A Handbook of Elmentary Rheology PDF

A HANDBOOK OF
ELEMENTARY
RHEOLOGY
Howard A. Barnes
Institute of Non-Newtonian Fluid Mechanics

University of Wales
2000
A HANDBOOK OF
ELEMENTARY
RHEOLOGY
Howard A. Barnes
Principal Scientist, and Science Area Leader
in Rheology and Fluid Mechanics
Unilever Research Port Sunlight
Wirral England
University of Wales
Institute of Non-Newtonian
Fluid Mechanics, Aberystwyth, 2000
Handbook of Elementary Rheology
Published by The University of Wales Institute of Non-Newtonian Fluid

Mechanics, Department of Mathematics, University of Wales Aberystwyth,
Penglais, Aberystwyth, Dyfed, Wales, SY23 3BZ.
First published 2000
Howard A. Barnes, 2000
ISBN 0-9538032-0-1
Printed by Cambrian Printers, Llanbadarn Road, Aberystwyth SY23 3TN
ii
FOREWORD
One of the principal objects ... in any department of knowledge is to find the point
of view from which the subject appears in its greatest simplicity, J. Willard Gibbs
In an earlier bookAn Introduction to Rheologywe stated that rheology is a difficult
subject. However, even though we then tried to make it as simple as possible, a number of
people still told us that they found the book hard going. This present book is meant to satisfy
this group. For this reason, the mathematical content has been kept to a minimum, and the
text is as descriptive as possible: there are no integrals and only the occasional differentiation
symbol, with all the equations in the simple algebraic form of A = B.
The book is suitable as a first introduction to the subject of Rheology for any student
of science or engineering. However, having worked in industry for thirty years, I am also
trying to meet the needs of my industrial colleagues whooften without any previous
training in the subject of rheologyfind themselves having to work on the formulation and
processing of non-Newtonian liquids. The contents of this book are expansions of lectures
that I have given to both these groups of people over the years, in introductory rheology
courses run by rheometer companies, training organisations and universities around the
world.
Before leaving this foreword I must mention two rheologists and two rheology
books. Dr. George Scott Blair, the grand old man of British Rheology, was the first author to
produce a book called Elementary Rheology (Academic Press, London, 1969), written as he
said, on Rheology, but not intended for rheologists! The present author worked in his
laboratory for a few months during a summer vacation as an undergraduate in the early
1960s. His parting words on that occasion were I hope you catch the Rheology bug. The
bug was truly caught, and this book is intended to spread the infection even further!
Then, An Introduction to Rheology for me marks my gratitude to Prof. Ken Walters
FRS, one of my co-authors of that book. He introduced me to the world of theoretical
rheology as a postgraduate student at the University of Wales, Aberystwyth and thereby
gave me of a wide appreciation of most aspects of rheology. Over 5000 copies of An
Introduction ... have now been sold and this present book has been written, in effect, as an
introduction to An Introduction ... .
This book is offered in the spirit of Skovoroda, the 18th century Ukrainian
philosopher who said that We must be grateful to God that He created the world in such a way that
everything simple is true and everything complicated is untrue.
My special thanks go to members of the Institute of Non-Newtonian Fluid Mechanics
for reading through the manuscript and making useful suggestionsdioch yn fawr!
Port Sunlight, January 2000
iii
A BRIEF GUIDE TO THE HANDBOOK

The handbook begins with a short introduction (chapter 1) to the viscosity of
everyday liquids, and shows the way in which rheology interacts with in-use
situations which most of us come across in everyday life at home and work. Then a
few chapters follow on the explanation of the different kinds deformation (chapter 2);
the use of graphs in general (chapter 3), and rheology graphs in particular (chapter 4).
A consideration of the simplest kind of liquid that we will meet comes next the
Newtonian liquid (chapter 5), and we then show how we can calculate its flow
behaviour in lots of simple geometries (chapter 6).
Two chapters cover rheology measurement (chapters 7 & 8), thus equipping us
for non-Newtonian behaviour in liquids. A discussion of such liquidswhere the
viscosity varies with shear rate (or shear stress)then follows (chapters 9 & 10), with
a special section devoted to very shear-thinning liquids that appear to have a yield
stress (chapter 11). To complete this part, we discuss the flow, i.e. creep, of solids
(chapter 12) under long-term stress.
Viscoelastic behaviour is then covered (chapters 13 and 14), first linear
viscoelasticity with its manifestations in the time and frequency domain . Then nonlinear displays of viscoelasticity are introduced, these effects being usually
encountered in steady-state flow, where overt normal-force effects such as the
Weissenberg rod climbing and the extrudate-swell phenomena are seen.
The origins of viscous and viscoelastic behaviour are then investigated for
suspensions (chapter 15) and polymer systems (chapter 16), showing the important
effects of colloidal interactions. Extensional (or stretching) flow results in a very
different interaction between microstructural elements such as rods and coils and the
surrounding stretching flow field, and this results in very divergent behaviour
compared to shear flow (chapter 17).
Two applied areas are then covered, first the rle of rheology in the surfactant
systems (chapter 18) found in so many everyday products, and then we give a short
overview of the rheology of food systems (chapter 19).
For those interested in buying journals or books on the subject of rheology, an
extensive list is given (chapter 20). The last chapter (chapter 21) considers the growing
importance of computing in rheology, where one can now predict the flow of
viscoelastic liquids in complex geometries.
The book concludes with an Appendix which covers the area of on-line or
process viscometry, which although not of great interest to all readers, is
nevertheless very important to some.
Note:
Where possible useful Internet websites are referred to in the text, but readers are encouraged
to explore such possibilities for themselves, starting at
http://www.ncl.ac.uk/rheology/esr/ (lots of links!)
http://www.ncl.ac.uk/rheology/bsr/noframes/index.html
http://www.rheology.org
http://www.engr.wisc.edu/centers/rrc/what.html
http://www.staff.ncl.ac.uk/chris.petrie/links.html
http://www.rheology.co.uk
iv
Contents
CONTENTS
FOREWORD AND GUIDE iii
1. THE VISCOSITY OF LIQUIDS - SOME SIMPLE IDEAS
1.1 Introduction . 1
1.2 Rheology, packaging and product delivery . 2
1.3 Where Rheology is important . 2
2.
WHAT IS FLOW AND DEFORMATION?

2.1 Introduction ..
2.2 Shear rate and shear stress .
2.3 The range of shear rates .
2.4 Basic dimensions and units
2.5 Everyday Rheology units
3. HOW TO INTERPRETE GRAPHICAL DATA

3.1 Introduction ..
3.2 Logarithmic plotting
3.3 Different ways of plotting the data ..
3.4 Plotting start-up ..
3.5 Long-distance running
3.6 Other running regimes
3.7 Summary
5
6
6
7
9
11
11
12
13
14
14
15
4. DRAWING RHEOLOGICAL GRAPHS

4.1 Introduction .. 17
4.2 Figure legends . 17
5. THE NEWTONIAN LIQUID
5.1 Introduction ..
5.2 Viscosity
5.3 Viscosity of common Newtonian liquids .
5.4 The variation of Newtonian viscosity with temperature ..
5.5 The effect of pressure on viscosity .
5.6 The everyday perception of viscosity
5.7 The limit of Newtonian behaviour .
References .
6. SOME EQUATIONS FOR THE FLOW OF NEWTONIAN LIQUIDS
6.1 Introduction ..
6.2 Flow in rotational viscometers
6.2.1 Narrow-gap concentric cylinder geometry ..
6.2.2 Small-angle cone-and-plate geometry
6.3 Flow in straight pipe circular pipes
6.3.1 Velocity profile and pressure drop in fully developed
laminar flow in a tube ....
6.3.2 Flow from large reservoirs into small tubes .
6.3.3 Filling a tube with constant suction
6.3.4 Tube emptying under a constant pressure
6.3.5 Flow through a packed bed of spheres in a tube
6.4 Spheres falling in Newtonian liquids
6.4.1 An isolated falling sphere
6.4.2 A falling/rising cloud of spheres
6.4.3 A sphere rolling down an inclined tube
6.5 Other important flows .
6.5.1 Drainage down a wall ..
6.5.2 Flow along a slot
6.5.3 Flow through an annulus .
19
19
20
21
22
23
23
24
25
25
26
26
27
27
28
29
29
29
30
30
30
31
31
31
31
32
6.5.4 A rod being pushed into a cup filled with liquid .

6.5.5 Flow between squeezed circular plates .
6.5.6 One tube moving inside another with the space between filled with liquid
6.5.7 An angled plate moving across a trapped liquid .
6.5.8 A cylindrical indenter being pushed into a viscous liquid .
6.5.9 A disc rotating in a sea of liquid .
6.6 Interaction between viscosity and density in flows
6.6.1 Flow into the entrance of a tube .
6.6.2 Turbulent flow in tubes
6.6.3 The size of the liquid jet coming out of a tube .
6.6.4 Stokes flow with inertia
6.6.5 Taylor vortices in concentric-cylinder flow ..
6.7 Some flows where surface tension is important .
6.7.1 Free coating
6.7.2 Levelling of surface undulations
6.7.3 Flow up a capillary tube ..
6.7.4 A large bubble moving in a tube .
7. VISCOMETRY
7.1 Introduction ..
7.2 Some important things to note about using viscometers ..
7.2.1 Calibration .
7.2.2 Artefacts ..
7.2.3 Wall effects ..
7.2.4 Evaporation .
7.2.5 Sedimentation/separation
7.2.6 Chemical damage
7.2.7 Mechanical damage
7.2.8 Sampling .
7.2.9 High-speed testing .
7.2.10 Suspended particles .
7.3 Viscometer design
7.3.1 General considerations .
7.3.2 Commercial controlled-stress viscometers
7.3.3 Evolution of the specifications of controlled stress viscometers ..
7.4 Non-simple viscometer geometries ..
7.4.1 Wide-gap concentric cylinders
7.4.2 The rotating parallel-plate viscometer ..
7.3.6 Tube/pipe viscometer ..
32
32
33
33
34
34
34
35
35
36
36
36
38
38
39
39
39
41
42
42
42
42
43
44
44
45
45
45
46
46
47
48
48
49
50
50
8. YOUR RHEOLOGY LABORATORY

8.1 Whats in a good rheology laboratory? .
8.2 Which problem?
8.3 Which rheometer? .
8.3.1 Standard laboratory viscometers .
8.3.2 Rheometers .
8.3.3 Elongational viscometers .
51
51
52
52
52
52
9. SHEAR-THINNING LIQUIDS
9.1 Qualitative features of flow curves
9.2 Mathematical descriptions of flow curves
9.2.1 The Cross model ..
9.2.2 The Bingham equation
9.2.3 The power-law equation .
9.2.4 The Sisko model
9.3 Fitting models to data .
55
58
58
58
59
60
60
10 . EQUATIONS FOR THE FLOW OF NON-NEWTONIAN LIQUIDS

10.1 Introduction
10.2 Flow in a straight circular pipe
10.2.1 Power-law liquids ..
10.2.2 Sisko model .
63
63
63
64
vi
Contents
10.2.3 Casson model ...

10.2.4 Bingham fluid ..
10.3 Drainage down a wall
10.4 A sphere moving under gravity in a power-law liquid
10.5 Flow in a slit (a very wide slot)
10.6 Radial flow between parallel discs .
10.7 Flow out of a tank through a long exit pipe ..
10.8 Flow of a power-law liquid between squeezing plates ....
10.9 Flow through a converging nozzle ..
10.10 Levelling of surface undulations under the action of surface tension .
10.11 A disc rotating in a power-law liquid ..
10.12 More formulas for liquids with an apparent yield stress
10.12.1 Flow in a mixer ..
10.12.3 Capillary rise for a yield stress liquid
References .
64
64
65
65
66
66
66
67
67
68
68
69
69
69
70
70
11. VERY SHEAR-THINNING OR YIELD - STRESS FLUIDS

11.1 Introduction .
11.2 The history of the yield stress .
11.3 Values of yield stress ..
11.4 The arrival of commercial controlled-stress rheometers ..
11.5 The vane geometry .
11.6 Some flow equations with yield stresses .
References ..
71
72
73
74
74
75
76
12. THE FLOW OF SOLIDS

12.1 Introduction .
12.2 Non-linear viscosity of solids .
12.3 Conclusion
References ..
77
78
79
79
13. LINEAR VISCOELASTICITY AND TIME EFFECTS

13.1 Introduction
13.2 Mechanical analogues of viscoelastic behaviour ..
13.3 Measuring linear viscoelasticity .
13.3.1 General .
13.3.2 Creep tests
13.3.3 The behaviour of model materials in creep tests
13.3.4 Oscillatory tests or mechanical vibrational spectroscopy .
13.3.5 Oscillatory tests with a Maxwell model ..
13.3.6 Double or multiple Maxwell models in oscillation
13.3.7 Oscillation with a Voigt model ..
13.3.8 The oscillatory response of real systems .
13.3.8.1 Viscous/terminal region simple Maxwell behaviour .
13.3.8.2 Polymer systems
13.3.8.3 Dispersions .
13.4 What can we get from oscillatory data?
13.5 The relationship between oscillatory and steady-state viscoelastic parameters .
13.5.1 General considerations .
13.5.2 Creep versus oscillatory testing ..
13.5.3 A simple link between creep and oscillatory parameters
13.6 Non-linear response in oscillatory testing
13.7 Using oscillatory data to monitor curing, and rebuilding after shear ..
13.8 Stress-relaxation testing
13.9 Start-up experiments
References .
81
82
84
84
84
86
88
89
90
92
92
94
95
96
98
100
100
100
100
101
102
103
104
105
14. NON-LINEAR VISCOELASTICITY

14.1 Everyday elastic liquids 107
14.2 Some visible viscoelastic manifestations 108
14.3 The proper description of viscoelastic forces . 110
vii
14.4 The measurement of viscoelastic forces .. 111

14.5 Examples of first normal-stress differences 112
14.6 Some viscoelastic formulas ..
14.6.1 General .
14.6.2 Extrudate flow .
14.6.3 Flow in a slightly inclined trough .
14.6.4 Pressure in holes and slits ..
14.6.5 Inertia in a cone-and-plate geometry
14.7 Postscript flow of a viscoelastic liquid through a contraction ..
References .
115
115
115
116
116
116
117
117
15 . THE FLOW OF SUSPENSIONS

15.1 Introduction .
15.2 Viscosity of dispersions and emulsions .
15.2.1 The continuous phase .
15.2.2 The effect of very low concentration of suspended particles
15.2.2.1 The effect of the continuous phase .
15.2.2.2 The effect of the disperse phase .
15.2.2.3 The effect of medium-to-high concentrations of particles .
15.3 Particle-size effects in concentrated dispersions .
15.4 Particle-shape effects in concentrated dispersions
15.5 Particle deformability
15.6 Particle interactions an overview .
15.7 The viscosity of flocculated systems
15.8 Thixotropy
15.9 Shear thickening .
15.10 Apparent wall slip
15.11 Very high concentration pastes .
15.12 Colloidal control of suspension viscosity .
15.13 The stability of suspensions
References .
119
119
119
120
120
122
122
125
126
126
126
129
131
133
133
136
137
138
139
16. POLYMER RHEOLOGY

16.1 Introduction .
16.2 Some of the different kinds of polymer chains ..
16.2.1 Long rods ..
16.2.2 Coils/strings .
16.3 Polymer solutions ..
16.4 Highly branched polymers
16.5 Living polymers or rod- or worm-like micelles .
16.6 Associative polymers .
16.7 A polymer solution as a standard non-Newtonian liquid ..
References ..
141
141
142
142
144
145
146
147
148
149
17. EXTENSIONAL FLOW AND EXTENSIONAL VISCOSITY

17.1 Introduction .
17.2 Where do we find extensional flows? .
17.3 Whats so special about extensional viscosities?
17.4 The Trouton ratio ..
17.5 Some examples of extensional viscosity curves .
17.6 The effect of time or extension on extensional viscosity .
17.7 Other kinds of extensional flows .
17.8 Viscometers to measure extensional viscosity ..
17.9 Commercially available extensional viscometers .
17.10 Some interesting applications
References .
151
152
153
155
156
158
159
160
162
162
163
18. RHEOLOGY OF SURFACTANT SYSTEMS

18.1 Introduction 165
18.2 Surfactant phases 165
18.3 Rheology of surfactant systems 167
viii
Contents
18.4 Necessary rheological properties . 168

19. RHEOLOGY OF FOOD PRODUCTS
19.1 Introduction .
19.2 Liquid-like food rheology ..
19.3 In-vivo flow fields ..
19.4 Rheology and food processing .
171
172
172
172
20. YOUR RHEOLOGY LIBRARY

20.1 Introduction
20.2 Rheology books in print ..
20.3 General Rheology .
20.4 Theoretical Rheology
20.5 Polymeric systems
20.6 Blood
20.7 Food .
20.8 Other specific systems .
20.9 Drag reduction ..
20.10 Geological subjects .
20.11 Proceedings of meetings (not mentioned elsewhere)
175
175
175
177
178
179
179
180
180
181
181
21. RHEOLOGY AND COMPUTING

21.1 Introduction
21.2 A brief description of how CFD works ..
21.3 The rheological description of a non-Newtonian liquid and CFD codes .
21.4 When should CFD be used? .
21.5 Some typical results of CFD calculations
21.6 A word of caution ..
183
183
184
184
184
185
APPENDIX: ON-LINE (PROCESS) VISCOMETRY

A.1 Introduction .
A.2 Desirable features of a good on-line viscometer
A.3 Types of measurement made by current instruments ..
A.4 The effect of non-Newtonian viscosity
A.5 Hygiene and cleaning .
A.6 General comments on actual representative instruments
A.7 Detailed description of a particular instrument
A.8 Conclusions .
References .
187
187
189
190
191
191
192
192
193
INDEX 195
ANSWERS TO SOME EXERCISES . 200
ix
CHAPTER 1: THE VISCOSITY OF LIQUIDS

- SOME SIMPLE EVERYDAY IDEAS
Everything should be as simple as possible, but not simpler, Einstein
1.1 Introduction
No one is really an absolute beginner when it comes to rheology, because the
flow properties of liquids play such an important part in our daily liveseither
deliberately or accidentally. This importance ranges, for instance, from the viscosity
of the blood flowing around our bodies, through the thickness of the liquids that we
swallow, to the grade of the oil that we put into our car engines. As consumers we
are not only told that products are new and improved, but also that they are thick and
creamy, because such rheological properties are more pleasing to the eye, mouth and
hand. Providing for our unmet needs as potential customers by adjusting the
rheology of products is an important part of the product development activity.
We use many words with a rheological connotation, e.g. viscosity,
consistency and texture, and they can have both positive and negative implications.
Hence, while thick and creamy might impart a positive inference; gluey, gelatinous,
sticky, tacky, gooey, gummy, cohesive, could well convey the opposite idea with regard
to liquids being too thick.
Watery and runny might communicate the idea of dilution, since in a
consumers mind a liquid with a high viscosity usually conveys (or reinforces) the
idea of concentration and strength, since a similar product with a lower viscosity
will often pass on the idea of weakness. On the other hand, many products, if
perceived as too thick will, in the consumers mind, be anticipated to have problems
in dissolving, i.e. dispersing.
In the context of everyday products it is a simple and useful exercise to go
around your house and garden and note the following examples of structured
liquids with interesting rheologies:
kitchen
bathroom cabinet
garage
pantry
tool shed
laundry liquids, surface cleaning liquids

personal products
oils, greases
sauces, salad dressings
paints, glues, mastics
Exercise 1: List other liquids with interesting flow properties, and ask why have
they been made like this?
A Handbook of Elementary Rheology
Exercise 2: Obtain samples of similar products with low and high viscosities
respectively, and carry out a small consumer test to see if viscosity is
connected with perceived strength and concentration in the minds of
average consumers.
1.2 Rheology, packaging and product delivery
Having considered the consistency of a product, we now need to think about
its packaging and means of delivery. These probably range over the following
sequence of packaging, consistency and delivery action:
Package
pressure cans
bottles, cans
tubes
tubs
Consistency
thin and watery
oily
pasty
soft-solid like
Action
Spray
Squirt
Squeeze
Scoop
Example
paint, hair spray
washing-up liquid, oil
toothpaste, tomato paste
face cream, Swarfega3
Clearly anyone interested in these areas should be thinking very carefully about
rheology.
1.3 Where rheology is important
As well as adjusting rheology to suit consumer tastes, it is sometimes
necessary to control the rheology for specific technical reasons. Some examples that
are easily appreciated are:
Slowing things down - draining on vertical surfaces
- suspending particles
- thickened sauces
- paint, bleaches
- paint, abrasive cleaners
- mayonnaise.
We often need a high viscosity to resist flow at the low forces posed by gravity, but at
the same time one needs a low viscosity for the same product under the high forces
experienced in pouring, squirting or brushing.
This is where the nonNewtonianviscosity varying with applied force or flow rateproperties are very
important. Some other examples where this is true are:
Hair shampoos that need to be thick enough to pour onto the hands for
transferring to the hair without flowing away through the fingers. The same is
also true of skin creams and other personal products.
Toothpastes that need to be viscous enough to sit on the brush without draining
into the bristles, but at the same time are easily extruded from the tube.
(However, some modern tooth-cleaning products are sold in bottles, and have a
much lower viscosity.)
Thixotropic or solid paints that have very special properties built into them so
that they appear to be solid after storing, but on brushing out or stirring, they thin
down. They will then re-thicken on standing.
Printing inks that need to spread easily when squeezed onto the type or transfer
surface, but once deposited/printed onto a surface, in order to maintain a sharp
edge, they must not spread by the action of surface tension or gravity.
2
Chapter 1: The viscosity of liquids
Ink in biros and roller-ball pens that must have a low viscosity as it is being
dispensed through the high-shear-rate roller-ball mechanism. However, when the
ink is sitting on the writing surface, its viscosity must be high to stop it spreading
far.
Concentrated liquids that often have suspended solid particles, either as crystals
above their solubility limit, or that have been deliberately added such as abrasive
particles: these have to be suspended for long periods of time. The product has
also to be pourable.
Cement pastes that have to be as concentrated as possible, but flowablethis can
be achieved by vibrating.
Exercise 3: think of other areas where the property of high viscosity at low stress
and low viscosity at high stress is important.
Epilogue:
Please note that the words fluid and liquid are often used interchangeably
(as sometimes in this book). However, while the word fluid covers both
gases and liquids, gases are not liquids. So all liquids are fluids, but not all
fluids are liquidsthink about it! And while youre thinking about it, dont
forget solids!
CHAPTER 2: WHAT IS FLOW AND DEFORMATION?

Insufficient facts always invite danger, Captain, Mr Spock
2.1 Introduction
Rheology has been properly defined as the study of the flow and deformation
of materials, with special emphasis being usually placed on the former. However,
we might ask the simple question what is flow?. If we carry water carefully in a
bucket, it is certainly moving, but it is not flowing, however if we pour out the water
it is flowing. What is the difference? In flow, elements of the liquid are deforming,
and adjacent points in the liquid are moving relative to one another.
There are two basic kinds of flow with relative movement of adjacent
particles of liquid; they are called shear and extensional flows. In shear flows liquid
elements flow over or past each other, while in extensional flow, adjacent elements
flow towards or away from each other, see figure 1 for illustrations of shear and
extensional deformation and flow respectively.
SHEAR FLOW
Adjacent particles moving

over or past each other
original shape
EXTENSIONAL FLOW
Adjacent particles moving away

from or towards each other
Figure 1: Particle motion in shear and extensional flows.
Exercise 1: When a liquid is being rubbed, poured or stirred, imagine that you are an
element of the liquid and consider whether you are experiencing shear
or extensional flow or both.
All flows are resisted by viscosity, so if we poured out a bucket of water, it
would flow much quicker than if we had poured out a bucket of motor oil. Liquids
are made to flow by imparting a velocityas in stirring with a spoonor else by
applying a force, as with squeezing a bottle of sauce. For a given velocity, the
5
resulting force increases when the viscosity is increased, whereas for a given force, the
velocity is reduced when the viscosity is increased.
Exercise 2: Try to think of a number of situations where flow is initiated by
imparting a velocity or else by applying a force.
2.2 Shear rate and shear stress
As we have seen in figure 1, simple shear flow is the continual movement of
particles of liquid over or past each other, while extensional (or elongational, or
stretching) flows are where particles of liquid flow towards or away from each other
(see below). In figure 2 we visualise shear flow alternatively as the movement of
hypothetical layers sliding over each other. In the simplest case the velocity of each
layer increases linearly with respect to its neighbour below, so that layers twice the
distance from any stationary edge move at double the speed, etc. The gradient of the
velocity in the direction at right angles to the flow is called the shear rate (sometimes
called the velocity gradient or strain rate), and the force per unit area creating or
produced by the flow is called the shear stress. In our simple example, the shear
rate is V/h and is described by the symbol & (pronounced gamma dot), while the
shear stress is given by F/A and is given the symbol (sigma).
hypothetical layers
Figure 2: Hypothetical layers in shear flow.

The shear rate has the units of velocity divided by distance, i.e. metres per
second / metres, leaving us with the units of reciprocal (i.e. one over) seconds, or s-1.
Shear stressforce per unit areahas the units of newtons per square metre, N m-2,
but in the SI system, stress, like pressure, it is given in units of pascals, Pa, in honour
of the famous French scientist. For reference, standard atmospheric pressure,
previously known to many as 14.7 pounds per square inch (psi), is 105 Pa.
2.3 The range of shear rates
The approximate value of the shear rate encountered in a wide variety of
circumstances found in everyday life or in industrial situations is shown in table 1.
Readers may relate these numbers to their own field of interest by simply dividing a
typical velocity in any flow of interest by a typical dimension. An example of this is
the average velocity of a liquid flowing in a pipe divided by the pipe radius, or the
velocity of a moving sphere divided by its radius.
6
Chapter 2: What is flow and deformation?
Situation
Sedimentation of fine
powders in liquids
Shear Rate
Range / s-1
10-6 - 10-3
Examples
Medicines, paints, salad
dressing
Levelling due to surface

tension
10-2 - 10-1
Paints, printing inks
Draining off surfaces

under gravity
10-1 - 101
Toilet bleaches, paints, coatings
Extruders
100 - 102
Chewing and swallowing
101 - 102
Polymers, foods
soft solids
Foods
Dip coating
101 - 102
Paints, confectionery
Mixing and stirring
101 - 103
Liquids manufacturing
Pipe flow
100 - 103
Pumping liquids, blood flow
Brushing
103 - 104
Painting
Rubbing
104 - 105
Skin creams, lotions
High-speed coating
104 - 106
Paper manufacture
Spraying
105 - 106
Atomisation, spray drying
Lubrication
103 - 107
Bearings, engines
Table 1: Typical shear rate ranges of various physical operations.

2.4 Basic dimensions and units
It is worth spending a little time on this subject, since it arises so often in
Rheology and is an area in which it is very easy to become confused, so that many
mistakes have been made by the unwary. We shall be following the SI (Systme
International d'Units) convention, which, for our purposes is built around the five basic
units of mass, length, time, temperature and amount of substance, as shown in table 2,
from which all others units of rheological interest are derived. Table 3 gives a number of
such units, which have special names, their symbols and their full equivalents.
The decade prefixesas shown in table 4should be used before units in order
to bring them into the sensible range of 0.1 to 1000 as a numerical value used with any
particular unit, e.g. 2 kN, but not 2000 N, and 10 Pa.s, but not 0.01 kPa.s. Note that all
prefixes representing 106 or greater are capitalised (e.g. M for 106), but all below 106 are
in lower case, e.g. m for 10-3. There should be no space between the prefix and the unit,
and combined prefixes should not be used, e.g. mmPa should be Pa. When units
have to be written out in full, any prefix should also be written out in full, beginning
with a lower-case letter, e.g. megahertz not Megahertz or Mhertz and millipascal
seconds not mpascal seconds.
7
Quantity
Unit Name
Unit Symbol
kilogram
kg
Length
metre
Time
second
Temperature
kelvin
Amount of substance
mole
mol
Mass
Table 2: SI base units used in Rheology

Quantity
Special Name
Symbol
Equivalent
Frequency
hertz
Hz
s-1
Force
newton
kg m/s2
Pressure or stress
pascal
Pa
N/m2
Work, energy, heat
joule
Power
watt
N m, kg m2 s-2
J/s
Table 3: Some derived SI units with special names

Factor
Prefix
Symbol
106
mega
103
kilo
102
hecto
101
deka
da
10-1
deci
10-2
centi
10-3
milli
Table 4: Usual SI decade prefixes

used in Rheology
10-6
micro
The convention followed in this book is that all physical quantities are in italic
type, while symbols for units are in normal type, so for instance we might say that ?the
force F is 15 N; the viscosity is 10 Pa.s and the shear rate & is 0.1 s-1. Symbols for unit
names derived from proper names begin with a capital letter but the unit names
themselves are not capitalised, i.e. N for newton, P for pascal and Hz for hertz. The unit
name should not be followed by a full stop, and plurals of unit symbols should not be
followed by an s, e.g. 3 kg but not 3 kg. or 3 kgs. If we are quoting temperature T in
absolute units, i.e. kelvin, then no symbol is needed, e.g. 200 K. However if you are
quoting the temperature in Celsius or similar, you should put C.
Chapter 2: What is flow and deformation?
Inserting a raised dot properly indicates multiplication of units, but a full stop
(period, US)as used hereis usually acceptable because some people only have access
to a limited character set on their keyboards. However leaving a space between the
units is also acceptable, so we can happily write either N.m, N.m or N m, but not Nm!
Here we shall keep to the full-stop/period convention (as in Pa.s) which is well
understood and easily typed, but when writing for particular companies or scientific
journals, you should check their preferred house-style. Division is indicated by the use
of a forward slash / or a negative exponent, but repeated use of a forward slash is not
allowed, so for acceleration we write m/s2 or m.s-2 but not m/s/s.
The numerical value and the unit symbol should be separated by a space, even
when used as an adjective in normal writing, e.g. 35 mm, not 35mm or even 35-mm. A
zero should always be placed in front of the decimal point for numbers less than one,
e.g. 0.3 J but not .3 J. The SI unit for volume, the litre, is best represented by L and not
the lower case l which is easily confused with the numeral 1 in most character fonts.
Exercise 3: Check out http://physics.nist.gov/Document/checklist.pdf on the Internet for
the US governments official view on the rules and style conventions for SI
units. Standard rheological nomenclature, as approved by the (American)
Society of Rheology may be found at www.umecheme.maine.edu.
2.5 Everyday Rheology units
Units that you will come across most frequently in Rheology are shown in table 5.
Quantity
Symbol
Units
Shear
(pronounced gamma)
Shear rate
& (pronounced gamma dot)
s-1
Shear stress
Pa
(pronounced sigma)
Shear viscosity
Pa.s
(pronounced eta, i.e. eater)
Table 5: Some commonly used rheological quantities and their units.
Note that the units of viscosity are Pa.s and not Pasecs!
Note that although it is good practice to write shear viscosity to distinguish it from
extensional viscosity, it is usually referred to simply as viscosity.
The Greek symbol (gamma) represents deformation in simple shear, this is
illustrated in figure 3, where shear stress (sigma) is also defined.
F, V
= F/ A
=/h
.
= V/ h
Figure 3: Definition diagram for shear flow.
Similarly, for uniaxial extension the proper measure of deformation is (epsilon) and
stress (sometimes called tension) e as shown in figure 4.
e = F / A
=/h
.
=V/ h
F, V
Figure 4: Definition diagram for uniaxial extensional flow.
Using the same diagrams we can also define deformation rate, where now the
deformation (delta) is continuous and V = d/dt, and & = dV/dh, and is called the
shear rate. The dot above the deformation symbols is used to signify differentiation
with respect to time, and follows Newtons calculus convention (Newtons dot), so
that is an amount of shear, and & is the rate of shear, or shear rate
dt (usually
referred to as d by dt), which is sometimes called the rate of strain or strain rate.
For Newtonian liquids in simple shear flows we can write
= & .
For uniaxial extensional or stretching flows, the equivalent symbols are e
& and e and then for Newtonian liquids,

e = e& .
Exercise 4: Convert a flow rate of 100 US gallons per hour into SI units (assume 1 US
gallon = 3.7854 x 10-3 m3).
10
CHAPTER 3: HOW TO INTERPRETE GRAPHICAL DATA

Few things are harder to put up with than the annoyance of a good example, Mark Twain
3.1 Introduction
One of the most important considerations in rheology is making sense of data
collected from viscometers or rheometers, especially when that data is to be plotted
visually, as in a graph. (Here we will follow the modern practicesee the latest Oxford
dictionariesand take the word data as singular, in the sense of a particular collection
of numbers.) In order to master some of the principles involved in constructing, using
and interpreting graphs, we will look at some interesting non-rheological data, viz. the
world records for men and womens running races (the data was collected from various
Internet sites). We will use the world records ranging from the 50 m indoor event,
through the various outdoor sprint distances, middle distances, the marathon, right
through to very long distances.
3.2 Logarithmic Plotting
Once we have compiled the particular data-set of interest, the first thing we do is
to plot the record times T as a function of race distance D, as shown in figure 1, but
noting the first lesson that we must reduce all the data to the same units, in our case
(mega-)metres and (mega-)seconds, and not a mixture of seconds, minutes and hours,
etc.
Time, T / Ms
10
8
6
4
2
0
0
Distance, D / Mm
Figure 1: Men's world records, time versus distance.
Once we have produced this plot we immediately see a substantial problem: if we want
to display all the data which covers a very wide range of distances50 to over 7,000,000
mon a simple linear plot, the points for distances less than 1,000,000 m are compressed
into the corner of the graph, and we cannot even see them! One answer is to plot the
results on a logarithmic basis, see figure 2, where we see that the spacing between 100 m,
1000 m and 10,000 m etc., is now the same on the distance axis. The facility of plotting
logarithmic-based graphs is available on most modern graph-plotting software
11
packages, or they can even be generated manually on specially printed logarithmic

graph paper.
7
Time, T / s
10
pe
t slo
uni
10
10
10
10
10
10
10
10
10
Distance, D / m
Figure 2: Men's world records, time versus distance,
logarithmic plot.
3.3 Different ways of plotting the same data
Average speed, V / m/s
Next we will show that the basic distance/time data can be manipulated and
plotted in different ways. For instance if we now divide the distance by the time for
each record, we get the average speed, V, in metres per second. For physiological
reasons this is now plotted as a function of timeagain on a logarithmic basisto see
how the body produces power (measured by velocity) as a function of time, see figure 3.
10
marathon
5
3
2
1
10
10
10
10
10
10
10
Time, T / s
Figure 3: Men's world records, average speed versus
time, logarithmic plot.
Notice that there is far more structure in the speed/time graph than its innocuous
time/distance equivalent and quite different regimes can now be identified. For
instance, we notice that for short distances, the average speed seems to rise from the 50
m event to a maximum at around 150 m and then drop off. The reason for this is quite
simple: because the runner starts from rest, he/she needs to accelerate up to maximum
speed, then, after a certain time, he/she begins to tire and hence slows down, see figure
4.
How can this data be best represented and what further information can we
extract from it? If we now plot the distance/time data over the range 50 to 400 m on a
linear basisas shown in figure 4three things are obvious. First, the data we have
plotted appears to extrapolate to a non-zero intercept on the time axis when the distance
is zero. We can call this time the acceleration time, To, (the more mathematically
12
Chapter 3: How to draw graphs
adventurous might like to work out what this means in terms of real features such as
force being mass times acceleration and resistance being proportional to speed).
50
2
T = T0 + D/Vmax + FD ...
Time, T / s
40
30
20
e=
lop
V ma x
32
12.
m/s
10
0
T0 = 1.67 s
0
100
200
300
400
500
Distance, D / m
Figure 4: Men's world records, time versus distance,
for short distances.
This acceleration time does contain a real time, that is the reaction time from the gun
being fired to start the timing to the runners legs receiving the nerve-impulse message
to commence running, i.e. a fraction of a second.
3.4 Plotting start-up
The fastest speed shown on the graph is the initial slope at zero on the time axis,
i.e.
dD
, which we shall call Vmax, and in this case it is nearly 14 m/s. Then the
dT T 0
curve slopes upwards off the tangential line dictated by this maximum speed, indicating
that running speeds are progressively reduced as fatigue sets in. We can now attempt
to describe this curve in the simplest possible mathematical way. We will try the
function T = To + D/Vmax + FD2, where To and Vmax are, as described above, the
acceleration time and maximum velocity, and F we shall call the fatigue factor, which
arises from the run-down of energy and the build-up of lactic acid in the body. This
equation can be fitted to the data using a regression package available on most
spreadsheet applications, e.g. Microsoft Excel. From such a fit we find that Vmax is 13.9
m/s and the acceleration time To is 1.8 s. We can subtract the acceleration phase and
reconstruct the average speed data to illustrate the effect of the fatigue factor alone.
This is shown in figure 5 where we have subtracted To from all the time data.
(T - T0 )/D / s/m
0.13
0.11
men
women
0.09
1/Vmax
0.07
0
100
200
300
400
500
Distance, D / m
Figure 5: The inverse of the corrected speed for men's
and women's world records for short distances,
versus distance.
13
This would be the result if the runner could start the race at his maximum speed, or
another way of combining these two numbers is to say that if the runner had started a
distance Vmax..To (i.e. about 25 m) before the line, then he would have been up to
maximum speed by the starting line (neglecting the very small amount of fatigue during
that time).
3.5 Long distance running
If we now consider the longer-distance races, we see an important change in
slope at around the marathon distance of approximately two hours, see figure 3. It so
happens that the end of a marathon race occurs near a critical point in terms of the
bodys store of energy, because after about two hours the glycogen stored in the blood
runs out and the body starts to burn fat. When this happens two things are important;
first, when his supply of glycogen runs out, the runner hits the wall, and second the
rate of conversion of fat to energy is thereafter slower, hence the greater slackening in
speed as a function of time above two hours, shown as a distinct change in the slope of
the curve shown in figure 3.
3.6 Other running regimes
Last we shall show how to plot data from parts of the original log graph that
seemed to show different regimes. If we take the middle-distance data from 1000 to
10,000 m, we see that this shows a logarithmic straight-line relationship, see figure 6.
2,000
Time, T / s
1,000
it
un
slo
pe
500
300
200
T = 0.078D
1.078
100
1,000
2,000
5,000
10,000
Distance, D / m
Figure 6: Men's world records, time versus distance for
middle distance, logarithmic plot.
This implies a power-law relationship of the form T = aDn , which in this particular case
gives T = 0.078D1.078. Again this kind of mathematical fit can be obtained from most
spread-sheet software packages using regression analysis of the data, this one was
obtained using Lotus Freelance Graphics. Remember that this relationship only works
within the distance range we have chosen.
If we choose another range and combination of data, say average speed as a
function of time for very long time events, we see another kind of logarithmic plot, see
figure 7. This time we can describe the relationship between average speed and time by
the expression V = cTm, but now we note that m is negative because the speed is a
decreasing function of time. The relationship is in fact given by V = 61.36T -0.265. If we
try to extrapolate this relationship outside its range, then large errors are introduced. For
instance, if we attempt to use the relationship to predict the speed down at T = 1 s, the
14
Chapter 3: How to draw graphs
result is very easy to calculate, because if T = 1, then 1 raised to any power is still 1, so
then V is predicted to be 61.36 m/s. In fact V is no more than 14 m/s!
7
5
V = 61.36T -0.265
3
2
1
4
10
10
10
Time, T / s
Figure 7: Men's world records, average speed versus

time for long distances, logarithmic plot.
The equivalent data for womens running records tell the same story, but now
we see that women are slower by about 13% (~ L) over almost all distances/times, see
figure 8 for some examples of this fact. In the plots of various ranges of distance and
time, the power-law indicesm and ndescribing men and womens performance are
virtually identical.
7
5
men
3
2
women
V = 54.14 T
1
4
10
-0.265
10
Time, T / s
10
Figure 8: Women's world records, average speed versus

time for long distances, logarithmic plot.
3.7 Summary
The lessons we have learned aboveeach of which has its own rheological
counterpartcan be summarised as follows
widely spread data is better plotted logarithmically,
the same data set can be plotted in a number of ways,
e.g. D vs. T ; T vs. D ; D/T vs. D ; D/T vs. T,
limited data can be extrapolated to zero distance to give an apparent intercept
on the time axis,
simple mathematical expressions can be fitted to the data, either linear,
power law or more complicated, and
mathematical fits must not be extrapolated outside the range of data
collection.
15
These points will be constantly referred to in the following chapters where

graphical representation of rheological behaviour is discussed at length.
Exercise: find some weight-lifting data on the Internet, and plot graphs of records
as a function of weight, using the ideas given above.
16
CHAPTER 4: DRAWING RHEOLOGICAL GRAPHS

First, get the facts, then you can distort them at your leisure, Mark Twain
4.1 Introduction
In the same way as we have described in Chapter 3 for data on running,
graphs of rheological quantities are also best plotted on logarithmic axes so that the
wide range of shear rates usually used and viscosities obtained are represented in the
optimum manner. In this case the axis divisions are in decades, and the better way
of representing these large numbers is to use decade exponents, i.e. 106, (the normal
decade number can be used if your plotting package only allows this form, but for
very high viscosities, these can be quite long, e.g. 1,000,000) since other ways of
representing this can be clumsy and confusing, e.g. E+6.
The axis labels can also be confusing and in order to overcome this the
standard format should show (in the following order)
the name of the quantity being plotted, e.g. Viscosity (note capitalisation),
its symbol, e.g. (remember, physical quantities are always in italics), and
its units, e.g. Pa.s, but shown as a division, so that the quantities plotted on
the
graph are then pure numbers.
Thus, a complete axis label would look like
Viscosity, / Pa.s
Shear stress, / Pa.
If the units being plotted are large, then the unit size should be increased, so for
instance, we might put
Tensile stress, e / MPa,
then the numbers plotted are in megapascals, the original quantity having been
divided by 106. Sometimes graphs are plotted to show qualitative relationships
only, and in that case the units should be omitted, hence
Viscosity, ,
Time, t
Strain, .
4.2 Figure legends

In addition to labelling the axes carefully, equal care should be taken in
composing the text for the figure legends. For instance, Figure 1 is not enough to
identify a graph, with the details of what is being plotted being described only in the
text. Figure legends should therefore be complete in themselves, so that a graph tell
the whole story, without the reader needing to refer back and forth to the place in the
17
text where the graph is mentioned. Therefore, put into the legend (or perhaps within
the graph itself), all the necessary information describing the variables and the
system being represented.
Do not put too many curves on one graph, remember a graph is meant to
enlighten and not to confuse! Use grids where necessary, but only where you want to
draw attention to local particular values of points. Also, if your are plotting coloured
graphs, beware of colours such as yellow are sometimes difficult to see, and do not
copy well.
yield
stress, 0
slope is plastic viscosity, p
10
Shear rate, / s1
Deformation,
steady- state
(iii) viscous
response
Time, t
Figure 3: Hypothetical creep curve for a Burger model.
-1
10
. -1
Shear rate, / s
10
Figure 2: Viscosity versus shear rate for a dilute polymer solution,

logarithmic scale.
Viscosity, / Pa.s
Figure 1: Hypothetical flow curve obeying the Bingham equation.
10
delayed
(ii) elastic
response
immediate
(i) elastic
response
Viscosity, / Pa.s
Shear stress,
10
10
10
10
10
10
10
200
300
500
Shear stress, / Pa
1,000
2,000
Figure 4: Viscosity versus shear stress for a very shear-thinning

suspension; logarithmic plot.
Anticipating the contents of future chapters, figures 1 - 4 illustrate the points

made here for some typical real and hypothetical rheological data.
Exercise 1: write down the graph axes for stress as a function of strain, and viscosity
as a function of time.
18
CHAPTER 5: THE NEWTONIAN LIQUID

