Anda di halaman 1dari 236

A Study of Non-Newtonian Viscosity and Yield Stress of Blood

in a Scanning Capillary-Tube Rheometer

A Thesis
Submitted to the Faculty
of
Drexel University
by
Sangho Kim
in partial fulfillment of the
requirements for the degree
of
Doctor of Philosophy
December 2002

ii

Acknowledgments

I wish to express my sincere gratitude to Dr. Young I. Cho, for his guidance
and inspiration during my entire tenure in graduate school. His experience and idea
have proven to be invaluable. I also wish to thank Dr. David M. Wootton for serving
as my co-advisor, and for his valuable suggestions and guidance on Biofluid
Dynamics.
I wish to express my appreciation to the members of my dissertation
committee, including: Dr. Ken Choi and Dr. Alan Lau from the MEM Department,
and Dr. Peter Lelkes from the School of Biomedical Engineering.
I am deeply indebted to Dr. Kenneth Kensey, Mr. William Hogenauer, and
Dr. Larry Goldstein from Rheologics, Inc. for providing valuable comments on the
test methods and data reduction procedure.
A sincere appreciation is extended to several colleagues whose friendship I
have cherished during my graduate studies, including: Dr.Wontae Kim, Dr. Sunghyuk
Lee, Chagbeom Kim, Giyoung Tak, Dohyung Lim, and Jinyong Wee.
Last but not least, I wish to thank my parents for their unbounded support
throughout my life. Their reliable provision of emotional, spiritual, and financial
support has allowed me to accomplish tasks that would have otherwise been
impossible.

iii

Table of Contents

LIST OF TABLES.....................................................................................................viii
LIST OF FIGURES ................................................................................................... x
ABSTRACT...............................................................................................................xiv
CHAPTER 1 INTRODUCTION ..............................................................................

1.1 Clinical Significance of Blood Viscosity....................................................

1.2 Motivation of the Present Study .................................................................

1.3 Objectives of the Present Study ..................................................................

1.4 Outline of the Dissertation ..........................................................................

CHAPTER 2 CONSTITUTIVE MODELS ..............................................................

2.1 Newtonian Fluid..........................................................................................

2.2 Non-Newtonian Fluid ................................................................................. 10


2.2.1

2.2.1.1

Power-law Model...................................................................... 11

2.2.1.2

Cross Model .............................................................................. 12

2.2.2

2.3

General Non-Newtonian Fluid........................................................... 10

Viscoplastic Fluid .............................................................................. 13

2.2.2.1

Bingham Plastic Model............................................................. 13

2.2.2.2

Casson Model............................................................................ 14

2.2.2.3

Herschel-Bulkley Model........................................................... 15

Rheology of Blood...................................................................................... 19
2.3.1

Determination of Blood Viscosity ..................................................... 19

iv
2.3.1.1

Plasma Viscosity....................................................................... 20

2.3.1.2

Hematocrit................................................................................. 20

2.3.1.3

RBC Deformability................................................................... 21

2.3.1.4

RBC Aggregation - Major Factor of Shear-Thinning


Characteristics........................................................................... 21

2.3.1.5

Temperature .............................................................................. 22

2.3.2

Yield Stress and Thixopropy ............................................................. 23

2.3.2.1

Yield Stress ............................................................................... 23

2.3.2.2

Thixotropy - Time Dependence ................................................ 24

CHAPTER 3 CONVENTIONAL RHEOMETRY: STATE-OF-THE-ART ........... 30


3.1

Introduction................................................................................................. 30

3.2

Rotational Viscometer ................................................................................ 34


3.2.1

Rotational Coaxial-Cylinder (Couette Type)..................................... 34

3.2.2

Cone-and-Plate................................................................................... 35

3.3

Capillary-Tube Viscometer......................................................................... 38

3.4

Yield Stress Measurement .......................................................................... 41


3.4.1

3.4.1.1

Direct Data Extrapolation ......................................................... 42

3.4.1.2

Extrapolation Using Constitutive Models................................. 43

3.4.2
3.5

Indirect Method.................................................................................. 42

Direct Method .................................................................................... 44

Problems with Conventional Viscometers for Clinical Applications ......... 46


3.5.1

Problems with Rotational Viscometers.............................................. 46

3.5.2

Problems with Capillary-Tube Viscometers...................................... 48

v
CHAPTER 4 THEORY OF SCANNING CAPILLARY-TUBE RHEOMETER.... 49
4.1

4.2

Scanning Capillary-Tube Rheometer (SCTR) ............................................ 49


4.1.1

U-Shaped Tube Set ............................................................................ 50

4.1.2

Energy Balance .................................................................................. 51

Mathematical Procedure for Data Reduction.............................................. 60


4.2.1

Power-law Model............................................................................... 60

4.2.2

Casson Model..................................................................................... 66

4.2.3

Herschel-Bulkley (H-B) Model ......................................................... 72

CHAPTER 5 CONSIDERATIONS FOR EXPERIMENTAL STUDY................... 81


5.1 Unsteady Effect ........................................................................................... 82
5.2 End Effect.................................................................................................... 87
5.3 Wall Effect (Fahraeus-Lindqvist Effect)..................................................... 90
5.4 Other Effects................................................................................................ 95
5.4.1

Pressure Drop at Riser Tube .............................................................. 95

5.4.2

Effect of Density Variation................................................................ 96

5.4.3

Aggregation Rate of RBCs - Thixotropy ........................................... 97

5.5 Temperature Considerations for Viscosity Measurement


of Human Blood..........................................................................................101
5.6 Effect of Dye Concentration on Viscosity of Water ...................................104
5.6.1

Introduction........................................................................................104

5.6.2

Experimental Method.........................................................................106

5.6.3

Results and Discussion ......................................................................107

CHAPTER 6 EXPERIMENTAL STUDY WITH SCTR.........................................112


6.1

Experiments with SCTR (with Precision Glass Riser Tubes) ....................112

vi

6.2

6.1.1

Description of Instrument ..................................................................113

6.1.2

Testing Procedure ..............................................................................114

6.1.3

Data Reduction with Power-law Model.............................................116

6.1.4

Results and Discussion ......................................................................117

Experiments with SCTR (with Plastic Riser Tubes)...................................130


6.2.1

Description of Instrument ..................................................................131

6.2.2

Testing Procedure ..............................................................................132

6.2.3

Data Reduction with Casson Mocel...................................................133

6.2.3.1

Curve Fitting .............................................................................134

6.2.3.2

Results and Discussion .............................................................135

6.2.4
6.3

Data Reduction with Herschel-Bulkley (H-B) Model .......................139

Comparison of Non-Newtonian Constitutive Models ................................158


6.3.1

Comparison of Viscosity Results.......................................................159

6.3.2

Comparison of Yield Stress Results ..................................................162

6.3.3

Effects of Yield Stress on Flow Patterns ...........................................164

CHAPTER 7 CONCLUSIONS AND RECOMMENDATIONS .............................180


LIST OF REFERENCES...........................................................................................184
APPENDIX A: NOMENCLATURE.........................................................................194
APPENDIX B: FALLING OBJECT VISCOMETER
- LITERATURE REVIEW ..............................................................197
APPENDIX C: SPECIFICATION OF CCD AND LED ARRAY............................200
APPENDIX D: BIOCOATING OF CAPILLARY TUBE........................................202
APPENDIX E: MICROSOFT EXCEL SOLVER.....................................................204
APPENDIX F: NEWTONS METHOD OF ITERATION .......................................206

vii
APPENDIX G: REPEATABILITY STUDY WITH DISTILLED WATER ............208
APPENDIX H: EXPERIMENTAL DATA ...............................................................210
VITA ..........................................................................................................................221

viii

List of Tables

2-1. Viscosity of some familiar materials at room temperature...............................

2-2. Range of shear rates of some familiar materials and processes ........................

5-1. Comparison of Punsteady and Pc for distilled water ........................................ 84


5-2. Comparison of Punsteady and Pc for bovine blood .......................................... 86
5-3. Density estimation............................................................................................. 99
6-1. Comparison of initial guess and resulting value using power-law model.........124
6-2. Comparison of initial guess and resulting value using Casson model ..............144
6-3. Comparison of initial guess and resulting value
using Herschel-Bulkley model ..........................................................................155
6-4. Comparison of four unknowns determined with Herschel-Bulkley model
for three consecutive tests..................................................................................157
6-5. Various physiological studies with non-Newtonian constitutive models .........167
6-6. Measurements of water viscosity ......................................................................169
6-7. Measurements of bovine blood viscosity ..........................................................171
6-8. Measurements of human blood viscosity ..........................................................173
6-9. Comparison of model constants, h y and y ...................................................175
6-10. Comparison of ht = and hst + h y ..............................................................176
H-1. A typical experimental data set of human blood obtained by a scanning
capillary-tube rheometer with precision glass riser tubes.................................210
H-2. A typical experimental data set of distilled water obtained by a scanning
capillary-tube rheometer with plastic riser tubes..............................................213
H-3. A typical experimental data set of bovine blood obtained by a scanning
capillary-tube rheometer with plastic riser tubes..............................................215

ix
H-4. A typical experimental data set of human blood obtained by a scanning
capillary-tube rheometer with plastic riser tubes..............................................218

List of Figures

2-1. Flow curve of a Newtonian fluid.......................................................................

2-2. Flow curve of power-law fluids......................................................................... 16


2-3. Flow curve of a Casson model .......................................................................... 17
2-4. Flow curve of viscoplastic fluids....................................................................... 18
2-5. Comparison of Newtonian plasma viscosity and
shear-thinning whole blood viscosity ............................................................... 26
2-6. Variation of the relative viscosity of blood and suspension with rigid spheres
at a shear rate > 100 s-1 ..................................................................................... 27
2-7. Rouleaux formation of human red blood cells photographed on a microscope
slide showing single linear and branched aggregates and a network................ 28
2-8. Elevated blood viscosity at low shear rates indicates RBC aggregation........... 29
3-1. Rheometers ........................................................................................................ 33
3-2. Schematic diagram of a concentric cylinder viscometer ................................... 36
3-3. Schematic diagram of a con-and-plate viscometer............................................ 37
3-4. Schematic diagram of a capillary-tube viscometer............................................ 40
3-5. Determination of yield stress by extrapolation.................................................. 45
4-1. Schematic diagram of a U-shaped tube set........................................................ 56
4-2. Fluid-level variation in a U-shaped tube set during a test ................................. 57
4-3. Typical fluid-level variation measured by a SCTR ........................................... 58
4-4. Liquid-solid interface condition for each fluid column
of a U-shaped tube set........................................................................................ 59
4-5. Fluid element in a capillary tube at time t ........................................................ 79
4-6. Velocity profile of plug flow of blood in a capillary tube................................. 80

xi

5-1. Pressure drop estimation for distilled water ...................................................... 83


5-2. Pressure drop estimation for bovine blood ........................................................ 85
5-3. Flow-pattern changes due to end effects ........................................................... 89
5-4. Migration of cells toward to the center of lumen (wall effect).......................... 92
5-5. Fahraeus-Lindquist effect due to the reduction in hematocrit in a tube with a
small diameter and the tendency of erythrocytes to migrate toward
the center of the tube......................................................................................... 93
5-6. Viscosity measurements for bovine blood with three different capillary tubes
with ID of 0.797 mm (with length = 100 mm), 1.0 mm (with length = 130 mm),
and 1.2 mm (with length = 156 mm) ................................................................ 94
5-7. Viscosity results for human blood with two different capillary tubes with
length of 100 mm (with ID = 0.797 mm) and 125 mm (ID = 0.797 mm) ........100
5-8. Schematic diagram of a U-shaped tube set for temperature measurement........102
5-9. Temperature measurement at a capillary tube during a viscosity test...............103
5-10. Schematic diagram of a scanning capillary-tube
rheometer (SCTR) system................................................................................109
5-11. Variations of both power-law index and consistency index of dye-water
solution due to effects of dye concentrations...................................................110
5-12. Viscosity data for dye-water solution with 6 different dye concentrations
at 25..............................................................................................................111
6-1. Schematic diagram of a scanning capillary-tube rheometer
with precision glass riser tubes .........................................................................121
6-2. Curve-fitting procedure with power-law model for mineral oil ........................122
6-3. Curve-fitting procedure with power-law model for human blood ....................123
6-4. Height variation in each riser tube vs. time for mineral oil...............................125
6-5. Viscosity measurement for mineral oil at 25 with a scanning
capillary-tube rheometer (SCTR) .....................................................................126
6-6. Height variation in each riser tube vs. time for human blood at 37. .............127

xii

6-7. Viscosity measurement (log-log scale) for human blood at 37 with


rotating viscometer (RV) and scanning capillary-tube rheometer (SCTR) ......128
6-8. Viscosity measurement (log-log scale) of unadulterated human blood
at 37, measured with scanning capillary-tube rheometer (SCTR) and
cone-and-plate rotating viscometer (RV), for two different donors .................129
6-9. Picture of a SCTR with plastic riser tubes.........................................................141
6-10. Heating pad for a test with unadulterated human blood...................................142
6-11. Curve-fitting procedure with Casson model for distilled water .......................143
6-12. Curve-fitting procedure with Casson model for donor 1..................................145
6-13. Curve-fitting procedure with Casson model for donor 2..................................146
6-14. Height variation in each riser tube vs. time for distilled water at 25............147
6-15. Viscosity measurement for distilled water at 25 ..........................................148
6-16. Height variation in each riser tube vs. time for bovine blood
with 7.5% EDTA at 25.................................................................................149
6-17. Viscosity measurement for bovine blood with 7.5% EDTA at 25 using
both rotating viscometer (RV) and scanning capillary-tube
Rheometer (SCTR) ..........................................................................................150
6-18. Height variation in each riser tube vs. time for human blood at 37 .............151
6-19. Viscosity measurement for human blood (2 different donors) at 37 ............152
6-20. Shear-stress variation vs. shear rate for human blood
(from 2 different donors) at 37.....................................................................153
6-21. Curve-fitting procedure with Herschel-Bulkley model for bovine blood ........154
6-22. Viscosity measurements of bovine blood with 7.5% EDTA
at 25, analyzed with Herschel-Bulkley model.............................................156
6-23. Test with distilled water at 25.......................................................................168
6-24. Test with bovine blood at 25 ........................................................................170

xiii
6-25. Test with unadulterated human blood at 37..................................................172
6-26. Wall shear stress at a capillary tube vs. shear rate............................................174
6-27. Variations of a plug-flow region at a capillary tube as a function of time
for bovine blood with 7.5% EDTA at 25.....................................................177
6-28. Velocity profiles at a capillary tube for bovine blood
with 7.5% EDTA at 25.................................................................................178
6-29. (a) Viscosity, (b) wall shear rate, and (c) wall shear stress
Plotted as a function of mean velocity at a capillary tube using
three non-Newtonian models for bovine blood with 7.5% EDTA ..................179
B-1. Falling cylinder viscometers .............................................................................199
C-1. Cross sectional view of SV352A8-01 module..................................................201
G-1. Repeatability study #1 ......................................................................................208
G-2. Repeatability study #2 ......................................................................................209

xiv

Abstract
A Study of Non-Newtonian Viscosity and Yield Stress of Blood
in a Scanning Capillary-Tube Rheometer
Sangho Kim
Professors Young I. Cho and David M. Wootton

The study of hemorheology has been of great interest in the fields of


biomedical engineering and medical researches for many years. Although a number
of researchers have investigated correlations between whole blood viscosity and
arterial diseases, stroke, hypertension, diabetes, smoking, aging, and gender, the
medical community has been slow in realizing the significance of the whole blood
viscosity, which can be partly attributed to the lack of an uncomplicated and clinically
practical rheometer.
The objectives of the present study were to investigate the theoretical
principles of a scanning capillary-tube rheometer used for measuring both the
viscosity and yield stress of blood without any anticoagulant, to experimentally
validate the scanning capillary-tube rheometer using disposable tube sets designed for
daily clinical use in measuring whole blood viscosity, and to investigate the effect of
non-Newtonian constitutive models on the blood rheology and flow patterns in the
scanning capillary-tube rheometer.
The present study introduced detailed mathematical procedures for data
reduction in the scanning capillary-tube rheometer for both viscosity and yield-stress
measurements of whole blood. Power-law, Casson, and Herschel-Bulkley models
were examined as the constitutive models for blood in the study. Both Casson and
Herschel-Bulkley models gave blood viscosity results which were in good agreement

xv
with each other as well as with the results obtained by a conventional rotating
viscometer, whereas the power-law model seemed to produce inaccurate viscosities at
low shear rates.
The yield stress values obtained from the Casson and Herschel-Bulkley
models for unadulterated human blood were measured to be 13.8 and 17.5 mPa,
respectively. The two models showed some discrepancies in the yield-stress values.
In the study, the wall shear stress was found to be almost independent of the
constitutive model, whereas the size of the plug flow region in the capillary tube
varies substantially with the selected model, altering the values of the wall shear rate
at a given mean velocity. The model constants and the method of the shear stress
calculation given in the study can be useful in the diagnostics and treatment of
cardiovascular diseases.

1
CHAPTER 1. INTRODUCTION

1.1. Clinical Significance of Blood Viscosity

The study of hemorheology has been of great interest in the fields of


biomedical engineering and medical research for many years. Hemorheology plays
an important role in atherosclerosis [Craveri et al., 1987; Resch et al., 1991; Lee et al.,
1998; Kensey and Cho, 2001]. Hemorheological properties of blood include whole
blood viscosity, plasma viscosity, hematocrit, RBC deformability and aggregation,
and fibrinogen concentration in plasma. Although a number of parameters such as
pressure, lumen diameter, whole blood viscosity, compliance of vessels, peripheral
vascular resistance are well-known physiological parameters that affect the blood
flow, the whole blood viscosity is also an important key physiological parameter.
However, its significance has not been fully appreciated yet.
A number of researchers measured blood viscosities in patients with coronary
arterial disease such as ischemic heart disease and myocardial infarction [Jan et al.,
1975; Lowe et al., 1980; Most et al., 1986; Ernst et al., 1988; Rosenson, 1993]. They
found that the viscosity of whole blood might be associated with coronary arterial
diseases. In addition, a group of researchers reported that whole blood viscosity was
significantly higher in patients with peripheral arterial disease than that in healthy
controls [Ciuffetti et al., 1989; Lowe et al., 1993; Fowkes et al., 1994].

2
Other researchers investigated correlation between the hemorheological
parameters and stroke [Grotta et al., 1985; Coull et al., 1991; Fisher and Meiselman,
1991; Briley et al., 1994]. They reported that stroke patients showed two or more
elevated rheological parameters, which included whole blood viscosity, plasma
viscosity, red blood cell (RBC) and plate aggregation, RBC rigidity, and hematocrit.
It was also reported that both whole blood viscosity and plasma viscosity were
significantly higher in patients with essential hypertension than in healthy people
[Letcher et al., 1981, 1983; Persson et al., 1991; Sharp et al., 1996; Tsuda et al., 1997;
Toth et al., 1999].

In diabetics, whole blood viscosity, plasma viscosity, and

hematocrit were elevated, whereas RBC deformability was decreased [Hoare et al.,
1976; Dintenfass, 1977; Hill et al., 1982; Poon et al., 1982; Leiper et al., 1982].
Others conducted hemorheological studies to determine the relationships
between whole blood viscosity and smoking, age, and gender [Levenson et al., 1987;
Bowdler and Foster, 1987; Fowkes et al., 1994; Ernst, 1995; Ajmani and Rifkind,
1998; Kameneva et al., 1998; Yarnell et al., 2000]. They found that smoking and
aging might cause the elevated blood viscosity. In addition, it was reported that male
blood possessed higher blood viscosity, RBC aggregability, and RBC rigidity than
premenopausal female blood, which may be attributed to monthly blood-loss
[Kameneva et al., 1998].

3
1.2. Motivation of the Present Study

The medical community has been slow in realizing the significance of whole
blood viscosity, which can be attributed partly to the lack of an uncomplicated and
clinically practical method of measuring whole blood viscosity. In most clinical
studies, mainly two types of viscometer have been available for general use:
rotational viscometers and capillary tube viscometer, as will be discussed in Chapter
3. These viscometers are used at laboratory only, and are not used in a clinical
environment. Until recently, the most immediate difficulty has been the lack of an
instrument that is specially designed for daily clinical use in measuring whole blood
viscosity.

1.3. Objectives of the Present Study

The objectives of the present study were 1) to investigate the theoretical


principles of a scanning capillary-tube rheometer (SCTR), which is capable of
measuring the viscosity and yield stress of blood without adding any anticoagulant, 2)
to validate the SCTR using disposable tube sets for clinical applications, and 3) to
investigate the effect of non-Newtonian constitutive models on the blood rheology
and flow patterns in the SCTR.
The present study introduced detailed mathematical procedures for data
reduction in the SCTR for both viscosity and yield-stress measurements of blood. In

4
experimental studies, distilled water (Newtonian fluid), bovine blood (non-Newtonian
fluid) with 7.5% EDTA, and unadulterated human blood (non-Newtonian fluid) were
used for the measurements of both viscosity and yield stress. Power-law, Casson, and
Herschel-Bulkley models were examined as constitutive models for blood in the study.

1.4. Outline of the Dissertation

Chapter 2 reviews the constitutive models applicable for non-Newtonian


characteristics including shear-thinning and yield stress.

Chapter 3 reviews the

conventional rheometers that measure either the viscosity or yield stress of a fluid. In
this chapter, only rheometers that can be applicable to clinical applications are
discussed. Chapter 4 introduces the theory of a scanning capillary-tube rheometer.
Chapter 5 discusses the considerations for the experimental study, which include
unsteady effect, end effect, wall effect, temperature analysis, dye concentration effect,
and other possible factors. Chapter 6 presents the results of experimental studies
performed with a scanning capillary-tube rheometer. Chapter 6 also reports the effect
of non-Newtonian constitutive models on the rheological measurements and flow
patterns of blood in a capillary tube. Chapter 7 gives conclusions of the study and
recommendations for future study.

5
CHAPTER 2. CONSTITUTIVE MODELS

This chapter reviews literature on non-Newtonian constitutive models, which


are applicable to the study of blood rheology. Viscous liquids including whole blood
can be divided in terms of rheological properties into Newtonian, general nonNewtonian, and viscoplastic fluids.

The characteristics of blood, which include

shear-thinning, yield stress, and thixotropy, are discussed in this chapter.

2.1. Newtonian Fluid

Fluid such as water, air, ethanol, and benzene are Newtonian. This means that
when shear stress is plotted against shear rate at a given temperature, the plot shows a
straight line with a constant slope that is independent of shear rate (see Fig. 2-1).
This slope is called the viscosity of the fluid. All gases are Newtonian, and common
liquids such as water and glycerin are also Newtonian. Also, low molecular weight
liquids and solutions of low molecular weight substances in liquids are usually
Newtonian. Some examples are aqueous solutions of sugar or salt.
The simplest constitutive equation is Newtons law of viscosity [Middleman,
1968; Bird et al., 1987; Munson et al., 1998]:

= &
where is the Newtonian viscosity and & is the shear rate or the rate of strain.

(2-1)

6
The Newtonian fluid is the basis for classical fluid mechanics. Gases and
liquids like water and mineral oils exhibit characteristics of Newtonian viscosity.
However, many important fluids, such as blood, polymers, paint, and foods, show
non-Newtonian viscosity.
Table 2-1 shows the wide viscosity range for common materials. Different
instruments are required to measure the viscosity over this wide range.

One

centipoise, 1 cP (= 10-3 Pas or 1 mPas), is approximately the viscosity of water at


room temperature. Shear rates corresponding to many industrial processes can also
vary over a wide range, as indicated in Table 2-2.

(a)

Shear stress

100

50

0
0

50

100

150

Shear rate

(b)

Viscosity

10

0
0

50
Shear rate

Fig. 2-1. Flow curves of a Newtonian fluid.


(a) Shear stress vs. Shear rate.
(b) Viscosity vs. Shear rate.

100

Table 2-1. Viscosity of some familiar materials at room temperature


[Barnes et al., 1989].
Liquid

Approximate Viscosity (Pas)

Glass

1040

Asphalt

108

Molten polymers

103

Heavy syrup

102

Honey

101

Glycerin

100

Olive oil

10-1

Light oil

10-2

Water

10-3

Air

10-5

Table 2-2. Range of shear rates of some familiar materials and processes
[Barnes et al., 1989].
Process

Range of
Shear Rates (s-1)

Sedimentation of fine powders


in a suspending liquid

10-6 10-4

Medicines, paints

Leveling due to surface tension

10-2 10-1

Paints, printing inks

Draining under gravity

10-1 101

Painting, coating

Screw extruders

100 102

Polymer melts, dough

Chewing and swallowing

101 102

Foods

Dip coating

101 102

Paints, confectionery

Mixing and stirring

101 103

Manufacturing liquids

Pipe flow

100 103

Pumping, blood flow

Spraying and brushing

103 104

Fuel atomization, painting

Rubbing

104 105

Application of creams and


lotions to the skin

Injection mold gate

104 105

Polymer melts

Milling pigments in fluid bases

103 105

Paints, printing inks

Blade coating

105 106

Paper

Lubrication

103 107

Gasoline engines

Application

10
2.2. Non-Newtonian Fluid

Any fluids that do not obey the Newtonian relationship between shear stress
and shear rate are non-Newtonian. The subject of rheology is devoted to the study of
the behavior of such fluids. Aqueous solutions of high molecular weight polymers or
polymer melts, and suspensions of fine particles are usually non-Newtonian.

2.2.1. General Non-Newtonian Fluid

In the case of general non-Newtonian fluids, the slope of shear stress versus
shear rate curve is not constant.

When the viscosity of a fluid decreases with

increasing shear rate, the fluid is called shear-thinning. In the opposite case where the
viscosity increases as the fluid is subjected to a high shear rate, the fluid is called
shear-thickening.

The shear-thinning behavior is more common than the shear-

thickening.
In general, the Newtonian constitutive equation accurately describes the
rheological behavior of low molecular weight polymer solutions and even high
molecular weight polymer solutions at very slow rates of deformation. However,
viscosity can be a strong function of the shear rate for polymeric liquids, emulsions,
and concentrated suspensions.

11
2.2.1.1. Power-law Model

One of the most widely used forms of the general non-Newtonian constitutive
relation is a power-law model, which can be described as [Middleman, 1968; Bird et
al., 1987; Munson et al., 1998]:

= m& n

(2-2)

where m and n are power-law model constants. The constant, m , is a measure of


the consistency of the fluid: the higher the m is, the more viscous the fluid is. n is a
measure of the degree of non-Newtonian behavior: the greater the departure from the
unity, the more pronounced the non-Newtonian properties of the fluid are.
The viscosity for the power-law fluid can be expressed as [Middleman, 1968;
Bird et al., 1987; Munson et al., 1998]:

= m& n 1

(2-3)

where is non-Newtonian apparent viscosity. It is well known that the power-law


model does not have the capability to handle the yield stress. If n < 1, a shearthinning fluid is obtained, which is characterized by a progressively decreasing
apparent viscosity with increasing shear rate. If n > 1, we have a shear-thickening
fluid in which the apparent viscosity increases progressively with increasing shear
rate. When n = 1, a Newtonian fluid is obtained. These three types of power-law
models are illustrated in Fig. 2-2.
One of the obvious disadvantages of the power-law model is that it fails to
describe the viscosity of many non-Newtonian fluids in very low and very high shear
rate regions. Since n is usually less than one, goes to infinity at a very low shear

12
rate (see Fig. 2-2) rather than to a constant, 0 , as is often observed experimentally.
Viscosity for many suspensions and dilute polymer solutions becomes constant at a
very high shear rate, a phenomenon that cannot be described by the power-law model.

