavour and texture due to the enzyme-
inhibiting action of the sugar (Fikiin and Fikiin, 1998, 1999). In addition, the take-up of food
additives (antioxidants, flavourings, aromas and micronutrients) is improved, which can result
in a better quality and extended shelf-life of the end product. Nevertheless, the available data
for the physical properties and the performance of such secondary fluids in food freezing
equipment are still too scarce. A set of predictive equations for the density, viscosity, thermal
conductivity and specific heat capacity of sugar-ethanol aqueous refrigerating media are given
by Fikiin et al. (2001).
Sugar-ethanol aqueous solutions and ice slurries suitable for immersion freezing of
fruits
Several important criteria must be taken into consideration when selecting liqueur-type
refrigerating liquids and ice-slurries pertinent for fruit freezing applications: (i) product
friendliness with regard to the end product quality, (ii) environmental friendliness, (iii)
suitable rheological properties in terms of good pumpability, and (iv) appropriate initial
freezing temperature (Fikiin et al., 2001). A number of test solutions were therefore prepared
at the St. Petersburg State University of Refrigeration and Food Technologies by using
distilled and deionized water, pure ethanol of food-admissible class and sugar (sucrose or
glucose). Solution compositions and freezing points are presented in Table 9.1. The estimated
uncertainty in the mass fraction data was ±0.1%.
Table 9.1. Compositions and initial freezing temperatures of the studied solutions
Solution Composition (mass fraction)

tf (°C)
Number water, xw ethanol, xe sucrose, xs
1 0.55 0.25 0.20 –28.0
2 0.60 0.25 0.15 –25.0
3 0.65 0.25 0.10 –22.0
4 0.55 0.20 0.25 –26.5
5 0.60 0.20 0.20 –23.5
6 0.65 0.20 0.15 –19.5
7 0.55 0.15 0.30 –23.0
8 0.60 0.15 0.25 –20.0
9 0.65 0.15 0.20 –16.5
Solution No. water, xw ethanol, xe glucose, xs tf (°C)
10 0.60 0.25 0.15 –24.5
An international taste panel considered that solutions No. 1, 2 and 10 posessed the best
sensory and physical properties, but some other solutions (such as No. 4, 5 and 6) were also
acceptable.
264
Predictive equations for the thermophysical and rheological properties of sugar-ethanol

aqueous solutions and ice slurries
A series of experiments were carried out and some theoretical approaches were employed to
determine basic thermal and rheological properties of the studied single- and two-phase
refrigerating media and to establish resulting regression relationships. The following
assumptions were made when deriving the predictive equations: (i) water is the only freezable
component of the solution (because, at atmospheric pressure, all the ethanol remains unfrozen
above –114.5°C), (ii) the temperature-dependent properties of the liquid solution are
extrapolated below the initial freezing point to estimate the unfrozen fraction properties, and
(iii) the ice fraction variation with the temperature conforms with the Raoult’s Law (Fikiin et
al., 2001).
Initial freezing temperature

Typical gradients of the time-temperature curves during slow freezing or thawing
(∼0.5°C/min) were registered to measure the initial freezing points of the studied solutions
(Table 9.1) and the following empirical equations were then established by a regression
analysis:
tf = a0 + a1. xe + a2.xs + a12.xe.xs (9.4)
where 0.15 ≤ xe ≤ 0.25, 0.10 ≤ xs≤ 0.30, and a0 = 10.9, a1 = –103.33, a2 = –51.67
and a12 = –66.67.
Density
The liquid densities were determined by a piezometric technique, which was previously tested
with reference liquids (water, ethanol and other well-investigated solutions). The
resulting regression equation is:
ρl = (b0 + b1 xe + b2 xs + b12 xe xs) + (c0 + c1 xe + c2 xs + c12 xe xs) t (9.5)
for tf ≤ t ≤ 21°C, 0.15 ≤ xe ≤ 0.25, 0.10 ≤ xs≤ 0.30 and coefficients are given in Table 9.2.
Table 9.2. Coefficients of Eq. (9.5)

b0 b1 b2 b12 c0 c1 c2 c12
981.8 –67.556 481.111 –255.556 0.30483 –3.10689 –2.06289 9.18444
Furthermore, the ice-slurry density can be expressed as follows:
ρis = [xi / ρi + (1 – xi)/ ρl ] –1 (9.6)
where xi is the mass fraction of ice, the ice density ρi = 916.8 (1 – 0.00015 t) and ρl is
determined by Eq. (9.5) for xs = xsin / [1 – (1 – tf / t) xwin ] and xe = xein / [1 – (1 – tf / t) xwin ].
Viscosity
Measurements of the liquid viscosity were carried out by using glass capillary viscometers
and the results obtained were approximated by the following equation:
265
ηl = D + E t + F t 2 (9.7)
where: D = d0 + d1 xe + d2 xs + d12 xe xs, E = e0 + e1 xe + e2 xs + e12 xe xs, F = f0 + f1 xe + f2 xs + f12

xes, tf ≤ t ≤ 21°C, 0.15 ≤ xe ≤ 0.25 and 0.10 ≤ xs ≤ 0.30. The empirical coefficients are given in
Table 9.3.
d0 –17.325 e0 1.556 f0 –0.04318
d1 79.222 e1 –6.162 f1 0.15612
d2 100.822 e2 –7.486 f2 0.18694
d12 –181.778 e12 15.611 f12 –0.40644
The model of Thomas (1965), valid for ice particle diameters between 0.1 and 45 µm, can
further be employed to evaluate the apparent viscosity of the ice slurry (please refer to
Chapter 3 for a detailed description of rheological ice slurry properties):
ηis = ηl [1 + 2.5 ϕi + 10.05 ϕi2 + 0.00273 exp (16.6 ϕi)] (9.8)
where the volumetric ice fraction ϕi = xi [ xi + (1 – xi) (ρi / ρl) ]–1 may vary between 0 and
0.625, while ηl is calculated by Eq. (9.7) for xs = xsin / [1 – (1 – tf /t) xwin ] and xe = xein / [1 – (1
– tf /t) xwin ].
Thermal conductivity
As is well-known, the thermal conductivity of a ternary solution does not comply with the
additive principle and cannot, therefore, be rigorously expressed through the properties of the
different constituents. Nonetheless, by using a quasi-binary assumption the overall liquid
conductivity could be considered as a compound function of the conductivities of ethanol and
water-sugar system. It turned out that the liquid thermal conductivity predominantly depends
on the ethanol content and could roughly be estimated on the basis of the ethanol fraction
only:
kl = (i0 + i1 xe + i2 xe2) + (l0 + l1 xe + l2 xe2) t (9.9)
for tf ≤ t ≤ 20°C, 0.10 ≤ xe ≤ 0.35 and equation coefficients are shown in Table 9.4.

i0 I1 i2 l0 l1 l2
0.50686 –0.573278 0.2401997 0.001347 –0.0028111 0.0011861
In accordance with the Maxwell model, the apparent thermal conductivity of the ice slurry
could be written in the form of Jeffrey (1973):
kis = kl (1 + 3 ϕi β + 3 ϕi2 β 2 χ) (9.10)
266
where χ = 1 + 0.25 β + 0.1875 β [ (α + 2) / (2 α + 3) ], β = (α – 1) / (α + 2), α = ki / kl ,

ki = 2.22 (1 – 0.0015 t), ϕi is as in Eq. (9.8), and kl is calculated by Eq. (9.9) for
xe = xein / [1 – (1 – tf / t) xwin ].
The dimensionless apparent ice-slurry viscosity, η* = ηis / ηl , and thermal conductivity, k* =
kis / kl , determined by Eqs (9.8) and (9.10) as a function of the volumetric ice fraction, ϕi, are
shown in Figure 9.9.
2,5
η∗ and λ∗
η∗ λ∗
1,5
0,5
0 0,05 0,1 0,15 0,2
Volumetric fraction of ice (ϕi)
Figure 9.9. Dimensionless apparent viscosity, η* = ηis / ηl, and thermal conductivity,
k* = kis / kl, versus ϕi. (The notation for thermal conductivity in the figure follows the symbol
tradition of Continental Europe, i.e. “λ” is used instead of “k”)
Specific heat capacity

General thermodynamic considerations make it possible to determine the specific heat
capacity of ternary solutions on the basis of the heat capacities of the solution components.
The data obtained for liquid sugar-alcohol aqueous solutions were fitted by the following
regression equation:
cp,l = (m0 + m1 xe + m2 xs + m12 xe xs) + (n0 + n1 xe + n2 xs + n12 xe xs) t (9.11)
for tf ≤ t ≤ 20°C, 0.15 ≤ xe ≤ 0.25, 0.10 ≤ xs ≤ 0.30 and empirical coefficients are presented in
Table 9.5.

m0 m1 m2 m12 n0 n1 n2 n12
4.192 –1.9272 –2.8172 0.0910 0.000020 0.005075 0.002675 0.000459
The apparent specific heat capacity of ice slurries and other water-containing systems can
easily be evaluated as follows (Fikiin and Fikiin, 1999b):
267
cp,is = cp,l – d [ xi L ] / d t (9.12)
where L = 334.2 + 2.12 t + 0.0042 t 2 , xi = xwin (1 – tf / t) and cl is determined through Eq.

(9.11) for xs = xsin / [1 – (1 – tf / t) xwin ] and xe = xein / [1 – (1 – tf / t) xwin].
It is obvious that t < tf in all the ice-slurry related equations, Eqs (9.6), (9.8), (9.10) and
(9.12).
A small-size prototype of HFM freezer for testing, demonstration and promotion of the HFM
technology is under investigation at the Technical University of Sofia and Interobmen Ltd. –
Plovdiv (Bulgaria). The hydrofluidization, conveyor and driving systems of the prototype are
shown in Figure 9.10.
Figure 9.10. Hydrofluidization, conveyor and driving systems of the HyFloFreeze® prototype
(Technical University of Sofia and Interobmen Ltd.): (a) top view, (b) overall view
The following initial design specifications were used when putting together the prototype
freezer:
(i) starting selection of fruits to be frozen: strawberries, raspberries, plums, apricots, morellos,
cherries, blueberries and cranberries; (ii) refrigerating media: unfreezable liquids or
pumpable ice slurries based on sugar-ethanol aqueous solutions, (iii) refrigerating medium
temperature: –25/–30°C, (iv) evaporation temperature of the refrigerant: –30/–35°C, (v)
throughput: 20-50 kg per hour (depending on the product); (vi) final temperature in the
frozen food centre: –12/–15°C; (vii) final average temperature of the product: –18/–20°C;
(viii) principle of operation: enhanced hydrofluidization throughout the whole freezer by
268
directed jets of fluidizing agent; and (ix) conveyor system: electronic regulation of the driving
motor revolutions for smooth variation of the conveyor speed and product residence time
depending on the individual freezing duration for each product.
The ice-slurry-based HFM technology possesses a series of advantages over the conventional
IQF modes, which can be summarised as follows (Fikiin and Fikiin, 1998, 1999; Fikiin et al.,
2001):
• Frozen fruit quality: high freezing rate; fine-grain crystal structure; sharp reduction of
the surface mass transfer; enzyme-inhibiting action of the sugar; easy incorporation of
antioxidants, flavourings, aromas and micronutrients and formulation of delicious
dessert-type frozen foods with extended shelf-life and improved nutritional value and
sensory properties.
• Energy and economic efficiency: higher refrigerating and evaporation temperatures;
possibility for single-stage refrigeration machine with reduced investment, maintenance
and energy costs; easy connection with systems for thermal energy storage; high
throughput and cost efficiency.
• Environmental friendliness: use of environmentally friendly secondary coolants (syrup-
type solutions or ice-slurries) and primary refrigerant closed in a small isolated system.
At some temperatures and concentrations, the pumping of sugar-ethanol-based slurry through

the hydraulic circuit and perforated bottom may be accompanied by intense foaming (Figure
9.5). Preliminary tests revealed that this adverse effect could be overcome by adding suitable
antifoaming agents.
Literature cited in chapter 9
1. Bel, O.; Lallemand, A. : Etude d’un fluide frigoporteur diphasique: 1 – Caractéristiques

thermophysiques intrinsèques d’un coulis de glace; et 2 – Analyse expérimentale du
comportement thermique et rhéologique. Int. J. Refrig., Vol. 22, No. 3: pp. 164-187,
1999.
2. Egolf, P.W.; Brühlmeier, J.; Özvegyi, F.; Abächerli, F.; Renold, P.: Properties of ice
slurry. Proc. IIR Conf. Applications of Natural Refrigerants, Aarhus, Denmark 1996,
IIF/IIR: pp. 517-526, 1996.
3. Fikiin, A.G.: Congélation de fruits et de légumes par fluidisation. Proc. Budapest Conf.,
IIF/IIR: pp. 197-203, 1969.
4. Fikiin, A.G. : Bases théoriques du procédé de fluidisation lors de l’intensification de la
congélation des fruits et des légumes. Proc. 15th Int. Congress Refrig., Venice, Vol. 4: pp.
221-230, 1979.
5. Fikiin, A.G.: Physical conditions of fluidized-bed freezing of fruits and vegetables.
Kholodilnaya Tekhnika / Refrig. Engng, No. 7: pp. 59-61 (in Russian), 1980.
6. Fikiin, A.G.: Method and system for immersion cooling and freezing of foodstuffs by
hydrofluidization. Invention Certificate No. 40164, Bulgarian Patent Agency INRA,
1985.
7. Fikiin, A.G.: New method and fluidized water system for intensive chilling and freezing
of fish. Food Control, Vol. 3, No. 3: pp. 153-160, 1992.
8. Fikiin, A.G.: Quick freezing of vegetables by hydrofluidization. In New Applications of
Refrigeration to Fruit and Vegetables Processing – Proceedings of IIR Conference,
269
Istanbul (Turkey), Refrigeration Science and Technology, International Institute of

Refrigeration, 1994-3: pp. 85-91,1994.
9. Fikiin, A.G.; Pham, V.H.: System for examination of heat transfer regimes during
hydrorefrigeration of foodstuffs. Invention Certificate No. 39749, Bulgarian Patent
Agency INRA, 1985.
10. Fikiin, A.G.; Ditchev, S.P.; Fikiina, I.K.: Principal parameters characterising the
fluidization of fruit and vegetable layers. Kholodilnaya Tekhnika / Refrig. Engng, No. 11,
pp. 33-37 (in Russian), 1966.
11. Fikiin, A.G.; Ditchev, S.P.; Karagerov, D.I.: Fluidized bed freezing system for fruits and
vegetables with various dimensions. Invention Certificate No. 10967, Bulgarian Patent
Agency INRA, 1965.
12. Fikiin, A.G.; Ditchev, S.P.; Karagerov, D.I.: Fluidized bed freezing apparatus AZF.
Kholodilnaya Tekhnika / Refrig. Engng, No. 7, pp. 55-58 (in Russian), 1970.
13. Fikiin K.A.: Novelties of Food Freezing Research in Europe and Beyond. Flair-Flow
Europe Synthetic Brochure for SMEs No.10 (ISBN: 2-7380-1145-4), INRA: Institut
National de la Recherche Agronomique, Paris (France), 55p., 2003.
14. Fikiin, K.A.; Fikiin, A.G.: Individual quick freezing of foods by hydrofluidization and
pumpable ice slurries. In Advances in the Refrigeration Systems, Food Technologies and
Cold Chain, Ed.: K. Fikiin, Proceedings of IIR Conference, Sofia (Bulgaria),
Refrigeration Science and Technology, International Institute of Refrigeration, 1998-6:
pp. 319-326 (published also in the AIRAH Journal, 2001, Vol. 55, No. 11, pp. 15-18),
1998.
15. Fikiin, K.A.; Fikiin, A.G.: Novel cost-effective ice-slurry-based technology for
individual quick freezing of foods by hydrofluidization. CD-Rom Proceedings of the
20th International Congress of Refrigeration, Sydney (Australia), ICR Paper No. 271,
1999a.
16. Fikiin K.A.; Fikiin A.G.: Predictive equations for thermophysical properties and enthalpy
during cooling and freezing of food materials. Journal of Food Engineering, Vol. 40, No.
1-2, pp. 1-6, 1999b.
17. Fikiin, K.A.; Fikiin A.G: Congelación individual rápida de alimentos por
hidrofluidificación y compuestos de hielo. Frío, Calor y Aire Acondicionado (Madrid),
Vol. 30, No. 334, pp. 22-27, 2002.
18. Fikiin, K.A.; Fikiin A.G.: Quick freezing of foods by hydrofluidization and pumpable ice
suspensions. Kholodilnaya Tekhnika / Refrig. Engng, No. 1, pp. 22-25 (in Russian),
2003a.
19. Fikiin K.A.; Fikiin A.G.: L'Ice Slurry (ghiaccio binario) per una surgelazione veloce
degli alimenti – Surgelamento singolo degli alimenti mediante idrofluidizzazione.
Industria & Formazione per il Tecnico della Refrigerazione e Climatizzazione (Milano),
No. 3, pp. 36-39, 2003b.
20. Fikiin, K.A.; Kaloyanov, N.G.; Filatova, T.A.; Sokolov, V.N.: Fine-crystalline ice slurries
as a basis of advanced industrial technologies: State of the art and future prospects.
Refrigeration Business (Moscow), No. 7, pp. 4-11 (in Russian), 2002.
21. Fikiin, K.A.; Tsvetkov, O.B.; Laptev, Yu.A.; Fikiin, A.G.; Kolodyaznaya, V.S.:
Thermophysical and engineering issues of the immersion freezing of fruits in ice
slurries based on sugar-ethanol aqueous solutions. Proceedings of the Third IIR
Workshop on Ice Slurries, Lucerne (Switzerland). International Institute of
Refrigeration, pp.147-154, 2001. (published also in EcoLibrium – Journal of AIRAH:
Australian Institute of Refrigeration, Air Conditioning and Heating, Vol. 2, No. 7, pp. 10-
15, 2003.)
270
22. Fleshland, O.; Magnussen, O. M.: Chilling of farmed fish. Proc. Aberdeen Conf., IIF/IIR,
pp. 185-192, 1990.
23. Hansen, T.; Wang, M.J.; Kauffeld, M.; Christensen, K.G.; Goldstein, V.: Application of
ice slurry technology in fishery. XXth Int. Cong. of Refr., Sydney, Australia, 19-24
September, 1999.
24. Jeffrey D.J.: Conduction through a random suspension of spheres, Proceedings of the
Royal Society London, Vol. A 335: pp. 355-367, 1973.
25. Lucas, T.; Raoult-Wack, A.L.: Immersion chilling and freezing in aqueous refrigerating
media: review and future trends. Int. J. Refrig., Vol. 21, No. 6: pp. 419-429, 1998.
26. Paul, J.: Binary ice as a secondary refrigerant. Proc. 19th Int. Congress Refrig., The
Hague, Vol. 4b: pp. 947-954, 1995.
27. Pearson, S.F.; Brown J.: Use of pumpable ice to minimise salt uptake during immersion
freezing. Proc. Oslo Conf., IIF/IIR: pp. 712-722, 1998.
28. Sokolov, V.N.; Fikiin K.A.; Kaloyanov, N.G.: Advantages, production and applications
of pumpable ice slurries as secondary refrigerants. BulKToMM Machine Mechanics,
Vol. 44, pp. 26-31 (in Russian), 2002.
29. Sørensen, B.S.: Resultater fra brug af kvalitetsindikatorer i modelforsøg, Workshop i
kvalitetsindikatorer, Danish Institute for Fisheries Research, 1999.
30. Thomas, D.G.: Transport characteristics of suspension – VIII: A note on the viscosity of
Newtonian suspension of uniform spherical particles, Journal of Colloid Science, Vol.
20: pp. 267-277, 1965.
31. Tressler, D.K.: Food freezing systems. In: Tressler, D.K., Van Arsdel, W.B., Copley,
M.J. The Freezing Preservation of Foods, Vol. 1, The AVI Publishing Co., Westport,
Connecticut: pp. 120-152, 1968.
32. Ure, Z.: Slurry-ice based cooling systems. Proc. Sofia Conf., IIF/IIR: pp. 172-179, 1998.
33. Woolrich, W.R.: Handbook of Refrigerating Engineering, Vol. 2: Applications. The Avi
Publishing Co., Westport, Connecticut, 434 , 1966.
34. Wang, M.J.; Hansen, T.M.; Kauffeld, M.; Goldstein, V.: Slurry Ice in Fish Preservation.
Infofish International 2, pp. 42-46, 2000.
271
CHAPTER 10. THE CONTROL OF ICE SLURRY SYSTEMS

