(International Centre For Mechanical Sciences 270) C. M. Rodkiewicz (Eds.) - Arteries and Arterial Blood Flow-Springer-Verlag Wien (1983)

INTERNATIONAL CENTRE FOR MECHANICAL SCIENCES
COURSES AND LECTURES - No. 270
ARTERIES
AND
ARTERIAL BLOOD FLOW
EDITED BY
C.M. RODKIEWICZ
UNIVERSITY OF ALBERTA
SPRINGER-VERLAG WIEN GMBH

This volume contains 199 fJgUres.
This work is subject to copyright.

All rights are reserved,
whether the whole or part of the material is concerned

specifically those of translation, reprinting, re-use of illustrations,
broadcasting, reproduction by photocopying machine
or similar means, and storage in data banks.
© 1983 by Springer-Verlag Wien
Originally published by Springer-Verlag Wien New York in 1983
ISBN 978-3-211-81635-6 ISBN 978-3-7091-4342-1 (eBook)

DOI 10.1007/978-3-7091-4342-1
PREFACE
With the ever increasing quality and volume of human knowledge we also arrive at a
better understanding of our own health problems. This is particularly true in medical and
paramedical research. Since diseases and defects of the human cardiovascular system remain
one of the most important causes of troubles and death, researchers are expanding
considerable enerf(y to understand this complex system.
Late Professor W. Olszak recognized t~e importance of this branch of learning and was
instrumental in the organization of the International Seminar on "Engineering and Medical
Aspects of the Arterial Blood Flow" in Udine, of which this book is part of
The book contents provide an exposition of the standard concepts and their
application as well as provide some recent research and its results. The material is arranged
in five chapters.
Chapter I concerns the blood rheology and its implication in the flow of blood. This is
studied as dominated by plasma viscosity, hematocrit and red cell properties, namely
aggregability and deformability. Quantitative models of highly concentrated suspensions,
which exhibit shear-thinning, thixotropy and viscoelasticity are discussed. Annular (two
phase) flow models are developed for analyzing blood flow in narrow vessels. Some
examples in clinical application are given.
Chapter II deals with the arterial walls. The geometry and mechanical properties of
blood vessels of the cardiovascular system are discussed in detail. Mathematical descriptions
are presented and compared with experimental work. The time dependence is treated both
in terms of quasilinear elasticity, continuous relaxation spectra, and general nonlinear
viscoelasticity theory.
Chapter III briefly introduces the dynamics of fluid filled tubes. Initial value problems
are considered f~r fluid filled elastic and viscoelastic tubes. Material properties are chosen
which are appropriate for biological applications. Numerical techniques for the inversion of
integral transforms, which arise in the analysis, are discussed and numerical results are
presented.
Chapter IV concerns the .small arteries and the interaction with the cardiovascular
system. This chapter describes and defines the properties of the microcirculation: the
;;
physical properties of the wall, the contractibility, the flow and pressure in the small vessels
and the transcappillary movement of the fluid. Special parts of the circulation (particularly
of coronary circulation) and the interaction between small arteries and the other
components of the circulation are described.
Chapter V deals with the flow in large arteries. Flow characteristics and governing
parameters for the Newtonian fluids are discussed. Basic differential equations and shear
stress expressions are set. This is followed with theoretical and experimental considerations
pertinent to the flow in straight tubes, curved passages, simple junctions, and the aortic
arch. Correlation with the atherosclerotic formations is indicated.
Czeslaw M. Rodkiewicz
CONTENTS
PREFACE i
IN MEMORIAM: WACLAW OLSZAK ix
CHAPTER I BLOOD RHEOLOGY AND ITS IMPLICATION IN FLOW OF BLOOD

1. GENERAL FEATURES OF BLOOD CIRCULATION IN NARROW VESSELS.
BASIC RHEOLOGY CONCEPTS
1 .1 Introduction 2
1.2 Constituents of Blood 3
1.3 Main factors in blood circulation and continuum models 7
1 .4 Basic rheological concepts 9
2. VARIABLES GOVERNING SUSPENSION RHEOLOGY, BRIEF SURVEY OF
EXPERIMENTAL METHODS
2.1 Variables governing mechanical behaviour of disperse
systems: Dimensional approach to suspension rheology 17
2.2 Experimental methods 20
2. 3 Viscometry 21
3. BLOOD AS A HIGHLY CONCENTRATED SUSPENSION: EFFECT OF
HEMATOCRIT
iv
3.1 Variables governing rheological properties of blood

and RBC suspensions under steady conditions: (1)
variations of viscosity with hematocrit at given
shear rate 28
3.2 Viscosity of extremely diluted and dilute suspensions 30
3.3 Viscosity of concentrated suspensions 32
3.4 Application to blood and RBC suspensions 45
4. BLOOD AS A STRUCTURED (SHEAR-THINNING) FLUID: (1) EFFECTS
OF RBC AGGREGATION AND DEFORMATION
4.1 Variables governing rheological properties of blood
and RBC suspensions under steady conditions: (2)
variations of viscosity with shear rate at given
hematocrit 50
4.2 Factors governing the RBC Deformation-Orientation 52
4.3 Factors governing the RBC Aggregation 54
5. BLOOD AS A STRUCTURED (SHEAR-THINNING) FLUID: (2) MODELS
AND APPLICATIONS
5.1 Simple rheological models for viscosity of non-
newtonian fluids 58
5.2 Extension of the newtonian model deduced from
optimization of energy dissipation to the description
of non-newtonian behaviour 60
6. BLOOD.AS A THIXOTROPIC-VISCOELASTIC FLUID: EFFECTS OF
MECHANICAL AND PHYSICO-CHEMICAL PROPERTIES OF THE RBC
MEr1BRANE
6.1 Some experimental results about thixotropy and visco-
elasticity of blood and RBC suspensions 76
6.2 Some models for thixotropy 81
6.3 Some models for viscoelasticity 84
6.4 Effects of mechanical and physico-chemical properties
of the RBC membrane 88
7. BLOOD FLOW IN NARROW VESSELS AS A TWO-PHASE (ANNULAR) FLOW:
BLUNTED VELOCITY PROFILE, PLASMA LAYER, FAHRAEUS AND
FAHRAEUS-LINDQVIST EFFECTS. TWO-FLUID MODELS
7.1 Experimental evidences in blood microcirculation 91
7.2 One-fluid models 98
7.3 Two-fluid models 99
8. CLINICAL HEMORHEOLOGY. CONCLUSION 106
REFERENCES 118
v
CHAPTER II THE ARTERIAL WALL 129
1. GEOMETRY AND STRUCTURE OF ARTERIES

1.1 Introduction 129
1.2 The Size of Arteries 130
1.3 The Geometry of the Large Vessels 131
1.4 The Area of Branches and Branching Angles 133
1.5 Structure of the Arterial Wall 134
1.6 Thickness of the Arterial Wall 138
2. EXPE.RIMENTAL MECHANICS OF THE ARTERIAL WALL

2.1 Quasistatic Load-extension Response 140
2.2 Time Dependence 142
2.3 Anisotropy 144
2.4 Compressibility of the Arterial Wall 145
2.5 Structural Basis of Mechanical Response 146
3. STRESS-STRAIN RELATIONS
3.1 Incremental Elasticity 150
3.2 Strain Energy Density 154
4. TIME DEPENDENCE
4 .1 Single Integral Representations 1 59
4.2 Descriptions of Time Dependence 160
4.3 Multiple Integral Representations 166
4.4 Time Dependence in Biological Tissues 170
REFERENCES 171
CHAPTER III DYNAMICS OF FLUID FILLED TUBES 179
1. INTRODUCTION
1.2 Formulation of Problem 181
1.3 Constitutive Equations of Tube Material 183
2. GOVERNING EQUATIONS
2.1 Tube Equations 185
2.2 Fluid Equations 186
2.3 Nondimensionalization Scheme 187
2.4 Nondimensional Form of Equations 189
vi
2.5 Solution of Fluid Equations 190

3. SOLUTION OF INITIAL VALUE PROBLEM
3.1 Dispersion Relations for Elastic Tube 192
3.2 Effect of Shear Deformation 196
3.3 Solution of Viscoelastic Tube 201
REFERENCES
CHAPTER IV SMALL ARTERIES AND THE INTERACTION WITH THE

CARDIOVASCULAR SYSTEM 217
1. INTRODUCTION
1.1 What is a small artery? 218
1 .2 Shear rate and shear stress and resistance to flow 222
1 .3 The arterial tree 225
1.4 The branching ratio 228
2. THE WALL OF SMALL ARTERIES
2.2 Arterial wall structure 231
2.3 Mechanical behaviour of arterial walls 234
2.4 Relations between vessel wall properties, flow and
·wave velocity - some "rule-of-the-thumb-equations" 240
2.5 Dynamic properties of small arteries 243
2.6 Linear modelling of the dynamic elastic behaviour 245
2.7 A note on incremental moduli 247
2.8 Critical closing pressure 250
2.9 Pressure-flow relation at the peripheral resistance 252
3. FREQUENCY DYNAMICS OF THE PERIPHERAL CIRCULATION
3.2 High frequency behaviour of a transmission line 259
3.3 The high frequency input impedance of arteries 265
3.4 Pressure and flow transmission 270
3.5 Pressure amplitude and peripheral pressure-flow
relation 276
3.6 Wave velocity 280
4. SMALL ARTERIES AND THE CONTROL OF FLOW AND PRESSURE
4.2 The low frequency input impedance 288
vii
4.3 Humoral autoregulation 295

4.4 Instability of flow and pressure in small arteries 297
4.5 Autooscillations and vasomotion 299
5. THE DOWNSTREAM EFFECTS OF SMALL ARTERIES
5.2 Relation between resistance vessels and
microcirculation 302
5.3 Hematocrit in small arteries, arterioles and
capillaries 304
5.4 Transport through the microcirculation 307
5.5 A new method and its application for the study of
the microcirculation 309
LITERATURE AND REFERENCES 315
CHAPTER V FLOW IN LARGE ARTERIES 327

1. INTRODUCTION
1 .1 Introduction 327
1 .2 Newtonian Behaviour of Blood 330
1 .3 Separation and Stagnation Regions 330
1 .4 Flow Governing Parameters 332
1 .5 Arterial Passage Classification 333
2. BASIC EQUATIONS
2.1 The Governing Equations 333
2.2 The Equations of Motion in Rectangular Coordinates 334
2.3 The Equations of Motion in Cylindrical Coordinates 335
2. 4 The Shear Stresses 336
3. FLOW IN STRAIGHT PASSAGES
3.2 Flow in a Rigid Tube of Elliptic Cross-Section 339
3.3 The Hagen-Poiseuille Flow 341
3.4 Pulsating Flow in Rigid Circular Tube 343
3.5 Entrance Length 351
3.6 Influence of Blood Cells 352
3.7 Arterial Pressure-Flow Relationship 354
4. FLOW IN CURVED PASSAGES
viii
4.2 Fully Developed Flow in a Curved Pipe 358

4.3 Unsteady Flow in Curved Rigid Tube 364
4.4 Unsteady Flow in Thin-Walled Curved Elastic Tube 366
4.5 Entrance Arc 367
4.6 Laminar Flow Downstream of a Bend 367
4.7 Drag on the Wall 371
5. FLOW IN JUNCTIONS
5.1 Simple Bifurcation 371
5.2 Symmetric Y Junction 385
5.3 Note on Non-Rigid Wall 387
6. AORTIC ARCH
6.1 Aortic Arch Flow Field 388
6.2 Early Atherosclerotic Lesions 394
LITERATURE AND REFERENCES 398
CONTRIBUTORS 413
IN MEMORIAM
WACLAW OLSZAK.
Prof. Olszak received Ph. D. degrees in Engineering and Mathematics at the Universities
of Paris and Vienna.
In 1952, he founded the Department of Mechanics and Continuous Media, which has
since become the most important and dynamic research center in Poland. Later he became
head of that Department. His research problems focused on the theory of elasticity and
plasticity, rheology and the theory ot structures. He was also the first in Poland to taclde
the problems of the theory of prestressed structures, and the results of his investigations
were published in two monographs.
Prof. Olszak also authored more than 300 scientific and technical papers, as well as 10
monographs, and was the editor-in-chief of several journals of the Polish Academy of
Sciences.
From 1936 till his death in 1980, Prof. Olszak participated in 5 international
congresses at which he presented his original research papers. In addition, he organized
several lecture tours in Canada and the United States, as well as in several countries in
Europe.
His -international reputation and authority in the field of plasticity has earned him
numerous honorary degrees all over the world. Prof. Olszak had received the title of
"Doctor Honoris Causa" from the Universities of Toulouse, Liege, G~gow and New
Brunswick, from the Technical Universities of Vienna, Dresden, Warsaw and Cracow and
from the Mining College in Cracow.
In addition, he held membership at the Academies of Sciences in Toulouse, Belgrade,
Stockholm, Helsinki, Turin, Halle, Vienna, Sofia, Boulogne, Paris, Madrid, Ljubljana and
Buenos Aires, honorary memberships in Budapest and Luxemburg, and a Fellowship in New
York.
During his career, Prof. Olszak founded many organizations all over the world. One of
his more noteworthy accomplishments has been participating in the foundation of the
International Center for Mechanical Sciences (CISM) in Udine, Italy, in 1969. The CISM is,
so far, the only institution which, on an international level, covers all domains of
Theoretical and Applied Mechanics. It has rapidly acquired a recognized international
reputation. To a great extent, this was due to Prof Olszak, who as CISM's rector, engaged his
scientific reputation and international experience to the benefit of this dynamic institution.
BLOOD RHEOLOGY AND ITS IMPLICATION IN FLOW OF BLOOD
Daniel QUEMADA
Laboratoire de Biorheologie
et d'HYdrodynamique Physicochimique
UNIVERSITE PARIS VII - 2, Place Jussieu
75251 Paris Cedex 05 (FRANCE)
SUMMARY : Rheological behaviour of blood lS studied as dominated by

plasma viscosity, hematocrit ~nd Red Cell properties, namely aggregabi-
lity and deformability. Quantitative models for highly concentrated sus-
pensions, which exhibit shear thinning, thixotropy and viscoelasticity,
are discussed. Annular (two-phase) flow models are developped for analy-
sing blood flow in narrow vessels. Some examples in clinical application
are given.
2 D. Quemada
1 - GENERAL FEATURES OF BLOOD CIRCULATION IN NARROW VESSELS.

BASIC RHEOLOGICAL CONCEPTS.
1.1 Introduction
Blood flow has been early considered as having a major i~po~tance ln
physio-pathology : for a long time, blood remaln as the vital fluid,
heart failure as death.
Transport of oxygen and nutrients that tissue cells need for their
metaboJic activity on the one hand and removal of carbon dioxide and
metabolic products, on the other hand, are the main functions of blood
circulation. Moreover 'J.nder 'rarying external conditions, peripheral (skin)
microcirculation maintains a statle te~perature within the body.
The rate of this exchan[e ~ainly depends on (i) red blood cell
(RBC) concentration (i.e. hematocrit::':) ln circ'J.lating blood anci (ii)
flow rate Q. It is therefore determined by the cardiac per:orEance. :or
a long time, the latter has been thought of as only related to the abili-
ty of heart to deliver the energy for blood pumping. It now appears more
and more that flow resistance dominates the cardiac performance, not
only through characteristic properties of blood vessels (non linear(visco-
elastic)mechanical properties ; geometrical factors - branching, tapering,
singularities - and vasa-control effects) but also through fluid proper-
ties. Indeed, rheological characteristics of blood (non-newtonian visco-
sity, thixotropy and visco-elasticity) and physico-chemical properties of
plasma, play a significant role in flow regulation in health and diseases.
Several authors , as S. CHIEN (1972) stressed that "hemorheological
factors were unjustly underestimated up to now, in comparison with car-
diac or vascular factors".
Indeed, it is not unreasonable to think that number of cardiovascular di-
seases-- currently attributed to some myocardiac or vascular pathologies
which would appear in response to direct action of some external factor--
could be in fact a secondary effect resulting from a primary hemorheolo-
gical disease due to this external factor acting directly on blood.
General Features of Blood Circulation 3
One goal of hemorheology is the analysis of direct effects that ma-

ny substances have on mechanical properties of blood and then, the study
of the consequences that these modifications have on blood flows. Clini-
cians and Physiologists, and therefore, Hydrodynamicists have to begin
their analysis with the study of these flow properties of blood. Before
classifying them, let us describe briefly the main characteristiree of
blood.
1.2 Constituents of blood.

A. Blood composition.
Blood is a very highly concentrated suspension (about 40 - 45 %in v~
lume fraction) of a variety of cells, Red Cells (RBC), White Cells (WBC),
Platelets ( 0 ) suspended in a continous phase called plasma.
Plasma is an aqueous solution of electrolytes 'and organic substances
mainly proteins. Relative proportions of cell elements and concentrations
of plasma components in blood are shown on Table 1 .1. Plasma behaves as
a newtonian fluid whose viscosity np depends primarily on temperature as
water does. Strong non newtonian properties of plasma reported in some
observations had been considered as resulting from air-plasma interfacial
TABLE 1.1- BLOOD CONSTITUENTS (5.106 Particles/mm3)

CELL ELEMENTS (Relative proportions)
Hhite cells (all kinds) 1
Platelets 30
Red cells 6oo
PLASMA (Weight fraction)
Water 0,91
Inorganic solutes 0,01
Proteins 0,07
Other organic substances 0,01
( 0 ) alternatively called, erythrocytes, leucocytes and thrombocytes,

respectively.
4 D. Quemada
effects and therefore must be considered as non-existing. However, it

can be expected, that due to the presence of long macromolecules, plasma
should exhibit non-newtonian properties at low shear rate and that using
a guard ring is required to avoid the formation of an interfacial film
of proteins.
20
Fig.1.1. Viscosity of RBC

content (from Data of CHIEN
et al. , 1970 ) .
0 40
B.General properties of blood.
As RBCs compose about 97% of the total cell volume, whole blood pro-
perties are (rheologically) dominated by RBC properties (excepted in so-
me pathologic cases as leukaemia in which the large number of white cells
increases both viscosity and shear-rate dependence of blood (DINTENFASS,
1965) ),
Hematocrit is the volume fraction of red cells plus trapped plasma
obtained by centrifuge of a.volume of whole blood prevented from clot-
ting by addition of anticoagulants. Blood rheological properties have
been found free from any effects that would be induced by these antico-
agulants(generally hepari~ or EDTA ( 0 )). The hematocrit value His normal-
ly a little greater than the true volume fraction ¢ , since a small volu-
me of plasma is trapped between the cells and it is advisable to use a
corrected hematocrit ¢ = 0.96 H ~n normal conditions (CHIEN and USA~I,
1971). Nevertheless, this correction strongly depends on RBC deformabi-
lity and becomes very important in the case of rigid particles for which
packing volume fraction is about 0.60. Such a value for the correction
factor was found in the case of hardened cell suspensions (CHIEN and
USAMI, 1971 ).
C. The red cell.

Human RBC can be considered as a partially inflated balloon
(BROCHARD, 1977), thus having a very high level in deformability, which
decreases with the age of the cell. Internal flui-d is a hemoglobin so-
lution (35 g/100 ml), the viscosity of which is about 6 to 8 cP (See
Fig. 1.1) . RBC membrane consists of a biomolecular leaflet (a lipid
bilayer, with inclusion of anchored proteins) supported by a more rigid
skeletal structure (the actin-spectrin network). RBC membrane carries a
negative electrical charge, equivalent to about 6.000 electronic charges
per cell.
The RBC biconcave discoid shape (in the case of mammals) results
from the ejection by the cell of its nucleus on entering the circula-
tion. This leads to a volume reduction of the cell, its area remaining
constant. RBC dimensions (Fig.1 .2) are greater than the critical ones for
which Brownian motion becomes important: Therefore, under steady condi-
tions, blood settling occurs. After one hour, a plasma column above the
settling RBC is formed, the height of which is about 1 em for normal
blood, but which can be ten (or more) times higher in diseases (in hos-
( 0 ) EDTA = disodium-diamino-ethane-tetra-acetate.
6 D. Quemada
pitals, this height is currently

used as a clinical index, the ESR ( 0 )
Although these sedimentation effects

can be neglected when blood flows ve-
ry rapidly, they are more and more
important as flow rate is decreasing .
Changes in osmotic conditions (nor-
mal ones correspond to those of an
aqueous solution containing about
Fig .1 . 2 Human RBC dimensions
0.9% by weight of NaCl) induce dras-
tic changes in RBC shape. Indeed,
RBC becomes spherical (and then is called spherocyte) after swelling in
solutions with lower NaCl concentration than normal ones. Further swel-
ling leads to hemolysis of the cell which eliminates hemoglobin towards
the plasma, the resul t being a ghost, i.e . a RBC reduced to its membrane
only . Under higher NaCl concentrations, on the contrary, the RBC volume
decreases, what leads to the formation of echinocytes, the membrane struc-
ture of which likely involves the actin-spectrum network.
D. Pl asma proteins and coagulation

Plasma proteins (See Table 1. 1) are mainly albumins and globulins
which control water exchanges between blood and tissues by their action
on osmotic balance . The remainder are lipoproteins and fibrinogen. The
latter plays a very important role in the clotting mechanism of blood .
However, under normal conditions, and in association with 6- globulins,
fibrinogen i s closely involved in the reversible aggregation of RBC
which fo2;'111 rouleaux, Fig.1 .] . We will see later that it i s one of the fun-
damental determinants of blood rheology .
Whole blood solidifies in the presence of air oxygen , or coming
--~---------------
(
0
) Erythrocyte Sedimentation Rate
into contact with extracorporeal surfaces. The f ormation of a clot would

be observed after an early increase in viscosity (DI NTENFASS, 1971 ). In
vitro, this clot mainly consists of a complex structure of RBC and fi-
brin which derives f rom the plasma fibrinogen by a chain reaction . Pla-
telets play a predominant role in clotting under in vivo conditions.
After complete formation, a contraction of the clot (COPLEY, 1960 ) begins,
which resembles syneresis, a similar retraction observed during gelation
process in colloids. After the clotting is achieved, the extra-fluid
is called serum, which ~s roughly plasma without fib rinogen .
Fi g .1 . 3 Phot omicrographs of r oule aux of

human RBC (a) which tend to f orm a more
or less complicated network (b).
1 .3. Main factors in blood circulation and continuum model

These blood properties play different roles ·as one considers dif-
ferent parts of the body.
Three main classes of blood circulation are usually sorted out, ac-
cording to the ratio of the vassel diameter, 2R,to the particle "size",
2a, ~ = R/a . These groups are the following:
8 D. Quemada
a) The capillary circulation ( 0 ) (~ ~ 1) 1n a very large number

( 1 . 2 x 109) of very narrow (3 ~m ~ 2R ~ 8 urn) and relatively long (~ ~
1mm) vessels . Inside the vessels, RBCs undergo a strong deformation.
Hence, the capillary circulation is dominated by RBC rheological proper-
ties i . e. the membrane visco-elasticity and t he internal fluid viscosity .
b) The systemic circulation (~ ~ 50) in few thousands of arteries
and veins (0 . 6 mm ~ 2R ~ 20 mm; 1 em~~~ 40 em), inside which blood
flow pulsatility promotes i mportant deformations of vessel walls . Thus
this circulation involves the rheological properties of vessel walls -
i.e . their non linear viscoelasticity - related to their microscopic
structure, with elastin, collagen and smooth muscle as the t hree major
wall materials (BERGEL, 1964) . Although it was currently accepted that
in'large vessels blood behaves as a newtonian fluid, it seems that , es-
pecially in veins , abnormal (i.e . non- newtonian) properties of blood
must be taken into account (FLAUD and QUENADA , 1980) .
c ) The microcirculation ( 0 } (3 ~ ~ ~ 50) in arterioles and venules
(about 108 in number; 10 urn~ 2R ~ 500 ~m 2 mm ~ 1 ~ 10 mm) which is
governed by these anomalous (non-newtonian) blood properties. Such pro-
pert ies are dominated by blood microscopic structure and associated with
the formation of a particle depleted wall layer (the so-called plasma
l ayer) anda particle rich axial core.The existence of such a two phase
(annular) flow leads to "anomalies" as Farhaeus Effect and Farhaeus-
Lindquist Effect .
I:.~~~~-~LI!z.<!.l2.~~'!.~~~-~~s_~l2.~t~q,!!.·
It must be noticed that this classification corresponds closely to
the limits in validity of the descript ion of the fluid as a quasi-homo-
geneous media , hence in validity of the hydrodynamic approximation . In
0
( ) Number of authors used the term microcirculation for circulation
both ( i) in capillaries and (ii)arteriols and venules . Hereafter we
prefer to limit the use of this t o the latter .
§ 1 .3a, the problem is confined to the mechanics of deformable bodies im-

mersed in a newtonian fluid (the plasma), in the presence of rigid walls.
In § 1 .3b, as a <<<R, blood can be considered as a homogeneous fluid, ha-
ving an effective (bulk) viscosity, more or less newtonian, (which howe-
ver depends on concentration). In§ 1 .3c, the problem is much more complex
since hydrodynamic approximation does not hold at this scale. Nevertheless,
on the artery (or vein) side, the plasma layer thickness o remains small
ln comparison with vessel rs.dius and the validity of the hydrodynamic e.p-
proximation is recovered when assuming that the actual flow can be des-
cribed as an annular two-phase flow of two different quasi homogeneous
fluids. As the ratio ~ = R/a decreases, the previous axial core, which
undergoes shear stresses ( 0 ) is progressively replaced by a more or less
axial single file of cells, in which cells follow each other keeping their
discoidal surfaces approximately perpendicular to the axis of the vessel.
Such a structure, named axial train by R.L. WHITMORE (1967), is still a
two-phase structure and it can be thought that, at least approximatively,
a continuous description by such an annular two-phase flow would be possi-
ble down to ~ +1 .
Anyway, using such two-phase flow models requires the knowledge of

bulk rheological properties of blood, which will be surveyed briefly after
recalling some basic rheological principles and experimental methods. Then,
general (more or less empirical) models will be studied with hope that
they may provide a basis for experimental and clinical data interpretation
of blood flow in narrow vessels (in vitro and, as far as possible, in vi-
vo).
1 .4. Basic rheological concepts.

A) Viscometric flows
As rheological behaviour of complex materials is very complicated,
the simplest flow conditions were used for their rheological characteri-
( 0 ) This annular flow can be called a sheared-core.

10 D. Quemada
zation. These conditions are achieved in viscometric flows, as simple

shear (which occurs in shear and pressure flows restrained by fixed boun-
daries i.e. Couette and Poiseuille flows) or as pure shear and axisymme-
tric distorsion (which occurs in elongational or compressional flows, as-
sociated with extension of sheets or filaments, respectively),
B) Steady shear flows : shear viscosity.

B.1 Newtonian viscosity.
Uniform plane shear is the flow obtained when two flat parallel pla-
tes with fluid in between are moved uniformily past one another, the dis-
tance h between them being kept constant and the fluid being assumed to
adhere to both plates (Fig.1 .4). A force per unit area (shear stress
a = F/S in dynes/cm 2 ) is required
to move one plate (the other being
stationnary) with the velocity V.
In ordinary fluids and under stea-
dy conditions, the fluid velocity
varies linearly from v = o (on the
stationary plate) to v V (on the
moving plate). One gets a plane
Couette Flow, which approximates
Fig.1 .4 Shear flow between two the Couette Flow between coaxial
parallel plates
cylinders (Radius R1 and R2= R1 + £)
when rotating the outer one, if the
gap width £ 1s very small in comparison with R1 ~ R2. The (constant) veloci-
ty gradient (or shear rate, in sec-1) is y = V/h.
a) For number of (ordinary) fluids, as water, oils, simple organic
fluids •.. , the shear stress a is found varying linearly with the shear
rate (Newton law):
a = ny (1 •1 )
the(constant) coefficient being the viscosity (in Poises):

n = .
(J
y
(1 .1 a)
Fluids to which ( 1 . 1) applies are called Newtonian fluids, the vis-

cosity of which is constant, i . e . does not depend on y (or a).
Furthermore, solutions and disperse systems (as suspensions , emul-
sions , ... )at least insi de a limited range of shear rate , show a newto-
nian viscosi ty which depends on parti cl e concentration (See Fig . 1. 5 from
GOLDSMITH , 1973) .
~ro ~~---.--.---.--.
10 ~~~~-+-+~~--~
2
0 .2 .4 .6 .8
Fig . 1 .5 Relative apparent viscosity nra of

particle suspensions, as a funct i on of volu-
me fract i on ~ · (Human blood 0; de-oxygenated
s i ckel cell a ; rigid spheres • ; rigid
dis us 6 ; droplets • ; (From GOLDEMITH, 1 973).
B. 2 Non newtonian viscosity .

However, any number of usual "fluids", as polymer solutions and
melts, inks , paints , slurries, colloids , bio- fluids ... deviate from
Newton law . Therefore, their "vi scosity", as the ratio of stress to
rate of shear (1 .1 a) is not a const ant , but depends on o or y .
or ( 1 • 2)
n or n(o) ( 1 • 3)
12 D. Quemada
Such fluids are called non-newtonian fluids, in which additional for-

ces arise from anisotropy of internal pressure. For the plane shear, the-
se forces called normal (or Weissemberg) forces, acting 1n the direction
normal to the streamlines, tend to separate the plates. In most cases the
viscosity (1 .3) decreases continuously when y (or cr) increases: the fluid
is then called a shear-thinning (or pseudoplastic) fluid. Fig. 1 .6 illus-
trates such a variation.On the contrary, it is called a shear-thickening
fluid,sometimes related,to dilatant effects observed in nearly close-
packed disperse systems, as \.,rater-sand mixtures.
a b
log~b:::_
,
,,
I
I log~
I
..·· .. .:1· "1 (X)
.. ······~,e'~~~'o~e~ ...... -
\"":........:::J •
... '?r Y 1),
Fig.1 .6 Non-newtonian (Shear-Thinning) Steady Behaviour a) cr=cr(y)
variations; b) Structural viscosity n=n(y), having the two limiting
values no=n(o) and noo= n(oo)·.(on set, curve n versus .y 'in logarith-
mic coordinates, exhibits an apparent plateau for y-+ o).
As most of these fluids are disperse systems, their pseudoplastic beha-

VIour has been attributed to structure modifications of the system under
shear stresses. For instance, at very low shear rate, particles form lar-
ge clusters (or floes, or aggregates), sometimes leading to a tri-dimen-
sional network, like in a gel. Such structures are progressively broken
down by shear stresses, as shear rate increases, which leads to the cor-
responding reduction of viscosity. At a given steady shear rate, a dyna-
mic equilibrium between formation and destruction of aggregates is to be
observed. If particles are not rigid, their deformation grows by increa-
sing the shear rate, that leads to increase the particle orientation in
the flow direction (the same effect is obtained with non spherical parti-
cles). Hence, an additional cause of lowering of viscosity as y increases
is found out. This discussion calls for the very important concept of
structural viscosity (first introduced in 1926 by OSTWALD and AUERBACH).
c) Steady elongational (or extensional) flow.

This type of flow (also called pure straining motion) corresponds to
the velocity field generated when an incompressible fluid is drawn out
into threads or sheets. For instance, for a thread parallel to the x-coor-
dinate the velocity components are, with K = ilv1 /ilx 1 , v 1 = KX
Vz =- !Kx2 , V3 =- !KX3 , and the material behaviour under this uniaxial
flow is described by the elongational (or Trouton) viscosity nE such as:
(1. 4)
For Newtonian fluids, nE = 3n , n being the (simple) shear viscosi-

ty. However, nE = nE(K), for non-newtonian fluids, and no relation bet-
ween nE(K) and n(y) can be predicted. Large enhancements of nE in sus-
pensions of elongated particles when compared to nE of the suspending
fluid, have been experimentally observed (WEINBERGER and GODDARD 1974).
These enhancements are much larger than the corresponding increases in
shear viscosity n .
Although nE is more difficult to measure than n and only few data
are thus available, such extensional effects could be of some importance
on blood circulation since extentional flows exist, at least somewhat
partly, at branching of vessels.
d) Unsteady shear flows: thixotropy and viscoelasticity

The existence of a shear dependent internal structure leads to time-
dependent effects, since structure modifications resulting from a sudden
change of shear rate (or stress) are not instantaneous.
14 D. Quemada
d 1 ) The response of a shear thinning material to the application

at t = 0 of a sudden step of shear rate (from 0 to Y1 for instance) is
shown on Fig. 1.7 . At t = 0, the system behaves without changing the
structure that it had at zero shear rate, i.e. it behaves like a fluid
b
a CT a
ao <To
c:r, o;
lt---1
t ..
to 0 t, 'ir 0 to t
c
"',,"'o
~
2t,
Fig.1 .7 Variations of cr(t) and n(t) for a thixotropic fluid
submitted to a step· of shear rate y , in relation with shear
rate dependence of steady shear stress and viscosity, cr(y) and
n( .y).
having a viscosity n0 = n(y= 0) (or like a gel if the structure at

y = 0 is a tri-dimensional network of aggregates, with n 0 ~ oo), Thus,
at t = O, the shear stress instantaneously reaches the value cro= noYl
(rrom 0 to A', Fig.l .7b). However, the zero shear structure cannot be
maintained at y= Yl, and it relaxes (from A' to A) towards the equili-
brium structure (i.e. the dynamic equilibrium at y = Yl), the viscosity
of which being n1 = n (y = Yl) and the corresponding shear stress,
a1 = n1Y1· This structure adaptation involves relaxation times, which cha-

racterize the relaxation of the structure. Such a time behaviour is called
thixotropy (Fig.i .8c) for shear-thinning materials, and rheopexy for shear
thickening materials.
j
a
1,1--1 GIVEN STEP IN SHEAR RATE
to t Responses for
I-- I
(S'
b cr., A NEWTONIAN FLUID
to t
<S'
c a; 1-r
to t
A THIXOTROPIC FLUID
(j
d
o;l-~ A VISCOELASTIC FLUID
to t
~
e
<S;l-~ A THIXOTROPIC AND
VISCOELASTIC FLUID
t
Fig.1 .8 Different types of non-newtonian behaviour under a step

of shear rate.
If the shear stress - shear rate relaxation a = a( y) lS not linear,

the type of time curve a(t) depends on the step height Yl . At low Yl
values, the response a(t) is very similar to the one of a newtonian
fluid (Fig.l .8b). As Yl increases, the difference ao - al 'between the
peak of the curve ao and the equilibrium value a1, grows.
d2) Thixotropy is not the only time behaviour currently observed. A

very large number of materials exhibit a viscoelastic behaviour, associa-
16 D. Quemada
ted with a purely elastic (instantaneous) response, followed by a delayed

one (retarded elasticity) (Fig.1 .8d). Nevertheless, the latter is very
often superposed to thixotropy and it appears very difficult to separate
these two types of effects (which moreover would be overshadowed by the
instrument dynamics). A typical time-curve cr(t) is shown on Fig.1 .Be.
Non-linearities still influence the form of the time curve:at very low
Yl , the response is very similar to the one for a pure viscoelastic (as
in Fig.1 .8d) and increasing the amplitude Yl , increases the height of
the peak ("overshoot").
Dynamical studies ,·i.e. analysis of the response (amplitude and
phase) in shear stress (or rate) induced in the sample of material under
applying an oscillatory shear rate (or stress) are ver,r usual ln Vlsco-
elastic media. For instance, a sinusoidal stress in observed in response
to an oscillatory shear rate:
• • iwt
Y = YICoswt = YIRe(e ) ( 1 • 5)
where Re(g) = real part of any complex number g, and i 2 = -1. This respon-
se in stress exhibits a phase difference with y such as:
cr = crocos ( wt + 'P) = cr1coswt ·
+ crzslnwt ·
, whlch ·
can be wrltten cr = Re ("'cre iwt)
using the complex amplitude of stress:
( 1 • 6)
This leads to introduce a dynamical viscosity as a complex number
( 1. 7)
where n1 = ~ and nz =~are the viscous and elastic components, res-

Yl Yl
pectively.
Variables Governing Suspension Rheology 17
2 - VARIABLES GOVERNING SUSPENSION RHEOLOGY - BRIEF SURVEY OF EXPERIMEN-

TAL METHODS".
2.1 Variables governing mechanical behaviour of disperse systems Di-

mensional approach to suspension rheology.
Attempt to predict flow behaviour of concentrated suspensions re-

quire to introduce "structural parameters" into rheological equations in
order to interrelate the rheological properties of the material and its
molecular or particulate structune. Progress in choosing such parameters
can be gained using a dimensional approach similar to that given by
KRIEGER (1963) for colloid suspensions.
The simplest system (from the structural point of view} is a sus-

pension of uniform rigid spheres in a newtonian fluid. For non-interac-
ting spheres, the viscosity n of this suspension will depend on the
eight following variables
0 particle variables (dimension)
sphere radius a (L)
density Pp (ML-3)
number per unit vmlume n (L-3)
0 suspending fluid variables
viscosity (ML-1T-1)
density (ML-3)
thermal energy (ML-2T-2)
(K being the Boltzman Constant)
0 flow variables
shear rate
"experimental" time
(or period}
18 D. Quemada
With n, these nine variables can be combined to form s1x dimension-

less groups :
the relative viscosity nr n/nF ( 2. 1)
the volume fraction <P = (4rr/3) na 3 (2.2)
the relative density Pr = PpfPF (2.3)
- the internal Reynolds number
- a reduced shear rate
.
Yr = TY
.
Rei= PFa 2 y/nF (2.4)
(2.5)
- a reduced time tr = t/T (2.6)
In (2.5) and (2.6), Tis a characteristic (internal) time of the

system. Taking T as the Brownian diffusion time
2 -1
T 'V a Dtr "' (2. 7)
where Dtr = KT/6~nFa and Drot = KT/8~nFa3 are the translational and
rotational Brownian coefficients for diffusion, respectively, leads to
(2.8)
where b = numerical constant. Then (2.5) appears_ as closely related to

translational and rotational (internal) Peclet numbers,Pet = a2 y/Dtr
and Per= y/Drot• respectively and (2.6),the inverse of the Deborah num-
ber,De.
Using these reduced groups leads to the general form of the relati-
ve viscosity equation :
(2.9)
In the case of neutrally buoyant particles, (pr=1) and assuming la-

minar Stokes flow (Rei< 1), (2.9) reduces to :
(2.10)
Steady laminar flow'at fixed Yr gives nr(~), that will be studied

in the next§ 3. Steady laminar flow at fixed~ leads to nr( yr), the
variation of which has been given on Fig. 1.6b for a shear-thinning ma-
terial, for instance. Nevertheless, in the case of very large particles,
as glass beads, Tis very high [T( ) ~ 6 b (a( )) 3 ] . Thus, even for
sec ~m
the smallest value available in viscometry, T >>> 1 : a constant vis-

cosity (close to the high shear limit) is observed, i.e. the fluid ap-
pears as newtonian. Likewise, T <<< 1 for very small particles, as mo-
lecules of a low molecular weight solute, giving again nr = cte (but
equal to the low shear limit), i.e. a newtonian behaviour.
Unsteady experiments as stress recovery after sudden stop of the

viscometer (See § 1.4D) will involve the same characteristic timeT
(as a relaxation time for thixotropic buildup of the structure at y= 0)
since nr = nr ( tr) in this case.
"Turbulence effects "can be occur if Rei becomes larger than a cri ti-
cal value Re~r.In fact, critical tube Reynolds number, pVR/n is found
1
at least two orders in magnitude lower than the usual critical value for
ordinary fluids.
In the case of more complex systems, each additional variable re-
quires a new dimensionless variable in the viscosity equation, as shown
in the following examples. For a suspension of ellipsoidal particles, the
axial ratio is ; 1!a1 of semi axes of rigid spheroid. Purely elastic par-
ticles (with a shear modulus G) suspended 1n a fluid of viscosity nF
have a relaxation time ( 0 ) Tm = nF/G that will be used for large defor-
( 0 ) that is the Maxwell relaxation time of a viscoelastic liquid (which

shows now viscosity and elasticity mix to give visco-elastic proper-
ties) (See § 6).
20 D. Quemada
mable particles (since Tin (2.8) is too large) to define t= ~y/G, as

the ratio of shear force to the elastic one. An emulsion involves two
additional parameters, the interfacial tension r and the internal visco-
sity of droplets ni that leads to the relaxation time T0 = nFa/r
and to the new dimensionless variables TnY and ni I np· Furthermore, ac~
counting for particle interactions involves ratio W/KT of some characte-
ristic energy W to the thermal one for instance, short range Van der
Waals interaction (energy ~ A/6 £a , where A is the Hamaker constant)
leads to take as new variable the ratio A/KT; long range forces as elec-
trostatic repulsion are governed by the surface potential (or surface
charge density) and by screening effects from the ionic double layer,
the thickness K; 1 of which (the so-called Debye length) enters in the
dimensionless group K 0 a.Comparing experiments performed at different

temperatures or ionic strengths allow estimation of effects that these
dimensionless parameters would have on the rheological behaviour.
2.2. Experimental methods.
Rheological studies of non-newtonian fluids are at first the mea-

surement of the shear stress-shear rate relationship under steady condi-
tions.
Structural changes involve time-dependent measnxements in which viscoe-
lastic and thixotropic effects are mixed, leading to difficulties in da-
ta interpretation. Nevertheless, as some structure knowledge is expected
from rheological measurements, complementary experiments are often car-
ried out to verify structural models introduced in rheological studies.
There follows a brief description of them.
(a) Microscopic observations under shear controlled conditions give 1n-
formation about (i) particle orientation and (eventually) deformation,
(ii) particle aggregation and their variations with shear rate. Quanti-
tative approaches have been performed using photometric aggregometry
(KLOSE et al, 1972), or laser back-scattering (MILLS et al, 1980).
(b)Sedimentation under controlled shear rates (COPLEY et al.1975; VIN-

CENT et OLIVER, 1977) exhibits variations of ESR closely related to the
aggregability of particles. (c) Filtration experiments (on polycarbonate
sieves with different mean pore diameters) have been believed as measu-
ring particle deformability (for ex. GREGERSEN et al., 1967) although
artefacts result from particle aggregates at the entrance of the pores.
(d) Doppler-Laser Velocimetry (for ex. BORN et al., 1978) allows us to
fit ·model velocity profiles on the experimental ones, giving some ir:-
sight into the plasma layer influence.
A brief survey of the operating principles and main characteristics
of usual viscometers will follow.
2.3 Viscometry
Steady measurements of blood rheological properties concern mainly

its shear viscosity. Two types of viscometers are currently used, capil-
lary (tube) viscometers and rotational viscometers.
A. Tube viscometers
Fluid flows down a tube from a reservo~r. The tube diameter 2R is
precisely known (nevertheless this diameter is always larger than about
300 pm, in order to avoid phase separation effects (See §7.1). Onere-
cords the time interval during which a fixed volume of fluid flows
through the tube under a given pressure gradient P = ~p/L. As shear ra-
te varies inside the tube, decreasing from its maximum value Ym at the
tube wall to zero at the tube axis, the measurement involves the (un-
known) flow curve between 0 and Ym· In the case of a newtonian fluid,
no difficulty arises since n = cte and the pressure-flow rate relation
(Poiseuille law) give the viscosity
1T RLf
n =- - P (2.11)
8 Q
22 D. Quemada
Corrections for end effects have been taken into account using
pairs of capillary tubes of equal radii and of different lengths (COKE-
LET, 1972).
In the case of non-newtonian fluids, eq.(2.11) 1s used to define
the apparent viscosity of the suspension na, which is the viscosity of
a fictive newtonian fluid which would give the same flow rate under the
same pressure gradient. In the case of homogeneous fluid (with local y
value as a single-valued function of local a-value), the true relation-
ship between y and a can be calculated from the values at the wall,aw
and Yw :
and (2.12)
Using the local shear rate y =-dv/dr, where v(r) is the axial component
of velocity, the flow rate expression can be integrated by parts :
Q = 21T JR v(r) r dr = - 1T JRr 2 ~; dr (2.13)

0 0
using the no-slip boundary condition at the wall r = R • v(R) = 0.

Changing the variable r into the local shear stress a = aw ~ trans-
forms (2.13) into
Q = (2.13a)
After differention of (2.13a) with respect to aw and using (2.12),

the Rabinowitsch-Moon ey equation is obtained which gives the true visco-
sity n (as the ratio aw I yw ) from the experimental data Q = Q(P) :
2
1TR4 (2.14)
It is worth noting that (2. 14) gives the value of the true viscosi-
ty n = n( Y) for Y = Yw·;, the true shear rate at the wall. Both n and Yw
differ respectively from na (by 2.11) andy
aw = 4Q/TIR , the latter being
3
the (fictive) wall shear rate of a fictive newtonian fluid (with viscosi-
ty na). Eq (2.11) is recovered if the fluid is assumed to be newtonian,
i.e. if dQ/dP = Q/P.
Nevertheless numerical derivative dQ/dP is required (since Q and P

are known as finite sets of values) and for the sake of simplicity re-
sults are often given using apparent viscosity na vs. apparent shear ra-
te Yaw· Moreover enhanced entry and exit effects can occur with vis-
coelastic fluids, depending on the average time ta of passage of the
fluid through the tube. (One needs ta > T, the relaxation time of the
fluid, in order to reach steady flow).
B. Rotational viscometers
The two maln types are the coaxial cylinder and the cone-plate vls-
cometers.
B.1- ~~~~~~!-~l!~~~~~-~~~~~~~~~~ (Fig. 2. 1)
The fluid is held between a cylindrical (inner) bob (radius = R,
height= h) and a coaxially mounted (outer) cup (radius= sR, s > 1).
The cup is rotated at constant angular speed n and the viscous drag
transmitted through the fluid is measured by the angular deflection of
the bob (suspended from a torsion wire) or by the torque 1 required to
maintain it at its original position.
When the gap between cup and bob lS small enough (s- 1 ~ 1), the
mean shear stress o and the mean shear rate Ya are defined as
a
1
0 = (2. 15)
a s 4TIR 2h
24 D. Quemada
(2.16)
L is the shear stress on the bob surface.

where crw
True viscosity (for newtonian fluids) or apparent viscosity (for

non newtonian fluids) is given by eq. (2.15) and (2.16) such as :
L
(2.1'7)
4rrR2h
Fig . 2 . 1 Coaxial cylinder

viscometer.
The bob (b) undergoes shear
forces resulting from the rota-
tion of the cup (c), driven
(angular speed n)by a motor
i (not represented here).
Torque measurements are perfor-
med by recording the current ~
through the coils (me) which
creates on the permanent magnet
(pm) an electromagnet ic moment
opposite to the torque moment.
This current is generated by
the photo-ele<!tri c cells (pc).
b ..
and amplifier (a) when the spot
s from the lamp (1) is reflec-
n ted by the mirror (m) is devia-
ted from the middle position between the cells.
For non newtonian fluids (restricted to the case when local shear
stress is an only function of local shear rate), the true viscosity can
be obtained from the following infinite series (KRIEGER and ELROD, 1953)
which give'3 th.e shea;· rate Yw at. the bob surface
.
Yw [ 1 + x ~
dx
+ l
3
x2 (d2y +
~
(~-2-1
dx
) +
•••
( 2. 18)
where -x_ = Ln-s x = Ln crw , and y Lnn
As in tube viscometers, end and edge effects can be eliminated u-

sing two bobs with different heights.
The fluid fills the gap between a cone of very large apex angle 26
(such as a = 90° - 6 ~ 4° for example) and a flat surface (plate) normal
to the cone axis. Cone apex just rests on the plate. One unit (cone or
plate) 1s rotated (angular speed n) and the viscous drag is measured as
a torque L on the other. If phase separation does not occur the fluid
inside the gap experiences a constant shear rate y= n/a • If R is the
"cone radius" the shear stress at the plate surface is
(2.19)
from which the apparent viscosity na (the true one for newtonian fluids)
can be calculted directly :
(2.20)
26 D. Quemada
Eq. (2.20) approximates the true

Cone viscosity within the limits of va-
lidity of constant shear rate y in-
side the gap (since y is very close
to yw, the shear rate at the plate
surface).
Fig. 2.2 Cone/Plate viscometer.
C. Some experimental problems in viscometry.

Different kinds of problems arise in viscometry of disperse systems,
which would limit the validity of measurements.
(i) First at all, the scale length of the viscometer (i.e. tube radius,
gap width between coaxial cylinders or cone and plate) must be greater
than about fifty times the particle diameter to avoid phase separation
effects. For blood, this requires a gap larger than about 500 ~m. Howe-
ver, this value will be too small at low shear rates since particle ag-
gregates are present, the lower the shear rate, the larger the aggregate
size. That leads to the formation of a marginal zone near the walls. Si-
multaneously, sedimentation of these aggregates occurs especially at low
shear rates and seems to give larger error in cone-plate viscometers
than in coaxial cylinder ones. Both phase separation and sedimentation
result in time dependent viscosity measurements under steady conditions,
(QUEMADA et al., 1981) then in difficulties in low shear viscometry.
Variables Governing S9spension Rheology 27
(ii) Interface effects come from the formation of a film of proteins at

the blood-air interface. Mechanical properties of this film can perturb
the viscosity measurement, especially in the case of the plasma. A guard
ring can prevent such an error (BROOKS et al., 1970).
(iii) As the onset of turbulence occurs in particle suspensions at cri-

tical Reynolds Number (~ 10) much smaller than Rec (~ 2000) for ordina-
ry fluids, it is necessary to insure that high shear rate measurements
remain free from such effects.
28 D. Quemada
3 . BLOOD AS A HIGHLY CONCENTRATED SUSPENSION EFFECT OF HEMATOCRIT.
3.1. Var~ables governing rheological properties of blood and RBC suspen-

sions under steady conditions: (1) variations of n with hematocrit at
given shear rate.
Apparent viscosity of blood, withdrawn with addition of anticoagu-
lant, varies as a function of the sample hematocrit H and the applied
shear rate Ya . Apparent relative viscosity is :
= = ( 3. 1 )
where np is the plasma viscosity. This function verifies the following

relations
an ra
> 0 < 0 (3.2)
ay a
that is (i) as for any suspension, blood viscosity increases when volume
fractiongrows and (ii) blood is a non-newtonian (shear-thinning) fluid.
Eq. (3.1) is in agreement with general conclusions of dimensional
analysis (See§ 2.1) for local values of the relative viscosity nr
nr(H , yr) under steady conditions. The present section 1s devoted to
the analysis of concentration effects on blood viscosity at fixed applied
shear rate, nr = nr(H) especially at very low or very high shear rates
when the suspension behaves as a newtonian fluid.
Fig. 3.1, from CHIEN et al., (1971 a) shows such variations nra(H)
for normal human RBC suspended in plasma and in Ringer solution ( 0 ) . Al-
though these viscosities are much smaller than those of suspensions of
( 0 ) which is a saline solution (0. 9% (w) of NaCl) containing small atnounts

Effect of Hematocrit 29
. s-1
y T = 3'7°C T = 20°C
0,052 ...
0,52 - -- - 0
5,2
52,0
- -
.'' .. '. •
6
1
(From S. CHIEN et al.,
0 40 80 HO 40 SOH ( 1971 ) )
Fig. 3.1 Relative apparent viscosity nra of normal RBC
suspended in saline (A) or plasma (B) as a func-
tion of hema!ocrit H , at different values of
shear rate y , and at two temperatures.
rigid particles at same volume fraction (See Fig. 1.5), rapid variations
are observed near the physiological level in hematocrit. Measurements
were performed in a coaxial-cylinder viscometer, at two temperatures
(20°C (points) and 37°C (curves)). These results give proof that the es-
sential part of temperature effects are involved in the plasma viscosity
., (CP)
2
104 y-1 (•K)-1

~~----------~-----------~----------~
3Z 33 34 35
Fig. 3.2 Temperature variation of plasma viscosity (different
protein content) (from COKELET, 19'72).
30 D. Quemada
since points measured at 20°C are very close to the experimental curve .
drawn from measurements at 37°C, although "plasma" viscosity of the for-
mer was 1.5 times the viscosity of the latter. In fact, this temperature
dependence of plasma viscosity results in main part from the similar va-
riation of water viscosity (See Fig. 3.2 from COKELET, 1972).
In order to understand and to quantify these variations we shall re-
turn in the following sections to main models which have been built to
gain a viscosity equation valid at high concentrations, after recalling
classical results on dilute suspensions.
3.2. Viscosity of extremely diluted and dilute suspensions.

In a very dilute system, the suspension viscosity n has been ob-
served higher than the suspending fluid viscosity nF . The relative vis-
cosity n , as a function of the volume concentration ~ , is given by :
r
(3.3)
where k1 is a particle shape dependent factor, thus which depends on

particle deformability (throughthe state of deformation reached in the
experiment).For hard spheres, EINSTEIN (1905), found theoretically k 1 =
kE = 2.5 .
Following the CHIEN's interpretation (1972), such an lncrease in
viscosity (n > 1) can be understood with the help of the concept of
r
the effective particle volume veff , first introduced by ANCZUROWSKI
and MASON ( 1967). This volume is defined as "the sum of the true particle
volume 11 and the volume of the surrounding fluid which behaves as if
p
it was a rigid extension of the particle". Roughly, this extension cor-
responds to the fluid volume perturbed by the presence of the particle,
which results in a squeezing of streamlines. As a consequence, velocity
gradient is enhanced, then viscous forces and thus energy losses are in-
creased, leading to a growth in bulk viscosity. For a rigid rod or disc
this volume v:ff is higher than ll~ff for a sphere having the same
volume v
, leading to k1 > 2.5 . On the contrary, k1 < 2.5 for a
p
liquid drop, which undergoes deformation and partial alignment, with ro-
tation of the internal fluid (See Fig. 4 in CHIEN's paper, 1972).
As eq. (3.3) contains only particle-fluid interactions discarding
any hydrodynamic action of one particle upon another, this equation only
holds at extremely low concentrations, that has been an important diffi-
culty in experimental verification of the Einstein's result. Above ~ ~
0.01 - 0.05 , deviation from linearity is observed and attributed to par-

ticle-particle interactions. The simplest interaction involves two sin-
gle particles and give the ~2 term in a power series for relative vis-
cosity :
n
r
=1 + k 1 ~ + k 2 ~ 2 + ... (3.4)
A number of theories calculated the ~2 coefficient, k2 , by different

methods, but leading to very different values (k 1 being assumed equal
to 2.5 for spheres). Accounting for hydrodynamicinteraction between
spheres by the method of successive reflections and doublet formation by
collisions, VAND (1948) obtained k2 = 7.349. Nevertheless, the most ri-
gorous calculation of shear viscosity recently performed (BATCHELOR and
GREEN, 1972) gave k2 = 5.2! 0.3 (this imprecision results from diffi-
culties in numerical evaluation of integrals).
Increasing concentration requires more terms in the series (3.4). No
theoretical calculation of coefficients k. exists and their determina-
l
tion can be only carried out by data fitting (see later).
In the case of deformable particles, as bubbles or liquid droplets
suspended in an unmiscible fluid, TAYLOR (1932) gave the expression of
k1 as a function of the ratio a= n./nF
l
, n.l being the internal vis-
cosity of particles
= (3.5)
a + 1
32 D. Quemada
where '( is the Taylor's factor. With r as the drop interfacial tension,
eq. (3.5) holds in the limit S =(~Fy a/r) << 1 , that is if particles
undergo very small deformation. From eq. (3.5), Einstein's result is reco-
vered with "rigid" particles suspension (ex » 1) , while gas bubble
emulsion (ex<< 1) leads to k1 + 1 .
3.3. Viscosity of concentrated suspensions.

For highly concentrated suspensions, phenomenological approaches ha-
ve been developped. They lead to (phenomenological ) viscosity equations,
and using them appears better than power series as (3.4). Indeed such re-
lations ~ = ~(~) must represent the strong increase in viscosity up to
infinity, as ~ tends towards its maximum (packing) value ~M (Fig. 3.3).
Fig. 3.3 Relative viscosity

~ versus volume frac-
1
- 1+~ tion~ of particles, in
concentrated suspensions
Infinite viscosity is
found at packing con-
centration ~M .
0~------------~------~
We shall give a detailled study of these equations and of some purely

empirical relations.
3.3.1. Phenomenological equations
A first group of studies refers to "effective medium" theories, con-
sidering the suspension with finite concentration ~ , a9 a fictive sus-
pending fluid,having a (unknown) viscosity ~(~) , to which an incremen-
tal fraction of. spheres, d~ , is added. The new suspension (concentra-
tion ~ + d~)is then considered as extremely diluted (concentration =
d~) comparing to the first one (considered as the suspending medium). Ap-
plying therefore the Einstein's result gives :
nr = n(~ + d~)/n(~) = (3.6a)
that yields the Arrhenius equation (ARRHENIUS, 1917)
n
r
= exp(kl ~) (3.6)
BRINKMAN (1952) analysed more cautiously the relation between d~
and the concentration of the(fictive)extremely dilute system.Assuming

the latt~r is obtained by adding one sphere (volume ~ ) to the N ones
p
contained in a suspension of volume V (thus ~ = N ~ P /V) , he calcula-
ted the concentration increment d~ from ~ + d~ = (N + 1 )~ /(V + 11 ) ,
p p
that is d~ =~ p (1 - ~)/(V + ~ ).
p
Applying the Einstein equation (3.6a)
to the dilute suspension having the volume concentration ~ /(V + ~ )
p p
led to
n(~ + d~) _ 1
n(~)
= kl __M_
1 - ~
(3.7a)
and gave, if n (O) = 1 ,

r
(3.7)
Eq. (3.7a), compared to (3.6a), shows that the particle addition is

equivalent to adding the amount v- d~ to the remaining liquid volume
V(1 - , therefore as an excluded volume effect.
~)
By a similar calculation, ROSCOE (1952) obtained (3.7) with an ef-

fective concentration ~eff which takes into account both particle size
distribution and effects of fluid trapping by aggregates. Putting ~eff =
A ~ in (3.7) gives
nr = (1 - A ~)-k 1 (3.8)
where A is a concentration dependent empirical factor, close to unity

at low concentration, but increasing up to 1.35 = (0.74)- 1 as ~
34 D. Quemada
increases and tends to the packing value ~M (close packing of spheres

is assumed to be reached here, that is ~M = 0 • 74).
It is worthy of note that although using Einstein equation for the fic-
tive dilute system, particle - particle interactions are taken into ac-
count through interactions between the added particle and the effective
suspending fluid, i.e, through the "effective medium" description.
MOONEY (1951), did not consider, as above, the high dilution limit.
He started from a suspension of spheres at ~1 , then having a viscosity
n1 = nF f(~1) where nF is the viscosity of suspending fluid and f(~)
appears as the relative viscosity. This factor must reduce to the Eins-
tein's relation (3.3) as ~1 + 0 . Adding a finite amount ~2 he obtai-
ned a new suspension, the viscosity of which was written n1+2 = n1 f($ 21 ),
where ~21 =
~2/(1 - A ~1) is the concentration of ~2 in the remai-
ning liquid filling the space not occupied by ~1 , with A as a crow-
ding factor. Therefore f(~1 + ~2) = f(~1).f(~21). Considering that
crowding influences mutually ~1 and ~2 , Mooney similarly took ~12
instead of ~1 and arrived to the following functional equation for

nr(cp) - f'(cp)
(3.9a)
the solution of which is :
= exp (3.9)
accounting for the Einstein's limit at ~ + 0 . Mooney extended eq, (3.9)

to polydisperse suspensions.
KRIEGER and DOUGHERTY (1967) claimed that crowding does not act sym-
metrically on ~1 and ~2 depending on the order of operating, one
must take ~1 and ~21 (or ~2 and ~12) • Therefore,one must write
instead of (3.9a) :
~1
= f(h) . f( )
1 - A~2
which is satisfied if :
nr = (3.10)
where k1 is again chosen from the dilute limit.