The beginning of wisdom is to call things by their right names, A Chinese proverb
5.1 Introduction
A Newtonian liquid is one for which the viscosityalthough varying with
temperature and pressuredoes not vary with deformation rate or time; nor does
such a liquid display any elastic properties or extensional anomaliesthese
deviations will be explained later. A Newtonian liquid is of course an idealisation,
but in many cases it is a very good representation of a large number of liquids under
normal everyday conditions, however see section 7 below. We shall now examine
the ideas of viscosity in general and then the Newtonian liquid in particular.
5.2 Viscosity
Long before viscosity was recognised as a quantifiable material property of a
liquid, there was obviously universal recognition that liquids could be viscous (from
the Latin viscum, the mistletoe, which exudes a gelatinous juice when squeezed).
Later, scientists knew qualitatively that viscosity was that property of a liquid which
was a measure of its resistance to flow, so that high viscosity liquids flowed slowly
and low viscosity liquids flowed quickly.
If we were able to explain to Isaac Newton (1642 - 1727) what we mean by
viscosity, he would well understand, but he would call it defectus lubricitatislack of
lubricity, or the internal friction of a liquid. Neither would he recognise the idea of
it having numbers with particular units associated with viscosity, because that came
a lot later. In fact it was Wiedemann (1856) who proposed Zahigkeits-Koeffizient, or
coefficient of viscosity as an absolute measure of the property. Then Hagenbach
(1860) measured the value for water at 10 C as 0.13351 in units of grams and square
metres; this compares very well with the present internationally accepted value of
0.1307 in the same units. In 1890 Couette introduced the concentric-cylinder
viscometer and found that the viscosity of water so measured agreed with
Poiseuilles value evaluated in tube flow. The first special unit for viscosity was the
Poise, signified by P, in honour of Poiseuille (1797 - 1869), the French physician who
had been interested in blood flow, and who, around 1840, had conducted extensive
work on flow in tubes. The everyday unit from this system was the centipoise, being
a hundredth of this value, i.e. the famous cP, with water being (very conveniently!)
nearly 1 cP at room temperature.
Viscosity measurements began to be used by colloid chemists when Thomas
Graham investigated the peptisation of silicic acid in the 1860s. Soon, with the
introduction of the Ostwald U-tube viscometer, viscometry became one of the most
widely used methods of investigating colloidal solutionsand to a large extent it still
is used for this purpose.
To be precise, we now say that shear viscosity is the ratio of shear stress to
shear rate, i.e. / & so the units of viscosity are pascals divided by reciprocal seconds,
19
A Handbbok of Elementary Rheology
giving us pascal seconds, or Pa.s. As this unit is only about 0.001 Pa.s for water at
room temperature, the prefix m is necessary, hence 1 mPa.s. The pascal-second unit
of viscosity is 10 times bigger than the older cgs equivalent, the Poise, but
numerically, 1 mPa.s 1 cP, which is precisely the viscosity of pure water at 20.2 C.
Usually the limiting values of the viscosity of non-Newtonian liquids at very
low and very high shear rate respectively (the so-called zero-shear and infinite-shear
viscosities) are symbolised by o and . It is also worth noting that the following
viscosity-related quantities are often quoted (with the appropriate symbols and
units)
the viscosity of the solvent or continuous phase, s , Pa.s;
dimensionless relative viscosity, r , which is the ratio of the viscosity of a
suspension to its continuous phase viscosity, /s ;
the dimensionless specific viscosity, sp , given by r 1 ;
the intrinsic viscosity, [] (with dimensions m3.kg-1) which is the specific
viscosity divided by the concentration c of a dispersed phase, (r - 1)/c ;
the kinematic viscosity, = / which in pre-SI days had the dimensions of
centi-Stokes, cS, for thin liquids, with water at 20 C ~ 1 cS.
Note: (i) in the engineering literature, the viscosity is often represented by the
Greek letter mu, , and
(ii) the word non-Newtonian is often spelt incorrectlythe prefix has a lower
case n and the word is hyphenated.
5.3 Viscosity of common Newtonian liquids
Many liquids display Newtonian behaviour under a wide range of shear
rates, and many more show this behaviour within limited ranges. If we limit
ourselves to everyday shear rates, then the liquids in table 1 show Newtonian
behaviour and encompass a large range of viscosities, best shownlike the race
distances of chapter 3on a logarithmic basis.
Liquid or gas
Approx. viscosity in Pa.s
Hydrogen
10-5
Air
2 x 10-5
Petrol
3 x 10-4
Water
10-3
Lubricating oil
10-1
0
Glycerol
10 (i.e.1)
Corn syrup
103
Bitumen
109
Table 1: Approximate viscosities of some common
Newtonian fluids.
Exercise: estimate the approximate viscosities of the following Newtonian liquids
- wine, baby oil, liquid honey.
It should be noted that even the international standard for the viscosity of
water (1.002 at 20 C) is only quoted to an accuracy of 0.25%. Commercially
20
Chapter 5: The Newtonian Liquid
available standard-viscosity liquids of the highest quality vary in the uncertainty to

which they quote the viscosity: this ranges from (a very optimistic) 0.2% for 5 mPa.s,
through 0.28% for 0.5 mPa.s to 1% for 500 Pa.s (ISO 9002). Cheaper variants will only
quote viscosities to an accuracy of a few percent. Everyday measurements of
viscosity cannot with much confidence be quoted as better than 5%.
5.4 The variation of Newtonian viscosity with temperature
The viscosity of all simple liquids decreases with increase in temperature
because of the increasing Brownian motion of their constituent molecules, and
generally the higher the viscosity, the greater is the rate of decrease. For instance,
while the viscosity of water reduces by about 3% per degree Celsius at room
temperature; motor oils decrease by about 5% per degree, while bitumens decrease
by 15% or more per degree. Many attempts have been made to describe the
viscosity/temperature dependence mathematically, so that by 1951 Partington had
listed nearly 50 [1]. The most widely-used expression is that due to Andrade [2]
log10 = A + B/T
where T is the temperature in degrees absolute (C + 273.15), i.e. K. Note that this
equation is also known by other names [3], for instance the Arrhenius law where B is
replaced by E/R where E is an activation energy for viscous flow and R is the
universal gas constant. The activation energy E is said to be a measure of the height of
a potential energy barrier associated with the force needed to produce elemental
quantum steps in the movement of molecules. In terms of what we said abovethat
the viscosity decreases more with increase in temperature if the viscosity is higher (see [4]
figure 2.1) we see that in the Andrade equation this implies that B increases with
increase in A, and this is shown for instance in figure 1 for a collection of 132 singlephase, organic liquids [5].
3,000
Value of B / K
2,500
2,000
1,500
1,000
500
0
7
8
Value of -A
10
11
Figure 1: The values of A and B in the Andrade viscosity/

temperature equation for a range of organic liquids.
As we see, a typical liquid has A ~ - 4.8 and B ~ 500 K; this means a viscosity of
around 0.8 mPa.s at room temperature (20 C) has a temperature variation of
viscosity of - 1.34% per degree Celsius. A liquid with a higher viscosity might have
say A = -10 and B = 3000 K, which would give a room-temperature viscosity of
around 1.7 Pa.s, and a viscosity-temperature sensitivity of - 7.84%/C. On the other
hand, a lower-viscosity liquid fitting this equation might have A = - 3.8 and B = 100
21
K; this would give a viscosity of 0.35 mPa.s (similar to petroleum) and a temperature
sensitivity of only around - 0.27%/C. In a severe winter diesel oil is said to freeze,
thus causing severe problems for transportation. In fact, all that happens is that the
viscosity increases to levels where the pumps cannot operate, thus stopping the flow.
Water over the range 0 - 100 C fits a slightly modified form of the Andrade
law, viz. log10 = A + B/(C - T), with A = - 4.5318, B = - 220.57 K and C = 149.39 K,
and at room temperature water changes viscosity at a rate of - 2.54% per degree
Celsius, [5]. This means that to perform reasonably reproducible experiments on
water-based liquids, we should be careful to control the experiment to within 1 C.
5.5 The effect of pressure on viscosity
The effect of pressure on viscosity is such that as the pressure goes up, the
viscosity increases; however the amount is quite small for the situations normally
encountered in everyday life. A useful rule-of-thumb is to remember that for most
single-phase, organic liquids,
(a) on average the viscosity approximately doubles when the pressure is
increased from atmospheric pressure to 100 MPa (1000 bar or atmospheres) and
(b) the increase is nearly linear with pressure up to that level [6], but the
dependence is greater than linear above that level, so that at around 200 MPa the
viscosity might increase tenfold, and at 400 MPa might be 100-fold etc., with the
precise values depending on the particular liquid being considered.
One important exception to the insignificance of pressure on the viscosity of
everyday materials is the very large increase in pressure and hence viscosity in the
contacts between intermeshing gears in car engines, etc. The pressures in this
situation are enormous during the very short times at the points of contact, and the
viscosity of oil then becomes as high as that of bitumen at normal pressure. To be
precise, at 10,000 bar, the viscosity of an oil could easily be 107 times greater than its
original value! It should also be remembered that in injection moulding of plastics,
the pressure can rise to 1000 bar. Then the viscosity of the melt can increase by a
factor of 10 - 100 depending on the molecular weight of the polymer.
Normalised viscosity, n
1.5
0 oC
1.0
10 oC
30 oC
0.5
0
2000
4000
6000
Pressure, P / bar
Figure 2: The viscosity of water normalised to the value at 0 oC
and atmospheric pressure as a function of pressure.
Water is one major anomaly with respect to the normal viscosity/pressure

rule, in that its viscosityat least from 0 C to around 30 Cfirst decreases with
increase in pressure, and only increases in viscosity at very high pressures, i.e. well
above 100 MPa (1000 bar or atmospheres). Figure 2 shows these points graphically:
this kind of dependence is also true for most water-based liquids.
22
Chapter 5: The Newtonian Liquid
5.6 The everyday perception of Newtonian viscosity

How does an average person judge viscosity without a viscometer? Although
an expert can be trained to judge the viscosity of a Newtonian liquid quite accurately
and thereafter give a reasonable estimate at its quantitative value, the same is not
true for an untrained person. The normal perception of the level of viscosity of
Newtonian liquids by such an untrained person is described by Stevens general
power-law of psycho-physical cognition [7], which for viscosity states that the
estimates are approximately proportional to the square root of the real numerical
value. Hence, if the viscosity of a 10 mPa.s Newtonian liquid was assigned unit
value, the perceived viscosity of a 1000 mPa.s liquid when compared with the lower
viscosity liquid would be seen as roughly 10 times higher, and similarly, a 106 Pa.s
liquid would be estimated as about 100 times more viscous. This should always be
borne in mind by those who are trying to produce a product with a particular
viscosity. If the potential consumer of the liquid product cannot differentiate to a
precise degree, then extensive efforts to adjust the viscosity to an exact value are
wasted.
5.7 The limit of Newtonian behaviour
At a high-enough shear rate, all liquids become non-Newtonian, but the
liquids we have been considering so far only do so at very high values of shear rate.
For instance, the values of the critical shear rates for glycerol and mineral oils are
above 105 s-1. The viscosity of a set of typical silicone oils is shown as a function of
shear rate in figure 3: these oils, which are often used as Newtonian standards,
become non-Newtonian at lower and lower values of shear rate as their molecular
weightand hence general viscosity levelis increased.
3
she
ar
str
~2
10
es s
00
0P
a
Viscosity, / Pa.s
10
10
10
-2
-1
10
10
10
Shear rate, / s
10
10
-1
Figure 3: Flow curves for a series of silicone oils.

Note the onset of non-Newtonian behaviour
at a shear stress of around 2000 Pa.
It has even been estimated that pentane becomes non-Newtonian above shear
rates of 5 x 106 s-1, while water would have to be sheared at an impossible 1012 s-1 in
order to see any non-Newtonian behaviour!
23
Exercise: Ask a number of untrained people to estimate how much thicker the
following liquids are than water (the approximate value of the real difference is
shown in brackets),
a motor oil, SAE 10W
( ~ 100)
washing-up liquid
( ~ 300)
glycerol
( ~ 1,500)
liquid honey
( ~ 10,000)
References
[1] Partington, J.E., An Advanced Treatise on Physical Chemistry, Vol. 2: The properties of
liquids, Longmans, Green & Co., London, 1951.
[2] E.N. da C Andrade, Phil. Mag. 17, 497 & 698, 1934.
[3] Scott Blair, G. W., A Survey of General and Applied Rheology, Pitman & Sons,
London, 2nd edition, 1949, p. 33
[4]Barnes, H A; Hutton, J F; Walters, K, An Introduction to Rheology, Elsevier,
Amsterdam 1989.
[5] D.S. Viswanath; Natarajin, G., "Data Book on the Viscosity of Liquids, Hemisphere,
New York, 1989.
[6] Hatschek, E., The Viscosity of Liquids, G. Bell and Sons, London, 1928, pp. 84
86.
[7] Stevens, S.S; Guirao, M, Science, 144, 1157 - 8, 1964.
24
CHAPTER 6: SOME EQUATIONS FOR THE FLOW OF

NEWTONIAN LIQUIDS
Look at it this way Steveevery equation will halve your sales!, Stephen Hawkings
publisher
6.1 Introduction
In this chapter we will look at a number of situations where, if we know the
(Newtonian) viscosity of a liquid, we are able to predict something about its flow in
certain simple geometries. First we will look at flows where the viscosity alone is
important, and then we will see what happens when we have to take the liquids
density into account, so that secondary flows and eventually turbulence are seen in
high-speed flows. Lastly, we will consider some flow situations where surface
tension as well as viscosity is important. Learning to use these equations is a very
useful precursor to looking at non-Newtonian equations. They can also be used to
obtain an order of magnitude estimate of many complicated flow situations.
Important: One of the greatest difficulties for beginners is to realise that consistent
units must be inserted into equations, and here we adhere to the SI
system, see Chapter 2, where all parameters are reduced to a seconds,
meters and kilograms basis.
6.2 Flow in rotational viscometers
The simplest possible flows are those where the shear rate is more or less the
same everywhere in the flowing liquid. These flows are in fact the ideal geometries
for viscometers for non-Newtonian liquids, where we want to measure the shear
viscosity at a given shear rate or shear stress.
cone and
plate
concentric
cylinders
cone
cone-cylinder
Figure 1: Viscometer geometries where the shear

rate is approximately the same everywhere.
25
Two of this class of flows will now be considered. The first is circular flow
between closely-fitting, narrow-gap concentric cylinders, and the second is circular
flow in a small-angle, cone-and-plate geometry, see figure 1. Also shown is the conecylinder or Mooney geometry where the shear rates in the concentric-cylinder and
the cone-and-plate parts of the geometry are arranged to be the same.
6.2.1 Narrow-gap concentric-cylinder geometry
If both cylinders have a large radius, and the gap between them is small, then
the shear rate is the almost the same everywhere in the liquid-filled gap. If the inner
cylinder radius is a1, and an outer cylinder radius is a2, then if the inner cylinder is
stationary and the outer is rotated, the shear rate & in the contained liquid is given
by
& =
a 2
a 2 a1
where is the rotation rate of the outer cylinder in radians per second, ( is also
equal to 2f, where f is the rotation rate in hertz, i.e. revolutions per second).
The shear stress is given by the force per unit area on the inner cylinder. If
the torque on the inner cylinder is T (otherwise known as the couple or moment,
with units of N.m), the turning force F on the inner cylinder is given by T/a1. Since
the surface area over which the force F acts is 2a1, where H is the cylinder height,
the shear stress is given by
T
.
2a12 H
If we insert this geometry into a suitable instrument, we have a viscometer!

We can then either apply a torque T to the inner cylinder and measure the rotation
rate (the controlled-stress mode), or else apply a rotation rate to the outer cylinder
and measure the resulting torque T on the inner cylinder (the controlled-strain mode).
This geometry forms the basis of an important viscometer geometry, but the simple
formula only works for small gaps, where a1 is only a few per cent different from a2.
A suitable formula for wider gaps will be considered later, because many liquids
with large suspended particles cannot be placed in very small gaps.
6.2.2 Small-angle cone-and-plate geometry
The second simple geometry we will consider is flow in a rotating cone-andplate (or cone-and-cone) geometry. The velocity at any point on the rotating surface
(either the cone or the plate) is given by r where r is the distance from the centre of
rotation and again is the rotation rate. If the cone angle is small (less than 4), then
the distance between the cone and plate at this point is given by r, where is the
angle between the cone and plate in radians (remember 2 radians = 360). Thus the
shear rate is now given by r/r or simply /. Since this quantity is the same
everywhere in the liquid, it follows that the shear rate is also the same everywhere.
In the same way as described above, we can now write down the shear stress as
3T/2a3, where a is the radius of the plate. Thus we have our second geometry to
place into our suitable apparatus, and we can apply rotation and measure torque or
26
Chapter 6: Some equations for the flow of Newtonian liquids
vice versa to either the cone or the plate. For practical reasons, the cone is usually
placed on the top, since it is easier to load a liquid sample onto a flat plate.
To help locate the cone relative to the plate, and to reduce wear, the cone is
usually truncated to remove the tip to the extent of a few tens of microns. This
makes no practical difference to the results, but does allow the use of liquids with
small suspended particles.
6.3 Flow in straight circular pipes
A pipe or a tube is a useful geometry for viscometers, but the study of flow in
this geometry is also of great importance in many industrial applications. Below we will
examine a number of important aspects of the subject.
6.3.1 Velocity profile and pressure drop in fully developed laminar flow through a
tube
We will consider first the laminar flow of a Newtonian liquid in a straight
circular piperadius a, length Lwell away from its entrance and exit. The flow rate is
Q and the pressure drop down the pipe is P. If we were able to visualise the velocity of
the fluid, we would find that it has a parabolic profile, with a value at any radius u(r)
given by
u(r ) =
P
a 2 r 2 .
4L
This implies that the shear ratethe velocity gradientvaries linearly in the tube from a
maximum at the wall, & w , to zero at the centre. This in turn shows that the shear stress
in the liquid also varies linearly from a maximum at the wall w (= Pa/2L) to zero in the
middle of the tube, see figure 2.
shear
rate
profile
du(r)
dr (r)
u(r)
r
shear
stress
profile
parabolic
velocity
profile
Figure 2: The physical parameters relating to flow

in a tube: flow rate Q, pressure drop P
along the pipe of length L and radius a.
27
This situation might seem paradoxical, in that the velocity is a maximum in the middle of
the tube, but the shear rate is zero, whereas the velocity is zero at the wall and the shear
rate is a maximum at the wall! If we follow a particle of liquid, then the time it takes to
move through the tube is a minimum at the centre and if it is right at the wall, its
velocity is zero, and theoretically it never gets through the tube!
These equations are based on the simple premise that there is no slip at the wall.
This concept is very important, and is true for most of the liquids that we will consider
here. However it is possible for very viscous materials such as polymer melts to lose
adhesion and rip away from the wall when they flow very quickly, but this is the rare
exception and certainly not the rule.
The overall equation governing laminar flow in a long pipe (L/a > 100) is
P=
8QL
a 4
where P is the pressure drop in pascals

Q is the flow-rate in cubic metres per second
L in the length of the pipe in metres
a is the radius of the pipe in metres
is the viscosity of the liquid in pascal seconds.
Remember:
105 Pa = 1 Bar = 14.7 psi
1 m3 /s = 1000 L/s = 15,852 US gal/min
1 Pa.s = 1000 cP.
This equation works for flows with Reynolds number (the ratio of inertial force to
viscous force), Re which is given by 2Q/ato around about 2300, and above this
value the flow becomes turbulent, see below.
Another way of writing the equation for flow in a tube is given by:
Pa
w
2L
=
& w 4Q 3
a
where we show the values of the wall shear stress Pa/2L and the wall shear rate 4V/a3
being divided to give the viscosity. This equation forms the basis of using the tube as a
viscometer. The wall shear stress is always given by Pa/2L, irrespective of the rheology
of the liquid, but the value of the shear rate 4Q/r3 does depend on the rheology, and has
to be suitably modified for non-Newtonian liquids [1].
6.3.2 Flow from large reservoirs into small tubes
The entry pressure drop Pe generated when a Newtonian liquid flows very
slowly (i.e. in creeping flow where Re <1) from a large radius reservoir into a smaller
tube is (approximately) given by
Pe =
28
0.6 Pa
L
where Pa/L is twice the wall shear stress w produced by the flow well inside in the small
tube. At the same time, the entry length Le within the small tube needed to achieve a
near steady-state velocity profile in creeping flow is given approximately by
Le
= 0.5 .
a
This means that under these circumstances, the velocity profile goes from an almost flat
shape at the entance to a fully developed parabolic profile within half a tube radius.
The volume of liquid V which flows out under its own weight in time t through
a vertical tube of radius a from a wide reservoir is given by [2]
a 4 tg (h1 h2 )
V=
h
8L ln 1
h2
where h1 and h2 are any two heights in the wide reservoir. This may be rewritten as the
efflux time T needed to drain a tank (but not the pipe below) as
T =
8LR 2
H
ln 1 +
4
L
ga
where H and L are now the height of the wide reservoir and the length of the pipe
respectively, and R is the radius of the reservoir.
6.3.3 Filling a tube with constant suction
The distance l that is filled when a suction of P is applied for time t to a tube of
radius a in contact with a liquid of viscosity is given by [3]
Pta 2
l =
.
4
6.3.4 Tube emptying under a constant pressure
The length of the liquid column l left when a liquid is blown from a tube of
length L at constant pressure P is given by [3]
l =
L -
Pt a 2
.
4
6.3.5 Flow through a packed bed of spheres in a tube

29
1
The pressure drop P across a well-packed bed of small particles in a tube of
diameter d, with a packing phase-volume (= 1 , where is the porisity), a bed length
L, for an average velocity v (= 4Q/d2) is given by the well-known Kozeny-Carman
equation
P =
180vL
d
(1 - )2
For this equation to apply without taking the tube walls into account, the ratio of tube
radius to that of the particle radius should be greater than 30. In some cases found in the
flow of dispersions, it is easier for the liquid to flow through the particles than it is for
the particles to flow with the liquid, and this equation then applies.
6.4 Spheres falling in Newtonian liquids
6.4.1 An isolated falling sphere
The movement of an isolated single solid sphere under the influence of gravity
in a liquid is described by Stokes law
2 g a 2
V =
9
where a is the radius of the sphere,
is the density difference between the spheres and the liquid,
is the viscosity of the liquid, and
g is the acceleration due to gravity.
If the moving sphere is itself a liquid, immiscible with the liquid through which
it is moving, then the velocity is given by the expression above multiplied by
3( + 1)
(3 + 1)
where is the viscosity ratio between the internal and external phases. Thus if the
viscosity ratio is very large, we revert to Stokes law, but if it is very small, the velocity
increases by 50%. For a gas bubble, the viscosity and density of the bubble may be
taken as zero, so is zero, and the rise velocity is then given by a2g/3 where is the
density of the liquid.
If the sphere is falling in a tube of radius R, it slows down relative to the Stokes
velocity, and an approximate correction factor for the amount of retardation is given by
(1 - 2.1R/r [1- (R/r)2]) for R/r < 0.32.
6.4.2 A falling/rising cloud of spheres
A cloud of solid spheres moving in a Newtonian liquid under the influence of
gravity travels with a velocity of [4]
2ga 2
(1 )5 0.25
V =
9
where is the phase volume of the spheres.
30
The precise value of the exponent depends on the particle size, being about 4.75
for particles bigger than a micron, and about 5.25 for submicron particles, see later under
15.13.
6.4.3 A sphere rolling down an inclined tube
The viscosity of a liquid can be calculated when a sphere of radius a rolls at a
velocity V down a tube of radius R filled with viscous liquid from the expression
4a 2 g sin R a
=
R
15V
5/ 2
where is the angle the tube subtends to the horizontal and is the density difference
between the sphere and the liquid.
When the sphere is small compared to the size of the tube, the formula simplifies
to
ga 2 sin
.
V
These equations can be used for the rolling ball viscometer [2], see figure 3.
Figure 3: The rolling ball viscometer.

6.5 Other important flows
6.5.1 Drainage down a wall
A liquid that was originally a layer of constant thickness on a vertical wall
eventually forms a parabolic wedge shape and the thickness hx as a function of the
distance x from top of the liquid sample is given by
hx =
x
.
gt
The maximum shear stress in this situationwhether for Newtonian or non-Newtonian

liquidsis at the wall, and is given by w = gh when the wall is vertical, and the
expression is multiplied by cos if the wall is at an angle with the vertical.
31
6.5.2 Flow along a slot

The pressure drop P for laminar flow along a long slot is given by
P=
QLr
bh 3
where b is the width and h the height. The value of r depends on the h/b ratio, so for
h/b = 1 it is 28.6; for 2 it is 17.5, for 3 it is 15.3; for 4 it is 14.2; for 5 it is 13.7; for 10 it is 12.8
and for infinity it is 12.
6.5.3 Flow through an annulus
The volumetric flow rate Q for laminar flow through an annulus of length L with
outer and inner diameters D and d is given by [2]
D2 d 2
P 4
4
Q =
D d
D
8L
ln
d
It can be shown that for a very narrow annular gap d D in the annulus this equation
reduces to the wide-slot formula above.
6.5.4 A rod being pushed into a cup filled with liquid
The viscosity of a very viscous liquid such as glucose syrup can be calculated
from the force F needed to push a rod of radius a1 centrally into a cup of radius a2 at a
velocity V as
F 2 1
(ln 1)
2 LV 2 + 1
where L is the depth of the rod (well away from the bottom of the cup) and is the ratio
of cup inner to rod outer diameters. This equation applies to the piston viscometer used
in on-line viscosity measurement. For a thin rod falling under its own weight in a large
cup under its own weight, the expression simplifies to
g a 22
a1
1 + ln
2V
a2
where is the density difference between the rod and the liquid. The wall shear stress
in that case is given by ga22/a1.
32
6.5.5 Flow between squeezed circular discs

The rate of change of the separation h with respect to time t is given by
dh
2F h3
=
dt
3 a4
where F is the force, a the disc radius and the liquid viscosity, see figure 4. This can be
integrated up from any position to another [2].
F, V
Figure 4 : The parallel-plate plastometer.

Where a given volume of liquid is used, initial radius Ro, and height ho, then if it
keeps within the plates,
3VRo4 ho2
.
F =
h5
6.5.6 One tube moving inside another with the space between filled with a liquid
The force F needed to push the inner tube length L through the outer cylinder
with velocity V with a liquid viscosity is given by
F=
2 LV
ln ( a 2 )
a1
where L is the length of the liquid band and a2 and a1 are outer and inner the radii
respectively. This is a simple viscometer for very viscous materials.
6.5.7 An angled plate moving across a trapped liquid
If a plate of length l and inlet and outlet heights defined by h = ho(1 + kx/l), then
the upward lift force, Fl is [5]
33
Fl =
6VBl
2k
ln(1 + k )
,
2 2
2 + k
k ho
2
and the drag force, Fd is
Fd =
VBl 2
kho
6k
4 ln(1 + k ) 2 + k
where B is the breadth of the plate, see figure 5.
Fl
Fd ,V
t
ican
lub r
B
Figure 5: The angled-plate lubrication geometry.
It is easy to show that the thrust force maximises at around k ~ 1.2, and then
Vl 2 B
Fmax
.
6ho2
This simple equation gives us the basis of lubrication, where an angled surface is forced
away from another with a viscous force generated inside the lubrication layer.
6.5.8 A cylindrical indenter being pressed into a very viscous liquid
The viscosity of a very viscous material such as asphalt may be conveniently
measured using a small flat cylindrical indenter. If the radius is a, the velocity V and the
force F in pascals, then the viscosity is simply given by
W
.
8Va
6.5.9 A rotating disc in a sea of liquid

The couple C on a thin rotating disc of radius a immersed in a sea of liquid is
given by
C=
32 3
a
3
where is the rotation rate in rad/s, and is the viscosity of the liquid. This forms the
basis of the simple Brookfield viscometer often used to measure an equivalent
Newtonian viscosity of non-Newtonian liquids.
34
6.6 Interaction between viscosity and density in flows

The energy needed to accelerate liquids in flow situations can have a marked
effect. This energy compared with the energy needed to overcome viscous forces can be
substantial, and the relative effects can be judged in different situations via an
appropriate Reynolds number, Re, which isas we have seen abovethe ratio of
inertial to viscous forces. In all such situations, the inertial force is proportional to the
density of the liquid.
6.6.1 Flow at the entrance of tubes
The entry pressure drop Pe when a Newtonian liquid flows quickly from a large
reservoir into the tube is given approximately by [6]
Pe =
01167
.
Q
0.6 Pa
(1 +
).
L
a
The entry length Le to achieve a near steady-state velocity profile in laminar flow
where inertia is important is given approximately by
Le
0.225Q
.
= 0.5 +
a
a
The velocity profile is dictated by a boundary layer that moves out from the wall as
illustrated in figure 6. The form of the equation governing this velocity-profile
development means that just below turbulent flow at a Reynolds number of say 2300, a
length of 138 diameters is needed to achieve fully-developed flow. (However for
turbulent flowsee next sectionlengths are much shorter because the relationship is
given by L/d = 4.4 Re0.167, so that just above the transition to turbulence, we only need 16
diameters to obtain fully developed flow [7].)
flat-profile
entry flow
fully developed parabolic flow
edge of boundary layer
Figure 6: The developing velocity profile in

laminar flow.
6.6.2 Turbulent flow in tubes
35
There is nothing as unstable as water, Gen. 49. 4, and as soon as we have any
reasonable flow rate in a pipe for water-like liquids, the flow becomes turbulent and
elements of fluid undergo local chaotic movement because inertial-driven
instabilities can no longer be damped out by viscosity. In this situation the following
equations are useful. For Reynolds numbers, Re, from around 2300 up to 60,000, the
pressure drop down the tube is given by [8]
P=
1
0.008LQ 2
4
.
R
,
e
a5
which can be rewritten as
w = 0.04 V 2 R
where w = Pa/2L, the wall shear stress, see above. The often-quoted friction factor
(f = 8/V2) is
f =
0.3164
Re
The ratio of the average to the maximum velocity in the flow is
Vav
1
,
=
Vmax 1 + 143
.
f
thus the velocity profile becomes flatter at higher flow rate when f decreases, as [9]
1
r 7
Vx = Vmax 1 .
a
Even though the bulk of the flow is turbulent, there is laminar sublayer at the wall, the
thickness of which is given by
5 8D
.
Re f
6.6.4 The size of the liquid jet coming out of a tube

If the Reynolds number Re (= 2aVav/) is less than 16, the eventual diameter of
Newtonian jet emerging from tubes swells slightly compared with the tube diameter; at
Re = 16 it is the same diameter as the tube and above 16 it contracts to below that value.
At very slow flows the swelling amounts to around 12 %, and at very high flow rate the
contraction is around 13%.
6.6.5 Stokes flow with inertia
36
If we used the Stokes equation to calculate the viscosity in a falling-ball

viscometer when the Reynolds number is high, the viscosity appears to be too large, and
we need to apply a correction to the formula. For small Reynolds numbers up to 2 this
is accounted for by a factor of 1/(1 + Re/16), where Re = 2V/. Figure 7 shows the
form of the inertia-driven secondary flows developing behind the sphere in this
situation which account for the extra energy in the flow that could be misinterpreted as
higher viscosity.
Re ~ 0.5
Re ~ 50
Figure 7: The streamlines around a sphere for

very slow and moderate velocities.
6.6.6 Taylor vortices in concentric-cylinder flow
For low-viscosity liquids flowing in concentric cylinders, with the inner cylinder
rotating at high velocities, we see the appearance inertia-driven secondary flow cells
called Taylor vortices, see figure 8. The onset of these vortices is controlled by the Taylor
number which is given by
a1 {a2 a1}4
T = 2 .
.
2
a2 a1
1
a2
2
where a1 is the inner radius

a2 is the outer radius
is the density of the liquid
is the rotational speed of the inner cylinder in rad/s
(= 2 where f is in revs. per second) and
is the fluid viscosity.
The onset of the vortices is governed by the critical Taylor number, Tc , which for small
gaps (a1/a2 > 0.95) is about 1700, but for larger gaps is given by
a
a
Tc = 17001 + 2 for 2 p 125
. .
a1
10a1
37
The cell pattern becomes wavy for a Taylor number 50% greater than the onset value, Tc
, and above that value soon becomes turbulent.
Figure 8: Taylor vortex secondary flow pattern between

rotating concentric cylinders.
On the other hand, if in this flow geometry the outer cylinder is rotated and the
inner is kept stationary, there is a sharp and catastrophic transition to turbulence
(without any Taylor-like secondary flows) at a simple Reynolds number, Re, of about
15,000, which is given by
Re =
(a2 a1 )a2 .
These two kinds of inertial-driven behaviour (secondary flow and turbulence)

limit the range over which concentric-cylinder viscometers can be used for low-viscosity
liquids.
6.7 Some flows where surface tension is important
The following are some examples of flows where there is an interaction between
the viscosity and the surface tension of Newtonian liquids of viscosity . In these
situations, the relative effect of surface tension may be judged from the so-called
Capillary number, Ca = V/, which is the ratio of viscous to surface tension forces.
6.7.1 Free coating
When pulling a thin sheet out of a Newtonian liquid at velocity V, a good
approximation for the final thickness of the liquid coating formed on the sheet h, is
given by
1
V 2 V
h 0.75
0.132
for values of the Capillary number up to 2, see figure 9.
38
V, F
Figure 9: A sheet being pulled out of a liquid.
6.7.2 Levelling of surface undulations

The rate at which the amplitude of a sinusoidal undulation on the surface of a
liquid decreases with time as it is pulled out by surface tension forces is given by [10]
3
da
h a 2 4
=.[
] (or halflife t2= 4/ h3)
dt
3
where is the wavelength of the sinusoidal undulation, see figure 10.
ten
sio
n
2a
very ex
pa n ded
vi ew o
f pa int
film
fac
e
sur
ion
tens
e
c
urfa
Figure 10: Definition diagram for the levelling

of surface undulations on a liquid.
6.7.3 Flow up a capillary tube
The height H that a liquid rises in a capillary tube of radius a, in time t,
(compared to the ultimate height Ho) is given by
t
H
= 1 e
Ho
8H o
2 cos
, where
ga
is the contact angle (this means that the height will decrease for mercury, since = 140
where is given by
and Ho the equilibrium height is given by

ga 2
).
39
6.7.4 A large flattened bubble moving up a tube

The approximate thickness of a Newtonian liquid layer trapped at the tube wall
by a large flattened bubble moving at a velocity U up/along a tube of radius a is given
by
= 0.643a (
3U
2
3
) .
Exercise: Calculate the rate of seepage of water through a crack 10 micron by 2 mm

which is at the bottom of the wall of a dam 2 m thick. The water is 10 m deep,
and remember that the pressure at a depth h is P = gh, where is the density
of water and g the acceleration due to gravity (9.8 m.s-2). Give the answer in
litres per day.
References
[1] Barnes, H A; Hutton, J F; Walters, K, An Introduction to Rheology, Elsevier, Amsterdam
1989.
[2] Bird R.B; Armstrong R. C; Hassager O, Dynamics of Polymeric Liquids, , Volume 1,
Fluid Mechanics, 2nd ed., Wiley, 1987
[3] Scott Blair, G.W, A Survey of General and Applied Rheology, 2nd Edition, Pitman &
Sons, London, 1949.
[4] Barnes, H A, Recent advances in rheology and processing of colloidal systems
Keynote Address in The 1992 IChemE Research Event, pp. 24-29, IChemE, Rugby,
1992, ISBN 0 85295 290 2.
[5] Tanner R.I, Engineering Rheology, Clarendon Press, Oxford, 1985.
[6] Mujumdar A.S; Mashelkar, R.A, Advances in Transport Processes, Wiley
Eastern Ltd., New Delhi, 1982.
[7] Gupta V; Gupta S.K, Fluid Mechanics and its Applications, Wiley Eastern Ltd.,
New Delhi, 1984.
[8] Pao R.H.F, Fluid Mechanics, Charles E. Merrill Books, Columbus, 1967.
[9] Darby R, Chemical Engineering Fluid Mechanics, Marcel Dekker, New York, 1996.
[10] Barnes, H A, Dispersion Rheology 1980, a survey of industrial problems and academic
progress, 190 pages, Royal Soc. Chem., Industrial Div., London, 1981.
40
CHAPTER 7: VISCOMETRY
It is a capital mistake to theorise before one has data, Arthur Conan Doyle
7.1 Introduction
Viscometry is the science of the measurement of viscosity*. Such viscometric
measurements generally have to do with applying either a force F or a velocity V at a
surface in contact with a contained test liquid. The response of this liquid to either
the velocity or the force is measured at that surface or at some other nearby surface
which is also in contact with the liquid. Examples of the geometries used for this
purpose include tubes, parallel plates, cone-and-plate arrangements, and concentric
cylinders. Sometimes artefacts arise whereby the presence of these surfaces
interferes with the local liquid microstructure, giving apparent slip effectsthese
will be discussed in detail later.
The aim of such viscometry is then
(a) to convert the applied force F to a shear stress , and
(b) to convert the velocity V to a shear rate & .
This is done using geometric constants, and is evaluated at a standard reference
point such as the immediate vicinity of the tube wall or the inner cylinder or the
outer edge of the cone and plate, etc. Then, for instance, the calculated shear stress
and shear rate & can be plotted as a function of one another, or else the shear stress
can be divided by the shear rate to give the viscosity .
In simple geometries the shear rateand hence the shear stressare the
same everywhere in the liquid, and cone-and-plate and narrow-gap concentriccylinder geometries are examples of this situation used in viscometers, see chapter 6.
In some other situations, either the shear rate or the shear stress varies in a known
manner independently of the rheology of the test liquid, and this information can be
used to reduce the measured data to viscosity/shear-rate or viscosity/shear-stress,
however we must make some assumption about the form of the viscosity/shear-rate
curve, for example that it approximately obeys a power-law relationship. Examples
of this known relationship are found in
(a) flow in a straight tube, where (see chapter 6) the shear stress varies linearly from
zero at the centre to Pa/2L at the wall,
(b) flow in the parallel-plate viscometer, where the shear rate varies linearly from zero
at the centre to /h at the edge of the plates, and
(c) flow in a wide-gap concentric-cylinder, where the shear stress falls off as the
inverse square of the distance r from the centre of rotation as T/2r2.
See APPENDIX for a discussion of on-line (process) viscometry.
41
Appropriate formulas to deal with these situations will be given at the end of this
chapter.
If the measured viscosity is constant with respect to shear rate, then the
liquid is said to be Newtonian as described above. However, as is usually the case for
structured liquids such as polymer solutions and suspensions, the viscosity decreases
with increasing shear rate: such liquids are described as shear thinning. Occasionally,
situations arise where the opposite is true, and the viscosity increases with increasing
shear rate: these are called shear-thickening liquids, see chapter 15.
7..
Some important things to note about using viscometers
7.2.1 Calibration
While it is possible to calibrate a viscometer from first principles by using
weights and pulleys etc., it is more usual to use previously-calibrated standard
liquids (mostly oils). These may be purchased from viscometer suppliers, or ordered
from laboratory-suppliers catalogues. They provide calibrated fluids starting (for
instance) from 0.29 mPa.s at 25 C, then in 18 steps that double the viscosity, arriving
eventually at 80 Pa.s; a typical cost would be around 66 per pint (~ $US100 for 0.5
kg). The fluids are NIST-traceable and ISO 9002 registered. The viscosity ranges
covered by these standardised liquids should match the ranges of viscosities that are
being routinely measured. In most cases, you will find that the calibration is quite
near that given by the viscometer manufacturer, and gross differences means that
you have a major mechanical or electrical problem.
7.2.2 Artefacts
Small zero errors and zero drift on the meters or on the electrical output of
viscometers due to mechanical and electronic changes are found more frequently
than actual calibration errors, and zero settings should be checked regularly (at least
once a month) by extrapolating low-speed data for standardised liquids.
7.2.3 Wall effects
Difficulties arises in smooth-walled viscometers because placing a structured
liquid next to the wall changes the local microstructure [1]. For a simple suspension
of smooth spherical particles, the spatial concentration of particles deep in the bulk of
the sample, well away from the wall, is random. However, right at the wall, the
particle concentration is zero. How do these two points join? The answer is that the
concentration rises rapidly as one moves away from the wall; shows a decaying
oscillatory behaviour, then it smoothes to the bulk concentration, see figure 1. This
whole process from zero to average concentration takes about five particles
diameters. The result is that the material near the wall is essentially different from
the bulk, however worse than this is the effective lubricating layer near the wall
where the particle concentration is first zero, and is small even up to half a particle
diameter. The phenomenon of lower concentration next to the wall is called wall
depletion, but is popularly known as slip, see chapter 15 section 10 for more details.
The effect is very particle-size dependent, and here particle size for
flocculated suspensions means floc size, which is shear-rate dependent. Hence, flocs
are biggest at low shear rate, and therefore for flocculated systems where this
problem is mostly seen, wall slip is a low shear rate phenomena. There are various
42
Chapter 7: Viscometry
tell-tale features that enable us to see even from superficial results if this effect is
present. These are shown in figure 2, where the measured flow curve for a
flocculated system is compared with the expected flow curve without slip.
x
s
no
Apparent viscosity, (log scale)
lip
x mm
x/5 mm
x/10 mm
Shear stress, (log scale)
Figure 1: Viscosity versus stress plot for a flocculated suspension

showing slip effects, as a function of (hypothetical)
concentric-cylinder gap.
Concentration of spheres / %
80
~ 64%
60
40
20
0
Distance out from the wall / 2a

Figure 2 : The local concentration of spheres as a function of
distance from the wall, measured in sphere diameters.
How do we eliminate these wall-depletion/slip effects? The answer is to take

the viscometer wall motion into the bulk of the liquid, which is done most easily by
either roughening or profiling the wall. Sandblasting with a coarse grit should be
enough, since this results in approximate 10 - 20 micron surface variations.
However, most rheometry manufacturers supply grooved geometries.
There is one wall effect seldom referred to, that is where the sample is fibrous.
In this case as well as upsetting the local concentration, a smooth wall upsets the
random orientation of the fibres. This too gives a lowering of the local viscosity,
since any alignment along the direction of flow makes flow easier.
To avoid
problems with fibrous suspensions, a vane geometry might be needed. The vane is
the ultimate in a profiled wall, and should be supplemented by some roughening of
the outer container, or else sliding in a gauze sleeve, see figure 3 and chapter 11.
43
Figure 3 : The vane and the vane-and-basket geometries.

7.2.4 Evaporation
Evaporation is often critical in cone-and-plate and parallel-plate geometries,
where drying at the edge of the samples leads to large errors in the measured torque,
given that the effect is at the greatest distance from the centre. The way around these
problems is two-fold: you either create a saturated atmosphere in the air next to the
sample, or else you can flood the same area with solvent, see figure 4.
cover
solvent trap
moat
solvent
liquid
overfill
test liquid
Figure 4: Various ways of overcoming evaporation in

viscometers.
7.2.5 Sedimentation/separation
Another problem that can arise with long-term shearing is enhanced
sedimentation. Even if the suspension continuous phase has been thickened to slow
down sedimentation at rest (as for instance in storage), when the sample is sheared,
the continuous phase viscosity is greatly reduced and its ability to keep particles in
suspension can be severely impaired. The only answer to this problem is to slowly
circulate the suspension, pumping it out of the bottom of the viscometer and
returning it to the top. The flow rate should be adjusted so that the average velocity
of the liquid down the viscometer gap is small compared with the velocity of the
wall, say 5%. This pumping/recirculating arrangement is suitable for naturally
unstable suspensions such as coal and mineral slurries in water.
44
7.2.6 Chemical attack

In some older instruments, chrome-plated brass geometries can be attacked
by a number of everyday liquids. Steel and plastic geometries (or even titanium in
certain circumstances) are usually recommended, but note the difference between
stainless and mild steels, since the latter can produce ions in solution which might
alter the viscosity of some aqueous liquids.
7.2.7 Mechanical damage
Accidental bending of spindles can lead to eccentric motion. Although this
might still give the correct answer for Newtonian liquids, the high/low shear rate
oscillatory flow then encountered in the gap, will give the wrong answer for nonNewtonian liquids. Always have a few spare sets of geometries arounddropping a
cone or cylinder can ruin a tight measurement schedule!
7.2.8 Sampling
Viscometers usually contain small samples, and these must be representative
of large batches of material, so care should be taken in obtaining samples.
7.2.9 High-speed testing
High-speed testing of low-viscosity liquids can lead to secondary flows, as
shown for cone-and-plate and concentric-cylinder geometries in figure 5.
-1
Viscosity, /Pa.s
10
-2
10
inertia
no inertia
-3
10
-1
10
10
10
Shear rate, / s-1

Figure 5: Inertial secondary flow patterns in concentriccylinder and cone-and-plategeometries.
Figure 6: Typical effect of inertia in a concentriccylinder and cone-and-plate geometry.
Little can be done about this phenomenon, but choosing a different geometry might
be helpful. Figure 6 shows the typical effect of high-speed testing using a lowviscosity liquid in a concentric-cylinder geometry, where the extra dissipation that
could be mistaken for an increase in viscosity is really due to Taylor vortices.
The resulting increased couple in the inertial case Ti, for a cone-and-plate
instrument, compared with that for no inertia T, is given for Newtonian liquids by
Ti
6 2 a 2
1 + 4
T
10
45
where is the density of the liquid of viscosity , and a is the cone radius whose
angle in radians is . From this equation it is easy to see that the correction becomes
important when
2 a 2
, which is a form of Reynolds number, exceeds 10. This
gives a useful guide to the highest shear rates available for testing liquids.
7.2.10 Suspended particles
Suspended particles can lead to two kinds of problem. First, large solid
particles can become jammed in narrow parts of certain geometries, such as the cone
and plate. However, the cone truncation often used to prevent wear of the tip, often
helps. Second, if the viscometer gap is not around five to ten times larger than the
largest particles, then the correct viscosity of the material is not being measured,
since the particle spatial distribution is being altered by the wall.
7.3 Viscometer design
7.3.1 General considerations
As we have seen, viscometers are instruments that can either apply a force and
measure a speed, or apply a speed and measure a force, to or from a simple geometry.
This might be as simple as a U-tube viscometer that measures the time taken for a
gravity-driven flow to move from one vertical position to another, or a similar situation
for flow from the hole at the bottom of a carefully manufactured cup called a flow-cup,
see figure 7.
cup crosssection
narrow
capillary
Figure 7 : A schematic diagram of a simple

U-tube viscometer and a flow-cup.
In both these situations, the shear stress experienced by the liquid being tested varies
in time as the vertical driving-force reduces;
in space because the flow is quite different at different places, and
in kind because the flow-cup is not a simple shear flow with the liquid being
squeezed into the hole as an extensional flow.
These cups are perfectly adequate for Newtonian liquids, once a correction has been
made for the effect of fluid density, but because of the variations mentioned, the
measurement of viscosity for non-Newtonian liquids becomes impossible. However, if
we want to apply a more well-defined situation, such as a fixed shear stress or shear rate
everywhere in the liquid, we need more control.
46
Originally, such controlled viscometers were based on an applied shear stress

which was generated by a weights-and-pulleys arrangement, as shown for instance
in figure 8.
Figure 8: Schematic diagram of the 1912 Searle controlledstress, concentric-cylinder viscometer (D. Hopkins).
Limitations were soon found with this type of instrument in terms of attaining low
shear ratesdue to bearing frictionas also were the limitations of prolonged
shearing at high shear ratesthe finite length of the string! These methods were then
superseded by electrically driven motors and the (what we now call) controlledstrain instruments had arrived, largely replacing the controlled-stress variants.
These became more and more sophisticated, and eventually with logarithmic
mechanical gear boxes were able to span a shear-rate range typically from ~ 10 -4 s-1
to ~10 4 s-1 (e.g. the Weissenberg Rheogoniometer).
7.3.2 Commercial controlled-stress viscometers
In the mid-1970s, a new generation of controlled-stress rheometers began to
appear, as opposed to the controlled-strain type of viscometer commonly used in
laboratories [2]. The first had been developed by Jack Deer and colleagues at the
London School of Pharmacy, who used air bearings and an air-driven turbine to
provide the torque. It was claimed that with such an instrument the application of a
known and controlled stress provides the rheologist with vital information of this
critical region in the form of a CREEP curve. These claims could be made because
they had introduced a specially designed air bearing ... and an air turbine drive
system for the application of torsional stress that is independent of rotational speed
... throughout the operational range. (Air bearings and air turbines had been and
still are used for dentists drills, but of course at much higher speeds, hence the
whine!) Using a pre-commercial version of the air-turbine rheometer in the creep
mode, it was found that the creep (i.e. zero shear) viscosity of wool fat BP at 25 C
over a shear rate range of 1.76 - 6.14 x 10-6 s-1 was constant (to within a small
experimental error) at ~ 2 x 106 Pa.s.
Then around 1980, commercial versions of the new generation of electrically
driven controlled-stress rheometers began to appear, still based on air bearings that
greatly reduced friction, but now using so-called drag-cup electrical motors that
allowed controlled stresses to be more easily applied, but still independent of
rotational speed. Along with these features came new ways of measuring smaller
and smaller rotation and rotation rates. The use of the latest optical-disc technology
47
now means that rotation rates as low as 10-8 rad/s (~ 1 revolution in 20 years!) can be
measured. Access to these ultra-low shear-rate regions is now called creep testing,
by analogy to the testing of solids under similar low-deformation-rate, long-time
conditions; albeit solids creep testing is usually performed in extension rather than in
shear.
7.3.3 Evolution of the specifications of controlled-stress viscometers.
Table 1 compares the specifications of some typical controlled-stress
viscometers as they have evolved over the last thirty years or so. It is clear that an
amazing increase in sensitivity and range has been brought about, especially in the
lower-and-lower minimum torques (and thus stresses) and rotation rates (and thus
shear rates) achievable.
Date
Typical
instrument
Torque
(N.m)
min.
max.
resolution
~ 1970
Air Turbine
rheometer
~ 1978
Deer
Rheometer
~ early 1980s
Carrimed
Mk1
~ late 1980s
Carrimed
CSL 100
~ 1999
TA Inst.
AR 1000
10-4
10-2
10-4
10-5
10-2
10-5
10-6
10-2
10-7
10-6
10-2
10-7
10-7
0.1
10-9
10-8
50
-
50
10-2
50
10-4
100
-
Ang. veloc.
(rad/s)
min.
(creep)
max.
50
resolution
-
Creep
(strain)
resolution
2 x 10-2
2.5 x 10-3
2.5 x 10-4
10-5
6.2 x 10-7
max.
1300
Table 1: Some typical commercial controlled-stress rheometer specifications.
7.4 Non-simple viscometer geometries
There are a number of popular viscometer geometries where the shear rate is not
the same everywhere. In order to convert the basic experimental data into
unambiguous viscosity/shear-rate data, an intermediate calculation step is needed. This
uses an assumption about the liquid, usually that at any particular value of shear rate,
the local viscosity/shear-rate data can be described by a power-law-type behaviour,
where the slope of the log/log curve is given by n. (For true power-law liquids, this is
the same as n, the power-law index.) The following are the necessary equations for
wide-gap concentric cylinders; the parallel-plate geometry and tubes used as
viscometers. In each case the viscosity data is related to a certain shear rate calculated at
some fixed point in the geometry, and n is related to the basic measured parameters.
48
The software provided by some viscometer manufacturers for calculating

viscosities in concentric-cylinder geometries only makes use of the narrow-gap
approximation (you should check this in your manual). However, this only applies to
gaps of at most a few percent of the outer-cylinder radius, and assumes that the shear
rate and shear stress are approximately constant throughout the gap. This is not the
case in many practical cases, and if for instance the inner cylinder is being rotated, the
value of the shear stress decreases as the inverse square of distance from the centre of
rotation as one moves away from the inner cylinder.
If non-Newtonian liquids are being measured and the narrow-gap
approximation is used, the data from different-sized gaps does not coincide. However,
the data can be corrected by the use of a number of lengthy analytical methods.
However, a simple correction of the data can be made as follows, using an empirical
correction factor worked out by the author, for the viscosity at the inner-cylinder,
uncorrected shear rate:
= ng [b + n' (1 b)]
where is the correct viscosity at the inner-cylinder shear rate, ng is the calculated
viscosity based on the narrow-gap approximation, n is the local power-law index at that
shear rate at which the viscosity is being evaluated, calculated as dlogT/dlog where T
and are the torque and rotational speed respectively, and, b is the gap ratio (innercylinder radius divided by the outer-cylinder radius). When this formula is used, data
for different gap ratios, b, which would have originally appeared as a set of different
curves if calculated using the narrow-gap approximation now lie on a single curve.
Table 2 shows the viscosity correction factors predicted by our simple equation for a
range of possible values of b and n.
Values of b n = 0.8
n = 0.7
n = 0.5
0.9
0.98
0.97
0.95
0.7
0.94
0.91
0.85
0.5
0.90
0.85
0.75
Table 2: Viscosity correction factors for non-Newtonian
liquids measured in wide-gap concentric cylinder viscometers.
Complete equations for the evaluation of the viscosity in the wide-gap situation
should be used for very shear-thinning liquids (say n < 0.3), i.e.
& =
2
n'
n(1 b )
and
2
n'
Cn( 1 b )
4 ri 2 L
where is the rotation rate of the inner cylinder,

C is the couple on the inner cylinder, and
ri is the radius of the inner cylinder of length L.
49
7.4.2 The rotating parallel-plate viscometer

In a parallel-plate geometry, the upper plate is usually rotated and the couple C
is like-wise measured on the upper plate. The shear rate varies from zero at the centre to
a/h at the edge of the plates, where a is the plate radius, h the gap and the rotation
rate in rad/s. The shear rate at the edge is given by
& a =
a
,
h
and the viscosity (evaluated at the edge shear rate & a ) is given by
3Ch
.
1 d log C
4
2 a 1 +
3 d log
For a power-law liquid dlogC/dlog is simply n.