By Jacques Guilpart and Laurence Fournaison
10.1 Introduction
As described in this Handbook, ice slurry systems are of utmost interest thanks to their wide
range of applications and to their good performance. Many authors have described the
advantages and drawbacks of this two-phase secondary fluid based on a pure thermal and
hydraulic analysis, but literature dedicated to the control of these systems is quite scarce. This
chapter aims to describe some possible and commonly used control principles for ice slurry
systems. These principles depend on considerations outlined in this chapter, a main concern is
the strategy of operational modes of the ice slurry system.
Two main operation mode strategies for an application of ice slurry systems exist:
The first consists of using the ice slurry tank as a simple energy storage system, without
paying attention to the ice-liquid phase separation inside the tank. In this case, the residual
pure liquid should be pumped from the bottom of the tank and used in a very classical
single-phase secondary fluid loop. Many systems dedicated to air conditioning are based
on this strategy, especially in Japan (Wang et al., 1999).
The second strategy consists of taking advantage of the latent heat contained in the ice
crystals of the slurry at the point of consumption or use. In this case, the slurry has to be
maintained as homogeneous as possible in order to be pumped in the secondary loop. Of
course, the design of this loop has to deal with the particular thermal and rheological
behaviour of the slurry. Many systems dedicated to supermarkets or food process
applications are based on this principle (Kauffeld et al., 1999; Wang et al., 2002).
Depending on the chosen strategy, this chapter aims to describe the classical concepts applied
to control ice slurry systems. It will be shown that, despite some particularities related to the
eventual use of ice slurry in the secondary loop and in heat exchangers, the control of these
systems remains nearly as simple as the control of classical secondary loops working with
single-phase fluids.
10.2 General organization of an ice slurry system
Compared with a classical secondary fluid loop, the energy storage capability of ice slurry
often calls for storage tanks designed to shave the peak demand in cooling. Consequently, an
ice slurry system comprises an ice slurry generator charging a storage tank. Depending on the
application the liquid or the ice slurry is then pumped from the storage tank to the consumer
units, which are in a classical hydraulic loop with discharge and return lines as shown in
Figure 10.1. In Figures 10.1a and 10.1b, two possibilities are represented with dotted lines:
the use of single-phase residual liquid pumped from the bottom of a non-agitated tank and the
use of ice slurry pumped from the middle of an agitated tank.
A “mono-tube dynamic®” circuit could also be installed (Figure 10.1b): as the temperature
during the melting of the ice remains almost constant in the slurry, this type of distribution
loop is well adapted to servicing the consumer units.
273
Ice generator
Ice generator
Storage tank
Storage tank
Secondary Primary Secondary Primary
refrigerant refrigerant refrigerant
refrigerant
loop loop
loop loop
Figure 10.1a Figure 10.1b
Ice generator
Storage tank
Secondary Primary
refrigerant refrigerant
loop loop
Figure 10.1c
Figure 10.1. Some possible designs of an ice slurry system
There are some additional differences between single-phase fluid and ice slurry secondary
loops besides the ice slurry generator and storage tank. The heat exchangers of the consumer
units working with ice slurries have to be supplied using small local pumps (classical
centrifugal pumps are sufficient). It has also to be emphasised that the control of these
terminal units cannot be based on three way valves or throttling devices, because of the risk of
blocking the flow channels with plugs of ice.
With the exception of these differences, an ice slurry system is designed as a classical
secondary loop. A major consequence of this is that if a secondary fluid loop is designed to
operate with ice slurry, it would work as well with a single-phase secondary fluid.
10.3 What is to be controlled?
Figure 10.1 shows that, whatever the organisation of the system is, four basic elements have
to be controlled in order to control the whole system: (i) the consumer units, (ii) the storage
tank, (iii) the ice slurry generator, and (iv) the primary refrigerating unit.
10.3.1 CONTROL OF CONSUMER UNITS
For the reasons cited above (necessity to use local feeding pumps), and considering that the
heat exchangers can work with ice slurry as well as single-phase coolants, the control of the
terminal consumer units can be easily set up: a simple ON-OFF control of the feeding pumps
gives reliable results, although a variable speed control of the pumps will result in energy
savings. A simple thermostat for measuring the process temperature (cold room, display
cabinet, etc.) is sufficient. The functional scheme of the control of terminal units is presented
in Figure 10.2.
274
X Y
terminal unit
T
Figure 10.2. Functional scheme of the control of terminal consumer units
The terminal supply pump can be combined with a solenoid valve, but in most cases this is
not mandatory.
Often the defrost strategy to be used is the same as those which are adopted for single-phase
fluids (clock, thermostat, differential pressure controller, electric resistances, whole loop
heating with an electric device or hot gas method, etc…). An example of the defrosting
system is presented in Figure 10.3. The heating device C should be an electric resistance-type
or a hot gas system.
Normal
Ice generator
B Defrost
Storage tank
cycle
C A OPEN CLOSE
B CLOSE OPEN
C OFF ON
A
Figure 10.3. Defrost system based on simultaneous defrosting of consumer units
However, a more suitable and adaptable defrosting strategy can be set up by defrosting
consumer units independently. This can be performed by locally warming the coolant in the
primary loop. On one hand this kind of set up is more expensive in terms of investment (as
every consumer unit has to be equipped with a heating device), but on the other hand better
energy efficiency of this alternative method leads to substantial savings. Figure 10.4
illustrates a possible design of such a defrosting system. In some applications, e.g. medium
product temperature at +5°C, the heating element might be avoided and defrosting is achieved
by simply stopping the ice slurry flow.
A B
Cooling demand
Defrost
terminal unit
Yes No
C
A OPEN CLOSE CLOSE
B ON OFF ON
C OFF OFF ON
Figure 10.4. Example of a local defrost set up
275
To conclude, the control of terminal consumer units in an ice slurry loop is easily realized and
is similar to the control principle applied in classical secondary refrigerant systems. It has to
be based on “ON-OFF” actuators to avoid the problem of plugging, which occurs if
proportional actuators such as three way valves or throttling devices are used.
10.3.2 CONTROL OF THE STORAGE TANK
Because the storage tank “only” accumulates or releases thermal energy, it should be
considered as a passive element, which does not need any specific control, except for safety
assessments. The list of safety devices presented below is not exhaustive and can be extended
according to the specific characteristics of the storage and the distribution devices:
Maximum power of the agitator. Different authors recommend a power between 25 to

70 W/m3 for maintaining an acceptable degree of homogeneity of the stored ice
(Kauffeld et al., 1999; Ben Lakhdar et al., 2001). In some applications no agitator is
required and ice-liquid phase separation is used as a process advantage (for instance,
for harvesting fairly dry ice at the top of the tank or using single-phase liquid in air
conditioning loops).
High-low level measurements in the tank.
Discharge pressure of the two pumps (distribution loop and ice slurry generator) to
detect eventual plugging or cutting off of the circuits.
Ice slurry temperature measurement in the tank and/or at the outlet of the tank. Note
that what this temperature represents is quite difficult to interpret, depending on the
homogeneity of the ice slurry.
To conclude, the control of the storage tank is not a major problem if one considers that it is
just a passive element. But the problem is totally different if one wants to control the stored
thermal energy inside the receiver: this consideration will be discussed below, at the same
time as the influence of the consequences of the ice and temperature distributions inside the
storage tank.
10.3.3 CONTROL OF THE ICE SLURRY GENERATOR
10.3.3.1 General considerations
The role of the ice slurry generator is to create ice crystals in the coolant. The ice mass
fraction at the outlet of the generator depends on different parameters and can be
approximately evaluated by:
1 Φ0 
x 0 = xi +  + c p ⋅ ∆T  (10.1)
L  m& 
where xo and xi are respectively the ice mass fractions at the outlet and inlet of the ice slurry
generator, cp is the specific heat capacity (kJ/kg K) of the carrier liquid (which should be
assumed to be constant as a first approximation), L is the latent heat of ice (333 kJ/kg), Φ0
(kW) is the refrigerating capacity of the primary refrigerating unit, m& (kg/s) is the mass flow
rate of the secondary refrigerant (the ice slurry) in the ice slurry generator and ∆T (K) is its
276
temperature drop. While the outlet temperature is greater than the melting point of the coolant,
ice fraction remains at zero, xo = xi = 0. In this case, the ice slurry generator works as a chiller
without incurring any damage.
Figure 10.5 shows the relationships between different input parameters for an ethanol-water
system. This chart is based on the calculation method for thermophysical properties proposed
by Guilpart et al. (1999) and Ben Lakhdar (1998). The values obtained with this method are
consistent with Bel and Lallemand (1999) in the two-phase domain and with Melinder (1997)
in the single-phase domain.
Figure 10.6 explains how to deal with Figure 10.5. For instance, if the intention is to work
with a 3% ice slurry (“In” at 3% ice in Figure 10.6) made from a 15% initial ethanol solution
(X = 0.15) the temperature has to reach –7.3°C (“Out 1” in Figure 10.6), corresponding to an
enthalpy of –79 kJ/kg. If this mixture is used at the inlet of an ice slurry generator designed to
Φ
provide the ratio 0 equal to –100 kJ/kg, the outlet enthalpy of –179 kJ/kg corresponds to a
m&
temperature of –12.5°C (“Out 2”) and to an ice concentration of 31% (“Out 3”).
Ice concentration (w/w)

0 0.1 0.2 0.3 0.4 0.5 0.6
0
-50
Enthalpy (kJ/kg)
-100
-150
X
X =
= 0.
1
X
0.
2
=
0.
3
0.
3
0.2
0. 1
-200
=
X
X=
X=
-35 -30 -25 -20 -15 -10 -5 0 5
Temperature (°C)
Figure 10.5. Thermal characteristics with main parameters of a water/ethanol mixture
277
Out 1 : -7.3 °C In : 3% ice

0 0.1 0.2 0.3 0.4 0.5 0.6
0
-50
∆h = 100 kJ/kg
-100
in
:X
=
-150 0.
15
-200
5
0. 1
X=
-35 -30 -25 -20 -15 -10 -5 0 5
Out 2 : -12.5 °C Out 3 : 31% ice
Figure 10.6. Use of the diagram presented in Figure 10.5 (cf. text)
10.3.3.2 Different possible control strategies
Three main control strategies of ice slurry generators may be envisaged. Today, all these
possible strategies are based on the measurement of a temperature, assuming that the
dependence between the temperature and the ice concentration is known. This assumption is
quite approximate (Fournaison et al., 2001), but it still remains the simplest and cheapest
solution. These three strategies are:
1. The control of the ice concentration at the outlet of the ice slurry generator. This
strategy is possible while using ON-OFF or HIGH-LOW control of the primary
refrigerating unit based on the outlet temperature measurement (see Figure 10.7a).
This strategy is quite safe for the ice slurry generator, but it does not take into account
the ice fraction contained in the storage tank.
2. The control of the ice concentration in the ice slurry tank, which is supposed to be the
same as at the inlet of the ice slurry generator. This strategy is an interesting solution,
only if the medium in the storage tank is perfectly agitated, which is seldom
encountered in practice (cf. Figure 10.7b).
3. The control of the ice concentration increase in the ice slurry generator while using
Φ
variable flow rates of the secondary fluid (modification of the 0 ratio at constant
m&
Φ0) and/or a high/low control of the primary refrigerant loop. The input values for this
control are the temperature decrease, necessarily associated with a low temperature
limit at the outlet (cf. Figure 10.7c).
278
ON / OFF ON / OFF
or or
HI / LOW HI / LOW
Ice generator
Ice generator
Storage tank
Storage tank
Primary Primary
refrigerant refrigerant
loop loop
(a) (b)
Ice generator
Storage tank
Primary
refrigerant
loop
(c)
Figure 10.7. Some different possible strategies to control an ice slurry generator
The first strategy is mainly used because it is inexpensive, easy to install and ensures safe
running conditions for the ice slurry generator. It is only an accurate method if the solute
concentration at the inlet of the ice slurry generator is constant and well known, and if the
temperature measurement is accurate enough. It is emphasized that these two conditions
sometimes remain difficult to be verified in industrial conditions. The behaviour of the
storage tank is particularly important: in this system, the ice-liquid phase separation leads to a
variation of ice concentration. Therefore the significance of the temperature measurement is
apparent.
As mentioned in the introduction, the problem of ice-liquid phase separation is somehow
related to the overall strategy of operating the ice slurry installation:
1. If ice slurry appears in pipes and heat exchangers the storage tank has to be a perfectly
agitated vessel to ensure a homogenous ice distribution (Figure 10.8a). This method is
used in applications near 0°C such as certain cold rooms, display cabinets and food
processing.
2. The ice slurry tank is only considered as an energy storage tank, and the secondary
loop is charged with a single-phase or low ice concentration coolant (Figure 10.8b and
10.8c). In this case ice-liquid phase separation implies a variation of the residual liquid
concentration, resulting in a variation of the melting point of the coolant. This system
is often used for applications requiring large storage tanks, such as air conditioning of
buildings for instance.
3. The highly concentrated ice at the top of the storage tank is harvested and either used
as “dry” ice (Figure 10.8d), or mixed with the liquid coming back from the process
(Figure 10.8e). The first case is often encountered in direct cooling in the food
industry, e.g., for chilling shrimps and/or fish (Wang et al., 2002). In these
applications, sea water (NaCl brine) is used and the brine is not recycled (the system is
279
charged with fresh sea water). The second case corresponds to marginal applications
for which the temperature of the process has to be variable.
Ice generator
Ice generator
Rich layer
Ice generator
Poor layer
a) perfectly agitated tank and use of b) Partially agitated tank and use of c) Storage tank without agitation
homogeneous ice slurry concentrated slurry and use of ice slurry at a low
concentration
Mixing device
Ice generator
Ice generator
Dry ice
drain
Fresh brine
d) Use of dry ice e) Mixing device
Figure 10.8. Different operational modes of ice slurry systems
In most cases the industrial applications lie between the “perfectly agitated” storage tank and
the “perfectly separated” rich and poor layers of ice slurry in the storage tank.
10.3.4 CONTROL OF THE PRIMARY REFRIGERATION UNIT
The primary refrigeration unit can be controlled in an on/off manner or by varying the
evaporation temperature gradually in order to adjust the ice making capacity according to
Equation 10.1. Any capacity control, used in conventional refrigeration units, might be
applied, e.g. speed controls of the compressor, thermostatic expansion valve, and automatic
solenoid expansion valve.
10.4 Influence of the temperature and solute concentration on the control parameters
If the ice slurry in the tank is perfectly agitated, the temperature of the slurry may be
considered to be constant. The temperature only depends on the nature of the freezing
depressant, on its initial concentration and on the ice mass fraction. According to the
solid/liquid equilibrium curve, this temperature perfectly defines ice concentration of the ice
slurry. Usually ice concentrations between 0.1 and 0.3 wt-% are used. Figure 10.9 shows for
an ethanol/water mixture the relationships existing among the temperature, the initial solute
concentration and the ice concentration.
280
1.2
1.0
0.8
Ice mass fraction
0.6
0.4
0.2
0.0
-50 -40 -30 -20 -10 0 10
Temperature (°C)
Initial ethanol mass fraction 0 0.05 0.1 0.15 0.2 0.25 0.3 0.4 0.5
Figure 10.9. Ice concentration versus temperature and solute concentration

for ethanol/water mixtures
According to Fournaison et al. (2001) and Figure 10.9, especially at low solute
concentrations, a low accuracy in the temperature measurements leads to an even lower
accuracy in the ice concentration. The initial solute concentration also plays a major role.
Consequently, the control of ice slurry systems has to be based on accurate knowledge of both
the temperature and the solute concentration. Table 10.1 describes the common accuracies
that should be expected in laboratory conditions and in industrial conditions:
Table 10.1. Expected accuracy in measurements under different conditions
Laboratory conditions Industrial conditions
Temperature measurement ± 0.1 K ± 0.5 K
Solute concentration measurement 1% 5%
These accuracies required show that the determination of the ice concentration by temperature
measurement is sometimes impossible, see also Figure 10.10.
0.3
Absolute error on ice concentration
0.2
0.1
Maximum absolute error
X i = 0.30 0.15
X i = 0.20 Xi=
X i = 0.10
X i = 0.05
0
-30 -20 -10 0
Température (°C)
Figure 10.10. Absolute error in ice concentration measurements under industrial conditions
when working with water/ethanol ice slurries
281
Figure 10.10 clearly shows that, e.g. for a solution of 5 wt-% ethanol at –2.5°C, it is almost
impossible to predict the ice concentration by temperature measurement. It also shows that:
• the maximum absolute error obtained in the ice concentration determination occurs close
to the freezing point,
• it is not reasonable to expect precise ice concentration determination, if the temperature is
two to three degrees below 0°C,
• the lower the temperature is, the better is the accuracy,
• a high solute concentration implies a low error in the ice concentration determination.
These considerations are also valid for other solutes as shown in Figure 10.11.
0.1
Maximum absolute error on ice concentration
0.09
0.08
0.07 Industrial conditions
0.06
0.05
ammonia
0.04 ethanol
propylglyc
0.03
0.02 Laboratory conditions
0.01
0
-35 -30 -25 -20 -15 -10 -5 0
Temperature (°C)
Figure 10.11. Maximum error in the ice concentration determination versus the temperature
for different operating conditions
Figure 10.11 indicates that, even under laboratory conditions and at low temperatures, the
maximum absolute error made in the ice concentration determination cannot be lower than
1% (6% in industrial conditions). The reduction in this maximum error is possible, if more
concentrated brines are used and in proportion the freezing point and consequently the ice
concentration at a given temperature are reduced.
Figure 10.11 also indicates that a higher temperature gives larger uncertainty in the ice
concentration measurement. It appears that the control of ice slurry systems tends to be
impossible at temperatures just below 0°C down to –3 or –4°C. This becomes evident when
the uncertainty in the ice concentration measurement is equal to or greater than the basic
value taken for controlling the system.
Figure 10.12 shows the operating range of an ethanol/water ice slurry system. The ice
concentration set-point is located between Xi = 0.1 and Xi = 0.3. This figure shows the degree
of freedom in ethanol concentration and temperature one has in reaching this target.
The dotted lines surrounding the curves for Xi = 0.1 and Xi = 0.3 represent the ice mass
fraction uncertainty due to the temperature and concentration measurements for the solution
under industrial conditions. It appears that the control of ice concentrations at temperatures
282
higher than –4.1°C and/or ethanol concentrations lower than 8% is very difficult, and
becomes even impossible at about 0°C.
0.2
0.18 Fre
ezin
gp
oin
0.16 Xi = t
0.1
ethanol concentration (W/W)
0.14
Xi =
0.12 0.3
0.1
Xi = 0.5
0.08
0.06
0.04
0.02
0
-10 -9 -8 -7 -6 -5 -4 -3 -2 -1
Temperature (°C)
Figure 10.12. Operating range of an ethanol/water ice slurry system
On the other hand, the lower the temperature is and the higher the solute concentration is, the
easier it becomes to control the ice concentration.
10.5 Consequences of the uncertainty in ice concentration on the control of the system
Whatever the operational mode of the ice slurry installation is, it usually involves an increase
in the solute concentration at the inlet of the ice slurry generator. This is the case for an
imperfectly agitated storage tank where ice, i.e. frozen water, accumulates. It is also the case
of a perfect ice separator (ice-rich and poor layers as illustrated in Figure 10.8). In all cases,
the residual mother liquor is enriched in solute. According to the discussion above, this solute
concentration shift helps to facilitate control of the ice slurry generator. At the same time, the
temperature set point is reached at a lower ice concentration and that leads to a safer running
condition.
In the worst case (cf. Figure 10.13), if the solute concentration reaches a too high a value, the
ice slurry generator would operate as a classical chiller, without any damage as mentioned
above. In this case, the performance degradation simply implies an increase in the running
time of the primary refrigeration loop.
Consequently, it can be said that, even if the ice concentration control in the ice generation
process is only approximately performed, the consequence on the global control of the system
is not significant. For instance, if using an initially 10% ethanol mixture, a control scheme
based on a classic thermostat (differential of ± 1 K) with a set point fixed at –6°C ensures a
safe operating range as shown in Figure 10.13.
283
0.2
0.18 Fre
ezin
C Possible running range
g poi (monophasic coolant)
0.16 nt
Xi =
0.1
ethanol concentration (W/W)
0.14
Ice slurry
Xi = running range
0.12 0.3
0.1
Xi = 0.5 B D
0.08
0.06
A
0.04
0.02
0
-10 -9 -8 -7 -5 -4 -3 -2 -1
-6
Set point +
- 1K
Figure 10.13. Operating range for an ice slurry loop working with an initial 10%
ethanol/water mixture with a thermostat set point at –6°C ± 1 K
The operating points shown in Figure 10.13 are as follows.

Point A: designed operating set point: ice concentration Xi = 25% at –6°C
Point B: maximum ice concentration (Xi = .35) obtained at the cut off of the thermostat.
Point C: if some ice is accumulated in the ice slurry tank or anywhere else in the secondary
loop, the concentration of the residual liquid increases. The ice slurry generator continues to
run until the temperature of –7°C is reached (cut off of the thermostat). No ice is produced if
the ethanol concentration reaches 15% or more (classical chilling mode).
Point D: if the thermal load is too large, there is no more ice in the secondary loop. The initial
ethanol concentration reaches the initial value of 10%. The ice slurry generator produces
some ice while the temperature is less than –4.2°C. If the thermal load is too high, the primary
refrigeration loop continues to run, even without ice production (classical chilling mode).
To conclude with this example, the ice slurry loop cannot work below 10% of ethanol or
below –7°C. Consequently, the maximum possible ice concentration is 35%, which is a safe
running condition. From Figure 10.13, it is easy to see that the running range of an ice slurry
loop is very small at temperatures above –3°C and that a very high ice concentration should
rapidly occur at this temperature.
The one situation to avoid is dilution of the initial mixture. This could happen, for instance,
while refilling the installation with pure water. This situation would indeed have a major
consequence leading to the system running with too high an ice concentration and the risk of
plugging of the circuits. It should be emphasized that low solute concentrations could also
lead to mechanical problems in certain ice slurry generators, which are not designed to be
operated as flake ice slurry generators.
284
10.6 Conclusions
The control of ice slurry secondary loops does not fundamentally differ from the control of
classical indirect cooling systems. It only has to be based on the use of small local pumps for
feeding the heat exchangers, and ON-OFF actuators to avoid the problem of plugging when
using proportional actuators as three way valves or throttling devices.
In practical situations, the control of the ice slurry generator should be based on the
measurement of the inlet, or preferably, of the outlet temperature, even if the consequent
accuracy on the ice concentration is only approximate. A judicious set up of the thermostat is
able to ensure a safe running condition of the ice slurry generator. One must nevertheless
emphasise that this control principle is not accurate enough to measure the refrigerating
capacity (and/or the energy) delivered by the system.
Special care has to be taken for systems running with low secondary fluid concentrations.
Under those conditions, plugging of the ice slurry distribution loop and possibly mechanical
damage to the ice slurry generator could occur. Therefore, it is not recommended to operate in
the ice slurry domain at temperatures higher than –3 to –4°C.
1. Bel O.; Lallemand A.: Study of a two phase secondary refrigerant. 1- intrinsic
thermophysical properties of an ice slurry. Int. Journal of Refrigeration, Vol. 22, No. 3,
pp. 164-187, 1999.
2. Ben Lakhdar M.A.: Comportement thermohydraulique d’un fluide frigoporteur
diphasique : le coulis de glace. Etude théorique et expérimentale. PhD thesis. INSA de
Lyon, 1998.
3. Ben Lakhdar M.A.; Blain G.; Compingt A. et al.: Direct expansion and indirect
refrigeration with ice slurry. Third IIR Workshop on Ice Slurries, Lucerne, May 2001.
4. Fournaison L.; Chourot J.M.; Faucheux M.; Guilpart J.: Ice fraction error calculation in
ice slurries. Third IIR Workshop on Ice Slurries, Lucerne, May 2001.
5. Guilpart J.; Fournaison L.; Ben Lakdhar M.A.: Calculation method of thermophysical
properties for ice slurries, application to Water / ethanol mixture. XXth IIR International
Congress of Refrigeration, Sydney, 1999.
6. Kauffeld M.; Christensen K.; Lund S.; Hansen T.: Experience with ice slurry. First IIR
Workshop on Ice Slurries, Yverdon les Bains, May 1999.
7. Melinder A.: Thermophysical properties of liquid secondary refrigerants, Tables and
diagrams for the refrigeration industry, IIF-IIR, Paris, 1997.
8. Wang M.J.; Lopez, G.; Goldstein, V.: Ice slurry for shrimp and processing. Fifth IIR
Workshop on Ice Slurries, Stockholm, May 2002.
9. Wang M.J.; Inoue Y.; Goldstein V.: Ice thermal storage in modern buildings, XXth IIR
International Congress of Refrigeration, Sydney, 1999.
285
CHAPTER 11. OPTIMIZATION OF ICE SLURRY SYSTEMS

by G. Grazzini, S. Pazzi
Ice slurry is used as a heat carrier and as a fluid useful for energy storage. The optimization of
ice slurry systems refers to systems with components producing temperature differences, heat
transfer and energy storage. There has not been much written about optimization of ice slurry
systems, hence this chapter presents a brief general guide to optimization in general and
optimization of energy systems in particular.
Optimisation is a design or managing method used to obtain the best value of the defined
system's goal (or goals). This can be expressed as an objective function which can be
minimized or maximized. A set of unknowns or variables, subject to constraints allowing
some values but excluding others, controls the values that the objective function can assume.
The systems considered always have physical constraints and often the user needs to optimize
a number of different objectives simultaneously.
11.1 Criteria
Many different criteria can be used to build the objective function. Some common criteria are
as follows.
To evaluate the economic suitability of a project it is necessary to build some evaluation

indices. These indices are used to evaluate the profitability of an investment and compare
alternative investments. The most commonly used are:
The Payback Period, which is defined as the length of time required for the cash
inflows received from a project to recover the original cash outlay of the initial
investment.
A Net Present Value (NP) criterion estimates the actual value of all the cash flows
during the lifetime of the project.
Internal Rate of Return (IRR) calculates an interest rate that is characteristic of the
project being considered. Mathematically the Internal Rate of Return (IRR) is the
interest rate that makes the Net Present Value of the investment equal to zero.
For more detailed information on the estimation of financial parameters related to plant
design, readers should consult Tsatsaronis (2000a), Bejan et al. (1996) and their references.
Economics often does not consider physical constraints. Energy is the physical parameter that
obviously appears when dealing with an energy system. The objective function then should be
the energy needs that have to be minimized. In cooling systems the parameter usually
considered is the COP (Coefficient Of Performance), defined as the ratio between the cooling
effect and the energy employed to obtain it. For irreversible systems a maximum exists for
this parameter (Grazzini, 1993) so the COP function can be assumed as the objective to be
maximized. When applied to complex systems the parameter should include all useful energy,
including pressure losses in pipes.
287
The COP is defined from the first law of thermodynamics. In order to evaluate the
temperature level of heat exchange, the second law also has to be considered. For this purpose
the concept of exergy was introduced. When two thermodynamic systems are placed in
communication, theoretically work can be developed as the two come into equilibrium. When
one of the two systems is conveniently idealized and used as a reference environment, the
other being some system of interest, exergy is defined as the maximum theoretical useful
work obtainable as the systems interact to equilibrium. Alternatively, exergy is the minimum
theoretical useful work required to bring a system from equilibrium with the environment to a
specified state.
For more detailed information on the exergy and exergy analysis, readers should consult the
references by Moran (2000), Moran and Shapiro (2000), Bejan et al. (1996), Kotas (1995),
Lozano and Valero (1993), Szargut et al. (1988).
A thermodynamic optimization produced a design that was very different from an

economically optimal one. The objective of thermoeconomic or exergoeconomic optimization
is to minimize costs, both economic and related to destruction and loss of exergy. The goal is
pursued with the association of a cost rate to each exergy transfer.
For more detailed information on thermoeconomic or exergoeconomic analysis, readers
should consult the references to work by Tsatsaronis (2000), Bejan et al.(1996), Lozano et. al.
(1994), Evans (1980).
Purely thermodynamic methods were proposed as the equipartition of entropy production

(Tondeur and Kvaalen, 1986) and entropy minimization (Bejan, 1996), but they appear to be
equivalent (Bejan and Tondeur, 1998). The same applies to finite-time, finite-seize, finite-
resources and endoreversible (exoirreversible) thermodynamics (Bejan, 1996).
Life Cycle Analysis (LCA) was developed in the 1970s to assess the environmental impact of
a process (for a review see Frankl and Gamberale, 1998). The method started from a balance
between the energy output and the energy used over the entire useful life of the system,
including the total amount of energy and materials employed in a fabrication process, in the
maintenance and in the decommissioning. LCA has been extended to include labour,
environmental damage and recycling, so actually the acronym is used very often for Life
Cycle Assessment considering all environmental impacts such as climate change, ozone
depletion, etc. (EPA, 1994; ISO, 1998). Being LCA based on First-Law Analysis it fails to
account for different types of energy carriers. Szargut (1990) introduced exergy in the
procedure and Sciubba (2001) proposed Extended Exergy Accounting incorporating the
above described approaches and thermoeconomic analysis.
11.2 Methods
Optimization problems are made up of three basic ingredients:
• An objective function which has to be minimized or maximized.