VAND (1948) solved the hydrodynamic equations for fluid flow around
a sphere. From the solution, he derived the Arrherius formula (3.6), in
the case of a suspension of non-interacting spheres, assuming that the
lue y of the shear rate to

.
disturbance of streaming caused by the particles changes the original va-
•
y + dy , the shear rate increment
.
dy
.
be:mg
related to the increase of viscosity dn by dn/n = - dy/y , which re-
sults from the conservation of the shear stress a = ny in a Couette
flow. Accounting for hydrodynamic interactions between spheres (calcula-
ted by thereflection method) led to change k1 from (3,6) into a concen-
tration dependent coefficient such as
k = k1 I (1 - Q~) (3,11a)
where Qv= 39/64 = 0.609 ... is an interaction constant. Finally, as col-

lisions between spheres create doublets, triplets, ,,, a mean shape fac-
tor k for the suspension as a mixture of particles of various shapes
(shape factors
. K.)
~
at different concentration ~-
~
, was written such as
=I K.
~
~·
~
that led to the viscosity equation (3,6), where k1
i=1
is changed into
(3,11b)
Therefore, n resulted in a power series like ( 3, 4) , Vand calculated the

r
life time of a doublet·and estimated its shape factor (including the ef-
fect of the immobilized liquid). Thus he obtained (3,4) for the viscosity
of a suspension of spheres, with k1 = 2,5 and k2 = 7.349,
A second group of attempts to nr = nrC~) results from application
of an energy principle which allows one to derive only general bounds on
36 D. Quemada
viscosity (KELLER et al., 1967), For instance, by the principle of mini-

mum entropy production, S. PRAGER (1963) obtained a lower bound for the
relative fluidity
Nevertheless a new viscosity equation can be obtained from an energy

principle, recalling that, as yet mentioned in §1.3, flows of concentra-
ted suspensions through narrow channels exhibit a two-phase structure,
Taking into account the possible occuring of such a phase separation
with a particle depleted marginal zone near the walls(i.e. for instance
in a pipe, a radial distribution in concentration, ¢ = ¢(r)), the va-
lidity of an optimization principle for viscous dissipation has been pos-
tulated (QUEMADA, 1977). As particle motions and flow of suspending
fluid are coupled by hydrodynamic interactions, a self-consistent solu-
tion - i.e. both v(r) and ¢(r) - is believed to exist ,where v(r) is
the velocity profile. As shear stresses induce structural changes, one
assumes in this approach the existence of a (structural) viscosity, n =
n(¢,y) will depend not only on properties of the state of the system but
also on flow conditions, the knowledge of which is thus required for de-
fining this viscosity function.
For fixed properties of the whole system (concentration, temperature,
suspending fluid characteristics ... ) and well-defined flow conditions,
such an energy principle states that among all stationary solutions (one-
phase or two-phase viscometric flows), the actual solution (v,¢) is the
one which minimizes the rate of energy dissipation (or more generally,
the rate of entropy production). Therefore, any unsteady process lS ex-
cluded from the present approach, especially mechanisms involved in the
formation of the actual two-phase flow structure.
Considering the flow through a pipe (0 ~ r ~ R) as an example, mini-
mization of the rate of viscous energy dissipation per unit length of
r
the tube
~ = 21T nv~ 2 r dr (3.12)
0
(where v' = dv/dr) is carried out, accounting for the flow conditions as
constraints upon v(r)
and ljl(r) . Eule.r-Lagrange's equations result
from applying the variational method to a functional which contains La-
grange multipliers associated to the above-mentioned constraints. Assu-
ming, for the sake of simplicity that the system behaves as a newtonian
fluid and that the actual concentration profile ljl(r) can be approximated
by a rectangular one, leads to decoupling of Euler-Lagrange's equations.
This gives (i) the Navier-Stokes 'equation fl5r the well-known two-fluid ve-
locity profile and (ii) a functional differential equation for the relati-
ve fluidity, F = llp I 11 , as a function o!· volume concentration tP This
equation relates values of F and its derivative F' = dF/dljl in
both the core ( 0 ~ r ~ eR , s subscript) and the marginal layer
(8 R ~ r < R , w subscript) , such as
(F' + F') (¢ - ¢ ) - 2(F -F) = 0 (3. 13)

s w s w ~ w
Eq. (3.13) can be solved in the two following limiting cases which
correspond to the one-fluid limits S+ 1 and S+ 0 (i) if ~w + 0, i.e.
for a particle free marginal layer and (ii) if ~s + ~M' i.e. for a
packed core (with zero fluidity).
(i) Particle free marginal layer. Thus, Fw = 1 (since ¢w = 0) and (3.13)
reduces to
2
F's - ki +- ( 1 - F s )
¢
=0
where
kl = lim (- dF w/d¢ )
w (3. 15)
¢-+0
w
1. e.appears as the intrinsic viscosity yet introduced in (3, 4), It can

be calculated from theoretical approaches carried out in the high dilu-
tion limit (for example, the Einstein's result for neutrally buoyant ri-
gid spheres, k1 = kE = 2.5). Integration of (3.14) leads to
(subscript s omitted) (3' 16)

38 D. Quemada
where K is an integration constant.

In semi-dilute systems ~ 2 << 1 , the result of OLIVER and WARD
(1953) is recovered
= (3.17)
Eq. (3.17) was found in fair agreement with different data, up to ~ =

0.25- 0.30 , i.e. in better agreement than (3.3), although these two
equations are formally equivalent in the Einstein's limit. Such a diffe-
rence could be understood through dif£erence in accounting for particle
interactions through (3.3) or (3.17). The value of K in (3.16) can be
taken from the coefficients k2 of the ~2 terms in the series (3.4).
The Batchelor and Green's k2 value ; k2 = 5.2 ± 0.3 leads to K=
1.05 : 0.3 .
For higher concentration one can determine both k1 and K from
'experimental data. Alternatively one can find K = K(k 1 ) - for a given
·k1 value - from a packing concentration limit, ~ + ¢M' for which
F(~ ) '"' 0 (3,18a)

M
Moreover many systems exhibit a smooth approach to zero of the slope

dF/d~ as ~ + ~ , such as
M
lim (dF/d~) + 0 (3. 18b)

~+q,
M
Using these conditions (3.18a and b) leads to K = ~2 and
= 2/~
M
(3.19)
(3.19) appears as a rather surprising result, since it relates the high

dilution limit k1 and the packing concentration. (In fact as we shall
see later, k1 in (3.19) is an effective intrinsic viscosity hereafter
named k which involves high concentration effects associated to
aggregate shape and suspending fluid trapping . Thus, k will be a ¢-de-

pendent structural variable, with k(O) = kE for spheres and k(¢M)=2/¢~
Eq.(3.19) leads to the very simple form
F = (1 - .lk
2
¢)2 ::: (1 -!
¢M
)2 (3.20a)
that is, for the relative viscosity
1
nr = ( 1 --
2
k ¢)-2 = (1 -1 )-2 (3.20)
¢M
Notice that (3.20) appears as a special case of the KRIEGER-DOUGHERTY

equation (3.10), with A= k/2.
In the vicinity of the packing concentration ¢M , such expressions
call for considering the "solidification" of a suspension, either as a
kind of phase transition of liquid-solid type (with a critical exponent
2 for the viscosity) or, may be preferable, as the sol+gel transition
observed in gelation of colloids.
(ii) Packed Core

However, increasing the feed concentration ¢ and before reaching
the above-mentionned sol-gel transition in bulk, a percolation transition
could occur if the core concentration ~s reaches the packing value ~M
at ¢* . Nearly above ~* , one has a very thin axial core with ¢ s "'
¢M and a very thick peripheral layer with ¢w ~ ¢ . As ¢ grows, the
core thickens at the expense of the peripheral layer. This percolation
transition is therefore of the same type than DE GENNES (1979) postulated
for suspensions of non-interacting hard spheres. Nevertheless such a tra~
sition is believed greatly promoted by attractive interactions between
spheres, the higher the latter, the clearer the evidence of this transi-
tion. At ¢ ~ ¢* , Fs = Fs (¢M) = 0 , and if one assumes that F's =0 (as
in 3.18b), (3.13) becomes
F' (¢ - ¢ ) + 2 F = 0 ( 3. 21)
w M w w
40 D. Quemada
, the solution of (3,21) is
4>w· 2.
( 1- -!
Fw = (3.22)
~M
which gives the fluidity of the suspension in which the core (as an infi-
nite aggregate in the flow direction) is suspended. Assuming that, at
• < 4>* , (3.17) holds
(3,22a)
an approximate value of ~* can be obtained by equalling (3.22) and (3.22a)

which gives as expected a ~M-dependent value of ~*
= 0,11
•* =
For spheres, for instance, one finds 4>* for close packing
for random close packing (•M = 0.637). Criti-
•*
(•M = 0.74) and 0.26
cal values of between 0.15 and 0.25, have been observed which de-
•*
pend on particle to pipe diameter ratio (KARNIS et al, 1966). Better esti-
mation of could be obtained using (3 .16) if K value was more preci-
sely known.
(iii) Discussion
As (3.22) is formally identical to (3.20a) this equation is believed
to have some general applicability. Furthermore, these expressions F.
~
=
(1- •/4>M) 2 , for i = s or w, verify the general equation (3.13).
Indeed, fitting (3.20) on various data using different values of k leads
to a satisfactory agreement, as shown on Fig. 3.4.
Nevertheless, more precise fittings give very interesting results.
Fig. 3.5 shows the comparison of (3.20) with the experimental data o~
EILERS (1941) on polydisperse suspensions of asphalt bitumen particles

(1.6 ~m < 2a < 8.1 ~m) and with that of VAND (1948) on nearly monodisper-
se suspensions of glass spheres (100 ~m < 2a < 160 ~m). Eq (3.20) is
~rei
")rei 100 I
I
100
10
.5 .9
Fig 3.4 Relative viscosity n Fig 3.5 Comparison between
versus volume fractioK Mooney eq(3.9) and
~ from different data. "Optimum Dissipation"
Curves from eq.(3.20) eq (3.20) fitting Vand's
with different k1 va- and Eiler's data.
lues.(From QUEMADA, 1977).
found in very fair agreement with the Eiler's data but not with the Vand's
one . It has been postulated (QUEMADA, 1977) that such a discrepancy
could originate from the difference between the true concentration ¢
and the effective one ¢eff , this difference resulting from trapped
fluid by transient doublets, triplets .... In polydisperse systems the
voids between large spheres are filled by smaller ones, that results
in a negligeable fluid trapping, i.e. ¢eff ~ ~ . Then a good fitting is
expected using a constant k1-value, close to kE . Morever, close pac-
king concentration can reach values higher than 0.74 (for monodisperse
spheres), up to ~M ~ 1 (in the limit of complete polydispersity). On the
contrary, monodisperse systems exhibit ¢eff larger than ~ , the ratio
¢eff/¢ depending on ~· As the model involves the actual volume of
42 D. Quemada
"particles", eq. (3.20) holds using this effective volume fraction cpeff
(with the suitable shape factor k1 for instance k1 = 2.5 for spheri-
cal aggregates). Nevertheless, cpeff is unknown and it is possible to
use cp in place of cpeff , introducing a new (concentration dependent)
coefficient (i.e. an effective intrinsic viscosity) k = k1¢eff/cp such as
(3.23)
such a ¢-dependence has been introduced in Vand theory (see above 3.11b)
accounting for shape factor and life-time of multiplets formed by colli-
sions. More precisely, fitting on Vand 1 s data (up to cp ~ 0.4) the k va~
lues deduced from (3.20), k = 2(1 - n-1/2)/cp
r
, gives k = 2.510 (1 +
0.91 ¢),where b1 = 0.91 is close to the Vand's theoretical value
1.08, and with k1 = 2.510 close to the Einstein's theoretical value for
spheres. Moreover, for cp ~ 0.4 , effect of doublets appears as the main
contribution to viscosity, since (3.23) reduces to a linear variation,
k
* +m:0.525
k :3.81
Fig. 3.6 displays these results.
• s The same fitting on Eiler's da-
3 ta gives k = 2.504 (1 + 0.03 cp),
• N hence with k1 very close to kE
•
*
and b1 << 1 . (i.e. negligeable
+m:0.637 trapping). Similar results have
k:2.5(1+ 0.8.) k:. 3.16 been obtained from data of
SAUNDERS (1961), including sur-
2.
+m:0.74
k :2.70
* factant layer effects, (QUEMADA,
1978a). On the other hand above
cp = 0.40, as shown on Fig. 3.6
the variation k(cp) changes for
4> stirred (S) or unstirred (N)
2~----------~~---------.
0 0.5 suspensions. The former gives
Fig, 3,6 Intrinsic viscosity vs.
k-values linearly increasing
concentration, from Vand's
data, Packing concentrations (i.e. negligeable triplet and
cp and corresponding intrin-
multiplet formation under
s~c viscosities k = 2/cpM.
stirred conditions) up to the value kc = 2/~C = 3.81 corresponding to

cubic packing ($C = 0.525) which is required to allow relative slipping
of spheres. For the latter, k decreases and tends to approach the random
close packing value, ($RCP = 0.625 , kRCP = 3.20) or the close packing
one (~CP = 0.74 , kCP = 2.70).
Coming back to Fig. 3.5 one can see fittings of the Mooney's equa-
tion (3.9) on Eiler's data and Vand's one. They are shown in moderate
agreement, with the reasonable packing value A-l = 0.70 for the Vand's
data, but the unacceptable value A-1 = 1.43 for theEiler's one. Diffi-
culties in applying the Mooney equation, have been yet stressed
(GILLESPIE, 1963) .
The above discussion has shown the importance of the intrinsic vis-
cosity k in (3.20) as a structural parameter, noticely in relation with
the packing fraction $M . In fact, it must be stressed that in general,
including the case of suspensions of deformable particles, the actual
structure of the suspension at a given $ - i.e. actual levels of parti-
cle aggregation and deformation - can be described by k($) or by an ac-
tual packing concentration
(3. 24)
which would be reached if particles at $ were packed without changing

their actual (aggregated and eventually deformed) state, leading to in-
finite viscosity. It must be kept in mind that this actual packing $p
will be usually different from the true one, $M , owing to structural
modifications which would occur during the (true) rising of the concentr~
tion. Therefore, the $-dependent value , $p =$p($) must verify

~p ( ~M) = ~M • It appears closely related to the inverse of the "ralati ve
sediment volume", as ROBINSON (1949) introduced.
The previous detailed discussion has been given to stress on

(i) the large applicability of eq. (3.20) as a very simple viscosity
equation,
44 D. Quemada
( ii) the importance of the k coefficient as an "effective intrinsic

viscosity" in which effective particle volume effects are included,
which are under influence of many factors (particle size and shape,
fluid trapping, ••. ), all of them being involved in the effective parti-
cle volume veff ,
(iii) the concentration dependence of k which results from (ii)
and finally
(iv) since veff depends on flow conditions, the possibility of exten-
ding ('3. 20)to the ·non-newtonian range through a shear-dependent intrinsic
viscosity k = k( y). (See later, § 4)
3.3.2. Empirical equations

Number of empirical equations n
r
= nr (~) have been proposed. Three
of them which give very good data fitting (and which support eq. (3.20) )
will be given here :
(i) Eiler's experiments(1941) were found fitted by the Eiler's equation:
with empirical packing value b = 0.78.

(ii) MARON et al (1957), by fitting the Ree-Eyring* (non-newtonian) equa-
tion on their data, found
nr = (1 - K ~)- 2 (3.26)
(iii) CASSON (1979) recently proposed the relation
= 1 - a p
(3.27)
nr (1 - 1.75 a ~)2
*See eq. ( 5 . 7)
where a is a "material parameter, defined as the ratio of the effecti-

ve hydrodynamic volume of the particle to its actual volume". Therefore
a appears as the voluminosity first introduced by HOUWINK, (1949) and
it is closely related to k in (3.20). In fact (3.27) does not differ
very much from the following expression, for low and moderate ~-values:
= ( 1 _ 1. 25 a P) 2 (3.27a)
( 1 - 1. 75 a <1>) 2
a
1 --<I>
2
Eq. (3.27a) transforms to the form of (3.20), if one takes
2.5 a
k = - 0.5 a <I>
~ 2.5 a(1 + 0.5 a <1> + ... )
Using average value a= 1.07 found by Casson, gives k = 2.68 (1+0.535<!>),

a result in agreement with those given above, noticely, k(<j>=O) = 2.5 a ,
i.e. a= <l>eff/<1> as expected. Values close to k = 2.7 were observed by
PAPIR et KRIEGER (1970) for suspensions of monodisperse colloidal spheres.
Moreover, the "crowding factor" (1 - 0.5 a cp) is very close to the one
(1 - 0.61 <j>) calculated by Vand (1949) from hydrodynamic interactions
between spheres.
3.4. Application to blood and RBC suspensions
The models given above were extensively used to fit blood and RBC
suspension data, obtained from artificially prepared samples*, since nor-
mal hematocrit is about 40 - 45 •
As hematocrit is not the true volume concentration <1> , a difficulty
arises for applying model equations to data. In fact, as it has been
pointed out, hematocrit measurements depend not only on centrifugation
* Separating RBC and plasma from centrifugation, removing the "buffy coat":
(platelets and white cells) and mixing remaining RBC and plasma in the
desired proportions.
46 D. Quemada
duration and speed (what can be fixed once for all) but also on particle
deformabili ty (and aggregabili ty). Therefore, a "corrected hematocrit"
H' as the true volUme fraction ~ has been defined. Under normal condi-
tions, this leads to ~ = 0.97 H , but to ¢ = 0.60 H for hardened cells
(CHIEN et al., 1971). This can be understood as follows. Let N be
number of particles suspended into the total volume V . Thus ~ = N~ /V ,
p
~ being the particle volume. After packing by centrifugation, the pac-
p
king fraction is~M = N ~ /H V , leading to
p
cp -2
= H ¢M (1 - .J! H) (3 .27)
cpr
Data are generally given using corrected hematocrit (See Fig. 3.1 for
instance).
At moderate concentrations (up to H ~.30) MAUDE and WHITMORE (1958)
verified the Oliver and Ward's equation (3.17) with k a 1.7- 1.9.
BAYLISS (1952) found that the same equation (3,17) holds for plasma or
serum using k = 0.06 ± 0.01 (gr/100 ml)-1 .
As hematocrit in pathologic situations varies about from,25 to
~0, need of an equation which covers the total range of variations in
H
led several authors to use equations for highly concentrated suspensions.
The more commonly used is the Arrhemius equation (3.6) which leads to a
linear dependence of Log n vs H. However, Fig. 3.1 shows that such an
r
equation fails at high hematocrit and at low shear rates. Therefore, po-
lynomial fitting has been performed. As an example, CHIEN et al. (1966)
gave fifth power polynomial fitting the data shown in Fig. 3. 1•.
For hardened RBC, the Brinkman-Roscoe equation (3.8) approximative-
ly holds as shown on Fig. 3.7 (from COKELET, 1972). The curve :
n = (1 - ~ )-2.5 (3.28)
r ~
proposed by LANDEL et al. (1965) using ~ =,635 fits the data of CHIEN
et al. (1967) and (1971) on the one hand and the data of BROOKS et al.
(1970) on the other hand.
1000 k:3.64
k:3.43
Fig. 3.7
Relative viscosity vs volume concentration
for hardened cells
• data of CHIEN et al. ( 1967)
..-data of BROOKS et al. ( 1970)
o data of CHIEN et al. ( 1971 )
curves--- eq. (3.20)
------- eq. (3.28a with~= 0.635 (LANDEL et al, 1965).
DlNTENFASS (1968) generalized the Brinkman-Roscoe equation (3.8) to

the form n = ( 1 - K "( 4> )- 2 · 5 , where '! is the Taylor factor, defined
r
in (3.5). and K a "packing coefficient or a coefficient of fluid
48 D. Quemada
immobilization". Then assuming K ~ = H (i.e. K = from (3.27)),

he wrote
(1 -'rH)-2.5 (3.29)
The validity of eq. (3.29) has been tested by DINTENFASS. However,

this proof is only qualitative as it results from superposing calcula-
ted and experimental curves given by DINTENFASS (Fig. 3.7 and 3.8 in
DINTENFASS, 1968 , as an example).
Coming back to the data for hardened cells shown on Fig. 3.8 , one
obtains better fittings using (3.20) than using (3.28). Moreover, diffe-
rent k values are related to the different suspending media (k ~ 3.43
. _,
1l !CP) (1\) ll ( g/'OOnt)
cprden (g/
0 2 3 4 5 6 1 8 9 11
Fig. 3.8 Viscosities of serum protein solutions. Data from

HARKNESS (1971). 1
Curves n = (1 - 2 [n] C)- 2 with fitted values
[~ (gr/100 ml)- 1 ; C = protein concentration (by
weight), (gr/1 00 ml).
in saline and in Ringer, k ~ 3.64 in water) that could be related to

small changes in the erythrocyte volume due to a slight swelling 1n
water and as a consequence in erythrocyte deformability.
As a viscosity equation valid for any suspension , eq. (3.20) must
be also valid for plasma and serum, as suspensions of proteins. Indeed,
this validity can be observed using data from HARKNESS (1971), who stu-
died the variations of viscosity of serum and of albumin, globulin and
fibrinogen solutions. As volume concentrations of particles are not known,
the product k ~ in (3.20) is replaced by the equivalent product [n]C ,
using the protein concentration by weight, C , and the (true) intrinsic
viscosity, [n] measured in (gr/100 ml)- 1 • Fitting by (3.20) with dif-
ferent [n]-values are shown in good agreement (Fig. 3.8). For plasma
and serum, such values agree with those given by BAYLISS (1952) who
found [n] = 0.06 :!: 0.01 (gr/100 ml)- 1 , using (3.17) with 2 ~ C~ 9
(gr/100 ml).
Although such a general agreement would suggest to give more at-

tempts to applying (3.20) to more data, the importance of non-newtonian
effects in blood rheology calls for studying now variations of n as
r
a function of y at constant ~ , that is the subject of the next sec-
tion.
50 D. Quemada
4. BLOOD AS A STRUCTURED (SHEAR THINNING) FLUID: (1) Effects of RBC

Aggregation and Deformation .
In the previous chapter, strong changes in variations of nr(H) when

the shear rate is lowered, has been shown on Fig. 3.1 . Analysis of the
specific effects the RBC Aggregation and RBC Deformation have upon steady
viscosity will be reviewed. Some models for rheological characterization
of blood will be discussed in the next chapter.
4.1. Variables governing rheological properties of blood and RBC suspen-

sions under steady conditions: (2) Variation of n with shear rate at
given hematocrit.
Shear thinning behaviour of blood, at constant hematocrit in the
range 40-50%, is a very marked effect, as shown on Fig. 4.1. from BROOKS
et al., 1970. Suspension of normal RBC 1n saline (NS) exhibits a lower
Fig.4.1. Comparison of the relative vis-

cosities of normal red cells suspended
in plasma and isotonic saline. NP : red
cells in plasma H = 53,5%; NS : red cells
in saline H = 52,S%~(From BROOKS et al.,
1971).
viscosity at low shear rate and, on the contrary, a high shear viscosity
higher than the corresponding blood viscosity (NP).
Effects of RBC Aggregation and Deformation 51
Fundamental determinants of blood viscosity were found by S. CHIEN

(1970). He compared non newtonian behaviour of the following RBC suspen-
sions at same Hematocrit (H=45):
(i) NP = normal RBC in plasma, where fibrinogen and ~-globulins promote
the reversible RBC aggregation by molecular bridging to form rouleaux;
(ii) NA = normal RBC in albumin-Ringer solution where no detectable aggre-
gation can be observed;
(iii) HA = hardened RBC (by glutaraldehyde) in albumin-Ringer solution.
The albumin concentration was chosen to give a suspending fluid vis-
cosity (nF) equal to the plasma viscosity (np = 1.2 cP), that is to get
same shear stress at given y .
Comparing viscosity dependences on shear rate (Fig.4.2) allows to
1Jr
1000
100
10
}............_ _....,._ _......._,---+-~·--·,.._ ____ 1_!!ec_-_,J_

.01 •1 1 10 100 1000
Fig.4.2. Relative apparent viscosity vs. shear ra-
te in three types of suspensions (H = 45%, np = 1.2 cP)
NP =normal RBC in plasma, NA =normal RBC in 11% albu-
min-Ringer, HA =Hardened RBC in 11% albUmin-Ringer.
(Adapted from CHIEN, 1970).
separate two fundamental processes in blood rheology:

52 D. Quemada
a) Reversible RBC aggregation at low shear rates, which gives alar-

ge increase of NP v~scosity in comparison with NA viscosity.
b) RBC deformation and correlative orientation at high shear rates~)
which leads to a same deformation and orientation of single cells in NP
(complete dispersion of rouleaux being achieved as in NA), hence to the
same reduction in NP and NA viscosities compared to the higher value for
HA suspension where such a process obviously cannot occur. Interpretation
of RBC deformation-orientation as the process which reduces the high
shear viscosity is confirmed by results shown in Fig.4.1:at high shear
rates, NP viscosity is lower than NS viscosity because of a lesser RBC
deformation in saline than that would be observed in plasma, since the
viscosity of the latter (nF ~ 0.9 cP) is lower than the viscosity of the
former ( np = 1 . 2 cP) .
These two fundamental processes can be studied directly by optical
means. Beyond microscopy, methods using light transmission (SCHMID-SCHON-
BEIN et al., 1973) or laser backscattering (MILLS et al., 1980) provide
quantitative measurements of RBC aggregation and deformation, both under
steady or transient conditions, in correlation with stress measurements.
Such observations support the great influence that the "structural
state" of blood has on blood viscosity. More precisely, they show that
the levels of RBC Deformation-Orientation, on the one hand, and of RBC
Aggregation on the other hand, appear as structural variables which
would allow us the rheological characterization of blood. Main factors
governing these processes will be briefly reviewed now following S.CHIEN
who gave moreexhaustive studies (S.CHIEN, 1972, 1979).
4.2. Factors governing the RBC Deformation-Orientation

RBC Deformation-Orientation is mainly controlled by the following
factors:
1) the shear stress a= npY that the plasma exerts upon the RBC. It
depends on plasma viscosity np and applied shear rate y. Raising np in-
(+) nevertheless small enough to avoid any hemolysis effects or the on-
set of turbulence.
creases RBC deformation, thus RBC alignment in the flow direction, that
leads to a decrease of relative viscosity. This decreasing in nr has
been observed using RBC suspended in saline, to which was added Dextran
at increasing concentrations (S.CHIEN, 1971). Raising t leads to similar
effects, i.e. increases deformation and orientation. These effects hence
constitute the major cause for shear-thinning in the moderately high
shear range (i.e. t above about 20 sec-1). In this domain, RBC membrane
exhibits a tank-tread like motion (M.FISCHER et al., 1978), i.e. a membra-
ne rotation around the internal content of the cell. One must notice that
at t higher than about 200 sec-1 , RBC deformation and alignment are "ery
close to the limits which were reached if it would be possible to arrive
at t + oo , avoiding hemolysis to occur.Therefore, a newtonian behaviour
will be recovered at high shear rate (See Fig.5.3). (MERRILL and PELLE-
TIER, 1967).
2) the hematocrit, which promotes RBC deformation by crowding effects
and RBC alignment, the higher the hematocrit the better the alig~~ent
(H.L. GOLDSMITH, 1968), therefore the greater the effective shear rate
between adjacent cells. Thus these effects are the more pronounced as
RBC has a higher degree of deformability. ,Disappearance of non-newtonian
behaviour at low hematocrit (H ~ 10%) would result from the smallness of
crowding and alignment of cells.
3) the pH of plasma, which changes the degree of cell alignment
through modifications of the RBC shape, and deformability (spherocytes
being more deformable than echinoaytes). For example in the case of
packed cells (H = 80%) at y= 60 sec-1, lowering of the pH from 7.4 to
_6.8 gives an increase in viscosity of the order of 60% (DINTENFASS at al.
1966).
4) the intrinsic deformability of RBC, which essentially depends on
the elasticity of the RBC membrane, the viscosity and the physico-chemi-
cal state of the internal fluid. The latter could be different from an
ordinary solution (a liquid crystal ?). Abnormalities of hemoglobin mole-
cules as in sickle cells (drepanocytes) containing HbSS, whose r~gidi-
54 D. Quemada
ty ( 0 ) is strongly increased under low pressure in oxygen, leads to very

high values in viscosity (DINTENFASS L., 1964; CERNY L.C. and al., 1974).
Furthermore RBC deformability was found dependent on its metabolic state
(noticely its ATP content, a depletion of which leading to an increase in
rigidity). RBC deformability is decreased in the presence of included bo-
dies, as cell nuclei or parasite (MILLER L.H., 1971).
4.3. Factors governing the RBC Aggregation.

Main factors controlling the RBC Aggregation are listed hereafter.
1) The shear rate t promotes opposite effects:
- on the one hand, aggregate formation by "hydrodynamic" collision
of two particles, located on adjacent fluid layers. These two particles
are in relative motion and then collide, they keep their "contact" over
a finite duration of time, allowing aggregation to take place;
- on the other hand, aggregate rupture under shear stresses exerted
on aggregate by the suspending fluid.
These opposite effects result in a dynamic equilibrium between re-
versible association and reversible dissociation of aggregates. At very
low y ,a tri-dimensional (gel-like) network of rouleaux is formed. As
y is increased, this network is dissociated into single rouleaux, the
size of which decreases, that leads to shear-thinning behaviour at low
shear. (If RBC aggregability lS sufficiently high one can observe rouleau
orientation by shear, giving an additional decrease in viscosity). At
about y
~50 sec- 1 (i.e. about cr = 2 dynes/cm2), all rouleaux are comple-
tely dispersed to single cells, the deformation of which is not very far
from its very high shear limit.
2) Physico-chemical properties of suspending fluid, play a very lm-
portant role through the ionic and macromolecular· contents of this fluid.
Indeed, rouleau formation at low y results from competition between cell
attraction due to macromolecular bridging, the ends of the macromolecular
chain being adsorbed on cell surfaces, and cell repulsion from electrosta-
tic forces. (Remember the lack in RBC Aggregation when RBC are suspended
( 0 ) SS-Hemoglobin is more Vlscous than the normal one, AA-Hb; that re-
sults from Hydrogen bonding and chain cyclization.
in Albumin-Ringer solution which does not contain fibrinogen, see Fig.

4.2, NA). Therefore, RBC aggregation will depend on the chain length
(that is for linear polymers closely related to the molecular weight M)
and on the ionic strength. Substances, as drugs, which modify equilibrium
conditions between these two opposite forces, will act as aggregative or
anti-aggregative substances. For example, Fig. 4.3. shows the decrease in
"Jap(cP)
60
H:45
40 NORHAL BLOOD
id. + SALICYLATE
20
0 ~ (sec-1)
1 10 20
Fig.4.3. Plot of the apparent viscosity vs appa-
rent mean shear rate in stationary flow for nor-
mal and unaggregated blood. (From HEALY and JOLY
1975).
low shear viscosity for blood+ Aspirin (J.C. HEALY et M. JOLY, 1975).
Evidence of this macromolecular bridging can be found in:
(i) rouleau formation in the presence of DeA~ran of different mole-
cular weight, which shows a constant interparticle distance d (between
adjacent cells) -thence which requires a cell deformation (mainly of the
concave part) - the distance d depending on M through the length of the
chain (S. CHIEN et al., 1971 c).
( ii) Adsorption isotherms of Dextrans (BROOKS, 1976) , which shows an
abrupt change in slope when aggregation starts (Fig. 4. 4.),. i.e. a factor
2 in slope.
56 D. Quemada
'?ra IONIC STRENGI'H

0.145M
O.lOOM
0.030M
no ox
(0.100M)
Fig.4.4.rsotherm of dextran 77.6 Fig.4.5. Relative viscosity as a func-

adsorption to human erythrocytes tion of shear rate for washed human ery-
at physiological ionic strength throcytes at the ionic strengths indica-
in terms of the number of molecu- ted. Dextran T 70 concentration 3g/100~
les adsorbed per cell, Na (From hematocrit 50.0 ± 0.2%; T = 25°C. (From
BROOKS , 1976) . BROOKS et al. , 1974).
(iii) Variation o~ RBC aggregation as a function of molecu-

level o~
lar weight M: Low molecular weight Dextran (as Dx 25, i.e., with M ~
25.000) exhibits absence of aggregation owing to Dx 25 is a too short mo-
lecule to overpass electrostatic repulsion. The contrary is observed using
high molecular weight dextrans (as Dx 150).
(iv) Variation of level of RBC aggregation as a function of ionic
strength (BROOKS et al., 1974): the latter governs the spatial extent of
electrostatic forces through the ionic double layer thickness, (KD)- 1
the higher the ionic strength, the thinner the ionic layer (see Fig.
4. 5.).
3) The hematocrit H influences the state of aggregation through the

collision frequency f of RBC. At low hematocrit, f is weak and aggrega-
tion ef~ects are negligible : newtonian behaviour is recovered at H be-
low about 1O% (See. Fig. 5. 3. ) . On the contrary, at high hematocrit , f
is too high and impedes the formation orr·ouleau and blood more and more
ressembles to a suspension of non aggregable particles. This leads to
consider ~BC aggregation as having its greater effects in the physiologi-

cal range of hematocrit .
4) The intrinsic aggregability of RBC depends on (see the above dis-
cussion):
- RBC membrane affinity for macromolecular adsorption
- Electric charge of the RBC membrane
RBC deformability, since rouleau formation requires RBC deforma-
tion.
In order to progress in quantitative interpretation of the rheologi-
cal behaviour in terms of structural characteristics of the materials,
we shall give now a brief survey of main models proposed for non-newto-
nian behaviour.
58 D. Quemada
5, BLOOD AS A STRUCTURED (SHEAR THINNING) FLUID: (2) MODELS AND APPLI-

CATIONS.
Inspection of fundamental determinants of blood rheology, carried

out in the previous section, calls for models as attempts for quantita-
tive evaluation of blood viscosity at different shear rates and under
various conditions. As far as possible, such models must involve these
fundamental determinants as structural parameters.
Before giving some examples with application to blood rheology,
some general models for non-newtonian behaviour will be reviewed.
5.1. Simple rheological models for viscosity of non-newtonian fluids.

For blood, SCOTT-BLAIR (1959) proposed to use the Casson relation
which was developped for pigment-oil sus-pensions (CASSON, 1959). This
relation, which gives the shear stress as a function of y, reads:
( 5. 1)
where cr 0 and K are"materia.J. constants" ,which depend on concentration.;cr0 is

the yield stress,K, the Casson viscosity. In the Casson theory, pigment
particles forming long chains by mutual attraction are for the sake of
simplicity treated as long rod-like aggregates, which are broken down by
shear stresses. At given shear rate y, a dynamic equilibrium is reached.
Casson found semi-empirically an equilibrium value for· the aggregate si-
ze, in terms of its axial ratio J depending both on y and on the cohesi-
ve force F between particles in aggregate. The maximum tension on an ag-
gregate compatible with no-rupture must be proportional to tJ 2 1eading to
J ~ y-t for a given F. Then he postulated the equilibrium J value given
by a linear relation J 'F 61 + 13 (nFy)-!. Then (i) adding energy dissi-
pation in fluid and solid parts.of the suspension and (ii) extending the
result to high concent~ations, he obtained eq. (5.1), with:
K = nF ( 1 - <I>) -q (5.2)
= (a a) 2[ (1 - <P ) - ~ - (5.3)
q
Models and Applications 59
with q = (aS 1 - 1), where a, S1 and S are structural parameters.

Number of phenomenological or empirical relations n(y) have been pro-
posed for general description of pseudo-plastic behaviour. Relations which
contain two parameters, as the well known power lawn =A .yn, can repre-
sent only a part of the involved range of y (Such relations will be not
considered hereafter). On the contrary, using three or more parameters
allows both low and high shear behaviour to be displayed.
Three parameter relations can be written in the general form:
n = + no - noo (5.4)
'' noo F(y)
where no and noo are the limiting viscosities as y + o andy + oo respective-

ly. F(y) is a dimensionless function of shear rate such as F(o)=1 and
F(oo) = oo, Table 5.1. gives some examples of F(y).
TABLE 5.1
Shear-dependence of F(y)
Ref Remarks F(y)
jwiLLAMSON ( 1929) empirical,A = Canst. 1 + Ay ( 5. 5)
. . . 2 4
CROSS (1965) sem1-emp1r1cal, m = 3' 5 1 + Atm (5.6)
REE-EYRING (1955) theoretical (with 2 campo- Ty/ Sinh-1(Ty) (5.7)
nents, T = relaxation time
For suspensions of colloidal spheres (radius = a) KRIEGER AND

DOUGHERTY (1959) gave a relation similar to eq. (5.4) :
n = nl+ n;~- n, (5.8)

1 + bor
where Or = oa 3/ K T (5.8a)
lS a reduced (dimensionless) shear stress, defined from the actual shear

stress o . In (5.8), n 1 and n 2 are the viscosities at the higher and lo-
wer shear stresses available respectively. These values are both dependent
upon the volume concentration~ in the form given by eq. (3.10). In their
60 D. Quemada
theory, Krieger and Dougherty applied reaction kinetics to doublet for-

mation. Nevertheless, they supposed that the concentration of doublets is
very small in comparison with the single particle one, assumption which,
strictly speaking, limits the use of eq. (5.8) to dilute systems.
5.2. Extension of the newtonian model deduced from optimization of

energy dissipation to the description of non-newtonian behaviour.
( QUEMADA, 1978 a).
Attempt to extend to the non-newtonian range the equation (3.20)
originated from the concept of structural intrinsic viscosity k = k(~),
yet introduced in § 3.3. Reversible aggregation of particles was conside-
red as responsible of such a ~ dependence. Moreover, the latter is belie-
ved to result from other ••structural" effects (in a large sense), as
thickness of interfacial layers (ionic double layer, surfactant layer)
and particle orientation, the greater the particle deformability, the more
orientated the particles. As most of these effects are shear dependent,
k must be also shear-dependent. Therefore, as an approximation~ it will be
supposed that validity of eq. (3.20) can be extended from newtonian to
non-newtonian range , assuming that these dependences, k = k(~,y) reflect
all non-newtonian effets. As resulting from dimensional analysis (§2.1)
this extension necessarily gives k = k(~,yr), where the reduced shear ra-
te Yr = Ty involves some characteristic time T.That leads to
(5.9)
Eq.(5.9) is a non-newtonian viscosity equation for very highly concentra-

ted systems. Although (3.20) has been only established in newtonian ran-
ge, such an extension is believed acceptable,since at very high ~ and
moderate flow rate ( 0 ), the two-phase flow model used in§ 3.3. can be
expected still valid, with a quasi-rigid core (where y << ,-1) surroun-
( 0 ) very low and very high flow rates lead to newtonian behaviour
ded by a very thin marginal layer (where y >> T- 1 ), therefore leading to

viscosities close to the (newtonian) low and high shear limits, respecti-
vely (See Fig.1.3).
The validity of (5.9) can be tested using various data. As an example
for RBC suspended in saline,variations of nr vs.H at different given shear
rates are shown on Fig.5.1. They can be represented by (3.20) with cons-
Fig. 5.1. Relative viscosity n, vs

volume concentration ~ at different
k:3 shear rates t [ sec- 1 ] = 671:3.2 (e),
170.8 (•), 1. 708 (t:.), 0.678 (o) (from
2.8 data of BROOKS and SEAMAN, 1971).
Theoretical variations nr=(1-!k~)- 2 ,
leading to a shear rate dependent in-
trinsic viscosity k = k(y).
10
2 4 6 8
tant k, the value of which varies with y but not with H since aggregation
does not occurs in saline.
At given concentration, the intrinsic viscosity k = k(yr) as a struc-
tural parameter will reach limiting values:
k 0 = k(o) and k 00
= k(oo) (5.10)
These values can be approximatively obtained using TY << 1 and TY >> 1

respectively, under steady conditions. In fact, if a uniform angular velo-
city is suddenly applied to a shear-thinning system, this structural para-
meter k becomes time-dependent, until, a steady state, characterized by
62 D. Quemada
(5.9) will be reached. As an exact time dependent description is not

available, a rate equation can be written that expresses the balance be-
tween aggregate build-up due to collisions and aggregate break-down in-
duced by shear. It is assumed that such two opposite processes can be
roughly described by shear dependent characteristic (relaxation) times
LA and LD respectively. For instance, if one assumes that the system
contains N "particles" of two kinds, aggregates and single (dispersed)
Particles, a rate equation which describes the dynamic equilibrium bet-
ween these two populations, can be written
where nA and ~ =N - nA are the number of aggregates and single particl-

es, respectively. One can define kin (5.9) by k~ = kA~A +~~Dusing
the intrinsic viscosities kA of aggregate and ~ of single particle sus-
pensions and the corresponding volume fractions ~ 1. = n.v
1 p
/V, (i = A,D),
where v
p
= single particle volume. At equilibrium
Writing kA = k0 = k(O) = intrinsic viscosity at zero shear rate and

~ = koo = k(oo) = intrinsic viscosity at infinite shear rate, leads to the
shear dependent intrinsic viscosity
k - koo
0
k
More generally, in order to avoid a too complicated description,

one can postulate the existence of a balance equation describing the
time evolution of the structure , as resulting from a dynamic equili-
brium between its building up and its breaking down, in a large sense,
taking these (reversible).processes in a large meaning (such as
orientat~pn ~ disorientation, deformation ~ undeformation, aggregation +

disaggregation). Using k as a structural variable leads to the following
balance equation:
(5.11)
where 'A and 'n are the relaxation times for formation and dissociation
of the structure.
Assuming the existence of such mean relaxation times, the steady
solution of (5.11) is again
k 0 - k IX>
k k<X> + 1 + e (5.12)
where:
e (5.13)
-1
In the case of dilute systems of rigid spheres (radius a), 'A can
be taken as the Browman collision frequency (SMOLUCHOWSKI, 1917) and
'n-1 as the frequency of shear induced collisions (GOLDSMITH and MASON,
1967), that leads to:
3
e -1 ,-1 = ~ 1 2p KT = 4nFa •
'n 1 A 1T 1T n a3
F
KT y (5 .13a)
Eq. (5.13a) exhibits a mean characteristic timeT such as e- Ty, in

aggrement with the result of dimensional analysis (.28), i.e. 8 - yiDrot.
64 D. Quemada
For concentrated suspensions, the shear dependence of 9 = S{y)is un-

known. However, according to the high and low shear limits 9(oo)+ oo and
9(0)+ o, one can take approximatively the following form:
( 5. 14)
where T =a characteristic time of the structure and p an empirical pa-

• -1
rameter. One may define a critical shear rate Yc = T when the order
of magnitude of the corresponding value of the shear energy-as the work
done by shear forces, crca 2= nFYca 2 on a distance comparable to the par-
ticle radius- becomes close to the thermal energy:
(5.14a)
in agreement with dimensional analysis (See§ 2.1)

Eq. (5.9), through (5.14), involves only one relaxation timeT ,
what,could be a too rough approximation since, in most cases, description
of complex systems requires several or sometimes a distribution of relaxa-
tion times. If the relaxation time spectrum is composed of several
"groups" which can be separated into different "flow units", each group
being characterized by a mean relaxation time Ti it will be possible to
use the general model of REE-EYRING (1955) in the form (See eq.(5.7)
table 5. 1):
( 5. 15)
Nevertheless, a satisfactory description of the non-newtonian beha-

viour is very often obtained using only one relaxation time(especially if
the latter is the only one whose reciprocal value lies in the shear
rate interval a~cessible to the measuring apparatus).Therefore, unless if
necessary ( 0 ) the simple form ( 5. 9 - 5. 12) will be kept for the non-newto-
nlan viscosity:
-2
(1 - 21 k~ ) • k =k 00
(5.16)
Using viscosity values at zero and infinite shear, n0 and noo, respec-
tively, transforms (5.16) to:
+ Jno - Jnoo where A (5.16a)

1 + A (-ry)P
Finally, it must be stressed that instead of a shear rate dependent

equation, the same model would lead to a shear stress dependent viscosi-
ty, as in (5.8). Indeed, 1n terms of effective medium, effects of parti-
cles interactions, occuring at high concentrations, can be taken into
account by using in (5.13a), the viscosity of the suspension as a whole
instead of the suspending fluid viscosity. This changes 8 , given in
(5.14) to a reduced shear stress, 8 ~(crr) where Or 1s defined as 1n
(5.8a). Put into (5.12), this new 8 transforms (5.9) to the simple rela-
tion:
Jno- Jnoo ( 5. 17)

1 + A(or)~
very similar to (5.8).

Both (5. 16aland (5.17) are 'related to pseudo-plastic behaviour, as
shown generally by (5.4). However, an important feature of the present
defined as in (3.24), now both~

.
model, eq. (5.9), is the existence of an actual packing concentration,
.
andy dependent. At y = O, for instance,
( 0 ) For instance, if aggregation of particles induced by shear becomes im-

portant, it would be necessary to introduced a new characteristic time T
into the rate constant for association, for instance TA- 1=-rA; 1[ 1+(T~y) 8 ].
'with some new exponent s and where TAo = TA( y = 0) .
66 D. Quemada
there exists a packing concentration ~

'~'Mo
If one increases ~ up to ~Mo the
zero shear viscosity n0 tends to infinite, and the pseudo-plastic behaviour
changes to a true plastic one, at ~ ~ ~Mo in the special case n = 1/2 (see
Chap. 7, and Fig. 5.6 and 5.7).
5.3. Application to blood and RBC suspensions
Most of analysis of non-newtonian data for blood and RBC suspensions
used the Casson equation (5. 1). MERRILL et al. (1965) found that this
equation correlates very well data for normal blood, but not for washed
red cells suspended in fibrinogen free plasma. Fig.5.2 shows variations
of a! versus y! (Casson plot).
a' 112 (dyneslcm2)1 12

.4
. ! .,
Flg. 5. 2 o vs y ror whole blood •
(H=41.9) and RBC suspended in de-
fibrinated plasma e (H=40.6).(From
MERRILL et al. 1965).
~ 1/2(sec-1/2)
0 1 2
Moreover Fig.5.3 from BROOKS et al (1970), shows same variations for

normal blood at different hematocrits. Fair agreement is observed, exclu-
ding parts of data corresponding to very low and very high shear rates,
more especially as H is large (for instance 1 ~ y! ~ 20 sec-! for normal
hematocrit). Pseudo-plastic behaviour results from no measurable yield
a tr 1/2(clynestcm2)112 H: •. 674
10 20
Fig.5.3Casson plot o'=f().!) for a blood with ACD

anticoagulant. Data of BROOKS et al-(1970) at dif-
ferent hematocrit. (Data points for plasma and
H = 8.25 RBC suspension are not shown for clarity:
these fluids appear to be newtonian. Casson law
(4.1) is satisfied except close toy= 0, where
pseudoplastic behaviour is observed (From G.R.
COKELET, 1972).
stress at low shear rate. At high shear rate, transition towards newto-
nian behaviour (which slightly appears on Fig. 5.3 for H = 67.4) has been
observed (MERRILL et al., 1967).
Few other attempts have been performed, as the use of an empirical

shear dependent Taylor coefficient inserted in the Brinkman-Rascoe equa-
tion (3.8) (DINTENFASS, 1971). However, as this shear dependence is assu-
med to result only from a shear dependent internal viscosity of the cell,
such an attempt is very questionable.
Applying eq. (5.16) first needs to know the value of p, the empiri-
cal determination of which can be obtained as follows. From data,
nr = nr(y) at given hematocrit, the effective intrinsic viscosity can be
calculated as:
(5.18)
68 D. Quemada
.
and then it can be inserted 1n (5.12), written as a linear function of
logy :
p log y+ p log t (5.19)
As an example, taking k(y) from the data of BROOKS and SEAMAN (1971)
for RBC suspension in saline, one obtains variatjons of Z =(k 0 -k)/(k-k00 )
against y, in logarithm coordinates, using koo = 1.80 and different k0
values (See Fig. 5.4). The linear variation is seen when k 0 = 3.60 with a
· Fig.5.4 Graphical representa-

tion of eq. (5. 19), using koo=
1.80 for different values of
1 k 0 : 3.24 (o), 3.42 (6), 3.60 (+)
3.96 (x), 4.50 (.).(From QUE}~DA
1978 a)
1 10
slope p close to 0.5 . More precisely, the best fit linear regression
eq.(5.19) is found for the following values:
t = 0.43 sec, p = 0.47
Number of such comparisons, especially for normal blood data,

yielded different values for k 0 , k 00 and t but 0.4 ~ p ~ 0.6 . Thus for
simplicity, the value p = 0.5 has been chosen. On the contrary, since p
values very close to 1 have been found for uniform colloidal spheres
(QUE~DA, 1978 b), values p # 1 where believed,in suspensions of non-
spherical particles, as resulting from polydispersity or/and particle sha-
pe. Indeed , in such systems, the non-linearity in y
dependence might re-
sult from variable angular velocities of particles, which spend more time
aligned with the flow than normal to it. (Alignment of particles means
alignment of faces for simple RBC as a disc, or alignment of axis for a
rouleau as a rod; notice that suspended discs and rods exhibit same pe-
riod of rotation under shear if their equivalent axis ratios are recipro-
cal (GOLDSMITH and MASON, 1967)).Furthermore, as we will see later(§ 8),
such a non linearity could also originate from the smaller value of the
shear rate in the core (where blunting of the velocity profile occurs)
than the value of the applied shear rate (that is the value which enters
ln data fitting equations).
As an example, result from fitting (5.16), with p =~,on the data
of MERRILL et al. (1965), is shown on Fig.5.5 (See Fig.5.2 for compari-
son). It must be stressed that the lack of any yield stress for blood
1
(1"2 I
t (dynes/cm 2)2 Fig.5.5 Non newtonian behaviour of
blood (D) and r~d cells in defibrinated
A
plasma ( •): Data from MERRILL et al.,
1965. (See Fig.5.2). Theoretical varia-
tions, according to eq. (5.17) with the
fitted parameters.
<P k"' ko 1"
Blood 0.419 1 .80 4.68 1.45 ---

RBC in
Defibr. 0.406 1 .80 3.96 0.01 --
Pl.
.1 1
does not mean that some true yield stress would not exist, but that if
it existed, it would be very smaller than the cr 0 value given by the Casson
plot.
The Casson equation can be recovered from (5.16) with p =~in two
limiting cases (QUEMADA, 1978 b):
a) In the "high shear rate" limit, TY » 1, (See Fig.5;6},if volume
70 D. Quemada
cr 112
~112
Fig.5.6 Recovering of the Casson Figo5o7 Recovering of the Casson

equation in the high shear rate li- equation in the very high concen-
mit of (5o 16) , TY » 1 (with a tration limit of (5.16), ~ + ~Mo
pseudo-yield shear stress, a~ )o (with a true yield shear stress,
a~ )o
fraction ~ is less than the packing value for particles .at rest .~M0=2/k0
Eq. (5o16) then turns into (5o1), written with a "Casson" viscosity K'
and a pseudo-yield stress a~ such as :
(5o20)
(5o21)
which are closely related to eqo (5o2) and (5.3) respectively. In fact
both eq. (5.2) and (5.3) apply to concentrated suspensions of deforma-
ble particles, like RBC suspensions or Blood, whose close packing con-
centration of disperse particles ~Moo= 2/kco is close tounity, i.e. koo"'2.
Moreover, MERRILL et al. (1965 a) found the exponent q in (5.2) and (5.3)
close to q = 2 (average value q = 1.97), which agrees to the exponent
value in (5.20) but that leads to a~~ ~ 2 (1-~)- 2 instead of ao ~ ~2(1-~)-4
in ( 5. 21) o
b) The pseudo-plastic behaviour can change into a plastic behaviour

if at rest the suspension form a very loose network of aggregates charac-
terized by a relatively low "packing" value ~M 0 .Then, increasing the
volume fraction, a ·~igh concentration" limit ~ + ~Mo= 2/k0 can be rea-
ched. In this· limit, the medium at rest forms a three dimensional network
like a (physical) gel, for which n0 -+ co (Fig. 5. 7. ), which still leads to
(5.1) with the same Casson viscosity (5.20), i.e.:
kco )-2 (5.22)

ko
and a true-yield-stress given from (5.11), by:
ai)
.
= nooYc (5.23)
As evidence a; is the shear stress required to disrupt the network struc-

ture (as in a shear induced gel-+ sol transition). Since the presence of
a yield stress is associated to the existence of a partial plug flow in
a tube, this high concentration limit.is believed to be reached inside
the core of a two phase structure. Therefore, the present model relates
the plastic behaviour to the packing transition. It has been suggested
(QUEMADA, 1981) that the same model associates the onset of non-newtonian
effects (i.e. pseudo-plastic behaviour) to the percolation threshold in-
troduced in §3.3 .• We shall return on this discussion in Chapter 7.
At y = Yc , the shear stress takes the critical value ac, such as:
(5.23a)
This result leads to a very simple graphical method for Yc determination

from Casson plot (See Fig. 5.7).
The relations (5.23) and (5.23a) hold approximately in the case of
pseudo-plastic behaviour, provided that no >> nco •
Interpretation of k 0 , ~ and T as parameters for rheological charac-

terization of blood can be deduced from analysing the data of CHIEN des-
cribed in§ 4.1. (See.Fig.4.2). Data fitting of eq. (5.16),with p = ~,
leads to values given in table 5.2 •
72 D. Quemada
TABLE 5.2 Rheological parameters for non-newtonian behaviour.
~ = 0.45 l1p = 1.2 cP

k"' ko 'sec ~:t-1o
NP 1. 78 4.20 0.2 0.476
NA 1. 78 3.29 0.04 0.608
HA 3.62 - - 0.552
ko "'Jr keD
4.21 NP
3.62
3.~1
1()-2 1 t 102
.2, .• 4 ~-~
~(M~)
Fig.5.8 shows the corresponding calculated curves. Such values sug-

gest the following interpretations.
(a) kco reflects RBC deformation and orientation, since (i) NP and
NA suspensions at -ry>>1 contain only single cells which experience same
deformation and orientation (because ~ and np have same values in both
cases) and (ii) these effects are quiteabsentfrom those in HA suspension
thus leading to k~p * k~A < k~ .
(b) k~P reflects RBC aggregation since, at TY << 1, (i) neither
shear deformation nor orientation of cells occur in NP or NA suspensions,
and (ii) aggregation effects appear in NP in comparison with NA, as the
latter from which plasma proteins (fibrinogen and globulins) are absent
does not undergo cell aggregation, that leads to k 0 NA < k NP

0
(c) Packing concentration, as ¢ = 2/k 0 for the NP suspension
Mo
¢M~A = 0.608, very close to the one, ¢ 0 = 0.61 ± 0.01 obtained for cen-
trifugal packing of cells hardened in acetaldehyde (CHIEN et al., 1971).
As deformation of normal cells by crowding can be expected small enough
to be of same order than deformation of hardened cells under high speed
centrifugation, such an agreement seems significant (Notice that the
value ¢
m
= 0.59 has been obtained (BURTON, 1966) for geometrical close
packing of discoids modeling underformed RBC). It has been proposed
(QUEMADA, 1981) that the value ¢MNP~ 0.48 for NP suspension could re-
sult from rouleaux packed to form a (loose) tri-dimensional network, the
mean mesh length of which containing about ten cells.
(d) Indeed, for suspension the packing concentration ¢~A = 2/k~
i.e. the packing value one would obtain if cells were packed keeping
unchanged their actual state (i.e. keeping their actual effective volume
unchanged) -takes the value 0.552 . This value is in close aggreement
with the value which has been theoretically found for rigid spheroid sus-
pension (same axis ratio than RBC) and which has been defined as the
critical concentration above which "the particle cannot undergo unimpeded
rotation because there is unsufficient space to move"(GOLDSMITH and MASON
1961). Furthermore, such a lowering in ¢MHA ln comparison with ¢;.[6A=o. 61
can be interpreted as a dilatancy effect, yet observed by SCHMID-SCHON-
BEIN (1915), on suspension of rigid cells, as shown on Fig. 5.9 . Data
fitting of (5. 16) for rigid cell suspension exhibits a strong shear thic-
kening effect, k~ > k~, resulting from dilatancy, while normal and crena-
ted cell suspensions show shear thinning effect, nevertheless with
k 0 ~kN0 and kC>kN
00 00
i.e. both reduced cell aggregability and deformability
of crenated cells compared to normal ones.
(e) Relaxation times ,NP ~ 0.2 and ,NA ~ 0.04 sec are more difficult
to be interpreted. As RBC (and a fortiori rouleaux) are too large parti-
cles to exhibit any Brownian effects, these times could be thought as
Maxwell relaxation times for non rigid particles, Ti 'V ni/Gi , involving
some internal or membrane viscosities ni and some elastic modulus Gi .
74 D. Quemada
")ccPJ
c
N
tf(sec-1)
0~~------~--
j ------~----
10
Fig.5.9 Red cells suspended in plasma.
Effect of RBC deformability, Data from
SCHlv!ID-SCHON"<EPI ( 1975) . Curves for
fitted values e£' rheological parameters
as follows.
<P ko koo Tsec 2/k0 2/koo

Normal N 0.40 4.88 1.99 1.08 0.41 1.01
Crenated c 0.40 4.56 2.62 0.78 0.44 0.76
Rigid R 0.39 3.23 3.97 1. 34 0.62 0.50
Values about 0.2 sec can be obtained from the data of SKALAK (1976) for
shear and dilatation deformations of the RBC membrane. Although single
cell deformability probably dominates the rouleau one, the agreement with
TNP seems non conclusive, unless the lower value TNA could be explained
through membrane property modifications associated to albumin molecules
adsorbed in the membrane.
The above discussion shows that close correlations can be expected
between the values of the rheological parameters and the fundamental
determinants of blood rheology, i.e. at given H, essentially RBC aggrega-

tion, RBC deformation and plasma content. Some clinical applications will
be given in Chap.8.
76 D. Quemada
6. BLOOD AS A THIXOTROPIC-VISCOELASTIC FLUID: EFFECTS OF MECHANICAL AND

PHYSICO-CHEMICAL PROPERTIES OF THE RBC MEMBRANE.
Up to now, we have been concerned with steady properties of suspen-

sions. Shear-thinning properties that most of concentrated disperse sys-
terns exhibit have been related, in the last section, to dynamic equili-
brium between the formation and the destruction of some internal ("micro-
scopic") structure. Moreover, time dependent behaviour observed in the
apparent viscosity not only involves this time dependent "microscopic"
structure (as aggregation or deformation-orientation of particles), but
also,as it will be seen in the next section, the time dependent collecti-
ve ("macroscopic") structure (as the annular two-phase flow in pipes) 1
which results from the development of a cell free layer,close to the cup
(and bob) walls (COKELET et al., 1963), for instance.
Since these processes involve characteristic times (as relaxation
times) , s·..:.ch systems will exhibit time dependent properties unrJer unstea-
dy conditions i.e. they are thixotropic materials. However, instantaneous
(reversible) deformation can be supported by the structure, the properties
of which are purely elastic before to be broken. Therefore under steady
conditions, the material possesses both stored energy (by deformation)
and dissipated energy (by creeping flow), what is the general feature
of viscoelastic materials . Additional viscoelastic effects may arise from
elongational deformation and elongational flow. Such effects, which have
been commonly observed in viscoelastic fluid flows through pipes presen-
ting an abrupt reduction in their circular section, can be expected to oc-
cur at the branching of two arterioles (compressible flow) or of two
venules (extensional flow), but seem not studied yet.
6.1. Some experimental results about thixotrop~' and visco-elasticity
of blood and RBC suspensions.

A) Transient experiments (Step function experiments)
As it has been shown in § 1, 2D, superposing thixotropy and visco-
elasticity would leave to different flow regimes, depending on the mean
value of shear rate. Such regimes had been observed with blood and RBC
Blood as a Thixotropic-Viscoelastic Fluid 77
suspensions in coaxial cylinder viscometers. (COKELET, 1972; HEALY and

JOLY, 1975).
Diagramatic representations of these regimes are shown on Fig.6.1
O'"(arbitrary '-lnits)
~:.05
Time
t
I~, Normal Blood

id +Salicylate
I
I
I
i
d t Time
Fig.6.1 Different types of time- Fig.6.2 Time curves of a (from J.C.

curves of the shear stress in res- HEALY and M.JOLY, 1975)
ponse of applying shear rate steps -------- normal blood
wi~h increasing amplitude.
-------- id. + salicylate
. -2 -1 -1)
At low values of mean rate ( Y~10 -10 sec a relatively slow increase
1n a is observed, that ressembles to what results from retarted elastici-
ty. After stopping the viscometer, stress relaxation occurs (Fig. 6.1a).
Increasing the shear rate ( y~ 1o-1-1sec-1) leads to a well-known shape
of curve, called an "overshoot", a general behaviour which can be belie-
ved as resulting from superposition of thixotropy and elasticity (i.e.
non-linear visco-elasticity). (See Fig. 6.1b and Fig. 1.5e). At higher
shear rates (y ~ 1-10sec-1) thixotropy dominates with, at t = 0, an
abrupt rising of a up to a peak value, then immediately followed by a
78 D. Quemada
slow decay down to a lower steady value (Fig.6.1c). Finally, at very high
shear rates (above 10-50 sec-1), the stress a rapidly rises to some equi-
librium value and remains there until stopping the viscometer, and then
quickly drops to zero, what is the common behaviour of pure liquids and
solutions.
Comparison of transient behaviour of normal blood (NB) with that of
disaggregated blood (DB); with sodium acetylsalicylate to disrupt rouleaux,
is shown on Fig. 6.2 a and Fig.6.2 b, (From HEALY and JOLY, 1975). At low
y , the curve shapes are similar, but with an important lowering in stea-
dy apparent viscosity. As expected, thixotropy completely disappears from
(DB), at intermediate y, and newtonian behaviour is recovered at higher
shear rates. (Comparison of steady viscosities of (NB) and (DB) samples
have been shown on Fig. 4.4). Such experiment seem to demonstrate that
RBC aggregation would be the main factor responsible of thixotropic be-
haviour of blood, through the formation of rouleau and of tridimensional
network of rouleaux. Nevertheless, since RBC deformation and orientation
are also responsible of shear-thinning at high hematocrits (See Fig.4.2),
one can expect that some thixotropic e~~ects should result ~rom these pro-
cesses.
Furthermore, several authors (COKELET et al., 1963; CHIEN et al.,
1966: COKELET, 1972) claimed that the interpretation of shear-time curves
would require to ·account for plasma layer formation near the walls of the
viscometer gap (and perhaps cell orientation). Such requirement cannot be
completely discarded noticely in pathological blood when plasma layer ta-
kes shorter times to be formed than in normal blood. Indeed as COKELET
( 1972) pointed out,.many factors are involved in recording the torque-
time curve, namely (1) effects due to acceleration of blood from rest,(2)
changes in fluid structure as the blood is sheared, i.e. thixotropy and
elasticity.(3) dynamic response of the instrument and (4) radial migration
of RBC. Moreover, at low shear rates, sedimentation effects can also oc-
cur (COPLEY et al., 1976; VINCENT and OLIVER, 1976,). Therefore,a complete
interpretation of these time curves, in order to get the relevant flow
properties, still rema1ns an open question.
Use of cyclic tests (as a linear increasing of y with time, up to a

maximum and then decreasing at the same rate down to zero) gives a hyste-
resis (open) loop, which contains more information than the step changes
in y (Fig.6.3, BUREAU at al., 19[8), and has the advantage to avoid some
Fig.6.3 Hysteresis loops obtained for linear increase of y ,

followed by the same linear decrease, for different rates a
a) tmax = 20 sec; a(sec-2): 1 = 5.1o-3 ; 2 = 14.5.10-3
3 = 42.10-3 .
b) tmax :6.5 sec; a(sec-2): 1 = 15.5.10-3; 2 = 45.10-3
3 155. 10-3 .
(Brom BUREAU et al., 197e).
of the above-mentioned difficulties.

B) Dynamic experimentals (Oscillatory experiments)
Taking a shear rate which varies sinusoidally with time leads to the
so-called dynamic response of the system, which can be charaterized by
the (complex) dynamical viscosity~ =nv- inE , nv and TlE being,the Vls-
cous and the elastic components, respectively (See§ 1, 2 D).
Number of experiments showed that nv and TlE are functions of the
fundamental factors governing blood viscosity- namely hematocrit, RBC
80 D. Quemada
aggregability and deformability, physico-chemical properties of the sus-

pending fluid- and the frequency f of oscillations. For instance, CHIEN
and his colleagues gave the variations of nv and nE vs f, at given hema-
tocrit, for RBC in plasma and RBC in Ringer from low frequency measure-
ments ln Couette system (CHIEN et al., 1975). Fig.6.4 shows the main fea-
tures of these variations, noticely the weak effect of cell aggregation
at high H values. Onthecontrary, at normal H, not only nv, as expected,
but nE also, are observed larger if cell aggregation is present than if
it is not present. This likely reflects that the network of rouleaux (in
plasma) has a stronger elasticity than the system of dispersed cells (in
Ringer).
RBC in Plasma RBC in Ringer

'' '
'
' ""
Cell cone.
95%
(P) 10_,
(b) ----45%
f (Hz)
Fig.6.4 Variations of viscous and elastic components of the
complex viscosity, ny (--)and nE (---),versus the frequency
f of oscillatory tests, for normal RBC suspended in plasma and
in Ringer (From CHIEN, 1979).
As blood is a shear thinning fluid, attention has been given to

oscillatory measurements ln the presence of a given shear rate. Changes
in response reflect similar changes observed in the shear rate dependent
stress-time curves from transient experiments. As an example, Fig.6.5
shows variations of the steady viscosity n8 and the viscous and elastic
- I
10-2 ---------;~ !~
- \1.
i'
Aggregation Dilatancy
10_4L-___CL_o_n_tr_oLI___-~~L-c_o_n_tr~o_l__~
3
Fig.6.5. Shear rate dependence of steady viscosity ns, and vis-

cous and elastic components nv and nE, of the complex viscosity
f = 2Hz. (a) Normal RBC in plasma, H = 43; (b) Hardened RBC in
plasma, B = 45 (From THURSTON, 1979 a).
components, nv and nE , respectively, versus the steady wall shear rate

or the rms value of the oscillatory shear rate at the wall, Ys or y,
respectively (THURSTON, 1979 a). The measurements has been performed at
f = 2Hz, in normal blood (Fig.6.5a) and suspensions of hardened RBC
(Fig.6.5b), using an oscillatory tube flow apparatus. A spectacular in-
version in the elastic component is observed and is interpreted at high
shear rates as resulting from dilatancy effects which originate from cell
hardening. Nevertheless, caution is needed in attempt to understand such
a complex system noticely the sharp lncrease of end effects that one
would expects as resulting from elongational flows at the entrance and
the exit of the tubes.
Such difficulties ln interpretation, and, more generally, in separa-
tion of thixotropic and viscoelastic effects, call for improvement of
rheological models for complex materials,nevertheless simple enough to be
available for clinical use, i.e. which avoids unnecessary mathematical
rigour. Some examples of such attempts will be given in the following.
6.2. Some models for thixotropy

As it has been stressed yet, shear-thinning behaviour implies thi-
xotropy, since structure modifications induced by a change in flow condi-
82 D. Quemada
tions need some time, what is measured by a relaxation time.

Many attempts at description of thixotropic behaviour of blood have
been proposed. As example, rheogoniometric experimental studies covering
a large range of shear rates - from 1o-3 to 103 sec-1 (COPLEY et al. 1973)
have been interpreted with the help of a rheological equation proposed by
HUANG as a model of isothermal structural change induced by shear stress
(HUANG, 1972). For viscometric flow, the shear stress-shear rate relation
can be writterr as a generalized (time dependent) Bingham equation:
cr = cr 0 +
.
y ( 6. 1)
where cr 0 , n, C, §, S and p are characteristic constants as defined in

the model. Very fair fittings have been obtained (HUANG et al., 1975).
A phenomenological approach of inelastic reversible thixotropic
fluids (CHENG and EVANS , 1965) can be tentatively used. The model is
based on two constitutive equations, namely:
(1) a scalar (rheological) state equation:
cr = n(X,y) • .Y (6.2a)
is assumed to exist, in which the viscosity depends both on y (shear·

thinning) and on A , a phenomenological parameter which characterizes
the structural state at time t ;
(2) a scalar rate equation:
dA •
- = g(A, y) (6.2b)
dt
describes the structure changes as a function of both y and the structure

reached at time t , through the A value. As an example, the Moore model
(MOORE, 1959) is based on the following form of (6.2), with A,B,C as mo-
del constants:
a =( n 1+ CA.) y (6.3a)
dA. = A - (A + By) A. (6. 3b)
dt
Under constant shear rate, the structure reaches an equilibrium

state, described by:
A
A.eq = A + Bt = + TY (T = B/A) (6.4a)
that gives a shear thinning viscosity (identical to the Hillamson's one.,.

See eq. ( 5 . 5) ) :
( 6. 4b)
Notice that A. values lie from 0 (fully broken-down structure at

y+ oo) to 1 (fully built-up structure at y= 0).
In the case of suspensions of deformable and aggregable particles,
a similar approach results from using the intrinsic viscosity kin (5.16)
as a structural variable which is governed by the phenomenological rate
equation (5.11). Indeed, using the structural parameter:
A. = k - koo = A. (t) (6.5)

ko- kp,
transforms (5.11) to
dA. (6.6)
- = A- A (1 + 8) A.
dt
where A= 'A 1 and e = TA/TD = (<y)p has been defined in (5.12- 14).
For p = 1, (5.6) is identical to (6.3b) from the Moore model. Time be-
haviour of the system can be deduced from (6.6) and (5.16). For instance
if the suspension starts from rest (A.(o) = 1), the shear stress response
to a shear rate step (0 + y 1 ) will be a= n(t)•yl, where n(t) is rela-
ted, through k(t) in (5.16) or A.(t) in (6.5) such as :
84 D. Quemada
-A( 1+8 )t
A(t) =A 00
+ (1-A 00 ) e (6.7)
where A00 lS the equilibrium value (at t+oo ) of A
A00 (6.7a)
+ e
involved in (5.16). Therefore (6.7) gives a correct reprensentation of

stress-time curves, as shown on Fig.6.1c. Notice that the relaxation
time of the stress is the shear dependent characteristic time for the
breakdown of the structure, which appears in the rate equation:
'R = A-1 ( 1+ 8)
-1
(6.8)
and which differs from the characteristic time T involved through 8

in the steady viscosity (6.4a). This corresponds to the frequently obser-
ved behaviour of the stress 1rhich, after cessation of flow under large
shear rates, can fall below that stress resulting from small shear rates.
Some difficulties ln these models can arise from the too crude ap-
proximation introduced by such rate equations, involving only two charac-
teristic times. As for viscoelasticity (see later), description of most
of complex fluids requires several characteristic times, sometimes forming
a continuous spectrum as in polymer solutions.
More sophisticated models, mainly developped for polymers, have been
based on reaction kinetics equations, as rate equations describing the
structure formation by linkage of particles (random chain) and the struc-
ture destruction by shear induced breaking of links (See, for instance,
RUCKENSTEIN and ME\HS, 1973). They could be tentatively used for blood.
6.3 Some models for viscoelasticity

Modelling of linear viscoelasticity lS currently performed by use of
mechanical models, as more or less complicated combinations of springs
and dashpots. The simplest viscoelastic fluid is the Maxwell fluid, re-
presented by a Maxwell element, i.e. a spring (having a Young (elastic)
modulus G) and a dashpot (having a viscosity n) (Fig.6.6.). The corres-

ponding constitutive equation (for stress
relaxation experiments) ls:
da + Q y (6.9)
G dt n
In the case of applying a shear rate step

from rest, the solution of (6.9) is:
a = n.Y ( 1 - e-tIT M) ( 6. 10)
MAXWELL element where TM = n/G lS the Maxwell relaxation

time. Eq. (6.10) defines a stress-time cur-
ve ln agreement with low shear rate obser-
Fig.6.6. Maxwell
element. vation when thixotropy has negligible ln-
fluence.
For a dynamic experiment, applying the oscillatir.g ~hear rate
y = Y1cos wt gives a dynamic response that allows to se-
parate the viscous and elastic components of the dynamic viscosity
·~ = nv - inE , respectively:
nv (6.11)
which involves TM through the dimensionless group WTM .