7.4.3 Pipe/tube viscometer
Tube or pipe viscometers take many forms, but they should all be able to give
the pressure-drop P as a function of flow rate Q for situations where the tube is long
enough to be able to neglect entrance and end effects, say L/a > 50. In this case we can
calculate the viscosity as a function of the wall shear rate, & w , which is given by
&w =
4Q 3 1 d log Q
+
.
a 3 4 4 d log P
The viscosity is then given by
a4 P
=
d log Q
2QL 3 +
d log P
where for power-law liquids, dlogQ/dlogP is simply 1/n.

References
[1] Barnes, H A, JNNFM, 56, 221 - 251 (1995).
[2] Barnes, H A, JNNFM, 81(1&2), 133 - 178 (1999).
50
CHAPTER 8: YOUR RHEOLOGY LABORATORY

The ultimate object is to measure something which we have already seen ...
to obtain a numerical estimate of some magnitude, James Clark Maxwell
8.1 Whats in a good rheology laboratory?
What do you expect to find in a good rheology laboratory? The answer
depends very much on the particular field of interest. Here we shall discuss what
we would anticipate for a number of industrial situations, but the same would be
true for academic studies in the same general area:
Areas of interest
Types of liquid
Personal products and cosmetics

Paints, coatings and inks
Polymers
Lubricants
Detergents
Pharmaceuticals
Food and biotechnology
Minerals
creams, lotions, pastes

suspensions
solutions, melts
oils, greases
suspensions, pastes
dispersions, lotions, creams
dispersions, emulsions, pastes
slurries, suspensions
This is not an exhaustive list (for which see Whorlows book [1]), but liquids found in
any other areas fall somewhere within those examples found in this list.
8.2 Which problem?
Obviously in each area, certain specific items of test equipment will be used,
such as small extruders, paint rigs, Stevens texture testers and penetrometers, etc.
However, here we are interested in those pieces of rheological equipment that we
should also find. The following important features are generally measured in
rheological studies for typical liquids listed above, with some specific examples:
51
Product Property
general thickness
suspendability
technical attributes
microstructural characteristic
Examples
- consumer-perceived flow properties
- long-term physical stability
- slow draining
- state of flocculation
Processing
Examples
machineability
- fly or misting in ink or coatings machines
process engineering - heat transfer, pump specification, mixing efficiency
8.3 Which rheometer?

The following are some general classes of rheometer:
8.3.1
Standard laboratory viscometers,
Standard laboratory viscometers have a shear rate range ~ 0.1 - 1000 s-1 . At
the moment these include the
Brookfield DV3
Haake VT7, VT500/550, RV1
Bohlin Visco 88
Rheometrics RM 180, 265
Reologica ViscoCheck
Physica MC1
Typical output: viscosity/shear-rate as a function of time of shearing for the shear
rate specified, so that the general flow curve and thixotropy can be studied for a
range of temperatures, usually in the range -10 to 100 C.
8.3.2 Rheometers,
Rheometers available today, and the typical viscoelastic parameter(s) they can
measure are shown in table 1.
8.3.3 Elongational viscometers
Rheometric Scientific RME (for polymer melts only)
Typical output: Extensional stress as a function of extensional strain rate/time.
Exercise: Find the Internet websites of the most prominent viscometer and
rheometer manufacturers, viz. Brookfield Engineering, Haake, Bohlin, TA
Instruments, Reologica Instruments AB, Rheometric Scientific and Physica,
start at http://www.staff.ncl.ac.uk/chris.petrie/links.html
52
Chapter 8: Your Rheology laboratory
Exercise: Imagine you are working in the following areas, and discuss the kinds of
rheometers you would like to use in setting up a laboratory to study
asphalt, blood flow, cosmetics, chocolate.
Normal Force (to give N1)*
Oscillation (to give G, G)
Controlled Stress (to give J(t))
Bohlin - CVO 50/120
Bohlin - CS, CVO 50/120,

DSR
CSR
Bohlin - CS, CVO 50/120,

DSR
CSR, CVR
Brookfield - RSCPS
Haake - RT20, RS75/150
Haake - RheoWave 1, RS
75/150
Haake RS1, RS 75/150
Physica - UDS 200
Physica - DSR 4000

UDS 200
Physica - DSR 4000

UDS 200
Reologica STRESSTECH/HR
VISCOTECH
Rheometrics
- ARES
SR 2000/5000
Rheometrics SR5/2000/5000
ARES
Rheometrics
SR5/ 2000/5000
ARES (controlled-rate
instrument using feed back)
TA(Carri-med) - AR1000N
TA(Carri-med) - AR 1000
CSL2 500
TA(Carri-med) - AR 1000
CSL2 500
* some instruments only give a qualitative value of normal force - check with manufacturer.
Table 1: Commercial rheometers available today and their available functions.

References
[1] Whorlow, R.W., Rheological Techniques, Second Edition, Ellis Horwood, New
York, 1992
The Deer Rheometer Mk 1
53
54
CHAPTER 9: SHEAR-THINNING LIQUIDS

[The viscosity] varies with the speed of shearing, Schwedoff (1890)
9.1 Qualitative features of flow curves
Employing the well-defined viscometers we discussed earlier, viz. the smallgap, concentric-cylinder or the cone-and-plate type, we are able to measure directly the
viscosity of structured liquids such as polymer solutions, emulsions, dispersions, etc., as
a function of either the applied stress or the applied shear rate. When we do this over a
wide-enough range of either shear rate or shear stress, we generally see the kind of
behaviour shown in figures 1 - 8, when, as described in chapter 4, we plot the results
on logarithmic axes.
Viscosity, / Pa.s
Li
qu
id
ab
ra
siv
Fa
ec
bri
lea
cw
ne
a sh
r
ing
liq
uid
10
10
10
Mustard
-1
10
-6
10
-4
-2
10
10
10
Shear rate,
.
/s
10
10
10
-1
10
-1
Figure 1: Viscosity/shear-rate curves for various products.
10
. -1
Shear rate, / s
10
Figure 2: Viscosity/shear-rate curve for a dilute solution

of polyacrylamide in glucose syrup.
Viscosity, / Pa.s
10
Viscosity, / Pa.s
10
Viscosity, / Pa.s
10
10
10
10
10
10
-2
xa
nth
an
gu
a lg
m
ina
te
10
10
10
Shear rate,
10
.
/s
10
-1
Figure 3: Viscosity/shear-rate curves for a typical printing ink.
-5
10
-1
10
Shear rate,
.
/s
10
-1
Figure 4: Viscosity/shear-rate curves for 1% aqueous

solutions of two natural polymeric thickeners.
At low-enough shear rates or shear stresses, the viscosity is constant (with a

value 0), but at some point it begins to decrease, and usually enters a straight-line
region on a logarithmic plot, which indicates power-law behaviour, see figure 9.
55

2
10
20
10
10
Viscosity, / Pa.s
Viscosity, / Pa.s
10
3.5
2
1
0.35
-1
10
-3
10
Show
er Ge
l
10
-3
10
-1
10
10
10
10
10
100
-1
Shear rate, / s
106
-1
10
Viscosity, / Pa.s
Viscosity, / Pa.s
10
104
.
/s
Figure 6: Viscosity/shear-rate curves for two personal products.
10
102
Shear rate,
Figure 5: Viscosity/shear-rate curves for 3% by weight polystyrene

in toluene: parameter - various molecular weights in
millions.
10
Bod
yW
ash
-2
-2
-2
10
10
Shear rate, / s
10
-1
Figure 7: Viscosity/shear-rate curves for 2% by weight,

polyisobutylene in decalin.
10
Ha
ir
10
Fa
bri
cc
10
10
Co
nd
itio
ne
r
on
dit
ion
er
-1
10
-8
10
-2
10
. -1
Shear rate, / s
10
Figure 8: Viscosity/shear-rate curves for two conditioners.
Viscosity, (log scale)
This decrease of viscosity with shear rate is called shear thinning and must be
distinguished from a decrease of viscosity with time of shearing which is called thixotropy,
see later. At some point well down the viscosity curve, we see the beginnings of a
flattening out, and if data at a high-enough shear rate or shear stress is available, then a
second constant viscosity region, (with value ), is usually seen. Thus we have the two
limiting Newtonian viscosities, 0 and separated by a power-law region.
Cross model
Power-law model
Sisko model
Shear rate, (log scale)

Figure 9: Definition diagram of the various models
and the ranges that they cover.
Sometimes the position of the typical behaviour along the shear-rate axis is such
that the particular measurement range used is too low to pick up the higher-shear-rate
part of the curve, see for instance figures 10 - 13 (note that the typical shear-rate range of
most laboratory viscometers is ~ 0.1 to 103 s-1). This is particularly true for polymer
melts and shampoos. On the other hand, if the shear-rate range is too high, then the
56
Chapter 9: Shear-thinning liquids
lower shear-rate behaviour is not seen, then the kind of behaviour shown in figures 14
and 15 is observed.
2
10
limit of
Newtonian
behaviour
10
10
10
0.8%
-1
10
0.5%
-2
0.25%
10
-3
10
10
-2
10
-1
10
10
10
10
. -1
Shear rate, / s
-2
10
10
Figure 10: Viscosity/shear-rate curves for a range of

silicone oils sold as viscosity standards.
10
10
. -1
Shear rate, / s
10
Figure 11: Viscosity/shear-rate curves for aqueous solutions of

locust bean gum.
10
10
4
3
2
10
Viscosity, / Pa.s
Viscosity, / Pa.s
1.8%
1.5%
Viscosity, / Pa.s
Viscosity, / Pa.s
10
1
-1
10
0.5
locust be
an gum
polysaccharide
guar gum
10
modified cellulose
carrageenan
10
-2
-3
10
10
-3
-1
10
10
10
10
. -1
Shear rate, / s
10
Figure 12: Viscosity/shear-rate curves for polystyrene

(Mw = 20M) in toluene: parameter - weight %.
-5
-3
10
-1
10
10
10
. -1
Shear rate, / s
10
Figure 13: Viscosity/shear-rate curves for 1% by wt. various

natural polymeric thickeners in water.
10
10
10
10
10
6
4
10
10
2
-1
10
-2
10
-3
-2
10 10-7
10
best-fit power law

best-fit Sisko model
Viscosity, / Pa.s
Viscosity, / Pa.s
This is often seen for dispersions and emulsions measured in typical laboratory
viscometers. Lastly, the situation sometimes arises where both the lower and higher
shear rate behaviour is difficult to see, and then only the power-law region is seen. This
is shown to be the approximate behaviour seen in figure 16. This is fairly typical of most
situations where the power-law is forced to fit the data, when careful scrutinytry
looking along the curve from the sidewill show that there is always some curvature.
10
-5
10
-3
10
-1
10
10
. -1
Shear rate, / s
Figure 14: Viscosity/shear-rate curves for 0.35% aqueous
Carbopol solution.
10
10
-2
10
10
Shear rate, / s
10
-1
Figure 15: Viscosity/shear-rate curves for aqueous starch

dispersions: ordinate weight %.
57
9.2 Mathematical descriptions of flow curves

All the features described above can be captured using simple equations
relating viscosity and shear rate via a minimum number of parameters. The
following are some examples of just some of the simpler forms of equations which fit
different parts of the flow curve, see figure 17 for an overall picture.
2
10
Viscosity, / Pa.s
10
10
Facial Wash
-1
10
Fabr
ic
-2
10
wash
ing l
iquid
po
we
rl
aw
00
-3
10
10
10
10
. -1
Shear rate, / s
10
Figure 16: Viscosity/shear-rate curves for two personal products.
-1
Shear rate, / s (Log scale)

Figure 17: Definition diagram of parts of the flow curve.
9.2.1 The Cross model

First, we will consider one equation that describes the whole curve: this is called
the Cross model, named after Malcolm Cross, an ICI rheologist who worked on dyestuff and pigment dispersions. He found that the viscosity of many suspensions could
be described by the equation of the form

1
=
o 1 + ( K& ) m
where, written in this particular way, K has the dimensions of time, and m is
dimensionless. When this model is used to describe non-Newtonian liquids, the degree
of shear thinning is dictated by the value of m, with m tending to zero describes more
Newtonian liquids, while the most shear-thinning liquids have a value of m tending to
unity.
If we make various simplifying assumptions, it is not difficult to show that the
Cross equation can be reduced to Sisko, power-law and Newtonian behaviour, see
below. There is another Cross-like model which uses the stress rather than the shear
rate as the independent variable, it has been called the Ellis or sometimes the Meter
model, and for some specific values of the exponent, it has been gives other names:
for an exponent of unity it has been called the Williamson or Dougherty and Krieger
model, while for an exponent of two it has been called the Philippoff model, etc.
The Carreau model is very similar to the Cross model, but with the whole of
the bottom line within brackets, i.e. (1 + (K & )2)m/2: the two are the same at very low
and very high shear rates, and only differ slightly at K & ~ 1.
9.2.2 The Bingham equation

58
For most structured liquids at high shear rates, 0 >> and K & >> 1, and
then it is easy to show that the Cross model simplifies to the Sisko equation
= +
0
.
(K& )m
For the most shear-thinning liquids, m is unity, and the equation above can
easily be recast as the Bingham equation by multiplying throughout by shear rate, so
o
+ & or = o + p &
K
where the Bingham parameters are the yield stress , which is now given by / K,
and the plastic viscosity p which is now given by .
Figure 18 shows the typical form of the results that would normally have been
measured and plotted graphically over the last 50 years. Over a reasonable range of
shear rates, the shear stress seemed to be a linear function of shear rate, but now
displaced upwards by a constant value, which is called the yield stress. This was found
by extrapolation to where the shear rate was zero. This seemed to show that there
would be no flow at all unless the stress was higher than this critical value. Bingham
investigated systems like this, and this simple straight-line-with-intercept-type
behaviour bears his name, and such liquids are called Bingham plastics, and as we have
said above, the intercept is called the Bingham yield stress o and the slope of the straight
line is called the plastic viscosity, p.
Shear stress,
p
y
Shear rate,
Figure 18: Definition diagram of the Bingham model.
9.2.3 The Power-law liquid

In many situations, 0 >> , K & >> 1, and is small. Then the Cross
equation (with a simple change of the variables K and m) reduces to the well-known
power-law (or Ostwald-de Waele) model, which is given by
= k& n or = k& n 1
where k is called the consistency and n the power-law index.
59
While Bingham linear plots are still favoured by many physical chemists,
most engineers were happy to utilise the next most obvious two-parameter model,
the power-law equation. In the power-law model, the consistency k has the strange
units of Pa.sn and the power-law index is dimensionless and usually ranges from 1
for Newtonian liquids towards 0 for very non-Newtonian liquids. This law alone is
sufficient to describe many non-Newtonian flows, and the use of it is described in
chapter 10 with many examples, which are analogous the Newtonian examples
shown above. The power-law description does well for most structured liquids from
shear rates around 1 to 103 s-1 or so, but at about 103 s-1, there is usually some
flattening. However, for engineering calculations, power-law predictions are quite
reasonable if limited to the medium range.
9.2.4 The Sisko model
Many real flows take place for structured liquids at shear rates where the
viscosity is just coming out of the power-law region of the flow curve and flattening
off towards . This situation is easily dealt with by simply adding a Newtonian
contribution to the power-law description of the viscosity, giving
= k& n 1 + , where K n = k ,
or in terms of shear stress
= k& n + & .
This is called the Sisko equation, and it is very good at describing the flow
behaviour of most emulsions and suspensions in the practical everyday shear rate range
of 0.1 to 1000 s-1.
Exercise: In figure 19, which models would you attempt to fit to each the flow curves
shown?
1
shampoo
polyme
r
10
m
gu
t
ur
gh
yo
n
ha
nt
xa
Viscosity, / Pa.s
10
-1
10
bloo
d
-2
10
10
-1
10
10
Shear rate,
10
.
/s
10
10
-1
Figure 19: Viscosity/shear-rate curves for blood, liquid

crystalline polymer, shampoo, yoghurt and
an aqueous xanthan gum solution.
9.3 Fitting data to equations

60
Let us suppose we have a set of shear-stress/shear-rate data ( , & ) which we

want to fit to an appropriate equation. First we plot the data on a linear basis to see if it
fits the Bingham equation. If it does, then all we need to do is to perform a linear
regression on the data, using a simple spread-sheet software package, or the software
usually provided nowadays with the viscometer/rheometer we have used. Then the
values of o and p can be used to predict flows of the liquid in other geometries, see
chapter 10.
If the linear plot shows curvature, the data should be plotted on a logarithmic
basis. From this plot we can see if we have a reasonable straight line, or a straight-line
and some curvature. If a fair straight line is seen, then the data can be submitted for
power-law regression and the k and n extracted and used for prediction. Last, if there is
considerable curvature, then the Sisko model, or if necessary the complete Cross
equation is indicated. However, here we have a problem, because the equation is nonlinear and simple regression analysis is inadequate. However, most modern
viscometers have suitable software for this purpose, otherwise other specialised
commercial mathematical software.
Chapter 10 will show how these simple non-Newtonian equationsBingham,
power-law and Siskocan be used to describe the flow of liquids in various geometries.
However a note of warning must be soundedthe equations should only be used to
predict behaviour within the range of the measurements taken to produce the various
fitting parameters. In simple terms this is done by finding a representative shear rate,
which is a typical size dimension and a typical velocity, so for a pipe we use the radius a
and the average velocity V derived form the volumetric flow rate Q, so V = Q/a2. Then
our representative shear rate would be given by V/a or Q/a3. Then the range of shear
rates in both situationsthe viscometer and the pipeshould be approximately in the
same range of decades, say 10 - 100 s-1.
61
CHAPTER 10: EQUATIONS FOR THE FLOW OF NONNEWTONIAN LIQUIDS

An equation is for eternity, Einstein
10.1 Introduction
Just as we have collected a series of simple mathematical solutions for flows
involving Newtonian liquids (Chapter 6), here we do the same for non-Newtonian
liquids. These are generally power-law equations, but some other flow laws are also
included. For most engineering applications the power law description is adequate if
the parameters for the flow law relate either to the local point where the calculation is
being made, or generally over the range being considered, see chapter 8, see also [1 3].
10.2 Steady-state flow in a straight circular pipe
10.2.1 Power-law liquids
If the liquid obeys a power-law flow law over the range of interest, then
1
3+
2kL
n ]n
P=
[Q.
3
R
R
where k is the consistency and n the power law index; Q is the flow rate through the
pipe of radius a, with a pressure drop of P over a length L.
The radial velocity profile u(r) across the tube is given by
1
n+ 1
n P n n + 1n r n
u (r ) =
a
1
n + 1 2 Lk
a
where r is the distance out from the centreline towards the pipe wall. The wall shear
rate is given by
&w =
Q
a3
1
3 + n .
We can use the tube as a viscometer, and plot of the wall shear stress w as a function of
& w . If the value of n only changes slowly over a range of flow rates, then this approach
can be used to deal piecewise with the whole flow-curve, even though it does not strictly
follow power-law behaviour over a wide range of shear rates.
63
The entry pressure needed for a power-law liquid to get into a pipe from a large
reservoir for slow (creeping) flow, see chapter 6, is less than that for Newtonian liquids
( P =
0.6 Pa
) by the factor
L
3(3n + 1)2
.
2(2n + 1)(5n + 3)
10.2.2 Sisko model

( = k &
If the liquid flowing in a pipe/tube can be described by the Sisko model

n-1 + ), then the pressure drop is given (to a very good approximation) by
L V 3n + 4s + 1
V 3n + 4s + 1
P = 2 [ k .
+ .
]
R R
n+ s
R
n+ S
where S =
V
k [4 ]
R
n -1
Where V is the average velocity in the pipe (= Q/a2) of radius a and length L. This
equation reduces to the power-law version (see relevant equation above) when is
zero. It is very good for describing the flow of most emulsions and dispersions through
pipes.
10.2.3 Casson fluid
The Casson fluid is essentially the same as its Bingham cousin, but with all the
components raised to the half power, so
= 0 2 + (k& ) 2 .
1
Then the flow rate in a pipe is given by

4
4 0
0
a 4 P
16 0
+
Q =
.
1
4
8 p L
7 w
3 w
21 w
Notice that this equation reduces to the Newtonian version at very high wall shear
stresses, when the wall shear stress w is much greater than the yield stress o.
Workers who handle molten chocolate favour the Casson model; blood is also
believed to conform to this model.
10.2.4 Bingham fluid
The Bingham model applied to pipe flow is one of the oldest mathematical flow
solutions, and it is called the Buckingham-Reiner equation:
64
Chapter 10: Equations for the flow of non-Newtonian liquids

4
4 o
Pa 4
1o
.
1
Q =
+
8 p L
3 w
3 w
If the ratio of stresses o/w is small (as when the yield stress is small anyway or the flow
stress is quite high), we can ignore the third term in brackets. The formula is then
simpler, and can be re-arranged to give P as a function of Q, hence
P=
8 p LQ
a
8 L o
.
3a
In order to accommodate a Bingham fluid flowing down a pipe with a

Newtonian slip layer (usually made up of the continuous phase), we need to add the
term
a3P
to the full equation for flow rate, where is the thickness of the
2 L s
Newtonian slip layer whose viscosity is s. In most practical cases where slip layers are
observed, they are usually less than one micron in thickness, but because the viscosity is
so low, they can still dominate the flow.
10.3 Drainage down a wall under the action of gravity
The surface velocity V of a power-law liquid draining down a vertical surface is
given by
V =[
1+ n
g 1 n
.h n
]n .
k
1+ n
where is the liquid density, g the acceleration due to gravity, k the consistency, n the
power-law index, and h is the liquid-film thickness.
10.4 A sphere moving in a non-Newtonian liquid
The various expressions found in the literature for the Stokes flow of a sphere of
radius a under the force of gravity in a power-law liquid have been simplified by the
present author to give the average shear rate and shear stress as
& av =
V
[9 - 7n] ,
2a
and
av =
ga
[1 + n] .
9
The viscosity is then given by:
2g a 2 1 + n
.
9V 9 7 n
65
These expressions simplify to the correct Newtonian values when n is set to 1.

We can invert these expressions so that given a known shear stress and a measured
viscosity at that stress, the velocity can be calculated.
10.5 Flow in a slit (a very wide slot)
The wall shear rate & w and wall shear stress w for the low of a power-law liquid in
a slot (see above, with b >> h ) are given by
& w =
2Q
1
2 + ,
2
bh
n
and
w =
hP
.
2L
The viscosity corresponding to these values of shear rate and shear stress is then given
by
P h3 b
.
1
4QL 2 +
This reduces to the expression for a very wide slot, see above.
10.6 Radial flow between parallel discs
If two circular discs are separated by a distance 2h, with a pressure-drop P for
radial flow between any two radii a1 and a2, the flow rate is given by
1
n
4 h Ph(1 - n)
Q =
.
2 + 1 n k a 21-n - a1-n
1
2
10.7 Flow out of a tank through a long exit pipe

The efflux time T of a liquid flowing from a tank through a long exit tube is
given by
1
1
1
12
1-
n
n
2kL
(H
+
L
)
R
L n
T = 3 + 1n
3
1
ga a
1
The height and radius of the tank are H and R and those of the outlet pipe L and a. This
approach assumes that the emptying is slow enough to leave no material on the walls of
the vessel.
66
10.8 Flow of a power-law liquid between squeezing plates

The power-law solution for squeezing flow between plates of radius a for a
power-law liquid of original thickness h0 and current thickness h at time t is given by
h0
h
n +1
n
1 h
= 1 + 1 + 0
n 2
n +1
n
Kt
where K is given by
n
K =
( n + 3) F 2n
,
k a n + 3 2n + 1
see figure 1.
F, V
Figure 1 : The parallel-plate plastometer.

The expression for calculating the apparent yield stress in a parallel-plate
plastometer is
5
o =
W he 2
2V
where W is the applied weight, V the volume of the sample, and he the equilibrium
distance between the plates. The plastic viscosity, p, can also be calculated, using
the speed of approach of the plates, see 10.8 above.
10.9. Flow through a converging nozzle
The pressure drop P along a tube of length L that reduces in radius from a1 to a2
for a power-law liquid at a flow-rate Q is given by
67

n
3 n
3 n
2kL Q 1
a1 a 2
P =
+ 3
.
3n n
a1 a 2
10.10 Levelling of surface undulations under the action of surface tension

If the surface undulations are assumed to maintain a sinusoidal cross-section, we
can say that
n 1
n
= a
n 1
n
0
1
3+ n
2n +1
n 2 n
(
n
1
)
h n t

(2n 1) k
where a is the amplitude of the undulation, a0 is the initial amplitude of the undulation,
is the wavelength, h is the mean film thickness, is the surface tension, k and n are the
power-law constants and t is time. The maximum stress m in that situation is given by
83.ah/3.
10.11
A disc rotating in a power-law liquid
One of the simplest and cheapest viscometers on the market today is the
Brookfield viscometer based on a thin disc rotating in large volume of liquid. While
the viscosity is normally quoted as the equivalent Newtonian viscosity as a function
of rotation speed in rpm for a given spindle, it is possible to evaluate the power-law
parameters k and n for suitable liquids using the following equations [4]
n=
d log C
,
d log
k =
3C
and
8 3 n
4
R
n
where C is the torque on the rotating disc,

is the disc rotation rate in rad/s, and
R is the radius of the disc.
It is also possible to define an average shear stress and shear rate for this
situation as
&av =
8
n
and
8
av = k
.
n
n
68
While these equations properly relate to true power-law liquids only, it is

possible to obtain a reasonable estimate of the true flow curve by correcting the data
collected at each point using the local value of dlogC/dlog for n.
10.12 More formulas for liquids with an apparent yield stress [5]
10.12.1 Flow in a mixer
When very non-Newtonian liquids are mixed in a vessel, a cavern of visibly
moving liquid is seen around the impeller, while liquid outside the cavern appears to
be motionless, see figure 2.
N ( = /2)
very
slow-moving
material
~ 0.5 Dc
well-mixed
cavern
D
Dc
Figure 2 : Cavern flow in a simple mixer.
The size of this mixing cavern is given by
1.36 Po N 2 D 2
Dc
D = 2

0
where Dc is the diameter of the cavern, D is the diameter of the impeller, N is the
impeller speed in revolutions per second (rps) and Po, the impeller power number is
given by P0 =
2 C
, where C is the couple on the impeller.
N 2 D 5

For a Bingham plastic between widely spaced concentric cylinders, at large
stresses (i.e. where C > 2o r22), the rotation rate (in rps) is given by
C
=
4
1
1 o a2
ln
2 2
a
a
p 1
p
1
2
69
where C is now the couple per unit height of the cylinder; a2 and a1 are the outer and
inner radii of the cylinders respectively. By plotting C against , the values of o and
p could be found.
10.12.3 Capillary rise of a yield-stress liquid
Recently, Kotomin [S. Kotomin, Moscow, personal communication, Sept.
1998] has suggested a simple means of measuring yield stress using two opentopped capillaries immersed in a yield-stress liquid. The value of yield stress o is
given by the simple equation
o =
g (r2 h2 r1 h1 )
h1 h2
where h1 and h2 are the equilibrium heights of the liquid of density

capillaries with radii a1 and a2, respectively.
in the
References
[1] Bird, R.B., Armstrong, R.C.; Hassager, O, Dynamics of Polymeric Liquids
Volume 1, John Wiley, New York, 1977.
[2] Macosko, C.W., Rheology Principles, Measurements and Applications, WileyVCH, New York, 1994.
[3] Tanner, R.I, Engineering Rheology, Clarendon Press, Oxford, 1985.
[4] Wein, O.J, JNNFM 1, 357 (1976).
[5] Barnes, H A, JNNFM, 81(1&2), 133 - 178 (1999).
70
CHAPTER 11: VERY SHEAR-THINNING OR YIELD STRESS FLUIDS

Difference of opinion leads to enquiry, and enquiry to truth, Thomas Jefferson
11.1 Introduction
Special attention has been paid over the years to very shear-thinning liquids
that appear to have a yield stress [1]. It was usually thought that at stresses below the
yield stress no flow takes place, and only elastic behaviour is seen. This is in fact
only half the story, since indeed although there are such materials which appear to
show this kind of behaviour, in reality there is as much happening in terms of flow
below as above the yield stress. Figures 1 10 (see [1] for details) show a number
of examples of such liquids, where the viscosity falls many orders of magnitude over
a narrow range of shear stress, and indeed when approaching this critical stress
region from regions of high stress it appears that the viscosity goes to infinity at a
certain minimum stress. However, careful and patient measurement below this stress
shows that the viscosity is still finite, and eventually levels off to a constant, but very
high value at low stress.
106
Viscosity, / Pa.s
x 1.5
x 1,000,000
Viscosity, / Pa.s
108
104
2
10
10
10
10
10
10
20
30
0.15%
0.20%
0.03%
-2
10
50
-1
10
10
x3
Viscosity, / Pa.s
x 100,000
Viscosity, / Pa.s
10
10
10
10
10
-2
-1
10
10
Shear stress, / Pa
Figure 3: Flow curve of penicillin broth.
10
10
x4
x 100,000
10
-2
10
Figure 2: Flow curves for various concentrations of Carbopol 980

in water, each box showing a million-fold drop in
viscosity for a three-fold increase in shear stress.
Figure 1: Flow curve of molten chocolate measured

in a vane geometry to eliminate slip.
10
Shear stress, / Pa
Shear stress, / Pa
10
10
-1
-1
10
10
10
Shear stress, / Pa
Figure 4: Flow curve of 6% iron oxide in mineral oil.
71
Here we take a little time to explain how the situation came about whereby
yield stresses were supposed, and how we explain it these days. We finally
comment on how to make proper use of the mathematical constant that is often
called the yield stress, but which has no physical reality, see Barnes [1] for fuller
details.
x2
6
Viscosity, / Pa.s
10
10
-2
10
x 1,000,000
10
x 100,000,000
Viscosity, / Pa.s
x9
6
10
10
-1
10
10
10
10
10
20
Shear stress, / Pa
30
50
100
200
300
Shear stress, / Pa
Figure 5: Flow curve of 10% suspension of Bentonite.
Figure 6: Flow curve of a mayonnaise.
x2
10
10
10
10
x2
10
200
300
500
1,000
10
2,000
Shear stress, / Pa
10
107
x 10,000
-3
-2
10
10
-1
10
10
Shear stress, / Pa
Figure 9: Flow curve of saliva at room temperature.
105
x 10,000
Viscosity, / Pa.s
10
-1
10
x2
x2
10
Figure 8: Flow curve of a toothpaste.
10
Viscosity, / Pa.s
Shear stress, / Pa
Figure 7: Flow curve of a flocculated ink.
10
x 20,000
10
10
Viscosity, / Pa.s
10
x 40,000
Viscosity, / Pa.s
10
103
101
101
102
Shear stress, / Pa
Figure 10: Flow curve of tomato paste at room temperature.
11.2 The history of the yield stress

A typical dictionary definition of the verb to yield would be to give way under
the action of force and this implies an abrupt and extreme change in behaviour to a
less resistant state. The yield stress of a solid material, say a metal like copper, is the
point at which, when the applied stress is increased, it first shows liquid-like
behaviour in that it continues to deform for no further increase in stress. Similarly,
72
103
Chapter 11: Very shear-thinning or yield-stress liquids
for a liquid, one way of describing its yield stress is the point at which, when
decreasing the applied stress; it first appears to show solid-like behaviour, in that it
does not continue to deform.
Anyone testing a very shear-thinning liquid (e.g. toothpaste, Carbopol,
tomato puree, etc.) in a viscometer with a limited range of shear ratessay the
typical ~ 1 to 1000 s-1 of a laboratory viscometerand plotting the results linearly, is
almost bound to come to the conclusion that the liquid has a yield stress. However,
measurement of the complete flow curveusing the appropriate equipmentshows
that no real yield stress exists. When plotted as, say, the logarithm of viscosity against
the logarithm of applied stress, the phenomenon is much better illustrated. Within
the limited range of measurable shear rates, it looked as though the viscosity of such a
yield-stress liquid was increasing asymptotically as the applied shear stress was
decreased. Indeed, for all practical purposes it looked as though there was a definite
stress where the viscosity did become infinite. These materials can easily show a
million-fold drop in viscosity over a very small region of increasing shear stress,
often around or even less than a decade, see again figures 1 10 [1].
The liquids that appear to have a yield stress are legion. Among them are
many original examples noted early in the century such as
clay, oil paint, toothpaste, drilling mud, molten chocolate, etc., then later, materials as
diverse as creams of all sorts, ketchups and other culinary sauces, molten filled
rubbers and printing inks, etc.,
were found to show similar behaviour. Nowadays such disparate systems as
ceramic pastes, electro-viscous fluids, thixotropic paints, heavy-duty washing
liquids, surface-scouring liquids, mayonnaise, yoghurts, purees, liquid pesticides,
bio-mass broths, blood, water-coal mixtures, molten liquid-crystalline polymers,
plastic explosives, foams, battery and rocket propellant pastes, etc.,
may be added to the extensive list.
11.3 Values of yield stresses
The values of apparent yield stresses measured for different sorts of liquidlike foods range over the following values,
ketchup and drilling muds
spaghetti sauce
mustard, and apple sauce
mayonnaise
tomato paste
~ 15 Pa
~ 25 Pa
~ 60 Pa
~ 90 Pa
~ 125 Pa
Within the limited time-scale of eating a meal, it is obvious what the concept
of a yield stress means to the non-expert, in terms of, for instance, the confined
spreading out of a liquid-like food when poured out, which seems to come to a
constant thickness, e.g. thick sauces and ketchup.
In food processing, given the limited time-scales of many processing
operations, the thickness of layers of liquid-like foodstuffs left on vertical walls after
drainage has taken place can be calculated from the yield stress o, for o = gh,
where is the fluid density, g is the acceleration due to gravity, and h is the layer
73
thickness. Having, for instance, emptied a vat of such a liquid product, we could then
calculate, at least in the short-to-medium term when the liquid does not appear to
flow, the thickness of the layer left on the wall of the vessel, which has to be removed
in any ensuing cleaning operations.
Sherman [2] listed the following values of yield stress (in pascals) and their
every-day significance for soft-solid, food-like, spreadable materials (such as
margarine and butter) -
5,000
10,000
20,000
80,000
100,000
10,000
20,000
80,000
100,000
150,000
means
very soft, not pourable

soft, but already spreadable
plastic, and spreadable
hard, but satisfactory [sic] spreadable
too hard, limits of spreadability
The yield stress of such materials can be easily calculated as o = W/a2 for a rightangled cone of radius a at the plane of penetration, where W is the loading of the
cone in newtons.
11.4 The arrival of commercial controlled-stress rheometers
As we have seen in chapter 7, the new generation of commercial electrically
driven, controlled-stress rheometers with the latest optical-disc technology now
means that rotation rates as low as 10-8 rad/s (~ 1 revolution in 20 years) can be
measured! This has opened up a new range of previously unobtainable flow
behaviour for structured liquids that seemed to have a yield stress. Measurement in
these ultra-low shear-rate regions is now called creep testing, by analogy with the
testing of solids under similar low-deformation-rate, long-time conditions; albeit
solids creep testing is usually performed in extension rather than in shear.
The reason why these types of rheometers lend themselves to measurement
of the whole curve is as follows: the complete range of shear rates needed to
encompass the whole range of flow can be very large, say 1010 decades. However,
the concurrent range of shear stress for the same conditions can be as low as 103. The
ability to apply this range of stress in a controlled fashion via the couple T applied to
a typical geometry is relatively easy compared to generating the equivalent range of
shear rates that would need to be applied. Measuring a wide range of shear rates is
much easier than generating them and this measurement is done very accurately
these days using optical discs.
11.5 The vane geometry
Last of all in our consideration of appropriate equipment to measure
materials that appear to have a yield stress, we must consider the steps necessary to
avoid wall-depletion or slip artefacts that very often arise in low shear rate
measurement of the kind of highly structured liquids that appear to have yield
stresses, see figure 11. Apart from roughening the walls of existing geometry
members, the most widely used geometry suitable for this purpose of eliminating
wall effects is the vane geometry (where the vane is used instead of an inner cylinder
of a viscometer / rheometer), with its diversity as to vane number and aspect ratio.
74
Chapter 11: Very shear-thinning or yield-stress liquids
It offers the possibility of inserting the thin vanes into an existing rested or stored
sample with the minimum of disturbance.
Figure 11: The vane and the vane-and-basket geometries.