2.2.1.2. Cross Model

As discussed in the previous section, the power-law model does not have the
capability of handling Newtonian regions of shear-thinning fluids at very low and
high shear rates. In order to overcome this drawback of the power-law model, Cross
(1965) proposed a model that can be described as [Ferguson and Kemblowski, 1991;
Cho and Kensey, 1991; Macosko, 1994]:

= & + 0
1 + m& n

(2-4)

where

0 and = viscosities at very low and high shear rates, respectively


m and n = model constants.
At an intermediate shear rate, the Cross model behaves like a power-law model as
shown in Fig. 2-3. However, unlike the power-law model, the Cross model produces
Newtonian viscosities ( 0 and ) at both very low and high shear rates.

13
2.2.2. Viscoplastic Fluid

The other important class of non-Newtonian fluids is a viscoplastic fluid.


This is a fluid which will not flow when a very small shear stress is applied. The
shear stress must exceed a critical value known as the yield stress for the fluid to flow.
For example, when opening a tube of toothpaste, we need to apply an adequate force
in order to make the toothpaste start to flow. Therefore, viscoplastic fluids behave
like solids when the applied shear stress is less than the yield stress. Once the applied
shear stress exceeds the yield stress, the viscoplastic fluid flows just like a normal
fluid.

Examples of viscoplastic fluids are blood, drilling mud, mayonnaise,

toothpaste, grease, some lubricants, and nuclear fuel slurries.

2.2.2.1. Bingham Plastic Model

Many types of food stuffs exhibit a yield stress and are said to show a plastic
or viscoplastic behavior. One of the simplest viscoplastic models is the Bingham
plastic model, and it can be expressed as follows [Bird et al., 1987; Ferguson and
Kemblowski, 1991; Macosko, 1994]:

= m B & + y

& = 0

when y ,

when y ,

where

y = a constant that is interpreted as yield stress

(2-5)
(2-6)

14
m B = a model constant that is interpreted as plastic viscosity.

Basically, the Bingham plastic model can describe the viscosity characteristics of a
fluid with yield stress whose viscosity is independent of shear rate as shown in Fig. 24. Therefore, the Bingham plastic model does not have the ability to handle the
shear-thinning characteristics of non-Newtonian fluids.

2.2.2.2. Casson Model

This model was originally introduced by Casson (1959) for the prediction of
the flow behavior of pigment-oil suspensions. The Casson model is based on a
structure model of the interactive behavior of solid and liquid phases of a two-phase
suspension [Casson, 1959]. The model describes the flow of viscoplastic fluids that
can be mathematically described as follows [Bird et al., 1987; Ferguson and
Kemblowski, 1991; Cho and Kensey, 1991; Macosko, 1994]:

= y + k &
& = 0

when y ,

when y ,

(2-7)
(2-8)

where k is a Casson model constant.


The Casson model shows both yield stress and shear-thinning non-Newtonian
viscosity. For materials such as blood and food products, it provides better fit than
the Bingham plastic model [Fung 1990; Cho and Kensey, 1991; Nguyen and Boger,
1992; Fung, 1993].

15
2.2.2.3. Herschel-Bulkley Model

The Herschel-Bulkley model extends the simple power-law model to include a


yield stress as follows [Herschel and Bulkley, 1926; Tanner, 1985; Ferguson and
Kemblowski, 1991; Holdsworth, 1993]:

= m& n + y

& = 0

when y ,

when y ,

(2-9)
(2-10)

where m and n are model constants.


Like the Casson model, it shows both yield stress and shear-thinning nonNewtonian viscosity, and is used to describe the rheological behavior of food
products and biological liquids [Ferguson and Kemblowski, 1991; Holdsworth, 1993].
In addition, the Herschel-Bulkley model also gives better fit for many biological
fluids and food products than both power-law and Bingham plastic models.

16

(a)

100
Shear stress

(b)
(c)
50

0
0

50

100

150

10
Viscosity

(c)

(b)
(a)
0
0

50

100
Shear rate

Fig. 2-2. Flow curves of power-law fluids.


(a) shear-thinning fluid ( n < 1).
(b) Newtonian fluid ( n = 1).
(c) shear-thickening fluid ( n > 1).

150

17

Newtonian regions

0
Viscosity
(log)

Power-law region

Shear rate (log)

Fig. 2-3. Flow curve of a Cross model.

18

100
(a)

Shear stress

(b)
50

mB

y
0
0

50

100
Shear rate

Fig. 2-4. Flow curves of viscoplastic fluids.


(a) Casson or Herschel-Bulkley fluid.
(b) Bingham plastic fluid.

150

19
2.3. Rheology of Blood

Blood behaves like a non-Newtonian fluid whose viscosity varies with shear
rate. The non-Newtonian characteristics of blood come from the presence of various
cells in the blood (typically making up 45% of the bloods volume), which make
blood a suspension of particles [Fung, 1993; Guyton and Hall, 1996]. When the
blood begins to move, these particles (or cells) interact with plasma and among
themselves.

Hemorheologic parameters of blood include whole blood viscosity,

plasma viscosity, red cell aggregation, and red cell deformability (or rigidity).

2.3.1. Determinants of Blood Viscosity

Much research has been performed to formulate a theory that accounts


completely for the viscous properties of blood, and some of the key determinants
have been identified [Dinnar, 1981; Chien et al., 1987; Guyton and Hall, 1996]. The
four main determinants of whole blood viscosity are (1) plasma viscosity, (2)
hematocrit, (3) RBC deformability and aggregation, and (4) temperature. The first
three factors are parameters of physiologic concern because they pertain to changes in
whole blood viscosity in the body.

Especially, the second and third factors,

hematocrit and RBC aggregations, mainly contribute to the non-Newtonian


characteristics of shear-thinning viscosity and yield stress.

20
2.3.1.1. Plasma Viscosity

Plasma is blood from which all cellular elements have been removed. It has
been well established that plasma behaves like a Newtonian fluid. Careful tests
conducted using both rotating and capillary tube viscometers over a range of shear
rates (i.e., from 0.1 to 1200 s-1) found no significant departures from linearity.
Therefore, its viscosity is independent of shear rate. Figure 2-5 illustrates this clearly
in the horizontal viscosity line for plasma [Dintenfass, 1971; Dinnar, 1981]. Since
blood is a suspension of cells in plasma, the plasma viscosity affects whole blood
viscosity, particularly at high shear rates.

2.3.1.2. Hematocrit

Hematocrit is the volume percentage of red blood cells in whole blood. Since
studies have shown normal plasma to be a Newtonian fluid [Fung, 1993], the nonNewtonian features of human blood undoubtedly come from suspended cells in blood.
The rheological properties of suspensions correlate highly with the concentrations of
suspended particles. In blood, the most important suspended particles are the red
blood cells (RBC). Hematocrit is the most important determinant of whole blood
viscosity [Benis et al., 1970; Thurston, 1978; Fung, 1993; Picart et al., 1998; Cinar et
al., 1999]. The effect of hematocrit on blood viscosity has been well documented.
All studies have shown that the viscosity of whole blood varies directly with

21
hematocrit at all cell concentrations above 10%. In general, the higher the hematocrit,
the greater the value of whole blood viscosity [Dintenfass, 1971; Dinnar, 1981; Chien
et al., 1987; Guyton and Hall, 1996].

2.3.1.3. RBC Deformability

Deformability is a term used to describe the structural response of a body or


cell to applied forces. The effect of RBC deformability in influencing general fluidity
of whole blood is clearly revealed in Fig. 2-6. This figure shows the relative viscosity
of blood at a shear rate >100 s-1 (at which particle aggregation is negligible, isolating
RBC deformability) compared with that of suspensions with rigid spheres. At 50%
concentration, the viscosity of a suspension of rigid spheres reaches almost infinity so
that the suspension is not able to flow. On the contrary, normal blood remains fluid
even at a hematocrit of 98%, on account of the deformability of its RBCs [Fung,
1993].

2.3.1.4. RBC Aggregation - Major Factor of Shear-Thinning Characteristic

Since red cells do not have a nucleus, they behave like a fluid drop [Dinnar,
1981]. Hence, when a number of red cells cluster together as in the flow of a low
shear rate, they aggregate together. Accordingly, human RBCs have the ability to

22
form aggregates known as rouleaux. Rouleaux formation is highly dependent on the
concentration of fibrinogen and globulin in plasma. Note that bovine blood does not
form rouleaux because of absence of fibrinogen and globulin in plasma [Fung, 1993].
Various degrees and numbers of rouleaux in linear array and branched network are
pictured in Fig. 2-7.
Figure 2-8 shows the relationship between blood viscosity and rouleaux
formation. Rouleaux formation of healthy red cells increases at decreasing shear
rates. As red cells form rouleaux, they will tumble while flowing in large vessels.
The tumbling disturbs the flow and requires the consumption of energy, thus
increasing blood viscosity at low shear [Fung, 1993]. As shear rate increases, blood
aggregates tend to be broken up, resulting in drop in blood viscosity (see Fig. 2-8). In
short, rouleaux formation increases blood viscosity, whereas breaking up rouleaux
decreases blood viscosity.

2.3.1.5. Temperature

Temperature has a dramatic effect on the viscosity of any liquid, including


whole blood and plasma. As in most fluids, blood viscosity increases as temperature
decreases [Fung, 1993; Guyton and Hall, 1996].

In blood, reduced RBC

deformability and increased plasma viscosity particularly elevate whole blood


viscosity at low temperatures [Barbee, 1973]. Consequently, precise control of the
sample temperature is necessary to measure viscosity accurately in vitro.

It is

23
preferable and is a standard in hemorheologic studies to carry out blood viscosity
measurements at body temperature of 37. Typically, blood viscosity increases less
than 2% for each decrease in temperature [Barbee, 1973].

2.3.2. Yield Stress and Thixotropy

2.3.2.1. Yield Stress

In addition to non-Newtonian viscosity, blood also exhibits a yield stress. The


source of the yield stress is the presence of cells in blood, particularly red cells.
When such a huge amount (40-45% by volume) of red cells of 8-10 microns in
diameter is suspended in plasma, cohesive forces among the cells are not negligible.
The forces existing between particles are van der Waals-London forces and
Coulombic forces [Cheng and Evans, 1965; Mewis and Spaull, 1976]. Hence, in
order to initiate a flow from rest, one needs to have a force which is large enough to
break up the particle-particle links among the cells.
However, blood contains 40-45% red cells and still moves relatively easily.
The healthy red cells behave like liquid drops because the membranes of red cells are
so elastic and flexible. Note that in a fluid with no suspended particles, the fluid
starts to move as soon as an infinitesimally small amount of force is applied. Such a
fluid is called a fluid without yield stress. Examples of fluid with no yield stress
include water, air, mineral oils, and vegetable oils. Examples of fluids having the

24
yield stress include blood, ketchup, salad dressings, grease, paint, and cosmetic
liquids.
The magnitude of the yield stress of human blood appears to be at the order of
0.05 dyne/cm2 (or 5 mPa) [Schmid- Sch&o&nbein and Wells, 1971; Walawender et al.,
1975; Nakamura and Sawada, 1988; Fung, 1993; Stoltz et al., 1999] and is almost
independent of temperature in the range of 10-37 [Barbee, 1973].

2.3.2.2. Thixotropy - Time Dependence

The phenomenon of thixotropy in a liquid results from the microstructure of


the liquid system. Thixotropy may be explained as a consequence of aggregation of
suspended particles. If the suspension is at rest, the particle aggregation can form,
whereas if the suspension is sheared, the weak physical bonds among particles are
ruptured, and the network among them breaks down into separate aggregates, which
can disintegrate further into smaller fragments [Barnes, 1997].
After some time at a given shear rate, a dynamic equilibrium is established
between aggregate destruction and growth, and at higher shear rates, the equilibrium
is shifted in the direction of greater dispersion. The relatively long time required for
the microstructure to stabilize following a rapid change in the rate of flow makes
blood thixotropy readily observable [How, 1996].
This effect on viscosity has been studied using a steady flow [Huang et al.,
1975]. At high shear rates, structural change occurs more rapidly than that at low

25
shear rates. In their study, the first step was from the no-flow condition to a shear
rate of 10 s-1.

They found that blood viscosity decreased over a period of

approximately 20 seconds at the shear rate of 10 s-1 before the final state was attained.
Next, when the shear rate stepped from 10 to 100 s-1, almost no time was required to
reach the microstructual equilibrium after the change of shear rate.
Gaspar-Rosas and Thurston (1988) also investigated on erythrocyte aggregate
rheology by varying shear rate from 500 s-1 to zero. Based on their results, it can be
concluded that the recovery of quiescent structure requires approximately 50 seconds
while the high shear rate structure is attained in a few seconds. In other words, in
order to minimize the effect of the thixotropic characteristic of blood on the viscosity
measurement between the shear rates of 500 and 1 s-1, at least 50 seconds should be
allowed during the test to have the fully aggregated quiescent state at a shear rate near
1 s-1.

Viscosity (cP)

26

Whole blood

Plasma
1
10

400

100
-1

Shear rate (s )

Fig. 2-5. Comparison of Newtonian plasma viscosity


and shear-thinning whole blood viscosity.

27

Relative viscosity

Suspension with rigid spheres


100

10
Normal blood
1
0.2

0.4

0.6

0.8

Particle volume fraction

Fig. 2-6. Variation of the relative viscosity of blood and suspension with rigid spheres
at a shear rate > 100 s-1 [Goldsmith, 1972].

28

Fig. 2-7. Rouleaux formation of human red blood cells photographed on a microscope
slide showing single linear and branched aggregates (left part) and a network (right
part). The number of cells in linear array are 2, 4, 9, 15 and 36 in a, b, c, d, and f,
respectively. [Fung, 1993; Goldsmith, 1972]

Relative viscosity

29

10
Normal blood
1
1

10
Shear rate (s-1)

400

Fig. 2-8. Elevated blood viscosity at low shear rates indicates RBC aggregation
(rouleaux formation). Blood viscosity decreases with increasing shear rates as RBC
aggregations breaks up to individual red cells.

30
CHAPTER 3. CONVENTIONAL RHEOMETRY: STATE-OF-THE-ART

This chapter reviews literature on conventional rheometries. Section 3.1


briefly introduces conventional rheometers. In sections 3.2 and 3.3, viscometers
commonly used for the viscosity measurements of fluids, which have been used for
hemorheology studies, are demonstrated. Section 3.4 provides conventional methods
of measuring yield stresses of fluids.

Section 3.5 presents the drawbacks of

conventional viscometers for clinical applications.

3.1. Introduction

Numerous types of rheometers have been used to measure the viscosity and
yield stress of materials [Tanner, 1985; Ferguson and Kemblowski, 1991; Macosko,
1994]. In the present study, rheometer refers to a device that can measure both
viscosity and yield stress of a material, whereas viscometer can measure only the
viscosity of the material. In addition, only shear viscometers will be discussed in the
study since the other type, extensional viscometers, are not very applicable to
relatively low viscous fluids, such as water and whole blood.
Typically, shear viscometers can be divided into two groups [Macosko, 1994]:
drag flows, in which shear is generated between a moving and a stationary solid
surface, and pressure-driven flows, in which shear is generated by a pressure
difference over a capillary tube. The commonly utilized members of these groups are

31
shown in Fig. 3-1. Numerous techniques have been developed for determining the
yield stress of fluids both directly and indirectly.
Most of these viscometers can produce viscosity measurements at a specified,
constant shear rate. Therefore, in order to measure the viscosity over a range of shear
rates, one needs to repeat the measurement by varying either the pressure in the
reservoir tank of capillary tube viscometers, the rotating speed of the cone or cup in
rotating viscometers, or the density of the falling objects. Such operations make
viscosity measurements difficult and labor intensive. In addition, these viscometers
require anticoagulants in blood to prevent blood clotting. Hence, the viscosity results
include the effects of anticoagulants, which may increase or decrease blood viscosity
depending on the type of anticoagulant [Rosenblum, 1968; Crouch et al., 1986;
Reinhart et al., 1990; Kamaneva et al., 1994].
Drag-flow type of viscometers includes a falling object (ball or cylinder)
viscometer and a rotational viscometer. However, the falling object viscometer is not
very convenient to use for clinical applications. In the case of the falling object
viscometer, the relatively large amount of a test fluid is required for the viscosity
measurement. In addition, since the testing fluid is at a stationary state initially, the
type of viscometer is not very applicable to a thixotropic fluid like whole blood. The
principle of the falling object viscometer is provided in Appendix B.
For the yield measurement of blood, most researchers have used indirect
methods rather than direct methods for practical reasons [Nguyen and Boger, 1983;
de Kee et al., 1986; Magnin and Piau, 1990]. Thus, the details of direct methods will

32
not be discussed in this chapter.

As indirect methods, data extrapolation and

extrapolation using constitutive models are introduced and discussed in this chapter.

33

Rheometers

Yield Stress
Measurements

Viscosity
Measurements

Drag
Flows

PressureDriven Flows

CapillaryTube
Viscometer

Falling/
Rolling
Object
Viscometer

Indirect
Methods

Data
Extrapolation

Rotational
Viscometer

Fig. 3-1. Rheometers.

Direct
Methods

Extrapolation
using
Constitutive
Models

34
3.2. Rotational Viscometer

In a rotational viscometer, the fluid sample is sheared as a result of the


rotation of a cylinder or cone. The shearing occurs in a narrow gap between two
surfaces, usually one rotating and the other stationary.

Two frequently used

geometries are Couette (Fig. 3-2) and cone-and-plate (Fig. 3-3).

3.2.1. Rotational Coaxial-Cylinder (Couette Type)

In a coaxial-cylinder system, the inner cylinder is often referred to as bob, and


the external one as cup. The shear rate is determined by geometrical dimensions and
the speed of rotation.

The shear stress is calculated from the torque and the

geometrical dimensions. By changing the speed of the rotating element, one is able to
collect different torques, which are used for the determination of the shear stressshear rate curve. Figure 3-2 shows a typical coaxial-cylinder system that has a fluid
confined within a narrow gap (

Ri
0.99 ) between the inner cylinder rotating at
Ro

and the stationary outer cylinder.


Once the torque exerting on either inner or outer cylinder is measured, the
shear stress and shear rate can be calculated as follow [Macosko, 1994]:

( Ri ) =

Mi
Mo
or ( Ro ) =
2
2Ri H
2Ro2 H

(3-1)

35

& ( Ri ) & ( Ro ) =

R
Ro Ri

when 1 >

Ri
0.99
Ro

(3-2)

where

Ri and Ro = radii of inner and outer cylinders, respectively


R=

Ri + Ro
2

M i and M o = torques exerting on inner and outer cylinders, respectively

H = height of inner cylinder


= angular velocity.

3.2.2. Cone-and-Plate

The common feature of a cone-and-plate viscometer is that the fluid is sheared


between a flat plate and a cone with a low angle; see Fig. 3-3. The cone-and-plate
system produces a flow in which the shear rate is very nearly uniform. Lets consider
a fluid, which is contained in the gap between a plate and a cone with an angle of .
Typically, the gap angle, , is very small ( 4 o ). The shear rate of the fluid depends
on the gap angle, , and the linear speed of the plate. Assuming that the cone is
stationary and the plate rotates with a constant angular velocity of , the shear stress
and shear rate can be calculated from experimentally measured torque, M , and given
geometric dimensions (see Fig. 3-3) as follows [Macosko, 1994]:

3M
and & =
.
3

2R

(3-3)

36

Ri

Ro

Fig. 3-2. Schematic diagram of a concentric cylinder viscometer.

37

Torque measurement device

Fluid
Cone

Plate

Fig. 3-3. Schematic diagram of a cone-and-plate viscometer.

38
3.3. Capillary-Tube Viscometer

The principle of a capillary tube viscometer is based on the Hagen-Poiseuille


Equation which is valid for Newtonian fluids. Basically, one needs to measure both
pressure drop and flow rate independently in order to measure the viscosity with the
capillary tube viscometer. Since the viscosity of a Newtonian fluid does not vary
with flow or shear, one needs to have one measurement at any flow velocity.
However, for non-Newtonian fluids, it is more complicated because the viscosity
varies with flow velocity (or shear rate).
In a capillary-tube viscometer, the fluid is forced through a cylindrical
capillary tube with a smooth inner surface. The flow parameters have to be chosen in
such a way that the flow may be regarded as steady-state, isothermal, and laminar.
Knowing the dimensions of the capillary tube (i.e., its inner diameter and length), one
can determine the functional dependence between the volumetric flow rate and the
pressure drop due to friction. If the measurements are carried out so that it is possible
to establish this dependence for various values of pressure drop or flow rate, then one
is able to determine the flow curve of the fluid.
For non-Newtonian fluids, since the viscosity varies with shear rate, one needs
to vary the pressure in the reservoir in order to change the shear rate, a procedure that
is highly time-consuming. After each run, the reservoir pressure should be reset to a
new value to obtain the relation between flow rate and pressure drop. In order to
determine the flow curve of a non-Newtonian fluid, one needs to establish the
functional dependence of shear stress on shear rate in a wide range of these variables.

39
Figure 3-4 shows the schematic diagram of a typical capillary-tube viscometer,
which has the capillary tube with an inner radius of Rc and a length of Lc . It is
assumed that the ratio of the capillary length to its inner radius is so large that one
may neglect the so-called end effects occurring in the entrance and exit regions of the
capillary tube. Then, the shear stress at the tube wall can be obtained as follows:

rPc
2 Lc

w =

Rc Pc
2 Lc

(3-4)

(3-5)

where

and w = shear stresses at distance r and at tube wall, respectively


r = distance from the capillary axis
Pc = pressure drop across a capillary tube.

It is of note that the shear stress distribution is valid for fluids of any rheological
properties.
In the case of a Newtonian fluid, the shear rate at tube wall can be expressed
by taking advantage of the well-known Hagen-Poiseuille Equation as:

& w =

4Q 4V
=
Rc3 Rc

where

& w = wall shear rate


Q=

Rc4 Pc
= Rc2 V = volumetric flow rate (Hagen-Poiseuille Equation)
8 Lc

V = mean velocity.

(3-6)

40

Compressed
air

Air
Test fluid

Reservoir
tank
Capillary tube
Lc

2 Rc

Collected
test fluid
Balance

Fig. 3-4. Schematic diagram of a capillary-tube viscometer.

41
3.4. Yield Stress Measurement

Whether yield stress is a true material property or not is still a controversial


issue [Barnes and Walters, 1985]. However, there is generally an acceptance of its
practical usefulness in engineering design and operation of processes where handling
and transport of industrial suspensions are involved. The minimum pump pressure
required to start a slurry pipeline, the leveling and holding ability of paint, and the
entrapment of air in thick pastes are typical problems where the knowledge of the
yield stress is essential.
Numerous techniques have been developed for determining the yield stress
both directly and indirectly based on the general definition of the yield stress as the
stress limit between flow and non-flow conditions. Indirect methods simply involve
the extrapolation of shear stress-shear rate data to zero shear rate with or without the
help of a rheological model.

Direct measurements generally rely on some

independent assessment of yield stress as the critical shear stress at which the fluid
yields or starts to flow.
The value obtained by the extrapolation of a flow curve is known as
extrapolated or apparent yield stress, whereas yield stress measured directly,
usually under a near static condition, is termed static or true yield value.

42
3.4.1. Indirect Method

Indirect determination of the yield stress simply involves the extrapolation of


experimental shear stress-shear rate data at zero shear rate (see Fig. 3-5).

The

extrapolation may be performed graphically or numerically, or can be fitted to a


suitable rheological model representing the fluid and the yield stress parameter in the
model is determined.

3.4.1.1. Direct Data Extrapolation

One of most common procedures is to extend the flow curve at low shear rates
to zero shear rate, and take the shear stress intercept as the yield stress value. The
technique is relatively straightforward only if the shear stress-shear rate data are
linear. With nonlinear flow curves, as shown in Fig. 3-5, the data may have to be
fitted to a polynomial equation followed by the extrapolation of the resulting curve fit
to zero shear rate. The yield stress value obtained obviously depends on the lowest
shear rate data available and used in the extrapolation. This shear rate dependence of
the extrapolated yield stress has been demonstrated by Barnes and Walters (1985)
with a well-known yield stress fluid, Carbopol (carboxylpolymethylene).

They

concluded that this fluid would have no detectable yield stress even if measurement
was made at very low shear rates of 10-5 s-1 or less. This finding should be viewed
with caution, however, since virtually all viscometric instruments suffer wall slip and

43
other defects which tend to be more pronounced at low shear rates especially with
yield stress fluids and particulate systems [Wildermuth and Williams, 1985; Magnin
and Piau, 1990]. Thus, it is imperative that some checking procedure should be
carried out to ascertain the reliability of the low shear rate data before extrapolation is
made.

3.4.1.2. Extrapolation Using Constitutive Models

A more convenient extrapolation technique is to approximate the experimental


data with one of the viscoplastic flow models. Many workers appear to prefer the
Bingham model which postulates a linear relationship between shear stress and shear
rate. However, since a large number of yield stress fluids including suspensions are
not Bingham plastic except at very high shear rates, the use of the Bingham plastic
model can lead to unnecessary overprediction of the yield stress as shown in Fig. 3-5
[Nguyen and Boger, 1983; de Kee et al., 1986]. Extrapolation by means of nonlinear
Casson model can be used from a linear plot of

1
2

1
2

versus & . The application of

Herschel-Bulkley model is less certain although systematic procedures for


determining the yield stress value and the other model parameters are available
[Heywood and Cheng, 1984].
Even with the most suitable model and appropriate technique, the yield stress
value obtained cannot be regarded as an absolute material property because its
accuracy depends on the model used and the range and reliability of the experimental

44
data available. Several studies have shown that a given fluid can be described equally
well by more than one model and hence can have different yield stress values
[Keentok, 1982; Nguyen and Boger, 1983; Uhlherr, 1986].

3.4.2. Direct Method

Various techniques have been introduced for measuring the yield stress
directly and independently of shear stress-shear rate data. Although the general
principle of the yield stress as the stress limit between flow and non-flow conditions
is often used, the specific criterion employed for defining the yield stress seems to
vary among these techniques. Furthermore, each technique appears to have its own
limitations and sensitivity so that no single technique can be considered versatile or
accurate enough to cover the whole range of yield stress and fluid characteristics.
Usually, the direct methods are used for fluids having yield stresses of greater than
approximately 10 Pa [Nguyen and Boger, 1983]. Therefore, as mentioned earlier, the
direct method is not very convenient to use for the yield stress measurement of blood
since the yield stress of human blood is approximately 1 to 30 mPa [Picart et al.,
1998].