As the structural viscosity does, a structural elasticity could orl-
ginate from changes in the kind of structural elements responsible of
elastic properties that would lead to a time-dependent shear modulus G(t).
In fact, as in fluids elastic effects are confined only to the early
part of the time curve, it will be possible in practice to replace G(t)
by its initial value G(o), corresponding to the initial structure of the
system.
Accounting for thixotropy (with non linear viscosity) can be carried
out by adding some elastic effects at short times to the previous shear
86 D. Quemada
thinning model. Characterizing these effects by a Maxwell time TM can be

estimated by initial values of n and G, TM = n(o)/G(o) . A model for
stress relaxation of a non-linear viscoelastic system then results from
generalizing eq.(6.9).
do
TM- + a = n(t )y• (6.12)
dt
where n(t) is the time-dependent structural viscosity, as given by (5. 16)

and (6.6) , and TM being estimated from the structure under the initial
conditions of the experiment, i.e. TM = n(o)/G(o). From (6. 12), the
shear stress response to a suddenly imposed shear rate step Yl will be
governed by a competition between the stress building up from retarded
elasticity (See (6.10)) and the stress relaxation coming from thixotropy
(See (6.6)). Since the relaxation time TR, given in (6.8), falls down as
Yl is increased, the "overshoot" behaviour, mentioned at the beginning
of Chap.6 (See Fig.6.1b) will appear as y increases. Nevertheless, it
must be stress again that complex systems could need a large number of
charater istic times,
Accounting for (i) shear thinning and (ii) non linear viscoelastici-
ty, THURSTON (1979b) proposed a very interesting model, based on the
Jeffreys'model- combining in parallel N Maxwell elements and a dash-pot
(Fig. 6.7.)- which gives a dynamic viscosity (under sinusoidal shear
rate):
n&
( 6. 13)
+ WT &
The steady (shear-thinning) viscosity ns results from (6.13) as

w -7- 0 :
To describe thixotropy, as "the manner ln which the model elements

kp kN I
-------- --------- . I
'Y), ")pi "~·I :2,'"
~t~
JEFFREY's model
Fig. 6.7. JEFFREY 1 s Model
change with state of dynamic equilibrium", Thurston assumed that each mo-
del element changes with shear rate according to:
( 6. 14)
involving ci.e srt!:le "degradation function" F( y T Q,) which keep each rela-
xation tir:~.e T 2. ill changed. In ( 6. 14), n 0 Q, and GoQ, are values at rest.
Further, Thr~ston proposed to take F in the form!
(6. 15)
where exponent A will be selected by best fitting of data.

Fig.6.8 illustrates the results from such a model fitted on data
given in Fig. 6.5a. The model contains 6 relaxing elements -thus 6 rela-
xations times. A fair agreement - taking into account that data cover
five decades - is observed.
88 D. Quemada
~P) 10
-
1
10-1 f= 2Hz • .
10-2 '( 'Ys
10- 2 10° (sec-1}
Fig.6.8 Theoretical variations of n8 , ny
and n from the Thurston's model (1979b)
E
6.4 Effects of mechanical and physico-chemical properties of the RBC

membrane.
Properties of RBC membrane, in relation with physico-chemical proper-
ties of the suspending fluid (noticely its macromolecular content ) are
as evidence involved in thixotropy and viscoelasticity of blood.
Quantitative correlations between these bulk properties and RBC ag-
gregability, on the one hand and RBC deformability, on the other hand,
can be obtained from comparative experiments. We shall limit here to give
some examples:
a) Fig. 6.2a and 6.2b have shown the effect of RBC aggregation upon
.
thixotropy of blood. Adding sodium salicylate to a blood sample at normal
hematocrit (i) reduces the viscosity at any y , (ii) suppresses complete-
ly the thixotropic behaviour.
b) Fig.6.9, from CHIEN et al. (1975), shows variations of viscoelas-
tic properties in dynamic experiments at f = 0.1 Hz. Differences in vis-
coelastic behaviour of normal blood (NP) and RBC suspended in albumin
ringer (NA) can be correlated with RBC aggregation, observed by micropho-
tography (COPLEY et al., 1975). As hematocrit is increased, the viscoelas-
tic behaviour of NP tends towards those of NA. They mainly result from
superposition of many factors, noticely mechanical properties of the RBC
membrane and viscosities of intra and extra cellular fluids.
'?v (cP) ") E (cP)
10 2 "" 100
"' "'
, , "'
/
" "' I
10
"' "' "' 10 /
I
"' "'
"' /
/
... .,
/
/
1
f =0,1 Hz 1
f =0,1 Hz
H H
10-1 10-1
40 60 80 100 40 60 80 100
Fig.6.9 Variations of nv and nE versus hematocrit for RBC ln

plasma and RBC in Ringer. (From CHIEN, 1979).(---- plasma,
- - - ringer) .
c) Fig.6. 10, from THURSTON (1979) shows the variations of vlscous

and elastic componentsnv and nE' versus y at f = 2 Hz, during a progressi-
ve hardening of RBC over a period of 11 days, leading day after day to a
progressive increase of nv, more especially nE , which presents a cha-
racteristic inversion at about 1 sec- 1 . Confirmation of this result has
been obtained from studying properties of osmotically modified RBC sus-
pended in plasma, where inversion is observed only for crenated cells
(i.e. hardened cells) leading to both enhanced nv and nE at high shear
rates.
Although these unsteady properties of blood are very important slnce
blood circulation is pulsatile, difficulties in their measurement
and interpretation show that much work is still needed before satisfacto-
ry understanding of the unsteady behaviour of blood will be gained.
90 D. Quemada
;As=::::---__ t'f)y
.~(b)
m-1
Properties of Osmotically Modified Cell
Suspensions in Plasma
Sample Osmolality Hematocrit
(mOSM/1)
A.Hypotonic (cells
215 0.43
swollen)
(a) B.Normal 310 0.33
C.Hypertonic 946 0. 19
(cells shrunken)
Fig.6.10(a) Variations of nv and nE during a progressive har-

dening of RBC by a~etaldehyie. The inversion in nE can be com-
pared to the one observed in osmotically crenated cells. (b)
(From THURSTON,1979a).
Blood Flow in Narrow Vessels 91
7. BLOOD FLOW IN NARROW VESSELS AS A TWO-PHASE (ANNULAR) FLOW: Blunted

velocity profile, plasma layer, Fahraeus and Fahraeus-Lindqvist Ef-
fects. Two-fluid models.
As it has been pointed out in §1, it is currently assumed that rheo-

logical properties of blood do not influence very much its flow in large
vessels, although some exception may result at low flow rates, near quasi
steady conditions and in the vicinity of changes of the cross section ar-
ea of the vessel (branching, stenosis or aneurism, ... ). On the contrary,
as the diameter of the vessel is lowered, blood rheological properties
appear more and more important from shear thinning, finally complicated
by occurence of phase separation in narrow vessels. However, as in vivo
studies depend on a too large number of unknown variables, one can hope
that precise in vitro analysis of blood flow through narrow tubes will
be able to allow some progress in understanding blood microcirculation
and further, in clinical applications.
7.1. Experimental evidences in blood microcirculation ( 0 ).
Table 7.1 briefly gives main flow characteristics of human microcir-

culation (adapted from SUTERA, 1977).
Experiments on steady blood flow exhibit four 'anomalous' features,
namely the blunting of velocity profile, the formation of the plasma
layer, the Fahraeus Effect and lastly the Fahraeus-Lindqvist Effect.
(i) A blunted velocity profile is observed near the axis of the
vessel, leading to a plug flow either at high hematocrit or at low -va-
lue of the vessel to cell diameter ratio, ~ = 2R/2a. Such a blunting
is shown on Fig. 7.1 of human nomal blood flowing through a narrow slit
(DUFAUX et al., 1980), H = 20 and 40 (the data has been obtained by Laser
( 0 ) restricted here to blood circulation in narrow vessels. Nevertheless,

circulation in capillaries shows some analogies through the "axial train"
structure.
92 D. Quemada
TABLE 7. 1. BLOOD FLOW IN HUJ.1AN

MICROCIRCULATION
V/Vo '7H=0.40
Vessel Diameter Blood ( 0)
• H =0.20 Diameter Ratio Velocity
(JJm) c (cm/s)
'·
Terminal 4-20
500-2000 60-250
Arteries
Sr'tall 10-60
70-500 4
.4rteries
Arteriol., 10-70 1-10 05
---- - - ·-·- -~------
Capilla. 4-10 2-1 01
Venules 10-11 0 -t--1=-14--r----o4~
0
j
Small
Veins 110-50~~-6;- r--;;- I
e ~m Terminal
500-5000 6o-6oo 2-10
Veins
0 100 200
( 0 ) Average peak values
Fig.7. 1. Velocity profiles in a
narrow slit (lxe=200]Jmx1. 1.cm).
Normal blood at H = .20 and .40
(From DUFAUX et al., 1980)
Doppler velocimetry).
Such results, accounting fer the fact that higher flow rates Q va-
lues are obtained under the same pressure gradient P, are comparable with
similar results on flow of rigid sphere suspensions through tubes (KARNIS
et al., 1966). Nevertheless, the independence of velocity profile against
flow rate (i.e. linearity of the pressure-flow rate relation) observed
by these authors seems to result from a too short range of flow rate va-
riations. Conversely, Fig. 7.2 shows these non-linearity in the P-Q rela-
tion, i.e. non-newtonian effects in the apparent viscosity.
( ii) In very narrow tubes ( 3 ~ ~ < 50) , the existence of a marginal
(plasma) layer near the wall is currently accepted, leading to consider
the flow as a two phase one, with a particle rich axial core surrounded
.
by a particle depleted (or a particleless)
existence,
~-;all layer. Such (much debated)
is now well-established after Bloch's photographic records
20 (JW (dynes/cm-2) H: .66
e:350p.m
10
'l)p=1.56
~a (s-1)
100 200
Fig.7.2 Pressure-Flow rate relation of blood flow in a narrow slit
(£xe = 1.1 em x 350 ~m) at various hematocrits. o = Pe/2 and
y = 4Q/£e 3 ( P =pressure gradient= P/10 em; Q =wflow rate). (From
DfiFAUX et al., 1980).
(BLOCH, 1962) of blood flow in var1ous animals. COPLEY and STAPLE (1962)
observed similar features. BUGLIARELLO et al., (1965) carried out measure-
ments in glass capillaries, using anticoagulated normal human blood at dif-
ferent hematocrits, changing the pressure gradient. They found that the
relative layer thickness, c/R, decreases increasing the hematocrit H and
/or decreasing the wall shear stress ow, whatever the radius may be excep-
ted at low hematocrit (H = 28%) for which o/R appears as very sensitive to
the R value. Values of o/R are found of about 0.05-0.1 at normal hemato-
crit.
Direct studies of single particle motions in Poiseuille flow
(GOLDSMITH and MASON, 1967) showed that rigid particle does not experien-
ce any radial motion if the flow is low enough, unless the particle
R¢ynolds number (see eq.(2.4)) is higher than about 10-4, then giving a
94 D. Quemada
radial accumulation of particles near the cylindrical surface at~0.5-0.6~
Nevertheless, such a "tubular pinch" ( SEGRE and SILBERBERG, 1962) will

not normally occur under physiological conditions. On the contrary, non
rigid particles as liquid droplets migrate towards the axis, whatever is
the value of flow rate.
At moderate concentration, radial migration is still observed (GOLD-
SMITH, 1971), but forces responsible of such a migration are likely coun-
terbalanced by crowding effects in the core. At high concentration the
latter (which correspond to forces involved in the packing of particles)
win on the former and, close to the wall, it remains only geometrical
exclusion of particles which leads to a reduced concentration within an
annulus) the thickness of which has same magnitude that the particle ra-
dius in the case of spherical particles (MAUDE et al., 1958). Up to now,
no adequate theory seems available to get a satisfactory description of
marginal layers. Whatever the real process for plasma layer formation
may be, this layer plays a very important role, as a lubricant layer,
which can be able to reduce significantly the apparent viscosity. As
VAND ( 1949) pointed out, "in the region of high concentrations, consl-
derable slip at the wall might develop due to the layers of low viscosity
along the walls which might finally completely overshadow the effects of
shear inside of the suspension, making the measurements useless".
(iii) A mean (tube averaged) hematocrit- 1t is f.::-und lesser than the
feed (reservoir) hematocrit Hf. It decreases as the tube diameter di-
minishes (Fahraeus effect)(FAHRAEUS, 1929).
As COKELET (1976) showed, the discharge hematocrit Hd (i.e. the
average hematocrit of the outflowing blood from the·tube) practically
equals the feed hematocrit Hf if E; ;;;;. 3. Below this value, Hd. < Hf, that
can be thought as resulting from some particle redistribution at the en-
trance of the tube. For 3 ~ $ 50,BARBEE and COKELET (1971) found that
E;
the tube relative hematocrit, Hr = Ht/Hf, varies linearly with Hr ,

the slope increasing as R decreases (See Fig.7.3). Similar variations but
with markedly different slopes from the ones expected were obtained with
v~y narrow tubes (down to 2R = 8.1 ~m) (COKELET, 1976). Thus, the follo-
wing approximate relation lS obtained
(7 .1)
~-::----------------~---·------~-----
11 H, 2211-1 ---------·-------·-----
.9 1541-1 ----~L--·------
128 1-1
99!-l••
---------------.~~---~~-·-
•• .
-------~.r--r-------•------
-~~~----~r-___L--
* .-:.*--*:....----· *
*----r _J...----~r--·-
75!-l--...
*• .J
591-1 ____.._-~·-----1- :» ·.
.5 ::----~ -------:;-L----.---1--------oL-------L-------J
0 10 20 30 40 50 60
Fig.7.3 Relative hematocrit Hr = Ht/Hf versus Hf for
different tube diameter D. (From BARBEE and
COKELET, 197 1 ) .
(iv) the apparent viscosity lowers as the tube diameter decreases (Fah-
raeus-Lindqvist Effect) (FAHRAEUS and LINDQVIST, 1931).
Number of studies of this effect have been carried out. We shall li-
mit here to display the results of HAYNES and BURTON (1959) in long tubes
(to avoid any entrance effects (L/R "' 90) with 67 1-1m~ R ~ 750 pm. They
observed that the higher the hematocrit, the stronger the na lowering.
(Fig. 7.4).
96 D. Quemada
H =80% ·
------------i
12
8 60
------------~
50
---------------1
4 40
--------~-----i
---------~~-- .
- - - - - - _h_\,J __ ....,.
TUBE RADIUS (mn)

0 .2 .4 .6 .8 .1 1.2
Fig. 7.4 Fahraeus-Lindqvist effect for human erythrocyte suspensions

of various hematocrits at 25.5°C. Smooth curves were fitted to the
points by the method of least squares. Asymptotic values for a tube
of infinite radius are indicated by broken lines together with their
standard errors of estimate. (From HAYNES and BURTON, 1959).
Similar results, expressed as wall shear stress Ow versus D=

C)./nR 3 = Ya/4 (where Ya is the apparent shear rate (see eq. ( 2. 12) )ha-
ve been obtained by BARBEE and COKELET (1971). Fig. 7.5 shows the data
obtained (i) with a large tube (where no phase separation occurs, hence
Ht = Hf.(ii) with a narrow tube, 29 ~min diameter, where Ht ~ Hf varies
as a function of Hf, as indicated. Comparing large and small tubes data
at Hf = 0.559, for instance, shows Fahraeus-Lindqvist effect as the lowe-
ring in Ow at any U value. These authors found an important empirical
fact from their data : the relation ow versus U for large tubes fits the
sm~ll tube data at a given feed hematocrit Hf, provided the tube hemato-
crit Ht was used in the relation Ow (U, H) instead of Hf as in large tube
that is
a (U (7.2)
w
::J
100~--------------------------------~
(Jw (dynejcm 2}
611~ TUBE
Ht= 0.559 • ~
10 ~·
~
~:~ Hf Ht
~~· 0.559 0.358

~·
6
I ... 0.494 0.310
~ ~:~·
~:
• O.L..22
• 0.359
0.259
0.216
r 0 0.290 0.171
* 0.220 0.127
• .0.156 0.088
U (sec-1)
j ~~--~~~U-----------~----------~
1 10 100 tJOO
Figure 7.5 Relationship between wall shear stress and average
blood velocity divided by the tube diameter. The points are ex-
perimental data obtained with 29~m tubes : the curves represent
data obtained with an 811 ~m tube. (From BARBEE and COKELET (1971)
98 D. Quemada
where Ht = Ht (Hf,R) is given 1n (7.1). For example (Fig. 7.5) the curve
crw vs U for large tube (~bout 800 ~m) at hematocrit H = 0.358 fits remar-
kably well the small tube data with feed hematocrit Hf = 0.559. Therefore
one can assert that, empirically, the main part of the Fahraeus-Lindqvist
effect results from the Fahraeus effect, i.e. the wall shear stress, as
function of U, Hf and R, must (at least approximatively) verify (7.2).
Again similar results hold in smaller tubes, down to 2R = 8.1 ~m (COKELET
1976).
7.2 ONE-FLUID MODELS
All the above properties have been observed in other systems, as rl-
gid sphere or disc suspensions. They call for some general model able to
explain them simultaneously.
Explanation of the blunted velocity profiles can be easily found in
the non-newtonian behaviour of blood since an/ay < o implies an enhan-
ced viscosity near the axis, n ~ n(y=O}=~and, on the contrary, a decrea-
sed viscosit~ near the wall, n~n(y==)<< no that results to a blunted
velocity profile. In terminal arteries such a non-newtonian one phase
flow model will be sufficient. Notice that, to some extent, such a flow
can be considered as a two phase flow, the plugged region ae a rigid co-
re, surrounded by a fluid annulus. Formation of the plug has been treated
as a percolation transition (de GENNES, 1979).
In narrow vessels anomalous properties of blood flow were believed
as resulting from the failure of the classical boundary condition for
fluid velocity, assuming the fluid studied had a slip velocity at the
wall (See for ex. JONES, 1966). Another explanation, as the non validity
of continuous description, 1s found in the so-called sigma-effect, where
the flow through the tube lS considered as the summation of contributions
from a finite number of coaxial layers of finite thickness (DIX and SCOTT
BLAIR, 1940) .
7.3- TWO-FLUID MODELS
The above models failed to give an overall and self-consistent ex-

planation of velocity profiles as well as the remaining experimental fea-
tures and nowadays two-fluid models appear as the more credible. The de-
crease of concentrat~on near the wall exagerates the blunting of v(r)
since, in addition to on/oy < o , one has on/oH > o . Thus a model of
blunted velocity pr~file will be very easily found.
Although the following results hold in other geometries (as narrow
slits, for instance, QUEMADA et al, 1980) we will limit our discussion
to flow through narrow tubes, taking again as a two-phase model, a core
(0 ~ r ~ SR, subscript s) surrounded by a peripheral layer (~R~ r~ R,
subscript w). In order to modelize the particleless (plasma) layer, we
shall take Hw = 0. Some "classical models will be discussed in the follo-
wing subsections.
7.3.1 -The simplest model is formed by two newtonian fluids, with cons-
tant viscosities ns = n {Hg) and nw = np
With appropriate boundary conditions (no slip velocity at the wall,
v (R) = 0 ; continuity of' shear stress at r= i3R ,cr (i3R) =cr (i3R); f'inite
w w s
velocity gradient on the axis, ( avs/ar) 0 =0 ),solving Navier-Stokes equa-
tions for the two-phases, allows to calculate the total volume flow rate
(BAYliSS, 1952 ; HAYNES, 1960 ; THOMAS, 1962) as the sum of core and wall
layer ones
(7.3)
with
2s (7.3a)
100 D. Quemada
(7.3b)
where A = 7TR3crR/ 4np and oR = ow (R) is the shear stress at the

wall. The ~pparent viscosity, na defined from the Poiseuille's law,
Q = 7TR3crR/4na lS then given by
na =
Assuming the plasma layer thickness o much smaller than R (what is

observed at normal hematocrits), i.e. o/R = 1- S << 1 , , leads to
the following good approximation of (7.4.) :
[1-4.§_ ns - 1) 1 (7.5)
R np
Since o mainly depends on particle-wall interaction, its value, in

the limit o « R should be very close to the one obtained neglecting
wall curvature, thus leading to assume that o does not depend on R. As
a consequence (7.5) gives the apparent viscosity as a decreasing function
of (1/R). Therefore (7.5) describes the Fahraeus-Lindqvi st effect, as
(3na/3R)< o.
Fig. 7.6 , from MIDDLEMAN (1972) displays fitting of eq.(7.5) on
the data of Fahraeus-Lindqvi st (1931). In tubes with relatively large
diameters, a linear variation is obtained, and n5 /np and o can be dedu-
ced from (7.5) as na= f(1/R) . Values of o about 0.7 ~mare found from
this data, and appear as limiting values at high wall shear stresses
( crw > 80 dynes/cm 2 ) and at normal hematocrit (which were not measu-
red).
Fitting eq. (7.4) on the data shown in Fig. 7.5., HAYNES (1960) ob-
tained the curves in Fig. 7.5. He calculated o for each H value : in
the normal range of H, he found o ~ 3 ~m. As evidence, such values, indi-
rectly calculated through a model, are very sensitive to the choice of
8 "'r the model and to the precision of

pressure flow rate measurements.
7.3.2 Discarding the assumption

that the core behaves as a newtonian
fluid, THOMA.S ( 1962) gave a self-
consistent treatment of the problem.
~ 0
He obtained formally the.same equa-
tions than (7.3), ~7.4), but where
1 ns is the apparent viscosity of the
0 60 -R mm-1 core, defined as if it was flowing
Fig.7.6 Data of Fahraeus-Lindq- through a fictive t~be of radius

vist (1931) plotted as a test SR, with a slip velocity Vs(SR)
of eq.{7.5). (From MIDDLEMEN,
and a shear stress cr 5 (SR) = ScrR
1972).
at the fictive wall. Written with
the help of relative fluidities, F = ~ , eq. (7.4) becomes :
n
(7.6)
where Fa = apparent relative fluidity for feed hematocrit Hr and under

wall shear stress crR and F5 = core relative fluidity, which depends on
core hematocrit Hs and shear stress ScrR . Moreover, Thomas accounted
for Fahraeus Effect which had been neglected in previous works.
From equation of continuity of the suspended phase :
it follows, with (7.3a) :
( j.8)
102 D. Quemada
If the function F= F (H,a) for fluidity in the absence of wall ef-

fects is available, the only unknowns in (7.6) and (7.8) will be Hs and
a. Thomas (uselessly) introduced the tube averaged concentration
Ht = a~Hs and solved eq. (7.6) and (7.8) by a relaxation method, taking
successive approximations to the value of a. Then, assuming aR high e-
nough to recover newtonian properties for the core, and taking (See eq.
(3.14) F =1 - kH for core fluidity, he applied his method to different
data. For rigid spheres of radius a, he found (o/a) = 0.71 - 0.76, in
satisfactory agreement with geometrical exclusion near the wall. For sus-
pensions of mammalian red cells, o was found equal to about 4 ~m, how-
ever decreasing or increasing as R is lowered. These results are not com-
pletely convincing ones, in one part because of the incompleteness of
the data (hence the approximation that Thomas made)and in other part be-
cause of the lack of non-newtonian effects in the core ( y going to zero
on the axis).
7.3.4. A non-newtonian behaviour for the core has been taken into ac-
count by CHARM and KURLAND (1962) and by CHARM et al. (1968), using a
power law fluid and later a Casson fluid eq.(5.1 ). They calculated the
pressure-flow rate relation, that reduces in our notations to the appa-
rent relative fluidity such as :
F
a
=1 - alt ( 1 - F ) + G (
s
a, £!J_
aR {7.9)
which differs from (7.6) by an additional term G which vanishes as yield

stress a0 tends to zero ( 0 ). From the existence of a0 , a partially plug
flow takes place, the radius of which is RPlug =Rao/aR. Eq.(7.9) was
( 0
) In the limit of x = ao/aR < 1, 4 3
one has G(a,x)"' 3 Ga! ]
ax [ l-7(~)
used by CHARM et al., to analyse their pressure-flow rate measurements

in a large number of capillaries of different sizes. Independently, the
Casson parameters K and cro, entering in Eq. (5.1) were determined as
functions of H by Couette viscometry. Furthermore, radial distribution
of cells, H(r) was assumed to be similar to the one observed by PALMER
(1965) in rectangular tubes. Fitting (7.9) on a data relative to flow
under a wide-range of pressure gradients and flow rates, through a large
number of tubes, different in size, led CHARM et al to find results in
good agreement with those from in vivo measurements (See table 7.2).
They deduced variations of the ratio o/2R vs H which appears scattered,
that would be due to (i) the failure of Casson law at low shear rates
and/or (ii) the choice of Palmer's radial distribution H(r), which is
questionable since very abrupt profiles have been deduced from analyzing
viscometric data (WATANABE et al. , 1963) .
TABLE 7.2 OBSERVE~ MARGINAL PLASMA LAYERS IN FROG MESENTARY VESSELS

AND CALCULATED FRCM FLOW OF HUMAN BLOOD IN GLASS WBES_
(BLOCH, 1962) (CHARM et al. , 1968)

From mesentary Human blood in glass tubes
(High speed photography,~ unknown (Calculated from pressure-flow rate
~ = 0.3-0.45)
Inside diameter o/D Inside diameter o/D
(].1m) (].1m)
71-80 0.0416 71.8 o.o44o

1 41-1 60 0.0280 155.0 0.0220
241-270 0.0230 256.0 0.0214
7.3.5. Alternatively, the non-newtonian viscosity equation (5.16) has

been used to describe the core behaviour, keeping (7.8) unchanged. Rheo-
logical parameters entering in (5.16) were determined with the help of a
low shear coaxial viscometer. The core hematocrit was calculated, after
104 D. Quemada
elimination of Fs between (7.6) and (7.8)
(7.10)
!!a=
Hf
For given R, crR and lif' experimental value of fluidity, F~ can be

obtained. Putting Fa= F~ into (7.10) leads to
(7.11)
Successive approximations for the value of 8 (as in the THOMAS'

method) are taken in (7.11 ), allowing to calculate Fs (Hs, 8crR) then
putting into (7.6), until the difference between the calculated value and
the experimental one, Fa- F'a• becomes unsignificantly different from
zero. Finally, after the value of 8 have been obtained for a given crR
the corresponding theoretical velocity profile can be calculated and com-
pared to the experimental one, measured tb.rough Laser-Doppler-velocimetry.
(DUFAUX et al., 1980, See Fig. 7.1).Such a model has been applied to nar-
row slits (thickness= h). Using again crR for the wall shear stress crR =
crw(h),Fig.7.7 gives variations ofF~ and of o= h(1 - 8), as functions of
crR (QUEMADA et al., 1980). It shows that, as crR increases from zero, o
increases, reaches a maximum and then decreases and seems to tend towards
zero at very high crR (limiting value Foo was used as crR ~ 00 ) , Similar fea-
tures has been found at different feed hematocrit and slit thickness.
Although these observations were made in narrow slits, it can be expected
that no dramatic differences would exist between slits and pipes. Indeed,
such a maximum on the curve o= o (crR) has been observed in 200 ~m diame-
ter tube, by microphotography under dark field illumination (DEVENDRAN
and SCHMID-SCHONBEIN, 1975).
Whatever the model may be, the importance of the (lubrificating)

plasma layer requires further work before significant conclusion about
its physiological importance would be drawn.
•
Apparent
Fluidity - ._.... ........ --
•1 Wall Layer
/
_:.tAt'
Thickness
)lm 3
0 2
.......
''- .... _
........
....... 1
0
i 1b 100
Wall Shear Stress dynes;an2
Fig.7.7(a)Apparen t fluidity F~ and (b) Wall layer thickness o vs.
wall shear stress aR . Normal Human Blood Flow ( ¢a= .57) through a
slit (350 ~m x 1. 1cm x 10cm).T=23°C. Rheological variables of the
same blood sample (from Couette Viscometry) :ko=3.24, k =1.68,
T=.135sec.(Points 0: from fitted curve on F~ data; Points o:from
theory).
106 D.Quemada
8 - CLINICAL HEMORHEOLOGY - CONCLUSION
Applying the analysis developped in ~he previous sections to blood

flow under pathological conditions comes up against many difficulties,
mainly (i) the extrapolation of results from the in vitro aonditions
tothein vivo ones and (ii) the increase in variability (i.e. standard
deviations) of measurements performed on patient blood, sometimes lea-
ding to contradictory resalts. As a consequence, number of clinicians
are not very convinced that quantitative hemorheology can help them in
their pra~tice.
Nevertheless, number of attempts at interpretation of microscopic

features of intravascular blood flow have been recently done in terms of
RBC deformability'and RBC aggregability. Indeed, due to the clinical im-
portance of platelet aggregation, clinicians believe the former more res-
ponsible of viscosity increasings than the latter. Now, it seems well
established that, at least in vitro, both these two main properties of
RBC are concerned, since both they can be affected independently or to-
gether by pathological processes. Especially, at low shear rate, not only
high levels in RBC Aggregation are observed, but also elevated values in
yield stress a 0 which can promote the stoppage of blood flow, i.e. blood
stasis. Many examples of application exist but only some of them will be
recalled hereafter.
Poor blood flow, instead of being only considered as resulting from

heart failure or from a blood vessel disorder, now appears as resulting
from high values of blood viscosity, which is observed in many diseases.
However, several attempts to correlate enhanced whole blood viscosity
and/or plasma viscosity to different variables,as Hand npJn various di-
seases did not give very convincing results (See for ex., ISOGAI et al.,
Clinical Hemorheology 107
MATSUDA et MURAKAMI, 1976 ; DINTENFASS and KAMMER, 1977) , although si-

gnifiant increases in viscosity have been found, especially at low shear
rates (For ex., see DINTENFASS, 1977, 1979). DORMANDY et al. (1973a)
studying patient5 with intermittent ciaudication, showed that blood
viscosity of these patients was significantly raised, compared to nor-
mals.
DINTENFASS (1971) gave an intensive study of the high viscosity syn-

droma in various diseases (ischaemic, sickle cell, hemolitic anemia,
polycythemia, shock ..... ) (See Table 8.1 , after DINTENFASS et al,
1966 b).
Fibrinogen concentration plays the most important role in RBC Aggre-

gability. Its lowering (afibrinogenaemia) leads to the recovery of new-
tonian properties and the lack of yield stress (MERRILL, 1969). Its in-
creasing leads to enhanced viscosity, as in heart failure (KELLOG and
GOODMAN, 1960). Similar effects can be expected to result from the pre-
sence of abnormal proteins, as macroglobulins (Waldenstrom 1 s disease) or
IgG1 myeloma (LINDSLEY et al., 1973). Polycythaemia remains a more obs-
cure disorder which is probably resulting from complex association of
the increase in Hematocrit and several factors. Same difficulties occur
in heart attack, in shocks Abnormal increase in viscosity of diabe-
tics have been shown, especially if they are associated with abnormal
circulation in retina (HOARE et al., 1976). Fig. 8.1
Changes in RBC Deformability, as in sickle cell disease lead again

to increase in viscosity at low oxygen pressure (i.e. de-oxygenated RBC).
(See CHIEN et al ; 1970). Changes in RBC shape (from pH modifications
for instance), in spherocytosis or in parasitosis (as malaria) also give
an enhanced viscosity.
108 D. Quemada
Table 8.1
BLOOD VISCOSITIES IN NORMAL CONTROLS AND IN PATIENTS SUFFERING FROM
THROMBOTIC AND OCCLUSIVE DISEASES
Viscosity of blood in poises
Shear_late Normal Normal Patients

sec women men
0.01
X 11.9 9.9 57.5
X ± SD 4.9-28.6 4.4-22.1 21.7-157
X ± 2SD 2.0-75 1 .9-49.4 7.8-428
0. 1
X 2.22 2. 10 7.23
x ::t SD 1.23-3.94 1.13-3.93 3.15-16.70
x ± 2SD 0.69-1.00 0.60-1.35 1. 36-38.3
1. 0
X 0.41 0.45 1 .00
x ± SD 0.28-0.59 0.27-0.82 0.60-2.00
x ± 2SD o. 19-D.87 0.15-1.41 0.24-4.00
-
7.2
X 0.13 0.15 0.22
x ± SD 0.09-0.18 0.11-0.22 o. 16-o. 32
x ± 2SD 0.06-0.26 0.08-0.31 0.11-0.45
29
x 0.01 0.095 o. 106
x ::t SD o.o6-o.o8 0. 01-0. 13 o.o8-o. 14
x ± 2SD 0.05-0.09 o.o6-o.n 0.06-0.19
118
x 0.048 0.056 0.065
x ± SD 0.045-0.051 0.048-0.065 0.041-0.101
x ± 2SD 0.043-0.054 0.041-0.076 0.026-0.158
(From DINTENFASS et al., 1966 b).

Number of the above-mentioned studies remained only qualitative

because they used averaged values of viscosity from large series of nor-
mals and patients. Although systematic increase in mean viscosities are
measured one observes some overlapping of normal and patients values
(See Table 8.1). These findings result from superposition of viscosity
changes with variations of rheological variables as H, np, RBC Aggrega-
tion and RBC Deformation ... for each donor, on the one hand, and onthe
other hand, to viscosity variations when one passes from one patient to
another. In fact, getting on a quantitative study needs t? reduce data
from each donor before to make some averaging procedure. CERNY et al.
(1974) gave nlood viscosity of patients with sickle cell trait disease.
Fig.8.3 shows viscosity variations for normal blood (HbAA) for two age
groups of people (young,12-29 years old ,and old )69-91 years old) as
well as for a sickle cell anemia (HbSS) (from CHIEN et al., 1970). Fig.
8.2 also contains the viscosity-shear rate curves for sickle cell trait
blood (HbSA) for five young people (4-41 years old) and an elder person·
(61 years old), which shows an 49 %.
Retinopathy ·"
,/
,.I'
/Dia~ics
./ ,'
/'No Relinopathy
> ~::::::>~..,.,
1
-/
•......0
u
~ ----ss
u
Shear Rate o.ns-1 10 ~SAYAAO
20 AAY
200 400 600 ~ (s-1)
Plasma Fibrinogen (mg%)
0 10 20 30
Fig. 8.1 Relationship between Fig. 8.2 Comparison of viscosities
viscosity and plasma fibrino- of different bloods.
gen (from HOARE et al, 1970) SS - Sickle cell anemia
SAO - Sickle cell trait-old
SAY - Sickle cell trait-young
AAO - Normal old ,AAY Nor~al young.
(from CERNY et al., 1974).
110 D. Quemada
Moreover, at low shear rates, the latter exhibits a variation very close
to the SS one.
SCHMID-SCHONBEIN and WELLS (1971) claimed that viscosity cannot be
sufficient for relevant clinical studies since "even the most sophistica-
ted rotational viscometers, while capable of measuring bulk viscosity
near stasis, are subject to artefacts introduced by the material under
study and present the net effect of several different mechanisms". There-
fore, one needs to employ simultaneously different tools and methods a-
mong them these authors proposed the "Rheoscope", a transparent counter-
rotating chamber, which can be used to quantify the kinetics of RBC aggre-
gation from light transmission analysis.
The rheoscope has been used in an extensive study of pathological
RBC aggregation (blood sludge) (SCHMID-SCHONEBEIN et al., 1973). Such a
method allowed photometric quantification of the kinetics of the RBC
Aggregation, in relation with changes in viscosity, according to values
given in Table 8.2.
While normal RBCs physiologically aggregate to form rouleaux, with end-
to-site attachment, RBCs, under pathological conditions, exhibit an in-
tensified aggregation, with side-to-side attachement of rouleaux, i.e.
both more compact and more shear stress resistant aggregates. Globulin
responsible of this pathological aggregation were identified.
A similar method, but using the back-scattered light (by multiple
reflexion) from a sheared blood sample, is believed to give more informa-
tion about the structure of aggregates and the degree of RBC orientation
at a given shear rate (MILLS et al, 1980). Fig. 8.3 shows a sketch of the
apparatus and Fig. 8.4 an example of time curves of reflectivity (as the
ratio of reflected flux in presence of aggregates to the one in the disa~
gregated state). Under steady shear, RBC are dissociated and oriented
(Fig. 8.4, part (a) which exhibits a constant value of reflectivity~
After abrupt stopping of the cup viscometer, random RBC orientation is

quickly reached (peak b) and then, RBC stack to form rouleaux at y= 0,
•leading
FILTER HeNe LASER
,/""" FILTER 6328 A

DETECTOR
RECORDER I
XY !....._ ___.
Fig. 8.3. Sketch of the apparatus for measurement of intensity of the
light back-scattered by the suspension (from MILLS et al,
1980).
to aggregation times about 5 sec. for normal blood and much higher values
for pathological blood.
rJr· 4.511% DxiO
b
I •
-PJ-l.l)s-1
~
10scc.
t.
Fig. 8.4. Variations of the reflectivity of the suspension versus time.
(N) Normal blood. (P) Pathological blood. (from MILLS et al,
1980).
112 D. Quemada
TABLE 8.2 COMPARISON OF THE APPARENT VISCOSITY OF NORMAL BLOOD AND OF

BLOOD WITH PATHOLOGICALLY INTENSIFIED AGGREGATION (PATIENTS RECOVERING
FROM MYOCARDIAL INFARCTION, E'.lJRGICAL TRAUMA) AS TESTED IN A GDM-VISCOME-
TER WITH GUARD RING (from SCHMID-SCHONBEIN et al, 1973).
Normal blood Patient's blood

(np = 1.2 ± 0.05 cP) (np = 1.46 ± o.2o cP)
n (cP) nr n (cP) nr
a a
!Hematocrit 40 ± 1 % n = 22 n = 15
-1
0.1 se.c -1
1 .0 sec -1
42.6 ± 11.4
13.8 ± 3.3
I 35.5 ± 9.5
11.5 ± 2.7
56.18±8.8
17.8 ± 3.6
38.8 ± 6.2
12. 1 ± 1.9
20.0 sec 5.0 ± 0.66 4. 15 ± 0.55 5.2 ± 0.5 3.59 ± 0.5
!Hematocrit 45 ± 1 % n = 20 n = 12
-1
0.1 sec_ 1
1 .0 sec _ 1
67.7 ± 12.5
17.3 ± 3.2
I 56.5 ±10.4
14.4 ± 2.6
71.76 ± 13.22149.02 ± 6.41
23.10±4.69 15.39 ± 2.13
20.0 sec 5.8 ± 0.74 4.8 ± 0.062 6.42 ± 0.88 4.3 ± 0.43
The values are grouped according to hematocrit values (40 ± 1 %,

45 ± 1 %) • In the patient group, low shear viscosity is indeed higher
than in the control group, but so is plasma viscosity. Relative apparent
viscosities are not higher.
All the above findings underline the importance of viscosity in

diseases. Nevertheless, values of viscosity, even measured at a standard
hematocrit, for different shear rates -especially for low ones - are ve-
ry often neither specific of the disease under study nor sufficiently
free of large variations for one patient to another one with same disea-
se. Furthermore, many questions about blood circulation in arterioles
and venules remain without response. Therefore, in spite of the fact that
blood complexity asks for the possibility to construct a rheological
model for blood, such a model appears very necessary, even if it is an
approximate one. As too much parameters enter in the problem, one must
try to adopt a more general viewpoint, considering as in the previous
sections that blood behaves as a high concentrated suspension which exhi-

bits shear-thinning from RBC Aggregation (RBCA) and Deformation (RBCD).
Applying general rheological laws - as Casson law for instance - is limi-
ted since these two fundamental processes \RBCA and RBCD) are not simply
contained into specific parameters. On the contrary, through the intrin-
sic viscosity given by eq. (5.16) assuming again all changes due to pa-
thology are included in it ( 0 ), these processes can be taken separately
into account by the single parameters k 0 and koo· When compared to normal
va1 ues, k n0 and kn 00 , · 1 y, var1a
respect1ve · t"1c~s o f k 0 an d k oo can b e expec t e d
to reflect variations of RBCA and RBCD, respectively. Tables 8.3 and 8.4
Table 8.3 Determination of Rheological Paraneters from

/':;. data of SCHMID-SCHONBEIN et al ( 1973)
0 data of SCHMID-SCHONBEIN et al ( 1971)
+ data of USAMI et al (1971)
I cp nf
cP
k 00 ~ sec
-1
Normal blood 0.40 1.2 1.84 4.65 2.23

/':;.
0.45 1. 2 2.07 4.33 1.88
Sickle cell anemia
0 4o 1.4 2.83 4.63 4.94
Packed cells
+ 0.88 1. 2 1.83 2.26 2.16
( 0 ) Such an assumption means that the form 3.17 for the relative viscosi-
ty holds again for pathological blood.
114 D. Quemada
depict such variations, using var1ous literature data. Values of Rheolo-

gical Parameters (RP) are given on Table 8.3 for :
(i) Normal blood, at ~ = 0.40 and 0.45 (from data of Schmid-Schon-
bein et al. 1973).
(ii) Sickle cells, i.e. RBC having abnormal Hemoglobin, hardened
when de-oxygenated, but which recovers same flexibility w~en oxygenated.
(from data of SCHMID-SCHONBEIN and WELLS 1971).
(ii{) Packed cells, at~= 0.88 (from data of USAMI et al. 1971).
On Table 8.4, comparison between n (or 1/n) calculated from these
RP-values, and those directly observed by SCHMID-SCHONBEIN et al. (1971)
is shown.
Tabl.e 8.4 Comparison of observed and calculated apparent viscosities

-observed values from measurements of SCHMID-SCHONBEIN et al (1971).
- calculated values,from eq. (5.16) (with r = 1/2) from RP values
given in·table 8.3.
apparent viscosity (cP)

Obs. calc. j obs. calc.
----·-- at y - 0.1 sec
Normal Blood at y = 230 sec
/::,
42.6 ± 11.6 35.3 3.6 0.3 ± 3·.0
67.3 ±12.3 68.2 4. 1 0.5 ± 4.1
Sickle Cell anemia at· y - o. 1 sec at y = 230 sec
0 68 ± 19 70.8 5. 1 ± 0.6 6.64
Packed cells
+
at y = 2.30 sec
8.61 0.86
I
apparent fluidity (Pois·~ 1)
at y = 230 sec
1. 51 2.64
(observed at ~ = o.95%)
% In the third case, RP-values are not known at ~ = 0.95 used in observa-
tions. Then, calculated values of fluidity, for~ = 0.88 are somewhat
higher than the observed ones.
As expected, RP-values appear as concentration dependent. Neverthe-

less their determination, at fixed ¢ , has been proposed as a gene-
ral rheological characterization (QUEMADA, 1976 b).
Large increases in RBCA have been found by fitting_ the data of SCHOLZ
et al (1975) for two groups of critically ill patients Group I = pa-
tients suffered from violent trauma or operative i:·.jury; Group II = pa-
tients with generalized septis.Table 8.5 shows significant increases in
k 0 (from enhanced RBCA) and slight increases 1n k 00 (which can be related
to RBCD lowering as a consequence of effects of transfusion of stored
blood to several patients). Dramatic increase in values of yield stress
a 0 = noo T, close to the values of the Casson's one (See eq.(5.1) ),i.e~lo
se to packing in the core; have been interpreted as increase in cohesion
of RBC Aggregates that could traduce prethrombotic situation for these
patients (QUEMADA, 1976a).
TABLE 8.5 Enhanced RBC Aggregation

'
H n ko k 00
T
sec
Go =n 00 /T He
(cP) (dynes/em~
Gri 32.2 1.28 6.24 2.08 1. 50 0.0193 32.1
Grii 30.2 1 .36 6.49 2.04 1. 54 0.0185 30.8
Normal 42.6 1. 19 4.33 1.61 7.80 0.0035 46.2
Moreover k 0 -values found for patients are critical values, in the

sense that each corresponding critical (packing) value of hematocrit,
H = 2/k is very close to the actual hematocrit of patient, that again
c 0
represents a very dangerous clinical situation, since blood stasis can
then promote complete stopping of blood flow (i.e. a shock) (QUEMADA,
1976 a).
Once a viscosity equation is available, models of blood circulation

1n narrow vessels, .taking into account the presence of the plasma layer,
116 D. Quemada
can be constructed, if one knows independently the Fahraeus Effect, i.e.

the changes of hematocrit in the vessel, compared with the feed hemato-
crit. Such models should be able to explain the role that RBCA plays in
the microcirculation, as L.E. Gelin early showed in a very illuminating
experiment using a X branching tube (GELIN, 1961).
Further aspects, especially physiological ones, have been given 1n
a recent review by 3CHMID-SCHONBEIN et al. (1979).
Number of transient or dynamic measurements have been performed on

pathological blood. (See for instance LESSNER et al., 1971 ; BUREAU et
al., 1976; COULTER and SING, 1974 ; THURSTON, 1976 ; CHIEN et al., 1975).
Clinical interest of such studies are evident but difficulties to
get available models render still incomplete the understanding of expe-
rimental data in terms of fundamental processes.
Some concluding remarks
In these lectures, I tried to analyse in detail some models as tools

to reach a comprehensive level in analysis of viscometric data. Indeed,
data analysis can be only performed with the help of models, which must
be both sufficiently complicated for significant describing of complex
fluids as blood, and simple enough to be of practical use, until clini-
cal routine. One has to hope improvements in the clinicians'belief in the
usefullness of models.
Macroscopic behaviour results in general from material properties at

a smaller scale, sometimes microscopic. However, in most cases, large
scale properties are not the simple addition of small scale properties
one observes "collective" properties, which can be entirely new ones. In
such a sense, some universal properties of disperse systems can be expec-
ted to exist. Therefore simple models for these general properties can
be developed in terms of a small number of variables and must be avai-
lable for blood characterizat ion. Conversely, although Blood and RBC sus-
pensions appear as very complex systems, they present under normal con-
ditions many advantages for the study of non-newtonian behaviour. Indeed,
the particles they contain have precise properties (in the sense of
small deviations from mean values) and are suspended in a fluid (the
plasma) having a well defined physico-chem ical state. The most important
features are the following :
(i) the size of RBC corresponds to a very narrow-calib ration
(ii)their extreme deformability (however associated to a large resistan-
ce to area changes) allows strong orientation effects by flow, dis-
carding too high y values
to avoid any inertial effect or hemolysis.
(iii)Aggregat es of RBC possess a characteristi c shape, the so-called
"rouleau-shap e", and the aggregation mechanism does not seem to de-
pend on the number of RBC in the rouleau, but mainly on RBC-RBC in-
teraction, the latter varying with physico-chem ical properties of
suspending fluid.
(iv)Relaxatio n times associated with Brownian motion are very large and
the observed characteristi c time should be either the Maxwell rela-
xation time of the collective (aggregated) structure or the one as-
sociated with the single RBC deformation (and the resulting orienta-
tion).
Clearly, much more studies are needed to establish unequivocally

the physiologica l significance of (i) the rheological properties of bulk
blood in vitro and blood flow in narrow vessels, (ii) the related models
and (iii) the variables involved in them.
Note added 1n proofs: A very complete information on hemorheology 1n di-
seases can be found in the recent book "Clinical Aspects of Blood Visco-
sity and Cell Deformabilit y"- LOWE G.D.O., BARBENEL J.C. and FORBES C.D.
(Eds.) Springer-Verl ag (Berlin), 1981.
118 D.Quemada
REFERENCES
ANCZUROWSKI E., MASON S.G. (1967), The kinetics of flowing dispersions.

III. Equilibrium orientations of rods and discs - J,Colloid Interface
Sci. 23, 533-546.
ARRHENIUS s. (1917), The viscosity of solutions- Biochem.J. ll• 112-113.
BARBEE J.H. and COKELET G.R. (1971), The Farhaeus effect.- Microvasc.Res.
l· 1-21.
BA1CHELOR G.K. and GREEN J.T. (1972), The determination of the bulk stress
in a suspension of spherical particles to order c2- J.Fluid.Mech.,
56, 401-427.
BAYLISS L. ( 1952)-; Rheology of blood and lymph, In Deformation and Flow
in Biolo~ical Systems, A. Frey-Wissling (Ed), North Holland Publ.,
Amsterdam- Chap.6, pp 355-415.
BERGEL D.H. ( 1972), The rheology of human blood vessels- In Biomechanics:
its foundations and objectives - Y.C. Fung, N.Perrone et M.Anliker
(Eds) Prentice Hall, Inc.New-Jersey, pp.63-103.
BLOCH E.H. (1962) , A quantitative study of the hemodynamics in the living
microvascular system - Amer.J.Anatomy,pp llQ, 125-153.
BORN G.V.R., MELLING A. and WHITELAW J.H.(1978), Laser doppler microscope
for blood velocity measurements - Biorheology, 12• 163-172.
BRINKMAN H.C. (1952), The viscosity of concentrated suspensions and solu-
tions -J.Chem.Phys., 20, 571.
BROCHARD F.(1977), Une bulle degonflee: le globule rouge- La Recherche
n.. 174-177.
BROOKS D.E., GOODWIN J.W. and SEAMAN G.V.F. (1970), Interactions among
erythrocytes under shear- J.Appl.Physiol.,28, 172-177.
BROOKS D.E. and SEAMAN G.V.F. (1971), Role of mutual cellular repulsions
in the rheology of concentrated red blood cell suspensions - In Theore-
tical and Clinical Hemorheology, H.H.Hertert and A.L. Copley (Eds),
Springer Verlag, Berlin, pp 127-135.
BROOKS D.E., GOODWIN J.W., SEAMAN G.V.F. (1974), Rheology of erythrocyte
suspensions: electrostatic factor,s in the dextran-mediated aggregation
of erythrocytes - B.~orheology ll• 69-77.
References 119
BROOKS D.E. (1976) Red cell interactions in low flow states- In Micro-
circulation, vol.I, J.Grayson and W.Zingg (Eds); Plenum Press, New-York
pp 33-52.
BUGLIARELLO G., KAPUR C., HSIAO G. (1965), The profile viscosity and
others characteristics of blood flow in a non-uniform shear field -
Proc.Four.Int.Congr.on Rheology,~. A.L. Copley (Ed), Interscience,
N.Y., pp 351-370.
CHIEN S., USAMI S., JAN K.M. (1971b), Fundamental determinants of blood
viscosity- In The Symposium on Flow, Dowdell (Ed.), Pittsburg.
CHIEN S., LUSE S.A., JAN K.M., USAMI S., MILLER L.H. and FREMOUNT H.
(1971c),Effects of macromolecules on the rheology and ultrastructure
of red cell suspensions - In Proc.6th Europ.Corrf.on Microcirculation,
Karger, Basel (Eds.), 29-34.
CHIEN S., USAMI S., DELLENBACK R.J., BRYANT C.A. and GREGERSEN M.I.(1971d)
Change of erythrocyte deformability during fixation in acetaldehyde -
In Theoretical and Clinical Hemorheology, Hartett H.H. and Copley A.L.
(Eds), Berlin, Springer-Varlag, pp 136-143.
CHIEN S.( 1972), Present state of blood rheology- In Hemodilution: Theore-
tical Basis and Clinical Applications.Messmer K., Schmid-Schonbein H.,
S.Karger (Eds.), Basel, 1-40.
CHIEN S., KING R.G., SKALAK, R., USAMI S. and COPLEY A.L. (1975), Visco-
elastic properties of human blood and red cell suspensions - Biorheolo-
gy, ~. 341-346.
CHIEN s., (1979), Blood rheology- In Quantitative Cardiovascular Studies
Hwang N.H.C., Gross D.R., Patel D.J. (Eds), Univ.Park Press, Baltimore
(USA), 241-287.
COKELET G.R., MERRILL F.W., GILLILAND E.R. and SHIN H. ( 1963). The rheolo-
gy of human blood measurement near and at zero shear rate - Trans.Soc.
Rheol. ,1, 303-317.
COKELET G.R. (1972) Rheology of blood- In Biomechanics, its Foundations
and Perspectives, Y.C.Fung, N.Perrone and M.Anliker (Eds), Prentice
Hall, Inc., Englewood Cliffs, N.J., pp 63-103.
120 D. Quemada
COKELET G.R. (1976) Macroscopic rheology and tube flow of human blood-
In Microcirculation Vol 1. J. Grayson and W.Zingg (Eds), Plenum
Press. New York, pp 9-32
COPLEY A.L., HUANG C.R., KING R.G. (1973) Rheogoniometric studies of
-1
whole human blodd at shear rates from 1000 to 0.0009 sec . Part 1.
Experimental findings - Biorheology _!Q, 17-22. Part II. Mathematical
interpretation. Biorheology, lQ, 23-28.
COPLEY A.L,, KING R.G., CHIEN S., USAMI S., SKALAK R. and HUANG C.R(1975)
Microscopic observations of viscoelasticity of h~an blood in steady
and oscillatory shear - Biorheology ~. 257-263.
COPLEY A.L., KING R.G., HUANG C.R. ( 1976) Erythrocyte sedimentation of
human blood at varying shear rates - Biorheology, l}, ~81-86.
COULTER Jr. N.A., MEGHA Singh (1971)~ Frequency dependence of blood vis-
cosity in oscillatory flow- Biorheology, ~. 115-124
CROSS M.M. (1965) Rheology of Non-Newtonian fluids : A new flow equation
for pseudoplastic systems -Colloid Sci 20, 417-437
DEVENDRAN T. and SCHMID-SCHONBEIN H.,(1975) Axial Concentration 1n Narrow
Tube Flow for Various RBC Suspensions as Function of wall shear stress
- Pflugers Arch. 355:R20
DINTENFASS L.,(1964), Rheology of the packed red blood cells containing
haemoglobins AA, SA, SS - J. Lab. Clin. Med. 64, 594-603.
DINTENFASS L., BURNARD E.D. (1966a) Effect of hydrogen ion concentration
on in vitro viscosity of packed red cells and blood at high hematocrits
- Med. J. Aust. 1, 1072-1078.
DINTENFASS L., JULLIAN D.O. and MILLER G. (1966b), Viscosity of Blood in
normal Subjects and in Patients Suffering from Coronary Occlusion and
Arterial Thrombosis -Am. Heart J., 11, 587-592.
DINTENFASS L., (1968), Inyernal viscosity of the red cell and a blood
viscosity eque.tion - Nature, Lond. 219, 956-957.
DINTENFASS L., (1969), The internal viscosity of the Red cell and the
structure of the red membrane. Considerations of the liquid Crystalli-
ne structure of the red cell interior and membrane from rheological
data- Mol.Cryst. ~. 101-107.
References 121
DINTENFASS L. (1971) Blood Microrheology Viscosity Factors in Blood Flow

-Ischaemia and Thrombosis- Butterworths. London.
DINTENFASS L. and KAMMER S. (1977) Plasma viscosity in 615 subjects. Ef-
fect of Fibrinogen, Globulin, and Cholesterol in Normals, Peripheral
vascular. Disease Retinopathy and Melanoma - Biorheology, ~' 24 7-251 .
DINTENFASS L. (1977) Blood Viscosity factors ln severe non diabetic and
diabetic retinopathy- Biorheology, ~' 151-157.
DINTENFASS L. (1979) Clinical applications of blood viscosity factors and
functions : especially in the cardiovascular disorders - Biorheology
..l2_, 69-84.
DORMANDY J.A. and EDELMAN J.B. (1973) High blood viscosity. An achological
factor in deep venous thrombosis - British Journal of Surgery, 60,
187-189.
DUFAUX J., QUEMADA D., MILLS P. (1980) -Velocity profiles measurements
by Laser-Doppler velocimetry (LDV) in plane capillaries. Comparison
with theoretical profiles from a two fluid model. in : Rheology, Vol 3
G. Astarita, G. Marruci and L. Nicolais (eds) Plenum Press,NY;pp561-566
DIX F.J. and SCOTT-BLAIR G.W. (1940) On the flow of suspensions through
narrow tubes- J. Appl. Physics. 11. 574-581.
FLAUD P., QUEMADA D.(1980) Role des effets non newtoniens dans l'ecoule-
ment pulse d'un fluide dans un tuyau viscoelastique -Revue Phys. Appl.
]2_, 223-233
FISCHER Th. M., SCHMID-SCHONBEIN H., STOHR M.,(1978)- Mechanical behaviour
of human red blood cells in the shear field of viscous dextran solu-
tion- In Cardiovascular and Pulmonary Dynamics. Jaffrin M.Y.(ed),Edi-
tions Inserm Paris pp. 243-256.
FAHRAEUS R. (1929) The suspension stability of the blood- Physiol. Rev.
_2_. 24 1-274
FAHRAEUS R. and LINDQVIST T. (1931) The viscosity of the blood in narrow
capillary tubes - Amer. J. Physiol. 96 : 562-568
GELIN L.E. (1961) Disturbances of the flow properties of blood and its
counter action in surgery - Acta Chirurgia, Scandinavia , 122,
287-295.
122 D. Quemada
De GENNES P.G. ( 1979). Conjectures on the transit_;_on from Poiseuille to

plug flow in suspensions- J. de Physique 40, 783-787.
GILLESPIE T. (1963) The effect of Aggregation and Liquid Penetration on
the viscosity of dilute suspensions of spherical particles - J.Colloid
Sci. ..:!..§_, 32-40.
GOLDSMITH H.L., MASON S.G. ( 1967) "The microrheology of dispersions" -
In Rheology: Theory and Applications. Eirich, (Ed) Acad.Press, N.Y.
pp. 85-250.
GOLDSMITH H.L., (1971) Deformation of human red cells 1n tube flow-
Biorheology l• 235-242.
GOLDSMITH H.L., (1968) The microrheology of red blood cell suspensions-
J. Gen. Physiol. ~. 5s-28s.
GOLDSMITH H.L., (1973) The microrheology of human erythrocyte suspensions
In Proceed. XIIIe Int. Cong. Theor. and Appl. Mech. E. Becker and
G.K. Mikhailov (eds), Springer Verlag, Berlin, pp 85-103.
GREGERSEN, M.I., USAMI S., CHIEN S. and DELLENBACK R.J. (1967) Characte-
ristics of torque-time records on heparinized and defibrinated ele-
phant, human and goat blood at low shear rates (0.01 sec- 1 ) : effects
of fibrinogen and Dextran (Dx 375) - Bibl. anat. 2: 276-281.
HARKNESS W. (1971) The viscosity of human blood plasma. Its measurement
in health and disease - Biorheology - ~. 171-193.
HAYNES R.H. and BURTON A.C. (1959) Role of the non Newtonian behavior of
blood in hemodynamics. Amer. J. Physiol. 197-943.
HAYNES R.H. (1962) The viscosity of erythrocyte suspension- Biophysics
~ 95-102
HEALY J.C., JOLY M.(1975l. Rheological behaviour of blood in transient
flow - Biorheology ~ 335-340.
HOARE E.M., BARNES A.J. and DORMANDY J.A. (1976) Abnormal Blood Viscosi-
ty in Diabetes Mellitus and Retinopathy - Biorheology, lJ, 21-25.
HOUWINK R. (1949) Macromolecular sols without electrolyte character. In:
Colloid Science, II, Reversible Systems, 153. Kruyt H.R. (Ed) Elese-
vier Publ., Amsterdam.
References 123
HUANG C.R., SISKOVIC N., ROBERTSON R.W., FABISIAK W., SMITHERBERG E.H.,
COPLEY A.L. (1975), Quantitative characterization of whole human blood-
Biorheology ~. 279-282.
ISOGAI Y., ICHIBIA K., IIDA A., CHIKATSU I. and ABE M. (1971), Viscosity
of blood and plasma in various diseases - In Theoretical and clinical
hemorheology, Hartett H.H. and Copley A.L. (Eds), Springer-Verlag,
(Berlin) pp 136-143.
KARNIS A., GOLDSMITH H.L. and MASON S.G. (1966), The kinetics of flowing
dispersions I: Concentrated suspensions of rigid particles. J.Coll.
Interface Sci., 22, 531-553.
KELLER J.B., RUBENFELD L.A. and MOLYNEUX J.E. (1967). Extremum principles
for slow viscous flows with applications to suspensions - J.Fluid.
Mech., 30, 97-125.
KELLOG F. and GOODMAN J.R. (1960), Viscosity of blood myocardial infarc-
tion - Circulation Research, ~ , 972-978.
KLOSE H.J., VOLGER B., BRECHTELSBAUER H., HERNICH I. and SCHMID-SCHONBEIN
H. (1972) - Microrheology and light transmission of blood I. The pho-
tometric quantification of red cell aggregation and red cell orienta-
tion - Pflilrers .Arch., 333, 126-132.
KRIEGER I.M. and ELROD H. (1953) -Direct dete¥mination of the flow cur-
ves of non-newtonian fluids II: Shearing rate in the concentric cylin-
der viscometer- J.Appl.Phys., 24, 134-140.
KRIEGER I.M. (in Surface and Coatings Related to Paper and Wood. R. Mar-
chessault, C.Skaar ed. Syracuse Univ.Press (1967) ) and T.J.DOUGHERTY,
some problems in the theory of colloids, (Ph.D.Thesis, Case Inst.Techn.
( 19 59) ) .
KRIEGER I.M. (1963), A dimensional approach to colloid rheology- Trans.
Soc.Rheol., l , 101-109.
KRIEGER I.M. and DOUGHERTY T.J. (1959). A mechanism for non-newtonian
flow in suspensions of rigid spheres - Trans.Soc.Rheol., }, 137-152.
LANDEL R.F., MOSER B.G. and BAUMAN A.J. (1965), Rheology of concentrated
suspensions. Effect of a surfactant - In Proceed. rvth Intern.Cong. on
Rheology, Part 2, Lee E.H. (Ed), Interscience, N.Y., PP 663-693.
124 D. Quemada
LESSNER A., ZAHAVI J., SILBERBERG A., FREI E.H., and DREYFUS P. (1971)
The viscoelastic properties of whole blood In Theoretical and Clini-
cal Hemorheology H.H. Hartert and A.L. Copley (eds.) Springer-Ver-
lag. New York, pp. 194-205
LINDSLEY H., TELLER D., NOONAN B., PETERSON M. and MANNIK M. (1973).
Hyperviscosity Syndrome in Multiple Myeloma. A reversible, concentra-
tion dependent Aggregation of the Myeloma Protein - The Amer of Medi-
cine, 54, 682-688.
MARON S.H. and SISKO A.W. (1957) Application of Ree-Eyring generalized
flow theory to suspensions of spherical particles: II. Flow in low
shear region- J. Colloid.Sci., ~. 99-107.
MAUDE A.D. and WHITMORE R.L. (1958) Theory of the Blood Flow 1n Narrow
Tubes- J. Appl.Physiol. ~: 105-113.
MATSUDA T. and MURAKAMI M. (1976). Relationship between fibrinogen and
blood viscosity - Thrombosi~ Research, Suppl.II, ~. 25-33.
MERRILL I. W. , MARGETTS W. G. , COKELET G. R. , BRITTEN A. , SALZMAN E. W. ,
PENNELL R.B. and MELIN M. (1955) Influence of plasma proteins on the
rheology of human blood. In : Proc.4th Inter.Cong.on Rheologg. A.L.
Copley (ed.) Pt, 4, Interscience(Wiley), New York. pp 601-12.
MERRILL E.W., PELLETIER G.A. (1967) Viscosity of human blood : transi-
tion from newtonian to non-newtonian- J. Appl. Physiol. 23,178-182.
MERRILL E.W., (1969) Rheology of blood- Physiol.Rev. 49 : 863-888.
MIDDLEMAN S. (1972) Transport phenomena in the cardiovascular system
Wiley-Interscience, N.Y. p. 91.
MILLER L.H., USAMI S. and CHIEN S. (1971) Alteration in the rheologic
properties of Plasmodium Knowlesi - infected red cells. A possible
mechanism for capillary obstruction - J.Clin.Invest. 50, 1451-1455.
MILLS P., QUEMADA D. and DUFAUX J. (1980) An optical method for studying
RBC orientation and aggregation in a Couette flow of Blood Suspension.
In: Rheology l G. Astarita, G. Marrucci, L. Nicolais (eds) Plenum
Press NY 1980. pp. 567-572.
MOONEY M.(1951) The viscosity of a concentrated suspension of spherical
particles - J.Colloid Sci. ~. 162-170.
References 125
MOORE F. (1959) (Cited by Cheng et Evans, 1965). Trans.Brit.Ceram.Soc.~,
470-492.
OSTWALD W., AUERBACH R. (1926) Uber die Viscositat kolloider Losungen 1m
Struktur, Laminar - und Turbulenzgebiet. kolloid z. ]&, 261-280.
PALMER A.A.(1968) Some aspects of plasma skimming. In:Hemorheology- A.L.
Copley (ed.) Pergamon Press, Oxford. pp. 391-400.
PRAGER S. (1963). Diffusion and viscous flow in concentrated suspensions.
Physica 29, 129-139 ( 1963).
QUEMADA D. (1976a). Red cell Aggregation and Thrombus formation: a rheolo-
gical approach - Proceedings of the 16th International Congress of He-
matology: Topics in Hematology. KYOTO, 1976. Excerpta Medica (Amster-
dam) 415, 733-736.
QUEMADA D. (1976 b). Some new results in rheology of concentrated disper-
se systems and blood. In : Proceedings of the VIIth International Con-
gress on Rheology. J. Kubat (ed.) Gothenburg, 1976, pp.628-629.
QUEMADA D. (1977) Rheology of concentrated.disperse system and minimum e-
nergy dissipation principle. I. Viscosity-concentration relationship.
Rheol. Acta 1£, 82-94.
QUEMADA D. (1978 a) Rheology of concentrated disperse systems, II. A model
for non newtonian shear viscosity in steady flows. Rheol. Acta fl,.
632-642.
QUEMPDA D. (1978b) Rheology of concentrated disperse systems. III. Gene-
ral features of the proposed non-newtonian model. Comparison with expe-
rimental data - Rheol. Acta fl, 643-653.
QUEMADA D., DUFAUX J., MILLS P.(1980) A two-fluid model for highly concen-
trated suspension flow through narrow tubes and slits: velocity profi-
les, apparent fluidity and wall layer thickness - In :Rheology, Vol.3
G.Astarita,G.Marrucci and L.Nicolais (eds) Plenum Press, NY pp.633-638
QUEMADA D. (1981) -A rheological model for studying the hematocrit depen-
dence of red cell-red cell and red cell-protein interactions in blood
Biorheology ~. 501-516 .
. QUEMADA D., MILLS P., DUFAUX J., SNABRE P., LAMBERT M. ( 1981) Sedimenta-
tion effects in viscometric measurements - in:Hemorheology and Diseases.
126 D. Quemada
J.F.Stoltz and P.Drouin(Eds)- Doin editeurs. Paris- pp 31-41.