Barnes [1] went further than simply using a vane when he introduced a closefitting wire-mesh cylinder inside the outer containing cylinder to prevent slip there
also, which can often occur, but is usually ignored, see figure 11. Of course, use of a
vane geometry at very high rotation rates is precluded due to secondary flow
developing behind the vanes.
Theoretical analyses of this geometry has been carried out by several workers
who all essentially concluded that for yield-stress/very-shear-thinning liquids, the
vane geometry acted as a circumscribing cylinder defined by the tips of the vane
blades, with the material inside the virtual cylinder essentially acting as a solid body,
and the material outside being sheared in the normal way [1]. This ensures that slip
is completely overcome at the rotating member.
11.6 Some flow equations with yield stresses
Many of the liquids whose flow curves we have just examined will obviously
appear to have a yield stress if scrutinised only over the shear-rate/shear-stress
region just above the large increase in viscosity. Much useful progress has been
made for these kinds of liquids by using simple, yield-stress-containing equations
that fit steady-state, shear-stress/shear-rate data quite well over a limited range of
shear rates. Then, if these equations are used to predict flow in any other situation,
where the same general range of shear rates applies, this is completely acceptable and uncontroversial, and can indeed be very effective.
The most popular equations that have been used to describe liquids with
yield stresses are the Bingham, Casson and Herschel-Bulkley (sometimes called the
generalised Bingham) models, i.e.
= o + & Bingham,
p
and
= o + & Casson,
p
= + k& n Herschel - Bulkley.
o
75
The Bingham model is the most non-Newtonian example of the Sisko model,
which is itself a simplification of the Cross (or Carreau model) under the appropriate
conditions [3].
A simple but versatile model that can cope with a yield stress, yet retain the
proper extreme of a finite zero and infinite shear-rate viscosity is the Cross model
with the exponent set to unity, so

1
=
,
0 1 + k&
the behaviour of which is shown in figures 12 and 13, plotted either linearly as stress
against shear rate or logarithmically in terms of stress versus shear rate or viscosity
versus stress.
3
60
oni
an,
10
Shear stress, / Pa
oni
an,
New
t
10
Approximately
constant stress, o/k
New
t
Shear stress, / Pa
10
40
30
20
10
10
6
10
50
10
-4
10
-2
10
.
Shear rate, / s
10
10
/ k = 10 Pa
0 0
-1
Figure 12: The Cross model with m = 1 and typical values of the
parameters, showing the apparent yield stress region
and upper and lower Newtonian regions.
200
400
600
.
Shear rate,
800
1,000
s-1
Figure 13: The Cross model with m = 1, plotted linearly

to show Bingham-type behaviour.
This behaviour is then very similar to figures 1 to 10, where we note the large drop in
viscosity for only a moderate increase in stress for all the results shown. Above the
yield stress where << 0, and k& >> 1 and it is easy to show that the equation
simplifies to
0
+ & ,
k
which is the same as the Bingham equation, but with 0 =
0
and p = 0 .
k
References
[1] Barnes, H. A, JNNFM, 81(1&2), 133 - 178 (1999).
[2] Sherman, P, Industrial Rheology, Academic Press, London, 1970.
[3] Barnes, H. A; Hutton, J F; Walters, K, An Introduction to Rheology, Elsevier,
Amsterdam 1989.
76
CHAPTER 12: THE FLOW OF SOLIDS

There are solids in rheology, even if they ... creep, Markus Reiner
12.1 Introduction
There is a widespread but mistaken belief that solids do not flow, but only
show elastic behaviour. However, all solids when stressed, while certainly
deforming elastically on a time-scale of less than a second, consequently creep (i.e.
continue to deform very slowly) over a very long time scale of days, months and
even years. If we replace the word creep with flow, we have moved smoothly
from material science into rheology, and we can happily talk about the viscosity of
solids, because, as we have seen so far, viscosity determines flow-rate.
Arnstein and Reiner [1] showed that solidified cements have viscosities of
1016 - 1017 Pa.s., while Sherman [2] later showed that ice-cream viscosities at low
stresses were in the order of 108 Pa.s. He quoted earlier measurements of the
viscosities of various English cheeses in the range 1010 - 1011 Pa.s, and for
concentrated oil-in-water emulsions with 65 % w/w disperse phase; the creep range
had viscosities of over 106 Pa.s. He even measured Madeira cake and came out with
viscosities of around 109 Pa.s! These kinds of numbers can be compared with the
more recent work of Harnett et al on the creep of butter [3], who found values of the
steady-state creep viscosity at 10 C to be in the range 2.4 11.0 x 1011 Pa.s. Similar
creep measurements were made on various kinds of solid soap by Pacor et al, who
found values of creep viscositydepending on formulation and moisture content
in the range 5 - 160 x 1011 Pa.s [4].
The approximate values of steady-state creep (zero-shear) viscosity for a
range of solid materials (mostly at room temperature) are shown in table 1, showing
typical examples for each viscosity decade from 1021 down to 106 Pa.s, arranged in
log10 decrements.
21 - the Earths mantle 17 - marble
13 - ice
20 - glass
16 - cement
12 - solid soap
19 - steel
15 - polyethylene 11 - asphaltic concrete
18 - mortar
14 - tin
10 - cheddar cheese
Table 1: Some steady-state creep viscosities of solid-like materials.
9 - bitumen
8 - ice cream
7 - resins
6 - gelatine gel
In these systems, we either have materials in a super-cooled, glassy state, and

hence their flow is the same as that of other liquids but is very slow, or else solid
polycrystalline materials where flow is usually via the integrated effect of atomic
jumping movements along grain boundaries.
77
12.2 Non-linear viscosity of solids

If solid materials are heated to temperatures of about half their melting point
(in degrees kelvin, K), they are still solid, but the long-term creep rates are then
considerably higher and usually within the measurable range, i.e. > 10-9 s-1. When
results from this type of experiment are collated, a simple picture emerges, so that
after a long-enough time, a steady creep rate is achieved; then
this creep rate, at low-enough applied stress, usually becomes directly
proportional to the applied stressin the language of the rheologist, this
proportionality signifies Newtonian behaviourand is governed by atomic
diffusion, while,
at higher stresses, the creep rate increases faster than the stress, and usually shows
a power-law-like behaviour.
Above the power-law region, other phenomena lead to different kinds of
creep behaviour, but among them is a situation where the creep rate (often quite
large by now) tends again towards a linear function of applied stressthe so-called
drag-controlled plasticityi.e. in rheological parlance an upper Newtonian
behaviour. If we express this response as would normally be done for a structured
liquid, the identical behaviour would be as shown in figures 1 - 9 of chapter 11.
The creep results of most solid materials, in the customarily observed linear
followed by power-law regions, can be fairly well described by the empirical Ellis
equation (omitting the usual high-shear-stress asymptotic viscosity), viz.
=
o
1

1+
c
where is the steady-state creep viscosity measured at a stress of ; is the zeroshear-rate viscosity; c is the critical stress which determining the region of shear
stress where the material shows significant departure from linearity, and m is
another material parameterthis time dimensionlesswhich determines the rate of
departure from linearity. (The results can be expressed equally well using the Cross
model, which is essentially the same as the Ellis model, but where the stress is
replaced by the shear rate).
Some further examples of this kind of behaviour are shown in figures 1 and 2.
16
20
10
14
100 microns
Viscosity, / Pa.s
Viscosity, / Pa.s
10
10
12
10
10 microns
10
10
1 micron
10
10
10
10
10
Stress, / Pa
10
10
Figure 1: Creep viscosity versus shear stress for magnesium oxide

at 1000 oC, for various values of grain size.
78
18
1000 microns
100 microns
10
16
10
10 microns
Typical conditions
in the Earth's upper
mantle
14
10
12
10
10
10
10
10
10
10
Stress, / Pa
Figure 2: Creep viscosity versus shear stress for olivine
(a rock) at 1000 oC, for various values of grain size.
Chapter 12: The flow of solids
The approximate parameters for other solid materials are collected together in table
2. The data has been read off graphs in Frost and Ashbys book [5] and has been
fitted to the Ellis model.
m
Material
Temp. C
0 Pa.s
0 Pa
Spinel, a mineral
1000
4.64 x 1016
7.4 x 107
1.2
16
7
1200
1.25 x 10
4.8 x 10
1.1
-Alumina
16
6
Zinc carbide
1600
2.00 x 10
1.5 x 10
2.9
Lithium fluoride
500
2.45 x 1014
2.7 x 106
4.6
14
6
Sodium chloride
400
3.02 x 10
1.7 x 10
4.3
304 Stainless steel 600
6.20 x 1015
2.5 x 107
3.6
15
6
Titanium
900
1.00 x 10
1.0 x 10
2.9
Ice
-8
1.02 x 1013
3.2 x 105
2.1
Table 2: Approximate values of the Ellis model parameters that
describe the creep of some solid materials.
Plasticine, which is considered to be a typical soft solid, also shows
characteristic non-Newtonian, steady-state liquid behaviour as seen in figure 3,
where the results of a creep test are compared to those obtained at the higher
deformation rates experienced in extrusion through a short orifice.
Viscosity, / Pa.s
10
10
10
10
12
controlled
stress
rheometer data
orifice
flow
data
10
10
10
10
Stress, / Pa
Figure 3: Flow curve as viscosity versus shear stress for
Plasticine at room temperature.
12.3 Conclusion
The conclusion of this chapter is that everything flows, and solids flow very
slowly [6]. Nevertheless their overall flow behaviour is essentially the same as that
of structured liquids. One thing worth noting about the degree of shear-thinning
seen for solids is that if we compare the steepness of the shear-thinning regions of the
very shear-thinning (yield-stress) liquids shown in chapter 11, they are much
steeper than for the solids shown here. This is because their viscosity often arises
from structures that are not space-filling (as for solids), but are composed of tenuous
arrangements such as chains and flocs that collapse above a critical shear stress to a
low viscosity, while solids are always space filling.
References
[1] Arnstein, A; Reiner, M, Civ. Engng., Sept 1945, p3.
[2] Sherman, P, Industrial Rheology, Academic Press, London, 1970.
[3] Harnett, M; Lambert, R.K; Lelievre, J; Macgibbon, A.K.H; Taylor, M.W; in
79
[4] Pacor, P; Lander, L.H; Papenhuijzen, J.M.P; Rheol. Acta, 9(3), 455, 1970
[5] Frost, H.J; Ashby, M.F, Deformation-Mechanism Maps, Pegamon Press, Oxford,
1982.
[6] Barnes, H A, Yield stress - a review, or - everything flows? JNNFM,
81(1&2), 133 - 178 (1999).
The mountains flow before the Lord
80
CHAPTER 13: LINEAR VISCOELASTICITY AND TIME EFFECTS

Time is nature's way of keeping everything from happening all at once, anon.
13.1 Introduction
In all structured liquids there is a natural rest condition of the microstructure
that represents a minimum-energy state. When these liquids are deformed, thermodynamic forces immediately begin to operate to restore this rest state, just like a
stretched spring will always seek to return to its unextended length. Also like a
spring, movement from the rest state represents a storage of energy, which manifests
itself as an elastic force trying to reproduce the static status quo. This kind of energy
is the origin of elasticity in structured liquids. Some examples of rest states are
unextended, isolated polymer coils with a random, undeformed configuration,

unstretched inter-entanglement polymer chains,
spatially random distribution of particles in a suspension at rest, or
undeformed particles in an unsheared emulsion.
Initially, the restoring force increases linearly with the distance that any
deformation takes the material away from its rest state, but eventually non-linearities
will be encountered. The rate of increase of force with deformation then diminishes,
until at very large deformations a steady-state situation arises and the elastic force
becomes constant. By the time that these large elastic forces are evident, the
microstructure has changed radically, and has become anisotropic. The elastic force
manifests itself at small deformations as various elastic moduli, such as the storage
modulus, and at steady state (i.e. very large deformations) by a normal-stress difference
(see chapter 14). Alongside these elastic forces are the ever-present viscous forces
due to the dissipation, and proportion to the rate not extent of deformation, so that
together these produce viscoelastic (i.e. viscous plus elastic) effects.
Most concentrated structured liquids shown strong viscoelastic effects at small
deformations, and their measurement is very useful as a physical probe of the
microstructure. However at large deformations such as steady-state flow, the
manifestation of viscoelastic effectseven from those systems that show a large
linear effectscan be quite different. Polymer melts show strong non-linear
viscoelastic effects (see chap. 14), as do concentrated polymer solutions of linear coils,
but other liquids ranging from a highly branched polymer such as Carbopol, through
to flocculated suspensions, show no overt elastic effects such as normal forces,
extrudate swell or an increase in extensional viscosity with extension rate [1].
If the stresses and strains in deformations are relatively large, then many time
effects are thixotropic in nature. The difference between the two situations
viscoelastic and thixotropicis that in the linear viscoelastic region the
microstructure responds over a certain time scale without changing, while in
thixotropy the microstructure does changeby breaking down or building upand
81
such changes take time. In this chapter we shall deal only with linear viscoelasticity,
and the resulting viscoelastic responses shown in various test situations.
Before embarking on an investigation of liquids that show obvious elastic
effects such as recoil and normal force, it is well to remember that all liquids show
elastic effects at a short-enough time or a high-enough frequency. Put in terms of a
Maxwell relaxation time (see next section), even what we would think of as simple
Newtonian liquids have elastic properties (see [2]). Table 1 shows a number of such
liquids compared to others that we would normally think of as elastic, giving the
modulus that would describe their behaviour at times less than their relaxation time,
which is also given. This is the converse of what we found in chapter 12, where we
saw that all solids flow, since here we are saying that all liquids show solid-like
properties under appropriate, if extreme, conditions.
Liquid
Viscosity
Modulus
Relaxation time
G / Pa
/ Pa.s
/s
Water
10-3
10-12
109
-9
An oil
0.1
10
108
A polymer solution
1
0.1
10
A polymer melt
105
10
104
15
5
A glass
>10
10
> 1010
Table 1: Values of approximate Maxwell parameters for a very wide range of liquids
(following Tanner [2]).
13.2 Mechanical analogues of viscoelastic liquids
For those not familiar with viscoelasticity, one of the simplest ways of
understanding the subject is to make use of simple mechanical models. These consist
of combinations of linear elastic and viscous elements, i.e. springs and dashpots, see
figure 1, that visualise the concepts that we discussed above.
G1
G
G
G2
Maxwell
Kelvin-Voigt
Burgers
Figure 1: The Maxwell, Kelvin-Voigt and Burgers models.
82
Chapter 13: Linear viscoelasticity and time effects
A spring is a representation of a linear elastic element that obeys Hookes lawi.e.

the strain is proportional to the applied stress. In a simple shear deformation the
constant of proportionality is the elastic modulus, G, so that
= G .
Notice that time does not come into this behaviour, so that if we apply a
strain to the unstrained model, the resultant stress suddenly appears, and then,
if we remove the strain, the stress falls immediately to zero (and vice versa when a
stress is applied).
Similarly, linear viscous response can be modelled using a dashpot.
A
plunger moving through a very viscous Newtonian liquid physically represents this.
The response of this element is described mathematically by the equation
= &.
Note that now only the rate of deformation matters, and if we switch on a
stress , the dashpot immediately starts to deform and goes on deforming at a
constant deformation rate & without any change with time until the stress is
removed, and then the deformation stops immediately (again vice versa if a shear
rate is applied). Note that again there are no time effects.
If we connect a spring and a dashpot in series we have the simplest
representation of a viscoelastic liquid, and such a combination is called a Maxwell
model, in honour of the famous Scottish physicist of the last century (1831 - 1879),
who introduced it in the 1870s.
If we connect the single elastic and viscous elements in parallel, we end up
with the simplest representation of a viscoelastic solid, and this has been named after
Voigt, the German physicist (1850 - 1919), while some workers call it the Kelvin or
the Kelvin-Voigt model, linking it with Lord Kelvin (1824 - 1907), another Scottish
physicist): to please as many as possible, we shall call it the Kelvin-Voigt model.
The Maxwell and Kelvin-Voigt models may be expanded to give either
multiple Maxwell models, which are usually combined in parallel, or else multiple
Kelvin-Voigt models, which are usually combined in series. However, if we combine
a Maxwell model and a Kelvin-Voigt model in series, we obtain a Burgers model,
named in honour of the Dutch physicist (1895 - 1981) of that name. This model is the
most complicated that we need to consider, since it describes all the basic features of
interest to us. Below we will see how all these various combinations of springs and
dashpots begin to describe real viscoelastic materials.
Of course, looking down a microscope, we would never see real springs and
dashpots, so what are the real physical entities being modelled by them? First, as we
saw above, elastic elements represent any physical system in the microstructure that
can store energy. This might be potential energy or entropic energy, so a bond or a
polymer segment being stretched would be an example of the former, while an
isolated polymer random coil being deformed from its spherical rest state being but
one example of the latter. A spring can also represent many other energy-storing
mechanisms. In each case, the movement of any entity through a liquid continuous
phase will dissipate energy and this can be modelled by a dashpot. Even in a
polymer melt where there is no liquid around the chains, there is friction as the
chains slide over one another.
To use these mechanical models we need to establish for any particular liquid
the number and configuration of springs and dashpots.
83
13.3 Measuring linear viscoelasticity

13.3.1 General
There are a number of ways of measuring linear viscoelastic response. One of
the simplest is the sudden application of a constant stress to the liquid being tested,
and the monitoring of the resulting strain thereafterthis is called creep testing.
Another frequently used method is oscillatory testing, i.e. applying an oscillating
stress or strain as an input to the liquid and monitoring the resulting oscillatory strain
or stress output. The same repetitive sinusoidal straining motion recurs over and
over again, with each cycle taking a certain time, and having a frequency that is
inversely proportional to that time.
One other method used occasionally is the sudden application of a constant
strain, and the monitoring of the consequent stress, which then decays away with
timethis is called a stress relaxation test. The only other test we need to mention
here is the application of a steady shear rate, as in the start-up of a simple viscometer
so not surprisingly this is called a start-up test. All these tests are shown in figure
2. The first two tests mentionedcreep and oscillatorywill now be dealt with at
some length, since they are the most frequently used tests, while the other two
stress relaxation and start-upwill only be mentioned briefly.
creep
apply a constant stress,

monitor resulting strain
oscillatory
apply an oscillatory stress/strain,

resolve resulting strain/stress
stress
relaxation
apply a constant strain,

monitor relaxing stress
start-up
apply a constant shear rate,

monitor resulting stress
Figure 2: Various tests that measure the time response

of a liquid.
13.3.2 Creep tests
For many years, metallurgists performed tests in which they stressed long,
thin samples of metal, and measured the (usually small) resultant deformation as a
function of time. In some cases the testing time was very long (months or even
years!), and they then found that the samples were still deforming long after the
initial elastic response; however the deformation they measured was so slow that the
phenomenon was called creep. Creep properties are very important in studying
certain practical situations where high stresses and long times (and sometimes high
temperatures) are involved, such as suspension bridges or pressurised nuclearreactor vessels. This kind of stress-controlled test was later used for softer materials
such as ice cream, butter and soap, and the name creep stuck, so that when the new
84
generation of controlled-stress instruments for structured liquids came along, they

had testing regimes called creep analysis.
The behaviour normally seen for typical viscoelastic liquidsrepresented for
convenience and simplicity by a Burgers modelin a creep test is shown in figure 3.
behaviour
Deformation,
e 0
delayed steady- state

(ii) elastic (iii) viscous
response
immediate response
(i) elastic
response
Time, t
equation
t
(t)
1 1 (1 - et/)
+
+
=
1
G 1 G2
model
Figure 3: The Burgers model in detail.

There is an initial elastic response, which for gel-like liquids can produce an
unwanted instrument artefact called ringing that arises from the large inertia of the
moving member of the rheometer. Thereafter there is a so-called delayed elastic
response where the deformation rate becomes slower and slower, ending up as a very
slow but steady-state deformation at the longest times, i.e. the material is in steady
flow. Sometimes it is difficult to be sure when this steady state has been reached. If
the controlled-stress instrument used does find it difficult to detect the steady-state
slope for structured liquids, it often measures a slope that is too steep, and the typical
result of this common artefact is seen in figure 4, where the long-time, steady-state
viscosity appears to decrease with decreasing shear stress.
The answer to this problem is to tighten the level of scrutiny with which the
instrument decides if steady state is achieved, and then the normal viscosity plateau
is seen.
The universal curve that describes the creep response of most structured
liquids and gels is shown in figure 5, with some examples shown in figure 6. The
behaviour is depicted over a very long time period, which, as ever, is best shown on
a logarithmic scale. The response for most materials measured under normal
conditions is such that we first see the immediate elastic response (the horizontal
dotted line), then we see the delayed elastic response, and eventually the viscous
flow predominates, and the slope approaches unity at very long times (see the
second dotted line).
85
ap
en
par
t slo
pe
actua
l slop
e
Viscosity (log scale)
Creep deformation, J(t)
Stress (log scale)
shearthinning
begins
here
Time, t / s
Figure 4: The artefact found in creep testing where the

rheometer reads the creep slope too early.
-1
slope = 0
elastic
response
Time, t (log scale)
viscous
response
Creep compliance, J(t) / Pa
1
=
pe
slo
(log) Creep compliance, J(t)
-4
10
-5
10
-6
10
PS, 105 oC
-7
10
-8
10
PC, 142.5 oC
-9
10
10
-4
10
10
10
Time, t / s
Figure 5: A typical creep curve plotted on a logarithmic basis. Figure 6: Creep curves for polystyrene and polycarbonate
melts [3].
13.3.3 The behaviour of model materials in creep tests

In a creep test, a simple elastic solida springshows an immediate
response to give a constant deformation (i.e. strain). On the other hand, a simple
Newtonian liquida dashpotwould show an ever-increasing strain, which
displayed on a graph of strain against time would be a straight line starting at the
origin, with the slope giving the shear rate, & .
If we now perform a creep test on a Maxwell model, the behaviour can be
described by the simple equation
1
t
= + ,
G
which, at very short times, is characterised by an immediate elastic response, = /G,
and at very long times, when t >> /G, by simple viscous behaviour, = t/. Here
/G is called the relaxation time .
If a creep test is performed on a Kelvin-Voigt model, the strain gradually
builds up to a constant value as described by
86
t

1 e .
Here is called the retardation time, since it characterises the retarded response of the
model, and its value is again given by /G. At very short times, the response is
viscous, and ~ t/. At t = , the strain has risen to ~ 63 % of its final, asymptotic
value of /G.
If we now stress the Burgers model in a creep test, it is easy to see what will
happen, see figure 3: first, the unrestrained spring G1 will instantaneously deform to
its expected extent, while the isolated dashpot will start to deform at its expected
rate. However, the spring in the Kelvin-Voigt element cannot immediately respond,
being hindered (i.e. retarded) by its dashpot. Nevertheless, it does begin to deform,
and eventually comes to its expected steady-state deformation. It is possible to
show that the overall deformation can be written down as:
1
1
=
+
G
G
2
1
1 e + t
where is the retardation time, now given by 2/G2.

Some examples of real materials whose creep deformation at low stresses are
described quite well by the Burgers model are
concentrated corn starch dispersions, see Liao et al. [4]
titanium dioxide dispersions, see Zupancic [5]
liquid crystals, see Shi et al [6]
Any number of extra Kelvin-Voigt elements can be added in series within the
Burgers model, and each will add one extra term to the creep equation, so that most
practical creep curves can be reasonably described, with the behaviour at the very
shortest times captured by G1 and at the longest times in steady flow by , so
t
t
1
1
1
(t )

1 e 3
= J (t ) =
+
1 e 2 +
G1
G2
G3
t
+ (...)... + .
If the long-time behaviour predicted by this model is extrapolated back to

zero time, the intercept is . If this is divided by the applies stress, we get the socalled equilibrium compliance, Je, which can be related to the elastic elements we
used by
Je =
1
1
1
o
=
+
+
+...... .
G1 G2 G3
In fact this extrapolated line is exactly the behaviour of a single Maxwell model,
where we have only a spring and a dashpot in series, with the spring elasticity G
given by
87
1
1
1
1
=
+
+ ..... ,
G G1 G2 G3
and a viscous element still given by . The deformation is then given by
Je =
1
t
+ .
G
-1
10
Viscosity, / Pa.s
Compliance, Jo / Pa
-1
For many applications, Je and are the most useful data to collect, the former being a
measure of elasticity at short times and the latter a measure of steady-state
deformation at long times. Both eventually become non-linear at large stresses,
figure 7 shows a typical example.
20%
40%
60%
-2
10
-3
10
10
10
Stress, / Pa
10
10
10
60%
40%
20%
10
10
Stress, / Pa
Figure 7: Non-linear creep properties of oil-in-water

emulsions with various amounts of oil [7].
13.3.4 Oscillatory tests, or mechanical vibrational spectroscopy

As well as doing creep tests over a range of time, we can equally perform
oscillatory tests over a range of frequency. Then short times correspond to high
frequencies, and long times relate to low frequencies. We shall find that for real
materials, just as the behaviour at short time is dominated by an elastic response, so
too is the behaviour at high frequency, and likewise the viscous response that
dominates at long times in creep, now shows up as being most import at low
frequencies.
If we perform an oscillatory test by applying a sine-wave-shaped input of
either stress or strain, we can then, using suitable electronic methods, easily resolve
(i.e. separate) the resulting sinusoidal strain or stress output into a certain amount of
solid-like response, which is in phase with the input, and a corresponding amount of
liquid-like response which is /2 (i.e. 90) out of phase with the input, see figure 8.
The solid-like component at any particular frequency is characterised by the
storage modulus, G, and the liquid-like response is described by the complementary
loss modulus, G. The units of both these moduli are pascals. The values of both
these parameters vary with applied frequency, , which is given by 2f, where f is
the frequency in hertz (Hz). As all modern rheometers are well able to handle the
complicated mathematical calculations necessary for determining these moduli, we
do not need to consider the calculation procedure here.
88
ideal -
Maxwell
input
(log) modulus
(log) modulus
oscillatory drive
output
strain
time
f = 2
time
lag
phase
measured couple
analysis
stress
G'
G''
(log) frequency,
typical
G'
G''
(log) frequency,
Figure 8: A schematic representation of oscillatory testing.

Note that the G and G data produced by rheometers is sometimes reported
in different ways, as for instance
the dynamic viscosity
' = G/ ,
the loss tangent
tan = G/G,
the complex modulus
G* =
G' 2
* = ( ') + 2 .

G '2 + G ''2 ,
1
the complex viscosity
Note: It is important to realise that there is no extra information in any of these other
derived parameters compared with that which is contained in the basic G and G
information. Hence, we shall only be using these basic G and G parameters.
In the same way that instrument mechanical inertia can cause problems at
short times in creep tests, so too can it cause difficulties at high frequencies in oscillatory
tests, especially with respect to the G data. Even fluid inertia can become important
at the latest highest frequencies of around 100 hertz. Fortunately, in good
rheometers, these effects are dealt with in the software and the displayed results
have had the inertial components eliminated, see figure 9 for a typical example of
viscoelastic results corrected for inertia.
13.3.5
Oscillatory tests with a Maxwell model
If we oscillate an ideal spring, at all frequencies G = G the spring modulus,

and G is zero. If, on the other hand, we oscillate a perfect dashpot, the loss
modulus is given by G = , and the storage modulus G is always zero.
However, the response of the simplest combination of these two elements in series
89
Storage modulus, G' / Pa
0.5
G'
0
-0.5
ted
rec
r
o
c
1.0
' = G''
0.5
-1.0
-1.5
10
20
30
40
50
Dynamic viscosity, ' / Pa.s
-1
Frequency, / s
Figure 9: Inertial effects in oscillatory flow

illustrated for a silicone fluid [8].
the Maxwell modelgives a far more complex and interesting behaviour over a
range of test frequencies, as shown in figure 10, which is given mathematically by
G ( )
and G ' =
2
2
1 + ( )
1 + ( )
2
G' ' =
G'
G''
Modulus, G' or G'' / Pa

(log scale)
where the relaxation time = / G.
Frequency, / s
-1
(log scale)
Figure 10: Typical Maxwellian behaviour in oscillatory

testing, with the crossover point indicated.
At very low frequencies, G is much larger than G, and hence liquid-like

behaviour predominates. However as the testing frequency is increased, G takes
over and solid-like behaviour prevails. The determinant of which kind of behaviour
is most significant is the value of the test frequency relative to the relaxation time .
This is a simple way to define a Deborah number, Dethe ratio of the relaxation time
to the test timeand a measure of De in this case is . Hence, low Deborah numbers
always indicate liquid-like behaviour, whereas high Deborah numbers means solid-like
response. At the midpoint, where G goes through a maximum; G = G, and this
takes place at a critical crossover frequency of = 1/.
As shown above, we can define the dynamic viscosity = G/, and then for
the Maxwell model we see that
90
' G'
+
=1
G
' = G '.
or put another way
This shows a linear relationship between and G over a range of frequencies,

which is a quick way to check if data really corresponds to the Maxwell model!
13.3.6 Double or multiple Maxwell models in oscillation
If we combine two Maxwell models in parallel, we begin to see features that
will later become quite familiar as we examine practical examples of oscillatory
curves. The overall behaviour is seen in figure 11, and is the simple sum of the two
Maxwell units, so
G ( )
G ( 2 )
G' = 1 1 2 + 2
.
2
1 + ( 1 )
1 + ( 2 )
2
1
2
G' ' =
+
2
2
1 + ( 1 )
1 + ( 2 )
and
Modulus, G or G / Pa
If the two relaxation times are different by more than a decade, then at very low
frequencies 1 dominates, and at very high frequencies, the elastic response is
governed by G1 + G2.
10
10
10
-1
10
-2
10
-3
10
-4
G''
G'
-3
10
10
-2
-1
10
10 10 10
Frequency, / s-1
10
10
Figure 11: Typical oscillatory response of a double Maxwell

model, with both spring moduli set at unity and the
time constants set at 1 and 0.1 seconds respectively.
The last in our series of models is the multiple Maxwell model obtained by
putting any number of Maxwell elements in parallel. The multiple Maxwell model is
to oscillatory flow what the multiple Burgers model is to creep testing, in that it
begins to represent real systems in a sensible way. The values of the overall storage
and loss moduli, G and G, at any frequency , are then given by the sum of i
contributions from i Maxwell elements
G ' ( ) =
Gi
i
2 i2
and G ' ' ( ) =
1 + 2 i2
G 1+
i
2 2
i
Here the dynamic response of a viscoelastic liquid over a range of frequencies

can be modelled by choosing a range of Maxwell elements with appropriate values of
G and chosen to cover the range used in the experiment for which G and G values
are available, as say = 10-3 s, 10-2 s, ... 103 s for a frequency range of 103 down to 10-3
rad/s, together with the appropriate values of Gi. It is obvious that at very high
91
frequencies the overall storage modulus G will be given by G1 + G2 + G3 ..., while at

the very lowest frequencies, the overall dynamic viscosity is given by 1 + 2 + 3
..., though 1 will usually dominate.
13.3.7 Oscillation with a Kelvin-Voigt model
The behaviour of a Kelvin-Voigt model is quite simple, since G = G and G
= , or ' = . We will not discuss the combination of Kelvin-Voigt models in
oscillatory flow, since this is rarely considered. However, as we shall see later,
Kelvin-Voigt-type behaviour is seen in at least one part of the viscoelastic spectrum
of real materials.
13.3.8 The oscillatory response of real systems
Modulus, G' or G'' (log scale)
The most general overall G, G response of real examples of structured

liquids is shown in figure 12. The exact values of the moduli and their position in the
frequency domain will of course vary, but the indicated overall qualitative behaviour
is usually seen if data is available over a wide-enough frequency range. A number of
specific regions can often be differentiated, namely
viscous/
terminal
transition
to flow
glassy
G''
leathery/
rubbery/
transition
plateau
Frequency, (log scale)

Figure 12: The various regions in the viscoelastic
spectrum of non-Newtonian liquids.
(a) the viscous or terminal region, where G predominates and viscous (flow)
behaviour prevails. All materials have such a region, even solids (because they creep
at long times), but the frequency where this is seen is often so low that most
oscillatory instruments cannot detect this part of the curve. At low-enough
frequencies, G is linear with increasing frequency and G is quadratic, and the
longest relaxation time, max is given by G/(G).
(b) the transition-to-flow region is so called because, when viewed from higher
frequencies (where elastic behaviour dominates and G > G), the loss modulus G ,
describing viscous or flow behaviour, becomes significant. The point where the two
moduli cross over is sometimes noted, and for a Maxwell model, this crossover
frequency is given by the inverse of the relaxation time .
(c) the rubbery or plateau region is where elastic behaviour dominates. While
in many cases we see what appears to be a flat plateau, there is always a slight
increase of G with frequency, but it can be as small as a few percent increase in
modulus per decade increase in frequency. The value of G is of course always
lower than that of G, but sometimes it can be considerably so. When the slope of the
92
storage modulus versus frequency curve (G,) is small, the value of G decreases
with increase in frequency, towards a minimum before rising again. The lower the
slope of the (G, ) curve, the deeper is the G valley, so that sometimes at the
minimum point, G/G is greater than 15. This fact is a good example of a very useful
approximation in linear viscoelasticity, that is, at any particular frequency
G d ln G
2 d ln
where the second term is the local slope of the logarithmic plot of G versus at that
frequency. If we express this slope as m, then we can also write
tan
m
.
2
(d) a leathery or higher transition crossover region is also seen, where, due to
high-frequency relaxation and dissipation mechanisms, the value of G again rises,
this time faster than G. Once more at G = G, a crossover frequency can be defined,
from which another characteristic time can be obtained.
(e) at the highest frequencies usually encountered in this form of testing, a
glassy region is seen, where G again predominates and continues to rise faster than
G.
When the typical range of frequencies available on most rheometers is used,
say 10-2 to 102 rad/s, it is quite usual to see only two of the regions described above.
The particular ones seen depend largely on the longest relaxation time, max , of the
material being tested, so that if max ~ 1, then the viscous and transition-to-flow
regions are seen. If max > 1, then often only the plateau region is seen, perhaps with
the higher transition point before the glassy region. On the other hand, for many
dilute polymer solutions, max < 1, and G always predominates.
If the range of testing is artificially extended by using different temperatures
(the so-called time/temperature superposition principle), then for many materials, it
is often possible to see all the regions described by superimposing the data using
suitable shift factors, see Ferry [9].
The behaviour of a real system, especially a gel or very high molecular weight
polymer melt, can be simplistically modelled by dividing it into Maxwell behaviour
at low frequencies giving way to Kelvin-Voigt behaviour at high frequencies, see
figure 13. In both cases we have crossover frequencies where the critical time is
given by = /G from the Maxwell or Kelvin-Voigt models respectively. In the
intermediate region, obviously elastic behaviour dominates, and we have the plateau
region seen for so many structured liquids, melts and gels.
We will now show various practical examples of viscoelastic behaviour in
oscillatory flow.
93
Modulus, G' or G'' (log scale)
KelvinVoigt-like
Maxwell-like
G''
G'
G'
crossover
frequency
G''
crossover
frequency
plateau
region
Frequency, (log scale)
Figure 13: Low and high frequency oscillatory behaviour

shown as Maxwellian and Voigt-like behaviour.
13.3.8.1 Viscous/terminal region - simple Maxwell behaviour
There are two specific examples of liquids that often show simple Maxwelllike behaviour within the normal measuring range (10-2 - 102 Hz), with the expected
shapes of the G and G curves and a single relaxation time, viz. associativethickener-type polymers (see chapter 16 for details) and worm-like surfactant
micellar systems, otherwise known as threadlike or rod-like micelles (linear or
branched). (The latter are also called living polymers because if they break under
large stresses, they can reform under conditions of rest or low stress (see chapter 18).)
The typical response of such a system is shown in figure 14, where a representative
relaxation time would be around 1 s, see [10].
10
10
10
10
G'
G''
10-2
10-1
100
101
Frequency, / s
102
103
-1
Figure 14: Oscillatory behaviour of an associative thickener.
Some particular examples of these two kinds of systems are:

associative polymers such as hydrophobic ethoxylated urethane (HEUR), see
Annable et al [11].
aqueous surfactant solutions containing thread-like micelles, see [12-16].
Occasionally other systems have also been reported as showing simple
Maxwell behaviour, such as
an water-in-oil emulsion, see Pons [17]
the childs toy Bouncing Putty , see Benbow et al [18] with, ~ 105 Pa s, G =
2.5 x 105 Pa.
94
some organic glasses, Benbow [19].
There are also some systems that show near-Maxwell behaviour, and an
example is given in figure 15, where in this case a surfactant phase responds with a
major relaxation time of around 10 s which is responsible for the prominent
maximum in G, see [20]. However the contributions from some other faster
relaxation mechanisms give a slower-than-expected fall-off of G with increase in
frequency.
10
10
10
10
10
G'
G''
-4
-2
10
10
10
10
10
-1
Frequency, / s
Figure 15: Oscillatory behaviour of the cubic phase of a

branched nonionic detergent.
13.3.8.2 Polymer systems

The particular oscillatory regions seen for polymer systems depend on their
concentration and molecular weight, as well as the range of frequencies used. The
behaviour of concentrated low-molecular-weight systems is shown in figures 16 and
17, where at low frequencies we see the characteristic viscous/terminal regions
which correspond more or less to Maxwell-type behaviour, with G dominating and
at the lowest frequencies being proportional to frequency, but eventually curving
over. This feature of the behaviour moves to lower and lower frequencies as the
concentration is increased, so that more and more plateau-like behaviour takes over
within the experimental window (10-2 - 102 Hz).
6
10
Moduli, G' or G'' / Pa
10
G
G
10
0.5%
G
G
0.4%
-1
10
0.3%
-2
10
10
G''
4
10
10
10
-3
10
-2
10
-1
10
Frequency, / s
10
10
-1
Figure 16: Oscillatory behaviour of aqueous xanthan gum

solutions, at various weight percentages, [21].
10
-3
10
-1
10
10
10
-1
Frequency, / s
10
Figure 17: Oscillatory behaviour of low molecular weight

polyethylene melt at 150 oC.
For higher molecular weight polymers, whether melts or more concentrated

solutions, a pronounced plateau region is seen. Examples of increasing molecular
weight and resulting deepening of the G valley and the introduction of other
95
associated features is shown in figures 18 - 21. Figures 18 20 are redrawn from

Haake Instruments promotional literature, and figure 21 arises from reference [22].
8
10
10
10
G''
10
10
G'
-2
10
-2
10
10
10
10
10
10
10
10
10
10
10
G
G''
-4
-2
10
10
10
10
10
10
-1
Frequency, / s
-1
Frequency, / s
Figure 18: Oscillatory behaviour of polydimethylsiloxane Figure 19: Oscillatory behaviour of 0.4 M broad molecular
o
o
weight polystyrene at 150 C.
of molecular weight 133,000 at 20 C.
7
10
G
10
10
G''
10
10
10
G
3
G'
10
10
-3
10
10
-1
10
-4
10
-2
10
10
-1
Frequency, / s
10
10
Frequency, / s
Figure 20: Oscillatory behaviour of PMMA, Mw = 0.12M,

o
polydispersity 1.08, at 185 C.
Figure 21: Oscillatory behaviour of narrow molecular weight

o
polystyrene, mol. weight 0.39 M, at 160 C.
10
10
375k
375k
132k
10
10
132k
75k
75k
10
Loss modulus, G'' / Pa
This deepening of the G valley with increase in molecular weight is particularly

well illustrated in figure 22.
-7
10
-2
10
-1
Frequency, / s
10
-7
-2
10
10
-1
Frequency, / s
Figure 22: Oscillatory behaviour of a polypropylene melts of

various molecular weights.
96
10
-1
10
10
G
(eff) = 0.783
10 G''
10
10
G''
-3
(eff) = 0.712
G
G''
(eff) = 0.638
G'
10

3
10
-1
10
G'
G''
10
10
10
10
-3
10
-1
Frequency, / s
Figure 23: Oscillatory behaviour of 84 nm PMMA particles

in decalin, at various effective phase volumes, [23]
10
10
Frequency, / s -1
Figure 24: Oscillatory behaviour of a nonionic hexagonal
liquid crystalline phase, [24].
13.3.8.3 Dispersions
All dispersions show some viscoelastic behaviour, and hence a measurable
value of G in the viscous region; figure 23 shows such a situation where G and G
begin to rise as expected at low frequencies, with G proportional to and G
proportional to 2, but as the concentration rises one moves to a plateau situation, see
also figures 24 - 26, where the situation is very analogous to concentrated polymer
solutions. Similarly, the possibility of a minimum in G and an apparent plateau in
G in a plateau or rubbery region is often seen, as in figures 27 - 32 showing the
response of a range of concentrated dispersions and gels.
5
10
10
G'
10
normal
G''
gelled
10
10
G'
G''
G''
1
10
10
-3
10
-1
10
-1
-1
10
10
Figure 26: Oscillatory behaviour of a typical oil-in-water

emulsion and a gloss paint [26].
G'
2
G''
G'
4
10
G''
10
10
Frequency, / s
Figure 25: Oscillatory behaviour of two toothpastes, [25].
10
10
Frequency, / s
10
gloss
paint
10
10
-1
10
10
o/w emulsion
-2
-1
10
10
Frequency, / s
10
-1
Figure 27: Oscillatory behaviour of a 3% by weight Laponite

dispersion in demineralised water.
10
10
-2
-1
10
10
10
10
Frequency, / s -1
Figure 28: Oscillatory behaviour of a lubricating grease at
room temperature, [27].
97

4
G'
3
10
G''
2
10
-3
10
10
-1
Frequency , / s
10
10
10
-2
30%
25%
20%
G'
~ x 100
G''
10
-1
10
-1
Frequency, / s
10
Figure 30: Oscillatory behaviour of a highly charged latex

suspension, at three volume fractions, [29].
10
-3
Figure 29: Oscillatory behaviour of a Bentonite,

polymer and aqueous electrolyte solution, [28].
G'
G''
1
10
10
10
G'
10
10
G''
-1
10
10
extrapolated
10
-5
-1
10
10
-1
Frequency, / s
10
Figure 31: Oscillatory behaviour of a polymer latex suspension;

particle size 0.4 microns and phase volume 60%, [30].
10
-3
10
-1
10
10
Frequency, / s-1
10
Figure 32: Oscillatory behaviour of a 5% crosslinked starch gel,

[31].
13.4 What can you get from oscillatory data?
plateau modulus, Gp
G'
2
1
G''
cross-over
frequency
Modulus, G' or G'' / Pa (log scale)
In qualitative terms the oscillatory curves give a fingerprint of the state of the
microstructure in the same way as does an NMR or an infrared spectrum. Just as
various mechanisms can be identified and quantified by the position on the
frequency axis and the height of the signal so too can oscillatory curves. A number
of examples are follow:
(a) At low-enough frequencies, it is possible to obtain 0, the zero-shear-rate
viscosity, since in that region, G = 0, see figure 33. This can be used in a number
of microstructural calculations, as well as for stability predictions. Also, the point
where G departs from being a linear function of frequency is a measure of the
departure of the steady-state (,) curve from linearity, at a shear rate given by the
same numerical value of the frequency in radians per second.
Frequency, / s
-1
(log scale)
Figure 33: Important features of low-frequency oscillatory

data.
98
(b) The crossover frequency at which G = G is a measure of the longest

relaxation time, and is given by the inverse of the numerical value in radians per
second. This can easily been shown as the case when the Maxwell model is used,
where the crossover coincides with the maximum value of G and at a frequency
such that = 1. This has been used as a measure of molecular weight.
(c) If the viscous-to-rubbery region does not overlap the transition-to-glassy
region, then a plateau modulus Gp can often be identified where the modulus G is
nearly constant. This too can be seen from the use of a single or a multiple Maxwell
model, where the modulus is dictated by purely elastic elements. In the ideal case
the loss modulus is then much smaller than the storage modulus by over an order of
magnitude. This is not always the case, and the plateau modulus is often defined as
the value of G when G is a minimum. In fact, the flatter the plateau in G, the
deeper will be the valley in G curve beneath it. If G does not have a deep
minimum, then sometimes one should look for a minimum in G/G (= tan ) instead,
to define the frequency necessary to evaluate Gp.
Once a plateau modulus Gp has been evaluated, it can be used for a number
of purposes, see figure 33. First, for polymeric systems, the value of the molecular
weight of the chain segments between temporary entanglements, Me, can be
evaluated, since Me = RT/Gp where R is the universal gas constant (8.314 J mol-1 K-1),
is the polymer density and T is the absolute temperature, see figure 34.
In disperse systems, the inter-particle force may be evaluated from the value
of Gp, since the plateau modulus is given by a mathematical function that contains
the radial distribution function and the force that particles exert on one another as a
function of their separation, see [32]. If one makes various approximations to the
radial distribution function such as a delta function or a cubic lattice, then the force
between the particles can be calculated. An example of the use of the plateau
modulus for emulsions is given by Sanchez et al [33].
At frequencies above the plateau modulus and loss modulus minimum, other
more complicated mechanisms begin to dictate the values of the moduli as both
moduli increase again. These are beyond the scope of this book, but anyone
interested should read Ferrys excellent book [35].
6
10
Gp, plateau
modulus
G'
10
x15
G''
10
10
10
-2
10
10
10
-1
Frequency, / s
10
10
Figure 34: Oscillatory behaviour of atactic polypropylene at