45

Fig. 3-5. Determination of yield stress by extrapolation [Nguyen and Boger, 1983].

46
3.5. Problems with Conventional Viscometers for Clinical Applications

3.5.1. Problems with Rotational Viscometers

Over the years, rotational viscometers have been the standard in clinical
studies investigating rheological properties of blood and other body fluids. Despite
their popularity, rotational viscometers have some drawbacks that limit their clinical
applicability in measuring whole blood viscosity. They include the need to calibrate a
torque-measuring sensor, handling of blood, surface tensions effects, and the range of
reliability.
The torque-measuring sensor can be a conventional spring or a more
sophisticated electronic transducer. In either case, the sensor requires a periodic
calibration because repeated use of the sensor can alter its spring constant. The
calibration procedure is often carried out at manufacturers laboratory because it
requires an extremely careful and elaborate protocol, requiring the viscometer unit to
be returned for service.
Another concern is the need to work with contaminated blood specimens.
After each measurement, the blood sample must be removed from the test section,
and the test section must be cleaned manually. Not only is this procedure timeconsuming, but also it poses a potential risk for contact with contaminated blood.
Surface tension effects arise in the use of the coaxial-cylinder viscometer
because surface tension is relatively high for blood and macromolecular solutions.
The contact area between the blood and an inner cylinder is not uniform along the

47
periphery. The bob (inner cylinder) is pulled in different directions and revealed in
fluctuating torque readings, introducing serious errors in viscosity measurement.
Another inherent difficulty in measuring whole blood viscosity using
rotational viscometers is the limited shear rate range. In the extremes of the reputed
range (whether high shear or low shear, depending on the instrument), the detected
torque values do not have sufficient accuracy. Usually, manufacturers recommend
discarding viscosity data if the torque is less than 10% of the maximum value of the
sensor. This restriction is a major concern. For example, in the case of Brookfield
rotational viscometer, the minimum shear rate is often limited at approximately 30-50
s-1 due to the 10% restriction.
There are other clinical, practical considerations in using the rotational
viscometer. For example, it is usually necessary to treat the blood sample with a
measurable amount of anticoagulant, such as ethylenediaminetetraacetic acid (EDTA)
or heparin, to prevent coagulation during viscosity measurements. The reason for this
is that the contact area among blood, rotational viscometer component, and air is
relatively large for the size of the blood sample, and it usually takes a relatively long
time to complete viscosity measurements over a range of shear rates. Treating blood
with such anticoagulants results in an altered sample, and subsequent viscosity
measurements do not reflect the intrinsic values of unadulterated blood.

48
3.5.2. Problems with Capillary-Tube Viscometers

There are some drawbacks in the use of conventional capillary-tube


viscometers for clinical applications. The range of shear rate is limited to high shears
over 100 s-1. Although one can produce viscosity data at lower shear rates below 100
s-1 with a sophisticated vacuum system, the capillary tube system is basically designed
and operated to obtain viscosity at the high shear range. Since it is essential to obtain
blood viscosity at low shear rates below 10 s-1, the traditional capillary tube
viscometer is not suitable for measuring the viscosity at low shear rates. However,
capillary-tube viscometer is simple in its design and uses gravity field to drive test
fluid such that there is no need for calibration.
It takes a relatively long time to complete viscosity measurements over a
range of shear rates because at each shear rate, a sufficient quantity of a fluid sample
must be collected for an accurate measurement of flow velocity.

After the

measurement at one shear rate, the pressure at the reservoir tank must be readjusted to
either increase or decrease shear rate. Then, the next shear rate case resumes. Thus,
anticoagulants must be added to whole blood for the viscosity measurement over a
range of shear rates.

49
CHAPTER 4. THEORY OF SCANNING CAPILLARY-TUBE RHEOMETER

Chapter 4 presents the theory of scanning capillary-tube rheometer (SCTR).


Mathematical procedures for both viscosity and yield-stress measurements were
demonstrated in detail using power-law, Casson, and Herschel-Bulkley (H-B) models.
Section 4.1 provides a brief introduction to the SCTR. In section 4.1.1, the
description of a U-shaped tube set is reported. In addition, this section shows how the
dimensions of the disposable tube set were determined. Section 4.1.2 demonstrates
the equations for the energy balance in the disposable tube set.
Section 4.2 provides the mathematical details of data reduction for both
viscosity and yield-stress measurements. Sections 4.2.1, 4.2.2, and 4.2.3 deal with
the mathematical modeling in the data reduction by using the power-law, Casson, and
H-B models, respectively. Especially, in sections 4.2.2 and 4.2.3, the yield stress as
well as the viscosity of blood was considered in the data reduction.

4.1 Scanning Capillary-Tube Rheometer (SCTR)

One of the drawbacks of using conventional capillary viscometers is that one


needs to change the pressure in the reservoir tank in order to measure the viscosity at
a different shear rate. Viscosity can only be measured at one shear rate at a time in
the conventional system. Similarly, in other types of viscometers such as rotating
viscometers and falling object viscometers, the rotating speed has to be changed or

50
the density of the falling object has to be changed in order to vary shear rate as
mentioned in Chapter 3. Such operations can make viscosity measurements time
consuming and labor intensive. Because of the time required to measure viscosity
over a range of shear rates, it is necessary to add anticoagulants to blood to prevent
clotting during viscosity measurements with these conventional viscometers. The
present study introduces an innovative concept of a new capillary tube rheometer that
is capable of measuring yield stress and viscosity of whole blood continuously over a
wide range of shear rates without adding any anticoagulants.

4.1.1 U-Shaped Tube Set

Figure 4-1 shows a schematic diagram of a U-shaped tube set, which consists
of two riser tubes, a capillary tube, and a stopcock. The inside diameter of the riser
tubes in the present study is 3.2 mm. The inside diameter and length of the capillarytube are 0.797 and 100 mm, respectively. The small diameter of the capillary tube,
compared with that of the riser tubes, was chosen to ensure that the pressure drops at
the riser tubes and connecting fittings were negligibly small compared to the pressure
drop at the capillary tube [Kim et al., 2000a, 2000b, and 2002].
Furthermore, the inside diameter of the capillary tube was chosen to minimize
the wall effect which is often known as Fahraeus-Lindqvist effect [Fahraeus and
Lingqvist, 1931]. The details of the wall effect will be discussed in Chapter 5. In the
present study, the wall effect was found to be negligibly small.

51
The length of the capillary tube (i.e., Lc = 100 mm) in the U-shaped tube set
was selected to ensure that the end effects would be negligible [Kim et al., 2000a,
2000b, and 2002]. The end effects at the capillary tube will be also reported in
Chapter 5. In addition, the capillary-tube dimensions in the SCTR were selected to
complete one measurement within 2-3 min, a condition that is desirable when
measuring the viscosity of unadulterated whole blood in a clinical environment.
Figure 4-2 shows sketches of the fluid levels in the U-shaped tube set as time
goes on. The fluid level in the right-side riser tube decreases whereas that in the leftside riser tube increases. As time goes to infinity, the two fluid levels never become
equal due to the surface tension and yield stress effects as shown in Fig. 4-2(c) (i.e.,

ht = > 0). While a test fluid travels through the capillary tube between riser tubes 1
and 2, the pressure drop caused by the friction at the capillary tube can be obtained by
measuring the fluid levels at riser tubes 1 and 2. In Fig. 4-3, a typical fluid-level
variation measured by the SCTR is shown. Points (a), (b), and (c) represent the three
moments indicated in Fig. 4-2 (i.e., at t = 0 , t > 0, and t = , respectively).

4.1.2 Energy Balance

Figure 4-4 shows the liquid-solid interface condition for each fluid column of
a U-shaped tube. A falling column (right side) always has a fully wet surface
condition, while a rising column (left side) has an almost perfectly dry surface
condition at the liquid-solid interface during the entire test. Therefore, the surface

52
tension at the right side was consistently greater than that at the left side since the
surface tension of a liquid is strongly dependent on the wetting condition of the tube
at the liquid-solid interface [Jacobs, 1966; Mardles, 1969; Kim et al., 2002]. The
height difference caused by the surface tension at the two riser tubes was one order of
magnitude greater than the experimental resolution desired for accurate viscosity
measurements. Thus, it is extremely important to take into account the effect of the
surface tension on the viscosity measurement using the disposable tube set.
The mathematical model of the flow analysis began with the equation of the
conservation of energy in the form of pressure unit, where the surface-tension effect
was considered between the two top points of the fluid columns at the riser tubes (see
Fig. 4-4). Assuming that the surface tension for the liquid-solid interface at each riser
tube remains constant during the test, one may write the governing equations as [Bird
et al., 1987; Munson et al., 1998]:

P1 +

s 2 V
1
1
V12 + gh1 = P2 + V22 + gh2 + Pc + ght = +
ds ,
s
1
2
2
t

where
P1 and P2 = static pressures at two top points

= density of fluid
g = gravitational acceleration
V1 and V2 = flow velocities at two riser tubes
h1 and h2 = fluid levels at two riser tubes

Pc (t ) = pressure drop across capillary tube


ht = = additional height difference

(4-1)

53
V = flow velocity

t = time
s = distance measured along streamline from some arbitrary initial point.

In Eq. (4-1), the energy emitted from LEDs was ignored since the energy transferred
from the LEDs, which can affect the temperature of a test fluid, was negligible small.
In order to ensure that the amount of the heat emitted from the LEDs is very small,
the temperature of bovine blood was measured during a room-temperature test. The
results showed no changes in temperature during the test, indicating that the energy
emitted from LEDs might be negligibly small.
For the convenience of data-reduction procedure, the unsteady term in Eq. (41),

s2
s1

V
ds , may be ignored under the assumption of a quasi-steady state. In order
t

to make the assumption, one should make sure that the pressure drop due to the
unsteady effect is very small compared with that due to the friction estimated from
the steady Poiseuille flow in a capillary tube.
The unsteady term can be broken into three integrations that represent the
pressure drops due to the unsteady flow along the streamlines at riser tube 1, capillary
tube, and riser tube 2 as [Munson et al., 1998]:

s2
s1

V
ds =
t

s1

s1

s 2 dV
s2 dV

dVr
c
r
ds +
ds +
ds ,
s
s
1
2
dt
dt
dt

(4-2)

where Vr and Vc are mean flow velocities at riser and capillary tubes, respectively.
Since the term of

V
is independent of streamlines, one can simplify the equation as:
t

54

s2
s1

V
ds =
t

dV r
dV
dV
dV
dV
l1 + c Lc + r l 2 = Lc c + (l1 + l 2 ) r ,
dt
dt
dt
dt
dt

(4-3)

where l 1 and l 2 are lengths of the liquid columns whereas Lc is the length of the
capillary tube as shown in Fig. 4-5. Using the mass conservation, Rc2 Vc = Rr2 Vr ,
the pressure drop due to the unsteady effect can be reduced as:
Punsteady =

s2
s1

R
V
ds = Lc r
t
Rc

2
dV

+ l1 + l 2 r ,
dt

(4-4)

where

Punsteady = pressure drop due to the unsteady flow


Rr and Rc = radii of riser and capillary tubes, respectively.

In the present experimental set up, l1 , l 2 , and Lc are measured to be


approximately 12, 4, and 10 cm, respectively. Since h1 (t ) and h2 (t ) are strongly
dependent on each other by the conservation of mass for incompressible fluids, Vr
must be equal to

dVr
dh1 (t )
dh2 (t )
and
. In order to calculate the term of
from the
dt
dt
dt

experimental values, one could use the following central differential method:

[h (t + t ) 2h1 (t ) + h1 (t t )] = [h2 (t + t ) 2h2 (t ) + h2 (t t )] .


dVr
= 1
dt
t 2
t 2

(4-5)

For the comparison of Punsteady with Pc , Punsteady was estimated through a curvefitting process. In order to obtain a smooth curve from raw data, the following
exponential equation was used.
2

dV

Error = r a e bt .
dt

(4-6)

55
Two constants, a and b , were obtained through a curve-fitting process, a leastsquare method, which minimized the sum of error for all experimental data points
obtained in each test.
Typical results showed that the magnitude of the pressure drop due to the
unsteady flow, Punsteady , was always less than 1% of that of pressure drop at capillary
tube, Pc , over the entire shear-rate range. This confirms that the assumption of a
quasi-steady state could be used for the present data procedure. The details of
experimental results will be discussed in Chapter 5.
Assuming a quasi-steady flow behavior, one may rewrite Eq. (4-1) as follows
[Bird et al., 1987; Munson et al., 1998]:

P1 +

1
1
V12 + gh1 (t ) = P2 + V22 + gh2 (t ) + Pc (t ) + ght = .
2
2

(4-7)

Since P1 = P2 = Patm and V1 = V2 , Eq. (3-7) can be reduced as:

Pc (t ) = g [h1 (t ) h2 (t ) ht = ] .

(4-8)

Note that h at t = contains a height difference due to the surface tension, hst ,
and an additional height difference due to the yield stress, h y , for the case of blood
(i.e., see Fig. 4-3).

The next section addresses the mathematical procedure of

handling the yield stress.

56

Open to air

3.2 mm
Riser tubes

Capillary tube
0.797 mm

100 mm

Fig. 4-1. Schematic diagram of a U-shaped tube set.

Stopcock

57

Riser tube 2

Riser tube 1

(a) at t = 0

(b) at t > 0

ht =

(c) at t =

Fig. 4-2. Fluid-level variation in a U-shaped tube set during a test.

58

Height

h1 (t )
ht =
h2 (t )

(a)

(b)
Time

(c)

Fig. 4-3. Typical fluid-level variation measured by a SCTR.


(a) at t = 0 , (b) at t > 0 , and (c) at t = .

59

Wet surface
condition

l1
Dry surface

condition

l2
2'

Lc

1'

Fig. 4-4. Liquid-solid interface conditions for fluid columns of a U-shaped tube set.

60

4.2 Mathematical Procedure for Data Reduction

In Chapter 2, we discussed the non-Newtonian characteristics of whole blood.


This section deals with non-Newtonian constitutive models for blood and their
applications to the SCTR. Since blood has both shear-thinning (pseudo-plastic) and
yield stress characteristics, three different constitutive models were used for the
viscosity and/or yield-stress measurements of blood in this study. Power-law model
was chosen to demonstrate the shear-thinning behavior of blood.

Casson and

Herschel-Bulkley (H-B) models were selected to measure both shear thinning


viscosity and yield stress of blood.
For the purpose of clinical applications, disposable tube sets can be used for
the viscosity and yield-stress measurements of blood. Since the disposable tube sets
have different surface conditions at riser tube 1 and 2 during the test, one needs to
mathematically handle surface tension and yield stress effects in order to measure the
viscosity and yield stress of blood using Casson or H-B model. The details of
mathematical method of isolating those two effects are shown in this section.

4.2.1 Power-law Model

It is well known that power-law model does not have the capability to handle
yield stress. As provided in Chapter 2, the relation among shear stress, shear rate, and
viscosity in power-law fluids may be written as follows:

61

= m& n ,

(4-9)

= m& n 1 .

(4-10)

Since n < 1 for pseudo-plastics, the viscosity function decreases as the shear rate
increases. This type of behavior is characteristic of high polymers, polymer solutions,
and many suspensions including whole blood.
We consider the fluid element in the capillary tube at time t as is shown in
Fig. 4-5. The Hagen-Poiseuille flow may be used to derive the following relationship
for the pressure drop at the capillary tube as a function of capillary tube geometry,
fluid viscosity, and flow rate [Fung, 1990; Munson et al., 1998]:
2 Lc
2 Lc & w 8Lc Q 8Lc Rr2 dh
2l
,
w =
Pc = =
=
=
r
Rc
Rc
dt
Rc4
Rc4

(4-11)

where

r = radial distance
l = length of fluid element

and w = shear stress and wall shear stress, respectively

& w =

4Q
= wall shear rate
Rc3

= Newtonian apparent viscosity


Lc = length of capillary tube
Q = Rr2

dh1
dh
dh
= Rr2 2 = Rr2
= volumetric flow rate.
dt
dt
dt

The above relationship is valid for Newtonian fluids whose viscosities are
independent of shear rate. For non-Newtonian fluids, the viscosities vary with shear

62
rate. However, the Hagen-Poiseuille flow within the capillary tube still holds for a
quasi-steady laminar flow. When applying a non-Newtonian power-law model to
whole blood, the pressure drop at the capillary tube can be described as follows
[Middleman, 1968; Bird et al., 1987; Fung, 1990]:
2 w Lc& w 2mLc & wn 2mLc
Pc =
=
=
Rc
Rc
Rc

3n + 1 Q
3

n Rc

,
=

2mLc
Rc

3n + 1 Rr2 dh
3

n Rc dt

(4-12)

where

w = power-law apparent viscosity


3n + 1 Q
3.
n Rc

& w =

It is of note that if n = 1 , Eq. (4-12) yields to Eq.(4-11). Applying Eqs. (4-8), (4-11),
and (4-12), one can rewrite the energy conservation equation as follows:

g {h1 (t ) h2 (t ) ht = } =

8Lc Rr2 dh
for Newtonian fluids,
dt
Rc4

2mLc
g {h1 (t ) h2 (t ) ht = } =
Rc

3n + 1 Rr2 dh
3

n Rc dt

(4-13)

for power-law fluids.

(4-14)

For convenience, one may define a new function, (t ) = h1 (t ) h2 (t ) ht = so that


Eqs. (4-13) and (4-14) become as follows:
d

= dt for Newtonian fluids,

(4-15)

63

1
n

= dt for power-law fluids,

(4-16)

where
dh
d dh1 dh2
=

= 2 2
dt
dt
dt
dt

gRc4
4 Lc Rr2
1

gRc n

2mLc

=
.
2
3n + 1 Rr

3
n 2 Rc
The above equations are the first-order linear differential equations. Since and
are constants, these equations can be integrated as follows:

(t ) = (0) e t

for Newtonian fluids,

(4-17)

n 1

n 1 n 1
(t ) = (0) n
t
n

for power-law fluids,

(4-18)

where (0) = h1 (0) h2 (0) ht = : initial condition.


Equation (4-18) can be used for curve fitting of the experimental data (i.e.,
h1 (t ) and h2 (t ) ) to determine ht = , the power-law index, n , and the consistency

index, m . A least-square method was used for the curve fitting. The data reduction
procedure adopted is as follows:
1. Conduct a test and acquire all data, h1 (t ) and h2 (t ) .
2. Guess values for m , n , and ht = .
3. Calculate the following error values for all data points:

64

Error = { (t )}Experimental value { (t )}Theoretical value

(4-19)

4. Sum the error values for all data points.


5. Iterate to determine the values of m , n , and ht = that minimize the sum of
error.
6. Let the computer determine whether a test fluid is Newtonian or not.
7. Calculate shear rate and viscosity for all data points as follows:

& w =

Rc
gRc
Pc =
(t )
2 Lc
2 Lc
1

for Newtonian fluids,

(4-20)

R
n gRc
n
& w = c Pc =
(t )
2mLc

2mLc

for power-law fluids.

(4-21)

When n becomes 1 ( 0.001), is equal to m , whereas when 0< n <1, the viscosity
is calculated from Eq. (4-10).
In order to obtain the velocity profile at the capillary tube, which changes with
time, using a power-law model, Eq. (4-21) can be used to derive it. Since & =

dV
,
dr

the velocity profile can be expressed as follows:


1

r
n
dV (t , r )
=
Pc (t ) ,
dr
2mLc

1
P (t ) n
P (t ) n
V (t , r ) = c r n dr = c
2mLc
2mLc

r
n + 1

n +1
n

+C ,

(4-22)

where C is a constant. Using no-slip condition on the capillary wall, V (t , Rc ) = 0 ,


the constant can be obtained as:

65
1

P (t ) n n nn+1
C = c
Rc .
2mLc n + 1

(4-23)

Finally, the velocity profile within the capillary tube can be expressed as follows:
1

n Pc (t ) n
Vc (t , r ) =

n + 1 2mLc

n +1
nn+1

Rc r n

(4-24)
n h1 (t ) h2 (t ) ht =
=

2mLc
n + 1

1
n

n +1

nn+1
Rc r n

where Pc (t ) = g [h1 (t ) h2 (t ) ht = ] . Note that if power-law index becomes zero,


n = 1 , then the above equation yields to the equation for the Newtonian velocity
profile as:
P (t )
Vc (t , r ) = c (Rc2 r 2 ) .
4 Lc

(4-25)

In order to determine the mean flow velocity at the riser tube, one has to find
the flow rate at the capillary tube first. The flow rate can be obtained by integrating
the velocity profile over the cross-sectional area of the capillary tube as follows:

Q(t ) = 2

Rc
0

Vc (t , r )rdr
1

3 n +1
n Pc (t ) n n
R
=

c
3n + 1 2mLc

(4-26)

1
n

3 n +1
n g [h1 (t ) h2 (t ) ht = ]
n
=

Rc
2mLc
3n + 1

Since Q(t ) = Rr2Vr (t ) , the mean flow velocity at the riser tube can be determined by
the following equation:

66
1

3 n +1

n Pc (t ) n Rc n
Vr (t ) =

2
3n + 1 2mLc Rr

(4-27)
1

3 n +1
n
c
2
r

n g [h1 (t ) h2 (t ) ht = ] n R
=

R
2mLc
3n + 1

where Rr is the radius of the riser tube.

4.2.2 Casson Model

The Casson model can handle both yield stress and shear-thinning
characteristics of blood, and can be described as follows [Barbee and Cokelet, 1971;
Benis et al., 1971; Reinhart et al., 1990]:

= y + k &

when y ,

when y ,

& = 0

(4-28)
(4-29)

where

and & = shear stress and shear rate, respectively

y = a constant that is interpreted as the yield stress


k = a Casson model constant.

Wall shear stress and yield stress can be defined as follows:

w =

y =

Pc (t ) Rc
,
2 Lc
Pc (t ) ry (t )
2 Lc

(4-30)

(4-31)

67
where ry is a radial location below which the velocity profile is uniform as shown in
Fig. 4-6, i.e., plug flow, due to the yield stress. Now, for the Casson model, Eq. (4-8)
becomes Pc (t ) = g [h1 (t ) h2 (t ) hst ] , indicating that the effect of the surface
tension is isolated from the pressure drop across the capillary tube. Using Eqs. (4-28)
and (4-29), one can obtain the expressions of shear rate and velocity profile at the
capillary tube as follows:

Pc (t ) ry (t )
dV
1 Pc (t ) r
,

& = c =

dr
k
2 Lc
2 Lc

Vc (t , r ) =

(4-32)

3
3
1 Pc (t ) 2
8 12

2
2
2

R
r
r
(
t
)(
R
r
) + 2ry (t )( Rc r )
c
y
c

4k
3
Lc

for ry (t ) r Rc , (4-33)

Vc (t ) =

1 Pc (t )
1

( Rc ry (t ) ) 3 ( Rc +
ry (t ) )
4k
3
Lc

for ry (t ) r .

(4-34)

For the purpose of simplicity, one may define two new parameters,

C (r ) =

ry (t )
r
, so that Eqs. (4-33) and (4-34) become as follows:
and C y (t ) =
Rc
Rc

1 Pc (t ) 2 r

Rc 1
Vc (t , r ) =
4k
R
Lc
c

1
2

8 ry 2 r
1
3 Rc R c

r
2
+ 2 y
Rc

r
1

Rc

1
3

Rc2 Pc (t )
8 2
2
=
1 C (r ) C y (t )1 C 2 (r ) + 2C y (t )(1 C (r ) )
4kLc
3

for ry (t ) r Rc , (4-35)

68

Pc (t )
Vc (t ) =
Rc
4kLc

( )

r
1 y

Rc


1 ry

Rc 1 +

3
R
c

for ry (t ) r ,

(4-36)

3
Rc2 Pc (t )
1

1 C y (t ) 1 +
C y (t )
4kLc
3

where C y (t ) =

ry (t )
Rc

y
w (t )

P ()
.
P(t )

In order to determine the mean flow velocity at the riser tube, one has to find
the flow rate at the capillary tube first. The flow rate can be obtained by integrating
the velocity profile over the cross-sectional area of the capillary tube as follows:
Rc

Q(t ) = 2 Vc r dr
0

Rc4 Pc (t )
8k

[(

Lc

16 2 y 12 Pc (t ) 1 2
(
) (
)
Rc
Lc
7

4 2 y
1 2 y 4 Pc (t ) 3
(
) (
) (
) ]
Lc
3 Rc
21 Rc

(4-37)

1 y 4
16 y 12 4 y
1 ( ) + ( ) ( )
21 w
3 w
7 w

Rc4 Pc
8kLc

Rc4 Pc
8kLc

16 12 4
1
4
Cy + Cy Cy

7
3
21

Since Q(t ) = Rr2Vr (t ) , the mean flow velocity at the riser tube can be determined by
the following equation:

69

Vr (t ) =

Rc4 Pc (t ) 16 2 y 1 2 Pc (t ) 1 2
[(
) (
) (
)
7
Lc
Rc
Lc
8kRr2
+

4 2 y
1 2 y 4 Pc (t ) 3
) (
(
) (
) ]
3 Rc
21 Rc
Lc

(4-38)

Rc4 Pc
16 1
4
1
[1 C y 2 + C y C y4 ]
2
7
3
21
8kRr Lc

where Rr is the radius of the riser tube.


For the purpose of simplicity, Eq. (4-38) can be rewritten to clearly display
the unknowns and the observed variables as:
Vr (t ) =

1
Rc4 g
16
[(h1 (t ) h2 (t ) hst ) (h y (h1 (t ) h2 (t ) hst )) 2
2
7
8kRr Lc

4
1
+ h y h y4 (h1 (t ) h2 (t ) hst ) 3 ]
3
21
where C y (t ) =

h y
P ()
=
.
P(t ) h1 (t ) h2 (t ) hst

(4-39)

Note that Eq. (4-39) contains two

independent variables, i.e., h1 (t ) and h2 (t ) , and one dependent variable, i.e., Vr (t ) .