REE T., EYRING H. (1955)- Theory of non-newtonian flow. I. Solid Plastic
System. J.Appl.Phys., 26, 793-804.
ROBINSON J.V. (1949) The viscosity of suspensions of spheres - J. Phys.

and Colloid Chem. 53, 1042-1056.
ROSCOE R. (1952) The viscosity of suspensions of rigid spheres- Brit. J.
Appl. Phys. }, 267-269.
RUCKENSTEIN E. and MEWIS J. (1973) Kinetics of Structural Changes in Thi-
xotropic Fluids -Colloid and Interface Sci 44, 532-541
SCHMID-SCHONBEIN H., WELLS R.E., GOLDSTONE J. (1971) Fluid drop-like beha-
viour of erythrocytes. Disturbance in pathology and its quantification
Biorheology l• 227-234.
SCHMID-SCHONBEIN H., WELLS R.E. (1971) Red cell aggregation and cell de-
formation : their influence on blood rheology in health and disease.
In : Theoretical and Clinical Hemorheology, Hartet H.H. and Copley A.L.
(eds) Berlin, Springer-Verlag, pp. 348-355.
SCHMID-SCHONBEIN H., GALLASCH G., VOLGER E., KLOSE H.J.(1973) Microrheolo-
gy and protein chemistry of pathological red cell aggregation (blood
sludge) studied in vitro- Biorheology 1Q, 213-227.
SCHMID-SCHONBEIN H. (1975) Erythrocyte rheology and the optimization of
mass transport in the microcirculation - Blood Cells£ 285-306.
SCHMID-SCHONBEIN H. (1976) Microrheology of erythrocytes, blood viscosity
and the distribution of blood flow in the microcirculation. In : In-
ternational Review of Physiology. Cardiovascular Physiology - A.C. Guy-
ton and A.W. Cowley (eds) University Park Press. Baltimore pp. 1-62
SCHMID-SCHONBEIN H., FISCHER T., DRIESSEN G., RIEGER H. -Microcirculation
In : Quantitative Cardiovascular Studies: Clinical and Research Appli-
cations of Engineering Principles, N.H.C. Hwang, D.R. Gross and D.J.
Patel (eds) Univ. Park Press, Baltimore (1979), Chap.8, 353.
SCHOLZ P.M., KARIS J.H., GUMP F.E., KINNEY J.M. and CHIEN S. (1975)
Correlation of blood rheology with vascular resistance in critically
ill patients- J.appl.Physiol. 39 1008-1011
References 127
SCOTT BLAIR G.W.(1959) An equation for the flow of blood, plasma and se-
rum through glass capillaries -Nature 183, 613-615.
SEGRE G. and SILBERBERG A. (1962) Behavior of macroscopic rigid spheres
1n Poiseuille flow. J. Fluid Mech. ~. 115-135 : 136-157
SUTERA S.P. (1978) Red cell motion and deformation in the microcircula-
tion- In : Cardiovascular and Pulmonary Dynamics. Jaffrin M.Y~d. Edi-
tions,Inserm Paris pp.221-242.
TAYLOR G1 (1932) The viscosity of a fluid containing small drops of ano-
ther fluid- Proc.Roy.Soc.(London) 138A 41-45
THOMAS H.W. (1963) The Wall Effect in Capillary Instruments, Biorheology
1 : 41-56.
THURSTON G.B. (1976) The viscosity and viscoelasticity of blood in small
diameter tubes- Microvasc. Res. l2:133-146
THURSTON G.B. (1979a) Erythrocyte Rigidity as a Factor in Blood Rheolo-
gy : Viscoelastic Dilatancy - J. of Rheology, 23, 703-719.
THURSTON G.B. (1979b) Rheological parameters for the viscosity viscoelas-
ticity and thixotropy of blood - Biorheology, ~' 149-162.
USAMI S., CHIEN S. and GREGERSEN M.I.(1971)- Viscometric Behavior of
Young and Aged Erythrocytes - In Theoretical and Clinical Hemorheolo-
gy, Hartett H~H. and Copley A.L.(eds) Springer-Verlag, Berlin,136-143.
VAND V. (1948) Viscosities of solutions and suspensions- J. Phys.Coll.
Chem. ~. 277-299.
VINCENT N.M., OLIVER D.R.(1977) Blood sedimentation at controlled shear
rates - Biorheology , ~' 51-58.
WEINBERGER C.B. and GODDARD J.D. (1974) Extensional flow behaviour of
Polymer solutions and particle suspensions in a spinning motion - In-
tern. J. Multiphase Flow, 1, 465-486.
WHITMORE R.L. (1967) -A theory of blood flow 1n small vessels- J. Appl.
Physiol., 22, 767-771.
CHAPTER II
THE ARTERIAL WALL
J.C. Barbenel,
Bioengineering Unit
University of Strathclyde,
Wolfson Centre,
Rottenrow
GLASGOW G4 ONW. U.K.
1. THE GEOMETRY AND STRUCTURE OF ARTERIES
1.1. Introduction
The pioneer of the quantitative study of the flow of blood through
tubes was Poissuille, who was both physicist and physician. He modelled
the flow of blood through the circulation by investigating the flow of
water through rigid cylindrical tubes. He was forced to use water as he
was unable to prevent the blood from clotting, and he was fortunate ~n
his substitution because blood shows non-Newtonian behaviour, which will
be dealt with elsewhere in this volume. He was also fortunate in using
rigid tubes. The real blood vessels have complicated non-cylindrical
geometries, and are highly extensible and non-linear. If he had used
blood and realistic models of blood vessels, it is unlikely he would

130 J .C. Barbenel
have produced the clear cut results he did.
This section will discuss the arterial wall, both geometry and
mechanical properties. The data from different sources gives values of
parameters, often markedly different and often conflicting. There are
several reasons for this.
Results have been obtained from measurements made both in living
and dead animals. The blood vessels are highly deformable and flaccid
when empty. The reported values of thickness, diameter, length, etc.,
may be open to some doubt as the methods used to measure these variables
often lead to changes in dimensions due to forces applied by the
measuring instruments. During life the large vessels are under a state
of longitudinal tension, and when cut transversely will retract.
Further retraction will occur if they are dissected free of the surrounding
tissue (McDonald, 1974. 1 ). Thus data obtained on excised specimens may
be distorted by the loss of this resting tension. Similarly during life
the vessels are filled with blood at a varying pressure, and it 1s often
difficult to decide what physiological significance some of the in vitro
data has. Finally there is usually a very considerable range of values
for the same feature of structure measured in different animals.
1.2. The Size of Arteries
TPe circulation within the arterial system is usually discussed in
terms of large and small arteries and arterioles. The dividing line is
usually arbitrarily set at vessels having a diameter of about 0.1 mm.

The Geometry and Structure of Arteries 131
The difference in size also reflects important differences in flow
conditions. The relative importance of the inertial forces can be
expressed ~n the Reynold's Number Re;
Ud (1)
Re
v
Where U ~s the mean flow velocity,
d the diameter of the vessel
and v the kinematic viscosity.
In the larger vessels Re is large and inertial forces more important than
the viscosity. In vessels of less than 0.1 mm diameter Reynold's number
is typically less than unity, and viscous forces become increasingly
important. The red blood cell has a diameter of 7-8 ~m and the 0.1 mm
diameter is also a point at which the structure of the blood becomes
important, and it becomes less realistic to treat the blood as a continuum.
The small vessels are extremely important, but most mechanical
testing has been carried out on the larger vessels and the results
extrapolated to the smaller structures.
1.3. The Geometry of the Large Vessels
The aorta is the major artery carrying the blood from the left
ventricle. The vessel is curved three dimensionally, first running in
the direction of the head, where it gives r~se to branches suppiying the
head and upper limbs. The aorta then curves to run caudally (Figure 1.1).
After this follows a nearly straight course, giving rise to arteries
supplying the organs in the abdominal cavity, and terminates by dividing

132 J.C. Barbenel
Arteries to Head
and Upper Limbs
Thorax
Intercostal Arteries
Diaphragm
Arteries to
Abdominal Orga~
Abdomen
Arteries to
Lower Limbs
Figure 1.1 Schematic diagram of aorta
into two major vessels supplying the lower limbs. In the dog, and
similar animals, a third terminal branch runs to the tail.
Most of the detailed information on artery size, diameter etc.,
has been derived from dogs. This is the most reliable data, but great
care must be used in attempting to apply or extrapolate the results to man.

The aorta tapers along its length, and in dogs it can be described
by the exponential equation
A A (2)
0
A is the aortic area at length x from the start of the vessel
where the area and radius ~s A and R B is a parameter, the value of

0 0
. h 1'~es b etween 2 - 5
wh ~c X l0- 5 (C aro et al, 1978 2 ). Details of the
cross sectional area of the arteries of dogs, at· selected sites, will
be found in McDonald, 1974.
1.4. The Area of Branches and Branching Angles
The cross sectional area of the aorta decreases regularly with
the distance from the heart, but because of the presence of branches,
the total arterial cross section actually increases. There is a
considerable body of data on the cross sectional area of dog arteries
obtained from casts made by expanding the arteries with elastomers.

3
This is discussed at length by Iberall, 1967 .
The angle between the arterial branches and the vessel from which
they arise is usually called the branching angle. There is a very wide
range of angles found in the circulation, but a lack of detailed
information. It has been known, however, that there is general
relationship between the branching angle and the relative size of branches
4
(d'Arcy Thompson, 1945 ):
(i) if the artery branches into two equal branches (e.g. the
common iliac arteries) the branching angles are symmetrical.

134 J.C. Barbenel
~
<1l
c0
....
+J
u
(!)'"C)
Cfl (!)
~
Cll
Cll4-l
0 0
H
u
~
<1l
+J
0
H
10 100 1000 10000

Vessel Diameters ~m
Figure 1.2 Cross sectional area of arterial tree (After Iberall,

1967)
(ii) if the branches are of unequal s1ze the smaller vessel has a
larger branching angle (e.g. internal and external iliac
arteries).
(iii) branches very much smaller than the parent artery have a
branching angle of about 90° (e.g. posterior intercostal
arteries).
1.5 The Structure of the Arterial Wall
The walls of arteries are neither uniform nor homogeneous but
consist of a variety of cells and extracellular proteins. The
arrangement of these components and the structure of the arterial wall is
described in the standard text books of histology (e.g. Bloom and

Fawcett, 1975 5), and a detailed description will be found in Benninghoff
(1930 6 ).
The dominant structural materials are the fibrous proteins, collagen
and elastin, and the associated glycosominoglycan ground substance. The
major cellular component is smooth muscle. The organisation of these
components in the blood vessel wall is similar in all arteries and they
are arranged in three concentric layers - the tunica intima, which ~s
the innermost layer and surrounded by the tunica media and surrounding
both is the tunica adventitia. Each layer is separated by a layer of
elastin, the internal elastic lamina demarcating the intima-media junction
and the external elastic lamina marking the junction between the media
and adventitia. In each of the three layers there is a dominant
structure and cell type.
The tunica intima consists of a single, thin (0.5 ~m) layer of
endothelial cells which forms a continuous lining to the circulatory
system. The cells are mechanically fragile but damaged or lost cells
can be replaced by regeneration. Beneath the endothelial cells there ~s
a thin layer of collagen fibres.
The internal elastic lamina beneath the adventia consists of a
dense layer of elastic fibres, which has also been described as an elastic
membrane.
The tunica media ~s usually the thickest layer of the arterial wall,
136 ].C. Barbenel
Figure 1.3 Diagramatic cross section of the arterial wall
A Tunica Intima, with a layer of endothelial cells

and connective tissue.
B Tunica Media, showing elastic laminae connected by

oblique smooth muscle cell.
C Tunica Adventia.
and contains elastin and collagen fibres, and smooth muscle cells. There
is a considerable variation in the amount of elastin and muscle in
different arteries and this leads to the description of arteries as
being either elastic or muscular. There is no sharp demarcation between
the two types of arteries but the major vessels nearer the heart are
The Geometry an<l Structure of Arteries 137
predominantly elastic and the distant arteries muscular.
The tunica media of a typical elastic artery consists of multiple
concentric layers of elastin separated by thin layers of collagen and
smooth muscle cells. The thickness of these layers of elastin is
relatively constant and they appear in the large arteries of all mammals
(Wolinsky and Glagov, 1967. 7 ). Thus the number of layers ~s approximately
proportional to the wall thickness. The elastic fibres in the layers are
arranged in a ·spiral pattern.
In the more distal arteries the media consists mainly of spirally
arranged smooth muscle cells, which are arranged in concentric layers
separated by small amounts of collagen and elastin. The number of layers
of muscles diminishes with the radius of the vessel.
The external elastic lamina is a thin layer of elastin, less clearly
defined than the internal lamina.
The tunica adventitia consists of loose connective tissues
containing both collagen and elastin fibres which are ~n a predominantly
longitudinal direction. The outer demarcation ~s not clearly defined and
merges with the tissue surrounding the vessel.
The nourishment of the intima and inner layers of the media ~s
derived from the blood flowing through the vessels, but arteries greater
·than 1 mm in diameter have small vessels called vasa vasorum supplying
the vessel wall.

138 J.C. Barbenel
Internal Wall
Diameter Thickness
Aorta 25 mm 2mm
4mm 1 mm
30 11m 20 11m
Figure 1.4 Relative thickness of relaxed arterial wall

(After Burton, 1962 8 )
Equivalent values for vessels in which the muscle of

the tunica media is contracted will be found in
Folkow and Neil (1971 9 ).
1.6. Thickness of the Arterial Wall
The wall thickness of the larger arteries varies with the size of
the animal. The results of Wolinsky and Glagov (1967) indicate a rapid
increase in thickness with body weight for small animals, but for large
animals, those which weigh more than 60 Kg, the thickness is relatively
constant. The diameter of the arterials also changes in a parallel

manner, and thickness to diameter ratios appear to be relatively constant,
being-:!= 0.1.
In the smaller peripheral arteries the wall becomes thinner, but the
thickness to diameter ratio actually increases, and for arterioles the
ratio may be 0.4.
The size of the arteries also change with age, and during growth
both wall thickness and diameter increase, but the ratio remains
relatively constant (Caro et al, 1978). The increase in thickness of the
wall is due to thickening of and an increase in number of the elastic
lamellae of the tunica media. After maturity has been reached there is
a continuing increase in diameter and wall thickness (Learoyd and Taylor,
1966 10 ).
2. EXPERIMENTAL MECHANICS OF THE ARTERIAL WALL
There is considerable disagreement on how the properties of the
arterial wall should be analysed. Before discussing these areas of
uncertainty this section will review, in descriptive terms, what is the
generally accepted view of the mechanics of the arterial wall.
The mechanical properties of the arterial wall have been investigated
both in quasistatic and dynamic test modes. Tests have been made on
arterial segments in vivo - that is with the specimen still in place in
the living experimental animal but the majority of results have been
obtained in vitro on excised arteries. These excised arteries may be
maintained at the in vivo length or allowed to relax before testing. In

140 J.C. Barbenel
addition there are numerous results which have been obtained from strips
on rings cut from the wall of excised vessels.
The results of these tests may differ in the magnitude of the
descriptive parameters obtained, but the general form of the load-
deformation response of the walls of the larger vessels have been clearly
established. Not unexpectedly, the response is similar to other soft
connective tissues and this reflects the fact that the behaviour of these
tissues is controlled by the properties, arrangement and mechanical
properties of the collagen and elastic fibres, and ground substance
common to all these tissues.
2.1. Quasistatic load-extension response
The load-extension response obtained from excised strips and the
equivalent pressure-radius results obtained from inflated arterial
segments are all markedly non-linear. There is an initial phase during
which small load increments produce large increases ln deformation
(Figure 2 .1) . The vessel becomes stiffer as the load (or deformation)
increases. The stiffness increases monotonically, although at an ever
decreasing rate. As well as being non-linear the vessels are non-elastic.
When the specimen is unloaded, load-deformation curve does not coincide
with that obtained for increasing load.

Experimental Mechanics 141
1 1.5
Extension Ratio, A
Figure 2.1 Typical load-deformation curve for excised artery.

Uniaxial Tension.
The vessel response also shows preconditioning, a progressive change
in response produced by repeated load or deformation cycling (Figure 2.2).
Typically the curvature of the load-deformation curve increases and the
magnitude of the hysteresis curve decreases as the cycles are repeated.
The alteration in behaviour grows progressively smaller as the number of
cycles increase, and after a relatively small number of cycles (usually
less than ten) a stable and repeatable response is obtained. As Fung
(198t~ has pointed out, the specimen is preconditioned only for the
specific test cycle used. If this is altered the specimen must be
preconditioned anew using the required test cycle. The majority of

142 J.C. Barbenel
1st cycle
5th cycle
Figure 2.2 Preconditioning in specimen of excised artery.

Uniaxial Tension.
experimental results which are reported in the literature has been
obtained on preconditioned tissue.
2.2. Time Dependence
Hysteresis is typical of materials which show time dependence, and
other aspects of this behaviour can be shown in arteries. An important
exception is the insensitivity of load-extension and hysteresis behaviour

-3 -1
to strain rate. Tests carried out in the range 10 - 1 s show little
12
change in response (Tanaka and Fung, 1974 ),and rate dependence appears
only at high rates (Collins and Hu, 1972 13 ). This feature is found in
other soft connective tissues.

Stress Relaxation at constant strain has been reported, mainly in
experiments made with strips and rings of tissue. The response is typical
of non-linear viscoelastic materials, with greater relaxation occurring
at higher strains (Figure 2.3).
The amount of relaxation is also site dependent, with muscular
arteries showing greater relaxation than elastic vessels.
The relaxation is most rapid at short times, and the rate of
relaxation decreases monotonically with time. If the stress is plotted
against the logarithm of elapsed time, an almost linear relationship is
usually obtained.
1.0
Cll
Cll
Q)
""'"""'
.u 0
(/) .._,
b
w-......
:>"""'
•.4 .u
.u .._,
"'
"""'Q)
b
t:>::
0.6
1 10 100
Time, s,
Figure 2. 3 Stress relaxation in excised artery
Creep behaviour has also been reported both 1n excised strips and in
pressurised arterial segments (e.g. Wiederhielm, 1965 14 ). As with stress
relaxation, the response is non-linear and site dependent.

144 ].C. Barbenel
Dynamic behaviour can be characterised in terms of the storage

modulus and phase angle. There is general agreement that the storage
modulus shows little frequency dependence above about 2Hz, but it
decreases slowly down to 10

-3
Hz. The ratio of the storage modulus to the
quasistatic value is site dependent. The ratio is higher (1.6-2) in the
peripheral muscular arteries, than in the central elastic vessels where
the ratio is 1.1.
The phase angle, which is commonly about 10°, also shows little
variation with frequency.
2.3. Anisotropy
When arteries are inflated they elongate, and if the wall is
incompressible, this implies that the vessel is anisotropic. Load-
elongation measurements on excised strips, the measurement of wall strains
during inflation of arterial segments, and wave transmission studies, have
all confirmed the anisotropy.

15
Patel and Fry (1969 ) showed that during inflation of arterial
segments to physiological pressures, the shearing strains which were
produced were very much smaller than the direct strains. The results
suggest that the wall shows elastic symmetry and that it is cylindrically
orthotropic. In these circumstances, the number of elastic constants
necessary to describe a compressible material is reduced to nine, which
may be taken as three orthogonal tensile moditi, three Poisson's ratios
and three shear moduli. If the wall material 1s incompressible, the
number is further reduced (Patel and Vaishnav, 198o 16 ).

The results have been confirmed by other workers and by other means,
and there is general agreement that the vessel walls show cylindrical
orthotropy. How to treat the walls quantitatively is less generally
agreed.
The results of Tanaka and Fung (1974), indicate the existence of
anisotropic time dependence. Strips of artery cut longitudinally showed
greater, and more rapid, relaxation than strips cut circumferentially.
2.4. Compressibility of the Arterial Wall
The arterial wall has a high water content and it has been thought
that it is incompressible. Lawton (1954 17 ) stretched strips of aorta in
saline and showed only a small change in volume, and concluded that the
aortic wall was essentially incompressible. Similar experiments by
Carew et al (1968 18 ) on inflated and stretched arterial segments showed
that the relative volume change on straining was less than 1%, and
confirmed the incompressibility of the tissue. Both experiments would
fail to demonstrate a volume change due to loss of fluid from the arterial
wall. Berge! (1972 19 ) suggests such a mobile component of the water
within the wall is unlikely. Such a change in volume due to fluid loss
does, however, appear to occur during in vitro tests on skin, and has
20
been reported for eat's skin (Veranda and Westmann, 1970 ), also for
human skin (Stark and Al Haboubi, 1981 21 ).
Tickner and Sacks (1967 22 ) made radiographic estimations of the
alteration in wall thickness during simultaneous inflation and elongation
of human and canine arteries and reported a change in volume. For the
146 J.C. Barbenel
volume to remain unaltered, the sum of the Poisson's ratios of the
material must be 1.5, and the data of Tickner and Sacks show that this is
not the case. The use of radiography to determine wall thickness may
lead to considerable errors. It has also been suggested that the results
were due to water loss and drying during the inflation process. In order
for the volume to decrease, the sum of two of the Poisson's ratios should
exceed unity. This certainly appears to occur in some tests (Tickner
and Sacks, Figure 7a) but is not a general result.
Although the evidence for arterial incompressibility may be less
convincing than much of the literature suggests, the wall can usefully be
considered incompressible.
2.5. Structural Basis of Mechanical Response
There has been considerable interest in interpreting the overall
mechanical properties of the soft connective tissues in terms of the
arrangement and properties of their constituents. The major structural
components of the arterial wall are the elastin and collagen fibres and
muscle cells found in the tunica media. At physiological distending
pressures, the components are well oriented and form a series of well
defined layers (Wolinsky and Glagov, 1964 23 ).
Elastin has been extracted from ligamentum muscle and tested in
uniaxial tension (Carton et al, 1962 24 ); the results showed considerable
non-linearity.
More recent work (Jenkins and Little, 1974 25 ) showed the material to
be more linear with a tensile modulus of c.0.5 M Pa. All results show,
Experimental Mechanics 14 7
however) that the material is highly extensible and almost elastic, with
time dependence being small (Gosline, 1976 26 ).
Collagen occurs as crimped fibres, and during fibre strengthening
the fibres may undergo considerable extensions. The nature of the crimp,
and therefore the magnitude of the extension is widely variable.
The straightened collagen fibres are stiffer than elastin, with a
tensile modulus about two orders of magnitude greater (Bergel, 1972;
Haut and Little, 1972 27 ). The fibres show considerable time dependence.
There is considerable disagreement about the mechanical properties
of the muscle cells in the wall, but it appears that the active
contractile force produced by the cells may be equivalent to a modulus of
the order of 0.1 M Pa.
Studies (Dobrin and Rovick, 1969 28) of the static properties of
elastic arteries in which the muscle tone was abolished by potassium
cyanide or heightened by nor-ephinerphrine showed the muscle to have a
significant effect on the circumferential modulus. There was almost no
effect on the longitudinal properties.
Between the muscle cells and the elastin and collagen fibres, there
is abundant glycosaminoglycen ground substance. This material is
associated with a large amount of fluid, and up to 70%, by weight, of the

29
vessel wall can be made up of water (Harkness, et al 1957 ). It is not
clear whether any of this water is mobile. If such a mobile component
exists it may be unable to take part in the load transmission mechanisms
of the arterial wall, and may be extruded on straining, thus making the
wall effectively compressible. Recent studies (McCrum and Darrington, 1976 30 )

148 J.C. Barbenel
suggest that when elastin is stretched, the material absorbs water, which
is then desorbed on releasing the fibres. This implies that there must
be some mobile water present. The authors suggest that this also means
that elastin must be considered to be compressible.
There is a considerable site variation of the ratio of elastin to
collagen, (McDonald, 1974). In the thoracic aorta there is about twice
as much elastin than collagen. At the diaphragm, this falls so that
there is about three times as much collagen as elastin, and this ratio
persists in the muscular arteries.
The geometry and properties of the elastin and collagen fibres has
led to the suggestion that the former is stretched during the initial
phase of extension of the wall and that during this period, the crimps are
lost from the collagen which is important at the stiffer, high extension
phase. Evidence to support this has been obtained from studies in which
either component was enzymatically degraded (Roach and Burton, 1957 31 ,

32
Hoffman et al, 1973 ).
3. STRESS-STRAIN RELATIONS
The walls of the arteries undergo large deformations, showing a
non-linear relationship between stress and strain, and are time dependent.
Most studies of the stress-strain response of blood vessels have assumed
that the effect of time dependence can be rendered insignificant if
suitable test methods are used. The favourite methods are the application
of constant loading or deformation rates (e.g. Tanaka and Fung, 1974) or
incremental measurements made after the majority of the creep or

Stress-Strain Relations 149
relaxation has occurred (e.g. Patel et al, 1969 33 ). It is usual in
these conditions to talk of the elastic properties of the arterial wall.
The problem of the large strain non-linearity can be tackled by the
use of a suitable non-linear strain measure. Fung(l972 34 ) showed that
the gradients of the load (I)/extension (A) plots obtained by constant
strain rate tests were directly proportional to the load,
I
~~ = a(T + 8) a,S constants
(3)
at.
hence T = ce -a
The exponential formulation was incorporated into a stored energy function
to describe arterial elasticity by Vito, (197T 5 ). A more detailed
study by Tanaka and Fung (1974) showed that the linearity was confined
to the large strain region, and did not describe the total stress-strain
curve.
There are two theoretical approaches by which general constitutive
relationships may be derived.
The pressure of blood in an artery during function is such that the
vessel normally operates in an initially strained state, with extensions
being as large as 70%. The strain variations about this resting state
are very much smaller (Patel et al, 1969). This suggests that
constitutive relations be sought in terms of incremental stress-strain
relations (Biot, 1965 36 ). The suitability of the method for application
to normal function is self-evident. There are, however, major
disadvantages if a wide range of strains are to be considered, as the

150 J.C. Barbenel
response around a large number of average strains will require to be
investigated. It may also be difficult to compare incremental moduli
obtained under different conditions.
The alternative approach in which general three dimensional stress-
strain relations are evaluated has relied heavily on the stored energy
function formulation which has been applied to elastomers. The theory
is very general, and allows the calculation of incremental moduli if
desired. It is, however, of considerably greater complexity than the
incremental approach.
3.1. Incremental Elasticity
Bergel (1961 3 ~investigated the variation of the radius of excised
dog artery subjected to different internal pressures. The vessel was
maintained at its in vivo length.
The vessel was assumed to be isotropic and incompressible. The
Young's modulus (E) of such a tube of fixed length is given by
2 2
lip 2(1-v )R. R
1 0 (4)
E
where lip 1s the change in internal pressure, liR the change in external
0
radius R , R. the internal radius and v Poisson's ratio, which was

0 1
taken as 0.5 (see Bergel 1972 for derivation).
The relationship in the above equation was adapted to provide an
incremental modulus Einc at a pressure p 2 (with radii Ro 2 and Ri 2 ) from
measurements made at a lower pressure p 1 (external radius Ro 1 ) and at a
higher pressure p 3 (external radius Ro 3 ) when

(5)
(R 3 - R 1 )(R 2 - R~)
0 0 0 1
The results indicate that the magnitude of the incremental modulus
was an increasing function of the internal pressure. The modulus
increased approximately fifteenfold with an increase of pressure from
40 to 220 rnrn Hg(S.2- 29k Pa).
The major limitation of the analysis was the assumption of isotropy.
Even if the vessel was isotropic 1n tqe re~ting state, the strains
produced by stretching it to the ~n vivo resting length would produce

unequal strains in the three principal directions and, therefore,
anisotropy.
The evaluation of the anisotropic incremental moduli for the aorta
was made by Patel et al (1969). The wall was once again treated as incom-
pressible but orthotropic. Tests were made on dogs in viva, with in vitna
comparisons corrected for arterial tethering affects (Patel and Fry, 1966 38 ).
The vessel was considered to be a thin walled cylinder and
described in cylindrical co-ordinates such that the z co-ordinate
corresponded to the centre line of the aorta, the r co-ordinate to the
radial direction and 8 to the circumferential direction. For an arterial
segment of length L, midwall radius R and thickness h subjected to an
internal pressure p and longitudinal tension f the incremental strains
in these three co-ordinate directions will be given by
h-h
!1h 0
e (6a)
r h h
0
l1R R-R
0 0
ee (6b)
R R
0 0
152 J.C. Barbenel
e (6c)
z
The stresses in these three directions were:
pR/h (7)
a f/21TRh
z
The equivalent incremental stresses were denoted by Pr' P6 and P .

z
The incremental normal strains and stresses are related by the
matrix equation
(8)
with Cij = Cji. The assumption of incompressibility requires that

0. The constants Cij ar.e related to the incremental
Young's moduli (E 6 , Ez and Er) and Poisson's ratios (vez' vzr' vre) by:
1 1 1 (9)
Ee = - - E =c- E = --
cee z r c rr
zz
and
vze vez
c = --
z.e Ee E
z
v zr v rz
czr = --
E E
(10)
r z
vre ver
ere = -Ee- E
r
The values of the Young's moduli were evaluated for the thoracic
aorta of dogs. The vessel was maintained at one of three chosen initial
lengths and the internal pressure increased in incremental steps of
20 em H2o (1.9k Pa). Each was maintained for 1 minute to allow creep,
and the radius and longitudinal load measured.
The pressure, radius, and longitudinal loads were used to calculate
the incremental strains and stresses. The incremental stress may be

31
calculated directly from the measured variables (Fung, 1977 ) but were
calculated by Patel et al from the stresses calculated, using equations
6 & 7 for each state of strain.
The results indicated that the incremental stresses and strains
were small compared to the average strains, thus justifying the use of
incremental theory.
?:: ?::
Ul
1.2
::l
......
::l
'"CI
0
~
......
...."' "'
~~
~ ~Er
1-1
tJ
~Er
;; 0.4
1.4 1.6 1.4 1.6

>.
"e z
Figure 3.1 Variation of incremental moduli with extension ratio.

154 J.C. Barbenel
The magnitude of the moduli increased with the extension ratios
Ae and Az as would be expected for a non-linear material. In general
E > Ee > E at physiological pressures.

z r
The vessel walls are clearly anisotropic. The errors produced by
40
treating an orthotropic tube as isotropic were analysed by Hardung (1964 )
and Bergel and Schultz (19714 ~used the results to show that there was good
agreement between the results of Bergel (1961) and Patel et al (1968).
3.2. Strain Energy Density or Stored Energy Function
The use of the incremental stresses and strains follows the Cauchy
approach to elasticity in which the stresses are expressed directly as a
function of the strains. An alternative approach, due to Green, is to
assume that there exists a function, the strain energy density, W, which
may be expressed as a function of the strains. The stresses are the
derivatives of the function with respect to the strains (Green and
Adkins, 196o 42 , Green and Zena, 1954 43 ).
The strain energy function may be expressed as a function of a
suitable strain measure E .. or the principal extension ratio A.

1J 1
l
w
(11)
w
For isotropic materials the strain dependence may be replaced by a
dependence on the principal strain invariants
(12)
where
I1 Ai + A; + A;
(AlA2)2 + (A2A3)2 + (AJA1)2 (13)
The principal stress-strain relations are obtained from
a.
aH
~ aAi
(14)
aw ail aw ai2 +3\~- ai3
+-
ail oA.
~
ar2 oA.~ ai3 oA.~
Stress-strain relations for specific deformations may be obtained
by manipulation of equations 14.
The third invariant r 3 is the relative change in volume. For
incompressible materials the term

aw
aT
is undefined. As a result the
3
stresses are indeterminant, and can be obtained only with respect to an
arbitrary hydrostatic pressure.
The application of this version of the theory to the arterial wall
is discussed by Tickner and Sacks (1964), but as we have seen the vessel
wall is not isotropic. More realistic models of wall anisotropy produce
considerably more complexity.
Vaishnav et al (1972 44 ) investigated the strain energy density of
the thoracic aorta of dogs on the assumption, used in the incremental
study (Patel et al 1969, See Section 3.1) that it was incompressible and
orthotropic. For such a material the strain energy density may be
expressed as (Green and Adkin, 1960, p.28)

2 2 2 (15)
W = W(e:ll' e:22' e:33' e:12' e:23' e:31)
156 J.C. Barbenel
Vaishnav et al used the Green strain components which, in the
notation used for the incremental study, are
2 = ~ (>, 2
e = ! (A. e
£ £ - 1); £ - 1) (16)
r z z
It was assumed that W could be approximated by a polynomial of the
form
The special case of the simultaneous inflation and longitudinal
extension was considered. Under such tests the principal strains are
very much smaller than the shear strains and Ere' e: 6z and e:zr were put
equal to zero. The incompressibility condition meant that the three
principal strains were interrelated and the polynomial series in
equation 11 was approximated by finite series in e: 6 and e:z. Three
series of second, third and fourth degree were proposed. These are
respectively
2 2
w Al£6 + Ble:6ez + Cle:z (18)
2 2
w A2£8 + B2e:eez + C2£z
(19)
3 2 2 3
+ D2£6 + E2e:6ez + F2e:6ez + G2e:z 1
and
2 2 3
w A3£6 + B3£6£z + C3£z + D3£6
2 2+G 3 4
+ E3e:6ez + F3£6£z 3£z + H3e:e (20)
3 2 2 3 4
+ 1 3e:6e:z + J3e:6e:z + k3e:ee:z + L3e:z
Experiments were made in vitro on the thoracic aorta of 13 dogs.

The vessel was precycled and elongated by a weight while the internal
pressure was incrementally increased. One minute after each pressure
increment the dimensions were recorded. The experimental data was used
to calculate the principal stresses and strains and the constants in
equations (18), (19) and (20) evaluated.
The relationship between the three constants in equation (18)
suggested that the vessel was stiffer in the circumferential direction
than longitudinally.
The incremental moduli may be predicted from the stored energy
function and these predictions were compared with experimental
measurements. The three and twelve constant theories &ave discrepancies
of up to 30% but the seven constant theory produced acceptable results.
The authors concluded that the seven constant theory was the most useful.
45 ) proposed an alternative exponential function
Fung et al (1979
(21)
which may be rewritten as
W= ~ exp [al (e:~8 - e::~) + a2 (e:~z - e::~) + 2a4 (e:8ae:zz - e::8S:z)] (22)
The strains e: * * are the result of arbitrary applied stresses

88 and e:zz
The function was evaluated experimentally on arteries of rabbits.
The vessels were subjected to longitudinal extension to approximately
in vivo length and then inflated, or stretched while uninflated. The
values of a * * were chosen to fall within the physiological range

88 and ozz
of function of the vessel.
158 J.C. Barbenel
The overall fit between the stress-strain relations predicted from
the exponential strain energy function and the experimental data was
reasonably good. The fit obtained from the seven constant polynomial
form discussed above was marginally better.
The major difference between the two forms of the stored energy
function appeared when the constants were evaluated at different arterial
sites. The polynomial constants showed wide variations and changes of
sign, but the exponential constants varied less erratically.
The stored energy functions evaluated by Vaishnav et al (1972) and
Fung et al (1979) are reduced forms of general three dimensional orthotropic
functions, which are appropriate to the physiological type of loading used
in the experimental test programmes. The loading is essentially two
dimensional as the longitudinal and circumferential stresses, a 2 and a 0 ,
are very much greater than the radial stress a . The principal stresses,
r
and strains, are also in the direction of the axes of elastic symmetry of
the vessel wall. The elimination of the radial strain, E , from the
r
three dimensional functions depend on the assumption that the vessel walls
are incompressible. Th€ walls of the large arteries are thin compared to
their radius, and it is possible (Green and Adkins, 1960) to develop two
dimensional stored energy functions for the two dimensional loading which
occurs during function. An example of the application of such a two
dimensional stored energy function, which does not depend on the
assumption of incompressibility of the arterial wall can be found in

46
Rachev (1980 ).
Time Dependence 159
4. TIME DEPENDENCE
The non-linear time dependence of the arterial wall was noted by
Roy (188o 47 ) who compared the behaviour with that found in rubbers.
There are numerous other reports of various aspects of time dependence
reported in the literature, and empirical descriptions of such variables
as the time course of stress relaxation.
There are two views of the non-linearity of the tissue. The first
assumes that the non-linearity of the load-extension response and time
dependence are separable leading to a single integral representation
(Fung, 1972). The alternative assumption that this simplification is
unjustified leads to the use of formulations for non-linear viscoelasticity
of the Green-Rivlin type. The former assumption will be discussed in this
section and multiple integral representations in Section 5.
4.1. Single Integral Representations
The stress cr(t) at time t after the application of a step extension
A(o) to a viscoelastic solid is given by
cr(t) = A(o) K(A,t) (23)
where K(A,t) is the relaxation function.
If the relaxation function is separable in extension and time this
reduces to
K(A,t) = G(t) K(e)(A) (24)
The nomenclature follows Fung (1972) and G(t) is a reduced
relaxation modulus and K(e)(A) an elastic response which is a function

160 J.C. Barbenel
only of the extension ratio A and is called the elastic response. If it
is assumed that a suitable modified superposition principle applies then
the stress produced by a continuous stretch history is analogous to the

48
Linear Boltzman's superposition integral (see FlUgge, 1975 ) and
a(t) roo G(t-T)

(25)
or a(t) = J:oo G(t-T) T(e)(T) dT
Equation 25 may be inverted to obtain the creep function relating
the time dependent extension and the applied step stress. In addition
the use of equation 25 leads to the prediction of the response to the
application of a sinusoidally varying deformation.
The elastic response T(e)(A) can be identified with the
characterisations of the overall load-extension response discussed in

Section 3.
4.2 Descriptions of Time Dependence
The time dependence of the tissues have been described in terms of
both discrete time constraints and continuous relaxation spectra.
Disarete time aonstraints lead to the characterisation of the stress
relaxation behaviour in terms of the sums of time varying exponentials.
That is
]
G(t)
(26)
or G(t) E C. exp(-t/T.)
i ~ ~
Time Dependence 161
.Fung (1972) used a normalised relaxation modulus and equation 5
becomes
G(t) ~ C. exp(-t/T.)/~ C. (27)

i ~ ~ i ~
There are three major drawbacks to using such a discrete time
constant characterisation.
It is usually impossible to obtain unique values of the parameters,
and the same data can give rise to quite different exponential terms if
the method of analysis is changed. Lanczos (1956 49 ) discussed this
point at some length, with examples, and concluded that the difficulty
lay in the extreme sensitivity of the parameters to small changes of the
data. He thought that the use of least squares or other statistics
would not provide a remedy.
The second drawback is the tendency for those deriving the time
constraints to return to the spring and dashpot mechanical analogues
which were used as conceptual aids by Kelvin and Maxwell and to identify
these imaginary mechanical analogues with real structures in the tissue
(e.g. Wiedehielm, 1965).
The greatest disadvantage is that models incorporating discrete
time constants generally predict a marked frequency dependence of the
dynamic viscoelastic parameters. This can be demonstrated by the
behaviour of the simplest model showing the essential features of the
arterial wall. The tissue shows an instantaneous elastic response and
an equilibrium stress after the application of a step deformation, and
the simplest model showing these features is the three parameter (or
162 J.C. Barbenel
standard) viscoelastic solid. The relaxation modulus of such a material
is given by
G(t) = G1 + Gell - exp(-t/T) I (28)
where Gi and Ge are the instantaneous and equilibrium moduli. Using the
general relations between the viscoelastic parameters it is possible to
use the relationship in equation 28 to predict the storage and loss
moduli G' (w) and G" (w) which the material will display in response to a
sinusoidal deformation. These are given by:
2 2
W T
G' (w) G + (G 1 - G )
e e 1 + w2 T2
G"(w) = (G - G )wT/1 +
I e
W
2 2
T
} (29)
Both parameters are frequency dependent, with the phase angle
showing a peak value at an angular frequency w equal to 1/T. This
behaviour is at variance with the frequency insensitivity of the phase
angle which has been shown experimentally (Section 2).
Continuous relaxation spectra replace the discrete time constants
with a continuous function. For such a system the stress relaxation
modulus G(t) in equation 26 becomes
G(t) = G + Joo A(T) exp(-t/T) dT (30)

e 0
where A(T) is the continuous relaxation spectrum. Alternatively the
logarithmic spectrum H(T) may be used (Ferry 197o 50 )
G(t) G
e
+ J+oo H(T) exp(-t/T)d2nT (31)
-oo
Time Dependence 163
The value of H(T), as a function of time, may be obtained from the
slope of a plot of the stress relaxation modulus against the logarithm of
elapsed time. To a first approximation this is (Alfrey, 1948 51 )
H(T) = - 2.303
1 IdG(t)
d~nt
I (32)
t=T
In terms of stress o(t) and initial extension A(o) equation 12 is
1 ldo(t)
H(T) = - 2.303 A(o) d~nt
I (33)
t=T
The relaxation in arteries is such that the stress falls linearly
with the log of time for several decades. This means both that the
approximation implicit in equations 31 and 32 ~s a good one, and that
the logarithmic spectrum is particularly simple, consisting of a spectrum
of constant value over the time region for which the relaxation is linear
and zero outside this. This spectrum is usually known as a box spectrum,
and its properties have been investigated both for polymers and for
52
tissues (Fung, 1972; Barbenel et al, 1973 ).
The spectrum can be characterised as having the value
l
H(T) = H(o)
(34)
= 0
Substituting equation 14 into equation 11 yields
T2
G(t) = Ge + H(o) J exp(-t/T) d~nT (35)
Tl
which can be integrated to yield (Tobolsky, 1960 53 )
G(t) (36)
164 J.C. Barbenel
where E. is the exponential integral.

l.
The series expansion for E. is

l.
2
X X
Ei (x) = Y + R.n x + (1.1! - 2 • 2! + ... ) (37)
where y is Euler's constant.
Thus for t < T1 << T2
(38)
for T2 < t
G(t) = G (39)
e
and in the middle range where Tl < t < T2
G(t) = Ge - H(o) (R.n l/T 2 - R.n t) (40)
which reproduces the experimental stress relaxation behaviour.
The upper limit T2 of the spectrum can be obtained from the time,
t 2 , at which the extrapolated linear central por·tion of the relaxation
curve is equal to the equilibrium modulus,
T2 = t2 exp y
(41)
1.781 t2
and similarly
T1 =1.781 tl (42)
These relationships can be used to obtain the box parameters fro.m
the ·experimental stress relaxation curve and to predict the dynamic
storage and loss moduli from the relationships (Olofsson, 1974 54 )

Time Dependence 165
G' (w) G + J+oo IH(o)w2T2 I 2 21dJI.n T

e -oo (l+w T )
(43)
l+w T2
2 2
G +' H(o) Jl.n
e 2 2 2
l+w Tl
and
G"(w) = J+oo IH(o)T I 2" 2 I dJI.n T

-oo (l+w T )
(44)
-1 -1
= H(o) (tan wT 2 - tan wT 1 )
In the time region of the non zero portion of the spectrum the phase
angle, o, is given by
1
i
tan -1,1T - G +
eiH(o),
(45)
-1 1T H(o)
tan
2G
e
The general form of the dynamic response predicted by the box
spectrum is similar to that found experimentally, and this led to the
suggestion of its use for describing time dependence in soft connective
tissues.
Tanaka and Fung (1974) showed that it is applicable to the aorta
and that the values of T2 varied from 94 s to 2480 s, with a major site
dependence. The values of Tl were all less than 1 s. These values also
imply that the load-extension response would be independent of strain

-3 -1
rate in the range 10 - 1 s
The single integral relationship between stress, and strain and
time produced by both the discrete time constant and continuous spectrum
model will predict a progressive alteration in response on repetitive
straining. No detailed analysis of the magnitude of this predicted

166 J.C. Barbenel
preconditioning has been made, but the nature of the experimental stress
relaxation, which has a prolonged course makes it probable that the
predicted preconditioning will be small, unlike the major changes found
experimentally.
In order to incorporate either model of time dependence into the
single integral representation of equation 25, it is essential to
confirm that the viscoelastic parameters are independent of the strain
magnitude. It is not clear that a critical test of this very important
condition has been made for artery, although it appears not to be
satisfied for in vivo skin.
These two unresolved features of the single integral representation
has led to the investigation of multiple integral forms.
4.3 Hultiple Integral Representations
Green and Rivlin (1957 55 , 196056 ) proposed a multiple integral
representation for the response of a general non-linear viscoelastic
material. For one dimensional behaviour this reduces to (Lockett, 1972 57 )
cr(t) J:.c 1(t-T 1 H<T 1 )dT 1 + Jl: c2 (t-Tl't-T 2 )i(T 1 );(T 2 )dT 1 dT 2 }
t (46)
+ JJJ G3 (t-T 1 ,t-T 2 ,t-T 3 )E(T 1 )€(T 2 )E(T 3 )dTl dT 2 dT 3+ •••
-co
where G1 , G2 and G3 ••• are first, second, third ... order kernels; it
can be assumed without any loss of generality that the kernels are
symmetric with respect to their arguments. The term E is a suitable
strain measure.
Time Dependence 167
To be usefuL the series must be finite and is usually terminated
after the third term.
The response of such a material can best be analysed by utilising
the step function ~t, which has the properties
Mt) 1 t > 0
Mt) 0 t < 0
The application of a step strain £(0)~t produces a stress cr(t) given
by:
(47)
It is apparent that the material is non-linear.
The response to a single one step test is given in the above
equation and it is clear that the response to a single step test is not
sufficient to determine the material constraints. The test may be
repeated with two additional strain values £(l)~t and £(2)~t and from
the three test results it is possible to calculate the values of
In order to obtain G2 and G3 when their arguments are unequal, it is
neces,sary to perform multiple step tests.
The second order kernel G2 can be evaluated using two step tests
£(t) = £(3)~t + £(4)~(t-t 1 ). The response is:

2 3
cr(t) £(3)G 1 (t) + £ (3)G 2 (t,t) + £ (3)G 3 (t,t,t)
2
+ £(4)G 1 (t-t 1 ) + £ (4)G 2 (t-t 1 , t-t 1 )
3
+ 2£(3)£(4)G 2 (t, t-t 1 )+ £ (4)G 3 (t-t 1 , t-t 1 , t-t 1 ) (48)
168 J.C. Barbenel
The first six terms are known from the single step tests and can be
evaluated. In order to evaluate the last three terms it is necessary to
carry out two additional two step tests in which t 1 is kept constant.
This allows the evaluation of values of three kernels
Thus G2 has been determined for unequal arguments, but only for one
value of the difference in argument. The number of values of the
kernel G2 (t, t-k) required to determine the complete function between
k=O and the value of k at which an equilibrium value occurs depends on
the form of G2 (t, t-k). This can only be determined by experiment, and
Lockett has suggested that a minimum of ten values of k is required.
As a result of the experiments to determine G2 (t, t-k) the values
of third order kernel are known when two of the arguments are equal. In
order to determine the value when all three are unequal, G3 (t, t-k, t-~)
it is necessary to carry out a single three step test, Once again it is
necessary to carry out multiple three step tests in order to cover a
range of values for k and ~. Lockett (1972) has calculated the number
of tests required to evaluate all three constants at n discrete values
of k and 2 . For n = 10, which Lockett suggests is a reasonable number,
the test programme requires 78 tests, and ideally this should be
increased to obtain replicate results.
The one dimensional relationship ~n equation 46 can be generalised
to produce a three dimensional theory for anisotropic materials. Cheung
and Hsaio (1972 58 ) presented a theoretical and experimental analysis
based on the assumption that the artery wall was transversely isotropic,
Time Dependence 169
for which there would appear to be little justification. A similar

59
study was made by Young et al (1977 ) who assumed the wall to be
incompressible and to be orthotropic.
The number of kernel functions depends on the order of the integral
equation. Young et al investigated a second order theory, and showed
for this case orthotropy required the evaluation of ten kernel functions.
By dropping second order kernel functions, a simpler first order equation
containing only four kernels was obtained. In the latter case it
appears the non-linearity of the response is determined by the strain
measure chosen. The lack of second order terms, and therefore the
presence of single integrals only, limits the nature of preconditioning
the theory can describe.
Tests were made on excised segments of the descending aorta of dogs,
·.and before testing the vessel was returned to its in vivo length and
repeatedly pressurised to a pressure of 19 kPa (200 em H20), which
suggests that the vessel was, in fact, preconditioned before testing
started. To evaluate the kernel functions data was obtained on stress
relaxation at 38 states of strain, which were obtained by longitudinal
extension and inflation of the vessel segment. Relaxation was measured
for lOOs. The results suggest that each step was considered in
isolation with results being presented in terms of a single time
argument.
The experimental data from which the parameters were obtained were,
not unsurprisingly, well described by both theories. The critical test
of their applicability is the prediction and experimental verification

170 J.C. Barbenel
of the response to a different state of strain.
4.4 Time Dependence in Biological Tissues
The time dependence of biological tissues has been described in
terms of classical viscoelastic theory. The appearance of terms of the
nature of ~(t-T) leads to a hereditary behaviour, such that the
influence of events occurring a long time before the current time at
which measurements are made will gradually diminish. It appears from
the response of preconditioned tissues that the application of a large
strain to the specimen may influence the current state of stress as much
as the current strain. It is not clear that such behaviour can be
contained within classical viscoelasticity. An alternative approach was
used by Chu and Blatz (1972 60 ), who postulated a cumulative microdamage
process to describe the mechanical behaviour of mesentery. The
irreversibility of the microdamage limits the applicability of the theory,
but investigations of alternatives to viscoelastic time dependence may
suggest how the effects of preconditioning may be incorporated into a
general theory of time dependence for tissues.

Fleferences 171
5. REFERENCES
1. McDonald, D.A., Blood flow in arteries, Edward Arnold Ltd.,
London, 1974.
2. Caro, C.G., Pedley, T.J., Schroter, R.C. and Seed, W.A., The
Mechanics of the circulation. Oxford University Press,
Oxford, 1978.
3. lberall, A.S., Anatpmy and steady flow characteristics of the
arterial system, Math. Bioscience l• 375-395, 1967.
4.. d'Arcy, W. Thompson, On growth and form, MacMillan, London,
pp. 948-957, 1945.
5. Bloom, W. and Fawcett, D.W., A textbook of histology. W.B. Saunders,
Philadelphia, 1975.
6. Benninghoff, A., Blutgafasse and Herz in 'Handbuch der
Microkopischen Anatomie', Springer Verlag, Berlin, Vol. Vl/1,
p.l-225, 1930.
7. Wolinsky, H. and Glagov, S., A lamellar unit of aortic medial
structure and function in mammals. Ciraulation Res. 20,
99-111, 1967.
8. Burton, A.C., Physical principles of circulatory phenomena.
Handbook of physiology, Amer. Physiol. Soc., Washington,
Vol 1/2, pp. 85-106, 1962.
9. Folkow, B. and Neil, E., Circulation, Oxford University Press,
London, 1971.
172 J.C. Barbenel
10. Learoyd, B.M. and Taylor, M.G., Alteration with age in the
viscoelastic properties of human arterial wall. Ci~culation Res.

18, 278-291, 1966.
11. Fung, Y.C., Biomechanics, Mechanical properties of living tissues,
Springer Verlag, New York, 1981.
12. Tanaka, T.T. and Fung, Y.C., Elastic and inelastic properties of
the canine aorta and their variation along the aortic tree.
J. Biomech. l• 357-370, 1974.
13. Collins, R. and Hu, W.C., Dynamic deformation experiments on aortic
tissue, J. Biomech. -~, 333-337, 1972.
14. Wiederhielm, C.A., Distensibility characteristics of small blood
vessels. Fed. Proc. 24, 1075-1084, 1965.
15. Patel, D.J. and Fry, D.L., The elastic symmetry of arterial
segments in dogs. Ci~c. Res. 24, 1-8, 1969.

16. Patel, D.J. and Vaishnav, R.N., Basic hemodynamics and its role in
disease processes, University Park Press, Baltimore, 1980.
17. Lawton, R.W., The thermoelastic behaviour of isolated aortic
strips of the dog. Ci~c. Res. 3 403-408, 1954.

18. Carew, T.E., Vaishnav, R.N. and Patel, D.J., Compressibility of the
arterial wall. CiPaulation Res. 22, 61-68, 1968.
19. Bergel, D.H., The properties of blood vessels -in 'Biomechanics-
it's foundations and objectives' ed. Fung, Y.C., Persone N.
and Anliker M. Prentice-Hall, New Jersey, pp. 105-139, 1972.

References 173
20. Veronda, D.R. and Westmann, R.A., Mechanical characterisation of
skin - finite deformations. J. Biomech. l• 111-124, 1970.
21. Stark, H.L. and Al-Haboubi, A., The relationship of width, thickness,
volume and load to extension for human skin, In Vitro~
Engineering in medicine ~. 179-183, 1980.
22. Tickner, E.G. and Sacks, A.H., A theory for the static elastic
behaviour of blood vessels. Biorheotogy ~. 151-168, 1967.
23. Wolinsky, H. ahd Glagov, S., Structural basis for the static
mechanical properties of the aortic media. Circutation Res. ~.
400-413, 1964.
24. Carton, R.W., Dainauskas, J. and Clark, J.W., Elastic properties
of single elastic fibres. J. Appl. Physiol. ll• 547-551, 1962.
25. Jenkins, R.B. and Little, R.W., A constitutive equation for
parallel fibred elastic tissue. J. Biomech. l• 397-402, 1974.
26. Gosline, J.N., The physical properties of elastic tissue in
'International review of connective tissue research'. ed.
Hall D.A. and Jackson, D.S. Academic Press, London Vol. 7.
pp. 211-250, 1976.
27. Haut, R.C. and Little, R.W., A constitutive equation for collagen
fibres. J. Biomech. ~. 289-298, 1972.
28. Dobrin, P.B. and Rovick, A.A., Influence of vascular smooth muscle
on contractive mechanics and elasticity of arteries. Amer. J.
Physiol. 217, 1644-1651, 1969.

174 J.C. Barbenel
29. Harkness, M.L.R., Harkness, R.D. and MacDonald, D.A., The collagen
and elastin content of the arterial wall in dogs. Proc. Roy.
Soo. B. 146, 541-551, 1957.

30. McCrum, N.G. and Darrington, K.L., The bulk modulus of solvated
elastin. J. Mat. Sci. ll• 1367-1368, 1976.
31. Roach, M.R. and Burton, A.C., The reason for the shape of the
distensibility curves of arteries, Can. J. Biochem. PhsioZ. 35,
681-690, 1957.
32. Hoffman, A.S., Grande, L.A., Gibson, P., Park, J.B., Daly, C.H.,
Borstein, P. and Ross, R., Preliminary studies on mechanochemical
-structure relationships in connective tissues using enzymolysis
techniques, in "Perspectives in biomedical engineering", ed.
Kenedi, R.M., MacMillan, London, pp.l73-176, 1973.
33. Patel, D.J., Janicki, J.S. and Carew, T.E., Static anisotropic
elastic properties of the aorta in living dogs. Circulation Res.
25, 765-779, 1969.
34. Fung, Y.C., Stress-Strain history relations of soft tissue in
simple elongation in 'Biomechanics - its foundations and
objectives' ed. Fung, Y.C. Perrone, N. and Anliker, M.
Prentice-Hall, New Jersey, pp. 181-208, 1972.
35. Vito, R., A note on arterial elasticity. J. Biomeche. l• 3-12, 1973.
36. Biot, M.A., Mechanics of incremental deformation. John Wiley,
New York, 1965.

References 175
37. Bergel, D.H., The static elastic properties of the arterial wall.
J. Physio"l. 156, 458-469, 1961.
38. Patel, D.J. and Fry, D.L., Longitudinal tethering of arteries in
dogs. Circ. Res. ~. 1011-1021, 1966.
39. Fung, Y.C., Rheology of blood vessels in 'microcirculation'
ed. Kaley, G. and Altura, B.M. University Park Press,
Baltimore, Vol. 1 pp. 299-324, 1977.
40. Hardung, V., Die bedeutung der anisotropie and inhomogenitat bei
der bestummung der elastizitat der blutgefasse II.
Angiotogica l• 185-196, 1964.
41. Bergel, D.H. and Schultz, D.L., Arterial elasticity and fluid
dynamics in 'Progress in biophysics and molecular biology'.
ed. Butler, J.A.V. and Noble, D. Pergamon Press, Oxford~.
3-36, 1971.
42. Green, A.E. and Adkins, J.E., Large elastic deformations and
non-linear continuum mechanics, Clarendon Press, Oxford, 1960.
43. Green, A.E. and Zerna, W., Theoretical Elasticity. Clarendon Press,
Oxford, 1954.
44. Vaishnav, R.N., Young, J.T., Janicki, J.S. and Patel, D.J.,
Non linear anisotropic elastic properties of the canine aorta.
Biophys. J. ~. 1008-1027, 1972.
45. Fung, Y.C., Fronek, K and Patitucci, P., Pseudo elasticity of
arteries and the choice of its mathematical expression.
Am. J. Physiot. 237, H620-H631, 1979.

176 ].C. Barbenel
46. Rachev, A.I., Effects of transmural pressure and muscular activity
on pulse waves in arteries. ASME Journa~ of Biomechanica~
Engineering 102, 119-123, 1980.

47. Roy, C.S., The elastic properties of the arterial wall, J. Physio~.
l· 125-159, 1880.
48. FlUgge, W., Viscoelasticity. 2nd Ed. Springer-Verlag, Berlin, 1975.
49. Lanczos, C., Applied analysis. Prentice-Hall, New Jersey, 1956.
50. Ferry, J.D., Viscoelastic properties of polymers. J~hn Wiley,
New York, 1970.
51. Alfrey, T., Mechanical behaviour of high polymers. Interscience,
New York, 1948.
52. Barbenel, J.C., Evans, J.H. and Finlay, J.B., Stress-strain-time
relations for soft connective tissues. 'Perspectives in
biomedical engineering'. ed. Kenedi, R.M., Macmillan, London,

pp. 165-172, 1973.
53. Tobolsky, A.V., Properties and structure of polymers, John Wiley,
New York, 1960.
54. Olofsson, B., Comparison of stress-activated models and linear
spectral models for visco-elasticity. Rheol. Acta. !l• 78-85,
1974.
55. Green, A.E. and Rivlin, R.S., The mechanics of non-linear materials
with memory. Arch. Rat. Mech. Ana~ • ..!_, 1-21, 1957.
56. Green, A.E. and Rivlin, R.S., The mechanics of non-linear materials
with memory. Arch. Rat. Mech. Ana~. ~. 387-404, 1960.

References 177
57. Locket~, F.J., Non-linear viscoelastic solids. Academic Press,
London, 1972.
58. Cheung, J.B. and Hsiao, C.C., Non-linear anisotropic viscoelastic
stresses in blood vessels. J, Biomeahias. i• 607-619, 1972.
59. Young, J.T., Vaishnav, R.N. and Patel D.J., Non-linear anisotropic
viscoelastic properties of canine arterial segments.
J. Biomeahs. 10, 549-559, 1977.
60. Chu, B.M. and Blatz, P.J., Cumulative microdamage models to
describe the hysteresis of living tissue. Ann. Biomed. Engng. l•

204-211, 1972.
CHAPTER I I I
DYNAMICS OF FLUID FILLED TUBES
J.B. Haddow
Department of Mechanical Engineering
T.B. Moodie, R.J. Tait
Department of Mathematics
University of Alberta
Edmonton, Alberta, Canada
1. INTRODUCTION
1 .1 Introduction
Models for describing pressure pulse propagation in arteries can,
with few exceptions, be separated into two categories depending upon
the emphasis placed on the fluid mechanics or the solid mechanics.
Models in the first category employ ideas first put forward by Euler
(1844) who proposed combining the equation of motion for an inviscid
fluid together with the continuity equation and a third equation relat-
ing the pressure in the tube to its cross-sectional area. Euler's
equations were first solved in the context of blood flow by Lambert

(1958) who used a pressure-area relation based on experimental data
rather than the one proposed by Euler. These ideas have been con-
siderably refined and developed to employ general pressure-area relations,
180 J.B. Haddow
include outflow through porous walls (Rudinger, 1966), and study the
development of shocks (Rudinger, 1970), (Teipel, 1973), (Forbes, 1979).
The second category, and the one we are concerned with is comprised
of those models employing a version of the shell equations for the tube
wall together with equations of motion for the fluid. An excellent re-
view of these models up to 1966 is included in the survey article by
Skalak (1966). Significant contributions in this category have been
made by Lamb (1898), Witzig (1914), Klip (1962), Rubinow and Keller (1968),
(1971 ), (1978), and many others. All of the models in this category are
analyzed in essentially the same way although the amount of detailed in-
formation obtained varies greatly from author to author. Solutions are
assumed in the form of travelling periodic waves and frequency equations
are obtained and plotted together with plots for the various mode shapes.
In no instance are either initial or boundary value problems solved for
models in this category.
It is our aim to outline a method of solution for an initial value
problem for fluid-filled tube models in this second category. We do not
make the claim that our results in their present form are directly ap-
plicable to the propagation of pressure pulses in arteries but rather
that these methods do indicate a viable means of analyzing the transient
response of various models in this category. The information so ob-
tained can then be compared with experimental results to test the
ability of such models to explain observed phenomena.
The first model considered is a straight, uniform, thin walled,
tethered, cylindrical, elastic tube filled with an incompressible in~
Introduction 181
viscid fluid. Viscosity terms can be 1ncluded in the analysis with

additional complication, however, preliminary work has shown that for
small disturbances superimposed on the flow in elastomer tubes the damp-
ing due to viscous effects in the tube wall is large compared with
that due to fluid viscosity when parameters appropriate for biological
applications are considered. We first present the governing equations
of the problem taking account of shear deformation of the tube wall and
discuss the validity of neglecting the shear deformation. Later we con-
sider the governing equations with shear neglected and extend the analysis
to consider viscoelastic tubes.
Although the theory is given for a linearly elastic or visco-
elastic tube, with small modification it can be adapted to consider small
perturbations of a tube subjected to finite deformation.
Results are presented for a tube with wall thickness to mean radius
ratio of 0.05 although the theory is valid for ratios up to approximately
0.15. As this ratio is increased the effect of shear deformation of the
tube wall becomes more important.
A slightly different approach to the problems considered here has
been presented by Tait, Moodie and Haddow (1981).
1.2 Formulation of Problem

A uniform thin-walled tube with its axis horizontal and in the
x-direction is considered. The radius, density, and wall thickness of
the tube are R, y, and h, respectively, and p is the density of the
fluid. The radius R is taken to be the inside radius of the tube
with only a small error introduced by using R in the shell equations in
182 J.B. Haddow
place of the mean radius R + h/2. The distinction between the mean
radius and inside radius of the tube could be incorporated in the
analysis with some extra complication but the effect on the results was
found to be negligible for h/R < 0.1.
It is assumed that an axially symmetric perturbation of the system
results in axial and radial perturbation displacements ux(r,x,t) and
u (r,x,t), respectively, of the fluid. Corresponding velocity perturba-
r
tions are aux/at and aur/at. It is further assumed that the pressure
on the outside of the tube is uniform and that the pressure perturba-
tion is small enough that a linear theory is valid. Since a tethered
tube is considered, there is no axial displacement of the tube wall.
It is well known that blood vessels in situ operate under considerable
axial constraint. We include shear deformation in the analysis and in-
corporate it along with rotatory inertia into a simple shell theory for
the fluid-filled tube.
In order to illustrate the method of solution of initial value
problems we will consider an unbounded tube to be perturbed by an
initial distribution of external pressure pe(x) which is suddenly re-
moved at time t = 0. For definiteness we consider the external loading
Pe = qo(x) [1 - H(t)] (1.2.1)
where H(t) is the unit step function and o(x) is the Delta function.
This is a concentrated circumferential loading of intensity q per unit
length of circumference as shown in Fig. 1.2.1.
Introduction 183
Pe = q8(x)
!
0
f
Fig. 1 .2.1 Concentrated circumferential loading
of intensity q per unit length of circumference
1 .3 Constitutive Equations of Tube Material

The tube wall is assumed to be isotropic, incompressible and
linearly elastic or viscoelastic. Consequently, for the elastic ma-
terial Poisson's ratio v = 0.5 and the constitutive equation is
(1.3.1)
where ~and ~are the stress deviator and strain tensors, respectively,
and~ is the shear modulus. The assumption of incompressibility is
appropriate for tubes in biological systems.
The viscoelastic model considered is the so-called standard visco-
elastic material. A spring-damper analogue model for this material is
shown in Fig. 1.3.1. In Fig. 1.3.1 the stiffness ).1 is analogous to
the impact or unrelaxed modulus of rigidity, n represents the viscosity,
and 0 < m < 1. The relaxation time T is given by
184 J.B. Haddow
T - --,.-;;11;<--T
-~(1-m)'
(1.3.2)
and the relaxed modulus by m~. Since the material is assumed to be in-
compressible the constitutive equation is given in differential form by
(1.3.3)
where tis the time. Eq. (1.3.3) can be obtained from eq. (1.3.1) by
replacing ~ by the operator
~(1-
a
at+ m)/(T at+ 1)
a (1.3.4)
The integral form of the constitutive eq. (1.3.3) is

de(t')
~ = Jt G(t-t') -dt' dt' ,
-oo
where the relaxation modulus G is given by
G(t) = 2{m~ + (1-m)~ exp(-t/T)}
As m+l, eq. (1.3.3) approaches the constitutive equation (1.3.1) for a

Hookean solid and as m+O, that for a Maxwell fluid.
miL
(1- m)J.t
Fig. 1.3.1 Spring-damper analogue for
standard viscoelastic material
Governing Equations 185
2. GOVERNING EQUATIONS
2.1 Tube Equations
When shear deformation and rotatory inertia of the tube wall are
taken into account, the equations of motion of the tube wall for a
tethered elastic tube, as obtained from a simplified shell theory, are:
2 3 2
D cOP + JlQh(aw _ 1/J) _ :r.b_ .DI!_ = o, (2.1.1)
~ ax 12 at2
a,,, a2 a2
kw + ]lQh(~- ~) + yh w - Pw = 0 , (2.1.2)
oX ax ~
where y is tube wall density, w is the radial displacement perturbation

of the tube wall, ljJ is the part, due to bending, of the slope aw/ax of
the tube wall, Pw = pi - Pe• pi and Pe are the internal and external
pressures at the wall, respectively, Q is the shear deflection coeffi-
cient, and
Eh 3
D= 2
12(1-v)
where E is Young's modulus. Except for notation, eqs. (2.1 .1) and
(2.1 .2) are the same as those used by Forrestal and Herrmann (1965).
Since it is assumed that the tube wall is incompressible, Poisson's
ratio v = 0.5, and E = 3]l. The shear coefficient Q is taken as Q = 0.91
which is the value suggested by Herrmann and Mirsky (1956) for v = 0.5.
If shear deformation of the tube wall is neglected the single
equation of motion of the tube wall for an elastic tube is
186 J.B. Haddow
(2.1.3)
The last term on the right hand side of eq. (2.1 .3) represents the
rotatory inertia. Actually, the effect of rotatory inertia is less
than that of shear deformation, however, it can be incorporated without
involving an additional equation of motion.
When the tube wall is composed of incompressible standard material,
the shear modulus ]..1 in eqs. (2.1.1); (2.1.2) and (2.1.3) is replaced by
the differential operator (1 .3 .4).
2.2 Fluid Equations
The effect of fluid viscosity is neglected since it is assumed
that damping arises mainly from the viscoelastic properties of the
tube wall. Consequently the equations of motion of the fluid are the
linearized Euler equations,
(2.2.1)
(2.2.2)
where r is the radial coordinate, p is the fluid density, p = p(r,x,t)

is the pressure perturbation. The internal pressure at the wall sur-
face is given by pi(x,t) = p(R,x,t). In addition to eqs. (2.2.1) and
(2.2.2) the continuity equation
au
lr j_
ar
( ru )
r
+ ---!. = 0
ax ' (2.2.3)
is required. Eliminating p from the Euler equations (2.2.1) and (2.2.2)

gives
(2.2.4)
and then eliminating ux from eq. (2.2.4) and the continuity equation
(2.2.3) gives
(2.2.5)
Eliminating ur and ux from eqs. (2.2.1), (2.2.2) and (2.2.3) gives
(2 .2 .6)
2.3 Nondimensionalization Scheme

Before solving the governing equations it is convenient to in-
troduce the following nondimensional variables:
r = r/R ' X = x/R ' t = tc/R '
(2.3.1)
188 J.B. Haddow
where c is the classical Korteweg-Moens wave speed and is given by
Eh ) 1/2
c = [
2 R(l-})
and for v = 0.5, which is the case considered here
(2.3.2)
Often when dynamic viscoelastic problems are considered the nondimen-

sionalization of quantities which have the dimension of time is based
on the relaxation timeT, however, in this work the nondimensionalization
is based on the wave speed (2.3.2) and the radius R so that the non-
dimensional relaxation time is
T = TC/R (2.3.3)
In eq. (1 .2 .1) the nondimensional forms of q and the Delta function are
q= + ,
pc R
o(x) = o(x) R , (2.3.4)
respectively.
Two additional quantities arise later, w, the circular frequency,
and k, the wave number, and the nondimensional forms are
k = kR , w= wR/c
Henceforth, nondimensional quantities are used, but for convenience

the bars are omitted.