463 oC, [34].
(d) In some cases it is difficult to measure the steady-state viscosity and

normal forces for very elastic liquids, since they often flow out of the measuring
geometry or else fracture. In this case it is possible to use oscillatory data to
reconstruct the steady-state data, at least for polymeric systems, see [1] and [36].
99
However, this is not possible (and often not necessary) for disperse systems such as
suspensions or emulsions, where the elastic forces in steady state are much smaller
than the equivalent values in oscillatory flow. In some cases such as a flocculated
suspension, the very elastic structure at low stresses gives way to a low viscosity,
inelastic liquid at high stresses.
13.5 The relationship between oscillatory and steady-state viscoelastic parameters
13.5.1 General considerations
The following identities are always true for viscoelastic liquids
( ) 0 = (& )& 0 and
G ' ( )
N (& )
= 1 2
.
2
2& & 0
0
These relationships mean that we can extend any oscillatory plot if steady-state data
such as zero-shear-rate viscosity from a creep experiment is available (or vice versa).
This then dictates the G curve, at least to the shear rate where the viscosity departs
from , which in turn prescribes the frequency at which G departs from .
There are various other formulas of this form that can be used for suitable
systems over a wider range of shear rates/frequencies, see [37] for details. By and
large they fit polymeric liquids, but often fail when used with colloidal systems.
13.5.2 Creep versus oscillatory testing
Which experiments are best, creep or oscillatory? This question is often
asked, and the answer isit depends! First, what do we need the answers for? If we
need to look at the long-term behaviour, creep is usually the best kind of test, since
the equivalent very low frequencies are difficult to generate. If on the other hand we
want to assess quantitatively how wobbly a material looks in a bottle, oscillatory
testing is ideal. Another important point is that the amplitude of oscillatory testing
can be limited, and the test will not exceed the critical amplitude where nonlinearities set in, and the material is broken down. This can take some time to
recover due to thixotropy. Hence slowly increasing the amplitude of testing and
carefully monitoring the output will define the linear region. In creep testing, the
amplitude increases relentlessly, and could easily take the material into the nonlinear region, see below.
13.5.3 A simple link between creep and oscillatory parameters
Is there any relationship between results from creep and oscillatory
experiments? In other areas of science, similar tests can be carried out that utilise
either time or frequency as parameters, e.g. NMR and various other spectroscopic
techniques, which are analogous to this form of mechanical spectroscopy. As in all
these areas, the relationship between data in the frequency and time domains is
complicated, see such books as [38]. However, in simple terms, we can write down
a relationship as follows according to a very approximate formulation worked out by
the author, see table 2.
100
Range of m
m 0.35
(solid like)
G at = 1/t
1/J(t)
G at = 1/t
m 0.35
(liquid like)
(1 m )
.
2 J (t )
1/J(t)
1.85 m
J (t )
Table 2: Approximate relationship between creep

and oscillatory parameters, where m is the local
log/log slope of the creep curve.
These approximate relationships give us a way of comparing and even
predicting results between the one test and the another, at least to the extent of being
able to check if the results are in reasonable agreement. The correspondence between
the inverse of the creep curve and oscillatory behaviour becomes obvious by
comparing figures 5 and 6 with figure 12.
13.6 Non-linear response in oscillatory testing
10
10
In oscillatory testing, great care has to be taken to ensure that the experiment
takes place in the linear region where G and G are independent of the maximum
amplitude or stress of the oscillation. A number of examples of non-linear testing are
shown in figures 35 - 40 where we see that the most sensitive parameters to breakdown is usually G, since it eventually falls faster, whereas when G does change it
can increase before decreasing.
The extent of the linear region can vary from a fraction of one per cent strain
up to a strain of unity for certain rubbery systems. In most cases the strain at which
non-linearities first become apparent are nearly independent of frequency. This
means that for controlled-stress experiments, results can be difficult to compare with
corresponding controlled-strain results unless both sets of results are reduced to the
same basis.
G'
G''
10
10-2
10-1
Strain,
100
101
Figure 35: Non-linear oscillatory behaviour of a hair gel

at 3 Hz, [39]
10
10
10
10
--- Carbopol
Laponite
G
G''
10
-2
10
-1
10
10
Strain,
Figure 36: Non-linear oscillatory behaviour of a Laponite
suspension at 1Hz compared with a similar
dispersion of Carbopol.
101
G'
10
G''
1
10
100
-2
102
-1
10
10
G''
101
-1
10
10
Strain,
Figure 37: Non-linear oscillatory behaviour of Bonjela,
a commercial oral preparation, at 1 Hz.
10
10
Stress, / Pa
10
Figure 38: Non-linear oscillatory behaviour of a commercial

sample of ketchup, see [40].
10
G'
10
10
G
4
10
G''
10
10
103
10
G'
G''
10
10
-4
10
-3
10
-2
10
Strain,
10
-1
-3
10
10
-2
10
Strain,
10
-1
Figure 40: Non-linear oscillatory behaviour of a Bentonite

suspension.
Figure 39: Non-linear oscillatory behaviour of a strong flour

dough at 10 Hz.
13.7 Using oscillatory data to monitor curing, and rebuilding after shear
Oscillatory testing is ideal for following the rebuilding of the microstructure
of sheared systems, or for tracking the setting or curing of gelling systems. This is
because when we operate in the linear G and G regions, the amplitude is small
enough so as not to interfere with the microstructure or the curing/setting
mechanism. An example is shown in figure 41, where the response changes from a
loss-modulus-dominated (G) viscous situation at short times, to a storage-modulusdominated (G) elastic situation at long times.
3
10
10
10
10
G'
-1
10
-2
10
10
10
10
Time, t / s
Figure 41: Oscillatory parameters during the gelation of a
guar-gum/borate dispersion.
Obviously the gelling times are crucial in, for instance, dental applications,
where patients will not tolerate very long curing times for moulding materials in
102
10
10
10
10
10
10
good
bad
their mouths. Besides, simple dental economics would dictate a reasonable time
which is long enough to manipulate the setting material around the teeth, but is not
too long for eventual curing, see figure 42 for an example of a good and bad material.
Epoxy resins can be allowed to set over longer times if the strengths produced are
better with slower setting, see figure 43, or again, time might be at a premium.
An example of rebuilding after shear is given by the silica gel in polydimethyl
siloxane sample shown in figure 44, here the complete G and G curves are given,
but again it is obvious that the G build-up is most significantas the case of curing
and settingshowing once again the sensitivity of this parameter to microstructure.
The development of the plateau region is clearly seen as an evidence of the formation
of a network. Another example, but for a single frequency (~ 1 hertz) is shown in
figure 45.
G''
G'
0
5.0
10.0
15.0
10
10
10
10
10
10
Time, t / min
G''
G'
0
25
50
75
100
125
150
Time, t / min
Figure 42: Oscillatory parameters during the curing of two

orthodontic silicone gels at 2 Hz, see [39].
Figure 43: Oscillatory parameters during the curing of an

epoxy resin, measured at 0.5 Hz, see [39].
At very short times, we have simple elastic behaviour, and = G, but then
the stress begins to fall, first of all it does so linearly (expanding the exponential to
the first term gives / = G(1 - t/ ...)). The stress then continues to die away, so that
when t = , the stress has dropped to 1/e (~ 37 %); for t = 2, to 1/e2 (~ 13.6 %), etc.,
and on a semi-logarithmic scale displays a straight line, see figure 46. Thus at the
shortest times the Maxwell model does not know it has liquid-like properties, and at
the longest times it has forgotten that it has solid-like properties!
10
20
2.5
1
10
0
-1
10
10
10
rebuilding
time in mins.
10
rebuilding
time in mins.
Loss modulus, G'' / Pa
10
20
2.5
1
0
10
-2
-1
10
10
Frequency, / s-1
10
G'
101
100
-1
10
G''
10
-2
10
-1
10
10
Frequency, / s-1
Figure 44: Oscillatory parameters during the rebuilding of a

silica gel in polydimethyl siloxane after shearing.
10
10
10
10
10
10
Time, t / s
Figure 45: Oscillatory parameters during the rebuilding of a
flocculated sepiolite suspension after shearing at
30 s, see [42].
103
13.8
Stress relaxation testing
Suddenly deforming a viscoelastic liquid to a predetermined strain results in

a stress that, once attained, begins to fall off rapidly with time. This decrease
eventually slows down, and sometimes appears to approach a steady-state value at
long times.
If the liquid under test corresponded to a Maxwell model, the behaviour is
given by
t
(t )
= Ge
where again = / G, see figure 46.
10
Normalised stress, /G
(log scale)
Normalised stress, /G
(linear scale)
1/e
t=
10
-1
10
-2
10
-3
Time, t (linear scale)
Time, t (linear scale)
Figure 46: Stress relaxation of a Maxwellian liquid plotted

as either a linear or a logarithmic function of the
normalised stress.
For real viscoelastic liquids, the stress relaxation curve is usually described by
a series of Maxwell models in parallel, so
t
(t )
= G1e + G2 e + G3 e .....
where the various parameters are usually obtained by non-linear curve fitting
programmes.
13.9
Start-up experiments
Normal viscometer-type operation consists of applying a constant shear rate

to a sample and monitoring the ensuing stress. If the liquid being tested is
viscoelastic, the stress can take a finite time to come to its equilibrium value. If the
shear rate is low and the steady-state viscosity is constant as a function of stress, then
the stress builds up slowly. If however the shear rate is high, the stress can
overshoot, going through a maximum before reaching a lower steady-state value,
because the response has moved into the non-linear region.
If we suddenly start deforming the simple Maxwell model at a shear rate of
& , the stress slowly builds up as
104
= & (1 e
).
At very short times this simplifies to = G & t, or = G which is simple elastic

behaviour. Then at t = , the stress is 1/e of its value at steady state, where it is
= & , i.e. purely steady-state viscous behaviour. The start-up of real viscoelastic
liquids may need to be modelled using a number of Maxwell elements. However, for
most realistic experiments using this kind of test, the response quickly enters the
non-linear region since the strain is continually increasing.
Last of all, for a Kelvin-Voigt model, if we apply a constant strain rate, & ,
then simply = & [Gt + ], so that at zero time, = & , i.e. viscous behaviour, and
at very long times, ~ G & t, which is complete elastic behaviour. However, this kind
of behaviour is rarely, if ever, seen in practice.
References
[1] Al-Hadithi, T. S. R; Barnes, H. A; Walters, K., Colloid and Polym. Sci., 270(1),
40-46 (1992).
[3] Kaschta, J; Schwarzl, F.R., in Theoretical and Applied Rheology, Proc. XIth Int.
Congr. on Rheol., ed. P. Moldenaars; Keunings, R., Elsevier Science, Amsterdam,
1992.
[4] Liao, H.J; Okechukwu, P.E; Damodaran, S; Rao, M.A., J. Texture Studies, 27(4),
403 - 418 (1996).
[5] Zupancic, A., Applied Rheology, 7(5), 219 - 226 (1997).
[6] Shi, H.Q; Chen, S.H; deRosa, M.E; Bunning T.J; Adams, W.W., Liquid Crystals
20(3), 277 - 282 (1996).
[7] Gladwell, N; Rahalkar, R.R; Richmond, P., Rheol. Acta, 25, 55 61 (1986).
[8] Jones, T.E.R; Davies, J.M; Barnes, H.A, in Advances in Rheology, Proc. Ixth Int.
Congr. on Rheology, Ed. B Mena et al, Universidad Nacional Autonoma de Mexico,
Vol. 4, pp. 45-52, 1984.
[9] Ferry, J.D., Viscoelastic Properties of Polymers, 2nd Edition, John Wiley, New
York, 1970.
[10] Fannum, G; Bakke, J; Hansen, F.K., Coll. & Polym. Sci., 271(4), 380 389 (1993).
[11] Annable, T; Buscall, R; Ettelaie, R; Shepherd, P; Whittlestone, D., Langmuir,
10(4), 1060 - 1070 (1994), and Annable, T; Buscall, R; Ettelaie, R; Whittlestone, D.,
Langmuir, 37(4), 695 - 726 (1993).
[12] Shikata, T; Hihon Reoroji Gakkaishi, 25(5), 255 - 265 (1997).
[13] Fischer, P; Rehage, H., Langmuir, 13(26), 7012 - 7020 (1997),
[14] Ali, A.A; Makhloufi, R., Physical Review, 56(4), 4474 - 4478 (1997),
[15] M.E. Cates, J. Phys.-Condensed Matter, 8(47), 9167 - 9176 (1996), and Structure
and Flow in Surfactant Solutions, 578, 32 - 50 (1994).
[16] Hoffmann, H., Structure and Flow in Surfactant Solutions, 578, 2 - 31 (1994),
[17] Pons, R; Erra, P; Solans, C; Ravey, J.C; Stebe, M.J., J. Phys. Chem., 97(47), 12320
- 12324 (1993)
[18] Benbow, J.J; Cogswell, F.N; Cross, M.M., Rheol. Acta, 15, 231 - 237 (1976)
[19] Benbow, J.J., Lab. Practice, 12, 6 (1973).
[20] Richtering, W; Linemann, R; Langer, J., Progress and Trends in Rheology, 4th
European Rheology Conference, Ed. Gallegos, C., Steinkopff, Darmstadt, 1994,
pp. 597 599.
[21] Carnali, J., J. Appl. Polym. Sci., 43, 929 941 (1991).
105
[22] Cassagnau, P; Montfort, J.P; Marin, G; Monge, P., Rheol. Acta, 32(2), 156 161
(1993).
[23] Mewis, J., unpublished IFPRI report.
[24] Cordobes, F; Gallegos, C., Proc. 5th European Rheology Conference, Stein-Koppf,
Darmstadt, 1998, pp. 519 & 520.
[25] Bohlin Instruments promotional literature.
[26] Gallegos, C., Proc. XIIth Congr. on Rheology, ed. Ait-Kadi et al, Canadian
Rheology Group, Laval University, 1996, pp. 785 & 786.
[27] Haake promotional literature.
[28] Marchal, P; Moan, M; Choplin, L., Progress and Trends in Rheology, 4th European
Rheology Conference, ed. Gallegos, C., Steinkopff, Darmstadt, pp. 669 671.
[29] van der Vorst, B; Toose, M; van der Ende, D; Mellema, J., Progress and Trends in
Rheology, 4th European Rheology Conference, Ed. Gallegos, C., Steinkopff,
Darmstadt, 1994, pp. 699 701.
[30] Mewis, J, unpublished IFPRI report.
[31] Gluck-Hirsh, J.B; Kokini, J.L., J. Rheol., 41(1), 129 139 (1997).
[32] Zwanzig, R; Mountain, R.D., J. Chem. Phys., 43, 4464 (1965).
[33] Sanchez M. C; Berjano, M; Guerrero, A; Brito, E; Gallegos, C., Can. J. Chem.
Eng., 76(3), 479-485 (1998).
[34] Echstein, A; Sulm, J; Friendrich, C; Maier, R. D; Sassmannhausen, R;
Bochmann, M; Mulhaupt, R., Macromolecules, 31(4), 1335 1340 (1998).
[35] Ferry, J.D., Viscoelastic Properties of Polymers, 2nd Edition, John Wiley, New York,
1970.
[36] Cox, W.P; Merz, E.H, J. Polym. Sci., 28, 619 (1958).
[37] Barnes, H. A; Hutton, J. F; Walters, K., An Introduction to Rheology, Elsevier,
Amsterdam 1989.
[38] Tschoegl, N.W., The Phenomenological Theory of Linear Viscoelastic Behaviour,
Springer-Verlag, Berlin, 1989.
[39] Haake Instruments promotional literature.
[40] Lapasin, R; Martorana, A; Grassi, M; Pricl, S., Proc. 5th European Rheology
Conference, ed. Emri, I, Steinkopff, 1998, pp. 579 & 580.
[41] Phan-Thien, N; Safari-Ardi, M., JNNFM, 74(1-3), 137-150 (1998).
[42] Mewis, J., unpublished IFPRI report.
106
CHAPTER 14: NON-LINEAR VISCOELASTICITY

In any real material ... rearrangements necessarily require a finite time
... [therefore] all real materials are viscoelastic, Nicholas W. Tschoegl.
14.1 Everyday elastic liquids
There are many situations encountered in everyday life around the home
where elastic (or to be more precise, viscoelastic) liquids are encountered, and their
viscoelasticity is very evident. They are either stringy when poured or wobbly when
shaken or swirled. Various other overt elastic properties are sometimes seen, such as
die swell as very viscoelastic liquids expand when squeezed out of tubes.
Among everyday liquids that show some of these effects are
egg-white
toothpaste
shower gels
half-set jelly
thickened bleaches
hair shampoos
oven cleaners (very caustic!)
low-calorie vinegraites
thixotropic paints
flour doughs
There are also more viscous gelled materials that show strong elastic
properties, such as Swarfega, Jungle Gel or similar hand-washing products, as
well as some kinds of greases, where the product is so elastic that it resonates or
rings in its plastic container when the container is struck.
In most of these cases the elastic behaviour arises from high-molecularweight polymers dissolved in the continuous phase, or else from the presence of
rodlike-micellar (living polymers) structured surfactant phases.
Although all these examples also manifest large elastic effects in linear
measurements as described in chapter 13, the same is not always true for non-linear
measurements. There are many colloidal systems that show large values of storage
modulus in small-amplitude testing that do not show any visible large-amplitude
elastic effects such as die swell or other similar phenomena (hence large G), see [1].
Examples of such systems include flocculated dispersions and emulsions, where the
linear viscoelasticity arises from the formation of chains of particles, which easily
break on shearing at large strains and in steady-state flows.
Exercise: collect some elastic liquids from your home, and rank them in terms of
their viscoelasticity, using whatever criteria seem appropriate.
107
14.2
Some visible viscoelastic manifestations
There are many situations where, if viscoelastic forces manifest themselves on

a grand scale, large changes can occur in flow and force patterns. Figures 1 to 7 show
some striking examples of these effects.
Newtonian
Viscoelastic
Figure 1: The Weissenberg effect, seen as the rod-climbing in an

elastic liquid, as opposed to the inertial dipping in
a Newtonian liquid.
elasticity
only
elasticity
+ inertia
inertia only
Figure 2: The effect of viscoelasticity in a cone-and-plate
configuration, in conjunction with the with the
competing inertial force.
108
Chapter 14: Non-linear viscoelasticity

Inelastic
liquid
Viscoelastic
liquid
The flow of a Newtonian and a viscoelastic liquid

Figure 3:
out of a tube, showing viscoelastic 'extrudate swell' .
Viscoelastic
liquid
Figure 4:The tubeless syphon: look, no tube!
Newtonian
Figure 5: The effect of viscoelasticity in a simple turbine

mixer configuration, with the effect of the competing
inertial force.
Viscoelastic
Figure 6: The Weissenberg effect, as seen within an elastic liquid,

compared with the inertial effect in a Newtonian liquid.
Clearly they are important in many practical and industrial applications. All these
effects arise essentially because of tension along the steamlines generated in the flow
of these viscoelastic liquids which are not present in Newtonian liquids, and this
tension is associated with the so-called first normal-stress difference.
Viscoelastic
liquid
Figure 7: The 'normal force' pump.

Figure 8 shows a simple but useful analogy to this situation. It compares a squeezed
solid rubber cylinder and a sheared viscoelastic liquid cylinder, where in both cases a
force emerges at right anglesor normalto the squeezing direction: hence we talk
about a normal force.
109
In the case of a viscoelastic liquid the squeezing action arises from the tension along
the circular streamlines which result in an inward strangulation force, otherwise
called a hoop stress [2].
soft rubber cylinder
squeezed
streamlines
in tension are like
stretched elastic bands
that create a hoop stress
Figure 8: An analogy to the normal force created in a sheared
cylinder of elastic liquid.
Another (but smaller) non-linear viscoelastic force called the second normal-stress
difference (see later for an explanation of this term) has a special rle to play in a
number of other situations such as
shear fracture in rotational instruments such as cone and plate

Taylor instabilities
die swell
flow in non-circular pipes
14.3 The proper description of viscoelastic forces

The viscoelastic forces that produce the remarkable manifestations illustrated
above are properly characterised in shear flow by the so-called first and second
normal-stress differences, N1 and N2, which occur in addition to the shear stress
(with which we are already familiar) note that occasionally N1 and N2 are called
the primary and secondary normal stress differences.
The complete stress
distribution in a flowing viscoelastic liquid may be written down formally as follows,
= & ,
xx yy = N 1 , and
yy zz = N 2
where , N1 and N2 are called viscometric functions. At this stage we do not need to
be too concerned about understanding the origins and mathematics of these forces,
110
but we do need to appreciate the general form of the curves shown later in this
chapter just as we have examined non-Newtonian viscosity.
The viscoelastic equivalents to viscositythe stress divided by the shear
rateare the so-called first and second normal-stress coefficients, 1 and 2 . These
are given by the first and second normal-stress differences divided by the shear rate
squared, so
1 =
N1
,
& 2
2 =
N2
.
& 2
and
approximate
power-law
region

Figure 9: An idealised plot of the shear stress (dotted line)
and first normal-stress difference (1st NSD, solid line)
as a function of shear rate for a wide range of shear rate
and concentration or molecular weight.
(log scale)
Viscosity, or 1st NSDC 1
(Log scale)
Shear stress or 1st NSD N1
The typical form of these coefficients is not unlike that of the non-Newtonian
viscosity, with a high plateau at low shear rate, giving way to a decreasing powerlaw form above a critical shear rate, and at high-enough shear rates it begins to
flatten out again. This means that at low-enough shear rates, both N1 and N2 increase
quadratically with shear rate. Then in the shear-thinning region, the coefficients
always drop off faster than the complementary viscosity. At higher shear rates, the
rise is not so steep, and is often nearly linear. At the highest shear rates, the
dependence increases again towards a quadratic dependence on shear rate. All these
features are shown in figures 9 and 10.
approximate
power-law
region
Shear rate, . (log scale)

Figure 10: An idealised plot of viscosity (dotted lines) and first
normal-stress difference coefficient (1st NSDC, solid lines)
as a function of shear rate for a wide range of shear rate
and various concentrations or molecular weights.
Generally speaking, the second normal-stress difference is a fraction of the first

usually less than a quarterand of the opposite sign.
14.4 The measurement of viscoelastic forces
There are a number of ways in which N1 and N2 can be measured as a
function of shear rate. The easiest and most convenient measurements can be made
using a combination of rotating cone-and-plate and parallel-plate tests. When a
viscoelastic liquid is sheared in these geometries, not only is the customary drag
force experienced, but a force now arises that tends to push the plates apart. This
force, F, when measured, can be used to evaluate the respective normal-stress
differences. The following formulas apply to the cone-and-plate, parallel-plate and
111
re-entrant cone-and-plate geometries respectively, see figure 11, (in all cases the
radius of the plates is a),
N1 =
2F
,
a2
N1 N 2 =
N2 =
2F
a2
h + a tan
F
N1
h
a2
cone and plate

h
1 d log F
,
1 +
2 d log &
F
2
+ve, h = 0
separated cone and plate - +ve, h +ve

parallel plates
= 0, h +ve
re-entrant cone and plate - -ve, h +ve
Figure 11: Various cone-and-plate and parallel-plate configurations

used to measure normal force, showing the values of the
gap and cone angle.
where F is, as stated above, the total normal force or thrust on the configurations, h
is the edge gap between the plates, and is the cone angle (which is either positive or
negative). In all the experiments, h is a variable. In the last formula, all the
evaluations are made at the same rotation speed , see [3].
The normal-force differences are evaluated in every case at the edge shear
rate, which is given by &a = a / h . From a combination of two or more of these
measurements, N1 and N2 can be separately determined.
14.5 Examples of first normal-stress differences
Particular examples of the first normal-stress difference N1 or its coefficient
1 are shown in figures 12 - 29, where a comprehensive collection of examples is
given for polymer solutions and melts, as well as an emulsion. These are compared
with either the equivalent shear stress or the viscosity. All the different possible
combinations of shear and normal-stress difference, and viscosity and normal-stress
coefficient are displayed to show the way that results are presented in the rheological
literature. The figures are set out to illustrate overall behaviour, with especial
emphasis on low shear-rate and mid-range (i.e. power-law) behaviour.
112

7
3
2
10
1
10
Viscosity, / Pa.s
or 1st NSD, / Pa
Viscosity, Pa.s or
1st NSDC 1 , Pa.s 2
10
1s
tN
SD
C
vis
cos
ity
10
10
-2
-1
10
10 3
10
10
-2
r str
shea
r
ea
lin
ess
10-2
10-1
10
10
10
.
-1
Shear rate, / s
10
Figure 13: The viscosity (dotted lines) and first normal-stress difference
(1st NSD, solid lines) as a function of shear rate at room
temperature, for 100, 200 & 500 ppm polyacrylamide
(AP-30) solutions respectively, [5].
Viscosity, Pa.s or
1st NSDC 1 , Pa.s 2
qua
dra
tic
10 4
qua
drat
ic
lin
ea
r
Shear stress, or
1st NSD, 1 / Pa
10 5
-1
10
10 .
10
Shear rate, / s -1
Figure 12: The viscosity and first normal-stress difference
coefficient (1st NSDC) as a function of shear rate at
room temperature for 20 wt. % polystyrene
(Mw ~ 2 M) in tricrecyl phosphate, [4].
500 ppm
200 ppm
100 ppm
10
10
10
10
10
extrapolated
10
10
10
10
10
-2
-2
-1
10
10
10
10
10
10
10
.
.
Shear rate, / s-1
Shear rate, / s -1
Figure 14: The shear stress and first normal-stress difference
(1st NSD) as a function of shear rate at room temperature
coefficient (1st NSDC, solid lines) as a function of shear rate
for a 0.5% hydroxyethyl cellulose solution, [6].
for solutions of polyvinyl acetate in benzyl alcohol, [7].
4
10
Viscosity, / Pa.s
or 1st NSD, 1 / Pa
10
10
7 wt. %
1
4.5 wt. %
10
10
10 0
10 1
10 2
.
Shear rate, / s-1
10 3
10
10
10
10
10 3
10 2
4 wt. %
10 1
0.6 wt. %
100
-2
10
10
10
10
-1
10
-2
10
-1
10 -1
10 0
10
. -1
Shear rate, / s
10
10
Figure 18: The viscosity (dotted line) and first normal-stress

difference (1st NSD, solid line) as a function of shear
rate for a HDPE melt at 200 o C, [10].
melt + filler
melt + filler
10-2
10
10
10
10
.
Shear rate, / s-1
(1st NSD, solid lines) as a function of shear rate at room
temperature, for 0.6 wt. % and 4 wt.% aqueous
polyacrylamide solutions, [9].
Viscosity, Pa.s or
1st NSDC 1 , Pa.s 2
qu
ad
rat
ic
10
Viscosity, / Pa.s
or 1st NSD, / Pa
Figure 16: The shear stress (dotted lines) and first normal-stress
difference (1st NSD, solid lines) as a function of shear
rate at room temperature for solutions of polyacrylate
in dipropylene glycol methyl ether, [8].
melt
1st NSDC
Shear stress, or
1st NSD, 1 / Pa
10
melt
10
-1
10
viscosity
10
0
.
Shear rate, / s -1
10
Figure 19: The viscosity and first normal-stress difference coefficient

(1st NSDC) as a function of shear rate for polystyrene melt at
232 o C, with and without rubber filler particles, [11].
113
10
10
10
10
10
10
r
ea
lin
0
-1
10
10
10
10
Shear rate, / s -1
qu
ad
rat
ic
10
Viscosity, / Pa.s
or 1st NSD, / Pa
10
qu
ad
rat
ic
Shear stress, or
1st NSD, 1 / Pa
10
-2
10
10
-1
10
Shear rate, / s
10
-1
10
10
10
10
Viscosity, / Pa.s
or 1st NSD, / Pa
10
qu
ad
rat
ic
Viscosity, / Pa.s
or 1st NSD, / Pa
Figure 20: The shear stress (dotted line) and first normal-stress
Figure 21: The viscosity and first normal-stress difference coefficient
difference (1st NSD, solid line) as a function of shear rate
(1st NSD) as a function of shear rate at room temperature
at room temperature for a solution of polystyrene in toluene.
for a solution of polyisobutylene in cetane, [12].
-2
-1
10
10
10
10
10
= 0.3
= 0.1
= 0.3
= 0.1
10
10
Figure 22: The viscosity (dotted line) and first normal-stress

difference (1st NSD, solid line) as a function of shear rate
at room temperature for a polystyrene melt, [13].
10
10
10
-1
Figure 23: The viscosity (dotted line) and first normal-stress difference
coefficient (1st NSD, solid line) as a function of shear rate
at room temperature for an emulsion of low-molecular weight
polybutene in glyerol, at two levels of phase volume, [14].
102
Viscosity, Pa.s or
1st NSDC 1 , Pa.s2
Viscosity, Pa.s or
1st NSDC 1 , Pa.s2
10
Shear rate, / s
Shear rate, / s -1
10
10
-2
10
-1
10
10
10
20,000 ppm
10
10
-2
10
-4
2,430 ppm
120 ppm
10
Shear rate, / s -1
10
10
10
Shear rate, / s -1
10
10
Shear stress, or
1st NSD, / Pa
Shear stress, or
1st NSD, 1 / Pa
Figure 24: The viscosity (dotted line) and first normal-stress difference Figure 25: The viscosity (dotted lines) and first normal-stress
coefficient (1st NSDC, solid line) as a function of shear rate
difference coefficient (1st NSDC, solid lines) as a
for a polypropylene melt at 240 o C, [15].
function of shear rate at room temperature for three
aqueous polyacrylamide solutions, [16].
5
10
10 3
10
10
10
10
Shear rate, / s -1
10
Figure 26: The shear stress (dotted line) and first normal-stress
difference (1st NSD, solid line) as a function of shear
rate at room temperature for a 1% aqueous solution of
polyacrylamide.
114
10
10
10
-1
10
10
Shear rate, / s -1
10
Figure 27: The shear stress (dotted line) and first normal-stress difference
coefficient (1st NSD, solid line) as a function of shear rate at
o
room temperature for a polypropylene copolymer at 230 C.
10
-2
-1
10
10
10
10
.
Shear rate, / s-1
10
10
10
10
10 6
PM
MA
PS
PS
10
PMMA
LD
PE
HDPE
10
LDPE
10
10
10
10
Stress, / Pa
10
10
Viscosity, / Pa.s
10
HD
PE
tic
qu
ad
ra
qu
ad
10
First normal stress difference,

N / Pa
1
ra
tic
10
Viscosity, / Pa.s
or 1st NSD, / Pa
10
Figure 28: The viscosity (dotted lines) first normal-stress difference

Figure 29: The viscosity (dotted lines) first normal-stress difference
(solid lines) as a function of shear rate for a 2.5% solution
(solid lines) as a function of shear stress for a number of
of polyisobutylene (Mw = 6.106) in tetradecane at 25 o C, [17].
polymer melts at 160 oC, [18].
When a filler is added to an elastic liquid, the viscosity always increases.

However, the opposite is true when it comes to the normal stress situation, where the
level of elasticity always decreases. One of the most obvious consequences of this
effect is that extrudate swell decreases as the filler loading is increased. The only
exception to this is for long, thin fibres, where an increase is seen, see [19] chapter 7
for a full discussion of this unexpected result.
14.6 Some viscoelastic formulas
14.6.1 General
Just as we wrote down some formulas that predicted the effect of Newtonian
and non-Newtonian viscosity in flow, see chapters 6 and 10, so now can we quote
some for the effect of N1 and N2. (Conversely, these situations may also be used to
evaluate the values of N1 and N2.)
14.6.2 Extrudate swell
The extent of extrudate or die swell of a viscoelastic liquid emerging from a long tube,
see figure 3, is given by the empirical Tanner equation [20]
De
1 N1
= 1 +
D
2 2
2

w
where D is the diameter of the tube, and De is the eventual diameter of the extrudate.
The shear stress and the normal-stress difference values are evaluated at the wall of
the tube.
Sometimes a numerical addition of 0.1 is added to the right-hand side of this
equation to bring it into better alignment with experimental findings. Die swell of up
to a few hundred percent have been observed for very elastic liquids.
As mentioned above, the addition of fillers (apart from long, thin fibres)
decreases the N1 level, and thus decreases the die swell. At levels of filler of about
40%, a typical polymer melt, which would swell by up to 40% when unfilled, will
show no swell at all [19].
115
14.6.3 Flow in a slightly inclined trough

If b is the bulge on the surface of an elastic liquid flowing down a channel inclined at
an angle to the horizontal, then
N 2 = gb cos .
This method can be used as a simple means of measuring N2.
14.6.3 Pressure in holes and slots
If PH is the extra pressure generated at the bottom of a hole set into a wall when an
elastic liquid flows past it, then its value for various geometrical configurations is
given by the following formulas:
a circular hole
PH =
1 N1
N
2 ,
3 n1
n2 w
a transverse slot
PH =
1
( N1 ) ,
2n1
an axial slot
PH =
N2
,
n2
with all values of normal-stress difference evaluated at conditions relevant to the

wall of the tube.
14.6.4 Inertia in a cone-and-plate geometry
When a low-viscosity viscoelastic liquid is tested at high rotational speed in a
cone-and-plate geometry, there is an inertial force acting outwards that tends to pull
the plates together, thus counteracting the normal-force effects which tend to push
the plates apart. When this effect is just measurable, the overall force is then given
by
F =
a2
2
(a )2
N
1
.
6
This can be used to correct the normal-force results. However, when the correction
becomes large, secondary flows are present in the cone-and-plate gap, and this
means that the results are then unreliable.
116
14.7 Postscript - flow of a viscoelastic liquid into a contraction.

One last item worth mentioning before we move on is the effect of
viscoelasticity on flow into a contraction. This has been a set piece for many years
for rheologists interested in viscoelasticity, and the possibility of calculating its effect
using formal mathematical methods or computer simulation. No one can attend a
large rheological conference these days without hearing someone talking on the
subject! It is often encountered in extrusion processes in the plastics industry, where
viscoelastic polymer melts are pumped into and out of short dies.
All else being equal, an increase in flow-rate for an elastic liquid, or an
increase in the level of elasticity (measured as, say, the first normal-stress difference)
produces the effect on the flow pattern in a contraction shown in figure 30. The
complex flow pattern shown on the right of the figure is very susceptible to flow
instabilities. The onset of these instabilities dictates the maximum flow-rate possible
for an extrudate emerging from the end of the die having a smooth, acceptable
surface. (See chapter 17 for a discussion of the role of extensional viscosity in this
kind of flow.)
increasing flow rate/elasticity
Figure 30: The flow of a viscoelastic liquid into a contraction; note

that Newtonian liquids do not have recirculating vortices
at high flow rates.
References
[1] Al-Hadithi, T. S. R; Barnes, H. A; Walters, K., Colloid and Polym. Sci., 270(1),
40-46 (1992).
[2] Reiner, M., Deformation, Strain and Flow, 3rd edition, H.K. Lewis, London, 1969.
[3] Barnes, H.A; Eastwood, A.R; Yates, B., Rheol. Acta, 14, 61-70 (1975).
[4] Mhetar, V.R; Archer, L.A., J. Rheol., 40, 549 571 (1996).
[5] Han, C.D; Yoo, H.J., J. Rheol., 24, 55 79 (1980).
[6] Walters, K., Rheometry, Chapman and Hall, London, 1975.
[7] Agarwal, S.H; Porter, R.S., J. Rheol., 25, 171 191 (1981).
[8] Joseph, D.D; Beavers, G.S; Cers, A; Dewald, C; Hoger, A; Than, P.T., J. Rheol.,
28, 325 345 (1984).
[9] Chang, D.H; Yoo, H.Y., J. Rheol., 24, 55 79 (1980).
[10] Macosko, C.W., Rheology Principles, Measurements and Applications, Wiley-VCH,
New York, 1994.
[11] Kruse, R.L; Southern, J.H., J. Rheol., 24, 755 763 (1980).
[12] Zapas, L.J; Phillips, J.C., J. Rheol., 25, 405 420 (1981).
[13] Sugeng, F; Phan-Thien, N; Tanner, R.I., J. Rheol., 32, 215 233 (1988).
[14] Han, C.D; King, R.G., J. Rheol., 24, 213 237 (1980).
[15] Schmidt, L.R., J. Rheol., 22, 571 588 (1978).
[16] Argumedo, A; Tung, T.T; Chang, K.I., J. Rheol., 22, 449 470 (1978).
117
[17] Schoonen, J.F.M; Swartjes, F.H.M; Peters, G.W.M; Baaijens, F.P.T; Meijer,
H.E.H., J. Non-Newton. Fluid. Mech., 79 (2-3), 529 - 561 (1998).
[18] Feguson, J; Kemblowski, Z., Applied Fluid Rheology, Elsevier Applied Science,
London, 1991.
[19] Shenoy, A; Rheology of Filled Polymer Systems, Kluwer Academic Publishers,
Dordrecht, 1999.
118
CHAPTER 15: THE FLOW OF SUSPENSIONS

The whole of science is nothing more than a refinement of everyday thinking, Einstein
15.1 Introduction
Many of the structured liquids that we come across in everyday life are
suspensions/dispersions of particles in a liquid, with examples ranging from mud to
blood, and custard to mustard. Other examples of the kind of suspensions we are
dealing with here, and the places you might find them are
under the sink

pantry or cupboard
garage or store-room
bathroom cabinet
liquid abrasive cleaners, fabric washing liquids,

sauces, meat-pastes, custard, mustard, mayonnaise,
paints, greases, mastics, fillers, glues, cements,
medicines, ointments, creams, lotions, and toothpastes.
Everyday experience tells us that if we increase the concentration of the

dispersed particles, the viscosity of a mixture increases, going from a free-flowing
liquid through to a paste, and eventually to a soft solid, as more and more particles
are added. A familiar example of this is when we add cement or flour to water.
What we need to know is, what influences the increase in viscosity in terms of the
character of the liquid phase and the added particles? For instance, what is the effect
of the shape of the particles; their size distribution; the interaction between them, and
their deformability, etc.? All these factors will be considered below.
N.B. We do not differentiate here between the words suspension and dispersion, but
some people speak about the latter as made up of small particles, where small might
mean sub-micron-sized so that colloidal factors are very significant.
Emulsions are dispersions of deformable liquid drops in a liquid continuous
phase. However, if they are composed of very small drops, they are very much like a
small-particle-sized dispersion since the drops are hardly deformable. As a result
they follow essentially the same kind of rules given below for solid dispersions.
15.2 Viscosity of dispersions and emulsions
15.2.1 The continuous phase
The viscosity of dispersions is first of all controlled by the continuous phase
liquid which might itself be Newtonian or non-Newtonian, and then the added
dispersed phase where the size, shape, amount and deformability of the particles of
dispersed material can vary considerably, as can the interaction between the
individual dispersed particles.
For many liquids the continuous phase is Newtonian, and might be water, as
found in most household or personal care products, modern paints, etc. On the
119
other hand, it might be a Newtonian oil, as in such oil-based products as in

lubricants, greases, or even Plasticine the well-known modelling clay. Some paints
and coatings are based on a Newtonian resin continuous phase.
15.2.2 The effect of very low concentration of suspended particles
Although much empirical progress had been made in the study of
dispersions, the first important theoretical work was done by Albert Einstein at the
beginning of this century, when he calculated that the viscosity of a dispersion with a
very small amount of material, , in the form of solid spherical dispersed particles, is
given by
= 0 (1 + [ ] )
where is the measured viscosity and 0 is the Newtonian continuous phase
viscosity: [] is called the intrinsic viscosity, which for spheres Einstein calculated to
be 5 . In the equation, is properly called the phase volume, because it represents
2
that volume of the dispersion occupied by the dispersed phase; it does not mean
weight fraction, because at this stage the particles are considered weightless.
15.2.2.1 The effect of the continuous phase
In all the suspension equations considered here, the viscosity is predicted to
be directly proportional to the viscosity of the continuous phase. This is an important
point to remember, because if we change the continuous-phase viscosity in any way,
then, all else being equal, the overall suspension viscosity changes proportionally by
the same amount. Thus, if the viscosity of the continuous phase is doubled, then so
too is that of the dispersion. This proportionality is important when we think about
the effect of temperature, concentration of soluble additives, etc.
The following concentrations of, for instance, sodium salts (% by weight)
approximately double the viscosity of water at 20 C: hydroxide (10.5), carbonate
(11.5), acetate (15), phosphate (21), sulphate (18), tartrate (19), chloride (25),
thiocyanate (35), and nitrate (37). The effect is obviously complicated, being a
function not only of molecular weight, but also the shape of the dissolved molecules
and the way in which they interact with the structure of water. Molecular weight
alone will eventually give a large effect, as for instance with dextran (Mw = 72,000
daltons), which only needs to be added at a level of 2.75 wt. % to double the
viscosity of water, whereas simple organic molecules like glucose, maltose and
sucrose have to be added at concentrations around 22 % to produce the same effect.
In this context it is worth remembering that sometimes chemical additives
can decrease the viscosity of water. Before the beginning of this century, Wagner [1]
discovered that some salts of potassium, rubidium, caesium, and ammonium
produce such a reduction at lower concentrations, before at higher concentrations
increasing the viscosity as normal. These are the so-called structure-breakers,
which interfere with the local intermolecular hydrogen bonding of water molecules.
One of the biggest decreases is produced by the addition of 36% by weight of
ammonium iodide to water at room temperature, when the viscosity decreases by
around 13.5%. The complete curves for ammonium chloride, bromide and iodide
are sketched out in figure 1 where the reduction is shown as a function of mol/L,
which can be converted into weight percent by noting that the molecular weights are
120
Chapter 15: The flow of suspensions
55.5, 98 and 145 respectively. This kind of reduction has also been seen when
potassium iodide was added to glycerol, with a 20 % by weight solution giving a
viscosity reduction of 20 %.
Viscosity, / mPa.s
0.95
0.90
am
0.85
mo
hlo
mc
niu
amm
0.80
0.75
oni
am
e
rid
ro
mb
mo
m
niu
mid
iod
ide
Electrolyte concentration, c / mole/L

Figure 1: Viscosity versus concentration for some
aqueous ammonium salt solutions at 25 oC, [2].
Last of all with respect to the viscosity of the continuous phase, we might
consider the addition of other miscible liquids to water, where the mixture properties
are complex, even though the mixture of the same liquid to an organic liquid will
give simple mixing. A good example of complex mixing is seen in the addition of
ethyl alcohol to water, when both have almost the same viscosity but their mixture
viscosities are very different, see figure 2. This will have an obvious effect on the
viscosity of dispersions of particles in such mixtures, and is an interesting way of
thickening a suspension!
Viscosity, / mPa.s
9
7
0 oC
10 oC
3
1
20 oC
30 oC
20
40
60
80
100
Weight per cent of alcohol

Figure 2: Viscosity of water/ethyl-alcohol mixtures as a
function of temperature, [3].
The effect of temperature on the overall viscosity isall else being equal
controlled by the viscosity/temperature variation of the continuous phase. This has
already been discussed in 5.4. This means for instance that the variation of viscosity
with temperature for water-based suspensions is around 3 % per degree, but it can be
much higher for some oil-based systems.
121
15.2.2.2 The effect of the dispersed phase

When we say that the Einstein equation accounts for dispersions with a very
small amount of suspended particles, we mean that the particles are so widely
spaced that they are completely unaware of each others existence. In reality this
means a concentration of no more than a few percent phase volume. This formula
was experimentally verified by Bancelin in 1911 [2] using particles of size 0.3, 1, 2
and 4 m, and no effect of the size of the particle was found.
On the other hand, shape does matter, since the increase in viscosity comes
from the diversion of streamlines in the flow as they are redirected around particles,
thus leading to an increase in the viscous energy dissipation, i.e. viscosity. When the
particles are non-spherical, this extra dissipation increases, and the incremental
amount of viscosity also increases. The measure of the increase is the intrinsic
viscosity [] and this increases when a sphere is pulled out towards a rod, or
squashed down towards a disc, with the former giving the greater increase in
viscosity. A typical example of a rodlike particle system is paper-fibre suspension,
while red blood cells are approximately disc-shaped. Simple formulas for both kinds
of particles have been derived by the present author [3] as
5
7 3
p
[ ] =
100
for rodlike (prolate) particles and
[ ] =
3
p
10
for disc-like (oblate) particles, and in both cases the axial ratio p is defined in such a
way that it is greater than unity.
The presence of electrical charges on the surface of particles leads to extra
energy dissipation due to flow distorting of the surrounding charge cloud. Von
Smouluchowski accounted for the effect mathematically as
2
1
})
= 0 (1 + 2.5{1 +
2
2 2 0 a
where is the relative permittivity of the continuous phase, is the electrokinetic
potential, is the specific conductivity of the continuous phase, and a is the radius
of the spherical particles. This effect can easily double the effective phase volume [4].
For the first time, we have introduced a size effect, since the smaller the
particle, the larger is the effect of the fixed-thickness electrostatic layer relative to the
particle size, and hence the greater the effective phase volume, giving the inverse
dependence on a2 shown in the equation.
15.2.2.3 The effect of medium-to-high concentrations of particles
Although the Einstein equation is an important starting point, it gives little
help for real situations, since individual particles are not aware of one anothers
existence. Many empirical equations followed Einsteins exact mathematical effort,
122
each of which sought to increase the concentration range into a more practical region.
One of the more useful is known as the Kreiger-Dougherty (K-D) equation [5], which is
given by
= 0 (1
[ ] m
)
m
where m is called the maximum packing fraction, which is that concentration where
just enough particles have been added for the viscosity to become infinite. For
instance, at rest this is often near the random close-packing limit of approximately 64
%. However, as we shall see, the value of m varies according to the prevailing
circumstances.
The K-D equation reduces to the Einstein equation in the limit of very small
phase volume. (The basic form = (1 + a)n goes back to Baker [6], but has been
given many names since.) Further examination of the K-D equation in the light of
practical experience shows that the product of [] is often around 2 for a variety of
situations. In the light of this fact we can simplify the equation to
= o (1
2
) .
m
0.9
0.8
0.7
0.6
0.5
b ea
f ine
co a
ds
sand
r se
san
d
0.2
0.4
0.6
0.8
1 / geometric standard deviation
100% monodisperse
1.0
100% polydisperse
Maximum packing fraction,m m
Now, surprisingly, the only real variable is m! This variable is a function of particle
size distribution (p.s.d.) and particle deformability, as well as the flow conditions. In
terms of the former, the wider the particle-size distribution, the higher is m. Figure
3 gives an example of data due to Wakeman [7] for the packing of dry powders that
illustrates this point.
1.0
Figure 3: The maximum packing fraction of a powder as a function

of its polydispersity (shown here as the inverse of
geometric standard deviation) for a log-normal distribution.
In the same figure we also see that the effect is mirrored for slightly non-spherical
particles such as sand grains. The particular kind of size distribution used in the
powder illustrationlog normalis the usual consequence of many size-reduction
operations such as the grinding up of solid particles or the break-up of liquid
droplets in liquid-mixing operations.
If we want to control or understand the viscosity of concentrated dispersions,
m is an important variable to manipulate. For instance, either a widening or
123
narrowing of the p.s.d. can be attempted, or else the deliberate mixture of particle
sizes have a large effect on m. If the particles are monodispersed, then mixtures of
particle sizes can reduce the viscosity if the size ratio is around 4:1, since the small
particles can fit into the holes left when the large ones touch. This size-mixing
manoeuvre is very effective for higher concentration suspensions, as shown in figure
4. If three sizes are mixed, under the same restriction of size ratio, the effect is
equally as good, see figure 5, [8]. After three sizes of particles, the restriction on size
becomes impractical. This exercise is used in the manufacture of concrete, when as
aggregate material as possible needs to be added while keeping the concrete quite
flowable.
60
0.2
80
0.4
100
Figure 4: Relative viscosity of a binary mixture of particles,

with size ratio > 4:1, for various total phase volumes.
0.6
0.8
1.0
40
Coarse fraction, %
Fine 0
0.2
0.4
20
0.2
0.4
45%
40%
30%
.3
67
50%
0.6
55%
0.8
40
20
1.0
60%
70 0
10 0
15 0
20 0
30 0
50 00
10
Relative viscosity,
Coarse
60
0.6
0.8
1.0 Medium
Figure 5: Relative viscosity of a ternary mixture of particles in

suspension, where the total phase volume is 66%.
The consequences of concentration, , are scaled by m, and they become very

critical as approaches m. This is easily understood by considering the situation
when powders are dispersed in a liquid in such kitchen operations as making
custard, where adding the critical final amount of water or milk transforms a stiff
paste into a flowing liquid. In simple dispersions this becomes true for phase
volumes above 50 %.
It is easy to show that the viscosity-concentration profile approaches a
situation where the viscosity would double or halve if the phase volume increased or
decreased by 1%. There are a number of ways this can happen by the particles
swelling or shrinking by that amount. This would mean that the particle size would
change by around J %. Since this very small change could not be measured on
any commercial particle sizer, it leaves us with the interesting possibility that the
viscosity of a product could change considerably for no obvious reason!
An interesting practical example of this was seen by the author when visiting a manufacturer
of polymer lattices [3]. The lattices were manufactured in one solvent, then dried off and
afterwards re-dispersed in a different solvent. Unfortunately, some of the second solvent
absorbed to a small extent into the polymer particles, but only very slowly. Hence the
viscosity increased continuously over the next few days due to the swelling, but for no reason
obvious to the manufacturer, since the small increase in size of the latex particles was
immeasurable.
124
15.3 Particle-size effects in concentrated dispersions

The effect of particle size of dispersions enters through a very important
mechanism which relates to the spatial arrangement of the particles.
When a
dispersion is at rest, the particles are randomly dispersed throughout the continuous
phase due to the perpetual action of Brownian motion. If the dispersion is sheared
at very low shear rates, then there has to be a great deal of co-operative movement to
allow these particles to move in the flow direction while maintaining the overall
random distribution, and thus the viscosity is high. However, when the dispersion is
sheared at a higher shear rate, the particles can be moved from the overall random
arrangement towards a situation where they begin to form into strings and layers,
see figure 6.
no shear
Figure 6: The formation of strings and layers in a suspension of

non-interacting 'Brownian' spheres under the action of
shear flow.
Then the average distance between the particles increases in a direction at right angles
to the flow direction, and it decreases along the flow direction. This change of spatial
arrangement makes their movement past each other much easier, thus lowering the
viscosity. This manifests itself in the K-D equation by a small but significant increase
in the value of m as the shear rate increases. Typical values of m are about 0.63 for
a dispersion of monodisperse, spherical particles at very low shear rates and about
0.71 at very high shear rates. This seemingly small change produces large effects at
high concentrations resulting in a large amount of shear thinning in concentrated
suspensions.
As the backwards driving force towards randomness is Brownian, hence of
thermal origin, we can see that shear thinning is easier for large particles where
Brownian motion is less effective and the shear forces are correspondingly more
important. When a small-particle-sized dispersion is sheared, the effect of Brownian
motion is longer lasting along the shear rate axis, and higher values of shear rate are
needed to produce the same amount of shear thinning, see figure 7.
Peclet number (log scale)

sm
all
pa
all
rti
cle
pa
rti
s
lar
cle
ge
s
pa
rti
cle
s

Figure 7: A schematic diagram of the viscosity as a function of
shear rate or Peclet number for large and small particle
size suspensions.
125
These effects can be properly scaled by plotting the so-called Pclet number instead
of the shear rate, which then collapses the data onto a single curve. The Pclet
number is described as the ratio of the hydrodynamic to the thermal force, so Pe =
a3 & /kT. This curve can usually be described by the Cross-type equation, see
chapter 9, with the shear rate term replaced by the Pclet number, see figure 7.
15.4 Particle-shape effects in concentrated dispersions
20
10
spheres
21:1
14:1
7:1
Relative viscosity,
spheres
grains
30
rods
plates
Relative viscosity,
The thickening effect of particles with respect to their shape follows the
descending order of rods > plates > cubes/grains > spheres, when the same phase
volume of particles is added to a liquid. This is illustrated in figure 8, where we see
the approximate order of thickening power of each shape, with rods/fibres being the
most efficient for any given phase volume. The aspect ratio of fibres is a controlling
factor in determining the precise effect, see figure 9. With regard to the viscosity
parameters we have considered, i.e., [] increases for non-spherical particles as we
have seen above for dilute suspensions, but m decreases. However, the product of
these two terms still does not differ greatly from 2, see [8].
3
2
1
10
20
30
40
50
Phase volume,
Figure 8: Viscosity as a function of phase volume for
various particle shapes.
10
20
30
40
Phase volume,
Figure 9: Viscosity as a function of phase volume for various
aspect ratio of fibres.
15.5 Particle deformability