• A set of unknowns or variables which affect the value of the objective function.
• An eventual set of constraints that allow the unknowns to take on certain values but
exclude others.
288
• An eventual set of constraints that allow the unknowns to take on certain values but
exclude others.
Optimization algorithms can be divided into two main categories: unconstrained and
constrained. The first category looks for the optimum of a specified objective function
reference to the range of variation of the input parameters. The latter category optimizes the
target function assuming a well defined set of constraints for the input parameters which
represent the independent variables for the objective function. When dealing with non-linearly
constrained optimization, other methods should be used. The sequential quadratic
programming (sequential QP) algorithm, for example, is a generalization of Newton's method
for unconstrained optimization in that it finds a step away from the current point by
minimizing a quadratic model of the problem (Fletcher, 2000). More detailed information on
the above mentioned optimization methods and other algorithms can be found, apart from the
quoted references, at the Web site: http://www-fp.mcs.anl.gov/otc/Guide/OptWeb/.
Genetic algorithms (Goldberg, 1989) optimize a function exploiting the principles of natural
evolution.
Evolutionary strategies (Baech, 1996; Baech et al., 1997) are based on the same principles of
genetic algorithms but differ from these in that the best individuals are not crossed.
Engineering problems are often characterized by a high degree of complexity which may
imply time-consuming computations for target function evaluation. That is why optimization
algorithms are commonly coupled to approximators such as Neural Networks (Dayoff, 1990)
or Lazy Learning (Aha, 1997, Atkeson, 1997) algorithms. The open literature presents several
studies dealing with the coupling of optimization algorithms and local or global
approximators to solve engineering problems (Cosentino et al., 2001).
1. Aha, D.W., Editorial, Artificial Intelligence Review, Special Issue on Lazy Learning ,
1997
2. Atkeson, C.G., Moore, A.W., Schaal, S., Locally weighted learning, Artificial Intell.
Review, 11: p. 11-73, 1997.
3. Baech, Th.: Evolutionary Algorithms in Theory and Practice, Oxford University Press,
New York, 1996.
4. Baech, Th.; Fogel, D.B.; Michalewicz Z. (eds): Handbook of Evolutionary Computation,
Oxford Univ. Press, New York, and Inst. of Physics Pub., Bristol, 1997.
5. Bejan, A., Entropy generation minimization., CRC Press Boca raton, FL, USA, 1996.
6. Bejan, A., Entropy generation minimization: the new thermodynamics of finite-seize
devices and finite-time processes. J. of Applied Physics 79: p.1191-1218, 1996.
7. Bejan, A.; Tsatsaronis, G.; Moran, M.J.: Thermal design and optimization, John Wiley &
Sons, Inc., New York, 1996.
8. Cosentino, R., Alsalihi, Z., Van Den Braembussche, R., Expert System for Radial Impeller
Optimization, ATI-CST-039/01, 4th Europ. Conf. on Turbom., March 2001, Florence,
Italy.
9. Dayoff, J., Neural Network Architectures. An Introduction, Van Nostrand Reinhold, 1990.
10. Evans, R.L., Thermoeconomic Isolation and Essergy Analysis, Energy 5, n.8-9, 1980.
289
11. Fletcher, R., Practical Methods of Optimization, John Wiley & Sons, New York, U.S. ,
2000.
12. Frankl, P., Gamberale, M., The methodology of LCA and its application to the energy
sector, Proc. Advances in En. Stud., P. Venere, Italy, 1998.
13. Goldberg, D.E.: Genetic Algorithms in Search, Optimization and Machine Learning,
Addison-Wesley, 1989.
14. Grazzini, G.: Irreversible refrigerators with isothermal heat exchanges, Int. J. of Refrig.,
16, No.2, pp.101-106, 1993.
15. Kotas T.: The Exergy Method of Thermal Plant Analysis, Butterworths, 1995.
16. Lozano M.A.; Valero A.: Theory of exergetic costs, Energy, Vol. 18, no.9, pp.939-960,
1993.
17. Moran, M.J.: Exergy analysis. In: F. Kreith ed. The CRC Handbook of Thermal
Engineering, 1-69-77, CRC Press, Boca Raton, FL, USA, 2000.
18. Moran, M.J.; Shapiro, H.N.: Fundamentals of Engineering Thermodynamics, 4th ed., John
Wiley & Sons, New York, 2000.
19. Sciubba, E., Beyond Thermo-Economics? The concept of Extended Exergy Accounting
and its Application to the Analysis and design of Thermal Systems, Int. J. Ex. 1, n.1,
2001.
20. Szargut, J., Analysis of cumulative exergy consumption and cumulative exergy losses. In:
Sieniytycz, S., Salamon, P., (Eds), Finite Time Thermodynamics and Thermoeconomics,
Taylor and Francis, New York, 278-302, 1990.
21. Szargut, J.; Morris, D.R.; Steward, F.R.: Exergy Analysis of Thermal, Chemical and
Metallurgical Processes, Hemisphere, New York, 1988.
22. Tondeur, D., Kvaalen, E., Equipartition of entropy production. An optimality criterion for
transfer and separation processes. Ind. Eng. Chem. Res. 26,50-56, 1986.
23. Tsatsaronis G.: Exergoeconomics. In: F. Kreith ed. The CRC Handbook of Thermal
Engineering, 1-88-94, CRC Press, Boca Raton, FL, USA, 2000.
290
CHAPTER 12. PRESENT AND FUTURE APPLICATIONS

by Savvas Tassou1, John Bellas, Issa Chaer, Thomas Davies, Thorsten Behnert, Ming-Jian
Wang, Vladimir Goldstein
12.1 Introduction
Recent improvements in ice slurry generator technology and advanced design concepts have
made ice slurries a competitive alternative to conventional secondary refrigeration systems
and sometimes even to conventional direct expansion refrigeration systems. Many ice slurry
plants have, for a number of years now, been successfully employed in many applications,
from comfort cooling and commercial refrigeration to industrial production processes and
medicine. These plants vary in size from small installations in specialist applications of a few
kW cooling capacity to very large systems for high rise buildings, ships and factories of MW
cooling demand.
This chapter provides examples of existing ice slurry cooling applications, and highlights
areas of research and development that could lead to future applications. The material
presented is a compilation of information published in the open literature by designers,
consultants, manufacturers and users of the technology. The examples cover applications
worldwide and more attention is placed on applications where detailed information and
analysis have been presented in the literature. In certain applications, reference is made to the
manufacturer of the ice slurry generator but this does not imply any endorsement or
judgement on the performance of the particular ice generation technology or product.
The applications presented in this chapter are grouped into three main sections: Comfort
cooling covers applications in office and institutional buildings as well as other commercial
and industrial facilities such as the Kyoto train station complex, and district and mine cooling.
The second section covers applications in the food industry including food processing,
storage, transport and retail. The third section covers special and future applications that may
result from ongoing research and development work.
12.2 Comfort Cooling
12.2.1 COMFORT COOLING OF BUILDINGS
Ice slurry cooling systems have been installed in many buildings world wide for air-
conditioning purposes. In most cases, the ice slurry production system has been combined
with a storage system to store energy during the night and use it during the day to meet the
peak load. The result is that smaller refrigeration systems (chillers) can be used, usually with a
capacity of 20-50% of the peak cooling load. Additionally, operational savings can be made
where off-peak electricity tariffs are available.
An ice slurry-based air conditioning system usually employs three independent circuits. A
refrigerant circuit, an ice slurry circuit between the ice slurry generator, the storage tank and a
heat exchanger and a chilled-water circuit between the ice slurry system and the load. An
1
The lead author of Chapter 12, Savvas Tassou, would like to acknowledge M. Tanino and Y. Kozawa for
contributing material on the Kyoto Station building, T. Kuriyama for material on the CAPCOM building, Z. Ure
for material on ice slurry applications, and P. Mueller Co. for the permission to use material and pictures from
their ice slurry projects.
291
alternative arrangement is to pump the slurry from the tank directly to the load. Depending on
the plant design single-phase fluid or ice slurry is distributed in the building or factory. A
third design concept is to use the ice slurry as the heat sink in the circulation of a common
refrigerant by gravity between the condenser and evaporator within a conventional
refrigeration system (Wang and Kusumoto, 2001).
In many cases, thanks to the low temperature of ice slurries a cold air distribution system has
been used, resulting in reductions in the size of the ducting and piping, and capital cost. The
above system configurations are discussed in greater detail in the application examples given
in the following sections.
12.2.2 APPLICATIONS TO OFFICE BUILDINGS
a) CAPCOM building
The CAPCOM building is situated in Osaka, Japan and houses the research and development
section of the software manufacturer CAPCOM. It was completed in March 1996 and
consists of 20 storeys with a total floor area of 16 784 m2. Figure 12.1 shows a schematic of
the building. The company’s offices are located from the 2nd to the 16th floor, occupying an
area of 570 m2 on each floor. The heating load of the offices was calculated to be 151 W/m2,
relatively higher than average, mainly due to the heat generated by the computers in the
offices.
Area covered by
17th Floor: Rest room, one VAV unit
16th Floor Outdoor units
Perimeter
Air-conditioner
3rd to 16th Floor

(Office Floors)
Office
Air-Conditioners
1th Floor
3rd Floor Perimeter
Air-conditioner
1st Floor
Plan of typical office floor

of the building
Figure 12.1. CAPCOM building
292
A gas-driven absorption refrigeration system is employed to provide cooling to the perimeter

areas. Cooling to the internal areas is provided by an ice slurry cooling system consisting of
two ice slurry units of 272 kW cooling capacity each. The ice generated by the ice slurry
machines, which are located on the 17th floor, is stored in two storage tanks. The ice slurry
from the tanks is pumped through an ice slurry separator which controls the ice fraction of the
slurry to 20%, to air handling units on each floor as shown in Figure 12.2. Each of the 15
office floors is served by two air handling units of 9000 m3/h flow capacity each, which in turn
serve six variable air volume (VAV) terminal devices.
Ice Slurry Separators

Ice maker Ice maker
Storage Storage
Tank Tank
Water/Alcohol Water/Alcohol
Pump Ice slurry Ice slurry
Floor 17 supply return
Floor 16
Floor 15
AHU AHU
Floor 2
Water and Hot Water and Hot

Water Coil Ice Slurry Coils Water Coil
Figure 12.2. A schematic diagram of the ice slurry cooling system
The low temperature of the ice slurry also offered the option of using a cooler air distribution
temperature of 12°C instead of 15°C, which is the normal design practice in Japan. This
reduced the air flow requirement from 41 m3/(m2h) down to 32 m3/(m2h), resulting in capital
and operating cost savings due to the smaller size of equipment and the lower power
293
requirement for air distribution. Moreover, the cold air supply temperature offered lower
humidity for the space and hence a more comfortable working environment. More details on
the building and ice slurry system can be found in Kuriyama and Sawahata (2001).
Power consumption for the ice slurry generators, pumps and air circulation fans was measured from
July 15 to September 27, 1996. The results were compared with predicted energy consumption for
an equivalent conventional ice thermal storage system and high temperature air distribution for the
building. The ice slurry plant was found to have a 22 MWh higher energy consumption than the
conventional plant, but hydraulic and air distribution costs were found to be lower by 47 MWh for
the ice slurry plant, resulting in 25 MWh savings which represent 4% savings in the overall energy
consumption of the building.
b) Herbis Osaka Building

The Herbis Osaka Building, in Osaka, Japan, was completed in February 1997. The building
has five underground floors, forty floors above ground and one tower floor. The total height
of the building complex is 181 m and the total floor area is 136 823 m2. It houses offices, a
hotel, exhibition halls, shops, parking places, etc. The building is described in detail by Wang
and Kusumoto (2001).
The heating and cooling requirements of the building are satisfied by a “Crystal Liquid Ice
Thermal Storage System with Heat Recovery” (CLIS-HR). The whole installation includes 31
units of heat recovery ice slurry generators of 260 kW capacity each and 16 sets of ice storage
tanks of either 70 m3 or 140 m3 capacity, having a total thermal storage capacity of 80 750
kWh.
A schematic diagram of the system is shown in Figure 12.3. The ice slurry system is utilised
to drive a gravity based refrigeration system using R134a as a refrigerant. The refrigeration
system employs two closed circuits, a warm and a cold circuit, each employing a condenser
coil, an evaporator coil and an expansion valve. The evaporator of the cold circuit and the
condenser of the warm circuit are of the plate fin type and are placed in the air handling unit
which provides conditioned air to the building.
Ice slurry generated by the super-chilling process is stored in a tank and from there it is
pumped to the condenser of the cold refrigerant circuit. Refrigerant vapour from the
evaporator coil condenses in the condenser of the cold circuit and the liquid refrigerant is then
expanded as it flows through the expansion valve back to the evaporator where it evaporates
and cools the air flowing through the air handling unit.
In the warm circuit, heat rejected by the super chiller ice slurry generator is pumped to the
evaporator coil of the warm refrigerant circuit. Refrigerant liquid from the condenser coil in
the air handling unit expands as it flows through the expansion valve and then evaporates at a
relatively high temperature in the evaporator by drawing heat from the hot water circulated
from the ice slurry generator. The refrigerant vapour from the evaporator then passes to the
condenser where it rejects heat to air flowing through the air handling unit for space heating.
294
Condenser for Air Cooling Mode

Ice Storage Tank
Air Cooling
Mode
Freon
(Vapour)
Super-
chiller
Pump
Ice
Freon Slurry
Cold (Vapour)
Air
Liquid Air Heating
Hot Hot
Mode Water
Air
Air
Conditioner
Condenser for Air Heating Mode
Figure 12.3. A schematic of the Herbis building thermal storage system
The natural circulation of refrigerant requires considerably less power for operation compared
to conventional vapour compression systems. From energy consumption records for the
building during the first three years of its operation, it was concluded that the ice thermal
storage system reduced the peak load of the building by about one third (Wang and
Kusumoto, 2001). It was also identified that 60% of the total annual cooling and heating
requirement could be met from night-time operation and thermal storage using off peak
electricity tariffs, which resulted in significant operating cost savings.
c) Techno-Mart 21 Complex, Seoul, Korea

The “Techno-Mart 21” complex (Figure 12.4) is one of the largest commercial buildings in
Korea, housing a 10-storey shopping centre and a 39 storey-office tower. The total floor area
is 260 100 m2. The peak cooling load of the building is 36 270 kW and is met by a 14 068 kW
ice slurry cooling system. The plant consists of eight scraped type ice slurry units (Figure
12.5) with a capacity of 1759 kW each and a 3400 m3 storage tank (Figure 12.6).
295
Figure 12.4. The “Techno-Mart 21” Figure 12.5. One of the eight ice-slurry units
building. (Information courtesy of P. (1,759 kW) installed in the building.
Mueller Co.) (Information courtesy of P. Mueller Co)
Each ice slurry unit consists of four ice slurry generators. The compressor/condenser package
consists of four screw compressors with water-cooled condensers. The units are located on
top of the storage tank and drop the ice slurry directly into the tank as shown in Figure 12.7.
According to the manufacturer, due to the large difference between peak and off peak
electricity tariffs, operation of the ice slurry thermal energy storage system during off peak,
leads to substantial energy savings which resulted in a payback period of 1.1 years compared
to a conventional cooling system (EPS Ltd., 2003a).
Figure 12.6. The inside of the tank showing Figure 12.7. Ice slurry generator
spray nozzles at the top of the tank discharge into the tank (Information
(Information Courtesy of Paul Mueller Co.) Courtesy of Paul Mueller Co.)
296
12.2.3 APPLICATIONS TO INSTITUTIONAL BUILDINGS
a) Stuart C. Siegel Centre, Virginia, USA

The Stuart C. Siegel Centre is a 17 600 m2 athletic complex at Virginia Commonwealth
University in Richmond, Virginia, housing a 7500 seat basketball arena and several other
facilities for sport activities and teaching purposes. The total peak design-cooling load of the
centre was calculated to be 4536 kW. The peak load occurs only when the arena reaches full
occupancy while for the rest of the operating hours the load is significantly lower. For this
reason, an ice slurry thermal energy storage system was employed consisting of 6 ice
machines of 225 kW capacity each (Figure 12.8).
Figure 12.8. Ice slurry generators at the Stuart Siegel Centre, Virginia.
(Information Courtesy of Paul Mueller Co.)
The ice slurry generated is stored in a tank and then pumped through a heat exchanger where
it chills the water circulated to the cooling coils of the air handling units down to 2.2°C
compared to 6.6°C for a conventional system. This low water temperature results in reduced
HVAC component sizes and capital and operating cost savings. Savings in peak demand
charges due to the utilisation of energy storage as well as savings arising from reduced pump
and fan power consumption amounted to USD 75 000 per year. A fuller description of the
system is given by Nelson et al. (1999).
b) Middlesex University, UK
The ice slurry air-conditioning system at the Bounds Green campus of Middlesex University
was (in 2003) the largest application of an ice slurry system in the UK. The system,
commissioned in 1999 has a capacity of 148 kW and replaced an old 372 kW direct expansion
refrigeration system. The choice of the ice slurry system was based on its overall
environmental performance including the use of a natural refrigerant in the primary plant.
Table 12.1 illustrates the projected capital and running costs of competing cooling systems at
the project planning stage (Everitt, 1999).
The capital cost of the ice slurry system was estimated to be only slightly higher than that of
an absorption cooling system but almost double that of a conventional vapour compression
plant. However, due to the utilisation of relatively cheap night-time electricity tariffs, the
annual fuel costs were projected to be two thirds those of the absorption system and only one
third those of the vapour compression system, giving a pay back period over the vapour
compression system of approximately eight years.
297
Table 12.1. Approximate comparative costs of competing cooling systems

Fuel cost
Capital Cost Fuel cost Capital + Fuel
Type over 20yrs
£ (Sterling) £ p.a. costs over 20 yrs
£
Ice slurry system with thermal
185 000 5518 110 000 295 000
storage
Absorption cooling system 179 000 8294 166 000 345 000
Conventional vapour
90 000 17 500 350 000 530 000*
compression plant
High specification vapour
compression plant 120 000 17 500 350 000 470 000
* includes more frequent replacement
The installed ice slurry plant (Figure 12.9) features two 75 kW ice slurry generators served
from an ammonia chiller. The fluid used in the ice slurry section is a water-urea solution. The
physical arrangement of the ice slurry system is illustrated in Figure 12.10. The ice slurry
generators are positioned at the top of two storage tanks so that the ice generated by each
machine can drop straight into the respective storage tank. The total storage capacity of the
two tanks is 72 m3. When fully charged, the tanks can provide cooling for up to 24 hours,
even if the cooling plant should fail.
The ice slurry is generated mainly overnight and is stored in the tanks. When cooling is
required, chilled solution from the bottom of the tank is pumped through a heat exchanger to
maintain the chilled water flow at the required temperature. Flow in the main chilled water
circuit is provided by a fixed speed pump supplying 11.2 l/s of water against a pressure of 350
kPa (Sims, 1999). A flow header with six circuits served by variable speed pumps connected
to the building management system (BMS) distributes chilled water to various parts of the
university building (lecture theatres, study rooms, laboratories and a computer centre). The
speed of the pumps is controlled by the BMS in response to variations in load. Operation
down to 50% of design flow is possible which should lead to significant energy savings
(Rendall, 1999).
The whole system was installed in two phases, I and II, and has been designed to allow for
future expansion. Presently the system operates in conjunction with an old direct expansion
chiller of 228 kW cooling capacity. When this chiller comes to the end of its life, chilled
water heat exchangers connected to the ice slurry plant will replace it.
Figure 12.9. Ice slurry machine on top of the storage tanks at the left of the picture
and air cooled condensers at the right. (Information Courtesy of Paul Mueller Co.)
298
System Air Cooled Chiller Inlet 10 °C

Return
Supply 6 °C Heat Air Cooled
Exchanger Ice Slurry Condensing Unit
Machine
System Supply
Ice Slurry Tank
Phase I
Inlet 6 °C
Supply 2 °C Heat Air Cooled