2.4 Nondimensional Form of Equations
The nondimensional forms of the elastic shell equations (2.1 .1)
and (2.1.1) are
(2.4.1.)
( 2 2 p
w + Q 1-v) (Et _ ~) + 11_ l.JL _ ...!i = 0 (2.4.2)
2 ax al 2p at2 2
and the nondimensional form of eq. (2.1.3) i·s
(2.4.3)
where f3 =a 2/12.
Nondimensional forms of the viscoelastic shell equations corres-
ponding to eqs. (2.4.1) and (2.4.2) are not given since shear deforma-
tion is neglected in the analysis of the viscoelastic tube. The non-
dimensional form of the viscoelastic tube equation, with shear neglected,
is
4 2 4
2{1_+ !!!_)(Sa w2 + w) =(,at+ l)(p - Y£4+ yaf3 a2w 2) (2.4.4)
at T at a T W p at.:; p ax at
for the incompressible standard material.

Nondimensional forms of the Euler equations are
(2.4.5)
190 J.B. Haddow
and the continuity equation is unchanged in form as in eq. (2.2.5).

2.5 Solution of Fluid Equations
A sol uti on, in the form of a propagating wave travelling in the x
direction, of
(2.5.1)
which is the nondimensional form of eq. (2.2.5), is sought. Consequently,
ur(r,x,t) = ~(r) ei(kx-wt) ,
where k is the nondimensional wave number, w is the nondimensional

circular frequency, and i =I- 1, is substituted in eq. (2.5.1) to ob-
ta in
(2.5.2)
Eq. (2.5.2) becomes
(2.5.3)
under the gauge transformation z = kr and the solution, which is regular

at r = 0, is
41(r) = GI 1 (kr), 0 ~r ~ 1,
where 11 is a modified Bessel function of the first kind and first

order and G is at most a function of k. It follows that
ur(r,x,t) = GI 1 (kr) e i(kx-wt) . (2.5.4)
More general solutions consist of terms of the form
00
ur(r,x,t) J G(k) r1 (kr) ei(kx-wt)dk, (2.5.5)
which are obtained by superposition of elementary solutions of the

type given by eq. (2.5.4).
Similarly it may be shown that general solutions of the nondimen-
sional form
(2.5.6)
of eq. (2.2.6) consist of terms of the form

00
p(r,x,t) = J H(k) I 0 (ir) ei(kx-wt) dk, (2.5.7)
where r 0 is a modified Bessel function of the first kind and order

zero. The functions G(k) and H(k) in eqs. (2.5.5) and (2.5.7) are
related through the second of eqs. (2.4.5) giving
iG(k) = kH(k) . (2.5.8)
Substituting r = 1 in eqs. (2.5.5) and (2.5.7) gives

192 J.B. Haddow
00
w = ur(l ,x,t) =I G(k) 11 (k) ei(kx~wt)dk (2.5.9)

~oo
00
Pi = p{l ,x,t) =I H(k) I 0(k) ei(kx-wt)dk . (2.5.10)
3. SOLUTION OF INITIAL VALUE PROBLEM

3.1 Dispersion relations for elastic tube
In order to obtain dispersion relations for the elastic tube when
shear is taken into account the so 1uti on for 1jl is taken to consist of
terms of the form,
00
1jl = J F(k) r1 (k) ei(kx-wt) dk. (3.1.1)

-oo
Substituting eqs. (3.1.1), (2.5.9) and (2.5.10) into eqs. (2.4.1) and
(2.4.2), and employing the relation (2.5.8) gives
i Q(1-v) kG+ [w 2 ~ - ~ Sk 2] F = 0 (3.1.2)
2 2
[1 +Q(l-v) k2 ~-~I] G+iQ(l-v) kF=O, (3.1.3)
~ - 2p 2k 2
where
In obtaining eqs. (3.1.2) and (3.1.3) it is assumed that the

external perturbation pressure Pe = 0 for t > 0.
The dispersion relations are obtained by equating the coefficient
Solution of Initial Value Problem 193
determinant of F(k) and G(k) in eqs. (3.1.2) and (3.1.3) to zero and
solving the resulting biquadratic equation in w, which is
w4[E;(~5a + i)]- w2[E;(k + ~ k3)

+ (Q(l-v) + Sk2)(kya + .!_)] + [Q(l-v) k
2 2p 2 2
+ Sk 3 + Q(1-v) Sk 5] = 0 . (3.1.4)
The roots of eq. (3.1.4), whicha--e real, are w = .!_ w1 (k), w = .!_ w2(k),
where w1 = (L-M) 112 , w
2 = (L+M) 112 and Land Mare easily obtained from
eq. (3.1.4). The contributions to the wave motion due to w1 (k) and
w2(k) are described as the first and second modes respectively. If the
shear deformation is neglected, $ = aw/ax, and there is only a single
mode of propagation, the single shell equation in nondimensional form
is given by eq. (2.4.3) and the corresponding dispersion relation is
w = W(k) = [ 2k(l+ k4 ) Jl/ 2 (3.1 .5)

1 + 2k{~
2p + sk~
A'realistic value of y/p for an elastomer tube filled with a

fluid 111hose density is close to that of water fs y/p = 1.1 and for
biolo.gical applications a realistic value of a is a= 0.1. The dis-
persion relations w
1 (k) and w
2(k) are shown graphically in Fig. 3.1 .1,
for a/p = 1.1, v = 0.5, a= 0.05 and a= 0.1, along with the dispersion
relation (3.1.5). It is evident that the second mode has a cut-off
frequency below which a wave does not propagate.
194 J.B. Haddow
a= 0.1
a= 0.05
Fig. 3.1.1 Dispersion relations
The phase.speeds for the first and second modes are given by
respectively, and the group speeds by

4 a= 0.05
10 20 30 40 50 60
k
Fig. 3.1.2 Phase and group speeds for a= 0.0.5
These are shown in Figs. 3.1.2 and 3.1.3 along with phase and group
speeds obtained from eq. (3.1.5). The results shown in Figs. (3.1.1),
(3.1.2) and (3.1.3) indicate that the dispersion relations and group
and phase speeds obtained by neglecting shear are good approximations
to those of the first mode for low wave numbers and, as expected, that
the lower the value of a = h/R, the greater the justification for
neglecting shear deformation of the tube wall. When an initial value
196 J.B. Haddow
problem is considered the quality of the approximation obtained by

neglecting shear depends on the initial conditions as well as on a.
a= 0.1
3 /c
/
/
/
/ c1
2
c1
10 20 30 40 50 60
k
Fig. 3.1.3 Phase and group speeds for a= 0.1
3.2 Effect of Shear Deformation on Solution

The ratios !F1/G1 I and !F 2/G 2 !, where the subscripts refer to the
first and second modes, are known as the mode shapes and are obtained
by substituting {W 1 (k)} 2 and {W 2(k)} 2, respectively for w2 in eqs.
( 3. 1. 2) or ( 3 .1. 3) , to give
F
___!!_=iS ,n=l,2 (3.2.1)
Gn n
where sn-
_
k
2
[ 2(J3k -Wn ~)
Q(l-v)
2
+1
J-1 (3 .2 .2)
In general a wave motion is composed of a superposition of the two modes

and this can be illustrated by considering an initial value problem
with the system quiescent at t = 0, so that the initial conditions are
of the form
w(x,O) = g(x) , ~(x,O) = f(x) , (3,2 .3)

. .
w(x,O) = 0 , lj!(x,O) = 0 . (3.2.4)
Because of initial conditions (3.2.4) the solution can- be expressed

in the form
J
00
2 i ( kx-W ( k) t ) i ( kx +W ( k) t
w= L G (k)[e n +e n ] dk , (3.2.5)
n=l -oo
n
2 Joo i(kx-W (k)t) i(kx+W (k)t)

L F (k)[e n +e n ]dk (3.2.6)
n=l -00
n
and it then follows from initial conditions (3.2.3) that

2 00 2 00
g(x) = 2 n~l J Gn(k) eikxdk , f(x) = 2 nil J Fn(k)eikxdk.

-00 00 ( 3 • 2 • 7)
The functions f(x) and g(x) can also be expressed in the form
00 00
g(x) = f G(k)eikxdk , f(x) = f F( k) ei kx dk , (3.2.8)

-00 -oo
00 00
g(x) e -kx dx , F(k) = 21IT f(x) e-kx dx . (3.2.9)

where G(k) = iiT f J
-oo
-oo
198 J.B. Haddow
Comparing eqs. (3.27) and (3.28) we obtain
(3.2.10)
(3.2.11)
Eqs. (3.2.9) are used to obtain G(k) and F(k) and eqs. (3.2.1), (3.2.10)
and (3.2.11) provide four equations which can readily be solved to
obtain the four unknowns Gn(k) and Fn(k), n = 1,2, which are in turn
substituted in eqs. (3.2.5) and (3.2.6) to obtain a formal solution
to the initial value problem. The integrals in eqs. (3.2.5) and
(3.2.6) can be evaluated numerically or otherwise.
In order to obtain the functions w(x,O) and ~(x,O) given by
initial conditions (3.2.3) the static form of eqs. (2.4.1) and (2.4.2)
with Pw = - Pe•
S 4+
2
dx
Q(~-v) (~~- ~) = 0, (3.2.12)
(3.2.13)
must be solved. In Eqs. (3.2.12) and (3.2.13), p is given by eq.

e
(1.2.1) with t < 0.
Applying standard Fourier transform techniques to equations
(3.2.12) and (3.2.13) gives
], (3.2.14)
(3.2.15)
where G(k) and F(k) are the functions defined by eq. (3.2.4) and
K = Q(l-v)/2. The Fourier transforms (3.2.14) and (3.2.15) can be in-
verted according to eqs. (3.2.7) to give
g(x) = - ~ s- 1/ 4 cosec28exp(-s 114 1xlcos8)[2cos28sin(e-s-l/ 4 1xlsin8)
+ sin(e+s- 114 1xlsin8)] (3.2.16)
f(x) = ~ s 112 cosec28exp{-S-l/ 4 1xlcose)sin(S-l/ 4 1xlsine)

(3.2.17)
where cos28 = s11 2;(2K). If shear deformation of the tube wall is
neglected the static displacement is obtained from the static equiva-
lent of eq. (2.4.3) again with pw =- qo{x), and is
w(x,O) = g{x) = 31 29 114 exp{-2 112 s- 114 1xl ){cos(2- 112s-l/ 4 1xl)
2 B
+ sin{2- 112 s- 114 1xi)J (3.2.18)
and the Fourier transform defined by the first of equations (3.2.9),

is denoted by "G(k) and is given by
G(k) = - _g_ ( 1 \ (3.2.19)

4n Sk 4 + 1 )
In Fig. 3.2.1 we show the initial displacements w(x,O) = g{x), for

a = 0.05 and a = 0.1, given by eq. (3.2.16) with q = 1. As before,
numerical results are presented for the parameters y/ p = 1 .1, v = 0. 5,
200 J.B. Haddow
Q = 0.91 and for q = 1.
0
0.75 1.00
-
0
X
-><
3:
-1
-~- · - Shear Neglected
-3
Fi g . 3 . 2 . 1 In i t i a 1 static d i s p1 a c erne n t due

to externa 1 pressure pe = qo ( x) with q = 1
Eqs. (3.2.1), (3.2.2), (3.2.10) and (3.2.ll) are used along with
(3.2.14) and (3.2.15) to obtain G1 (k) and G2 (k). The contributions
G1 (k) and G2 (k), of the first and second modes, respectively, to the
tube wall displacement w, for a given wave number k, are shown graphi-.
A
cally in Figs. 3.2.2 and 3.2.3 along with G(k)/2 given by equation
(3.2.19). If the effect of shear on the displacement w of the tube
wall is negligible, then w1 (k) "'W(k), G1 (k) » G2 (k), so that G1(k)"'

A
G(k)/2 "'G(k)/2 for the range of k outside of which G1 (k) and G2 (k)
are negligible. The results shown in the figures indicate that for
the initial value problem considered the effect of shear deformation
is negligible for a= 0.05. For a = 0.1 the effect of shear deforma-

0.08
a= 0.05
0.07
0.06
0.05
0.03
0.02
0.01
10 20 30 40 50 60
k
Fig. 3.2.2 Contributions G1 (k) and G2(k) to the
radial displacement w for a = 0.05 and q = 1
tion is more significant but still small enough that its neglect may
be justified. If a> 0.1 then shear deformation of the tube wall
should be included in the analysis. In the following sections it is
assumed that neglect of shear deformation is justified in the analysis.
3.3 Solution for Viscoelastic Tube
In this section the solution, for a viscoelastic tube composed of
the incompressible standard material, is obtained. The constitutive
equation is given by eq. (1.3.3) and the nondimensional governing
equation is eq. (2.4.4). Shear deformation is neglected and the elastic
202 J.B. Haddow
a= 0.1
0.01
0 o~--~10--~W._.__3~0~-4~0----S•O--~~
k
Fig. 3.2.3 Contributions G1 (k) and G2 (k) to the
radial displacement w for a = 0.1 and q = 1
solution is obtained as a 1imiting case as m + 1.

It is assumed that the circumferential line loading
Pe = qo(x) [1-H(t)] , (3.3.1)
is applied at timet=- oo so that at timet= 0, wand all the time

derivatives of ware zero. It follows then that the initial deflected
form is given by the solution of
-:-f
d4
dx
(x,O) + w(x,O) = - in o(x) . (3.3.2)
Eq. (3.3.2) is the same as that for an elastic tube with shear modulus
equal to the relaxed modulus of the viscoelastic tube. The deflected
form w(x,O) at t = 0, obtained from eq. (3.3.2) is given by
-1/2 -1/4 -1/2 -1/4 -1/2 -1/4

w(x,O)/w 0 = exp(-2 S jxj){cos(2 S lxl)+ sin(2 S jxj)}
(3.3.3)
where w0 = w(O,O).
The nondimensional intensity of 1 ine loading is given by
_
q - - 2
3;2sl/4 m w
0
Since the system is initially at rest eq. (3.3.2) and
. +
w(x,O ) = 0
provide two initial conditions. A third initial c~ndition is required,

but this is not obtained explicitly, but is obtained in the Fourier
transform domain.
We adopt the following form for the Fourier transform f(k) of a -
function f(x),
00
f(k) = J_ J f(x) e-ikx dx , (3.3.4)

2n
00
so that f(x) = J f(k) eikx dk . (3.3.5)
Eqs. (3.3.4) and (3.3.5) are consistent with eqs. (3.2.8) and (3.2.9).
Henceforth a superposed tilde denotes the Fourier transform as defined
by eq. (3.3.4).
204 J.B. Haddow
The initial value problem considered involves temporal attenua-

tion rather than spatial attentuation of the pe.rtarbation. Consequently
we express the radial deflection w(x,t) and pressure p(r,x,t) as Fourier
integrals
w(x,t) = Joo w(k,t) eikx dk , (3.3.6)
00
p(r,x,t) = J p(r,k,t) eikx dk (3.3.7)

-oo
The Fourier transform of eq. (2.5.6) is
2 -
~ + !_ ~ - k 2- = 0 (3.3.8~
-:----z r dr
dr P '
so that
-
p(r,k,t) = P0(k,t)I 0(kr) . (3.3.9)
It follows from the Fourier transform of the second of eqs. (2.4.5)

and eq. (3.3.9) that
(3.3.10)
and for r = 1 this gives
2-
aw= - 'P 0 ( k, t) kI 1 ( k) (3.3 .11)
at 2
It follows from eq. (3.3.9) that
(3.3.12)
and eliminating p0(k,t) from eqs. (3.3.11) and (3.3.12) gives
(3.3.13)
We multiply the Fourier transform of eq. (2.4.4) by ki 1 (k) and use

eq. ( 3 . 3 .1 3) to obtain
3~
[I 0 (k) + ypa (Sk 2+1) ki (k)] ~

1 at3
4at
2~
+ l_ [I 0 ( k) + Y9:. ( Sk2+1.) kI l ( k) ]
1 P
2 k 4 ~ p Clp
+ ~ I1 (k)(Sk+l)w=-ki 1 (k)[-f+ate], (3.3.14)
where ;e = rn [1- H(t)].

Then, by integrating this equation over - E ~ t ~ E and, letting
E + 0 , we obtain w(k,O+) = w(k,O-), ~(k,O+) = ~(k,O-) = 0 and
-2 312ms 114ki 1 (k)

w(k,O+)/w 0 = - - - - - - - - - - - (3.3.15)
n[I0(k) + y~k (l+Sk 2) I1 (k)]
For t > 0 we can write eq. (3.3.14) as
3~ 1 "2~ "~ ~
Cl w + - ~ + A ~ + !!!. Aw = 0 (3.3.16)
~ 1 at2 at 1
and the initial conditions are

206 J.B. Haddow
(3.3.17)
(3.3.18)
and eq. (3.3.15). In eq. (3.3.16),
Substituting w= Be~t. where B is a constant, in eq. (3.3.16) gives

the cubic equation
~3 + lT ¢2 + A¢ + ~ A = 0
T
(3.3.19)
When 0 < m < 1 it is easily shown that the real parts of the roots
will be negative for all real k as they must be from purely physical
considerations. The condition for one real root and two complex con-
jugate roots is
1 1
where a = 3 (3A - 2 ),
T
1 2
and b = 27T ( 27mA - 9A + "2)
T
It may then be shown that if~~ m < 1, eq. (3.3.19) has precisely
·~·--------------------------------------------------------
one real negative root and two complex conjugate roots for all real k
and' > 0 and we consider this case. The real root of (3.3.19) is then
b b2 31/21/3 b b2 a31/21/3 1
'h = [{~2+ (T+IT) } + {~2- (4+27) } ] - 3c ( 3 · 3 · 20 )
and the complex conjugate roots are
,/, = - !._ (!._ + ,/, ) + ir_ (3.3.21)

'~'2 2 T '~'1 2 '
(3.3.22)
where y is the positive square root of
Putting the roots in the form
cp 1 = - x1 ( k) , cp 2 , 3 = - x2 (k) .:_ iw(k) ,
x1 , x2 , and w are positive and functions of k which can be deduced

from eqs. (3.3.20), (3.3.21) and (3.3.22). The solution of eq.
(3.3.15) is then given by
Using the initial conditions (3.3.15), (3.3.17) and (3.3.18) it is a

routine calculation to obtain A1 , A2 and A3 for a given value of k.
The solution of the governing equation (2.4.4), with Pe given by
eq. (3.3.1 ), is then given formally by
208 J.B. Haddow
w(x,t)/w 0
-oo
It may be shown that A1 , A2 , x1, x2, ~nd ware even functions of k

while A3 is an odd function.
Evaluation of the Fourier integral (3.3 .24), that is inversion
of the Fourier transform (3.3.23) appears to be intractable analytically,
consequently results have been obtained numerically applying the Fast
Fourier Transform algorithm, as described by Brigham (1974).
Since the Fourier transform given in eq. (3.3.23) is an even
function of k the Fourier integral (3 .3 .24) can be rewritten as
00
-X t -X t
w(x,t)/w 0 2 J [A 1e 1 + e 2 (A 2coswt+A 3sinwt)] coskxdk.
0
This form is more convenient for numerical evaluation.

Results are shown graphically in terms of nondimensional quanti-
ties in Figs. 3.3.2 to 3.3.5, for y/p = 1.1, which is a realistic
value for an elastomer tube filled with biological fluid, and a = 0.05.
The initial displacement w(x,O)/w0 is shown in Fig. 3.3.1. In Figs.
3.3.2 and 3.3.3 w(x,t 1 )/w 0 is shown for t 1 = 0.5, 1.0, 2.0 and m = 0.5.
Two different relaxations times, T = 1.0 and T = 10 are considered. In

Fig. 3.3.4 the results for an elastic tube are shown and they can be
compared with those form = 0.9 and T = 1 shown in Fig. 3.3.5.
As m ~ 1 the standard viscoelastic model approached that for an

elastic material and this is indicated by the results shown in Figs.
3.3.4 and 3.3.5. The results for the elastic tube we obtained by
putting m = 1 in the program for the viscoelastic tube. It is clear

that the influence of-m is greater than of the relaxation time T.
For biological applications m = 0.5 is a realistic value and the

response of the tube wall is a rapidly damped dispersed wave form,
1.00
0.75
0
~ 0.50
-'
-x
0
0.25
~
X
-0.25
Fig. 3.3.1 Initial displacement

210 . J.B. Haddow
0.75
0
-·0
~
........
0.50
10
-><
~
0.25 t = 0.5
0.75
0
~
........
- 0.50
0
,.....
.:s. 0. 25 t = 1.0
~.
0.75
0'
-
~ 0.50
........
0
C\1 t 2.0
-><
0.25 =
~
0
X
-0.25
Fig. 3.3.2 w(x,t)/w 0 for m = 0.5, T = 1, a = 0.05

0
3:
........
iO
ci
-x 0.25 t = 0.5
3:
0
3:
-,....
0
- x 0.25 t = 1.0
0.75
0
3:
-
........
0
0.50
-
C\J
x- t = 2.0
3:
0
X
-0.25
Fig. 3.3.3 w(x,t);w 0 form= 0.5, T = 10, a= 0.05
212 J.B. Haddow
0.50
0
~
-
........
1.0
d 0.25 t = 0.5
-><~
~
0
-
~
........
0
t = 1.0
-><~
0.50
0
~
-
........
0 0.25
C\1 t = 2.0
-x
~
0
X
-0.25
Fig. 3.3.4 w(x,t)/w 0 for elastic tube, a = 0.05

0.50
0
:;
-
.........
1.0
0 0.25 t = 0.5
-:;
x
0
0.50
0
:;
-
.........
0 0.25
-:x;
T'""
0.50
0
:;
-
.........
0 0.25
-x
C\1
:;
0
-0.25
Fig. 3.3.5 w(x,t)/w 0 form= 0.9, < = 1, a= 0.05
214 J.B. Haddow
ACKNOWLEDGEMENT
The authors are indebted to Helen Wozniuk (Mechanical Engineering
Department Secretary) for preparing the manuscript.
LITERATURE AND REFERENCES

1. Brigham, E.O. (1974), The Fast Fourier Transform, Prentice Hall.
2. Euler, l. (l844), Principia pro motu sanguins per arterias
detarminado, opera posthuma mathematica et physica anno 1844
detecta, eiderant P.H. Fuss et N. Fuss. Petropoli: Apud Eggers
et Socios, Vol. 2, 1862.
3. Forbes, L.K. {1979), A note on the solution of the one-dimensional
unsteady equations of arterial blood flow by the method of charac-
teristics, J. Austral. Math Soc. 21 (Series B), 45.
4. Forrestal, M.J. and Herrmann, G. (1965), Response of a submerged
cylindrical shell to an axially propagating step wave, J. Appl.
Mech. 32, 788.

5. Hermann, G. and Mirsky, I. (1956), Three dimensional and shell theory
analysis of axially symmetric motions of cylinders, J. Appl. Mech.
23, 563.
6. Klip, W. (1962), Velocity and Damping of the Pulse Wave, Martinus
Nyhoff, The Hague.
7. Lamb, H. (1898), On the velocity of sound in a tube, as affected
by the elasticity of the walls, Manchester Lit. Phil. Soc. Mem.
Proc . 42 ( 9) , 1.
8. Lambert, J.W. (1958), On the nonlinearities of fluid flow in non-
rigid tubes, J. Franklin Inst., 266, 83.
References 215
9. Rubinow, S.I. and Keller, J.B. (1968), Hydrodynamic aspects of the

circulatory system, In Hemorheology, Proc. 1st. Int. Conf. Univ.
of Iceland, A.L. Copley, Ed., Pergamon.
10. Rubinow, S.I. and Keller, J.B. (1971 ), Wave propagation in a fluid
filled tube, J. Acoust. Soc. Am. 50, 198.
11. Rubinow, S.I. and Keller, J.B. (1978), Wave propagation in a visco-
elastic tube containing a viscous fluid, J. Fluid Mech. 88, 181.
12. Rudinger, G. (1966), Review of current mathematical methods for
the analysis of blood flow, Biomedical Fluids Symposium, ASME,
New York.
13. Rudinger, G. (1970), Shock waves in mathematicsl models of the
aorta, J. Appl. Mech. 37, 34.
14. Skalak, R. (1966), Wave propagation in blood flow in biomechanics,
Proc. Symp. Appl. Mech. Div. ASME, Y.C. Fung, Ed.
15. Tait, R.J., Moodie, T.B. and Haddow, J.B. (1981 ), Wave propagation
in a fluid-filled elastic tube, Acta Mechanica, 38, 71.
16. Teipel, I. (1973), Michtlineare wellenausbreitungsvorgange in
elastischen leitungen, Acta Mechanica, 16, 93.
17. Witzig, K. (1914), Uber erzwungene Wellenbewegungen zaher,
inkompressibler Flussigkeiten in elastischen Rohren, Inaugural
Dissertation, Universitat Bern, K.J. Wyss, Bern.
CHAPTER IV
SMALL ARTERIES AND THE INTERACTION WITH
THE CARDIOVASCULAR SYSTEM
THOMAS KENNER
Physiologisches Institut
Universitat Graz
Graz, Austria
I. INTRODUCTION
The purpose of this paper is the presentation and
discussion of problems which are related to the reactions
of small resistance vessels and their upstream and downstream
effects. It is intended to give a short and broad overlook
over this field. The function of small artuies cannot be
understood without uonsidering the interaction with other
parts of the cardiovascular system. These small muscular
vessels are executing control simultaneously on local blood
flow and on arterial blood pressure. This dual task some-
times leads to contradicting trends; e.g. during physical
exercise we find vasodilatation in order to increase local
blood flow, and at the same time vasoconstriction in order

218 T. Kenner
to prevent breakdown of the arterial blood pressure. Hydro-
dynamic phenomena are closely intertwined with metabolically
or neurally mediated smooth muscle reactions.
In this article many references are made to the work of
E.Wetterer and his pupils, and to the book by Wetterer and
Kenner (1968). More recent work from our group has been re-
viewed during a Satellite Symposium to the World Congress of
Physiology in Budapest (Kenner et al. 1982).
The work described in this article has been supported
by the Austrian Science Research Fund.
I .I What is a small artery?
Before I was invited to discuss blood flow in small
arteries I certainly did not realize the difficulty to de-
fine the term "small artery". An agreement on the gross
anatomical nomenclature is found for all vessels except just
the small arteries which, in some sources (e.g. Noordergraaf
1978) may be listed under the name "terminal arteries". In
other sources terminal arteries and small arteries are
assumed as different entities. In general, the successive
anatomical orders of arteries are called: aorta, large
arteries, main arterial branches, terminal branches, small
arteries and arterioles. From the arterioles rise the capil-
laries and thence the venoles and the venous part of the
system.
Introduction 219
Schmid-Schonbein (1976) cites a paper by Mall who in
1888 examined the structure of the canine mesenteric vascula-
ture and points to the quite important fact that "Malls data
are quoted in practically all textbooks of physiology through-
out the world (often as treated by Green (1944)). Important
physiological concepts . . . . . are based on these archaic fin-
dings."
Noordergraaf (1978) contrasts in a table Greens com-
piled canine data with those by Wiedeman (1962,1963) who
measured and counted microvessels in the bat wing. Noorder-
graafs table lists terminal arteries as having a diameter
of 600 rum according to Green and 19 rum according to Wiedeman.
The corresponding diameters of arterioles are 20rum (Green),
7 rum (Wiedeman), and those of capillaries 8 Fm (Green) and
3.7 ('lm (Wiedeman).
The whole problem seems to be complicated because of
large differences in different vascular beds. In any case it
seems to me necessary to point out that very often useful and
good data are missing, particularly for the most basic and
obvious phenomena, because nobody realizes that such data
have not been carefully measured with appropriate methods.
Fortunately in most vascular areas reasonably similar
vascular pattern can be found. One example of the terminal
bed of the bat wing is shown in fig. I (Mayrovitz et al.1976).
This figure demonstrates how small arteries give rise to
arterioles and how several capillaries branch off from one

220 T. Kenner
arteriole. In many publications on arterial flow and es-
pecially in textbooks a dichotomic branching scheme is ex-
plicitely or implicitely assumed. Actually, dichotomic bran-
ching plays an important role in single branches of larger
or medium size arteries.
A MAIN ARTERY ENTERING WINO COl
l
AI i .........,.. ART[RilS ..-z.--" +
IS !II ALL
Az SMALL ARTERY I 2)
As
AIIIT[fUOL.E.S
ID•IIALL.
-
...."'
;;;
...
141 lSI Ill
~ lw
.
ITARTL.taPCS' CAP' II'CV
SMALL ARTIRY 2 "• "' .., 5 z
"'
c 12 111 ALL.
;;;
>
c"' >
Fig.l Topological model of arterial branching

(Mayrovitz et al.1976)
A more general scheme of a series - parallel network model
is shown in fig.2
Fig.2 Series - parallel network (Popel 1980)

A arterioles,CAP capillaries,V venoles
Introduction 221
From a functional viewpoint Lee and Nellis (1974) could show
that the microcirculatory transit times can be modeled best
by an array of vessels as shown in fig.3
-·8-11,
TERMINAL VEIN
Fig. 3 Distribution of mean transit times 1n

a vascular network (Lee and Nellis :97--J
The schematic outlines of microvascular beds sho~~ i~
the preceding figures represent the three possible vie~?oi~ts
of presentation: fig. 1 shows the anatomical viewpoint "'·hi~h
is valid for one particular example. Fig.2 is a more general
scheme, compromising the exact and the restricted view with
an attempt of generalization. Fig. 3 represents a theoretical
simplification of an actually rather more complex three di-
mensional network (Metzger 1973, Popel 1980).
It seems ~hat we can describe the microcirculation in
most general terms as consisting of structural 11

modules 11 of
vessels the pattern of which is repeated in a certain tissue.
Fig. 3 represents such a repeating pattern. Another morpho-

222 T. Kenner
logic pattern of such microvascular moduli was described by
Lipowsky and Zweifach(1974).
In conclusion we can define the region of "small arte-
ries" as extending between larger and medium size arteries
on one hand and the arterioles and the microvascular moduli
on the other hand. We therefore have to include vessels with
a diameter between about 20 (llm and I 00 fUm.
In a functional sense this region is essential for the
control and biological variability of the peripheral resis-
tance, especially if we include the arterioles and the pre-
capillary sphincters - small muscular ring structures around
the entrance of true capillaries. Capillaries themselves do
not contain ·contractile elements. In other· words, if we dis-
cuss the function of the small arteries we must not exclude
the downstream part of the microcirculation.
1.2 Shear rate and shear stress and resistance to flow
Assuming laminar Poiseuille flow the pressure gradient
along a small artery is related to the flow q and the average
velocity v by
2
q v r 'II" (I)
and
Since the shear stress~, at the vessel wall is proportional
to the shear rate dv/dr by

Introduction 223
Ts = 'YJ dv/dr (3)
and since in laminar flow
(4)
(r. is the radius at the vessel wall), we find the following
1
relations for the shear stress at the radius r:
Ts=- (~p/L1x)(r/2) (5)
or
'ts= - 4 ~v/r (6)
or
Ts = - 4 '? q I ( r 31T ) (7)
The values of the shear rates at the vessel wall r. are in

1
-I
the order of 100 to 200 s in the ao~ta and atQ fg~nd_to
rise slightly towards smaller vessels. The largest shear
rates are reported for capillaries.Schmid-Schonbein (1976)

-I
cites values of about 300 s . The corresponding shear
stresses are quite small and range between less than I to 10

2
Pa. (I Pa corresonds to I 0 fbar = I 0 dyn/ em ) , Data about
values of shear rates and shear stresses in vivo differ in
different sources (e.g Whitmore 1968 and Caro et al. 1978

-I
report shear rates in capillaries up to 1000 s ). Shear
stresses in the normal circulation, in any case, are 3 to 4
magnitudes smaller than the values of the arterial pressure
(mean value 13.3kPa 133 mbar) and - as will be discussed
in section 2.2- 4 to 5 magnitudes smaller than the circum-
ferential stress in the arterial wall. As Patel et al.find,
shear stress values above 40 Pa = 400 ubar may damage the

224 T. Kenner
vascular endothelium.
The resistance to flow of one vessel segment of length
Llx equals
R (8)
and, therefore
R (9)
The total resistance of n equal parallel vessels of resistan-
ce R. is
l.
Rtotal Ri/n (I 0)
The "hydraulic hindrance" Ax/((n.2r) 4 ) (I I)
applied by Schmid-Schonbein (1976) is a magnitude proportio-
nal to the resistance R of n parallel vessels:

tot a 1
"hydraulic hindrance" = Rtotal1T' j ( 128 '?) (I 2)
In this short summary of simplified equations it should
also be mentioned shortly that blood shows non-Newtonian be-
haviour: the relation between shear rate and shear stress
cannot be described by the linear relation eq.3. Several
attempts to describe this relation have been made. Two equa-
tions cited from Noordergraaf (1978):

s
T5 = b(dv/dr) + C ( 13)
where C is the yield stress necessary to initiate flow. b and
s are constants. The following Casson equation describes
the behaviour of blood in a wide range of shear rates:
T.l/2 = k(dv/dr)l/2 + cl/2 (I 4)

s
Introduction 225
I .3 The arterial tree
According to Vadot (1967) 9 orders (O to 8) of vessel
sizes can be distinguished from the aorta to the capillaries.
As mentioned above, in many single arterial branches, especi-
ally in the extremities, in the mesentery and in the lung,
the main type of branching appears to be dichotomic. However,
the overall count of vessels of each order shows that from
each parent level about IS daughter vessels are branching off.
If z is the order number of each level, n , D and 1

z z z
are number, diameter and length, respectively, of each level,
then the following model relations can be assumed:

z
n n k ( I )
z 0 n
z
D DokD ( 2)
z
z
1 lokl ( 3)
z
k are dimensionless constants describing the behaviour of
each variable from level to level.
Fig.4 after Vadot (1967) shows the distribution of
vessels as calculated for a 60 kg adult with an assumed res-
ting cardiac output of 4.5 1/min. The corresponding equa-
tions of this concrete example are as follows:
n (15.4) 2 ( 4)
z
D 2 (0.376) 2 ( 5)
z
1 =59 (0.475) 2 ( 6)
z
with z = 0,1,2, . . . . . 8.
The "hydraulic hindrance" of this system is

226 T. Kenner
Fig. 4 Relation between log n (left, n: number o~ vessels),

log L (right, L length of vessel segment) and log D (ordina-
te, D: diameter of the vessel) in the human circulation.
The plot can be modeled by eq. 4 to eq.6. Closed circles:
arteries. Open circles : veins. After Vadot (1967)
"hydraulic hindrance" = 1 /(n D 4 ) = 3.69 (1.56)z (7)

z z z
The highestvalue of the resistance to flow, therefore, is
found in the capillary level (level 8 according to the model).
The resistance of the small arteries and arterioles (levels
5 to 7) corresponds to 48% of the total peripheral resistan-
ce of the model whereas the estimated resistance of the capil-
lary bed alone makes 37% of the total. The small rest of 15%
is due to the resistance of medium and larger arteries (lev-
els 0 to 4). - I t is important to note that these values
may be approximately valid for resting state. We can expect
marked biologieal variability and large changes during exer-
cise (see section 4.5).
The shear rate as well as the shear stress are propor-
tional to
l' L..o> D -3n -1 ( 8)

s z z
Insertion of the corresponding numerical value of the model
Introduction 227
assumption leads to
't"s ._., ( 1 • 2 4 ) z ( 9)
indicating an increase of the shear rate and of the shear
stress from level 0 to level 8 by the factor 5.6, a value
which is reasonably well 1n agreement with experimental data.
The Reynolds number 1s proportional to

Re ......., D -In -I (I 0)
z z
and after insertion of numerical values:
Re ......, ( 0 . I 7 5) 2
( I I )
Thus the model predicts a decrease of the Reynolds number by

-6
the factor 10 from the aorta to the capillary level. This
is in agreement with values given for the dog by Whitmore
(1968). Sources of data for the morphological properties of
the canine arterial system are found in Patel et al.(1963).
For human beings data are given by Patel et al. (1964),
lberall (1967) and Westerhof (1969). A survey of literature
is found in McDonald (1974).
The number of levels in the arterial tree deserves some
more discussion. An interesting and illustrative comparison
is possible for the pulmonary arterial tree. Vadot (1967)
counts 9 levels similar to the systemic arterial tree.
Lefevre (1982) mentions 20 levels of branching points which
are mainly dichotomous, an assumption which was earlier pro-
posed and discussed by Pollak et al. (1968). On one hand it
seems surprising that a morphologic distinction between 9

228 T. Kenner
and 20 levels seems to be difficult to decide. On the other
hand it is easy to understand that the final overall effect
in terms of the number of branches of 8 quadruple branching
points is equal to 16 dichotomic branching points.
As far as the systemic arterial tree is concerned, a
classification by McDonald (1974) assumes 9 levels similar
to Vadot (1967). Iberall (1967) introduced a model with II
levels.
The following statement by McDonald (1974) is certainly
quite correct: " The large numbers of X-ray arteriographs
that are made annually in all major medical centres would
suggest a large source of data which has not, to my know-
ledge, been systematically organised."
I .4 The branching ratio
In bifurcations the total cross section area of both
daughter vessels in most examples is somewhat larger than
the cross section area of the parent vessel. The socalled
branching ratio
(I)
is quoted to have values around 1.1 to 1.2 at the aortic bi-
furcation (Wetterer and Kenner 1968) and values close to
1.26 at many major arterial bifurcations (McDonald 1974).
This particular value I .26 of the branching ratio has
some interesting implications. It can be shown that for a

Introduction 229
given volume, resistance as well as energy dissipation are
minimized if the branching ratio of a bifurcation corresponds
to the value 1.26. Rosen (1967) showed that the minimization
of the following cost functional F can be used for this cal-
culation:
( 2)
where R is the resistance of the vessel, 1 0 ,r 0 and 1 1 ,r 1 are
length and radius of parent and daughter vessel. Using cal-
culus of variation one obtains for the case of symmetric
branching the following condition for the minimization ot F:

3 3
r 2r 1 ( 3)
0
or
r I . 26 r I ( 4)
0
or
2 2
d = 2r 1 /r 0 = I. 2 6 ( 5)
Thus , the branching ratio I .26 corresponds to the optimal
value.
Gessner (1981) recently pointed out that the condition
of equal wall shear stress in all vessels leads to the same
result. Since
( 6)
as shown in section 1.2 (eq.7), ~s the shear stress ~n the
parent vessel, the shear stress ~n each daughter vessel ~s
(7)
if both are related to the total flow q through the branches.
If both shear stresses are assumed to be equal, we again

230 T. Kenner
obtain
3 3
r0 = 2r 1 (8)
In the case of two unequal daughter vessels both conditions,
minimization of resistance and equating of wall shear stress
lead to the more general condition

3
r (9)
0
We had observed in section 1.3 that, actually in vivo,
a slight increase of the shear stress from the aorta towards
the smaller vessels can be observed. However, this increase
is not very marked.
There are some additional conditions and aspects concer-
ning branching and bifurcations, which are worthwhile to be
mentioned. Interesting is the behaviour of the characteris-
tic impedance (for definition see section 2.3 and 3.2) and
the influence of wave reflections (see section 3.5). Wetterer
and Kenner (1968) could show, that - assuming equal geometry
and elastic properties of parent and daughter vessels- the
condition
d 1 • 15 ( 1 0)
guarantees reflection free condition at a bifurcation.

Al:terial Wall 231
2. THE WALL OF SMALL ARTERIES
2.1 Introduction
The physical behaviour of arteries can be observed in
vivo or examined in vitro under the following conditions:
Vessels can be freely extensible like mesenteric arteries or
coiled uterine arteries. In this sase it will be observed
that during pressure pulsations each pulse pressure increase
will lead to a lengthening of the freely distensible segment.
As the other extreme, vessels may be longitudinally constrai-
ned. Here length is constant. Since arteries have the com-
mon property that increasing pressure distends them both
in longitudinal and in circumferential direction we will
observe that the force in length direction which is related
to the constraint , increases as the pressure decreases.
If an arterial segment is cut out of a tissue it retracts by
30 to 40% and is some cases even more. During transition of
an arterial pressure pulse the longitudinal wall stress may
either show pulsatile decrease or increase, depending on the
mean arterial pressure and on the contractile condition of
the arterial smooth muscles.
2.2 Arterial wall structure
The inner surface of any artery is lined with endothelial
cells which usually tend to be arranged with their large
ahis in the direction of the blood stream. An example of
endothelial cell lining in a small artery is schematically

232 T. Kenner
shown in fig.S.
Fig . S Endothelial lining of a small artery (left),

layers of a small artery (right)
The inner layer of the artery including the endothelial cells
and a small subendothelial layer is called the tunica intima.
This subendothelial layer contains in most small arteries a
more or less expressed internal elastic membrane which appears
folded in the relaxed state (as arteries usually are fixed
undistended for preparation for microscopic slides). The next
large layer - the tunica media - is the main load-carrying
layer of small arteries. It contains layers of elastic fiber
nets and smooth muscles which appear to interconnect these
elastic fibers. One of the best descriptions is found in
Benninghoff (1930). Besides the eaastic fibers and smooth
muscles the tunica media also contains collagen fibers. We
can distinguish elastic and muscular arteries. In general
small arteries belong to the group of muscular arteries.As
the percentage of smooth muscles inc~eases from larger ar-
teries towards smaller arterUs and arterioles the content
of elastic tissue decreases from about 45% to 20% accor-
ding to Noordergraaf (1978). The content of collagen fibers

Arterial Wall 233
decreases from about 30% in the aorta to about 20% or some-
what less in the arterioles. The smooth muscle cells increa-
se in total mass from about 25% in the aorta to around 60%
in the arterioles .
The media is surrounded by the socalled tunica adventi-
tia (or simply adventitia) which extends into the surrounding
tissue. It seems important for experimental measurements
and for the calculation of tensions and stresses, that the
determination of a clea~ limit between the vessel wall and
the surrounding tissue is 1 quite often,rather difficult.
The overall relation between pressure and radius of a
small artery can be characterized under quasistatic conditi~
ons as being S-shaped as shown in fig . 6 which shows the beha-
viour of a small rat carotid artery (Weizsacker and Pascale
1982).
Fig.6 Pressure-diameter curves from a rat carotid

artery (Weizsacker and Pascale 1982)
The artery was examined under longitudinal constraint with

234 T. Kenner
different prestretch in longitudinal direction. The length
of the segment was adjusted between 1.4 and 2,6 em. The larger
the longitudinal stretch the smaller the diameter.
The S-shape of the curves can be explained by a three
phase process. The tissue components show following sequence
of extension: flattening and stretching of elastic fibers,
stretching of smooth muscles. Finall~ the extension is limit
ted by the extension and stretching of collagen fibers.
2.3 Mechanical behaviour of arterial walls
Most structures of the arterial wall are arranged in
spiral loops as indicated in fig. 7 (Kenner 1967).
Fig.7 Arterial fiber structure and paralellogram

of forces in the wall. Kenner (1967)
The details of this structural arrangement of fibers may
differ from area to area. However, the general outline seems
to be quite similar (Benninghoff 1930, Kenner 1967, Rhodin
1980, Hudetz and Monos 1982, Weizsacker and Bescale 1977,1982

Arterial Wall 235
The forces in the vessel wall can be described by the
following equations if the wall thickness (h) is small com-
pared to the diameter (2r).
The circumferential force carried by a segment of
length 1 is
r 1 p ( 1)
the corresponding circumferential stress per unit cross sec-
tion area is
6'1 p r/h (2)
where r is the radius and h the wall thickness. Since, in
the relaxed condition, r/h ~ 10 the circumferential wall
stress has about tO times the magnitude of the arterial
pressure.
In the longitudinal direction the total force F 2 is
composed of two components

2
r 'TJ' P + F2ext (3)
where F 2 is the force which is related to,or generated by,

ext
the longitudinal constraint of the vessel. As has already
been mentioned in the introduction the effect of this force
can be demonstrated if a vessel is cut out from the tissue.
Then it retracts by 30 to 40% or even more depending on the
contraction state of the smooth muscles.
Per unit cross section area the corresponding stresses
carried by the material are:
p r/ (2h) + F2
ext
I (2r 1T' h) (4)
236 T. l(enner
The spiral structure of the vessel walls leads to an 1n-
teresting observation concerning the passive behaviour (van
Loon et al. 1977,Weizsacker and Pascale 1977,1982),which is
shown in fig. 8.In a rat carotid artery the length-force
relation of the vessel was recorded. It depends on the in-
ternal pressure in such a way that below a certain length the
incr~sing pressure tends to decrease the longitudinal force
F2 whereas above this particular length the effect is re-