Particle deformability means that for the same amount of dispersed material,
the viscosity at high concentrations is lower than the equivalent dispersion of solid
particles, since the droplets are deformable and can thus accommodate each other at
rest and squeeze past each other during flow, so that first the maximum packing
fraction m is higher and the intrinsic viscosity [] is lower [9].
15.6 Particle interactions - an overview
The simplest picture we can form of any dispersion of submicron particles, is
of the particles being free to approach and depart from each other under the action of
the ever-present Brownian motion, with no more resistance than that presented by
the movement through the intervening viscous liquid. However for such small
particles, significant interparticle forces can change this picture. For instance the
ubiquitous van-der-Waals attraction force between particles, if not counteracted, will
cause the particles to stick together on touching. This attractive force arises from
correlated atomic motions in neighbouring particles and is always present in all
particle-particle interactions. For particle separations in the nanometre range, the
126
van-der-Waals interaction potential present between two equal spheres can be

described simply by
vdW =
Aa
12(r a )
where A is the relevant Hamaker constant for the particular situation, a is the radius
and r is the centre-to-centre separation of the particles.
This force is usually significant in the range 1-10 nanometre for colloid-sized
particles (~ 1 m). It does not increase indefinitely at very small distancesas
indicated by the simple equation abovebut approaches a minimum.
If the
particles are completely unprotected, the force minimum at these very small
distances (the so-called primary minimum) can be large enough to give more-or-less
permanent particle contactthis is called coagulation. In most situations, however,
we are usually interested in the shallower secondary minimum caused by the
combined effects of the van-der-Waals attraction and some repulsive force arising
from the presence of a protecting moiety at the particle surface or an electrostatic
repulsion between neighbouring particles carrying the same sign charge.
A strong repulsive force between particles arises from the repulsion between
like charges on the surfaces of adjacent particles. The approximate operating
distance of the electrostatic force resulting from charges adsorbed on particles in an
aqueous environment is 1/ where is known as the double-layer thickness. The
thickness of this layer is strongly dependent on the electrolyte concentration and is
given approximately (in nanometre units) by
1/ = 0.3 c-1/2 [z]-1
where c is the electrolyte concentration in mol/L, and [z] is the valency of the
electrolyte. This means that (assuming [z] = 1) the following is true for aqueous
liquids
electrolyte concentration, c = 10-5, then double layer thickness 1/ 100 nm
c = 10-3,
1/ 10 nm
-1
c = 10 ,
1/ 1 nm
If the repulsive forces are larger and longer-range, then they are able to
counteract the van-der-Waals attractive forces; thus giving a (colloidally) stable
dispersion. Polymers are availablecalled block copolymerswhich have one end
soluble in the continuous phase, and the other end insoluble and able to absorb onto
the particle surface. These can protect particles that would otherwise flocculate - the
protruding polymer loops and strands overlap to hold the particles far enough apart
so that they are not strongly attractedthis is called steric repulsion. The system can
then be regarded as non-interacting, but for very small particles, the effective phase
volume must be increased to account for any polymer layer.
As well as producing repulsion by introducing a polymer onto the surface of
the particles, one can combine both effects using absorbed polyelectrolytes, i.e.
polymers with charge distributed along the chains. The particular nature of the
forces arising is different in each case but the overall effect is always the samethey
produce a repulsive force. We need say no more about normal polymers at this
stage with respect to the details of these force, but it is useful to note that the
127
approximate distance of action of a polymer adsorbed at random points along its

backbone (a homopolymer) is approximatelywhen expressed in nanometresone
tenth of the square root of its molecular weight. If however the polymer is attached
at one end only, the distance is about twice that value, thus for instance, for a
terminally anchored polymer of molecular weight 10,000, then thickness of the layer
would be around 20 nanometres, while the randomly attached equivalent would
have a thickness of about 10 nanometres.
Another attractive force between particles has very surprising origins. If high
molecular weight polymers are present in the continuous phase of a suspension, then
the only effect we normally expect is that due to increasing the continuous-phase
viscosity. However, another effect can also occur under certain circumstances. This
arises because when particles come close to one another, and the gap between them
is quite small, any large polymer molecules (or surfactant micelles) in the vicinity are
excluded from the small space. The result is a depleted region with respect to
polymer concentration and this leads to a local difference in osmotic pressure. The
overall effect is that the two adjacent particles producing the small gap are pulled
together. This pull can be so strong that it exceeds any repulsive forces, and with the
addition of the van-der-Waals force, it can produce flocculation of the particles. The
effect is called depletion flocculation, and it can also be caused by very small particles
or detergent micelles present in the dispersion.
In summary we can say that the overall effect of the attractive van-der-Waals
and depletion forces with the various repulsive forces shown in figure 10, where two
extremes are shownvery large attractions due to van-der-Waal plus depletion
forces, and the large repulsion present if we have very large surface charge and low
electrolyte in aqueous systems.
Attraction
0 Repulsion
DISPERSION
ary
ond
c
e
fs
th o m
p
e
d imu
min
FLOCCULATION
d
van
rW
aal
Distance from surface

Figure 10: Force-distance curves for colloidal forces: van der
Waals attraction forces balanced against increasing
steric or electrostatic repulsion forces.
However, most systems have a mixture of both effects, so that we have the
appearance of a secondary minimum. If this secondary minimum is shallow (only a
few kTs), then the particles remain dispersed, but if it is much larger than that, then
the particles stick together, that is they flocculate.
All these kinds of interactions lead to an increased viscosity. However,
although these colloidal forces dictate the form of the dispersion at low shear rate,
when the shear rate is greater than some critical value, the viscosity begins to
decrease. Eventually, the viscosities approach similar values at high shear rates,
128
largely dictated by the hydrodynamic effects that we have previously seen, see figure
11.
COLLOIDAL FACTORS
shape
separation
surface
size (distribution)
salt
shape
separation
size (distribution)
HYDRODYNAMIC FACTORS

Figure 11: Flow curve of a suspension of colloidal particles.
In the figure we see that colloidal effects dictate the viscosity at low shear rates, and
then the five Ss are important ,
shape of the particles and their

size distribution, plus the
separation of particles (i.e. concentration), their
surface properties and the possible presence of
salt in the continuous aqueous phase.
However, at very high shear rates, hydrodynamic effects largely control the
situation, and only the first three factors are important, with some effect due to the
presence of surface charge in the increase of the effective phase volume, but the effect
is smaller than at low shear rates.
15.7 The viscosity of flocculated systems
While particles with a large overall repulsion are very interesting from a
scientific point of view, giving us the crystal-type effect that can have very beautiful
optical effects, they are rarely important in real products, since these systems cannot
tolerate the presence of salts and surfactants in solution. However, a far more
widespread situation is where flocculation is present, at least to some extent,
resulting from an overall attraction between the particles. The individual flocs
making up such dispersions can be easily broken down by shear, as for instance
during vigorous shaking. This situation will now be described in some detail,
especially its rheological consequences.
The depth of the secondary minimum formed due to the addition of the
repulsive and attractive forces will vary from a few kT unitswhich means no
flocculation because Brownian motion will keep the particles apartto ~ 10 - 20 kT.
Forces in this latter range mean that typical flows can break-up any flocs present,
although they will reform under more quiescent conditions.
The size and
architecture of the flocs formed play a major rle in determining the rheology and
physical stability of the suspension.
Shearing hard enough will result in the flocs being reduced to the primary
particles, but shearing at a low shear rate results in the partial breakdown or
reformation of flocs. The form of the floc depends on the interaction force and to
some extent on the flow history attending the floc formation. Depending on its
129
form, strength and duration, any flow will progressively break down these flocs:
thereafter, further rest reforms them under the action of Brownian motion. (This
process accounts for thixotropy as well as shear thinning in flocculated suspensions
and it is completely reversible.)
We will spend a little time considering the nature of flocs. A floc is a
collection of particles where the spatial arrangement often follows some simple law.
It might be that the average particle concentration is the same throughout the floc,
but usually the concentration decreases from the centre towards the outside. This is
due to the way that the flocs are formed.
The simplest way of describing such a floc is to use fractals. In our case this
simply means that the concentration falls off according to a power law with distance
from the flocs centre. Then we can write down the following,
R
p 0
a
where p is the number of particles in a floc, a is the radius of the particles, R0 is the
radius of the smallest enclosing sphere for the floc, and D is called the fractal
dimension, see figure 12.
smallest
enclosing
sphere
D<2
D > 2.5
Figure 12: Fractal flocs: the dimensions and extremes

of fractal flocs (shown in 2D).
D can vary from almost 3 for the densest floc down towards 1 for a very open linear
floc with a few radiating arms. The value of D depends on the particle interaction
forces. If a suspension is sheared to completely deflocculate the particles, and then
the flocs are allowed to reform under quiescent conditions, the architecture will
depend on the forces in such a way that if the attractive forces are large, say around
10 kT, with D often less than 2. Here the particles hit and stick, giving a very open
structure. However if the attractive force is small, say a few kTs, the particles hit and
roll, forming a denser floc, with D greater than 2.5. Obviously small changes in D
will give large changes in floc structure.
From the mathematical relationship above we can redefine the phase volume
as the effective phase volume, assuming for the sake of simplicity that the phase
volume is now that defined by the enclosing spheres of the flocs of radius R0. Then it
is easy to show that
3
eff = p D
130
where is the total (real) phase volume of the particles, and eff is the effective phase
volume of the flocs. It is obvious now that the effective phase volume of the
flocculated suspension is much higher than the real phase volume of the particles,
and hence the viscosity of the suspension is much higher.
From this equation we also see the sensitivity of viscosity to interparticle
force, for the exponent 3/D - 1 will vary from 0 when D is 3 (for a small attractive
force), to 0.5 for a D of 2 (for a large force). This would mean that for p = 100, the
effective phase volume would vary by a factor of 10 for D going from 3 to 2.
However we also see another important fact from this equationthe effective phase
volume of the suspension now depends on the size of the flocs, via the value of p, the
number of particles per floc. For example if D = 2 (for most real flocculated
suspensions D usually varies in the range 2 0.25), the effective phase volume
decreases by a factor of 10 when the number of particles in a large floc is reduced
from 10,000 to 100 under the action of shearing.
The most obvious way of reducing the number of particles in a floc is to
subject the floc to shear. If such a shear field is applied to a flocculated suspension,
then the floc size is approximately [10]
Rs =
R0
c
1 + (b& )
where Rs is the radius of the sheared floc; b is a constant with the dimensions of time,
and c is a dimensionless constant, and & is the applied shear rate. From this
relationship we can rewrite the effective phase volume, insert it into the K-D
equation and arrive at a simple expression for the viscosity of a flocculated
suspension as
= 0 1
R0
c
a (1 + (b& )
3 D

. .
m
The effect of shear rate on viscosity in this equation is very strong, and demonstrates
the typical situation with flocculated systems where severe shear thickening is seen,
giving what looks like a yield stress.
15.8 Thixotropy
Thixotropy is the change of viscosity with time of shearing rather than rate of
shearing, and is generally viewed as a troublesome property that one could well do
without. All the advantages previously claimed for thixotropy are in fact better seen as
the result of a high degree of shear thinning [11].
Thixotropy comes about first because of the finite time taken for any shearinduced change in microstructure to take place. Microstructure is brought to a new
equilibrium by competition between, on the one hand the processes of tearing apart by
stress during shearing, and on the other hand build-up due to flow- and Brownianmotion-induced collision, over a time that can be minutes. Then, when the flow ceases,
the Brownian motion (the only force left) is able to slowly move the elements of the
microstructure around to more favourable positions and thus rebuild the structure: this
can take many hours to complete. The whole process is completely reversible.
131
shear stress,
If we refer back to the viscosity of flocculated suspensions, we had assumed

there that if we started to shear the suspension at a particular shear rate, then we
assumed that the floc radius immediately adjusted itself to that value appropriate to the
shear rate. This of course is not the case, and a finite time is required for such changes to
take place. During this time, if the shear rate has been changed from a lower to a higher
shear rate, then initially the viscosity is that which is pertinent to the floc radius of the
lower shear rate, then slowly the average radius of the flocsand consequently the
viscositydecreases. On the other hand, if the shear rate is decreased, the average floc
size gradually increases, and the viscosity subsequently increases also, see figure 13.
bre
akdow
n
up
build-
shear rate,
Time of shearing, t
Figure 13: The behaviour of a thixotropic liquid, initially at rest,
and then subjected to a high shear rate followed by a
lower shear rate.
In fact, any shear-induced change in the microstructure of a suspension takes

time to occur. This is true for the simplest changes such as the transition from the
random-at-rest situation for simple suspensions to the breakdown of flocs to primary
particles when flocculated suspensions are sheared hard. The former change is often
over before the relatively slow mechanics of most viscometers could detect it, but the
latter kind of change can take hours or even days to complete!
Thixotropy is a function of time and shear rate, and therefore cannot be
properly accounted for in experiments where both these variables are changed
simultaneously. This is true in loop tests where either shear rate or shear stress is
varied in a triangular fashion with time, increasing linearly from zero to a maximum
and back again. The measured results for the up- and down-curves are quite
different for the first few loops, and the area between these curves is sometimes used
as a measure of thixotropy, but these produce numbers that are impossible to relate
to quantitative description of the phenomenon. For this reason, loop tests are not
recommended for the serious study of thixotropy.
The best experiments to properly measure thixotropy are those where the
sample to be tested is sheared at a given shear rate until equilibrium is obtained, then
as quickly as possible the shear rate is changed to another value. The typical response
to such a step-wise change from one steady-state condition to another is, in terms of the
viscosity, often characterised by the so-called stretched exponential model:
t
= e, + ( e, - e,0 )(1 - e )
where e,0 is the viscosity at the commencement of shearing; e, the viscosity after
shearing for an infinite time and is a time constant. Note that this formula accounts for
the fact that the eventual viscosity will be higher than the original value when we move
132
from a higher to a lower shear rate, and the viscosity recovers. The value of is a
function of both the original and the final shear rate.
Further details on the subject of thixotropy may be found in a review published
by the author [11].
15.9 Shear thickening
s
rea
inc
~ 55%
s
ha
np
ei
~ 50%
,
me
olu
ev
~ 45%
In very specific circumstances, the viscosity of a suspension can increase with

increasing shear rate as well as decrease. After the flow field has organised the
particles into strings and layers, there can come a time when, if the shear-rate/shearstress is high enough, the particle ordering will break down and clumps of particles
are formed. The form of these shear-induced clumps may be temporary strings
aligned across the flow, or permanent or transient random groupings or strings of
particles. The result of this disruption is an increase in viscosity. The form of this
increase is shown in figure 14.
locus of critical shear rate, c

Figure 14: Schematic representation of the flow curve of a typical
shear-thickening suspension, for various phase volumes.
The increase in viscosity follows on from the normal decrease of viscosity with
increasing shear rate seen with all suspensions. The increase then seen becomes
more and more abrupt as the concentration is increased. Since the viscosity is
double-valued with respect to shear stress, strange things can happen when stresscontrolled rheometers are used to measure this phenomena. For instance, if a stresssweep programme is used, the shear rate will increase, but eventually decrease.
The overall situation with respect to the effect of particle size is shown in
figure 15, where particle sizes from very small to quite large are responsible for the
onset of shear thickening moving from very high to very low shear rates. In fact, the
critical shear rate is approximately proportional to the inverse of the particle size
squared. More details on this interesting but troublesome phenomenon may be found
in a review written by the author [12].
15.10 Apparent wall slip
An apparent slip or lubrication occurs at the wall in the flow of any multi-phase
systems if the disperse phase moves away from smooth walls. This arises from steric,
hydrodynamic, viscoelastic and chemical forces present in suspensions flowing near
smooth walls and constraints acting on the disperse phase particles immediately
adjacent to the walls, see [13].
133

Critical shear rate, / s
-1
10
coating
pumping
10
60%
50%
40 %
mixing
slow
stirring
-3
10
latices
10
-2
resins
-1
10
clays
10
starch
0
10
10
Particle size in microns

Figure 15: Critical shear rate 'map' for the onset of shear thickening
as a function of particle size and phase volume, for various
operations and materials.
Concentration of spheres / %
The enrichment of the boundary near the wall with the continuous (and usually lowviscosity) phase means that any flow of the suspension near such a boundary is easier
because of the lubrication effect. Because this effect is usually confined to a very
narrow layerwith typical thickness of 0.1-10 mit so resembles the slip of solids
over surfaces that it has historically been described by the same terminology. The
restoring effect for all the forces that cause an increase in concentration as particles
move away from walls is usually osmotic, and this will always limit the movement of
particles away from the walls, thus also limiting the effective slip.
How do these effects occur? When any suspension of particles is placed next
to a smooth wall, the original microstructure is locally disturbed. For a simple
suspension at rest where the particles are randomly dispersed in space, the
concentration of particles undergoes a damped oscillatory variation as one moves
away from the wall, see figure 16, where the concentration is at the maximum
packing fraction, so that the effect is enhanced. The new distribution has two effects,
first that the variation in concentration does not die out until about five particle radii
away from the wall, and secondly that the average particle concentration is zero at
the wall and less than average for a small distance away from the wall.
80
~ 64%
60
40
20
0
Distance out from the wall / 2a

Figure 16 : The local concentration of spheres as a function of
distance from the wall, measured in sphere diameters.
As well as these staticessentially geometriceffects that produce depletion

at the wall, there are also dynamic effects which enhance the phenomenon. The
existence of a shear rate and/or a shear rate gradient in the fluid next to the wall (as
in a pipe) results in a further movement of particles away from the wall, towards
areas of lower shear rates such as the centre of pipes. Solid particles, emulsion
droplets and polymer molecules all show this tendency.
134
no
sli
p
Apparent viscosity, (log scale)
The result of these combined static and dynamic depletion effects is an

effective lubricating layer at the wall. This can be modelled simply as a particle-free
layer of about half a particle-radius wide. Hence the lubricating effect is larger for
larger particles, and in this context large particles include flocs, so that at low shear
rates, slip effects are stronger than at higher shear rates for flocculated suspensions.
Slip manifests itself in such a way that viscosity measured in different-sized
viscometer geometries gives different answers if calculated the normal way - in
particular the apparent viscosity decreases with decrease in geometry size (e.g. tube
radius). Also, in single flow curves, unexpected lower Newtonian plateaus are
sometimes seen, with an apparent yield stress at even lower stresses. Sudden breaks
in the flow curve can also be seen. The typical effect of slip on flow curves measured
in viscometers/rheometers is shown in figure 17. The effect is obviously very gapsize dependent, since the slip layer becomes more and more important as the gap
size decreases.
x mm
x/5 mm
x/10 mm

Figure 17: Viscosity versus stress plot for a flocculated
suspension showing slip effects, as a function of
concentric-cylinder gap.
What do we need to do about slip? First we might want to properly describe

the slip in viscometers because it might also be happening in the situation we are
dealing with, such as flow down smooth-walled process pipework. In this case we
can for instance write down the following equation for flow in a pipe with a very
thin Newtonian slip layer in a pipe of radius a,
4V
4Q
= & w + s
3
a
a
where Q is the overall flow rate and Vs is the velocity at the edge of the slip layer.
The effect of this is shown in figure 18, where we see the possibility of extrapolating
the apparent wall shear rate (4Q/a3) in pipes of different diameter but at the same
values of wall shear stress w (= Pa/2L), in order to extract the no-slip shearstress/shear-rate data as well as obtaining the slip-velocity/shear-stress function
from the slope of the curves. This type of graph is called a Mooney plot, and enables
us to extract both the wall shear rate, & w , and slip velocity, Vs , as a function of wall
shear stress from simple plots of Q/a3 vs. 1/a for fixed values of w in different pipes.
135
Q/a
(linear scale)
increasing wall shear stress
x
x
x
x
x
Inverse of tube radius, 1/a (linear scale)

Figure 18: The apparent wall shear rate as a function of inverse
diameter for various wall shear stresses.
Rather than characterising slip, we might want to eliminate it. This can often be
done by sandblasting or otherwise roughening or profiling (with ribs etc.) the
surfaces where the test liquid comes in contact with the viscometer. In fact, any
regular surface undulations of the order of 10 m should eliminate slip effects in
most cases. See chapter 7 for a further discussion of the elimination of slip, especially
by the use of the vane and basket geometry.
In summary we can say that slip becomes more and more important as
concentration is increased, with greater difference between the slip layer and the
plug viscosities,
particle size is increased (flocs included),
viscometer gap size (or tube radius) is decreased and
(usually) shear rate is decreased, see figure 17.
15.11 Very high concentration pastes
In many situations, the concentration of a suspension is so high that the flow
properties in shear are dominated by the shearing thin slip layer and an almost
unsheared solid plug, while for any extensional flow that the paste encounters (such as
flow into an orifice or the entrance of a tube) still demands the deformation of the
whole paste. In these situations the flow of a typical paste into and along a tube,
such as are found in ceramic and battery pastes etc., can be characterised by the
formula
D
P = 2 o + 1V m ln 0 + 4 o + 1V n L
D
D
where the first term describes the entry flow into the tube (dominated by extensional
flow), and contains the yield stress of the bulk material, o, see [14]. The second term
describes the slip flow along the tube, and is characterised by a possible (but not
always necessary) slip yield stress, o , with a power-law index in each case, given by
m and n respectively . In the simplest case, m and n are unity, and the slip yield
stress, o , is zero.
Cone penetrometry of paste-like materials can be used to measure the
apparent extensional yield stress using the equation
136
o =
F
L tan 2
2
where F is the applied force, is the half angle of the cone, and L is the penetration
depth.
15.12
Colloidal control of suspension viscosity
Some very significant changes can be brought about by the alteration of

colloidal conditions, since the overall force between particles can be made more
attractive or repulsive in nature. We will now consider a few examples of such
manipulation.
First, we see the effect of the addition of electrolyte to a suspension of highlycharged spheres. Particle surface charge results in a large repulsion between
adjacent particles, causing them to form into a pseudo-crystalline state with particles
sitting on regular lattice positions (but vibrating due to the incessant Brownian
motion). In this state, it is difficult for particles to pass over each other in shear flow,
jumping from one lattice position to the next, hence out of one potential well into
another. The result of this is a high viscosity, see figure 19. If electrolyte (i.e. salt) is
now added to the aqueous phase, the charges on the particle are shielded, and the
interparticle repulsion is reduced and the viscosityat least at low shear rates
drops significantly. There might also be an accompanying small drop in viscosity at
high shear rate, see 15.2.2 above.
If enough salt is added, the particles are eventually neutral with respect to
one another, and just bounce off each another if they collide. The viscosity is then at
a minimum. If even more salt is added, the electrostatic repulsion is swamped, the
van-der-Waals attractive force dominates, and the particles flocculate. The viscosity
then increases again, as in figure 19.
repulsion
neutral
3
attraction
add salt

Figure 19 : Flow curves resulting from the addition of electrolyte
(salt) to the aqueous phase of a suspension of
charged particles.
Our second example is the change of surface forces brought about by change
in solution pH. The hydroxyl groups on many mineral particleschalk, titanium
dioxide, quartz, etc.become charged (in terms of degree and sign) according to the
solution pH. Thus at very low pH we might have a large positive charge on the
particle surfaces, while at very high pH the situation might completely reverse to a
large negative charge. Between these values the charge can pass through zeroat
the so-called isoelectric point (some examples of the isoelectric point in terms of pH
137
zeta potential
- 40 mV 0 + 40 mV
are, titanium dioxide ~ 8, alumina ~ 10, calcium carbonate ~ 12) . The effect on the
viscosity is shown in figure 20. At the isoelectric point the system flocculates and
becomes very shear thinning.
1
2
3
0 pH value 13

Figure 20: Flow curves resulting from changing the surface
charge (via solution pH) on the suspended particles
of a concentrated suspension.
Last, we will consider the effect of adding surface-active material or very high
molecular weight polymers that absorb at one end or at intervals onto the suspended
particle surface (block co-polymers). If we begin with a flocculatedand hence high
viscositysuspension, and add sufficient of this kind of material, then after vigorous
stirring the viscosity drops significantly as the particle surface gets covered. The
electrostatic or the steric repulsion (or both) causes the state of flocculation to reduce
or disappear with its accompanying reduction in viscosity.
One other interesting effect is found with either shearing emulsions or
grinding particles in a mill. If a certain amount of surface active material or
polymers are adsorbed to reduce the viscosity, then when the particle size is reduced,
the surface area increases greatly (as 1/a2), and hence the surface coverage
(concentration per unit area) decreases, and hence the interparticle repulsion
decreases. This can, under the appropriate conditions, lead to flocculation, and
hence an increase in viscosity. Consequently, processing operations which involve
size changes can greatly affect viscosity.
In summary, we note the following colloidal factors at our disposal that we
could use to change the viscosity of a given suspension
continuous-phase electrolyte concentration

change of surface charge by changing pH
surface coverage of particles by surface active agents
adsorption of block copolymers
changing total surface area by size reduction
15.13 The stability of suspensions

The equation that we gave in 6.4.2 for the velocity of a cloud of spheres
moving in a liquid under the influence of gravity, is sometimes applicable to the
stability of non-flocculated suspensions or emulsions, viz.
V =
138
2ga 2
(1 )5 0.25
9
where
is the density difference between the suspended and the continuous

phases,
a is the average particle radius,
g is the acceleration due to gravity (9.8 m /s2),
is the phase volume of the suspended phase, and
is the continuous phase viscosity.
If either the continuous phase is Newtonian, or else has a low-shear rate

Newtonian plateau controlling the velocity, then the equation can be used. If the
particle size a is less than a micron, then the final exponent in the equation is ~5.2,
while for particles greater than a micron, it is lowered to ~4.8. Figure 21 shows the
effect graphically, and illustrates that for a phase volume of 50%, the sedimentation
( > 0) or creaming ( < 0) velocity reduces to around 0.05 or 0.025 times the
velocity of a large or small isolated sphere respectively.
0.10
creaming
sedimenting
Normalised Stokes velocity, V
1.00
supe
r-m
icro
n
sub
-m
icr
on
~0.05
~0.025
0.01
10
20
30
40
50
Phase volume, / %
Figure 21: Normalised Stokes velocity as a function of phase
volume for super- and sub-micron particles respectively.
Here the overall velocity V is the same as that of the interface between the
concentrating suspension and the clear liquid left behind.
Exercise: consider what happens if salts are added to water to above their solubility
limit, bearing in mind the effect of crystal shape.
References
[1] Wagner, J., Zeit. Physik. Chem., 5, 31 (1891).
[2] Bancelin, M., Comp. Rend., 152, 1382 (1911).
[3] Barnes, H A, Dispersion Rheology 1980, a survey of industrial problems
and academic progress, Royal Soc. Chem., Industrial Div., London, 1981.
[4] Garcia-Salinas, M. J; de las Nieves, F.J., Progr. Coll. Poly. Sci., 110, 134 138
(1998).
[5] Krieger, I.M; Dougherty, T.J., Trans. Soc. Rheol., 3, 137 152 (1959).
[6] F. Baker, Trans. Chem. Soc., 103, 1655 (1913).
[7] Wakeman, R., Powder. Tech., 11, 297 299 (1975).
[8] Barnes, H.A; Hutton, J.F; Walters, K, An Introduction to Rheology, Elsevier,
Amsterdam 1989.
139
[9] Barnes, H.A, Colloids and Surfaces A: Physicochemical and Engineering Aspects,
91, 89-95 (1994).
[10] Barnes, H.A, Recent advances in rheology and processing of colloidal systems
Keynote Address in The 1992 IChemE Research Event, pp. 24-29, IChemE, Rugby,
1992, ISBN 0 85295 290 2.
[11] Barnes, H.A, JNNFM, 70(1/2), 1 - 33 (1997).
[12] Barnes, H.A, J. Rheol., 33(2), 329-366 (1989).
[13] Barnes, H.A, , JNNFM, 56, 221 - 251 (1995).
[14] Benbow, J; Bridgwater, J, Paste Flow and Extrusion, Clarendon Press, Oxford,
1993.
140
CHAPTER 16: POLYMER RHEOLOGY

One cannot help but be in awe ... contemplating ... the marvellous structure of reality,
Einstein
16.1 Introduction
The flow of polymer systems, whether solutions or melts, has fascinated
rheologists for many years. Since these systems often display very viscoelastic
properties, they have attracted more-than-expected attention from a wide range of
workers. We find polymer solutions in many every-day situations, especially among
the macromolecular (i.e. polymeric) solutions found in the human body such as
mucus, saliva, blood proteins, joint lubricating fluids, etc. In terms of natural
products, various gums and resins have proved very useful in many practical
situations. Then man-made polymers represent a huge industry, whose products
pervade every part of our lives, ranging from consumer-product packaging through
to the innumerable plastic articles of every-day life.
Sources of well-known polymeric material vary considerably, for instance
trees and plants

seeds
seaweed
fruit
grains
microbial
animal
petroleum
gum-arabic, cellulose
Guar gum, locust bean gum
carageenan, alginates, agar
pectins
starches
xanthan gum
gelatin, keratin and
acrylic acid, polyacrylamide, and many plastics.
The special properties of polymers arise because although macromolecules

are long and intertwined, they are nevertheless small enough to be subject to
Brownian motion that tends to randomise their configuration, giving linear chains
that ball-up to an approximate spherical shape. The ever-present Brownian
randomising is in completion with flow effects that seek to straighten isolated
polymer chains and comb-out entangled chains. All these effects combine to give
long-range interactions that persist to high shear rates. Also, when linear polymers
are stretched, their unique configuration allows for the transition to highly aligned
strings in the stretch direction. This gives a very high resistance to stretching
manifested in a large extensional viscosity, see later.
16.2
Some of the different kinds of polymer chains
In order of complexity of their configuration, the general forms of

macromolecules described in the following paragraphs are often encountered.
141
16.2.1 Long rods

Some polymers are naturally rod-like, for instance xanthan gum (Mw ~ 2 x
106, with a typical length up to 1 micron and an aspect ratio, L/d, of ~ 600), while
others can become rod-like. An example of the latter is a polyelectrolyte in a very
low electrolyte solution, where the electrostatic charges along the polymer chains
repel each other and force the convoluted chain to become straight. For rod-like
polymers with an aspect ratio of up to around 200, the relative increase in viscosity
compared with the solvent (sp = ( s)/s) is given by the approximate relationship
[1]
1 L
s 2.5 +
10 d
when the shear rate is low and the rods are randomised by Brownian motion.
However, when they become aligned along or across the flow at high shear rates, the
specific viscosity in both cases drops to approximately 2.5, thus becoming
independent of aspect ratio L/d. This is an obvious source of shear thinning, but also
shows that the viscosity difference between low and high shear rate increases with
aspect ratio. (Of course the Brownian motion is always trying to randomise the
rods, and the result is an accompanying elastic restoring force.)
The extensional viscosity (see chapter 17 for details) of semi-dilute suspensions
of long rod- or fibre-like entities (xanthan gum being a good example) is given by [2]
2
4
d
e = 3 s 1 + .
6
9 ln
This shows that the aspect ratio plays a dominant part in determining the extensional
viscosity.
16.2.2 Coils/Strings
These types of polymers can either be linear or branched, but because they
are usually made up of segments that are freely jointed, they can take up many
transient configurations under the action of Brownian motion. At a low-enough
concentration, when linear coils are isolated and non-interacting, they take up an
approximate spherical shape, the size of which is dictated by thermodynamics, see
figure 1. When they are sheared, these spheres become ellipses which are (slightly)
aligned to the flow. At the highest shear rates, the coils can be unwound into strings
completely aligned with the flow.
At high concentrations, the chains are completely intertwined and entangled
(see figure 2), and being under the incessant action of Brownian motion, they move
around like snakes in a snake pit, continuously sliding over each other, forming and
disengaging from individual entanglements as they move. Polymer melts are of
course made up of these chains alone, while concentrated solutions have solvent
between the chains which strongly influences the overall rheology.
142
Chapter 16: Polymer Rheology

entanglements
linear
branched
Figure 1: Isolated polymer coils subject to Brownian motion:

note the space-filling efficiency of a small amount of
material, with solvent immobilised within the coils.
Figure 2: Concentrated coils overlapping, and highly entangled.

Note that entanglements are continually forming,
disappearing and reforming, but at any one instant
there is a transient network.
Melts and concentrated solutions are elastic because tension in the chains between
entanglements persists while the entangled chains slide over each other. The
transient networks present in such situations are elastic on a short time-scale, but as
the stretched chains slide over each other and loose entanglements, they relax and
show normal viscous behaviour.
The controlling factors for polymer rheology include concentration, molecular
weight distribution, and degree of branching, as well as temperature and pressure.
Of course there is also the obvious difference between a melt and a solution, with
concentration playing no part in melt rheology (except for additives to the melt such
as plasticisers). Below a critical molecular weight, Mwe, the viscosity of a molten
polymer is almost directly proportional to the weight average molecular weight,
Mwav. However above this value of molecular weight, Mw, the proportionality
dramatically increases to a power-law relationship of around 3.4 [3], so
0 = K [Mwav ]3.4 .
The critical molecular weight is the point at which entanglements become important,
and below that value the chains may be considered as non-penetrating spheres. This
critical molecular weight depends on the chain configuration, but apart from some
important exceptions, it is of the order of tens of thousands.
Obviously
entanglements make flow more difficult, hence the higher dependence on molecular
weight. However, at very high shear rates, the entanglements disappear as the flow
combs them out; the dependence on molecular weight disappears, and the viscosity
is only dependent on concentration, see figure 3.
complete entanglement
g
en radu
de tan al
al velo glem loss
ig
nm pm en of
en ent ts &
t
of
complete alignment

Figure 3: The flow curve for a polymer solution showing
the two extremes of complete entanglement and
complete alignment.
143
All the individual curves for a series of molecular weights, temperatures and shear
rates can often be incorporated into a single master curve by plotting the nondimensional quantity
& Mw
against 0
, where is a constant near unity, is
T
0
the polymer density at T, the temperature in degrees Kelvin.

All else being equal, broadening the molecular-weight distribution generally
results in a polymer melt becoming non-Newtonian at lower values of shear rate, so
by the time we come to those shear rates relevant to extruding etc., the polymer is
easier to handle with a lower viscosity.
A small degree of branching of the polymer chains generally decreases the
viscosity of a melt with the same average molecular weight since the branched chain
is more compact. However, if the branching is extensive, for a reasonably high
molecular weight (~106) the viscosity at low shear rates may be much higher, given
the difficulties for these complicated interpenetrating structures to move. Imagine
the structure as a collection of stars, i.e.6. Nevertheless, in almost all cases branching
results in a lower viscosity at high shear rates.
16.3
Polymer solutions
10
10
10
10
4
3
2
20
Viscosity, / Pa.s
Viscosity, / Pa.s
The viscosity of polymer solutions, as well as depending on shear rate and

temperature (and in some extreme cases pressure), also depends on the molecular
weight distribution and the concentration, see figures 4 and 5.
-1
0.5
10
10
10
3.5
-1
10
2
1
0.35
-3
-3
-3
10
-1
10
10
10
. -1
Shear rate, / s
Figure 4: The flow curve for polystyrene in toluene,
Mw = 20M, for different concentrations of polymer
(shown as wt.%)
10
10
-3
-1
10
10
Shear rate, / s
10
-1
Figure 5: The flow curve for 3% w/w polystyrene in toluene,

for different molecular weights (shown in millions).
The solvent properties are also important since the quality of the solvent can vary for
organic liquids from good to bad, and have the effect shown in figure 6, with the
effect on the viscosity being then obvious.
In aqueous systems, some polymers have charges distributed along the
chains. In that case, the amount of electrolyte (salt) in the aqueous solvent will affect
the overall shape of the chain, see figure 7. In the absence of electrolyte, the charges
are unshielded and repel one another. This results in the chain being stretched out.
However, when electrolyte is present, the charges are shielded and their effect is
suppressed, resulting in the chain tending to shrink towards its natural random
configuration. This change from a rod to a sphere obviously causes the viscosity to
decrease.
144
10
good solvent
bad solvent
low electrolyte level
Figure 6: The effect of solvent quality on polymer coil dimensions.
high electrolyte level
Figure 7: The effect of electrolyte on the coil dimensions

of a polyelectrolyte.
= 3.5
slope
(log scale)
Zero-shear-rate viscosity,
As in polymer melts, there is a point at which entanglements take place in polymer

solutions. However, in the solution case it is controlled by concentration as well as
molecular weight. Figure 8 shows the dependence, which changes from a low to a
high dependence, being governed by the critical concentration c*.
slop
e=1
N.B. 2
3.5
> 10
so doubling concentration
increases viscosity by over
an order of magnitude!
.5
c*
concentration (log scale)
Figure 8: The zero-shear-rate viscosity of a polymer solution
versus polymer concentration, showing the critical
overlap concentration c*.
The intrinsic viscosity of many polymer solutions is given by the so-called

Mark-Houwink equation [3]
[ ] = KM
where M is the molecular weight and K and are constants for particular kinds of
polymers, for instance varies from 0.5 for a random coil to 2 for a rigid rod.
16.4
Highly branched polymers
When highly branched chains are sheared, they too unwind, but they cannot
form very long strings because of the branching. For this reason they do not give
stringy liquids, see chapter 17. This is important for thickening consumer products,
and for this reason the highly-branched Carbopol polymers are highly prized for
thickening without causing stringiness. Highly cross-linked polymers like these are
called micro-gels for obvious reasons. At quite low concentrationsoften less than
1% by weightthese microgel particles form a space filling network, which give a
145
high viscosity, but at the same time is extremely shear thinning, giving the
appearance of a high apparent yield stress, see chapter 15.
16.5
Living polymers or rod- or worm-like micelles
If the right molecular proportions and electrolyte concentrations are chosen,

mixtures of many surfactant molecules form themselves into rods or worm-like
threads, which are sometimes called living polymers, see chapter 18. Figure 9 shows
a schematic representation of these micelles.
Figure 9: A schematic diagram of worm-like micelles.
At concentrations of a few weight percent, these rods/worms behave in exactly same

way as the equivalent macromolecules, and entangle with each other to form
transient networks. The end result is strong viscoelastic behaviour. This is often
seen in transparent hair shampoos, shower gels, thickened bleaches, etc. The rods so
formed give the best possible thickening for a given amount of surfactant. In their
efforts to use this efficient thickening, some manufacturers overdo the viscoelastic
effect and produce liquids that are very stringy and verge on being unpleasant to
use.
Living polymers have one extra property that the normal polymer system
lack, and hence their name, as we now show. If either polymer chains or surfactant
worm-like micelles are subjected to a high extensional force, they can be snapped in
half. While polymer fragments always remain thereafter at this lower (halved)
molecular weight, the broken surfactant micelles can reform under more quiescent
conditions, since they are thermodynamic objects held together by reversible bonds.
If elastic liquids are needed to withstand continual damage, and yet survive, then
these micellar systems are much more mechanically robusta good example of this
is in turbulent drag reduction. For this reason they have been suggested for dragreducing fluids in central-heating systems. On the other hand, they are more prone
to chemical breakdown if electrolytes or metal ions enter the system.
Like all fibrous microstructures, the viscosity is high when the system is
randomly dispersed in space and highly entangled, but once the flow has aligned the
microstructures, the viscosity is quite low. This change from very high to very low
viscosity takes place over a narrow range of stress, see figure 10. This is the situation
for shampoos and shower gels.
146

Figure 10: The form of the flow curve for worm-like micelles,
e.g. shampoos.
16.6 Associative polymers

These are an interesting class of polymers which have hydrophobic (waterhating) pendant (side) and/or terminal groups on their hydrophilic (water-loving)
chains, making them similar in some respects to surfactant molecules. These
end/side-groups naturally clump together, either with similar groups nearby on the
same chain, or else with similar groups on adjacent chains, the latter interaction
producing a loose network. Each group only spends a given amount of time in such
an association before Brownian motion dislodges it, but sometime later the same
randomising force brings the end-group into another association. Thus if we take a
snapshot at any time, there is a gel-like network (showing viscoelastic properties),
but over time, there is flow (see [4]). The molecular weight of each polymer chain
can be quite low (104 - 105 daltons), but the overall effect is to produce a high
viscosity. The kind of network formed is visualised in figure 11.
intramolecular associations increase,
intermolecular associations decrease.
inc
rea
sin
gs
all associations
decrease
he
ar
-ra
t
e/s
tre
ss
Figure 11: The interactions between the hydrophobic endgroups

in associative polymers, [5].
It is also possible for shearing to produce new junctions between polymer

chains at the expense of those between end-groups along particular chains, and
hence the viscosity can increase after the low-shear-rate Newtonian plateau by up to
100%. However, the normal effect of shearing soon takes over afterwards when
associations are lost due to large stresses and then the viscosity decreases as usual,
see figure 12: this effect has also been seen in extensional flow [6]. These kind of
147
thickeners are often used in water-based coatings, drilling fluids, liquid laundry
additives and personal-care products.
Viscosity, / Pa.s
100
10-1
10-2
10-3
-2
10
-1
10
10
10
10
Shear stress, / Pa
Figure 12: The typical form of the flow curve for an
associative polymer solution, [5].
Obviously, the sites where the hydrophobic ends can interact can be blocked by
ordinary surfactant molecules and the viscosity decreases, or interestingly, the
associations can be strengthened by the addition of electrolyte. Thus the usefulness
of associative polymers is very formulation dependent.
16.7 A polymer solution as a standard non-Newtonian liquid
1st normal-stress coefficient
10
-1
10
viscosity
-3
10
10
10
10
10
-1
10
-2
10
-3
-2
10
10
Shear rate, / s
-1
10
10
G''
ear
lin
G'
qu
ad
ra
tic
10
Modulus, G or G / Pa
Viscosity, / Pa.s or 1 / Pa.s
It is worth giving especial consideration to a particular polymer solution

which has been extensively researched around the world and has now become a
commercially-available, non-Newtonian standard liquid. It is a 2% w/v solution of
poly(isobutylene) (molecular weight 4.3 .106 daltons) in a mixture of cis- and transdecalin. It is marketed by the Pure and Applied Chemistry department, University
of Strathclyde, Scotland, as Standard Fluid SUA1.
Data is provided on the complete rheological characterisation, relative to
steady-state shear stress and first and second normal stress differences as a function
of shear rate, as well as the complementary oscillatory data, see figures 13 and 14.
The measured extensional properties are also available, see figure 15 [7].
-3
10
-2
10
10
-1
10
Frequency, / s
-1
10
10
Figure 13: The viscosity and first normal-stress coefficient

Figure 14: The linear viscoelastic parameters G' and G'' as a
as a function of shear rate for the SUA1 standard polymer solution.
function of frequency for the SUA1 standard polymer
solution.
148
Extensional viscosity, e / Pa.s
10
10
10
10
10
10
-2
10
-1
10
10
10
10
-1
Extension rate, / s
10
Figure 15 : The extensional viscosity as a function of

extension rate for the SUA1 standard polymer
solution.
References
[1] Scheraga, H.A., J. Chem. Phys., 23, 1526 (1955).
[2] Acrivos, A; Shaqfeh, E.S.G., Phys. Fluids, 31(7), 1841 1844 (1988).
[3] Ferry, J.D., Viscoelastic Properties of Polymers, 2nd Edition, John Wiley, New York,
1970.
[4] Annable, T; Buscall, R; Ettelaie, R; Whittlestone, D., J. Rheol., 37(4), 695-726
(1993).
[5] English, R.J; Gulati, H.S; Jenkins, R.D; Khan, S.A., J. Rheol., 41(2), 427 - 444
(1997).
[6] Kennedy, J.C; Meadows, J; Williams, P.A., J. Chem. Soc. Faraday Trans., 91(5),
911 916 (1995).
[7] Hudson, N.E; Jones, T.E.R., JNNFM, 46, 69 88 (1993).
149
150
CHAPTER 17: EXTENSIONAL FLOW AND VISCOSITY

There was things which he stretched, but mainly he told the truth, Mark Twain
in The Adventures of Huckleberry Finn
17.1 Introduction
Although most viscometric and rheological studies are carried out in simple
shear flows such as rotational viscometers, real flows experienced by real liquids are
very often extensional (stretching or elongational) in nature, and for some liquids
there can be a very large difference between their shear and extensional viscosities.
To recap, in extensional flow the elements of a fluid are stretched out or
squeezed down rather than sheared. A good example of this is found in the flow in
and out of a short tube, see figure 1, which is the kind of flow experienced when
liquids such as ketchup, washing-up liquid and skin lotion are squeezed from plastic
bottles, or when toothpastes, meatpastes and processed cheese are squeezed from
tubes.
intense extensional flow

Figure 1: The deformation of a droplet from a larger to a smaller
diameter tube, where the entrance region is dominated
by extensional flow.
The history of a typical element of fluid in such a flow is shown in figure 1 as it

moves, initially unsheared, along the middle of the first tube (remember the shear
rate at the tube centre is zero), but then it has to stretch and deform in order to get
into the smaller tube. Once it is well inside the smaller tube the stretching ceases.
Then as the element emerging from the tube, it experiences compression along the
flow axis as it tries to return to its original shape. If the same deformable drop is
suspended in a liquid in simple shear flow, it is not deformed as much, see figure 2,
and rotates in the flow at a rotation rate (in radians per second) equal to half the
shear rate. For these reasons shear flows are called weak flows, while extensional
flows are called strong flows. It is also obvious that extensional flow is irrotational,
since our test element did not undergo any rotation in deforming.
151
at rest
shearing
45
Figure 2: The motion of a droplet in shear flow.

If our test element is a small fibre initially aligned at an angle to the flow lines
as it approaches the entrance to the smaller tube, the stretching flow it experiences as
it enters the small tube tends to align it along the flow axis, and as it emerges from
the smaller tube it would tend to move towards its off-axis angle again, see figure 3.
intense extensional flow

Figure 3 : The motion of a long fibre in and out of a small tube.
17.2 Where do we find extensional flows?