There are three unknown parameters to be determined through the curve fitting in Eq.
(4-39), namely hst , k , and h y . hst is h due to the surface tension, k is the
Casson constant, and h y is h due to the yield stress.
Once the equation for the mean flow velocity, Vr (t ) , was derived, one could
determined the unknown parameters using the experimental values of h1 (t ) and h2 (t ) .
A least-square method was used for the curve fitting. For the Casson model, there
were three unknown values, which were k , hst , and y . Note that the unknown
values were assumed to be constant for the curve-fitting method. Since h1 (t ) and

70

h2 (t ) are strongly dependent on each other by the conservation of mass for


incompressible fluids,

dh1 (t )
dh (t )
must be equal to 2 . Therefore, it was more
dt
dt

convenient and accurate to use the difference between the velocities at the two riser
tubes, i.e.,

d (h1 (t ) h2 (t ))
dh (t )
dh (t )
, than to use 1
and 2
directly. In order to get
dt
dt
dt

the difference between the two velocities from the experimental values, one could use
the central differential method as follows:
d (h1 (t ) h2 (t ) ) [h1 (t + t ) h2 (t + t )] [h1 (t t ) h2 (t t )]
=
.
dt
2 t

(4-40)

Using Eq. (4-39), the derivative of the velocity difference can be determined
theoretically as follows:
d (h1 (t ) h2 (t ) )
= 2Vr (t )
dt
=

1
Rc4 g
16
[(h1 (t ) h2 (t ) hst ) (h y (h1 (t ) h2 (t ) hst )) 2
2
7
4kRr Lc

4
1
h y h y4 (h1 (t ) h2 (t ) hst ) 3 ]
3
21
(4-41)

where Vr (t ) is the mean flow velocity at the riser tube.


In order to execute the curve-fitting procedure, one needs to have a
mathematical equation of Vr for the Casson model. Eq. (4-40) and (4-41) were used
for the curve fitting of the experimental data to determine the unknown constants, i.e.,
k , hst , and h y . Note that Eq. (4-41) could be applicable for both Casson-model

71
fluids and Newtonian fluids regardless of the existence of the yield stress. The data
reduction procedure adopted is as follows:
1. Conduct a test and acquire all data, h1 (t ) and h2 (t ) .
2. Guess values for the unknowns, k , hst , and h y .
3. Calculate the following error values for all data points.

Error = {2V (t )}Experimental values {2V (t )}Theoretical values

(4-42)

4. Sum the error values for all data points.


5. Iterate to determine the unknowns that minimize the sum of the error.
6. Calculate wall shear rate and viscosity for all data points as follows:

& w (t ) =
w (t ) =

gRc
2kLc

h1 (t ) h2 (t ) hst h y

gRc [h1 (t ) h2 (t ) hst ]


.
2 & w (t ) Lc

(4-43)

(4-44)

Note that when h y becomes approximately zero (i.e., resolution of 8.3 10 5 ), the
non-Newtonian viscosity, , is reduced to k , a Newtonian viscosity. Furthermore,
the relation between wall viscosity and shear-rate can be obtained from Eqs. (4-43)
and (4-44) as follow:

w (t ) = k +

y
& w (t )

4k y

& w (t )
(4-45)

gRc h y

2 Lc
=k+
& w (t )

2kgRc h y
Lc

& w (t )

where k and h y are the fluid properties to be determined using the Casson model.

72
Yield stress could be also determined through the curve-fitting method from
the experimental data of h1 (t ) and h2 (t ) by using the Casson model. Since the
pressure drop across the capillary tube, Pc (t ) , could be determined using Eq. (4-8),
Pc () represents the effect of the yield stress on the pressure drop. The relationship
between the yield stress, y , and Pc () can be written by the following equation:

y =

Pc () Rc gh y Rc
.
=
2 Lc
2 Lc

(4-46)

Therefore, once h y is obtained using a curve-fitting method, the yield stress can be
automatically determined.

4.2.3 Herschel-Bulkley (H-B) Model

For a Herschel-Bulkley (H-B) model, the shear stress at the capillary tube can
be described as follows [Tanner, 1985; Ferguson and Kemblowski, 1991; Macosko,
1994]:

= m& n + y

& = 0

when y ,

when y ,

where

and & = shear stress and shear rate, respectively

y = a constant that is interpreted as yield stress


m and n = model constants.

(4-47)
(4-48)

73
Since the H-B model reduces to the power-law model when a fluid does not have a
yield stress, the H-B model is more general than the power-law model.
For the H-B model, wall shear stress and yield stress can also be defined as
follows:

w =

y =

Pc (t ) Rc
,
2 Lc
Pc (t ) ry (t )
2 Lc

(4-49)

Pc () Rc gh y Rc
,
=
2 Lc
2 Lc

(4-50)

where ry is a radial location below which the velocity profile is uniform due to the
yield stress (see Fig. 4-7). Using Eqs. (4-47)-(4-50), one can obtain the expressions
of shear-rate outside of the core region as:
Pc
dV
& = c =
dr 2mLc

1
n
(r ry ) n

for ry (t ) r Rc .

(4-51)

The velocity profile outside of core region can be obtained by integrating Eq. (4-51)
as:
1

n +1
n +1

n Pc (t ) n
n
n
(
)
(
)
Vc (t , r ) =
R
r
(
t
)
r
r
(
t
)

c
y
y

n + 1 2mLc

for ry (t ) r Rc .

(4-52)

Since the velocity profile inside of the core region is a function of time, t , only, the
profile can be obtained using a boundary condition, Vc (t , r ) = Vc (t ) at r = ry .
1

n +1
n Pc (t ) n
n
(
)
Vc (t ) =
R

r
(
t
)

c
y

+
n
1
2
mL

for ry (t ) r .

(4-53)

74
Again, for the purpose of simplicity, one may define two new parameters, C (r ) =

and C y (t ) =

ry (t )
Rc

r
Rc

, so that Eqs. (4-52) and (4-53) become as follows:

1
n +1
n +1

n +1
n Rc Pc (t ) n ry (t ) n r ry (t ) n

Vc (t , r ) =

Rc
Rc
n + 1 2mLc
Rc

n +1
n +1
n +1

n Rc Pc (t ) n
n
n
(
)
(
)
1
C
(
t
)
C
(
r
)
C
(
t
)
=

y
y
n + 1 2mLc

for ry (t ) r Rc , (4-54)
1

n +1
n Rc Pc (t ) n
Vc (t ) =

n + 1 2mLc

ry (t )

1
Rc

n +1
n

for ry (t ) r .

(4-55)

1
n

n +1
n +1
n Rc Pc (t )
(1 C y (t ) ) n
=

n + 1 2mLc

In order to determine the mean flow velocity at the riser tube, one has to find
the flow rate at the capillary tube first. The flow rate can be obtained by integrating
the velocity profile over the cross-sectional area of the capillary tube as follows:
Rc

Q(t ) = 2 Vc r dr
0

Pc
=
2mLc

2 n +1
n +1
n n 2
n
n

(
)
(
)
(
)

[
r
R

r
+
R
+
r

r
y c y
c
y
c
y
n + 1

n
2
ry (Rc ry )
2n + 1

2 n +1
n

n
2
(Rc ry )
3n + 1

3 n +1
n

]
(4-56)

75

Pc
=
2mLc
2R

= R

3 n +1
n
c

3 n +1
ry
n n

[ Rc n
n + 1
Rc

3 n +1
n
c

n ry

2n + 1 Rc

n Pc

n + 1 2mLc

ry (t )
Rc

1 y
Rc

1 y
Rc

2 n +1
n

n +1
n

2R

+R

3 n +1
n
c

3 n +1
n
c

r
r

1 + y 1 y
Rc R c

ry
n

1
3n + 1 Rc

3 n +1
n

2 n +1
n

2 n +1
n +1
n 2
n
[C y (1 C y ) + (1 + C y ) (1 C y ) n

n
2
C y (1 C y )
2n + 1
where C y (t ) =

2 n +1
n

y
w (t )

3 n +1
n

n
2
(1 C y )
3n + 1

h y
P ()
=
.
P (t ) h1 (t ) h2 (t ) hst

Since Q(t ) = Rr2Vr (t ) , the mean flow velocity at the riser tube can be
determined by the following equation:

3 n +1

R n n Pc
Vr (t ) = c 2

Rr n + 1 2mLc

n +1
2 n +1
n 2
[C y (1 C y ) n + (1 + C y ) (1 C y ) n

n
2
C y (1 C y )
2n + 1

2 n +1
n

(4-57)
3 n +1
n

n
2
(1 C y )
3n + 1

Equation (4-57) can be rewritten to clearly display the unknowns and the observed
variables as follows:

76
3 n +1

R n n g [h1 (t ) h2 (t ) hst ] n

Vr (t ) = c 2

2mLc
Rr n + 1

h y

[
h1 (t ) h2 (t ) hst
h y

h y

1
h1 (t ) h2 (t ) hst

+ 1 +
h1 (t ) h2 (t ) hst

h y

n +1
n


1
h1 (t ) h2 (t ) hst

h y
n
2

2n + 1 h1 (t ) h2 (t ) hst

2 n +1
n

h y

1
h1 (t ) h2 (t ) hst

h y
n
2
1
3n + 1 h1 (t ) h2 (t ) hst

3 n +1
n

(4-58)

2 n +1
n

Note that Eq. (4-58) contains two independent variables, i.e., h1 (t ) and h2 (t ) , and
one dependent variable, i.e., Vr (t ) .

There are four unknown parameters to be

determined through the curve fitting in Eq. (4-58), namely m , n , hst , and h y .
Once the equation for the mean flow velocity, Vr (t ) , was derived, one could
determined the unknown parameters using the experimental values of h1 (t ) and h2 (t )
by using the same curve-fitting method of determining unknowns as in the case of the
Casson model. In the case of the H-B model, there were four unknown values, which
were m , n , hst , and h y . Note that the unknown values were assumed to be
constant for the curve-fitting method.
Using Eq. (4-58), the derivative of the velocity difference can be determined
theoretically as follows:

77
d (h1 (t ) h2 (t ) )
= 2Vr (t )
dt
3 n +1

2 Rc n n g [h1 (t ) h2 (t ) hst ] n

2mLc
Rr2 n + 1

h y

[
h1 (t ) h2 (t ) hst

h y


1
h1 (t ) h2 (t ) hst

h y

+ 1 +
h1 (t ) h2 (t ) hst

n +1
n

h y

1
h1 (t ) h2 (t ) hst

h y
n
2

2n + 1 h1 (t ) h2 (t ) hst

2 n +1
n

h y

1
h1 (t ) h2 (t ) hst

h y
n
2
1
3n + 1 h1 (t ) h2 (t ) hst

3 n +1
n

2 n +1
n

]
(4-59)

where Vr (t ) is the mean flow velocity at the riser tube. In order to execute the curvefitting procedure, one needs to have a mathematical equation of Vr for the H-B model.
Eq. (4-58) and (4-59) were used for the curve fitting of the experimental data to
determine the unknown constants, i.e., n , m , hst , and h y . Note that Eq. (4-59)
could be applicable for H-B fluids, Shear-thinning fluids, and Newtonian fluids
regardless of the existence of the yield stress.
After iterations for the determination of the unknowns that minimize the sum
of the error, wall shear rate and viscosity for all data points can be calculated as
follows:
gRc
& w (t ) =
2mLc

n
(h1 (t ) h2 (t ) hst ) h y

1
n

(4-60)

78

w (t ) =

gRc [h1 (t ) h2 (t ) hst ]


.
2 & w (t ) Lc

(4-61)

Note that when h y becomes approximately zero (i.e., resolution of 8.3 10 5 ), the
H-B model is reduced to power-law model. In addition, when n becomes 1, the
mathematical form of the H-B model yields to Bingham plastic [Tanner, 1985], which
can be described as follows:

= m B & + y
& = 0

when y ,

(4-62)

when y ,

(4-63)

where

and & = shear stress and shear-rate, respectively

y = a constant that is interpreted as the yield stress


m B = a model constant that is interpreted as the plastic viscosity.

Similar to the Casson model, the relationship between wall viscosity and
shear-rate using the H-B can be expressed as follows:

w (t ) = m& w (t ) n 1 +

y
& w (t )
(4-64)

gRc h y

2 Lc
n 1
= m& w (t ) +
& w (t )

where m , n , and h y are the fluid properties to be determined using the H-B model.

79

Capillary tube

Rc

l
(a) Motion of a cylindrical fluid element within a capillary tube.

2rl

Pr 2

Flow direction

( P P)r 2

(b) Free-body diagram of a cylinder of fluid.

Fig. 4-5. Fluid element in a capillary tube at time t .

80

Capillary tube

ry

Rc

Fig. 4-6. Velocity profile of plug flow of blood in a capillary tube.

81
CHAPTER 5. CONSIDERATIONS FOR EXPERIMENTAL STUDY

Chapter 5 presents the issues and considerations in the experimental study


with a scanning capillary-tube rheometer (SCTR). Theoretical and experimental
issues involved in the viscosity and yield stress measurements of fluids, such as
distilled water, bovine blood, and human blood, are examined.
Sections 5.1, 5.2, and 5.3 address the major assumptions in the study that may
affect the rheological measurements in the SCTR: unsteady effect, end effect, and
wall effect, respectively. In addition, section 5.4 reports other possible factors that
include pressure drop at riser tubes, effect of density variation of blood, and
thixotropic effect.
In section 5.5, the temperature consideration for the viscosity measurement of
human blood in the SCTR during a test is discussed. The temperature of human
blood was checked to see if it could be maintained at a body temperature of 37
during a viscosity measurement.
In section 5.6, the study on the effect of dye concentration on the viscosity of
distilled water is presented. The objective of the study was to see whether or not the
viscosity of distilled water could be altered by the addition of dye.

82
5.1. Unsteady Effect

In order to make the assumption of a quasi-steady flow behavior during a test


with the SCTR, one needs to make sure that the pressure drop due to the unsteady
state is negligibly small compared to that due to the friction through a capillary tube.
Distilled water and bovine blood were analyzed for the unsteady effects on the
viscosity measurements of the fluids.
Figure 5-1 shows the pressure drops due to both unsteady flow and friction at
the capillary tube obtained by using Eqs. (4-4), (4-6), and (4-8) in the case of distilled
water. Usually, the pressure drop due to the unsteady flow was less than 3 Pa at the
beginning of a test while pressure drop due to the friction at the capillary tube was
greater than 250 Pa (see Table 5-1).
As shown in Fig. 5-2, in the case of bovine blood, the pressure drop due to the
unsteady flow was also much smaller than that at the capillary tube. Typically, the
pressure drop due to the unsteady flow was less than 1.2 Pa at the beginning of a test
while pressure drop due to the friction at the capillary tube was greater than 700 Pa.
Furthermore, as shown in Table 5-2, the magnitude of the pressure drop due to the
unsteady flow, Punsteady , was always less than 1% of that of the pressure drop at the
capillary tube, Pc , over the entire shear-rate range.

This confirms that the

assumption of a quasi-steady state could be used for the present data reduction
procedure.

(a)

Pressure drop (Pa)

83

Punsteady

3
2
1
0
0

10

20

30

(b)

Pressure drop (Pa)

Time (s)

400

Pc

300
200
100
0
0

10

20

30

Time (s)
Fig. 5-1. Pressure drop estimation for distilled water. (a) Pressure drop due to an
unsteady flow in a test with distilled water. (b) Pressure drop at a capillary
tube in a test with distilled water.

84

Table. 5-1. Comparison of Punsteady and Pc for distilled water.

Punsteady

Time (s)

Punsteady (Pa)

Pc (Pa)

0.5

2.89

245.46

1.18

2.61

222.08

1.18

1.56

132.60

1.18

0.93

79.40

1.17

10

0.26

22.98

1.13

15

0.07

6.85

1.02

20

0.02

2.02

0.99

Pc

100 (%)

(a)

Pressure drop (Pa)

85

0.8

Punsteady
0.4

0
0

30

60

90

120

150

(b)

Pressure drop (Pa)

Time (s)

800

Pc

600
400
200
0
0

30

60

90

120

150

Time (s)
Fig. 5-2. Pressure drop estimation for bovine blood. (a) Pressure drop due to an
unsteady flow in a test with bovine blood. (b) Pressure drop at a capillary
tube in a test with bovine blood.

86

Table. 5-2. Comparison of Punsteady and Pc for bovine blood.


Punsteady

Time (s)

Punsteady (Pa)

Pc (Pa)

0.5

0.54

705.21

0.08

0.52

688.11

0.08

0.39

535.93

0.07

10

0.28

399.14

0.07

30

0.067

139.24

0.05

60

0.008

42.63

0.02

120

15.27

Pc

100 (%)

87
5.2. End Effect

Figure 5-3 shows the flow-pattern changes due to end effects at both (a)
entrance and (b) exit of a capillary tube. Due to the sudden contraction and expansion,
additional pressure drops can occur at the both ends. The most common method used
to estimate these minor pressure drops is to use the loss coefficient, K L , which is
defined as [Munson et al., 1998]:
KL =

PEnd
1
Vc 2
2

(5-1)

so that

PEnd = K L

1
Vc 2
2

(5-2)

where PEnd is the pressure drop due to the end effects and Vc is the mean velocity at
the capillary tube.
With the present experimental set-up, the velocity in the capillary tube was
approximately 16 times greater than that in the riser tube. Therefore, the energy loss
by secondary flow patterns or eddies in the entrance and exit of the capillary tube
may appear to be significant in a high shear zone. In the case of a laminar flow, the
loss coefficient was reported to be approximately 2.24 [Ferguson and Kemblowski,
1991]. Using the value of the loss coefficient, the pressure loss due to the sudden
changes in geometry, PEnd , became only 1.79 Pa (for distilled water) and 1.88 Pa
(for bovine blood) for the maximum shear rate of 400 s-1 at a corresponding velocity
of 0.04 m/s. In contrast, the pressure drops across the capillary tube at the maximum

88
shear rate were 245 Pa (for distilled water) and 705 Pa (for bovine blood), indicating
that the loss due to the secondary flow patterns or eddies at both entrance and exit
could be neglected.
In these end regions (see Fig. 5-3), the flow is changing from (or to) its
previous (or future) distribution outside the capillary tube. The length of an end
region is generally a function of tube geometry and some dynamic parameters. The
entrance length, the length of tube required to achieve the fully developed simple
shear flow, can be estimated by using the following equation [Middleman, 1968]:
Le
0.035 Re
D

(5-3)

where Le is the entrance region, D ( 0.8 mm) is the inner diameter of a capillary
tube, and Re is the Reynolds number. The maximum Reynolds number from a

typical run in the present study was approximately 28.6. The entrance length in the
capillary tube used for the present study was estimated to be 0.0008 m using the
above equation. Generally, the ratio of the entrance length to the capillary-tube
length,

Le
, should be the order of 0.01 in order to assume the effect of entrance
Lc

length to be negligible [Middleman, 1968]. Since the ratio was 0.008 in the present
study, it is reasonable to assume that the entrance length effect is negligibly small.

89

(a) Sudden contraction at entrance of a capillary tube

(b) Sudden expansion at exit of a capillary tube

Fig. 5-3. Flow-pattern changes due to end effects [Munson et al., 1998].

90
5.3. Wall Effect (Fahraeus-Lindqvist Effect)

Apart from end effects, other sources of error in a capillary-tube rheometry


should be considered. The wall effect is one of the most important error sources
[Barnes, 1995; Missirlis et al., 2001]. For example, during the flow of a suspension, a
thin layer of the solvent whose viscosity is lower than the viscosity of the suspension
solution may be formed near the capillary wall, and this wall effect becomes more
significant with the decrease in the capillary diameter [Dinnar, 1981; Ferguson and
Kemblowski, 1991; Fung, 1993].
In the case of blood flow, as shown in Fig. 5-4, the wall effect can be
described as a tendency for RBCs to move toward the center of the tube or blood
vessel [Thomas, 1962; Picart et al., 1998]. The plasma-rich zone next to the solid
wall, although very thin, has an important effect on blood rheology. In other words,
the plasma-rich layer near the wall must affect the measurement of blood viscosity by
any instrument with a solid wall. The reduction in the RBC concentration in this
layer near the wall decreases the measured value of blood viscosity, resulting in
erroneous viscosity results.
Thus, the apparent viscosity of whole blood decreases with the decrease in
tube diameter. However, as shown in Fig. 5-5, the wall effect is reported to be
negligibly small when the tube diameter is greater than approximately 0.4 mm
[Fahraeus R and Lindquist, 1931; Dintenfass, 1971; Dinnar, 1981; Stadler et al.,
1990; Pries et al., 1992] or 0.8 mm [Haynes, 1960; Barbee, 1971]. Note that the

91
apparent viscosity of blood decreases to a value close to plasma viscosity if the
diameter of the capillary tube decreases below 0.1 mm [Benis et al., 1970].
In order to check whether or not the present capillary tube diameter (with
0.797 mm ID) was large enough to prevent the wall effect, two additional capillary
tubes, whose diameters were 1.0 mm (with length = 130 mm) and 1.2 mm (with
length = 156 mm), were used for the viscosity measurements of bovine blood with
7.5% EDTA at a room temperature of 25. As shown in Fig. 5-6, the experiments
performed with three different capillary tubes with ID of 0.797 mm (the standard size
of the SCTR), 1.0 mm, and 1.2 mm provided almost identical viscosity results,
confirming that the wall effect was negligibly small for the present capillary tube with
ID equal to 0.797 mm.

92

Arterial wall
Flow

RBCs

Cell-free region

Fig. 5-4. Migration of cells toward to the center of lumen (wall effect).

93

Blood viscosity (cP)

1
10

100

400

800

Tube diameter (microns)


Fig. 5-5. Fahraeus-Lindquist effect due to the reduction in hematocrit in a tube with a
small diameter and the tendency of erythrocytes to migrate toward the center of the
tube [Fahraeus and Lindquist, 1931; Dintenfass, 1971; Dinnar, 1981; Stadler et al.,
1990; Pries et al., 1992]

94

Viscosity (cP)

100
0.797 mm
1.0 mm
1.2 mm
10

1
1

10
100
-1
Shear rate (s )

1000

Fig. 5-6. Viscosity measurements for bovine blood with three different capillary tubes
with ID of 0.797 mm (with length = 100 mm), 1.0 mm (with length = 130 mm), and
1.2 mm (with length = 156 mm).

95
5.4. Other Effects

5.4.1. Pressure Drop at Riser Tube

Since the small diameter of a capillary tube was selected to make sure that the
pressure drop at the capillary tube could be dominant, the pressure drops at riser tubes
should be negligibly small. It has been suggested that the pressure drop in the
reservoir should be estimated by using a power-law model as [Marshall and Riley,
1962; Metzger and Knox, 1965; Macosko, 1994]:
Pr =

Pc Lr
R
Lc ( r )
Rc

(5-4)

n +3
n

where
Pr = pressure drop in reservoir

Lr = wetted length in reservoir


Rr = radius of reservoir.
The power-law model is one of the simplest models, which can be used to
show non-Newtonian behavior of blood.

Furthermore, the power-law model

generally provides almost identical viscosity results with both Casson and HerschelBulkley models at the shear rates between approximately 300 and 30 s-1.

The

viscosity results of blood obtained with those models will be discussed in Chapter 6
in detail. Typically, the power-law index, n , for healthy human blood is 0.75-0.85 at
a body temperature of 37.

96
Considering the reservoir in Eq. (5-4) as the riser tubes in the present system
and n = 0.8 for human blood, one can obtain the following relation between pressure
drops at capillary and riser tubes using Eq. (5-4).

Pr

1
Pc
500

(5-5)

Therefore, in the case of human blood, the sum of the pressure drops at riser tubes is
approximately 0.2 Pa at a shear rate of 30 s-1 while the pressure drop at the capillary
tube is approximately 93 Pa.
It could be argued that Casson or Herschel-Bulkley model would have a larger
pressure drop than the power-law model at a lower shear rate. Therefore, we want to
examine whether or not the pressure drop at the riser tube is still negligibly small for
Casson or Herschel-Bulkley model compared to that at the capillary tube at a very
low shear rate by looking at the upper bound of the error. It is rather obvious that the
pressure drop at the riser tube at a low shear rate (i.e., below 30 s-1) should be smaller
than 0.2 Pa. Lets consider a shear rate of 1 s-1. The pressure drop at the capillary
tube at & = 1 s-1 is approximately 15 Pa. Therefore, the pressure drop at the riser tube
is less than 1.33% (i.e.,

0.2
= 0.0133 ). Hence, it is reasonable to assume that the
15

pressure drops at the riser tubes can be ignored compared to the pressure drop at the
capillary tube.

97

5.4.2. Effect of Density Variation

In order to measure the viscosity of blood by using the SCTR, one needs to
know the density of blood. However, in the case of human blood, it is not very
convenient to measure the density of blood for each viscosity measurement.
Therefore, the following relation [Chien et al., 1987] between hematocrit (Hct as a
dimensionless fraction) and blood density ( in kg/m3) was used for the estimation
of the density of human blood.

= 1026 + 67 Hct

(5-6)

Table 5-3 shows the density of blood corresponding to hematocrit. In normal


hematocrit concentrations, i.e., 35-45% [Guyton and Hall, 1996], the density variation
is less than 1%, which barely affects the viscosity results of blood.

5.4.3. Aggregation Rate of RBCs Thixotropy

As discussed in Chapter 2, the thixotropic effect on blood viscosity may be


more significant at low shear rates than high shear rates. In the SCTR, the shear rate
varies from high (approximately 400 s-1) to low (1 s-1) values. At least 50 seconds is
required during a test to have the fully aggregated quiescent state at a shear rate near
1 s-1 [Gaspar-Rosas and Thurston, 1998].
In the viscosity measurement of human blood with the SCTR, a typical test
duration in which the shear rate decreased from 10 to 1 s-1 was longer than 60 seconds.

98
It is reasonable to assume that the 60-second period is long enough to cause
aggregations if the aggregations were going to take place. To further validate the
above assumption, a longer capillary tube (125-mm length) was used. Since the test
duration increased in the longer capillary tube, an anticoagulant (7.5% EDTA) was
added to avoid the blood clotting. As shown in Fig. 5-7, the viscosity results obtained
by using the longer capillary tube showed excellent agreements with those obtained
by using the capillary tube with 100-mm length (also with 7.5% EDTA). Therefore,
it is concluded that, in the present system, the thixotropic effect of blood on the
viscosity measurement is negligibly small.

99

Table. 5-3. Density estimation


Hematocrit
(%)

Density
(kg/m3)

35

1049.5

40

1052.8

45

1056.2

50

1059.5

100

100

Viscosity (cP)

100 mm
125 mm
10

1
1

10

100

1000

-1

Shear rate (s )

Fig. 5-7. Viscosity results for human blood with two different capillary tubes with
length of 100 mm (with ID = 0.797 mm) and 125 mm (ID = 0.797 mm).

101

5.5. Temperature Considerations for Viscosity Measurement of Human Blood

For unadulterated human blood, the temperature of a SCTR was controlled


during the test at a body temperature of 37C by using preheated disposable tube sets
and a heating pad installed inside the SCTR. In order to check whether or not the
temperature of blood was maintained at a body temperature, a special U-shaped tube
set was prepared for the experiment. Figure 5-8 shows the special U-shaped tube set
which is basically the same as a standard U-shaped tube set except three additional
thermocouples placed at both ends and on the outside surface of the capillary tube.
The temperatures of blood at three predetermined points were measured
during a viscosity test. Figure 5-9 provides the temperature measurement results for
human blood during the test. The temperature of blood at the exit of the capillary
tube was maintained at approximately 38, whereas that at the entrance was
gradually increased from about 36 at the beginning of the test and reached 37.5
at the end of the test. Therefore, it is reasonable to say that the temperature of human
blood flowing through the capillary tube during the test was maintained at a body
temperature of 37 with 1.