ext
versed. We call this unique point in the diagram "the crossing
point".
a length (mm)
~~~~~~~~~~------~
1.40 2.02 2.64
Fig. 8 Longitudinal force F 2 ext versus length,
pressure range 0 to 200 mm Hg, pressure
steps 20 mm Hg.Wei~sacker and Pascale,I982
In the rat the behaviour can be observed in different vessels
as examined currently in our department. Van Loon et al.(I977)
have observed the same phenomenon in dog arteries. An expla-
nation is given in the following sketch, fig. 9 (Kenner 1967).
As a vessel is distended by internal pressure, the fibers tend
to adjust themselves into the direction of the parallelogram

Arterial Wall 237
of forces in the wall. The equilibrium condition will be de-
termined below (eq.9). If an artery is extended over the
equilibrium length,the increasing internal pressure will
tend to "contract" the artery. If the length of the artery 1 s
below the equilibrium point then increasing internal pressure
will extend or relax the artery as explained in fig.9.
Fig. 9 Scheme, explaining effect of pressure

on length-forc es in an arterial wall
The following simplified calculation estimates the
average angle of the spiral fibers in the equilibrium state
if these fibers are assumed as unrestraine d by elastic for-
ces. Under the condition of free extension of an unrestraine d
vessel segment the external length force F 2ext is assumed
as zero:
(5)
F2ext = 0
then from eq. 3 and 4 follows
II 2 (6)
Using the relations shown in fig. 7 we can, furthermore ,
derive
(7) 6"1 F 1 /hl 2 F cos! I (hl sin f)

and
(8) 62 F 2 /hl 1 F sinf/(hl cos!)

238 T. Kenner
Therefore, the equilibrium condition is

2
= tg 1 (9)
and with the condition of free extension (no constraint)
J= 35.2° , a value which is reasonably
close to measured values (Kenner 1967, Kenner et al. 1 n2,
Hudetz and Monos 1982).
The interaction between forces and extensions in three
orthogonal directions is described by the Poisson ratio (see
Fung 198l,Kenner 1967). The Poisson ratios of the arterial
wall tissue show anisotropic behaviour and depend on the
extension. At small values of the extension, near~no inter-
action between orthogonal directions aan be observed. The
Poisson numbers under this condition may be close to zero.
During extension the Poisson numbers incEease. In an extended
spiral fiber structure like that shown in fig. 7 Poisson
ratios of larger than I can ~observed (Kenner 1967).
Due to the elastic distensibility of the spiral fiber
structure, a freely extensible arterial segment wnich is
continuously filled and extended by fluid will stretch in
longitudinal as well as in circumferential direction. In
contrast, a rubber tube will, under the same conditions ex-
tend only in circumferential direction becaus~ in tms case,
the material is homogeneous The interaction between ortho-
gonal directions is described by the value 0.5 of the Poisson
number in all directions (Wetterer and Kenner 1968).

Arterial Wall 239
Fig. 10 shows schematically the relation between pressu-
re and external longitudinal force F 2 in a rat carotid

ext
artery under three different conditions. Below the crossing
point - expkined in fig.8- an increase of the distending
pressure decreases the external constraining force F 2 .

ext
length force ('5 .
F: ei-~
2 ext '(\\~n
equilib.
Fig. 10 Longitudinal force as a function of the

distending pressure . Parameter: length
Exactly at the crossing point, the external force is inde-
pendent of pressure. If the vessel segment is extended beyond
the crossing point, an increase of the internal pressure will
further increase the external constraining force as was
schematically explained in fig. 9. We assume that the
crossing point is close to the natural length of the vessel.
It seems possible, however, that the contractile status of
t~vascular~smooth muscles may influence the position of the
crossing point.
It can be expected that the three possibilities of
longitudinal force variation to changes in arterial pressure
may also be observed during pressure pulse transients.

240 T. Kenner
Wetterer and Kenner have already in 1968 critizised naive re-
ports - usually badly supported by experiments - which from
time to time mean to "prove" the existence of fast contrac-
tions of vascular smooth muscles which follow the pulse wave
and are interpreted as kind of peristaltic movement. These
reports, including a recent one on rhythmic contractile
activity of the in vivo rabbit aorta, can be explained by
the interactions between pressure and length force as demon-
strated in fig. 10. Of course it is not surprising that acti-
vation or paralysis of smooth muscles influence the phase
relation between pressure and length force if a shift of the
crossing point is produced.
2.4 Relations between vessel wall properties, flow and wave
velocity - some "rule of the thumb - equations"
Although in small vessels the effect caused by the
viscosity of the fluid and of the wall material (both of
which will be discussed later) is rather marked, some simple
relations should be mentioned here. The relations which are
valid only for ideally elastic frictionless tubes, neverthe~
less, appear useful to explain some fundamental phenomena
and to check orders of magnitude (Wetterer and Kenner I 96 8).
The relation between pressure amplitude ~p and flow
amplitude ~q in an ideally elastic tube without reflections
is defined as characteristic impedance:

Arterial Wall 241
z .lip I Aq (I)
Provided viscosity can be neglected, the eharacteristic im-
pedance is related to the wave velocity c in the tube by
z c e 1A ( 2)
2
where A = r ~ is the cross section area of the vessel. The
properties of the vessel which is filled with fluid of den-
sity ~ can be described by the compliance C per unit length
c 6AI .1p (3)
and by the socalled effective mass per unit length
( 4)
The wave velocity of an ideally elastic tube c is related
to these magnitudes by
2
c = JIMC ( 5)
We define the velocity amplitude of a flow pulse as the
peak amplitude of the velocity averaged over the cross sec-
tion area of the tube:
Llv LlqiA (6)
The equations described above yield the following rela-
tion between pulsatile variations of the cross section area
( 6A), the velocity pu ls~ amplitude ( Llv), and the wave velo-
city c (Wetterer and Kenner 1968):
f1v I c = JAI A 2 /Jrlr (7)
If,in a pulsating artery, pressure and diameter is si-
multaneously recorded we observe a relation between the two
variables which indicates by its gross steepness the elastic
distensibility of the tube. Usually the record shows a hys-

242 T. Kenner
teresis loop in which pressure leads diameter. As will be
discussed below the area of the loop is proportional to
the viscous energy loss produced by viscous friction within
the wall material. In elastic arteries the loop usually has
a smaller opening than in small muscular arteries. An exam-
ple taken from Wetterer et al. (1977) is shown in fig. II.
p 0
(mml
3.50
'5'5 p (mm Hg)100
0 t (sec) 0.5
Fig. II Pressure (p) and external diameter (D) of

a common carotid artery of a dog. Wetterer
et al. (1977)
In this example we find that the relative radius pulsation is
in the order of magnitude of 2% in the femoral artery of a
dog, corresponding to an area pulsation of 4%. Thls is in
agreement with eq. 7 if the ratio equals
.!lv I c 80(cm/s)/2000(cm/s) (8)
These values agree reasonably well with acwal measurements
in the dog, although the wave velocity is rather high.(See
also section 3.6). We have to mention here that the proper-

Arterial Wall 243
ties of the aorta-iliaca-femoral tube change in several ways
from the aortic root towards the periphery (Wetterer and
Kenner 1968, Kenner 1979). As shown schematically in fig.12
25 /
/
/
/
I
I
20
15
10 \ C(xJ
\C(OJ
5
,~y rcrr?J c(x)
C(Q)
'~·'-'
50 100 150cm
Fig. 12 Properties of the aorta-iliaca-femoral

tube; geometric and elastic tapering
the compliance C, and the cross section area A decrease,
whereas the pulse wave velocity c and the characteristic im-
pedance Z increase towards the periphery (as estimated for
a human being). The tapering effect explains the high wave
velocity in smaller muscular arteries.
2.5 Dynamic properties of small arteries
There is an abundant literature about the dynamic elas-
tic properties of blood vessels. Recent reviews include Cox,
Monos and Kovach,Hudetz and Monos, Busse et al. ,Newman and

244 T. Kenner
Greenwald, in Kenner et al. (1982); furthermore Dobrin 1978
and Basar 1981.
Busse et al. ( 1982) have recently reexamined the dynamic
behaviour of the rat tail artery under quasistatic and dyna-
mic condition, both under extreme relaxation by papaverin
and during activation by noradrenalin. Under quasistatic
conditions,i.e . during slow increase and decrease oi pres-
sure 1 the activation leads 1 besides a contraction 1 to a marked
widening of the hysteresis loop as shown in fig 13,left.
350 //--_/ ,/
re
500
,, .---?-~- --~--
i ~'""'""""
l~ml
re
i!Jm]
//
450
II Papaver~ne
1 Hz
300
7 1Hz
II I
~
400
0 5 10 15 0 5 10 15
P lkPa] P lkPa]
Fig. 13 Quasistatic pressure-radiu s curves of

the rat tail artery (Busse et al. 1982)
The superimposed small dynamic loops, recorded at 1 Hz did
not change their slope very markedly. A similar observation
was also reported by Newman and Greenwald (1982).This obser-
vation implies that there are two different mechanisms which
have to be considered in order to explain the viscoelastic
behaviour of small vessels with and without activation of
smooth muscle contraction. Busse et al.(1982) assume that
the widening of the hysteresis loop during muscle contracti-

Arterial Wall 245
on 1s due to the time- and load-dependent reattachment of
cross bridges of contractile elements, actin and myosin in
the smooth muscles. Therefore the pattern of the quasistatic
hysteresis loop can be attributed to the properties of the
contractile element (CE) of the smooth muscles. The quasi-
static elastic modulus, visually expressed in fig. 13 by the
steepness of the radius-pressure relation is markedly lower
during contraction than in the relaxed state.
The steepness and, therefore, the dynamic elastic modu-
li during the small 1 Hz cycles seems to depend mainly on the
mean pressure and is not much influencedby the contraction
of the vessels. The behaviour of these small dynamic loops
can be ascribed to the properties of some passive elastic
elements which can be modeled as between in series to the con-
tractile element.
2.6 Linear modeling of the dynamic elastic behaviour
As a first approximation of the description we assume
piecewise linearity and consider only small extensions
E = dl/1 ( I )
Between E and the stress d and the elastic properties we
can observe the following general relation:
+ 'l')w d€ Idt ( 2)
Complex Fourier transformation and division by ~ leads to
E + j W~w (3)
246 T. Kenner
E is the complex modulus of elasticity. Ed is the dynamic
modulus of elasticity. ~w is the coefficient of wall visco-
sity. The equation implies the application of a model as
shown in fig. 14 which is usually called a Voigt solid(Fung
1981). If the magnitudes Ed and ~w are allowed to be func-
tions of the frequency W then any
viscoelastic model can be fitted
into the above equation. This means
that both components of the model
in fig.14 are assumed as frequency
dependent.
In my opinion a certain confu-
Fig.l4 sion may possibly be caused by the
sometimes indiscriminate exchange of the v~scous (dashpot)
and the contractile element (CE) in models given the same
names. Fig. 15 shows, for comparison, Voigt and Maxwell mo-
dels from Noordergraaf (1978) left and from a paper by Cox
(1982) right.
VOIGT MAXWELL
~
SE
PE
CE PE CE
A. MAXWELL 13. VOIGT
Fig.15 Models with and without contractile element

Arterial Wall 247
Certainly, whenever the use is properly defined, the 4iffe-
rence is not a mistake but, nevertheless, confusing. A possi-
ble confusion may be even increased because the actual Max-
well element is a model of an elastoviscous fluid.It seems
to me obvious that in a model of a smooth muscle the series
elastic element (SE in fig.l5) which describes the elastic
behaviour of the dynamic 1 Hz cycles in fig. 13 ought to be
a viscoelastic element. The parallel elastic element (PE)
describes the properties of the muscle during relaxation.
From the description of Busse et al.(l982) it appears that
PE as well as CE exhibits complex properties during contrac-
tion.
In terms of the model of fig.l5 Busse et al. (1982) de-
scribe the following results in a small artery: Ed and ~w
increase with increasing wall stress. At a given wall stress,
Ed is virtually independent of the frequency while ~wdecrea
ses markedly with increasing frequency. This behaviour of ~w
is called thixotropy or pseudoelasticity and agrees basical-
ly with Bergels wellknown observations (1961). Smooth muscle
activation does not markedly change either Ed or ~w in the
small muscular arteries.
2.7 A note on incremental moduli
According to a report by O.Frank (1906), W.Roentgen
( 1876) - the discoverer of the X-rays- was the first to reco-
gnize the problems related to the description of highly QX-

248 T. Kenner
tensible materials. O.Frank himself was the first to intro-
duce an incremental modulus of elasticity. As discussed by
Kenner (1967) and by Wetterer and Kenner (1968), there are
two possibilities to define Franks incremental moduli. The
difference apparently, was not recognized by Frank (1906,
1920).
Since, for certain problems, the difference plays a
role, we discuss the problem in the following.
The two definitions of incremental moduli are the
"force related modulus"
E (dF/A) (1/dl) (I)
and the "stress related modulus"
dO' dl/1 (2)
with F force, A cross section area of the material, d = F/A
stress. The interrelation between the two moduli in the case
of one dimensional stretch is given by the equation
dO'= dF/A C:S dA/A (3)
and, therefore, valid for extensions in one direction and
with the assumption that the material is incompressible:
E + 6 (4)
The problem which has been solved in a paper by Kenner
(1967) is the fact that the relation of the two modulus
systems in two dimensional extensions (of thin-walled incom-
pressible arteries) can only be described by systems of rna~
trix equations. Both systems of matrix equations can be re-

Arterial Wall 249
lated to the stress energy function of the material.
For a longitudinally constrained vessel the following
sim~lified equations describe the circumferential extension
(index I) in terms of the force related modulus system (length
direction index 2, flij are the Poisson numbers which are de-
pendent on the direction and differ slightly between both
modulus sys terns) :
dr/r dFI/(AIEI) - fli 2 dF 2 1(A 2 E 2 ) ( 5 a)
dlll = 0 - fl2IdFII(AIEI) + dF 2 1(A 2 E 2 ) (Sb)
Further insertion yields
drlr [dFII(AIEI)](I - fUI2 fU2I) (6)
and
dFIIAI d(lrp)lhl (r dp + p dr)lh (7)
Insertion of these equations leads finally to the definiti-
on of the circumferential incremental modulus in a length
constrained tube (in the force related definition):

(8)
The corresponding equation for the stress related modulus
is
CI = ~ ( r d pI d r + 2 p) I ( I - fU I 2 fU 2 I ) ( 9)
According to Wetterer and Kenner (I968) two facts are note-
worthy; I) The usual definition of the incremental modulus
E. which is also valid for thin walled tubes, corresponds

1nc
to the force related modulus EI (e.g.Bergel I96I, Monos and
Kovach 1982). 2) The usual omission of the factor p (pressure)
which comes into the equation from the correct differentia-

250 T. Kenner
tion of the circumferential force (or stress - in the stress
related modulus system) may produce an error of 10 to 20% in
the calculated modulus.
Busse et al. (1982) have used the force related incre-
mental modulus to describe the dynamic properties of small
vessels taking into account finite wall thickness (h). r. and

~
re are internal and external wall radius, respectively. f is
the Poisson number and 'f is the phase angle between sinu-
soidal pressure and radius variation. The correct definition
of the incremental dynamic modulus according to Busse et al.
(1982) then is as follows: The real part
Ed (W) = (( 2 r 2 r . ) I h ( r ~ + r 2 ) ) ( I -.u 2 ) ( p + r . (.1p I L1r . ) c o s cp ( Io)

e ~ ~ e 1 ~ ~ J
and the imaginary part of the modulus
VJ')Iw= ( ( 2 r 2 r . ) I h ( r 2. +r 2 ) ) ( I -JU 2 ) r . (t.Jp

I e ~ ~ e 1
Jl
~
I Ll r . ) n . n IP
~ J
(I I)
The assumption that the Poisson~mber is invariant, inde-
pendent of stretch, isotropic and equal 0.5 in all directions
is certainly not correct in vessel walls, and introduces ano-
ther error - which, by the way, is quite frequently found in
the literature, However, the experimental determination of
Poisson numbers is rather difficult.
2. 8 Critical closing pressure
Most small muscular arteries and arterioles have the
capability of complete closure by contraction. Wetterer and
Kenner (1968) have extensively discussed the concept of a
critical closing pressure. Burton (1951) had assumed that

Arterial Wall 251
during a decrease of the internal pressure or during an in-
crease of active wall tension the radius may become smaller
and smaller until at a certain moment "instability" is rea-
ched as the radius shrinks to zero. On the one hand we ac-
tually can observe a zero-flow pressure intercept. On the
other hand the original concept by Burton (1951,1962) is mis-
leading because of the somewhat abstract and ideal assumption
that a vessel may in the contracted state have a zero radi-
us.Furthermore, the use of a "Laplace" tension defined as a
force per unit length does not take into account that the
vessel wall material has a finite cross section. Wetterer and
Kenner (1968) have critizised the reference to "Laplaces law"
which may be appropriate for the description of soap bubbles
but not for the analysis of blood vessels with finite wall
thickness. It seems particularly important that the mean ra-
dius of the tunica media never can become zero even in the
maximally contracted state.
Meanwhile similar critique has also come from other au-
thors as summarized by Gow (1980). In conclusion we can assu-
me that a closure of an artery is possible without any "insta-
bility". The endothelium then is found to be extremely folded.
An opening is possible at the same pressure as the closing
pressure because at some point upstream the lumen of the con-
tracted vessel may extend conically into the next larger open
segment. Large arteries do not have the capability of comple-
te closure.
252 T. Kenner
Burton (I9SI) in any case has the merit to have initia-
ted a discussion on a very important topic. Closure of ves~
sels and its relation to the properties of the vessel wall
plays a very important role under pathologic conditions as e.
g. the development of hypertension.
It should be added here that the closure of small ves-
sels can be brought about by three processes: I) the first
process is the more or less concentric closure by the con-
traction of the media. 2) In striated muscles and particular-
ly in the myocardium the tissue pressure generated by the
muscle contraction leads to a collapse of small vessels. In
other words, under these conditions small arteries have.to
be considered as collapsible vessels. The two processes are
schematically shown in fig.16 3) In some small vessels
swelling of endothelial cells has been observed.
collal"e ~
Fi g . I 6 Co n s t r i c t i on and c o 11 a p se o f a s ma 11 v e s s e 1
2. 9 Pressure-flow relation at the peripheral resistance
Many measurements of the peripheral pressure-flow re-
lation have been made in local arterial beds and in the
whole arterial system as summarized by Wetterer and Kenner

Pressure-Flow Relation 253
1968,McDonald 1974,Kenner 1979, Kenner et al. 1982. One exam-
ple of a typical quasistatic measurement in the lung circula-
tion by Wezler and Sinn 1953 is shown in fig.l7.

flow ~W~ljmin o.oo21Sf>l 885
80 .---.----'-.---r 0.001~P 36 S.,--.,--p----f-1.,..._....- 0.0125? 2·358
o.oo•p268s
0.000095?364
Fig.l7 Pressure-flow relatio~ (Wezler a.Sinn 1953)
The experiments were made in the plasma perfused rabbit lung.
Besides each curve from single measurements the equation of
the parabola ~s written which is used to approximate the func-
tion. The shape of the curves uan be explained by the passive
elastic distensibility of the small arteries and arterioles.
Green et al.(l944) had first proposed a description of
the pressure-flow relation in the form of

n
q = a p ( I)
As discussed by Ronniger (1955) the following relations can
be derived from this equation: the absolute value of the re-

n-1
sistance 1/R b = q/p = a p (2)
a s
and the differential value of the resistance
n-1
= dq/dp = a n p (3)
254 T. Kenner
and the relation between the two
Rd/Rabs = 1/n (4)

With respect to the results described below it seems
that in many examples the pressure-flow relation of periphe-
ral arterial beds can be well described by a straight line
with the slope 1/Rd and the pressure intercept pc.With some
caution the index c may be interpreted as closure. The follo-
wing fig. 18 indicates this behaviour.
p
Fig.18 Pressure-flow relation
The function can be described by the following equation:
(5)
For the measurement in the whole circulation, Wetterer
and Pieper (1953) have developed a method which has recently
been revived and extensively used by Dujardin (1982). The
principle of the method is based on the indirect measurement
of diastolic outflow from the arterial system during the ar-
tificial introduction of slow volume variations with a pump
connected to a large artery. The frequency of the pump is
in the order of 0.2 Hz. Since the variations of pressure and
volume are rather slow, a simple windkessel model is used for
the evaluation (Wetterer and Kenner 1968,Dujardin 1982) as

shown in fig. 19.
1 Q~
___T...J--_ __,T Pc;
Fig. 19 Windkessel model
In this model the zero pressure intercept is simulated by
a battery. In the notation of Dujardin (1982) the non linear
compliance of the windkessel is
C = C (I
0
- k p) (6)
where k is a constant and p is the pressure. The total flow
into the windkessel is then described by the equation
(7)
During each diastole the input into the system equals the
pump flow qp • The first term on the right side of the wind-
kessel equation is always the outflow qR through the periphe-
ral resistance. With respect to the second term we introdu-
ce the following approximation and abbreviation:

dp/dt ;:;; ..!lp/D s (8)
where D is the duration of the diastole and ~p is the pres-
sure difference during diastole. Then the windkessel equati-
on reads during diastole as follows:
(9)
Fig. 20 shows an original recording from Wetterer and Pieper
(1953) in a dog. At 90° the maximum volume has been pumped
into the aorta; 270° indicates the maximum volume in the pump.
In the original method Wetterer and Pieper (1953) used a pair
of pulses with the same mean diastolic pressure (pulse A and

256 T. Kenner
B).For both pulses the outflow qR is the same.Furthermore, if
180° :270° QO
0....._1-'---
Fig. 20 Blood pressure during sinusoidal pumping

(Wetterer and Pieper 1953)
pulse A is recorded during positive pump flow, pulse B cor-
responds to the negative pump flow of the same magnitude.The
windkessel equation then yields
and (I O)
where the indices A and B relate to the two pulses. From the
equations (eq.IO) we find two solutions.One permits to calcu-
late the total compliance C of the arterial system.The other
solution yields the total peripheral flow qR during diastole.
Q mt/sec
20 A
c
10
I
I
I
: I
o I
,./Pet 1
0.___
1/ ~I
___,.::..---L-+-----f--~
p
50 TOO 150 mmHg
Fig.Zl Pressure-flow relation (Wetterer a.Pieper 1955)
Fig. 21 shows an example of the relation between pressure
and flow in an anesthetized dog before and after administra-
tion of adrenalin (A) as measured by Wetterer and Pieper
(1953,1955). It is interesting to note that the pressure-
flow relation actually is found to be straight and that vaso-
constriction leads to a shift of the whole relation more or
less parallel. This indicated that Rd may, during vasocon-
striction, stay constant. Dujardin (1982) used a parameter
estimation method for the evaluation of the windkessel equa-
tion. Fig.22 shows the pressure-flow relation measured in
normal rats (WKY,circles) and in spontaneously hypertensive
rats (SHR,squares). In this example the slope of the pressure
flow relation is different for the two conditions indicating
an increase of the differential value of the resistance Rd
in spontaneous hypertensive rats compared to normal rats.
a SHR
o WKY
100 200 300.

Fig. 22 Pressure-flow relation in rats(Dujardin
1982)
The method described in this section allows clearly to
characterize the overall behaviour of the small arteriei.

258 T. Kenner
3. FREQUENCY DYNAMICS OF THE PERIPHERAL CIRCULATION
3.1 Introduction
With some exceptions we have so far restricted our dis-
cussion to quasistatic phenomena. However, the mechanics of
the normal circulation is characterized by pulsatile activity
of the heart. It is important to note that a characterisuic
dimensional relation exists between heart rate and the pro-
portions of the arterial system (Kenner, 1979). The normal
resting heart rate of animals is reciprocally related to the
length dimension of the body. Therefore, head period, time
constants of diastolic arterial pressure decay and autooscil-
lation periods of the arterial pressure are proportional to
each other and to the cube root of the body mass. Therefore,
the contour of the arterial pulses is similar in all animals.
Likewise impedance and pressure transmission functions are
similar - under the condition that the frequency is normali-
zed with respect to the heart rate.
Fig.23 shows the dependence of the aortic value of the

,
unsteadiness parameter (Witzig 1914, Womersley 1957) ~
rywe;~· (I )
which is proportional to the square root of the aortic radius.
Impedance, defined as complex relation between pressure
and flow, and pressure transmission function,defined as com-
plex relation between two pressures, are well established
functions used to describe the properties of the arterial

Frequency Dynamics 259
system. Both functions depend on two basic components: the
characteristic impedance and the peripheral resistance.

(X
25
20 -- --
Ox •
Hone
.
--- -
ex =15v"f
15 .,.. 1 .... .,.. Man
/ Dog
10 ~Rhesu~monkey
--~ abbit
5
1
t
/ t.Cat
t.Guinea Pig
Rat
1
tMouse
0 0·5 1·0 J.5 2·0 2-5 em
Fig.23 Parameter~ as function of the radius of the

aorta (Kenner 1972)
In discussing frequency dynamics we will have to cover
I) the frequency region of the heart rate and its harmonics
and 2) the low frequency region. - In our discussions so far
the small arteries, the arterioles and the microcirculation
have been assumed as time invariant localized elements which
determine the magnitude of the peripheral resistance. This
part of the article will extend the view and include reacti-
ons and responses of the small arteries and resistance ves-
sels which are mediated by nerves or by autoregulatory mecha-
nisms.
Summaries of frequency dynamics of the peripheral cir-
culation are given by Attinger (1973) and Basar (1981).
3.2 High frequency behaviour. of a transmission line
In spite of the possibilities to calculate flow and
pressure in rather complex tube systems using increasingly

260 T. Kenner
detailed and accurate models and methods the fundamental ana-
lysis of a 4-terminal network transmission line still is
highly useful for the study of pressure and flow in arteries.
A 4-terminal network element as discussed in the follo-
wing is shown in fig.24. zl
Fig. 24 A 4-terminal network element
Indices 1 and 2 refer to input and output, respectively.The
relation between pressure and flow at input and output can
be described by the matrix equation
where
(::) (::l A
( 1)
1
A ( (2)
1 I zt
If the 4-terminal element has the infinitesimal length dx
then the transfer matrix A for one such element is
A (3)
By a matrix operation it is possible to calculate the matrix
A for a chain of such infinitesimal elements which, put to-
gether in series, constitutes a transmission line of length L.
( co•h fL Z •inh fL)

A = (4)
(1/Z)sinhfL cosh 1 L
where Z is the characteristic impedance:

(5)
The complex transmission coefficient:
B + jk (6)
The two components of the transmission coefficient are the
real part B, the damping coefficient, and the imaginary part
k , the phase coefficient. The latter is related to the wave
velocity c by
c W/k (7)
The role of the transmission coefficient is best described
by the relation between complex pressure PI at the entrance
and P 2 at the end of a tube with a matching impedance Z at
its end: P
2
= P
I
e- tL (8)
In general the pressure-flow relation at the end of a
tube can be described by a (real or complex) impedance R:
(9)
and after multiplication of the matrix equation we find the
complex pressure transmission function:
+ (Z/R)sinhfL (I 0)
Furthermore, the flow transmission function:
+ (R/Z) sinh f L ( I I )
and the input impedance

cosh tL + (Z/R)sinh fL
z ( I2)
sinh fL + (Z /R) cosh ( L
In order to apply these equations we have to describe
the longitudinal impedance z 1 and the transverse (circumfer-
ential) impedance z . Fig.25 shows a generalized 4-terminal

t
262 T. Kenner
element of a viscoela stic leaking tube.
Fig. 25 4-termina l element of a viscoela stic

leaking tube
The longitudi nal impedance is composed of resi~tance to f~ow
R and effective mass M per unit length
R + jwM ( 13)
As rough approxim ation one possible assumptio n is
R ( 14)
The effective mass per unit length is defined as
M (I 5)
where A = r
2
11' and eis the density of blood. By introduci ng
the unsteadin ess paramete r« as defined in section 3.1 :
jWM(I - j 8/t~h (16)
For the transvers e impedance we consider viscoela stic
distensi bility of the vessel wall. The complianc e C is
c 1/(1/Cd + . WR
J w
) ( 17)
Cd is the dynamic complianc e and Rw is the internal frictio-
nal resistanc e of the vessel wall. - The electric analog for
a parallel spring and dashpot is a capacitor and a resistor
in series as shown in fig.25.
In addition, leakage by outflow from the tube through
small side branches is taken into account by the outflow ad-
mittance G. Altogethe r we can write:

1/(G + j WCd/(1 + jtg~ )) ( 1 8)
The tg of the phase angle between pressure and radius vari-
ation due to the viscoelastic wall properties equals
tg ~ (19)
Under the condition that the resistance to flow R is small
and can be neglected, and that tg 1 as well as G are small,
we find the characteristic impedance for larger vessels
with the abbreviation Z VM/cd'. As shown in section 2.5,

0
tg 1 is usually rather small and is independent of frequency
since, as mentioned ~wdecreases with increasing frequency.
For the phase canstant k we find
( 2 1)
Therefore, the phase velocity in a reflection free tube is

A
c = W/k = 1 /~Mcd' (22)
and independent of the frequency. The damping coefficient is
13 Z G/2
0
+ ktg 1 /2 (23)
Since k increases proportional to the frequency , the damping
factor B in a viscoelastic tube increases with frequency.
In small arterial vessels we can assume leakage G, the
viscous wall resistance R and the effective mass M to be

w
negligible compared to the other components (Gross 1977).
Under these conditions we find as approximate solutions - C
is the compliance of the wall:
B + jk = Vj w RC' (24)
264 T. Kenner
Damping as well as the phase constant increase with the fre-
quency : ~WRC/2 (25)
k VwRC/2 (26)
Therefore also the wave velocity is rather small and increa-
ses with increasing frequency (Caro et al.1978; see sect.3.6).
c = ~2 w/Rc' ( 2 7)
The characteristic impedance turns out as a complex function
(28)
The amplitude of the impedance decreases with increasing fre-
quency, in contrast to the characteristic impedance of large
vessels. The phase angle is constant -45°. - It should be
noted that only one solution of the square root which permits
a positive real part is physically sound. Actually the charac-
teristic impedance measured in small vessels agrees well with
this prediction. Fig. 26 is taken from Wetterer et al. 1977
and shows the characteristic impedance of the femoral artery
of a dog. IZf ldyn 61't/cm'l

t5 ·•o'
- • P,.: 1401t1M H9
··-• P.,• 90~t~MH9
IO
l ' 6 1 10 11 14 f tHrl
lrociJ
-I
1 4 6 I tO 11 14 f IHal
Fig. 26 Characteristic impedance of the femoral

artery of a dog (Wetterer et al. 1977)
Gross (1977) has compared measurements of the pressure ampli-

tude in the microcirculation with results of a simulation in
a model, which consists of 5 levels of dichotomic branching
microvessels. Fig. 27 shows the calculated relative pre~re
amplitude in the vessels of this network (Gross 1977). The
pressure decay is mainly due to the marked damping by viscous
friction of fluid resistance.

1.0,__ __
0.8
,v
2
Q.
E
0 0.6
v
a
~
.;:
0.4
0.2
Fig. 27 Decay of pulsatile
Capillaries
pressure amplitude in a net- Arterioles Venule
work model of the microcircu- 01 4

lation (Gross 1977) Level
3.3 The high frequency input impedance of arteries
In this connection "high frequency" means heart rate
and its harmonics. Besides that the input impedance always
comprises the zero frequency value, i.e. the mean absolute
value of the peripheral resistance at the end of the examined
arterial bed (R b ). -As far as the pulsatile component of

a s
the pressure-flow relation is concerned, the input impedan-
ce also contains information about Rdiff as explained in
fig. 28 (Kenner 1978). The low frequency part of the impe-
dance extends between the points Rdiff and the heart rate.
266 T. Kenner
The contour of this part contains
information about reactions of
the arterial bed as will be dis-
cussed in part 4. In fig. 28 no
'~'--f--._ low frequency reaction is assumed
! --z----:-1- - the peripheral resistance is
he•l"t constant. The high frequency part

rJ~te
Fig. 28 Input impedance. of the input impedance of arteries
Schematic plot
tends to the value of the charac-
teristic impedance Z with increasing frequency. Particularly
in large vessels there may be maxima and minima of the input
impedance express~ng resonance phenomena (Wetterer and Kenner
1968,Kenner 1972,1979, W~terhof et al. 1979). At least one mi-
nimum may be observed more frequently somewhat above. the
heart rate. For the simulation of this minimum a manometer
model (see below) may be used. The input impedance of a hom-
geneous large tube with a resistance at its end shows several
resonance maxima. The rather flat high frequency part of the
arterial impedance in vivo is due to tapering and branching
of the arteries (Kenner 1972). A recent summary about aspects
concerning the arterial input impedance was presented by
O~Rourke (1982).
Windkessel models have been used to simulate input im-
pedance of the whole arterial system or of arterial branches.
Two of the more important models are the "improved windkessel"
model by Broemser and Ranke (1930) -recently sometimes called

Frequency Dynamics 26 7
"westkessel" (Noordergraaf 1978) - and the manometer model
by Broemser (1932). The input impedance of the improved wind-
kessel model tends towards the series resistance R 1 (in.the
nomenclature of fig. 29 which is taken from Wetterer and
Kenner 1968). This makes the model useful to simulate arteri-
al branches since R 1 simulates the effect of the characteris-
tic impedance Z (as indicated in fig. 28). The second model,
the manometer model includes the effect of the effective
mass and accounts for one minimum of the input impedance
which in some in vivo recordings can be observed (Westerhof
et al. 1979, 0' Rourke 1982).
c-~
r~, c.L :\
_Jr
Fig. 29Left: "improved windkessel" by Broemser and
Ranke (1930). Right: manometer model by
Broemser (1932)
There are two possibilities to plot the frequency depen-
deuce of the input impedance. The usual way is the seperate
plotting of modulus and phase. Ronniger 1954 proposed locus
plots for the combined display of moduli and phases - which
has some advantages if one is used to the representation.
Fig. 30 shows a comparison between both plots . The example
of an actual measurement in a hydrodynamic elastic tube mo-
del is shown. The right upper part gives a simple explana-
tion for the construction of the locus plot from the modulus
(amplitude) and phase data. The input impedance of a non
branching tube with rather small damping shows a series of

268 T. Kenner
60° ----+---+
I. oo 1----1--f
t ·200
~ 0 ~--~--++--+-~-+-+--+--.H-~
+ -20°~,-~~--+~-~-+-;--+--r~-~
2,29Hz
0 ,_
Fig. 30 Input impedance of a homogeneous tube. Left:

phase and amplitude, right: locus plot
maxima and minima due to resonance. For a comparison, in fig.
31 the locus plots of the input impedance of the two wind-
kessel models from fig. 29 are shown. Highly characteristic
is the negative half circle which is passed through in clock-

im
wise direction with increasing
c
frequency. Fig. 32 shows a re-
cording of the input impedance of
the aorta of a dog during diffe-

om
d rent experimental conditions in
order to demonstrate the rather
flat high frequency part. The
values of the zero frequency
(R b ,in the fig. 32 indicated

a s
Fig. 31 Input impedance as Z ) are written in the insert
0
of windkessel and
manometer (Kenner et al.1968).
For a comparison, finally a typical locus plot of the high

frequency part of the input im-
pedance of a dog is shown in

)
-&
+60
+40 I
I
fig. 33. In this figure the
•20
o~------~7t~-------------- points of the absolute and dif-

-Control (Zo"296l
-2o
-40
---- Chamber (Zo"2 54)
ferential resistance value are
- - Gen anoxoa (Zo~3 64)
-60
t:.
indicated. The harmonics of the
liN I
001:
~~/'
t' ) .
t~, ---~
input impedance are arranged in
... .. - - -· ... , .. '

010 \
005
~...
~ - -·•·-·
a half circle which resembles
·-t- --r-----
the half circles of the models
-~----
5 '0 15 20 clsec
shown in fig. 31. - A correspon-
Fig. 32 Input impedance of ding locus plot including an in

ascending aorta of
a dog (Kenner et al. vivo recorded low frequency
1968)
part is shown in fig. 47 in section 4.2 One consequence of
the typical pattern of the high frequency part of the input
impedance is the fact, that the phas~ of flow always precedes

IM
the phase of the pressure. In
j
the early days of hemodynamic
modeling this fact was sometimes
called "Franks rule" in honour

RA mm Hg /ml/sok
Fig. 33 Input impedance of of O.Frank (1899) who was the
ascending aorta of
a dog (Kenner 1975) first to study pressure and
flow in windkessel models.
The input impedance of a small artery is shown in fig.
34 (Busse et al.l979). Pressure and flow was recorded in a
small muscular branch. The modulus of the impedance decrea-

270 T. Kenner
ses markedly with increasing
frequency while the phase angle

IZI
,.!~]k [~1
s
is negative and almost constant
.w/
in the frequency region up to
kJ~:JL
30 Hz. The pattern resembles to
• o+-'"'"""T,-T"IO-f-1111--il
''"• + - - - - - - - i the characteristic impedance
·&P. ....-.--~--
shown in fig. 26 , and is simi-
Fig. 34 Pressure and flow lar to a "windkessel pattern".

(left) and input im-
pedance of a small For the explanation and modeling
artery (right).From
Busse et al. (1979- Busse et al. have assumed taper-
Abstract Euromech
118) ing of small vessels. However,
the influence of the characteristic impedance as discussed
ab0ve in section 3.2 should also be considered. -In any case
it seems important to note the "windkessel- like" properties
of the small peripheral arteries which certainly is important
for modeling the arterial system.
3.4 Pressure and flow transmission
As the arterial pressure and flow pulses are transmitted
along the tube system, both pressure and flow waves show
characteristic changes. An example from a linear transmission
line model (Kenner 1979) is demonstrated in fig. 35. This
figure shows from above to below: the central arterial pres-
sure pulses,the arterial pulses along the system, the flow
pulses normalized with the local characteristic impedance.

Fig.35 Pressure and flow along transmission line

model. Description in the text
In this model (Kenner 1965, 1979) the main arterial line is
simulated by an inhomogen~ous linear transmission line. The
tapering of the line is simulated by 3 reflections at the
locations 11,23 and 61 within a total length of 100 units.
The chosen reflection factors are 0.2,0.5 and 0.3; the re-
flection factor at the end of the tube is assumed as 1.0
corresponding total reflection. The outflow from the system
is assumed by leakage along the tube (side branches).The lea-
kage generates a damping of 0.006 per unit. The total increa-
se of the characteristic impedance from central to periphery
is 8-fold In agreement with in vivo recordings we find the
following characteristics in fig. 35. The pressure amplitudes
increase from central to periphery. The contours of the pres-
sure pulses change in a very typical pattern, in that the
secondary (dicrotic) wave gets more marked. The flow ampli-
tude decreases from central to periphery (this is concealed
in fig.35 since the flow amplitudes are normalized by multi-
plication with the characteristic impedanc~. Towards t~peri-

272 T. Kenner
phery a marked negative wave appears in the flow tracing.The
reflection factor at each reflection site is defined as
k = (Z I - Z )/(Z I + Z ) (I)
n n+ n n+ n
In our example the index n = I,2,3,4,5 indicates the number
of the tube segment proximal to the reflection site. z5 cor-
responds to the peripheral resistance and is set infinite.
In the following we restrict ourselves to the pressure
transmission function (PTF) which is calculated by estimation
of the complex quotiebt (or, in other word~ the modulus and
the phase angle of the quotient of the harmonic components)
of two pressure pulses
(2)
If pressures are recorded in a reflection free transmission
line there will be no change of the amplitudes of any frequen-
cy. The phase shift entirely depends on the local wave velo-
city. - If two pressures are recorded in a tube with reflec-
tions the pattern will be more complicated because the phase
velocity as well as the amplitudes are influenced by periphe-
ral wave reflection. - In section 3.2 the pressure transmiss-
ion function in a homogeneous tube with a resistance at its
end was theoretically derived. In order to demonstrate the
most simple case we assume now a friction-less flow so that
the damping coefficient B = 0. Under this condition the
pressure transmission function can be written as
PI[P 2 = cos kL + j(Z/R)sin kL (3)

where L is the length of the tube and the transmission func-
tion is estimated between entrance and peripheral resistance.
The equation shows, that the basic shape of the function is
an ellipse. A recording in the same homogeneous elastic tube
model and under the same condition as was shown in fig. 30 is
displayed in fig. 36. Here again the two possible ways of
plotting are shown.
5 oo•
t oo• / ---
3oo• ~ I>"'
~2 2oo•
/ --
/
~
too• r--- ----- --

0
.vi'
1,2
1,0 ' "'- ~ -·--

I \ I
~
0,8 '' r\
\ I L~ I'-
0.6
O,t
\ J v
0.2
0 ,_ 2 3 5 6Hz
Fig. 36 Phase and modulus (left) and locus plot

(right) of the pressure transmission func-
tion P 1 /P 2 in a homogeneous tube (Wetterer
and Kenner 1968)
The locus plot shows the characteristic ellipse, somewhat
distorted by the viscoelastic damping. The relation between
the two axes of the ellipse corresponds to the quotient
F Z/R (4)
Therefore the pressure transmission function between~the en-
trance and the end of a vascular segment is a very sensible
indiuator for changes of the perip~eral resistance R. Vaso-

274 T. Kenner
dilatation (decrease of R) increases the ellipse to a circle
and 'vasoconstriction leads to a flattening of the ellipse.
It shou!~ be mentioned that tapering makes the ellipses
excentric as can be seen in all in vivo recordings. Further-
more, if the transmission function is measured between entran-
ce PI of an arterial segment and some point Px which is loca-
ted proximally to P 2 then the pattern of the transmission
function is much more complex than an ellipse. The pattern is
highly characteristic dependent on the relative location of
Px with respect to PI and P 2 , as was first pointed out by
Ronniger (I954) and Lindner and Ronniger (I955). Fig.37 shows
3 oo•
I'
2oo• /
~
~
,,Xt Ioo• - -- /""'_ /

""'
__ /
i/ v
0
------ --- r--- -·-- ·-
--J ---\' --..,...,..

~ ""
3 -----·- ---
2 -----· -- -2
I ""o ..<>' ~
0 ,_ I 2 3 ~ 5 6Hz
Fig. 3 7 Pressure transmission function P. 1 /P where

P was measured in the middle betwe~n PI and
d~e end of the tube (P 2 ). (Wetterer and
Kenner 1968)
a characteristic example from the same model as in figs. 30
and 35. In the example of fig. 37 the pressure transmission

function was measured between the entrance of the tube and
the middle of its length. The following equation explains the
way to analyze the distortion of the locus plot (Lindner and
Ronniger 1955). If one writes
P IP ( 5)
I X
it is clear that both functions P 1 /P 2 and Px/P 2 are ellipses
since both describe the transmission from an entrance to the
end of a segment. The division of the two functions leads to
the rather typical pattern shown in fig.37 and in the follo-
wing figures. The location of the loop p£ the locus plot is
characteristic for the location of Px with respect to P 1 and
Fig. 38 (Kenner 1975, MODEL CAT

7cm
1979) shows left the esti-
: ~!~
mated pressure transmissi-
;y.",
on function in the trans-
mission line model descri-
bed in the beginning of
this section (Kenner,l975). 1.0
The estimation was performed 28cm

2
from the entrance to 1/3 rd,
~
from the entrance to 2/3 rd 1.0
and from the entrance to
the end of the model shown
at the bottom of the figure.
The result is compared (right)

276 T. Kenner
with measurements performed in a cat. In other animals and
in man simmlar result.s have been obtained (Wetterer and
Kenner 1968). In the cat in which the transmission functions
shown in fig. 38 were measured, we have examined the effect
of vasoconstriction {NOR = noradrenalin ) and vasodilatation
(ACCH = acetylcholin ). The result is shown in fig. 39.
CONTROL
P:2f0/t50
'
Fig. 39 Pressure transmission ascending aorta to
femoral artery in a cat {Kenner 1979)
As mentioned above, vasodilatation is characterized by chan-
ging the elliptic shape of the transmission function to a mo-
re circular shape. Vasoconstriction narrows the ellipse. Due
to a reduction of damping, small "reflection-loops" may beco-
me apparent which indicate that the peripheral recording is
not exactly located at the end of the segment.
3.5 Pressure amplitude and peripheral pressure-flow relation
There are two main factors leading to a peripheral in-
crease of the pressure amplitude: tapering and peripheral
reflection, The effect of tapering is due to the increase of
the characteristic impedance along the aorta (see fig.J2).In
a frictionless and damping-free tube we would expect an asymp-

totic value of the peripheral amplification of the pres-
sure amplitude as follows (Kenner, 1972):
JJpz/ Llpl VZz/zl' (I)
The second component of the pressure amplification is the pe-
ripheral reflection. At the site of the reflection we can ex-
pect (2)
in a homogeneous tube; if .dp 1 is the amplitude of the incom-
ing pressure wave, and k is the local reflection factor, then
Llp 2 is the actual amplitude recorded.
Kenner and van Zwieten (1982) have discussed the impli-
cations of changes of the distribution of the pressure ampli-
tude along the aorta in relation to the peripheral reflection.
Fig. 40 shows the distribution of pressure amplitudes along
the aorta of 8 cats (A ascending aorta,M mesenteric artery,
B iliac bifurcation, F femoral artery). The total amplificat-
ion is more than 2-fold which is in agreement with combined
effects of tapering and reflection. We observed that vaso-
L:Jp mm Hg dilatation, produced by
60 different drugs may lead
to different changes with

50
respect to the pressure
40 amplification. We assume
A M 8 F that a decrease of the
Fig. 40 Distribution of pressure amplitudes along the aorta.

Details see text (Kenner and van Zwieten 1982)
peripheral resistance decreases reflection , and thus also,

278 T. Kenner
amplification. This pattern can be actually found with acetyl-
choline. Other vasodilating drugs like e.g. hydralazin inter-
restingly lead to an increase of the pressure amplification.
This result shows that we are still far away from fully under•
standing the physics of the peripheral pressure-flow relation
in small arteries and arterioles.
The peripheral reflection factor k is defined as
k = (Rd - Z)/(Rd +Z) (3)
where Rd is the differential value of the resistance. From
our observation of the possibility of discordant behaviour of
mean pressure and pressure amplitude (amplification) we may
consider that absolute and differential value of the periphe-
ral resistance must not always change in the same direction.
In a highly schematized drawing we have tried to express this
in fig. 41. In both

Q Q
examples shown in the
'
/Y
'\
/~
figure the absolute va-
lue of the resistance

/
is assumed to decrease I
....
,.. .., p //
/
p
(shift to the left). In
the left diagram the Fig. 41 Variations of the peri-

pheral pressure-flow relation
slope Details see text (Kenner and van
Zwieten 1982)
of the pressure-flow re-
lation changes into the same direction. In the right diagram
the slope changes into the opposite direction. It will be

necessary, of course , to confirm this interpretation by ano-
ther method.
A further interesting aspect concerning the peripheral
pressure-flow relation is shown in fig. 42 which shows a de~
~il from fig.35: pressure and flow in the periphery of the
tube model (Kenner 1979). The
outflow from this model is ass-
umed to take place through small
side branches (leakage). There-
flection factor at the very end
is one. So outflow at the end
is zero. A few units of the model
proximally to the end a flow Rig. 42 Pressure (top)

and flow (bottom) in the
pattern can be observed which periphery of the tube mo-
del (Kenner 1979)
closely resembles the time de-
rivative of the pressure pulse. The record is also very simi-
lar to pressure and flow in a small artery (fig. 34). This
phenomenon leads again from a slightly different viewpoint to
the fact that the periphery behaves like a windkessel. In
other words, the pressure and flow pattern in figs.34 and 42
can, formally,be interpreted by the windkessel equation:
q p/R + C dp/dt (4)
which implies in the case of k = I and R oO:
q c dp/dt (5)
The impedance, therefore, is

280 T. Kenner
P/Q = J/j we (6)
and has a modulus which decreases with frequency and a nega~
tive phase angle for all frequencies of - 90°, very similar
to the pattern shown in fig, 34.
3.6 Wave velocity
The speed with which a signal (pulse) is transmitted
through an arter¥ depends on the following factors:
I) Elastic modulus of the vessel wall
2) Radius of the vessel
3) Wall thickness
4) Damping and viscosity
By influencing one or more of these factors the wave speed
shows correlations with the following magnitudes or functio-
nal changes: pressure inside the vessel (better: transmural
pressure difference). Contractile status of the vascular
smooth muscles. Age and pathologic changes of vessel wall
geometry and composition.
The wave velocity can be described by "Webers equation"
c = {<dp/dr)(r/2e >' (I)
In the case of small extensions the pressure radius relati-
onship is described by Youngs modulus E. Insertion into the
above equation yields "Moens-Korteweg" equation:
<2)
rn elastic tubes with large extensibility we have to use
Wave Velocity 281
an incremental modulus Et as discussed in seation 2.6. In
the following equation a force-related incremental modulus
is used (Wetterer and Kenner 1968, Kenner 1967):
c
v e><
\ I< I I 2 Et h
r(l-r)
2 - p> (3)
In these two equations the effect of friction and of viscous
energy losses is neglected.
In small arteris, arterioles and capillaries the wave
velocity is markedly influenced by fluid viscosity, as dis-
cussed in section 3.2. There the wave velocity was found to
be c (4)
The wave velocity in large arteries has the same order
of magnitude in animals of different size. Towards the peri-
phery as the content of muscle fibers in smaller arteries in-
creases, the wave velocity increases as long as the effect
of the fluid viscosity is relatively small. In the dog the
following values of the wave velocity can be observed, accor-
ding to Caro et al. 1978): central aorta 400- 600 cm/s ,
abdominal aorta 600 - 750 cm/s, femoral artery 800 - 1000
cm/s. All these velocities may be higher if the arterial pres-
sure increases (see below). In small vessels the influence
of the viscous resistance is marked and decreases the wave-
velocity. In capillaries values as low as 10 cm/s are repor-
ted (Caro et al. 1978).
Vasoconstriction increases the wave velocity by decrea-
sing the radius. However, the question whether an increase

282 T. Kenner
of smooth muscle tone ceteris paribus would increase or de-
crease the modulus of elasticity of the vessel wall is still
lacking a general answer. Probably in some cases the modulus
does not measurably change, in some other vessels a decrease
of the modulus is actually possible.
The elastic modulus of the vessel walls in any case in-
creases if the arterial pressure rises. One example by Wetterer
and Pieper (1953) of the relation between arterial pressure
and wave velocity is shown in fig. 43. There is a general
trend of the wave velocity in the aorta to increase with ine
creasing pressure. However, during Sympatol effect (Sy) the
wave velocity is shifted towards lower values indicating a
decrease of elastic modulus as a response to the sympathomi~
j
1000 or---,---_,~sro.---_,---,~~~oo~~---r---~~~~m~m~Hg
I •••J
metic drug.
I
.. . . ••
1 1
em/sec
~.~+t.t
/ • ....•"i-:••..-::cr---t
t
800t----r-------t--- ' •
I ·+··i~. "i
600 r-- ·-- - 00
_i __. ·+~t:~<.j-1"!:.-;+----t----t----t------i
I
+
o
~·~~
I ++ I
~
ttl ~ ;:, ,
400 id?J J
dlo I I I •R
~or I
0 so 100 ··1~so:;----'----::z~oo;;-;·1"'o·r'
p dyn I cm 2
Fig. 43 Relation between arterial pressure and wave velocity
in the aorta of a dog. R control, Vk vagus cooling
Sy Sympatol injection, VkSy same during vagus cooling
(Wetterer and Pieper 1953)
If in an artery the transmission of a pulse is measured se-
veral additional factors influence the observed result. Usu-
ally the transmission is measured at the pulse front at the

Wave Velocity 283
mean arterial pressure value or at 1/5 th of the pulse height.
Using this method the influence of reflections seems minimal
although it cannot be excluded completely.
A marked influence of reflected waves can be observed
if the phase velocities of harmonic frequency components are
measu~ed from the pressure- or flow- transmission function.
The equation for the pressure transmission function (eq. 4
in section 3.4) shows that the phase transmission time for
the first harmonic frequency usually is much smaller than the
phase transmission time for higher harmonics. A typical exam-
ple from a cat is shown in fig. 44.(Kenner and van Zwieten
1982). The figure shows im

2
the locus plot of the ,
.,.--
pressure transmission
3....-"'
4
I
I
function before (full
I 1.0
line) and after the in- \
\
I
jection of the vaso- A~
dilating drug hydrala- ' ..... ..... 5
zin (dashed line), Fig. 44 Pressure transmission function

before and after vasodilatation
Two observations (Kenner and van Zwieten 1982)
are pertinent: The phase angle of the first harmonic is much
smaller than 1/2 of the phase angle of the second harmonic.
Thus, the phase velocity of the second harmonic is much smal-
ler than the phase velocity of the first harmonic. The phase
velocity is defined by the following equation:

284 T. Kenner
( W ~~ ) X (5)
w is the frequency and t/J is the corresponding phase angle. w1

is the frequency of the first harmonic of a pulse wave and n
is the order number of the harmonic frequency. +n is the phase
angle of the n-th harmonic frequency. In general the phase
velocities of lower harmonics are higher than those of higher
harmonics, due to the influence of reflections. During vaso~
dilatation and decrease of the blood pressure both the trans-
mission times and the phase angles increase - and the corres-
ponding velocities decrease.
A worthwhile method for measuring transmission times is
the calculation of the cross correlation function (CCF) of
two trains of pulses (a central and a peripheral one). The
time lag of the first maximum of the CCF corresponds to
the weighted transmission time of the pulses (Kenner 1979).
As the CCF of harmonic. functions equals the sum of the CCFs
of each harmonic frequency, the weighted transmission time
rCCF is close to the phase transmission time of the first
harmonic which is always the largest component of the arteri-
al pulse. If the CCF is calculated after differentiation of
the pulses with respect to time then the transmission time
TccFD corresponds to a weighted transmission time of higher
harmonics. Fig. 45 shows the dependence of T.F ,rc and

ront CF
rCCFD on arterial blood pressure in a rabbit and a cat. The
blood pressure variation was brought about by infusion of

Wave Velocity 285
noradrenalin. In agreement with the pattern demonstrated in
fig. 44 the transmission times for lower harmonics depart
from those of higher harmonics the more the higher the blood
pressure and the more pronounced the peripheral vasoconstric-
tion,
msec
KKF
100 200rrrnHg 200
Fig. 45 Transmission times measured along the aorta of a

cat (left) and a rabbit (right) using three different methods:
L propagation measured at the front of the pulse wave.KKF:
using cross correlation. KKFD: same after differentiation.
It can be proposed that the continuous recording of trans-
mission times or wave velocities can be used for the assess-
ment of arterial blood pressure. - In order to summarize the
definitions of different velocities the following equations
are presented: The phase velocity without reflections
c = W/k (6)
The group velocity in absence of reflections
c dW /dk (7)
g
Due to the rather unique skewed distribution of harmonics in
pulse waves, the group velocity of arterial pulses does not
play a large role in practical considerations. The phase ve-
locity ( 8)
286 T. Kenner
as mentioned above. Finally, the apparent phase velocity at
a certain location is defined as
c wdx/df (9)
app
The signal velocity is influenced by the flow velocity in a
tube (v) and, as Anliker et al.(1968) proved, corresponds to

+ (I 0)
dx/dt c - v
The influence of age and blood pressure on the aortic
wave velocity was examined by Schimmler (1965). Fig.46 shows
a slightly modified diagram
(Kenner 1972) - the dashed li- c(ml c)

nes which, in the original
16 200
are distorted due to selecti- /
,/
on by death of the patients,

, /160 p
were straightened. The aortic , 1.0
12
wave velocity in humans is shown 120
100
as a function of age and blood
pressure. The wave velocity was
measured from A.carotis to the

8
~-
femoral artery in 2500 persons.
4
--+--+-t--+--+-t--Aa• Years
25 35 45 55 65 75
Fig. 46 Aortic wave velocity
as a function of age and mean
arterial pressure. Data from
2500 persons. After Schimmler
1965.
Low Frequency Impedance 287
4. SMALL ARTERIES AND THE CONTROL OF FLOW AND PRESSURE

4.1 Introduction
We have in part 3 introduced the concept of frequency
dynamics. The low frequency part - below the heart rate - re-
presents effects of slow reactions of the living vessel wall
to changes of pressure and flow. The more pronounced and in~
teresting reactions in this frequency region are active and
related to metabolism. The reactions are initiated by autore-
gulation (local effects to changes in flow and pressure) or
by neural control which, in most cases , is centrally coordi-
nated. The low frequency reactions also comprise passive ef-
fects as slow stre~ relaxati~n and creep.
In the small vessels all the influences which possibly
can change pressure and flow come together as indicated in
fig. 47 after Folkow and Neil 1971. Here also the interacti~
ons take place between extrinsic influences like drugs and
neural and autoregulatory feedback mechanisms.
Fig. 47 Summary of the regulatory influences in the region

of the small arteries.
288 T.Kenn~
4.2 The low frequency input impedance
All the discussions in the preceding sections were some-
haw.related to the question how the properties of small resis-
tance vessels influence pressure and flow in the circulation.
The pressure-flow relation in the resistance vessels was dis-
cussed as a nonlinear function (2.8) and as related to the
characteristic impedance of small arteries (3.2).Furthermore,
we have attempted to distinguish absolute and differential va-
lue of the resistance (3.3) and the upstream influence of the
resistance due to reflections (3.4 to 3.6). - The method by
Wetterer and Pieper (1953) to determine the pressure-flow re-
lation of the peripheral resistance used pressure and volume
variations artificially induced by a pump with a frequency
of around 0.2 Hz. - The question now of course arises whether
this particular frequency shows a particular and different
behaviour than other frequencies in the region below the
heart rate (Kenner 1971,Attinger 1973,Basar 1981,1982, Sipke-
ma et al. 1982).
The first who had examined low frequency properties of
the circulation in an extended range of frequencies was M.G.
Taylor (1966) who used the most advanced method of random
excitation and spectral analysis. Kenner (1971) performed
experiments using sinusoidal variation of the blood volume
of anesthetized dogs to determine low frequency input impe-
dance ( the complex pressure-flow relation) of the aorta.

Fig.48 shows a slightly schematized comparison of the aortic
input impedance of a dog wmth the low frequency part display-
ing regulatory phenomena (upper part) and without, i.e. after
paralysis of vascular reactivity.
'"'j
R..,
3 RE
'"'j
RA mmHg/ml/stk
Fig. 48 Input impedance of the aorta of a dog.

Upper part including regulatory pheno-
mena. Kenner 1975)
With respect to the small vessel reactivity, a futher impor-
tantstep was the application of sinusoidal perfusion in order
to examine the low frequency impedance of different arterial
beds which was performed by Basar and coworkers. Basar rec,
cently summarized his results in a book(Basar 1981 ,1982).
Rubenstein et al. (1973) used pseudorandom binary sequences
(PRBS) to test and analyze local hemodynamic control.
Following these studies Kenner and Ono (1971) and Kenner
and Bergmann (1975) have examined the low frequency input
impedance of different arterial beds. Fig. 49 shows the lo-
cus plot of the input impedance of the femoral artery of a

290 T. Kenner
dog. The circles show the

lm
high frequency impedance
part, measured from pressure
and flow pulses. The filled
circles show the low frequen-
cy part measured with intro-
ducing artificial sinusoidal-
ly varying perfusion of the
Fig. 49 Input impedance of a r t e r y w i t h b 1 o o d . The square

the femoral artery of a dog
(Kenner and Bergmann 1975) indicates the absolute value
of the zero frequency resistance.
Fig. 50 shows the locus plot of the low frequency im-
pedance of the renal artery of a dog at three different per-
fusion pressures. Circles: 80

lm
mm Hg, filled circles: 150
mm Hg , triangles: 200 mm Hg.
This figure indicates that
the reactions are pressure-
dependent and, in a wider
range of pressure, nonlinear.
Trying to explain in
Fig. 50 Input impedance of short these impedance pattern

the renal artery of a dog.
Low frequency part at three we have to assume that the
different mean pressures
(Kenner and Bergmann 1975) interaction between pressure
and flow represents regulatory phenomena: in local peri-

pheral beds autoregulation. In the general circulation, re-
presented by the aortic input impedance, in addition reflex
reactions, particularly baroreceptor reflexes, play an impor-
tant role.
Autoregulation in the small resistance vessels has the
following physiological correlates. !)Metabolic effects: if
the local £low decreases the supply with oxygen and the trans-
port of metabolic products is reduced. These changes will,
directly or indirectly dilate vessels in order to increase
flow and vice versa. 2) Myogenic autoregulation: if the ves-
sel wall is quickly stretched, a contractile response may be
observed (Bayliss effect). Both these effects will be summa-
rized below as "contractile response C " to increase of flow.

q
3)~Autoregulation of arterial pressure. Similar to the re-
action of the baroreceptor sytstem therE! seems to be a local
reactive "vasodilatation response D " to pressure increase

p
(Kenner and Ono 1971).
We describe the resistance R of the small vessels as
a function of pressure and flow.
p q R (I)
and R = R(p,q) (2)
In order to derive a linear approximation we use the follo-
wing procedure which recently has also been used by Hatakey-
ama (1982). Differentiation of the two equations yields
dp = q dR + R dq (3)
292 T. Kenner
and dR = ( ~RI~p)dp + ( "bRI"bq)dq (4)
and after insertion

+ ( q I R) ( "'bR I ~q)
~ R (5)
dq
(pI R) ( ?JR I ~p )
The two factors in the right side can be dnterpreted as
gains of the resistance control system; The following two
functions are defined: the gain of flow control
cq (qiR) ( ~Ria-q) (6)
As explained above C is supposed to represent contractile

q
responses to flow increase. The gain of the pressure control
D =- (piR)( ~RI~p) (7)

p
The negative sign takes into account that an increase of
the pressure dilates the vessel and thus decreases the re-
sistance. The two effects C and D are reciprocal as far

q p
as the influence on the resistance is concerned (Kenner and
Ono 1971). A more general formulat~on including dilating
effects of flow and constricting effects of pressure was
given by Kenner (1971). For practical applications the way
shown here seems sufficient for complete description of phe-
nomena.
We now assume that the gains can be expressed as fre-
quency dependent functions and chose as simplest possibility
first order transfer functions (in Laplace formulation):
Cl(1 + ST ) (8)
q
and Dl(1 + ST ) (9)
p
Here we write the Laplace frequency operator s instead of
Low Frequency Imped:mce 293
the Fourier operator j in order to express the necessity
to calculate solutions in the time domain. Using these func-
tions, the low frequency impedance can be written as
dp
+ C/(1 + s 't )
q
R (10)
dq
+ D/(1 + sr )
p
C and D are dimensionless constant values of the gains. Tq
and T are time constants. R is the (mean) zero frequency

p
value of the impedance. (R actually contains information ab-
out the values Rabs and Rd and about the high frequency part
of the impedance. Fig. 51 shows a comparison of calculated
locus plots (left) and locus plots measured in different
arterial beds of a dog (Kenner and Ono 1971).
Rilin/flio A) 1. Flj mes

Tjl
j
0.001
0005
00' 001
6) 001 fern
0005
j
C) 1 .
ren
'[I
Fig. 51 Low frequency impedance. Left: calculated. Right:

measured. mes mesenteric artery, fern femoral arte-
ry, ren renal artery. Description see text.
(Kenner and Ono 1971)
294 T. Kenner
A)Left: Pure autoregulation of flow, Dp = O, Cq 3/(1 + Ss).
Right: low frequency impedance of the mesenteric artery.
B) Left: Autoregulation of pressure, D = 3/(1 + 100s),C =0.

p q
Right: Low frequency impedance of the femoral artery.
C) Left: Combined autoregulation of pressureD = 3/(1+1000s)

p
and flow C = 3/(1 + Ss). Right: low frequency impedance of
q
the renal artery.
It can be shown by partial fraction expansion or by in-
sertion that the equation for the low frequency impedance
(eq.10) can also be written in the following form:
~
dq
= R ( 1 + A/(1 + s "C )
q
- B/(1 + s "t'.))
p
( 1 I)
where
r•p = t' I <1 + D)

p
( 1 2)
and A and B are constants related to the gain values C and D
and to the time constants (Kenner and Bergmann 1972). The
equation can be solved by inverse Laplace transformation
into the time domain. The following equation de~ibes the
pressure response to a step flow increase dq

step
-t/ 't -t/ t!
( 1 J) dp(t) dqstepR(t+A(1-e q) - B(1-e P~
Examples of the shapes of calculated responses to step chan-
ges to flow (di,upper part) and pressure (dp,lower part) is
shown in fi9. 52 (Kenner and Ono 1971).
From these results we can conclude that reactions of
vessels to changes in pressure and flow can be described by
a combination of pressure and flow control gains. The fre-

Low Frequency Impedan<;e 295
quency region of these metabolic and myogenic control reac-
tions extends down from t~heart rate towards ab~ut 0.001 Hz
or even lowe~ (there may be a smooth transition towards cir-
cadian rhythms or even longer periods). We did not discuss
here in detail the region of nervous mechanisms as barorecep-
tor reflexes which are most pronounced in the frequency re-
gion between 0.01 and I Hz.
di
~[
di
I~--
Fig. 52 Calculated responses
using eq. 13. Upper part pres-
sure response (dp) to a flow
(di) step. Lower part: flow
resonse to a pressure step.
Kenner and Ono(I971) dp
~[
0 3 min
4.3 Humoral autoregulation
Flow in arteries carries information and humoral signals
especially to the small vessels. For a substance with very
slow decay rate we can expect the concentration in the blood
to be rather independent of flow. For a substance whose de-
cay rate is fast compared with the flow rate the concentra-
tion on the site of action becomes dependent on flow.Kenner
and Ono (1972) have described experiments to demonstrate a
possible mode of autoregulation based on these considerati-

296 T. Kenner
ons. If a vasoactive substance is secreted intm an artery at
a constant rate, the concentration X at the site of action
is inversely proportional to the flow. We assume that the re-
sistance some distance downstream from the site of pcoducti-
on of the substance is influenced by its concentration X:
R(X) = p/q (I )
Differentiation leads to
2
(aR/cX)dX dp/q - p dq/q (2)
and, if the concentration of the substance in the small arte-
ries varies inversely with the flow we can write
dX = - k dq (3)
where k is a constant proportional to the rate of production
of the substance. Therefore
dp/dq R(l - k (4R/~X)(q/R)) (4)
If we describe the second term on the right side as the con-
trol gain of the vasoactive substance and chose a first order
transfer function- positive C for vasoconstriction, negative

X
D for vasodilatation , e.g.

X
cX = c ( I + s '["" ) (5)
we find by inverse Laplace transformation the pressure res-
ponse to a flow step

- t /T:x
(6) dp(t) = R dqstep(l - C (1 - e ))
This equation indicates that the pressure in the arterial
bed increases immediately after the flow step and then de-
clines back to its former value (during constant flow per-

fusion) because of the reductwon of the resistance by the
dilution of the vasoactive substance in the increased volume
flow.
A similar reaction may be found for the effect of a
pressure step on local flow if the vasoactive substance is
..
vasodilating. Then the solution in the time domain is
-t/""Cx
dq(t) = (dp /R) (I D (I - e )) (7)
step + D
The equation describes the following phenomenon: if, under
the presence of a continuous secretion of a vasodilating sub-
stance the perfusion pressure is increased, the flow follows
with a transient increase and then exponentially declines
back to the control value. If the vasodilating gain is close
to one , there is a perfect autoregulation of flow- i.e. the
flow is kept constant under these conditions.
Flow regulation of this kind may be possible in vascu-
lar areas where vasoactive substances are secreted, like
renin in the afferent arterioles in the kidney, or where
substances are generated by local effects, like certain
prostaglandines during the process of thrombosis.
4.4 Instability of flow and pressure in small arteries
Using the model described in the preceding sections
Kenner and Ono (1972) have examined the interaction of flow
and pressure in the course of drug induced reactions. On the
one hand we were able to quantify the interaction of the

298 T. Kenner
of the effect of vasoactive drugs and the baroreceptor re-
flex. On the other hand we observed oscillations of pressure
and flow as shown in fig. 53, particularly during the infusi-
on of the vasodilating substance acetylcholin in dogs.
The tracings show from top to
bottom: femoral venous oxygen
saturation, femoral arterial
flow, femoral venous flow,
flow in the common carotid
artery, systemic arterial
pressure. The period of intra-
venous infusion of acetylcho-
lin (ACh) is indicated at
1001'9/rnnACh the bottom. During infusion
Fig. 53 Autooscillations of acetylcholin slow oscill-

during acetylcholin- infusion
in a dog (Kenner and Ono 1972) ations of the femoral flow
and, at the same time higher frequent waves (Mayer waves)
can be observed in the arterial pressure tracing. All these
observations indicate the generation of instability of flow
and pressure by the infusion of the drug.
An interpretation was attempted by the application of
the following equation of the low frequency impedance:
dpI dq = R( I + c 1(1 + s t'c ) - DI (I + s Td ) ) ( I)
with C and L gain and time constant of a vasoconstricting

c
effect (neural control or autoregulation). D and Ld are gain
and time constant of the vasodilating influence of the in-

fused sub~tance. - I f the magnitudesof both gains C and D
are equal and if the time constant of the dilating mechani-
sm is smaller than the time constant of the constricting me-
chani sm, then instability may occur at the frequency
wi,.= yi /rc rd ( 2)
At this frequency small random pressure oscillations may
start "infinite", i.e. very large flow oscillations because
the low frequency input impedance (eq. I) tends towards zero
at this frequency. - We therefore assumed, that acetylcholin
may trigger a change of the time constant rd which then
starts the oscillations as shown in fig.53.
4.5 Autooscillations and vasomotion
As discussed by Basar (1981) it is certainly a mistake
to assume that all oscillations in the cirulation may be
explained by instabilities of local or general (neural) con-
trol. The smooth muscles of the vascular tunica media of
most small arteries has an immensely dense innervation which
generates an adrenergic vasoconstrictor tonus. Tonus is a
very fuzzy term which indicates contractile status depending
on the conditions: isometric stress if shortening is not
possible, or, isotonic shortening if constant stress is gi-
ven. In any case sympathetic activity under normal resting
conditions accounts for about I/2 of the vascular smooth mus-
cle tone as can be judged from the decrease of vascular re-
sistance after paralysis of the sympathetic nerves - resis-

300 T. Kenner
tance falls to about 1/2 after such a procedure (by the way
another fuzzy expression of the meaning of the term "tonus"
of vascular smooth muscle).
If, during muscular exercise the resistance of small
vessels in the muscles decreases to 1/10 or even less of its
resting value, vasodilating metabolites and autoregulation
play a major role besides the decrease of the sympathetic
innervation. At the same time about 65 to 75% of the existi-
ng capillaries open up which have been closed or collapsed
during rest (Folkow and Neil 1971).
The smooth muscles of probably all vessels have the
capability to produce spontaneous excitations and contracti-
ons. In small vessels and arterioles and in precapillary
sphincters the "vasomotion" may periodically occlude the
lumen completely. The frequency of these contractions is re-
ported between 10 and 60 per min (Noordergraaf 1978).
Since parallel vascular areas exhibit a synchronous va-
somotion, autoregulatory phenomena may be described and ex-
plained as a statistical process. The frequency of the oscil-
lations and the duration of closure varies under different
local conditions in such a way as to provide autoregulatory
control as described in section 4.2.
In larger vessels the autooscillations are related to
contractile responses. Fig. 54 shows an example of a myogen-
ic response of an isolated small vessel (rat tail artery)

Downstream Effects 301
to a change of the internal pressure (Busse et al. 1982).
The small artery exhibits
spontaneous activity. As the

r.455] ,. • ;
P50442~
.
l~ml ~ , 1
pressure is increased the fre-
lmm Hgl I JO
J ..
~
l ' ,
'
0 lI
quency of the oscillations
r. 470 rises. At the same time the
hJml 1 •
P
lmmHgl
1o]4 J rrVVVt-~ radius of the vessel under-
goes a transient reaction,

JO ·-Ho-t+-Hij:±±±±
the shape of which correspon-
Fig. 54 Myogenic response of ds closely to the double ex-

a small artery to a
pressure step. ponential function mentioned
(Busse et al. 1982)
in section 3.7 and shown in
fig. 52 - it is obvious that the reaction of the radius fol-
low a similar contour as the reaction of flow to a pressure
step.
5. THE DOWNSTREAM EFFECTS OF SMALL ARTERIES
5.1 Introduction
We do not really know why in all mammals the arterial
blood pressure has the same order of magnitude (Kenner 1979).
We know that the pressure decay from its arterial mean va-
lue of about 100 mm Hg (13.3 kPa) to the capillary pressure
which is in the order of IS to 30 mm Hg is due to the flow
resistance of small arteries, arterioles and precapillary
sphincters. The average value of the capillary pressure

302 T. Kenner
is found to correlate with the value of the colloid osmotic
pressure of the plasma. In mammals the relation between arte-
rial pressure and colloid osmotic pre sure is 5/1.
Since all the important exchange processes of fluids
and metabolites take plaee in capillaries, the functional
aim of the small vessels is directed towards controlling ca-
pillary flow and pressure, and thus also, transcapillary
fluid exchange.
5.2 Relation between resistance vessels and microcirculation
The small arteries, arterioles and precapillary sphioc-
ters act together as precapillary "arterial" resistance K .

a
We have already discussed the main influences which act upon
the state of contraction of the smooth muscles in the walls
of these vessels. The resistance vessels open into the net-
work of capillaries. These exchange vessels are tubes formed
by a single layer of endothelial cells and a basement membra-
ne, and have no capability of contraction. The blood from the
capillaries flows in~the venoles and the small veins, ves-
sels with some smooth muscles in their tunica media so that
a certain variability "f this postcapillary "venous" resis-
tance R can be anticipated. R is smaller than R and does

v v a
not contribute much to the total peripheral resistance.How-
ever, both R and R are important since the ratio of these

a v
two resistances determines the magnitude of the hydrostatic
pressure p in the capillaries. The latter can be calculated

c
if the arterial pressure p and the venous pressure p are
a v
given using the equation (see Folkow and Neil 1971)
+
( I)
+ R /R
v a
Besides diffusion and active transport by pinocytosis there
is a fluid flux through the capillary wall which is related
to the gradient of hydrostatic pressures in the capillary p

c
and in the surrounding tissue pt' and the corresponding g~a
dient of colloid osmoti~ 1 pressures in the inver~direction.
Vt is the colloid osmotic pressure in the tissue spaces, Tpl
is the colloid osmotic pressure in the plasma. k is the fil-
tration coefficient, S is the filtration surface area of the
capillary wall, ~ is the socall~reflection coefficient of
the capillary wall with respect to a certain solute. Molecu-
les which cannot pass the pores of the wall are reflected
with a factor of 6 = I. Under this condition the osmotic
pressure gradient can exert its maximum effect.
The flux Jf through the capillary walls into the tissue
space can be calculated according to the classical socalled
Starling hypotherls (Lee and Kenner 1982):
Jf=kS(p -p- <S(lT- lT)) (2)

c t pl t
Whenever the sum of these terms on the right side is positi-
ve then filtration occurs and vice versa. It can be assumed
that under steady state condition mainly filtration occurs
in an amount which corresponds to about 0.1% of the plasma

304 T. Kenner
inflow into the capillaries. Most of the filtered fluid is