There are many everyday and industrial situations where there is a large
element of extensional flow. Some of the following are typical,
flow into and out of orifices,

break-up of liquid jets, droplet break-up, atomisation and spraying,
expanding bubbles, in foam and doughs,
unwanted atomisation (misting) in coating and printing operations,
drinking through a straw (entry flow) and swallowing,
blow-moulding and other inflation operations for plastic components,
man-made fibre spinning, and even spinning spiders webs,
flow through porous media, and some
rapid squeezing, forming and stamping operations.
In all these situations we can estimate the extension rate by calculating the
appropriate velocity gradient along the flow direction. This is done, as usual, by
dividing an appropriate velocity difference by an pertinent distance, as for instance in
our example above, by dividing the difference of average velocity between two
tubes, see figure 1, by the radius of the smaller tube.
17.3 Whats so special about extensional viscosities?
What is the effect of stretching and alignment on the microstructure of real
systems, and what resistances to flow (i.e. viscosity) do they show? First consider
152
Chapter 17: Extensional flow and viscosity
one of the simpler forms of stretching flowuniaxial extensional flow. This is the
equivalent of pulling out a filament or thread of liquid, see figure 4, where the
velocity gradient is in the flow direction, which is quite different from a shear flow
where the velocity gradient is at right angles to the flow direction. In the same way
that we have a stress and a deformation rate in shear flow, so now we have the
extensional stress (also called the tension), e and the extension (or elongational or
stretching) rate, & , with the ratio of the stress and rate giving the uniaxial extensional
(or elongational) viscosity, i.e. e =
& .
F, V
Tension, = F/A
.
Extension rate, = V/L
Figure 4: Definition diagram for uniaxial extensional flow:
note that the far end of the liquid thread is anchored.
The first major difference between the shear and the extensional viscosity
(apart from a mathematical feature which dictates that for all liquids at very low
shear rate, e = 3) is best illustrated by considering a single fibre suspended in a
Newtonian liquid. If the liquid with the suspended fibre is subjected to a shear flow,
the fibre will align along the flow axis, and then the disturbance it presents to the
surrounding liquid is a minimum. On the other hand, if the same suspended fibre is
subjected to an extensional flow, it again aligns, but now the resistance it presents to
the flow is maximised, since the liquid being stretched tries in turn to stretch the fibre.
This is reflected in the viscosity of a suspension made up of such suspended fibres in
which the viscosity will decrease as a flow aligns the fibre in shear flow, but will
increase the viscosity as it aligns in extensional flow, see figure 5. The eventual
difference between the shear and extensional viscosities at high deformation rates
increases with increasing aspect ratio of the fibres [1].
If the fibre concentration is much higher, there are many contacts between
individual fibres, and the number of these contacts then dominates the viscosity. In
fact the situation is very similar to a polymer solution or melt, in which case the
contacts are intermolecular entanglements. In both cases we want to know about
the dynamics of these contacts/entanglements as, for instance a function of extension
rate.
At rest the transient network formed as the fibres or chains become entangled
and disentangled under the action of Brownian motion, and stretching strongly
interferes with the dynamics of this transient network. If the inverse of the extension
153
viscosity (log scale)
3x
fully aligned by flow
rest state - random
.
. -1
Deformation rate, or / s (log scale)
Figure 5: The shear (dotted line) and extensional (solid line)
viscosities of a dilute fibre suspension, compared at
comparable deformation rates.
Shear or extensional
viscosity, or e
rate is greater than the average lifetime of the entanglements forming the network,
then momentarily the polymer segment between two entanglements is acting as part
of an elastic solid, and the stress increases considerably, until the entanglements
would rupture. If, in this situation, the average number of entanglements is constant,
then it can be shown that the extensional viscosity increases dramatically with
extension rate while the shear viscosity remains constant [2]. However, in reality, as
the deformation rate increases, the number of entanglements (and possibly their
lifetime) begins to decrease with increasing extension rate, then the viscosityafter
increasingdrops again, and at the same point the shear viscosity also begins to fall.
The overall effect is shown in figure 6. This explains the behaviour of many polymer
solutions and melts. This shows another difference between shear and extensional
viscosities.
(log scales)
.
Deformation rate, or
Figure 6: The shear and extensional viscosities of a polymer

network, network dynamics effects shown as solid lines,
and the effect of loss of junctions shown as dotted lines.
Last we note the most spectacular effect in extensional flow that is seen in
very dilute solutions of high-molecular-weight, linear polymers, where the
properties are dictated by the isolated individual polymer coils. At low shear and
extension rates, the polymer chains are balled up into a loose spherical shape, and
the viscosity is low. However, at a critical rate of extensional flow, the polymer
suddenly unwinds into a long, stretched string. This is called the coil-stretch
transition, and results in a spectacular increase in extensional resistance, to give a
viscosity that can be up to thousands of times higher than the equivalent viscosity in
shear flow, see figures 7 and 8. This is best seen when strings of liquids are pulled
154
3x
101
10-1
10-3
.
. -1
Deformation rate, or /s
(log scale)
viscosities of a dilute solution of linear polymer,
showing the coil-stretch transition from coiled to
unwound configuration.
coil
stretch
viscosity, or e / Pa.s
coiled
linear
chain
103
Stretched
linear chain
coil-s
t
trans retch
ition
viscosity (log scale)
out of the surface of such a solution, and it is virtually impossible to break them, due
to their very high viscosity. One practical application of this effect is the suppression
of atomisation in such liquids, see figure 8 for a real example of this phenomenon for
the ICI aviation-fuel antimisting agent FM9.
10-1
100
101
102
103
.
. -1
Deformation rate, or /s
104

viscosities of an aviation fuel with an anti-misting
agent added.
Recap.
The mechanisms that result in the extensional viscosity being much bigger than
the shear viscosity are
aligning particles with the flow results in maximum resistance
strong interaction of extensional flows and network dynamics
coil-stretch transition giving very long, aligned particles
A good example of what happens when two of these effects are present is shown in
figure 9, where the network dynamics dictate the first increase and decrease in
extensional viscosity, and coil-stretch accounts for the second increase, see [3].
2.0
1.0
0.5
10
10
Extension rate, /
s-1
10
Figure 9: Extensional viscosity versus extension rate for a solution of

polybutadiene in decalin, showing the extension/tension
thickening, extension/tension thinning and coil stretch
regions respectively.
17.4 The Trouton ratio

The difference between the two kinds of viscosities being considered is best
illustrated by introducing the so-called Trouton ratio, defined as the ratio of the
viscosities as measured at any particular shear rate and root-three ( 3 ) times the
equivalent extension rate. In all cases this ratio starts at 3 for small deformation
155
rates, and then at some critical point it can begin to rise. This is illustrated in figure
10. The rise can be quite spectacular in the case of the coil-stretch transition for dilute
solutions of high-molecular-weight linear polymers.
dilute linear coils
Trouton ratio
(log scale)
st
nge
f lo
e o time
e rs
Inv xation
rela
1000's
melts 5 - 100
er
polym
inelastic liquids
.
Deformation rate, or 3
Figure 10: The Trouton ratio (extensional viscosity divided by

shear viscosity) for various kinds of liquid.
For some polymer melts the rise can also be marked. In all cases the Trouton ratio
levels off at very high deformation rate. The eventual level of the Trouton ratio for
the coil-stretch type polymers can be thousands, while for polymer melts it can vary
from say 5 to 100, with typical values being around 15. High values of Trouton ratio
are coincident with high degrees of elasticity, as seen in branched polymer melts and
linear stretched coils. Colloidal systems where there are no alignable entities in the
microstructure will always have a Trouton ratio of around 3, i.e., they show little
elastic behaviour. Heavily-branched polymeric thickeners such as Carbopol do not
show a high Trouton ratio, and thicken very well without being perceived as
stringy.
The fact that the Trouton ratio eventually approaches a constant value means
that the shear and extensional viscosity flow curves are parallel in that range, and if
the behaviour is power-law, then the power-law index is the same in both cases.
(Notice this from the polymer melt examples shown later in the figures 13 - 20.)
17.5
Some examples of extensional viscosity curves
101
10
0p
pm
10-1
50
102
100
125
-1
10
10
.
ppm
Deformation rate, or / s-1
10
10-2
Figure 11: The shear (dotted lines) and extensional viscosities

(solid lines) versus deformation rate for aqueous
polyacrylamide solutions. Note the (thousand-fold)
different viscosity scales, [4].
156
101
m
0 pp
200
103
0 pp
101
Shear viscosity, / Pa.s
104
750
Figures 11 - 13 (see [4] and [5]) show examples of the different kinds of
uniaxial extensional viscosities as a function of extension rate, in some cases
compared to the equivalent shear viscosities.
100
10
-1
10-2
10
10
10
Deformation rate, or /
10
s-1
Figure 12: The shear (dotted lines) and extensional (solid lines)
flow curves for aqueous solutions of polyacrylamide
(tension thickening) and callogen( tension thinning)
Rheometrics promotional literature.
The obvious difference between the two kinds of viscosity is most marked in those
cases where the extensional viscosity increases with increasing extension rate, which
even if the viscosity eventually begins in decrease at much higher extension rates,
still results in the extensional viscosity being very much larger than the shear
viscosity.
6
10
HDPE
PS
PS PP
PP
10
10
HDPE
LLDPE
10
LLDPE
10
10
-1
10
10
10
10
10
10

Figure 13: The shear (dotted lines) and extensional (solid lines) flow
curves for typical polymer melts at 190 oC.
lin
ly
br
an
ch
ed
ear
.
. -1
.
Extension rate, (log scale)
(log scale)
BRANCHED
Extension rate, / s-1
Figure 15: The effect of temperature and molecular weight on the

shear (dotted lines) and extensional (solid lines) viscosity
of a polymer melt.
(log scale)
Extension rate, / s-1
d
wd wd
w d
m w
dm m
w
d m
oa
br arro
oa ow
br narr
n
d
wdmw
wd d
m mw
dmw
oa ro roadrow
br nar
b ar
n
Extensional viscosity, / Pa.s
Figure 14: The effect of branching on the extensional viscosity

of polymer melts.
LINEAR
molecular weight
gh
temperature
Extensional viscosity, e
(log scale)
hi
Extensional viscosity, e
(log scale)
The degree of branching of the polymer melt is crucial in this situation, as well as
molecular weight and temperature. In the case of polymer melts, the presence and
kind of fillers (spherical or alignable, i.e. sheets or fibres) is a strong factor in
dictating the form of the extensional flow curve. All these general effects are
illustrated in figures 14 - 20, shown either in general terms or with specific examples.
50%
35%
x3
0%
x3
x3
50%
35%
0%
Shear stress / tension, or e (log scale)
(log scale)
Figure 17: The shear and extensional viscosities of an LDPE melt

Figure 16: The effect of branching and molecular weight
with various landings of glass beads.
distribution on extensional viscosity of polymer melts.
157
The overall situation is summarised in figure 21, which shows that behaviour
ranges from simple suspensions and emulsions, where the extensional viscosity is
only a small multiple of the shear viscosity, to the coil-stretch situation of a possible
thousand-fold difference. The behaviour at the highest (relative) extension rates is
shown in figure 22, where a concentrated rod-like polymersay xanthan gumis
stretched very quickly, see [6]. Both shear and extensional viscosities flatten out as
the polymer chains are completely untangled and aligned, and show and
approximately forty-fold difference.
10
40%
25%
60%
40%
0%
25%
0%
10 -3
10-2
10-1
10. 0 . 10 1
10 2
LDPE
50/50
HDPE
LDP E
HDPE
50/50
LDPE
HDP E
.
-1
Figure 20: The shear and extensional viscosities of two

polymer melts and their 50/50 blend. Note that low
density polyethylene is highly branched.
10
10
10
10
-1
-1
10
10
HDPE
158
(log scale)
pol
ym
er s
p ol
x3
ol u
ti
o ns
ym
er m
el t s
coll
shear v
o
id al
iscosit
y
sy st
em s

Figure 21: The extensional flow curves of various kinds of liquid.
10
viscosities of 0.04% xanthan gum in a water/syrup
mixture.
17.6
-1
Figure 19: The shear and extensional viscosities of three

polymer melts.
Deformation rate, or / s
Extensional viscosity,e
(log scale)
Figure 18: The effect of mica filler level on the shear (dotted lines)
and one- third of the extensional (solid lines) viscosity
of a polymer melt.
LDPE
NYLON
shear
10
60%
extensional
10
10
The effect of time or extension on extensional viscosity
Extensional viscosity, / Pa.s
The typical response of materials with interesting extensional flow curves (i.e.
polymer melts) is shown in figure 23 for linear polymer melts and in figure 24 for
branched polymer melts. Both take quite a time to come to steady state, showing
first the typical viscoelastic response seen in shear flow. The most important point
about this data is that it represents not only long times to steady state, but also large
extensions. In a fibre-pulling experiment, this can mean an extension of the fibre to
over 1000 times its original length! This makes obtaining steady-state values very
difficult for very viscoelastic systems.
Time, t (log scale)
Figure 23 : The extensional viscosity build-up for relatively

inelastic polymer melts, as a function of time for
different extension rates, projected onto the
extensional flow curve.
17.7
Time, t (log scale)
Figure 24: Extensional viscosity build-up for very elastic polymer

melts, as a function of time for different extension rates,
projected onto the extensional flow curve.
Other kinds of extensional flows
Apart from the uniaxial extensional flow considered above, there are two
other kinds of flows where the material is stretched, see figure 25.
e.g. thread
pulling
e.g. balloon
inflation
uniaxial
biaxial
e.g. cylinder
inflation
planar
Figure 25: The three different kinds of extensional flow with an

example of each.
First we have
biaxial extensional flow, where a sheet of liquid is pulled out in two directions: this is
the kind of deformation experienced by an portion of a balloon or a bubble being
inflated.
Here we have stretching in two directions, and compression in one
direction as the thickness of the element decreases.
Then we have
159
planar extensional flow, as seen for instance when a long cylinder is inflated, and the
stretching only takes place in one direction, but the thickness of the sample
decreases, i.e. it is effectively compressed, with the third dimension unaffected.
These two kinds of extensional flows are less demanding on microstructures such as
individual polymer chains, since uniaxial extension demands that a threedimensionally random chain be pulled towards a straight extended chain.
All these extensional flows have different values of viscosity associated with
them, so that in the simplest case for Newtonian liquids - or for all liquids at low
enough shear/extension rates - which are as follows
Extensional flow
Uniaxial
Biaxial
Planar
17.8
Extensional Viscosity
3
4
6
compared with
shear viscosity =
Viscometers to measure extensional viscosity
How do we measure extensional viscosity? Apart from a limited number of

applications where it is possible to form filaments and sheets and pull them in one or
two directions, the answer is that it is much more difficult to perform extensional
viscosity measurements than their shear viscosity equivalents. For polymer melts, it
is possible to chill the clamps used to hold the melt, and so secure the sample, see
figure 26. Sometimes, it is even possible to use a super-glue to attach the sample to
the moving clamps!
clamping device
flexible delivery pipe
Pull-down
force F
liquid thread
fibre or thread
Wind-up
velocity, V
knife
fibre pulled under constant tension

or at constant extension rate
Figure 26: Two thread-pulling methods of measuring
extensional viscosity.
Another problem is that even for the simplest stretching flowuniaxial

extensionwhen we have succeeded in clamping the sample, the nature of
extensional flow is such that we need to increase the length of the sample
exponentially in order to ensure a constant extension rate. Second, when we are
doing this, the cross-section of the sample is continually decreasing, and then if we
have decided to initiate the flow by applying a set force, the stressforce divided by
cross-sectional areais constantly increasing as the cross-section decreases. A
number of ingenious ways have been found to get around this problem, and it is
160
possible to apply a constant stress and to measure the deformation, or else to apply a
constant extension rate and measure the resulting stress.
However, when we have mobile solutions, the methods described above are
impossible, and we have to try to create an extensional flow within a flowing system.
This has been done in a number of ways, as shown in figure 26. The method used
depends on the liquid of interest being spinnable or not spinnable. If the solution
is spinnable, then it is possible to wind up the liquid thread on a rotating drum. In
all these cases the average extension rate is measured using a camera system to
record the profile of the stretched liquid.
Where a liquid is not spinnable, it is possible to use the opposing jets system.
Here two identical jets are immersed in a liquid, quite near to, but opposing each
other, an equal amount of liquid being withdrawn through each. The flow in the
liquid between the jets contains a high degree of extensional flow, and by noting the
jet orifice size and the flow rate as well as the reaction on one of the jets, then a
measure of the extensional viscosity can be obtained, see figure 27.
test liquid
Pressure,
P
flow-rate, Q
Reaction measured Test liquid pumped
on one arm
equally up both arms
Figure 27 :Two flow geometries for measuring extensional
viscosity; the opposed orifice device and the sharpedged orifice.
In all these measurements there is one great recurring problemone is never

sure if steady state has been achieved. In simple rotational shear instruments, the
attainment of steady state is just a matter of waiting. However in extensional
instruments the amount of deformation is usually small compared with that
necessary to reach equilibrium, which for some polymeric systemsas we have
seencan sometimes only be reached at over 1000 times extension of the
undeformed sample at its rest state.
There is one simple experiment that can be performed to obtain a fair
estimate of the extensional viscosity in practical situations, this is by measuring the
entrance pressure into a capillary of radius a, see figure 27. Then a reasonable
expression for the extensional stress is the entrance pressure drop Pe and an equally
simple expression for the extension rate is Q/a3. The next level of sophistication is
to carry out a power-law analysis, this was done by Cogswell [7] who the suggested
that
e =
&
3
(n + 1) Pe and & = w w
8
2 e
161
where is the wall shear stress inside the small pipe, Pa/2L, and &w is the wall
shear rate inside the small pipe,
17.9
4Q 3n + 1
.
a 3 4 n
Commercially available extensional viscometers
As we see, these procedures are quite difficult, and thus it is not surprising
that few commercial instruments are available to measure extensional viscosity. The
present choice is quite small, being limited to Rheometric Scientifics RME
extensional viscometer for polymer melts.
Anyone interested in measuring
extensional viscosity of other systems will have to resort to home-made instruments.
17.10
Some interesting applications
There is a natural tendency for liquid threads to break up due to surface

tension forces. This break-up is governed by the extensional viscosity of the liquid
thread under the prevailing stretching conditions. If the extensional viscosity is then
high, it will oppose any local thinning tendency as a thread is stretched, see figure 28.
This resistance is especially enhanced if the stretching takes place in the region where
the viscosity is not only high, but is also increasing with extension rate, since the
more a thread thins, the greater is the extension rate, and then the greater is the
resistance. This is true in at least two situations, first for dilute solutions of highmolecular-weight linear polymers. In this case atomisation of watery liquids is
inhibited, see figures 7 and 8, since ligaments of liquid being ripped apart as they fly
through the air in the atomisation process undergo strong extensional flow, and the
resisting force will be high. In this case the situation is as illustrated in figure 28
where the break-up normally produces small drops called satellites. These are
inhibited and a clean break is seen, with the suppression of satellite drops, see also
figure 8. If the extensional viscosity is very high, atomisation is completely
suppressed.
low
viscosity
satellite
drop
medium
viscosity
high
viscosity
Figure 28 : The break-up of liquid filaments when pulled out.
Then, secondly, in the production of man-made fibres, polymer melts are

extruded through many-holed spinnerets, and the hardening threads are quickly
wound up unto drums, i.e. they are stretched. If the extensional viscosity of the stillmolten stretched portion of the threads is an increasing function of extension rate,
the tendency for the thread to neck and break is inhibited, see the right-hand side of
162
figure 28. In this way the maximum spinning rate is increased for high extensional
viscosity.
For mobile liquids with high extensional viscosities, it is possible to perform
the inverted siphon experiment, where the spinnable liquid is sucked out of a surface
and the tube drawn up to a suitable height, see figure 29. This is the ultimate
demonstration of tension along the streamlines and non-breaking of liquid threads!
Last we note the effect of large Trouton ratios on flow through contractions,
see figure 30. For this kind of flow, the flow pattern is not predetermined by the
geometry, as for instance in simple pipe flow, and in contractions the liquid can
choose its own type of flow, so as to minimise the overall energy dissipated in the
system. In order to do this, the change in velocity from the large tube into the small
tube is more spread out for the high Trouton ratio liquid, so that the average
extension rate is lower, thus lowering the overall energy dissipation.
distance over which most of the velocity change takes place
stretching flow
low Trouton ratio
Figure 29: Cross-section of the inverted (tube-less) siphon.
high Trouton ratio
Figure 30 : The flow

flow for polymer
, patterns in converging
.
melts with low and high Trouton ratios respectively,
e.g. LDPE and HDPE.
References
[1] Brenner, H., Int. J. Multiphase Flow, 1, 195 341 (1974).
[2] Lodge, A.S., Elastic Liquids, Academic Press, London, 1964.
[3] Ferguson, J; Kemblowski, Z., Applied Fluid Rheology, Elsevier Applied Science,
London, 1991.
[4] Barnes, H.A; Hutton, J.F; Walters, K., An Introduction to Rheology, Elsevier,
Amsterdam 1989.
[5] Barnes, H.A, Polym. Mat. Sci. Enging, 61, 30-37 (1989).
[6] Zirnsak, M.A; Boger, D.V; Tirtaatmadja, V., J. Rheol. 43(3), 627-650 (1999).
[7] Cogswell, F.N, Polym. Eng. Sci., 12, 64 (1972).
163
164
CHAPTER 18: RHEOLOGY OF SURFACTANT SYSTEMS

If its thick, its strong, a consumer interviewed by the author
18.1 Introduction
The rheology of surfactant (otherwise called detergent) systems is important
for such products as
washing-up liquids,
shower gels,
hair shampoos
liquid fabric surfactants,
liquid abrasive cleaners and
rinse conditioners,
as well as playing a role in other household commodities such as toothpastes, or

even thickened bleaches. The rheology of such products is important in terms of
consumer preference for thick and creamy liquids, as well as technical aspects that
have to be built in, such as slowing down drainage on vertical surfaces by increasing
the viscosity, or high viscosity for ease of handling of such products as hair shampoo.
18.2 Surfactant phases
Although surfactant molecules are quite small, they can aggregate together to
give large structures of up to 1m in size. (None of these systems are essentially
permanent, since the individual surfactant molecules on their outside are constantly
interchanging with those in solution.) These remarkable structures give rheological
effects that mirror those seen in polymer solutions, or in particulate suspensions with
apparent yield stresses. They result from the surfactant molecules schizophrenic
nature in that their head-groups love water (hydrophilic), while their organic tails
hate water but love oil (hydrophobic or oliophilic). When a surfactant is dispersed in
water, up to a certain low concentration, the molecules disperse monomerically.
However, above this level, we see the beginning of the formation of associations of
molecules, which are called micelles: this particular concentration is called the critical
micelle concentration, or the c.m.c.
The particular shape of the micellar structure that forms above the c.m.c is a
function of the shape of the surfactant molecules involved; the electrolytic nature of
the water phase and the temperature. Because the tails hate water, the molecules
aggregate in such a way as to shield the tails from water by clumping together.
There are various shapes that result from this clumping, where the head-groups also
line up.
165
The structures formed include
spherical micelles
rod-like micelles
lamellar phase sheets
lamellar phase droplets
These structures (which are sometimes called vesicles) are shown schematically in
figures 1 - 4 , see also 16.5.
Figure A
1: diagrammatical representation of spherical micelles;
some shown in cross-section .
166
Chapter 18: Rheology of Surfactant systems
Figure 2: A diagrammatical representation of some

rod-like micelles.
Figure 3: A diagrammatical representation lamellar phase sheets.
167
water layer
surfactant layer
N.B. only a few layers shown
Figure A
4: diagrammatical representation lamellar-phase
droplets,
with a cross-sectional views of two droplets .
18.3
Rheology of surfactant systems
Clearly the interesting surfactant structures just mentioned will all have
important rheological effects. First, the formation of spherical micelles results only in
Newtonian behaviour over the normal range of shear rates (<104 s-1), given that the
spheres are only a few times greater in diameter than the length of the surfactant
molecule, which is around 50 nm. However due to the high phase volumes
involved, the viscosity can be many times greater than water. These structures are
often present in washing-up liquids, which have viscosities around a few hundred
millipascal seconds.
The next degree of complexity comes whenfor certain mixtures of
surfactant moleculesthe structures are present as long rod- or worm-like
structures: they are also called living molecules, see chapter 16. These worm-like
structures behave as polymers, so they greatly increase the viscosity of the micellar
solution via entanglements. They also impart significant viscoelastic properties.
However, a special feature is the high degree of shear thinning due to entanglements
being lost as the micelles align in the flow. Typically, these systems have a surfactant
concentration of less than 10%. Products such as some washing-up liquids, and most
shampoos, shower gels, hair shampoos and thickened bleaches rely on this kind of
organisation of molecules. The length of these worms can be up to 1 m, but
because the diameters is only a few times the length of the constituent molecules,
their length to diameter can be very large, but they are always transparent (a simple
check).
The next molecular organisation is more complex. This involves many layers
of alternating molecules, intercalated with layers of wateri.e. lamellar phase. It is
possible for rafts of these layers to float around in water, but for most real products,
they present themselves as fully enclosed multi-layered onion-like droplets. The
number of layers can run into hundreds, and these so-called lamellar drops (or
vesicles) can be up to several microns in diameter.
The distinctive feature of these structures is the very large amount of water
that they trap between the surfactant layers. For a surfactant concentration of
around five percent, the total phase volume of the drops can be greater than 70%,
168
Chapter 18: Rheology of Surfactant systems
thus giving space-filling, close-packed spheres. These systems are thus very shear
thinning and appear to have yield stresses, and hence are good for suspending
particles as in liquid abrasive cleaners. These structures also account for the high
viscosity of low-concentration rinse conditioners.
18.4 Necessary rheological properties
What are the rheological properties of surfactant-base liquids necessary to
fulfil our expectations of them? The following examples might be cited:
(a) lowering viscosity for a high concentration of surfactant:
washing up liquids
concentrated rinse conditioners
liquid fabric-washing liquids
spray cleaners
(b) keeping viscosity high enough for ease of handling:

shower gels
hair shampoos
liquid soaps
(c) increasing viscosity to slow down run-off from vertical surfaces:
thickened bleaches
(d) suspension of particles without visible sedimentation:
liquid abrasive cleaners
liquid fabric-washing liquids.
These desired properties are best brought about by the manipulation of the
micelle structures formed by the proper choice of mixtures of surfactants as well as
the correct processing procedures needed to assemble the more complex structures.
Other chemical species present in real surfactant formulations (perfumes,
hydrotropes etc.) can also play a major role in manipulating micellar form if present
at significant levels.
Exercise: Produce a prescription for the rheology of a liquid abrasive cleaner that has
to suspend solid particles, but has to readily drain down a vertical surface.
169
170
CHAPTER 19: RHEOLOGY OF FOOD PRODUCTS

Science is nothing but perception, Plato
19.1 Introduction
Apart from the smell, taste and colour of food, their flow properties are high
on the list of consumer-perceived attributes. Also, eating and drinking food
products are probably the most common applied rheological experiences that the
average person encounters. Who has not appreciated the thickness of cream; the
creaminess of soup and the texture of creamed potato? The perception of these
verbally-expressed properties arises from the interaction between the rheology of the
food product and movement in the mouth, tongue and throat. This is a very
complex operation, but it is something that is done without real thought but
nevertheless from it arises a perception of the food expressed in a wide variety of
words. The problem facing the food scientist interested in rheology is first obtaining
a proper rheological description of the food being considered and second to describe
the flow field in the eating, drinking and swallowing operations.
The processing side of food rheology is how to transform the basic
ingredients, viz.
Sugars
Polysaccharides
Lipids
Proteins
Colours
Flavours
Vitamins
Preservatives
Minerals
Water
into a palatable form with the correct microstructure that delivers an acceptable
rheology.
19.2 Liquid-like food rheology
Liquid-like food is usually made up of a complex mixture of dissolved
natural polymers and dispersed insoluble biological material, usually in an aqueous
phase with various electrolytes. The rheology varies over the range
thin liquids
viscous liquids
very viscous liquids
elastic liquids
pastes
soft solids
e.g.
drinks, consomms, oils

creamed soup, mayonnaise
ketchup, yoghurt
dough, some soups
meat paste, processed cheese
pats, spreads.
171
For many of these products, we all know that they can be too thick, or too thin and
thus, either way in the situation being considered, unacceptable.
Obviously the rheology of these materials is very diverse, and even at a
superficial level, the level of viscosity as well as the degree of non-Newtonianness
varies over a wide range, see figure 1.
"Viscosity"
syrup
ice
cream
salad yoghurt margarine
cream
tomato
ketchup pate
soft
cheese
honey
salad
dressing
cream
gravy creamy
soup
tomato
juice
oil
consomm
water
"Non-Newtonianness"
Figure 1: A 'map' of food qualitative rheology.
19.3 In-vivo flow fields

The flow field experienced by a liquid as it is ingested is complex. For the
few seconds that the liquid is in the mouth, it obviously experienced shear flows, but
also as the tongue is pressed against the pallet, or when the liquid flows around the
mouth and moved through small orifices it also encounters extensional flow. For
simple non-Newtonian liquids, the average shear rate in the mouth has been
estimated as 50 s-1; the typical extension rates are similar.
The interaction between these flow fields and the perceived experience of
eating and drinking is complicated, but it is possible to relate verbal descriptors such
as creamy or thick to measured rheological properties.
A high degree of viscoelasticityexperienced as stringinessis obviously
unacceptable to consumers, but some viscoelasticity is inevitable, given the kinds of
high molecular-weight polymers such as proteins that are dissolved in the aqueous
phase.
19.4 Rheology and food processing
Whilst the rheology of finished food products is crucial for a successful brand,
there are as many rheology problems encountered in manufacturing. These include
mixing, pumping, as well as flow in heat exchangers and even atomisation of
powder precursors in spray dryers.
The kind of mixing that can cause a problem is due to severe shear thinning
that resembles a yield stress. It is then possible to use the cavern theory approach
described in chapter 10. Then we can be assured that the well-mixed zone fills the
whole mixer.
Shear thickening in food systems is found for concentrated starch dispersions.
Often in the early parts of the production of a food product, this particular situation
is encountered, and problems can arise. By following the shear thickening map
shown in chapter 15, these difficult areas can be avoided.
172
Chapter 19: Rheology of food products
Exercise: Using whatever rheometers you have at your disposal, measure the
rheological properties of tinned custard.
173
174
CHAPTER 20: YOUR RHEOLOGY LIBRARY

The man who does not read good books has no advantage over the man who can't read them, Mark Twain
20.1 Introduction
There are a number of journals that are primarily concerned with rheology, which you
should seriously consider subscribing to:
Applied Rheology
Rheologica Acta
Journal of Rheology
Journal of Non- Newtonian Fluid Mechanics
Rheology Abstracts
Your librarian or else your usual journal supplier will be able to arrange this for you.
20.2 Rheology Books in print
To help you move on from being a mere beginner in rheology, you should start your
own Rheology library. To assist you, here a list of all the currently available books that you can
buy on the subject rheologymore than 160from which you can begin to build up your
collection. Unless otherwise stated, the approximate UK price in pounds sterling of hardback
versions for the books is given: sometimes cheaper paperback versions are available.
The list is drawn up according to particular areas of interestGeneral Rheology,
Theoretical Rheology, Polymeric Systems, Blood, Food, Other Specific Systems, Drag Reduction, Other
Meetings, Geological Subjectsand titles appear within each section in alphabetical order. The
abbreviations used for publishers names are well known in the book trade, and the information
supplied should be sufficient for any reputable book supplier to obtain copies of the books
listed.
20.3 General rheology
Advances in the flow and rheology of non-Newtonian Fluids, Eds. D.A. Siginer, D. De Kee &
R.P. Chhabra, Elsevier, 1999, $380
Applied fluid flow, Cheremisinoff N.P, Marcel Dekker, 1979, $125
Applied fluid Rheology, Ferguson J, Kemblowski Z, Chapman & Hall, 1991, 65
Bubbles, drops and particles in non-Newtonian fluids, Chhabra R.P, CRC Press,
175
1992, 179
Crack and contact problems for viscoelastic bodies, Graham G.A, Springer Vienna,
1995, 47
Creep and relaxation of nonlinear viscoelastic materials, Findley W.N,
Dover Pubns., 1989, $12.95
Developments in non-Newtonian flows (meeting), Siginer D.A, ASME, 1994, $74
Developments & applications of non-Newtonian flows 1995 (meeting), Siginer D.A,
ASME, 1995, 72
Encyclopedia of fluid mechanics: Rheology and non-Newtonian flows, Vol. 7:
Cheremisinoff N, Gulf Publishing, 1988, 157
Engineering Rheology, Tanner R. I, Oxford UP, 1985, 65
English-Chinese dictionary of Rheology, Science Press Staff, Fr. & Eur., 1990, $49.95
Free liquid jets and films: hydrodynamics and Rheology: interaction of mechanics and mathematics,
Yarin A, John Wiley & Sons, A$246
Fundamentals of heat transfer in non-Newtonian fluids, Chen J,
Amer. Soc. Mech. Eng., 1991, $30
Interfacial transport processes and Rheology, Edwards D, Butterworth, 1991, 65
(An) Introduction to Rheology, Barnes H.A, Hutton J.F, Walters K, Elsevier, 1989, Df 230
(An) Introduction to Rheology, Laba D, Micelle Press, Weymouth, 1997, 11.50
Mechanics of viscelastic fluids, Drozdov A , John Wiley Ltd, 1998, 60
Non-Newtonian Flow in the Process Industries, Chhabra, R.P, Richardson J, ButterworthHeinnemann, 1999, 35.
Non-Newtonian fluid mechanics, Bohme G, North-Holland, 1987, $256
Non-Newtonian Fluids: Fluid Mechanics Mixing & Heat Transfer, Wilkinson W. L,
Elsevier Science, 1999 (ISBN: 0080418635), ?
Optical rheometry of complex fluids, Fuller G.G, OUP, 1995, $60
Physics of flow (A level), Open University staff, Heinemann Educational, UK, 1997, 8.50
Large deformation of materials with complex rheological properties at normal and high pressures,
Levitas V.I, Nova Science, 1996, $97
Progress & trends in Rheology (meeting), Giesekus H, Springer Verlag, 1988, $217
Rheological Measurement, Collyer A.A & Clegg D.W, Elsevier Applied Science, 1988, 490
Rheological modelling: thermodynamical and statistical approaches (meeting),
Casas-Vasquez J, Springer-Verlag, 1991, 30
Rheological phenomena in focus, Boger D.V & Walters K, Elsevier, 1993, $149.50
Rheological techniques, Whorlow R.W, Ellis Horwood, 1992, 83.95
Rheology : An Historical Perspective, Tanner R I, Walters K, Elsevier Science, 1998, $201.00
Rheology, principles, measurements and applications, Macosko C.W,
VCH, 1994, 66
Rheology and elastohydrodynamic lubrication, Jacobson B.O, Elsevier, 1991, $175.75
Rheology & non-Newtonian flow, Harris J, Bks Demand (reprint soon), $107.20
Rheology for Chemists An Introduction, Goodwin J.W & Hughes, R.W, Royal Society of
Chemistry, Cambridge, 2000, 27.50.
Rheology Fundamentals, Malkin A. Ya, (Institute of Physics), ChemTec Publishing, 1995, $135
Rheology of disperse systems (meeting), Mill C, Franklin, 1959, $108
Techniques in rheological measurement, Collyer A.A, Chapman & Hall, 1993, 75
The structure and rheology of complex fluids, Larson R.G, Oxford, 1999, $79.95
176
Chapter 20: Your Rheology library
Synthesis & degradation-Rheology & extrusion, Cantow H, J Springer Verlag, 1982, $62
Transport phenomena, Bird R. B, Wiley 1960, $57.50
Transport properties of fluids: thermal conductivity, viscosity and diffusion coefficient,
Kestin J, Taylor & Francis, 1988, 96
Understanding Rheology, Morrison F.A, OUP, 2001, 50.
(The) vitreous state: thermodynamics, structure, Rheology, crystallisation, Gutzow I,
Springer-Verlag, 1995, 86
Viscoelastic machine elements: elastomers and lubricants in machine systems,
Moore F, Butterworth Heineman, 1993, 75
Viscoelasticity of Engineering Materials, Hadda Y, Chapman & Hall, London, 1995, 75
20.4 Theoretical Rheology
Advances in constitutive laws for engineering materials (meeting), Fan J, Elsevier,
1990, $345
Advances in finite deformation problems in materials processing and structures,
Chandra M, Amer. Soc. Mech. Eng, 1992, $74
Applications of supercomputers in engineering: fluid flow and stress analysis applications:Vol 2,
Bebbia C.A, Comp. Mechanics, 1989, 69
Approach to Rheology through multivariable thermodynamics, Hull H.H, Hull, 1981, $16
Computer programs for rheologists, Gordon G.V, Hanser Garner, 1994, 60
Constitutive equations for anisotropic & isotropic materials, Smith G. F., Elsevier 1994 $203.25
Constitutive laws & microstructure, Axelrad D, Springer Verlag, 1988, $56.95
Constitutive laws for engineering materials, Qesai C.S, ASME, 1991, $200
Constitutive laws of deformation, Chandra J, Soc. Indus. & Applied Mathematics,
1987, $40.75
Constitutive equations for engineering materials (meeting) Vol.1, Chen W.F,
Elsevier, 1994, $359.50
Constitutive laws of plastic deformation and fracture (meeting), Krausz A.S,
Kluwer, 1990, 87.95
Elements of stability in viscoelastic fluid, Dunwoody J, Longman, 1990, 22
(The) finite element method: Solid and fluid mechanics: nonlinear problems: Vol. 2,
Zienkiewicz O.C & Taylor R.L, McGraw-Hill, 1991, 29.95
Fluid dynamics of viscoelastic liquids: Applied mathematical sciences: Vol 84,
Joseph D.D, Springer-Verlag, 1990, 69
Functional & numerical methods in viscoplasticity, Ionescu I.R, OUP, 1993, $85
Numerical simulation of non-Newtonian flow, Crochet M.J, Davies R & Walters K, Elsevier, 1984,
$168.75
Mechanics of non-Newtonian fluids, Schowalter W.R, Franklin (Pergamon Press Reprint),
1978, $146
Mechanics of viscoelastic fluids, Zahorski S, Kluwer, 1982, 106.50
Numerical simulations of heat transfer and fluid flow on a personal computer, Kotake S,
Elsevier Science, 1995, $85.75
Numerical analysis of viscoelastic problems: recherches en mathematiques appliques,
Le Tallec P, Springer-Verlag, 1990, $25
177
Numerical methods for non-Newtonian fluid dynamics, Vradis G.C, ASME, 1994, $35
(The) phenomenological theory of nonlinear viscoelastic behavior: An introduction,
Tschoegl N.W, Springer-Verlag, 1989, 79.50
Thermodynamics & constitutive equations, Grioli G, Springer Verlag, 1985, $30
Thermodynamics of irreversible processes: applications to diffusion and Rheology,
Kuiken G.D.C, John Wiley Ltd, 1994, 35
Thermodynamics & Rheology, Verhas J.O, Kluwer, 1996, 74
The thermodynamics of Rheology (or inside the thermodynamic black box), Hull H.H, Hull, 1995, 20
20.5 Polymeric systems
Advances in polymer science, Nijenhuis K, Springer Verlag, 1997, 130
Amorphous polymers and non-Newtonian fluids: Volumes in Mathematics and its application,
Dafermos C, Springer-Verlag, 1987, $29
Analytical polymer Rheology: Structure-Processing-Property Relationships,
Rohn C, Hanser Garner Pubns., 1995, $125
Chemorheology of thermosetting polymers, May C.A, Am. Chemical, 1983, $49.95
Constitutive equations for polymer melts and solutions, Larson R,
Butterworth Heinneman, 1988, 45
Dynamics of Polymeric Liquids, Bird R.B, Armstrong R. C & Hassager O, Volume 1, Fluid
Mechanics, 2nd ed., Wiley, 1987, 115
Dynamics of Polymeric Liquids, Bird R.B, Hassager O, Armstrong R. C & Curtiss F.C, Volume 2,
Kinetic Theory, 2nd ed., Wiley, 1987, 115
Flow and Rheology in polymer composite processing, Advani S, Elsevier, 1994, $203.25
(The) flow of high polymers: continuum & molecular Rheology, Middleman S, Bks Demand
(reprint soon), $72.70
(An) introduction to polymer Rheology and processing, Cheremisinoff N,
CRC Press , 1992, 95
Mechanical properties of polymers & composites, Neilsen L.E, 2nd ed., Marcel Dekker, 1994, $165
Melt Rheology and its role in plastics processing: theory and application, Dealy J & Wissbrun K.F,
V Nostrand Reinhold, 1990, 61.50
Nonlinear viscoelastic effects in flows of polymer melts & concentrated polymer solutions,
Leonov A.I, Elsevier, 1993, 72
Nonlinear phenomena in flows of viscoelastic polymer fluids, Leonov A.I & Prokunin A.N,
Chapman & Hall, 1994, 95
Polymer alloys and blends: thermodynamics and Rheology, Utracki L, Hanser, 1990, 60
Polymer melt Rheology: a guide for industrial practice, Cogswell F.N,
Technomic Pub Co, 1996, $110
Polymer mixing and extrusion technology, Cheriminoff P.N, Marcel Dekker, 1987, $170
Polymer Rheology: theory & practice, Yanovsky Y, Chapman & Hall, 1993, 75
Polymer Rheology, Nielsen L.E, Marcel Dekker, 1977, $125
Polymer Rheology, Lenk R. S, Chapman & Hall, 1978, 95
Polymer Rheology and processing, Collyer A.A, Chapman & Hall, 1990, 75
Polymers as Rheology modifiers, Schultz D & Glass J.E, Amer. Chem. Society, 1991, $84.95
Principles of polymer engineering Rheology, White J.L, John Wiley, 1990, 75
Principles of Rheology for polymer engineers, White J. L, JohnWiley, 1990, $75.95
178
Rheological fundamentals of polymer processing (meeting), Covas J.A, Kluwer, 1995, 394
Rheology and processing of liquid crystal polymers, 2, Acierno D & Collyer A. A, Chapman & Hall,
1996, 69
Rheology of filled polymer systems, Shenoy, A.V, Kluwer, 1999, 90
Rheology of polymeric systems: principles & applications, Carreau P.J, De Kee D. C. R, &
Chhabra R.P, Hanser-Gardner, 1997, $197.5
Rheology for polymer melt processing, Piau J.M, Elsevier Science, 1996, $243
Rheology of polymer melts (IUPAC meeting), Kahovec J, Huethig & Wepf, 1992, 28.50
Rheology processing and properties of polymeric materials (meeting), Acierno D,
Huethig & Wepf, 1993, 56
Rheology of polymers, Vinogradov G.V, Springer-Verlag, 1981, A$99.20
Rheometers for molten plastics: a practical guide to testing & property measurement,
Dealy J.M, Chapman & Hall, 1981, $82.95
(The) Theory of Polymer Dynamics, Doi M, Clarendon Press, 1989 (reprint), 25
Thermoplastic melt Rheology and processing, Shenoy A.V & Saini D.R, Marcel Dekker, 1996, $175
Viscoelastic properties of polymers (3rd Ed), Ferry J.D, John Wiley Inc, 1980, 120
20.6 Blood
Biorheology: proceedings of the second international congress, Copley A,
Butterworth-Heinemann, 1975, $47
Blood cells, Rheology and aging, Platt D, Springer-Verlag, 1987, $89.70
Blood Rheology, Lowe G.D, Bailliere Tindall, 1987, 27.50
Blood viscosity in heart disease & cancer, Dintenfass L, Franklin (Pergamon Press Reprint),
1981, $88
Blood viscosity:Hyperviscosity and hyperviscosaemia, Dintenfass L, Kluwer, 1985, 148.75
Clinical blood Rheology, Lowe G.D, CRC Pr, 1988, $137
Flow properties of blood & other biological systems (meeting), Copley A, Franklin
(Pergamon Press Reprint), $205
Hemorheology (meeting), Copley A, Franklin (Pergamon Press Reprint), 1968, $378
Hemorheology & thrombosis (meeting), Copley A, Franklin (Pergamon Press Reprint),
1976, $198
(On the) Rheology of blood & synovial fluids, Chmiel H, Bks Demand, (reprint soon), $50.80
Rheology, hemolysis, gas & surface interactions, Ghista D.N, S Karger, 1980, $78.50
Rheology of the circulation, Whitmore R. L, Franklin (Pergamon Press Reprint)
1968, $99
White blood cells: morphology and Rheology as related to function, Bagge U,
Kluwer, 1982, 2.95
White cell Rheology & inflammation, Messmer K, S Karger, 1985, $57
20.7 Food
Dairy Rheology: a concise guide, Prentice J.H, VCH, 1992, 66
Dough Rheology and baked product texture, Faridi H.A, Aspen, 1989, 75
179
Feeding and texture of food (meeting), Vincent J.F.V, Cambridge UP, 1991, 60
Food texture : measurement and perception , Rosenthal A.J, Aspen Publ., 1999, $142
Food texture: Instrumental and sensory measurement (meeting), Moskowitz H.R,
Marcel Dekker, 1987, $138
Food texture and viscosity, Bourne M.C, 1994, Academic Press, $85
Measurements in the Rheology of foodstuffs, Prentice J. H, Chapman & Hall, 1984, 59
Rheological methods in food process engineering, Steffe J. F, Freeman Press, 1992, $65
Rheology of Fluids and Semisolid Foods : Principles and Applications, Rao M.A, Aspen Pub., 1999,
$125.00
Rheology of foods, Borwankar R, Elsevier Adv. Technol., 1993, 75
Rheology of industrial polysaccharides: theory and applications, Lapasin R & Pricl S,
Rheology of Foods, Borwankar R & Shoemaker C.F, Elsevier Science, 1992, $135.00
Rheology of food, pharmaceutical and biological materials with general Rheology (meeting), Carter R.E,
Rheology of wheat products, Faridi H, Amer. Assn. Cereal, 1986, $41
Texture measurements of food, Kramer A & Szczesniak A.S, Kluwer, 1973, 49.95
Viscoelastic properties of foods, Rao M.A & Steffe J.F, Chapman & Hall, 1992, 85
20.8 Other specific systems
Asphalt Rheology: relationship to mixture, Briscoe O, ASTM,1987, $34
Clay-water interface & its rheological implications, Guven N, Clay Minerals, 1992, $15
Coatings Fundamentals: suspension rheology, Higgins B, Tappi Press, 1997, $30
Drilling mud & cement slurry Rheology manual, French Oil & Gas Ind. Assoc. Publications
Staff, St Mut, 1982, $250
Electro-rheological flows 1993, Siginer D.A, ASME, 1993, $37.50
Electrorheological fluids: proceedings of the 2nd international conference on ER fluids, Carlson J.D,
Technomic, 1990, $145
Engine oils - Rheology & tribology: 1995 international congress & exposition meeting, various, Soc.
Auto. Engineers, 1995, $49.00
Free liquid jets and films: hydrodynamics and rheology, Yarin A, Addison-Wesley Longman Higher
Education, 1993, 83
Free surface flows with viscosity, Tyvand P. A, WIT Press/Computational Mechan., 1997, 84
Glass: science and technology: viscosity and relaxation, Uhlmann D.R,
Academic Press, 1986, 144
High pressure Rheology of lubricants using the rotating optical micro-viscometer,
Wong P.L, ASME, 1996, ?
Paste flow and extrusion, Benbow J & Bridgwater J, Clarendon Press, 1993, 47
Paint flow and pigment: a rheological approach to coating and ink technology, Patton T,
John Wiley Inc, 1979, 180
Rheological and thermophysical properties of greases, Vinogradov G.V & Froishteter G.B,
Gordon & Breach, 1989, 92
Rheological properties of cosmetics and toiletries, Laba D, Marcel Dekker, 1993, 165
Rheological properties of lubricants, Briant Jean, Gulf Publishing, 1989, 61
Rheology of emulsions (meeting), Sherman P, Franklin, Pergamon Press Reprint, 1963
180
Rheology & tribology of engine oils, various, Soc. Auto. Engineers, 1992, $29
Rheology of fresh cement (meeting), Banfill P.F.G, E & FN Spon (C&H), 1991, 59
Rheology of liquid crystals, Khabibullaev P.K, Allerton Press, 1994, $95
Rheology of lubricants, Davenport T, Elsevier Applied Science, 1973, $39.75
Viscosity solutions and applications, Bardi M, Springer Verlag, 1997, 28
20.9 Drag reduction
Drag reduction by additives: review and bibliography, White A & Hemmings J.A.G,
Brit. Hydromechanics Research Assn., 1976, 17
Drag reduction in fluid flows, Sellin H.J & Moses R, Horwood Ellis, 1989, 39.95
Drag reduction in turbulent flows by additives, Gyr A & Bewersdorff H-W, Kluwer, 1995, 79
Emerging techniques in drag reduction, Kwing-So C, Prasad K.K & Truong T.V,
Mech. Engineering Publications, 1996, 98
Fluid mechanics and its applications: Proceedings of the 4th european drag reduction meeting,
Coustoils E, Kluwer, 1990, 70.95
Recent developments in turbulence management: Proceedings of the 5th drag reduction in
engineering flows meeting, Choi K.S, Kluwer, 1991, $139.50
Structure of turbulence and drag reduction: IUTAM Symposium, Gyr A,
Spinger Verlag, 1990, 78
20.10 Geological subjects
Deformation mechanisms, Rheology and tectonics, Knipe R.J,
Geol. Soc. Pub. House, 1990, 85
Earth Rheology, isostasy & eustasy: proceedings of earth Rheology (meeting),
Morner N-A, Bks Demand (reprint soon), $177
Experimental rheology of clayey soils, Meschyan S.R, A A Balkema, 1995, 66
Glacial isostasy, sea-level and mantle Rheology, Sabadini R, Kluwer, 1991, 142.25
Geomaterials: Constitutive Equations & Modelling, Darve F, Elsevier, 1990, $117
International conference in Rheology and soil mechanics, Keedwell M.J,
E. & F.N Spon (C&H), 1988, 75
Mudflow rheology and dynamics, Coussot, P, A A Balkema, 1997, 60
Proceedings of the international symposium on Rheology & soil mechanics (meeting), Symposium
Staff, Springer Verlag, 1966, $113.95
Rheological fundamentals of soil mechanics, Vyalov S.S, Elsevier Science, A$224.50
Rheological parameters of soils and design of foundations, Ter-Martirosyan Z,
A.A. Balkema, 1993, 50
Rheology and soil mechanics, Keedwell M.J, E. & F.N Spon (C&H), 1984, 79
Rheology of the Earth, Ranalli G, Chapman & Hall, 1995, 29.95
Rock and soil Rheology, Cristescu N, Springer-Verlag, 1988, $327
Rheology of solids & of the Earth, Karato S, OUP, 1989 $110
Rock Rheology: mechanics of elastic and inelastic solids, vol. 7,
Cristescu N, Kluwer, 1988, 106.75
181
20.11 Proceedings of meetings (not mentioned elsewhere)