Exchanger Ice Slurry Condensing Unit
Machine
Ice Slurry Tank
Phase II
Figure 12.10. A schematic diagram of the ice slurry cooling system at Middlesex University
12.2.4 APPLICATIONS TO COMMERCIAL BUILDINGS
Kyoto Station Building

The Kyoto Station Building shown in Figure 12.11 is a complex 19-storey building having a
total floor area of 238,000 m2. The building houses a train station, a hotel, a theatre and
several commercial facilities. The layout of the building is shown in Figure 12.12.
The cooling needs of the building are satisfied by two separate systems (Ise et al., 2001):
• An absorption system driven by steam generated by a co-generation system, and
• An ice slurry plant powered by electricity from the co-generation system and
supplemented from the city’s electrical network.
299
Figure 12.11. An aerial view of the Kyoto Station Building
Figure 12.12. Layout of the Kyoto Station Building
The ice slurry plant incorporates two sets of brine chillers of 1400 kW cooling capacity each,
four sets of shell and tube super-coolers of 700 kW capacity each and two concrete tanks for
the storage of ice slurry. The tanks are located below the building’s basement and have a
storage capacity of 400 m3 each. Figure 12.13 shows a schematic diagram of the plant.
Ice slurry is produced by the method of super-cooling. Water is super-cooled to a temperature

of –2°C in the super-cooling heat exchanger. As it exits the super-cooler, a small portion,
around 2.5%, changes into ice, while the remainder stays in the liquid phase. The ice slurry
produced drops into the tank and due to the density difference, the ice is accumulated at the
upper part of the tank while the water occupies the lower part, remaining at a temperature of
approximately 0°C. The ice slurry production/storage takes place during the night (23:00 h to
08:00 h), and the ice slurry in the tank reaches progressively an ice concentration of 23% with
a cooling capacity of about 40 GJ.
300
During daytime operation, water from the bottom of the tank is pumped to heat exchangers
where it chills the water circulated in the cooling circuits throughout the building. The fluid
temperature in these circuits is 7°C and the return temperature is 10°C.
The electrical power consumption for night-time ice production was estimated to be 3960
MWh/year, which represents only 4.6% of the building’s total energy consumption. A
comparison between the absorption and ice-slurry system showed the ice-slurry system to
have 50% lower running costs compared to the absorption system.
Super-cooled 0.5 °C Water

Super-cooler Heat exchanger
Water –2 °C
-3 °C
-6 °C
Pipe Releaser of super-cooling stateBrine Pre-heater
Refrigerator
Pump
Supply of
Ice-slurry £
Ice Layer nozzle Je Water 0 °C

t (Tap Water)
Water Layer
Figure 12.13. Schematic diagram of the ice slurry plant at Kyoto Station Building
12.2.5 MINE COOLING
One of the first industrial sectors to introduce ice slurries to satisfy cooling needs was the
mining industry. The continuously increasing depth of the mines, the larger machines and the
larger rock surfaces, resulted in higher temperatures in the tunnels and the pumping costs of
chilled water increased significantly (Sheer et al., 1984; Hager et al., 1991). Nowadays, mines
operate at depths of more than 3000 m where temperatures exceed 50°C. Ice slurries, having
much greater cooling capacity than chilled water, can effectively provide the same cooling
load with much lower flow rate offering significant savings in pumping cost and size of
equipment. Kidd (1995) also indicated that in mine cooling, when water is pumped
underground, its potential energy is converted into heat. For the case of a pilot study for the
Vaal Reefs mine, this heat represented a temperature rise of 2.34°C per 1,000 m of pipe run
which would reduce considerably the cooling capacity of the water. In the case of ice slurry
the temperature will be maintained fairly constant by the melting ice.
301
Ophir and Koren (1999) describe the ice slurry plant at the Western Deep Level Gold Mine in
South Africa. The ice slurry is produced by a Vacuum Freezing Vapour Compression process.
Four units are employed with a 3 MW cooling capacity each, serving working areas at depths
of 4000 m below ground. Each ice slurry unit consists of a large vapour compressor of 320
m3/sec volumetric displacement and a compression ratio of 8:1. Gravity is used for the
vertical transportation of ice slurry whereas horizontal transportation is achieved by a
pneumatic pumping system.
12.3 Food Processing
Ice slurries can be applied to many food processing applications. These include rapid cooling
of vegetables and fish, processing of meat and dairy products, food storage, display and
distribution.
12.3.1 RAPID COOLING OF VEGETABLES
In the rapid cooling of vegetables, ice slurry can be pumped at a low temperature, close to
0°C, to cooling coils and air is blown across the coils by centrifugal or axial flow fans onto
the vegetables which are laid out in trays or a moving belt, in a similar way to a blast freezer.
An example of a vegetable cooling facility is one in Boston Lincolnshire, UK. The plant
consists of a 88 kW ice slurry generator and a 10 m3 storage tank and is designed to meet a
peak load of 180 kW (EPS Ltd., 2003b).
12.3.2 FISH PROCESSING
Ice slurry is an excellent medium for fresh fish cooling. Wang et al. (1999) reported that tests
carried out with artificial and real fish to compare the cooling performance of ice slurry and
flake ice showed ice slurry to perform better than the traditionally used flake ice in preserving
the quality of the product.
In an ice-slurry fish cooling system, the fish is totally covered by the slurry leaving no air
pockets between the product and ice as is the case with flake ice. As a result, the cooling of
the fish is faster and the growth of bacteria is slower, resulting in longer product life.
Comparative tests between ice slurry and flake ice for the cooling of different fish have been
carried out by a number of investigators (Wang et al., 1999; Paul, 2002). Although the rate of
cooling is a function of the shape and size of the fish as well as its thermophysical properties,
in every case ice slurry was found to produce faster cooling. Figure 12.14 shows comparative
curves for plaice (flatfish) cooled in boxes immersed in ice slurry (Paul, 2002). The results
showed that the time required to cool the plaice down to 2°C was more than 3 times faster
than the time required with flake ice.
302
Temperature in the Center 18

16
Flake Ice
14
of the Fish, C
o
12
Ice Slurry
10
8
6
4
2
0
0 10 20 30 40 50 60
Time, min
Figure 12.14. Cooling rate of ice slurry compared to flake ice (Paul, 2002)
Ice slurry plants for the cooling of fish are in operation in many countries. An example of
such a plant is the prawn cooling plant at the Isle of Harris in Scotland (EPS Ltd., 2003b). The
system consists of a 10 kW ice slurry unit and a 6 m3 storage tank containing seawater. The
chilled seawater is essential to maintaining the quality of live prawns which are stored in the
tanks. Cooling of the seawater in the tank is provided by the ice slurry system which is
pumped through a heat exchanger to cool the seawater, bringing the temperature in the tank
very rapidly down to 8°C. After this, the temperature is reduced down to 4°C more slowly,
using a specially designed controller. The 10 kW ice slurry generator and the storage capacity
of the tank can meet a peak cooling load of 47 kW. Operation at night with off-peak
electricity can also lead to energy savings compared to conventional non-storage refrigeration
systems.
A plant described by Wang et al. (1999), shown in Figure 12.15, consists of two ammonia
based ice slurry generators, a 62 m3 storage tank with a top harvester machine, a 1.2 m3 ice re-
mixing tank and ice slurry distribution network. The ice is generated in the ice machines and
stored in the storage tank where the ice crystals are accumulated at the top, forming an ice
rich layer with 70% ice concentration. Once the storage tank is full of ice, brine is injected
into the tank and pushes the ice layer against the rotating blades of the harvester machine
which removes the ice crystals from the ice bed into a delivery chute. The ice is then
discharged into the ice-mixing tank, which is equipped with an agitator, where it is mixed
with brine forming ice slurry of 30 to 35% ice concentration. The mixture is then pumped to
several locations to accommodate the cooling requirements of the processing and transport
operations.
303
Figure 12.15. The ice slurry system of the fish processing plant
The overall benefits of using ice slurry in the fish processing industry are obvious. While ice
slurry is used through the entire operating process, a constant temperature of fish is
maintained from the moment fish are received, to the moment they are finally canned. In
addition, fish are chilled at a much faster rate than before. As a result, the quality of the
product, such as the texture, colour and bacteria retardation is greatly improved. Automated
fish handling with ice slurry improves the productivity of the plant, as the plant is able to deal
with peak load without inducing too many shifts per day.
Ice slurry systems have also been successfully implemented in either small boats or large
vessels. Conventional on-board cooling technologies are often based on refrigerated seawater
(RSW) systems and/or different types of fresh ice (e.g. flake ice). With RSW systems onboard
large vessels, fish are kept in fish holds with a circulation of seawater, at a mean temperature
of about 2 to 4°C. Some disadvantages of the system include: the volume of space taken up by
the refrigeration machinery; temperature level; and salt uptake, especially in smaller species
such as sardine.
Onboard smaller ships, fresh ice is normally loaded at the harbour, stored and eventually
mixed with the catch. The tendency of fresh ice to agglomerate limits in many cases the
quality of icing, since the ice has to be loosened by manual means before mixing. Mechanical
loosening with motor driven snails is also reported to be a major cause of injury on smaller
fishing boats in some countries. Furthermore, large blocks of ice mixed with fish often result
in uneven cooling rates. Last but not least, larger pieces of ice with its sometimes sharp edges
often damage the fish.
In addition, bacterial growth is extremely dependent on temperature. As a rule of thumb, the

shelf life of fish kept at 0°C is more than double that of fish kept at 5°C; the closer to the
freezing point of the fish, the better the product quality. Since bacterial growth rates cannot be
stopped, but only decelerated, it is also important to minimize the bacterial level by a rapid
304
drop of temperature in each part of the process. Nevertheless, care should be exercised to
avoid freezing of the fish as it may result in decreased saleability.
On fishing vessels, ice slurry can provide significant improvements over conventional chilling
methods, such as RSW or flake ice. Since ice slurry consists of small crystals, either in dry
form or pumpable slurry, it provides excellent chilling of fish while avoiding freezing and at
the same time preventing fish from bruising, because of the soft and flexible nature of ice
slurry. In addition ice slurry is easy to dose and mix with the catch.
12.3.3 MEAT PROCESSING
Paul (1999) describes an ice slurry system installed in a meat-processing factory in Germany
to cover all the cooling and air conditioning requirements of the 3800 m2 site. This includes
cooling for processing and storage of the meat, display cabinets and comfort cooling for the
offices. The peak cooling demand was calculated to be 385 kW. The installed plant has a 240
kW capacity and consists of two ice slurry generators of 120 kW each and a 36 m3 tank.
The capital cost of the plant was 435 000 Euros, while an equivalent direct expansion plant
would have cost 390,000 Euros. However, the annual operating cost of the ice slurry system
was found to be 89,000 Euros compared to a projected operating cost for a conventional plant
of 109,000 Euros, resulting in a payback period of just over two years.
12.3.4. DAIRY PROCESSING
An ice slurry plant for cheese processing is described by Gladis (1997). The plant in Hanford,
California, USA, is installed in a dairy factory with a daily production of 90 000 kg of
cheddar cheese. The cooling loads of both slow and rapid cooling processes such as cheese
starter cooling, whey protein concentration, filtration vessels, etc., are estimated to be 2546
kWh per day. The peak cooling demand occurs at 7:00 am when the initial cooling of the
cheese starter takes place from 85°C down to 25.5°C. This process requires 265 kW of
cooling load. Other simultaneous cooling demands in the factory bring the total peak load to
396 kW. This peak load occurs for only one hour during the day while for the remaining 23
hours the total cooling load varies between 56 and 148 kW, representing less than 40% of
peak load.
To reduce the peak demand, an ice slurry system was installed which consisted of an ice
slurry generator of 106 kW capacity and a 24.6 m3 storage tank, capable of storing 763 kWh
of energy (Figure 12.16). The load profile of the plant is shown in Figure 12.17.
The ice slurry plant operates continuously and generates ice slurry of 5 to 10% ice
concentration from an initial solution of water with 7% propylene glycol. The generated ice
slurry is pumped to the storage tank where, due to density differences, the ice crystals and
chilled solution separate and occupy the upper and lower parts of the tank, respectively. A
chilled solution of approximately 0°C is pumped from the bottom of the tank to the heat
exchangers and the “warm” solution returns to the tank and distributed through spray nozzles
over the ice layer.
305
Ice slurry
generator
return R-22 Vapour
Compressor/
PHE
Condensing
Ice slurry
unit
R-22 Liquid
Spray Headers
supply
Ice Slurry Tank
Figure 12.16. A schematic of the cheese processing ice slurry plant
450 350
400 Cooling Load
Energy stored in the tank, kWh

300
350
kWh stored
Cooling Load, kW
250
300
200
250
200 150
150
100
100
50
50
0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
hour of the day
Figure 12.17. Load profile and storage tank capacity over a 24-hour period
12.3.5 BREWERY
An ice slurry plant has been installed for process cooling and bottling the beer in Kempten,
Germany. This brewery produces 13 different kinds of beer. Nearly all German breweries use
ammonia as refrigerant in the production and the cold storage of the beer. As a result of the
geographic location in the centre of the city, and the safety and environmental requirements of
the city, the brewery decided to decrease the amount of ammonia used in their process. This
was the reason for the brewery to use ice slurry as a cold storage and transportation medium.
306
In this way the amount of ammonia has been decreased and the ammonia charge of the
cooling plant remains completely in the machinery room. All security devices against the
occurrence of damages can also be limited to the machinery room. The ice storage system is
placed at a distance of 50 m from the machinery room. There are two different circuits for
producing and delivering the ice slurry as shown in Figure 12.18.
Figure 12.18. Circuit schematic of the brewery plant
For the ice generation two orbital rod ice slurry generators are used with 350 kW of cooling
capacity. A pump discharges brine from the bottom of the storage tank to the top of the ice
slurry generators. The discharged brine must be completely ice free for this type of ice slurry
generator. Otherwise the ice slurry generators could freeze and stop the ice production. At the
outlet of the ice slurry generator a secondary pump discharges the newly formed ice slurry
back to the storage tank. The secondary pump is required, because the ice slurry generators
are not pressurized.
The volume of the storage tank is approximately 70 m³. The storage tank has two functions.
The first is to accumulate the ice and to increase the ice concentration in the storage. The
main result is that the ice slurry generators only produce ice during night time, and the ice
concentration can be increased up to a maximum of 30 wt-%. Another function is to decouple
the ice generation process from the consumer circuits. In order to obtain a homogeneous
dispersion of the carrier fluid and ice crystals, jet nozzles and stirrers are used. From the
storage tank, ice slurry is pumped through the consumer circuit at an ice concentration of 10
wt-%. This ice concentration is high enough to supply all the consumers with chilling. As a
result of the low ice concentration, ice slurry acts like a Newtonian fluid.
The consumer circuit is a two string system. There are three different branches for the
consumers; each of them is supplied by a pump. Different heat exchangers are used in this
plant. A plate heat exchanger is used for cooling the wort, an in-process product of the beer
production. The wort must be cooled down in a short time and a small temperature range must
be kept. The outlet temperature of ice slurry from these heat exchangers is high and without
any ice crystals. To keep the temperature of the wort within the targeted temperature range, a
307
part of the warm brine can be mixed by a 3 way valve with the cooled inlet stream. As a result
of the warm outlet the wort cooler has its own return pipe.
The fermentation and storage of the beer take place in special vessels, cooled by welded half-
round tubes. Each vessel has its own pipe connected to the consumer circuit.
The control of the plant is executed by a Siemens SPS, which also controls the refrigeration
plant. Ice concentration in the storage tank and the pipe connected to the consumer branch is
determined from the temperature measurement. The accuracy of estimating the ice
concentration by measuring the temperature depends strongly on the initial additive
concentration. So, it is essential to analyse the initial additive concentration. If the
temperature in the storage tank remains under a certain preset value, i.e. the ice concentration
remains above the desired value, the ice production process will be stopped. Experience has
shown that this control meets the demand of an industrial plant.
12.3.6 RETAIL FOOD APPLICATIONS
Traditional centralized refrigeration systems in large retail food stores require considerable
amounts of refrigerant. These systems are prone to leakage due to the large number of joints
and the long distribution pipe-work. The use of ice-slurry as a secondary fluid in retail food
stores can reduce considerably the refrigerant charge in the primary system without
significantly increasing the overall power consumption, due to the much smaller quantities of
slurry that has to be pumped in the secondary circuit compared to traditional single-phase
secondary systems.
Tassou et al. (2001) compared the performance of a refrigerated display cabinet evaporator
coil both with ice slurry and R22. The results indicated that even though the temperature of
ice slurry entering the coil was at least 2°C higher than the temperature of R22, the ice slurry
was able to provide slightly better heat transfer performance than R22.
Lueders (1999) presented a feasibility study for a supermarket that would employ an ice
slurry thermal storage system to serve the medium and high temperature loads of the
refrigerated fixtures and the HVAC requirements of a retail food store. A direct expansion
system would be used only for the low temperature frozen food display cases (–23°C). Heat
rejection from the low temperature DX circuit would be accomplished through a water-cooled
condenser served from the ice slurry system. The ice slurry plant which will employ ammonia
as primary refrigerant would consist of two ice slurry loops with separate ice slurry tanks. The
first loop would be connected to the “high temperature” ice slurry tank and will accommodate
the HVAC requirements of the building as well as the condensing requirements of the low
temperature DX system using a liquid solution at –1°C pumped from the bottom of the tank.
The second loop would be connected to the “medium temperature” ice slurry tank and would
accommodate the loads of the medium and high temperature display cabinets (–4°C and
4.5°C, respectively) using slurry of 15% ice concentration, at temperatures between –9°C and
–12°C. The proposed plant arrangement is shown in Figure 12.19.
308
Ice cream cases, – HVAC Cases

23 οC 22 οC -4 οC
Freon “Low”
Dairy cases
temperature loop
4.5οC
Glycol Ice Slurry

-1 οC -9 οC/-12 οC
Water-cooled
condenser “High” “Medium”
temperature temperature
Ice slurry Ice slurry
tank tank
Receiver
Ice slurry
t
Figure 12.19. A schematic diagram of a supermarket ice slurry system
The author made a comparison of pumping requirements, flow rates and piping costs between
the ice slurry system and a system using a single-phase secondary fluid as a heat transfer
medium. It was proposed that the ice slurry system would result in an 82% reduction in
secondary fluid flow (10.2 m3/h instead of 55.4 m3/h) producing 82% savings in pumping
costs and 51% savings in piping costs.
It was estimated that the capital cost of the ice slurry system would be 89% higher than that of
a conventional DX system (369 751 USD compared to 195 840 USD). The ice slurry system
was also found to produce a 3.5% increase in the energy consumption of the plant but it
would be able to shift 80% of the peak hour energy consumption to off peak hours which
would result in 49% savings in electricity costs resulting in a payback period of 4.4 years.
A similar system has been proposed by Davies and Lowes (2002a), who suggested that using
ice slurry to condense and sub-cool the refrigerant liquid of a conventional retail refrigeration
system down to 0°C would lead to an increase in the COP of the vapour compression
refrigeration system and produce running cost savings of the order of 20%. Controlling the
flow of ice slurry in the condenser will control the condensing pressure of the refrigeration
system and if off peak electricity is used to charge the ice slurry storage, the energy cost
savings could rise to 30% compared to ambient air heat rejection. If ice slurry is produced at
temperatures down to –10°C, the same system could also be used to satisfy the refrigeration
load of the medium and high temperature cabinets in the store directly. With around 120 000
supermarkets of more than 400 m2 floor area in the world, a widespread application of such a
system could lead to significant reductions in the energy consumption of the sector (Davies
and Lowes, 2002a).
309
One ice slurry systems manufacturer has recently developed an ice-slurry system for the
preservation of fresh produce, which is traditionally stored and preserved in ice (Sunwell,
2003). In 2002 the company delivered a few deepchillTM ice generation and distribution
systems to a large Canadian supermarket chain in Ontario. In this system ice crystals formed
inside an ice slurry generator with a 23 kW ice making capacity are pumped into an insulated
3.5 m3 storage-dispenser tank where they remain suspended in water. Dry ice crystals from
the top of the tank at a maximum rate of 3 ton/h are then mixed with a small amount of water
and the mixture containing 60% ice is pumped with a positive displacement pump (3.7 kW
power) through a piping system to the display cases where it is spread over the display surface
with a flexible hose. The ice delivery control at the individual display counters allows the
operator to fill the display cases by simply turning on the delivery system at the counter and
pumping the ice slurry into the cases (Figure 12.20). After drainage ice crystals form a
shining ice bed in the cases.
Traditionally, supermarket display cases for seafood and produce are commonly filled early in
the morning before the store is opened by manually shovelling ice into the trays. Pumping ice
slurry directly into the display cases at the counter eliminates any possible contamination of
the ice due to transporting and shovelling. The transportation of the ice in a sealed
environment guarantees a clean product at the counter. In addition, automation of the ice
distribution process with the ice slurry system means a significant reduction in labour and less
chance of work-related injuries, which result in lower cost.
Figure 12.20. High density ice slurry being discharged into display cases
(Photo courtesy of Sunwell)
310
12.4 Possible future applications
12.4.1. ICE PIGGING
In many industrial applications, components like pipes, ducts and heat exchangers, require
frequent and efficient internal cleaning. This process, commonly known as “pigging” is
usually accomplished using piston rings or other mechanical methods. However, mechanical
pigging cannot perform well in cleaning components with complex geometry. Experimental
and theoretical work carried out at Bristol University, UK (Quarini, 2002), to investigate the
cleaning efficiency of ice slurry, showed slurry to be able to “pig” even very complex
geometries. The tests were conducted with ice slurry of 10% ice concentration using either
salt or sugar as freezing point depressants. Fouling materials tested included jam and fats
(food industry), toothpaste (pharmaceutical product) and fine silt and sand (river water cooled
exchangers). The ice slurry proved capable of removing fouling materials from pipes with
sharp bends, orifice plates and even from heat exchangers. Further work is being carried out
but the first results are very encouraging for a new field of application.
12.4.2 MEDICAL APPLICATIONS
Ten to twelve minutes after a cardiac arrest, brain cells start dying rapidly because of lack of
blood flow to the brain. However, when the cells are cooled, their metabolism and their
chemical processes slow dramatically. Researchers at Argonne National Laboratory’s Energy
Laboratory working on developing methods for rapid cooling of the blood to sustain the heart
and brain cells after cardiac arrests have proposed a new possible medical application for ice
slurries (Argon National Laboratory). In this approach, a specially formulated ice-particle
slurry of high fluidity engineered using chemicals compatible with human tissues would be
pumped into the patient’s carotid arteries, jugular veins and lungs to rapidly lower the
temperature of vital heart and brain tissues and induce localised hypothermia. In simulated
head and heart cooling the technique has been shown to be able to cool the brain’s and heart’s
core temperatures down to 25°C in ten minutes which is much faster than conventional
methods. More work, however, is required on production and delivery of ice slurry before the
technique becomes a standard approach in helping cardiac arrest and stroke victims.
Ice slurry can also be used to treat bruising and relieve pain during sport or other activities. It
can be used as a cold pack, providing better contact with the body than traditional ice packs,
or for immersion cooling of bruised feet, hands, elbows, etc.
12.4.3 ARTIFICIAL SNOW
Paul (2002) notes that ice slurry could be an excellent kind of artificial snow for indoor
skiing. It can provide a better skiing surface than the commonly used snow substitutes, with
less energy consumption. A large-scale project in Germany is in progress for the construction
of a 30,000 m2 skiing hall. All the cooling requirements of the centre, from snow production
to air conditioning will be catered by ice slurry plants.
311
12.4.4 TRANSPORT REFRIGERATION
Another application proposed by Paul (2002) is the use of ice slurry as the cooling medium in
refrigerated vehicles. Since the efficient cooling of products with ice slurry is well
established, vehicles with double walls containing ice slurry in the gap could be an alternative
to conventional refrigerated vehicles. The ice-slurry vehicles could operate in a network with
ice slurry filling and discharging stations. A feasibility study by the author showed the idea to
be viable. This technology has been used since 1994 in refrigerated railroad cars in China.
Here the ice slurry is charged into compartments under the roof of each car.
A similar system, where chilling is provided by ice slurry contained in a double-walled