recently assumed to enter the lymph vessels (lntaglietta 1977).
However, it should be mentioned that the numbers about
the absolute and relative values of the transcapillary flux
vary from tissue re tissue. The classical viewpoint according
to the Starling hypothesis assumes that filtration occurs
at the arterial end of the capillary and that reabsorption
is found at the venous end of the capillary and at the venole.
Thus, a "paracapillary. flow" can be assumed which supposedly
amounts to about 1% of the plasma inflow to the capillary.
The difference between filtered and reabsorbed fluid enters
the lymph. - In non steady state, e.g. during changes of the
capillary pressure or during transition of osmotically hyper-
tonic fluids through the
cap.;·'ary net, there is no

Venole
doubt at all that reabsorpti- . -- -- ....
on may occur. A schematic
drawing which explains the
phenomena in the microcircu- Arteriole

lation is shown in fig. 55 Fig. 55 Fluid exchange in
the microcirculation
5.3 Hematocrit in small arterks, arterioles and capillaries
One of the surprising observations about the properties
of the microcirculation is the extremely low hematocrit.Fig.
56 from Lipowsky et al. (1980) shows the hematocrit as a

Hmocro / Hsys
VENOUS
function of the vessel diame-
•
AITfiiAL
• •• ter: the smaller the vessel,
the smaller the value of the
hematocrit in comparison to
the "feed" hematocrit in the
large vessels (Klitzman and
Duling 1979, Lipowsky et al.
1980). Other authors have ma-
de similar observations as
10 eo 'IO •o lO Jo •o 1 to 10 10 •o so 6o 10
summarized by Lee and Kenner
VESSEL DIAMETER tllftll
Fig. 56 Microvessel hematocrit 1982. - Thus, the microvessel
as a function of the vessel di-
ameter. Lipowsky et al. 1980 hematocrit may be as low as
3% ,given a normal large vessel hematocrit of 45%.
The described phenomenon can most clearly be seen in
resting muscle. During stimulation and activation of the
muscle, the flow as well as the hematocrit increased in the
capillar~es to values close to the large vessel"feed"hemato-
crit. The observation that increase of flow through a vascu-
lar area simultaneously leads to an increase of the micro-
vascular hematocrit seems very important for the function of
the microcirculation.
Four, more or less interconnected, reasons may be cited
for the interpretation of the effect (Fung 1981, Lee and
Kenner 1982) 1) Shunt flow of blood with high hematocrmt
through some thoroughfare channels other then capillaries.
2) Reduction of capillary hematocrit due to the entry con-

306 T. Kenner
dition at branching sites. As shown in fig. 57 from Fung (1981)
the erythrocytes tend to move into the faster channel at bran-
chings. In other words, the more flow. the more erythrocytes
move into a small vessel.
3) Decrease of hematcilcrit
-o at the wall of a larger tu-
0 be where the capillary

0
0 siphons off the blood leads
to an effect which is cal-
led plasma skimming. It

SHEAR
DISTRIBUTION
can be observed that ery-
)~ RESULTANT
throcytes tend to move away
'\
from the vessel walls, so
Fig. 57 The branch with the fas- that acell free plasma
ter stream gets the red blood
cells. Fung (1981) layer is generated.
4) Connected with this observation is the Fahraeus-Lindqvist
effect. The red cells in the core stream of the small vessels
move faster than the plasma along the wall.
For all these reasons, the hematocrit in the small ves-
sels is reduced. At the same time also the viscosity m£ the
blood in the capillaries is low. However, as Lee and Kenner
(1982) have reported, the explanation of the low hematocrit,
including the search for possible artifacts, is still unsa-
tisfactory.
5.4 Transport through the microcirculation
If an indicator solution is injected into some vessel
the solution is dispersed as it moves along the vessel, I) by
convection,particularly by the behaviour of the velocity pro-
file of the fluid; 2) by diffusion and 3) by splitting of the
stream through branchings of the vessels. The effects of the
profile development and the in-
fluence of the side branch on
the distribution of an indicator
is schematically shown in fig.
58 from Lee (1980). He explained
Fig. 58 Development of an that different parts of the pro-

indicator profile in a tube
at .a branching point. file reach the branching points
Lee (1980)
at different times.Furthermore,
the passage times are different in different branches so that
at the venous end of a vascular bed the dispersion of the in-
dicator can be found as mentioned above.
The effect of diffusion is marked in small tubes, where
the product of velocity and diameter is small compared to the
diffusion coefficient (Taylor 1953).
Lee (1980) has described the effects of convection and
diffusion in socalled "stream tubes". A stream tube is a func-
tionally defined transport unit which can be described by an
appearance time t and a transfer function h. Fig.59 from Lee

a
(1980) shows the transport function (=transfer function h of
the stream tubes) for the arterial, capillary and venous seg-
308 T. Kenner
ment of a stream tube. The capillary part of the transport
function is markedly narrowed by the noticeable mixing effect
of the "Taylor diffusion" (Taylor 1953). Due to this diffusi-
on the indicator profile front in a capillary is nearly flat
so that paradoxically the dispersion is minimized.
The overall effect on trans-
port by the microcirculation can
be described through convolution
of the three functions shown in
fig. 59, and further by convolut-
ion of the result with the distri-
TIME, lee bution function of the transit
Fig. 59 Transfer function times through all the branches

of-the arteriolar,capilla-
ry and venular streamtube (i.e. all the parallel stream tu-
segment of the microcircu-
iation (Lee 1980) bes) of the examined vascular seg-
ment. As an approximate description of the downstream trans-
fer function corresponding to the transient on the venous
side after an arterial impulse injection of an indicator can
be given by the following simplified exponential function
(Lee and Kenner 1982):

-.B(t-t )/t
a a
h = I (t - t
a
)(13/t )e
a
( I)
I (t - t
a
) is the unit step function starting at appearance
time. Using this function we also describe the arterial trans-
ient of an indicator after intravenous injection and passage
through the lung circulation. In this particular case which

will be discussed below with respect to the continuous recor-
ding of blood density, the overall mixing in the heart and
in the lung circulation is summarized by the described appro-
xi ma t·e e quat ion ( e q • 1 ) •
5.5 A new method and its application for the study of the
microcirculation
Kenner et al.(1977) have introduced a method for the con-
tinuous recording of the blood density into physiological
and clinical research. As summarized in Kenner et al.(1982)
the technique is based on the mechanical oscillator principle
by Leopold et al. (1977).
The physiological use of this technique is based on the
differences between the densities of blood and its components
and on the fact that the mechanical oscillator permits the_
continuous recording of density with an accuracy of 10- 5 g/ml.
The assumption is made and, under the most physiological con-
ditions can be proved, that the following linear mixing equa-
tion is valid for biological fluids including blood as sus-
pension of erythrocytes in plasma:
( 1)
The density of the mixed fluid~ can be calculated from the
densities D. and the volumes V. of its i components. The fol-

~1 1
lowing table shows the densities of some important components
of the blood and of solutions entering the blood or used for

310 T. Kenner
injection as test solution.
erythrocytes 1085 - 1095 g/1
plasma 1015 - 1020 g/1
whole blood 1035 - 1055 g/1
interstitial fluid 1000 - 1005 g/1
isotonic NaCl (37°C) 1000 g/1
20% hypertonic mannitol (37°C) 1061 g/1
Isotonic NaCl corresponds to an osmolar concentration of 310
mosmol/1. Hypertonic 20% mannitol corresponds to 1110
mosmol/1.
The injection of isotonic solutions can be used as any
other indicator dilution method for the determination of car-
diac output and blood volume (Kenner et al. 1980) ,Furthermo-
re, the injection of hypertonic solutions can be used to de-
termine the filtration properties of the microcirculation.
During the passage of the hypertonic solution through the
capillaries fluid will be osmotically shifted from the extra-
vascular space towards the blood.
10 ml NoCI-sol 09"/. 1v 10ml monn1t.o1-sol 20"/o 1v

blood dens1t~ 37"C
(g/1)
1046
Fig. 60 Density transients

in arterial blood of a dog. 1045
Details see text,
Kenner et al.(1982) 30 s
Fig. 60 shows the arterial density transient observed in the

carotid artery of an anesthetized 20 kg dog after intra-
venous injection of 10 ml of isotonic NaCl solution (left).
The transient is quite similar to dye dilution or to thermo-
dilution transients.
Once the dilution transient of an injected solution of
known density has been recorded, it is possible to calculate
the expected transientLlo (t) for any other solution with

\ exp
known density. In fig. 60 the calculated expected density
transient after intravenous injection of 20% hypertonic man-
nitol solution is shown as a dashed line (right). -The ac-
tual observed transientfl{>obs(t) is "distorted" by the os-
motic influx of extravascular fluid (Kenner 1980,1982). The
observed transient thus is composed of two parts written on
the right side of the following equation:
.6~obs(t) = L1o1;.exp ( t) + Lleosmot (t) (2)
A model of the process has been presented by Lee and
Kenner (1982). As discussed in section 5.4 the transient of
the injected solution can be described by a simple exponen-
tial function. After intravenous injection of a hypertonic
solution the bolus passes through the lung and appears in
the artery as modeled in the upper part of fig. 61 as
"hypertonic disturbance" which is the same as L1o\exp (t) in
eq. 2 (Kenner et al. 1980).
The passage of the "hypertonic disturbance" leads by its
osmotic pressure to a fluid flux from the lung to the blood.
This fluid flux is shown in the lower part of fig. 61.

312 T. Kenner
The flux corresponds td the time derivative of the weight
of the lung as measured in a gravimetric experiment. This
influx of extravascular fluid, furthermore is proportional
to the "osmotic density transient" Llo"osmo t(t) and shows the
characteristic biphasic shape which was also observed in
the original recording of fig. 60. The biphasic shape of
L1 ee"p(t)
.I k---~
----DI~UR~CE 1
.I v~
I I
0.5 1.0 1.5
L'1eosmot(t) TIME. min.
Fig. 61 Hypertonic disturbance and osmotic density

transient according to a model by Lee and
Kenner (1982)
the density transient can be explained by the fact that
the total mass of the solutes in the lung tissue does not
change. During the transition of the hypertonic bolus, there-
fore, the lung tissue through the extraction of fluid trans-
iently becomes hyper~onic. After passage of the bolus a re-
versed flux into the tissue transiently increases the blood
density.
These experimen~s permit the determination of the trans-
port and the filtration properties of small vessels.
The effect on the pre- and post-capillary resistances
Ra and Rv of vasoactive substances injected into the circula-
tion can be observed indirectly by our method from the trans-

ient response of the arterial density. After the injection
of vasoconstricting drugs we found a transient decrease of
the density as shown in fig. 62 (Kenner et al. 1977 b).
mm Hg
180
140
g/cml
1 046l
1.045
1044 1 mon
Fig. 62 Blood pressure (upper tracing) and arterial blood

density in a cat after injection of I fg angioten~
sin I I (Kenner et al. 1977 b)
This decrease of the density indicates a relatively more
marked constriction of the precapillary resistance which
reduces the capillary pressure and thus leads to an influx
of interstitial fluid. Following this transient the densi-
ty tends to rise with rising pressure. We also observed this
parallelity of density and pressure during experiments in
dogs and during measurements in humans during hemodialysis
(Kenner et al. 1977 b). Particularly marked is thf entry in-

,.
to the blood of interstitial fluid immediately after blee-
ding. - Here a very fast reduction of blood density is a ty-
pical response (Kenner et al. 1977).

314 T. Kenner
6. SUMMARY
In this article I have tried to present an overlook
over the interesting problems related to blood flow in small
artertts. Again, it should be stated that this part of the
circulation and its effects cannot be understood without
looking at the circulation as a whole.
Small arteriu are mainly resistance vessels and thus
are responsible for the adjustment of the blood pressure
on one hand and of transcapillary fluid exchange on the
other hand.
Small arteries have some properties in common with
large vessels, and some properties in common with capil-
laries, especially as far as flow of erythrocytes and plas-
rna is concerned.
My v.rewpoint in presenting these topics is certainly
a personal one and I hope the reader will excuse the empha-
sis on our own work.

Literature and References 315
Anliker, M., Histand, M.B. and Ogden, E. (1968), Dispersion
and attenuation of small artificial pressure waves in the
canine aorta, Circulat.Res.23, 639.
Attinger,E.O. and Attinger, F. (1973), Frequency dynamics
of the peripheral vascular blood flow, Ann. Rev.Biophys.
Bioengineering. 2, 7.
Basar, E. (1981) , Vasculature and circulation, Elsevier/
North Holland , Amsterdam-N.Y.-Oxford.
Basar, E., Basar-Eroglu, c., Demir, N., Turner, N. and Weiss,
C. (1982). The overall myogenic coordination in circula-
tory dynamics, pp. 509, from Kenner et al. (l.c.)
Bauer,R.D., Busse, R. (1979), The arterial system, dynamics,
control theory and regulation. Springer, Berlin-Heidel-
berg.
Bauer, R.D.,Busse,R., Schabert,A. and Wetterer,E. (1982),
The role of elastic and viscous wall properties in the
mechanics of elastic and muscular arteries, pp.373,from
Kenner et al. (l.c.)
Benninghoff, A., (1930), BlutgefaBe und Herz. from: Hand-
buch der mikroskopischen Anatomie, Bd. VI/I pp.I,Berlin.
Bergel, D.H. (1961), The dynamic elastic properties of the
arterial wall. J. Physiol. (London) 156, 458.
Broemser,Ph. (1932) , Beitrag zur Windkesseltheorie des
Kreislaufs. Zeitschr. Biol. 93, 149.

316 T. Kenner
Broemser, Ph. and Ranke, F. (1930),Uber die Messung des
Schlagvolumens des Herzens auf unblutigem Weg. Zeitschr.
Biol. 90, 467.
Burton, A.C.(l951), On the physical equilibrium of small
blood vessels. Amer.J.Physiol. 164, 319.
Burton, A.C. (1962), Physical principles of circulatory phe-
nomena: the physical equilibrium of heart and blood
vessels, from Handbook of physiology. Sect. 2, Circula-
tion , Vol. I, pp. 85 , Washington. D.C.
Busse,R.,Bauer,R.D., Burger,W.,Sturm,K. and Schabert,A.
(1982), Correlation between amplitude and frequency
of spontaneous rhythmic contractions and the mean circum-
ferential wall stress of a small muscular artery, from
Kenner et al. pp.363 (l.c.)
Busse,R., Sturm,K., Schabert,A. and Bauer,R.D. (1982 b),
The contribution of tqe parallel and series elastic
components to the dynamic properties of the rat tail
artery under two different smooth muscle tones. Pflugers
Arch. 393, 328.
Caro,C.G.,Pedley,T.J.,Schoter,R.C. and W.A.Seed,(l978),
The mechanics of the circulation. Oxford University Press
N.Y.-Toronto.
Cox,R.H. (1982), Determination of the mechanical properties
of the contractile system in arterial smooth muscle
models. pp.317, from Kenner et al. (l.c.)

Dobrin,P.B.,(1978), Mechanical properties of arteries.
Physiol. Rev. 58, 397.
Dujardin,J.P.L. and Scott,D.L. (1982), The dynamic arterial
pressure flow relationship and total arterial compliance
in spontaneously hypertensive and normal rats. pp. 199,
from Kenner et al. (1. c.)
Folkow,B. and Neil,E. (1971), Cir~lation, Oxford University
Press, N.Y.-London-Toronto.
Frank,O. (1899), Die Grundform des arteriellen Pulses.
Zeitschr. Biol. 37, 483.
Frank,O. (1906), Die Analyse endlicher Dehnungen und die
Elastizitat des Kautschuks. Ann. Physik. 21, 602.
Frank,O. (1920), Die Elastizitat der BlutgefaBe. Zeitschr.
Biol. 71,255.
Fung, Y.C. (1981), Biomechanics, mechanical properties of
living tissues. Springerverlag, N.Y.-Heidelberg-Berlin.
Gessner,U. (1981) personal communication.
Gow.,B.S. (1980), Circulatory correlates: vascular impedan-
ce, resistance , and capacity. pp 353, from Handbook of
Physiology. Vol.II Vascular smooth muscle. Washington D.C.
Green,H.D. (1944), Circulation: Physical principles, from
O.Glasser (ed.) Medical Physics. Year Book Publ.,Chicago.
Green, H.D., Lewis,R.N.,Nickerson,N.D. and Heller,L.(1944),
Blood flow, peripheral resistance and vascular tonus,
with observations on relationship between blood flow and
cutaneous temperatures. Amer.J.Physiol. 141,518.

318 T. Kenner
Gross, J.F.,(1977), The significance of pulsatile micro-
hemodynamics. pp. 365 1 from G.Kaley and B.M.Altura (eds.)
Microcirculation Vol.1, University Park Press, Baltimore.
Gross,J.F. and Popel,A.S. (1980), Mathematics of microcircu-
lation phenomena. Raven Press, N.Y.
Guyton,A.C. and Cowley,A.W.,(1976), Cardiovascular Physiolo-
gy II., University Park Press , Baltimore.
Hatakeyama, I.,(1982) , Hydrodynamic amplification in blood
vessels and cardiovascular dynamics. pp.181, from Kenner
et al. (l.c.)
Hudetz,A.G. and Monos,E.,(1982), A structural model for non-
linear anisotropic behaviour of the arterial wall.
p p . 3 3 7 . from Kenner e t a 1. ( 1. c . )
Iberall, A.S.,(1967) ,Anatomy and steady flow characteris-
tics of the arterial system with an introduction to its
pulsatile characteristics. Math. Biosci. 1,375.
Intaglietta, M., (l977), Transcapillary exchange of fluid
in single microvessels. pp. 197 , from Kaley and Altura
(l.c.)
Kaley,G. and Altura,B.M. (eds.) (1977). Mircocirculation
Vol.I., University Park Press, Baltimore.
Kenner,T. (1967), Neue Gesichtspunkte und Experimente
zur Beschreibung und Messung der Arterienelastizitat.
Archiv.Kreislaufforschung.54,68.
Kenner,T. (1971), Dynamic control of flow and pressure in
the circulation. Kybernetik, 9,215.

Kenner,T. (1972), Flow and pressure in the arteries. from
Fung,Y.C. et al. (eds.) , Biomechanics, its foundations
and objectives. Prentice Hall. Englewood Cliffs N.J.
Kenner,T. (1974) , Beziehungen zwischen Dynamik und Re-
gulation des Arteriensystems. Verb. Dtsch. Ges. Kreis-
laufforschung. 40, 41.
Kenner,T. (1975) , The central arterial pulses. Pflligers
Arch. 353, 67.
Kenner, T. ( 197 8) , Models of the arterial system. from
R.D.Bauer and R.Busse (eds.). The arterial system. pp.80,
Springerverlag, Berlin-Heidelberg.
Kenner,T. (1979) , Physical and mathematical modeling in
cardiovascular systems. pp.41 1 from N.H.C.Hwang et al.
Quantitative cardiovascular studies. University Park-
Press.
Kenner,T. and Bergmann H. (1975), Frequency dynamics of
arterial autoregulation. Pflligers Arch. 356,169.
Kenner,T., Busse,R. and Hinghofer-Szalkay,H. (1982)
Cardiovascular system dynamics - models and measurements,
Plenum Press. N.Y-London.
Kenner,T., Hinghofer-Szalkay,H., Leopold,H. and Pogglitsch,
H. (1977 b). The relation between the density of blood
and the arterial blood pressure in animal experiments ·
and in patients during hemodialysis. Zeitschr. Kardiol.
66, 399.
320 T. Kenner
Kenner,T., Hinghofer-Szalkay,H.,Moser,M and Leopold,H.(1982)
The application of the continuous recording of blood den-
sity for hemodynamic measurements. pp. 431, from Kenner
et al. (l.c.)
Kenner,T., Moser,M. and Hinghofer-Szalkay,H.(1980) , Determi-
nation of cardiac output and transcapillary fluid exchan-
ge by continuous recording of blood density. Basic Res.
Cardiol. 75, 501.
Kenner,T. and Ono,K. (1971) , Reciprocal autoregulation of
blood flow and blood pressure. Experientia 27,528.
Kenner,T. and Ono,K, (1971) , The low frequency input impe-
dance of the renal artery. Pflugers Arch. 324, 155.
Kenner,T. and Ono,K. (1972) , Humoral autoregulation of
blood flow and blood pressure. Experientia 28, 528.
Kenner, T. and Ono,K. (1972) , Analysis of slow autooscilla-
tions of arterial flow. Pflugers Arch. 331, 347.
Kenner,T. and Ono,K. (1972) , Interaction between circula-
tory control and drug-induced reactions. Pflugers Arch.
331,335.
Kenner,T., Ueda,M, Huntsman,L. and Attinger,E.O. (1968) ,
Effects of local and general hypoxia on iliac flow.
Angiology 5,345.
Kenner,T., van Zwieten,P.A. (1982), Use of hemodynamic
analysis for the interpretation of the mode of action
of vasoactive drugs. from Kenner et al. (1. c.)
Klitzman,B. and Duling, B.R. (1979), Microvascular hemato-

crit and red blood cell flow in resting and contracting
striated muscle. Amer.J.Physiol. 273, H 481.
Lee,J.S. (1980), Micro-macroscopic scaling. pp.159 from
Gross and Popel (l.c.)
Lee, J.S. and Kenner,T. (1982) , Microvascular dynamics.
from Kenner et al. pp. 413, (l.c.)
Lee, J.S. and Nellis,S. (1974) , Modeling studies on the
distribution of flow and volume in the microcirculation
of cat mesentery. Ann. Biomed. Eng. 2, 206.
Lefevre, J. ( 1982) , Teleonomical representation of the
pulmonary arterial bed of the dog by a fractal tree.
pp. 137, from Kenner et al. (l.e.)
Leopold, H., Jellinek,R. and Tilz,G. (1977) , The applicati-
on of the mechanical oscillator technique for the deter-
mination of the density of physiological fluids. Biomed.
Technik. 22, 231.
Lindner,A. and Ronniger,R. (1955), Zur Darstellung der
Beziehungen zwischen zentralen und peripheren Pulsen als
Ortskurven. Arch. Kreislaufforschung. 22, 72.
Lipowsky,H.H., Usami,S. and Chien,S. (1980) , In vivo mea-
surements of "apparent viscosity" and microvascular hema-
tocrit in the mesentery of the cat. Microvasc. Res.19,
297.
Lipowsky,H.H. and Zweifach,B.W. (1974), Network analysis
of microcirculation of cat mesentery. Microvasc. Res.
7' 7 3.
322 T. Kenner
Mayrovitz,H.N.,Wiedeman,M.P. and Noordergraaf,A. (1976)
Analytical characterization of microvascular resistance
distribution. Bull. Math. Biphys. 38,71.
McDonald,D.A. (1974) , Blood flow in arteries, 2 nd ed.
Edward Arnold, London.
Metzger,H. (1973) , Geometric considerations in modeling
oxygen transport processes in tissue. Advances Exp.Biol.
Med. 3 7 6, 6 6 I .
Monos,E. and Kovach, A.G.B. (1982), Biomechanics of isola-
ted canine splenic artery. pp. 32 7, from Kenner et al. ( 1. c.)
Newman,D.L. and Greenwald,S.E. (1982) The effect of smooth
muscle activity on the static and dynamic properties of
the rabbit carotid artery. pp. 393 , from Kenner et al.
(l.c.)
Noordergraaf,A. (1978) , Circulatory systems dynamics.
Academic Press, N.Y.
o·Rourke, M.F. (1982), Vascular impedance in studies of
arterial and cardiac function. Physiol. Rev. 62, 570.
o· Rourke, M. F. ( 1982) , Vascular impedance - a call for
standardization. pp. 175, from-Kenner et al. (l.c.)
Patel, D.J. ,Austen,W.G. and Greenfield,J.C. (1964) , Impe-
dance of certain large blood vessels in man. Ann. N.Y.
academy Sci. 115,1129.
Patel,D.J. ,FreitastF.M. ,Greenfield,J.C, and Fry,D.L. (1963),
Relationship of radius to pressure along the aorta in
living dogs. J.Appl. Physiol. 18,1111.

Patel,D.J.,Vaishnav,R.N. and Atabek,H.B. (1979) , Local
mechanical properties of the vascular intima and adja-
cent flow fields. p. 215, from Hwang et al. Quantitative
cardiovascular studies. University Park Bress.
Pollak,G.H.,Reddy,R.V. and Noordergraaf,A. (1968), Input
impedance, wave travel and reflections in the pulmonary
arterial tree. Studies using an electric analog. IEEE
transact. BME. 15,151.
Popel,A.S. (1980) Mathematical modeling of convective and
diffusive transport in the microcirculation. pp.63 1 from
Gross and Poper (l.c.)
Rhodin, J.A.G. (1980), Architecture of the vessel wall.
pp. 11 from Handbook of Physiology, Vol. II, Vascular
smooth muscle, Washington D.C.
Ronniger,R.,(I954) , tiber eine Methode der Ubersichtlichen
Darstellung hamodynamischer Zusammenhange. Arch. Kreis-
laufforschung. 21,127.
Ronniger,R., (1955). Zur Theorie der physikalischen Schlag-
volumenbestimmung. Arch. Kreislaufforschung. 22,332.
Rosen,R. (1967), Optimality principles in biology.
Butterworths, London.
Rubenstein,H.J.,Kenner,T. and Ono,K. (1973), Pseudorandom
test technique for the characterization of local hemo-
dynamic control. Pflligers Arch. 343,309.
Schimmler, W.,(l965), Untersuchungen zum Elastizitatspro-
blem der Aorta. Arch.Kreislaufforschung ,47,189.

324 T. Kenner
Schleier,J. (1918). Der Energieverbrauch der Blutbahn.
Pfliigers Arch. 173,172.
Schmid-Schonbein,H. (1976) , Microrheology of erythrocytes,
blood viscosity and the distribution of blood flow in
the microcirculation, pp.1 1 from Guyton and Cowley (l.c.)
Sipkema,P. and Westerhof,N. (1982), Peripheral resistan~e
and low frequency impedance of the femoral bed. pp.501,
from Kenner et al. (l.c.)
Taylor,G. (1953), Dispersion of soluble matter in solvent
flowing slowly through a tube. Proc.Royal Soc.Lond.
Ser.A. 219,186.
Taylor,M.G., (1966), Use of random excitation and spectral
analysis in the study of frequency dependent par~meters
of the cardiovascular system. Circulat. Res. 18,585.
Vadot,L. (1967), Mecanique du coeur et des arteres.
L~expansion scientifique Franc., Paris.
Van Loon,P.,Klip,W. and Bradley,E.L. (1977), Length-force
and volume-pressure relationship of arteries.
Biorheology, 14,181.
Weizsacker,H.W. and Pascale,K. (1977), Das Kraft-Dehnungs-
verhalten von Rattenkarotiden in Langsrichtung bei ver-
schiedenem Innendruck und seine modulmaBige Deutung.
Basic Res.Cardiol. 72,619.
Weizsacker,H.W. and Pascale,K. (1982), Anisotropic passive
properties of blood vessel walls. pp.347, from Kenner
et al. (1. c.)

Westerho£,N.,Bosman,F.,De Vries,C.J. and Noordergraaf,A.(I969)
J. Biomechanics 2,121.
Westerh6f,N.,Sipkema,P.,Elzinga,G,Murgo,J. P. and Giolma 7 J.P.,
(1979), Arterial impedance, pp. Ill, from Hwang et al.
Quantitative cardiovascular studies, University Park Press.
Wetterer,E. ,Bauer,R.D. and Busse.R. (1977) , Arterial dyna-
mics. INSERM-Euromech 91, Cardiovascular and pulmonary
dynamics,Vol.71 ,pp.17.
Wetterer,E. and Kenner,T. (1968), Grundlagen der Dynamik
des Arterienpulses. Springerverlag, Berlin-N.Y.Heidel-
berg.
Wetterer,E. and Pieper,H. (1953 a), tiber die Gesamtelasti-
zitat des arteriellen Windkessels und ein experimentelles
Verfahren zu ihrer Bestimmung am lebenden Tier. Zeitschr.
Biol. 106,23.
Wetterer,E.and Pieper,H. (1953 b), Messungen am Arterien-
system in vivo wahrend erzwungener periodischer Volumen-
schwankungen. Verh.Dtsch.Ges.Kreislaufforschung 19,259.
Wezler,K. and Sinn,W.,(1953), Das Stromungsgesetz des Blut-
kreislaufs. Ed.Cantor,Aulendorff i. Wlirttemberg.
Whitmore.R.L. (1968) ,Rheology of the Circulation. Pergamon
Press, Oxford-London-N.Y.
Wiedemann.M.P. (1962), Lengths and diameters of peripheral
arterial vessels in living animals. Circulat.Res. 10, 686.
Wiedeman, M.P. (1963), Dimensions of blood vessels from dis-

326 T. Kenner
tributing artery to collecting vein. Circulat.Res. 12,375.
Witzig,K. (1914), Uber erzwungene Wellenbewegungen zaher,
inkompressibler FlUssigkeiten in elastischen R5hren.
Dissertation, Bern.
Womersley,J.R. (1957), An elastic tube theory of pulse
transmission and oscillatory flow in mammalian arteries.
WADC Report, TR 65-614.

CHAPTER V
FLOW IN LARGE ARTERIES
Faculty of Engineering
The University of Alberta
Edmonton, Alberta, Canada
1. INTRODUCTION
1.1 Introduction
For centuries, the world within himself fascinated man as much as
his near and distant environment. In particular the cardiovascular sys-
tern was the object of attention of scientific observers like Aristotle
and Leonardo da Vinci. However, the concept of the Circulation of the
Blood was clearly presented by W. Harvey in 1628, in his famous De Motu
Cordis et Sanguinis in Animalibus. The evidence provided was almost com-
plete and reached into the present day understanding, except that Harvey
could not see the passage of blood from the peripheral arteries to the
veins. He speculated that there must be "pores" at these locations.
These "pores" were in 1661 identified by Malpighi as the capillaries
(K.D. Keele, 1978). Later in 1733 Stephen Hales, the Vicar of Teddington,
328 C.M. Rodkiewicz
published the first measurements of arterial blood pressure.

The circulatory system, with the heart as the driving force of the
double-action pump type, serves to transport and deliver to the tissues
those substances which are essential for maintenance of function and to
remove the by-products of metabolism. The vessels that arise from the
two ventricles and serve as the delivery roots constitute the pulmonary
and systemic arteries. The systemic arteries are considered to be the
primary distributing conduits, however, the smaller arteries, and in
particular the arterioles, maintain the blood pressure and are instru-
mental in regulation of flow rate to the respective tissues. Elaborate
drawing of the maze of arteries running through the body can be found in
any anatomy text book. Originating at the aorta these distributing con-
duits subdivide the flowing blood into many streams by the Y type bifur-
cations, most of which are non-symmetric. The aorta resembles an in-
verted U with one end leadin9 out of the heart and the other end leading
down into the abdomen. 1 Normally it has three channels branching from the
top of the curve to the head and upper body.
The heart produces a periodic or pulsatile flow on the arterial side
of the circulatory system. The amplitude of the flow pulse is largest in
the aorta and becomes gt·adually smaller as the system branches. The ar-
terial vessels are subjected to higher pressure and pressure variation,
and they are thicker and contain more elastin than the venous system.
Despite the extra stren9th and elasticity of the arterial walls, it seems
likely that a system under the continual wear and tear of a pulsatile flow
would be subject to many disorders. Such is the case; arterial disease
Introduction 329
is a very great problem.

Among diseases of the arterial tree atherosclerosis is the most com-
mon and the most important. Under the general terms of atherosclerosis
are included several types of tissue changes. Despite the abundance of
relevant information atherosclerosis is little understood except in rather
broad terms. Its apparently complex pathology so far defies precise ex-
planation. There are few theses explaining the etiology of the athero-
sclerotic formations. These have been discussed, for example by
Constantinides (1965).
One of the basic features of atherosclerosis is that it occurs pre-
dominantly at specific sites of the arterial net. It seems appropriate,
therefore, to consider its association with the blood flow phenomena,
which is implicated here in the etiology. The regions most susceptible
to the atherosclerotic lesions are portions of larger arteries such as
bends and junctions of various geometry. Although it is understood that
hemodynamic forces have something to do with the lesion sites, the exact
mechanism of inflicting these lesions has not yet been elucidated.
Considering the fluid-flow point of view, one cannot fail to notice
that the above mentioned sites of lesions are also locations of develop-
ing stagnation and separation regions. Some correlation between the oc-
currence of the atherosclerotic plaque and the angle of bending and
branching have been noted by Schneck and Gutstein (1966). The possible
correlation between turbulent flow, thrombosis, ctnd arterial lesions has
been also pointed out by Mitchell and Schwartz (1965). Rodkiewicz (1975)
postulated that the atherosclerotic formations of the aortic arch commence
and develop at the locations where there are developed or developing sep-
330 C.M. Rodkiewicz
aration and stagnation regions. This could be extended to other geo-

metrical configurations, for examples bifurcations.
1.2 Newtonian Behavior of Blood

In the case of fluid motion, such as the flow in the cardiovascular
system, the influence of viscosity is such that, under the assumption of
the no-slip condition at the wall, the frictional forces retard the motion
of the fluid in a layer near the wall. In this layer, the boundary layer,
the velocity of the fluid increases from zero at the wall to its full
value at a distance from the wall. These velocity gradients across the
fluid stream give rise to the frictional shearing stresses (particularly
important at the wall}, which when integrated yield the viscous drag on
the wall. To overcome this drag the heart has to expend associated energy.
According to Newton's law of friction the shear stress at the wall is pro-
portional to the product of the absolute viscosity (viscosity describes
the resistance of a fluid to shear when the fluid is subjected to a tan-
gential stress) and the rate of deformation. Though blood is recognized
to be a non-Newtonian fluid it may be considered a Newtonian fluid when
flowing through conduits of a larger diameter. Consequently, for such
cases Newtonian fluids have been used as transport media when modelling
the blood flow. Weiting (1968), for example, found that a 36.7% glycerol-
aqueous solution was a good hydraulic analog for blood.
1.3 Separation and Stagnation Regions

In some cases the decelerated fluid particles do not follow the di-
rections suggested by the containing walls. The boundary layer, in the
Introduction 331
downstream direction, may increase significantly and the flow next to the
wall may become reversed. In such cases the decelerated fluid stream be-
comes separated from the wall as indicated in Fig. 1.3.1. This is as-
sociated with additional energy losses. The boundary layer separation
SHEAR STRESSES AT THE

WALL NEAR THE SEPARATION
POINT Se
Fig. 1.3.1 Diagrammatic Representation of Flow Near a Point of

Boundary Layer Separation.
exists in regions with an adverse pressure gradient and the likelihood of

its occurrence increases at flow dividing elements, or in flows around the
curved walls. At the point of separation, Se, the shear stress is zero,
but on each side of it the shear stresses are finite and act in the di-
rection toward the point, or line segment, Se (converging). A prerequisite
for a separation to appear is a prior development of a low shear stress
region.
Another pertinent case, so called free stagnation flow, is shown in
332 C.M. Rodkiewicz
Fig. 1.3.2. Fluid impinges on the wall creating a stagnation point, or
SHEAR STRESSES
AT THE WALL
NEAR THE
STAGNATION
POl NT St
Fig. 1.3.2 Diagrammatic Representation of Flow Near a

Stagnation Point.
line segment, St. Along the streamline which leads to the stagnation
point the pressure increases in the direction of flow. On each side of
that point the shear stresses are finite and act in the direction away
from the point St (diverging).
1.4 Flow Govern1ng Parameters

According to Kuchar and Ostrach {1965), and also Kuchar and Scala
(1968), the flow in a segment of arterial tree of a fixed geometry is de-
scribed by three dimensionless parameters: Reynolds number, frequency
parameter, and the amplitude parameter. These are, respectively,
1/2 I
Re = UD/v, a= a(n/v) , A= U /U (1.4.1,1.4.2,1.4.3)
Introduction 333
where D = unstressed characteristic diameter, a = unstressed character-

I
istic radius, U = characteristic longitudinal velocity, U = peak to peak

amplitude of the fluctuating component of longitudinal velocity, v =
kinematic fluid viscosity, and n = pulse rate.
1.5 Arterial Passage Classification

It is proposed to classify the large arteries into four major geo-
metrical types each with a descriptive set of dimensionless parameters.
The first class is considered to be straight arteries with the possibility
of a converging or diverging nature. The second class introduces curva-
ture to the straight artery, again with the possible complication of con-
vergence or divergence. The third class is composed of tubes with bi-
furcations or branches, and includes the possibilities that the tubes may
be curved, and converging or diverging. The fourth class are the most
complicated systems such as the aortic arch which combines all the aspects
of the first three groups.
2. BASIC EQUATIONS
2.1 The Governing Equations
The differential form of the law of conservation of mass can be
written in the form
¥t-+v·(pV)=O (2.1.1)
where p is density, t is time, and Vis velocity vector. Blood may be

considered as an incompressible fluid and, therefore, the above equation
may be reduced as follows
334 C.M. Rodkiewicz
v·v = o (2.1.2)
The differential form of the principle of momentum, for constant

fluid properties and a Newtonian fluid, may be written as
DV -t: 1 ]..! 2+ (2.1.3)

Dt ~ T - p 'i/p + p 'i/ V
where f denotes the external force, p is pressure, l..l is absolute viscosity,

and
(2.1.4)
2.2 The Equations of Motion in Rectangular Coordinates

Equations of motion (2. 1.2) and (2.1.3), for later reference, may
now be -written in rectangular coordinates. If x,y,z denote a three-di-
mensional system of coordinates, and Vx' Vy' Vz denote the velocity com-
ponents in the corresponding directions, then the following system of
equations is obtained:
av av av
_!+_y+-z=O (2.2.1)
ax ay az
av av av av
-2.+v -2.+v -2.+v x_g _.!_~
at x ax y ay z az x p ax
(2.2.2)
Basic Equations 335
(2.2.3)
av z av av av z _ 1 an
~
+V x -ax 2 +Vy -ay 2 +V z az- -
gz - p ..:::.J:.
az
(2.2.4)
where v is kinematic viscosity (v = w/p) and various g's represent accel-

eration components of the gravitational field.
2.3 The Equations of Motion in Cylindrical Coordinates

Equations of motion (2. 1.2) and (2.1 .. 3), for the passages of circu-
lar cross-section, may conveniently be written in cylindrical coordinates.
If r, e, z denote the radial, azimuthal, and axial coordinates respect-
ively, of a three-dimensional system of coordinates, and vr' ve, vz de-
note the velocity components in the corresponding directions, then the
following system of equations is obtained:
l ~ (rV ) +lave + avz = (2.3.1)

r ar r r ae az 0
336 C.M. Rodkiewicz
(2.3.2)
(2.3.3)
2 2
a 1a 1- a ve 2 avr a ve
+ v [ - ( - - ( rV ) ) + -2- -...,.- + -2 - + -...,.-]
ar r ar e r aeL r ae azL ·
2 2
(2.3.4)
1 a avz 1 a vz a vz
+ v [ - . , (r - ) + - 2 -...,.- + ~]
r or ar r aeL dZ
In the above, the term V~/r when multiplied by p, yields the centrifugal
force. Similarly, expression p VrV 6 /r is the Coriolis force.
2.4 The Shear Stresses

The components of the local shear stress associated with Equations
(2.2.1) through (2.2.4) may be written in the following way.
avx ~
'xy = 'yx = ~ <ay + ax ) (2.4.la)
(2.4.lb)
Basic Equations 337
av av
Tzx = Txz = ~ (ax 2 + azx) (2.4. lc)
Similarly the local shear stress components associated with Equations

(2.3.1) through (2.3.4) are as follows.
=T = ~ [r ~ ( ve) + l avr] (2.4.2a)

er ar r r ae
(2.4.2b)
(2.4.2c)
A double-index scheme has been utilized. The first subscript de-

notes the direction of the normal to the plane containing the stress,
while the second subscript indicates the direction of the stress itself.
For example Tzx is the value of the shear stress acting in a plane whose
normal is parallel to the z direction, while the stress itself is parallel
to the x direction.
3. FLOW IN STRAIGHT PASSAGES

3.1 Introduction
In the case of a simple straight artery of constant diameter the
velocity and pressure distributions would be given by the Poiseuille re-
lationship provided the flow is fully developed. However, in the arteries
in which blood may be assumed to behave like a Newtonian fluid, this con-
dition most likely does not exist. Since the flow rate must be the same
338 C.M. Rodkiewicz
for every passage section the decrease in the volume of flow near the wall,
which is due to friction, must be compensated by a corresponding increase
near the axis. Consequently, the boundary layer is formed under the in-
fluence of an accelerating external stream. Such a steady state flow will
never separate from a regular wall and the shear stress at the wall will
tend to transform asymptotically to the value which would exist if
Poiseuille flow could be established. In the pulsatile case adverse pres-
sure gradients may be present and, depending on the fluctuation frequency
and amplitude, reversed flow at the wall could appear. The associated
minimum wall shear stress would be equal to zero and the wall drag would
keep changing its direction.
For the steady state flow in the converging and diverging passages
the incompressible fluid accelerates and decelerates, respectively. In
the latter case, when channel divergence is sufficiently significant, the
boundary layer may not be able to cope with the adverse pressure gradient
and the flow may separate from the wall. In a closed circuit it will re-
attach further downstream, and will enclose a separation region. At the
locus of separation points the shear stress at the wall is equal to zero.
The wall shear stresses within the separation region, which generates
vortices shed into the main stream (not to be identified·with the onset
of turbulence), are in the direction opposite to the direction of the
swifter main stream.
In general the separation region should not vanish in the pulsatile
flow. It should, however, change its shape and shift back and forth in
accord with the pulsation frequency.
Flow in Straight Passages 339
At this point it is of interest to mention a study done by M.J.

Martin (1960) which illustrates the presence of atherosclerotic plaque in
the carotid sinus. This is where channel divergence has been observed.
3.2 Flow in a Rigid Tube of Elliptic Cross-Section

Consider a steady and fully developed flow in which the body forces
may be neglected. Let the z-direction be the direction of flow. For such
a case Equations (2.2.1} through (2.2.4) reduce to the form:
~+ ~= 0 (3.2.1)
ax ay
au au - .!_ .££. + \) (a 2u + a2u)

u-+v-= (3.2.2)
ax ay p ax
~ a/
2 2
u av + v ~ = - .!_ .££. + \) (a v + a v) (3.2.3)
ax ay P ay ~ a/
2 2
u aw + v aw = _.!_EE.+v (~+a w) (3.2.4)
ax ay p az
ai ai
where for Vx' Vy' Vz we have written u, v, w respectively.

Differentiating expression (3.2.2) with respect to y, and expression
(3.2.3) with respect to x and eliminating the pressure terms by subtract-
ing one of these equation from the other, one obtains
(3.2.5)
340 C.M. Rodkiewicz
Let us now introduce the stream function~ (x,y), so that Equation (3.2.1)
is satisfied automatically by the following expressions
u - d~
- ay'
v =
- ax
d~ (3.2.6)
Now, with the aid of (3.2.6), Equation (3.2.5) can be written in terms of
the stream function, namely
(3.2. 7)
The u and v velocity components are equal to zero on the boundary.

Therefore, according to expressions (3.2.6) function ~ is equal to a con-
stant on the boundary. But function ~ = constant also satisfies Equation
(3.2.7) and, consequently, function~= constant is the solution. This
in turn means that the u and v velocity components are equal to zero at
all points of the flow cross-section. If so, Equations (3.2.2) and (3.2.3)
indicate that ap/ax = 3p/3y = 0, and Equation (3.2.4) assumes the follow-
ing form
2 2
a w +a w = l~ =CONSTANT
;;! al1l dz
(3.2.8}
Equation (3.2.8) yields
1
w = -------.2..-- ('2 x2 + Y2 - ,2 a2) ~
I\ dzI\ (3.2.9}
2 11 (1 + A )
where A = b/a, and 2a and 2b are the major and minor axes of a cross-
section, respectively. Also
(3.2.10)
In addition, using the following definition for the friction

coefficient
f = Twall (3.2.11, 3.2.12)

1 2
2 P w mean
we obtain (tis length of tube):
8
f =-
Re (3.2.13)
where
w L
Re = mean L = K Aa (1 +A),
v ' (1 + A2)
1- A
. . . ' m=m
3.3 The Hagen-Poiseuille Flow

The solution for flow through a tube of elliptic cross-section can
be specialized to the case of flow through a tube of circular cross-section.
Letting a= b, we obtain A= 1, and from Equation (3.2.9) the velocity
component parallel to the axis becomes
342 C.M. Rodkiewicz
1 2 2 dn
w = - (r
4 ~
- a )=-
dz
(3.3.1)
where r 2 = x2 + y2. For this case m = 0, K = 1, L =a, and the Reynolds

number based on the tube radius becomes Re = wmeana/v. The velocity
over a cross-section is distributed in the form of a paraboloid of rev-
olution. This type of laminar flow occurs as long as the tube Reynolds
number has a value which is less than the critical Reynolds number. For
large Reynolds numbers the flow may become turbulent and the present
analysis becomes invalid.
At this time it should be indicated that the generally adopted
practical definition of the friction factor for pipe flow calculations,
which differs from definition (3.2.11), is the Darcy-Weisbach equation,
namely
R- v2
h =f --
D 2g
(3.3.2)
where h is the head loss (meter-newton per newton); R, is length of pipe;

D is pipe diameter; V is mean velocity; and g is the gravitational con-
stant. In Equation (3.2.10) the pressure gradient dp/dz is a constant,
and when used in conjunction with Equation (3.3.2) yields
f = 64 (3.3.3)
Re
where Re = VD/v.
3.4 Pulsating Flow in Rigid Circular Tube

Blood in the arterial system moves under the influence of the vari-
able pressure gradients. In order to better understand the pulsatile
blood flow some simplified analytical models have been developed. At this
time we will involve ourselves with the solution of S. Uchida (1956),
which is similar to that given by J.R. Womersley (1955). In the limit
these solutions reduce to the steady state Hagen-Poiseuille equations
provided in Section 3.3.
The associated simplifying assumptions were: the artery is re.garded
as a rigid tube of constant diameter; the velocity distribution across
the flow section is independent of the coordinate along the direction of
flow; the velocity is purely axial everywhere; there are no body forces;
density is constant; viscosity is constant; the flow is laminar; and the
fluid is a Newtonian fluid. It is most unlikely that all of these con-
ditions are met in the arterial tree - yet the solutions serve a useful
purpose. For such a flow, equations (2.3.1) through (2.3.4) reduce to a
single equation used by Uchida (1956) and by Womersley (1955) as the
starting point, namely
2
aw = _ .!_ EB. + v (~ + .!_ aw) (3.4.1)
at p az ar~ r ar
where r is radius, z is the flow direction and w· is velocity in z direc-

tion.
The pressure gradient in (3.4.1) becomes a function of time only
and it was expressed by a Fourier series as follows
344 C.M. Rodkiewicz
- l~ ~
p az = po +n=l (P en cosn t + Psn sin nt) (3.4.2)
where P0 corresponds to the Hagen-Poiseuille flow pressure gradient, and

Pen and Psn are constants representing the amplitudes of the time contri-
buting terms. The corresponding expression for the velocity distribution
was assumed to be
00
w = W + E (Wen cos nt + Wsn sin nt) (3.4.3)

o n=l
where W0 is the Hagen-Poiseuille velocity distribution given by expression

(3.3.1), and Wen and Wsn are constants representing the amplitudes of the
time contributing terms.
Expressions (3.4. 1) through (3.4.3) yield the dimensionless velocity
distribution (k = ln7V ; note that ak = a) given by
= 2{1 - ~} + E Pen [ ____§___!__ cosnt + 8 (1 -A) sin nt]

aL n=l Po (ka)2 (ka)2
p
+ I ~ [ ~ sin nt - 8( 1 - A) cos nt ] (3.4.4)
n=l Po (ka)2 (ka)2
where
ber ka berkr + beika bei _:_

A = -----..,------=-.......---=-.:....:_...;.: kr (3.4.5a)
2
ber ka + bei ka 2
B = bei ka ber kr - berka bei kr

(3.4.5b)
ber 2ka + be1· 2ka
and where Wis the mean velocity of the W0 distribution. The associdted
pressure gradient becomes
2a £E_= 64 [ 1 + E (Pen cosnt + Psn sinnt)] (3.4.6)

az Re n=l P0 P0
where Re = 2aW/v.
In conclusion if a pressure gradient is presented as a Fourier
series, then the corresponding velocity can be computed from equation
(3.4.4). It will be noted that the periodic part of the velocity distri-
bution is governed by the parameter (ka).
Uchida (1956) also obtained velocity distributions for the limiting
cases of a very small and a very large magnitude of the parameter (ka).
In the latter case the motions near the center of the tube and ·near the
wall of the tube were discussed.
For the case when (ka) < < 1 expression (3.4.4) reduces to the
velocity distribution given by a paraboloid of revolution as is the case
for the steady Hagen-Poiseuille flow, namely
1 2 2 1 an (3.4. 7)
w = - (a - r ) (- - =-)
4v p az
It is seen that the magnitude of velocity varies in phase with the pres-
sure gradient.
When the parameter (ka) > 10 the motion near the center of the tube
346 C.M. Rodkiewicz
was given as
r2 00 Pen 8
~ = 2 (1 - 2 ) + L --cos (nt - ~)
w a n=l ~ (ka) 2
00
Psn 8
+L sin (nt - ~) (3.4.8)
n=l ~ ~
and near the wall by
2
w= 2(1 - ;-)
a
co Pen 8 -k(a - r)/12" k

+ L - { sin nt - I~ e sin[nt -- (a-r)]}
n=l Po ~ r 12
psn 8 a -k(a-r)/2 k
+ L: -p- ~ {-cosnt + 1- e cos[nt --(a - r)] }
o (ka) r ~
(3.4.9)
Expression (3.4.8) indicates that fluid at the center of tube flows

with a phase lag of 90° relative to the wave of the pressure gradient.
In addition at this location the amplitude diminishes with increasing
frequency. On the other hand expression (3.4.9) shows that the maximum
velocity occut·s in the neighbourhood of the wall for this case of a very
high frequency. A few examples of the simple periodic pulsations given by
1 an
- - :::.z:.. = p
p az o + p·en cosnt (3.4.10)
are provided in Fig. 3.4.1 through Fig. 3.4.4.

( ~W )/(...&t..)
P
r
a
0
-2
(a) Fluctuating Velocity
Component
cos nt
(b) Pressure Gradient Wave
(c) Fluctuating Velocity Component
Fig. 3.4.1 Velocity Profile for {ka} = 1. After S. Uchida, 1956.

With permission.
348 C.M. Rodkiewicz
r
( ~W )/( _fg)
P 1 a
0

Component
=ot~~ rr nt 2rr
-1
(L) Pressure Gradient Wave
0
(c) Fluctuating Velocity Component
Fig. 3.4.2 Velocity Profile for (ka) = 3. After S. Uchida, 1956.

With permission.
r
a
(a) Fluctuat;ng VelocHy
Component
1~ ~
cos nt -~~~ nt 2rr
r
a
(b) Pressure Grad;ent Wave
( ~)/( Pen)
W P0
(c) Fluctuat;ng Veloc;ty Component

With permission.
350 C.M. Rodkiewicz
r
a
Component
'"'"'~~~v<::
-1
rr
nt
2rr
r
a
(b) Pressure Gradient Wave
-0.10
(c) Fl uc tua ting Velocity Component
With permission.
The solution obtained may be used in finding the instantaneous shear

stress acting on the wall. This is given by~ au/ar, which is to be
evaluated at r = a. Furthermore, the instantaneous drag may be found by
integrating the shear stress over the area of interest.
3.5 Entrance Length

In the previous sections we have considered the fully developed
flows. However, the arterial passages are of such geometrical config-
uration that, although the fully developed flow considerations help in
understanding the basics of the flow phenomena, the actual flow character-
istics will be somewhat altered.
To illustrate: Consider a straight tube with a well-rounded entrance
where the velocity profile is nearly uniform. Owing to viscous effects,
a boundary layer will be formed on the wall. At some infinite distance
downstream the boundary layer reaches the center of the pipe. Shortly
thereafter the velocity at the center of the tube reaches 99 percent of
the maximum of its final parabolic profile. This constitutes the en-
trance length, L The flow from then on is considered to be fully de-
veloped and is assumed to follow Poiseuille eauations. According to
\'lhite (1974), such condition is reached when
R./D .. 0.08 Re + 0.7 (3.5.1)
The volume of flow must be the same for every cross-section of the
rigid tube. Consequently, within the entrance length, the decrease in
the rate of flow near the wall which is due to viscous effects must be
352 C.M. Rodkiewicz
compensated by a corresponding increase near the axis. Thus the boundary

layer in this case is formed under the influence of an accelerating ex-
ternal stream. Such a steady state flow will never separate from a reg-
ular wall and the shear stress at the wall will transform asymptotically
from its initial value to the value which would exist if Poiseuille flow
could be established. In the pulsatile case adverse pressure gradients
may be present and, depending on fluctuation frequency and amplitude, re-
versed flow could appear. The associated minimum wall shear stress would
be equal to zero and the wall drag would periodically change its direction.
In view of the above it may be recognized that in the large arteries,
where blood may be assumed to behave like a Newtonian fluid, the fully
developed flow conditions most likely does not exist.
3.6 Influence of the Blood Cells

If one considers that in blood there would be normally more than
5 x 10 3 million red cells per 1 m~ occupying 40 to 45 percent of the
total volume, it becomes necessary to investigate the influence of the
presence of these particles on the flow characteristics. Though the other
blood particles, at times, produce important effects in the flow charac-
teristics, the red cells are so numerous that they largely determine the
macroscopic flow pattern. There should be no doubt, that in order to
reach a full understanding of the phenomena the effect of the presence
of the corpuscles should be considered.
Many investigators disregard the action of the individual particles
and consider blood as being a continuous substance and adopt a continuum
model of fluid. Such a model possesses appropriate continuum properties
which are so defined as to ensure that, on the macrosopic level, the be-
havior of the model duplicates that of the real fluid. The mean proper-
ties of the fluid element are in the limit assigned to a point so that
we may ultimately adopt a field representation for the continuum proper-
ties. This is not necessarily always the case in arterial blood flow.
However, at least in the large vessels the assumption of a pure liquid
is a sufficiently close approximation. The continuum assumption is made
particularly inaccurate by the presence of a cell-depleted or cell-free
layer at the wall of blood vessel.
In essence there exists a mechanism which causes radial migration of
suspended corpuscles. This migration is predominantly in the direction
away from the wall. However, migration out from the tube axis has been
also observed. It has been attributed to the fact that a cell may possess
a degree of rigidity. In such circumstances, depending on the magnitude
of the Reynolds number, there appears to exist an equilibrium radial
position at which migration across lines of flow ceases (tubular "pinch
effect").
According to Goldsmith (1972) there are three main effects which
may be noted at high particle concentrations: (a) the velocity distri-
bution is no longer parabolic (the profile becomes blunted in the tube
center where there is a region in which particles move with the same
speed); (b) the particle path has an erratic component in the direction
normal to the flow; (c) particle deformation in blood occurs to a degree
which cannot be attributed to shear alone.
There are indications that similar flow characteristics are present
354 C.M. Rodkiewicz
in pulsatile flows as in steady flow. However, we may expect the mag-

nitudes to change. For example, the "equilibrium position" may move
closer to the wall. Nevertheless, what actually occurs in vivo is still
left to speculation.
3.7 Arterial Pressure-Flow Relationship

The arterial pressure-flow relationship is the central problem in
haemodynamics. The driving force is the pressure gradient. Most of the
time fluid will flow in the direction of favorable pressure gradient
(decreasing pressure). However, at times it is possible for the fluid
to continue flowing against adverse pressure gradient (increasing pres-
sure}, until the kinetic energy of a particle in that direction is de-
pleted. That is to say that the flow curve may be lagging behind the
pressure curve.
The timing of flow reversal was shown by D.A. McDonald (1955). Fig.
3.7.1 reproduces a volume flow curve and the pulse pressure in the femoral
artery of a dog. The femoral artery was chosen because more reports were
available on the flow pattern in this artery of the dog than on any other.
Pressure was measured by the use of capacitance manometers, recording
through a tube inserted into branches of the femoral artery and adjusted
so that the ends lay flush with the wall of the main vessel. Direct
measurements of flow have been made by following the movement of injected
bubbles of oxygen recorded by high-speed cinematography. The bubble fills
the tube and travels at the mean velocity.
It is shown in Fig. 3.7.1 that the maximum favorable pressure gra-
dient is reached during the rising phase of the pulse wave at point a.
b
0
:I:
E
E
en
:::::,.
E PI
::::J
en
en
~ ....
CD
u:: a..
CD
!!l
::::J
ll.
Fig. 3.7.1 The pulse pressure and the volume flow curves in the femoral
artery of a dog. After McDonald, 1955. With permission.
However, due to fluid inertia and possibly wall effects the. maximum for-
ward rate of flow appears a little later. From point a to point b the
favorable pressure gradient decreases from its maximum value to zero,
which sets demand on the fluid to decelerate. Between points band d the
adverse pressure gradient goes from zero to zero with its maximum value
at point c. This pressure gradient is present for a sufficiently long
time to reverse the flow. Here again, the maximum negative flow is de-
layed with respect to the maximum point of adverse pressure gradient.
Similar response of the flow curve to the pulse pressure curve is for the
d-e-f secondary diastolic pressure hill.
Curves such as those shown in Fig. 3.7. 1 may be represented mathema-
356 C.M. Rodkiewicz
tically by a series of sine shape waves which are harmonics of a funda-

mental wave. Individual components do not necessarily begin their cycle
at the zero point. Each pressure gradient term generates the correspond-
ing flow term which is lagging in phase. McDonald (1955) presented a
graphical analysis of one such correlation according to the derivations
of Womersley (1955). The sum of the first four Fourier components of the
pressure gradient and the corresponding sum of the flow terms are repro-
duced in Fig. 3.7.2. The harmonics were of the form Mcos(nt + ~). Each
pressure gradient wave generated a corresponding flow curve also of the
,-
1 \
\ Pressure Gradient
I
I
\
\
\
\
Fig. 3.7.2 The sum of the first four harmonics of the pressure gradient
and the flow. After McDonald, 1955. With permission.
sine wave form but lagging in phase. Qualitative comparison of Fig. 3.7. 1
and Fig. 3.7.2 indicates some disagreement in the latter part of the cycle.
However, inclusion of the fifth and sixth harmonics would move the two
corresponding curves significantly closer.

In conclusion it was stated that in the femoral artery the flow
oscillates in the.same way as the pressure gradient but with a phase lag
which varies throughout the cycle. A large forward flow (7 to 15 times
the mean flow rate) occurs during systole followed by a smaller back-flow
phase and a subsequent forward flow during diastole.
4. FLOW IN CURVED PASSAGES

4.1 Introduction
Flow through curved arteries has been a subject of much concern in
blood flow studies related to the atherosclerosis. Many parts of the
human arterial tree are curved with various different radii. Martin
(1960) reports the existence of S-shaped arteries in the vertebral
cerebral system. Also it may be appropriate to refer here to the work of
Mitchell and Schwartz (1965) who indicated that different segments of the
same artery show considerable difference in disease severity. In parti-
cular, in the ''tortuous" arterial segments striking increases in severe
disease have been found.
When a fluid flows through a curved passage a secondary flow occurs
in planes perpendicular to the axis of the tube. For a given velocity
profile some fluid particles possess a higher velocity than others. The
centrifugal force created by motion around the bend, then, is larger for
the faster moving particles than for the slower ones, leading to the
emergence of a secondary flow directed outward in the centre and inward
at the wall. Such flows were originally studied by Dean (1927), White
(1929), Adler (1934), Nippert (1929), and Richter (1930). These experi-
358 C.M. Rodkiewicz
mentally show that the influence of curvature is stronger in laminar than

in turbulent flow.
4.2 Fully Developed Flow in a Curved Pipe

At this time we will involve ourselves with the presentation of
Cuming (1952). The right-handed orthogonal coordinates system has been
Fig. 4.2. 1 Orthogonal System of Axes
Upper Wall
OuterWall lnnerWall
---r-------+.~-----+-------------+C
1+--..,.....:;._-1 /K-----l)J;~I
Lower Wall
Fig. 4.2.2 Elliptic Section

Flow in Curved Passages 359
used in connection with the assumed elliptical cross-section (2a and 2b

major and minor axes, respectively): 0 ~along the major axis; 0 n along
the minor axis and 0 ~ along the curved central axis of the passage of
the curvature K (see Fig. 4.2.1 and 4.2.2.) The corresponding velocity
components along these axes were taken as u, v and w, respectively.
Equations of motion could be presented in the dimensionless form by
the following substitutions: ~=ax, n = Aay (A= b/a), ~ = az,
+ + 2 2
u+ = au/v, v+ = av/v, w = aw/v, p = a p/(p v ). Substitution into
Eouations (2.2. 1) through (2.2.4) yields
(4.2.1)
+ + + +2
+ ~ + y_~ _ Kaw (4.2.2)
u ax A ay 1 + Kax
(4.2.3)
+ +
+ Ka (~ _ l ~)
1 + Kax ax A ay
+ + + + + '"\ + ... 2 +
+ + '!___ + KaU +
u ax
3W
A ay
~ W
1 + KaX
__ ...----:--'---- ~
+ KaX az ;;z-
a W
(4.2.4)
2 + + 2 2 +
+-1 aw + Ka aw Ka W
A2 7 + KaX ax - (l + KaX)2
Approximate solution of Equations (4.2.1) through (4.2.4) have been

obtained by Cuming (1952), by letting
360 C.M. Rodkiewicz
+ 2 I.
u = KaRe u
+ 2 I
v = KaRe v
+ _ 2 I II
w = (w + KaRe w + Kaw ) Re/2
+ - 2
p = p + KaRe p
I
Re = - A2 ~
+ A2 az
where p and ware the pressure and axial velocity as indicated in Section
3.2 (flow through a straight elliptic pipe), respectively; and where Re
is the Reynolds number for the flow also through a straight pipe. Equat-
ing coefficients of the zero and first power in K, one obtains
(4.2.5)
I 2 I 2 I
= ap av 1 a u
0
- I ay
1 +
7 -I axay (4.2.6)
(4.2.7)
2
4) x + !_·w
,, II II _
0 = _ 2 (1 + + 1 a w + aw
A.. ax2 "'17 ax
(4.2.8)
I I
~ + .!_~v- = 0 (4.2.9)
ax A ay
Eliminating pressure between Equations (4.2.5) and (4.2.6}, and in-

troducing stream function via Equation (4.2.9), velocity components u and
v are obtained, namely
2
U =~
.L
(1 - x2 - Y2 ) [ (1 - X2 - y 2) ( C1 + C2 X2 + 3 C3Y 2 )
A
(4.2.10)
2 2
(1 - x - y ) [2(c 1 + c2 x + c3y )
2 2
(4.2.11)
- C2 ( 1 - X 2 - y2 ) ] xy
where
c1 = A4(375 + 820A 2 + l,ll4A 4 + 212A 6 + 39)._8 )/360(5 + 2A 2 + A4 ) G(A)
c2 = -A 4(75 + 2A 2 + 3)._ 4)/360 G(A)
c3 = -)._ 4(15 + 26)._ 2 + 39)._ 4)/360 G(A)
G(A) = 35 + 84A 2 + ll4A 4 + 20A 6 + 3)._ 8
Equations (4.2.10) and (4.2. 11) indicate a set of streamlines as in-

dicated in Fig. 4.2.3. The secondary flow ·consists of two opposed vortex
Outer Inner
Wall Wall
Fig. 4.2.3 Secondary flow

362 C.M. Rodkiewicz
motions in the top and bottom halves of the tube. In addition it has
been found that the pressure along the minor axis is equal to the arith-
metic mean of the pressures at x = 1 and x = -1. Furthermore, it has been
demonstrated that for large and small magnitudes of the A ratio the sec-
ondary flow diminishes from its maximum value at A = 2.2. Variation of
the velocity component u across the central plane (-1 ~ x ~ 1, y = 0) for
various A ratios is shown in Fig. 4.2.4.
Ka = 0.2
Re = 100
Fig. 4.2.4 Variation of Velocity u+ for various A Ratios

(- 1 < x < 1, y = 0) After Cuming, 1952
With the permission of the Controller of Her Majesty's
Stationery Office.
Substitution of the velocity w,as expressed in section 3.2, into
I II
Equations (4.2.7) and (4.2.8) yields solutions for w and w . These in-
dicate that, to the degree of approximation considered, the axial velocity
I
component is modified by two curvature terms: the first of these w ,

associated with the square of Re, causes the velocity to increase in the
outer half of the bend and to decrease in the inner half of the bend (ef-
II
feet associated with flow in a curved pipe); the second term w has re-
versed influence (effect associated with the flow in a curved channel).
For values of A around unity the influence of the first term is predomi-
nant. However, with increasing A the second term begins to dominate.
The associated results are shown in Fig. 4.2.5. It can be seen that as A
increases the point of maximum axial velocity shifts from the outside of
w+
Fig. 4.2.5 Variation of Velocity w+ for Various A Ratios

(- 1 < x < 1, y = 0) After Cuming, 1952
With the permission of the Controller of Her Majesty's
Stationery Office.
364 C.M. Rodkiewicz
the bend over to the inside of the bend. For the special case when A
ratio is unity we obtain the solution for the flow through a curved pipe
of circular section.
The above solution represents the first order modification to the
zero order approximation which is the same as the flow in a straight tube.
Ito (1950) obtained the second order modification which is presented in
Fig. 4.2.6. In practice the secondary flow of this order is added to
the secondary flow of the type shown in Fiq. 4.2.3.
Fig. 4.2.6 The Calculated 2nd Approximation Streamlines. After Ito,

1950. With permission.
4.3 Unsteady Flow in Curved Rigid Tube

An interesting phenomenon was reported by Lyne (1970) who studied
the fully developed unsteady viscous fluid flow in a curved rigid tube of
a circular cross-section. Of special interest was the secondary flow in-
duced in the plane of the cross-section of the tube. In order to simplify
the problem, the radius of pipe curvature was assumed to be large in re-
lation to its own radius, and the pressure gradient was assumed to be
sinusoidal in time with zero mean.

Lyne defined a nondimensional parameter S given by
S2 = 2v/(na 2) (4.3.1)
which may be interpreted as the ratio of Stokes' layer of thickness

(2v/n) 112 to the radius of the pipe (note that S = ~a). It was found
that for the sufficiently small values of S (this implies that the viscous
effects are confined to a thin layer next to the wall, while the core of
the flow is inviscid) the secondary flow in the central zone of the tube
is in the opposite sense (see Fig. 4.3.1) to that predicted by the steady
state analysis (see Fig. 4.2.3). That is the secondary flow due to the
centrifugal effect is confined to the thin boundary layer region, which
in turn drags the fluid in the central region.
Outer Inner
Wall Wall
fig. 4. 3.1 The Stream1 ines for Small s. After Lyne, 1970. With
permission.
366 C.M. Rodkiewicz
4.4 Unsteady Flow in Thin-Walled Curved Elastic Tube

In the study of etiology of arterial diseases it is relevant to
understand the mechanics of pulsatile flow not only in rigid tubes but
also in elastic tubes. Chandran et al (1974) presented their investiga-
tion on the oscillatory flow in thin-walled tubes which correlates with
findings of Lyne (1970). This work is the object of the present section.
Equation (2.3.1) through (2.3.4) have been written in the toroidal
coordinate system and supplemented by the corresponding equations of
motion for the tube. To include the effect of the surrounding tissues
on the arteries, a spring-mass restraint has been introduced. Due to the
assumption of small radial tube displacements the boundary conditions
were linearized by equating the velocity components at the undisturbed
tube radius rather than at the instantaneous tube radius.
An order of magnitude study has shown that a number of terms in the
governing equations could be neglected. Pressure disturbance was assumed
to be sinusoidal in time and the equations were linearized before a wave
propagation analysis was attempted. Furthermore, the perturbation tech-
nique (perturbation parameter was equal to aK) was used to introduce the
small curvature effect on the solution for the similar flow through a
straight elastic tube. The solutions were evaluated numerically.
The results were presented for a small perturbation parameter,
namely aK = 0.1. A comparison was made of the axial velocity profile of
the oscillatory flow in a straight elastic tube with that of curved elas-
tic tube. In conclusion it was reported that the maximum axial velocity,
for the case of time dependent flow in curved tube may be shifted towards
the center of curvature. This shift is opposite to that for the steady
flow in curved tubes (see Fig. 4.2.5).
4.5 Entrance Arc

The fully developed flow in a curved passage is defined as the
region in which the velocity no longer depends upon the distance in the
axial direction. It is the outcome of the flow reconfiguration which
takes place within the entrance arc. In a curved tube, and for a steady
state, this entrance arc is measured by the number of degrees around the
curve from the tube entrance to the location where the fully developed
flow has established itself. Scarton et al (1977} quote the following
formula for the entrance arc:
49 (Re) 113 (a K) 112 (4.5.1)
It may be noted that, for example, in the human aortic arch the en-
trance arc, most of the time, will be in excess of the angular distance
from the aortic root to the branching site at the upper surface of the
aortic arch. It will be also recognized that in our arterial system
there are present numerous successive branching sites. In addition the
axis of curvature of practically any arterial tube is not necessarily in
the same plane. Consequently, it is most unlikely that in the arterial
tree of the human the fully developed flow could exist.
4.6 Laminar Flow Downstream of a Bend

Measured and calculated velocity profiles were reported for the
water flow downstream of a 15 degree bend in a round glass tube at a
368 C.M. Rodkiewicz
o oo Plane
oo •30° Plane
30°
esoo Plane
t:..90° Plane
Jl
b. b. b. b. b.'
b. •••
• 10
b. • e
b. .• e
e
~:..•
• ee b.
•e 1.00
.!
• e !
e
••0
.g
b.
0.50 •e
b. b.
0 0
0
[ r/R]
Fig. 4.6.1 Experimental Velocity Profile at Initial Section

After Gosman et al., 1975. With permission.
Reynolds number of 300 by Gosman et al (1975). As water is a Newtonian

fiuid and blood flowing in the large arteries behaves like a Newtonian
fluid, this investigation is pertinent to the present subject.
The measurements were obtained using laser-Doppler anemometry and
provided information regarding the development of the axial velocity from
the exit of the bend to the fully developed situation. Initial flow at
the entrance to the bend was fully developed. The results are shown in
Fig. 4.6.1 through 4.6.3. The calculation procedure was based on a finite-
difference solution of the three-dimensional, steady, boundary-layer form
of the Navier-Stokes equations given by Eqn. 2.2.1 through 2.2.4. Com-
parison between the measurements and calculations revealed close agreement
and indicated the high precision of both calculation and measurement tech-
niques. The maximum deviation of the calculated velocity values from the
measured quantities was 2.5%.
Fig. 4.6.1 shows the measured velocity profiles across four diameters
at the initial station which was located approximately 5 tube diameters
downstream of the centre of the bend. These values were used as initial
conditions for the solution of the appropriate governing equations. Note
the double peak velocity profile for the plane perpendicular to the plane
of the bend. Fig. 4.6.2 shows velocity values computed and measured across
diameters normal to the plane of the bend and at six consecutive axial
locations. These profiles are symmetric and indicate that the fully-
developed condition was reached within 20 tube diameters of the centre of
the bend or within 15 tube diameters of the initial station. The velocity
profiles computed and measured in the plane of the bend are shown in
370 C.M. Rodkiewicz
t 1.0
~ 10.8
0.6
0.4
0.2
or-~~~~~~~~~
-0.2
-0.4
-0.6
-1.0 L-..---o=:::::....:...-oc::--O::::..__--o-::,__--cF--
5.0 10.0 15.0 20.0
[ x/D] = 0 2·0
Fig. 4.6.2 Flow Development in Normal to the Bend Plane

After Gosman et al., 1975. With permission.
-1.0
0~~4-~~~~4-~
1.0
0.8
0.6
o Experiment
e Predictions
Fig. 4.6.3 Flow Development in the Bend Plane. After Gosman et al.,
1975. With permission.
Fig. 4.6.3 As could be expected the maximum of the velocity is shifted

towards the outer wall of the bend. Again here the fully-developed flow
is reached within 20 tube diameters of the centre of the bend.
4.7 Drag on the Wall

A vector addition of velocity components yields the magnitude and
direction of the absolute velocity of a participating fluid particle.
These give the velocity profiles and in turn the shear stresses which can
be computed with the aid of the expressions provided in Section 2.4.
Integration of the shear stresses, over some specified area, produces the
local time dependent drag which at the wall assumes the following form
(4.7.1)
where 'o is the shear stress evaluated at the wall and A is the wall area.
To overcome such a flow resisting drag a driving force in the form of a
corresponding pressure gradient must be applied.
5. FLOW IN JUNCTIONS
5.1 Simple Btfurcatton
The manner in which the flow divides at a simple arterial junction
(where a si:ngle branch leaves the straight-through parent tank) was
studi.ed in terms of the si.'gni.fi,cant di:mensionless parameters by Rodkiewicz
and Howell (1971). They found that, for certai,n values of these para-
meters, more flutd goes into the si,de branch than straight through. In
order to have a better understanding of this phenomenon and in order to
fi.'nd wh.y and when the mass flow ratio y (the ratio of the rate of flow in
372 C.M. Rodkiewicz
-----t-
rD
~~_ __._j_
(a) With Sharp Inside Edges
~~ e
t
----r- D
__,~_ __._l
(b) With Rounded Inside Edges
Fig. 5.1.1 Experimental Bifurcations

low inJunctions 373
he side branch to the flow rate in the main branch} becomes greater than
ne, the study of the steady-state flow characteristics in an arterial
unction has been undertaken by Rodkiewicz and Roussel (1973}. Their
ange of study was: 1000 ~ Re ~ 5000 and 0.40 ~ B ~ 1 (B = d/D; see Fig .
. 1.1}. In order to enable observations and photography, the bifurcations
ere made of acrylic. Sixteen junctions were tested; e= 8/90 = l/3,
/9, 7/9, 1, and S = 0.4, 0.6, 0.8, 1.0. The isolated basic geometry
epresents the usual idealization of the actual arterial flow problem.
lthough the bifurcations inside the body generally have a tapered main
essel and a possibly more gradual curvature to the side branch, these
dditional variables were not considered. Hydrogen bubble technique was
sed to visualize the flow distribution within the bifurcation. The
lectrolysis direct current was brought to the electrodes from the gener-
tion unit which allowed the operator to have control of the frequency and
he duration of the pulses producing the bubbles.
The dependence of the mass flow ratio y on the pertinent dimension-
ess parameters is reproduced in Fig. 5.1.2 through 5.1.5. These graphs
ndicate that the y ratio decreases when the Reynolds number increases.
t could be attributed to the inability of the fluid to negotiate the
urn, due to an increase in its momentum in the mainline direction. The
raphs also give the variations of y with diameter ratio B and the angle
atio e.
It has been established that, for the equal resistance discharge, a
hange in the y ratio is due to a variation in size and location of two
ndependent (one in the main branch and one in the side branch) separation
374 C.M. Rodkiewicz
... • p = 1.0
• ...• •p=O.B
•p=0.6
......
D
TP=0.4
o p = 1.0 Reproducibility Test
·~...
il
...
.
\ ~
•
'' A...
.•••
0 1.2 ~.
...... ·~~~
+"'
a:s
a: • ........ ... ......
~ ••••cf'll...
?--
1.0
•• ...... ...
... ...o-•
~
• •• ••
0.8 • •• ••
0.6
0·4o..__---'---l-'oo-o--'-2__.oo_o___.__3o. . .o_o__.__4o. .o_o_...__s__,ooo

.
Reynolds Number
Fig. 5.1.2 Mass Flow Ratio y Versus Reynolds Number Re as a Function

of S for 8 = 1/2 (8 = 30 deg)
Flow in Junctions 375
1.4
.....
'.. \._. . ......._
1.2
•• "'4
·~l
......
... ...............:
1.0
... • •
••
JttliiiA ...... •-r·~c-
- -
0 0.8 •••• •••••

.....
tU
ex:
,....
0.6
0.4
• {3 = 1.0
•{J=O.B
0.2 •P=0.6
'Y p = 0.4
1000 2000 3000 4000 5000

Reynolds Number
Fig. 5.1.3 Mass Flow Ratio y Versus Reynolds Number Re as a function

of B for 8 = 5/9 (e = 50 deg)
376 C.M. Rodkiewicz
OL-~--~~--~~--._~--._~~
0 1000 2000 3000 4000 5000

Reynolds Number
Fig. 5. 1.4 Mass Flow Ratio y Versus Reynolds Number Re as a Function

of 8 fore= 7/9 (e = 70 deg)
1.4
1.2
....
•
1.0
•
. ,....•'"""" . .•...
..... ····-····· ....
• •fll.llf• ....,
••
• ......
0.8
·-
.....0as
a:
~
~ ...
>-- 0.6
~
~~
~~
~
~~~~ .....~~~
~~
~~~ ~
~ ~
0.4
•P= 1.0
•p=O.B
0.2 •P=0.6
~ p=0.4
1000 2000 3000 4000 5000

Reynolds Number

of S for 8 (e = 90 deg}.
378 C.M. Rodkiewicz
Top View
Section A-A Section 8-8
Separated
Separated Region
Region
Fig. 5. 1.6 Physical description of the flow. After Rodkiewicz and

Roussel (1973).
regions. This is shown in Fig. 5.1.6. Defining the thicknesses of the

separation regions by Td and To (Fig. 5. 1.6), one can observe that, when-
ever TD increases, Td decreases, and vice versa (see Fig. 5. 1.7 through
5. 1.9). Since growth of the separation region in the main branch re-
stricts the flow in this branch, the mass flow ratio increases. The sep-
aration region in the main branch was found to grow when the Reynolds
number decreases (Fig. 5. 1.7), when the diameter ratio increases (Fig.
5. 1.8) and when the angle of branching decreases (Fig. 5. 1.9). At lower
Reynolds numbers this separation region is so thick that the y ratio be-
comes greater than 1, i.e. more fluid goes into the side branch than via
the straight-through main branch.
Inside the separation region of the main branch vortices are gen-
erated. This is shown in 5.1.6 and 5.1. 10. The vortices have been ob-
served to be "banana-shaped". This shape enables, in the early life of
a vortex, some particles to be transferred from the main-branch separation
region into the side branch, as sketched in Fig. 5.1.6. Inside the side
branch the flow becomes double-helicoidal, which is typical of flows in
curved pipes.
The transition from laminar flow to turbulent flow may occur for a
Reynolds number greater than 2000. The turbulence, flattening the vel-
ocity profile in the main branch, enables more fluid to flow into the
side branch than in the laminar case for the same Reynolds number. This
increases the mass flow ratio y. The effect is particularly pronounced
for large values of 8 and S, as can be seen in Fig. 5. 1.5.
It may be of interest to mention that Roussel et al. (1973)
380 C.M. Rodkiewicz
(a) Re = 1400 (b) Re = 1970
Fig. 5. 1.7 Reynolds number dependence (S = 1.00, 8 = 1/3, e = 30 deg)
(a) 13 = 0.40 (b) s = 0.60

Fig. 5. 1.8 Diameter ratio dependence (8 = 5/9, e =50 deg, Re = 1600)
(a) a=S/9 (b) a=7/9
Fig. 5.1.9 Angle of branching dependence 13 = 0.80, Re = 1160

Fig. 5. 1.10 Vortex formation (e = l/3, e = 30 deg, s = 1.00, Re = 1520)
investigated the question of the importance of the edge roundness at the

junction (see Fig. 5. 1.1). His range of study was 1000 2 Re ~ 5000, and
it was found that earlier conclusions (Rodkiewicz and Roussel, 1973)
were valid whether the bifurcation had sharp or rounded edges. However,
for Reynolds numbers less than or equal to 2000 (see Fig. 5.1. 11 and
5.1.12) y is slightly higher when the inside edges of the bifurcation
are rounded. This is due to the fact that the point of separation in
the side branch is pushed slightly downstream in the case of the rounded
edge bifurcation.
It has been emphasized (Rodkiewicz and Hsieh, 1975) that flow
characteristics for free discharge cases have significance predominantly
as an experimental base state. A full understanding of such a state is
helpful in analyzing the effect of various kinds of added resistances
downstream of the junction. Such is the case in blood flow where the
junction flow properties are usually controlled by the system which
follows the dividing element. Rodkiewicz and Hsieh (1975) demonstrated
one of these cases by extending the straight-through section on one of
382 C.M. Rodkiewicz
1.8 ...
•• Bifurcation
.tJ • {3 = 1.0 0
1.6 I}
Round .& {3 = 0.8 a Sharp
•fl
@> 4:1
Edged •
...
{3=0.6 0 Edged
{3 = 0.4 v
ce ...
1.4 ai
0" ... ~
·~l!
~ aa~i
-0 1.2 ... \ +.a.&·~ i~
as 0~ ~
a: oo• ~ t!t'il•u~
%• ...~
""'" 1.0 ... 0~~ ~!a~
t.o
0.8
~
w
.... C)~
~~~~
' ~
v..,
Vv ......
0.6 Vv~..-
v~v•
...... v
0.4
0 1000 2000 3000 4000 5000
Reynolds Number