102

Capillary Surface
Thermocouple

Exit
Thermocouple

Entrance
Thermocouple
Thermometer

Fig. 5-8. Schematic diagram of a U-shaped tube set for temperature measurement.

103

Temperature ()

40
38
36
Entrance
Exit
Capillary Surface

34
32
30
0

50

100

150

200

Time (s)
Fig. 5-9. Temperature measurement at a capillary tube during a viscosity test.

104

5.6. Effect of Dye Concentration on the Viscosity of Water

5.6.1. Introduction

Figure 5-10 shows a schematic diagram of a SCTR, which consists of two


charge-coupled devices (CCDs) that are positioned vertically, two light-emitting
diodes (LEDs), two riser tubes and a capillary tube, a stopcock, and a data-acquisition
system. The essential feature in the SCTR is to use two riser tubes, where initial fluid
levels are different: one riser tube has a higher fluid level than the other one. Thus, at

t = 0, the fluid begins to fall from the riser tube with the high level to the riser tube of
low level by gravity. Since the flow rate depends on the pressure head between the
two fluid levels, the flow rate gradually decreases with time as the difference between
the two fluid level decreases with time. Since the flow rate can be estimated from the
time rate of change of the fluid level, one can estimate both flow rate and pressure
drop from the measurement of two fluid levels. Then, one can calculate shear rate
from the flow rate data and shear stress from the pressure drop data, respectively.
From the shear rate and shear stress, one can determine the viscosity of the liquid.
Thus, the most important experimental variable in the operation of the SCTR
is the measurement of two fluid levels in the riser tubes. As shown in Fig. 5-10, the
present SCTR uses an optical detector (i.e. CCD sensors and LED array) to measure
the fluid-level variations in the riser tubes. The optical detector works as follows: as
an opaque fluid level rises in the riser tube, the opaque fluid blocks the passage of the
light emitted by the LED. Accordingly, the number of the CCD sensors that receive

105
the light from the LED becomes smaller. Computer software records the changes in
the number of CCD sensors that receive the light from the LED. Since the number of
the CCD sensors that dont receive the light from the LED is directly proportional to
the fluid level, one can determine the fluid level. In other words, the instantaneous
fluid levels are recorded in the form of pixel numbers (i.e., CCD sensors) versus time
in a computer data file through an analog-to-digital dataacquisition system. The
fluid level data from the two riser tubes were analyzed to determine the viscosity of
the fluid.
Therefore, it is essential to have an opaque fluid for the present SCTR
operation so that the light from the LED can be blocked by the opaque fluid as the
fluid level increases, and vice versa. Of course, one can use a laser light so that a
transparent fluid can be used as demonstrated by Kim et. al. (2000b). However, the
cost of a SCTR using such a laser-based system became prohibitively expensive,
making such a system economically unattractive.
In order to use the SCTR using CCD-LED arrangements for the viscosity
measurement of a transparent fluid, one may add dye to the fluid in order to make the
fluid opaque. However, the addition of a dye to a transparent fluid may alter the
viscosity of the fluid. Furthermore, the addition of the dye may make a transparent
Newtonian fluid such as water a non-Newtonian fluid if the concentration of the dye
is sufficiently large [Kim and Cho, 2002].
Therefore, the objective of the study is to investigate the effect of dye
concentration on the viscosity of distilled water in the SCTR. More specifically, the

106
present study plans to determine the maximum concentration of dye below which the
viscosity of the dye-water solution is not altered.

5.6.2. Experimental Method

Although distilled water is a Newtonian fluid, the aqueous solution of dyewater may exhibit the non-Newtonian characteristics for a sufficiently large dye
concentration. Thus, in order to investigate the viscosity characteristics of a dyewater solution, the present study used a non-Newtonian model to reduce experimental
data. In the previous chapters, various non-Newtonian models have been introduced
for the determination of blood viscosity with the SCTR, which include power-law
model, Casson model, and Herschel-Bulkley (H-B) model.
However, it is not very convenient to use the Casson model when a fluid
shows only shear-thinning characteristics without yield stress. The H-B model is
reduced to a power-law model for the case of fluids with no yield stress. Thus, in the
present study, a power-law model was used for the viscosity analysis of the dye-water
solution. The procedure of data reduction with power-law model will be discussed in
Chapter 6. Therefore, only the experimental results will be provided and discussed in
the next section.

107

5.6.3. Results and Discussion

In this study, six different concentrations (0.5, 1, 2, 3, 4, and 7% by volume)


of dye were used for the viscosity measurement of dye-water solution at 25. The
dye in a liquid form was purchased at a grocery store, which was a vegetable dye
produced from brand name, McCormick. For the validation of the method to reduce
data for the SCTR, the viscosities of dye-water solution with different dye
concentrations were compared with well-accepted reference data for water at 25
[Munson et al., 1998].
Figure 5-11 shows the variations of both power-law and consistency indices
of the dye-water solution for six different dye concentrations. Both indices were
determined through a curve-fitting method. Rectangular symbols indicate the powerlaw index whereas triangular symbols indicate the consistency index. As shown in
the figure, both indices started to vary when the amount of dye used became greater
than 2% by volume. When the dye concentration was less than 2%, the power-law
index was exactly one. The values of the consistency index for 0.5%, 1%, and 2% of
dye concentration cases were 0.890, 0.878, and 0.888, respectively. The distilled
water viscosity, which is given in the literature as a function of temperature, was
estimated to be 0.892 cP at 25 [Munson et al., 1998; Kim et al., 2002]. When the
dye concentration was greater than 2%, the power-law index decreased from n = 1 at
2% to n = 0.913 at 7%, whereas the consistency index increased from k = 0.89 at 2%
to k = 1.68 at 7%. The present results indicated that the effect of dye concentration

108
on the viscosity of the dye-water solution was negligibly small when the amount of
dye used was less than 2% by volume.
Figure 5-12 shows the viscosity data for the dye-water solution with six
different dye concentrations.

At a high shear-rate of 500 s-1, even with high

concentrations (i.e., 3, 4, 7%) of dye, the results showed that the effect of dye
concentration on the viscosity of water was very small. However, at low shear-rates
such as 1 and 10 s-1, the viscosity of the dye-water solution dramatically increased
with increasing dye concentration.
In the present experiment, the maximum concentration of dye, under which
the viscosity of the dye-water solution did not change, was approximately 2% by
volume. Compared with the reference data for water at 25 [Munson et al., 1998],
the test results obtained with 0.5%, 1%, and 2% of dye concentrations gave less than
2% error in the entire shear-rate range.

109

LED array

CCD 1

Riser tube 1

Riser tube 2
CCD 2

Computer
system
for data
collection

Test fluid

Capillary tube

Three-way stopcock

Fig. 5-10. Schematic diagram of a scanning capillary-tube rheometer (SCTR) system.

1.04

1.8

1.6
1.4

0.96

1.2
0.92

0.88

0.8

0.84

0.6
0

2 3 4 5
6 7
Dye concentration (%)

Consistency index, k
(cPs n-1 )

Power-law index, n

110

Fig. 5-11. Variations of both power-law index and consistency index of dye-water
solution due to the effects of dye concentrations.

111

2.5
at 500 1/s
at 100 1/s
at 10 1/s
at 1 1/s

Viscosity (cP)

2
1.5
1
0.5
0
0

Dye concentration (%)

Fig. 5-12. Viscosity data for dye-water solution with 6 different dye concentrations
at 25.

112
CHAPTER 6. EXPERIMENTAL STUDY WITH SCTR

Chapter 6 presents the results of viscosity and yield stress measurements with
the scanning capillary-tube rheometer (SCTR). Experimental tests were performed
with mineral oil, distilled water, bovine blood with 7.5% EDTA, and unadulterated
human blood.
Section 6.1 provides the viscosity results of both mineral oil and human blood
produced with the SCTR (with precision glass riser tubes) using the power-law model
for data reduction.
Section 6.2 gives the test results of distilled water, bovine blood, and human
blood obtained with the SCTR (with plastic riser tubes). Casson and HerschelBulkley models were used for data reduction to handle the yield stress of blood.
Section 6.3 reports the effect of the three models on the viscosity and yieldstress measurements of blood with the SCTR as well as on the flow patterns of blood
such as velocity profile and wall shear stress in a capillary tube.

6.1. Experiments with SCTR (with Precision Glass Riser Tubes)

The present study measured the viscosity of unadulterated blood at body


temperature, 37. Blood is a fluid consisting primarily of plasma and cells such as
erythrocytes, leukocytes and platelets.

Erythrocytes (i.e., red blood cells, RBC)

constitute the majority of the cellular content and account for almost one half of the

113
blood volume. The presence of such a high volume of red blood cells makes blood a
non-Newtonian fluid whose viscosity varies with shear rate. Whole blood viscosity
decreases

as

shear

rate

increases,

phenomenon

called

shear-thinning

characteristics. In other words, whole blood behavior may be described using a


power-law model, a Casson model, or a Herschel-Bulkley model. In the present
study, the power-law model was chosen for simplicity.
In order to demonstrate the validity of the scanning capillary-tube rheometer,
the viscosity data were compared with data obtained using a cone-and-plate rotating
viscometer. Since the rotating viscometer produces only 7 data points at relatively
high shear rates due to the torque requirement, the accuracy of the new scanning
rheometer at a low shear rate range was demonstrated by comparison with the
viscosity of a standard-viscosity oil (a Newtonian fluid) from Cannon Instrument
Company (State Park, PA).

6.1.1. Description of Instrument

Figure 6-1 shows a schematic diagram of the scanning capillary-tube


rheometer, which consists of two charge-coupled devices (CCDs) that are positioned
vertically, two light emitting diodes (LEDs), two riser tubes and a capillary tube both
made of precision glass, two stopcocks, a transfer tube made of tygon, and a
computer acquisition system. The inside diameter of both the transfer and riser tubes

114
used in the present tests were 3 mm. The inside diameter and length of the capillary
tube were 0.797 mm and 100 mm, respectively.
The essential feature in the scanning capillary-tube rheometer is the use of an
optical detector (i.e., CCD sensors and LED array) to measure the fluid level
variations in the riser tubes, h1 (t ) and h2 (t ) , every 0.02 s. The instantaneous fluid

levels were recorded in a computer data file through an analog-to-digital data


acquisition system in the form of pixel numbers vs. time. Since 12 pixels are equal to
1 mm, one could determine the actual height changes in the riser tube with an
accuracy of 0.083 mm.

6.1.2. Testing Procedure

Typical tests are conducted as follows: The system was turned on and
connected to a computer.

The software on the computer was executed, and

communication with the viscosity measurement system was properly established. At


that point, the computer was ready to acquire data from the CCD sensors in the
system. The experimental test run was initiated with a venipuncture on the patient
using a 19-gauge stainless steel needle. Fresh blood was first directed from the first
stopcock to the second stopcock to collect blood into the syringe. About 5 ml of
blood was collected in the syringe for tests with a cone-and-plate viscometer
(Brookfield DV-III) and hematocrit measurements, and the syringe was then removed
from the system. Approximately 0.5 ml of this fresh blood from the syringe was

115
immediately transferred to the sample cup of the Brookfield rotating viscometer that
was maintained at a constant temperature of 37 by a water bath connected to the
cup.
The viscosity measurements with the rotating viscometer were completed
within approximately 1 minute from the time when the blood left the human body.
Blood clotting rapidly developed inside the cone-and-plate test section. The rate of
blood clotting with time critically depends on the thrombotic tendency of a particular
individuals blood. As soon as blood began to clot, the rotating viscometer flashed an
EEEE sign indicating an overloaded torque, and thus tests were stopped. This
usually happened within 2 minutes of the test. During the viscosity measurement
with the Brookfield rotating viscometer, hematocrit values were determined with a
microhematocrit centrifuge (International Clinical Centrifuge).
Immediately following the removal of the syringe, the experiment with the
scanning capillary-tube rheometer was continued with the second stopcock turned to
a position to allow blood flow to both the capillary tube and riser tube 2. When blood
reached a predetermined height of 300 pixels in the riser tube 2, the second stopcock
was shut to stop further blood flow into the riser tube 2, and the first stopcock was
then turned to direct blood flow into the riser tube 1 up to a height of 1000 pixels. At
t = 0, the data acquisition system was enabled, and both stopcocks were adjusted to

allow blood to flow from the riser tube 1 to tube 2 as driven by the gravity head. Of
note is that the initial pixel difference of 700 was chosen to produce the maximum
shear rate of approximately 400 s-1. If a higher shear rate is desired, an initial pixel
difference greater than 700 can easily be selected.

116
For the purpose of calibration, the present study used the scanning capillarytube rheometer to measure the viscosity of mineral oil, which had a standard
Newtonian viscosity of 9.9 cP at 25. In the tests with human blood and mineral oil,
the capillary tube and major portions of the transfer tube in which test fluids were
actually flowing through the capillary tube were placed in a water bath maintained at
37 and 25, respectively.

6.1.3. Data Reduction with Power-law Model

The mathematical procedure for data reduction using a power-law model was
discussed in Chapter 4. The least-square method was used for curve fitting of the
experimental data and Eq. (4-18) in order to determine the power-law index, n , and
the consistency index, m . A standard software package (Excel-Solver, Microsoft;
see Appendix E), which has a formula known as a Newtons method (see Appendix
F) [Microsoft Corporation, 57926-0694; Harris, 1998; John, 1998; Brown, 2001], was
used for iterations to determine the values of n and m that minimize the sum of error
(see Eq. (4-19)).
The analysis of data reduction for a mineral oil is introduced in Fig. 6-2.
Figure 6-2(a) shows both experimental values of (t ) and theoretical values of (t )
that were obtained with initial guesses of the two unknowns, whereas Figure 6-2(b)
shows the curve-fitting result after iterations to minimize the sum of error. The initial
guesses for n and m were 0.8 and 8 (cPsn-1), respectively, in the case of the mineral

117
oil (see Fig. 6-2(a)). The resulting values of n and m were determined to be 1 and
9.91 (cPsn-1), respectively, by the iterations using the Excel-Solver.
The analysis of data reduction for human blood is shown in Fig. 6-3. For the
human blood case, initial guesses of the two unknowns of n and m were 0.8 and 6
(cPsn-1), respectively (see Fig. 6-3(a)). After iterations to minimize the sum of error,
the unknowns, n and m , were determined to be 0.83 and 9.27 (cPsn-1), respectively
(see Fig. 6-3(b)). The initial guesses and resulting values of n , m , and ht = for
both mineral oil and human blood were reported in Table 6-1.

6.1.4. Results and Discussion

Both mineral oil and unadulterated human blood were used in the present
study. The former was specially ordered as a dyed viscosity-standard fluid (i.e., 9.9
cP at 25) from Cannon Instrument Company (State Park, PA), and the latter was
obtained from donors. For comparison purpose, the viscosity of the human blood was
also measured by using a cone-and-plate rotating viscometer (Brookfield model DVIII) at 37. The rotating viscometer used in the present study had an LV-type spring
torque with a CP-40 spindle. In order to maintain the preset temperatures, a water
bath (PolyScience model 2LS-M) was used, which controlled the temperature with an
accuracy of 0.1.
Figures 6-4 and 6-5 show test results obtained with the mineral oil at 25.
Figure 6-4 shows the fluid level variations in the riser tubes, h1 (t ) and h2 (t ) . Both

118
fluid levels converge gradually from the initial fluid level difference and eventually
reach an equilibrium fluid level. In the case of mineral oil, 14.5 mm of an initial fluid
level difference was used to ensure that viscosity measurements at a low shear rate
range were accurate. It is of note that, for mineral oil, ht = was found to be zero.
Figure 6-5 shows viscosity results for the mineral oil at 25 obtained with
the SCTR. The power-law index of the mineral oil was determined to be 1.0 by a
computer program (Excel-solver), confirming that it was a Newtonian fluid. Based
on the present viscosity measurement method, the viscosity of the mineral oil was
found to be between 9.86 and 9.91 cP at 25, which was a 0.5% difference in the
whole range of shear rates from the standard viscosity of 9.9 cP at the same
temperature.
Figure 6-6 shows height variations in each riser tube as a function of time for
fresh human blood at 37. In the case of human blood, about 58 mm of initial fluid
level difference was used for the viscosity measurement so that one could obtain the
accurate viscosity of human blood over a wide shear rate range as low as 1 s-1. In
order to finish a test without using anticoagulants, the test should be completed within
3-4 minutes. Otherwise, blood may begin to clot. In the present study, one test run
took less than 2 minutes. For human blood, the trends of fluid level variations were
very similar to those for mineral oil. However, ht = for human blood was not zero
but a finite value, which depended on the individual donor. The minimum and
maximum values of ht = were found to be 3.86 mm and 6.22 mm, respectively,

119
among 8 donors. These values of ht = represent the thixotropic characteristics of
blood that result in the yield stress.
Figure 6-7 shows the viscosity of unadulterated human blood at 37, which
was measured with both the SCTR and the cone-and-plate rotating viscometer (RV).
Closed circle symbols indicate viscosity data measured with the SCTR while triangle
symbols indicate those measured with the RV. The viscosity of the unadulterated
human blood measured with the present SCTR was based on a calculation method
that determined the power-law index, n , and consistency index, m . In the case
shown in Fig. 5, the values of n and m were 0.828 (dimensionless) and 9.267

(cP s n -1 ) , respectively.
Compared with the measured data using the RV, the present test results from
the SCTR gave excellent agreement with those measured by the RV (i.e., less than
5% difference) in a shear rate range between 30 and 375 s-1. However, as the shear
rate decreased below 30 s-1, the RV was not recommended by Brookfield for the
measurement of blood viscosity. More specifically, the shear stress should vary from
a minimum of 10% to 100% of the full range of the torque sensor used in the
rotational type viscometer at a given shear rate for reasonably accurate viscosity
measurements [Brookfield, 1999]. Therefore, the minimum shear rate at which the
RV could be used for the viscosity measurement of human blood was 30 s-1.
Blood clotting in the RV was the other reason that one could not obtain more
than 7 data points. One could see the effects of blood clotting on viscosity as testing
time passed beyond 1 minute with the RV. Since only 0.5 ml was used for the RV
test, the blood contact area with the surface of the cone-and-plate was much bigger

120
than that in the case of the SCTR, a condition that might have caused rapid blood
clotting.
Figure 6-8 shows the viscosity of unadulterated human blood for two different
donors at 37, whose hematocrits were Hct = 41 and 46.5. Furthermore, human
blood from 8 donors was tested for viscosity measurements in the present study.
Every result using SCTR gave good agreement with that from the RV at high shear
rates but had a different slope with respect to shear rate individually. The viscosity
for the case with Hct = 46.5 was consistently greater than that for the case with Hct =
41. The difference between the two viscosity data was very small at high shear rates
greater than 300 s-1 whereas the difference was significant (i.e., greater than 200%) at
a low shear range, indicating the significance of low shear viscosity data.
In fact, it is well known that slip at the wall occurs in the flow of two-phase
systems because of the displacement of the disperse phase away from solid surfaces
[Barnes, 1995; Picart et al., 1998a, 1998b]. In the case of blood, a significant amount
of slip appears at low shear rates when the size of RBC (red blood cells) is relatively
large compared to wall roughness. For a smooth geometry like a glass tube, however,
the slip effect begins to be considerable from as low as 0.5 s-1 [Picart et al., 1998a].
Therefore, whole blood that was used in the present study did not show large slip
effects since the lowest shear rate data used was 1 s-1. In order to get reliable
viscosity data below 0.5 s-1, it may be necessary to use a rough surface capillary.

121

LED Array
CCD
Riser tube 2

CCD
Riser tube 1

First
stopcock

Computer
system
Transfer tube
Water bath

Blood from vein


Blood
Second
stopcock

Capillary tube

Fig. 6-1. Schematic diagram of the scanning capillary-tube rheometer with precision
glass riser tubes.

122

16
Experimental data

12

Theoretical data

(mm) 8

n = 0.8
m = 8 (cPsn-1)

4
0
0

20

40

60

80

100

120

Time (s)

(a) With initial guess values

16
Experimental data

12

Theoretical data

(mm) 8

n =1
m = 9.91 (cPsn-1)

4
0
0

20

40

60

80

100

120

Time (s)

(b) With final resulting values

Fig. 6-2. Curve-fitting procedure with power-law model for mineral oil.

123

60
Experimental data

40

Theoretical data

(mm)
n = 0.8
m = 6 (cPsn-1)

20

0
0

20

40

60

80

100

120

Time (s)

(a) With initial guess values

60
Experimental data

Theoretical data

40

(mm)

n = 0.83
m = 9.27 (cPsn-1)

20

0
0

20

40

60

80

100

120

Time (s)
(b) With final resulting values

Fig. 6-3. Curve-fitting procedure with power-law model for human blood.

124

Table. 6-1. Comparison of initial guess and resulting value using power-law model.

Distilled Water

Human Blood

Initial Guess

n = 0.8
m = 8 (cPsn-1)
ht = = 0.03 mm

n = 0.8
m = 6 (cPsn-1)
ht = = 3 mm

Resulting Value

n =1
m = 9.91 (cPsn-1)
ht = = 0.0271 mm

n = 0.83
m = 9.27 (cPsn-1)
ht = = 3.86 mm

125

60

Height (mm)

Riser tube 1
40
Riser tube 2
20

0
0

50

100

150

Time (s)

Fig. 6-4. Height variation in each riser tube vs. time for mineral oil (9.9 cP viscositystandard oil).

126

Viscosity (cP)

12
10

SCTR
8
6
0

20

40

60

Shear rate (s -1)

Fig. 6-5. Viscosity measurement for mineral oil at 25 with a scanning capillarytube rheometer (SCTR).

Height (mm)

127

100
90
80
70
60
50
40
30
20
10
0

Riser tube 1

Riser tube 2

50

100

150

Time (s)
Fig. 6-6. Height variation in each riser tube vs. time for human blood at 37.

128

Viscosity (cP)

100

Hematocrit : 40.5

SCTV
RV

10

1
1

10

100

1000

Shear rate (s -1)

Fig. 6-7. Viscosity measurement (log-log scale) for human blood at 37 with
rotating viscometer (RV) and scanning capillary-tube rheometer (SCTR).

129

Viscosity (cP)

100
SCTV
RV

Hematocrit : 46.5
10

Hematocrit : 41
1
1

10

100

1000

Shear rate (s -1)

Fig. 6-8. Viscosity measurement (log-log scale) of unadulterated human blood at


37, measured with scanning capillary-tube rheometer (SCTR) and cone-and-plate
rotating viscometer (RV), for two different donors.

130

6.2. Experiments with SCTR (with Plastic Riser Tubes)

A new U-shaped scanning capillary-tube rheometer (SCTR) has been


developed from the concept of a conventional capillary-tube viscometer. In general,
the capillary-tube viscometer is an attractive technique for several reasons. The basic
instrument is relatively inexpensive, easy to construct and simple to use
experimentally. Temperature control is relatively easy. The flow in a capillary tube
most closely simulates the blood flow in physiological conditions compared with the
flows in rotating viscometers and falling-object viscometers. Most of all, since the
capillary-tube viscometer uses the gravity as the driving force, it does not require
periodic calibration of any components. However, significance of end effect, wall
effect, and surface-tension effect should be carefully considered.
The end and wall effects can be made negligibly small by selecting
appropriate dimensions for the capillary and riser tubes. However, unlike the case of
the SCTR with precision glass riser tubes, the effect of the surface tension is a unique
and critical factor in using U-shaped disposable tube sets.

Since inexpensive

disposable capillary-riser tube sets should be used for clinical applications, it may not
be easy to strictly control the surface quality of riser tubes within a certain limit.
Therefore, the effects of the surface tension at the riser tubes as well as the properties
of a testing fluid had to be considered in the viscosity and yield stress measurements
with the SCTR.
The resistance associated with air-liquid interfaces in plastic riser tubes of a
small diameter in the SCTR can be a significant part of the pressure head applied to

131
the SCTR, particularly at low shear rates [Jacobs, 1966; Mardles, 1969; Einfeldt and
Schmelzer, 1982]. The meniscus resistance depends on the surface tension of the
fluid-air interface and on the reciprocal of the internal radius of the riser tube.
Throughout the development of a U-shaped scanning capillary-tube rheometer
concept, the focus has been on how to isolate the effects of surface tension and yield
stress in obtaining low-shear-rate viscosity for non-Newtonian fluids like blood. This
study attempted to measure the viscosity of unadulterated blood at a body temperature
of 37.

In the present study, both Casson and Herschel-Bulkley models were

selected for the viscosity and yield stress measurements of blood since both models
have a yield stress term.

6.2.1. Description of Instrument

Figure 6-9 shows a photograph of the SCTR with plastic riser tubes, which
consists of two charge-coupled devices (CCD 1 and CCD 2) that are positioned
vertically, two light-emitting diodes (LEDs), two riser tubes made of acrylic plastic
and a capillary tube made of glass, a stopcock, and a data-acquisition system. The
inside diameter of the riser tubes used in the study was 3.2 mm. The inside diameter
and length of the capillary tube were 0.797 mm and 100 mm, respectively. The small
diameter of the capillary tube, compared with that of the riser tubes, was chosen to
ensure that the pressure drop at the capillary tube was significantly greater than those
at the riser tubes and connecting fittings.

132

6.2.2. Testing Procedure

Tests with distilled water and bovine blood were performed at the room
temperature of 25.

Riser tube 1 was first filled with the test fluid to the

predetermined height of approximately 550 pixels by using a syringe. Once the


desired level was reached, the stopcock was turned to a position to allow test fluid to
flow to the capillary tube. When the fluid reached the predetermined height of 100
pixels in riser tube 2, the stopcock was shut to stop further fluid movement into riser
tube 2. Next, the syringe was removed from the SCTR system, and then the stopcock
was turned to allow fluid to move from riser tube 1 to riser tube 2 by gravity. As
shown in Fig. 6-9, two CCD sensors were used: one located at the lower (left) side
and the other located at the upper (right) side. The difference between the locations
of the two CCD sensors was 500 pixels, a number that was taken into account in the
height measurements.
For the purpose of comparison, approximately 0.5 ml of bovine blood from
the syringe was immediately transferred to a cone-and-plate test cup of a rotating
viscometer (Brookfield model DV-III with LV-type spring torque with CP-40
spindle) that was maintained at a constant temperature of 25 by a water bath
connected to the cup. For unadulterated human blood, the temperature of the SCTR
was controlled during the test at a body temperature of 37 by using preheated
disposable tube sets and a heating pad (shown in Fig. 6-10) installed inside the SCTR.
After the temperature in the SCTR was stabilized, the viscosity measurement was
initiated with a venipuncture using a 19-gauge stainless steel needle. About 5 ml of

133
extra blood was collected for hematocrit measurements, which were determined with
a microhematocrit centrifuge (International Clinical Centrifuge) for both bovine
blood and human blood.
Viscosities of distilled water (a Newtonian fluid), bovine blood containing
7.5% EDTA, and unadulterated human blood were measured over a range of shear
rates. Because the CCD sensor requires an opaque fluid, dye was added to the
distilled water.