container and recharged at service stations, was in operation on certain trains of the Swiss
railway company. Here trolleys were chilled by ice slurry in the walls of the trolleys.
Since autumn 2004 an experimental unit is producing ice slurry for delivery vans in Austria.
The question came into discussion because some end users announced problems with their
existing systems in 1999. Since then several systems have been examined.
Some of the main points in question were:

o reducing service and maintenance costs,
o reducing the weight of the delivery vans,
o easy handling,
o use of existing refrigeration plants if possible,
o environment and sustainable technology (no use of HFCs if possible),
o fast availability of refrigeration energy,
o multi-usage of the van because of different temperature requirements,
o weight savings – reduction in the weight of the delivery van in comparison to existing
systems and hence increasing the payload,
o no additional technology (compressors, valves, pumps, etc.) used in the van.
Ice slurry offered advantages in many of the above topics. The main idea was to “refuel ice
slurry like gasoline at a gas station”.
Figure 12.21. Charging the unit Figure 12.22. Refrigerated body for delivery vans
312
The first test van in Austria is a small delivery van for frozen goods (Figures 12.21 and
12.22). This vehicle was chosen because it requires the highest cooling capacity due to the
many stops during the operation where doors are opened and goods are taken out. The
temperature is not allowed to rise above –18°C during the entire delivery period.
The tests yielded interesting details. The design of the heat exchanger had to be adapted to
practical requirements. In addition, changes were made to the ice generator and the pump. The
lowest ice slurry temperature achieved so far is about –39°C. The tests have demonstrated the
use of ice slurry for freezing applications, i.e. –18°C compartment temperature. The tests will
be extended thanks to the promising results.
The tests also showed that the heat exchanger required a special design, different to that used
for other purposes in the past. A new heat exchanger is now under development.
12.4.5 THE USE OF ICE SLURRY FOR FIRE FIGHTING
The properties of ice slurries may be exploited to great effect when they are used in place of
water to fight Class A fires i.e. non-fuel fires, which constitute the vast majority of fires
(Lowes, 2002). Because ice slurries can easily be pumped up to ice concentrations of 30%,
existing fire fighting equipment such as hoses and nozzles can be used in an ice-slurry fire
fighting system provided the slurry is stored and pumped from an appropriately stirred
holding tank. Slurry ice may also be used in conjunction with foaming agents. Most portable
water pumps used by fire brigades use the water they pump as the coolant for the engine. In
the case of ice slurry it would be better to employ an electric or air-cooled engine driven
pump, to minimise melting in the slurry.
Comparisons between the performance of water and ice slurry when tackling similar fires
demonstrated that ice slurry extinguished the fire in less than half the time taken with water,
and that the gas temperature in the test room was dramatically lower when using ice slurry.
The amount of steam formed when using ice slurry was said to be very low and dense. Only a
small volume of ice slurry was needed to extinguish the fire compared with the volume of
water normally used. The tests were carried out using ice slurry produced from a water
solution of 7% propylene glycol concentration giving an ice production temperature of –2°C.
The slurry was stored in a stirred 1 m3 polypropylene shipping container and standard fire
brigade suction hoses were employed to withdraw the slurry into a portable fire pump. The
slurry was delivered at 9 bar through a standard fire brigade hose of 25 mm internal diameter.
It has also been proposed that if sodium bicarbonate is used as the freezing point depressant in
the manufacture of ice slurry and the resulting slurry is pumped onto a fire together with a
weak solution of acetic acid, a copious carbon dioxide foam would be produced which may
further enhance the effectiveness of the ice slurry in extinguishing fires (Davies and Lowes,
2002b).
12.5 Conclusions
In this chapter a number of applications of ice slurry technology to different sectors of

commerce and industry have been described. The chapter also highlighted opportunities for
possible future applications. Over recent years, the field of applications has increased due to
313
advances in technology and increased understanding of the thermodynamic and

thermophysical properties of ice slurries. The spread of applications covers all continents but
the majority are concentrated in Europe and Japan. In Japan the interest has mainly been in air
conditioning with thermal storage and electrical load shifting to off-peak periods. In Europe,
the applications vary widely from air conditioning to immersion and process cooling.
The capital cost and energy consumption of ice slurry generators are factors that currently
inhibit a wider application of the technology. With further research and development and
increase in the number of applications, the capital cost of the equipment is likely to decrease
and this will increase further the attractiveness of the technology to potential users.
1. Argonne National Laboratory: http/www.ipd.anl.gov/biotech/programmes/biomedical/ice-

slurry.html
2. Davies, T.W.; Lowes, A.R.: A high efficiency refrigeration system, Proc. of a conference
on New Technologies in Commercial Refrigeration, IIR, Urbana, USA, pp. 93-100, 2002.
3. Davies, T.W.; Lowes, A.R.: Fire extinguishing system, UK Patent Application 0222022.6,
2002.
4. EPS Ltd.: 2003a, http://www.epsltd.co.uk/slurry_sites.pdf
5. EPS Ltd.: 2003b, http://www.epsltd.co.uk/slurryice1.htm
6. Everitt, N.: Ice Pick, Heat. Vent. Rev., vol.39, no.11, p. 19, 1999.
7. Gladis, S.: Ice Slurry Thermal Energy Storage for Cheese Process Cooling, ASHRAE
Transactions , Vol.103, Part 2, 1997.
8. Hager, M.; Kamper, T.: The Hydraulic Transport of Ice for Mine-Cooling. Powder
Handling & Processing, Vol. 3, no. 4, pp. 317-325, 1991.
9. Ise, H.; Tanino, M.; Kozawa, Y.: Ice Storage System in Kyoto Station Building.
Information booklet for the Technical Tour of the Fourth IIR Workshop on Ice Slurries,
pp. 11-16, Nov. 2001.
10. Kidd, J.: Slurry Ice Production in Gold Mining, The South African Mechanical
Engineering, Vol. 45, pp. 11-13, 1995.
11. Kuriyama, T.; Sawahata, Y.: Slurry Ice Transportation and Cold Distribution System,
Information booklet for the Technical Tour of the Fourth IIR Workshop on Ice Slurries,
pp. 1-6, Nov. 2001.
12. Lowes, A.R.: Fire-fighting apparatus and a method of fighting fire, UK Patent Application
0206813.8, 2002.
13. Lueders, D.: Proc. of the 21st Annual Meeting of the International Meeting of Ammonia
Refrigeration, pp. 301-326, 1999.
14. Nelson, K.; Pippin, J.; Dunlap, J.: University Ice Slurry System, presented at the IDEA
College-University Conference New Orleans, Louisiana, U.S.A, 1999.
15. Ophir, A.; Koren, A.: Vacuum Freezing Vapor Compression Process (V.F.V.C.) for mine
cooling, Proc. of the 20th International Congress of Refrigeration, IIF/IIR, vol. III, paper
508, Sydney, 1999.
16. Paul, J.: Innovative Applications of Pumpable Ice Slurry, presented before the Institute of
Refrigeration at London, UK, 2002.
17. Paul, J.: Generation and Utilisation of Liquid Ice (Binary Ice), Proc. of the 20th
International Congress of Refrigeration, IIF/IIR, vol. III, paper 495, 1999.
18. Rendall, S.: Future Air Conditioning, Bldg. Serv. Environ. Engr., vol. 22, no.11, p. 12,
1999.
314
19. Sheer, T.; Correia, R.M.; Chaplain E.J.; Hemp R.: Research into the use of ice for cooling
deep mines, Proc. of the 3rd International Mine Ventilation Congress, pp. 277-282, 1984.
20. Sims, B.: Going Green, Building Services Journal, CIBSE Supplement, pp. 8-9, 1999.
21. Sunwell, http://www.sunwell.com/index1.html, 2003.
22. Tassou, S.A.; Chaer, I.; Bellas, J.; Terzis, G.: Comparison of the performance of Ice
Slurry and Traditional Primary and Secondary Refrigerants in Refrigerated Food Display
Cabinet Cooling Coil, 4th IIR Workshop on Ice Slurries, pp. 87-95, 2001.
23. Quarini, J.: Ice-Pigging to Reduce and Remove Fouling and to Achieve Clean-in- Place,
Applied Thermal Engineering, Vol. 22, pp. 747-753, 2002.
24. Wang, M.J.; Hansen, T.M.; Kauffeld, M.; Christensen, K.G.; Goldstein, V.: Application
of Ice Slurry Technology in Fishery, Proc. of the 20th International Congress of
Refrigeration, IIF/IIR, vol. IV; paper 569, Sydney, 1999.
25. Wang, M.J.; Kusumoto, N.: Ice Based Thermal Storage in Multifunctional buildings, Heat
and Mass Transfer, Vol. 37, pp. 594-604, 2001.
315
CHAPTER 13. CONCLUSIONS AND RECOMMENDATIONS FOR FURTHER R&D

By Michael Kauffeld, Masahiro Kawaji and Peter W. Egolf
Multifunctional (thermal) fluids and suspensions, also named “intelligent fluids”, are new
classes of fluids with improved (thermal) properties. These fluids can be designed to
optimally fulfil particular objectives. In refrigeration and air conditioning this can be
enhanced thermal conductivity of a fluid, better heat transfer characteristics, a higher thermal
energy storage capacity, temperature stabilization, reduced pressure drop, etc. The use of
phase change materials (PCM) as a dispersed phase in a continuous carrier fluid leads — due
to its phase change (latent heat) — to very high energy densities, whereby the pumpability of
the fluid is still maintained. Such substances are named “phase change slurries (PCS)”. Ice
slurry is the best known PCS. It is restricted to temperatures below 0oC.
Ice slurries and other phase change slurries as well, are fluids with dispersed particles, which
undergo phase change at the melting temperature of the dispersed phase. If the fluids are
mixtures, very often a temperature glide occurs, which is apparent as the enthalpy h changes
continuously as a function of the temperature T. The energy to destroy a solid crystal during a
melting process is stored in the material, and when the material — in the reverse process — is
solidified, this amount of thermal energy (latent heat) is released again. In a water/ice
transition the stored energy is 333 kJ/kg. Because the concentration of ice particles in a carrier
fluid is usually less than 50% in technical applications, the specific enthalpy due to phase
change will typically be smaller than 170 kJ/kg for ice slurries. In addition to this latent heat,
a smaller fraction of sensible heat for cooling or heating of the carrier fluid and the ice
crystals is involved. Thus, the energy storage of ice slurries can be eight to ten times higher
compared with the conventional storage technology using chilled water, depending on the
operational temperature range of the engineering systems. Furthermore, the pumping power
can be reduced by up to 80% and at the same time heat transfer can be increased twofold
when compared with single-phase liquid used as a secondary refrigerant. In fact, heat transfer
for melting ice slurry is in the same range as for evaporating HFC refrigerants.
One of the characteristics of ice slurries is that the particles disappear in the melting process
and have to be created again by a special ice slurry generator. In storage tanks the particles
experience significant buoyancy forces, which cause stratification of the ice at the top of the
tank. Therefore, a mixing device or other special piece of equipment is usually necessary to
create a homogeneous ice suspension in the storage tank, which guarantees safe system
operation without the occurrence of tube clogging. The simplicity of freezing water with an
environmentally friendly additive (alcohols, salts, etc.) and obtaining very high specific
enthalpies make the application of ice slurries a promising technology for the future.
Pervading the studies of the IIR Working Party on Ice Slurries and from the material in the
foregoing chapters of this book, a few topics appear to be of relevance to future research and
development:
1. Heat transfer in melting and freezing ice slurries with a range of additives
2. Ice slurry generators
3. Principles of ice storage operation
4. Suspensions with melting temperatures above 0°C
5. Air-conditioning systems with low supply air temperature
6. New applications
317
1. Heat transfer of melting ice slurry

The thermal behaviour of ice slurry in heat exchangers is important to understand in
successfully utilising the distributed ice slurry. On the other hand, the heat transfer
characteristics of ice slurry during generation are important for the optimum design of ice
slurry generators.
Future R & D should cover an experimental parametric study of the influences of ice
concentration, additives, heat flux and velocity on the heat transfer coefficients.
In relation to melt-off rates in storage tanks, a heat transfer model for hot fluid flow through a
packed bed of ice slurry would also be useful in the design of thermal ice slurry systems.
The effect of the time behaviour of ice particle size on the heat transfer properties should be
quantified.
2. Ice slurry generators

Many different ice slurry production techniques are known. All of them have advantages and
drawbacks. The one feature that all of them have in common is the high price of the ice
generator itself. In many ice slurry systems, the ice generator is the single most expensive
component. Its price is often a barrier to people/companies when applying ice slurry systems.
Future research and development should focus on less expensive and more reliable ice slurry
production methods.
3. Principles of ice storage operation

Compared with traditional (harvest and coil-based) ice storage systems, ice slurry offers a
more dynamic type of storage. However, the application of ice slurry systems still remains
relatively limited due to lower temperature as compared with pure water ice from e.g.
harvesting systems, costly equipment and more complex storage layout. Intensive work is
needed to develop more cost-effective ice slurry generators working at close to 0°C. To bring
the technology one step closer to its real potential, it is of utmost importance to focus on the
storage principles. Today, systems are in many cases constrained by their mode of operation
and also are often unsuitable to fit into existing buildings and storage facilities.
If the full potential of ice slurry is to be exploited, i.e. pumping ice slurry to heat exchangers,
the storage tank must be either homogeneously mixed or a top harvester type where
subsequent mixing with a carrier fluid is employed. The homogeneously mixed storage is
used in many small ice slurry installations. The major drawbacks of this storage principle are
relatively low maximum ice content (35 wt-%) allowed and a relatively high cost of the
mixing devices when applied to large-scale applications.
Future R & D should yield design principles that can be applied to existing storage facilities,
e.g. both cylindrical and rectangular tanks, in heterogeneous, partially homogeneous or fully
homogeneous mode. The maximum ice content of these storage systems must be quantified
and in the case of the heterogeneous storage, the melt-off capacity, extraction temperatures
and the feasibility of ice slurry extraction should be investigated at different:
• flow rates,
• return temperatures,
• diffuser designs,
• internal mixing principles.
318
4. Suspension/emulsion systems with melting temperatures above 0°C

For many purposes, it is difficult to obtain good economy in ice bank plants, as the ice has to
be produced at a lower temperature than that required at the cooling coils (point of cold
utilization). Exergy is lost in this way in the production of ice.
The selection of the optimum design for a storage system is based on a balance among
savings in energy during production, savings in capital expenditure on refrigerating machines,
reduction in operating costs and increases in investment in the storage tank as well as possibly
increased energy consumption due to a lower evaporation temperature in the ice generator. As
the storage tank may amount to a significant part of the total capital expenditure, it is essential
that a high energy density can be obtained in the storage tank and that latent heat is utilised if
the size has to be kept at a satisfactory level. It will thus be optimal to use a phase-change
material (PCM) above 0°C in the same manner as ice slurry. A number of materials are
available which meet the requirements for high energy density and melting point temperatures
between e.g. 6°C and 10°C.
Future R & D should investigate thermo-physical and thermodynamic properties as well as
practical handling aspects of PCM suspensions/emulsions. The properties of the most
applicable PCMs should be explored experimentally.
The practical investigations must focus on feasible storage design, emulsification processes,
bacteriology, material compatibility, surface deposition, rheology, heat transfer during phase
change (melting and crystallization) and suspension properties.
5. Air-conditioning systems with low supply air temperature

The supply air temperature of air conditioning systems can be reduced below the typical
12°C to 15°C. Typical low-temperature air supply temperatures are between 2°C and 7°C.
Several researchers and companies have published investigations and results from installed
systems, which show improved overall energy efficiency for cold air distribution systems.
The power saved on the air side (reduced fan power due to reduced air-flow) is much greater
than the increase in compressor power consumption due to a lower evaporation temperature.
The installation costs are also reduced due to smaller fans, smaller ducts and reduced floor
height (some authors quote a gain of stories and hence floor area in the same overall building
height when applying cold air distribution systems). Typically, low-temperature air is
produced with direct evaporation of refrigerant or with cold water very close to its freezing
point.
Ice slurry could be used in a beneficial manner to produce low-temperature air. Future R & D
work should concentrate on ice slurry generation principles, which result in ice slurry
temperatures close to 0°C. Heat transfer to the air should be studied, as it is expected that the
ice slurry will have better heat transfer coefficients than cold water. Air coils might have
smaller tubes with ice slurry facilitating lower pressure drop on the air side and hence reduced
fan power.
6. New applications
In addition, new applications outside the traditional refrigeration and air conditioning area
may require more research. Chapter 12 mentions ice pigging for cleaning of complex systems,
the in-situ cooling of the heart or brain after cardiac arrests and strokes, artificial snow
319
production and fire fighting. But the refrigeration area still offers new applications for ice
slurry, e.g. transport vehicle cooling.
This Handbook presents current knowledge on ice slurries and the application of ice slurries.
The material has been compiled by more than 50 leading scientists in this field from around
the world. We all hope that the material presented will be helpful to others and will assist in
increasing the acceptance of this very promising technology.
320
Appendix 1: List of Authors

(indicated by code in brackets – LA = Lead Author, CA = Co-Author)
Derrick Ata-Caesar (CA 4) Tom Davies (CA 5 - Reviewer)

University of Applied Sciences of Western The University of Exeter, The Harrison Bldg.
Switzerland School of Engg. and Computer Science
Route de Cheseaux 1 North Park Road
CH-1401 Yverdon-les-Bains, Switzerland Exeter, EX4 4QF, UK
E-mail: Derrick.Atta-Caesar@eivd.ch E-mail: T.W.Davies@exeter.ac.uk
Michel Barth (CA 5) Christian Doetsch (CA 3)

Association Française du Froid Fraunhofer Institut UMSICHT
17, rue Guillaume Apollinaire Osterfelder Strasse 3
75006 – Paris, France D-46047 Oberhausen, Germany
E-mail: michel.barth@free.fr E-mail: Doe@umsicht.fhg.de
Jean-Pierre Bedecarrats (CA 4) Peter W. Egolf (Editor, CA 1, LA 4 and CA

LTE – Université de Pau 13)
Avenue de l’Université, BP 1155 University of Applied Sciences of Western
64013 – Pau Cedex, France Switzerland
E-mail: jean-pierre.bedecarrats@univ-pau.fr Route de Cheseaux 1
CH-1401 Yverdon-les-Bains, Switzerland
Thorsten Behnert (CA 12) E-mail: peter.egolf@eivd.ch
AximaRefrigeration GmbH
Kemptener Str. 11-15 Kostadin Fikiin (LA 9)
D-88131 Lindau/Bodensee, Germany Refrigeration Science and
E-mail: thorsten.behnert@axima.eu.com Technology Division
Technical University of Sofia
John Bellas (CA 12) 8 Kliment Ohridski Blvd.
Brunel University BG-1756 Sofia, Bulgaria
Department of Mechanical Engineering E-mail: agf@tu-sofia.bg
Uxbridge – Middlesex, UB83PH, UK
E-mail: john.bellas@brunel.ac.uk Laurence Fournaison (CA 2 and 10)
Cemagref
Issa Chaer (CA 3 and 12) Parc de Tourvoie
Brunel University BP 44
Department of Mechanical Engineering F-92163 Antony, France
Uxbridge Middlesex, UB8 3PH, UK E-Mail: Laurence.fournaison@cemagref.fr
E-mail: issa.chaer@brunel.ac.uk
Beat Frei (CA 2, 3, 6 and 7)
Kim Gardø Christensen (CA 3 and 12) University of Applied Sciences Lucerne
Danish Technological Institute Technikumstrasse 21
Teknologiparken CH-6048 Horw, Switzerland
Kongsvang Allé 29 E-mail: Bhfrei@hta.fhz.ch
DK-8000 Aarhus C, Denmark
E-mail: kim.gardo.christensen@teknologisk.dk Vladimir Goldstein (CA 12)
Sunwell Technologies Inc.
Viktor Cucarella (CA 5.2.7) 180 Caster Avenue
Technical University Dresden Woodbridge, Ontario L4L 5Y7, Canada
Faculty of Mechanical Engineering E-mail: vgold@sunwell.com
Institute of Material Science
01062 Dresden, Germany Giuseppe Grazzini (LA 11)
E-mail: victor.cucarella@mailbox.tu-dresden.de Università di Firenze
Dipartimento di Energetica "S.Stecco"
Via S. Marta 3
321
IT-50139 Firenze, Italy Yoshiyuki Kozawa (LA 7, CA 5 and 11)

E-mail: ggrazzini@ing.unifi.it Takasago Thermal Engineering Co., Ltd.
3150 Iiyama
Jacques Guilpart (LA 10) 243-0213 Atsugi, Kanagawa Pref., Japan
Cemagref E-mail: YoshiyukiKozawa@tte-net.co.jp
Parc de Tourvoie
BP 44 Adrien Laude-Bousquet (CA 5)
F-92163 Antony Cedex, France ALB
E-mail: Jacques.guilpart@cemagref.fr 70, avenue de l’Europe
69480 – Anse, France
Torben Hansen (CA 2, 3 and 9, LA 6) E-mail: adrien.laude-bousquet@wanadoo.fr
Danish Technological Institute
Teknologiparken M. Leprieur (CA 5)
Kongsvang Allé 29 Dinan District – Pole Cristal
DK-8000 Aarhus C, Denmark 34 rue Bertrand Robidou - BP 357
E-mail: torben.hansen@teknologisk.dk 22106 – Dinan Cedex, France
E-mail: m.leprieur@pole-cristal.tm.fr
Robert Huhn (LA 7)
Dresden University of Technology Jeroen Meewisse (CA 5)
Institute of Power Engineering Delft University
D-01062 Dresden, Germany Michiel de Ruyterweg 37
E-mail: rhuhn@metrs1.mw.tu-dresden.de NL-2628 BX Delft, The Netherlands
E-mail: j.w.meewisse@wbmt.tudelft.nl
Cecilia Hägg (CA 2)
Royal Inst. of Technology, KTH Åke Melinder (LA 2 - Reviewer)
Dept. Energy Technology/ETT Royal Inst. of Technology, KTH
S-10044 Stockholm, Sweden Dept. Energy Technology/ETT
E-mail: cecilia@energy.kth.se S-10044 Stockholm, Sweden
E-mail: ake@energy.kth.se
Rob Jans (CA Appendix 2)
Fri Jado D. Mito (CA 5)
Oude Kerkstraat 2 Takasago Thermal Engineering Co., Ltd.
NL-4878 AA Etten-Leur, The Netherlands 3150 Iiyama
E-mail: Rob.Jans@net.HCC.nl 243-0213 Atsugi, Kanagawa Pref., Japan
E-mail:
Michael Kauffeld (Editor and CA 1, 5, 6, 7, 8
and 9, LA 13) Ebbe Nørgaard Jensen (LA 8)
University of Applied Sciences Danish Technological Institute
Moltkestr. 30 Teknologiparken
D-76133 Karlsruhe, Germany Kongsvang Allé 29
E-mail: Michael.kauffeld@fh-karlsruhe.de DK-8000 Aarhus C, Denmark
E-mail: ebbe.norgaard@teknologisk.dk
Masahiro Kawaji (Editor, CA 1, 4 and 13, LA 5)
S. Pazzi (CA 11)
University of Toronto
Università di Firenze
200 College St.
Dipartimento di Energetica "S.Stecco"
Toronto, Ontario, M5S 3E5, Canada
Via S. Marta 3
E-mail: Kawaji@ecf.utoronto.ca
IT-50139 Firenze, Italy
Andrei Kitanovski (LA 3, LA 4, CA 2) E-mail: s.pazzi@ing.unifi.it
University of Applied Sciences of Western Pepijn Pronk (CA 5)
Switzerland Delft University of Technology
Route de Cheseaux 1 Faculty of Mech Engg. and Marine Techn.
CH-1401 Yverdon-les-Bains, Switzerland Section Refrig. and Indoor Climate Control
E-mail: andrej.kitanovski@eivd.ch Mekelweg 2
NL-2628 CD Delft, The Netherlands
E-mail: p.pronk@wbmt.tudelft.nl
322
Savvas Tassou (LA 12)