of S for 8 = l/3 (e = 30 deg)
1.4 Bifurcation
• f3 = 1.0 D
Round • f3 = 0.8 Sharp

£..
1.2 • Edged • p = 0.6 o Edged
T f3 = 0.4 V
1.0
...a:s
0 0.8
a:
~
0.6
0.4
0.2
o~~--~~--~--~~--~~~~~
0 1000 2000 3000 4000 5000
Reynolds Number
Fig. 5.1.12 Mass Fow Ratio y Versus Reynolds Number Re as a Function

of a for 6 = 1 (e = 90 deg}
384 C.M. Rodkiewicz
their junctions, i.e. adding resistance to the main flow. It was found
that though there was significant shift in the pressure magnitude, the
basic character of the curves remained the same. Also, the separation
regions did not vanish. It is unfortunate that the available data was
for a Reynolds number which is much higher than the upper limit of the
desired range. However, in principle, arterial pressure distribution
within the junction should be analogous.
It was mentioned in the Introduction that atherosclerosis has been
observed to occur predominantly at specific sites in the arterial system.
Wesolowski et al (1962, 1965) point out that the atherosclerotic lesions
are in the regions of the turbulent eddies which are in the vicinity of
arterial bends and junctions. It is thought by Rodboard (1956),
Downie et al (1963), and Fry (1968) that deposition of blood particles
occurs at these sites, which locations have large variations in the local
turbulent flow shearing stresses. Particularly the work of Fry (1968,
1969) shows the damage done to the endothelial cells by the high turbu-
lent flow, which was generated by insertion of a plug channel into the
descending thoracic aorta of a dog. Lynn et al. (1970) describe the
velocity and shear fields in the arterial bifurcation. Fox and Hugh
(1966) made steady state observations based on the open channel water
table models: the right angle branch, the "Y" junction, the divergent
channel, and the curved channel segments. Lee and Fung (1970) examined
the steady laminar flow through a tube with an axisymmetric constriction.
Just downstream of the constriction they reported a region of reverse
flow. This region was thought to be a significant factor in the deposi-
Flow inJunctions 385
tion process which has been studied by Friedlander and Johnstone (1957).
One of the good examples associating fluid flow distribution with the
atherosclerotic plaques distal to orifices of the intercostal arteries in
dogs was presented by Texan (1972}.
5.2 Symmetric Y Junction

There have been numerous studies dealing with the symmetric Y jun-
ctions. One of these was done by Kandarpa and Davids (1976). Though the
focus of their work was on the two~dimensional flow, the flow field they
presented (velocity, pressure and shear stress distribution) represents
qualitatively the situation which one may expect in the circular cross-
section bifurcations. In fact, it has been observed by Crow (1969) that
in principle the flow patterns are similar in circular and rectangular
geometries. However, it will be recognized that in the two-dimensional
passages one is not able to observe the dramatic doublehelicoidal patterns
and the double peak velocity profi'les which may be present in a plane per~
pendicular to the plane of the Y junction (Brech and Bellhouse, 1973).

The typical steady state velocity profiles reported by Kandarpa and
Davids (1976) are shown in Fig. 5.2.1. A short distance into the tapered
segments, along the outer wall, the boundary layer separates and then re-
attaches. Variation of wall shear stress with the Reynolds number and
with the distance is shown in Fig. 5.2.2. It will be noted that there is
a change of sign of the outer wall shear stress as the flow goes through
the separation point and then when it reattaches. Along the branch inner
wall, a peak shear stress occurs at the leading edge (the smaller the bi~
furcation angle the greater the shear stress) and decreases to a value
386 C.M. Rodkiewicz
Separation Region
12
Fig. 5.2.1 Separation region in bifurcated channel. After Kandarpe &

Davids, 1976. With the permission of Pergamon Press Ltd.
Reynolds No.
bi.
-2100
-1.0 ---1050
- - - - 525
o·
,...·
X
CiJIN -0.5
~li
CD'
-en
.s=
Distance em
Fig. 5.2.2 Variation of wall shear stress with Reynolds number along
outer (a) and inner (b) walls. After Kandarpa & Davids, 1976.
With the permission of Pergamon Press Ltd.
which is attained along both walls. Modelling was made so that branch
dimensions were approximating those of the iliac of a dog. Prediction of
the leading edge shear stress was around 55 dyne/cm 2 for an entrance Rey-
nolds number of 1050.
5.3 Note on Non-Rigid Wall

If we now relax the assumption of rigidity of the wall the interior
arterial surface will be subjected to pulls back and forth at the loca-
tions of the shearing stresses. This action, say around the indicated
separation arcs (see Fig. 1.3.1), will be repeated a great many times
(converging shear stresses). Restoring forces will be effective as long
as the tissue is resilient. However, if for some reason the tissues show
fatigue, creep could begin. One could draw a parallel here of a sagging
face. The effect of it would be creation of a ridge which is dramatically
illustrated in Fig. 5.3. 1. This immediately extends the separation region
FREE
STREAM
ELASTIC
WALL
Fig. 5.3.1 Diagrammatic Representation of the Effect Due to Convergi.ng

Shear Stresses
388 C.M. Rodkiewicz
and initiates the depository tendencies, should such tendencies not be

present before.
The diverging time and position dependent shear stresses at the
stagnation arcs (see Fig. 1.3.2) may bring the same effect, but on each
side of the arc. The tissue at the stagnation arc will be in tension at
all times and under the momentum flux of the impinging fluid molecules.
Possible wall deformation is indicated in Fig. 5.3.2. It may be noted
POTENTIAL
FREE SEPARATION SITES
STREAM
Fig. 5.3.2 Diagrammatic Representation of the Combined Effect Due to

Fluid Impact and Diverging Shear Stresses
that should such deformation become permanent additional separation sites

may be created which may perpetuate the phenomenon irreversibly.
6. AORTIC ARCH
6.1 Aortic Arch Flow Field
Meisner and Rushmer (1962) reported steady state flow patterns ob~
served in a flat rectangular model of the outflow tract of the left ven-
tricle with rigid 11 0pen valves 11 • The flow was essenttally two-dimensional.
Existence of some separation regions was indicated. The knowledge of the
Aortic Arch 389
RIGHT
SIDE
Fig. 6. 1.1 Proposed Shape of the Aortic Arch Model
separation and stagnation regions within the aortic arch, the site which
according to Spain (1966) is among those positions which are favored by
the atherosclerotic plaque, appeared to be important. Rodkiewicz (1975)
determined where such regions can be expected in the system with specific
conditions and configuration of the aortic arch of the human, and fol-
lowed it with the experimental localization of the early atherosclerotic
lesions in the rabbit. The shape of the open channel model of the aortic
arch used (AA = Ascending aorta, BA = Brachiocephalic artery, LA Lower
aortic arch, LC = Left common carotid, LS = Left subclavian, LTA Lower
thoracic aorta, RC = Right common carotid, RS = Right subclavian, UA =
Upper aortic arch, UTA= Upper thoracic aorta) is shown in Fig. 6.1 .1.
Research into the detailed flow patterns in the human aortic arch is
difficult to perform in vivo. Similarly, due to the complexity of the
390 C.M. Rodkiewicz
geometry, the flow through the region does not lend itself easily to
theoretical studies. Thus it appears fruitful to gain the basic knowledge
of the flow in such a branching system by using physical models. Several
studies have followed this course. In the Rodkiewicz (1975) study, the
model has a geometry based on the physiological measurements, even though
it is two-dimensional. Furthermore, the flow parameters could be con-
trolled independently. Thus, it is believed that this model simulated
some of the flow phenomena which occur in the aortic arch region.
At this point it is convenient to present the average shape of the
aortic arch of a rabbit which is shown in Fig. 6.1.2. It has been deter-
mined by superimposing enlarged photographic transparencies of a number of
rabbits studied in the project (Rodkiewicz, 1975). A comparison of Fig.
6.1.1 and Fig. 6. 1.2 indicates a great similarity in the geometry of the
LTA
Fig. 6. 1.2. Aortic Arch of the Rabbit.

Aortic Arch 391
Fig. 6.1.3. Experimental Flow Pattern in the Aortic Arch Model
aortic arches of humans and rabbits. Consequently, flow similarities

should also exist. In addition, under certain laboratory conditions
rabbit lesions imitate relatively quickly the advanced stages of human
atherosclerosis (Constantinides, 1965).
Experimental results related to the flow pattern are diagrammatically
shown in Fig. 6. 1.3. At each of the upper aortic branch (BA, LC, LS)
entrance sites there is one stagnation point (St-2, St-3, St-4), and one
separation point (Se-3, Se-4, Se-5) followed by a separation region where
the fluid turns back in the layer next to the wall. Separation and stag-
nation points are on the right and left walls, respectively. These points
actually represent segments of the circumferential length in the 3-dimen-
sional model. The phenomenon was isolated as shown in Fig. 6. 1.4. The
392 C.M. Rodkiewicz
dark areas indicate fluid that is free of the suspended particles which
here are in the process of evacuation. The white areas represent the
separation regions. One should study Fig. 6.1.4 in conjunction with
Fig. 6. 1.3 where all the regions are indicated graphically.
In the 2-dimensional model an additional very shallow separation
region, Se-2, has been observed on the outer wall and at the end of AA.
Furthermore, there is a stagnation region St-1 at the junction of RS and
RC. It is interesting to note that the generation and shedding of vor-
tices is present at all the separation regions. These are vividly de-
picted in Fig. 6. 1.4 but not indicated in Fig. 6.1.3.
Opposite to the upper branching sites, at the maximum curvature of
Fig. 6. 1.4 Selected Flow Photographs

Aortic Arch 393
this two-dimensional model, there appears to be a relatively large sep-

aration region. It begins at the separation arc Se-1 at the end of AA,
and is terminated at the arc of reattachment within UTA. The reattach-
ment site can be seen readily in the photographs of Fig. 6.1.4. In steady
flow, for a given Reynolds number, reattachment arcs oscillate to and fro
as the vortices are swept downstream. The vortex formation of this sep-
aration region is particularly well seen in Fig. 6.1.4. These reattach-
ment arcs are rather difficult to locate, particularly in the side
branches where the separation regions are shallow. The flow and the shear
stress configuration in the reattachment sites are similar to those shown
in Fig. 1.3.2.
With pulsatile flow the configuration, in principle, is the same.
Comparison of steady and pulsatile flow in a 2-dimensional model indicates
that the separation and stagnation regions exist in similar locations
most of the time. However, the location of the separation arcs now os-
cillate and the flow distribution within the region itself undergoes cyc-
lic modifications. It was interesting to note that when the instantaneous
net flow rate was decreasing, the layers next to the wall, within the
separation regions, have been seen to move·very swiftly in the upstream
direction.
At this time it should be indicated that there has been a controversy
regarding existence of the separation region at the maximum curvature of
the aortic arch (i.e. separation region commencing at Se-1). The pre-
sently described observations were made on an open channel model (2-dim-
ensional). However, in a curved convergent tube (3-dimensional), like
394 C.M. Rodkiewicz
the main trunk of the aortic arch, no separation should be present when
there are no branches. On the other hand, sufficiently significant
branching should generate this separation.
This hypothesis has recently been supported by steady and pulsatile
flow tests in a 3~0 model (Pelot and Rodkiewicz, 1982). For steady flow,
there is evidence of separation at the point of maximum curvature when
the Reynolds number exceeds 1000. The separation is dependent on the
presence of the branches which act as diffusers, lowering the velocity in
the arch and causing the adverse pressure gradient required for separation.
For pulsatile flow, the backflow in the separation zone persists through-
out the period. The 3-D model experiments also confirmed the existence
of Se-2, Se-3, St-2, St-3 and St-4 (the other points not being considered).
6.2 Early Atherosclerotic Lesions

It has been postulated (Rodkiewicz, 1975} that the atherosclero tic
· formations within the aortic arch of the human may commence at the sites
of possible separation and stagnation regions. As a consequence of the
geometry, flow, and functional similarity, male rabbits were used in the
experiment.
Animals were kept in separate cages and were allowed water and food
as desired. The basic control diet was "Master Baby Rabbit" food (Maple
Leaf Mills Ltd., Canada) containing, besides minerals and vitamins, 18%
of crude protein and 2.5% of crude fat. To make the "cholesterol test
diet" cholesterol was added to this basic diet, 5 g per 100 g of dry diet.
Control rabbits were fed basic diet and sacrificed for the study of the
aorta after the sampling of blood at the end of four weeks. "Cholesterol
Aortic Arch 395
fed" rabbits were on a cholesterol rich diet for periods of 3, 4, 5, 6, 7
and 8 weeks. At the end of a given feeding period, each rabbit had its
blood sampled from the median auricular artery and then was sacrificed by
the intra-arterial injection of sodium pentobarbital. The aorta of each
animal was then opened and prepared for photography. Serum cholesterol
was determined in the blood of control and cholesterol-fed animals fol-
lowing the method of Abell et al. (1952).
Average daily food and water consumption was respectively, 100 g and
194 ml per animal in control rabbits and 117 g and 195 ml in cholesterol
fed rabbits. In the first four weeks of experiments the body weight of
controls increased by 22.3% and of cholesterol fed animals by 22.8%. In
respect to food and water consumption and body-weight-increase two groups
of rabbits did not differ significantly. The cholesterol rich diet re-
sulted in a significant increase of total serum cholesterol. The average
serum cholesterol was 38 mg/100 ml of serum in control animals. It in-
creased to 1380 mg/100 ml of serum after three weeks of feeding on a
cholesterol rich diet. It averaged 2300 mg/100 ml of serum in animals
fed the cholesterol rich diet for 8 weeks.
The experimental results are reproduced in Fig. 6.2.1 through 6.2.6.
After three weeks of cholesterol rich diet the aortic arches indicated
some shadows in the surface, in comparison with the perfectly clear aortas
of the control rabbits. These shadows, as indicated in Fig. 6.2.1, have
been found exactly in all four stagnation and five separation regions
formed on the open channel model.
At the end of the fourth and fifth week, Fig. 6.2.2 and 6.2.3, it
was found that shadows extended to cover almost the upper half of the
396 C.M. Rodkiewicz
Fig. 6.2. 1. Atherosclerotic Form- Fig. 6.2.2. Atherosclerotic

ations after 3 weeks on the diet. Formations after 4 weeks on the
diet.
IHAOOW
D
rn "-"''X 0 U•• THIC"
SHADOW
D·~ OlS-l'HtCJ:
AI'~~ 030•• THM:.:
-"""X OlO- THICK
Fig. 6.2.3. Atherosclerotic Form- Fig. 6.2.4. Atherosclerotic

ations after 5 weeks on the diet . formations after 6 weeks on the
diet.
Aortic Arch 397
D SHAOOW
D
• . . . .,
AMOl OIJ- IHICl
• .,.Ol 0 - IHICl
o ·~- IHIO<
Fig. 6,2.5. Atherosclero tic form- Fig. 6.2.6 . Atherosclero tic

ations after 7 weeks on the diet. formations after 8 weeks on the
diet.
aortic arch and the rather long and narrow area at the aorta's maximum
curvature. However, at the locations where the shadows originated thicker
layers appear and grow as the time progresses. This can be seen in Fig.
6.2.4, 6.2.5, and 6.2.6 which represent rabbits after 6, 7 and 8 weeks on
the diet, respectively.
One notes that the atherosclero tic formations did not appear at the
sites where separation or stagnation regions were not expected. Further-
more, surprisingly the traces very closely reflect the flow distribution
reported in Fig. 6.1 .4.
In summary the experimental evidence verified that atherosclero tic
formations in the aortic arch of the rabbit commence and develop at the
398 C.M. Rodkiewicz
locations where there are expected separation and stagnation regions.

As a consequence of the geometry, flow and functional similarity, it was
assumed that the atherosclerotic formations within the aortic arch of
the human will begin and develop at similar locations. Furthermore, it
was speculated that the same rules apply to the separation and staqnation
regions of, the various Y junctions.
The above described report refrains from the speculation of what is
atherosclerosis and what is its exact generating mechanism. These
questions merit special and substantially more inter-disciplinary re-
search. It could well be that the phenomenon is the effect of the action-
reaction inter-relation of a number of elements of which only a part of
the full story was isolated and reported.
Abell, L.L., Levy, B. B., Brodie, B.B. and Kendall, F.E. (1952), A simpli-
fi'ed method for· the estimation of total cholestral in serum and demon-
stration of its specificity, J Bto! Chern, 195, 357.
Agrawal, Y.C. (1975), Laser .velocimeter study of entrance flows in Curved
pipes, U. of California Berkeley Report FM-75-1.
Amyot, J.S., Francis, G.P., Kiser, K.M. and Falsetti, H.L. (1970),
Measurement of sequential velocity development in the aorta, ASME
Paper 70-~IA/BHF-13.
Anderson, B. and Porje, I.G. (1946), Study of Ph. Broemser's manometer
theory for oscillations in the aorta, Acta Phy~~oL Seand, 12, 3.
Atabek, H.B. (1964), End effects in Pulhatife stood Flow, Attinger, E.O.,
ed., McGraw Hill, New York, 201.
Attinger, E.O. (1963), Pressure transmission in pulmonary arteries re-
lated to frequency and geometry, C~e R~, 12, 623.
Attinger, E.O. (1966), Hydrodynamics of blood flow, in Advane~ ~n
References 399
Hy~o~cience, vol. 3, Chow, V.T., ed., Academic Press, New York, 111.
Attinger, E.O., Ann~. A. and McDonald, D.A. (1966}, Use of Fourier series
for the analysis of biological systems, &i.ophy~ J, 6, 291.
Austin, L.R. and Seader, J.D. (1973}, Fully developed viscous flow in
coiled circular pipes, AICHE J, 19, 85.
Back, L.D. (1975}, Theoretical investigation of mass transport to arterial
walls in various blood flow regions - I and II, Mathematical &i.o-
~cienc~, 27, 231.
Back, L.D., Radbill, J.R. and Crawford, D.W. (1977}, Analysis of pulsatile,
viscous blood flow through diseased coronary arteries of man, J &i.omech,
10, 339.
Bard, P. (1956}, Medical Phy~-iologu, 1Oth ed., C. V. Mosby, St. Louis.
Batchelor, G.K. (1970}, An Intnoduction to Flu-id Vynamie6, Cambridge Univ.
Press, New York, 148.
Bell, G., Davidson, J.N. and Scarborough, H. (19~5), Textbook on Phy~-io-.
logy and &i.ochem~~lj, 6th ed., E &S Livingston, Edinburgh.
Bellhouse, B.J. and Talbot, L. (1969), The fluid mechanics of the aortic
valve, J Flu-id Mech, 35, 721.
Benditt, E.P. and Benditt, J.M. (1973), Evidence for a monoclonal origin
of human athercsclerotic plaQues, P4oc Nat Acad Sci USA, 70, 1753.
Benedict, J.V., Harris, E.H. and Von Roseberg, E.U. (1970), An analytical
investigation of the cavitation hypothesis of brain damage, ASME·Paper
70-BHF-3.
Berne, R.M. and Levy, M.N. (1972}, Ca4dlov~c~ Phy~-iology, C.V. Mosby,
St. Louis.
Blumenthal, H.T. (1967}, Hemodynamic factors in the etiology of arther-
iosclerosis, in Cow~y'~ A4th~o~cl~o~~. 2nd ed., Blumenthal, H.T.,
ed., C.C. Thomas, Springfield, Illinois, 510.
Boussinesq, M.J. (1868), Memoire sur 1 'influence des frottements dans les
mouvements resuliers des fluids, J Math Put~ Appl, ser. 2, 13, 377.
Boussinesq,. M.J. (1872), Influence de forces centrifuges sur le mouvement
perm. varie de l'eau dans les canauz larges, Soc Ph-<.lom Butt, 8, 77.
Brech, R. and Bellhouse, B.J. (1973), Flow in branching vessels,
Ca!U:Li.ov~ R~, 7, 593.
400 C.M. Rodk.iewicz
Carlsten, A. and Grimby, G. {1966}, The Cbr..c.ul.a:to!t!J Rupon6e to MU6c.ula!t

Ex~el6e ~n Man, C.C. Thomas, Springfield, Illinois.
Caro, C.G. {1966}, The dispersion of indicator flowing through simplified

models of the circulation and its relevance to velocity profile in
blood vessels, J Phy~~eat, 185, 501.
Caro, C.G. {1973), Transport of material between blood and wall in
arteries, in A.th~ogenu~: I~ng Fac.toM, CIBA Foundation Sym-
posium 12 {new series), Porter, R. and Knight, J., eds., Associated
Scientific Press, Amsterdam, 127.
Caro, C.G., Fitz-Gerald, J.M. and Schroter, R.C. (1971), Atheroma and
arterial wall shear. Observatio~. correlation and proposal of a shear
dependant mass transfer mechanism for atherogenesis, P!toc. Roy Soc. Lon,
Ser. B, 177, 109.
Cassanova, R.A., Giddens, D.B. and Mabon, R.F. {1975), A comparison of
stenotic fluid dynamics in steady and pulsatile flow, 1975 ASME Bio-
mechanics Symposium, New York, 27.
Chandran, K.B., Swanson, W.M. and Ghista, D.N. {1974), Oscillatory flow
in thin-walled curved elastic tubes, Ann Biomed Eng, 2, 392.
Constantinides, P. (1965), Exp~ental A.th~o~c.l~o~~. Elsevier,
New York, 14.
Cox, R.H. (1969), Comparison of linearized wave propagation models for
arterial blood flow analysis, J Biomec.h, 2, 251.
Cox, R.H. {1970), Wave propagation through a Newtonian fluid contained
within a thick-walled, viscoelastic tube: The influence of wall com-
pressibility, J Biomec.h, 3, 317.
Crow, W.J. {1969), Studies of arterial branching models using flow bire-
fringence, Ph.D. Thesis, Univeristy of Florida, Florida.
Cuming, H.G. (1952), The secondary flow in curved pipes, G!t 8!tit A~o Ru
Counc. RepoJt:t& and Memo!tanda, No. 2880.
Daly, B.J. {1976), A numerical study of pulsatile flow through stenosed
canine femoral arteries, J Biomec.h, 9, 465.
Davis, W. and Fox, R.W. (1967), An evaluation of the hydrogen bubble
technique for the quantitative determination of fluid velocities within
clear tubes, J o6 Ba6~c. Eng, T!tan6 ASME, 89, 771.
Dean W.R. (1927), Note on the motion of fluid in a curved pipe, Phit Mag,
4, 208.
References 401
Dean, W.R. (1928), The streamline motion of fluid in a curve pipe, Phil
Ma.g, 5, 674.
Deshpande, M.D., Giddens, D.P. and Mabon, R.F. (1976), Steady laminar
flow through modelled vascular stenoses, J Biomech, 9, 165.
Deutsch, S. and Phillips, W.M. (1971), The use of the Taylor-Couette
stability problem to validate a constitutive equation for blood,
Bionheotogy, 14, 253.
Downie, H.G., Murphy, E.A., Rowsell, H.C. and Mustard, J.F. (1963),
Extracorporeal circulation: A device for the quantitative study of
thrombus formation in flowing blood, C~nc Re6, 12, 441.
Duncan, L.E. (1963), Mechanical factors in the localization of atheromata,
in Evalua.tion on the a.theno~eienotic pia.que, Jones, R.J., ed., Univer-
sity of Chicago Press, Chicago, 171.
Ehrlich, L.l-1. and Friedman, M.H. {1977), Steady convective diffusion in
a bifurcation, IEEE Tna.~ Biomed Eng, BME-24, 12.
Ellard, D., Huth, C. and Scott, A. {1968), A biomedical heat exchanger
for localized cooling of the brain, Unpublished-Design project for
Mec.E. 463, U of Alberta, Edmonton.
Eustice, J. (1910), Flow of water in curved pipes, Pnoc Roy Soc Lon, ser.
A, 84, 107.
Eustice, J. (1911), Experiments on streamline motion in curved pipes,
Pnoc Roy Soc Lon, ser. A, 85, 119.
Evans, R.L., Hosie, K.F., Kooiker, R.H., Perry, J. and Stish, R.J. (1960),
Reflections in model and arterial pulse waves, J Appl Phy~~ol, 15,
258.
Ferguson, G.G. and Roach, M. {1972), Flow conditions at bifurcations as
determined in glass models, with reference to the focal distribution
of vascular lesions, in Ca.ndiav~euta.n Fluid Vy~~, vol. 2, Bergel,
D.H., ed., Academic Press, London, 141.
Fernandez, R.C., DeWitt, K.J. and Botwin, M.R. (1976), Pulsatile flow
through a bifurcation with applications to arterial disease, J Biomech,
9, 575.
Forstram, R.J., Blackshear, P.L. Jr., Dorman, F.D., Kreid, O.K. and
Kihara, K. {1971), Experimental study of a model blood flow in
channels, ASME Paper 71-WA/BHF-5.
Fox, J.A. and Hugh, A.E. (1966), Localization of atheroma: A theory based
on boundary layer separation, BnLt Hea.nt J, 28, 388.
402 C.M. Rodkiewicz
Friedlander, S.K. and Johnstone, H.F. (1957), Deposition of suspended

particles from turbulent gas streams, Ind Eng Chern, 49, 1151.
Fry, D.L. (1968), Acute vascular endothelial changes associated with in-
creased blood velocity gradients, C~e Reo, 22, 165.
Fry, D.L. (1969), Certain histological and chemical respon~.es of the vas-
cular interface to acutely induced mechanical stress in the aorta of
the dog, C~e Reo, 24, 93.
Fry, D.L. (1973), Response of the arterial wall to certain physical fac-
tors, in Ath~ogeneo~: Initiating Facto~, CIBA Foundation Symposium
12 (new series), Porter, R. and Knight, J., eds., Associated Scientific
Press, Amsterdam, 93.
Fry, D.L., Mallos, A.J. and Casper, A.G.T. (1956), A catheter tip method
for measurement of the instantaneous aortic blood velocity, C~e Reo,
4, 627.
Fung, Y.C. (1969), Blood flow in the capillary bed, Prepared for a General
Lecture at the Technical Conference on Biomechanics, U. of Michigan,
Ann Arbor, Michigan.
Fung, Y.C. (1971), Biomechanics: A survey of the blood flow problem, in
Advaneeo .in App.Ued Meehan.iu.,, vol. 11, Yih, C.S., ed., Academic Press,
New York, 65.
Geer, J.C. and McGill, H.C. (1967), The evolution of the fatty streak, in
A.th~Mcl..~o:Ue VMeula!t V~eMe, Brest, A.N. and Moyer, J.H., eds.,
Appleton-Century-Crafts, New York, 8.
Gerrard, J.H. (1971), An experimental investigation of pulsating turbulent
water flow in a tube, J Ftu.id Meeh, 46, 43.
Gilman, S.F. (1955), Pressure losses in divided-flow fittings, Heating
P.ip.ing and~ Cond, 27, part 1, no. 4, 141.
Glagov, S. (1972), Hemodynamic risk factors: Mechanical stress, mural
architecture, medical nutrition and the vulnerability of arteries to
atherosclerosis, in The Pathogeneo~ o6 Ath~o¢cl..~o¢~, Wissler, R.W.
and Geer, J.C., eds., William &Wikins, Baltimore, 164.
Goldsmith, H.L. (1972), The flow of model particles and blood cells and
its relation to thrombogenesis, in Pnogneo¢ .in Hemo¢ta¢~ and Thnom-
bo¢~, vol. 1, Spaet, T.H., ed., Grune and Stratton, New York, 97.
Gorman, J. (1977), A runr.ing argument, The Sueneeo, 17, no. 1, 10.

Gosman, A.D., Vlachos, N.S. and Whitelaw, J.H. (1975), Measurement and
calculation of laminar flow downstream of a bend in a round tube and
Fleferences 403
the prognosis for similar investigations of blood flow in venules,

ASME Paper 75-APMB-7.
Greenfield, H. (1969), The design of artificial heartvalves and other
medical problems by computer, Vata Sheet, Official publication of the
College of Engineering, University of Utah.
Greenfield, J.C. Jr. and Fry, D.L. (1962), Measurement errors in esti-
mating aortic blood velocity by pressure gradient, J Appl Phy¢iol,
17, 1013.
Guest, M.H., Bond, T.P., Cooper, R.G. and Derrick, J.R. (1963), Red blood
cells: Change in shape in capillaries, Scienee, 142, 1319.
Gutstein, W.H. and Schneck, D.J. (1967}, In vitro boundary layer studies
of blood flow in branched tubes, J Ath~o¢el~ Re¢, 7, 295.
Gutstein, W.H., Schneck, D.J. and Marks, J.O. (1968), In vitro studies
of local blood flow disturbance in a region of separation, J Ath~o¢el~
Re¢, 8, 381.
Guyton, A.C. (1960), Function o6 the Human Body, W.B. Saunders,
Philadelphia.
Guyton, A.C. (1971}, Textbook o6 Medieal Phy¢iology, 4th ed., W.B. Saunders,
Philadelphia.
..
Hagen, G. (1839), Uber die bewegung des wassers in engen zylindrischen,
Rohnen Pogg Ann, 46, 423.
Hardung, V. (1962), Propagation of pulse waves in visco-elastic tubings,
in Handbook o6 Phy~iology: C~eulation, Sec. 2, val. 1, Hamilton, W.F.
ed., American Physiological Society, Washington D.C., 107.
Hawthorne, W.R. (1951), Secondary circulation in fluid flow, Pnoe Roy Soe
Lon, ser. A, 206, 374.
Hoeber, T.W. and Hochmuth, R.M. (1970), Measurement of red cell modulus
of elasticity by in vitro and model cell experiments, ASME Paper
70-BHF-4.
Huang, H.K. (1970}, Theoretical analysis of flow patterns in single-file
capillaries, ASME Paper 70-BHF-10.
Hugh, A.E. and O'Malley, A.W. (1975), Correlation of intra-arterial con-
trast stasis with flow patterns at constrictions, branches and bends:
An experimental model, Clin Radial, 26, 505.
Ito, H. (1950), Theory on laminar flows through curved pipes of elliptic
and rectangular cross-sections, The Repont¢ o6 the 1n¢t High Speed
404 C.M. Rodkiewicz
Meeh, Tokohu University, 1, 1.

Kandarpa, K. and Davids, N. {1976), Analysis of fluid dynamic effects on
atherogenesis of branching sites, J Biomeeh, 9, 735.
Kaufman, B., Davey, T.B., Smeloff, E.A., Huntley, A.C. and Miller, G.E.
(1968), Development of mechanical heart assists, ASME Paper
68-WA/BHF-4.
Keele, K.D. (1978), The life and work of William Harvey, Endeavoun, New
Series, 2, No. 3, 104.
Keulegan, G.H. and Beij, K.H. {1937), Pressure losses for fluid flow in
curved pipes, J Re6 Natt Bun Stand, 18, 89.
Kiser, K.M., Falsetti, H.L., Yui, K.H., Resistarits, M.R., Francis, G.P.
and Carroll, R.J. (1976), Measurements of velocity wave forms in the
dog aorta, J Fluid~ Eng, T~~ ASME, 98, 297.
Kline, K.A. and Allen, S.J. (1971), Deformation of red cells in shear
fields, ASME Paper 71-WA/BHF-11.
Krovetz, L.J. {1965), The effect of vessel branching on haemodynamic
stability, Phy~ Med Riot, 10, No. 3, 417.
Krueger, J.W. , Young, D.F. and Cholvin, N.R. (1970), An in vitro study
of flow response by cells, ASME Paper 70-BHF-9.
Kuchar, N.R. and Ostrach, S. (1965), Flows in the entrance region of cir-
cular elastic tubes, FTAS/TR-65-3, Case Western Reserve University,
Cleveland, Ohio.
Kuchar, N.R. and Scala, S.M. (1968), Design of devices for optimum blood
flow, ASME Paper 68-Df-52.
Langhaar, H.L. (1942), Stead)' flow in the transition length of a straight
tube, J Appt Meeh, 9, A55.
Lee, J.S. and Fund, Y.C. (1968), Experiments on blood flow in lung alveoli
models, ASME Paper 68-WA/BHF-2.
Lee, J.S. and Fung, Y.C. (1970), Flow in locally constricted tubes at low
Reynolds numbers, J Appt Meeh, 37, 9.
Lew, H.S. (1973), The dividing streamline of bifurcating flows in a two-
dimensional channel at low Reynolds number, J Biomeeh, 6, 423.
Lighthill, M.J. (1972), Physiological fluid dynamics: A survey, J Fluid
Meeh, 52, 475.
Fleferences 405
Lighthill, Sir J. (1975), Mathematieai Bio6£uiddifnamie6, Soc Indus. Appl.

Math., Philadelphia.
Lyne, W.H. (1970), ~nsteady flow in a curved pipe, J Fluid Meeh, 45, 13.
Lynn, N.S., Fox, V.G. and Ross, L.W. (1970), Paper presented at 63rd
Annual Meeting, AICHE, Chicago.
Malindzak, G.S. (1956), Reflection of pressure pulses in the aorta, Med
Re6 Eng, 6, 4th quarter, 25.
Malindzak, G.S. and Stacy, R.W. (1965), Dynamic behaviour of a mathematical
analog of the normal human arterial system, Am~ J Med Elee, 4, no. 1,
28.
Malindzak, G.S. and Stacy, R.W. (1966), Dynamics of pressure pulse trans-
mission in the aorta, Ann NY Acad Se, 128, Art, 3, 921.
Mark, F.F., Bargeron, C.B. and Friedman, M.H. (1975), An experimental in-
vestigation of laminar flow in a rectangular cross-section bifurcation,
Applied Physics Lab, John Hopkins U, Silver Spring, Maryland.
Martin, J.D. and Clark, M.E. (1966), Theoretical and experimental analyses
of wave reflections in br·anched flexible conduits, Eng Meeh Re6, ASCE,
441.
McDonald, D.A. (1955), The relation of pulsatile pressure to flow in
arteries, J Phif~~ol, 127, 533.
McGill, H.C., Geer, J.C. and Strong, J.P. (1963), Natural history of
human atherosclerotic lesions, in Ath~o~el~o~~ & 1~ O~gin,
Sandler, M. and Bourne, G.H., eds., Academic Press, New York, 39.
Medical College, New York Hospital (1970) Filtering blood may avert by-
pass brain damage, JAMA, 212, 1450.
Meisner, J.E. and Rushmer, R.F. (1973), Eddy formation and turbulence in
flowing liquids, C~e Re6, 12, 455.
Meisner, J.E. and Rushmer, R.F. (1968), Production of sounds in distensible
tubes, C~e Re6, 12, 651.
Mitchell, J.R.A. and Schwartz, C.J. (1965), The localization of arterial
plaques, in Ant~ V~~e, Blackwell Scientific Publications, Oxford
50.
Mueller, T.J., Llyod, J.R. Chetta, G.E. and Galanga, F.L. (1975), Effect
of test section geometry on the occluder motion of caged-ball prosthetic
heat valves, 1975 ASME Biomechanics Symposium, New York, 55.
406 C.M. Rodkiewicz
Mullinger, R.N. and Manley, G. (1967), Glycosaminoglycans and atheroma in

iliac arteries, J Ath~o4cl~ Re&, 7, 401.
Nerem, R.M. and Seed, W.A. (1972}, An in vivo study of aortic flow dis-
turbances, C~va4~ Re&, 6, 1.
o•Brien, V., Ehrlich, L.W. and Friedman, M.H. (1976} Unsteady flow in a
branch, J Fluid M~h, 75, 315.
Park, S.D. and Lee Y. {1971}, Diabatic turbulent flow in the entrance
region of concentric annuli, Eng J, 54, No. 6.
Patel, D.J. and Janicki, J.S. (1966}, Catalogue of some dynamic analogies
used in pulmonary and vascular mechanics, J Med Re& Eng, 5, 30.
Pedley, T.J., Schroter, R.C. and Sudlow, M.F. (1971), Flow and pressure
drop in systems of repeatedly branching tubes, J Fluid M~~, 46, 365.
Pelot, R.P. and Rodkiewicz, C.M. (1982}, Aortic Arch Separation and Flow
Patterns - In Vitro Study, Dept. Report No. 27, Dept. of Mech. Eng.,
University of Alberta, Edmonton.
Poiseuille, J. (1840, 1841), Recherches exp~rimentalles sur le mouvement
des liquids dans les tubes de tres-petits diam~tres, Compte& R~~,
11, 961, 1041, and 12, 112.
Porg~. I.G. {1964). Hemodynamics of the ascending aorta, in Pu£4atite
Blood Flow, Attinger, E.O., ed., McGraw Hill, New York, 237.
Reif, T.H., Nerem, R.H. and Kulacki, F.A. (1976}, An in vitro study of
transendothelial albumin transport in a steady state pipe flow at high
shear rates, J Fluid Eng, T~n& ASME, 98, 488.
Resch, J.A., Okabe, N., Loewenson, R.B:, Kimoto, K., Katsuk-i, S. and
Baker, A.B. (1969}, Pattern of vessel involvement in cerebral athero-
sclerosis, J Ath~o4cl~ Re&, 9, 239.
Rodbard, S. (1956}, Vascular modifications induced by flow, Am~ H~ J,
51, 926.
Rodkiewicz, C.M. (1974}, Atherosclerotic formations in the light of fluid
mechanics, Pkn~ l974 F~ Eng Con6, Montreal.
Rodkiewicz, C.M. (1974}, Bifurcation characteristic Reynolds number,
J Hyc1Jr.a.u.tic. Re&, 12, 241.
Rodkiewicz, C.M. (1975), Localization of early atherosclerotic lesions in
the aortic arch in the light of fluid flow, J·Biom~~ 8, 149;
Rodkiewicz, C.M. and Hung, R. (1976), Flow division dependence of some
large arterial junctions on frequency and amplitude, Departmental
References 407
Report No. 5, Dept. of Mech. Eng., U of Alberta, Edmonton, Alberta.

Rodkiewicz, C.M. and Kalita, W. (1978), Flow characteristics of the curved
pipe junctions, ·Departmental Report No. 12, Dept. of Mech. Eng., U of
Alberta, Edmonton, Alberta.
Rodkiewicz, C.M. and Roussel, C.L. (1973), Fluid mechanics in a large
arterial bifurcation, J F~ Eng, Tnan6 ASME, 95, 108.
Rodkiewicz, C.M. and Wong, S.L. (1974), On the mass flow division in
curved junctions, ASHRAE T~an6, 80, Part 2, 280.
Rogers, V.A. and Moskowitz, G.D. (1971), Analysis for hydrodynamic model
of human systemic arterial circulation, ASME Paper 71-WA/AUT-13.
Rosenhear, L. (1973), Lami~ Bound~y Lay~, Oxford University Press,
London.
Roussel, C.L. and Rodkiewicz, C.M. (1973), Edqe effect in large arterial
bifurcation, J Flt.1idJ.J Eng, Tnan6 ASME, 95, 334.
Rowe, M. (1970), Measurements and computations of flow in pipe bends,
J Fluid Meeh, 43, 771.
Sato, Y. (1963), Separating and uniting flows in a branch pipe, J Japan

Soe Meeh Eng, 66, No. 537.
Scarton, H.A., Shah, P.M. anc Tsapogas, M.J. (1975), The role of hemo-
dynamics in early atheroma in the aorta, 1975 ASME Biomechanics Sym-
posium, New Yor_k, 15.
Scarton, H.A., Shah, P.M. and Tsapogas, M.J. (1977), Relationship of the
spatial evolution of secondary flow in curved tubes to the aortic arch,
in Meehan.ic.J.J -in Eng-ine~n.g, Dubey, R.N. and Lind, N.C., eds., U of
Waterloo Press, Waterloo, Canada, 111.
Scherer, P.~/. (1973), Flow in axisyrr.metrical glass model aneurysms,
J Biomeeh, 6, 695.
Schlichting, H. (1968), Bound~y Lay~ Theo~y, 6th ed., McGraw Hill,

New York.
Schneck, D.J. (1977), Pulsatile blood flow in a channel of small expo-
nential divergence, III - Unsteady flow, J FfuidJ.J Eng, Tlta.n6 ASME,
99, 33.
Schneck, D.J. and Gustein, W.H. (1966), Boundary-layer studies in blood
flow, ASME Paper 66-WA/BHF-4.
Schraub, F.A., Kline, S.J., Henry, J., Runstadler, P.W. Jr. and Littell,
408 C.M. Rodkiewicz
A. {1965}, Use of hy~rogen bubbles for quantitative determination of

time dependant velocity fields in low-speed water flows, J Bah~e
Eng, TJta.YL6 ASME, 87, 429.
Schultz, D.l. {1972), Pressure and flow in large arteries, in C~o
va6e~ Fluid Vy~, vol. 1, Bergel, D.H. ed., Academic Press,
london, 281.
Seed, W.A. and Wood, N.B. {1971), Velocity patterns in the aorta,
C~ova6 Re-6, 5, 319.
Shah, P.M., Tsapogas, M.J. Scarton, H.A., Jindal, P.K. and Wu, K.T. {1976)
Predilection of occlusive disease for the left iliac artery, J
CaJLdiova6 e SuJr.g, 17, 420.
Sharma, M.G. and Hollis, T.H. {1976), Rheological properties of arteries
under normal and experimental hypertensive conditions, J Biomeeh,
9, 293.
Singh, M.P. {1974), Entry flow in a curved pipe, J Fluid Meeh, 65, 517.
Skalak, R. and Branemark, P.I. {1969}, Deformation of red blood cells in
capillaries, S~enee, 164, 717.
Snyder, H.F., Rideout, V.C. and Hillestad, R.J. {1966}, Computer modelling
of the human systemic arterial tree, J Biomeeh, 1, 341.
So, R.M.C. (1976}, Entry flow in curved channels, J Fluid Eng, TJta.YL6
ASME, 98, 305.
Spain, D.M. (1966), Atherosclerosis, Sc. Am, 215, 49.
Spencer, M.P. and Denison, A.B. (1956), The aortic flow pulse as related
to differential pressure, C~e Re-6, 4, 476.
Stehbens, W.E. {1974), Changes in the cross-sectional area of the arterial
fork, Ang~ology, 25, 561.
Stehbens, W.E. {1974), Hemodynamic production of lipid deposition, intimal
tears, mural dissection and thrombosis in the blood vessel wall, P~oe
Roy Soe Lon, ser. B, 185, 357.
Strehler, E. and Schmid, P. {1970), Nomogram for determining normal aortic
diameter (aortic arch) and aortic biological age in 2-m chest x-rays,
in S~enti6~e Table-6, 7th ed., Diem, K. and lentner, C., eds., J.R.
Geigy S.A., Basel, Switzerland.
Szmidt, E.W., Bieliczenko, J.A., Bogatyriew, J.W., Kniaziew, M.D. and
Pokroweskij, A.W. {1973}, The Obliterative occlusions of the carotid
arteries and their surgical treatment {in Russian), K~g~, 49, 3.
References 409
Taylor, G.I. (1929), The criterion for turbulence in curved pipes, PJtoc.
Roy Soc. Lon, Ser. A, 124, 243.
Taylor, M.G. (1957),· An approach to an analysis of the arterial pulse
wave, Phy~ Med Biol, 1, 258.
Taylor, M.G. (1963), Wave travel in arteries and the design of the car-
diovascular system, in Ptt.iJ.Ja.:tUe stood Ftow, Attinqer, E.0., ed.,
~1cGraw Hi 11 , New York, 343.
Texan, M. (1963), The role of vascular dynamics in the development of

atherosclerosis, in Athe~to~deJto~.Lo a.nd Iu 0Jt,[gin, Sandler, M. and
Bourne, G.H., eds., Academic Press, New York, 167.
Texan, M. (1967), Mechanical factors involved in atherosclerosis, in
AtheJtMdeJto:tic. V~c.ulalt V.Lo~e, Brest, A.N. and Moyer, J.H., eds.,
Appletcn-Century-Crafts, New York, 23.
Texan, M. (1972), The hemodynamic basis of atherosclerosis, further ob-
servations: The ostial lesson, Butt N Y Acc.d Med, 48, 733.
Texan, M. (1974), Atherosclerosis: Its hemodynamic basis and implications,
Sympo~ium on AtheJto~deJto~.Lo, Medical Clinics of North AMerica, 58,
No.2, 257.
Thompson, J. (1876), On the orioin of windings of rivers in alluvial
plains with remarks on the flow of water round bencs in pipes, PJtoc.
Roy Soc. Lon, 25, 5.
Thompson, J. (1877), Experimental demonstration in respect to the origin
of windinq of rivers in alluvial plains, and to the mode of flow of
water round bends of pipes, PJtoc Roy Soc Lon, 26, 356.
Thompson, J. (1878), On the flow of water in uniform r~gime in rivers and
other open channels, PJtoc Roy Soc. Lon, 28, 114.
Thompson, J. (1879), Flow round river bends, PJtov I~t Mech Eng~, 2, 456.
Thurston, G.B. (1975), Viscoelastic resonance and impedance for blood flow
in larqe tubes, 1975 AS~1E Biomechanics Symposium, New York, 39.
Timm, C. (1942), Der stromungsverlauf in einem modell der menschichen
aorta, Z Biol, 101, 79.
Turnstall, M.J. and Harvey, J.K. (1968), On the effect of a sharp bend in
a fully developed turbulent pipe flow, J Fluid Mec.h, 34, 595.
Tuttle, W.W. and Schottelius, B.A. {1965), Textbook o6 Phy~iology, 15th
ed., C.V. Mosby, St. Louis.
410 C.M. Rodkiewicz
Uchida, S. {1956), The pulsating viscous flow superposed on the steady

laminar motion of incompressible fluid in a circular pipe, Z Ang~ndte
Math Phyll.ie!A, 7, 403.
Vazsonyi, A. (1944), Pressure loss in elbows and duct branches, T~
ASME, 66 , 177 .
l·lalton, K.W. and Williamson, N. {1968), Histological and irrmunofluorescent

studies on the evolution of the human atheromatous plaque, J Ath~o.llel~
Rell, 8, 599.
Wells, H.K., Winter, D.C., Ne~lscn, A.W. and McCarthy, T.C. (1977), Blood
velocity patterns in coronary arteries, J Biomeeh Eng, T~nll ASME,
99, 26.
Wesolowski, S.A., Fries, C.C., Sabini, A.M. and Sawyer, P.N. (1962), The
significance of turbulence in hemic systems and in the distribution
of the atherosclerotic lesion, SUkg~y, 57, 155.
Wesolowski, S.A., Fries, C.C., Sabini, A.M. and Sawyer, P.N. {1965),
Turbulence, intimal injury and atherosclerosis, in Biophyll.ieal
Meeha.n.illrn.6 .in V.Meuialt Homeo.6:t.a..6.i.6 a.nd 1n:tlta.vMc.ulaJc. Th!Lombo.ll.ill,
Sawyer, P.N., ed., Appleton-Century-Crofts, New York, 147.
White, C.M. (1929), Streamline flow through curved pipes, P~oe Roy Soe
Lon, ser, A, 123, 645.
White, F.M. (1974}, V~eoU6 Fluid Flow, McGraw Hill, New York.
Whitmore, R.L. (1968), Rheology o6 the C~eulation, Pergamon Press,
New York.
Wieting, D.W. (1968), A method of analysing the dynamic flow characteris-
tics of prosthetic heart valves, ASME Paper 68-WA/BHF-3.
Wieting, D.W., Akers, W.W., Feola, M., and Kennedy, J.H. (1970), Analysis
of a variable volume intra-aortic balloon pump in a mock circulatory
system, ASME Paper 70-BHF-5.
Wintrobe, M.M. (1967), Cl.in.iea.l Hematology, Lea and Febiger, Philadelphia.
Wormersly, J.R. (1955), Method for the calculation of velocity, rate of
flow and viscous drag in arteries wnen the pressure gradient is known,
J Phy.ll.iol, 127, 553.
Yao, L.W. and Berger, S.A. (1975}, Entry flow in a curved pipe, J Fluid
Meeh, 67, 177.
Yellin, E.L., Peskin, C.S. and Frater, R.W.M. {1972), Pulsatile flow
across the mitral valve: Hydraulic, electronic and digital computer
References 411
simulation. ASME Paper 72-WA/BHF-10.

Young. D.F. ~nd Tsai, F.Y. (1972), Flow characteristics in models of
arterial stenoses, I. - Steady flow. II. - Unsteady flow, J B£omeeh,
6, 395 and 547.
Zareckij, W.W., Sandrikow, W.A., Wychowskaja. A.G., Sablin, J.N. and
Lemieniew, J.N. (1973), The study of the blood flow in the pathologic
structures of the peripheral blood vessels (in Russian). K~g~,
49, 24.
Zechmeister, A. (1969), Calcification of epicardial stretches of bridged
coronary arteries in man, J Ath~o~ct~ R~, 9, 121.
CONTRIBUTORS
The experience and background of the contributors differs widely

and a brief biography of each is provided to give one an awareness of
the qualifications of the authors.
Daniel Quemada of the Universite Paris 7, France, contributed
Chapter I. He is full professor of Physical Hydrodynamics and Biomech-
anics at the University of Paris VII and head of the "Laboratory of
Biorheology and Physico-Chemical Hydrodynamics" sponsored by the C.N.R.S.
He is a former student of the Ecole Normale Superi~ure, one of the
most prestigious centers of high education in France, where he completed
its curriculum in 1954. He owns, since 1964, a Ph. D. (Docteur d 1 Etat)
on plasma physics from the University of Orsay and became Assistant
Professor in the Physics Department of the University of Paris in 1966.
Between 1964 and 1972, he conducted research on plasma dynamics in the
414 C.M. Rodkiewicz
outer space with special interest in MHO models of magnetosphere. As a

result of this work, he authored a book on "Waves in Plasma" (in French)
and a number of scientific papers on the subject.
Since 1971, he directed his interest towards problems related with
biorheology and biomechanics and developed a research group whose con-
tributions to this matter has become recognized worldwide.
He is a member of the board of editors of Biorheology, of the
British Society of Rheology, of the Soci~t~ Francaise de Physique and of
the Soci~t~ de Chimie-Physique. His contribution to the field of blood
flow comprises more than twenty papers treating from the pulsatory flow
in large vessels to the modelization of microflow of dispersed systems
of deformable particles.
Joseph Barbenel of the University of Strathclyde, Scotland, con-
tributed Chapter II. He received his B.D.S. London, L.D.S., R.C.S.
(Eng.) in 1960 with a prize in Anatomy and Dental Anatomy. From 1960 to
1962 he served as Lt. R.A.D.C. and later Capt. R.A.D.C. in the National
Service in Malaya. During those two years, he provided a complete ser-
vice for Army personnel and families. In addition, he was in charge of
the dental departments of BMH, Kinrara and BMH, Kamunting for approxim-
ately 7 months.
Dr. Barbenel joined the University of St. Andrews in Dundee as a
student in 1963, and graduated in 1966 with a B.Sc., receiving a Junior
Honours Physics Medal as well. He became a student at the Bioengineer-
ing Unit of the University of Strathclyde in 1966, and in 1967 received
his M.Sc., (Strathclyde) with a MRC award for Further Education in the
Contributors 415
Medical Sciences. From 1967 to 1969 he was a lecturer at the Dept. of

Dental Prosthetics, Dental School and Hospital, University of Dundee.
Dr. Barbenel became Senior Lecturer at the Bioengineering Unit, of
Stratclyde in 1970, and in 1973 he was Head of Tissue Mechanics Divis-
ion. In that year he also became a M. Inst.P. He received his Ph.D.
(Strathclyde) in 1979. From November to December 1979, he was visiting
Professor at the National Engineering University, Lima, Peru.
Jim B. Haddow of the University of Alberta, Canada, is the princi-
pal author of Chapter III. He took his B.Sc. a~ the University of St.
Andrews, Scotland, in 1951 and his Ph.D. degree in Engineering at the
University of Manchester in 1960. From 1955 - 58 and from 1961 to the
present he has been a member of the Mechanical Engineering Department at
the University of Alberta except for one year as a visiting professor at
the University of New South Wa 1es.
The co-authors with Dr. Haddow are B. Moodie and R.J. Tait, both
also with the University of Alberta.
Bryant Moodie received his B.Sc. degree in mathematics from Carl-
ton University, Ottawa and his M.Sc. and Ph.D. degrees in mathematics
from the University of Toronto. From 1970 - 72 he was a research as-
sociate at Carleton University, from 1972 - 73 a visiting lecturer at
the University of Dundee, Scotland and is presently Associate Professor
of Mathematics at the University of Alberta, Edmonton and recipient of a
McCalla Research Professorship.
R.J. Tait took his Ph.D. degree in applied mathematics at Glasgow
with I.N. Sneddon in 1962. He joined the department at Edmonton the
416 Contributors
same year and apart from a year at Norman, Oklahoma, and another at the
Mathematics Institute at Warwick is, at present, still there.
Thomas Kenner of the Physiologisches Institut der Universitat Graz
contributed chapter IV. He is full professor and chairman of the phys-
iological institute in Graz. He received his M.D. in 1956 at the Uni-
versity of Vienna, Austria, and has since that time performed clinical
studies and experimental and theoretical research on various aspects of
applied and theoretical hemodynamics. Together with E. Wetterer he pub-
lished a monography on the arterial pulse in 1968. From 1968 to 1971
he performed research on circulatory control at the division of biomed-
ical engineering of the University of Virginia. In 1972, he was appo-
inted full professor and chairman at the position he still holds in
Graz. His int~rest is currently directed towards the relation between
dynamics and conti·ol of blood flow, towards the analysis of stochastic
phenomena in the circulation, and towards the application of a new
method for the continuous measurement of blood density for the analysis
of fluid exchange in the microcirculation.
Czelaw M. Rodkiewicz of the University of Alberta, Canada, contrib-
uted Chapter V. He is a full professor in the Mechanical Engineering
Department and has conducted research on various aspects of blood flow.
He received his Diploma in Mechanical Engineering at the Polish Univer-
sity College in London, Great Britain, and his M.Sc. degree at the
University of Illinois, U.S.A. He earned his Ph.D. degree at the Case
Institute of Technology in Cleveland, Ohio, U.S.A.
Dr. Rodkiewicz has done extensive research into fluid mechanics
Contributors 417
and heat transfer, specifically in the domains of hypersonic flight,

lubrication and blood flow. He has written numerous papers on the sub-
ject and given lectures at various conferences and invited lectures and
tours in mafly countrie·s. Dr. Rodkiewicz is listed in the "Who•s Who in
the World" and many other biographical books.

(International Centre For Mechanical Sciences 270) C. M. Rodkiewicz (Eds.) - Arteries and Arterial Blood Flow-Springer-Verlag Wien (1983)

Uploaded by

Copyright:

Available Formats

(International Centre For Mechanical Sciences 270) C. M. Rodkiewicz (Eds.) - Arteries and Arterial Blood Flow-Springer-Verlag Wien (1983)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(International Centre For Mechanical Sciences 270) C. M. Rodkiewicz (Eds.) - Arteries and Arterial Blood Flow-Springer-Verlag Wien (1983)

Uploaded by

Copyright:

Available Formats

INTERNATIONAL CENTRE FOR MECHANICAL SCIENCES

COURSES AND LECTURES - No. 270

SPRINGER-VERLAG WIEN GMBH

This work is subject to copyright.

whether the whole or part of the material is concerned

© 1983 by Springer-Verlag Wien

Originally published by Springer-Verlag Wien New York in 1983

ISBN 978-3-211-81635-6 ISBN 978-3-7091-4342-1 (eBook)

CHAPTER I BLOOD RHEOLOGY AND ITS IMPLICATION IN FLOW OF BLOOD

3.1 Variables governing rheological properties of blood

CHAPTER II THE ARTERIAL WALL 129

1. GEOMETRY AND STRUCTURE OF ARTERIES

2. EXPE.RIMENTAL MECHANICS OF THE ARTERIAL WALL

CHAPTER III DYNAMICS OF FLUID FILLED TUBES 179

2.5 Solution of Fluid Equations 190

CHAPTER IV SMALL ARTERIES AND THE INTERACTION WITH THE

4.3 Humoral autoregulation 295

CHAPTER V FLOW IN LARGE ARTERIES 327

4.2 Fully Developed Flow in a Curved Pipe 358

SUMMARY : Rheological behaviour of blood lS studied as dominated by

1 - GENERAL FEATURES OF BLOOD CIRCULATION IN NARROW VESSELS.

One goal of hemorheology is the analysis of direct effects that ma-

1.2 Constituents of blood.

TABLE 1.1- BLOOD CONSTITUENTS (5.106 Particles/mm3)

( 0 ) alternatively called, erythrocytes, leucocytes and thrombocytes,

effects and therefore must be considered as non-existing. However, it

Fig.1.1. Viscosity of RBC

C. The red cell.

pitals, this height is currently

Although these sedimentation effects

D. Pl asma proteins and coagulation

into contact with extracorporeal surfaces. The f ormation of a clot would

Fi g .1 . 3 Phot omicrographs of r oule aux of

1 .3. Main factors in blood circulation and continuum model

a) The capillary circulation ( 0 ) (~ ~ 1) 1n a very large number

§ 1 .3a, the problem is confined to the mechanics of deformable bodies im-

Anyway, using such two-phase flow models requires the knowledge of

1 .4. Basic rheological concepts.

( 0 ) This annular flow can be called a sheared-core.

zation. These conditions are achieved in viscometric flows, as simple

B) Steady shear flows : shear viscosity.

the(constant) coefficient being the viscosity (in Poises):

Fluids to which ( 1 . 1) applies are called Newtonian fluids, the vis-

Fig . 1 .5 Relative apparent viscosity nra of

B. 2 Non newtonian viscosity .

Such fluids are called non-newtonian fluids, in which additional for-

..·· .. .:1· "1 (X)

As most of these fluids are disperse systems, their pseudoplastic beha-

c) Steady elongational (or extensional) flow.

For Newtonian fluids, nE = 3n , n being the (simple) shear viscosi-

d) Unsteady shear flows: thixotropy and viscoelasticity

d 1 ) The response of a shear thinning material to the application

having a viscosity n0 = n(y= 0) (or like a gel if the structure at

a1 = n1Y1· This structure adaptation involves relaxation times, which cha-

Fig.1 .8 Different types of non-newtonian behaviour under a step

If the shear stress - shear rate relaxation a = a( y) lS not linear,

d2) Thixotropy is not the only time behaviour currently observed. A

ted with a purely elastic (instantaneous) response, followed by a delayed

This leads to introduce a dynamical viscosity as a complex number