The amount of dye used for distilled water was less than 1%

concentration by volume, and the dye-effect on the viscosity of distilled water was
negligibly small at this concentration. Bovine blood was purchased from Lampire
Biological Laboratories, Inc., and human blood was obtained from two healthy male
donors who were 29 and 51 years of age. For comparison purposes, a reference value
was used for the distilled water while the viscosity of the bovine blood was
independently checked by using the cone-and-plate rotating viscometer.

6.2.3. Data Reduction with Casson Model

The fluid level data from the two riser tubes were analyzed to determine the
viscosities of distilled water (a Newtonian fluid) and blood (a non-Newtonian fluid).
In order to measure blood viscosity using a U-shaped SCTR, one needs to isolate the
effects of both surface tension and yield stress on the viscosity of blood. The details
of mathematical procedure for curve-fitting using the Casson model have already

134
been introduced in Chapter 4. Thus, in this section, the procedure to determine the
unknown values for the Casson model is discussed.

6.2.3.1. Curve Fitting

As discussed in Chapter 4, there are three unknown values, i.e., k , hst , and

h y , to be determined through the iterations using the same software package (Excel
Solver; see Appendix E) used in the power-law model case. The least-square method
was used for curve fitting of the experimental data and Eq. (4-41) to obtain the three
unknowns involved in the Casson model.
The procedure of data reduction for distilled water is shown in Fig. 6-11.
Figure 6-11(a) shows mean-velocity variations at a riser tube which were obtained
experimentally and theoretically. In the case of theoretical values, the initial guesses
for the three unknowns were used to estimate the values. In Fig. 6-11(b), the curvefitting results after iterations to minimize the sum of error that were calculated by Eq.
(4-42) are shown. The initial guesses and final values of k , hst , and h y for
distilled water are shown in Table 6-2.
Figures 6-12 and 6-13 show the curve-fitting procedures for human blood
obtained from two donors. As shown in Table 6-2, the same initial guesses of the
three unknowns were used for the two different bloods. However, the resulting
values of the three unknowns in the Casson model for the two donors were very
different, validating the present curve-fitting method.

135

6.2.3.2. Results and Discussion

Figures 6-14 and 6-15 show test results obtained with distilled water at 25.
Figure 6-14 shows the fluid-level variations in the two riser tubes, h1 (t ) and h2 (t ) .
Both fluid levels converged gradually from the initial difference to an equilibrium
state. Even for the distilled water, ht = was not zero due to the difference in the
surface tension between the two riser tubes. Unless the wetting conditions of the
liquid-solid interface at the two riser tubes are exactly same, the surface-tension
difference always exists.
Figure 6-15 shows viscosity results from two tests for distilled water at 25
obtained with the SCTR, together with the reference data for comparison. The values
of hst were determined to be approximately 4 mm for both tests whereas the values
of h y were determined to be zero by a computer program (Microsoft Excel-solver),
validating the data reduction procedure involving the yield stress. Based on this
viscosity measurement method, the viscosity of the distilled water was found to be
between 0.876 and 0.878 cP at 25. The solid line indicates the viscosity of the
distilled water calculated by using the so-called Andrades equation. Since the water
viscosity data are given in the literature as a function of temperature, the exact
viscosity of water at 25 was calculated using the Andrades equation [Munson et al.,
1998]:

= De

(6-1)

136
where D and B are given as 9.93 10 4 mPas and 2026.57 K, respectively, for water
in a temperature range between 20 and 30.
Based on Eq. (6-1), the water viscosity was estimated to be 0.892 cP at 25.
The test results obtained with the SCTR gave less than 2% error in the entire shear
rate range, validating the test methods and data reduction procedure. Thus, it is
concluded that it is extremely important to consider the effect of the surface tension
on the viscosity measurement using a gravity driven, U-shaped capillary-tube system.
Figure 6-16 shows height variations at the two riser tubes as a function of time
for bovine blood with 7.5 % EDTA at a room temperature of 25. The trends of
fluid-level variations for the bovine blood were very similar to those for the distilled
water. As expected, ht = for the bovine blood was not zero but a finite value that is
slightly greater than that for the distilled water. The height difference due to surface
tension, hst , and the height difference due to yield stress, h y , were determined to
be in the range of 5.7-6.1 mm and 0.52-0.59 mm, respectively. The hematocrit of the
bovine blood was measured to be 35 percent.
Figure 6-17 shows the viscosity of the bovine blood with 7.5% EDTA at 25,
which was measured with both the SCTR (indicated by circles and triangles) and the
rotating viscometer (RV; indicated by diamonds). Compared with the measured data
using the RV, the test results from the SCTR gave excellent agreement within 3% in a
shear rate range between 15 and 300 s-1. However, as the shear rate decreased below
15 s-1, the viscosity data measured from the RV seemed to be incorrect. More
specifically, the torque for viscosity measurements with the RV should be greater
than 10% of the full scale (as suggested by Brookfield) at a given shear rate for

137
reasonably accurate viscosity measurements. In the case of the bovine blood, the
minimum shear rate that Brookfield recommended, based on the 10% criterion, was
approximately 30 s-1. In contrast, the SCTR gave a consistent viscosity measurement
over a range of shear rates as low as a shear rate of 1 s-1. Related to the uncertainty in
measuring blood viscosity at low shear rates, wall slip is an issue.

Since the

minimum shear rate in this study was 1 s-1, as discussed earlier, one could assume that
the slip effect was negligibly small [Picart et al., 1999a].
Figures 6-18 and 6-19 show test results of fresh, unadulterated human blood at
a body temperature of 37. The test was completed within 2-3 min to avoid blood
clotting, which might have altered the viscosity of the blood. Figure 6-18 shows
height variations in the two riser tubes as a function of time for the fresh human blood
at 37. The value of ht = for the fresh human blood had a finite value, which
depended on individual donors. The values of hst for donors 1 and 2 were found to
be approximately 8.5 and 9.6 mm, respectively.
In the use of the SCTR, two phenomena of particular importance should be
pointed out in regards to clinical hemorheology: one is the carry-over effect and the
other is the surface tension effect. First, at the completion of a measurement, a thin
layer of the test blood sample was always retained on the tube wall, unless the tube
was very carefully washed and dried. This residual layer can be called carry-over.
Hence, for unadulterated blood-viscosity measurements, it was necessary to use
disposable capillary-riser tube sets to avoid the carry-over phenomenon. Second, the
difference between surface tensions at the two riser tubes may vary from one

138
disposable set to another. Although the riser tubes were made of the same material
(acrylic plastic), the surface tensions were slightly different from set to set.
The values of h y representing the effect of the yield stress for donors 1 and 2
were found to be 0.7 and 0.2 mm, respectively. These results were consistent with
hematocrit data for donors 1 and 2, which were 42 and 35 percent, respectively.
Donor 2 (a physician) practices therapeutic bloodletting periodically, which explains
why the hematocrit of donor 2 was unusually low. It is of note that the value of h y
represents the thixotropic characteristics of fresh human blood that is closely related
to the yield stress.
A thixotropic blood exhibits a high viscosity when first sheared from rest.
The viscosity continues to decrease as shearing continues. Thixotropy is usually a
result of the partial destruction, by shearing, of the internal liquid structure. While at
rest, the internal structure made of suspended cells and plasma may form to create
aggregations of RBCs, for example. This phenomenon is generally referred to as
structure viscosity.
In fact, the phenomenon of RBCs aggregations at low shear rates is well
known but not well understood so far. The forces leading to aggregations are weak,
so if a sample of normal blood is subjected to increasing shear rate, the aggregates
progressively break up and are generally monodispersed at a shear rate greater than
10 s-1. However, it is important to note that there could be two kinds of yield stresses:
a start-up yield stress and a stopping yield stress [Cho and Choi, 1993]. The yield
stress determined in this study was the stopping yield stress, an important
phenomenon in clinical hemorheology and studies of cardiovascular disease. In the

139
viscosity measurement with whole blood, a typical test duration in which the shear
rate decreased from 10 to 1 s-1 was longer than 60 seconds. It is reasonable to assume
that the 60-second period is long enough to cause aggregations if the aggregations
were going to take place.
Figure 6-19 shows the viscosities of the two donors at 37 measured with the
SCTR. The viscosities for donor 2 were significantly lower than those for donor 1
due to the difference in hematocrit. In addition, the viscosity curve for donor 2 was
flatter than that for donor 1 since donor 2 had a smaller value of h y and a lower
hematocrit. Figure 6-20 shows shear stress variations against shear rate for both
donors. Like the case of viscosity, shear stress for donor 2 was significantly lower
than that for donor 1 due to differences in hematocrit and yield stress. The values of
the yield stress determined from Casson model were approximately 14 mPa and 5
mPa for donor 1 and 2, respectively. The difference between the viscosity data
increased as the shear rate decreased indicating that the viscosity was more influenced
by hemorheological parameters such as hematocrit and yield stress at low shear rates
than at high shear rates.

6.2.4. Data Reduction with Herschel-Bulkley (H-B) Model

The detailed mathematical procedure for curve-fitting using a HerschelBulkley (H-B) model was provided in Chapter 4. As in the case of the Casson model,
the H-B model can also handle the yield stress of blood. However, in the data

140
reduction with the H-B model, there are four unknown values to be determined
through the curve-fitting technique, whereas the Casson model has only three
unknowns.
The four unknown values, i.e., m , n , hst , and h y , are to be determined
through a least-square method using the same software package (Excel-Solver,
Microsoft; see Appendix E) that uses the formula of Newtons method. In the
process of curve fitting, Eq. (4-59) was used for theoretical values. In the case of the
H-B model, the experimental data of bovine blood with 7.5% EDTA were used to
validate the method of data reduction.
Figure 6-21(a) shows mean-velocity values at a riser tube which were
obtained both experimentally and theoretically. In the case of theoretical values, the
initial guesses for the four unknowns, which are shown in Table 6-3, were used. In
Fig. 6-21(b), the curve-fitting results after the iterations to determine the four
unknowns are shown. The initial guesses and final values of m , n , hst , and h y
for bovine blood with 7.5% EDTA using the H-B model are shown in Table 6-3.
Figure 6-22 shows the viscosity of the bovine blood with 7.5% EDTA at 25,
which was measured with the SCTR. Three consecutive tests were performed with
bovine blood.

As expected, the H-B model also produced very accurate and

repeatable results. The final values of four unknowns for each test are reported in
Table 6-4.

The effects of constitutive models on the viscosity and yield stress

measurement of blood will be further discussed in the next section.

141

CCD-LED arrays

Riser tube 1
Test Fluid

Stopcock

Riser tube 2

Capillary tube

Fig. 6-9. Picture of a SCTR with plastic riser tubes.

142

Rheometer
System

Heating
Pad

Fig. 6-10. Heating pad for a test with unadulterated human blood.

143

0.025
Experimental data
Theoretical data

0.02

2Vr 0.015
(m/s)
0.01
0.005
0
0

10

20

30

40

Time (s)

(a) With initial guess values


0.025
Experimental data

0.02

Theoretical data

2Vr
0.015
(m/s)
0.01
0.005
0
0

10

20

30

40

Time (s)

(b) With final resulting values

Fig. 6-11. Curve-fitting procedure with Casson model for distilled water.

144

Table. 6-2. Comparison of initial guess and resulting value using Casson model.

Initial Guess

Resulting Value

Distilled Water

Human Blood

k = 1 (cPs)
h y = 0.5 mm

k = 1 (cPs)
h y = 0.5 mm

hst = 3 mm

hst = 5 mm

k = 0.878 (cPs)
h y = 0.00
hst = 4.02 mm

Donor 1

Donor 2

k = 2.743 (cPs)
h y = 0.74 mm

k = 2.121 (cPs)
h y = 0.16 mm

hst = 8.47 mm

hst = 9.58 mm

145

0.008

2Vr
(m/s)

Experimental data
Theoretical data

0.006
0.004
0.002
0
0

20

40

60

80

100

120

Time (s)

(a) With initial guess values


0.008
Experimental data
Theoretical data

0.006
2Vr
(m/s)

0.004
0.002
0
0

20

40

60

80

100

120

Time (s)

(b) With final resulting values

Fig. 6-12. Curve-fitting procedure with Casson model for donor 1.

146

0.008
Experimental data
0.006
2Vr
(m/s)

Theoretical data

0.004
0.002
0
0

20

40

60

80

100

120

Time (s)
(a) With initial guess values

0.008
Experimental data
2Vr
(m/s)

Theoretical data

0.006
0.004
0.002
0
0

20

40

60

80

100

120

Time (s)
(b) With final resulting values

Fig. 6-13. Curve-fitting procedure with Casson model for donor 2.

147

0.1

Height (m)

0.08

Riser tube 1

0.06
0.04
Riser tube 2

0.02
0
0

50

100

150

Time (s)

Fig. 6-14. Height variation in each riser tube vs. time for distilled water at 25.

148

Viscosity (cP)

1.5
Test 1
Test 2
Reference

1.3
1.1
0.9
0.7
0.5
0

100

200
300
-1
Shear rate (s )

400

Fig. 6-15. Viscosity measurement for distilled water at 25. 1 cP = 1 mPas.

149

0.1
Riser tube 1

Height (m)

0.08
0.06
0.04

Riser tube 2

0.02
0
0

30

60
90
Time (s)

120

150

Fig. 6-16. Height variation in each riser tube vs. time for bovine blood with 7.5%
EDTA at 25.

150

Viscosity (cP)

100
Test 1
Test 2
RV
10

1
1

10
100
-1
Shear rate (s )

1000

Fig. 6-17. Viscosity measurement for bovine blood with 7.5% EDTA at 25 using
both rotating viscometer (RV) and scanning capillary-tube rheometer (SCTR).
Hematocrit was 35. 1 cP = 1 mPas.

Height (m)

151

0.1
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0

A : Donor 1
B : Donor 2
B
A

30

60
90
Time (s)

120

150

Fig. 6-18. Height variation in each riser tube vs. time for human blood at 37.

152

Viscosity (cP)

100

Donor 1
Donor 2
10

1
1

10

100

1000

-1

Shear rate (s )
Fig. 6-19. Viscosity measurement for human blood (2 different donors) at 37.
Hematocrits for donors 1 and 2 were 42 and 35, respectively. 1 cP = 1 mPas.

153

Shear stress (mPa)

10000
1000
100
10

Donor 1
Donor 2

1
1

10

100

1000

Shear rate (s -1)

Fig. 6-20. Shear-stress variation vs. shear rate for human blood (from 2 different
donors) at 37.

154

0.005
2Vr
(m/s)

Experimental data

0.004

Theoretical data

0.003
0.002
0.001
0
0

40

80

120

Time (s)
(a) With initial guess values

0.005
Experimental data

0.004

Theoretical data

2Vr
(m/s) 0.003

0.002
0.001
0
0

40

80

120

Time (s)
(b) With final resulting values

Fig. 6-21. Curve-fitting procedure with Herschel-Bulkley model for bovine blood.

155

Table. 6-3. Comparison of initial guess and resulting value


using Herschel-Bulkley model.
Bovine Blood

Initial Guess

n = 0.8
m = 5 (cPsn-1)
h y = 1 mm

hst = 6 mm

Resulting Value

n = 0.875
m = 8.6 (cPsn-1)
h y = 1.2 mm

hst = 5.8 mm

156

Viscosity (cP)

100
Test #1
Test #2
Test #3
10

1
1

10

100

1000

Shear rate (s-1)


Fig. 6-22. Viscosity measurements of the bovine blood with 7.5% EDTA at 25,
which were analyzed with Herschel-Bulkley model.

157

Table. 6-4. Comparison of four unknowns determined with Herschel-Bulkley model


for three consecutive tests.
Test #1

Test #2

Test #3

n = 0.875
m = 8.6 (cPsn-1)
h y = 1.2 mm

n = 0.876
m = 8.7 (cPsn-1)
h y = 1.2 mm

n = 0.872
m = 8.76 (cPsn-1)
h y = 1.24 mm

hst = 5.8 mm

hst = 5.8 mm

hst = 5.8 mm

158

6.3. Comparison of Non-Newtonian Constitutive Models

It is well known that blood has both shear-thinning (pseudoplastic)


characteristics and yield stress. The present study examined the capability of three
constitutive models in handling data from the scanning capillary-tube rheometer
(SCTR) for the viscosity and yield-stress measurements of blood. Power-law model
was chosen for the shear-thinning behavior of the blood. Casson and HerschelBulkley (H-B) models were selected to measure both the shear-thinning viscosity and
yield stress of the blood.
As shown in Table 6-5, many researchers used non-Newtonian constitutive
models for the investigations on blood rheology and flows which include power-law,
Casson, and H-B models. In addition, it has been pointed out that the selection of a
constitutive model could be very significant in analyzing blood flows [Tu and Deville,
1996; Siauw et al., 2000]. Siauw et al. (2000) performed a comparative study of nonNewtonian models for the prediction of unsteady stenosis flows by using power-law
and Casson models, whereas Tu and Deville (1996) used H-B, Bingham, and powerlaw models for their study on stenosis flows.
The objective of the present study was to investigate whether or not the results
of blood rheology and flow in the SCTR could be significantly altered by constitutive
models. Hence, the study investigated the effect of the three models on the viscosity
and yield-stress measurements of blood using the SCTR as well as on the flow
patterns of blood such as velocity profile and wall shear stress in a blood vessel.

159

6.3.1. Comparison of Viscosity Results

Figures 6-23(a) and 6-23(b) show the calibration results obtained in the SCTR
with distilled water at 25. Figure 6-23(a) shows the fluid-level variations in the
two riser tubes, h1 (t ) and h2 (t ) . Both fluid levels converged gradually from the
initial difference to an equilibrium state. Even for the distilled water, the difference
in the two fluid-levels at t = , ht = , was not zero due to the difference in the
surface tension between the two riser tubes.
Figure 6-23(b) shows viscosity results for distilled water at 25 obtained
with the SCTR using the three different non-Newtonian constitutive models, together
with the reference data for comparison. The values of hst were determined to be
approximately 3.3-3.4 mm for the three models, whereas the values of h y were
determined to be zero for all models, validating the data reduction procedure
involving the yield stress. The viscosity of distilled water was found to be between
0.884 and 0.905 cP at 25 for all three models. As shown in Table 6-6, the test
results obtained with the SCTR gave less than 2% error over the entire shear rate
range, validating the experimental procedure and data reduction method using the
three non-Newtonian constitutive models. Furthermore, the viscosity measurement of
water confirmed that one needs to consider the effect of the surface tension on the
viscosity measurement in the SCTR.
Figures 6-24(a) and 6-24(b) show test results obtained with bovine blood at
25. Figure 6-24(a) shows height variations at two riser tubes as a function of time

160
for the bovine blood with 7.5% EDTA as an anticoagulant. The trends of fluid-level
variations for the bovine blood were very similar to those for the distilled water. As
expected, ht = for the bovine blood was not zero, but a finite value that is slightly
greater than that for the distilled water. The hematocrit of the bovine blood was
measured to be 35%.
Figure 6-24(b) shows the viscosity of the bovine blood at 25, which was
measured with both the SCTR and a rotating viscometer (RV). The SCTR results
obtained by using the three constitutive models showed very good agreement among
themselves at shear rates higher than approximately 30 s-1. However, the power-law
model maintained the rate of the viscosity change (the slope of a viscosity curve)
constant whereas both the Casson and H-B models increased the rate of viscosity
change as shear rate decreased, indicating the existence of yield stress in the bovine
blood. Moreover, the viscosity results using the Casson model showed very good
agreement with those using the H-B model.
The test results from the SCTR using the three non-Newtonian models gave
excellent agreement with the measured data using the RV within 5% in a shear-rate
range between 30 and 300 s-1. However, as the shear rate decreased below 30 s-1, the
viscosity data measured from the RV seemed to be incorrect. More specifically, the
torque for viscosity measurements with the RV should be greater than 10% of the full
scale (as suggested by Brookfield) for reasonably accurate viscosity measurements.
In the case of the bovine blood, the minimum shear rate that Brookfield
recommended, based on the 10% criterion, was approximately 30 s-1. In contrast, the
SCTR gave a consistent viscosity measurement over a range of shear rate as low as a

161
shear-rate of 1 s-1. In addition, the viscosities measured using both the Casson and HB models in the SCTR seemed to be more accurate than the power-law viscosity at
low shear rates. Table 6-7 provides the viscosity results at the several shear rates,
which were produced by using the three non-Newtonian constitutive models.
Figures 6-25(a) and 6-25(b) show test results of fresh, unadulterated human
blood at a body temperature of 37. The test was completed within 2-3 minutes to
avoid blood clotting, which might have altered the viscosity of the blood. Figure 625(a) shows height variations in the two riser tubes as a function of time for the fresh
human blood at 37. The value of ht = for the fresh human blood was measured to
be approximately 8.5 mm. The hematocrit of this donors blood was 42%.
Figure 6-25(b) shows the viscosity of the human blood at a body temperature
of 37, which was measured with the SCTR. The SCTR results obtained by using
the three constitutive models showed very good agreement among themselves at
shear rates higher than approximately 20 s-1. However, like the bovine blood case,
the power-law model maintained the rate of the viscosity change constant whereas
both the Casson and H-B models increased the rate of viscosity change as shear rate
decreased, indicating the existence of yield stress in the human blood. Table 6-8
shows the viscosity results of the unadulterated human blood for the three constitutive
models.
For cases of the bovine blood and fresh human blood, the power-law model
maintained constant slopes in the viscosity curve, while both the Casson and H-B
models showed rapid increases in the viscosity as the shear rate decreased. However,

162
the viscosities obtained from the three different constitutive models gave good
agreement at a high shear rate zone.

6.3.2. Comparison of Yield Stress Results

Over the years, many researchers have reported on the yield stress of blood.
The measurements of the rheological properties of fluids having yield stress are
summarized by Nguyen et al. (1983). According to their classifications, the yieldstress measurement with the SCTR can be classified as indirect methods rather than

direct methods since the yield stress of a fluid can be obtained in the SCTR by using
constitutive models. Bingham plastic, Casson, and H-B models were used in their
study to describe the rheological behavior of yield-stress fluids.
Figures 6-26(a), 6-26(b), and 6-26(c) show the results of shear stress versus
shear rate for distilled water, bovine blood, and human blood, respectively. In the
case of the distilled water, the shear stresses obtained using the three different nonNewtonian models were almost identical.

The yield stress can be graphically

described as the intersecting point where the shear stress-shear rate curve meets with
the y-axis (i.e., at zero shear rate). Table 6-9 shows the yield-stress results together
with the model constants of the three constitutive models. As expected, the distilled
water shows no yield stress, while both the bovine and human bloods show finite
values of the yield stress for the cases of Casson and H-B models. Note that the yield

163
stress values for the Casson and H-B models were calculated using the following
equation:

y =

gh y Rc
2 Lc

(6-2)

The yield stress of the human blood was consistently greater than that of the
bovine blood although the human blood was tested at a high temperature of 37. It
might be due to both the difference in hematocrit and the RBC aggregations of the
human blood at low shear rates. The yield stress values of the human blood with
hematocrit of 42% were measured to be 13.8 and 17.5 mPa for the Casson and H-B
models, respectively. Note that the yield stress measured in the present study was the
stopping yield stress, an important phenomenon in clinical hemorheology and
treatments of cardiovascular disease. The yield stress values vary from 1 mPa to 30
mPa for normal human blood with hematocrit of 40% [Chen et al., 1991; Picart et al.,
1999a], supporting the validity of the method of the present yield-stress measurement
using the SCTR.
The yield stress obtained with the H-B model was consistently greater than
that obtained with the Casson model in the cases of both bovine and human bloods as
shown in Table 6-9. In order to evaluate which model produces more accurate yield
stress results, the experimentally measured values of ht = were compared with those
of hst + h y determined analytically through the curve-fitting procedure for the
bovine and human bloods (see Table 6-10). The value of the fluid-level difference in
riser tubes 1 and 2 at a time of 180 seconds was taken as a measure of ht = . Hence,

164
the experimental values (i.e., ht = ) should be bigger than those (i.e., hst + h y ) to
be obtained analytically.
As shown in Table 6-10, the values of hst + h y obtained with the Casson
model were consistently smaller than the experimentally measured values while the
values obtained with H-B model were bigger. Based on the comparison, one may
conclude that the Casson model does a better job in determining the yield stress of
blood than the H-B model. However, it is of note that both models produced almost
identical values of hst , thus almost identical surface tensions, for the bovine and
human blood.

6.3.3. Effect of Yield Stress on Flow Patterns

Figure 6-27 shows the variations of C y (t ) for the bovine blood with 7.5%
EDTA, indicating that the plug-flow region grows at the capillary tube with
increasing time. Due to the difference in yield stress values for the Casson and H-B
models, the size of the plug-flow region estimated from the two models start to differ
after approximately 30 seconds. Note that the H-B model predicts a much larger
plug-flow region than the Casson model.
Figures 6-28(a), 6-28(b), and 6-28(c) show velocity profiles at the capillary
tube for the bovine blood at room temperature of 25, which were plotted at three
mean velocities of 3, 0.3, and 0.03 cm/s, respectively. At a relatively high velocity of
3 cm/s, the three constitutive models predicted identical velocity profiles. However,

165
as the mean velocity (i.e., shear rate) decreased, the clear deviation among the three
models started to appear near the center of the tube at 0.3 cm/s and became bigger at
0.03 cm/s, a phenomenon which can be attributed to the difference in yield stress
values. Therefore, it can be concluded that the yield stress plays an important role in
the determination of both the blood viscosity and velocity profiles in a blood flow.
Figure 6-28(c) also shows that the size of the plug-flow region at the center of the
tube for the H-B model is much larger than that for the Casson model at
approximately 0.03 cm/s.
The shear rate, viscosity, and shear stress obtained with both the Casson and
H-B models for the bovine blood were plotted as a function of the mean velocity at
the capillary tube in Figs. 6-29(a), 6-29(b), and 6-29(c), respectively. Wall shear
stress represents the friction exerted on the vessel wall by moving blood. It has been
shown in a number of studies that wall shear stress may play an important role in
endothelial cell morphology and functions influencing the production of substances
such as nitric oxide, prostacyclin, and endothelin [Baldwin and Thurston, 1995;
Usami et al., 1995; Fung, 1996; Mitsumata et al., 1996; Samijo et al., 1998; Frame et
al., 1998; Cotran et al., 1999; Kensey and Cho, 2001].
As shown in Fig. 6-29(c), the value of the wall shear stress is almost
independent of the selection of a constitutive model. Due to the difference in the size
of the plug-flow regions, the wall shear rates for the Casson model at low mean
velocities were much lower than those for the H-B model, resulting in consistently
higher wall viscosity for the Casson model as shown in Fig. 6-29(b). Since the wall
shear stress can be calculated from the product of viscosity (shown in Fig. 6-29(b))

166
and wall shear rate (shown in Fig. 6-29(a)) at a given mean velocity, the difference in
the wall shear stress between the two models is very small.