Patrick Reghem (CA 3) Brunel University
University of Pau Kingston Lane
Avenue de l'Université Uxbridge, UB8 3PH, UK
F-64000 Pau, France E-mail: Savvas.Tassou@brunel.ac.uk
E-mail: Patrick.reghem@univ-pau.fr
Oleg B. Tsvetkov (CA 2)
Paul Rivet (CA appendix 2-A2, CA 12) Chairman of the St. Petersburg
MC international SA Regional Branch Academician
16/18 Avenue Morane Saulnier Lomonosov str. 9
F-78941 Velizy, France 191002 St. Petersburg, Russia
E-mail: Privet@Mc-international.fr E-mail: obereg@softrex.com
Pedro D. Sanz (CA 5) Didier Vuarnoz (CA 2 and 3)
Department of Engineering University of Applied Sciences of Western
Instituto del Frío (CSIC) Switzerland
Calle José Antonio Nováis, 10 Route de Cheseaux 1
Ciudad Universitaria CH-1401 Yverdon-les-Bains, Switzerland
E-28040 Madrid, Spain E-mail: didier.vuarnoz@eivd.ch
E-mail: psanz@if.csic.es
Ming-Jian Wang (CA 5, 9 and 11)
Osmann Sari (CA 4) Sunwell Technologies Inc.
École d'Ingénieurs du Canton de Vaud 180 Caster Avenue
Route de Cheseaux 1 Woodbridge, Ontario L4L 5Y7, Canada
CH-1401 Yverdon-les-Bains, Switzerland E-mail: wang@sunwell.com
E-mail: Sari@eivd.ch
Hartmut Worch (CA 5.2.7)
Alen Sarlah (CA 4) Technical University Dresden
University of Ljubljana Faculty of Mechanical Engineering
Faculty of Mechanical Engineering Institute of Material Science
Askerceva 6 01062 Dresden, Germany
1000 Ljubljana, Slovenia E-mail: hworch@rcs.urz.tu-dresden.de
E-mail: alen.sarlah@fs.uni-lj.si
Thomas Zwieg (LA 2.1 and 2.2)
Evangelos Stamatiou (CA 4 and 5) Danish Technological Institut
University of Toronto Kongsvang Allé 29
200 College St. DK-8000 Aarhus C, Denmark
Toronto, Ontario, M5S 3E5, Canada E-mail: thomas.zwieg@teknologisk.dk
E-mail: stamtio@chem-eng.utoronto.ca
Françoise Strub (CA 4)

Laboratoire La TEP
UFR Sciences & Techniques
BP 1155 - Avenue de l'Université
64013 – Pau Cedex, France
E-mail: francoise.strub@univ-pau.fr
Masayuki Tanino (CA 5 and 11)

Takasago Thermal Engineering Co., Ltd.
3150 Iiyama
243-0213 Atsugi
Kanagawa Pref., Japan
E-mail: Masayuki_Tanino@tte-net.co.jp
323
Appendix 2: List of Selected Plants

Table A1. List of ice slurry applications in various countries of the world
(based on private communications by the IIR Ice Slurry Working Group)
Country Approximate Typical applications

number
Argentina 8–10 Fishing vessels, fish processing plant
Australia 3–4 Fishing vessels, fish processing plant, air conditioning
Barbados 1–2 On-board fish processing
Canada 7–8 Fish processing plants, produce packing, air conditioning
Chile 20–25 Fishing vessels, fish processing plants
China 7–8 Fishing vessels, railway cargo cooling, air conditioning
Cuba 1–2 Hospital air conditioning and refrigeration
Denmark 2–3 Air conditioning, process cooling
Ecuador 1–3 Salmon farming
Faeroe Island 7–8 Fishing vessels
France 7–10 Air conditioning, supermarkets, fishing vessels, cold stores,

catering
Germany 15–17 Replacement of CFC air-conditioning plants, retrofit of CFC

air-conditioning systems, meat processing plants, breweries,
food vendors, fish processing
Great Britain 8–10 Fishing vessels, air conditioning, cold stores, food cooling
Guatemala 1–2 Shrimp processing
Iceland 50–60 Fishing vessels
Japan 250–260 Building air conditioning, fisheries, cold storage, dairy

production
Korea 2 Building air conditioning, poultry processing plant
Mexico 2–3 Fish processing plants, railcar chilling
Namibia 4–6 Fishing vessels,
New Zealand 7–9 Fishing vessels, thermal storage
325
Country Approximate Typical applications

number
Norway 20–30 Fishing vessels, fish processing plants
Philippines 4–6 Poultry processing plants, chemical crystallization
Russia 1–3 Aircraft air conditioning
South Africa 22–23 Fishing processing plant, fishing vessels, mine cooling
Spain 8–9 Fishing vessels
Switzerland 4–6 Supermarkets, trolley cooling in passenger trains, ground

cooling of airplanes, pharmaceutical production park
The 15–20 Fishing vessels, cold store, supermarkets

Netherlands
USA 15–20 Fish processing plants, air conditioning (thermal storage),

penguin enclosure, fire protection
326
Table A2. a) France: scraped surface technology (reference LGL)

by P. Rivet
Year of Specific technical data Refrigeration Number of

Type of installation Power Secondary fluid
installation Capacity ice slurry
Refrigerant Ice storage (kW) max (kW) generators
Mini supermarket 8 m3 Ethanol, –5°C
1996 R22 14 and 4 20 and 5 2
(demonstration) 0.5 m3 Ethanol, –30°C
1997 Fishing vessels R22 4 See water
Supermakets (cold
1997 R507 130 200 8 Ethanol, –4.8°C
stores)
1998 Air conditioning R22 1 m3 10 20 1 Ethanol, –1.5°C
Cold storage plant 0°C,
1998 Cold storage hall R22 28 m3 80 195 4 Ethanol, –2.4°C
+12°C
1998 Fishing vessels R22 5 Sea water
Kitchen, refrigeration
2000 R404A 8 m3 1 Ethanol, –3.5°C
of meals
Table A2. b) France: brushed surface technology (reference LGL)

by P. Rivet
Supermarket, cold storage

2000 R22 6 m3 50 1 Prop. Glycol, –4.5°C
plant
Dairies (milk and serum
2001 R404A 30 m3 120 300 2 Water-ethanol, –4.5°C
cooling)
Supermarket (CO2), cold
2001 R404A/CO2 50 CO2
storage plant, ice machine
327
Table A2. c) Worldwide: scraped surface technology installed by SUNWELL

by Ming-Jian Wang
Year of Primary Ice Ice Storage Cold Output for Cold Output for
Type installation Location
Installation Refrigerant Production Direct Cooling Indirect Cooling
Building air conditioning
1990 Japan R134A 32x130 kW 41680 kW-h Chilled water
for Crystal Tower, Osaka
Broccoli packing plant, Slurry at about
1990 Canada R22 48 ton/day 2x50 ton ice
cooling and packing broccoli 20% ice fraction
Shrimp processing plant, Slurry at 30-40%
1993 U.S.A. R707 20 ton/day 60 ton ice
shrimp cooling and preservation ice fraction
Poultry processing plant Slurry at 30-40%
1997 Philippines R22 24 ton/day 42 ton ice
cooling chicken ice fraction
Transport refrigeration Ice slurry with salt
1994 China R22 85 ton/day 2 x 250 ton
cooling railroad cars (5-25% fraction)
Arctic habitat
1999 U.S.A. R22 16 ton/day 20 ton ice Dry crystal ice
Crystal ice for penguin enclosure
Ice slurry with
Fish processing plant, 15%/50% ice
1999 Japan R404A 96 ton/day 40 ton ice
Saury cooling and packing fraction;
Chilled seawater
Salmon processing plant Slurry at 35-50%
2001 Norway R404A 15 ton/day 25 ton ice
Salmon cooling ice fraction
Slurry at 60% ice
Supermarket
2002 Canada R404A 3.5 ton/day 3.2 ton ice fraction, water
Ice for display cases
drained afterwards
328
Table A2. d) Switzerland: Scraped surface technology

by Peter W. Egolf
Year of Specific technical data Refrigeration Number of

Type of installation Power Secondary fluid
installation capacity ice slurry
Refrigerant Ice storage (kW) max (kW) generators
Migros Lancy Onex,
Propylene glycol at 13%, -
1995, 1999 display cabinets & storage R507 1 m3 90 92.4 26
4.5°C
plants
2001 Pilot system NH3 1.5 m3 10.9 15.5 4 Ethanol 10%, -3.8°C
329
IIF-IIR – Handbook on Ice Slurries – 2005NOMFICHIER
Table A2. d) Ice-slurry Production and Storage Systems in Japan

by Yoshiyuki Kozawa, Masayuki Tanino
Liquefied ice systems

Ice manufacturing mechanics Company / Characteristics
(1) Continuous manufacturing of ice slurry by using scraped Sunwell Japan Ltd.
surface technology, Takenaka Corporation
(2) Storage of uniform sherbet in a tank by feeding of ice Sekisui Plant System Co., Ltd.
slurry.
(1) Heat Storage Medium:

Water/Ice/ Alcohol
(2) Designed Max. IPF: 50%
(3) Saving energy and cost by
feeding of ice slurry to
secondary systems, Power
of Pump: 30% of Cold
Water Feeding, Size of
Pipe: 50% of Cold Water
System
(4) Cold heat source in building
air-conditioning, by means of
natural circulation of Freon
(Tokyo Electric Power Company,
1996)
(5) Number of systems
installed: 160
(Takenaka Corporation, 2000)
328
Table A2. d) Ice-slurry Production and Storage Systems in Japan (continued)
(1) Ice slurry manufacturing by means of repeated icing and Shimizu Corporation
de-icing modes in evaporator tubes,
(2) Feeding of ice slurry into storage tanks.
Water/Ice/Inorganic
Compounds
(3) Melting process with high
performance due to
transporting of ice slurry
(4) Heat storage medium
without rotting and algae
installed: 1; volume of tank:
75 m3
(Tokyo Electric Power

Company, 1996)
331
(1) Ice slurry manufacturing by means of ice and de-ice cycles in Maekawa MFG. Co., Ltd.
ice makers, or by using rounding poles in ice makers,
(2) Storage of uniform sherbet in a tank by feeding of ice slurry,
(3) Feeding of ice slurry with certain ice content into secondary
Water/Ice/ Alcohol
systems from ice storage tanks.
(6) Saving energy and cost by
feeding ice slurry to secondary
systems, Power of Pump: 30% of
Cold Water Feeding, Size of
Pipe: 50% of Cooled Water
System
(7) High response to heat load of air
conditioning by controlling ice
content
(8) Number of systems installed for
air conditioning: 1; volume of
tank: 19 m3.
Number of systems installed for
warehouse cooling: 3
(Tokyo Electric Power Company,
1996)
(1) Continuous manufacturing of ice slurry with about 100 µm ice Mitsui Engineering &
particles, Shipbuilding Co., Ltd.
Water/Ice/Alcohol
(3) Non-adhering ice particles
on wall and/or surrounding
ice particles by using the
special type of brine
installed: 5; total volume
of tanks: 242 m3
Company, 1996)
332
Supercooled water systems

(1) Making supercooled water Takasago Thermal Engineering Co., Ltd.

(–2°C) in a shell & tube-type
supercooling heat exchanger,
(2) Continuous manufacturing of (1) Heat Storage Medium: Tap water and ice
ice slurry by release of (2) Designed Max. IPF: 40%
supercooling, (3) Flexible system arrangement, especially no restriction as
(3) Feeding of ice slurry into to shapes and locations of ice storage tanks
storage tanks. (4) Temperature of tank wall > 0°C Non-freezing of water in
concrete walls
(Tokyo Electric Power Company, 1996)
(5) Number of systems installed: 23; total volume of tanks:
8520 m3
(by Takasago Thermal Engineering as of 2002)
333
Ice manufacturing mechanics Characteristics

(1) Making supercooled water (–2°C) in supercooling heat exchanger,
(2) Continuous manufacturing of ice slurry by release of supercooling due to an (1) New type of
ultrasonic wave, this system
(Mito, D. et al.,
2002)
(1) Making of supercooled water (–2°C) in a plate-type super-cooling Shinryo Corporation

heat exchanger,
(2) Continuous manufacturing of ice slurry by release of supercooling
due to the vibration of leaf spring, (1) Heat Storage Medium:
(3) Feeding of ice slurry into storage tanks. Tap water and ice
(2) Designed Max. IPF:
50%
(3) Compact-type
supercooling heat
exchanger
(Nagato, H., 2001)

installed: 6
334

(1) Making supercooled water (–2°C) in a plate-type evaporator, Shimizu Corporation

(2) Continuous manufacturing of ice slurry by release of
supercooling due to additional cooling by Peltier elements,
(3) Feeding of ice slurry into storage tanks. (1) Heat Storage Medium: Tap
water and ice
(3) Compact-type supercooling
heat exchanger
(4) Supercooled ice making heat
pump unit
(5) Number of systems installed:
1; volume of tank: 23 m3
(Kurihara, T., Kawashima, M.,

2001)
(1) Making of supercooled water in an evaporator, Mitsubishi Heavy Industries,

(2) Continuous manufacturing of ice slurry by release of supercooling Ltd.
in storage tanks.
(1) Heat Storage Medium: Tap

water and ice
(3) Oil-less turbo-type
refrigeration unit, High
COP (=120% of usual
refrigeration unit
installed: 1; volume of
tank: 87 m3
(Kakutani, S., Osa, N., 1999)
335
(1) Making of supercooled water in a Nippon Steel Corporation

plate-type super-cooling heat
exchanger,
(2) Continuous manufacturing of ice (1) Heat Storage Medium: Tap water and ice
slurry by release of supercooling in (2) Designed Max. IPF: 40%
storage tanks. (3) Compact-type supercooling heat exchanger
(4) Number of systems installed: 15
(1) Making of supercooled water Daikin Industries, Ltd.

in an evaporator,
(2) Continuous manufacturing of (1) Heat Storage Medium: Tap water and ice
ice slurry by release of (2) Designed Max. IPF: 40%
supercooling due to the (3) Ice storage unit of multi-zone air- conditioners for building uses
additional cooling in a (4) Number of units installed: 2000
separated tank, (Information from Tokyo Electric Power Company [1999,6])
(3) Feeding of ice slurry into
storage tanks.
336
(1) Making of supercooled water Toshiba Corporation

in a plate-type evaporator,
(2) Continuous manufacturing of
ice slurry by release of (1) Heat Storage Medium: Tap water and ice
supercooling in storage tanks. (2) Designed Max. IPF: 40%
(3) Compact-type supercooling heat exchanger
(4) Ice storage unit of multi-zone air- conditioner for
building uses
(Toshiba Corporation, 1998)
(5) Number of units installed: 160
(Reply of Toshiba Corporation
to a questionnaire [1999, 1996])
337
Ice harvest systems

(1) Repeating of freezing and harvesting on ice Obayashi Corporation
making plates grouped into 4,
(2) Storing of harvested ice in a water tank,
(3) Transporting of ice slurry after adjusting ice (1) Heat Storage Medium: Tap water and
content to secondary systems. ice
(3) Number of systems installed: 13; total
volume of tanks: 835 m3
(Tokyo Electric Power Company, 1996)
338
Ice scratch systems

(1) Instantaneous freezing of water flowing on the inner surface of a Taikisha Ltd.
cylindrical evaporator,
(2) Making ice flakes of 0.5 mm thickness by scratching with a rotating
scrubber, (1) Heat Storage
(3) Storing of ice flakes in a water tank. Medium: Tap water
and ice
60%
installed: 2; total
volume of tanks: 7 m3

Company, 1996)
(1) Making of thin film ice of 2 mm thickness, Sanki Engineering Co., Ltd.
(2) Scratching a thin film ice mechanically into a storage tank.
(1) Heat Storage

Medium: Tap water
and ice
30%
installed: 2; total
volume of tanks: 75
m3

Company, 1996)
339
Ice manufacturing mechanics Company /

Characteristics
(1) Freezing of sprayed water to ice of about 1 mm thickness on a
rotating evaporator drum, Hazama Corporation
(2) Scratching down of ice by a blade installed outside the drum
into a storage tank.
Tap water and ice
40%
installed: 1; volume of
tank: 357 m3

Company, 1996)
340
Other Systems installed in Japan

(1) Freezing of water on low temperature organic compound Toshiba Corporation

droplets sprayed into a water tank due to direct heat
exchange,
(2) Accumulating gradually of fragile ice from the free (1) Heat Storage Medium: Tap water
surface of water in a tank. and ice
(3) Simple ice manufacturing
(Hashiguchi, M. et al., 1995)
(4) Number of systems installed: 1;

volume of tank: 960 m3
(Watanabe, Y. et al., 1997)
(1) Ice manufacturing in the evaporator of a lithium bromide Taikisha Ltd.

solution type absorption refrigerator.
(1) Heat Storage Medium: Tap water

and ice
(3) Effective use of waste heat and co-
generation systems
(Yoshida, T., Sasao, H., 1993)
(4) Results of introduction: Laboratory
stage as of 1999.6
(Reply of Taikisha to questionnaire)
341
Dutch supermarket with ice slurry technology

by Rob Jans
The project Paradijsvogel Supermarkt was Fri-Jado’s fourth ice slurry (flo-ice®) project and
the third in Holland. The supermarket is located in Amsterdam and owned by Mr Slot, who is
a franchisee under the C1000 banner (Schuitema). The installation has been in operation since
autumn 2000.
It was a market introduction project, because Fri-Jado offers a mature, fully developed
technique, proven energy savings up to 20% and patented techniques with regard to
generation and application.
The objectives of the projects were as follows:

• demonstrating technical fulfilment;
• reduction of annual direct emission of synthetic refrigerant: from 9.25 kg (comparable
conventional installation) to 1.2 kg (ice slurry system);
• energy saving of at least 16% compared with a conventional installation.
In the supermarket with a sales area of 1250 m2 three different circuits were used:
• a circuit for cold generation HFC-404A is used;
• an ice slurry distribution circuit for cooling (ethanol/water ice slurry at –5°C);
• a separate conventional circuit for freezing.
Ice slurry is generated in a scraped-surface ice slurry generator and transferred to a cold buffer
where the ice slurry is stored. In the daytime the generation of the cold buffer continues. The
buffer gradually decreases because of a larger cold demand than the installation generates in
the daytime.
Technical details:
Object-load (25°C/60%) kW 98.0

Compressor capacity (–12°C/+43°C) kW 80.1
Refrigerant HFC-404A
Liquid supply Gravity
Cold-storage tank m3 6
kWh 154
Ice scrapers n 5
kWh 65
Liquid flow through objects m3 11
Maximum ice concentration % 30
Low-temperature-system Cascade Refrigerant
Object load (25°C/60%) kW 11.4
342
The most important results of the project are:

• Mechanical problems concerning the scrapers and adjustment problems, which arose in
previous projects, have been solved definitively;
• One malfunction has occurred, which originated with a supplier. This experience is taken
into account with future deliveries of the parts concerned;
• Energy savings add up to 17.6%; the savings on the energy bill, which will be calculated
at the end of 2000, were even higher due to a lower night rate;
• In accordance with a theoretical approach a reduction of 86.2% in direct emissions is
effected compared with a conventional installation. A practical approach (registration of
added refrigerant) results in a leak percentage of 0%.
Consequently, the project was technically successful and the objectives set were realized.
Therefore, the project offers an outstanding reference to demonstrate the good functioning of
the ice slurry technology used.
However, the implementation of ice slurry installations with scraper technology is not
acceptable from a commercial point of view. Therefore Fri-Jado will start a feasibility study
concerning ice slurry with vacuum technology, in which the expensive scrapers are replaced
by a vacuum system. The resulting price reduction is a focus of the study.
Finally, there are good possibilities to introduce natural refrigerants in Dutch supermarkets,
because the use of refrigerants is limited to the engine room.
Literature cited in Appendix 2
1. Hashiguchi, M.; Kamakura, K.; Okazaki, T.; Watanabe, Y.; Yamashita, K.; Takayanagi,
M.: Development of an Ice Storage System Using Direct-contact Heat Transfer between
Water and Water-insoluble Antifreeze, Proc. of Annual Conf. of the Soc. of Heating, Air-
conditioning and Sanitary Engineers of Japan: pp. 505-508, 1995.
2. Kakutani, S.; Osa, N.: Development of High Density Heat Transportation System, Jour.
of the JSRAE, 74-856, p. 28-3, 1999.
3. Kurihara, T.; Kawashima, M.: Dynamic ice storage system using supercooled water,
Introductions of heat pump and thermal storage unit for commercial use, Fourth IIR
workshop on ice slurries, pp. 61-69, 2001.
4. Mito, D.; Mikami, Y.; Tanino, M.; Kozawa, Y.: A new ice-slurry generator by using
actively thermal-hydraulic controlling both supercooling and releasing water, Fifth IIR
Workshop on Ice Slurries, Stockholm, May 2002.
5. Nagato, H.: A dynamic ice storage system with a closed ice making device using
supercooled water, Fourth IIR workshop on ice slurries, pp. 97-103, 2001.
6. Takenaka Corporation, Catalogue of “Full Storage”, 2000.
7. Tokyo Electric Power Company, Heating and Cooling for Air-conditioning by Heat-
pump, 54, 1996.
8. Watanabe, Y.; Shingu, H.; Goto, K.; Yoshino, H.; Tamaya, S.; Nakatsu, Y.:
Characteristics of melt-ability of Huge Ice Storage System for Gas-turbine inlet air
cooling, Proc. of Annual Conf. of the Heat transfer Soc. of Japan, pp. 715-716, 1997.
9. Yoshida, T.; Sasao, H.: Development of Absorption Ice-maker for Ice-slurry, Part1;
Experiment of Ice-maker, Proc. of Annual Conf. of the Soc. of Heating, Air-conditioning
and Sanitary Engineers of Japan, pp.401-404, 1993.
343
Appendix 3: List of symbols