3rd European Rheology conference (Edin.), Oliver D.R, Chapman & Hall, 1990, 79
Progress and Trends in Rheology V : Proceedings of the Fifth European
Rheology Conference, Eds. Emri I & R. Cvelbar R, Dr Verlag Steinkipff Dietrich,
1998, $113.00
IUTAM symposium on numerical simulation of non-isothermal viscoelastic liquids,
Dijksman J.F, Kluwer, 1995, 61.50
Theoretical and applied Rheology (meeting), Moldenaers P & Keunings R, Elsevier, 1992, $394.50
Viscoelastic materials: basic science and clinical applications: proceedings of the second international
symposium of the Northern Eye Institute, Rosen E.S, Pergamon Press, 1989, 110
Exercise : Log on to the Internet, go to the Amazon Books Website - http://www.amazon.com/
or amazon.co.uk - or any similar large book-seller site, and search for any new
rheology titles.
Liquids
182
Mechanics
Y
LOG
of
O
RHE
Fluid
FLOW
CHAPTER 21: RHEOLOGY AND COMPUTING

The world has arrived at an age of cheap complex devices of great
reliability, and something is bound to come of it, Vannevar Bush (1945)
21.1 Introduction
The great possibilities opening up following the latest developments in
computing mean that many things that were previously not feasible in rheological
calculations are now possible. This is especially true with respect to calculating the
steady and unsteady flow of rheologically complex fluids in complex geometries,
which go far beyond the examples of liquids and geometries given in chapter 10.
The particular technique we are referring to is Computational Fluid Dynamics, or
simply CFD. The most familiar form of CFD is modern weather forecasting, where
the local dynamics and thermodynamics of the atmosphere are calculated using a
mathematical description of the flow and physical properties of air, together with the
latest knowledge of wind speed, pressure and temperature at certain points, i.e.
weather stations and balloons. The output is in terms of the spatial distribution of air
velocity, temperature and pressure, and thus the distribution of rain, fog, frost and
snow. Some of the largest computers in the world are used for this purpose.
21.2 A brief description of how CFD works
The conservation of mass, momentum and energy govern all fluid flows.
These basic principles are expressed in the so-called Navier-Stokes equations. CFD
involves the use of a numerical code on a suitable computer to solve these
relationships as a series of partial differential equations.
The flow domain of interest has to be subdivided into a number of volumes,
or cells, to form the computational grid. The conditions at all the boundaries of the
grid must then be specifiedthe boundary conditions, e.g. in-flow, outflow, no-slip,
velocity or pressure. The next step is to specify the properties of the liquid flowing
through the domainthis is called the constitutive equation, which might be as simple
as a power-law description of viscosity/shear-rate or might be as complicated any of
the most advanced models that account for viscoelasticity and time effects such as
thixotropy, see below. The completion of these three steps makes it possible to solve
the flow of the liquid throughout the domain.
After the solution stage, the results may be visualised as colour contours or
vector plots of, for example, velocity, pressure, temperature, stress components, etc.
It is also possible to track virtual particles flowing through the flow domain to
produce streamlines. It is then possible to create videos of results of moving virtual
particles.
183
21.3
The rheological descriptions of a non-Newtonian liquid and CFD codes
Usually mathematical models (or constitutive equations) that are of the

following kind can describe the rheological properties of a liquid:
inelastic
elastic
- Newtonian, power-law, Cross, Herschel-Bulkley, Bingham

- differential or integral viscoelastic models, e.g. Oldroyd B.
Popular CFD codes presently available include:

CFX
from AEA Technology
FLUENT
FIDAP
POLYFLOW
from Fluent Ltd
Star CD
from Computational Dynamics
PHOENICS
from CHAM
All the codes listed provide a wide range of inelastic models, which
adequately cover such liquids as dispersions and emulsions, but the kinds of liquids
found in the polymer, personal-product, detergent, pharmaceutical and general
chemicals industries might need viscoelastic models, and these presently are only
provided by the POLYFLOW code.
21.4
When should CFD be used?
There is no better way to investigate the flow of complex fluids than to

perform careful experiments on well-characterised fluids in suitably instrumented
experimental geometries. CFD is not a cheap way of avoiding such proper, wellplanned experiments, since the necessary software, hardware and dedicated trained
personnel necessary for its use can be quite expensive. However, it is a very useful
adjunct to experimentation, and should be seen as another useful tool for those
interested in rheology.
Very often, pre-experiments using CFD will save a lot of time in planning
other work, and once one is certainvia practical validationthat a computer code
is giving a sufficiently good picture of reality, then systematic changes in both
geometry and liquid can be performed with some confidence. This optimisation
exercise is then very effective and can be economical in terms of resources that would
otherwise have to be used to build and test new geometries, as well as in mixing and
characterising new liquids.
21.5
Some typical results of CFD calculations
Figure 1 shows typical example of a flow geometry that has been meshed,
and figure 2 shows a typical output of such an exercise.
184
Chapter 21: Rheology and computing
Figure 1: A typical CFD mesh, showing refinement

where needed.
21.6
Figure 2: A schematic CFD output, showing flow-lines.

Pressure, stress or velocity contours could also
be shown.
A word of caution
CFD calculations are only as good as the mathematical representation of the

liquid and the particular calculation procedure being used. The calculation
operations used nowadays are however very robust, and given that the mesh used is
fine enough in the right places (so it can be established that the results are
independent of grid size), then we can be confident that the calculations will
converge to a reasonable result. However the greatest problem remaining is the
correct choice of a constitutive equation to adequately describe the liquid in a
mathematical form. The hierarchy of rheological behaviour we are trying to capture
begins at the general viscosity level, and moves up to the precise form of shear
thinning behaviour, through possible elastic effects, and then to extensional flow
anomalies. Most difficult of all is the inclusion of thixotropic effects at all these levels
of sophistication.
Exercise: Look up the CFD website at http://www.cfd-online.com/Resources/ for a
comprehensive list of codes available around the world.
185
186
APPENDIX: ON-LINE (PROCESS) VISCOMETRY

The mere act of observing something changes the nature of the thing observed, Hiesenberg
A.1 Introduction
The advantagesfor both batch and continuous process operationsof being able to
measure the viscosity of liquid products (or product precursors such as slurries) on-line, during
their manufacture are obvious:
time and effort are saved in obtaining the current or batch-end viscosity, which is otherwise
obtained by taking a sample to an off-line viscometer, which has to be carefully loaded and
subsequently cleaned.
a control function can be introducedif means are available to alter the viscosity by process
or chemical variables; and
a continuous monitor of viscosity build-upand hence structureis available for batchmixing operations.
While the term 'process viscometer' is generic, the terms in-line and on-line need to be
defined: here 'in-line' is used to describe the situation where all the liquid whose viscosity is
being measured passes through or around the process viscometer, while 'on-line' means that
only part of the liquid is taken and passed through the viscometerfor instance in a bypass loop.
We shall usually use the term on-line for convenience as it is commonly used for this class of
instrument, but it should be noted that the correct place for a viscometer must be carefully
chosen, bearing in mind the effect of flow-rate and temperature, so that sometimes it is best to
have the viscometer in-line and sometimes on-line.
Here we review briefly the principles on which good practice is based in the area of
process viscometry. We will not go over the same ground as Dealy [1] or Cheng et al [2], the
former giving some principles, but few viscometer types, while the latter gives lots of
viscometers with little 'consumer-preference rating', and no principles given. Here we shall give
an outline of the principles; an overview of the kinds of instruments commercially available and
then give some of our own experience with one particular viscometer, the Brookfield TT100.
A.2 Desirable features of a good on-line viscometer
A good on-line viscometer should conform to the three R's of all process
instrumentation:
Representative samples of the liquid must continually pass through the responsive
viscometer at a reasonable speed, but preferably without noticeably affecting the viscometer
reading in doing so;
187
Relatability must exist between the readings obtained on-line and those obtained from the
definitive off-line instrument for all liquids of interest;
Reliability and robustness of operation is needed for the rigors of a typical factory
environment.
Equivalent Newtonian viscosity, Pa.s
In order to have a representative sample, liquid must either be taken from a fast-flowing
part of the process stream or the instrument be placed directly in the stream. The electromechanical response of any viscometer is usually quite fast, but the overall response to a change
of viscosity of the liquid stream depends on the rate at which the liquid is substituted by fresh
material in the measuring volume of the viscometer. Of course this takes us into a compromise
situation because a good flow-through might mean that the flow rate affects the viscosity
reading, and then any fluctuations in flow rate are incorrectly interpreted as changes in
viscosity, see figure 1. For instance, if we want to place a measuring element directly in the flow
stream (say a vibrating element) in order to get an almost instant indication of viscosity change,
then it is unlikely to be the kind of measurement that is insensitive to flow rate and it is also
unlikely that the viscosity reading easily relates to a conventional off-line measurement.
10
10
10
-1
10
-2
10
-5
-3
-1
10
10
Velocity of liquid steam, V / m/s
10
Figure 1: The equivalent Newtonian viscosity indicated by the

Nametre vibrating-sphere in-line viscometer as a function of
stream velocity.
On the other hand, to make a proper viscosity/shear-rate measurement on-line requires

that the liquid be introduced into, and then removed from, a carefully controlled measuring
volume which is part of a well-designed viscometer. As we have said, this can take time to
carry out, and usually leads to a delay in responding to current viscosity changes in the flow
line. What kind of viscometer one chooses depends on ones need for instant information or
otherwise. However, data handling systems such as Connoisseur (Predictive Control Ltd.,
Northwich, UK) exist that can make up for slow instrument response because the form of the
response curve is known and reproducible. This means for instance that if we know the
response after say 30 seconds, we can predict where the viscosity will end up. Figure 2 shows a
typical response of this kind.
Relatable readings demand that we know quite a lot about the fluid and the flow or else
we do the same measurement on-line as off-line to be able to relate the readings, see below.
Reliable and robust machines are designed to withstand the kind of mechanical and
chemical environments found in factory situations. They must be mounted in a relatively
vibration-free position, and should not be attacked chemically by the liquids flowing through
them.
188
Percentage change in
viscosity reading
Appendix: On-line (process) viscometry
100
80
60
40
20
0
Switch on
50
100 150 200 250
0
Time since viscosity change / s
Figure 2: The (normalised) response of a Brookfield TT100 to

doubling the viscosity, showing 60% of the change in
~ 30 s.
All these factors mean that the study of on-line viscometry is a subject apart, with its
own rules and regulations. This is especially true since most liquids of interest are temperature
sensitive and non-Newtonian (and possibly viscoelastic) in nature.
A.3 Types of measurement made by current instruments
On-line viscometers use basically the same kinds of measurements made on off-line
instruments, i.e.,
the drag on rotating geometries or stationary objects in the flow
e.g. Contraves Convimat 101/105, APPA Systems Iso-torq, Brookfield TT100;
the speed of falling/pulled objects
e.g. Cambridge piston viscosity sensor;
the pressure-drop or flow-rate down conduits or through orifices
e.g. Porpoises P5, home-made devices;
the energy absorbed by vibrating elements, spheres, rods and blades
e.g. Nametres Viscoliner, AccuTraxs Electronic vibrating blade, Yamaichis oscillating
cylinder series, plus many others.
Because most liquids of industrial importance are non-Newtonian, we know that the
measured viscosity can vary because of the
type of flow (shear or extension viscosity);
the rate of oscillation (frequency-dependent viscosity or elasticity) and/or
the speed of flow (deformation-rate-dependant viscosity).
The result is that the one number usually produced as the output of a process viscometer,
purporting to be the measured 'on-line' viscosity, can vary enormously for a non-Newtonian
liquid depending on the type of measurement.
Also the measurement will be affected to some extent or other by the need to replenish
continuously the fluid in the vicinity of the measuring element with fresh fluid. If this is done
continuallyas it is in most casesthen there is a strong possibility that the flow could affect
the viscosity reading by, for instance, increasing the overall shear rate or even for vibrating
189
elements give an oscillatory-plus-steady flow which is very difficult to interpret. This can be
controlled, accounted for or replaced by a sampling method that ensures no-flow when
measurements are made.
One particular problem in the use of vibrational instruments is that if the liquid being
measured is viscoelastic to any degree, then care should be taken in analysing the results,
because any correlation sought with a conventional viscometer might not be found. The signal,
if not split into in-phase and out-of-phase, will include (often non-linear) viscous and elastic
components. Because the frequency of these devices is often very high, i.e. ultrasonic, the
degree of elasticity can be very high, and hence its contribution to the overall signal is large. In
this case it is difficult to relate to an off-line viscosity measurement on a conventional
viscometer.
One point that should always be borne in mind is the effect of temperature on viscosity,
because the temperature of the liquid being measured on-line is often above that used off-line.
Therefore some compensation should be inserted. This is often built into the viscometer but can
also be accommodated in any computerised data logging system being used, so that an
equivalent standard-temperature viscosity can be displayed. For some measurements however,
it might be necessary to pass the material through a heat exchanger to bring the material to the
desired temperature for measurement.
A.4 The effect of non-Newtonian viscosity
Many on-line instruments are calibrated with Newtonian liquids, so that the results are
presented as equivalent Newtonian viscositythis can be far from the truth! Very few
viscometers give the actual unambiguous (non-Newtonian) viscosity as a function of the
variable, be it shear rate, frequency, etc. As Dealy has rightly said "Most commercial process
rheometers are designed for use with Newtonian fluids, and when non-Newtonian materials are involved,
special problems arise." [1].
In cases where we only want a signal that is proportional to some measure of 'thickness',
then almost any viscosity reading will suffice. However if anyone has gone to the lengths of
putting in an on-line viscometer, this usually means that we have for a long period been
accumulating information from off-line measurementsand this is usually the case as few
companies install on-line viscometers before off-linethen they will want correspondence
between the on-line and off-line reading. This can of course, for the right kind of flow-curve, be
accomplished by a simpler method on-line than off-line, when a one-to-one correlation can be
established, but this is by no means always the case.
If the particular product is not very non-Newtonian, a single-point measurement might
be sufficient. Even then however we must be sure that the flow type is the same. For instance if
our experience off-line is built up from steady-state, simple-shear flow curves, then a too-fast
measurement might mean the measurement is not made under steady-state conditions. Equally
if the on-line flow has an appreciable extensional component, then problems can arise for some
liquids, especially polymer solutions. Also if a vibrational mode is used, then it is probable that
some non-linear oscillatory function is being measured. All these facts could mean that we end
up with no simple one-to-one correlation between on-line and off-line. Hence the safest way is
to duplicate on-line what is done off-line. This is possible nowadays for most situations.
Last, we should mention one recurring problem with any kind of viscometrywall slip.
To be more correct, we should speak about the depletion of the dispersed phase of suspensions
and emulsions that takes place at a smooth wall. This leads to a lubricating layer at the wall that
can greatly interfere with the measurement. This can be overcome by roughening or profiling
the surface of the measuring element [3].
190
A.5 Hygiene and cleaning

Cleaning is an important issue for process viscometers. Anyone with experience of real
factory operations knows that process instrumentation is expected to work immediately and it
is never expected to need adjustment or cleaning, but is used day-in, day-out for months or
even years. The best designs of viscometer anticipate this and ensure that little build-up of
debris occurs, and if it does, it is easily removed. This will be achieved by either washing the
system through with a suitable solvent or by having a built-in cleaning-in-place capability (CIP).
The design of the viscometer housing is crucial in this context. Brookfield have introduced a
special housing for their TT100 viscometer which was supplied by APV of New York.
A.6 General comments on actual representative instruments
All on-line viscometers have some element of novelty in terms of the way they measure
viscosity or how they ensure that liquid is replenished in the measuring region. However, at
the same time, because they are in the flow-line, there is some degree of compromise. For
instance, for the Brabender Convimeter, the novelty lies in the nutation of the rotating member
which precesses inside the measuring cell without rotating. However, although this novelty
ensures that fresh liquid is drawn automatically into the measuring cell and measured liquid is
expelled, the actual flow inside the measuring cell is very complicated, and does not necessarily
reach steady state. The Contraves Convimats novelty lies in the magnetic drive through the
wall, ensuring complete isolation from the outside. However, the jewelled mounting which
holds the driven cylinder often wears out, and lacks robustness. Also, the flow geometry is not
well defined, so no true shear rate can be assigned for non-Newtonian liquids, and the flow past
the measuring element is too fast.
The Namtre Viscoliner (vibrating sphere) is a simple device, easily placed in the flow
line, and easy to clean. However, the measurement one obtains is dependent on the flow rate
past the vibrating sphere, and the number obtained is at best an oscillatory parameter, possibly
the loss modulus, G, which for the materials usually being measured, cannot necessarily be
correlated with off-line measurements (the same is true for the Bendix vibrating reed instrument
and the many others of this kind). However, it is rarely the case that the output signal is
resolved into its elastic and viscous components.
The Cambridge piston viscosity sensor (Cambridge Applied Systems, Cambridge, MA,
USA) is ingenious in that the piston is lifted to the top of a slot let into the side wall of a pipe,
and allowed to drop, and the cycle then being repeated. The speed of fall is then related to the
viscosity. The novelty here is that the action of the cylinder moving up and down which pumps
fresh liquid in and out of the tube containing the piston. However, the one number so obtained
can only yield the equivalent Newtonian viscosity.
Whether or not one can live with the compromise associated with any instrument
depends on the nature of the liquids being measured. Thixotropic, shear-thinning liquids are
very difficult to measure even in the best off-line viscometer under ideal laboratory
circumstances. To measure such liquids on-line is extremely difficult. While most instruments
are calibrated to give an equivalent Newtonian viscosity, this is not always very useful for difficult
materials. If on the other hand the liquids of interest are easy to measure, being virtually time
independent and only slightly non-Newtonian, than almost any method will suffice, since a
reliable correlation can be established between on-and off-line measurements.
A.7 Detailed analysis of a particular instrument
191
The Brookfield TT100 is a typical on-line instrument that fulfils the criteria set out above,
in that it follows the 3Rs, i.e., representative sampling, relatability, reliability and robustness of
operation. It is manufactured by the Brookfield Engineering Company of the USA, who are the
longest-standing manufacturers of commercial rotational viscometers in the world. It is a
concentric-cylinder instrument that can operate in- or on-line, and has an open-ended variant
that can be mounted into the wall of a vessel - the TT200.
The TT100 is mounted in a chamber through which the liquid of interest flowsthe
general arrangement is shown in schematic cross-section in figure 3. The speed of the motor
driving the outer cylinder can be varied or fixed and the torque produced on the inner cylinder
causes a small twist of the sealed torque-tube supporting it. This small twist is transmitted
through the wall without friction and is measured on a rotation transducer mounted outside the
chamber.
torque tube
seal
Figure 3: A schematic cross-section of the Brookfield TT100 in-line
viscometer.
Some of the flow finds its way down through the gap between the cylinders either by
natural means because the flow resistance is asymmetric or as directed using a flow-deflector
that drives most the flow downwards, so enhancing enhances the flow between the cylinders.
The limits of the viscosity that can be measured by this instrument are set by the onset of
instabilities when the viscosity is too low and the torque-tube being twisted too far when the
viscosity is too high. The shear rate can be set to any value between about 1 and 1000 s-1, being a
combination of the effect of the motor speed and the gap between the cylinders. The latter
value is quite small for most pourable liquids and therefore the shear rate is virtually constant
throughout the gap. The response time depends on the complete replacement of all the liquid
in the measuring volume.
The TT100 has a very effective CIP set-up. Inlet ports for hot water are positioned to
wash the cylinders and keep the all-important gap clean. A special APV hygienic housing is
now available that ensures easy cleaning by washing through with cleaning fluids.
A.8 Conclusions
On-line viscometers should be carefully chosenthe amount of care depending on the
complexity of the liquid. The viscosity of Newtonian liquids can be measured by any kind of
viscometer. If the non-Newtonian liquid of interest is simple, then we will be able to correlate
any in- or on- line viscosity with its carefully measured off-line equivalent. If however there is
no simple correlation, we need to do our best to measure on-line in the same way as we are
measuring off-line. For instance a number of suitable concentric-cylinder type instruments are
192
now available. Even then, care has to be taken to ensure there is no interference from the flow;
either by choice of viscometer or care in regulating the flow through. In the latter situation,
single speed or multiple speed models are available. Computer control of course allows us any
degree of complexity in measuring.
In terms of response time, we have a compromise situation: fast response times usually
come from devices such as vibrating elements (e.g. the Nametre) placed directly in the flow
line, however these might be difficult to relate to off-line measurement, and are certainly very
sensitive to the local flow rate. On the other hand, the best devices that virtually duplicate offline measurements are usually much slower in response because of the relatively long time
taken to replace the liquid being measured (e.g. the Brookfield TT100). Thus the actual choice
of device must be made very carefully.
References
[1] Dealy, J.M, Rheol Acta, 29, 519 (1990).
[2] Cheng D C-H; Hunt, J A and Madhvi, P, Status Report on Process Viscometers:
Current Applications and Future Needs, Warren Spring Laboratory, Rep. No. W85034, 1985,
ISBN 0 85624 363 9. (Now available from AEA Technology, Harwell, UK).
[3] Barnes, H.A, JNNFM, 56, 221 - 251 (1995).
193
INDEX
computational fluid dynamics, see CFD
cone penetrometry, 136
confectionery, 7
consomm, 172
constitutive equation, 183
contraction flow, 117
Contraves Convimat, 191
Couette, 19
cream, 172
creep tests, 84, 86, 88, 89, 100
Cross model, 56, 58
cubic phase detergent, 95
curing, 102
dashpots, 83
delayed elasticity, 85
depletion forces, 128
dextran, effect on viscosity of water, 120
die swell, see extrudate swell
dip coating, shear rate in, 7
disc rotating in power-law liquid, 68
disc-shaped particles, 122
double-layer thickness, 127
dough, 102
drag reduction, 180
drainage down a wall, 7, 31, 65
drilling fluids/muds, 73, 148
Einstein, Albert, 120
elastic moduli, 81
elastic response, 85
Ellis equation, 78, 79
elongational flow, see extensional flow
engines, shear rate in, 7
entry length in tube flow, 35
epoxy resin, curing of, 103
extensional flow, 5, 150, 152, 159, 160
extensional flow, biaxial, 159, 160
extensional flow, planar , 159, 160
extensional viscosity, 142
extrudate swell, 109, 115
fabric conditioner, 56
fabric washing liquid, 55, 58
face cream, 2
facial wash, 58
figure legends, 17
fine sand, particle size distribution, 123
flocculated ink, 72
flocculated systems, 43
flocculation, 128
flow along a slot, 31
flow at the entrance of tubes, 35
flow between squeezed circular discs, 32
flow curves, 55
flow in a slit, 66
flow in a mixer, 69flow
in a straight circular pipe, 63
flow in a straight tube, 27, 28, 39
abrasive cleaners, 2
absorbed polyelectrolytes, 127
alginate, 55
ammonium salts, effect on viscosity of
water, 120
Andrade law, 21, 22
antimisting additive, ICI, 155
apple sauce, 73
Applied Rheology, 175
Arrhenius law, 21
aspect ratio of fibres, 126
associative polymer, 94, 147, 148
atomisation, shear rate in, 7
axial slot, pressure in, 116
axis labels, 17
balloon inflation, 159
battery pastes, 136
beads, particle size distribution, 123
bearings, shear rate in, 7
Bentonite suspension, 72, 97, 102
Bingham equation, fluid or plastic,
59, 61, 64, 69, 75, 76
bleaches, 2
block copolymers, 127
blood and blood flow, 7, 60, 179
body wash, 56
Bonjela, 101
Bouncing Putty, 94
Brabender Convimeter, 191
branching, effect of, 157
break-up of liquid filaments, 162
Brownian motion, 21
brushing, shear rate during, 7
Buckingham-Reiner equation, 64
Burgers model, 83, 85, 86, 87, 91
butter, 74
capillary number, 38
Carbopol, 57, 71, 73, 81, 101, 145, 156
carrageenan, 57
Carreau model, 75
Casson fluid, 64, 75
cavern flow in mixers, 69, 172
cement pastes, 3
centipoise definition, 19
centi-Stokes, definition, 20
ceramic pastes, 136
CFD, 183
CFD mesh, 185
chewing, shear rate in, 7
CIP (cleaning-in-place), 192
circular hole, pressure at bottom of, 116
clays, 134
coarse sand, particle size distribution, 123
coating and coatings, 7, 134, 148
coil-stretch transition, 154, 155
195
Laponite suspension, 97, 101

latex suspension, 97, 98, 134
leathery region, 92
levelling due to surface tension, 7, 39
liquid abrasive cleaner, 55
liquid laundry additives, 148
liquid-like behaviour, 90
living polymers, 107, 146
locust bean gum, 57
logarithmic plotting, 11
log-normal size distribution, 123
loop tests, 132
loss modulus, 88
lotions, 7
lubricating grease, 97
lubrication, 7
maltose, 120
margarine, 74, 172
Mark-Houwink equation, 145
maximum packing fraction, 123, 126
Maxwell model, 83, 89, 90, 91, 103, 104
mayonnaise, 2, 72, 73
mechanical vibrational spectroscopy, 88
medicines, 7
mica filler in polymer melt, 158
micelles, spherical, 166
micro-gels, 145
mixing, 7, 134
modified cellulose, 57
molecular weight, effect of, 157
molten chocolate, 71
mustard, 55, 73
Navier-Stokes equations, 183
network dynamics, 155
Newton, Isaac, 19
Newtonian behaviour, limit of, 23
Newtonian liquid, 19, 58
Newtonian viscosity, equivalent, 190
non-linear response, 100
normal force, 53
normal-force pump, 109
normal-stress coefficients, 111
normal-stress difference, 81
normal-stress difference, second, 110
normal-stress difference, first, 109, 112
nuclear-reactor vessels, 84
nylon melt, 158
oil, 172
oliophilic interactions, 165
organic glasses, 94
orthodontic silicone gels, 103
oscillatory flow and testing, 84, 92, 100
Ostwald-de Waele, see power law
paint, 2, 7, 97
paper manufacture, 7
parallel-plate plastometer
flow in inclined trough, 115

flow out of a tank, 66
flow through a converging nozzle, 67
flow through an annulus, 32
flow through packed bed, 29
flows, strong, see extensional flow
flows, weak, 151
food processing, 172, 179
food rheology, 171
fractals, 130
gelling systems, 102
geological subjects, 181
geometry, cone-and-plate, 111
geometry, parallel-plate, 111
geometry, re-entrant cone, 111
glassy region, 92
glucose, 120
gravy, 172
greases, 107
guar gum, 57
guar-gum/borate dispersion, 102
hair conditioner, 56
hair gel, 101
hair shampoo, 2
hair spray, 2
Hamaker constant, 127
HDPE melt, 113, 115, 157, 158, 163
HDPE/LDPE melt, 158
Herschel-Bulkley model, 75
high-speed coating, shear rate in, 7
holes, pressure in, 115
honey, 172
Hookes law, 82
hoop stress, 110
hydrophobic interactions, 165
hydroxyethyl cellulose, 113
ice cream, 172
indenter into a viscous liquid, 34
inertial effects, 89
in cone-and-plate, 116
inflation of cylinders, 159
ink, 3
inverted (see tube-less) siphon, 163
iron oxide suspension, 71
ISO 9002, 21
isoelectric point, 137
isolated falling sphere, 30
jet emerging from tubes, 36
Journal of Non-Newtonian Fluid
Mechanics, 175
Journal of Rheology, 175
Kelvin-Voigt element, 83, 86, 87, 92, 104
Kozeny-Carman equation, 29
Kreiger-Dougherty equation, 123
lamellar phase drops, 166-168
lamellar phase sheets, 166, 167
196
Index
rodlike (prolate) particles, 122

rodlike-micelles, 107, 146, 166
rods or fibres, long, 142, 152
rotating disc in a sea of liquid, 34
rubbery region, 92
rubbing, shear rate in, 7
salad dressing, 7, 172
saliva, 72
satellite drops in atomisation, 162
sauce, 2
secondary minimum, 127, 128
sedimentation of small particles, 7, 138
sepiolite suspension, 103
shampoo, 56, 60, 146
sharp-edged orifice, 161
shear flow, 5
shear rate, definition and range of, 6
shear stress, 6
shear thickening, 133, 172
shear thinning, 56
shower gel, 56, 146
SI units, 7
silica gel, 103
silicone oil, fluid, 57, 89
Sisko model, 5658, 60, 61, 64
skin cream, 7
slip layer, 65
slots, pressure in, 115
sodium salts, effect on viscosity of water,
120
soft cheese, 172
solid-like behaviour, 90
soup, creamy, 172
spaghetti sauce, 73
spectroscopy, mechanical, 100
sphere in a non-Newtonian liquid, 65
sphere rolling down inclined tube, 31
spheres, cloud of, falling/rising, 30, 138
spray drying, shear rate in, 7
spraying, shear rate in, 7
spreadable materials, 74
springs, 83
starch, 134
start-up experiments, 84, 104
steric repulsion, 127
stirring, 7
Stokes flow, law, 30, 65, 88
Stokes flow with inertia, 36
stress relaxation test, 84, 103
stretched exponential model, 132
stretching flow, see extensional flow
strong flow, see extensional flow
structure-breakers for water, 120
SUA1 polymer solution, 148, 149
sucrose, effect on viscosity of water, 120
surface tension, 68
surface undulations, 68
(or squeeze-film), 33
particle deformability, 126
particle interactions, 122, 126
particles, binary mixtures, 124
particle-size distribution, 123
pastes, flow of, 136
pat, 172
Pclet number, 125, 126
penetrometers, 51
penicillin suspension, 71
personal-care products, 148
pipe flow, 7, 64
plastometer, parallel-plate, 67
plateau modulus, 99
plateau region, 93
Poiseuille, 19
polyacrylamide, 55, 113 115, 156
polyacrylate, 113
polybutene, 114
polydimethylsiloxane melt, 96
polyethylene melt, 95
polyethylene,low density (LDPE), melt,
115, 157, 158, 163
LLDPE melt, linear LDPE, 157
polyisobutylene in cetane, 114
polyisobutylene in decalin, 56
polyisobutylene tetradecane, 115
polymethylmethacrylate (PMMA) melt,
96, 115
polymer lattices, suspension of, 124
polymer melts, 15, 56
polypropylene (PP) melt, 77, 96, 99, 114,
115, 157
polysaccharide, 57
polystyrene (PS) melt, 96, 113, 114, 157
polystyrene in toluene, 56, 57, 114, 144
polyvinyl acetate, 113
power-law liquid, 56, 57, 58, 59, 61, 63,
66, 67
primary minimum, 127
printing ink, 2, 7, 55
psycho-physical cognition, 23
pumping, 134
radial distribution function, 99
radial flow between parallel discs, 66
rebuilding after shear, 102
relaxation test, 84, 103
resins, 134
Rheologica Acta, 175
Rheology Abstracts, 175
rheometer, air turbine, 47, 48
rheometer, Carrimed Mk1, 48
rheometer, controlled-stress, 47, 53, 74
rheometer, Deer, 48
rheometers, 52
rod pushed into liquid in cup, 32
rod climbing, 108
197
viscometer, inertia in a concentriccylinder, 45

viscometer, in-line, Nametre vibratingsphere, 188
viscometer, low-viscosity liquids, 45
viscometer, mechanical damage, 45
viscometer, narrow-gap concentriccylinder, 26
viscometer, on-line, 187
viscometer, parallel-plate, 41, 50
viscometer, pipe/tube, 50
viscometer, process, 187
viscometer, rolling-ball, 31
viscometer, rotational, 25
viscometer, Searle controlled-stress, 47
viscometer, sedimentation/separation, 44
viscometer, small-angle cone/plate, 26
viscometer, standard laboratory, 52
viscometer, U-tube (Ostwald), 19, 46
viscometer, wide-gap concentric-cylinder,
41, 49, 69
viscometry, on-line (process), 187
viscosity, -alumina, 79
viscosity, air, 20
viscosity, asphaltic concrete, 77
viscosity, bitumen, 20, 77
viscosity, butter, 77
viscosity, cement, 77
viscosity, cheddar cheese, 77
viscosity, continuous phase, 20
viscosity, corn syrup, 20
viscosity, definition, 19
viscosity, Earths mantle, 77
viscosity, effect of pressure, 22
viscosity, effect of temperature, 21
viscosity, everyday perception, 22
viscosity, flocculated suspension as, 129,
131
viscosity, gelatine gel, 77
viscosity, glass, 77
viscosity, glycerol, 20, 23
viscosity, hydrogen, 20
viscosity, ice cream, 77
viscosity, ice, 77, 79
viscosity, infinite-shear, 20
viscosity, international standard, 20
viscosity, intrinsic, 20
viscosity, kinematic, 20
viscosity, lithium fluoride, 79
viscosity, lubricating oil, 20
viscosity, Madeira cake, 77
viscosity, magnesium oxide, 78
viscosity, marble, 77
viscosity, mineral oils, 23
viscosity, mortar, 77
viscosity, Newtonian liquids, 20
surfactant phases, 165

s
solutions, 94
suspended particles, 46
stability, 138
suspension viscosity control, 137
swallowing, 7
Swarfega, 2
syrup, 172
tank drainage, 29
Tanner equation, 115
Taylor vortices, 36, 37, 45
terminal region, 92
terminally anchored polymer, 127
ternary mixture of particles, 124
texture testers, 51
thixotropic paints, 2
thixotropy, 56, 82, 130, 131
thread pulling, 159, 160
toilet bleaches, 7
tomato juice, paste, pure, ketchup,
2, 72, 73, 101, 172
toothpaste, 2, 72, 73, 97
transient network, 153
transition region, 92
transition-to-flow region, 92
transverse slot, pressure in, 116
Trouton ratio, 155, 156
tube emptying under pressure, 29
tube filling under suction, 29
tube flow, 64
tubeless siphon, 10, 163
turbulent flow, 35
uniaxial extension, definition diagram, 10
van-der-Waals forces, 127, 128
vane geometry, 43, 74
vane-and-basket geometry, 44, 75
velocity profile in circular pipes, 27
vesicles, 168
viscoelasticity, 81
viscometer, Brookfield, 34, 68
viscometer, calibration, 42
viscometer, Cambridge piston, 191
viscometer, chemical attack, 44
viscometer, concentric-cylinder, 38
viscometer, cone-and-plate, 45, 46
viscometer, controlled-stress, 48
viscometer, effect of gap, 46
viscometer, elongational, 52
viscometer, evaporation in, 43
viscometer, extensional, opposedorifice/jet, 161
viscometer, flow-cup, 46
viscometer, high-speed testing, 45
viscometer, in- line, Brookfield TT100,
187, 189, 191, 192
198
Index
viscosity, olivine, 78
viscosity, petrol, 20
viscosity, plastic, 7
viscosity, Plasticine, 79
viscosity, relative, 20
viscosity, resins, 77
viscosity, silicone oils, 23
viscosity, sodium chloride, 79
viscosity, solid soap, 77
viscosity, solvent, 20
viscosity, specific, 20
viscosity, spinel, 79
viscosity, steel, 77, 99
viscosity, tin, 77
viscosity, titanium, 79
viscosity, uniaxial extensional (or
elongational), 153, 159,
160
viscosity, water, 20, 172
viscosity, zero-shear, 20
viscosity, zinc carbide, 79
viscous region, 92
wall effects or slip, 42, 133
washing-up liquid, 2
water/ethyl-alcohol mixtures, 121
Weissenberg effect, 108, 109
Weissenberg Rheogoniometer, 47
worm-like micelles, see rod-like micelles
xanthan gum solution, 55, 60, 95, 158
yield stress, 67, 69, 71
yield stress, flow equations with, 75
yoghurt, 60, 172
199
ANSWERS TO SOME EXERCISES

Chap 2, Exercise 4: Answer, 1.0515 x 10-4 m3/s.
Chap 6, Exercise: Answer, approximately 10-3 litre (or 1 ml) per day.
Chapter 9, Exercise:
Shampoo
Polymer
Yoghurt
Blood
Xanthan gum
Cross model
Sisko model
Sisko model
Cross or Sisko model
Cross model
Hint: look for turning points in the flow curves.

Chapter 15, Exercise: The viscosity increases smoothly as electrolyte is dissolved
until the solubility limit is reached (for that particular temperature). Crystals are then
formed, and the solution becomes a suspension. The continuous phase viscosity is
given by the electrolyte level at saturation, and the suspended crystals contribute via
their phase volume and shape. Temperature changes can change crystal shape
(morphology).
Chapter 18 Exercise: The velocity of a cloud of 50 m diameter particles (phase
volume 0.3 and density difference with respect to the continuous phase 1000kg/m3)
is given by
V =
2ga 2
(1 )4.75
9
If we can live with 1 mm per year as an acceptable sedimentation rate, then using the
equation it is possible to show that the zero shear rate viscosity 0 needs to be greater
than around 105 Pa.s. This can be checked on a controlled-stress rheometer.
For a 1 mm thick layer of the liquid to drain down a wall at a superficial velocity greater
than 1mm/s, if we assume the density of the liquid is 1500 kg/m3, then the wall shear
stress is given by 15 Pa. The order of magnitude of the shear rate is then given by the
velocity divided by the thickness, i.e. ~ 1 s-1, thus the viscosity around unit shear rate
has to be less than 15 Pa.s. This can be checked using a laboratory viscometer.
200
ABOUT THE AUTHOR

Howard Barnes studied Physics as an undergraduate at the University of
Wales Aberystwyth, and then did a PhD there with Prof. Ken Walters FRS
in the Applied Mathematics department. Since then, he has spent nearly
30 years at Unilever Research Port Sunlight where he is now Science Area
Leader in Rheology and Fluid Mechanics. His particular interest is the
relationship between the formulation, processing, microstructure and
rheology of liquid products. Among his awards for his efforts in this area
are a D.Sc. and a visiting professorship from the University of Wales; prizes from the Royal
Society of Chemistry and the Institution of Chemical Engineers, and an OBE from Queen
Elizabeth II for services to science and technology. He is a fellow of the Royal Academy of
Engineering; a past president of the British Society of Rheology, and an associate member of
The University of Wales Institute of Non-Newtonian Fluid Mechanics. He is the author of
some 60 publications in the area, and together with Professor Ken Walters and Mr John
Hutton, he is a co-author of An Introduction to Rheology.
Website: www.rheology.co.uk
ABOUT THE INSTITUTE OF NON-NEWTONIAN FLUID MECHANICS
The University of Wales Institute of Non-Newtonian Fluid Mechanics
(UWINNFM) came into being in 1991, bringing together expertise from
three constituent colleges of the University of Wales. Its Director is
Professor Ken Walters FRS and there are eight other permanent members
of staff, in addition to several Associate members, one of whom is
Professor H.A. Barnes.
Much of the research carried out within the Institute is driven by the needs
of Industry, although significant funding has also been raised from UK Government funding
agencies and the EC.
The UWINNFM runs a popular series of annual workshops.
Website: www.aber.ac.uk/~nniwww/
ISBN 0-9538032-0-1

A Handbook of Elmentary Rheology PDF

Uploaded by

Copyright:

Available Formats

A Handbook of Elmentary Rheology PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Handbook of Elmentary Rheology PDF

Uploaded by

Copyright:

Available Formats

A HANDBOOK OF

Institute of Non-Newtonian Fluid Mechanics

Handbook of Elementary Rheology

Published by The University of Wales Institute of Non-Newtonian Fluid

Handbook of Elementary Rheology

A BRIEF GUIDE TO THE HANDBOOK

WHAT IS FLOW AND DEFORMATION?

3. HOW TO INTERPRETE GRAPHICAL DATA

4. DRAWING RHEOLOGICAL GRAPHS

Handbook of Elementary Rheology

6.5.4 A rod being pushed into a cup filled with liquid .

8. YOUR RHEOLOGY LABORATORY

10 . EQUATIONS FOR THE FLOW OF NON-NEWTONIAN LIQUIDS

10.2.3 Casson model ...

11. VERY SHEAR-THINNING OR YIELD - STRESS FLUIDS

12. THE FLOW OF SOLIDS

13. LINEAR VISCOELASTICITY AND TIME EFFECTS

14. NON-LINEAR VISCOELASTICITY

Handbook of Elementary Rheology

14.4 The measurement of viscoelastic forces .. 111

15 . THE FLOW OF SUSPENSIONS

16. POLYMER RHEOLOGY

17. EXTENSIONAL FLOW AND EXTENSIONAL VISCOSITY

18. RHEOLOGY OF SURFACTANT SYSTEMS

18.4 Necessary rheological properties . 168

20. YOUR RHEOLOGY LIBRARY

21. RHEOLOGY AND COMPUTING

APPENDIX: ON-LINE (PROCESS) VISCOMETRY

CHAPTER 1: THE VISCOSITY OF LIQUIDS

laundry liquids, surface cleaning liquids

A Handbook of Elementary Rheology

Chapter 1: The viscosity of liquids

A Handbook of Elementary Rheology

CHAPTER 2: WHAT IS FLOW AND DEFORMATION?

Adjacent particles moving

Adjacent particles moving away

Figure 1: Particle motion in shear and extensional flows.

A Handbook of Elementary Rheology

Figure 2: Hypothetical layers in shear flow.

Chapter 2: What is flow and deformation?

Levelling due to surface

Paints, printing inks

Draining off surfaces

Toilet bleaches, paints, coatings

Chewing and swallowing

Mixing and stirring

Pumping liquids, blood flow

Skin creams, lotions

Atomisation, spray drying

Table 1: Typical shear rate ranges of various physical operations.

A Handbook of Elementary Rheology

Table 2: SI base units used in Rheology

Work, energy, heat

Table 3: Some derived SI units with special names

Table 4: Usual SI decade prefixes

Chapter 2: What is flow and deformation?

A Handbook of Elementary Rheology

Figure 4: Definition diagram for uniaxial extensional flow.

& and e and then for Newtonian liquids,