167

Table. 6-5. Various physiological studies with non-Newtonian constitutive models.


Researchers

Power-law

Chakravarty and Datta (1992)

Siauw et al. (2000)

Tu and Deville (1996)

Liepsch and Moravec (1984)

Walburn and Schneck (1976)

Casson

Herschel-Bulkley
x

x
x

Rohlf and Tenti (2001)

Misra et al. (2000, 2002)

Das and Batra (1995)

Dash et al. (1996)

Walawender et al. (1975)

Rodkiewicz et al. (1990)

Misra and Kar (1991)

Chakravarty and Datta


(1989, 1992)

168

100

(a)

Height (mm)

80
Riser tube1

60
40

Riser tube 2

20
0
0

30

60

90

120

150

Time (s)
1.5

(b)

power-law
H-B
Casson
Reference

Viscosity (cP)

1.3
1.1
0.9
0.7
0.5
0

100

200

300

400

Shear rate (s -1)

Fig. 6-23. Test with distilled water at 25. (a) Fluid-level variations in two riser
tubes. (b) Viscosity results.

169

Table. 6-6. Measurements of water viscosity.


Model

Viscosity (cP)

*Error

Power-law

0.905

1.46%

Herschel-Bulkley (H-B)

0.884

0.90%

Casson

0.886

0.68%

*Error is based on the comparison with the reference value


(0.892 cP at 25)

170

100

(a)

Height (mm)

80

Riser tube 1

60
40
Riser tube 2

20
0
0

30

60

90

120

150

Time (s)
100

Viscosity (cP)

(b)

power-law
H-B
Casson
RV

10

1
1

10

100

1000

-1

Shear rate (s )
Fig. 6-24. Test with bovine blood at 25. (a) Height variations in riser tube vs. time
for bovine blood with 7.5% EDTA. (b) Viscosity measurement for bovine blood with
7.5% EDTA using a rotating viscometer (RV) and a scanning capillary-tube
rheometer (SCTR).

171

Table. 6-7. Measurements of bovine blood viscosity.


Shear rate
(s-1)

Viscosity
(cP)
from RV

Power-law

Casson

H-B

300

4.43

4.39

4.49

4.28

150

4.78

4.75

4.84

4.71

90

5.11

5.03

5.18

5.09

45

5.75

5.44

5.85

5.71

30

6.25

5.7

6.38

6.2

15

8.81

6.16

7.7

7.21

7.5

17

6.67

9.7

8.9

7.4

14.5

12.8

Lower than 3

8.38
(at 1 s-1)

22.5
(at 1.35 s-1)

18.55
(at 1.55 s-1)

Viscosity (cP) from SCTR

172

(a)

100

Height (mm)

80

Riser tube 1

60
40
Riser tube 2

20
0
0

30

60

90

120

150

Time (s)
(b)

Viscosity (cP)

100
power-law
H-B
Casson
10

1
1

10
100
-1
Shear rate (s )

1000

Fig. 6-25. Test with unadulterated human blood at 37. (a) Height variations in riser
tubes vs. time. (b) Viscosity results.

173

Table. 6-8. Measurements of human blood viscosity.


Viscosity (cP) from SCTR

Shear rate (s-1)

Power-law

Casson

H-B

300

3.89

4.11

4.09

150

4.45

4.47

4.57

90

4.93

4.85

4.97

45

5.67

5.59

5.63

30

6.15

6.16

6.15

15

7.06

7.65

7.3

7.5

8.12

9.95

9.1

Lower than 5

9.76
(at 3 s-1)

14.73
(at 3.33 s-1)

12.9
(at 3.3 s-1)

Lower than 3

12.17
(at 1 s-1)

27.26
(at 1.18 s-1)

21.9
(at 1.32 s-1)

174
(a)

Shear stress (mPa)

1000
power-law
H-B
Casson

100
10
1
1

10

100

1000
Shear stress (mPa)

(b)

W ater

Bovine Blood
100
power-law
H-B
Casson

10
1
1

10

100

(c)

Shear stress (mPa)

1000
Human Blood

100
power-law
H-B
Casson

10
1
1

10
Shear rate (s -1)

100

Fig. 6-26. Wall shear stress at a capillary tube vs. shear rate. (a) for distilled water at
25. (b) for bovine blood at 25. (c) for human blood at 37.

175

Table. 6-9. Comparison of model constants, h y , and y . (Note that [m] = cPsn-1)
Power-law

H-B

n =1
m = 0.905
h y = 0

n =1
m = 0.884
h y = 0

y=0

y=0

Bovine blood
(25)

n = 0.8866
m = 8.3771
h y = 0

n = 0.8753
m = 8.599
h y = 0.8 mm

h y = 0.52 mm

y=0

y = 16.4 mPa

y = 10.7 mPa

Human blood
(37)

n = 0.7991
m = 12.171
h y = 0

n = 0.8601
m = 8.9721
h y = 0.85 mm

h y = 0.67 mm

y=0

y = 17.5 mPa

y = 13.8 mPa

Distilled water
(25)

Casson
k = 0.886 cP
h y = 0

y=0
k = 3.7302 cP

k = 3.2896 cP

176

Table. 6-10. Comparison of ht = and hst + h y .

Bovine blood
(25)

Human blood
(37)

H-B

Casson

6.5 mm

6.5 mm

6.6 mm
( hst = 5.8 mm

6.26 mm
( hst = 5.74 mm

h y = 0.8 mm)

h y = 0.52 mm)

ht =
(experimental)

9.13 mm

9.13 mm

hst + h y

9.25 mm
( hst = 8.4 mm

9.07 mm
( hst = 8.4 mm

h y = 0.85 mm)

h y = 0.67 mm)

ht =
(experimental)

hst + h y
(analytical)

(analytical)

177

1
Casson
H-B

0.8

Cy

0.6
0.4
0.2
0
0

50

100
Time (s)

150

200

Fig. 6-27. Variations of a plug-flow region at a capillary tube as a function of time for
bovine blood with 7.5% EDTA at 25.

178

6
3 cm/s
Velocity (cm/s)

(a)

4
power-law
Casson
H-B

2
0
0

0.01

0.02

0.03

0.04

(b)

Velocity (cm/s)

0.6
0.3 cm/s
0.4
power-law

0.2

Casson
H-B

0
0

0.01

0.02

0.03

0.04

(c)

Velocity (cm/s)

0.06
0.03 cm/s
0.04
power-law
Casson
H-B

0.02
0
0

0.01

0.02

0.03

0.04

Radius (cm)
Fig. 6-28. Velocity profiles at a capillary tube for the bovine blood with 7.5% EDTA
at 25: (a) at a mean velocity of 3 cm/s. (b) approximately 0.3 cm/s.
(c) approximately 0.03 cm/s.

179

-1

Wall shear rate (s )

(a)

1000
100
10
Casson
H-B

1
0.1
0.01

0.1

10

(b)

Viscosity (cP)

1000
Casson
H-B

100
10
1
0.01

0.1

10

(c)

Wall shear stress


2
(dyne/cm )

100
10
1
0.1

0.01
0.01

Casson
H-B
0.1
1
Mean velocity (cm/s)

10

Fig. 6-29. (a) Wall shear rate, (b) Viscosity, and (c) Wall shear stress. Plotted as a
function of mean velocity at a capillary tube using three non-Newtonian models for
bovine blood with 7.5% EDTA. Note that 1 dyne/cm2 = 102 mPa.

180
CHAPTER 7. CONCLUSIONS AND RECOMMENDATIONS

The present study introduces a scanning capillary-tube rheometer to measure


liquid viscosity over a range of shear rates continuously from high to low shear rates
(as low as 1 s-1). The feasibility and accuracy of the new viscosity measurement
technique has been demonstrated for a standard-viscosity oil and unadulterated
human bloods by comparing the results obtained with a power-law model against an
established viscosity measurement technique. One of the advantages of this new
rheometer is that one can measure the viscosity of whole blood without using
anticoagulants.

In addition, the viscosity measurement of whole blood can be

completed within 2 minutes in a clinical setting, rendering viscosity results over a


wide range of shear rates. The viscosity data from the new rheometer gave excellent
agreement with those measured within 1 minute by the rotating viscometer. The
rotating viscometer could not be used after 1 minute of use with an unadulterated
blood sample due to blood clotting.
The present study introduces a mathematical method to isolate the surfacetension and yield-stress effects on the viscosity measurement in using a SCTR. The
feasibility and validity of the method to reduce data for the SCTR have been
demonstrated for distilled water and bovine blood by comparing with reference data
and the results from a Brookfield cone-and-plate rotating viscometer. The viscosity
of unadulterated human blood has also been measured using the SCTR.

181
The effect of the surface tension was taken into account by using an additional
term, hst , while the effect of the yield stress was considered as a model constant in
either Casson or Herschel-Bulkley model. For the SCTR using gravity as a driving
force, it was necessary to consider the effect of the surface tension even for
Newtonian fluids. Furthermore, in the case of a thixotropic liquid like whole blood,
the surface-tension effect should be isolated from the yield-stress effect to obtain
accurate viscosity data over a range of shear rates using the SCTR. In order to avoid
the influence of the carry-over phenomenon on viscosity measurements, disposable
tube sets were used for tests with fresh human blood.
The present study also investigated the effect of dye concentration on the
viscosity of a dye-water solution using a SCTR. In the experiment, six different
concentrations (0.5, 1, 2, 3, 4, and 7% by volume) of dye were used at 25C. When
the dye concentration was greater than 2%, the viscosity of the dye-water solution
could be significantly altered particularly at low shear rates. Based on the experiment
with the SCTR, one can conclude that the maximum 2% concentration of dye by
volume can be used to make a transparent aqueous solution opaque for the operation
of the SCTR.
The present study investigated the effects of three non-Newtonian constitutive
models on the viscosity and yield stress measurements in a scanning capillary-tube
rheometer: power-law, Casson, and Herschel-Bulkley models.
For a Newtonian fluid (i.e., distilled water), all three models produced
excellent viscosity results.

For non-Newtonian fluids (i.e., bovine and human

bloods), both Casson and H-B models gave viscosity results which are in good

182
agreement with each other as well as with the results obtained by a conventional
rotating viscometer, whereas the power-law model seemed to produce inaccurate
viscosities at low shear rates due to its inability to handle the yield stress of blood.
The yield stress values obtained from the Casson and H-B models for the
human blood were measured to be 13.8 and 17.5 mPa, respectively. The two models
showed some discrepancies in the yield-stress values. The results from the Casson
model seemed to be more accurate than those from the H-B model.
The ability to estimate the wall shear stress in various arterial vessels could be
a significant step in clinical hemorheology. In the present study, the wall shear stress
was found to be almost independent of a constitutive model, whereas the size of the
plug flow region varies substantially with the selected model, altering the values of
the wall shear rate at a given mean velocity. The model constants and the method of
the shear stress calculation given in the study can be useful in the diagnostics and
treatment of cardiovascular diseases.

Recommendations for Future Research

The present study developed a new rheometer that was specially designed for
measuring unadulterated human blood. However, the measurement was not
strictly in vivo. It would be very useful to develop a method to measure the
viscosity of human blood in vivo.

183
-

The present study focused on the method of measuring the viscosity of


unadulterated human blood. As discussed in Chapter 2, whole blood could be
affected by several factors such as RBC deformability and aggregation. The
effects of RBC deformability and aggregation on the blood viscosity should
be studied.

The present study measured the viscosity and yield stress of human blood
without adding any anticoagulants. The study on the effect of thrombotic
tendency of each individual person on both viscosity and yield stress of blood
should be conducted.

The two yield stress models, Casson and Herschel-Bulkley models, gave
different yield stresses for blood in the present study. It is not very clear
which model is more accurate.

An experimental method of measuring

velocity profiles should be developed to determine the more accurate model


for characterizing blood sample.

184

LIST OF REFERENCES

Ajmani, R.S., Rifkind, J.M., Hemorheological changes during human aging,


Gerontology, 44 (1998) 111-120.
Baldwin, A.L., Thurston, G., Endothelial cell morphology in relation to blood flow
and stretch in arterioles and veins, Biorheology, 32 (1995) 112.
Barbee, J.H., The effect of temperature on the relative viscosity of human blood,
Biorheology, 10 (1973) 1.
Barbee, J.H., Cokelet, G.R., The Fahraeus effect, Microvasc. Res. 34 (1971) 6.
Barnes, H.A., Thixotropy-a Review, J. Non-Newtonian Fluid Mech., 70 (1997) 1-33.
Barnes, H.A., The yield stress-a review- everything flows?, J. Non-Newtonian Fluid
Mech. 81 (1999) 133.
Barnes, H.A., A review of the slip (wall depletion) of polymer solutions, emulsions
and particle suspensions in viscometers: its cause, character, and cure, J. NonNewtonian Fluid Mech., 56 (1995) 221-251.
Barnes, H.A., Hutton, J.F., Walters, K., An Introduction to Rheology, Elsevier,
Amsterdam, 1989.
Barnes, H.A., Walters, K., The yield stress myth? Rheol. Acta, 24 (1985) 323-326.
Benis, A.M., Usami, S., Chien, S., Effect of hematocrit and inertial losses on
pressure-flow relations in the isolated hind paw of the dog, Circulation Res. 27 (1970)
1047.
Bird, R.B., Armstrong, R.C., Hassager, O., Dynamics of Polymeric Liquids, Vol. 1,
Wiley, New York, 1987.
Bowdler, A.J., Foster, A.M., The effect of donor age on the flow properties of blood.
Part I: Plasma and whole blood viscosity in adult males, Exp Gerontol 22 (1987) 15564.
Briley, D.P., Giraud, G.D., Beamer, N.B. et al., Spontaneous echo contrast and
hemorheologic abnormalities in cerebrovascular disease, Stroke, 25 (1994) 1564.

185
Brookfield Engineering Laboratories, Inc., Rheology/Methodology,
Reference, Manual No. SB 197-600-B398, 1999.

Seminar

Brown, A.M., A step-by-step guide to non-linear regression analysis of experimental


data using a Microsoft Excel spreadsheet, Comp. Prog. Meth. Biomed., 65 (2001)
191-200.
Casson, N., Rheology of Disperse System, Pergamon, London, 1959.
Chakravarty, S., Datta, A., Dynamic Response of stenotic blood flow in vivo, Mathl.
Comput. Modelling, 16 (1992) 3-20.
Chakravarty, S., Datta, A., Effects of stenosis on arterial rheology through a
mathematical model, Mathl. Comput. Modelling, 12 (1989) 1601-1612.
Chakravarty, S., Datta, A., Dynamic Response of arterial blood flow in the presence
of multi-stenoses, Mathl. Comput. Modelling, 13 (1990) 37-55.
Chen, H.Q., Zhong, L.L., Wang, X.Y., Zhou, T., Chen, Z.Y., Effects of gender and
age on thixotropic properties of whole blood from healthy adult subjects, Biorheology,
28 (1991) 177-183.
Cheng, D.C.H. and Evans, F., Phenomenological characteristization of the rheological
behavior of inelastic reversible thixotropic and antithixotropic fluids, British J
Applied Physics, 16 (1965)1599-1617.
Chien, S., Dormandy, J., E. Ernst, A. Matrai, Clinical Hemorheology, Martinus
Nijhoff Publishers, Dordrecht, 1987.
Cinar, Y., Demir, G., Pac, M., Cinar, A.B., Effect of hematocrit on blood pressure via
hyperviscoesity, American Journal of Hypertension, 12 (1999) 739-743.
Ciuffetti, G., Mercuri, M., Mannarino, E., et al., Peripheral Vascular disease:
Rheologic variables during controlled ischemia, Circulation, 80 (1989) 348.
Cho, Y.I., Choi, E., The rheology and hydrodynamic analysis of grease flows in a
circular pipe, Tribology Trans. 36 (1993) 545.
Cho, Y.I., Hartnett, J.P., Viscoelastic effects in the falling ball viscometer, J. Rheol.
24 (1980) 891.
Cho, Y.I., Kensey, K.R., Effects of the Non-Newtonian viscosity of blood on
hemodynamics of diseased arterial flows: Part 1, Steady flow, Biorheol. 28 (1991)
241.

186
Cho, Y.I., Kim, W.T., Kensey, K.R., A new scanning capillary tube viscometer, Rev.
Sci. Instrum. 70 (1999) 2421.
Cotran, R.S., Kumar, V., Collins, T., Robbins Pathologic Basis of Disease, 6th ed.,
WB Saunders, Philadelphia, 1999.
Coull, B.M., Beamer, N., Garmo, P., et al., Chronic blood hyperviscosity in subjects
with acute stroke, transient ischemic attack, and risk factors for stroke, Stroke, 22
(1991) 162.
Craveri A., Tornaghi G., Paganardi L., Hemorrheologic disorders in obese patients:
study of the viscosity of the blood, erythrocytes, plasma, fibrinogen and the
erythrocyte filtration index. Minerva Med. 78 (1987) 899906.
Cross, M.M., Rheology of Non Newtonian Fluids: A New Flow Equation for
Pseudoplastic Systems, J. Colloid Sci., 20 (1965) 417.
Crouch, J.D., Keagy, B.A., Schwartz, J.A., Wilcox, J.R., Johnson, G., Current
Surgery, 43 (1986) 395
Das, B., Batra, R.L., Non-Newtonian flow of blood in an arteriosclerotic blood vessel
with rigid permeable walls, J. Theor. Biol. 175 (1995) 1-11.
Dash, R.K., Jayaraman, G., Mehta, K.N., Estimation of increased flow resistance in a
narrow catheterized artery-A theoretical model, J. Biomechanics, 29 (1996) 917-930.
de Kee, D., Mohan, P., Soong, D.S., Yield stress determination of styrenebutadienestyrene triblock copolymer solutions, J. Macromol. Sci., B25 (1986) 153.
Dinnar U., Cardiovascular Fluid mechanics, CRC Press, Boca Raton, FL, 1981.
Dintenfass, L., Blood microrheology-Viscosity factors in blood flow, ischaemia and
thrombosis, Appleton-Century-Crofts, New York, 1971
Dintenfass, L., Blood viscosity factors in severe non-diabetic and diabetic retinopathy,
Biorheology, 14 (1977) 151-157.
Dintenfass, L., Modifications of blood rheology during aging and age-related
pathological conditions, Aging, 1 (1989) 99-125.
Einfeldt, J., Schmelzer, N., The theory of capillary viscometers taking into account
surface tension effects, Rheol. Acta. 21 (1982) 95.
Ernst, E.. Haemorheological consequences of chronic cigarette smoking. J Cardiovasc
Risk 2 (1995) 435439.

187
Ernst, E., Matrai, A., Marshall, M., Blood rheology in patients with transient ischemic
attacks, Stroke, 19 (1988) 634.
Fahraeus, R., Lingqvist, T., Viscosity of blood in narrow capillary tubes, Am. J.
Physiol. 96 (1931) 562.
Ferguson, J., Kemblowski, Z., Applied Fluid Rheology, Elsevier Science, London and
New York, 1991.
Fisher, M., Meiselman, H.J., Hemorheological factors in cerebral ischemia, Stroke,
22 (1991) 1164.
Fowkes, F.G.R., Pell, J.P., Donnan, P.T., et al., Sex differences in susceptibility to
etiologic factors for peripheral atherosclerosis: importance of plasma, fibrinogen, and
blood viscosity, Arterioscler Thromb., 14 (1994) 862.
Frame, M.D.S., Chapman, G.B., Makino, Y., Sarelius, I.H., Shear stress gradient over
endothelial cells in a curved microchannel system, Biorheology, 35 (1998) 245.
Fung, Y.C., Biomechanics: Motion, Flow, Stress, and Growth, Springer-Verlag, New
York, 1990.
Fung, Y.C., Biomechanics: Mechanical Properties of Living Tissues, Springer, New
York, 1993.
Fung Y.C., Biomechanics: Circulation, Springer, New York, 1996.
Gaspar-Rosas, A., Thurston, G.B., Erythrocyte aggregate rheology by transmitted and
reflected light, Biorheology, 25 (1988) 471.
Goldsmith, H.L., The microrheology of human erythrocytes suspensions. In:
Theoretical and Applied Mechanics, edited by Becker, E., Mikhailov, G.K., Proc. 13th
IUTAM Congress, Springer, New York, 1972
Grotta, J., Ostrow, P., Fraifeld, E., Hartman, D., Gary, H., Fibrinogen, blood viscosity,
and cerebral ischemia, Stroke, 16 (1985) 192.
Guyton, A.C., Hall, J.E., Textbook of Medical Physiology, 9th edition, W.B. Sanders
Company, Philadelphia, 1996.
Harris, D., Nonlinear least-squares curve fitting with Microsoft excel Solver, J. Chem.
Ed., 75 (1998) 119.
Haynes, R.H., Physical basis of the dependence of blood viscosity on tube radius, Am.
J. Physiol. 212 (1960) 1193.

188
Heywood, N.I., Cheng, D.C.H., Comparison of methods for predicting head loss in
turbulent pipe flow of non-Newtonian fluids, Trans. Inst. Meas. Control, 6 (1984) 33.
Herschel, W.H., Bulkley, R., Measurement of consistency as applied to rubberbenzene solution, Proc. Am. Soc. Test. Matls., 26 (1926) 621.
Hildebrand, F.B., Introduction to Numerical Analysis, 2nd Ed., Dover, New York,
1987.
Hill, M.A., Court J.M., Mitchell, G.M., Blood rheology and microalbuminuria in type
1 diabetes mellitus, Lancet 2 (1982) 985.
Hoare, E.M., Barnes, A.J., Dormandy, J.A., Abnormal blood viscosity in diabetes and
retinopathy, Biorheology, 13 (1976) 21-25.
Holdsworth, S.D., Rheological models used for the prediction of the flow properties
of food products: a Literature Review, Trans IChemE, 71 (1993) 139.
How, T.V., Advanced in Hemodynamics and Hemorheology, Vol. 1, JAI Press,
London, 1996.
Huang, C.R., Siskovic, M., Robertson, R.W., Fabisiak, W., Smitherberg, E.J., Copley,
A.L., Quantitative characterization of thixotropy of whole human blood, Biorheology,
12 (1975) 279.
Isaacson, E., Keller, H.B., Analysis of Numerical Methods, Dover, New York, 1994.
Jacobs, H.R., Meniscal resistance in tube viscometry, Biorheol. 3 (1966) 117.
Jan, K., Chien, S., Bigger, J.T., Observations on blood viscosity changes after acute
myocardial infarction, Circulation, 51 (1975) 1079.
John, E.G., Simplified Curve Fitting using spreadsheet add-ins, Int. J. Engng. Ed., 14
(1998) 375-380.
Kameneva, M.V., Antaki, J.F., Watachi, M.J., and Borovetz, H.S., Biorheology, 31
(1994) 297.
Kameneva, M.V., Garrett, K.O., Watach, M.J., Borovetz, H.S., Red blood cell aging
and risk of cardiovascular diseases, Clinical Hemorheology and Microcirculation, 18
(1998), 67-74.
Kensey, K.R., Cho, Y.I., The Origin of Atherosclerosis, Vol. 1, EPP Medica, New
Jersey, 2001.

189
Keentok, M., The measurement of the yield stress of liquids, Rhol. Acta, 21 (1982)
325-332.
Kim, S., Cho, Y.I., The effect of dye concentration on the viscosity of water in a
scanning capillary-tube viscometer, J. Non-Newtonian Fluid Mech. 2002 (submitted)
Kim, S., Cho, Y.I., Hogenauer, W.N., Kensey, K.R., A method of isolating surface
tension and yield stress effects in a U-shaped scanning capillary-tube viscometer
using a Casson model, J. Non-Newtonian Fluid Mech. 103 (2002) 205-219.
Kim, S., Cho, Y.I., Jeon, A.H., Hogenauer, B, Kensey, K.R., A new method for blood
viscosity measurement, J. Non-Newtonian Fluid Mech. 1939 (2000a) 1-10.
Kim, S., Cho, Y.I., Kensey, K.R., Pellizzari, R.O., Stark, P.R., A scanning dualcapillary-tube viscometer, Rev. Sci. Instrum. 71 (2000b) 3188.
Lee A.J., Mowbray P.I., Lowe G.D., Blood viscosity and elevated carotid intimamedia thickness in men and women: the Edinburgh Artery Study, Circulation 97
(1998) 14671473.
Leiper, J.M., Lowe, G.D.O., Anderson, J. et al., Effects of diabetic control and
biosynthetic human insulin on blood rheology in established diabetics, Diabetes res.,
1 (1982) 27.
Letcher, R.L., Chien, S., Pickering, T.G., et al., Direct relationship between blood
pressure and blood viscosity in normal and hypertensive subjects: role of fibrinogen
and concentration. Am J Med 70 (1981) 11981202.
Letcher, R.L., Chien, S., Pickering, T.G., et al., Elevated blood viscosity in patients
with borderline essential hypertension. Hypertension 5 (1983) 757762.
Levenson, J., Simon, A.C., Cambien, F.A., Beretti, C., Cigarette smoking and
Hypertension, Arteriosclerosis, 7 (1987) 572.
Liepsch, D., Moravec, S., Pulsatile flow of non-Newtonian fluids in distensible
models of human arteries, Biorheology, 21 (1984) 571.
Lowe, G.D.O., Blood rheology in arterial disease, Clinical Science 71 (1986) 137-146.
Lowe, G.D.O., Drummond, M.M. Lorimer, A.R., et al., Relation between extent of
coronary artery disease and blood viscosity, British Medical Journal, 8 (1980) 673.
Lowe, G.D.O., Fowkes, F.G.R., Dawes, J., et al., Blood viscosity, fibrinogen, and
activation of coagulation and leukocytes in peripheral arterial disease and the normal
population in the Edinburgh artery study, Circulation, 87 (1993) 1915.

190
Macosko, C.W., Rheology; Principles, Measurements, and Applications, Wiley-VCH,
New York, 1994.
Magnin, A., Piau, J.M., Con-and-plate rheometry of yield stress fluids-Study of an
aqueous gel, J. Non-Newtonian Fluid Mech., 36 (1990) 85-108.
Mardles, E.W.J., The flow of liquids through fine capillaries and narrow channels:
The meniscus resistance, Biorheol. 6 (1969) 1.
Marshall, D.I., Riley, D.W., J. Appl. Polym. Sci. 6 (1962) 546.
Metzger, A.P., Knox, J.P., Trans. Sci. Rheol., 9 (1965) 13.
Mewis, J., Spaull, A.J.B., Rheology of concentrated dispersions, Advances in Cooloid
and Interface Science, 6 (1976) 173-200.
Microsoft Corporation, Microsoft Excel Users Guide, Version 5, Document Number
XL 57926-0694.
Middleman, S., The Flow of High Polymers, Interscience, New York, 1968.
Misra, J.C., Ghosh, S.K., Flow of a Casson fluid in a narrow tube with a side branch,
International Journal of Engineering Science, 38 (2000) 2045-2077.
Misra, J.C., Kar, B.K., A mathematical analysis of blood flow from a feeding artery
into a branch capillary, Mathl. Comput. Modelling, 15 (1991) 9-18.
Misra, J.C., Pandey, S.K., Peristaltic transport of blood in small vessels: Study of a
mathematical model, Computers and Mathematics with Applications 43 (2002) 11831193.
Missirlis, K.A., Assimacopoulos, D., Mitsoulis, E., Chhabra, R.P., Wall effects for
motion of spheres in power-law fluids, J. Non-Newtonian Fluid Mech., 96 (2001)
459-471.
Mitsumata, M., Hagiwara, H., Yamane, T., Shu, K., Yoshida, Y., Morphology and
function of endothelial cells under the blood flow, Biorheology, 33 (1996) 424.
Most, A.S., Ruocco, N.A., Gewirtz, H., Effect of a reduction in blood viscosity on
maximal myocardial oxygen delivery distal to a moderate coronary stenosis,
Circulation, 74 (1986) 1085.
Munson, B.R., Young, D.F., Okiishi, T.H., Fundamentals of Fluid Mechanics, Wiley,
New York, 1998.