List of symbols Chapter 2
Symbol: Unit [Rat. Inch Unit [SI]

Pounds]
a Parameter, coefficient - -
B Coefficient - -
C Coefficient - -
c Volumetric fraction - -
cp Specific heat capacity Btu lb-1 °F-1 J kg-1 K-1
C Volume concentration - -
C Electrical conductivity µS/inch µS/cm
CD Coefficient of drag - -
d Distance ft or microns m or microns
D Diameter ft or microns m or microns
E Coefficient - -
f Frequency s-1 s-1
f Function - -
F Force lb ft s-2 kg m s-2
g Function - -
g Gravity ft s-2 m s-2
G Gibbs free energy Btu/lb J/kg
h Enthalpy Btu/lb J/kg
H Height of liquid column ft m
H Enthalpy Btu J
He Hedström number - -
k Boltzmann constant Btu lb-1 F-1 J kg-1 K-1
k Spring constant lb s-2 kg s-2
K Ripening constant ft³ s-1 m³ s-1
l length ft m
m, M Mass lb kg
p Pressure psi kPa
Pr Prandtl number - -
q Power Btu/h W
Q Refrigeration capacity TR W
r Resistivity Ω inch Ω cm
r, R Radius ft or microns m or microns
Re Reynolds number - -
R2 Coefficient of determination - -
s Entropy Btu lb-1 F-1 J kg-1 K-1
t Time s s
T Temperature °F °C
U Uncertainty of measurement - -
v Velocity ft s-1 m s-1
V Volume ft3 m3
x Concentration (weight fraction) - -
x Distance ft m
345
Greek symbols:
α Heat transfer coefficient Btu h-1 ft-2 °F-1 W m-2 K-1
δ Infinitesimal difference - -
∆ Difference - -
λ Thermal conductivity Btu h-1 ft-1 °F-1 W m-1 K-1
γ Specific surface free energy Btu lb-1 ft-2 J kg-1 m-2
µ Chemical potential Btu lb-1 J kg-1
ν Kinematic viscosity ft² s-1 m² s-1
ω Angular velocity s-1 s-1
Ω Atomic volume ft³ m³
ρ Density lb ft-3 kg m-3
τ Time s s
Θ Wetting angle - -
ξ Temperature compensation for % K-1 % K-1
electrical conductivity
χ Specific ice slurry property - -
Subscripts:
0 Overall
1 Different states
2 Different states
add Additive
c Coriolis
C Catalyst
cf Carrier fluid
end Final state
fl Fluid
ice Ice
hw Hot water
ls Latent melting
L Liquid
mix Mixing heat
n Nucleus
R Resonance
slurry Ice Slurry
S solid
V volume
w Water
Superscript:
* Diluted
346
List of symbols Chapter 2.5 (Tsvetkov)
c – volume fraction of ice

k – thermal conductivity, Wm-1 K-1
a, A, B – coefficients
Greek symbols:
η – dynamic viscosity, Pa⋅s
ϑ – temperature, 0 C
α , β , δ – coefficients
Index
* – effective
2,s – ice
1 – liquid
List of symbols Chapter 3.1
C [-, %] concentration
D [m] pipe diameter
L [m] pipe lenght
p [Pa] pressure
v [ms-1 ] velocity
Greek symbols:
η [Pas] dynamic viscosity
τ [Pa] shear stress
τ0 [Pa] yield stress
Index
app apparent
B Bingham
C Casson
eff effective
K arbitrary constant
L liquid
max maximum
n power index
∞ limiting viscosity at ininite shear rate
0 limiting viscosity at zero shear rate
347
Standard Subscripts
A Surface m2 C Cross section

C Concentration m 3 m -3 /kg kg -1 Cf Carrier fluid
C Constant - e Ethanol
cp Specific heat J kg -1 K-1 h Hydraulic
D Diameter m in Inlet
f Friction factor - i Ice
h Enthalpy density J kg -1 is Ice slurry
h Heat transfer coefficient W m-2 K-1 ln Logarithmic
H Height m m Mean value
k Thermal conductivity W m-1 K-1 V Volumetric
L Length m w Wall, water
m Mass kg
m& Mass flow kg s -1
n Power law index -
m
Dimensionless numbers
P Wetted perimeter
q& Heat flux density W m-2 +
Χ Cameron number
q’’ Heat flux density W m-2
Gr Graetz number
r Radial coordinate m
He Hedström number
R Radius m
Nu Nusselt number
R Thermal resistance K/W
Pr Prandtl number
T Temperature K
Re Reynolds number
t Time s
u Velocity m s -1
U Velocity m s -1
U Overall heat transfer coef. W m-2 K
v Velocity m s -1
V Velocity m s -1
V Volume m3
x,z Downstream coordinate m
y Vertical coordinate m
Greek symbols
α Thermal diffusivity m2 s -1
α∗ Channel aspect ratio -
δ Thermal bound. lay. thick m
εh Thermal eddy diffusivity m2 s -1
φ Angle °
Φ Local ice fraction -
η Viscosity Pa s
Θ Correction factor -
ρ Density kg m-3
ϑ Τemperature °C
τ0 Critical shear stress Pa
ξ Mass concentration kg kg -1
348
Subscripts
–1 –1
c Specific heat capacity (kJ kg K )
D Diameter of ice particles (µm) e ethanol
x, X Mass fraction (kg kg–1 ) d dry matters
h Enthalpy per unit mass (kJ kg–1 ) i ice
k Thermal conductivity (W m–1 K–1 ) is ice slurry
L Latent heat of water freezing/thawing (kJ kg–1 ) l liquid solution
t, T Temperature (o C or K) oil fat matters (oil)
tf Initial freezing temperature (o C) p at constant pressure
s sugar
w water
Greek letters Superscripts
η Dynamic viscosity (mPa s) in initial

λ Thermal conductivity (W m–1 K–1 )
ρ Density (kg m–3 )
τ Time (s)
ϕ Volumetric fraction (m3 m–3 )
Acronyms
CFC ChloroFluoroCarbon HFC HydroFluoroCarbon

COP Coefficient of Performance HFM HydroFluidisation Method
HCFC HydroChloroFluoroCarbon IQF Individual Quick Freezing
349
APPENDIX 4: LIST OF FIGURES
CHAPTER 2
Figures 2.1 Critical nucleus radius in homogeneous nucleation
2.2 Geometrical and surface energy balances in heterogeneous nucleation
2.3 Nucleation of an ice embryo on a crystalline substrate with a lattice misfit of
approximately 10%. Dislocations (arrows) and elastic strain are introduced
into the nucleus [x, Flet]
2.4 Ice crystals formed heterogeneously in a shell-and-tube ice slurry generator
under forced flow conditions
2.5 a) Mean Feret diameter growth dependency on additive, ice concentration
and storage type shows clearly the Ostwald ripening effect
b) Particle Feret diameter growth rate dependency on additive, ice
concentration and storage type
2.6 Ice crystal morphology during a freezing/melting cycle, a), b), c), h), and i)
from EIVD-TiS laboratory, Switzerland; d) from DTI, Denmark, and e), f),
g) from Delft Univ., Netherlands (Pronk 2002)
2.7 Ice particle shapes estimated from projected images: a) a spheroidal ice
particle shape composed of two hemispheres connected by a cylinder of the
same radius (applied by Vuarnoz and Egolf [2001], b) disk-shaped ice
particle model proposed by Pronk (2002)
2.8 Ice particle size distribution data obtained for a 10 w/w% water-ethanol ice
slurry after correction with a view factor. Shown are the data for a)
Ci=10.3% and b) Ci=21.4% ice concentration (Vuarnoz, 2001 and Egolf,
2002)
2.9 Freezing point temperature (ϑF) as a function of additive concentration (CA)
2.10 Freezing point diagram (schematic)
2.11 Ice concentration as function of temperature for two types of aqueous
solutions
2.12 Schematic enthalpy phase diagram (NaCl-H2O)
2.13 Enthalpy phase diagram for sodium chloride-water
2.14 Enthalpy phase diagram for ethylene glycol-water
2.15 Enthalpy phase diagram for propylene glycol-water
2.16 Enthalpy phase diagram for ethyl alcohol-water
2.17 Enthalpy phase diagram for ammonia-water
2.18 Enthalpy phase diagram for potassium formate-water
2.19 Enthalpy phase diagram for calcium chloride-water
CHAPTER 3
Figures 3.1 Eight different series of measurements of the effective viscosity, performed
at CEMAGREF, DTI and UASCS with rotary and online viscometry. The
solid curve shows a model calculation presented by Christensen et al.
(1997) (see Frei and Egolf 2000)
3.2 Rheogram for typical fluids with and without a minimum yield stress (τ0)
and shear thickening (n > 1) or shear thinning behaviour (n < 1)
3.3 Casson parameters ηC and τC with a dependence on the viscosity of the
Carrier fluid (2 mPas < ηL < 10 mPas) and the ice particle concentration
(0-45%) (Doetsch, 2001)
351
3.4 Flow patterns of suspensions with S < 1 (see list of symbols)

3.5 Visualization of ice slurry flows only allows a qualitative determination of
flow patterns (10% water-ethanol mixture, dp ~ 0.1 mm, Di = 44.6 mm, case
A: C = 5%, v = 0.1 m/s, case B: C = 10%, v = 0.2 m/s). Work performed by
P. Reghem et al. at Pau University
3.6 Visualization of ice slurry flows does not allow a very accurate
determination of flow patterns (10 % water-ethanol mixture, dp = 0.2 ~ 0.4
mm, Di = 20 mm, rectangular pipe, case A: C = 7.6%, v = 0.54 m/s, case B:
C = 7.5%, v = 0.23 m/s, case C: C = 7.5%, v = 0.15 m/s). Work performed
by Didier Vuarnoz et al. at EIVD
3.7 Flow pattern diagrams from three different experiments performed at LTE -
Pau University and the Danish Technological Institute (DN 50, 10% water-
ethanol)
3.8 Flow pattern “diagrams” performed at the Korean Institute of Energy
Research (Di = 24 mm, 6.5% ethylene glycol-water, d p = 0.27 mm) (Dong
Won Lee et al., 2002)
3.9 Velocity profiles of ice slurry flows (7% ethanol-water mixture, Cr = ~20%,
DN 50, dp ~ 0.1 mm, dashed lines are approximate predictions (Kitanovski
et al., 2002)
3.10 Ice particle velocity profiles for ice slurry flow (11% ethanol-water mixture,
Cr = 12, 10, 11%, DN 25, dp ~ 0.1 mm) measured by Vuarnoz et al. (2000)
3.11 Concentration profiles of ice slurry flows in a horizontal pipe calculated
with eq. 3.40 (D = 27.2 mm, dp = 1 mm, 10% water-ethanol) (Kitanovski et
al., 2002)
3.12 Deposition velocities for ice slurry flow – theoretical model (10% ethanol-
water) (Kitanovski, 2002)
3.13 Deposition velocities obtained from an empirical function (eq. [3.43]).
(Kitanovski et al., 2002)
3.14 Flow pattern diagram for ice slurry flows obtained by applying a theoretical
model (DN 50, 10% water-ethanol) (Kitanovski et al., 2002)
3.15 Critical Reynolds number as a function of the Hedstrom number (Eq. 3.55 +
3.56)
3.16 Darcy´s friction factor dependence on the Reynolds and the Casson numbers
(Doetsch, 2001)
3.17 The specific pressure drop R in a heat exchanger decreases approximately
exponentially. In this example R changes from the inlet to the outlet by
more than a factor of ten! “Example” denotes a step-by-step downstream
calculation through the heat exchanger. “Model” denotes results obtained by
applying Eq. (3.76) (Egolf et al., 2001)
3.18 The specific pressure drop in a charge - and discharge - loop alters as a
function of time and evaporator temperature at the same velocity and ice
mass fraction (Frei and Egolf, 2000)
CHAPTER 4
Figures 4.1 Heat transfer from the wall to an ice-slurry in a vertical rectangular duct
with a second kind boundary at the wall
4.2 Local ice fraction distribution measurement at L/Dh≈19
4.3 Calorimeter (680 W) for on-line ice fraction measurements
4.4 Local temperature and velocity measurements (side view)
352
4.5 Velocity profile data for ΦV=0.0-6.0%, <uis>=0.080m/s, Recf,L~2200

4.6 Velocity profile data for Φv=6.0-9.0%, Recf,L~2100 & 4000
4.7 Typical ice slurry velocity profiles near a heated wall
4.8 Ice slurry temperature distributions for Recf=3800 and α*=H/W=1:12
(H=25.4 mm)
4.9 Ice slurry temperature distributions for Recf=6600 and α*=H/W=1:8 (H=12.7
mm)
4.10 Effect of Φv (6-11%) on the ice fraction distribution at Uo/<uis > ~ 1.8
4.11 Effect of Φv on the ice fraction distribution at Uo/<uis > ~ 1.3
4.12 Effect of Φv on ice fraction distribution at Recf ~ 3800 near a heated wall
4.13 Average Nusselt number for ice slurry flows through a pipe
4.14 Nusselt number data for an ice slurry flow heated with a heat flux of 15
kW/m2 and 40 kW/m2 (Knodel et al., 2000)
4.15 Calculated heat transfer coefficients obtained by using Equations (4.19) and
(4.20)
4.16 The local heat transfer coefficient calculated by applying Eq. (4.21) - (4.23)
4.17 Test section schematic and temperature measurement locations (Sari et al.,
2000)
4.18 Numerically calculated temperature profiles compared to measured profiles
4.19 The Nusselt number as a function of the Hedström number for laminar flow
4.20 The Nusselt number as a function of the Hedström number for turbulent
flow
4.21 Local heat transfer coefficients measured in a horizontal rectangular channel
of height 60 mm. Other parameters are given in the figures (from Kawanami
et al., 1998)
4.22 Flow patterns according to Kawanami et al. (1998) and some predictions of
the IPF (ice packing factor) for different channel heights: a: H = 60 mm, b:
H = 40 mm, c: H = 20 mm
4.23 Mean heat transfer coefficients calculated using Eqs. (4.33) and (4.34)
4.24 Experimentally and numerically determined temperature profiles for a heat
flux of 1800 W/m2 in a 1-meter long rectangular heat exchanger at a mass
flow rate of 1.2 kg/s. The measurement locations were: x = 0.25 m, 0.5 m,
0.75 m and 1 m. Trapezoidal symbols denote experimental data, while small
triangles and the fitted solid curves represent numerically calculated profiles
4.25 Temperature profiles for a heat flux of 7200 W/m2 in a 1-meter long
rectangular heat exchanger. The axial locations of the measurements are: x =
0.5 m and 1 m. Trapezoidal symbols denote experimental data and the solid
curves represent numerically calculated profiles
4.26 Ratios between the thermal eddy diffusivity and the thermal diffusivity at
different axial locations in a rectangular heat exchanger for different heat
fluxes. The data correspond to the results shown in Figures 4.24 and 4.25.
The dotted line was drawn in analogy to the solid line. It must be confirmed
by more measurements
4.27 Geometrical dimensions of a rectangular duct
4.28 Nucf,m as a function of Recf,m and Xs,m at Thw,m ~ 41.0oC and α*=1:8
(Stamatiou, 2003)
4.29 Nucf,m as a function of Recf,m and Xs,m at Thw,m ~ 14.8oC and α*=1:8
(Stamatiou, 2003)
4.30 Schematic of the flow model (Kawanami et al., 2001)
353
4.31 Local (above) and mean (below) heat transfer coefficients for the concave
and convex curved walls (Kawanami et al., 2001)
4.32 Melting behaviour of an ice slurry with an ice fraction of Ci=20 mass-% and
heat flux of q& =2.4 kW/m2 after: a) 5 min, b) 15 min, c) 25 min
4.33 Melting mechanisms as described and explained by Kawanami et al. (1999)
4.34 Melting behaviour and temperature distribution for Ci=20 mass%, a) q·t =
2.88x106 J/m2; d) q = 800 W/m2; b and e) 1600 W/m2, c and f) 2400 W/m2
(Kawanami et al., 1999)
4.35 Comparison of experimental results and proposed empirical correlations for
convective heat transfer to ice slurries (Stamatiou and Kawaji, 2004b)
4.36 An alternative heat transfer correlation for turbulent ice slurry flow
(Stamatiou and Kawaji, 2004b)
4.37 Comparison of local Nusselt number data with predictions for laminar non-
Newtonian heat transfer to ice slurries (Stamatiou and Kawaji, 2004b)
CHAPTER 5
Figures 5.1a Scraped-surface ice slurry generator (Courtesy of Sunwell)
5.1b Orbital rod ice slurry generator (Courtesy of Paul Muller Co.)
5.1c Scraped-surface ice slurry generator (Courtesy of Integral)
5.1d Scraped-surface ice slurry generator with helical screw – originally designed
for the production of flake ice (Courtesy of Ziegra)
5.2 Schematic of an ice slurry generation system
5.3 Double-walled tubular ice slurry generators with an envelope
5.4 Disk type scraped-surface ice slurry generator with an envelope
5.5 The internal circuits of a refrigerating disk
5.6 Photo of an 80-kW ice slurry generator with staggered disks
5.7 Staggered brushes
5.8 Immersed disk type scraped-surface ice slurry generator
5.9 Ice slurry generation with water as refrigerant (also known as vacuum ice)
5.10 Vacuum ice system with water vapour removal by freeze-out (Zakeri, 1996)
5.11 Schematic diagram of direct contact ice slurry generator (Wobst, 1999)
5.12 Ice slurry generator based on direct contact evaporation (Coldeco®)
5.13 Direct contact liquid coolant type ice slurry generator
5.14 Current and possible systems of supercooled brine method in Japan
5.15 Ice slurry production process using the supercooled water method in an ice
storage system
5.16 Actual numbers of equipment and installed refrigeration capacity of ice
storage systems using the supercooled method in Japan (System (A), system
(B) and system (C) of Figure 5.14)
5.17 Schematic of a hydro-scraped ice slurry generator
5.18 Representative scheme of the INP arrangement at the cell membrane of
INA-bacteria. Hydrophilic residues act as nucleating sites, matching an ice-
like structure. Neighbouring hydrophobic residues surround the nucleation
site, preventing ice sticking on the cell membrane
5.19 Aluminium oxide modified by applying the sol-gel technology. Red colour
(wavy line in the centre) represents hydrophilic residues and blue colour
(two outer wavy lines on either side of the red curve) represents
hydrophobic residues
5.20 Sketches of the ice formation process over different coating materials
354
5.21 Particle/Ice Removal Mechanism

5.22 Fluidized bed heat
5.23 High-pressure shift freezing of an agar gel cylinder from 210 MPa and
–21ºC
5.24 Refrigeration circuit for recuperative ice slurry generator A turbo-chiller
plate of about 10 kW capacity, 1.5 m wide, 1 m high made from two sheets
of 1 mm thick stainless steel plate welded together, with refrigerant
entry/exit at the top and at the bottom to the right
5.25 A turbo-chiller plate of about 10 kW capacity, 1.5 m wide, 1 m high made
from two sheets of 1 mm thick stainless steel plate welded together,with
refrigerant entry/exit at the top and at the bottom to the right
CHAPTER 6
Figures 6.1 Example of reduced total pipe length by using a single pipe system
6.2 Centralized warm liquid defrosting systems
6.3 Local warm liquid defrosting systems
6.4 Separated flow regimes. Left: measurement of separation velocities in a
horizontal 50 mm pipe based on visual inspection. Right: ice concentration
profiles (dashed lines) and velocity profiles (solid lines) for moving bed,
stratified and homogeneous flow (top to bottom). (Hansen and Noergaard).
(This reference is missing in the List of References)
6.5 T-junction with a vertical upward facing branch without flow, i.e. the valve
is closed
6.6 Vertical T-junction without flow in the left branch
6.7 Flow in parallel valve sections, experiment 14A. (Hansen, Radosevic,
Kauffeld, 2002)
6.8 Enthalpy of ice slurry depends on ice fraction and additive concentration
6.9 Transport capacity (kW of cooling at full melt off) in pipes (Drør : Inner pipe
diameter)
6.10 The experimental set-up at the Danish Technological Institute and a sketch
of a Grundfos CR pump
6.11 Performance curves of a Grundfos CR 2-50 pump working with ice slurry
6.12 Stage efficiency and power consumption curves of a Grundfos CR 2-50
pump working at 2800 rpm with ice slurry
6.13 Stage efficiency of a Grundfos CR5-11 pump working at 2800 rpm with ice
slurry
6.14 Design of centrifugal pump to be immersed in to the ice storage
6.15 Control loop circulation pump (combined UPE and a SP pump: UPESP1A-
3)
6.16 Screw pump
6.17 Pump characteristics for the screw pump in Fig. 6.16
6.18 Electrical
.
power input, P (denoted “E” in the figure) against volumetric flow
rate V (denoted “Q” in the figure) for the screw pump shown in Figure 6.16
6.19 Overall efficiency for the screw pump shown in Figure 6.16
6.20 Side channel pump
6.21 Pump total head for the side channel pump shown in Figure 6.20
6.22 Pump total head for the side channel pump shown in Figure 6.20
6.23 Overall efficiency of the side channel pump show in Figure 6.20
355
6.24 Measured performance curves compared to manufacturer’s data based on

Jeffrey’s viscosity model (Grundfos pump CR 2-50 at 2800 rpm)
6.25 Measured performance curves compared to manufacturer’s data based on
Jeffrey’s viscosity model (30 wt-% ice, Grundfos pump CR 2-50 with
variable speed). Cross bars indicate manufacturer’s tolerance
6.26 Relative performance curves of the tested centrifugal pumps at 2800 rpm
6.27 Comparison of measured relative performance of the CR5 centrifugal pump
(2800 rpm) with and without ice particles
6.28 Density of the ice slurry during the 24-hour test, 20% ice = 969 kg/m3
.
6.29 24 hour performance test of the CR5-11(Grundfos) at 0.33 x V 0 with 20 wt-
% ice
6.30 Typical locations of the ice slurry clustering, photo taken from inlet side
6.31 Flow rate over time of the pump shown in Figure 6.30 with 10 and 30 wt-%
ice fraction
6.32 Test section used in the experiments. The dots show the pressure
measurement locations
6.33 Pressure taps (Dimensions in millimetres)
6.34 Typical axial pressure profile along a piping system containing a fitting
6.35 Experimental results on pressure loss measurements in ABS-fittings
6.36 The left figure shows a sketch of the possible flow pattern with ice slurry
when the flow is highly turbulent, while the right sketch shows a possible
flow pattern when the flow is laminar or of Bingham type
6.37 Measured pressure loss coefficients for 90° bends with 10 and 30 wt-% ice
fraction in a water/glycol carrier fluid. The diagram shows a clear relation
between the loss coefficients and the Reynolds number. As the Reynolds
number is increased the value of the loss coefficient approaches the
tabulated value. Even the experiments without ice slightly show this trend
6.38 The loss coefficients plotted against the Reynolds number and the ice
concentration. This correlation has a R2 value of 0.96 compared to 0.84 for
the correlation shown in Figure 6.37. Therefore it is concluded that the
presence of solid particles in the fluid influences the loss coefficient
negatively by a factor of 2 to 3
CHAPTER 7
Figures 7.1 Overview of different categories of storage strategies (Hansen et al., 2002)
7.2 Examples of some typical storage layout principles (Hansen et al., 2002)
7.3 Ice storage process in a tank (seen from the side)
7.4 Ice melting process in a tank
7.5 Overview of ice storage model in a tank
7.6 Overview of the ice melting model in a water-spraying system
7.7 Comparison of prediction and measurement in an ice storage process
7.8 Rate of utilisation of the volume of the ice storage tank
7.9 Design conditions for ice storage
7.10 Comparison of experimental and predicted results for cold water exit
temperature from a flat water tank
356
CHAPTER 8
Figures 8.1 Illustration of a theoretical temperature difference between the two media
that exchange energy. All the ice is melted in the heat exchanger
8.2 A district cooling system where CBE´s are used as intermediate heat
exchangers between the central refrigeration unit and the buildings to be
cooled
8.3 Sketch of a laboratory test plant where a CBE was used with ice slurry as a
condenser in a cascade system with propane in the top cycle and carbon
dioxide in the bottom cycle. The installation is designed for a supermarket
application. The condensing pressure of the carbon dioxide was two bar
lower than with the original single-phase media
8.4 Dimensions of the three different heat exchangers tested
8.5 Sketch of test rig
8.6 Heat transfer coefficients on the ice slurry side obtained for all three CBE´s
and plotted against the Reynolds number based on average ice slurry
properties between the inlet and outlet conditions. The ice concentration
shown in the legend is the inlet condition – the outlet concentration is
always 0 wt-%. The channel number, “ch”, denotes channels with ice slurry
8.7 Heat transfer coefficients plotted against the heat flux and Reynolds
number. R2 = 0.97 for the grid shown
8.8 Accuracy of heat transfer correlation
8.9 Pressure difference plotted against the Reynolds number. The ice
concentration on the label shows the inlet condition – the outlet
concentration is always 0 wt-% - and “ch” is short for channels with ice
slurry
8.10 Accuracy of pressure drop correlation
8.11 Pressure drop data plotted against the heat exchanger capacity. The ice
concentration in the legend shows the inlet condition – the outlet
concentration is 0 wt-%. The comparison is made with equal temperature
differences on both primary and secondary sides and equal mass flow rates
on the secondary side
8.12 Pumping power (Ptransport) required to transport the ice slurry through the
CBE against the transferred heat capacity (Qperformance). The pump efficiency
coefficient used to calculate the transport effect is assumed to be constant at
0.1
8.13 The relation between the Reynolds number and ice concentration where the
flow begins to pulsate. If stable flow conditions are desired the point of
operation should be above the dotted line. The experiments were carried out
without any heat flux from the secondary side
8.14 Sketch of test rig. The ice slurry air coil is labelled 1
8.15 Heat transfer results for both coils. The ice concentration shown in the
legend is the inlet ice concentration. The outlet ice concentration varied
from 0-25 wt-%. The Reynolds number was calculated based on the average
ice slurry properties between the inlet and outlet conditions. Furthermore the
heat flux also varied between the experiments
8.16 Heat transfer results for both coils. The area refers to the inner surface area.
The Nusselt number is the average of the inlet and outlet conditions
8.17 Ice slurry side heat transfer results for both coils. The calculated values are
based on the correlation described in section 8.1.4 and average fluid
357
properties between the inlet and outlet. The air-side HTC is calculated from
a correlation supplied by the manufacturer
8.18 Pressure drop results. The pressure loss in the connections is at maximum 8
kPa and included in the values shown. The ice concentration shown in the
legend is the inlet ice concentration. The outlet ice concentration varied
from 0-25 wt-%. The Reynolds number was calculated based on the average
ice slurry properties between the inlet and outlet conditions. Furthermore the
heat flux varied between the experiments
8.19 Comparison of measured and calculated pressure drop
8.20 Comparison of internal heat transfer coefficients between a single-phase
fluid and ice slurry. The outlet ice concentration varied from 0-15 wt-%
8.21 Comparison of pressure drop between a singlephase fluid and ice slurry.The
outlet ice concentration varied from 0-15 wt-%
8.22 Illustration of appropriate/inappropriate header design
8.23 Schematic of temperature development through a CBE and further
downstream when superheating (SH) occurs
8.24 Location of thermocouples at the outlet of 6 parallel passes, just before the
return header. Thermocouple no. 3 was not used. There were 14 passes in all
8.25 Surface temperatures of six outlet tubes in 29 experiments.The ice slurry
temperature was measured with a Pt 100 probe in the liquid both at the inlet
and outlet. Those two measurements are labelled as “Ice On” and “Ice Off”
respectively. SH stands for superheat and is the only measurement that
refers to the right y-axis. The number at the top of the diagram identifies the
number of experiments. Hence the data points in the columns under each
number (1-29) correspond to one experiment
8.26 Super-heating of carrier fluid at the outlet. The parameter on the x-axis is
the melt off rate, and the ice concentration in the legend is the inlet
concentration. The numbers 3/8 and 12 identify the two different air coils
tested and refer to the diameter of the tubes
8.27 Superheating (SH) of the carrier fluid in all the CBE experiments plotted
against the average Reynolds number
8.28 The experimental results on time dependency in the C-type (B5) heat
exchanger (CBE) with constant mass flow rate and full melt off from an
inlet concentration of 20 wt-%. Heat transfer coefficient (HTC) refers to the
left y-axis labelled “h” and the superheat refers to the right y-axis labelled
“SH”
CHAPTER 9
Figures 9.1 Possible arrangements of a HFM-based freezing system combining the
advantages of both air fluidization and immersion food freezing techniques
(Fikiin and Fikiin, 1998, 1999a): (1) charging funnel; (2) sprinkling tubular
system; (3) refrigerating cylinder; (4) perforated screw; (5) double bottom;
(6) perforated grate for draining; (8) sprinkling device for glazing; (7 and 9)
netlike conveyor belt; (10 and 11) collector vats; (12) pump; (13 and 14)
rough and fine filters; (15) cooler of refrigerating medium; (16) refrigeration
plant
9.2 Experimental temperature histories during HFM freezing of some kinds of
(a) fish and (b) vegetables, when using sodium chloride solution (without
358
ice slurry) as a fluidizing agent (Fikiin, 1992; Fikiin and Fikiin, 1998,
1999a)
9.3 Different foods immersed in slurries with various ice concentrations: (a)
fruits; (b) vegetables; (c) chickens; (d), (e) and (f) fish (Fikiin et al., 2002)
9.4 Schematic diagram of an ice-slurry-based hydrofluidization system
HyFloFreeze® (Fikiin and Fikiin, 1998, 1999a)
9.5 HyFloFreeze® prototype: hydrofluidized bed of highly turbulent ice
slurry(Fikiin, 2003)
9.6 Fish chilling in ice slurry (Fikiin et al., 2002): (a) land-based systems and
(b) on-board systems (Liquid IceTM, Brontec, Iceland)
9.7 Enthalpy-temperature curve from differential scanning calorimetry of cod
fillets (Sørensen and Spange, 1999) and calculated thermophysical
properties
9.8 Variation in centre temperature (left graph) measured in an artificial fish
model (43 mm diameter) cooled in a container with ice slurry at –3°C and
with flake ice at 0°C, together with calculated temperatures (initially 20%
ice in the container). Temperature measurement (right graph) in horse
mackerel cooled in a container with ice slurry and with flake ice (initially
15% ice in the container)
9.9 Dimensionless apparent viscosity, η* = ηis / ηl, and thermal conductivity, k*
= kis / kl, versus ϕi. (Note the symbol used for thermal conductivity in the
figure follows the German symbol tradition, i.e. “λ” is used instead of “k”)
9.10 Hydrofluidization, conveyor and driving systems of the HyFloFreeze®
prototype (Technical University of Sofia and Interobmen Ltd.): (a) top
view, (b) overall view
CHAPTER 10
Figures 10.1 Some possible designs of an ice slurry system
10.2 Functional scheme of the control of terminal consumer units
10.3 Defrost system based on simultaneous defrosting of consumer units
10.4 Example of a local defrost set up
10.5 Thermal characteristics with main parameters of a water/ethanol mixture
10.6 Use of the diagram presented in Figure 10.5 (cf. text)
10.7 Some different possible strategies to control an ice slurry generator
10.8 Different operational modes of ice slurry systems
10.9 Ice concentration versus temperature and solute concentration for
ethanol/water mixtures
10.10 Absolute error in ice concentration measurements under industrial
conditions when working with water/ethanol ice slurries
10.11 Maximum error in the ice concentration determination versus the
temperature for different operating conditions
10.12 Operating range of an ethanol/water ice slurry system
10.13 Operating range for an ice slurry loop working with an initial 10%
ethanol/water mixture with a thermostat set point at –6°C ± 1 K
CHAPTER 12
Figures 12.1 CAPCOM building
12.2 A schematic diagram of the ice slurry cooling system
359
12.3 A schematic of the Herbis building thermal storage system

12.4 The “Techno-Mart 21” building (Information courtesy of P. Mueller Co.)
12.5 One of the eight ice-slurry units (1,759 kW) installed in the building
(Information courtesy of P. Mueller Co.)
12.6 The inside of the tank showing spray nozzles at the top of the tank
(Information courtesy of Paul Mueller Co.)
12.7 Ice slurry generator discharge into the tank (Information courtesy of Paul
Mueller Co.)
12.8 Ice slurry generators at the Stuart Siegel Centre, Virginia (Information
Courtesy of Paul Mueller Co.)
12.9 Ice slurry machine on top of the storage tanks at the left of the picture and
air cooled condensers at the right (Information Courtesy of Paul Mueller
Co.)
12.10 A schematic diagram of the ice slurry cooling system at Middlesex
University
12.11 An aerial view of the Kyoto Station Building
12.12 Layout of the Kyoto Station Building
12.13 Schematic diagram of the ice slurry plant at Kyoto Station Building
12.14 Cooling rate of ice slurry compared to flake ice (Paul, 2002)
12.15 The ice slurry system of the fish processing plant
12.16 A schematic of the cheese processing ice slurry plant
12.17 Load profile and storage tank capacity over a 24-hour period
12.18 Circuit schematic of the brewery plant
12.19 A schematic diagram of a supermarket ice slurry system
12.20 High density ice slurry being discharged into display cases (Photo courtesy
of Sunwell)
12.21 Charging the unit
12.22 Refrigerated body for delivery vans
360
APPENDIX 5: LIST OF TABLES
CHAPTER 2
Tables 2.1 Critical radius and number of molecules per embryo for homogeneous
nucleation of ice (Vali, 1995)
2.2 Parameters E, B and C
2.3 Effective particle diameters (Turian et al., 2002)
2.4 Coefficients in equation (2.40) for k * k 1
2.5 General characteristics of the listed additives
CHAPTER 3
Tables 3.1 Formulae for viscosities of Newtonian suspensions as a function of the
volume fraction C of the solid phase (Darby 1986)
3.2 Different models for laminar suspension flows (Darby, 1986; Hanks, 1986;
Steffe, 1992)
3.3 The maximum packing fraction of various arrangements of dispersed
spheres (Barnes, 1989)
3.4 Friction factors for Bingham, Power Law and Casson models. Laminar flow
is assumed (Darby, 1986)
3.5 Empirical correlations for the friction factor of ice slurry flows
3.6 Coefficients for empirical correlations for initial concentrations of the
additive ethanol (Kitanovski, 2002)
3.7 Coefficients for empirical correlations for two different additives (from
Table 3.5)
CHAPTER 4
Tables 4.1 Some models for the effective thermal conductivity of suspensions
4.2 Dimensionless numbers relevant to ice slurry fluid dynamics and
thermodynamics
4.3 Test parameters (after Jensen et al., 2000)
4.4 Parameters of the experiments performed by Snoek and Bellamy (1997)
CHAPTER 5
Tables 5.1 Typical specifications for ice slurry generators
5.2 Percentage of ice instantaneously produced in different adiabatic and
isentropic HPSF processes, obtained by mathematical modelling
5.3 Approach to the experimental case. Supercooling and percentage of ice
instantaneously obtained by a heat balance calculation
CHAPTER 6
Tables 6.1 Centrifugal pumps tested with ice slurry; the manufacturer’s (Grundfos)
product name/number is used
6.2 List of components
6.3 Pressure loss coefficients calculated with reference to the dynamic pressure
in the actual cross section including the length of the fitting. For the reducer,
361
the reference dynamic pressure is downstream of the fitting. For the

expansion, the reference dynamic pressure is upstream of the fitting. Danvak
(1992) and VDI (1991) refer to the respective Danish and German
engineering reference books
CHAPTER 7
Table 7.1 Ratio of effective heat transfer area
CHAPTER 8
Tables 8.1 Measuring equipment
8.2 Empirical values obtained to correlate heat transfer results
8.3 Empirical values obtained to correlate pressure drop results
8.4 Measuring equipment
8.5 Coil dimensions
8.6 Equivalent lengths of return bends. The values represent the equivalent
length of each bend
CHAPTER 9
Tables 9.1 Compositions and initial freezing temperatures of the studied solutions
9.2 Coefficients of equation (9.5)
CHAPTER 10
Table 10.1 Expected accuracy in measurements under different conditions
362
Michael Kauffeld
Prof. Dr.-Ing. Michael Kauffeld, born in 1962, studied mechanical engineering at the University of
Hannover, Germany and at the Catholic University of America, Washington, D.C. He holds a Master
degree and Doctorate in mechanical engineering from the University of Hannover. Since 1986, he has
worked with a variety of development projects within the field of refrigeration. First at the National
Institute of Standards and Technology (NIST), USA, where he investigated energy savings through the
use of zeotropic refrigerant mixtures in air conditioners and heat pumps. Then at the University of
Hannover, Germany, where he continued to work with refrigerant mixtures and in addition studied the use
of air as refrigerant. From 1992 to 1994, he was responsible for various heat exchanger development
projects at a Norwegian aluminium manufacturer. He joined the Danish Technological Institute (DTI) in
1994, where he worked as project manager. Most of his work at DTI has focused on natural working
fluids, such as air, water, ammonia, carbon dioxide, hydrocarbons and ice slurry. For 4 years, he acted as
leader of the Ice Slurry Center, which joined the forces of fifteen companies, a university institute and the Danish Technological
Institute in order to investigate the potentials of ice slurry in refrigeration systems. He has worked on the formulation of refrigerant
management plans for Uruguay, Jamaica and Syria. Since 1997, he serves on the UNEP Refrigeration, Air Conditioning and Heat
Pumps Technical Options Committee and served on the UNEP TEAP Task Force on Global Warming. He is president of the Ice
Slurry Working Group and vice president of commission B2 of the International Institute of Refrigeration (IIR). Michael Kauffeld
has written more than 100 scientific publications and filed several patents. Since 1 September 2002, he is professor for
thermodynamics, refrigeration and air conditioning at the Karlsruhe University of Applied Sciences.
Masahiro Kawaji
Masahiro Kawaji is a Professor in the Department of Chemical Engineering and Applied Chemistry at the
University of Toronto in Canada. He received B.A.Sc. from the University of Toronto in 1978, M.S.
(1979) and Ph.D. (1984) from the University of California, Berkeley. After working on nuclear reactor
thermal hydraulics research at the Japan Atomic Energy Research Institute for three years, he moved to
the University of Toronto as an Assistant Professor in 1986. He became a full Professor in 1993 and
served as an Associate Chair of his Department during 1995-97. In the past nineteen years at the
University of Toronto, Professor Kawaji has conducted both fundamental and applied research covering
various multi-phase flow and heat transfer problems including those related to HVAC and refrigeration,
ice-slurry production and utilization, nuclear and chemical reactors, heat exchangers, Kraft Recovery
boilers, mini and microchannels, and microgravity-related phenomena. His work has been published in
more than 170 refereed journal and conference papers. He is a fellow of the Chemical Institute of Canada
and received the Jules Stachiewics Medal for contributions to the heat transfer research. He is currently
serving as vice-president of the Ice Slurry Working Group of the International Institute of Refrigeration (IIR), and a member of the
scientific committees of the International Center for Heat and Mass Transfer and many international conferences. He is also serving as a
regional editor of International Journal of Transport Phenomena and on the editorial boards of International Journal of Multiphase Flow
and Journal of Process Mechanical Engineering.
Peter W. Egolf
Peter W. Egolf made an apprenticeship in the enterprise Gebrüder Sulzer AG in Aarau as a heating
designer. Then he studied at the University of Applied Sciences of Central Switzerland, where he
obtained his engineer’s diploma with a prize of the Swiss Association of Heating and Air-conditioning
Companies. After working in an industrial R&D laboratory at Hesco in Rüti he studied physics at the
Swiss Federal Institute of Technology (ETH Zurich). In 1984 he obtained his diploma contributing
with a work in Dynamical Meteorology. Then he entered a R&D division at Gebrüder Sulzer AG in
Winterthur, where he invented and investigated new (industrial) air conditioning systems. At an
advanced age he had the opportunity to make a PhD in a Noble Laureate family (Nernst–Mendelsohn
–Olsen–Rohrer) in low-temperature physics, studying nonlinear entropy and suface waves of super-
fluid He II. During this time he had a collaboration with I. Khalatnikov (Director of the Landau
Institute for Theoretical Physics in Moscow) on the Langrange-Hamilton description of the free
surface of the quantum fluid He II. During his PhD time, in 1988, he won a Research and Innovation
Exhibition Award of the Swiss Federal Institute of Technology and obtained his PhD at this institute
in 1990. After that a ten years employment at the Swiss Federal Institute for Materials Testing and Research followed in the field of
energy and buildings. Egolf became a specialist for phase change materials (PCM) research, developing a macroscopic model: The
Continuous-Properties Model for Melting and Freezing (1994) and obtained - together with Heinrich Manz – an award of building
technology for a translucent solar wall system for day lighting and thermal heat storage with PCM. Also in 1994 he invented the
Difference-Quotient Turbulence Model. In the same year he entered the field of research on ice slurries and he won in 1996, together
with Joseph Brühlmeier and Osmann Sari, the Swiss Technology Award. Simultaneously they were awarded a Special Prize of the
Swiss Bank Society for their approach to this environmental friendly technology. In 1998 Egolf created the Working Party on Ice
Slurries of the International Institute of Refrigeration (IIR) and was its First President. Since 2000 he is the head of the Theory and
Numerics Division (SIT) of the Thermics Institute at the University of Applied Sciences of Western Switzerland. Recently he
initiated a second new Working Party of the IIR, namely on Magnetic Refrigeration (at Room Temperature), and he is currently
serving as its President. He was a main organizer of eight scientific conferences. Furthermore, he is the regional French-Swiss
coordinator of the Swiss national competence network BRENET and a regional Editor of the International Journal of Refrigeration.
Egolf is a member of the Swiss Physical Society (SPG) and the Swiss Association of Cold (SVK). He is author of the IIR Technical
Note on Ice Slurry, a Promising Technology. At present - for the Edition Elsevier - he is writing a book with a mathematical
background and the title: Ice Slurries.

Ed. - Michael Kauffeld .... - Handbook On Ice Slurries - Fundamentals and Engineering PDF

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Ed. - Michael Kauffeld .... - Handbook On Ice Slurries - Fundamentals and Engineering PDF

Hochgeladen von

Copyright:

Verfügbare Formate

Handbook on

INTERNATIONAL INSTITUTE OF REFRIGERATION

Copyright and Disclaimer

© Copyright 2005 International Institute of Refrigeration

1. Summary of Ice Slurry Technology .................................................................... 11

2. Ice Creation, Thermophysical Properties of Ice Slurries

2.5.2 Viscosity ....................................................................................................... 39

3. Fluid Dynamics ......................................................................................................... 55

4. Thermodynamics and Heat Transfer ................................................................. 83

4.5.2 Pipes ............................................................................................................. 95

5. Ice Slurry Production ........................................................................................... 125

6. Transport .................................................................................................................. 167

6.4.7 Problems at standstill .................................................................................... 191

7. Storing/Melting and Mixing ............................................................................... 203

8. Melting Ice Slurry in Heat Exchangers ........................................................... 223

8.2.3 Correlation for heat transfer ....................................................................... 239

10. The Control of Ice Slurry Systems .................................................................... 273

11. Optimization of Ice Slurry systems ................................................................. 287

12. Present and Future Applications ...................................................................... 291

12.3 Food processing .................................................................................................... 302

13. Conclusions and recommendations for further R&D ............................... 317

Appendix 1: List of authors .................................................................................... 321

An indirect system can be improved if phase-change materials1 are applied instead of

This Handbook summarizes the current knowledge on ice slurry. Approximately 50

Karlsruhe, Toronto and Yverdon, Michael Kauffeld,

CHAPTER 1. SUMMARY OF ICE SLURRY TECHNOLOGY

1.2 Existing technologies

1.2.1 HETEROGENEOUS NUCLEATION

1.2.2 HOMOGENEOUS OR SPONTANEOUS NUCLEATION

1.3 Modern ice slurry applications

1.4 International Institute of Refrigeration (IIR) Working Party on Ice Slurries

1.4.1 SCIENTIFIC CHALLENGES

1.4.2 TECHNICAL CHALLENGES

1.4.3 OBJECTIVES OF THE WORKING PARTY

1. Listing and studying current scientific and technical problems.

3. Establishing and continuously updating a reference list of research publications

The working party currently consists of approximately 70 international scientists and

1.5 Research institutes

1.6 Conclusions and outlook on the use of ice slurry

1.7 Other phase-change slurries (PCSs)

1.7.1 DRY ICE/CARBON DIOXIDE SLURRY

1.7.2 CLATHRATE SLURRY

1.7.3 MICRO-EMULSION SLURRY

see e.g. (Hadjieva, 2003).

1.7.4 SHAPE-STABILIZED PCM SLURRY

1.7.5 MICROENCAPSULATED PCM SLURRY

Micro-encapsulation techniques were well developed, mainly by the pharmaceutical indu-

1.7.6 POLYETHYLENE PELLETS

Literature cited in Chapter 1

CHAPTER 2. ICE CREATION, THERMOPHYSICAL PROPERTIES OF ICE

2.1 The formation of ice

2.1.2 METASTABLE STATE – SUPERCOOLED WATER

From 2.4 and 2.5, the following expression is obtained:

The radius, r, at which the slope ∂(∆G)/∂r = 0 is

where r* corresponds to the above mentioned

Figure 2.1. Critical nucleus radius in homogeneous nucleation

Table 2.1. Critical radius and number of molecules per embryo

nL kT ⎛ ∆GA ⎞ ⎡ 16πγ iw3Tm 2 ⎤

In practice, the majority of nucleation phenomena in liquids take place at degrees of

Figure 2.2. Geometrical and surface energy balances in heterogeneous nucleation

The above treatment of heterogeneous nucleation is based entirely on simplified

Figure 2.3. Nucleation of an ice embryo on a crystalline substrate

A special case of heterogeneous nucleation is the secondary nucleation, where nucleation is