191
Nakamura, M., Sawada, T., Numerical study on the flow of a non-Newtonian fluid
through an axisymmetric stenosis, ASME J. Biomech. Eng., 110 (1988) 137-143.
Nguyen, Q.D., Boger, D.V., Yield stress measurement for concentrated suspensions, J.
Rheol., 27 (1983) 321.
Nguyen, Q.D., Boger, D.V., Measuring the flow properties of yield stress fluids,
Annu. Rev. Fluid. Mech. 24 (1992) 47.
Park, N.A., Irvine Jr., T.F., The falling needle viscometer. A new technique for
viscosity measurements, Wrme-und Stoffbertragung 18 (1984) 201.
Persson, S.U., Gustavsson, C.G., Larsson, H., Persson, S.. Studies on blood rheology
in patients with primary pulmonary hypertension. Angiology, 42 (1991) 836842.
Picart, C., Piau, J.M., Galliard, H., Human blood shear yield stress and its hematocrit
dependence, J. Rheol. 1 (1998a) 42.
Picart, C., Piau, J.M., Galliard, H., Carpentier, P., Blood low shear rate rheometry:
influence of fibrinogen level and hematocrit on slip and migrational effects,
Biorheology, 35 (1998b) 335-353.
Poon, P.Y.W., Dornan, T.L., Orde-Peckar, C. et al., Blood viscosity, glycaemic
control and retinopathy in insulin dependent diabetes, Clin. Sci., 63 (1982) 211.
Pries, A.R., Neuhaus, D. Gaehtgens, P., Blood viscosity in tube flow: dependence on
diameter and hematocrit, American J. Physiology, 263 (1992) H1770-H1778.
Reinhart, W.H., Haeberli, A., Stark, J., and Straub, P.W., J. Lab. Clinical Med., 115
(1990) 98
Resch K.L., Ernst E., Matrai A.,.Can rheologic variables be of prognostic relevance in
arteriosclerotic diseases? Angiology 42 (1991) 963970.
Rosenblum, W.I., Blood, 31 (1968) 234
Rosenson, R.S., Viscosity and ischemic heart disease, J. Vascular Medicine and
Biology, 4 (1993) 206.
Rodkiewicz, C.M., Sinha, P., Kennedy, J.S., On the application of a constitutive
equation for whole human blood, Transactions of the ASME, 112 (1990) 198.
Rohlf, K., Tenti, G., The role of the Womersley number in pulsatile blood flow a
theoretical study of the Casson model, J. Biomechanics, 34 (2001) 141-148.

192
Sajjadi, S.G., Nash, G.B., Rampling, M.W., Cardiovascualr flow modeling and
measurement with application to clinical medicine, Clarendon Press, Oxford, 1999.
Samijo, S.K., Willigers, J.M., Barkhuysen, R., Kitslaar, P.J.E.H.M., Reneman, R.S.,
Brands, P.J., Wall shear stress in the human common carotid artery as function of age
and gender, Cardiovascular Research, 39 (1998) 515.
Schmid- Sch&o&nbein H., Wells, R.E., Rheological properties of human erythrocytes and
their influence upon anormalous viscosity of blood, Physioly Rev, 63 (1971) 147-219.
Sharp, D.S., Curb, J.D., Schatz, IJ., et al., Mean red cell volume as a correlate of
blood pressure. Circulation 93 (1996) 16771684.
Siauw, W.L., Ng, E.Y.K., Mazumdar, J., Unsteady stenosis flow prediction: a
comparative study of non-Newtonian models with operator splitting scheme, Medical
Engineering & Physics, 22 (2000) 265-277.
Skinner, S.J., J. Appl. Polym. Sci. 5 (1961) 55.
Smith, W.C.S., Lowe, G.D.O., Lee, A.J., Tunstall-Pedoe, H., Rheological
determinants of blood pressure in a Scottish adult population, J. Hypertension, 10
(1992) 467.
Stadler, A.A., Zilow, E.P. Linderkamp, O., Blood viscosity and optimal hematocrit in
narrow tubes, Biorheology, 27 (1990) 779-788.
Stoltz, J.F., Singh, M., Riha, P., Hemorheology in Practice. IOS Press, 1999.
Tanner, R.I., Engineering rheology, Oxford University Press, New York, 1985.
Thomas, H.W., The wall effect in capillary instruments: An improved analysis
suitable for application to blood and other particulate suspensions, Biorheology, 1
(1962) 41-56.
Thurston, G.B., Effect of hematocrit on blood viscoelasticity and in establishing
normal values, Biorheology, 15 (1978) 239-249.
Toth, K., Kesmarky, G., Vekasi, J., et al., Hemorheological and hemodynamic
parameters in patients with essential hypertension and their modification by alpha-1
inhibitor drug treatment. Clin Hemorheol Microcirc 21 (1999) 209216.
Tsuda, Y., Satoh, K., Kitadai, M., et al., Chronic hemorheological effects of the
calcium antagonist nilvadipine in essential hypertension. Arzneimittelforschung 47
(1997) 900904.

193
Tu, C., Deville, M., Pulsatile flow of non-Newtonian fluids through arterial stenoses,
J. Biomechanics, 29 (1996) 899-908.
Uhlherr, P.H.T., A novel method for measuring yield stress in static fluids, In Proc.
4th Natl. Conf. Rheol., Aust. Sco. Rheol., (1986) 231.
Usami, S., Wung, S.L., Chien, S., The response of endothelial cell to shear flow,
Biorheology, 32 (1995) 231.
Walawender, W.P., Chen, T.Y., Cala, D.F., An approximate Casson fluid model for
tube flow of blood, Biorheol. 12 (1975) 111.
Walburn, F.J., Schneck, D.J., A constitutive equation for whole human blood,
Biorheology, 13 (1976) 201.
Wildemuth, C.R., Williams, M.C., A new interpretation of viscosity and yield stress
in dense slurries: coal and other irregular particles, Rheol. Acta 24 (1985) 75-91.
Yarnell, J.W.G., Sweetnam, P.M., Rumley, A., Lowe, G.D.O., Lifestyle and
hemostatic risk factors for ischemic heart disease: the Caerphilly Study. Arterioscler
Thromb Vasc Biol 20 (2000) 271279.
Young, D.M., Gregory, R.T., A Survey of Numerical Mathematics, Vol. 1, Dover,
New York, 1988.

194

APPENDIX A. Nomenclature

English Letters

dimensionless radius at = y , r

Cy

dimensionless radius at = y ,

gravitational acceleration [ms-2]

height [m]

fluid height [m]

fluid height difference [m]

constant for Casson model [cP]

KL

loss coefficient

length of tube [m]

length of fluid element [m]

consistency index in Power-law


and Herschel-Bulkley models [cPsn-1]

torque [Nm]

power-law index

pressure [Pa]

pressure drop [Pa]

radius [m]

ry

R
R

195
R

mean radius [m]

radial distance [m]

ry

radial distance at = y [m]

Re

Reynolds number

volumetric flow rate [m3s-1]

distance measured along stream line [m]

time [s]

flow velocity [ms-1]

mean flow velocity [ms-1]

terminal velocity [ms-1]

Greek Letters

shear stress [Pa] or [cPs-1]

&

shear rate [s-1]

Newtonian viscosity [cP]

non-Newtonian viscosity [cP]

constant viscosity near zero shear rate [cP]

constant viscosity near infinite shear rate [cP]

density [kgm-3]

angular velocity [rads-1]

196

angle [rad]

Subscripts

atm

atmosphere

Bingham plastic model

capillary tube

entrance

End

end effects

fluid

inner cylinder

loss

needle

outer cylinder

reservoir or riser tube

st

surface tension

time

unsteady

unsteady state

wall

yield stress

infinity

197

APPENDIX B. Falling Object Viscometer Literature Review

Falling Cylinder/Needle

Figure B-1 shows a schematic diagram of a falling cylinder moving in a fluid


filled in another cylinder. In case that the gap between two cylinders is very small,
then simple shear can be obtained as indicated in Fig. B-1(a). As the cylinder is
dropped in a closed cylinder, the displaced fluid must flow back, which results in the
velocity profile shown in Fig. B-1(b). Typically, a small diameter needle is dropped
in a large cylinder of the test fluid. After the needle falls for a distance great enough
to allow the fluid to reach a steady state, the terminal velocity, v , is determined by
timing between two marks. This generally limits the technique to transparent fluids.
Assuming a wide gap, i.e., << 1, the relations for shear stress, , and the
Newtonian viscosity, , are as follows [Macosko, 1994]:

( n f ) gR

(B-1)

2
( n f ) gR 2
2v

(1 + ln )

where

n = density of needle

f = density of fluid
g = gravitational acceleration

(B-2)

198

R = inner radius of outer cylinder


v = terminal velocity of needle
Park and Irvine (1988) gave relations for a power-law fluid. They also demonstrate
that one can easily change the density of the needle and thus the shear stress, , by
using a hollow tube filled with various amounts of dead weight. In that way, one can
obtain non-Newtonian viscosity, , as a function of shear rate, & .

199

Fig. B-1. Falling cylinder viscometers.


(a) open ends for high viscous samples
(b) closed end, free falling

200

APPENDIX C. Specification of CCD and LED Array

Description

Syscans SV352A8-01 Contact Image Sensor (CIS) is a black/white linear


image sensor module, which is originally designed for scanning a document. Figure
C-1 illustrates the cross sectional view of the SV352A8-01 module. The module
consists of a LED light source to illuminate the document, a one-to-one erect graded
index micro lens array to focus the document image on the photo detector array, an
array of linear MOS image sensors to convert the image into an electronic signal, a
glass cover to protect the sensor array, micro lens array, and LED light source from
dust, 8-pin connector for input/output connections and a protective case to house all
of the components.

Key Features

Compact size: 12 mm height 15 mm width 70 mm length


Resolution: 12 dots/mm (304dpi)
Scanning length: 2.2 inch or 3.2 inch
Scanning speed: 2.5 ms/line
Single power supply (+5V)

201

Cover Glass
Plastic Housing

Rod Lens
LED Light Source
Connector

Sensor Array

PCB Substrate

Fig. C-1. Cross sectional view of SV352A8-01 module.

202

APPENDIX D. Biocoating of Capillary Tube

For the experiment with unadulterated human blood, capillary tubes used in a
SCTR have been coated with biocompatible materials. The coating work was carried
out by a company called Biocompatibles in Farnham, Surrey (U.K.). The following
procedure was employed to coat the inner surface of the capillary tube:
1. Prepare 100 ml of a 10mg/ml PC1036 solution in 99% Hexane and 1%
Ethanol.
2. Clean the capillary tube lumen by using a 20ml syringe to pull and push the
Hexane through the lumen vertically.
3. Pass compressed air vertically through both ends of the capillary tube lumen
for 2 seconds at a flow rate of 30-35 liters per minute to remove any
remaining traces of Hexane.
4. Coat the capillary tube by using a 20 ml syringe to pull and push the PC1036
polymer solution through the lumen vertically.
5. Immediately after coating, pass compressed air vertically through both ends of
the capillary tube lumen for 2 seconds at a flow rate of 30-35 liters per minute
to remove any remaining traces of PC1036 polymer solution.
6. Place the capillary tube horizontally in an incubator preheated to 70 for 4
hours to allow the coating to cure.

203
The capillary tube lumen was reported to be free from blockage after the coating
procedure. In addition, the thickness of the polymer coating cured on the inner
surface of the capillary tube was reported approximately 1 m.

204

APPENDIX E. Microsoft Excel Solver

A powerful tool that is widely available in a spreadsheet format provides a


simple means of fitting experimental data to both linear and non-linear functions
[Microsoft Corporation, 57926-0694].

The curve-fitting method by using Excel

Solver is well-known and widely used in scientific researches [Harris, 1998; John,
1998; Brown, 2001]. The procedure and its mode of operation are very easy and
obvious.

Frequently in engineering, science and business, data is collected and

plotted as a graphical representation of the variables involved. The next step is to


create an association between the variables by connecting the points with a line.
Once drawn, the line is examined and a model which best fits the data points assumed
when the theoretical solution for the data points is not available. Then, this is fitted
and used to replace the existing set of data points as the appropriate model. However,
in case that the theoretical solution is available, this procedure can also be used to
determine the unknowns in the solution.
In order to fit a curve to a data series, using the excel solver is simplicity itself.
If a data series contains the x and y values, and an appropriate model has been
available. Fitting the chosen model is then as follows [Harris, 1998; John, 1998;
Brown, 2001]:
1. Enter the known x and y values as a data series onto the spreadsheet.

205
2. Add a further column containing a chosen model.

The parameters

(unknowns) of the chosen model are estimated and located in any free cells.
These are the Change cells.
3. Add a further column which expresses the squared error between the known

y values and the assumed model values.


4. Sum the squared error column in an appropriate free cell.
5. Evoke Solver by selecting the Tolls menu and Solver to present the Solver
dialogue box.
6. In the dialogue box, enter the sum of the squared error cell as the target cell.
7. Set the Equal to option to Min.
8. Enter the selected Change cells to the By changing cells box.
9. Include any constraints and modify the options as necessary.
10. Select the Solver button to initiate the curve fitting.
The values of the assumed model parameters (unknowns) will then be adjusted in
each of the Change cells until the Target cell value is a minimum. Excel Solver
uses Newtons method of iteration to determine the best combination of unknowns
that fit into the model [Microsoft Corporation, 57926-0694].

206

APPENDIX F. Newtons Method of Iteration

The Newton method is one of the most widely used methods for root finding.
It can be used for the problem to find solutions of a system of non-linear equations
[Young and Gregory, 1988; Isaacson and Keller, 1994; Hildebrand, 1987]. Consider
the general problem of fitting a function of the following type:

y = f ( X ; A)

(F-1)

where X (Variables) = ( x1 , x 2 ,..., x n ) and A (Parameters) = (a1 , a 2 ,..., a m ) .

Choosing the parameters, A = (a1 , a 2 ,..., a m ) , which minimize the sum of


error, the least-square error function, E ( A) , can be expressed by using the following
equation:
l

E ( A) = f ( X ( j ) ; A) y ( j )

(F-2)

j =1

A necessary condition that the parameters corresponding to a minimum is that they


are a stationary point.

Therefore, the following system of equations should be

satisfied:

E
E
E
= 0,
= 0 , , and
= 0.
a1
a 2
a m

(F-3)

Note that some or all of the equations in Eq. (F-3) may be non-linear. Applying the
chain rule to the definition of the error function E , one may rewrite Eq. (F-3) in the
following forms:

207

[f (X

( j)

[f (X

( j)

; A) y ( j )

j =1
l

] f ( Xa

; A)

] f ( Xa

; A)

( j)

=0

; A) y ( j )

j =1

( j)

=0

(F-4)

[f (X
l

( j)

; A) y ( j )

j =1

] f ( Xa

( j)

; A)

=0

Considering the non-linear cases, the standard form for these problems is Eq. (F-4).
F1 ( A) = 0
F2 ( A) = 0

i.e.,

Fm ( A) = 0

F ( A) = 0

(F-5)

where F and A are vectors.


Supposed that an initial approximation, A0 = (a1( 0 ) , a 2( 0 ) ,..., a m( 0 ) ) , to a solution
of the system is provided, the Newtons method can be used. The Newtons method
is based on the Taylor Expansion, which can be expressed in matrix form as follows:
F ( A) = F ( A0 ) + J ( A0 ) ( A A0 ) + higher order terms

(F-6)

where J is the Jacobian matrix whose elements are evaluated at A0 . Since F ( A)


should vanish, and the higher order terms can be assumed to be negligible, Eq. (F-6)
can be reduced to:
J ( A0 ) ( A A0 ) = F ( A0 )

(F-7)

The above equation is a linear system of equations, so one can solve it for A1 A0 .

208

Viscosity (cP)

APPENDIX G. Repeatability Study with Distilled Water

1
0.98
0.96
0.94
0.92
0.9

Test #1
Test #2
Test #3
Test #4
Test #5
Reference (0.892 cP)

0.88
0.86
0.84
0.82
0.8
0

100

200
300
Shear rate (s -1)

Fig. G-1. Repeatability study #1.

400

500

Viscosity (cP)

209

1
0.98
0.96
0.94
0.92
0.9
0.88
0.86
0.84
0.82
0.8

Test #1
Test #2
Test #3
Test #4
Test #5
Reference (0.892 cP)

100

200

300

400

Shear rate (s -1)

Fig. G-2. Repeatability study #2.

500

210

APPENDIX H. Experimental Data

Table H-1. A typical experimental data set of human blood obtained by a scanning
capillary-tube rheometer with precision glass riser tubes. One out of 100 data points
is selected from an original data set.
Time (s)
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
18.00
20.00
22.00
24.00
26.00
28.00
30.00
32.00
34.00
36.00
38.00
40.00
42.00
44.00
46.00
48.00
50.00
52.00
54.00
56.00
58.00
60.00

Pixel Number at Riser Tube 1


1183
1040
1036
1036
1004
976
946
922
907
894
885
891
882
875
868
863
859
854
851
847
845
842
840
838
836
834
833
832
830
829
828

Pixel Number at Riser Tube 2


399
460
509
550
583
609
632
650
666
679
690
700
707
714
720
725
730
734
737
740
742
743
745
747
748
750
751
752
753
754
754

211

Table H-1. Continued.


Time (s)
62.00
64.00
66.00
68.00
70.00
72.00
74.00
76.00
78.00
80.00
82.00
84.00
86.00
88.00
90.00
92.00
94.00
96.00
98.00
100.00
102.00
104.00
106.00
108.00
110.00
112.00
114.00
116.00
118.00
120.00
122.00
124.00
126.00
128.00
130.00

Pixel Number at Riser Tube 1


827
826
825
825
824
823
822
822
821
820
820
819
819
818
818
818
817
817
816
816
816
815
815
814
814
814
813
813
813
812
812
812
812
811
811

Pixel Number at Riser Tube 2


755
756
756
757
757
757
758
758
758
759
759
759
759
759
760
760
760
760
760
760
760
760
760
760
760
760
760
760
759
759
759
759
759
759
759

212

Table H-1. Continued.


Time (s)
132.00
134.00
136.00
138.00
140.00
142.00
144.00
146.00
148.00
150.00
152.00
154.00
156.00
158.00
160.00
162.00
164.00
166.00
168.00
170.00
172.00
174.00
176.00
178.00
180.00

Pixel Number at Riser Tube 1


811
811
811
810
810
810
809
809
809
809
809
808
808
808
808
809
808
808
808
808
807
808
807
807
807

Pixel Number at Riser Tube 2


759
758
758
758
758
758
758
758
758
758
758
758
758
758
758
759
758
758
758
758
758
758
758
758
758

213

Table H-2. A typical experimental data set of distilled water obtained by a scanning
capillary-tube rheometer with plastic riser tubes. One out of 100 data points is
selected from an original data set.
Time (s)
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
18.00
20.00
22.00
24.00
26.00
28.00
30.00
32.00
34.00
36.00
38.00
40.00
42.00
44.00
46.00
48.00
50.00
52.00
54.00
56.00
58.00
60.00
62.00
64.00
66.00
68.00
70.00

Pixel Number at Riser Tube 1


568
411
292
220
178
152
137
128
122
118
117
115
115
114
112
112
112
112
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111

Pixel Number at Riser Tube 2


130
288
407
478
522
548
561
571
576
580
583
584
585
585
585
586
586
586
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587

214

Table H-2. Continued.


Time (s)
72.00
74.00
76.00
78.00
80.00
82.00
84.00
86.00
88.00
90.00
92.00
94.00
96.00
98.00
100.00
102.00
104.00
106.00
108.00
110.00
112.00
114.00
116.00
118.00
120.00
122.00
124.00
126.00
128.00
130.00
132.00
134.00
136.00
138.00
140.00
142.00
144.00
146.00
148.00
150.00

Pixel Number at Riser Tube 1


111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111
111

Pixel Number at Riser Tube 2


587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587
587

215

Table H-3. A typical experimental data set of bovine blood obtained by a scanning
capillary-tube rheometer with plastic riser tubes. One out of 100 data points is
selected from an original data set.
Time (s)
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
32
34
36
38
40
42
44
46
48
50
52
54
56
58
60
62
64
66
68
70

Pixel Number at Riser Tube 1


550
520
481
455
430
405
384
365
349
332
319
304
288
278
268
259
251
244
237
231
224
218
213
208
203
198
193
190
187
184
181
179
176
174
172
170

Pixel Number at Riser Tube 2


115
152
182
210
235
259
280
299
316
331
345
359
370
381
390
399
408
415
422
428
435
440
445
450
455
458
462
465
469
472
475
477
480
482
485
487

216

Table H-3. Continued.


Time (s)
72
74
76
78
80
82
84
86
88
90
92
94
96
98
100
102
104
106
108
110
112
114
116
118
120
122
124
126
128
130
132
134
136
138
140
142
144
146
148
150

Pixel Number at Riser Tube 1


168
166
164
162
161
159
158
156
155
154
153
152
150
150
149
148
147
146
146
145
144
144
143
143
142
142
141
141
140
140
139
139
139
138
138
138
138
137
137
137

Pixel Number at Riser Tube 2


489
491
492
494
496
497
498
500
501
502
503
504
505
506
507
508
509
510
511
511
512
512
513
513
514
515
515
516
516
517
517
517
518
518
518
518
519
519
519
519

217

Table H-3. Continued.


Time (s)
152
154
156
158
160
162
164
166
168
170
172
174
176
178
180

Pixel Number at Riser Tube 1


137
136
136
136
136
135
135
135
135
134
134
134
133
133
159

Pixel Number at Riser Tube 2


520
520
520
520
521
521
521
522
522
522
522
523
523
523
555

218

Table H-4. A typical experimental data set of human blood obtained by a scanning
capillary-tube rheometer with plastic riser tubes. One out of 100 data points is
selected from an original data set.
Time (s)
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
18.00
20.00
22.00
24.00
26.00
28.00
30.00
32.00
34.00
36.00
38.00
40.00
42.00
44.00
46.00
48.00
50.00
52.00
54.00
56.00
58.00
60.00
62.00
64.00
66.00
68.00
70.00

Pixel Number at Riser Tube 1


584
535
484
446
411
380
356
335
315
299
283
272
261
252
244
237
230
224
219
214
210
206
201
196
193
192
189
187
186
184
182
181
179
178
177
176

Pixel Number at Riser Tube 2


133
185
235
273
311
341
366
387
408
423
437
450
460
469
476
482
491
497
502
507
510
513
517
520
522
525
527
529
531
533
534
535
537
538
539
540

219

Table H-4. Continued.


Time (s)
72.00
74.00
76.00
78.00
80.00
82.00
84.00
86.00
88.00
90.00
92.00
94.00
96.00
98.00
100.00
102.00
104.00
106.00
108.00
110.00
112.00
114.00
116.00
118.00
120.00
122.00
124.00
126.00
128.00
130.00
132.00
134.00
136.00
138.00
140.00
142.00
144.00
146.00
148.00
150.00

Pixel Number at Riser Tube 1


176
175
174
173
172
171
171
170
170
169
169
168
168
167
167
167
166
166
166
165
165
165
165
164
164
164
163
163
163
163
162
162
162
162
162
161
161
161
161
160

Pixel Number at Riser Tube 2


541
542
543
544
545
545
546
546
547
547
548
548
549
549
550
550
550
551
551
551
551
552
552
552
552
552
552
553
553
553
553
553
553
553
553
553
554
554
554
554

220

Table H-4. Continued.


Time (s)
152.00
154.00
156.00
158.00
160.00
162.00
164.00
166.00
168.00
170.00
172.00
174.00
176.00
178.00
180.00

Pixel Number at Riser Tube 1


160
160
160
160
160
160
160
160
159
159
159
159
159
159
159

Pixel Number at Riser Tube 2


554
554
554
554
554
555
555
555
555
555
555
555
555
555
555

221

VITA
Sangho Kim
Education
Drexel University
Doctor of Philosophy in Mechanical Engineering
Master of Science in Mechanical Engineering
Kyungpook National University
Bachelor of Science in Mechanical Engineering

Philadelphia, PA
12/2002
Taegu, Korea
02/1997

Journal Publications
S. Kim, Y.I. Cho, K.R. Kensey, R.O. Pellizzari and P.R. Stark
A scanning dual-capillary-tube viscometer
Review of Scientific Instruments, Vol. 71, No. 8, August 2000, 3188-3192
S. Kim, Y.I. Cho, A.H. Jeon, B. Hogenaeur and K.R. Kensey
A new method for blood viscosity measurement
J. Non-Newtonian Fluid Mechanics, 94 (2000) 47-56
S. Kim, Y.I. Cho, W.N. Hogenaeur and K.R. Kensey
A method of isolating surface tension and yield stress effects in a U-shaped scanning
capillary-tube viscometer using a Casson model
J. Non-Newtonian Fluid Mechanics, 103 (2002) 205-219
S. Kim and Y.I. Cho
The effect of dye concentration on the viscosity of water in a scanning capillary-tube
viscometer
J. Non-Newtonian Fluid Mechanics, 2002 (submitted)
S. Kim, Y.I. Cho, and W.N. Hogenaeuer
Non-Newtonian constitutive models for the viscosity and yield-stress measurements of
blood using a scanning capillary-tube rheometer
Biorheology, 2002 (submitted)

Conference Publications

S. Kim and Y.I. Cho


A new method of measuring blood viscosity with a U-shaped scanning capillary-tube
viscometer using a Casson model
Proceedings of the IEEE 28th Annual Northeast Bioengineering Conference, April 20-21,
2002, 253-254
Y.I. Cho and S. Kim
A new scanning capillary tube viscometer for blood viscosity measurement
Proceedings of the First National Congress on Fluids Engineering, September 1-2, Korea,
2000, 5-8

US Patents
U.S. Patent No. 6,428,488
U.S. Patent No. 6,412,336
U.S. Patent No. 6,322,524
U.S. Patent No. 6,402,703
U.S. Patent No. 6,450,974

Anda mungkin juga menyukai